CELL

Molecular Cell
Previews
Histone Demethylation and Timely DNA Replication
Erica L. Gerace1 and Danesh Moazed1,2,*
of Cell Biology Hughes Medical Institute Harvard Medical School, Boston, MA 02115, USA *Correspondence: danesh@hms.harvard.edu DOI 10.1016/j.molcel.2010.11.036
2Howard 1Department
It is well-established that silent regions of the genome replicate late during S phase. In this issue of Molecular Cell, Black et al. (2010) uncover a conserved role for the JMJD2 family of histone demethylases in promoting replication within silent chromatin regions that contain histone H3 lysine 9 methylation and HP1.
The faithful replication of the genome during each cell division is a vital process that is highly coordinated and tightly controlled. In eukaryotic cells, origins of DNA replication, the sites where replication begins, are found throughout the genome and re at different times throughout S phase. This timing is inuenced by local chromatin structure and correlates with cell type-specic differences in gene expression and chromatin structure. Although there are exceptions, in general, transcriptionally active portions of the genome are replicated earlier in S phase than those found in silent, heterochromatic regions (MacAlpine and Bell, 2005). Histone H3 lysine 9 (H3K9) methylation is a conserved modication that is found in heterochromatin from ssion yeast to human, and denes a binding site for Heterochromatin Protein 1 (HP1) family members, which help make heterochromatin inaccessible. Recent studies suggest that heterochromatin is more dynamic than previously appreciated and can be broached by machineries that transcribe and repair DNA (Kwon and Workman, 2008). However, how access to heterochromatin or other silent regions is regulated is poorly understood. The discovery of histone lysine demethylases suggested new mechanisms for regulation of chromatin structure and, indeed, these enzymes have already been identied with numerous roles in regulation of transcription. Histone lysine demethylation is catalyzed by two distinct classes of evolutionarily conserved enzymes (Shi and Whetstine, 2007). JmjC domain proteins are metalloenzymes that catalyze oxidative reactions. Several years ago, a JmjC domain protein of the JMJD2 family was found to catalyze the demethylation of trimethylated H3K9, H3K36, and H1.4K26 (Fodor et al., 2006; Klose et al., 2006; Trojer et al., 2009; Whetstine et al., 2006). Depletion of JMJD-2 in C. elegans results in the activation of a DNA damage-induced apoptosis (Whetstine et al., 2006), but the molecular basis of this event had not been established. In this issue of Molecular Cell, Black et al. (2010) describe a role for histone demethylation in cell cycle progression and DNA replication. They report that the expression of JMJD2A is cell cycle regulated with peak expression at the G1/S transition dropping off in S phase and lowest at G2/M. The overexpression of JMJD2A, but not catalytically inactive JMJD2A, which has a mutation at a key metal-coordinating residue required for demethylase activity (Whetstine et al., 2006), in human cell lines led to faster progression through S phase. In addition, the overexpression of JMJD2A resulted in both the early replication of a latereplicating satellite region, sat2 of chromosome 1 (Chr1 sat2), as well as faster recovery of these cells after treatment with hydroxyurea (HU), which causes replication stress. Similarly, mutating or knocking down the gene for the worm homolog, jmjd-2, resulted in increased sensitivity to HU. These results suggest that expression of JMJD2A/JMJD-2 during S phase facilitates DNA replication. Consistently, Black et al. also demonstrated that more single-stranded DNA was observed during S phase in cells ectopically expressing JMJD2A, which suggests increased presence of replication forks. However, it remains unknown whether the faster S phase effects are at the level of initiation and/or fork progression. Analysis of jmjd-2/ worms revealed a decreased number of cells within the mitotic zone and an increased number of RAD-51 foci, an indication of stalled or collapsed replication forks. Furthermore, direct in vivo visualization of DNA replication using pulse-chase labeling experiments with cy3-dUTP/ALEXA-488-dUTP in the adult germlines of jmjd-2/ worms showed that replication was slowed, suggesting that in vivo, JMJD-2 likely affects replication timing. This observation also explains the previously observed increased apoptosis in jmjd-2 mutants (Whetstine et al., 2006), a phenotype that Black et al. (2010) also show can be rescued by knockdown of the p53 and ATR homologs, CEP-1 and ATL-1, respectively. However, the mutation of the ATM homolog did not provide a rescue of the observed phenotype, indicating that the increased apoptosis is mediated through the ATR pathway, which is activated after replication stress, but not the ATM pathway, which is typically activated in response to double-strand breaks. How does JMJD2A-mediated histone demethylation contribute to DNA replication timing? The overexpression of JMJD2A causes the redistribution of HP1 proteins (Klose et al., 2006) likely by removing the chromatin mark to which HP1 proteins bind, trimethylated H3K9. So, potentially JMJD2A may regulate chromatin structure and replication timing via effects on HP1 localization. Indeed, using micrococcal nuclease digestions of isolated nuclei, Black et al. (2010) observed increased DNA accessibility in cells overexpressing catalytically active JMJD2A, specically during S phase. In
Molecular Cell 40, December 10, 2010 2010 Elsevier Inc. 683
Previews
Molecular Cell
addition, in these cells, the regions and the mating type (1) wild-type chromatin structure of the locus (Hayashi et al., 2009). JMJD2A late-replicating locus Chr1 Surprisingly, Black et al. sat2 was found to be more (2010) found that deletion of HP1 HP1 H3K9me3 open corresponding to its the worm HP1g homolog observed earlier replication. (HPL-2) had the same deori These data support the nolayed replication phenotypes tion that altering chromatin as jmjd-2/ cells, suggesting accessibility can regulate that intricacies in the balance (2) JMJD-2 depletion cell cycle progression and between the two worm HP1 replication timing, providing proteins are also important another way to regulate cell for proper replication timing. HP1 HP1 HP1 cycle that is independent or We can therefore look forcomplimented by altered ward to future studies that transcriptional programs. will explore possible direct ori ori Interestingly, it appears that links between different HP1 JMJD2A/JMJD-2 act by proteins and activation or antagonizing a specic HP1 silencing of DNA replication. (3) JMJD2A overexpression isoform. Most cells contain two or more HP1 homologs, REFERENCES JMJD2A HP1a, -b, and HP1g in human, HP1 JMJD2A and HPL-1 and -2 in worms. HP1 Black, J.C., Allen, A., Capucine, The overexpression of HP1g, V.R., Forbes, E., Longworth, M., Tschop, K., Rinehart, C., Quiton, but not HP1a or HP1b, supJ., Walsh, R., Smallwood, A., et al. pressed the JMJD2A-depenori (2010). Mol. Cell 40, this issue, dent increased number of 736748. cells in late S phase, an Fodor, B.D., Kubicek, S., Figure 1. Human and C. elegans JMJD2A/JMJD-2 Antagonize effect that depended on Yonezawa, M., OSullivan, R.J., HP1g/HPL-2 and Promote DNAReplication both the chromodomain and Sengupta, R., Perez-Burgos, L., (1) HP1g binds trimethylated histone H3 lysine 9 (H3K9me3), maintaining Opravil, S., Mechtler, K., Schotta, chromoshadow domain of a closed chromatin structure, which inhibits the ring of origins of replication G., and Jenuwein, T. (2006). Genes or slows the progress of replication forks. Demethylation of H3K9me3 results HP1g. Also, at the Chr1 sat2 Dev. 20, 15571562. in dissociation of HP1g, thus promoting DNA replication and cell cycle locus, the overexpression of progression. (2) In JMJD-2-depleted germline nuclei in C. elegans, H3K9me3 Hayashi, M.T., Takahashi, T.S., JMJD2A decreased HP1g levels and HPL-2/HP1g association increase, leading to delayed replication Nakagawa, T., Nakayama, J., and origin ring. Potential spreading of HPL-2/HP1g may also prevent the ring localization. In worms, the Masukata, H. (2009). Nat. Cell Biol. of distally located origins of replication. (3) During S phase or when JMJD2A / jmjd-2 phenotypes of in11, 357362. is overexpressed, demethylation of H3K9me3 and HP1g displacement allow creased RAD-51 foci, deorigin ring resulting in faster progression through S phase. Klose, R.J., Yamane, K., Bae, Y., creased number of nuclei in Zhang, D., Erdjument-Bromage, the mitotic zone, slowed replication, and at specic heterochromatic regions, anal- H., Tempst, P., Wong, J., and Zhang, Y. (2006). increased apoptosis are all rescued by ogous to human satellite repeats. The Nature 442, 312316. the depletion of HP1 homolog HPL-2, ndings also add to an emerging body of Kwon, S.H., and Workman, J.L. (2008). Mol. Cells establishing a conserved antagonistic evidence that suggests distinct roles for 26, 217227. relationship between these two proteins. different HP1 proteins in regulation of MacAlpine, D.M., and Bell, S.P. (2005). ChromoTogether the results suggest that DNA replication and other heterochro- some Res. 13, 309326. JMJD2A/JMJD-2 modulate replication matin functions, which may relate to Motamedi, M.R., Hong, E.J., Li, X., Gerber, S., timing by opposing reduced DNA acces- protein interactions and/or genomic loca- Denison, C., Gygi, S., and Moazed, D. (2008). sibility to the replication machinery in tion. For example, the two ssion yeast Mol. Cell 32, 778790. DNA domains that are decorated by homologs, Swi6 and Chp2, play distinct Shi, Y., and Whetstine, J.R. (2007). Mol. Cell 25, 114. H3K9 trimethylation and HP1g (Figure 1). roles in replication of heterochromatin It remains to be determined whether the and gene silencing (Hayashi et al., 2009; Trojer, P., Zhang, J., Yonezawa, M., Schmidt, A., Zheng, H., Jenuwein, T., and Reinberg, D. (2009). activation of DNA damage checkpoint Motamedi et al., 2008). Swi6HP1 speci- J. Biol. Chem. 284, 83958405. / and apoptosis in jmjd-2 cells result cally recruits the Cdc7 kinase to promote Whetstine, J.R., Nottke, A., Lan, F., Huarte, M., from a widespread failure to complete origin ring and early replication of het- Smolikov, S., Chen, Z., Spooner, E., Li, E., Zhang, DNA replication or from a failure to do so erochromatin at pericentromeric DNA G., Colaiacovo, M., et al. (2006). Cell 125, 467481.
eM eM eM eM eM Me
684 Molecular Cell 40, December 10, 2010 2010 Elsevier Inc.
Molecular Cell
Previews
Hunting for Alternative Disulde Bond Formation Pathways: Endoplasmic Reticulum Janitor Turns Professor and Teaches a Lesson
Deborah Fass1,*
1Department of Structural Biology, Weizmann Institute of Science, Rehovot, Israel 76100 *Correspondence: deborah.fass@weizmann.ac.il DOI 10.1016/j.molcel.2010.11.034
In this issue of Molecular Cell, Ron and colleagues (Zito et al., 2010b) show that an enzyme responsible for cleaning up hydrogen peroxide in the endoplasmic reticulum can contribute productively to disulde bond formation.
Since the discovery of the sulfhydryl oxidase enzyme Ero1 in yeast more than a decade ago, our notion of oxidative protein folding in the endoplasmic reticulum (ER) has been the following: Ero1, by reducing molecular oxygen to hydrogen peroxide, generates disulde bonds de novo. Ero1 then oxidizes protein disulde isomerase (PDI), and PDI in turn oxidizes diverse ER substrates. Mammals have two Ero1 paralogs, Ero1a and Ero1b, and approximately 20 PDI family oxidoreductases, many of which may be oxidized by Ero1 enzymes (Schulman et al., 2010). RNA transcripts corresponding to both Ero1a and Ero1b are distributed broadly across human tissue types, with Ero1b expressed at particularly high levels in the pancreas (Pagani et al., 2000). The presence of Ero1 in many cell types is consistent with a fundamental role for Ero1 in oxidative protein folding in mammals, as has been demonstrated for yeast. The recent observation that viable mice can be produced when both Ero1a and Ero1b are disrupted thus came as a great surprise (Zito et al., 2010a). These mice are diabetic due to defects in insulin production and secretion, but numerous other essential functions requiring disulde bond formation seem to proceed apace. With the nding that the bestknown catalysts of disulde bond formation in the mammalian ER are less essential than anticipated, the search was on for alternative mechanisms to keep oxidative protein folding going in the absence of the Ero1 enzymes. In fact, several other known enzymes that generate disuldes are candidates to explain the viability of the Ero1 double-knockout mice. Some of these enzymes are conveniently ER localized and have already been shown to interface with ER oxidoreductases. For example, the membrane-embedded protein vitamin K epoxide reductase (VKOR) reduces vitamin K by concomitant oxidation of PDI family members in the ER (Wajih et al., 2007). Though VKOR or a VKOR paralog may in principle take some of the reduced thiol load off an Ero1-depleted ER, it was recently shown that VKOR interacts specically with membraneanchored PDI family members and much less with luminal PDI-like proteins such as PDI itself (Schulman et al., 2010). Another candidate disulde catalyst is quiescin sulfhydryl oxidase (QSOX). QSOX is primarily localized to the Golgi apparatus (Mairet-Coello et al., 2004), but it presumably traverses the ER to get there. In vitro, QSOX is a poor catalyst of PDI-family protein oxidation but an effective oxidant of unfolded proteins (Kodali and Thorpe, 2010). Subtle changes in QSOX levels or trafcking might rescue an underoxidizing ER. Rather than examining known disulde catalysts, Zito et al. (2010b) instead went shing. Using a PDI mutant with a single active-site cysteine as bait, they angled for new potential PDI oxidants, which they identied by mass spectrometry. The idea behind this experiment, which has been used to identify other dithiol/disulde exchange pathways in the past, is that disulde-bonded complexes can form between mutant PDI and its potential oxidants, but the complexes have trouble disassembling without a second PDI cysteine to attack the interprotein disulde bond and liberate newly oxidized PDI. In wild-type cells, both Ero1a and Ero1b were trapped in complex with PDI as expected (Zito et al., 2010b). Another major hit was peroxiredoxin IV (PRDX4). PRDX4 was previously shown to be ER localized and, in its janitorial role, to break down hydrogen peroxide generated by Ero1a activity (Tavender et al., 2008; Tavender and Bulleid, 2010). After hooking PRDX4 with the PDI mutant, though, Zito et al. (2010b) asked whether cells derived from the Ero1 double-knockout mice might be hypersensitive to depletion of PRDX4. If the sole role of PRDX4 were to clean up peroxide produced by Ero1 activity, then Ero1 double-knockout cells should be less dependent on PRDX4. However, if PRDX4 contributes productively to disulde formation in cells lacking Ero1 enzymes, then these cells should be more sensitive to depleting PRDX4 activity. Indeed, by a number of measures, depletion of PRDX4 exacerbated the modest defects in disulde formation seen in the Ero1a/Ero1b knockout cells, precisely the effect one would expect if PRDX4 is an important disulde catalyst in this context (Figure 1). Participation of PRDX4 in disulde formation via PDI is supported by a study in press demonstrating rapid oxidation of PDI family proteins by PRDX4 in vitro (Tavender et al., 2010). In wild-type cells, the ability of a partnership between Ero1 and PRDX4 to get two disuldes out of a single oxygen molecule seems a paradigm of efciency. Ero1 generates one disulde and a molecule of hydrogen peroxide, and PRDX4 uses
Previews
Molecular Cell
occurs or is regulated, and a thorough analysis of the levels of reductionoxidation-active small molecules in the Ero1 knockout mice remains to be performed. Finally, enzymes such as VKOR and QSOX, which do not directly interact with PDI, may nevertheless affect the PDI redox state indirectly. They may channel oxidizing equivalents to the glutathione pool through other PDI family proteins, through re-reduction of nonnative disuldes in folding proteins, or through as yet undiscovered pathways.
REFERENCES Kodali, V.K., and Thorpe, C. (2010). Antioxid. Redox Signal. 13, 12171230. Mairet-Coello, G., Tury, A., Esnard-Feve, A., Fellmann, D., Risold, P.Y., and Griffond, B. (2004). J. Comp. Neurol. 473, 334363.
Figure 1. Allegory for the Newly Discovered Activity of PRDX4 in Promoting Disulde Formation in Cells Lacking Ero1
The sulfhydryl oxidase Ero1 (sloppy professor) reduces oxygen to hydrogen peroxide to drive oxidation of PDI. Ero1 thereby generates a potential mess of reactive oxygen species as a byproduct of disulde bond formation in the secretory pathway. PRDX4 (perceptive janitor) was previously identied as an ERlocalized peroxiredoxin, an enzyme that cleans up hydrogen peroxide by further reducing it to water. Such peroxireductase activity by PRDX4 requires a source of electrons. PRDX4 could gain those electrons by oxidizing PDI and thus contribute to an Ero1-independent pathway for oxidizing secretory proteins albeit requiring another source of hydrogen peroxide.
Pagani, M., Fabbri, M., Benedetti, C., Fassio, A., Pilati, S., Bulleid, N.J., Cabibbo, A., and Sitia, R. (2000). J. Biol. Chem. 275, 2368523692. Schulman, S., Wang, B., Li, W., and Rapoport, T.A. (2010). Proc. Natl. Acad. Sci. USA 107, 15027 15032. Tavender, T.J., and Bulleid, N.J. (2010). J. Cell Sci. 123, 26722679. Tavender, T.J., Sheppard, A.M., and Bulleid, N.J. (2008). Biochem. J. 411, 191199. Tavender, T.J., Springate, J.J., and Bulleid, N.J. (2010). EMBO J., in press. Published online November 5, 2010. 10.1038/emboj.2010.273. Wajih, N., Hutson, S.M., and Wallin, R. (2007). J. Biol. Chem. 282, 26262635. Zito, E., Chin, K.T., Blais, J., Harding, H.P., and Ron, D. (2010a). J. Cell Biol. 188, 821832. Zito, E., Melo, E.P., Yang, Y., Wahlander, A., Neubert, T.A., and Ron, D. (2010b). Mol. Cell 40, this issue, 787797.
the peroxide to generate a second disulde. When PRDX4 functions as a backup disulde catalyst in the absence of Ero1a and Ero1b, however, it has to use another source of peroxide. Identifying that source was beyond the scope of the current study, leaving a future challenge for cell biologists. Although in some senses the method used by Zito et al (2010b) to nd alternative ER oxidants was unbiased, it still
assumes a central and direct role for PDI. Might there be other contributors to protein oxidation in the ER that do not operate through PDI? If so, the strategy used in this study would miss them. One possible mechanism to compensate for defects in processing a reducing load on the ER would be to decrease glutathione synthesis or to curtail glutathione import into the ER lumen. Currently, little is known about how glutathione import
Molecular Cell
Previews
Noxa: A Sweet Twist to Survival and More
Alfredo Gimenez-Cassina1,2 and Nika N. Danial1,2,*
of Cancer Biology and Division of Metabolism and Chronic Disease, Dana-Farber Cancer Institute, Boston, MA 02115, USA of Pathology, Harvard Medical School, Boston, MA 02115, USA *Correspondence: nika_danial@dfci.harvard.edu DOI 10.1016/j.molcel.2010.11.037
2Department 1Department
The BCL-2 family member Noxa induces apoptosis by antagonizing the prosurvival protein MCL-1. In this issue of Molecular Cell, Lowman et al. (2010) uncover a glucose-dependent phosphoregulatory mechanism that inactivates Noxas apoptotic function and triggers its capacity to modulate glucose metabolism.
Mounting evidence in recent years indicates that the homeostatic crosstalk between nutrient metabolism and apoptosis is governed by shared regulatory components. For example, metabolic byproducts of nutrient breakdown can serve as signaling messengers that modulate survival pathways, and proteins with canonical roles in cell death/survival may have additional physiologic functions in nutrient metabolism (Sengupta et al., 2010; Vander Heiden et al., 2009). These cross-regulatory mechanisms ultimately guarantee that cellular survival, division, and repair are attuned to cellular metabolic ux and energy charge. The report by Lowman et al. in this issue describes a glucose-sensitive signaling pathway that phosphorylates proapoptotic Noxa and imparts cytoprotective effects by inactivating its apoptotic function and enhancing glucose utilization (Lowman et al., 2010). The complex and selective networks of interactions among the pro- and antiapoptotic members of the BCL-2 family are central to the regulation of the mitochondrial pathway of apoptosis (Chipuk et al., 2010). The BH3-only subclass of proapoptotic proteins share sequence homology only within the BCL-2 homology (BH) 3 domain, which serves as a minimal death domain required for their interaction with other BCL-2 family members. The apoptotic activity of BH3only molecules is regulated by tissueand signal-specic mechanisms. The BH3-only protein Noxa was identied as a phorbol myristate acetate (PMA) response gene in adult T cell leukemia and later discovered as a p53-inducible gene in response to g irradiation (reviewed in Ploner et al., 2008). Noxas apoptotic function is selectively induced by DNA damage, ischemia, and glucose deprivation. Moreover, in light of the signicant species-specic differences in Noxa protein sequence, speciesspecic regulatory mechanisms may modulate Noxas function. Lowman et al. (2010) provide evidence for posttranslational regulation of human Noxa (hNoxa) in leukemia cell lines and primary activated T cells. Initial clues that a potent inhibitor of the atypical cyclin-dependent kinase-5 (Cdk5) induces apoptosis in a Noxa-dependent manner prompted the authors to test Noxa as a direct target of this kinase. In pursuit of both the regulatory mechanisms and functional consequences of this modication, the authors uncovered a link to glucose metabolism and signaling. Specically, they found that glucose is required for Cdk5 phosphorylation of hNoxa on Ser13 upstream of its BH3 domain, and this modication in turn imparts dual benets by inactivating Noxas apoptotic function while triggering a previously unknown effect on glucose metabolism. Glucose-mediated control of Noxa is also in line with its selective role in glucose deprivationinduced death in leukemia cell lines, suggesting that changes in Noxa phosphorylation in this setting may inform cells of glucose availability and determine whether the cells will undergo apoptosis or utilize glucose and survive. Initial biochemical data suggest that Noxas dual functions are dictated by its dynamic recruitment to two distinct macromolecular complexes, the constituents of which are being identied. The precise aspect of glucose metabolism per se that is quantitatively altered by Noxa modication is yet to be determined. However, initial studies rule out changes in glycolysis and point to a possible effect on preferential channelling of glucose to the pentose phosphate pathway (PPP). This idea would be best validated through metabolic ux analysis using glucose tracers, which can also help determine the exact branch of PPP activated by Noxa. Furthermore, identication of potential metabolic enzymes that may associate with phosphorylated Noxa in the macromolecular survival complex may shed mechanistic insights into Noxas metabolic function. Phosphorylation-dependent modication of another BH3-only protein, BAD, has also been previously linked to glucose metabolism (Danial, 2008), albeit different from Noxa in multiple aspects. BAD toggles between metabolism and apoptosis through phosphorylation of a conserved serine residue within its BH3 domain that neutralizes its apoptotic function while stimulating glucose phosphorylation and oxidation (Danial et al., 2008). In endocrine tissues such as pancreas and liver, this is mediated through activation of glucokinase (hexokinase IV) and plays an important role in systemic glucose homeostasis. BADs effect on glucose phosphorylation is likely not restricted to liver and pancreatic b cells (Deng et al., 2008); however, its precise metabolic role in nonendocrine cells remains to be determined (Danial, 2008). The different mode of molecular engagements and the distinct aspect of glucose metabolism regulated by Noxa and BAD reinforce both the specialization and complexity of networks linking apoptosis and metabolism. A potential role for Noxa in regulation of ux through PPP may have physiologic implications within the context of both activated primary T lymphocytes and leukemic cells. Beyond provision of ribose sugars for DNA synthesis, intermediates
Previews
of PPP can inuence protein glycosylation or generation of complex oligosaccharides. These carbohydrate-based modications may be especially pertinent to specialized lymphocyte function, including migration and homing. In addition, NADPH production through PPP can be used for fatty acid and cholesterol synthesis. Cholesterol-rich lipid rafts are implicated in T cell receptor signaling. NADPH levels can also inuence the pool of reduced glutathione and cellular ROS detoxication capacity (Schafer et al., 2009). One of the intriguing ndings in the report by Lowman et al. (2010) is the presence of MCL-1L in both of the macromolecular complexes that mediate Noxas functions. The specic interaction mode of Noxa and MCL-1L in the apoptotic versus survival complexes is likely distinct. While Noxas apoptotic activity involves direct engagement of its BH3 domain with MCL-1L, the BH3 domain is not required for Noxa phosphorylation, and the interaction of phosphorylated Noxa with MCL-1L within the survival complex appears to be indirect. The exact mechanisms underlying Noxas binding interactions within these complexes, the involvement of potential phosphatases that counteract Noxa phosphorylation, and the nature of nutrient cues that may superimpose on this cross-regulation will provide important insights into dynamic regulation of Noxas bifunctional activities in apoptosis and metabolism. In addition, how and if recruitment of MCL-1L to the survival complex synergizes with Noxas metabolic role and whether this may have implications for the control of MCL-1s stability, which is sensitive to glucose metabolism (Zhao et al., 2007), pose exciting questions for future investigation. Given the critical role of BCL-2 proteins in apoptosis, it is perhaps not surprising that a wide spectrum of regulatory mechanisms regulate their function. However, much remains to be learned about how cellular survival/death decisions receive input from other homeostatic pathways. The ndings of Lowman et al. (2010) are yet another indication that the expanding functional interaction networks of BCL-2 proteins, which may include cell-deathand non-cell-death-related partners, hold powerful clues.
REFERENCES
Molecular Cell
Chipuk, J.E., Moldoveanu, T., Llambi, F., Parsons, M.J., and Green, D.R. (2010). Mol. Cell 37, 299310. Danial, N.N. (2008). Oncogene 27 (Suppl 1 ), S53S70. Danial, N.N., Walensky, L.D., Zhang, C.Y., Choi, C.S., Fisher, J.K., Molina, A.J., Datta, S.R., Pitter, K.L., Bird, G.H., Wikstrom, J.D., et al. (2008). Nat. Med. 14, 144153. Deng, H., Yu, F., Chen, J., Zhao, Y., Xiang, J., and Lin, A. (2008). J. Biol. Chem. 283, 2075420760. Lowman, X.H., McDonnell, M.A., Koloske, A., Odumade, O.A., Jenness, C., Karim, C.B., Jemmerson, R., and Kelekar, A. (2010). Mol. Cell 40, this issue, 823833. Ploner, C., Koer, R., and Villunger, A. (2008). Oncogene 27 (Suppl 1 ), S84S92. Schafer, Z.T., Grassian, A.R., Song, L., Jiang, Z., Gerhart-Hines, Z., Irie, H.Y., Gao, S., Puigserver, P., and Brugge, J.S. (2009). Nature 461, 109113. Sengupta, S., Peterson, T.R., and Sabatini, D.M. (2010). Mol. Cell 40, 310322. Vander Heiden, M.G., Cantley, L.C., and Thompson, C.B. (2009). Science 324, 10291033. Zhao, Y., Altman, B.J., Coloff, J.L., Herman, C.E., Jacobs, S.R., Wieman, H.L., Wofford, J.A., Dimascio, L.N., Ilkayeva, O., Kelekar, A., et al. (2007). Mol. Cell. Biol. 27, 43284339.
Molecular Cell
Review
The Chromatin Signaling Pathway: Diverse Mechanisms of Recruitment of Histone-Modifying Enzymes and Varied Biological Outcomes
Edwin Smith1 and Ali Shilatifard1,*
1Stowers Institute for Medical Research, 1000 East 50th Street, Kansas City, MO 64110, USA *Correspondence: ash@stowers.org DOI 10.1016/j.molcel.2010.11.031
Posttranslational modications of histones are coupled in the regulation of the cellular processes involving chromatin, such as transcription, replication, repair, and genome stability. Recent biochemical and genetic studies have clearly demonstrated that many aspects of chromatin, in addition to posttranslational modications of histones, provide surfaces that can interact with effectors and the modifying machineries in a context-dependent manner, all as a part of the chromatin signaling pathway. Here, we have reviewed recent ndings on the molecular basis for the recruitment of the chromatin-modifying machineries and their diverse and varied biological outcomes.
Introduction Although every cell within our body bears the same genetic information and the same set of genes, only a small subset of genes is transcribed in a given cell at a given time. The molecular mechanism underlying this cell-/stage-specic transcriptional control has been the subject of intense study for many years. The genetic information encoded in our DNA is packaged within nucleosomal arrays forming what is referred to as chromatin. A nucleosome contains 146 bp of DNA, wrapped twice around an octamer composed of two copies of histones H3 and H4 and two copies of histones H2A and H2B. Initial structural studies using electron microscopy demonstrated that nucleosomes are found in arrays forming a series of beads on a string, with the beads being the individual nucleosomes and the string the linker DNA (Kornberg, 1974; Kornberg and Lorch, 1999). Histone proteins contain a exible amino-terminal tail and a characteristic histone fold, a globular domain that mediates substantial interactions between histones to form the nucleosome. High-resolution X-ray crystallography demonstrated that histone N-terminal tails protrude outward from the nucleosome (Luger et al., 1997), and biochemical studies have conrmed that such histone tails are extensively posttranslationally modied (Bhaumik et al., 2007). Reported posttranslational modications of histones so far include acetylation, phosphorylation, methylation, monoubiquitination, sumoylation, and ADP ribosylation (Berger, 2007; Bernstein et al., 2007; Campos and Reinberg, 2009; Kouzarides, 2007; Shilatifard, 2006; Weake and Workman, 2008; Workman and Kingston, 1998). For almost every modication identied, there are also machineries involved in their removal (Bhaumik et al., 2007). Histone modications can have a variety of functions; they can change the charge of a residue to disrupt protein-DNA, protein-protein, and nucleosome-nucleosome interactions, and they can form binding surfaces for a variety of proteins (Campos and Reinberg, 2009; Kouzarides, 2007). Histone modications that form binding surfaces for other proteins have been implicated in several epigenetic processes. For example, HP1 can bind to the histone H3 tail methylated on lysine 9 as part of the process of stable maintenance of heterochromatic silencing. Polycomb (Pc) and Eed proteins can bind to histone H3K27-methylated tails at different stages of the cell cycle to help maintain the silencing of developmental genes during differentiation (Hansen et al., 2008; Kouzarides, 2007; Margueron et al., 2009). However, histone modications are not always epigenetic. The addition and removal of histone acetylation and phosphorylation can be transient events that are associated with the initiation and repression of genes. For example, histone H2B monoubiquitination is added and then quickly removed during the process of gene activation (Henry et al., 2003). Histone H3K4 methylation is associated with counteracting Pc silencing of developmental genes when implemented by Trithorax, but is also found on many housekeeping genes, where its function is unknown. Thus, the same modication can have both epigenetic and nonepigenetic functions. Recent studies have demonstrated that the posttranslational modications of histones do not represent a code and are no different than the posttranslational modications associated with any other proteins in the cell (Lee et al., 2010; Schreiber and Bernstein, 2002; Sims and Reinberg, 2008). The posttranslational modications of histones are part of signaling pathways, and their readout is context dependent, with the biological outcomes dictated by many variables. In this review, we analyze recent reports on the different molecular mechanisms of recruitment and biological outcomes for histone-modifying machineries. These studies demonstrate that a combination of many factors, including DNA elements, protein-protein interactions, stage and origin of cells, and posttranslational modications on transcription factors and histones, regulate the diverse biological outcome associated within the chromatin signaling pathway. The Coactivator and Corepressor Models of Recruitment Although it was known for decades that histone acetylation was associated with actively transcribed genes, it was not known if
Review
histone acetylation was a consequence of transcription or was involved in the activation process. Purication of a histone acetyltransferase from the transcriptionally active macronucleus of the ciliated protozoan Tetrahymena thermophila identied a protein homologous to yeast Gcn5, previously described as a coactivator of transcription (Berger et al., 1992; Brownell et al., 1996; Marcus et al., 1994). Coactivators were postulated to be factors required for bridging sequence-specic binding proteins and the basal transcriptional machinery (Berger et al., 1990; Pugh and Tjian, 1990). Subsequently, other previously known coactivators were shown to be histone acetyltransferases or to participate in complexes with histone acetyltransferases, including CBP/p300, SRC-1, and ACTR (Bannister and Kouzarides, 1996; Chen et al., 1997; Ogryzko et al., 1996; Spencer et al., 1997). Both Gcn5, as part of the SAGA histone H3 acetyltransferase complex, and Esa1, as part of the NuA4 histone H4 acetyltransferase complex, were shown to be recruited by certain acidic activators (Utley et al., 1998). Importantly, another Gcn5-containing complex, the ADA complex, did not interact with these activators, demonstrating specicity in the recruitment of different histone-modifying activities by distinct activators. Concurrent with the discovery of a histone acetyltransferase as a transcriptional activator was the discovery of a known transcriptional repressor, Rpd3, as a histone deacetylase (Taunton et al., 1996). Rpd3 and related enzymes are found to act as corepressors with sequence-specic transcriptional binding factors such as nuclear hormone receptors that can recruit SMRTNCOR histone deacetylase complexes (Alland et al., 1997; Hassig et al., 1997; Heinzel et al., 1997; Kadosh and Struhl, 1997; Nagy et al., 1997). The same repressor can recruit different deacetylase complexes. For example, IKAROS is a critical regulator of hematopoiesis that can interact with both the SIN3A and Mi-2-NuRD deacetylase complexes to repress lineage-specic genes (Kim et al., 1999; Koipally et al., 1999). Together, these studies not only demonstrated a functional role for histone-modifying activities as key regulators of transcription, but also provided the rst paradigm of how histone-modifying activities could be recruited to chromatin (Figure 1). Recruitment of Histone Modiers by RNA Polymerase II While histone acetylation and other histone modications can function at the promoter during transcriptional activation, histone modications can also have roles in the body of genes. The trimethylation of histone H3 on lysine 36 is mediated by the Set2 enzyme in yeast and animals, and this modication peaks in the middle and the 30 end of genes associated with the elongating RNA polymerase II (Krogan et al., 2003b; Shilatifard, 2006). Purication of the Set2 enzyme from yeast identied the large subunit of Pol II, Rpb1, as an interactor of Set2 (Krogan et al., 2003b; Xiao et al., 2003). This interaction depends on the phosphorylation of Rpb1s C-terminal repeat domain on serine 2, a marker of elongating Pol II. Thus, Set2 is recruited to gene bodies through an interaction with elongating Pol II (Figure 2). In contrast to the direct physical interaction between Set2 and RNA Pol II, other histone-modifying activities can be recruited subsequently to transcription initiation through the interaction with the polymerase-associated factor (Paf1) complex (Krogan
Molecular Cell
Figure 1. Coactivators and Corepressors Are Histone-Modifying Enzymes

(A) The histone acetyltransferase, Gcn5, as part of the SAGA complex, is recruited to genes by a variety of transcriptional activators such as the sequence-specic binding protein Gal4. Acetylation of histone H3 through recruitment of Gcn5 by Gal4 synergizes with nucleosome remodeling activities, leading to the recruitment of RNA Pol II. (B) The histone deacetylase HDAC1 can be recruited to genes as part of the SMRT-NCOR complexes by sequence-specic DNA-binding proteins such as retinoic acid receptors (RXR/RAR). Deacetylation of histones facilitates reassembly of compact chromatin for transcriptional repression.
et al., 2003a; Ng et al., 2003a; Wood et al., 2003). The Paf1 complex is associated with elongating Pol II. It was found in a global proteomic analysis in S. cerevisiae (GPS) to be required for proper H3K4 methylation (Krogan et al., 2003a; Wood et al., 2003). Paf1 directly interacts with COMPASS (complex of proteins associated with Set1), the sole H3K4 methyltransferase in yeast, and Paf1 is required for the recruitment of COMPASS to chromatin (Wood et al., 2003). These early studies in yeast set the paradigm that the Paf1 complex plays a role as a platform for the requirement of histone-modifying machinery to the elongating Pol II (Gerber and Shilatifard, 2003) (Figure 2). We now know that the Drosophila and mammalian Paf1 complexes also function as a platform for the recruitment of Set1/TRX/ MLL-containing complexes in Drosophila and human cells (Tenney et al., 2006; Wang et al., 2008). In addition to the recruitment of histone-modifying activities to chromatin, complexes can be recruited to chromatin, but remain
Molecular Cell
Review
Figure 3. Recruitment of Histone-Modifying Activities by Pre-existing Histone Modications Figure 2. Recruitment of Histone-Modifying Activities by RNA Pol II
The histone H3K4 methyltransferase, Set1, as part of COMPASS, is recruited to genes through interaction with the Polymerase-associated factor (Paf1) complex (1) (Krogan et al., 2003a; Wood et al., 2003). H3K4 trimethylation is found at the start sites of the transcription of active genes, as well as poised genes with stalled RNA Pol II (Gilmour, 2009). Upon release of RNA Pol II by phosphorylation of its C-terminal domain (CTD) at serine 2 (P-Ser2), numerous factors associate with the phosphorylated C-terminal domain, including the histone H3K36 methyltransferase Set2 (Xiao et al., 2003) (2). Set2 can be recruited through interactions with both the CTD of Pol II and the Paf1 complex (Krogan et al., 2003b; Xiao et al., 2003). Methylation of histone H3K36 in the body of actively transcribed genes (see Figure 2) can be bound by the chromodomain-containing protein, Eaf3, as part of the Rpd3S deacetylase complex. This process helps to reposition nucleosomes in the wake of transcribing RNA Pol II. In the absence of Rpd3S, nucleosomes are misplaced, resulting in the exposure of cryptic start sites that can lead to the formation of aberrant transcription. Thus, recruitment of Rpd3S by a histone modication placed by passing RNA polymerase complexes can maintain transcriptional delity. It has recently become apparent that at most genes, Rpd3S can be recruited directly through the interaction with the serine 5 phosphorylated form of the RNA Pol II CTD (Govind et al., 2010; Drouin et al., 2010). These ndings indicate that the relative contribution of H3K36 methylation and CTD phosphorylation on the recruitment of Rpd3S can be gene specic and context dependent.
in a relatively inactive state. The Paf1 complex appears to regulate several activities in this manner. Histone H2B can be monoubiquitinated by the Rad6-Bre1 E2/E3 ubiquitin ligase complex. This monoubiquitination is needed for the higher methylation states of H3K4 and H3K79, and Paf1 is required for H2B monoubiquitination (Krogan et al., 2003a; Wood et al., 2003). However, both Rad6-Bre1 and Dot1 appear to be recruited independently of Paf1 (Wood et al., 2003). Paf1 stimulates the enzymatic activity of Rad6-Bre1, perhaps after recruitment of these complexes to chromatin. The subsequent monoubiquitination of H2B stimulates the activity of Set1/COMPASS and Dot1 to generate the trimethylation of H3K4 and H3K79 (Dover et al., 2002; Krogan et al., 2003a; Wood et al., 2003). Several recent mammalian studies have also conrmed the generality of these early observations made in yeast regarding the role of the Paf1 complex in the regulation of histone H2B monoubiquitination by Rad6Bre1, H3K4 methylation by Set1/COMPASS, and H3K79 methylation by Dot1 (Kim et al., 2009; McGinty et al., 2008; Pavri et al., 2006; Zhu et al., 2005). Recruitment by the Preceding Histone Modications Many of the posttranslational modications of histones can enhance chromatin binding by other proteins through a variety of protein domains: bromodomains, found in several transcriptional activators, can preferentially bind peptides with acetylated lysines; 14-3-3 and forkhead domains can bind phosphorylated serines and threonines; chromodomains, MBT repeats, and PHD ngers can discriminate among lysines that are mono-, di-, or trimethylated; and tudor domains can recognize methylated arginine or lysine residues (Maurer-Stroh et al., 2003; Taverna et al., 2007; Yaffe and Elia, 2001). The importance of the interaction between histone modications and proteins containing these modules is exemplied by the occurrence of mutations within these domains in human disease (Matthews et al., 2007; Pena et al., 2006). Modications can help form a landing platform for proteins and their complexes to aid in recruitment to chromatin through recognition of the modied residue. One example of this is the recruitment of deacetylase complexes to the body of
transcriptionally active genes that require the chromodomaincontaining protein Eaf3. Eaf3 preferentially binds to H3K36 di- and trimethylated states. While Eaf3 is a component of both histone acetyltransferase and deacetylase complexes, loss of Eaf3 leads to an increase in acetylation in the body of genes, which is proposed to have the effect of opening up the chromatin structure to allow cryptic transcription (Carrozza et al., 2005; Keogh et al., 2005) (Figure 3). An important issue is the relative contribution of histone modications in recruitment of other factors to chromatin. Since Eaf3 is also a component of the NuA4 histone acetyltransferase complex, one might expect both of the Eaf3 complexes to compete for binding to H3K36-methylated nucleosomes. However, Rco1, a component of Rpd3S but not of NuA4, interacts with nucleosomes (with or without histone modications) and is required for the interaction of Rpd3S with nucleosomes methylated at H3K36, indicating that both histone modication-dependent and independent mechanisms are important for targeting Rpd3S to the coding regions of transcribed genes (Li et al., 2007). Recently, Rpd3S has been demonstrated to be recruited to genes through interactions between the Rco1 subunit and the serine 5 phosphorylated form of the CTD of RNA Pol II. This nding suggests that the interaction with H3K36 methylated nucleosomes by Eaf3-Rco1 is subsequent to the initial recruitment of the complex by the CTD of RNA Pol II (Govind et al., 2010). Supporting these observations, genome-wide proling of Rpd3S in a Set2 deletion by Robert and colleagues demonstrated that H3K36 methylation had a modest effect on the recruitment of Rpd3S to transcribed regions (Drouin et al., 2010). Animals have at least two homologs of Eaf3: its ortholog, MRG15, and a paralog, MSL3. MRG15 participates in the NuA4-like Tip60 complex as well as the histone deacetylase complexes (Kusch et al., 2004; Lee et al., 2009; Moshkin et al., 2009; Spain et al., 2010). MSL3 is part of an H4K16-specic histone acetyltransferase complex. MSL3 can interact with H3K36-methylated nucleosomes and mediates acetylation in the ORFs of transcribed genes (Sural et al., 2008). Although
Review
a balance of histone acetylation/deacetylation in ORFs is probably important for such processes as enhancing transcription elongation and maintaining transcription delity, how the MRG15 and MSL3 complexes are precisely targeted to the promoters or gene bodies likely involves factors other than just H3K36me3 binding. Rpd3S has not been extensively characterized in metazoans, although complexes with MRG15 and histone deacetylases have been reported (Lee et al., 2009; Tominaga et al., 2003; Yochum and Ayer, 2002). Importantly, the metazoan homologs of Rco1, PHF12 in mammals and CG3815 in Drosophila, are found in Rpd3S-like complexes, although the relative contribution of this subunit to the recruitment of Rpd3S by direct interaction with RNA Pol II or indirect interaction with the H3K36 methylation deposited by Set2 remains to be determined. Direct Recruitment by DNA Sequences In the coactivator model, the DNA-bound transcription factors recruit histone-modifying complexes. However, the most direct way to recruit a histone-modifying activity to particular genes is through recognition of specic DNA sequences by the histone-modifying complexes. This process is perhaps at the heart of, and the basis for, all recruitment to chromatin and can be considered the main step in the epigenetic regulation, as well (Berger et al., 2009). Here, we review the role of DNA elements that form target sites for histone-modifying complexes that extend sequence-dependent recruitment mechanisms beyond the simple coactivator/corepressor model. The PRE in Polycomb-Mediated Repression The identication of the Polycomb group of genes (PcG) (Ringrose and Paro, 2004) was made by the discovery of mutations in several genes in Drosophila that led to the ectopic expression of the Hox genes and a transformation of segmental identity. The PcG genes are required for the repression of Hox genes in regions where these genes have not been activated. The repression can be stably maintained for several cell generations after the transcription factors that initially determined cell identity are no longer present, constituting a form of epigenetic memory (Berger et al., 2009). Cloned regions near the Hox genes were shown to confer proper repression of reporter genes in a PcGdependent manner, thus being known as Polycomb response elements (PREs). PREs are like enhancers in that they can function from large distances from genes and are sometimes found in introns of the genes they regulate. However, unlike canonical enhancer-binding proteins, PcG proteins are expressed ubiquitously and are downstream of the decision of whether a target gene is to be activated or repressed (Ringrose and Paro, 2004). PcG proteins are found in at least three distinct complexes (Schuettengruber et al., 2007; Schwartz and Pirrotta, 2007; Simon and Kingston, 2009; Wang et al., 2004). Polycomb repressive complex 2 (PRC2) contains the Suppressor of zeste 12 (Su[z]12) and Extra sex combs (ESC), as well as the histone methyltransferase Enhancer of zeste (E[z]) (EZH1 and EZH2 in mammals) that implements H3K27 methylation. H3K27 methylation has been shown to form a binding site for the Polycomb protein (Pc), a component of the Polycomb repressive complex 1 (PRC1). PRC1 also includes the E3 ubiquitin ligase RING,
Molecular Cell
Posterior sex combs (PSC, or Bmi-1 in mammals) and Polyhomeotic (PH). A third complex, the PHO-repressive complex, consists of Pleiohomeotic (PHO, similar to YY1 in mammals) and Sfmbt (MBTD1 in mammals), a protein that can recognize H4K20 mono- and dimethylated states. Of the known PcG proteins, PHO, which bears several zinc ngers, is clearly a sequence-specic DNA-binding protein and is required for recruitment of PRC2 to chromatin in Drosophila. However, PHO binding is not sufcient for recruitment, and other factors have been implicated in PRC2 localization in Drosophila. While statistically signicant enrichment of sequences such as the PHO/YY1 consensus can be found in several PREs, a sequence-based denition of PREs has remained elusive (Ringrose et al., 2003). One outstanding question is whether PREs exist in mammals. Although PRC2 is conserved from ies to mammals, YY1 appears to be the only mammalian homolog of the Drosophila transcription factors implicated in sequence-specic binding to elements commonly found within Drosophila PREs. However, JARID2, a protein with very weak DNA-binding activity, was recently identied as a component of PRC2 that is required for recruitment to chromatin. The mechanism by which JARID2 recruits PRC2 is not known (Landeira et al., 2010; Li et al., 2010; Pasini et al., 2010; Peng et al., 2009; for review, please see Herz and Shilatifard, 2010). Unlike other PRC2 components, JARID2 shows obvious tissue-specic differences in expression, being particularly highly expressed in embryonic stem cells. Additionally, JARID2 seems to have different effects on different genes, perhaps reecting the diversity in the PRC2 complexes (Herz and Shilatifard, 2010). A full understanding of the role of JARID2 in the regulation of PRC2 activity and gene expression will require further detailed and comprehensive biochemical and in vivo studies using organisms amenable to genetic manipulations. Recently, Kingston and colleagues identied a 1.8 kb region that conferred PcG responsiveness at the HoxD locus on a reporter gene (Woo et al., 2010). This region contained YY1 binding sites (Wang et al., 2004) as well as an evolutionarily conserved sequence within the HoxD locus (Beckers and Duboule, 1998). In addition to these two features, this putative mammalian PRE contains a CpG island (Illingworth and Bird, 2009). In mammals, PRC2 binding is highly correlated with the presence of CpG islands, and CpG islands are largely predictive of the presence of PRC2, suggesting that CpG islands behave as PREs in mammalian cells (Ku et al., 2008). CpG islands are regions enriched for CpG dinucleotides that are associated with promoters of 60%70% of genes (Illingworth and Bird, 2009). Most CpG dinucleotides in the mammalian genome are cytosine methylated, but when they are sufciently clustered together in islands, they are typically unmethylated, except during transcriptional silencing associated with imprinting, X inactivation, or silencing of tumor suppressor genes during oncogenesis. Genes that are silenced during development are not associated with the CpG methylation of their promoters, even if associated with CpG islands (Baylin and Bestor, 2002; Illingworth and Bird, 2009). Therefore, it is likely that PcG function, without DNA methylation, is the major repressor of many developmentally regulated genes, perhaps through recognition
Molecular Cell
Review
Figure 4. Recruitment by DNA Sequences
(AC) In humans, CpG islands are found near the promoters of 60%70% of genes (Illingworth and Bird, 2009). While cytosines within CpG dinucleotides are frequently methylated, in CpG islands, cytosines are largely unmethylated. A number of proteins bearing a zinc nger of the CXXC class preferentially bind unmethylated CpG islands (A). These proteins are associated with a variety of histone-modifying activities, including the H3K4 methyltransferases, MLL and Set1/COMPASS, and the H3K36 demethylase, KDM2A (Clouaire and Stancheva, 2008). MBD1 has two zinc ngers that bind methylated DNA and a third of the CXXC class that binds unmethylated DNA. MBD1 is associated with H3K9 methyltransferase activities and is thought to help coordinate silencing by DNA and histone methylations (Sarraf and Stancheva, 2004). A recent study demonstrated that CpG islands were major sites of recruitment of CXXC1 (Cfp1) and H3K4 methylation via mammalian COMPASS, independent of the transcriptional status of the nearby gene (B) (Thomson et al., 2010). In another recent manuscript, KDM2A, an H3K36 demethylase, was found to be targeted to CpG islands via its CXXC motif (C). Since H3K36 methylation can recruit histone deacetylases (Figure 3), removal of H3K36 methylation in CpG islands could maintain CpG islands in an open, permissive state (Blackledge et al., 2010).
of CpG islands; however, PcG proteins can also synergize with DNA methylation machineries during silencing associated with imprinting and X inactivation (Illingworth and Bird, 2009). How is PRC2 targeted to CpG islands? CpG-rich sequences can be recognized by proteins bearing a CXXC domain, a type of zinc nger that can recognize CpG sequences. However, CXXC domains are not found within PRC2 components, suggesting that a novel domain recognizes these sequences or that other motifs found within these islands help recruit PRC2. PRE versus TRE While Hox genes can be silenced with PcG complexes, they can be activated with the help of the trithorax group of proteins (trxG) (Eissenberg and Shilatifard, 2010). Trithorax, like its mammalian counterpart MLL, is an H3K4 methyltransferase. H3K4- and H3K27-methylated nucleosomes are usually found to occur in a mutually exclusive pattern throughout the genome. This can perhaps be explained by the competition for the same recruitment sites by the enzyme complexes that implement these modications. Experimental evidence indicates that PREs and Trithorax response elements (TREs) substantially overlap (Ringrose and Paro, 2007). DNA-binding factors that are important for recognizing TREs include the GAGA factor (also known as Trl), which recognizes GA-rich sequences (Farkas et al., 1994), and PHO and Zeste, although other combinations of factors are likely to be important for recruitment (Schwartz et al., 2006). In mammals, the trithorax homologs MLL1 and MLL2 have CXXC domains within their polypeptide, suggesting that the CpG character within a PRE may constitute a distinct sequence that helps recruit the competing trxG activity to the PREs/TREs. Interestingly, recent work from two laboratories has shown that the CXXC domains are involved in targeting distinct types of chromatin-modifying activities to chromatin, conrming the impor-
tance of the CpG islands in recruiting histone-modifying activities to chromatin while also raising the question of which other features determine the specicity demonstrated for these diverse activities. CpG Islands Recruit Multiple Modiers The CXXC domain binding of CpG was originally discovered in MBD1, a protein isolated for its homology to the methylated CpG DNA-binding protein, MeCP2, but also having three zinc ngers of the CXXC type (Cross et al., 1997). Subsequently, it was determined that while two of the CXXC domains within MBD1 bind methylated DNA, many CXXC domains are specic for unmethylated CpG sequences (Birke et al., 2002; Lee et al., 2001), including a third CXXC domain within MBD1 (Jrgensen et al., 2004) and the DNA methyltransferase DNMT1 (Pradhan et al., 2008). Thus, multiple chromatin modiers could potentially be recruited by CpG-rich sequences (Figure 4A). Two recent studies have tested this idea and found a critical role for CXXC domains in recruiting histone-modifying activities to chromatin (Blackledge et al., 2010; Thomson et al., 2010). In one test of the role of the CXXC domains in targeting CpG islands, Bird and colleagues studied the recruitment of CXXC1 (Cfp1), a component of mammalian Set1/COMPASS, the major H3K4me3 in mammals (Lee and Skalnik, 2005; Miller et al., 2001; Thomson et al., 2010; Wu et al., 2008). Bird and colleagues performed genome-wide proling of CXXC1 in mouse brain. They found CXXC1 to be localized at 80% of the CpG islands, 90% of which were enriched for H3K4me3. Half of the CXXC1negative CpG islands were previously reported to be sites of Polycomb and H3K27me3 occupancy. Comparing CXXC1 occupancy at Xist, a gene transcribed on only one of two X chromosomes in females, CXXC1 was found to associate exclusively with the transcribed, unmethylated CpG copy. Together, these
Review
ndings suggest that CpG islands could help recruit mammalian Set1/COMPASS through CXXC1s afnity for unmethylated CpG sequences (Figure 4B). To experimentally test the role of CpG islands in recruiting CXXC1, Bird and colleagues created two embryonic stem cell lines carrying a promoterless construct with eGFP and puromycin sequences that contain CpG dinucleotide densities similar to those found in CpG islands. In one cell line, the articial CpG island was targeted to the 30 end of Nanog, while in the other, the construct was targeted to the 30 end of Mecp2, an X-linked gene. When located adjacent to Nanog, the cassette remained free of CpG methylation, despite lacking detectable Pol II. H3K4me3 and CXXC1 occupancy tracked CpG density within the cassette and around the insertion site. Interestingly, the cassette inserted next to the Mecp2 gene showed two-thirds CpG methylation, but still was found to be bound by CXXC1. Bisulte sequencing of the CXXC1 and H3K4me3 chromatin immunoprecipitation (ChIP) DNA demonstrated that only a third of the immunoprecipitated cassette was CpG methylated, strongly suggesting that unmethylated CpG sequences are recruitment sites for mammalian Set1/COMPASS. However, point mutations within the CXXC domain of CXXC1 would be required to demonstrate the direct role of this protein in binding CpG and the ensuing regulation of H3K4 methylation by mammalian Set1/COMPASS. While Set1/COMPASS is the major H3K4 trimethylase in mammalian cells (Wu et al., 2008), in other studies it has been demonstrated that the loss of CXXC1 leads to an increase in global H3K4 trimethylation levels in embryonic stem cells (Tate et al., 2009). Based on this observation, Skalnik and colleagues have suggested that CXXC1 functions by restricting Set1/ COMPASS methyltransferase activity. In contrast, Bird and colleagues nd that H3K4me3 levels are reduced at the CpG islands in the absence of CXXC1. It will be interesting to learn where in the genome H3K4me3 is increasing upon loss of CXXC1. An important nding by Bird and colleagues is that H3K4 methylation implemented by mammalian Set1/COMPASS is independent of Pol II and transcription, while ndings in yeast have shown that transcription is required for proper H3K4 trimethylation (Krogan et al., 2003a; Ng et al., 2003b; Shilatifard, 2006). One possible explanation for this apparent difference is that the interaction of Pol II with CpG islands might be transient and not as easily detectable as the product of the process, histone H3K4 trimethylation. Therefore, sensitive RNA-seq methods, such as global run-on sequencing (GRO-seq) (Core et al., 2008), could reveal transcription within these CpG islands, thus explaining the association of CXXC1 and mammalian COMPASS at these CpG islands. Interestingly, the yeast homolog of CXXC1, Cps40, has a PHD nger in common, but lacks a CXXC domain. However, the Drosophila homolog of CXXC1, CG17446, contains both the PHD and CXXC domains. CpG islands have not been studied in Drosophila, although their existence has been predicted (Takai and Jones, 2002). It is notable that Drosophila Trithorax, unlike its mammalian counterpart, MLL, lacks a CXXC domain, further demonstrating that although these H3K4 methyltransferase complexes are largely conserved in composition and function, some aspects of their recruitment can differ, perhaps reecting
Molecular Cell
differences in the size or complexity of the genome in which they are found. Another class of histone-modifying enzymes bearing a CXXC domain is KDM2A/B. KDM2A/B is a histone demethylase that preferentially uses H3K36me2 as a substrate (Tsukada et al., 2006). H3K36me2 is a modication that has previously been linked to gene silencing (Bender et al., 2006) and can recruit histone deacetylases (Li et al., 2009; Youdell et al., 2008), suggesting that removal of H3K36me2 could facilitate the formation of open chromatin. Klose and colleagues tested the role of the CXXC domain in targeting KDM2A. First, they tested the DNAbinding specicity of KDM2As CXXC domain and found that it preferentially binds to unmethylated CpG sequences (Blackledge et al., 2010). Genome-wide proling demonstrated that KDM2A was highly enriched at annotated CpG islands. Major peaks of KDM2A binding not corresponding to CpG islands were shown to have a high CpG content with little DNA methylation, as assessed by bisulte sequencing. Since most CpGs outside of the CpG islands are unmethylated, Klose and colleagues likely found novel CpG islands that were previously unnoticed due to the statistical criteria used in the annotation process. Sites of strong KDM2A binding were also found to be depleted for H3K36me2, suggesting that KDM2A localization to CpG islands results in active demethylation of H3K36me2 (Figure 4C). Indeed, knockdown of KDM2A results in increased H3K36me2 at some CpG islands; however, very little alteration in transcription was reported when KMD2A levels were reduced by RNAi, suggesting that H3K36 dimethylation at CpG islands at the promoters does not have a major transcriptional regulatory role. KDM2A and the highly related KDM2B have been shown to be highly concentrated in nucleoli of cells where they can repress ribosomal RNA transcription (Frescas et al., 2007; Tanaka et al., 2010). Interestingly, rDNA accounts for 20% of the unmethylated CpG sequences in the mouse genome (Bird et al., 1985), suggesting that KDM2A and KDM2B are targeted in part through CpG recognition for rDNA transcription by RNA Pol I. The studies by the Bird and Klose groups were both in agreement that recruitment to unmethylated CpG islands via the CXXC domain was uncorrelated with transcriptional activity, suggesting that unmethylated CpG content is sufcient for recruitment. Since MBD1 and the DNA methyltransferase DNMT1 also have CXXC ngers that recognize unmethylated CpG, it will be important to ask how various CXXC-associated activities compete or coexist with each other on the same site and how this is regulated during development. It would not be surprising if future studies in this regard will nd that the interactions of the CXXC nger-containing proteins with their target site are context dependent and require other cellular signals for proper function. The studies by the Bird and Klose groups should stimulate investigations into the role of CpG islands in transcription and the function of histone-modifying activities in this process. Same Modication, Different Complexes, and Distinct Outcomes One of the best-studied histone modications is the methylation of H3 on lysine 4 (H3K4). The rst methyltransferase complex for H3K4 to be identied was COMPASS in yeast (Krogan et al.,
Molecular Cell
Review
2002; Miller et al., 2001; Roguev et al., 2001; Shilatifard, 2006). The original interest in purifying COMPASS was based on the similarity between yeast Set1 and the human MLL protein involved in leukemia, and indeed, human MLL forms a COMPASS-like complex with methyltransferase activity for H3K4 (Eissenberg and Shilatifard, 2010; Hughes et al., 2004; Miller et al., 2001; Shilatifard, 2006). While yeast has just one methyltransferase for H3K4, humans have at least six methyltransferases that can methylate H3K4, including MLL1-4 and Set1A/B, which are found in COMPASS-like complexes (Cho et al., 2007b; Eissenberg and Shilatifard, 2010; Hughes et al., 2004; Lee and Skalnik, 2008; Lee et al., 2007a; Shilatifard, 2008; Wu et al., 2008). COMPASS-like complexes are dened as having a Set1 or MLL-related protein, and the core subunits Cps60/ASH2, Cps30/WDR5, and Cps50/RBBP5. In addition to the common subunits, the Set1 and MLL complexes each have unique subunits. MLL1 and MLL2, which are related to Drosophila Trithorax (TRX), interact with the tumor suppressor Menin, which helps target MLL1/2 to Hox and other targets for transcription activation (Hughes et al., 2004; Wang et al., 2009). MLL3 and MLL4, which show high similarity to Drosophila Trithorax-related (TRR), contain NCOA6, PTIP, and PA-1, which may help target MLL3/4 to hormone-responsive genes, together with UTX, an H3K27 histone demethylase (Cho et al., 2007b; Issaeva et al., 2007; Patel et al., 2007). Two subunits found in the SETD1A and SETD1B complexes, but not in the MLL1-4 complexes, are CXXC1 and WDR82 (Lee and Skalnik, 2008; Lee et al., 2007a; Wu et al., 2008). These subunits are homologous to the COMPASS subunits Cps40 and Cps35 (Miller et al., 2001). Cps35/WDR82 has a unique role in mediating the crosstalk between histone H2B ubiquitination and H3K4 methylation (Lee et al., 2007b; Wu et al., 2008; Zheng et al., 2010). The COMPASS and MLL1-4 complexes can be recruited by different mechanisms to chromatin, with varying functional consequences associated with H3K4 methylation. COMPASS can be recruited to the actively transcribed genes through its interaction with the Paf1 complex and RNA Pol II, which is sufcient for H3K4 monomethylation (Gerber and Shilatifard, 2003; Krogan et al., 2003a; Wood et al., 2003). The Paf1 complex may also be involved in recruiting Rad6-Bre1 through a direct Paf1-Bre1 interaction, leading to ubiquitinated H2B within this region (Wood et al., 2003). Cps35/WDR82 interacts with COMPASS in a histone H2B monoubiquitination-dependent manner, and this interaction on chromatin converts COMPASS to a trimethylation-competent complex (Lee et al., 2007b; Zheng et al., 2010). Deletion of COMPASS subunits or mutation of H3K4 to alanine has little effect on transcription levels in yeast (Miller et al., 2001), consistent with the recruitment and stimulation of COMPASS activity following gene activation (Krogan et al., 2003a; Ng et al., 2003b) (Figure 5A). In contrast to COMPASS, the MLL complexes can function as transcriptional activators. Loss of MLL leads to loss of H3K4 methylation and activation of transcription at Hox and other loci (Hughes et al., 2004; Wang et al., 2009). MLL1-4 complexes appear to be coactivators of nuclear receptors (Cho et al., 2007a; Dreijerink et al., 2009; Eissenberg and Shilatifard, 2010; Goo et al., 2003). H3K4 methylation by the MLL complexes could
Figure 5. The Same Histone Modication Implemented by Different Complexes Can Have Different Consequences
(A) Set1/COMPASS is recruited to chromatin of transcriptionally active genes through interaction with the Paf1 complex (1) (Krogan et al., 2003a), where it is only capable of implementing H3K4 monomethylation. Upon H2B monoubiquitination (H2Bub) by Rad6/Bre1 (2), WDR82 (Cps35 in yeast) interacts with chromatin and associates with COMPASS (3), enabling the complex to trimethylate H3K4 (4), whose presence at promoters is a mark of actively transcribed genes (see Figure 2). (B) MLL is part of a COMPASS-like complex that shares many subunits but, importantly, lacks WDR82, and therefore, its methyltransferase activity is likely to be independent of H2Bub. MLL, like its Drosophila homolog trithorax, is well known for its role in activating Hox genes. Recruitment of MLL via Menin and sequence-specic binding factors could help activate transcription of its target loci by mediating trimethylation of H3K4 (1), which can form a binding site for TAF3, a component of TFIID in the preinitiation complexes (PIC) (2), thereby mediating recruitment or stabilization and the basal machinery (PIC) and RNA Pol II (3) (Vermeulen et al., 2007; Wang et al., 2009). Thus, H3K4 trimethylation implemented by the MLL complexes is instructive for transcription (Wang et al., 2009), while H3K4 trimethylation implemented by Set1/ COMPASS may be recruited subsequent to initiation of transcription by PIC and RNA Pol II through the H2Bub-WDR82 pathway (Wu et al., 2008; Wang et al., 2009).
activate genes through recruitment of H3K4me3-binding proteins (Figure 5B). For example, the basal transcription factor can be directly recruited by H3K4me3 via the Taf3 subunits PHD nger (Vermeulen et al., 2007). Alternatively, H3K4me3 at the promoter could help recruit histone acetyltransferase complexes or nucleosome remodeling complexes, which harbor subunits with PHD ngers capable of recognizing the H3K4me3 state (Kouzarides, 2007; Ruthenburg et al., 2007). Thus, the timing and mechanism of recruitment to a gene can inuence the biological readout of H3K4 methylation (Wang et al., 2009). Noncoding RNAs as Recruitment Vehicles for Histone Modiers In addition to protein and DNA-based recruitment of histone modiers, noncoding RNAs have recently been shown to be major regulators of chromatin and transcription. While RNAs are well-known integral components of ribosomes and spliceosomes for the translation and splicing of mRNAs, the role of RNA molecules in transcriptional regulatory complexes has only recently gained widespread notice. One of the rst
Review
noncoding RNAs associated with transcriptional regulation was the Xist RNA, whose transcription is required for somatic silencing of an X chromosome in mammalian females, part of a process of dosage compensation to equalize expression from X-linked genes between XY males and XX females (Brockdorff et al., 1992; Brown et al., 1991). Inactivation of the X chromosome is associated with accumulation of repressive histone marks and the compaction of the X chromosome into the Barr body (Chow and Heard, 2009; Senner and Brockdorff, 2009). Silencing initiates and spreads in cis from the site of Xist transcription to encompass most of the X chromosome. The ability of certain forms of Xist RNA to interact with PRC2 complexes provides a possible molecular explanation for the requirement of Xist for H3K27 methylation and gene silencing (Zhao et al., 2008). Equalization of transcription between males and females of Drosophila also involves noncoding RNAs. roX1 and roX2 are transcribed from the X chromosome in males and are required for the 2-fold increased expression of the male X chromosome to equalize transcription with the two X chromosomes from females. Unlike Xist, roX RNAs can act in trans (Kelley et al., 1999). They form a ribonucleoprotein complex with the MSL proteins that includes the histone acetyltransferase MOF. MOF mediates H4K16 acetylation in the transcribed region of genes, which is proposed to facilitate the decompaction of the X chromosome to facilitate higher rates of transcription elongation by Pol II (Smith et al., 2001). The targeting of the roX-MSL complex to its X-linked targets is comprised of two steps: (1) recognition of high-afnity sites for the MSL1-MSL2 subunits and (2) spreading to nearby actively transcribed genes (Alekseyenko et al., 2008). The nature of the high-afnity sites is unknown, perhaps consisting of low-afnity sites within a specic context (Alekseyenko et al., 2008; Fauth et al., 2010; Straub et al., 2008). Spreading from high-afnity sites to nearby genes could be facilitated by the MLE subunit, an RNA helicase whose closest mammalian homolog has been shown to interact with RNA Pol II (Nakajima et al., 1997). MLEs presence in the roX-MSL complex is dependent on the integrity of the RNA component (Ilik and Akhtar, 2009; Smith et al., 2000). Thus, roX RNAs could function in MSL spreading by interacting with the RNA helicase MLE, which in turn associates with RNA Pol II at active genes, allowing the roX-MSL complex to spread from one active gene to another. The enrichment of the MSL complexes in the middle and the 30 end of the transcribed genes could be facilitated by interactions between the chromodomain of the MSL3 subunit and H3K36me3 implemented by Pol II-associated Set2 (Sural et al., 2008). Placing high-afnity binding sites for MSL1-MSL2 on autosomes allows a similar spreading to nearby autosomal genes, which becomes even more pronounced if particular spliced forms of roX RNAs are ectopically expressed (Park et al., 2005). While the roX RNAs are associated with the upregulation of transcription, many noncoding RNAs recruit repressive histone-modifying activities similarly to Xist. For example, Air and Kcnq1ot1 are noncoding RNAs expressed from the paternal chromosome that are required for silencing of neighboring genes in cis (Mancini-Dinardo et al., 2006; Sleutels et al., 2002). Both have been shown to associate with the H3K9 methyltransferase,
Molecular Cell
G9a, while Kcnq1ot1 is also associated with PRC2 (Nagano et al., 2008; Pandey et al., 2008). Although Xist, Air, and Kcnq1ot1 all work to silence genes in cis, it is becoming apparent that noncoding RNAs can silence genes on other chromosomes. HOTAIR, a noncoding RNA transcribed from the HOXC cluster, recruits PRC2 to silence genes in the HOXD cluster (Rinn et al., 2007). Separate regions of HOTAIR interact with the PRC2 and LSD1/CoREST complexes (Tsai et al., 2010). LSD1 is an H3K4 demethylase that can demethylate H3K4me2, but not H3K4me3, and associates with the CoREST histone deacetylase complex (Lee et al., 2005; Shi et al., 2004). In addition to a role in developmental regulation, HOTAIR is also implicated in promoting metastasis. HOTAIR is frequently upregulated in metastatic tumors, and its high expression is associated with a poor prognosis (Gupta et al., 2010). Importantly, PRC2 components are required for the matrix invasiveness of cells ectopically expressing HOTAIR, suggesting that HOTAIR is mistargeting PRC2 components when overexpressed (Gupta et al., 2010). The potential regulatory scope of noncoding RNAs is just being realized. Genome-wide analyses have identied over 1000 large intergenic noncoding RNAs (lincRNAs) based on signatures of H3K4me3 at transcription start sites and H3K36me3 in transcribed regions (Guttman et al., 2009). RNA immunoprecipitation with PRC2 antibodies identied 24% of the expressed noncoding RNAs as interactors (Khalil et al., 2009). RNAi knockdown of six of these lincRNAs showed significant overlap between genes upregulated by the knockdown of PRC2 components, but no signicant overlap among the six noncoding RNAs (Khalil et al., 2009). One of these, TUG1, is induced upon DNA damage in a P53-dependent manner and is required for the repression of a set of genes involved in cell-cycle regulation (Guttman et al., 2009; Khalil et al., 2009). Subsequent studies aimed at nding P53-induced noncoding RNAs identied lincRNA-p21 as a gene required for repression of genes in the P53 pathway (Huarte et al., 2010). While lincRNA-p21 is located near p21, which is also a repressor in the P53 pathway, their gene targets do not signicantly overlap. LincRNA-p21 interacts with PRC2 similarly to many other lincRNAs, but it also interacts with hnRNP-K (Huarte et al., 2010). hnRNP-K, when interacting with P53, can activate genes. When associating with lincRNAp21, hnRNP-K mediates repression of genes; hnRNP-K localization to these targets requires intact lincRNA-p21 (Huarte et al., 2010). The corepression of genes by hnRNP-K and lincRNAp21 provides a hint as to why so many different noncoding RNAs associate with PRC2 for transcriptional repression. Distinct noncoding RNAs could form unique scaffolds for histone-modifying complexes to associate with gene-specic targeting factors. P53 alone is involved in the upregulation of approximately 30 noncoding RNAs under a variety of conditions and cell types, indicating that the gene regulatory potential for these large noncoding RNAs is enormous (Huarte et al., 2010). Recent work suggests that short RNA transcripts from the PRC2 target genes can form short hairpin structures that directly recruit PRC2 (Kanhere et al., 2010), and small RNAs transcribed from LINE elements on the mammalian X chromosome could play a key role in X inactivation (Chow et al., 2010), indicating that both large and small RNAs contribute to these silencing events.
Molecular Cell
Review
Multiple Targeting Modes within a Complex For many of the complexes weve discussed here, several mechanisms exist for recruiting the same complex. PRC2, for example, can be recruited by its previously deposited H3K27me3 mark for epigenetic memory, it can be recruited through DNA sequences such as PREs, and it can be recruited through interacting with noncoding RNAs. The opposing activity of PRC2 is mediated in part by MLL, which also displays multiple modes of targeting to chromatin. For example, MLL complexes can be recruited through MLLs CXXC domain to CpG islands (Ayton et al., 2004; Cierpicki et al., 2010). Translocation of the MLL gene with a variety of other genes results in chimeric proteins comprised of an N-terminal portion of MLL that includes the CXXC domain. Mutations in MLLs CXXC that prevent DNA binding caused an increase in DNA methylation at the Hoxa9 locus, reduced Hoxa9 expression, and reduced transformation of bone marrow cells by the MLL-AF9 chimera (Ayton et al., 2004; Bach et al., 2009; Cierpicki et al., 2010). It has also recently been proposed that MLL can be recruited to genes through a direct interaction between its CXXC domain and the PAF complex. Remarkably, it was found that only MLL1, and not other Set1-related proteins, could interact with the Paf1 complex (Milne et al., 2010; Muntean et al., 2010). At this time, it is unclear how these data can be reconciled with demonstrated functional differences for the Set1 and MLL complexes in mammals (Wang et al., 2009; Wu et al., 2008). Another domain that may be important for targeting MLL to chromatin is its third PHD nger (PHD3), which specically recognizes H3K4me3 (Chang et al., 2010). MLL also contains AT hooks, which are found in proteins that bind AT-rich sequences (Reeves and Nissen, 1990), although their requirement for recruitment of MLL to Hox or other loci isnt known. Aside from domains that bind DNA sequences or histone modications, MLL could also be recruited to genes through the coactivator model due to its interaction with Menin-LEDGF and their recruitment by nuclear hormone receptors (Dreijerink et al., 2009; Eissenberg and Shilatifard, 2010). Summary and Future Perspectives Histone modications play an indisputably important role in transcription and other DNA-templated processes. However, the identity of the enzyme and the mechanism of recruitment can inuence the effect of the modication. This is exemplied by the diverse functions of the highly related Set1/COMPASS and MLL COMPASS-like complexes, whose biological roles are reected in part by their mode of recruitment. Another example is the differing role of H3S10P at the FOSL1 gene, which when mediated early during activation by MSK1 at the promoter, or later at the enhancer by PIM1, leads to distinct downstream events (Zippo et al., 2009). Furthermore, histone-modifying enzymes can themselves be multifaceted transcription factors, with only one aspect being the ability to modify histones. For example, distinct phenotypes are found for loss of MLL and loss of just the histone-modifying Set domain within MLL, the latter of which yields viable offspring (Terranova et al., 2006). Thus, rather than simply writing or erasing a code that is waiting to be read, histone-modifying activities are integral components of gene regulatory networks in a larger chromatin signaling pathway. A better understanding of the multiple modes of recruitment of these histone-modifying activities is therefore essential for understanding the gene regulatory processes in which they are engaged. The growing number of noncoding RNAs involved in this process provides a challenging yet promising avenue for future research. Identication of the factors that bind to these RNAs could help us better understand how histone-modifying enzymes are targeted for gene activation or repression.
ACKNOWLEDGMENTS We thank Laura Shilatifard for editorial assistance. The studies in the Shilatifard laboratory are supported in part by grants from the National Institute of Health: R01GM069905, R01CA150265, and R01CA89455. REFERENCES Alekseyenko, A.A., Peng, S., Larschan, E., Gorchakov, A.A., Lee, O.K., Kharchenko, P., McGrath, S.D., Wang, C.I., Mardis, E.R., Park, P.J., and Kuroda, M.I. (2008). A sequence motif within chromatin entry sites directs MSL establishment on the Drosophila X chromosome. Cell 134, 599609. Alland, L., Muhle, R., Hou, H., Jr., Potes, J., Chin, L., Schreiber-Agus, N., and DePinho, R.A. (1997). Role for N-CoR and histone deacetylase in Sin3mediated transcriptional repression. Nature 387, 4955. Ayton, P.M., Chen, E.H., and Cleary, M.L. (2004). Binding to nonmethylated CpG DNA is essential for target recognition, transactivation, and myeloid transformation by an MLL oncoprotein. Mol. Cell. Biol. 24, 1047010478. Bach, C., Mueller, D., Buhl, S., Garcia-Cuellar, M.P., and Slany, R.K. (2009). Alterations of the CxxC domain preclude oncogenic activation of mixedlineage leukemia 2. Oncogene 28, 815823. Bannister, A.J., and Kouzarides, T. (1996). The CBP co-activator is a histone acetyltransferase. Nature 384, 641643. Baylin, S., and Bestor, T.H. (2002). Altered methylation patterns in cancer cell genomes: cause or consequence? Cancer Cell 1, 299305. Beckers, J., and Duboule, D. (1998). Genetic analysis of a conserved sequence in the HoxD complex: regulatory redundancy or limitations of the transgenic approach? Dev. Dyn. 213, 111. Bender, L.B., Suh, J., Carroll, C.R., Fong, Y., Fingerman, I.M., Briggs, S.D., Cao, R., Zhang, Y., Reinke, V., and Strome, S. (2006). MES-4: an autosomeassociated histone methyltransferase that participates in silencing the X chromosomes in the C. elegans germ line. Development 133, 39073917. Berger, S.L. (2007). The complex language of chromatin regulation during transcription. Nature 447, 407412. Berger, S.L., Cress, W.D., Cress, A., Triezenberg, S.J., and Guarente, L. (1990). Selective inhibition of activated but not basal transcription by the acidic activation domain of VP16: evidence for transcriptional adaptors. Cell 61, 11991208. Berger, S.L., Pina, B., Silverman, N., Marcus, G.A., Agapite, J., Regier, J.L., Triezenberg, S.J., and Guarente, L. (1992). Genetic isolation of ADA2: a potential transcriptional adaptor required for function of certain acidic activation domains. Cell 70, 251265. Berger, S.L., Kouzarides, T., Shiekhattar, R., and Shilatifard, A. (2009). An operational denition of epigenetics. Genes Dev. 23, 781783. Bernstein, B.E., Meissner, A., and Lander, E.S. (2007). The mammalian epigenome. Cell 128, 669681. Bhaumik, S.R., Smith, E., and Shilatifard, A. (2007). Covalent modications of histones during development and disease pathogenesis. Nat. Struct. Mol. Biol. 14, 10081016. Bird, A., Taggart, M., Frommer, M., Miller, O.J., and Macleod, D. (1985). A fraction of the mouse genome that is derived from islands of nonmethylated, CpG-rich DNA. Cell 40, 9199.
Review
Birke, M., Schreiner, S., Garca-Cuellar, M.P., Mahr, K., Titgemeyer, F., and Slany, R.K. (2002). The MT domain of the proto-oncoprotein MLL binds to CpG-containing DNA and discriminates against methylation. Nucleic Acids Res. 30, 958965. Blackledge, N.P., Zhou, J.C., Tolstorukov, M.Y., Farcas, A.M., Park, P.J., and Klose, R.J. (2010). CpG islands recruit a histone H3 lysine 36 demethylase. Mol. Cell 38, 179190. Brockdorff, N., Ashworth, A., Kay, G.F., McCabe, V.M., Norris, D.P., Cooper, P.J., Swift, S., and Rastan, S. (1992). The product of the mouse Xist gene is a 15 kb inactive X-specic transcript containing no conserved ORF and located in the nucleus. Cell 71, 515526. Brown, C.J., Ballabio, A., Rupert, J.L., Lafreniere, R.G., Grompe, M., Tonlorenzi, R., and Willard, H.F. (1991). A gene from the region of the human X inactivation centre is expressed exclusively from the inactive X chromosome. Nature 349, 3844. Brownell, J.E., Zhou, J., Ranalli, T., Kobayashi, R., Edmondson, D.G., Roth, S.Y., and Allis, C.D. (1996). Tetrahymena histone acetyltransferase A: a homolog to yeast Gcn5p linking histone acetylation to gene activation. Cell 84, 843851. Campos, E.I., and Reinberg, D. (2009). Histones: annotating chromatin. Annu. Rev. Genet. 43, 559599. Carrozza, M.J., Li, B., Florens, L., Suganuma, T., Swanson, S.K., Lee, K.K., Shia, W.J., Anderson, S., Yates, J., Washburn, M.P., and Workman, J.L. (2005). Histone H3 methylation by Set2 directs deacetylation of coding regions by Rpd3S to suppress spurious intragenic transcription. Cell 123, 581592. Chang, P.Y., Hom, R.A., Musselman, C.A., Zhu, L., Kuo, A., Gozani, O., Kutateladze, T.G., and Cleary, M.L. (2010). Binding of the MLL PHD3 nger to histone H3K4me3 is required for MLL-dependent gene transcription. J. Mol. Biol. 400, 137144. Chen, H., Lin, R.J., Schiltz, R.L., Chakravarti, D., Nash, A., Nagy, L., Privalsky, M.L., Nakatani, Y., and Evans, R.M. (1997). Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300. Cell 90, 569580. Cho, Y.W., Hong, T., Hong, S., Guo, H., Yu, H., Kim, D., Guszczynski, T., Dressler, G.R., Copeland, T.D., Kalkum, M., and Ge, K. (2007a). PTIP associates with MLL3- and MLL4-containing histone H3 lysine 4 methyltransferase complex. J. Biol. Chem. 282, 2039520406. Cho, Y.W., Hong, T., Hong, S., Guo, H., Yu, H., Kim, D., Guszczynski, T., Dressler, G.R., Copeland, T.D., Kalkum, M., and Ge, K. (2007b). PTIP associates with MLL3- and MLL4-containing histone H3 lysine 4 methyltransferase complex. J. Biol. Chem. 282, 2039520406. Chow, J., and Heard, E. (2009). X inactivation and the complexities of silencing a sex chromosome. Curr. Opin. Cell Biol. 21, 359366. Chow, J.C., Ciaudo, C., Fazzari, M.J., Mise, N., Servant, N., Glass, J.L., Attreed, M., Avner, P., Wutz, A., Barillot, E., et al. (2010). LINE-1 activity in facultative heterochromatin formation during X chromosome inactivation. Cell 141, 956969. Cierpicki, T., Risner, L.E., Grembecka, J., Lukasik, S.M., Popovic, R., Omonkowska, M., Shultis, D.D., Zeleznik-Le, N.J., and Bushweller, J.H. (2010). Structure of the MLL CXXC domain-DNA complex and its functional role in MLL-AF9 leukemia. Nat. Struct. Mol. Biol. 17, 6268. Clouaire, T., and Stancheva, I. (2008). Methyl-CpG binding proteins: specialized transcriptional repressors or structural components of chromatin? Cell. Mol. Life Sci. 65, 15091522. Core, L.J., Waterfall, J.J., and Lis, J.T. (2008). Nascent RNA sequencing reveals widespread pausing and divergent initiation at human promoters. Science 322, 18451848. Cross, S.H., Meehan, R.R., Nan, X., and Bird, A. (1997). A component of the transcriptional repressor MeCP1 shares a motif with DNA methyltransferase and HRX proteins. Nat. Genet. 16, 256259. Dover, J., Schneider, J., Tawiah-Boateng, M.A., Wood, A., Dean, K., Johnston, M., and Shilatifard, A. (2002). Methylation of histone H3 by COMPASS requires ubiquitination of histone H2B by Rad6. J. Biol. Chem. 277, 2836828371. Dreijerink, K.M., Lips, C.J., and Timmers, H.T. (2009). Multiple endocrine neoplasia type 1: a chromatin writers block. J. Intern. Med. 266, 5359. Drouin, S., Laramee, L., Jacques, P.E., Forest, A., Bergeron, M., and Robert, F. (2010). DSIF and RNA Polymerase II CTD Phosphorylation Coordinate the Recruitment of Rpd3S to Actively Transcribed Genes. PLoS Genet. 6, e1001173. Eissenberg, J.C., and Shilatifard, A. (2010). Histone H3 lysine 4 (H3K4) methylation in development and differentiation. Dev. Biol. 339, 240249. Farkas, G., Gausz, J., Galloni, M., Reuter, G., Gyurkovics, H., and Karch, F. (1994). The Trithorax-like gene encodes the Drosophila GAGA factor. Nature 371, 806808. Fauth, T., Muller-Planitz, F., Konig, C., Straub, T., and Becker, P.B. (2010). The DNA binding CXC domain of MSL2 is required for faithful targeting the Dosage Compensation Complex to the X chromosome. Nucleic Acids Res. 38, 32093221. Frescas, D., Guardavaccaro, D., Bassermann, F., Koyama-Nasu, R., and Pagano, M. (2007). JHDM1B/FBXL10 is a nucleolar protein that represses transcription of ribosomal RNA genes. Nature 450, 309313. Gerber, M., and Shilatifard, A. (2003). Transcriptional elongation by RNA polymerase II and histone methylation. J. Biol. Chem. 278, 2630326306. Gilmour, D.S. (2009). Promoter proximal pausing on genes in metazoans. Chromosoma 118, 110. Goo, Y.H., Sohn, Y.C., Kim, D.H., Kim, S.W., Kang, M.J., Jung, D.J., Kwak, E., Barlev, N.A., Berger, S.L., Chow, V.T., et al. (2003). Activating signal cointegrator 2 belongs to a novel steady-state complex that contains a subset of trithorax group proteins. Mol. Cell. Biol. 23, 140149. Govind, C.K., Qiu, H., Ginsburg, D.S., Ruan, C., Hofmeyer, K., Hu, C., Swaminathan, V., Workman, J.L., Li, B., and Hinnebusch, A.G. (2010). Phosphorylated Pol II CTD recruits multiple HDACs, including Rpd3C(S), for methylation-dependent deacetylation of ORF nucleosomes. Mol. Cell 39, 234246. Gupta, R.A., Shah, N., Wang, K.C., Kim, J., Horlings, H.M., Wong, D.J., Tsai, M.C., Hung, T., Argani, P., Rinn, J.L., et al. (2010). Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis. Nature 464, 10711076. Guttman, M., Amit, I., Garber, M., French, C., Lin, M.F., Feldser, D., Huarte, M., Zuk, O., Carey, B.W., Cassady, J.P., et al. (2009). Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals. Nature 458, 223227. Hansen, K.H., Bracken, A.P., Pasini, D., Dietrich, N., Gehani, S.S., Monrad, A., Rappsilber, J., Lerdrup, M., and Helin, K. (2008). A model for transmission of the H3K27me3 epigenetic mark. Nat. Cell Biol. 10, 12911300. Hassig, C.A., Fleischer, T.C., Billin, A.N., Schreiber, S.L., and Ayer, D.E. (1997). Histone deacetylase activity is required for full transcriptional repression by mSin3A. Cell 89, 341347. Heinzel, T., Lavinsky, R.M., Mullen, T.M., Soderstrom, M., Laherty, C.D., Torchia, J., Yang, W.M., Brard, G., Ngo, S.D., Davie, J.R., et al. (1997). A complex containing N-CoR, mSin3 and histone deacetylase mediates transcriptional repression. Nature 387, 4348. Henry, K.W., Wyce, A., Lo, W.S., Duggan, L.J., Emre, N.C., Kao, C.F., Pillus, L., Shilatifard, A., Osley, M.A., and Berger, S.L. (2003). Transcriptional activation via sequential histone H2B ubiquitylation and deubiquitylation, mediated by SAGA-associated Ubp8. Genes Dev. 17, 26482663. Herz, H.M., and Shilatifard, A. (2010). The JARID2-PRC2 duality. Genes Dev. 24, 857861. Huarte, M., Guttman, M., Feldser, D., Garber, M., Koziol, M.J., KenzelmannBroz, D., Khalil, A.M., Zuk, O., Amit, I., Rabani, M., et al. (2010). A large intergenic noncoding RNA induced by p53 mediates global gene repression in the p53 response. Cell 142, 409419. Hughes, C.M., Rozenblatt-Rosen, O., Milne, T.A., Copeland, T.D., Levine, S.S., Lee, J.C., Hayes, D.N., Shanmugam, K.S., Bhattacharjee, A., Biondi, C.A., et al. (2004). Menin associates with a trithorax family histone methyltransferase complex and with the hoxc8 locus. Mol. Cell 13, 587597.
Molecular Cell
Molecular Cell
Review
Ilik, I., and Akhtar, A. (2009). roX RNAs: non-coding regulators of the male X chromosome in ies. RNA Biol. 6, 113121. Illingworth, R.S., and Bird, A.P. (2009). CpG islandsa rough guide. FEBS Lett. 583, 17131720. Issaeva, I., Zonis, Y., Rozovskaia, T., Orlovsky, K., Croce, C.M., Nakamura, T., Mazo, A., Eisenbach, L., and Canaani, E. (2007). Knockdown of ALR (MLL2) reveals ALR target genes and leads to alterations in cell adhesion and growth. Mol. Cell. Biol. 27, 18891903. Jrgensen, H.F., Ben-Porath, I., and Bird, A.P. (2004). Mbd1 is recruited to both methylated and nonmethylated CpGs via distinct DNA binding domains. Mol. Cell. Biol. 24, 33873395. Kadosh, D., and Struhl, K. (1997). Repression by Ume6 involves recruitment of a complex containing Sin3 corepressor and Rpd3 histone deacetylase to target promoters. Cell 89, 365371. Kanhere, A., Viiri, K., Araujo, C.C., Rasaiyaah, J., Bouwman, R.D., Whyte, W.A., Pereira, C.F., Brookes, E., Walker, K., Bell, G.W., et al. (2010). Short RNAs are transcribed from repressed polycomb target genes and interact with polycomb repressive complex-2. Mol. Cell 38, 675688. Kelley, R.L., Meller, V.H., Gordadze, P.R., Roman, G., Davis, R.L., and Kuroda, M.I. (1999). Epigenetic spreading of the Drosophila dosage compensation complex from roX RNA genes into anking chromatin. Cell 98, 513522. Keogh, M.C., Kurdistani, S.K., Morris, S.A., Ahn, S.H., Podolny, V., Collins, S.R., Schuldiner, M., Chin, K., Punna, T., Thompson, N.J., et al. (2005). Cotranscriptional set2 methylation of histone H3 lysine 36 recruits a repressive Rpd3 complex. Cell 123, 593605. Khalil, A.M., Guttman, M., Huarte, M., Garber, M., Raj, A., Rivea Morales, D., Thomas, K., Presser, A., Bernstein, B.E., van Oudenaarden, A., et al. (2009). Many human large intergenic noncoding RNAs associate with chromatinmodifying complexes and affect gene expression. Proc. Natl. Acad. Sci. USA 106, 1166711672. Kim, J., Sif, S., Jones, B., Jackson, A., Koipally, J., Heller, E., Winandy, S., Viel, A., Sawyer, A., Ikeda, T., et al. (1999). Ikaros DNA-binding proteins direct formation of chromatin remodeling complexes in lymphocytes. Immunity 10, 345355. Kim, J., Guermah, M., McGinty, R.K., Lee, J.S., Tang, Z., Milne, T.A., Shilatifard, A., Muir, T.W., and Roeder, R.G. (2009). RAD6-Mediated transcription-coupled H2B ubiquitylation directly stimulates H3K4 methylation in human cells. Cell 137, 459471. Koipally, J., Renold, A., Kim, J., and Georgopoulos, K. (1999). Repression by Ikaros and Aiolos is mediated through histone deacetylase complexes. EMBO J. 18, 30903100. Kornberg, R.D. (1974). Chromatin structure: a repeating unit of histones and DNA. Science 184, 868871. Kornberg, R.D., and Lorch, Y. (1999). Twenty-ve years of the nucleosome, fundamental particle of the eukaryote chromosome. Cell 98, 285294. Kouzarides, T. (2007). Chromatin modications and their function. Cell 128, 693705. Krogan, N.J., Dover, J., Khorrami, S., Greenblatt, J.F., Schneider, J., Johnston, M., and Shilatifard, A. (2002). COMPASS, a histone H3 (Lysine 4) methyltransferase required for telomeric silencing of gene expression. J. Biol. Chem. 277, 1075310755. Krogan, N.J., Dover, J., Wood, A., Schneider, J., Heidt, J., Boateng, M.A., Dean, K., Ryan, O.W., Golshani, A., Johnston, M., et al. (2003a). The Paf1 complex is required for histone H3 methylation by COMPASS and Dot1p: linking transcriptional elongation to histone methylation. Mol. Cell 11, 721729. Krogan, N.J., Kim, M., Tong, A., Golshani, A., Cagney, G., Canadien, V., Richards, D.P., Beattie, B.K., Emili, A., Boone, C., et al. (2003b). Methylation of histone H3 by Set2 in Saccharomyces cerevisiae is linked to transcriptional elongation by RNA polymerase II. Mol. Cell. Biol. 23, 42074218. Ku, M., Koche, R.P., Rheinbay, E., Mendenhall, E.M., Endoh, M., Mikkelsen, T.S., Presser, A., Nusbaum, C., Xie, X., Chi, A.S., et al. (2008). Genomewide analysis of PRC1 and PRC2 occupancy identies two classes of bivalent domains. PLoS Genet. 4, e1000242. Kusch, T., Florens, L., Macdonald, W.H., Swanson, S.K., Glaser, R.L., Yates, J.R., 3rd, Abmayr, S.M., Washburn, M.P., and Workman, J.L. (2004). Acetylation by Tip60 is required for selective histone variant exchange at DNA lesions. Science 306, 20842087. Landeira, D., Sauer, S., Poot, R., Dvorkina, M., Mazzarella, L., Jrgensen, H.F., Pereira, C.F., Leleu, M., Piccolo, F.M., Spivakov, M., et al. (2010). Jarid2 is a PRC2 component in embryonic stem cells required for multi-lineage differentiation and recruitment of PRC1 and RNA Polymerase II to developmental regulators. Nat. Cell Biol. 12, 618624. Lee, J.H., and Skalnik, D.G. (2005). CpG-binding protein (CXXC nger protein 1) is a component of the mammalian Set1 histone H3-Lys4 methyltransferase complex, the analogue of the yeast Set1/COMPASS complex. J. Biol. Chem. 280, 4172541731. Lee, J.H., and Skalnik, D.G. (2008). Wdr82 is a C-terminal domain-binding protein that recruits the Setd1A Histone H3-Lys4 methyltransferase complex to transcription start sites of transcribed human genes. Mol. Cell. Biol. 28, 609618. Lee, J.H., Voo, K.S., and Skalnik, D.G. (2001). Identication and characterization of the DNA binding domain of CpG-binding protein. J. Biol. Chem. 276, 4466944676. Lee, M.G., Wynder, C., Cooch, N., and Shiekhattar, R. (2005). An essential role for CoREST in nucleosomal histone 3 lysine 4 demethylation. Nature 437, 432435. Lee, J.H., Tate, C.M., You, J.S., and Skalnik, D.G. (2007a). Identication and characterization of the human Set1B histone H3-Lys4 methyltransferase complex. J. Biol. Chem. 282, 1341913428. Lee, J.S., Shukla, A., Schneider, J., Swanson, S.K., Washburn, M.P., Florens, L., Bhaumik, S.R., and Shilatifard, A. (2007b). Histone crosstalk between H2B monoubiquitination and H3 methylation mediated by COMPASS. Cell 131, 10841096. Lee, N., Erdjument-Bromage, H., Tempst, P., Jones, R.S., and Zhang, Y. (2009). The H3K4 demethylase lid associates with and inhibits histone deacetylase Rpd3. Mol. Cell. Biol. 29, 14011410. Lee, J.S., Smith, E., and Shilatifard, A. (2010). The language of histone crosstalk. Cell 142, 682685. Li, B., Gogol, M., Carey, M., Lee, D., Seidel, C., and Workman, J.L. (2007). Combined action of PHD and chromo domains directs the Rpd3S HDAC to transcribed chromatin. Science 316, 10501054. Li, B., Jackson, J., Simon, M.D., Fleharty, B., Gogol, M., Seidel, C., Workman, J.L., and Shilatifard, A. (2009). Histone H3 lysine 36 dimethylation (H3K36me2) is sufcient to recruit the Rpd3s histone deacetylase complex and to repress spurious transcription. J. Biol. Chem. 284, 79707976. Li, G., Margueron, R., Ku, M., Chambon, P., Bernstein, B.E., and Reinberg, D. (2010). Jarid2 and PRC2, partners in regulating gene expression. Genes Dev. 24, 368380. Luger, K., Mader, A.W., Richmond, R.K., Sargent, D.F., and Richmond, T.J. (1997). Crystal structure of the nucleosome core particle at 2.8 A resolution. Nature 389, 251260. Mancini-Dinardo, D., Steele, S.J., Levorse, J.M., Ingram, R.S., and Tilghman, S.M. (2006). Elongation of the Kcnq1ot1 transcript is required for genomic imprinting of neighboring genes. Genes Dev. 20, 12681282. Marcus, G.A., Silverman, N., Berger, S.L., Horiuchi, J., and Guarente, L. (1994). Functional similarity and physical association between GCN5 and ADA2: putative transcriptional adaptors. EMBO J. 13, 48074815. Margueron, R., Justin, N., Ohno, K., Sharpe, M.L., Son, J., Drury, W.J., 3rd, Voigt, P., Martin, S.R., Taylor, W.R., De Marco, V., et al. (2009). Role of the polycomb protein EED in the propagation of repressive histone marks. Nature 461, 762767. Matthews, A.G., Kuo, A.J., Ramon-Maiques, S., Han, S., Champagne, K.S., Ivanov, D., Gallardo, M., Carney, D., Cheung, P., Ciccone, D.N., et al. (2007). RAG2 PHD nger couples histone H3 lysine 4 trimethylation with V(D)J recombination. Nature 450, 11061110.
Review
Maurer-Stroh, S., Dickens, N.J., Hughes-Davies, L., Kouzarides, T., Eisenhaber, F., and Ponting, C.P. (2003). The Tudor domain Royal Family: Tudor, plant Agenet, Chromo, PWWP and MBT domains. Trends Biochem. Sci. 28, 6974. McGinty, R.K., Kim, J., Chatterjee, C., Roeder, R.G., and Muir, T.W. (2008). Chemically ubiquitylated histone H2B stimulates hDot1L-mediated intranucleosomal methylation. Nature 453, 812816. Miller, T., Krogan, N.J., Dover, J., Erdjument-Bromage, H., Tempst, P., Johnston, M., Greenblatt, J.F., and Shilatifard, A. (2001). COMPASS: a complex of proteins associated with a trithorax-related SET domain protein. Proc. Natl. Acad. Sci. USA 98, 1290212907. Milne, T.A., Kim, J., Wang, G.G., Stadler, S.C., Basrur, V., Whitcomb, S.J., Wang, Z., Ruthenburg, A.J., Elenitoba-Johnson, K.S., Roeder, R.G., and Allis, C.D. (2010). Multiple interactions recruit MLL1 and MLL1 fusion proteins to the HOXA9 locus in leukemogenesis. Mol. Cell 38, 853863. Moshkin, Y.M., Kan, T.W., Goodfellow, H., Bezstarosti, K., Maeda, R.K., Pilyugin, M., Karch, F., Bray, S.J., Demmers, J.A., and Verrijzer, C.P. (2009). Histone chaperones ASF1 and NAP1 differentially modulate removal of active histone marks by LID-RPD3 complexes during NOTCH silencing. Mol. Cell 35, 782793. Muntean, A.G., Tan, J., Sitwala, K., Huang, Y., Bronstein, J., Connelly, J.A., Basrur, V., Elenitoba-Johnson, K.S., and Hess, J.L. (2010). The PAF complex synergizes with MLL fusion proteins at HOX loci to promote leukemogenesis. Cancer Cell 17, 609621. Nagano, T., Mitchell, J.A., Sanz, L.A., Pauler, F.M., Ferguson-Smith, A.C., Feil, R., and Fraser, P. (2008). The Air noncoding RNA epigenetically silences transcription by targeting G9a to chromatin. Science 322, 17171720. Nagy, L., Kao, H.Y., Chakravarti, D., Lin, R.J., Hassig, C.A., Ayer, D.E., Schreiber, S.L., and Evans, R.M. (1997). Nuclear receptor repression mediated by a complex containing SMRT, mSin3A, and histone deacetylase. Cell 89, 373380. Nakajima, T., Uchida, C., Anderson, S.F., Lee, C.G., Hurwitz, J., Parvin, J.D., and Montminy, M. (1997). RNA helicase A mediates association of CBP with RNA polymerase II. Cell 90, 11071112. Ng, H.H., Dole, S., and Struhl, K. (2003a). The Rtf1 component of the Paf1 transcriptional elongation complex is required for ubiquitination of histone H2B. J. Biol. Chem. 278, 3362533628. Ng, H.H., Robert, F., Young, R.A., and Struhl, K. (2003b). Targeted recruitment of Set1 histone methylase by elongating Pol II provides a localized mark and memory of recent transcriptional activity. Mol. Cell 11, 709719. Ogryzko, V.V., Schiltz, R.L., Russanova, V., Howard, B.H., and Nakatani, Y. (1996). The transcriptional coactivators p300 and CBP are histone acetyltransferases. Cell 87, 953959. Pandey, R.R., Mondal, T., Mohammad, F., Enroth, S., Redrup, L., Komorowski, J., Nagano, T., Mancini-Dinardo, D., and Kanduri, C. (2008). Kcnq1ot1 antisense noncoding RNA mediates lineage-specic transcriptional silencing through chromatin-level regulation. Mol. Cell 32, 232246. Park, Y., Oh, H., Meller, V.H., and Kuroda, M.I. (2005). Variable splicing of noncoding roX2 RNAs inuences targeting of MSL dosage compensation complexes in Drosophila. RNA Biol. 2, 157164. Pasini, D., Cloos, P.A., Walfridsson, J., Olsson, L., Bukowski, J.P., Johansen, J.V., Bak, M., Tommerup, N., Rappsilber, J., and Helin, K. (2010). JARID2 regulates binding of the Polycomb repressive complex 2 to target genes in ES cells. Nature 464, 306310. Patel, S.R., Kim, D., Levitan, I., and Dressler, G.R. (2007). The BRCT-domain containing protein PTIP links PAX2 to a histone H3, lysine 4 methyltransferase complex. Dev. Cell 13, 580592. Pavri, R., Zhu, B., Li, G., Trojer, P., Mandal, S., Shilatifard, A., and Reinberg, D. (2006). Histone H2B monoubiquitination functions cooperatively with FACT to regulate elongation by RNA polymerase II. Cell 125, 703717. Pena, P.V., Davrazou, F., Shi, X., Walter, K.L., Verkhusha, V.V., Gozani, O., Zhao, R., and Kutateladze, T.G. (2006). Molecular mechanism of histone H3K4me3 recognition by plant homeodomain of ING2. Nature 442, 100103. Peng, J.C., Valouev, A., Swigut, T., Zhang, J., Zhao, Y., Sidow, A., and Wysocka, J. (2009). Jarid2/Jumonji coordinates control of PRC2 enzymatic activity and target gene occupancy in pluripotent cells. Cell 139, 12901302. ` Pradhan, M., Esteve, P.O., Chin, H.G., Samaranayke, M., Kim, G.D., and Pradhan, S. (2008). CXXC domain of human DNMT1 is essential for enzymatic activity. Biochemistry 47, 1000010009. Pugh, B.F., and Tjian, R. (1990). Mechanism of transcriptional activation by Sp1: evidence for coactivators. Cell 61, 11871197. Reeves, R., and Nissen, M.S. (1990). The A.T-DNA-binding domain of mammalian high mobility group I chromosomal proteins. A novel peptide motif for recognizing DNA structure. J. Biol. Chem. 265, 85738582. Ringrose, L., and Paro, R. (2004). Epigenetic regulation of cellular memory by the Polycomb and Trithorax group proteins. Annu. Rev. Genet. 38, 413443. Ringrose, L., and Paro, R. (2007). Polycomb/Trithorax response elements and epigenetic memory of cell identity. Development 134, 223232. Ringrose, L., Rehmsmeier, M., Dura, J.M., and Paro, R. (2003). Genome-wide prediction of Polycomb/Trithorax response elements in Drosophila melanogaster. Dev. Cell 5, 759771. Rinn, J.L., Kertesz, M., Wang, J.K., Squazzo, S.L., Xu, X., Brugmann, S.A., Goodnough, L.H., Helms, J.A., Farnham, P.J., Segal, E., and Chang, H.Y. (2007). Functional demarcation of active and silent chromatin domains in human HOX loci by noncoding RNAs. Cell 129, 13111323. Roguev, A., Schaft, D., Shevchenko, A., Pijnappel, W.W., Wilm, M., Aasland, R., and Stewart, A.F. (2001). The Saccharomyces cerevisiae Set1 complex includes an Ash2 homologue and methylates histone 3 lysine 4. EMBO J. 20, 71377148. Ruthenburg, A.J., Allis, C.D., and Wysocka, J. (2007). Methylation of lysine 4 on histone H3: intricacy of writing and reading a single epigenetic mark. Mol. Cell 25, 1530. Sarraf, S.A., and Stancheva, I. (2004). Methyl-CpG binding protein MBD1 couples histone H3 methylation at lysine 9 by SETDB1 to DNA replication and chromatin assembly. Mol. Cell 15, 595605. Schreiber, S.L., and Bernstein, B.E. (2002). Signaling network model of chromatin. Cell 111, 771778. Schuettengruber, B., Chourrout, D., Vervoort, M., Leblanc, B., and Cavalli, G. (2007). Genome regulation by polycomb and trithorax proteins. Cell 128, 735745. Schwartz, Y.B., and Pirrotta, V. (2007). Polycomb silencing mechanisms and the management of genomic programmes. Nat. Rev. Genet. 8, 922. Schwartz, Y.B., Kahn, T.G., Nix, D.A., Li, X.Y., Bourgon, R., Biggin, M., and Pirrotta, V. (2006). Genome-wide analysis of Polycomb targets in Drosophila melanogaster. Nat. Genet. 38, 700705. Senner, C.E., and Brockdorff, N. (2009). Xist gene regulation at the onset of X inactivation. Curr. Opin. Genet. Dev. 19, 122126. Shi, Y., Lan, F., Matson, C., Mulligan, P., Whetstine, J.R., Cole, P.A., Casero, R.A., and Shi, Y. (2004). Histone demethylation mediated by the nuclear amine oxidase homolog LSD1. Cell 119, 941953. Shilatifard, A. (2006). Chromatin modications by methylation and ubiquitination: implications in the regulation of gene expression. Annu. Rev. Biochem. 75, 243269. Shilatifard, A. (2008). Molecular implementation and physiological roles for histone H3 lysine 4 (H3K4) methylation. Curr. Opin. Cell Biol. 20, 341348. Simon, J.A., and Kingston, R.E. (2009). Mechanisms of polycomb gene silencing: knowns and unknowns. Nat. Rev. Mol. Cell Biol. 10, 697708. Sims, R.J., 3rd, and Reinberg, D. (2008). Is there a code embedded in proteins that is based on post-translational modications? Nat. Rev. Mol. Cell Biol. 9, 815820. Sleutels, F., Zwart, R., and Barlow, D.P. (2002). The non-coding Air RNA is required for silencing autosomal imprinted genes. Nature 415, 810813.
Molecular Cell
Molecular Cell
Review
Smith, E.R., Pannuti, A., Gu, W., Steurnagel, A., Cook, R.G., Allis, C.D., and Lucchesi, J.C. (2000). The drosophila MSL complex acetylates histone H4 at lysine 16, a chromatin modication linked to dosage compensation. Mol. Cell. Biol. 20, 312318. Smith, E.R., Allis, C.D., and Lucchesi, J.C. (2001). Linking global histone acetylation to the transcription enhancement of X-chromosomal genes in Drosophila males. J. Biol. Chem. 276, 3148331486. Spain, M.M., Caruso, J.A., Swaminathan, A., and Pile, L.A. (2010). Drosophila SIN3 isoforms interact with distinct proteins and have unique biological functions. J. Biol. Chem. 285, 2745727467. Spencer, T.E., Jenster, G., Burcin, M.M., Allis, C.D., Zhou, J., Mizzen, C.A., McKenna, N.J., Onate, S.A., Tsai, S.Y., Tsai, M.J., and OMalley, B.W. (1997). Steroid receptor coactivator-1 is a histone acetyltransferase. Nature 389, 194198. Straub, T., Grimaud, C., Gilllan, G.D., Mitterweger, A., and Becker, P.B. (2008). The chromosomal high-afnity binding sites for the Drosophila dosage compensation complex. PLoS Genet. 4, e1000302. Sural, T.H., Peng, S., Li, B., Workman, J.L., Park, P.J., and Kuroda, M.I. (2008). The MSL3 chromodomain directs a key targeting step for dosage compensation of the Drosophila melanogaster X chromosome. Nat. Struct. Mol. Biol. 15, 13181325. Takai, D., and Jones, P.A. (2002). Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc. Natl. Acad. Sci. USA 99, 37403745. Tanaka, Y., Okamoto, K., Teye, K., Umata, T., Yamagiwa, N., Suto, Y., Zhang, Y., and Tsuneoka, M. (2010). JmjC enzyme KDM2A is a regulator of rRNA transcription in response to starvation. EMBO J. 29, 15101522. Tate, C.M., Lee, J.H., and Skalnik, D.G. (2009). CXXC nger protein 1 contains redundant functional domains that support embryonic stem cell cytosine methylation, histone methylation, and differentiation. Mol. Cell. Biol. 29, 38173831. Taunton, J., Hassig, C.A., and Schreiber, S.L. (1996). A mammalian histone deacetylase related to the yeast transcriptional regulator Rpd3p. Science 272, 408411. Taverna, S.D., Li, H., Ruthenburg, A.J., Allis, C.D., and Patel, D.J. (2007). How chromatin-binding modules interpret histone modications: lessons from professional pocket pickers. Nat. Struct. Mol. Biol. 14, 10251040. Tenney, K., Gerber, M., Ilvarsonn, A., Schneider, J., Gause, M., Dorsett, D., Eissenberg, J.C., and Shilatifard, A. (2006). Drosophila Rtf1 functions in histone methylation, gene expression, and Notch signaling. Proc. Natl. Acad. Sci. USA 103, 1197011974. Terranova, R., Agherbi, H., Boned, A., Meresse, S., and Djabali, M. (2006). Histone and DNA methylation defects at Hox genes in mice expressing a SET domain-truncated form of Mll. Proc. Natl. Acad. Sci. USA 103, 6629 6634. Thomson, J.P., Skene, P.J., Selfridge, J., Clouaire, T., Guy, J., Webb, S., Kerr, A.R., Deaton, A., Andrews, R., James, K.D., et al. (2010). CpG islands inuence chromatin structure via the CpG-binding protein Cfp1. Nature 464, 10821086. Tominaga, K., Leung, J.K., Rookard, P., Echigo, J., Smith, J.R., and PereiraSmith, O.M. (2003). MRGX is a novel transcriptional regulator that exhibits activation or repression of the B-myb promoter in a cell type-dependent manner. J. Biol. Chem. 278, 4961849624. Tsai, M.C., Manor, O., Wan, Y., Mosammaparast, N., Wang, J.K., Lan, F., Shi, Y., Segal, E., and Chang, H.Y. (2010). Long noncoding RNA as modular scaffold of histone modication complexes. Science 329, 689693. Tsukada, Y., Fang, J., Erdjument-Bromage, H., Warren, M.E., Borchers, C.H., Tempst, P., and Zhang, Y. (2006). Histone demethylation by a family of JmjC domain-containing proteins. Nature 439, 811816. Utley, R.T., Ikeda, K., Grant, P.A., Cote, J., Steger, D.J., Eberharter, A., John, S., and Workman, J.L. (1998). Transcriptional activators direct histone acetyltransferase complexes to nucleosomes. Nature 394, 498502. Vermeulen, M., Mulder, K.W., Denissov, S., Pijnappel, W.W., van Schaik, F.M., Varier, R.A., Baltissen, M.P., Stunnenberg, H.G., Mann, M., and Timmers, H.T. (2007). Selective anchoring of TFIID to nucleosomes by trimethylation of histone H3 lysine 4. Cell 131, 5869. Wang, L., Brown, J.L., Cao, R., Zhang, Y., Kassis, J.A., and Jones, R.S. (2004). Hierarchical recruitment of polycomb group silencing complexes. Mol. Cell 14, 637646. Wang, P., Bowl, M.R., Bender, S., Peng, J., Farber, L., Chen, J., Ali, A., Zhang, Z., Alberts, A.S., Thakker, R.V., et al. (2008). Parabromin, a component of the human PAF complex, regulates growth factors and is required for embryonic development and survival in adult mice. Mol. Cell. Biol. 28, 29302940. Wang, P., Lin, C., Smith, E.R., Guo, H., Sanderson, B.W., Wu, M., Gogol, M., Alexander, T., Seidel, C., Wiedemann, L.M., et al. (2009). Global analysis of H3K4 methylation denes MLL family member targets and points to a role for MLL1-mediated H3K4 methylation in the regulation of transcriptional initiation by RNA polymerase II. Mol. Cell. Biol. 29, 60746085. Weake, V.M., and Workman, J.L. (2008). Histone ubiquitination: triggering gene activity. Mol. Cell 29, 653663. Woo, C.J., Kharchenko, P.V., Daheron, L., Park, P.J., and Kingston, R.E. (2010). A region of the human HOXD cluster that confers polycomb-group responsiveness. Cell 140, 99110. Wood, A., Schneider, J., Dover, J., Johnston, M., and Shilatifard, A. (2003). The Paf1 complex is essential for histone monoubiquitination by the Rad6-Bre1 complex, which signals for histone methylation by COMPASS and Dot1p. J. Biol. Chem. 278, 3473934742. Workman, J.L., and Kingston, R.E. (1998). Alteration of nucleosome structure as a mechanism of transcriptional regulation. Annu. Rev. Biochem. 67, 545579. Wu, M., Wang, P.F., Lee, J.S., Martin-Brown, S., Florens, L., Washburn, M., and Shilatifard, A. (2008). Molecular regulation of H3K4 trimethylation by Wdr82, a component of human Set1/COMPASS. Mol. Cell. Biol. 28, 7337 7344. Xiao, T., Hall, H., Kizer, K.O., Shibata, Y., Hall, M.C., Borchers, C.H., and Strahl, B.D. (2003). Phosphorylation of RNA polymerase II CTD regulates H3 methylation in yeast. Genes Dev. 17, 654663. Yaffe, M.B., and Elia, A.E. (2001). Phosphoserine/threonine-binding domains. Curr. Opin. Cell Biol. 13, 131138. Yochum, G.S., and Ayer, D.E. (2002). Role for the mortality factors MORF4, MRGX, and MRG15 in transcriptional repression via associations with Pf1, mSin3A, and Transducin-Like Enhancer of Split. Mol. Cell. Biol. 22, 7868 7876. Youdell, M.L., Kizer, K.O., Kisseleva-Romanova, E., Fuchs, S.M., Duro, E., Strahl, B.D., and Mellor, J. (2008). Roles for Ctk1 and Spt6 in regulating the different methylation states of histone H3 lysine 36. Mol. Cell. Biol. 28, 49154926. Zhao, J., Sun, B.K., Erwin, J.A., Song, J.J., and Lee, J.T. (2008). Polycomb proteins targeted by a short repeat RNA to the mouse X chromosome. Science 322, 750756. Zheng, S., Wyrick, J.J., and Reese, J.C. (2010). Novel trans-tail regulation of H2B ubiquitylation and H3K4 methylation by the N terminus of histone H2A. Mol. Cell. Biol. 30, 36353645. Zhu, B., Zheng, Y., Pham, A.D., Mandal, S.S., Erdjument-Bromage, H., Tempst, P., and Reinberg, D. (2005). Monoubiquitination of human histone H2B: the factors involved and their roles in HOX gene regulation. Mol. Cell 20, 601611. Zippo, A., Serani, R., Rocchigiani, M., Pennacchini, S., Krepelova, A., and Oliviero, S. (2009). Histone crosstalk between H3S10ph and H4K16ac generates a histone code that mediates transcription elongation. Cell 138, 11221136.
Molecular Cell
Article
Epigenetic Instability due to Defective Replication of Structured DNA
Peter Sarkies,1 Charlie Reams,2 Laura J. Simpson,1 and Julian E. Sale1,*
Research Council Laboratory of Molecular Biology, Hills Road, Cambridge CB2 0QH, UK of Cambridge Computer Laboratory, William Gates Building, 15, J.J. Thomson Avenue, Cambridge CB3 0FD, UK *Correspondence: jes@mrc-lmb.cam.ac.uk DOI 10.1016/j.molcel.2010.11.009
2University 1Medical
SUMMARY
The accurate propagation of histone marks during chromosomal replication is proposed to rely on the tight coupling of replication with the recycling of parental histones to the daughter strands. Here, we show in the avian cell line DT40 that REV1, a key regulator of DNA translesion synthesis at the replication fork, is required for the maintenance of repressive chromatin marks and gene silencing in the vicinity of DNA capable of forming G-quadruplex (G4) structures. We demonstrate a previously unappreciated requirement for REV1 in replication of G4 forming sequences and show that transplanting a G4 forming sequence into a silent locus leads to its derepression in REV1-decient cells. Together, our observations support a model in which failure to maintain processive DNA replication at G4 DNA in REV1-decient cells leads to uncoupling of DNA synthesis from histone recycling, resulting in localized loss of repressive chromatin through biased incorporation of newly synthesized histones.
INTRODUCTION Multicellular organisms must maintain gene expression states, and therefore cell identity, epigenetically through cell division (Corpet and Almouzni, 2009). It is proposed that this is achieved through posttranslational modication of the histone proteins around which the DNA is wrapped in chromatin. During replication, histones are displaced by the replicative helicase and then randomly distributed to the nascent daughter DNA strands in a process coordinated by histone chaperones, notably Asf1 and Caf1 (reviewed in De Koning et al., 2007). To avoid a reduction in nucleosome density, recycled histones are combined with newly synthesized histones. The modications on the parental histones can then be copied to the new histones (Bannister et al., 2001; Hansen et al., 2008; Lachner et al., 2001; Margueron and Reinberg, 2010). In order for this process to be viable as a mechanism of maintaining gene expression states, it is essential that there be coordination between histone eviction by the replicative helicase and the synthesis of new DNA. Without this coordination, parental histones will not be deposited near to their original locations, resulting in loss of the epigenetic information carried by their posttranslational modications.
DNA replication is susceptible to interruptions caused by, for example, DNA damage. In turn, replication arrest interrupts histone recycling (Jasencakova et al., 2010). The Y family DNA polymerase REV1 plays an important role in vertebrates in maintaining replication fork progression on damaged DNA templates (Edmunds et al., 2008; Jansen et al., 2009), a role it fulls by coordinating other specialized polymerases that are able to bypass DNA damage directly in a process known as translesion synthesis (Guo et al., 2003; Ross et al., 2005). In REV1-decient cells, this results in damage bypass taking place predominantly in postreplicative gaps (Edmunds et al., 2008), which form when replication restarts downstream of a block. Importantly, the DNA synthesis associated with the lling of these gaps, which have been estimated to be between about 400 and 3000 bp in length and visualized to persist up to 20 kb behind the fork (Lehmann, 1972; Lopes et al., 2006), will be uncoupled from bulk DNA replication and therefore from the replicative helicase and histone recycling. Indeed, in budding yeast, gap lling can be deferred to G2 (Daigaku et al., 2010; Karras and Jentsch, 2010). Thus, chromatinization of these regions will be likely to exhibit a bias toward the deposition of newly synthesized histones and therefore result in the formation of a tract of nucleosomes lacking key epigenetic marks present in the parental strands. We postulated that REV1-decient cells would be more liable to loss of coordination between the replicative helicase and DNA synthesis, and that this might lead to loss of chromatin modications through the preferential incorporation of new, unmodied histones during gap lling. To test this hypothesis, we have taken advantage of the chicken b-globin locus, in which the histone modications associated with developmentally regulated expression have been extensively studied (reviewed in Felsenfeld, 1993; Felsenfeld et al., 2004) and the genetic tractability of the chicken cell line DT40 (Buerstedde and Takeda, 1991). We therefore set out to ask initially whether loss of REV1 had any impact on the ability of DT40 to maintain repression of the b-globin locus. RESULTS Loss of H3K9 Dimethylation in the b-Globin Locus of rev1 Cells Is Associated with an Increase in Marks of New Histone Deposition To test the hypothesis that rev1 cells might lose epigenetic information, we used chromatin immunoprecipitation (ChIP) to examine the histone modications at the b-globin locus (Figure 1A), which has been previously shown to be silent in nonerythroid cells (Litt et al., 2001a), and specically in DT40 (Litt et al.,
Molecular Cell 40, 703713, December 10, 2010 2010 Elsevier Inc. 703
Molecular Cell
REV1 Maintains Epigenetic Stability at G4 DNA
FR gene HSA
constitutively condensed chromatin
-globin domain HS4
Figure 1. Epigenetic Dysfunction in the b-Globin Locus of rev1 DT40

(A) Map of the region of the chicken b-globin locus studied in this paper (Litt et al., 2001b) between the folate receptor (FR) gene and rst of the b-globin genes, r. HSA and HS4 are DNase hypersensitive sites that correspond with chromatin domain insulator sequences (Felsenfeld et al., 2004). The distance markers and location of the ChIP primers are indicated below the diagram (see also Table S2). (B) H3K9 dimethylation (H3K9me2) in the WT (solid line) and pcnaK164R (dashed line). In all cases, the specic ChIP signal was normalized to total H3 then to the signal at HS4 (21.37). Error bars represent the standard error of the mean. (C) H3K9 dimethylation (H3K9me2) in rev1 cells derived in our laboratory (solid line, rev1 Cambridge) (Simpson and Sale, 2003) and independently the laboratory of JeanMarie Buerstedde (dashed line, rev1 Munich) (Arakawa et al., 2006). (D) Increased levels of acetylation of the N terminus of H4 in rev1 cells. (E) Acetylation of H3 at K9 and K14 in WT and rev1 cells. (F) Trimethylation of H3 at K4 in WT and rev1 cells. (G) Loss of DNA methylation at the r-globin promoter in rev1 cells. Loss of DNA methylation renders the r-globin promoter sensitive to restriction by HpaII. Amplication of the r-globin promoter by qPCR after HpaII restriction allowed the fraction of DNA remaining uncleaved, and therefore methylated, to be determined. Amplication was normalized to BamHI digested genomic DNA, then further normalized to set the WT level at 1. The amplied region does not contain any BamHI sites. Error bars represent the standard deviation. See also Figure S1.
5 ChIP primers
10
15 kb
20
25
30
35
.7 19 6 .3 7 21 22.37 .1 9
8. 9 10 9 .3 5
.1
13
.8
15
B
H3K9me2 Fold enrichment relative to HS4
6 5 4 3 2 1 0 0 5 10 15 kb 20 25 30 35
17
WT pcnaK164R
C
6 5 4 3 2 1 0 0 5 10 15
rev1 Cambridge rev1 Munich
20 kb
25
30
32
.0
0
35 WT rev1
D
H4 N-terminal tail ac Fold enrichment relative to HS4
1.2 1.0 0.8 0.6 0.4 0.2 0 WT rev1
E
1.2
H3K9/K14ac Fold enrichment relative to HS4
1.0 0.8 0.6 0.4 0.2 0
Associated with loss of H3K9me2, there was enrichment in acetylation of the H4 N-terminal tail in rev1 compared to WT DT40 (Figure 1D). ChIP primer pair (kb) ChIP primer pair (kb) As the increase in H4 N-terminal acetylation F G 1.2 WT was not associated with enrichment of other 1.2 rev1 1.0 marks of canonical transcriptional activation, 1.0 0.8 H3K9/14ac and H3K4me3, previously observed 0.8 0.6 0.6 at this locus (Litt et al., 2001a; Litt et al., 2001b) 0.4 0.4 (Figures 1E and 1F), it is consistent with enrich0.2 0.2 ment of newly synthesized histones in rev1 cells 0 0 relative to the WT (Lande-Diner et al., 2009; WT rev1 Sobel et al., 1995). Interestingly, we also ChIP primer pair (kb) observed an increase in H3K56 acetylation (Figure S1 available online), which is a robust marker of newly synthesized H3 in yeast 2001b). Consistent with this, we found an enrichment of the (Li et al., 2008) but whose signicance in vertebrates remains repressive H3K9me2 modication in wild-type (WT) DT40 cells a subject of debate. Despite not observing histone marks assoacross the constitutively condensed chromatin domain and at ciated with transcriptional activation, we did observe signicant the promoter of the nearby r-globin gene (Figure 1B). Compa- loss of DNA methylation at the r-globin promoter in rev1 cells rable enrichment was evident in a cell line, pcnaK164R, in which (Figure 1G). PCNA cannot be ubiquitinated and which is defective in postreplicative gap lling (Arakawa et al., 2006; Edmunds et al., 2008) Derepression of the r-Globin Locus in REV1-Decient (Figure 1B). In contrast, we observed a greatly reduced enrich- Cells ment of H3K9me2 in two independently generated rev1 lines Enrichment of H3K9me2 across the promoter of the r-globin (Arakawa et al., 2006; Simpson and Sale, 2003) (Figure 1C). gene is associated with its silencing (Litt et al., 2001a). Therefore,
98 98 0 9 9 7 7 .0 .1 .1 .3 8. 8. .3 32 13
13
21
8. 98
.1
.3 7
704 Molecular Cell 40, 703713, December 10, 2010 2010 Elsevier Inc.
13
21
32
.0 0
Relative HpaII protection
21
32
.0
Molecular Cell
600 500 400 300 200 100 0
Fold increase in -globin mRNA relative to WT
200
100
ii
iii
iv
Figure 2. Derepression of r-Globin Expression in rev1 Cells

(A) Derepression of r-globin expression in rev1 cells. Comparison of r-globin expression in different DT40 mutants. Expression, monitored by qRT-PCR with primers RhoExpF and R (Table S2), is given as the fold increase over the WT level, which is set at 1. rev1 #162 and #217 are two independent rev1 clones derived in our lab (rev1 Cambridge). Error bars show the range. (B) Effect of complementation with human REV1 on r-globin derepression. Increase in expression of r-globin in rev1 cells, and rev1 cells complemented with hREV1, relative to WT: i, rev1 cells cultured for >3 months; ii, rev1 complemented with hREV1 at 4 weeks, which is as soon as practically possible, and then cultured for >3 months; iii, rev1 cells from i, with established derepression of r-globin, complemented with hREV1 and cultured for 3 weeks; iv, as iii, but cultured for 3.5 weeks; and v, as iii, but cultured for 5 weeks. Five weeks in culture corresponds conservatively with 70 cell divisions.
we predicted that loss of this modication in rev1 cells would lead to increased expression of the gene. Quantitative PCR revealed an approximately 100-fold increase in the expression of the r-globin gene in rev1 cells relative to WT cells (Figure 2A). Consistent with the enrichment of H3K9me2, no increase in r-globin expression was seen in the pcnaK164R line. Moreover, a mutant lacking XRCC3, defective in homologous recombination, also showed no increase in r-globin expression. Interestingly, mutants defective in the translesion polymerases Polh and Polz (REV3) showed only small increases in r-globin expression, which correlates with their mild phenotypes when assessing the progression of replication forks on damaged DNA templates (Edmunds et al., 2008; Jansen et al., 2009). Reintroduction of a human REV1 complementary DNA (cDNA), which we have previously shown to complement all phenotypes of the chicken rev1 line (Edmunds et al., 2008; Ross et al., 2005), was unable to reverse the r-globin derepression once established over the course of 5 weeks in culture (conservatively 70 cell divisions), suggesting that the kinetics of restoration of repressive marks is, at best, slow (Figure 2B). However, a rev1 line in which hREV1 had been reintroduced at an early stage but then cultured for several months did not exhibit r-globin derepression.
Spontaneous DNA Damage Is Unlikely to Explain the Loss of Repressive Histone Marks in REV1-Decient Cells Taken together, these data are consistent with the hypothesis that gap-lling modes of DNA replication predominate in rev1 DT40 cells, leading to the replacement of parental modied histones with newly synthesized histones devoid of repressive marks. However, the effect appears without the introduction of exogenous DNA damaging agents. We therefore asked whether spontaneous DNA damage could be sufcient to lead to this phenotype. To address this question theoretically, we developed a computer model simulating inheritance of histone modications across cell division through copying of modications from parental histones to newly synthesized histones (see the Experimental Procedures). We also simulated replication fork stalling, occurring with a dened probability per cell division, that leads to gap-lling DNA synthesis accompanied by a tract of newly synthesized histones with length equal to the length of the gap. We could vary the probability of replication fork stalling and the length of the gap to examine the effect these parameters might have on epigenetic stability. Our model showed that it was possible to obtain loss of histone modications with random replication fork stalling using a gap length consistent with that observed in vivo, but only at a very high frequency of stalling (one stall every 10 kb) (Figure 3A). Even assuming stalling at the maximum level possible from estimates of the frequency of spontaneous damage, approximately one lesion per 60 kb, (Lindahl, 1996), the mean length of gap would have to be greater than 8 kb (approximately 40 nucleosomes) in length to achieve a 40% loss of histone modications in our model (Figure 3B). This is considerably greater than the largest current estimate for the length of postreplicative gaps (Lopes et al., 2006). We therefore considered the possibility that replication forks may stall more frequently at specic sites within the b-globin locus. Such sites exist widely in all genomes and are frequently found where the DNA sequence can form secondary structures (reviewed in Mirkin and Mirkin, 2007). Indeed, the chicken r-globin gene has been previously shown to contain a region in the second intron in which replication forks are slowed or blocked (Prioleau et al., 2003). We therefore simulated the effect of a xed stall and found that, with a probability of stalling of 0.4 or above per cell division, a stable tract of lost histone modications approximately equal to the length of the gap had developed after 30 cell divisions (Figure 3C). G Quadruplex Formation by the r-Globin Second Intron Sequence We therefore examined the sequence in the second intron of r-globin and noted, at the site identied by Prioleau et al. (2003), a sequence that corresponds to a consensus for a G quadruplex (G4) DNA (Figure 4A). Indeed, the region of the chicken b-globin locus we studied in this work contains three G4 sequences, the other two residing in the constitutively condensed region (Figure 4A). G4 DNA, of the general sequence G3-5-L1-7-G3-5-L1-7-G3-5-L1-7-G3-5 (where L can be any base), can form a variety of secondary structures at physiological salt concentrations, whose stability exceeds that of duplex DNA, both in vitro and in vivo (Lipps and Rhodes, 2009; Maizels,
Fold increase in -globin mRNA relative to WT
re WT v1 #1 6 re v1 2 re #2 v1 1 M 7 pc unic na h K1 pc 64 na R K1 rev 64 1/ R po l re v3 xr cc 3
Molecular Cell
A
Marks lost (%)
40
30
20
et al., 1989). In vitro, the 29bp G4 sequence from the r-globin intron forms a K+-dependent quadruplex structure as shown by circular dichroism spectroscopy, with diagnostic positive peaks at 210 nm and 265 nm (Kypr et al., 2009) (Figure 4B). This structure is dependent on four G-rich blocks in the oligonucleotide (Figure 4B). REV1 Is Required for Efcient Replication of G Quadruplex-Forming DNA on the Leading-Strand Template Translesion synthesis has been previously implicated in the repli cation of G4 DNA (Betous et al., 2009; Youds et al., 2006). We therefore asked whether REV1 assists in replication of this specic G4 DNA sequence. To do this, we took advantage of a replicating plasmid assay (Szuts et al., 2008), which can measure the efciency of replication as a change in the number of ampicillin resistant colonies recovered normalized to a control kanamycin-resistant plasmid not containing the G4 sequence. We found a striking reduction in the efciency of replication of the plasmid when the G4 sequence was placed on the leadingstrand template, but not on the lagging-strand template (Figures 4C and 4D), implying a role for REV1 in the replication of this sequence. This observation is consistent with the presence of two strong origins mapped 30 of the r-globin G4, which would place the G4 on the leading-strand template (Prioleau et al., 2003). The colonies that we recovered from transfection of rev1 cells did not show loss of the G4 sequence. This suggests that other mechanisms are able to compensate for the loss of REV1 but that in rev1 cells these are not able to efciently counteract loss of the plasmid. Using this assay, we were able to dissect the contribution of the different domains of REV1 to the replication of the G4 containing plasmid. REV1 has two principal activities. It is a deoxycytidyl transferase (Nelson et al., 1996) and also has a noncatalytic function in which its C terminus plays a crucial role in the coordination of TLS by other polymerases (Guo et al., 2003; Ross et al., 2005). Full-length human REV1 complemented the G4 replication defect completely, as it does other rev1 phenotypes (Edmunds et al., 2008; Ross et al., 2005) (Figure 4E). A DT40 line harboring a REV1 BRCT domain deletion (Figure S2) also exhibited no defect. However, consistent with our observations on replication fork progression (Edmunds et al., 2008), REV1 lacking its C-terminal 100 amino acids did not complement, implicating the polymerase-interacting region in the replication of G4 DNA. Interestingly, a catalytically dead mutant complemented only to about 50% of full-length REV1, suggesting that the catalytic activity also plays a role in G4 DNA replication, perhaps by the incorporation of a nontemplated C opposite G, leading to disruption of the G4 structure (Figure S3). Loss of Transcriptional Repression in rev1 Cells Is Associated with G4-Forming DNA In the light of the requirement for REV1 in replication of this specic G4 sequence, we speculated that the phenotype observed at the r-globin gene could represent a more general loss of repression in the vicinity of sequences with the potential to form G4 structures. In order to test this, we used a microarray to identify targets signicantly upregulated in rev1 cells relative
10
0 0 20 40 60
Cell divisions
B
Kb / stall for 40% modification loss
60 50 40 30 20 10 0 0 1000 2000 3000 4000 5000
Gap Length (bp)
Length of nucleosome tract in which marks are lost
Gap length (nucleosomes)

4 5 6 7
0.0
0.2 0.4 0.6 0.8 probability of stall at fixed site
1.0
Figure 3. Computational Simulation of Loss of Histone Modications in Response to Formation of Postreplicative Gaps
(A) Percentage of marks lost as a function of time with a xed stall probability. The graph shows a representative time course for histone modication loss. For this simulation, the postreplicative gap length was set at 1 kb, the probability of two place copy at 0.25 and the probability of fork stalling at 0.025 per nucleosome. This corresponds to one stall every 8 kb. Each data point represents an average of 30 simulations. (B) Postreplicative gap length necessary to produce 40% loss of histone marks in 30 generations. A lower estimate for spontaneous stalling intervals of 60100kb would clearly place the necessary postreplicative gap length much higher than any current in vivo estimate (see the main text). The x axis scale assumes an internucleosome distance of 200 bp. (C) Length of nucleosome tract in which marks are lost as a function of the probability of stalling per replication cycle at a xed point. Data is shown for a postreplicative gap length of four to seven nucleosomes, with the variance in length set to 0.
2006). These structures are characterized by stacks of planar arrays of four Hoogsteen bonded dG bases coordinated by a monovalent metal ion (Sundquist and Klug, 1989; Williamson
Molecular Cell
A
FR gene HSA constitutively condensed chromatin HS4 -globin locus
C
25 20 15 10 5 0
p<0.005 n=4
Ampr/Kanr
n=6
G4 on leading strand template ggggagtaaaagggagcggggtgctgggg
WT
rev1
B
7 6 5 4 3 2 1 0 -1 -2 -3
with K+ without K+ G3
D
12
CD (mdeg)
Ampr/Kanr
10 8 6 4 2 0
n=3
n=3
G4 on lagging strand template

220 240 260 280 300 320
(nm)
WT
rev1
p<0.002
30 25
n=4 n=4
p<0.007
n=4 n=3
Ampr/Kanr
20 15 10 5 0
n=6 n=4
n=4
r hRev1 EV : 1
71 r A] hRev1 EV : 1 [1 -
v1
37
v1 B
70
re
Figure 4. REV1 Is Required for Replication of G Quadruplex-Forming DNA on the Leading-Strand Template
(A) Position of G4 DNAs (ovals) in the region of the b-globin locus studied in this work. The sequence of the r-globin G4 DNA is shown. (B) Circular dichroism spectroscopy of the r-globin G quadruplex forming sequence. Renaturation of the minimal 29 bp G4 oligonucleotide (GQCDG4) in the presence (black line) or absence (light gray line) of K+ ions. Renaturation of a truncated r-globin G4 sequence (GQCDG3) in the presence of K+ ions (mid gray line). (C and D)Replication efciency, shown as the ratio of Ampr to Kanr E. coli colonies, for the r-globin G4 DNA on the leading- (C) and lagging- (D) strand template of pQ (Szuts et al., 2008). Error bars represent standard error of the mean. p values were calculated using the unpaired t test (two-tailed). (E) Replication efciency of the leading-strand template G4 in rev1 mutants. Complementation is with full-length, catalytically inactive (D570AE571A) and C-terminally truncated (1-1137) human REV1. The BRCT mutant is an endogenous deletion of amino acids 69116 of REV1 (Figure S2). The WT and rev1 data from Figure 4C are shown again for comparison. Error bars represent standard error of the mean. See also Figures S2 and S3.
to the WT. Having validated the upregulation of four transcripts in two further independent rev1 lines by qPCR (Figure S4), we selected genes from the array that were expressed at a low level in WT cells, but that were increased more than 1.4-fold in rev1 with a t test p value of <0.075, and examined the sequence 1500 bp either side of the annotated transcriptional start site for the presence of G4-forming sequences, using the Quadnder server (Sca-
ria et al., 2006). As a control, we analyzed a set of genes with similarly low levels of expression in WT but that were not upregulated in rev1 cells. At least one G4 sequence within the 3 kb window was found in 38% of the control set (Table S1), a similar gure to that obtained in previous analyses of G4 sequences near promoters in the chicken (Du et al., 2007) and human genomes (Eddy and Maizels, 2008; Huppert and Balasubramanian, 2007).
r hRev1 EV : 1 [D 5
pc
na K 16 4
re
AE
11
Molecular Cell
LYSC
Sa
Ex1
Sp
Ex2
Ex3 Ex4
Figure 5. Derepression of Silenced Loci in rev1 Cells Is Associated with G4 DNA

(A) Introduction of the r-globin G4 DNA into the LYSC locus. The genomic locus was amplied as a SalI (Sa)ClaI (C) fragment with primers LYSCSalF and LYSCClaR. A linker DNA (G4TplantSph1) containing a BamHI (B) site and the G4 sequence was introduced into the SphI (Sp) site in the rst intron of the LYSC gene, so as to be on the feature strand. Correct insertion of the G4 DNA was conrmed by sequencing with primer LYSCG4seq. A bidirectional origin has been demonstrated in the CpG island at the 30 end of the gene (Phi-van and Stratling, 1999) meaning that the introduced G4 structure will form on the leading-strand template. A puromycin-resistance selection cassette was inserted into the BamHI site. This was then removed by transient expression of Cre recombinase. (B) Southern blot of BamHI-digested DNA showing targeting of one allele of LYSC producing at 5 kb band. (C) Conrmation of the presence of the G4 sequence by PCR using primers (LYSCG4F and R) annealing either side of the expected insertion of the G4 DNA (plus the remnants of the loxP recombination sites). (D) qRT-PCR for LYSC expression in clones of WT and rev1 harboring the r-globin G4 DNA in the LYSC locus expressed as the fold change relative to unmanipulated WT DT40. (E) Map of the chicken LYSC and GAS41 loci. E, transcription enhancer element; S, transcription suppressor element; MAR, matrix attachment region (adapted from Myers et al., 2003). The positions of the three pairs of ChIP primers are indicated. (F) H3K9 dimethylation (H3K9me2) at the LYSC locus normalized to that at the constitutively active GAS41 promoter in WT and rev1 cells harboring the r-globin G4 DNA in one allele of the LYSC locus (lysc+/G4). Error bars represent standard error of the mean. (G) H4 N terminal acetylation at the LYSC locus in WT and rev1 lysc+/G4 cells, normalized to the GAS41 promoter.
Sthn probe
loxP
Ex1
1 kb
B
WT lysc:G4
C
lysc:G4 WT
LYSC expression (fold change relative to wild type)
kbp >20 5.3
bp 1000 500
lys W c +/G T
LYSC
GAS41
5 MAR CpG island
3 MAR
0 ChIP primers
10 kb
1 SC U LY
15
pr
F
H3K9Me2 Fold enrichment relative to prGAS41
4 3 2
WT lysc+/G4 rev1lysc+/G4
G
H4 N-terminal tail Ac Fold enrichment relative to prGAS41
1.6 1.2 0.8
pr
locus into the developmentally regulated lysozyme C gene. LYSC, another well-studied locus 0.4 (Myers et al., 2006) that is silent in chicken 1 lymphocytes, lacks any endogenous G4 0 0 sequence and is unaffected by loss of REV1 in DT40. Like the r-globin gene, it has a strong origin of replication at its 30 end (Phi-van and ChIP primer pair ChIP primer pair Stratling, 1999). Using homologous recombination, we inserted the G4-forming sequence and a puromycin resistance selection cassette into Contrastingly, 71% of the targets upregulated in rev1 cells con- the rst intron of LYSC approximately 500 bp from the transcriptained at least one predicted G4-forming sequence in the 3 kb tional start site, a position equivalent to that seen in the r-globin window. Therefore, there is a statistically signicant (p < 0.001) locus (Figures 5A5C). Having screened for successful inteassociation between G4 sequences in the vicinity of the promoter grants in both WT and rev1 backgrounds, we removed the selecand increased gene expression in rev1 cells. tion cassette by Cre-loxP recombination and cultured the clones for 4 weeks. We detected a marked increase in expression of lysozyme in 7 of 14 clones of rev1 cells carrying the same G4 Introduction of the 29 bp r-Globin G4 DNA into a Silent DNA integration (Figure 5D). Such an effect was not observed Locus Confers Susceptibility to Derepression in rev1 in any of 12 WT clones harboring the same integration of the rCells In order to test our hypothesis more directly, we conducted an globin G4 DNA in the LYSC locus (Figure 5D). To examine experiment to transplant the G4 sequence from the r-globin whether this increase in expression was correlated with the
SC SC 41 U U AS LY LY G pr pr
AS
WT lysc+/G4 rev1lysc+/G4
41
lys rev c +/G 1
20
pr
pr
AS
41
Seln. Cassette
loxP
Ex2
Ex3 Ex4
40 30 20 10 0
Molecular Cell
recycled histones
Figure 6. A Model for Loss of Repressive Histone Marks at Sites with G4-Forming Potential in rev1 Cells
Replication is depicted arresting at a G4 DNA on the leading-strand template. Parental histones are shown as light-gray circles, with repressive epigenetic marks represented as gray bars. New histones are shown in black. If REV1 is present, the fork can replicate through the G4 DNA, maintaining processive DNA synthesis and histone deposition. It is not clear whether the presence of REV1 prevents the formation of the structure or assists in its unwinding (see also Figure S2). In the absence of REV1 the fork remains arrested at the G4 DNA, resulting in a postreplicative gap. The DNA synthesis associated with the resolution of this gap and of the G4 DNA is accompanied by new histone incorporation resulting in a tract of chromatin lacking the parental epigenetic marks.
new histones
G4 unwinding + REV1 at the fork
G4 unwinding + PCNA-ubiquitin or HR-dependent gap filling
copying of preexisting marks to adjacent nucleosomes
epigenetic changes predicted by our model, we examined the H3K9 dimethylation and H4 N-terminal tail acetylation at the LYSC promoter in a rev1 lysc+/G4 clone exhibiting a >30-fold increase in lysozyme messenger RNA (mRNA). Compared to a WT lysc+/G4, the level of H3K9me2 is decreased, while the level of H4 N-terminal acetylation is increased. Thus, the r-globin G4 sequence can trigger loss of repression when inserted into a silenced locus in rev1, but not WT, cells. DISCUSSION In this work, we demonstrate a link between two important facets of chromosomal replication, the replication of structured DNA and the faithful maintenance of a repressive chromatin environment. Our observations suggest a model (Figure 6) in which the absence of REV1 leads to uncoupling of histone recycling from DNA synthesis at sites capable of forming G4 structures. In turn, this results in the repeated loading of newly synthesized histones that ultimately leads to a permanent loss of repressive epigenetic marks. The role of REV1 in Replication of G4 DNA Sequences capable of forming G4 DNA are abundant throughout the vertebrate genome but are highly enriched at telomeres, the immunoglobulin gene switch regions, and the vicinity of the transcription start site of genes. There is also increasing evidence for the formation of such structures in vivo (reviewed in Lipps and Rhodes, 2009; Maizels, 2006). G4 DNA can block replicative DNA polymerases in vitro (Woodford et al., 1994), and there is now good evidence, particularly from the study of telomeres, that they form and can slow or block replication in vivo (Schaftzel et al., 2001; Sfeir et al., 2009). The exact correlation between the replication slow zone in the r-globin second intron identied by Prioleau et al. (2003) and a robust G4-forming sequence
provides further evidence that these sequences pose a challenge to the replicative machinery, even in normal cells. TLS has been previously implicated in the replication of G/C tracts in both C. elegans and human cell lines. Deletion of either Polk or Polh in a dog-1 worm (decient in the FANCJ helicase) resulted in an increased frequency of small deletions in G/C tracts (Youds et al., 2006). More recently, RNA interference (RNAi)-mediated knockdown of Pols h, i, and k has been shown to sensitize cells to the G4-stabilizing compound telomestatin and to result in elevated DNA damage associated with the human c-MYC promoter, which contains a G-rich sequence capable of forming G4 structures (Betous et al., 2009). However, the precise role of TLS in replication of structures that actually contain no damaged DNA remains unclear. In particular, it remains to be shown whether REV1 collaborates with the helicases that have been demonstrated to unwind G4 DNA, such as FANCJ, BLM, and WRN (Fry and Loeb, 1999; London et al., 2008; Sun et al., 1998; Wu et al., 2008), and whether similar epigenetic instability at G4 DNA is triggered by loss of these helicases. It is noteworthy that cells lacking either BLM or WRN exhibit altered expression of genes harboring sequences with G4-forming potential (Johnson et al., 2010), although this effect was mechanistically ascribed to regulation of transcription. A potential clue to the role of REV1 may come from our observation that not only the C-terminal polymerase-binding region, but also the catalytic activity of the enzyme is required for fully effective replication of the r-globin G4 DNA. REV1 is a deoxycytidyl transferase (Nelson et al., 1996) but also a template G-dependent DNA polymerase (Haracska et al., 2002). This suggests a possible model (Figure S3) in which the ability of REV1 to produce a tract of dC bases with minimal reference to the template may destabilize the G4 structure through base pairing between newly synthesized dC and template dG. The C terminus of the protein could then coordinate the handoff to
Molecular Cell
other TLS polymerases, allowing extension of this dC-rich primer and replication of the G4 sequence. Replication Impediments and Epigenetic Stability Recycling of parental histones is likely to play a key role in the propagation of epigenetic memory but requires tight coupling between histone displacement and redeposition in order that the register between histone marks and underlying DNA sequence is not lost. Such coupling is likely to be mediated by histone chaperones, notably Asf1 (reviewed in Annunziato, 2005; Groth et al., 2007; Margueron and Reinberg, 2010). Indeed, very recent evidence suggests that Asf1 can buffer histones displaced by hydroxyurea-induced replication arrest, leading to the suggestion that replication stress may jeopardize proper chromatin restoration and thereby trigger epigenetic changes in daughter cells (Jasencakova et al., 2010). Here, we provide evidence that replication impediments can indeed lead to epigenetic change, although in our model it is failure to use recycled histones during gap lling rather than unscheduled deposition that underlies a loss of epigenetic information. Our computer model predicts that histone mark propagation is likely to be sufciently robust to deal with such gaps occurring sporadically, as would be caused by DNA damage. At levels of damage compatible with cell survival, the model suggests that spreading of histone methylation back into the demethylated gap would result in healing of the repressive chromatin environment. However, for G4 DNA, it seems likely that the repeated deposition of newly synthesized histones swamps this ability to restore the pre-existing chromatin environment and ultimately leads to derepression. It is also conceivable that spreading of histone demethylation can occur if the tract of demethylated histones is sufciently long. It seems likely that this spreading would be limited by chromatin domain insulators, such as those marked by HSA and HS4 (Figure 1A), resulting in switching of the histone methylation state of the whole domain. Such behavior has been proposed on theoretical grounds (Dodd et al., 2007) and may explain how only two identiable G4 DNA sequences in the condensed chromatin region can nonetheless lead to loss of H3K9 dimethylation across the whole 15 kb domain. It is noteworthy that derepression in the absence of REV1 is not seen in all silenced loci. Notably, the expression of HOX genes, known to be under the control of the polycomb repressive complexes, does not appear to be affected by loss of REV1 (data not shown). This may reect different mechanistic approaches to the generation and maintenance of particular forms of silencing, which may include specic DNA signals, the use of RNA interference, and histone recycling. Conversely, the heterochromatin region in the b-globin locus does appear to be affected, even though it has recently been proposed that RNAi plays a role in repression of this sequence (Giles et al., 2010). It will therefore be interesting to explore the relationship between histone recycling and mechanisms such as RNAi in the initiation and maintenance of gene repression, possibly using REV1 deciency as a tool. Finally, dysregulation of gene expression is common in many cancers. Our observations suggest one possible mechanism by which a single mutation in a pathway promoting genetic stability could lead to more widespread epigenetic instability.
EXPERIMENTAL PROCEDURES DT40 Strains, Culture, and Transfection DT40 cells were propagated and transfected as previously described (Simpson and Sale, 2003). DT40 mutants used in this work have also been described previously (Arakawa et al., 2006; Edmunds et al., 2008; Kawamoto et al., 2005; Ross et al., 2005; Simpson and Sale, 2003; Takata et al., 2001). Chromatin Immunoprecipitation Chromatin immunoprecipitation (ChIP) was performed as described (Aparicio et al., 2005) using formaldehyde (FA) crosslinking to trap protein-DNA complexes, with minor modications detailed in the Supplemental Experimental Procedures. PCR primers for ChIP qPCR are listed in Table S2. Antibodies for ChIP The following antibodies were used: anti-H3K9me2, Millipore ChIPAb+ catalog number 17-648; anti-H3K9/K14ac, Millipore ChIPAb+ catalog number 17-615; anti-H3K4me3, Cell Signaling Technology catalog number 9727; anti-H3, Cell Signaling Technology catalog number 2650; and anti-acetylH4, Millipore ChIPAb+ catalog number 17-630. This polyclonal antibody recognizes acetylation of H4K5, 8, 12, and 16 and has been previously used (as Upstate catalog number 06-598) to monitor acetylation of H4 during histone deposition (LandeDiner et al., 2009). A negative control for ChIP was provided by normal rabbit IgG (Millipore). ChIP qPCR and Data Analysis Quantitative PCR was performed in real time with SYBR green. ChIP DNA (2.5 ml) was used in each reaction, with 400 nM primer mix and 12.5 ml 2xSYBR-green qPCR ready-mix (Invitrogen). The reaction was carried out on a ABI Prism real-time cycler with the following program: 50 C for 2 min, 90 C for 10 min, and 40 cycles of 90 C for 15 s (denaturation), 60 C for 1 min (annealing and extension). Each reaction was performed in duplicate. ChIP results were normalized to the positive control anti-H3 antibody with the formula 2(Ct(Ab)-Ct(H3)). In control experiments to initially validate the protocol, immunoprecipitation with the normal rabbit IgG antibody recovered extremely low amounts of material (less than 0.1% of the H3 signal and less than 0.05% of input). Data from the b-globin locus was further normalized to the hypersensitive site (HS4) of the b-globin locus, which has been previously shown to contain high levels of H4 and H3 N-terminal acetylation (Litt et al., 2001b). This was found to allow reproducible comparison between different extracts. For the LYSC locus, normalization of the specic ChIP signal was to total H3, then to the promoter of the adjacent, constitutively active GAS41 locus. Absolute enrichment relative to total H3 of H3K9me2 and H4 N-terminal acetylation at the GAS41 promoter was found to be similar in WT and rev1 cells. qRT-PCR RNA was extracted with Trizol (Invitrogen) according to the manufacturers instructions. cDNA was prepared with 5 mg mRNA with Super RT (HT Biotechnology, Cambridge, UK) and oligodT primer in a nal volume of 40 ml. qPCR reactions were performed as described above, with 2.5 ml cDNA per reaction. Quantitation was relative to b-actin (cDNA diluted 1/100) with the exception of LYSC, which was relative to the adjacent GAS41 gene (see Table S2 for primer sequences). The efciency of amplication was veried to be close to 1 (i.e., a Ct change of 1 reported a 2-fold change in concentration of template) for the control primers with a standard curve of cDNA dilutions. DNA Methylation Analysis Five micrograms of genomic DNA, quantied by nanodrop spectrophotometer, was cut with 5 units HpaII, BamHI, or MspI for 6 hr at 37 C. After phenol/chloroform extraction, DNA was precipitated overnight at 20 C with sodium acetate and ethanol before analysis with qPCR. HpaII and MspI are isoschizomers, but HpaII is blocked by CpG methylation whereas MspI is not; therefore, we could verify the assay by amplifying from the hypersensitive site showing no HpaII enrichment relative to MspI. Data from the r-globin promoter was normalized to BamHI digested DNA to control for differences in genomic DNA preparations (there are no BamHI sites within the expected amplicon).
Molecular Cell
Circular Dichroism Spectroscopy Oligonucleotides corresponding to the full-length r-globin G4 and the shorter sequence lacking the rst run of Gs (Figure 4A) were synthesized and puried by desaltion (Sigma), resuspended in TE buffer, and diluted to a nal concentration of 10 mM before heating to 95 C for 5 min to denature secondary structure. At this point, either 2M KCl was added to a nal concentration of 100 mM or an equivalent volume of nuclease-free H2O was added. The oligos were then left to cool overnight at room temperature. Spectroscopy was performed on a Jasco J-810 spectrometer at room temperature with a bandwidth of 2 nm, a response of 1 s, a data pitch of 0.2 nm and a scanning speed of 50 nm/min. Scans were performed over the range 200 to 320 nm. Curves for each oligo were processed by subtracting the trace produced by TE buffer (with or without KCl) and smoothing with the software provided by the manufacturer. Replicating Plasmid Assay For creation of the G4-containing replicating plasmid, oligonucleotides coding the r-globin G4 sequence (RGG4LeadF and RGG4LeadR) were ligated into pQ1 (Szuts et al., 2008) as an EcoRI fragment. Sequencing was used to select plasmids where the G4 DNA was on the leading-strand template relative to the Gal origin (Figure 5). A second plasmid placing the G4 on the lagging-strand template was made with oligonucleotides with EcoRI and PstI cohesive ends (RGG4LagF and RGG4LagR). Oligonucleotide sequences are given in Table S2. The plasmid replication assay was performed as described (Szuts et al., 2008) with minor modications: 1 mg control G4-free pQ2 plasmid (conferring kanamycin resistance) was used per transfection along with 1 mg G4-containing plasmid, (conferring ampicillin resistance); DpnI-digested plasmid was used to transform Invitrogen E-shot electrocompetent cells (catalog number 18290-015); and an equal amount of cells was plated out onto kanamycin and ampicillin plates. Computer Modeling The computer simulation Zippee is a Java applet that will run on most web browsers with the latest implementations of Java. (We have tested it on Internet Explorer 7, Firefox 3, and Safari 4). It can be found at http://www.cl. cam.ac.uk/calr3/zippee/. The Java code is available on request. A detailed description of the algorithm can be found in Supplemental Information. Microarray Microarray analysis was performed on three independent rev1 and WT lines simultaneously. RNA was extracted with Trizol (Invitrogen). The cDNA labeling and microarray hybridization was carried out by the Cambridge University School of Clinical Medicine Department of Metabolic Science with the Affymetrix Chicken Genome array. For analysis of the data, targets showing increased mean probe intensity in rev1 relative to the WT were sorted according to t value. Targets were selected whose initial mean probe intensity value was below 7 and whose upregulation was at least 40% and signicant at p < 0.075. We sorted the list by decreasing change in expression relative to WT and downloaded 1500 kb either side of the transcriptional start site as annotated in Ensembl (as at 31.3.2010), discarding targets that either had signicant unsequenced regions or could not be found in the Ensembl database. For our control set, we took the targets with the highest t value (i.e., no change between WT and rev1) and selected for those that were expressed at a level below a mean WT probe intensity level of 7. From this list, targets were selected at random and analyzed in the same way as the upregulated set. The sequence data from both sets was then analyzed with the Quadnder server (http://miracle.igib.res.in/quadnder/) (Scaria et al., 2006) to search for potential G4 forming motifs on either strand according to the consensus G(3-5)L(1-7)G(3-5)L(1-7)G(3-5)L(1-7)G(3-5). We used Fishers Exact Test to determine whether there was a signicant difference between the numbers of G4 motifs found in the upregulated and the control sets. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, four gures, and two tables and can be found with this article online at doi:10.1016/j.molcel.2010.11.009.
ACKNOWLEDGMENTS We would like to thank Ian McFarlane and his team at the University of Cambridge School of Clinical Medicine Microarray Facility for carrying out the microarray hybridization and guidance on analysis of the results, Guilherme Santos for help with the CD spectroscopy, and Shunichi Takeda and Jean-Marie Buerstedde for sharing DT40 lines. We also thank Daniela Rhodes, Cristina Rada, K.J. Patel, and members of the Sale lab for helpful discussions and critical comments. Work in the laboratory is supported by the Medical Research Council and Association for International Cancer Research. Received: May 14, 2010 Revised: July 30, 2010 Accepted: September 10, 2010 Published: December 9, 2010 REFERENCES Annunziato, A.T. (2005). Split decision: what happens to nucleosomes during DNA replication? J. Biol. Chem. 280, 1206512068. Aparicio, O., Geisberg, J.V., Sekinger, E., Yang, A., Moqtaderi, Z., and Struhl, K. (2005). Chromatin immunoprecipitation for determining the association of proteins with specic genomic sequences in vivo. Curr. Protoc. Mol. Biol. Chapter 21, Unit 21 23. Arakawa, H., Moldovan, G.L., Saribasak, H., Saribasak, N.N., Jentsch, S., and Buerstedde, J.M. (2006). A role for PCNA ubiquitination in immunoglobulin hypermutation. PLoS Biol. 4, e366. Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thomas, J.O., Allshire, R.C., and Kouzarides, T. (2001). Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain. Nature 410, 120124. Betous, R., Rey, L., Wang, G., Pillaire, M.J., Puget, N., Selves, J., Biard, D.S., Shin-ya, K., Vasquez, K.M., Cazaux, C., and Hoffmann, J.S. (2009). Role of TLS DNA polymerases h and k in processing naturally occurring structured DNA in human cells. Mol. Carcinog. 48, 369378. Buerstedde, J.M., and Takeda, S. (1991). Increased ratio of targeted to random integration after transfection of chicken B cell lines. Cell 67, 179188. Corpet, A., and Almouzni, G. (2009). Making copies of chromatin: the challenge of nucleosomal organization and epigenetic information. Trends Cell Biol. 19, 2941. Daigaku, Y., Davies, A.A., and Ulrich, H.D. (2010). Ubiquitin-dependent DNA damage bypass is separable from genome replication. Nature 465, 951955. De Koning, L., Corpet, A., Haber, J.E., and Almouzni, G. (2007). Histone chaperones: an escort network regulating histone trafc. Nat. Struct. Mol. Biol. 14, 9971007. Dodd, I.B., Micheelsen, M.A., Sneppen, K., and Thon, G. (2007). Theoretical analysis of epigenetic cell memory by nucleosome modication. Cell 129, 813822. Du, Z., Kong, P., Gao, Y., and Li, N. (2007). Enrichment of G4 DNA motif in transcriptional regulatory region of chicken genome. Biochem. Biophys. Res. Commun. 354, 10671070. Eddy, J., and Maizels, N. (2008). Conserved elements with potential to form polymorphic G-quadruplex structures in the rst intron of human genes. Nucleic Acids Res. 36, 13211333. Edmunds, C.E., Simpson, L.J., and Sale, J.E. (2008). PCNA ubiquitination and REV1 dene temporally distinct mechanisms for controlling translesion synthesis in the avian cell line DT40. Mol. Cell 30, 519529. Felsenfeld, G. (1993). Chromatin structure and the expression of globin-encoding genes. Gene. 135, 119124. Felsenfeld, G., Burgess-Beusse, B., Farrell, C., Gaszner, M., Ghirlando, R., Huang, S., Jin, C., Litt, M., Magdinier, F., Mutskov, V., et al. (2004). Chromatin boundaries and chromatin domains. Cold Spring Harb. Symp. Quant. Biol. 69, 245250.
Molecular Cell
Fry, M., and Loeb, L.A. (1999). Human werner syndrome DNA helicase unwinds tetrahelical structures of the fragile X syndrome repeat sequence d (CGG)n. J. Biol. Chem. 274, 1279712802. Giles, K.E., Ghirlando, R., and Felsenfeld, G. (2010). Maintenance of a constitutive heterochromatin domain in vertebrates by a Dicer-dependent mechanism. Nat. Cell. Biol. 12, 9499, sup pp 91-96. Groth, A., Rocha, W., Verreault, A., and Almouzni, G. (2007). Chromatin challenges during DNA replication and repair. Cell 128, 721733. Guo, C., Fischhaber, P.L., Luk-Paszyc, M.J., Masuda, Y., Zhou, J., Kamiya, K., Kisker, C., and Friedberg, E.C. (2003). Mouse Rev1 protein interacts with multiple DNA polymerases involved in translesion DNA synthesis. EMBO J. 22, 66216630. Hansen, K.H., Bracken, A.P., Pasini, D., Dietrich, N., Gehani, S.S., Monrad, A., Rappsilber, J., Lerdrup, M., and Helin, K. (2008). A model for transmission of the H3K27me3 epigenetic mark. Nat. Cell Biol. 10, 12911300. Haracska, L., Prakash, S., and Prakash, L. (2002). Yeast Rev1 protein is a G template-specic DNA polymerase. J. Biol. Chem. 277, 1554615551. Huppert, J.L., and Balasubramanian, S. (2007). G-quadruplexes in promoters throughout the human genome. Nucleic Acids Res. 35, 406413. Jansen, J.G., Tsaalbi-Shtylik, A., Hendriks, G., Gali, H., Hendel, A., Johansson, F., Erixon, K., Livneh, Z., Mullenders, L.H., Haracska, L., and de Wind, N. (2009). Separate domains of Rev1 mediate two modes of DNA damage bypass in mammalian cells. Mol. Cell. Biol. 29, 31133123. Jasencakova, Z., Scharf, A.N., Ask, K., Corpet, A., Imhof, A., Almouzni, G., and Groth, A. (2010). Replication stress interferes with histone recycling and predeposition marking of new histones. Mol. Cell 37, 736743. Johnson, J.E., Cao, K., Ryvkin, P., Wang, L.S., and Johnson, F.B. (2010). Altered gene expression in the Werner and Bloom syndromes is associated with sequences having G-quadruplex forming potential. Nucleic Acids Res. 38, 11141122. Karras, G.I., and Jentsch, S. (2010). The RAD6 DNA damage tolerance pathway operates uncoupled from the replication fork and is functional beyond S phase. Cell 141, 255267. Kawamoto, T., Araki, K., Sonoda, E., Yamashita, Y.M., Harada, K., Kikuchi, K., Masutani, C., Hanaoka, F., Nozaki, K., Hashimoto, N., and Takeda, S. (2005). Dual roles for DNA polymerase h in homologous DNA recombination and translesion DNA synthesis. Mol. Cell 20, 793799. Kypr, J., Kejnovska, I., Renciuk, D., and Vorlckova, M. (2009). Circular dichroism and conformational polymorphism of DNA. Nucleic Acids Res. 37, 17131725. Lachner, M., OCarroll, D., Rea, S., Mechtler, K., and Jenuwein, T. (2001). Methylation of histone H3 lysine 9 creates a binding site for HP1 proteins. Nature 410, 116120. Lande-Diner, L., Zhang, J., and Cedar, H. (2009). Shifts in replication timing actively affect histone acetylation during nucleosome reassembly. Mol. Cell 34, 767774. Lehmann, A.R. (1972). Postreplication repair of DNA in ultraviolet-irradiated mammalian cells. J. Mol. Biol. 66, 319337. Li, Q., Zhou, H., Wurtele, H., Davies, B., Horazdovsky, B., Verreault, A., and Zhang, Z. (2008). Acetylation of histone H3 lysine 56 regulates replicationcoupled nucleosome assembly. Cell 134, 244255. Lindahl, T. (1996). The Croonian Lecture, 1996: endogenous damage to DNA. Philos. Trans. R. Soc. Lond. B Biol. Sci. 351, 15291538. Lipps, H.J., and Rhodes, D. (2009). G-quadruplex structures: in vivo evidence and function. Trends Cell Biol. 19, 414422. Litt, M.D., Simpson, M., Gaszner, M., Allis, C.D., and Felsenfeld, G. (2001a). Correlation between histone lysine methylation and developmental changes at the chicken b-globin locus. Science 293, 24532455. Litt, M.D., Simpson, M., Recillas-Targa, F., Prioleau, M.N., and Felsenfeld, G. (2001b). Transitions in histone acetylation reveal boundaries of three separately regulated neighboring loci. EMBO J. 20, 22242235.
London, T.B., Barber, L.J., Mosedale, G., Kelly, G.P., Balasubramanian, S., Hickson, I.D., Boulton, S.J., and Hiom, K. (2008). FANCJ is a structure-specic DNA helicase associated with the maintenance of genomic G/C tracts. J. Biol. Chem. 283, 3613236139. Lopes, M., Foiani, M., and Sogo, J.M. (2006). Multiple mechanisms control chromosome integrity after replication fork uncoupling and restart at irreparable UV lesions. Mol. Cell 21, 1527. Maizels, N. (2006). Dynamic roles for G4 DNA in the biology of eukaryotic cells. Nat. Struct. Mol. Biol. 13, 10551059. Margueron, R., and Reinberg, D. (2010). Chromatin structure and the inheritance of epigenetic information. Nat. Rev. Genet. 11, 285296. Mirkin, E.V., and Mirkin, S.M. (2007). Replication fork stalling at natural impediments. Microbiol. Mol. Biol. Rev. 71, 1335. Myers, F.A., Chong, W., Evans, D.R., Thorne, A.W., and Crane-Robinson, C. (2003). Acetylation of histone H2B mirrors that of H4 and H3 at the chicken b-globin locus but not at housekeeping genes. J. Biol. Chem. 278, 36315 36322. Myers, F.A., Lefevre, P., Mantouvalou, E., Bruce, K., Lacroix, C., Bonifer, C., Thorne, A.W., and Crane-Robinson, C. (2006). Developmental activation of the lysozyme gene in chicken macrophage cells is linked to core histone acetylation at its enhancer elements. Nucleic Acids Res. 34, 40254035. Nelson, J.R., Lawrence, C.W., and Hinkle, D.C. (1996). Deoxycytidyl transferase activity of yeast REV1 protein. Nature 382, 729731. Phi-van, L., and Stratling, W.H. (1999). An origin of bidirectional DNA replication is located within a CpG island at the 300 end of the chicken lysozyme gene. Nucleic Acids Res. 27, 30093017. Prioleau, M.N., Gendron, M.C., and Hyrien, O. (2003). Replication of the chicken b-globin locus: early-ring origins at the 50 HS4 insulator and the rand betaA-globin genes show opposite epigenetic modications. Mol. Cell. Biol. 23, 35363549. Ross, A.L., Simpson, L.J., and Sale, J.E. (2005). Vertebrate DNA damage tolerance requires the C-terminus but not BRCT or transferase domains of REV1. Nucleic Acids Res. 33, 12801289. Scaria, V., Hariharan, M., Arora, A., and Maiti, S. (2006). Quadnder: server for identication and analysis of quadruplex-forming motifs in nucleotide sequences. Nucleic Acids Res. 34 (Web Server issue), W683W685. Schaftzel, C., Berger, I., Postberg, J., Hanes, J., Lipps, H.J., and Pluckthun, A. (2001). In vitro generated antibodies specic for telomeric guanine-quadruplex DNA react with Stylonychia lemnae macronuclei. Proc. Natl. Acad. Sci. USA 98, 85728577. Sfeir, A., Kosiyatrakul, S.T., Hockemeyer, D., MacRae, S.L., Karlseder, J., Schildkraut, C.L., and de Lange, T. (2009). Mammalian telomeres resemble fragile sites and require TRF1 for efcient replication. Cell 138, 90103. Simpson, L.J., and Sale, J.E. (2003). Rev1 is essential for DNA damage tolerance and non-templated immunoglobulin gene mutation in a vertebrate cell line. EMBO J. 22, 16541664. Sobel, R.E., Cook, R.G., Perry, C.A., Annunziato, A.T., and Allis, C.D. (1995). Conservation of deposition-related acetylation sites in newly synthesized histones H3 and H4. Proc. Natl. Acad. Sci. USA 92, 12371241. Sun, H., Karow, J.K., Hickson, I.D., and Maizels, N. (1998). The Blooms syndrome helicase unwinds G4 DNA. J. Biol. Chem. 273, 2758727592. Sundquist, W.I., and Klug, A. (1989). Telomeric DNA dimerizes by formation of guanine tetrads between hairpin loops. Nature 342, 825829. Szuts, D., Marcus, A.P., Himoto, M., Iwai, S., and Sale, J.E. (2008). REV1 restrains DNA polymerase z to ensure frame delity during translesion synthesis of UV photoproducts in vivo. Nucleic Acids Res. 36, 67676780. Takata, M., Sasaki, M.S., Tachiiri, S., Fukushima, T., Sonoda, E., Schild, D., Thompson, L.H., and Takeda, S. (2001). Chromosome instability and defective recombinational repair in knockout mutants of the ve Rad51 paralogs. Mol. Cell. Biol. 21, 28582866.
Molecular Cell
Williamson, J.R., Raghuraman, M.K., and Cech, T.R. (1989). Monovalent cation-induced structure of telomeric DNA: the G-quartet model. Cell 59, 871880. Woodford, K.J., Howell, R.M., and Usdin, K. (1994). A novel K(+)-dependent DNA synthesis arrest site in a commonly occurring sequence motif in eukaryotes. J. Biol. Chem. 269, 2702927035.
Wu, Y., Shin-ya, K., and Brosh, R.M., Jr. (2008). FANCJ helicase defective in Fanconia anemia and breast cancer unwinds G-quadruplex DNA to defend genomic stability. Mol. Cell. Biol. 28, 41164128. Youds, J.L., ONeil, N.J., and Rose, A.M. (2006). Homologous recombination is required for genome stability in the absence of DOG-1 in Caenorhabditis elegans. Genetics 173, 697708.
Article
Regulation and Rate Enhancement during Transcription-Coupled DNA Repair
Laura Manelyte,1 Young-In T. Kim,1 Abigail J. Smith,1 Rachel M. Smith,1 and Nigel J. Savery1,*
1DNA-Protein Interactions Unit, School of Biochemistry, University of Bristol, Bristol BS8 1TD, UK *Correspondence: n.j.savery@bristol.ac.uk DOI 10.1016/j.molcel.2010.11.012
Molecular Cell
SUMMARY
Transcription-coupled DNA repair (TCR) is a subpathway of nucleotide excision repair (NER) that is triggered when RNA polymerase is stalled by DNA damage. Lesions targeted by TCR are repaired more quickly than lesions repaired by the transcription-independent global NER pathway, but the mechanism underlying this rate enhancement is not understood. Damage recognition during bacterial NER depends upon UvrA, which binds to the damage and loads UvrB onto the DNA. Bacterial TCR additionally requires the Mfd protein, a DNA translocase that removes the stalled transcription complexes. We have determined the properties of Mfd, UvrA, and UvrB that are required for the elevated rate of repair observed during TCR. We show that TCR and global NER differ in their requirements for damage recognition by UvrA, indicating that Mfd acts at the very earliest stage of the repair process and extending the functional similarities between TCR in bacteria and eukaryotes.
INTRODUCTION Transcription-coupled DNA repair (TCR) is a subpathway of nucleotide excision repair (NER) that targets DNA lesions in the template strand of active genes and plays an important role in maintaining genome integrity (Hanawalt and Spivak, 2008). TCR is initiated when RNA polymerase (RNAP) stalls at a lesion and typically results in damage being repaired more quickly than it would be in nontranscribed regions that are repaired by the global NER pathway. All organisms in which TCR has been detected express a transcription-repair coupling factor (reviewed in Hanawalt and Spivak, 2008; Svejstrup, 2002). These proteins have been characterized in humans (CSB), yeast (Rad26), and bacteria (Mfd) and are all ATP-dependent enzymes that interact with the transcription apparatus and with components of the NER apparatus. NER has a broad substrate specicity, and the lesions that it targets are detected via multistep processes (Shuck et al., 2008; Truglio et al., 2006a). Damage recognition during global NER in bacteria is performed by a complex of UvrA and UvrB proteins. The stoichiometry of this complex is uncertain, with
UvrA2UvrB2 tetramers being detected in some studies (Kad et al., 2010; Malta et al., 2007; Verhoeven et al., 2002) and UvrA2UvrB trimers in others (Orren and Sancar, 1989; Pakotiprapha et al., 2009). Both UvrA and UvrB participate in damage detection. UvrA is responsible for initial recognition and loads UvrB onto DNA, possibly by partially unwinding the DNA close to the lesion. UvrB inserts a b-hairpin between the two strands, clamping one against an adjacent protein domain (Truglio et al., 2006b). Limited ATP-dependent DNA translocation by UvrB may be required for nal lesion verication. At this stage, UvrA dissociates from the complex and is replaced by UvrC, which cuts the damaged strand on either side of the lesion. The short oligonucleotide that results is displaced by UvrD helicase, and the gap in the damaged strand is lled by DNA polymerase I and DNA ligase. During TCR, the inability of RNAP to transcribe a noncoding lesion in the template strand causes it to stall, thus denying repair proteins access to the damage (Selby and Sancar, 1990). Mfd is recruited to the stalled transcription complex via a specic interaction with RNAP and removes RNAP from the DNA (thus overcoming the inhibitory effect that RNAP stalled at a lesion has on repair) (Park et al., 2002; Selby and Sancar, 1993; Smith and Savery, 2005). As TCR is faster than global NER, Mfd and/or RNAP must also affect the rate at which the Uvr proteins undertake repair. The mechanism by which this rate enhancement occurs is not known, but Mfd can interact with UvrA, and this interaction is presumed to be important for TCR (Selby and Sancar, 1993, 1995a). It has been suggested that this interaction recruits UvrA to damaged DNA or that it promotes dissociation of UvrA from a UvrA:UvrB:DNA preincision complex (Selby and Sancar, 1993, 1994). Models for strand-specic repair mechanisms that are independent of Mfd-UvrA interactions have also been proposed (Kunala and Brash, 1995; Patel et al., 2004). Mfd is an eight-domain monomeric protein (Deaconescu et al., 2006) (Figure 1A). Domain 4 is the RNAP-binding domain, and domains 5 and 6 comprise a helicase superfamily 2 DNA translocation module. Domains 1a, 2, and 1b are structurally homologous to 3 of the 5 domains of UvrB and are termed the UvrB homology module (BHM). In UvrB, these domains are involved in interaction with UvrA and UvrD, in damage recognition, and in ATP hydrolysis (Truglio et al., 2006a). The motifs responsible for damage recognition and ATP hydrolysis are absent from the BHM of Mfd, and the sequence similarity between the two proteins is greatest in domain 2 (D2) (Deaconescu et al., 2006; Selby and Sancar, 1993). Although little is known about the Mfd:UvrA interaction, the UvrB:UvrA interaction has been characterized by mutagenic studies (Truglio et al., 2004) and by
Molecular Cell
Mechanism of Transcription-Coupled DNA Repair
Figure 1. Mfd and UvrA Proteins

(A) Structure of E. coli Mfd (PDB ID 2EYQ) (Deaconescu et al., 2006). Residues mutated in this study are shown in space-lling representation, and the truncated proteins Mfd DBHM and Mfd DD7 are shown as black bars. (B) Structure of B. stearothermophilus UvrA (PDB ID 2R6F) (Pakotiprapha et al., 2008). A single monomer is shown, and the dimerization interface is indicated by a dotted line. SigI and SigII domains contain ABC ATPase signature motifs. Residues equivalent to those mutated in this study are shown in space-lling representation (labels show E. coli residues and numbering). The truncated proteins UvrA D131-248 and UvrA D290400 are shown as black bars, with dotted lines indicating regions missing from the constructs.
cocrystallization of UvrB D2 in complex with its partner domain from UvrA (Pakotiprapha et al., 2009). By analogy to UvrB, the surface of Mfd D2 that is likely to interact with UvrA can be deduced (Deaconescu et al., 2006; Pakotiprapha et al., 2009). In the crystal structure of apo-Mfd, this surface is largely buried in an intramolecular contact with domain 7 (D7), suggesting that interdomain rearrangements would be necessary to allow the crystallized form of the protein to interact with UvrA (Deaconescu et al., 2006). Interdomain rearrangements involving D7 are also thought to be necessary for activating the DNA translocation activity of Mfd when it binds to RNAP, as deletion of either D7 or domains 13 relieves the autoinhibition of motor activity that is observed in the isolated full-length protein (Murphy et al., 2009; Smith et al., 2007). The mechanism of RNAP displacement by Mfd is becoming clearer. The ATP-dependent DNA translocase activity of Mfd is activated when the protein binds to a stalled transcription complex (Smith et al., 2007). This pushes RNAP forward and destabilizes the interactions that hold the transcription complex in place (Park et al., 2002). In contrast, the mechanism by which the rate of DNA repair is enhanced during TCR remains poorly understood. In this work, we have investigated the determinants of TCR in Mfd, UvrA, and UvrB using mutants in which individual functions of each protein were specically impaired. We show that the BHM of Mfd is essential for the enhancement of repair rate but that the regulation of UvrA binding and DNA translocation by D7 is not. We also show that the damage specicity of UvrA is at least partially redundant during TCR but that TCR retains the need for damage recognition by UvrB. RESULTS The UvrA Interaction Surface Is Involved in the Autoregulation of Mfd The interaction between UvrA and the BHM of Mfd is presumed to be essential for the elevated rate of repair that is observed during TCR, but this hypothesis has not been tested experimen-
tally. To address this issue, we constructed Mfd derivatives in which the putative UvrA interaction surface was either disrupted by substitutions or removed entirely by deletion of the BHM. Mfd R165A R181A F185A (Mfd D2AAA) carries alanine substitutions of three residues on the surface of Mfd D2 that, on the basis of homology with UvrB, have been highlighted in previous studies as potential UvrA-interacting residues (Assenmacher et al., 2006; Deaconescu et al., 2006). In the structure of Mfd, residues R165 and R181 are buried in the interface of D2 and D7, whereas F185 is exposed (Figure 1A). These three residues are conserved between Mfd and UvrB in E. coli and are all highly conserved among Mfd proteins from different species. Mfd DBHM is an N-terminally truncated derivative of E. coli Mfd that lacks domains 1a, 1b, and 2 (Figure 1A). We puried Mfd D2AAA and Mfd DBHM and examined their ATPase, DNA translocation, and RNAP displacement activities (Figure 2). Mfd DBHM had deregulated ATPase and DNA translocase activities and displaced stalled transcription complexes from DNA more quickly than WT Mfd did. It thus behaved like previously characterized Mfd derivatives in which the autoinhibition of the protein was disrupted by N- or C-terminal truncations (Murphy et al., 2009; Smith et al., 2007). Mfd D2AAA was unable to translocate DNA in the absence of RNAP and displaced RNAP from DNA at a rate similar to the WT protein. However, the ATPase activity of Mfd D2AAA was considerably higher than WT, and although its maximal activity was lower than that of Mfd DBHM, it was comparable to that of Mfd DD7 ($110 min1) (Smith et al., 2007). The properties of Mfd D2AAA suggest that the substitutions within the UvrA interaction surface may affect the interaction of D2 with D7 and thus partially relieve the autoinhibition of the translocase domains. To investigate this possibility further, we determined the effect of disrupting the highly conserved cluster of residues within D7 that interact with D2 (Deaconescu et al., 2006). We constructed a derivative of Mfd in which residues E1045, D1048, and R1049 were all substituted with alanine (Figure 1A) and examined the activities of the puried protein (Figure S1). We found that Mfd E1045A
Molecular Cell
Figure 2. Biochemical Analysis of Mfd Derivatives Containing Mutations within the UvrB Homology Domain
(A) DNA translocation activity. Displacement of a triplex-forming oligonucleotide (TFO) from supercoiled pSRTB1 plasmid was monitored by EMSA. The graph shows the percentage of TFO displaced at each time point, normalized for the amount of triplex present at t = 0. Data are the mean of at least three independent experiments and are shown with standard deviation. (B) RNAP displacement activity. Transcription complexes were stalled by nucleotide starvation on a PCR product carrying the T7A1 promoter, and displacement by Mfd was monitored by EMSA. The graph shows the percentage of the transcription elongation complex present at t = 0 that was displaced at each time point. Data are the mean of at least three independent experiments and are shown with standard deviation. (C) ATPase activity. Rates were measured at 37 C using an NADH-coupled assay. Data are the mean of at least three independent experiments and are shown with standard deviation.
D1048A R1049A retained the ability to displace stalled RNAP and exhibited the elevated ATPase rate and robust RNAP-independent translocase activity that is a signature of derepressed Mfd derivatives. Taken together, our results indicate that the derepression of Mfd ATPase/translocase activity, which has previously been reported for truncated Mfd derivatives lacking one or more domains, can be effected in the full-length protein by amino acid substitutions in either of the surfaces that constitute the D2:D7 interaction. The UvrB Homology Module of Mfd, but Not Domain 7, Is Essential for TCR In Vitro To determine which aspects of Mfd function are required for strand-specic repair, we established a patch-synthesis assay to monitor repair of UV-induced lesions in vitro. RNAP, Mfd, repair proteins, and all necessary cofactors were incubated with UV-irradiated plasmid DNA, and repair was monitored by the incorporation of [a-32P]dATP into repair patches. The template (transcribed) and nontemplate strands of a reporter cassette downstream of the strong T7A1 promoter were distinguished by denaturing gel electrophoresis following asymmetric cleavage by restriction enzymes (Figure 3A). The nontemplate strand is repaired by the global NER pathway and is not a substrate for TCR (Selby and Sancar, 1993) (Figure S2A). The efciency of repair of the template strand relative to the rate of global NER can thus be deduced by comparing the amount of radiolabel incorporated into the template and nontemplate strands. In the absence of RNAP, repair patches formed in the two strands with approximately equal efciency, as both strands are repaired by global NER (Figure 3B, lane 1). As expected from previous studies of TCR (Selby and Sancar, 1993), tran-
scription in the absence of Mfd specically inhibited repair of the template strand (Figure 3B, lane 2), and transcription in the presence of WT Mfd specically enhanced repair of the template strand (Figure 3B, lane 3). The Mfd-dependent strand bias in our assay was sensitive to UvrA concentration (Figure S2B), as reported previously (Selby and Sancar, 1995b). When Mfd DBHM or Mfd D2AAA was used in the assay, there was little strand bias in the synthesis of repair patches (Figure 3B, lanes 5 and 6), but Mfd DD7 catalyzed the preferential repair of the template strand almost as efciently as WT Mfd did (Figure 3B, lane 4). We conclude that all three mutant proteins displaced transcription complexes stalled at lesions, thus preventing the inhibition of template strand repair, but that only Mfd DD7 retained the ability to enhance the repair of the template strand. The simplest interpretation of these results is that the interaction of D2 of Mfd with UvrA is essential for the mechanism by which repair is accelerated during TCR, and that substitution of conserved residues within the D2 surface or deletion of the BHM disrupts this interaction. The observation that Mfd DD7 can catalyze strand-specic repair indicates that autoregulation of UvrA binding and DNA translocation by Mfd is dispensable for the mechanism by which strand-biased repair is achieved. Further support for this proposal comes from the observation that the deregulated Mfd E1045A D1048A R1049A mutant can also catalyze strand-specic repair (Figure S1D). The UvrB Homology Module of Mfd, but Not Domain 7, Is Essential for TCR In Vivo To determine whether the results obtained in the in vitro TCR assay reect the situation within a cell, we used a primer extension assay to detect repair in vivo. UV-induced lesions can block DNA polymerase, and analysis of the primer extension products
Molecular Cell
Figure 3. Effect of Mfd Derivatives on TCR In Vitro

(A) TCR was reconstituted in vitro using puried proteins and a UV-irradiated plasmid substrate. The plasmid contained a reporter cassette anked by BsrGI and SphI sites, downstream of the T7A1 promoter. Repair was monitored by incorporation of [a-32P]dATP into repair patches, and cleavage by BsrGI and SphI to produce nontemplate strand (NT) and template strand (T) fragments. The gure shows the analysis of the products by denaturing gel electrophoresis. TCR reactions contained RNAP, UvrA, UvrB, UvrC, UvrD, DNA ligase, DNA polymerase, and where indicated, Mfd or an Mfd derivative. Global NER reactions (GGR) were performed as for TCR reactions, except that RNAP and Mfd were omitted. The control reaction in lane C was performed as for the GGR reaction, except that UvrA, UvrB, UvrC, and UvrD were omitted. (B) Quantication of strand bias. The gure shows the ratio of incorporation of radioactive label into the T and NT strands under the conditions indicated. Data are the mean of at least three independent experiments and are shown with standard deviation.
allows detection of lesions at single-nucleotide resolution (Wellinger and Thoma, 1996). Our experiments monitored the generation and repair of lesions within the template strand of
a plasmid-borne lacI gene in mfd+ and mfd cells (Figure 4A). Because the primer annealed upstream of the promoter, the experiment allowed analysis of both the untranscribed promoter DNA and the transcribed region of the lacI gene. Mfddependent TCR does not occur in untranscribed regions or in the promoter-proximal regions that are transcribed prior to dissociation of the sigma subunit of RNAP (Selby and Sancar, 1995b). In these experiments, the DNA between the primer and approximately +15 thus acted as a control region in which repair was independent of Mfd, and the effect of Mfd and Mfd mutants could be assessed in the region downstream of approximately +15. Comparison of the primer extension products obtained from samples taken before and immediately after irradiation revealed the presence of premature termination products unique to the irradiated DNA (Figure 4A). The pattern of the predominant damage-induced bands correlated with the location of pyrimidine dimers and previously published analysis of UV-induced lesions in the lacI gene (Chandrasekhar and Van Houten, 1994; Kunala and Brash, 1995). However, the primer extension assay detects any damage that stalls Taq DNA polymerase, and it is likely that not all of the lesions detected were pyrimidine dimers. The rate at which individual lesions were repaired (detected as the disappearance of the primer extension product terminating at that location) differed considerably. Some lesions within the transcribed region of the lacI gene were repaired much more rapidly in the presence of Mfd than in its absence. Other lesions showed little repair in either strain, even after 30 min. One region in which an Mfd-dependent effect was clear was a pair of lesions that produced bands at +30/+31. These correspond to a known mutation hotspot in cells lacking Mfd (Kunala and Brash, 1992; Oller et al., 1992) and were chosen as the test lesions for assessing the effects of Mfd mutants. To quantify the effect of Mfd and Mfd derivatives on repair, and to avoid the effects of lane-to-lane variations, we calculated the ratio of the intensity of the +30/+31 bands to the intensity of a group of bands from +1 to +7 at each time point. Lesions from +1 to +7 of genes are not subject to TCR, and we reasoned that if the lesions at +30/+31 were repaired at an enhanced rate, the ratio would decrease over time, whereas if the lesions at +30/+31 were repaired by the same global NER pathway as the +1 to +7 region, the ratio would remain constant as repair progressed. In agreement with these predictions, the ratio remained constant throughout the course of the experiment in cells that lacked Mfd (Figure 4A, group 2). In strains that expressed WT Mfd from either a chromosomal or plasmid-borne allele, the ratio decreased within the rst 10 min, reecting preferential repair of the lesions at +31/32, and then remained constant for the remainder of the experiment (Figure 4A, groups 1 and 3). To examine the effects of Mfd mutants on TCR in vivo, experiments were conducted using mfd cells transformed with plasmids carrying the mutant mfd alleles. Preferential repair of the lesions at +30/+31 was observed in cells containing Mfd DD7 (Figure 4A, group 6), but not in cells containing Mfd DBHM or Mfd D2AAA (Figure 4A, groups 4 and 5). TCR occurred over a longer period with Mfd DD7 than with the WT protein, which may reect the sequestration of UvrA by the unshielded D2 of Mfd DD7 (Deaconescu et al., 2006; Selby and Sancar, 1995a).
Molecular Cell
In vivo
Figure 4. Effect of Mfd Derivatives In Vivo

(A) Primer extension analysis of TCR in strains transformed with a plasmid carrying a lacI gene. Samples were taken before UV irradiation (No UV), immediately following irradiation (0), and at intervals after incubation in the dark at 37 C. Plasmid DNA isolated from each sample was analyzed using a primer that annealed to the template strand upstream of the lacI promoter. The gel image shows the primer extension products obtained from cultures of strain UNCNOMFD transformed with pET21 (mfd) or pETMfd2 (mfd+). The bands reect the location of lesions within the lacI template strand. The rst transcribed base and direction of transcription are indicated by a bent arrow. A dotted line marks the transition between the region repaired by GGR and the region repaired by TCR at +15. Box brackets indicate the regions whose intensity was quantied. The chart shows the relative intensities of bands in regions 1 (+30/+31) and 2 (+1 to +7). The four bars in each group represent samples taken at 0, 10, 20, and 30 min postirradiation. Group 1: AB1157 transformed with pET21a. Group 2: UNCNOMFD transformed with pET21a. Groups 36: UNCNOMFD transformed with pETMfd2 derivatives carrying the mfd alleles indicated. Data are the mean of at least three independent experiments and are shown with standard deviation. (B) Mutation frequency decline. The frequency of reversion of UV-irradiated argE3(Oc) strains to Arg+ was determined in samples that had either been added to rich medium immediately after irradiation (mt fq [unstarved]) or that had been held in a medium lacking amino acids for 30 min prior to addition of rich medium (mt fq [starved]). A mt fq (unstarved) to mt fq (starved) ratio of greater than 1 is indicative of MFD. Bar 1: AB1157. Bar 2: UNCNOMFD transformed with pET21a. Bars 36: UNCNOMFD transformed with pETMfd2 derivatives carrying the mfd alleles indicated. Data are the mean of at least three independent experiments and are shown with standard deviation.
A phenotype linked to TCR in E. coli is mutation frequency decline (MFD), which is a reduction in the frequency of certain suppressor mutations when cells are transiently starved of amino acids after UV irradiation (Selby and Sancar, 1994). The Mfd protein is essential for this phenotype, which is thought to be due to enhanced repair of lesions within tRNA genes. We examined the effect of Mfd mutants on the MFD phenotype by determining the frequency with which Arg+ revertants arose in argE3(ochre) strains. As expected, in cells expressing WT Mfd from either a chromosomal or plasmid-borne allele, the mutation
frequency was lower in cells that were starved for 30 min after UV irradiation than in cells that were not starved (Figure 4B, lanes 1 and 3). Cells that lacked Mfd did not exhibit MFD in response to starvation (Figure 4B, lane 2). Mfd DD7 was able to restore the MFD phenotype to mfd cells, but cells expressing Mfd DBHM or Mfd D2AAA showed little or no starvation-dependent MFD (Figure 4B, lanes 46). The results of the in vivo TCR and MFD assays conrm those of the in vitro TCR assays, i.e., the BHM (and specically the UvrA interaction surface of D2) is essential for the
Molecular Cell
Figure 5. Interactions with UvrA

(A) Effect of UvrB and Mfd on the GTPase activity of UvrA. Rates were measured at 37 C using an NADH-coupled assay and 2 mM GTP. Data are the mean of at least three independent experiments and are shown with standard deviation. (B) Bacterial two-hybrid analysis: A fragment of UvrA containing the UvrB interaction domain was fused to the a subunit of RNAP, and a fragment of Mfd containing D2 was fused to the lcI protein. Interaction between the fusion proteins recruits RNAP to a lacZ reporter construct, and b-galactosidase activity reects the strength of the interaction. The chart shows the specic b-galactosidase activity measured in KS1 cells transformed with pRA02 and pRA03 derivatives encoding the indicated combinations of fusion proteins. indicates expression of lcI or a subunit without additional fused domains. Data are the mean of at least three independent experiments and are shown with standard deviation.
m
enhanced repair of lesions within the template strand and for MFD, but the regulatory role of D7 is dispensable for these processes. Substitutions within Mfd D2 Disrupt the Mfd:UvrA Interaction When UvrA and UvrB interact, nucleotide hydrolysis by UvrA is inhibited and nucleotide hydrolysis by UvrB is increased (Caron and Grossman, 1988). The inhibition of UvrA by UvrB can be observed if GTP hydrolysis is monitored, because UvrB does not hydrolyze GTP (Caron and Grossman, 1988) (Figure 5A, compare bars 1 and 2). We found that domains 1a, 2, and 1b of UvrB (which are conserved in Mfd) are sufcient to regulate UvrA (Figure S3A), and so we investigated whether Mfd regulates the NTPase activity of UvrA. Mfd has no detectable GTPase activity (data not shown), and even at high concentration, WT Mfd had no effect on the GTPase activity of UvrA (Figure 5A, bar 3). Because the UvrA interaction surface is obscured in full-length Mfd by its interaction with D7, we also examined the effect of Mfd DD7 (in which D7 is deleted) and Mfd E1045A D1048A R1049A (in which the D2:D7 interface is disrupted by substitutions within D7). Both of these mutant proteins inhibited the GTPase activity of UvrA (Figure 5A, bars 4 and 6, and Figure S3B), which supports the hypothesis that in WT Mfd, the UvrA interaction surface is masked by its interaction with D7, and that a conformational change enables the proteins to interact.
The ability of Mfd DD7 to repress UvrA GTPase activity was abolished when the truncated Mfd contained the D2AAA substitutions (Figure 5A, compare bars 4 and 5). The D2AAA substitutions also disrupted Mfd:UvrA interactions in a bacterial two-hybrid system in which the strength of the interaction between protein fragments is reected by b-galactosidase activity from a lacZ reporter construct (Figure 5B, compare bars 1 and 2). These results indicate that the D2AAA substitutions disrupt the Mfd:UvrA interaction and support the proposition that the UvrA interaction surface of Mfd is similar to that dened in UvrB. Mutations in UvrA Have a Differential Effect on TCR and Global NER During global NER, UvrA must recognize DNA damage, load UvrB onto the damaged DNA, and then dissociate from the preincision complex. Structural analysis of UvrA has aided in the identication of the regions responsible for interacting with DNA and with UvrB, and it has also revealed the architecture of the two ATPase sites (termed proximal and distal) within each UvrA monomer (Figure 1B) (Pakotiprapha et al., 2008, 2009; Timmins et al., 2009). The roles of these activities in global NER have been studied extensively, but their importance for TCR has not been investigated. As the process leading to the formation of the preincision complex is different in TCR and global NER, disruptive mutations within UvrA may fall into three classes: (1) mutations that affect a function that is equally important for global NER and TCR, (2) mutations that affect a function that is required for global NER but not for TCR, and (3) mutations that affect a function that is required for TCR but not for global NER. We puried UvrA derivatives carrying substitutions that disrupt the proximal ATPase activity (K37A), the distal ATPase activity (K646A), DNA damage recognition (G502D, which likely exerts its effect indirectly), DNA binding activity (R712A R722A R724A
Molecular Cell
Figure 6. Effect of UvrA Derivatives on TCR In Vitro

(A) TCR was analyzed using a patch-synthesis assay as described in Figure 3. Reactions contained 4 nM UvrA. (BD) Quantication of relative repair of the template and nontemplate strands in reactions conducted with 4nM, 16 nM, or 64 nM UvrA. The gure shows the ratio of incorporation of radioactive label into the T and NT strands under the conditions indicated. Data are the mean of at least three independent experiments and are shown with standard deviation.
mutants: UvrA D131-248 lacks the UvrB binding domain (also likely to be the Mfd binding domain), and UvrA D290-400 lacks the insertion domain attached to the distal ATPase module (Pakotiprapha et al., 2008). To facilitate purication, the UvrA mutants carried an N-terminal His-tag, which has no effect on UvrA function in NER or TCR (Manelyte et al., 2009) (Figure S4). We checked the activity of the mutant proteins by measuring their GTPase activities, their interaction with UvrB (assessed by inhibition of GTPase activity), and their DNA binding activity (Table S1). We tested the global NER and TCR activities of WT and mutant UvrA proteins in the patch-synthesis repair assay at concentrations of 464 nM (Figure 6). We found no mutants that could catalyze global NER but not TCR. Two of the mutant proteins (UvrA R712A R722A R724A R730A and UvrA D131-248) were unable to catalyze detectable global NER or TCR at any of the concentrations tested (data not shown), indicating that the UvrB-binding domain of UvrA and the ability to bind to DNA are essential for both pathways. UvrA K646A showed detectable levels of repair only at 64 nM, and preferential repair of the template strand was observed in the presence of Mfd and transcribing RNAP. Disruption of the distal ATPase activity thus decreases the efciency of both global NER and TCR, but the distal ATPase is not essential for either process. UvrA R216A E222A showed a level of strand bias that was similar to the optimal level obtained with WT UvrA, indicating that the disruption of the UvrB/Mfd interaction surface affected global NER and TCR equally under the conditions of our assay. Three of the mutant proteins (UvrA D290-400, UvrA K37A, and UvrA G502D) showed a signicantly enhanced bias toward repair of the template strand under all conditions tested. As these mutants are all expected to cause loss of function, the simplest interpretation of these results is that these mutations have a greater detrimental effect on global NER (which repairs the nontemplate strand) than on TCR (which repairs the template strand). We conclude that the insertion domain, the proximal ATPase activity, and the DNA damage recognition activity of UvrA play a less important role in TCR than in global NER. We also analyzed the ability of three mutant UvrB proteins to catalyze TCR. Each was defective in a function related to damage recognition. The proteins were UvrB K45A (containing a substitution in the Walker A box that abolishes ATPase activity [Seeley and Grossman, 1989]), UvrB Y95A Y96A (containing substitutions of residues at the base of the b-hairpin that are important for DNA damage recognition [Moolenaar et al., 2001]), and UvrB Y101A F108A (containing substitutions of residues at the tip of the b-hairpin that are important for the strandseparating/clamping activity of UvrB [Moolenaar et al., 2001]). In a patch-synthesis assay, none of these proteins supported detectable NER or TCR (data not shown). The absence of global NER activity is consistent with previous studies of these proteins, and the absence of TCR indicates that none of the properties of UvrB that are disrupted by the substitutions are made redundant by the action of RNAP and Mfd. The Insertion Domain of E. coli UvrA Is Involved in DNA Damage Recognition A derivative of Bacillus stearothermophilus UvrA that lacked the insertion domain exhibited no apparent functional defects, and it
R730A), and the UvrB interaction surface (R216A E222A) (Pakotiprapha et al., 2009; Thiagalingam and Grossman, 1991; Wang and Grossman, 1993) (Figure 1B). We also puried two deletion
Molecular Cell
Figure 7. Activity of UvrA D290-400 in Incision and DNA Binding Assays

(A) Nicking of a 50 bp duplex containing a single uorescein (FldT) adduct. The substrate was incubated at 37 C with UvrA, UvrB, and UvrC, and aliquots were removed and quenched at the times indicated. The samples were analyzed by denaturing acrylamide gel electrophoresis, and the proportion of the radiolabeled DNA fragment that had been nicked was quantied. Data are the mean of three independent experiments and are shown with standard deviation. (B) Nicking of a UV-irradiated supercoiled plasmid. The plasmid was incubated at 37 C with UvrA, UvrB, and UvrC, and aliquots were removed and quenched at the times indicated. The samples were analyzed by native agarose gel electrophoresis, and nicking was detected as the conversion of supercoiled plasmid to nicked open circle DNA. Data are the mean of three independent experiments and are shown with standard deviation. (C and D) Binding of UvrA and UvrA D290-400 to a 50 bp duplex containing either a single FldT adduct or no lesion. The indicated concentrations of protein were incubated with 1 nM DNA for 20 min at 37 C. Samples were analyzed by EMSA. Data were tted to a one-site binding equation, and the calculated dissociation constants = 10.5 1 nM and Kdno lesion = 20.8 4 nM. Data are the
were as follows: WT UvrA, KdFldt = 5.6 1 nM, Kdno lesion = 43 17 nM and UvrA D290-400, KdFldt mean of three independent experiments and are shown with standard error.
was concluded that the insertion domain was not critical for UvrA function in vitro (Pakotiprapha et al., 2008). In contrast, analysis of Deinococcus radiodurans UvrA2 (a class II UvrA of uncertain function) suggested that its insertion domain is involved in DNA binding and damage recognition (Timmins et al., 2009). To understand why deletion of the insertion domain of E. coli UvrA (a class I UvrA) differentially affected TCR and global NER in our patch-synthesis assay, we examined the specicity of the DNA binding and global NER activities of UvrA D290-400. First, we analyzed the ability of UvrA D290-400 to support incision of a 50 bp duplex carrying a uorescein-dT (FldT) adduct (Figure 7A). In agreement with the results of similar experiments by Pakotiprapha et al. (2008), we observed no difference between the activity of WT UvrA and UvrA D290-400 in this assay. We then analyzed the ability of UvrA D290-400 to support incision of UV-induced lesions in a supercoiled plasmid DNA template (Figure 7B). In contrast to the results with the short duplex, incision was signicantly slower in the presence of UvrA D290-400 than in the presence of WT UvrA. UvrA D290-400 was also defective in the incision of UV-induced lesions from relaxed closedcircular plasmid templates, indicating that the topology of the template is not important for the effect (data not shown). We measured the afnity of WT UvrA and UvrA D290-400 for a 50 bp undamaged duplex and a 50 bp duplex containing a FldT adduct and found that deletion of the insertion domain decreased the ability of UvrA to distinguish between the two substrates (Figures 7C and 7D). The reduced ability of UvrA D290-400 to catalyze global NER on the plasmid templates used in our patch-synthesis and incision assays likely results from the nonproductive binding of the
protein to the undamaged DNA present in these templates, and we suggest that no defect was found in the NER activity of B. stearothermophilus UvrA lacking an insertion domain because the analysis was conducted with a short oligonucleotide that lacked nonspecic competitor DNA. We conclude that the insertion domain of E. coli UvrA is involved in damage recognition and that the model of DNA binding and damage recognition proposed for class II UvrAs (Timmins et al., 2009) is applicable to class I UvrAs. DISCUSSION When the product of the mfd gene was identied as the bacterial transcription-repair coupling factor, a striking feature was that a region toward the N-terminal region of the protein was similar in sequence to a region of UvrB (Selby and Sancar, 1993). It was found that this region of Mfd interacted with UvrA, and it has since been widely assumed that this interaction is responsible for the enhanced rate of repair that is observed during TCR (Hanawalt and Spivak, 2008; Roberts and Park, 2004; Selby and Sancar, 1993; Truglio et al., 2006a). In this work, we have tested this hypothesis experimentally. Our results show that the BHM of Mfd is essential for the Mfd-dependent enhancement of repair, although it is dispensable for the process by which RNAP is removed from the site of damage. The effect on TCR of deleting the BHM can be recapitulated in a fulllength protein by substitution of conserved residues on the surface of D2 of the protein, and these substitutions disrupt the Mfd:UvrA interaction. Our results support the conclusion that the interaction of Mfd with UvrA is essential for TCR and
Molecular Cell
that this interaction is similar in nature to the well-characterized UvrA:UvrB interaction. In addition to its role in interaction with UvrA, the BHM of Mfd is also involved in autoregulation of the protein through its interaction with D7. The surface of D2 that interacts with UvrA overlaps with the surface that interacts with D7, and we found that substitution of residues within the area of overlap led to derepression of Mfd ATPase activity. Alanine substitutions of the residues within D7 that interact with D2 also led to derepression. The observation of disrupted autoregulation in these full-length variants of Mfd provides evidence that the interaction between D2 and D7 maintains the integrity of an inhibitory clamp that constrains the movement of the motor domains unless Mfd is bound to RNAP (Murphy et al., 2009). While the substitutions within D7 led to derepression of both ATPase activity and DNA translocase activity, the substitutions within D2 caused derepression of only ATPase activity. This suggests that the DNA translocase activity of Mfd can be regulated independently of the ATPase activity. It is possible that the substitutions made in D7 abolish the interdomain interaction, allowing the two halves of the inhibitory clamp to behave completely independently, whereas the substitutions made in D2 allow some interdomain movement at the interface without enabling the two parts of the inhibitory clamp to separate. The nding that Mfd DD7 is able to catalyze TCR both in vitro and in vivo indicates that the regulatory mechanisms in which D7 is involved are not essential features of the TCR mechanism. Thus, although the autoinhibition of the DNA translocase activity is likely to benet the cell by preventing nonproductive ATP hydrolysis and unscheduled displacement of RNAP, the mechanism by which repair is accelerated during TCR does not require a regulatable translocase activity. Similarly, while the ability of D7 to block the UvrA interaction surface in free Mfd appears to prevent sequestration of UvrA into nonproductive complexes (Deaconescu et al., 2006; Selby and Sancar, 1995a), the mechanism of TCR is not dependent on the ability of D7 to compete with UvrA for binding to the BHM. To understand the mechanism by which the Mfd:UvrA interaction enhances the rate of repair, we searched for UvrA and UvrB mutants that were differentially affected in global NER and TCR. We found three UvrA mutants that were more defective in global NER than they were in TCR. The property that links the three is defective damage recognition. UvrA K37A carries a substitution within the Walker A motif of the proximal ATPase site that is thought to abolish ATP hydrolysis by this site without signicantly affecting ATP binding (Thiagalingam and Grossman, 1991). One reported consequence of this substitution is that it abolishes the ability of UvrA to discriminate between UV-damaged and undamaged DNA (Thiagalingam and Grossman, 1991; Timmins et al., 2009). UvrA G502D is also unable to discriminate between UV-damaged and undamaged DNA (Wang and Grossman, 1993). It was originally studied because it lay within a putative DNA binding helix-turn-helix motif, but the crystal structure of B. stearothermophilus revealed that the residue is located well away from the likely path of the DNA and is likely to exert its effects on damage specicity indirectly (Pakotiprapha et al., 2008). UvrA D290-400 lacks the insertion domain, which is important for damage recognition in
UvrA2 (Timmins et al., 2009). Our results show that the insertion domain is also important for DNA damage specicity in a class I UvrA of the type involved in NER and suggest an explanation for the fact that an earlier study found no defect in NER when UvrA lacked the insertion domain (Pakotiprapha et al., 2008). The differential effects of these three UvrA mutants on global NER and TCR indicates that Mfd, or a combination of Mfd and a stalled transcription complex, negates the need for UvrA to detect DNA damage in the way that it does during global NER. This in turn suggests that Mfd acts at the rst step of the repair process, enhancing the rate of repair by altering the way in which UvrA loads onto the damaged DNA, rather than at a later stage such as destabilizing the UvrA:UvrB:DNA complex. The discovery that the initial damage recognition step of bacterial NER (damage detection by UvrA) is dispensable for TCR raises parallels with eukaryotic repair pathways. The TCR and NER pathways in bacteria and eukaryotes follow broadly similar strategies but appear to have evolved independently (Ogrunc et al., 1998). The initial damage recognition step during global NER in eukaryotes involves the XPC protein, which is dispensable for TCR (Mu and Sancar, 1997). Our data now show that the involvement of RNAP and a transcription-repair coupling factor circumvents the need for mechanisms that are important for locating DNA lesions in nontranscribed DNA in both bacteria and eukaryotes.
EXPERIMENTAL PROCEDURES Strains, Plasmids, and Proteins Details of strains, plasmids, and proteins used are given in Supplemental Experimental Procedures. Assays for DNA Translocation, RNAP Displacement, and NTPase Activity DNA translocation by Mfd was assayed by monitoring the displacement of a triplex-forming oligonucleotide from supercoiled pSRTB1 plasmid as described in Smith et al. (2007). Displacement of stalled transcription complexes by Mfd was analyzed by EMSA as described in Smith et al. (2007), except that the DNA substrate carrying the T7A1 promoter was a PCR fragment generated from plasmid pSRT7A1 (Smith et al., 2007). ATPase activity was measured using an ATP-NADH-coupled spectrophotometric assay as described in Smith et al. (2007). GTPase activity was measured using the same protocol as ATPase activity, except that 2 mM GTP was used in place of ATP, NADH concentration was 200 mM, reaction volume was 500 ml, and absorbance was measured using a spectrophotometer. The concentrations of proteins used in GTPase assays are indicated in the relevant legends. In Vitro Patch-Synthesis Assay for TCR Strand-specic repair was analyzed using a modication of the repair synthesis assay reported in Selby and Sancar (1993). A supercoiled DNA template containing randomly located UV-induced photoproducts was incubated for 20 min with UvrA, UvrB, UvrC, UvrD, DNA polymerase I, DNA ligase, dNTPs, rNTPs, NADH, and [a-32P]dATP. For TCR reactions, RNAP and Mfd were also present. The template contained an $170 bp reporter cassette downstream of the T7A1 promoter. Repair patches were detected by excising the reporter with BsrGI and SphI and analyzing the products by denaturing gel electrophoresis. Refer to Supplemental Experimental Procedures for full protocol. Primer Extension Analysis of TCR In Vivo Strains AB1157 (mfd+) and UNCNOMFD (mfd) transformed with pET21a (Novagen), pETMfd2, or pETMfd2 derivatives encoding mutant Mfd proteins
Molecular Cell
were grown to mid-log phase, resuspended in M56 minimal salts, and irradiated with 40 J/m2 254 nm UV light. Plasmid DNA was isolated at intervals, digested with RsaI, and analyzed by primer extension using Taq DNA polymerase and a radiolabeled primer that anneals to the template DNA strand from 104 to 85 upstream of the lacI gene carried on pET21a and pETMfd2. The products were analyzed by denaturing gel electrophoresis. Refer to Supplemental Experimental Procedures for full protocol. MFD Assay Strains AB1157 (mfd+ argE3(Oc)) and UNCNOMFD (mfd argE3(Oc)) transformed with pET21a, pETMfd2, or pETMfd2 derivatives encoding mutant Mfd proteins were grown to mid-log phase, resuspended in M9 minimal medium that lacked amino acids, and irradiated with 20 J/m2 254 nm UV light. Aliquots of irradiated culture were added to rich media either immediately or after 30 min incubation in M9 minimal medium that lacked amino acids. After overnight incubation, the frequency of Arg+ mutants in each sample was determined using selective growth media. Refer to Supplemental Experimental Procedures for full protocol. Bacterial Two-Hybrid Assay Bacterial two-hybrid assays were performed as described in Manelyte et al. (2009) using reporter strain KS1 transformed with the indicated combinations of pRA02 and pRA03 derivatives encoding rpoA-uvrA and cI-mfd fusions. Assays for DNA Binding and Incision Activities of NER Proteins DNA binding by UvrA was studied by EMSA. The substrates were blunt-ended 32 P-labeled 50 bp duplexes with or without a single FldT adduct. UvrA proteins were incubated for 20 min at 37 C with 1 nM of substrate and 2 mM ATP. The ability of UvrA and UvrA D290-400 to support the steps of NER up to and including the point of incision by UvrC was assessed by monitoring the cleavage of either a 32P-labeled 50 nt oligonucleotide containing a single FldT adduct or a tritiated 4.2 kb 3H-labeled plasmid containing randomly located UV-induced lesions. Reactions contained DNA substrate, UvrA, UvrB, UvrC, and ATP and were incubated at 37 C. In each case, reaction products were analyzed by quantifying the labeled DNA following gel electrophoresis. Refer to Supplemental Experimental Procedures for full protocols. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, Supplemental References, four gures, and one table and can be found with this article online at doi:10.1016/j.molcel.2010.11.012. ACKNOWLEDGMENTS This work was funded by research grants BB/E004695/1 and BB/F007361/1 from the BBSRC. We are grateful to Andy Yuan, Ann Hochschild, and the E. coli genome project at the University of Wisconsin, Madison for providing strains. We are also grateful to Seth Darst, Lars Westblade, and Mark Szczelkun for useful discussions and comments on the manuscript. Received: May 12, 2010 Revised: August 3, 2010 Accepted: September 14, 2010 Published: December 9, 2010 REFERENCES Assenmacher, N., Wenig, K., Lammens, A., and Hopfner, K.P. (2006). Structural basis for transcription-coupled repair: the N terminus of Mfd resembles UvrB with degenerate ATPase motifs. J. Mol. Biol. 355, 675683. Caron, P.R., and Grossman, L. (1988). Involvement of a cryptic ATPase activity of UvrB and its proteolysis product, UvrB* in DNA repair. Nucleic Acids Res. 16, 1089110902. Chandrasekhar, D., and Van Houten, B. (1994). High resolution mapping of UV-induced photoproducts in the Escherichia coli lacI gene. Inefcient repair
of the non-transcribed strand correlates with high mutation frequency. J. Mol. Biol. 238, 319332. Deaconescu, A.M., Chambers, A.L., Smith, A.J., Nickels, B.E., Hochschild, A., Savery, N.J., and Darst, S.A. (2006). Structural basis for bacterial transcriptioncoupled DNA repair. Cell 124, 507520. Hanawalt, P.C., and Spivak, G. (2008). Transcription-coupled DNA repair: two decades of progress and surprises. Nat. Rev. Mol. Cell Biol. 9, 958970. Kad, N.M., Wang, H., Kennedy, G.G., Warshaw, D.M., and Van Houten, B. (2010). Collaborative dynamic DNA scanning by nucleotide excision repair proteins investigated by single-molecule imaging of quantum-dot-labeled proteins. Mol. Cell 37, 702713. Kunala, S., and Brash, D.E. (1992). Excision repair at individual bases of the Escherichia coli lacI gene: relation to mutation hot spots and transcription coupling activity. Proc. Natl. Acad. Sci. USA 89, 1103111035. Kunala, S., and Brash, D.E. (1995). Intragenic domains of strand-specic repair in Escherichia coli. J. Mol. Biol. 246, 264272. Malta, E., Moolenaar, G.F., and Goosen, N. (2007). Dynamics of the UvrABC nucleotide excision repair proteins analyzed by uorescence resonance energy transfer. Biochemistry 46, 90809088. Manelyte, L., Guy, C.P., Smith, R.M., Dillingham, M.S., McGlynn, P., and Savery, N.J. (2009). The unstructured C-terminal extension of UvrD interacts with UvrB, but is dispensable for nucleotide excision repair. DNA Repair (Amst.) 8, 13001310. Moolenaar, G.F., Hoglund, L., and Goosen, N. (2001). Clue to damage recognition by UvrB: residues in the beta-hairpin structure prevent binding to nondamaged DNA. EMBO J. 20, 61406149. Mu, D., and Sancar, A. (1997). Model for XPC-independent transcriptioncoupled repair of pyrimidine dimers in humans. J. Biol. Chem. 272, 75707573. Murphy, M.N., Gong, P., Ralto, K., Manelyte, L., Savery, N.J., and Theis, K. (2009). An N-terminal clamp restrains the motor domains of the bacterial transcription-repair coupling factor Mfd. Nucleic Acids Res. 37, 60426053. Ogrunc, M., Becker, D.F., Ragsdale, S.W., and Sancar, A. (1998). Nucleotide excision repair in the third kingdom. J. Bacteriol. 180, 57965798. Oller, A.R., Fijalkowska, I.J., Dunn, R.L., and Schaaper, R.M. (1992). Transcription-repair coupling determines the strandedness of ultraviolet mutagenesis in Escherichia coli. Proc. Natl. Acad. Sci. USA 89, 1103611040. Orren, D.K., and Sancar, A. (1989). The (A)BC excinuclease of Escherichia coli has only the UvrB and UvrC subunits in the incision complex. Proc. Natl. Acad. Sci. USA 86, 52375241. Pakotiprapha, D., Inuzuka, Y., Bowman, B.R., Moolenaar, G.F., Goosen, N., Jeruzalmi, D., and Verdine, G.L. (2008). Crystal structure of Bacillus stearothermophilus UvrA provides insight into ATP-modulated dimerization, UvrB interaction, and DNA binding. Mol. Cell 29, 122133. Pakotiprapha, D., Liu, Y., Verdine, G.L., and Jeruzalmi, D. (2009). A structural model for the damage-sensing complex in bacterial nucleotide excision repair. J. Biol. Chem. 284, 1283712844. Park, J.S., Marr, M.T., and Roberts, J.W. (2002). E. coli Transcription repair coupling factor (Mfd protein) rescues arrested complexes by promoting forward translocation. Cell 109, 757767. Patel, S., Venkatesh, K.V., and Edwards, J.S. (2004). An integrated mechanistic model for transcription-coupled nucleotide excision repair. DNA Repair (Amst.) 3, 343348. Roberts, J., and Park, J.S. (2004). Mfd, the bacterial transcription repair coupling factor: translocation, repair and termination. Curr. Opin. Microbiol. 7, 120125. Seeley, T.W., and Grossman, L. (1989). Mutations in the Escherichia coli UvrB ATPase motif compromise excision repair capacity. Proc. Natl. Acad. Sci. USA 86, 65776581. Selby, C.P., and Sancar, A. (1990). Transcription preferentially inhibits nucleotide excision repair of the template DNA strand in vitro. J. Biol. Chem. 265, 2133021336. Selby, C.P., and Sancar, A. (1993). Molecular mechanism of transcriptionrepair coupling. Science 260, 5358.
Molecular Cell
Selby, C.P., and Sancar, A. (1994). Mechanisms of transcription-repair coupling and mutation frequency decline. Microbiol. Rev. 58, 317329. Selby, C.P., and Sancar, A. (1995a). Structure and function of transcriptionrepair coupling factor. I. Structural domains and binding properties. J. Biol. Chem. 270, 48824889. Selby, C.P., and Sancar, A. (1995b). Structure and function of transcriptionrepair coupling factor. II. Catalytic properties. J. Biol. Chem. 270, 48904895. Shuck, S.C., Short, E.A., and Turchi, J.J. (2008). Eukaryotic nucleotide excision repair: from understanding mechanisms to inuencing biology. Cell Res. 18, 6472. Smith, A.J., and Savery, N.J. (2005). RNA polymerase mutants defective in the initiation of transcription-coupled DNA repair. Nucleic Acids Res. 33, 755764. Smith, A.J., Szczelkun, M.D., and Savery, N.J. (2007). Controlling the motor activity of a transcription-repair coupling factor: autoinhibition and the role of RNA polymerase. Nucleic Acids Res. 35, 18021811. Svejstrup, J.Q. (2002). Mechanisms of transcription-coupled DNA repair. Nat. Rev. Mol. Cell Biol. 3, 2129. Thiagalingam, S., and Grossman, L. (1991). Both ATPase sites of Escherichia coli UvrA have functional roles in nucleotide excision repair. J. Biol. Chem. 266, 1139511403. Timmins, J., Gordon, E., Caria, S., Leonard, G., Acajjaoui, S., Kuo, M.S., Monchois, V., and McSweeney, S. (2009). Structural and mutational analyses
of Deinococcus radiodurans UvrA2 provide insight into DNA binding and damage recognition by UvrAs. Structure 17, 547558. Truglio, J.J., Croteau, D.L., Skorvaga, M., DellaVecchia, M.J., Theis, K., Mandavilli, B.S., Van Houten, B., and Kisker, C. (2004). Interactions between UvrA and UvrB: the role of UvrBs domain 2 in nucleotide excision repair. EMBO J. 23, 24982509. Truglio, J.J., Croteau, D.L., Van Houten, B., and Kisker, C. (2006a). Prokaryotic nucleotide excision repair: the UvrABC system. Chem. Rev. 106, 233252. Truglio, J.J., Karakas, E., Rhau, B., Wang, H., DellaVecchia, M.J., Van Houten, B., and Kisker, C. (2006b). Structural basis for DNA recognition and processing by UvrB. Nat. Struct. Mol. Biol. 13, 360364. Verhoeven, E.E.A., Wyman, C., Moolenaar, G.F., and Goosen, N. (2002). The presence of two UvrB subunits in the UvrAB complex ensures damage detection in both DNA strands. EMBO J. 21, 41964205. Wang, J., and Grossman, L. (1993). Mutations in the helix-turn-helix motif of the Escherichia coli UvrA protein eliminate its specicity for UV-damaged DNA. J. Biol. Chem. 268, 53235331. Wellinger, R.E., and Thoma, F. (1996). Taq DNA polymerase blockage at pyrimidine dimers. Nucleic Acids Res. 24, 15781579.
Molecular Cell
Article
Structure and Biological Importance of the Spn1-Spt6 Interaction, and Its Regulatory Role in Nucleosome Binding
Seth M. McDonald,1,2 Devin Close,1,2,3 Hua Xin,1 Tim Formosa,1,* and Christopher P. Hill1,*
of Biochemistry, University of Utah School of Medicine, Salt Lake City, UT 84112-5650, USA authors contributed equally to this work 3Present address: Bioscience Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA *Correspondence: tim@biochem.utah.edu (T.F.), chris@biochem.utah.edu (C.P.H.) DOI 10.1016/j.molcel.2010.11.014
2These 1Department
SUMMARY
Eukaryotic transcription and mRNA processing depend upon the coordinated interactions of many proteins, including Spn1 and Spt6, which are conserved across eukaryotes, are essential for viability, and associate with each other in some of their biologically important contexts. Here we report crystal structures of the Spn1 core alone and in complex with the binding determinant of Spt6. Mutating interface residues greatly diminishes binding in vitro and causes strong phenotypes in vivo, including a defect in maintaining repressive chromatin. Overexpression of Spn1 partially suppresses the defects caused by an spt6 mutation affecting the Spn1 interface, indicating that the Spn1-Spt6 interaction is important for managing chromatin. Spt6 binds nucleosomes directly in vitro, and this interaction is blocked by Spn1, providing further mechanistic insight into the function of the interaction. These data thereby reveal the structural and biochemical bases of molecular interactions that function in the maintenance of chromatin structure.
INTRODUCTION Spn1 and Spt6 are transcription factors that interact with one another and are each essential for viability in yeast (Clark-Adams and Winston, 1987; Fischbeck et al., 2002). S. cerevisiae Spn1 is a 410 residue, 46 kDa protein with a central core domain (residues 140300) that is anked on both sides by regions that are predicted to be disordered (Ward et al., 2004). Spt6 is a 1451 residue, 168 kDa protein whose core (residues 3001250) likely resembles the structure of the bacterial Tex protein (Johnson et al., 2008) with an acidic N-terminal extension that is expected to be unstructured (Ward et al., 2004) and a C-terminal domain (CTD) that adopts an SH2 fold (Dengl et al., 2009; Maclennan and Shaw, 1993). Spn1 and Spt6 interact stably with one another, and they and their interaction have been implicated in
several aspects of gene expression (Krogan et al., 2002; Lindstrom et al., 2003; Yoh et al., 2007; Yoh et al., 2008). Spt6 was originally identied in a screen for factors that alter normal initiation of transcription (Clark-Adams and Winston, 1987; Denis, 1984; Neigeborn et al., 1987; Simchen et al., 1984). Subsequently, Spt6 was implicated in a variety of biological processes in organisms ranging from yeasts to humans, including embryogenesis in Zebrash (Keegan et al., 2002; Kok et al., 2007), multiple stages of development in Drosophila (Ardehali et al., 2009), gut morphogenesis in C. elegans (Nishiwaki et al., 1993), signal transduction in mammals (Baniahmad et al., 1995; Shen et al., 2009), and HIV transcription regulation and mRNA processing in human cells (Vanti et al., 2009; Yoh et al., 2007). The role in transcription initiation has been ascribed to the ability of Spt6 to chaperone histones to promote reassembly of nucleosomes in the wake of RNA polymerase II (RNAPII), thereby reestablishing the default repressive chromatin state that prevents inappropriate initiation of transcription (Adkins and Tyler, 2006; Bortvin and Winston, 1996; Cheung et al., 2008; Kaplan et al., 2003). While of profound importance, maintaining repressive chromatin appears to be just one of Spt6s roles. For example, Spt6 also promotes elongation by RNAPII (Hartzog et al., 1998; Kaplan et al., 2005; Kaplan et al., 2000; Lindstrom et al., 2003) on nucleosome-free DNA templates in vitro (Endoh et al., 2004; Hartzog et al., 1998; Keegan et al., 2002; Yoh et al., 2007), as well as on chromatin templates in vivo (Ardehali et al., 2009). Together, these data indicate that Spt6 plays a number of mechanistically distinct roles during transcription. The SPN1 gene was originally identied as a key regulator of transcription from genes that are regulated postrecruitment of RNAPII (Fischbeck et al., 2002). Spn1 was also identied as a protein that interacts with Spt6 and has been reported to bind with Spt6 in some but not all of Spt6s functional states (Lindstrom et al., 2003; Yoh et al., 2007; Zhang et al., 2008). For example, Spt6 can be coimmunopuried with three distinct Spt4/5-RNAPII complexes, whereas Spn1 is found in only two of these complexes (Lindstrom et al., 2003). The CYC1 gene of S. cerevisiae provides an example of how the Spn1-Spt6 interaction contributes to postrecruitment regulation (Zhang et al., 2008). RNAPII is constitutively bound to the CYC1 promoter, but is kept from elongating because it interacts with Spn1, which in turn inhibits the Swi/Snf nucleosome remodeling
Molecular Cell
Spn1-Spt6 Structure/Function
complex from promoting transcription. During activation, Spt6 binds to Spn1, and repression of Swi/Snf recruitment is relieved. Spn1 is also needed to achieve normal recruitment of the histone methyltransferase HYPB/Setd2 (Yoh et al., 2008) and the elongation factor TFIIS (Ling et al., 2006; Zhang et al., 2008) to RNAPII complexes traversing active genes. HYPB/ Setd2 methylates histone H3K36, which in turn recruits Rpd3type histone deacetylases to restore chromatin to the repressive hypoacetylated state and block inappropriate transcription (Yoh et al., 2008). In contrast to their antagonistic relationship in activating postrecruitment initiation, Spn1 and Spt6 each contribute toward restoration of repressive chromatin. Human Spn1/IWS1 also binds the protein arginine methyltransferase PRMT5, which methylates the elongation factor Spt5 and thereby regulates its interaction with RNAPII (Liu et al., 2007). Spn1 can additionally function through interactions with pathway-specic regulatory factors, such as the Arabidopsis steroid hormone responsive transcription factor BES1, which recruits Spn1 to the promoter and transcribed regions of activated genes (Li et al., 2010). Spn1 therefore contributes in several ways to the appropriate functioning of RNAPII. In addition to their roles in regulating transcription, Spt6 and Spn1 also collaborate to promote mRNA processing and export. Spt6 is required for proper 30 end formation by preventing premature 30 processing at upstream polyadenylation signals (Bucheli and Buratowski, 2005; Kaplan et al., 2005). Further, mammalian Spt6 can bind the Ser2-phosphorylated RNAPII CTD, enhancing recruitment of RNA processing/export factors (Yoh et al., 2007, 2008), and Drosophila Spt6 copuries with the RNA processing exosome complex (Andrulis et al., 2002). Both SPN1 and SPT6 have also been implicated in mRNA splicing in S. cerevisiae (Burckin et al., 2005), and binding of human Spn1/IWS1 to the RNA export factor REF1/Aly is important for recruitment of REF1/Aly to the body of the c-Myc gene during transcription (Yoh et al., 2007). Spt6 and Spn1 and their interaction with one another therefore play pivotal roles in dening the composition of RNAPII elongation complexes, maintaining the structure of chromatin, and modulating the production of mature mRNA transcripts. To advance mechanistic understanding of their functions, we have determined the structural basis of the Spn1-Spt6 interaction. We also demonstrate the importance of this interface in vitro and in vivo, and show that Spn1 negatively regulates binding of Spt6 to nucleosomes. RESULTS Mapping of the Spn1-Spt6 Interface Full-length S. cerevisiae Spt6 and Spn1 proteins were poorly behaved, but deletion of much of their presumably unstructured N-terminal regions (Ward et al., 2004) allowed us to observe coelution of a complex of recombinant Spn1(120410) and Spt6(2061451) by size-exclusion chromatography in sodium chloride concentrations up to 300 mM (data not shown). Spt6 (2391451) also bound Spn1, whereas Spt6(3151451) did not. Further truncations revealed that Spt6(239268) is sufcient for Spn1 binding (Figures 1A and 1B). This 30 residue segment of
Spt6 is predicted to be unstructured, and comes from a region that is N-terminal to the region expected to resemble the structure of the bacterial protein Tex (Johnson et al., 2008). Spn1 (148293) includes most of the Spn1 residues that are predicted to be structured and retained the ability to bind Spt6. The slightly larger Spn1(141305) fragment was previously shown to complement a deletion of SPN1 (Fischbeck et al., 2002), indicating that this core domain provides the major function(s) of Spn1 in vivo. Isothermal titration calorimetry (ITC) was used to measure binding afnities of Spn1(148293) for two different Spt6 constructs: Spt6(2391117), the largest Spt6 construct that remained soluble at sufcient concentrations for these experiments, and Spt6(239268), the smallest construct tried that retained full binding afnity. In both cases the binding displayed 1:1 stoichiometry and the mean binding constant (KD) was 170 nM (Figures 1C and 1D and see Table S1 available online). This indicates that the 30 residue segment of Spt6, Spt6 (239268), is sufcient to recapitulate the binding energy observed for larger Spt6 constructs. Crystal Structures of the Spn1 Core The crystal structure of Spn1(148307) was determined by the single-wavelength anomalous diffraction method using data collected to 3.0 A resolution from a selenomethioninesubstituted crystal (Figure 2A, Table 1). This unrened model was used in molecular replacement calculations with 2.15 A data from a native crystal that belonged to a different space group, and the native structure was rened to Rwork/Rfree values of 18.5%/22.4%. Residues 148295 were clearly observed in the electron density, as were four nonnative N-terminal residues that remained after TEV digestion. The 12 C-terminal residues were disordered and are not included in the nal model. Spn1(148307) forms a right-handed superhelical bundle of eight helices (named H18, Figure 2A). Surprisingly, this structure resembles the domains of the RNA processing factors Pcf11 (S. cerevisiae) (Meinhart and Cramer, 2004) and SCAF8 (human) (Becker et al., 2008) that bind the RNAPII CTD (Figure S1A). Spn1 is reported to associate with RNAPII (Zhang et al., 2008), and the structural similarity suggested that Spn1 might bind the RNAPII CTD. We have not, however, observed binding in a uorescence polarization assay between the Spn1 core and synthetic peptides, either phosphorylated or unphosphoryalated, that span more than two heptad repeats of the RNAPII CTD (data not shown). Crystal Structure of an Spn1-Spt6 Complex The structure of an Spn1(148293):Spt6(239268) complex was determined by molecular replacement and rened to Rwork/Rfree values of 18.6%/24.4% against 2.15 A resolution data (Figure 2B, Figure S2, Table 1). There are two complexes in the asymmetric unit that superimpose closely with an rmsd of $0.5 A over all 170 native ordered Ca atoms. Residues 148 292 of Spn1 are clearly visible in the electron density, as are six nonnative N-terminal residues that remain after TEV digestion. Residues 239263 of Spt6 are also clearly dened in the electron density as well as four nonnative N-terminal residues. There is no clear electron density for the ve C-terminal residues of Spt6 and the one C-terminal residue of Spn1, and these are
Molecular Cell
Figure 1. The Spn1 Core Binds 30 Residues in the Spt6 N-Terminal Region
(A) Spt6 and Spn1 domain organization and their interacting regions. Constructs assayed for binding are indicated above (Spt6) or below (Spn1) with their N- and C-terminal residues given. Constructs that showed binding are colored black, while those that did not show binding are colored red. (B) Overlays of size-exclusion chromatograms with Spn1 in black, Spt6 in orange, and binding experiments in blue. Chromatograms are scaled on the y axis to allow comparison of the elution volume proles of each protein or binding experiment. Spn1 has a 6-fold lower molar extinction coefcient, and so at equimolar concentrations, the Spn1 peak appears approximately six times smaller than the Spt6 peak (blue, middle panel). The x axis shows the 50100 ml elution volume. (C) Raw ITC data of the titration of Spn1(148293) into Spt6(2391117) (top) and the titration of Spt6 (239268) into Spn1(148293) (bottom). (D) Representative binding isotherms of the titration experiments from (C). See text and Table S1 for thermodynamic parameters.
not included in the nal model. Spn1 retains the same globular eight helix-bundle fold in the complex, showing no notable conformational changes upon binding Spt6 (rmsd $0.5 A on 135/145 pairs of Ca atoms). Binding of Spt6 to Spn1 does not mimic the binding of RNAPII CTD peptides to Pcf11/SCAF8, as it involves a face of Spn1 that is structurally distinct from the binding surface of Pcf11/SCAF8 (Figure S1B). The structure of the Spn1(148293):Spt6(239268) complex reveals an extensive interface in which the Spt6 residues drape across the structured Spn1 core domain as two helices, H1 (residues 239249) and H2 (256265), that are connected by a short extended segment. There are multiple contacts along the length of the Spt6 segment that bury a total of 1790 A2 of accessible surface area upon complex formation. Spt6 H1 and the connecting segment, in particular, include a high fraction of conserved residues and contact a conserved patch on the Spn1 surface, thereby indicating that this interface is likely to be preserved across eukaryotes (Figures 2C2E, Figures S1C and S1D). Distinctive hydrophobic and polar interactions are made throughout the length of the Spt6 peptide (Figure 3). Starting at
the Spt6 peptide N terminus, the H1 residues M245, I248, and F249 contact a hydrophobic pocket formed by Spn1 residues L256, G262, I266, I286, and W289. The C-terminal carbonyl groups of Spt6 H1 form water-mediated hydrogen bonds with Spn1 R263 guanidinium, whose extensive network of contacts also includes direct hydrogen bonds with the carboxylate of D254 in the extended region of Spt6. The aromatic side chain of Spt6 Y255, from the extended central region, makes van der Waals contacts with Spn1 P227, V264, and F267, while its hydroxyl group forms a hydrogen bond with the carboxylate of Spn1 E226. The Spt6 H2 residues W257, A258, L259, and I261 contact two hydrophobic patches formed by Spn1 P227, G231, V264, and F267. Finally, the Spt6 E262 carboxylate hydrogen bonds with Spn1 S271 and R273, and K272 forms polar contacts with the amide of Spt6 N263. These extensive interactions are consistent with the strong specic binding observed between Spn1 and Spt6. Mutations that Disrupt Spn1-Spt6 Complex Formation In Vitro To validate the relevance of the interface seen in the crystal structure, Spn1(148293) and Spt6(239268) variants were expressed and puried, and binding afnities were measured by ITC (Figure 4, Figure S3, Table S1). The Spt6-F249K protein bound Spn1 with a mean KD of 11 mM, a reduction in afnity compared to the WT interaction of about 60-fold. The afnity of the Spn1-R263D protein for Spt6 (mean KD of 30 mM) was reduced about 170-fold compared to WT. No binding was
Molecular Cell
Figure 2. Structures of the Spn1 Core and Its Complex with Spt6
(A) The structure of the Spn1 core shown as a cartoon representation in two orthogonal views. The polypeptide chain is colored as a blue to red rainbow from N to C terminus. Secondary structures and the N and C termini are labeled. (B) Structure of the Spn1-Spt6 complex. The Spn1 core is shown as a surface electrostatic (5 kT/e) representation in the same views as (A). Spt6 is colored yellow and the N and C termini and the helices are labeled. Spn1 K192 has been implicated in interactions with RNAPII (Zhang et al., 2008) and lies in the basic conserved pocket indicated. Remnants of the afnity tags that remain in the crystallized Spn1 and Spt6 proteins are not shown in any of the gures and do not contribute to the Spn1 interface. (C) Same as (B), except that the Spn1 surface and Spt6 peptide are colored according to conservation. (D) Spn1 core amino acid sequence. Secondary structural elements are indicated above, and were dened with ESPript (Gouet et al., 2003). Residues that make direct contact across the interface are indicated with a black square or with an asterisk if they were mutated in this study. Numbering and residue identities shown here refer to the S. cerevisiae protein. Coloring represents degrees of conservation, dark green (high), light green (medium), in an alignment of proteins from Saccharomyces cerevisiae, Schizosaccharomyces pombe, Caenorhabditis elegans, Drosophila melanogaster, Danio rerio, and Homo sapiens. Amino acid sequences were aligned using the T-coffee multiple sequence alignment method (Notredame et al., 2000) and slightly adjusted manually in light of the structure (Figure S2). (E) Same as (D) but for Spt6. High conservation (red), medium conservation (yellow).
detected for the Spn1-F267E protein at concentrations expected to give a reliable estimate of afnity around 40 mM. These observations validate the crystallographic interface and demonstrate that single point mutations are sufcient to signicantly reduce binding afnity in vitro. Mutating the Spn1-Spt6 Interface Causes Profound Effects In Vivo To determine the physiological importance of the Spn1-Spt6 interaction, we introduced spt6-F249K, spn1-R263D, and spn1-F267E mutations into the genomes of yeast cells such that each mutant protein was expressed from its native promoter as the sole source of the affected protein. spt6-F249K caused a moderate growth defect at low temperatures, and this was strongly enhanced at elevated temperatures (Figure 5A). Strains with the spn1-R263D mutation grew normally at all temperatures, while those with the spn1-F267E mutation were normal at low temperatures but failed to grow at high temperatures. Each of the three mutations caused a defect in transcription initiation site selection due to defective chromatin repression, as indicated by the Spt phenotype (growth of a strain with the lys2-128v allele on medium lacking lysine (Simchen et al., 1984; Figures 5A5C). The strength of the defect was different for each mutation, being weakest for spt6-F249K, stronger for
spn1-R263D, and strongest for spn1-F267E. This order correlates precisely with the level of perturbation of binding observed with these mutant proteins in vitro (Figure 4, Table S1), strongly supporting the importance of the Spn1-Spt6 binding interface detected in our crystal structure in maintaining a repressive chromatin state in vivo. Each mutation in the Spn1-Spt6 binding interface disturbed the interaction to a different extent in vitro (Figure 4, Table S1), but each individual mutation was tolerated in vivo (Figure 5). If the phenotypes caused by individual mutations result from partial disruption of binding, then combining the mutations should lead to enhanced defects. Consistent with this prediction, cells with both spt6-F249K and spn1-R263D mutations were viable but severely impaired for growth (Figure 5A), and cells with both spt6-F249K and spn1-F267E were inviable (Figure S4D). The severity of the defect caused by combining mutations therefore correlates with the level of disruption of binding by the individual mutations, suggesting that the Spn1-Spt6 interaction is essential for viability. We were unable to detect an interaction between Spt6 and Spn1-F267E in vitro (Figure 4), but the viability of the spn1-F267E strain suggests that this mutant retains some binding. Combining spt6-F249K with spn1-K192N was also lethal (Figure S4D); this mutation does not directly affect the Spn1-Spt6 interface (Figures 2B and 2C) but has been shown to decrease the
Molecular Cell
Table 1. Data Collection and Renement Statistics Spn1 Core SeMet Data Collection Beamline Space group Unit cell dimensions a, b, c (A) a, b, g ( ) Resolution (A) Wavelength (A) I/sI Completeness (%) Rsym (%) Redundancy Renement Resolution (A) Number of reections Rwork/Rfree (%) Number of protein atoms Number of solvent atoms Rmsd bond lengths (A)/angles ( ) 4/c most favored/allowed (%) Values in parentheses are for the highest-resolution shell. 29.62.15 13,847 18.5/22.4 1,216 144 0.007/1.001 99.3/100.0 27.72.15 30,096 18.6/24.4 2,903 268 0.012/1.279 99.2/100.0 SSRL 9-2 P3112 61.9, 61.9, 240.4 90, 90, 120 353.0 0.97923 (Peak) 23.2 (2.8) 90.8 (92.3) 5.2 (47.6) 4.2 (4.1) Home source P3212 61.3, 61.3, 116.05 90, 90, 120 302.15 1.54178 22 (4.1) 99.8 (98.4) 6.1 (31.9) 5.8 (3.8) Home source P21212 105.9, 68.7, 73.9 90, 90, 90 302.15 1.54178 19.7 (3.1) 100.0 (100.0) 6.4 (54.4) 5.3 (5.0) Spn1 Core Native Spn1-Spt6 Complex
interaction between Spn1 and RNAPII (Zhang et al., 2008), and our results with the recombinantly expressed protein suggest that Spn1-K192N protein is unstable (data not shown). Synthetic growth defects therefore support the importance of the Spn1Spt6 interaction, but defects outside this interface can also cause additive growth defects. Another strategy for determining the importance of the Spn1Spt6 interface is to test the effect of overexpressing one partner. If decreased afnity of Spt6-F249K protein for Spn1 is responsible for growth defects in vivo, these defects might be suppressed by increasing the level of Spn1 protein. We tested this by transforming an spt6-F249K strain with high-copy plasmids containing variants of SPN1. As shown in Figure 5C, increasing the level of normal SPN1 suppressed the temperature sensitivity and partially corrected the Spt phenotype caused by spt6-F249K. Consistent with retention of partial binding, overexpression of spn1-R263D also had an effect but suppressed less efciently, rescuing growth at 36 but not at 38 and having no effect on the Spt phenotype. Spn1-K192N is not active at elevated temperatures (Zhang et al., 2008; and Figure 5A), and overexpression of this allele also did not rescue the phenotypes caused by spt6-F249K. Elevated SPN1 copy number did not suppress the temperature sensitivity or Spt phenotypes caused by the spt6-1004 allele (data not shown), which is a deletion of the helix-hairpin-helix domain within the Tex-like core of Spt6 (Kaplan et al., 2005). The suppression of spt6 defects by increased Spn1 is therefore at least partly specic for a mutation that alters the Spn1-Spt6 interface, and supports the importance of this interaction in an essential function in vivo.
Spt6 Binds Nucleosomes Directly and Is Inhibited by Spn1 Spt6 has been shown to bind both (H3-H4)2 tetramers and H2AH2B dimers (Bortvin and Winston, 1996), leading to models in which Spt6 acts as a histone chaperone during nucleosome eviction and redeposition. Following our earlier prediction (Johnson et al., 2008) that the N-terminal region of Spt6 binds nucleosomes, we examined puried Spt6 constructs for nucleosomebinding activity in an electrophoretic mobility shift assay (EMSA). Interestingly, we found that intact Spt6 does bind nucleosomes in this assay (Figure 6), but does so only in the presence of the small HMGB family member Nhp6 (Stillman, 2010). This is similar to the requirement for addition of Nhp6 to observe complexes between nucleosomes and a different histone chaperone, FACT (Formosa et al., 2001). As with FACT, this suggests that Spt6 may form stable complexes with nucleosomes only after the nucleosome has been partially destabilized by Nhp6. This requirement appears to be physiologically relevant, because loss of Nhp6 in vivo exacerbated the growth defects caused by any of several mutations in Spt6 (Figure 5C, Figure S4A). Histone chaperones do not necessarily bind intact nucleosomes; in fact, the Asf1-(H3-H4) interaction is incompatible with histone contacts within the nucleosome (Antczak et al., 2006; English et al., 2006; Natsume et al., 2007). Binding to both free histones and to nucleosomes therefore might indicate that Spt6 makes multiple distinct contacts with nucleosomes and their components during different steps in chromatin maintenance. Spt6(2391451) bound nucleosomes while Spt6 (3151451) did not (Figure 6, compare lanes 2 and 3). Thus, the
Molecular Cell
Figure 3. Details of the Spn1-Spt6 Interface

Stereoview of the Spn1-Spt6 interface. Residues mutated in this study are indicated in bold font and underlined. Water molecules are shown as red spheres, and polar interactions are indicated by black dashes.
Figure 4. Mutations at the Spn1-Spt6 Interface Disrupt Binding In Vitro

Representative ITC binding isotherms for the indicated mutant Spt6(239268) and Spn1(148293) proteins, Spt6-F249K (diamonds), Spn1-R263D (circles), and Spn1-F267E (squares). A WT isotherm is shown for reference (triangles). See text and Table S1 for thermodynamic parameters.
region of Spt6 required for nucleosome binding (residues 239 314) contains the region required for Spn1 binding (239268), suggesting that these two interactions may be mutually exclusive. Consistent with this possibility, adding Spn1(120410) to the nucleosome-binding assay inhibited the formation of Spt6nucleosome complexes (Figure 6, compare lanes 2 and 5). Spn1 did not form stable complexes with nucleosomes itself, so the inhibition is unlikely to be caused by competition between Spn1 and Spt6 for a common binding site on nucleosomes. Spn1 did not interact with Nhp6 genetically (Figures S4B and S4C), and inhibition of Spt6-nucleosome complex formation by Spn1 could not be overcome by increasing the concentration of Nhp6 (data not shown), making it unlikely that the inhibition is caused by sequestering of Nhp6 by Spn1. Instead, Spn1 appears to prevent Spt6 from binding to nucleosomes directly by blocking the Spt6 binding domain. Supporting this interpretation, the Spn1-R263D variant with reduced afnity for Spt6 did not block formation of Spt6-nucleosome complexes efciently (Figure 6, lane 7). Further, the Spt6-F249K mutation affects a region important for both nucleosome binding and Spn1 interaction (Figures 4 and 6, compare lanes 2 and 8). Spt6-F249K protein was impaired for nucleosome binding, but the residual binding was partially resistant to the addition of Spn1 (Figure 6, lanes 8 and 9). These results show that Spt6 residue F249 contributes to both nucleosome binding and to Spn1 binding, and that Spn1 binding can block an interaction between Spt6 and nucleosomes. The Spn1-Spt6 interaction can therefore provide a switch that controls the interaction of Spt6 with nucle-
osomes, and the proper functioning of this switch is important for maintaining normal chromatin structure. DISCUSSION We have determined a crystal structure of the ordered central domain of Spn1 and shown that it binds a 30 residue segment of Spt6. We have also determined a crystal structure of an Spn1-Spt6 complex that reveals that Spt6 binds in an extended/helical conformation that drapes the Spt6 residues along one face of Spn1. We further used site-directed mutations to disrupt this interaction and demonstrated that the interaction observed in solution depends on residues located at the interface seen in the crystal structure. Binding is not accompanied by conformational changes in Spn1. In contrast, Spt6(239268) is almost certainly unstructured in isolation and becomes ordered upon binding Spn1. Indeed, the rst 300 residues of Spt6 are likely to be unstructured in isolation (Ward et al., 2004), whereas much of the remainder of the protein appears to comprise multiple recognizable structural domains that likely fold against each other to form an elongated structure (Johnson et al., 2008). The extended and inherently exible nature of the Spn1 binding sequence of Spt6 presumably explains why dramatic mutations in the interface substantially weaken but do not completely eliminate this interaction, as localized perturbations can be accommodated by conformational changes that do not propagate across the entire interface. Moreover, the $35 residues separating the Spn1-binding residues from the ordered
Molecular Cell
Figure 5. The Spn1-Spt6 Interaction Is Signicant for the Essential Activities of Each Protein In Vivo
(A) Isogenic strains from the A364a genetic background and with the relevant genotypes indicated (Supplemental Information) were grown to saturation in rich medium, then aliquots of 10-fold serial dilutions were placed on solid medium and incubated as labeled. SC is synthetic medium (complete or lacking tryptophan or lysine as noted) and YPAD is rich medium. Growth on medium lacking lysine reveals the Spt phenotype, reporting here on aberrant transcription initiation from the lys2-128v allele. The strain with the spt6-F249K allele grows slowly on SC lys at 25 , but the Spt phenotype is more robust at 30 (B). (B) WT and spt6-F249K strains were transformed with a high-copy number vector or the same vector carrying the version of SPN1 noted (Supplemental Information). Multiple transformants were tested to insure that the phenotypes detected are typical, then one clone of each was grown to saturation in synthetic medium lacking tryptophan to select for retention of the plasmids. Aliquots of 10-fold serial dilutions were placed on solid synthetic medium as in (A) and incubated as indicated. (C) As in (A). nhp6-D indicates deletion of both NHP6A and NHP6B.
Molecular Cell
Figure 6. Spt6 Binds Nucleosomes and Is Competed by Spn1

EMSA visualizing the signal from the labeled DNA incorporated into the nucleosomes (Cy5, 633 nm). All reactions included 150 fmoles of nucleosomes, 3 mM Nhp6, and 2 mM of the indicated Spt6 or Spn1 proteins. Migration positions of bound and unbound nucleosomes are indicated. Addition of Spn1 reduced the amount of Spt6-nucleosome complex from 37% of the total signal in lane 2 to 3.2% in lane 5 (9% remaining), but Spn1-R263D only reduced it to 28% (76% remaining). Spt6-F249K formed a lower amount of stable complex (18%), but Spn1 was less effective at inhibiting this (6% bound, or 36% of the original signal remaining comparing lanes 8 and 9) than it was with WT Spt6 (9% of the original signal remaining).
region of Spt6 likely provide a tether whose exibility may be required to allow the Spn1-Spt6 complex to form in multiple functional contexts. Spt6 and Spn1 were found to copurify in high-throughput screens (Gavin et al., 2002; Krogan et al., 2002), and further studies suggested that their interaction promotes normal activation of genes regulated after recruitment of RNAPII (Fischbeck et al., 2002; Yoh et al., 2007; Zhang et al., 2008). However, Spt6 and Spn1 have also been implicated in several other distinct and overlapping roles, and each has been implicated in interactions with multiple factors. We have veried a functional interplay between Spt6 and Spn1 proteins in S. cerevisiae by demonstrating that cells cannot tolerate certain partial loss-offunction alleles of both genes simultaneously, and by showing that the defects caused by the spt6-F249K allele, which impairs binding with Spn1, can be suppressed by overexpressing SPN1. Our results therefore show that a short region of Spt6 is important for interacting with Spn1 during at least one of the essential processes mediated by these proteins. One of the primary functions ascribed to Spt6 is as a histone chaperone that promotes the reassembly of nucleosomes following passage of RNAPII (Adkins and Tyler, 2006; Bortvin and Winston, 1996; Cheung et al., 2008; Kaplan et al., 2003). Spt6 was previously shown to bind histones and to promote
nucleosome deposition in vitro (Bortvin and Winston, 1996), and we now show that it can also bind to intact nucleosomes. The dependence of this interaction upon the presence of Nhp6 is consistent with our genetic studies and with an equivalent dependence of the unrelated FACT histone chaperone for its interaction with intact nucleosomes (Formosa et al., 2001). We found that nucleosome binding requires a section of Spt6 that overlaps with the Spn1 binding site, that Spn1 antagonizes Spt6-nucleosome binding in vitro, that either SPT6 or SPN1 mutations that affect the Spn1-Spt6 interaction cause the Spt phenotype, and that the effects of the spt6-F249K allele can be suppressed by overexpressing Spn1. These results show that the Spn1-Spt6 interaction disrupts Spt6-nucleosome binding and that this disruption has a positive role in maintaining normal chromatin. An attractive explanation is that Spn1 is actively engaged in the nucleosome reassembly process, possibly by disengaging Spt6 from nucleosomes to allow multiple rounds of reassembly (see the graphical abstract available online). Another possibility is that Spn1 binding inuences the balance between Spt6s functional roles in nucleosome reassembly and mRNA processing. Regardless of the precise mechanistic details, Spn1 appears to be important for Spt6-mediated nucleosome reassembly in vivo. Nucleosome reassembly is just one of several processes in which Spt6 and Spn1 have been implicated. For example, Spn1, in complex with Spt6, has been reported to interact with a variety of other factors that function in mRNA processing, nucleosome modication, or transcription, including REF1/Aly, Setd2, and RNAPII (Yoh et al., 2007, 2008; Zhang et al., 2008). Our structural data provide insight into how Spn1 might accommodate simultaneous interactions. Spt6 binds against a region of the Spn1 surface that is rich in conserved residues, but other regions of Spn1 display a similar level of conservation and are therefore good candidates for binding surfaces for other proteins (Figure 2C, Figure S1C). This includes residues that are immediately adjacent to the Spt6 binding surface but extend beyond contacts with Spt6, which may indicate that other factors can bind Spn1 cooperatively or competitively with Spt6. Finally, K192, whose mutation to asparagine impairs Spn1 function (Zhang et al., 2008) and appears to function in interactions with RNAPII, is located in a conserved pocket that is formed at the ends of H1, H2, and H4 (Figures 2B and 2C), suggesting another potential binding surface. In summary, we have determined the structural basis for the interaction between Spn1 and Spt6, and shown that this interaction is important in vivo and that it regulates Spt6-nucleosome binding in vitro. Overall, our data indicate that Spn1 is important for Spt6-mediated nucleosome reassembly, perhaps by regulating the process or by providing a switch that drives disengagement. This does not, however, seem to encompass all of the functional roles in which these two proteins and their interaction with each other participate (Lindstrom et al., 2003; Yoh et al., 2007, 2008; Zhang et al., 2008). Sequence conservation indicates that other surfaces on the Spn1 core domain are good candidates for mediating functionally important interactions. Moreover, interactions with Setd2 and REF1/Aly have been mapped to the N- and C-terminal regions, respectively, which extend beyond the Spn1 core domain and are predicted to be
Molecular Cell
unstructured (Ward et al., 2004; Yoh et al., 2007, 2008). This use of inherently exible segments, including the N-terminal region of Spt6 that binds the Spn1 core, may provide a mechanism that allows exibility in a crowded transcriptional environment.
EXPERIMENTAL PROCEDURES See the Supplemental Information for protein expression and purication, strains used, strain construction, plasmid construction, and tests of the effects of combining spn1 and spt6 mutations. Crystallization and Structure Determination All crystals were grown at 20 C by sitting drop vapor diffusion. Se-Spn1 (148307) drops comprised 2 ml of 8 mg.mL1 protein with 2 ml of well solution (0.1 M Bis-Tris propane [pH 7.0], 1.4 M Li2SO4). Native Spn1(148307) drops comprised two parts 7.5 mg.mL1 protein and one part well solution (0.01 M MgCl2, 0.05 M HEPES [pH 7.0], 1.6 M [NH4]2SO4). Spn1(148293)-Spt6(239 268) drops comprised two parts 13 mg.mL1 protein and one part well solution (0.2 M Mg[CH3CO2]2, 0.1 M MES [pH 6.5], 20% PEG 8000). Crystals were cryoprotected in a solution of the reservoir made up with 30% glycerol, and cooled by plunging into liquid nitrogen. All data were processed using HKL2000 (Otwinowski and Minor, 1997). Phases were determined for Se-Spn1(148307) by single-wavelength anomalous diffraction. Phenix (phenix.autosol) (Adams et al., 2010) located four out of six possible selenium positions and computed a map into which a model was built. This unrened model was used in molecular replacement using PHASER (McCoy et al., 2005) to determine the structure of native Spn1(148307) at 2.15 A resolution. This subsequently rened model was used to determine the Spn1(148293):Spt6(239268) structure by molecular replacement. In all cases, model building, renement,and validation were performed using Coot (Emsley et al., 2010), Phenix (Adams et al., 2010), and MolProbity (Chen et al., 2010), respectively. Renement of the Spn1(148293)-Spt6(239268) complex included TLS restraints (Painter and Merritt, 2006). Electrostatic potential surfaces were calculated using APBS (Baker et al., 2001). Figures of molecular structures were generated using PyMol (DeLano, 2002). Size-Exclusion Chromatography Binding Assay Puried recombinant proteins were mixed at equimolar concentrations (2 mM) and incubated for 2 hr at 4 C. The protein mixture was concentrated to 15 mM and chromatographed on a 120 ml Superdex 200 16/60 column (GE Healthcare) in 15 mM Tris (pH 7.5), 200 mM NaCl, 5% glycerol, 0.5 mM EDTA, and 2 mM 2-mercaptoethanol. Isothermal Titration Calorimetry Amino acid substitutions were made by site-directed mutagenesis and veried by DNA sequencing. Puried recombinant proteins were dialyzed overnight at 4 C against 2 L of degassed ITC buffer (20 mM Tris [pH 7.5], 150 mM NaCl, 5% glycerol, 2 mM 2-mercaptoethanol, 0.5 mM EDTA). Titrations for all reactions were done at 25 C on an iTC200 (Microcal), including an initial injection of 0.4 ml (which was omitted from data analysis), and all injections were spaced 180 s apart. For Spt6(2391117) reactions, the titrations were carried out with 18 injections of 1.8 ml 76 mM Spn1(148293) into 8.2 mM Spt6(239 1117). For Spt6(239268) reactions, the titrations were with 18 injections of 2 ml 74 mM Spt6(239268) into 8 mM Spn1(148293). For Spt6-F249K reactions, the titrations were with 18 injections of 1.8 ml 1.52 mM Spt6(239268)-F249K into 168 mM Spn1(148293). For Spn1-R263D reactions, the titrations were with 18 injections of 1.8 ml 3.5 mM Spt6(239268) into 389 mM Spn1(148 293)-R263D. For Spn1-F267E reactions, the titrations were with 18 injections of 2.0 ml 5.2 mM Spt6(239268) into 578 mM Spn1(148293)-F267E). In all cases, three independent experiments were performed. Data were analyzed using Origin software (Microcal), and the stoichiometry (N), association constant (KA), and change in enthalpy (DH) were obtained by tting the isotherm to the one-site binding model. Other thermodynamic parameters were calculated using the following relationships:
K1 = KD and RTlnKA = DH TDS: A
Nucleosome Preparation and Gel Mobility Shift Binding Assay A 146 bp sea urchin 5S rDNA fragment labeled with Cy5 was generated by PCR and gel puried. Xenopus laevis histone H2A-S113C was labeled with Oregon Green 488-maleimide and then assembled into nucleosome core particles as described (Xin et al., 2009). Reactions contained 15 nM nucleosomes, 2 mM Spt6 or Spn1 proteins, 100 mM NaCl, 0.8 mg/ml HSA, 9.7% glycerol, and 3 mM S. cerevisiae Nhp6a. Following incubation at 30 C for 15 min, samples were subjected to electrophoresis on native polyacrylamide gels (4.5% acrylamide [acr:bis, 37.5:1], 0.5 X TBE, 5% glycerol, 2 mM MgCl2 at 160 V for 6 hr at 4 C). The gels were scanned using a Typhoon imager at 670 BP30/Red(633 nm) for Cy5-DNA and 520 BP40/Blue(488 nm) for Oregon Green 488-H2A, and the amount of signal in the bound form quantied with ImageQuant Software (GE Health Sciences). ACCESSION NUMBERS Coordinates and structure factor amplitudes have been deposited for the Spn1 and Spn1-Spt6 crystal structures have been deposited in the Protein Data Bank under ID codes 3o8z and 3oak, respectively. SUPPLEMENTAL INFORMATION Supplemental Information includes one table, four gures, Supplemental Experimental Procedures, and Supplemental References and can be found with this article online at doi:10.1016/j.molcel.2010.11.014. ACKNOWLEDGMENTS Portions of this research were carried out at the Stanford Synchrotron Radiation Laboratory, a national user facility operated by Stanford University on behalf of the U.S. Department of Energy, Ofce of Basic Energy Sciences. The SSRL Structural Molecular Biology Program is supported by the Department of Energy, Ofce of Biological and Environmental Research, and by the U.S. National Institutes of Health (NIH), National Center for Research Resources, Biomedical Technology Program, and the National Institute of General Medical Sciences. We thank David Stillman and Warren Voth for helpful discussions and Charisse Kettelkamp and Laura McCullough for technical assistance. This work was supported by NIH grants to C.P.H. and T.F. Received: May 26, 2010 Revised: August 11, 2010 Accepted: September 21, 2010 Published online: November 18, 2010 REFERENCES Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010). PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213221. Adkins, M.W., and Tyler, J.K. (2006). Transcriptional activators are dispensable for transcription in the absence of Spt6-mediated chromatin reassembly of promoter regions. Mol. Cell 21, 405416. Andrulis, E.D., Werner, J., Nazarian, A., Erdjument-Bromage, H., Tempst, P., and Lis, J.T. (2002). The RNA processing exosome is linked to elongating RNA polymerase II in Drosophila. Nature 420, 837841. Antczak, A.J., Tsubota, T., Kaufman, P.D., and Berger, J.M. (2006). Structure of the yeast histone H3-ASF1 interaction: implications for chaperone mechanism, species-specic interactions, and epigenetics. BMC Struct. Biol. 6, 26. Ardehali, M.B., Yao, J., Adelman, K., Fuda, N.J., Petesch, S.J., Webb, W.W., and Lis, J.T. (2009). Spt6 enhances the elongation rate of RNA polymerase II in vivo. EMBO J. 28, 10671077.
Molecular Cell
Baker, N.A., Sept, D., Joseph, S., Holst, M.J., and McCammon, J.A. (2001). Electrostatics of nanosystems: application to microtubules and the ribosome. Proc. Natl. Acad. Sci. USA 98, 1003710041. Baniahmad, C., Nawaz, Z., Baniahmad, A., Gleeson, M.A., Tsai, M.J., and OMalley, B.W. (1995). Enhancement of human estrogen receptor activity by SPT6: a potential coactivator. Mol. Endocrinol. 9, 3443. Becker, R., Loll, B., and Meinhart, A. (2008). Snapshots of the RNA processing factor SCAF8 bound to different phosphorylated forms of the carboxylterminal domain of RNA polymerase II. J. Biol. Chem. 283, 2265922669. Bortvin, A., and Winston, F. (1996). Evidence that Spt6p controls chromatin structure by a direct interaction with histones. Science 272, 14731476. Bucheli, M.E., and Buratowski, S. (2005). Npl3 is an antagonist of mRNA 30 end formation by RNA polymerase II. EMBO J. 24, 21502160. Burckin, T., Nagel, R., Mandel-Gutfreund, Y., Shiue, L., Clark, T.A., Chong, J.L., Chang, T.H., Squazzo, S., Hartzog, G., and Ares, M., Jr. (2005). Exploring functional relationships between components of the gene expression machinery. Nat. Struct. Mol. Biol. 12, 175182. Chen, V.B., Arendall, W.B., 3rd, Headd, J.J., Keedy, D.A., Immormino, R.M., Kapral, G.J., Murray, L.W., Richardson, J.S., and Richardson, D.C. (2010). MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr. 66, 1221. Cheung, V., Chua, G., Batada, N.N., Landry, C.R., Michnick, S.W., Hughes, T.R., and Winston, F. (2008). Chromatin- and transcription-related factors repress transcription from within coding regions throughout the Saccharomyces cerevisiae genome. PLoS Biol. 6, e277. 10.1371/journal. pbio.0060277. Clark-Adams, C.D., and Winston, F. (1987). The SPT6 gene is essential for growth and is required for delta-mediated transcription in Saccharomyces cerevisiae. Mol. Cell. Biol. 7, 679686. DeLano, W.L. (2002). The Pymol Molecular Graphics System (Palo Alto, CA: Delano Scientic). Dengl, S., Mayer, A., Sun, M., and Cramer, P. (2009). Structure and in vivo requirement of the yeast Spt6 SH2 domain. J. Mol. Biol. 389, 211225. Denis, C.L. (1984). Identication of new genes involved in the regulation of yeast alcohol dehydrogenase II. Genetics 108, 833844. Emsley, P., Lohkamp, B., Scott, W.G., and Cowtan, K. (2010). Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486501. Endoh, M., Zhu, W., Hasegawa, J., Watanabe, H., Kim, D.K., Aida, M., Inukai, N., Narita, T., Yamada, T., Furuya, A., et al. (2004). Human Spt6 stimulates transcription elongation by RNA polymerase II in vitro. Mol. Cell. Biol. 24, 33243336. English, C.M., Adkins, M.W., Carson, J.J., Churchill, M.E., and Tyler, J.K. (2006). Structural basis for the histone chaperone activity of Asf1. Cell 127, 495508. Fischbeck, J.A., Kraemer, S.M., and Stargell, L.A. (2002). SPN1, a conserved gene identied by suppression of a postrecruitment-defective yeast TATAbinding protein mutant. Genetics 162, 16051616. Formosa, T., Eriksson, P., Wittmeyer, J., Ginn, J., Yu, Y., and Stillman, D.J. (2001). Spt16-Pob3 and the HMG protein Nhp6 combine to form the nucleosome-binding factor SPN. EMBO J. 20, 35063517. Gavin, A.C., Bosche, M., Krause, R., Grandi, P., Marzioch, M., Bauer, A., Schultz, J., Rick, J.M., Michon, A.M., Cruciat, C.M., et al. (2002). Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 415, 141147. Gouet, P., Robert, X., and Courcelle, E. (2003). ESPript/ENDscript: extracting and rendering sequence and 3D information from atomic structures of proteins. Nucleic Acids Res. 31, 33203323. Hartzog, G.A., Wada, T., Handa, H., and Winston, F. (1998). Evidence that Spt4, Spt5, and Spt6 control transcription elongation by RNA polymerase II in Saccharomyces cerevisiae. Genes Dev. 12, 357369.
Johnson, S.J., Close, D., Robinson, H., Vallet-Gely, I., Dove, S.L., and Hill, C.P. (2008). Crystal structure and RNA binding of the Tex protein from Pseudomonas aeruginosa. J. Mol. Biol. 377, 14601473. Kaplan, C.D., Morris, J.R., Wu, C., and Winston, F. (2000). Spt5 and spt6 are associated with active transcription and have characteristics of general elongation factors in D. melanogaster. Genes Dev. 14, 26232634. Kaplan, C.D., Laprade, L., and Winston, F. (2003). Transcription elongation factors repress transcription initiation from cryptic sites. Science 301, 1096 1099. Kaplan, C.D., Holland, M.J., and Winston, F. (2005). Interaction between transcription elongation factors and mRNA 30 -end formation at the Saccharomyces cerevisiae GAL10-GAL7 locus. J. Biol. Chem. 280, 913922. Keegan, B.R., Feldman, J.L., Lee, D.H., Koos, D.S., Ho, R.K., Stainier, D.Y., and Yelon, D. (2002). The elongation factors Pandora/Spt6 and Foggy/Spt5 promote transcription in the zebrash embryo. Development 129, 16231632. Kok, F.O., Oster, E., Mentzer, L., Hsieh, J.C., Henry, C.A., and Sirotkin, H.I. (2007). The role of the SPT6 chromatin remodeling factor in zebrash embryogenesis. Dev. Biol. 307, 214226. Krogan, N.J., Kim, M., Ahn, S.H., Zhong, G., Kobor, M.S., Cagney, G., Emili, A., Shilatifard, A., Buratowski, S., and Greenblatt, J.F. (2002). RNA polymerase II elongation factors of Saccharomyces cerevisiae: a targeted proteomics approach. Mol. Cell. Biol. 22, 69796992. Li, L., Ye, H., Guo, H., and Yin, Y. (2010). Arabidopsis IWS1 interacts with transcription factor BES1 and is involved in plant steroid hormone brassinosteroid regulated gene expression. Proc. Natl. Acad. Sci. USA 107, 39183923. Lindstrom, D.L., Squazzo, S.L., Muster, N., Burckin, T.A., Wachter, K.C., Emigh, C.A., McCleery, J.A., Yates, J.R., 3rd, and Hartzog, G.A. (2003). Dual roles for Spt5 in pre-mRNA processing and transcription elongation revealed by identication of Spt5-associated proteins. Mol. Cell. Biol. 23, 13681378. Ling, Y., Smith, A.J., and Morgan, G.T. (2006). A sequence motif conserved in diverse nuclear proteins identies a protein interaction domain utilised for nuclear targeting by human TFIIS. Nucleic Acids Res. 34, 22192229. Liu, Z., Zhou, Z., Chen, G., and Bao, S. (2007). A putative transcriptional elongation factor hIws1 is essential for mammalian cell proliferation. Biochem. Biophys. Res. Commun. 353, 4753. Maclennan, A.J., and Shaw, G. (1993). A yeast SH2 domain. Trends Biochem. Sci. 18, 464465. McCoy, A.J., Grosse-Kunstleve, R.W., Storoni, L.C., and Read, R.J. (2005). Likelihood-enhanced fast translation functions. Acta Crystallogr. D Biol. Crystallogr. 61, 458464. Meinhart, A., and Cramer, P. (2004). Recognition of RNA polymerase II carboxy-terminal domain by 30 -RNA-processing factors. Nature 430, 223226. Natsume, R., Eitoku, M., Akai, Y., Sano, N., Horikoshi, M., and Senda, T. (2007). Structure and function of the histone chaperone CIA/ASF1 complexed with histones H3 and H4. Nature 446, 338341. Neigeborn, L., Celenza, J.L., and Carlson, M. (1987). SSN20 is an essential gene with mutant alleles that suppress defects in SUC2 transcription in Saccharomyces cerevisiae. Mol. Cell. Biol. 7, 672678. Nishiwaki, K., Sano, T., and Miwa, J. (1993). emb-5, a gene required for the correct timing of gut precursor cell division during gastrulation in Caenorhabditis elegans, encodes a protein similar to the yeast nuclear protein SPT6. Mol. Gen. Genet. 239, 313322. Notredame, C., Higgins, D.G., and Heringa, J. (2000). T-Coffee: a novel method for fast and accurate multiple sequence alignment. J. Mol. Biol. 302, 205217. Otwinowski, Z., and Minor, W., eds. (1997). Processing of X-ray Diffraction Data Collected in Oscillation Mode (New York: Academic Press). Painter, J., and Merritt, E.A. (2006). Optimal description of a protein structure in terms of multiple groups undergoing TLS motion. Acta Crystallogr. D Biol. Crystallogr. 62, 439450. Shen, X., Xi, G., Radhakrishnan, Y., and Clemmons, D.R. (2009). Identication of novel SHPS-1-associated proteins and their roles in regulation of insulin-like
Molecular Cell
growth factor-dependent responses in vascular smooth muscle cells. Mol. Cell. Proteomics 8, 15391551. Simchen, G., Winston, F., Styles, C.A., and Fink, G.R. (1984). Ty-mediated gene expression of the LYS2 and HIS4 genes of Saccharomyces cerevisiae is controlled by the same SPT genes. Proc. Natl. Acad. Sci. USA 81, 2431 2434. Stillman, D.J. (2010). Nhp6: a small but powerful effector of chromatin structure in Saccharomyces cerevisiae. Biochim. Biophys. Acta 1799, 175180. Vanti, M., Gallastegui, E., Respaldiza, I., Rodriguez-Gil, A., Gomez-Herreros, F., Jimeno-Gonzalez, S., Jordan, A., and Chavez, S. (2009). Yeast genetic analysis reveals the involvement of chromatin reassembly factors in repressing HIV-1 basal transcription. PLoS Genet. 5, e1000339. 10.1371/journal.pgen. 1000339. Ward, J.J., Sodhi, J.S., McGufn, L.J., Buxton, B.F., and Jones, D.T. (2004). Prediction and functional analysis of native disorder in proteins from the three kingdoms of life. J. Mol. Biol. 337, 635645.
Xin, H., Takahata, S., Blanksma, M., McCullough, L., Stillman, D.J., and Formosa, T. (2009). yFACT induces global accessibility of nucleosomal DNA without H2A-H2B displacement. Mol. Cell 35, 365376. Yoh, S.M., Cho, H., Pickle, L., Evans, R.M., and Jones, K.A. (2007). The Spt6 SH2 domain binds Ser2-P RNAPII to direct Iws1-dependent mRNA splicing and export. Genes Dev. 21, 160174. Yoh, S.M., Lucas, J.S., and Jones, K.A. (2008). The Iws1:Spt6:CTD complex controls cotranscriptional mRNA biosynthesis and HYPB/Setd2-mediated histone H3K36 methylation. Genes Dev. 22, 34223434. Zhang, L., Fletcher, A.G., Cheung, V., Winston, F., and Stargell, L.A. (2008). Spn1 regulates the recruitment of Spt6 and the Swi/Snf complex during transcriptional activation by RNA polymerase II. Mol. Cell. Biol. 28, 13931403. Note Added in Proof An independent study (M.-L. Diebold, M. Koch, E. Loeliger, V. Cura, F. Winston, J. Cavarelli, and C. Romier, EMBO J, 10.1038/emboj.2010.272) has also determined a very similar structure of a Spn1-Spt6 complex; the results of the two studies are highly complementary.
Article
Conserved Antagonism between JMJD2A/KDM4A and HP1g during Cell Cycle Progression
Molecular Cell
Joshua C. Black,1 Andrew Allen,1,5 Capucine Van Rechem,1,5 Emily Forbes,1,5 Michelle Longworth,1 Katrin Tschop,1 Claire Rinehart,2 Jonathan Quiton,3 Ryan Walsh,1 Andrea Smallwood,4 Nicholas J. Dyson,1 and Johnathan R. Whetstine1,*
1Massachusetts General Hospital Cancer Center and Department of Medicine, Harvard Medical School, 13th Street, Charlestown, MA 02129, USA 2Department of Biology 3Department of Mathematics and Computer Science Western Kentucky University, 1906 College Heights Boulevard, Bowling Green, KY 42101-1078, USA 4Ludwig Institute for Cancer Research UCSD, 9500 Gilman Drive, La Jolla, CA 92093, USA 5These authors contributed equally to this work *Correspondence: jwhetstine@hms.harvard.edu DOI 10.1016/j.molcel.2010.11.008
SUMMARY
The KDM4/JMJD2 family of histone demethylases is amplied in human cancers. However, little is known about their physiologic or tumorigenic roles. We have identied a conserved and unappreciated role for the JMJD2A/KDM4A H3K9/36 tridemethylase in cell cycle progression. We demonstrate that JMJD2A protein levels are regulated in a cell cycle-dependent manner and that JMJD2A overexpression increased chromatin accessibility, S phase progression, and altered replication timing of specic genomic loci. These phenotypes depended on JMJD2A enzymatic activity. Strikingly, depletion of the only C. elegans homolog, JMJD-2, slowed DNA replication and increased ATR/p53-dependent apoptosis. Importantly, overexpression of HP1g antagonized JMJD2Adependent progression through S phase, and depletion of HPL-2 rescued the DNA replication-related phenotypes in jmjd-2/ animals. Our ndings describe a highly conserved model whereby JMJD2A regulates DNA replication by antagonizing HP1g and controlling chromatin accessibility.
INTRODUCTION In eukaryotes, DNA is packed into a highly ordered chromatin structure composed of DNA, histones, and other chromosomal proteins. Posttranslational modication (PTM) of histones plays an instrumental role in integrating extra- and intracellular signals to regulate DNA-dependent processes. The dynamics associated with histone PTMs are critical determinants of stem cell fate and maintenance, tissue patterning, and the inhibition of tumor growth (Bernstein et al., 2007; Fraga and Esteller, 2005; Moss and Wallrath, 2007). Multiple lysine (K) residues on the tails of histone H3 and H4 have been shown to be sites for methylation. The site and degree of methylation (mono-, di-, or tri-) are
critical for heterochromatin formation and maintenance, replication timing, and gene expression (Chen et al., 2008; Grewal and Jia, 2007; Maison et al., 2002; Wu et al., 2005). Moreover, aberrant methylation can result in human diseases such as cancer (Fraga and Esteller, 2005; Moss and Wallrath, 2007). The FAD-dependent amine oxidases and the JmjC-containing histone demethylases regulate methylation balance with specicity for particular lysines, and for the degree of methylation (Cloos et al., 2008; Klose et al., 2006; Lan et al., 2008; Whetstine et al., 2006). For example, the JMJD2/KDM4 family demethylates trimethylated H3K9, H3K36, and H1.4K26 (Fodor et al., 2006; Klose et al., 2006; Trojer et al., 2009; Whetstine et al., 2006). There are four JMJD2 family members in humans, JMJD2A-D, and one in C. elegans, JMJD-2 (Whetstine et al., 2006). We previously demonstrated that the depletion of JMJD-2 increased DNA double-strand breaks (DSBs) in the adult germline, which resulted in p53-dependent apoptosis (Whetstine et al., 2006). These observations underscore the importance of balancing methylation levels so that DNA damage and p53dependent apoptosis are avoided. At least part of the function of histone methylation is mediated through recruitment of Heterochromatin Protein 1 (HP1) family members. There are three mammalian HP1 proteins, HP1a, HP1b, and HP1g, which are recruited to H3K9me3 and H1.4K26me3 (Bannister et al., 2001; Daujat et al., 2005; Lachner et al., 2001; Trojer et al., 2009). While HP1a and HP1b are generally localized to heterochromatin, HP1g is present in both heterochromatic and euchromatic domains (Kwon and Workman, 2008; Vakoc et al., 2005). Emerging evidence suggests HP1 proteins are important regulators of DNA replication, which is exemplied through regulation of pericentric heterochromatin. Pericentric heterochromatin is the region that surrounds the centromere and is highly condensed and H3K9me3 enriched (Chen et al., 2008; Grewal and Jia, 2007; Weidtkamp-Peters et al., 2006; Wu et al., 2005). Deletions of the yeast H3K9me3 methyltransferase (CLR4) and SWI6 (yeast HP1 homolog) result in altered S phase progression and replication stress (Biswas et al., 2008; Grewal and Jia, 2007; Kim et al., 2008). Additionally, SWI6 localizes to pericentric regions and heterochromatin foci in a Chromatin Assembly
Molecular Cell
JMJD2A/JMJD-2 Antagonize HP1 during Cell Cycle
Factor-1 (CAF-1, a chaperone important for histone assembly during replication)-dependent manner (Dohke et al., 2008). SWI6 is also required for proper replication timing of pericentric heterochromatin in ssion yeast (Hayashi et al., 2009). Similarly, in mammalian cells, HP1a, HP1b, and H3K9me3 are important for condensing pericentric regions to maintain transcriptional repression and to prevent this region from being aberrantly replicated (Bannister et al., 2001; Czvitkovich et al., 2001; Nakayama et al., 2001; Quivy et al., 2008). Mammalian HP1a and HP1g also directly interact with components of the replication fork, including CAF-1 (Huang et al., 1998; Quivy et al., 2008). The HP1a and CAF-1 association is critical for maintaining the late replication of pericentric heterochromatin during S phase (Quivy et al., 2008). Interestingly, no role has been ascribed to HP1g in DNA replication. Heterochromatin and H3K9me3-enriched regions tend to replicate late during S phase (e.g., satellites), which has been conrmed by genome-wide and locus-specic analyses (Mendez, 2009). Additional studies have further demonstrated that chromatin accessibility is a critical determinant of DNA replication timing (Hansen et al., 2010; Karnani et al., 2010; MacAlpine et al., 2004, 2010). For example, active gene expression and DNase I hypersensitivity are associated with earlier DNA replication. However, active transcription is not required to replicate earlier as altering histone acetylation at specic loci changes replication timing without altering transcription (Cimbora et al., 2000; Schubeler et al., 2000). These observations strongly suggest that an open chromatin conguration is a strong indicator of whether a region will replicate in early or late S phase. In this study, we demonstrate that the histone tridemethylase JMJD2A regulates cell cycle progression by increasing chromatin accessibility and antagonizing HP1g occupancy. Overexpression of catalytically active JMJD2A resulted in faster S phase progression, corresponding to a more open chromatin state, increased single-strand DNA (ssDNA) generation, increased BrdU incorporation, and loss of HP1g at target loci. Strikingly, HP1g overexpression, but not HP1a or HP1b, abrogated JMJD2A-dependent replication changes at both a cellular and locus-specic level. Antagonism by HP1g was dependent on both the chromodomain and the chromo shadow domain. Consistent with our observations in human cell culture, loss of JMJD-2 in C. elegans results in increased DNA damage, slower DNA replication, and ATR/p53-dependent apoptosis. Furthermore, the loss of the C. elegans HP1 homolog HPL-2 rescued the increased DNA damage, the impaired DNA replication, and the ATR/p53-dependent apoptosis observed upon JMJD-2 depletion. Taken together, the data in both human cell culture and C. elegans demonstrate a highly conserved and unappreciated role for JMJD2A/JMJD-2 in DNA replication and S phase progression by altering chromatin accessibility and antagonizing HP1g/HPL-2. RESULTS JMJD2A Protein Levels Are Regulated over Cell Cycle To determine if JMJD2A/KDM4A was regulated throughout the cell cycle, we immunoblotted whole-cell extracts from asynchronous cells, cells synchronized at G1/S and S with hydroxyurea
(HU) arrest and release and cells in G2/M (nocodazole arrest) (Figure 1A). Flow cytometry conrmed the cell cycle proles (see Figures S1A and S1B available online). JMJD2A peaked during G1/S with a decrease during S phase reaching the lowest levels during G2/M (Figure 1A). JMJD2A transcript levels remained comparable over the cell cycle (Figures S1C and S1D), which suggests JMJD2A protein levels are posttranscriptionally regulated. JMJD2A expression was similar following nocodazole release (peaking after 67 hr, G1/S), demonstrating that HUrelated DNA damage did not result in the JMJD2A protein changes (Figure 1B). JMJD2A Overexpression Alters Cell Cycle Progression We ectopically expressed JMJD2A to determine its effect on cell cycle progression (Figure 1B). We created stable HEK293T cells expressing control, JMJD2A, or catalytically inactive JMJD2AH188A (Whetstine et al., 2006). These cell lines exhibited modest 2- to 3-fold overexpression of both JMJD2A and JMJD2AH188A (Figure 1B and Figure S1E) that did not result in decreased global levels of H1.4K26me3 or H3K9/36me3 (Figure S1F). Interestingly, ectopic expression of JMJD2A resulted in stabilization of both endogenous and exogenous JMJD2A throughout the cell cycle (Figure 1B). JMJD2A overexpression, but not JMJD2AH188A, resulted in faster S phase progression (Figure 1C; note red arrow at 10 and 12 hr postnocodazole). A graphical representation of the average percentage of cells reaching late S phase is shown in Figure 1D. To gain insight into JMJD2A-dependent faster S phase progression, we analyzed the amount of ssDNA generated following release from nocodazole, which indicates if additional replication forks are present (Jasencakova et al., 2010). At 6 and 10 hr postnocodazole release, we observed an increase in ssDNA in JMJD2A but not JMJD2AH188A overexpressing cells (Figures 1E and 1F). These data are consistent with an increase in replication forks in JMJD2A cells, which promote faster S phase progression, at least in part, by increasing replication initiation. We further tested the relationship between JMJD2A and DNA replication by conducting BrdU immunoprecipitations of specic genomic loci in asynchronous cells. JMJD2A overexpression, but not JMJD2AH188A, resulted in increased BrdU incorporation at chromosome 1 satellite 2 (Chr1 sat2), a region enriched for H3K9me3 and HP1g (Hansen et al., 2010; Zeng et al., 2009) (Figure 1G, p = 0.0218). Similar results were obtained with a region near the X centromere (designated X cen, Figure S1G). In contrast, an ALU repeat on chromosome 19 with minimal H3K9me3 was not regulated by JMJD2A (Figure S1H) (Zeng et al., 2009). Since JMJD2A-overexpressing cells progressed through S phase faster, had increased replication forks, and increased BrdU incorporation at Chr1 sat2, we reasoned that JMJD2A could be impacting the timing of replication of specic regions. To determine if JMJD2A altered replication timing, we sorted live cells labeled with BrdU by ow cytometry, collected multiple S phase fractions (Figure S1I), and conducted BrdU IPs. As predicted, ectopic expression of JMJD2A resulted in earlier replication of Chr1 sat2 (Figure 1H) without impacting the replication timing of the b-actin locus (Figure S1I).
Molecular Cell
Figure 1. Overexpression of JMJD2A Allows Faster Progression through S Phase

(A) JMJD2A protein levels are regulated during cell cycle in whole-cell extracts from HU-arrested cells. (B) JMJD2A protein levels are regulated during cell cycle in whole-cell extracts from nocodazole release. (C) JMJD2A overexpression results in faster S phase progression using ow cytometry. The red arrow highlights the faster accumulation in late S phase. (D) Graphical summary of the average percentage of control, JMJD2A, or catalytically inactive JMJD2AH188A cells in late S phase. (E) JMJD2A cells have increased ssDNA (indicative of replication forks) 6 and 10 hr post nocodazole release. (F) Quantitation of (E). Error bars represent the SEM of biological replicates analyzed in duplicate.
Molecular Cell
Figure 2. JMJD2A/JMJD-2 Expression Levels Alter Sensitivity to the Replication Inhibitor HU

(A) JMJD2A or JMJD2AH188A cells do not have different growth rates compared to control cells. (B) JMJD2A overexpression increases recovery from HU, while overexpression of catalytically inactive JMJD2AH188A increases cell sensitivity to HU. (C) Depletion of JMJD2A by two different shRNAs does not alter cell growth rates. (D) Depletion of JMJD2A by two different shRNAs increases cellular sensitivity to HU. For (A)(D), error bars represent the SEM of nine points from at least three biological replicates. (E and F) When compared to wild-type N2 worms, jmjd-2/ and jmjd-2 (RNAi)-treated worms show an increased embryonic lethality in response to HU. These experiments were conducted at least three independent times with at least 15 worms per genotype per experiment. Error bars represent the SEM.
JMJD2A Levels Impact the Cellular Response to Replication Stress Factors involved in DNA replication often modulate response to replication stress (e.g., sensitivity to HU; Biswas et al., 2008; Kim et al., 2008). Therefore, we tested whether alterations in JMJD2A protein levels would affect the cellular response to HU. We treated control, JMJD2A-expressing, or JMJD2AH188A-expressing cells with or without 1 mM HU for 24 hr before being released into fresh media (Figures 2A and 2B). Even though ectopic expression of JMJD2A leads to faster S phase progression, JMJD2A-overexpressing cells have similar growth curves when compared to control cells,
suggesting that overexpressing cells have a delay in another part of the cell cycle (e.g., G2/M) (Figure 2A). However, ectopic expression of JMJD2A resulted in faster recovery from HU treatment (Figure 2B). Conversely, overexpression of catalytically inactive JMJD2AH188A resulted in increased sensitivity to HU (Figure 2B). JMJD2A depletion with two different shRNA constructs also resulted in increased HU sensitivity with no change to standard growth (Figures 2C and 2D; expression levels shown in Figures S2A and S2B). Since JMJD2A depletion mimics the increased HU sensitivity of JMJD2AH188A cells, we hypothesize that catalytically inactive JMJD2A acts as a dominant negative. JMJD2 Proteins Impact DNA Replication in C. elegans Since JMJD2A altered cell cycle progression, DNA replication, and sensitivity to DNA replication stress in human cells, we determined if these processes were regulated in vivo by evaluating the jmjd-2(tm2966) loss-of-function allele in C. elegans (referred to as jmjd-2/). This allele has a 672 base pair deletion that is followed by an 18 base pair insertion, resulting in a frameshift that truncates the protein 25 amino acids into the JmjC domain, which eliminates enzymatic activity (Figure S3A).
(G) JMJD2A overexpression drives replication at chromosome 1 satellite 2 as measured using BrdU DNA IP. Error bars represent the SEM of four independent experiments. JMJD2A cells were statistically signicantly different from control (p = 0.0218) while JMJD2AH188A were not (p = 0.4009) using a two-tailed Students t test. (H) Overexpression of JMJD2A leads to earlier replication of Chr1 sat2 (compare control to JMJD2A cells: for early S, p = 0.004; for mid-S, p = 0.005). The graph shows the fraction of Chr1 sat2 replicated in each fraction compared to total across all fractions (BrdU incorporation of each fraction compared to input). Error bars represent the SEM.
Molecular Cell
We determined whether jmjd-2/ or jmjd-2 (RNAi)-treated worms had increased sensitivity to replication stress like the JMJD2A-depleted cells. As expected, both the jmjd-2/ and the jmjd-2 (RNAi)-treated worms were more sensitive than wild-type worms to a dose range of HU (Figures 2E and 2F). Similar to the jmjd-2 (RNAi), jmjd-2/ worms also had increased H3K9me3 and H3K36me3 levels and increased meiotic DSBs as determined by the number of RAD-51 foci per nucleus in midpachytene nuclei (Figures S3BS3D; Whetstine et al., 2006). We also observed an increase in the number of RAD-51 foci in the mitotic zone, which likely reects stalled or collapsed replication forks (Figure 3A; Garcia-Muse and Boulton, 2005). The increased RAD-51 foci in the mitotic zone were accompanied by larger, morphologically distinct nuclei that were phenotypically similar to HU-treated nuclei (Figure S3E; Garcia-Muse and Boulton, 2005). These nuclei were greater than 4.2 microns in diameter, which is signicantly larger than the 3.5 micron average of wild-type nuclei. Often, these HU-like nuclei had increased RAD-51 foci (Figures S3E and S3F). Furthermore, jmjd-2/ worms had less nuclei in the mitotic zone when compared to wild-type worms (Figure 3B, 175 3 nuclei versus 209 2 nuclei, respectively; p < 1 3 1015). The number of nuclear diameters between the distal tip cell and the transition zone was unchanged when compared to wild-type worms (20.4 0.4 versus 20.8 0.3, p > 0.3), suggesting impaired proliferation of a structurally normal mitotic zone. To directly measure DNA replication in the jmjd-2/ and wildtype adult C. elegans germline, we injected Cy3-dUTP and scored the number of nuclei that incorporated Cy3-dUTP in the rst ten nuclear diameters. This region most likely reects the germ cell nuclei in mitotic S phase (Figure S4A; JaramilloLambert et al., 2007). The jmjd-2/ worms had reduced Cy3dUTP incorporation in the rst ten nuclear diameters of the mitotic zone (Figures 3C and 3D and Figure S4B). The jmjd2/ worms consistently had more unlabeled nuclei when compared to wild-type worms (Figures 3C and 3D and Figure S4B). Using the same exposure settings, we also noted less overall signal per nucleus when compared to the wild-type worms. Since decreased nucleotide incorporation could reect less proliferation without impacting the timing of replication, we directly measured replication timing by utilizing a pulse-chase method in adult germlines (Jaramillo-Lambert et al., 2007). Cy3-dUTP was injected into adult germlines and after a 4 hour recovery Alexa 488-dUTP was injected into the same germline and allowed to incorporate for an additional hour. We then quantied the number of nuclei labeled with both Cy3- and Alexa 488dUTP in each germline (designated dual labeled nuclei). At the same time, we scored the number of nuclei containing overlap of Cy3- and Alexa 488-dUTP incorporation at the same position within the chromosomes (yellow, designated colabeled nuclei). If replication is slower, there will be an increase in the number of germlines with dual labeled nuclei, as well as an increase in colabeled (yellow) chromosomal regions. We observed a signicant increase in the number of germline nuclei that contained both Cy3-dUTP and Alexa 488-dUTP (Figures 3E and 3F), as well as colabeled regions within the same nucleus in jmjd-2/ animals (Figures 3G and 3H). These results strongly indicate
that JMJD-2 is required for the normal timing of DNA replication in vivo. Similar to JMJD-2-depleted worms, jmjd-2/ worms had increased germline apoptosis that was mediated by CEP-1 (the p53 homolog), which established a link to DNA damageinduced apoptosis (Figure 4A; Whetstine et al., 2006). Since loss of JMJD-2 slowed replication and increased damage, we reasoned that the increased apoptosis was a result of the replication stress kinase ATL-1 (the C. elegans ATR homolog) signaling to CEP-1. ATL-1 has been shown to act upstream of CEP-1 in C. elegans to trigger DNA damage-induced apoptosis (Stergiou et al., 2007). Consistently, ATL-1 depletion [atl-1 (tm853) mutant or atl-1(RNAi); Garcia-Muse and Boulton, 2005] rescued the increased apoptosis observed in JMJD-2-depleted worms (Figure 4B). In contrast, the depletion of ATM-1 (PI3 kinase involved in DNA damage response to ionizing radiation) with either atm-1 (RNAi) or the atm-1(gk186) mutant worms did not provide a signicant rescue (Figure 4C). JMJD2A Overexpression Alters Chromatin Accessibility over Cell Cycle Chromatin structure impacts DNA replication timing (Mendez, 2009). Since JMJD2A regulates histone methylation and has been shown to relocalize HP1 upon high levels of overexpression (Klose et al., 2006), we hypothesized that JMJD2A overexpression alters chromatin organization and DNA replication. If enough genomic regions are altered, we would observe modest changes in global chromatin structure. Therefore, we evaluated chromatin accessibility by micrococcal nuclease (MNase) digestion. Ectopic expression of JMJD2A, but not JMJD2AH188A, produced a higher ratio of mononucleosomes to hexanucleosomes in asynchronous cells (Figure 5A). Quantitation of seven independent biological experiments revealed a modest but signicant 1.9-fold increase in mononucleosomes following 15 min of MNase digestion (Figure 5B, p = 0.032). We then evaluated chromatin accessibility during S phase. Control and JMJD2A-overexpressing cells were synchronized at G1/S with HU and nuclei were collected from arrested and released cells (2.5 and 5 hr) so that chromatin structure could be monitored over the course of S phase (Figure 5C). JMJD2Aoverexpressing cells had more open chromatin in the G1/ S-arrested cells and at each subsequent point through S phase (Figure 5C). We hypothesized that JMJD2A alters the accessibility of specic genomic regions and in turn promotes DNA replication and S phase progression. To test this hypothesis, we examined the accessibility of Chr1 sat2 and X cen by Southern blot analysis of the MNase digests. We observed that sat2 was more accessible to MNase in asynchronous cells overexpressing JMJD2A (Figures 5D and 5E). X cen also exhibited an increase in chromatin accessibility, albeit more modest than sat2 (Figures S5A and S5B). Further, sat2 became more accessible during the progression of S phase in JMJD2A-overexpressing cells (Figure 5F). Interestingly, in JMJD2A-overexpressing cells, sat2 exhibited the strongest increase in accessibility 2.5 hr into S phase, which suggests this locus becomes more accessible earlier during S phase. These results demonstrate that the increased DNA replication observed at sat2 and X cen corresponds with
Molecular Cell
Figure 3. In Vivo Labeling of DNA Synthesis in Wild-Type and jmjd-2/ Worms

(A) jmjd-2/ worms have an increased frequency of RAD-51 foci within the mitotic zone. Scale bar, 10 mm. (B) jmjd-2/ mutants have a decreased number of nuclei in the mitotic zone. Error bars represent the SEM from scoring at least 30 germlines of each genotype (p < 1 3 1015 by two-tailed Students t test). (C) jmjd-2/ mutants have decreased replication in the mitotic zone as measured using Cy3-dUTP to label nuclei undergoing DNA replication. The rst ten nuclear diameters of the mitotic zone are shown. The distal tip cell is on the left. A comprehensive panel of germlines scored is available in Figure S6. Cy3dUTP is shown in white and DAPI is shown in blue. Scale bar, 2 mM. (D) Quantitation of 36 wild-type and 30 jmjd-2/ germlines scored blind for Cy3-dUTP incorporation. (E) Representative nuclei from N2 or jmjd-2/ worms double injected with Cy3-dUTP and 488-dUTP. Arrowheads indicate regions labeled with either Cy3- or 488-dUTP (dual labeled nuclei). Chevrons demonstrate chromosome regions labeled with both Cy3- and 488-dUTP (colabeled nuclei). Scale bar, 2 mM. (F) Quantication of dual labeled nuclei. (G) Magnied (200%) image of jmjd-2/ nuclei from (E). (H) Quantication of the colabeled nuclei in N2 and jmjd-2/ worms.
Molecular Cell
Figure 4. CEP-1 and ATL-1 Depletion Rescues the Increased DNA Damage-Based Apoptosis in JMJD-2-Depleted Worms
(A) CEP-1 depletion by either cep-1 (RNAi) or cep-1(lg12501) rescues the increased apoptosis observed in either jmjd-2/ or jmjd-2 (RNAi)-treated worms. (B) ATL-1 depletion by either atl-1 (RNAi) or atl-1(tm853) rescues the increased apoptosis observed in either jmjd-2/ or jmjd-2 (RNAi)-treated worms. (C) ATM-1 depletion by either atm-1 (RNAi) or atm-1(gk186) does not rescue the increased apoptosis observed in either jmjd-2/ or jmjd-2 (RNAi)-treated worms. Each experiment was conducted at least three independent times with at least 30 germlines per genotype per experiment. The SEM is shown for each experiment. ** indicates p < 0.0001, and R indicates rescue and no signicant difference from wild-type control by Students t test (p > 0.05).
a more open chromatin environment (Figures 1G and 1H and Figure S1G). HP1g/HPL-2 Antagonize JMJD2A/JMJD-2 Cell Cycle Phenotypes High overexpression of JMJD2A results in HP1 redistribution (Klose et al., 2006). Several studies have clearly demonstrated an important and highly conserved role for HP1 proteins in DNA replication (Grewal and Jia, 2007; Quivy et al., 2008; Wu et al., 2006). Based on these previous studies and the data presented here, we hypothesized that JMJD2A antagonizes HP1 proteins, resulting in faster DNA replication and S phase progression. We rst determined whether HP1 protein expression was regulated during cell cycle in a similar manner to JMJD2A. We observed that HP1g was lowest during G1/S and S phase and highest during G2/M (Figure 6A). This pattern was reciprocal to JMJD2A (Figures 1A and 6A). HP1b also had a similar pattern during cell cycle; however, HP1a resembled the expression pattern of JMJD2A (Figure 6A). To test the hypothesis that JMJD2A and HP1 proteins antagonize each other during S phase, all three HP1 isoforms were overexpressed in JMJD2A-overexpressing cells. S phase progression was monitored by ow cytometry and plotted based on the percentage of cells in early and late S phase 12 hr after release from nocodazole (expression levels in Figure S6A). As demonstrated earlier (Figure 1C), JMJD2A-overexpressing cells proceeded into late S phase (L) faster than control cells (Figure 6B, late S [L]; p = 0.000064). Overexpression of HP1a or HP1b did not alter the faster progression of JMJD2A cells. However, HP1g overexpression completely abrogated the increased progression of JMJD2A cells into late S phase (compare JMJD2A + control to JMJD2A + HP1g for late S [L]; p = 0.0055). Consistent with increased JMJD2A displacing HP1g, knockdown of HP1g also modestly increased progression through S phase (Figures S6B and S6C), albeit less dramatically than JMJD2A overexpression. We then determined if the divergent N terminus, the chromodomain, or the chromo shadow domain of HP1g was important for mediating the rescue of the JMJD2A phenotype. While overexpression of HP1g lacking the N terminus still antagonized JMJD2A (p = 0.0047 with respect to JMJD2A + control), the chromodomain or the chromo shadow domain deletions were unable to compensate for JMJD2A overexpression (Figure 6B; expression levels shown in Figure S6A).
Molecular Cell
Figure 5. JMJD2A Overexpression Increases Chromatin Accessibility

(A) Representative micrococcal nuclease (MNase) digestion analysis of HEK293T cells overexpressing GFP-control or GFP-JMJD2A. (B) Quantitation of seven biological replicates represented in (A). The percentage of mononucleosomes relative to total DNA in each lane was normalized to the amount of the highest visible nucleosome (hexanucleosome). Error bars represent the SEM. For 15 min, p = 0.032. (C) MNase digestion analysis of chromatin from GFP control or GFP-JMJD2A cells during S phase (24 hr in HU [0 Hour] and washed and released for 2.5 hr [earlymid S] and 5 hr [Mid/Late S]). Black triangle represents MNase digest time titration of 1, 3, 6, 9, and 15 min. Values below the graph depict the ratio of mononucleosomes to tetranucleosomes. (D) MNase digestion of GFP-Control, GFP-JMJD2A, or GFP-JMJD2A catalytic mutant (H188A) HEK293T cells. (E) Chr1 sat2 is more accessible in asynchronous JMJD2A-overexpressing cells. Southern blot for Chr1 sat2 of (D). (F) Chr1 sat2 is more accessible throughout S phase in JMJD2A-overexpressing cells. Southern Blot for Chr1 sat2 of (C).
We further veried the antagonism between HP1g and JMJD2A during DNA replication of Chr1 sat2. HP1g overexpression antagonizes the increased DNA replication at sat2 in JMJD2A-overexpressing cells (p = 0.0128), while HP1a and HP1b did not have a statistically signicant impact (Figure 6C; HP1a, p = 0.7654; HP1b, p = 0.1711). Overexpression of HP1 isoforms had no impact on replication at ALU 19 (Figure S6D). Since HP1g and JMJD2A antagonism could be indirect, we evaluated HP1g occupancy during S phase at Chr1 sat2 in control and JMJD2A cells. HP1g-bound chromatin was immunoprecipitated from control and JMJD2A cells in early-mid S (released for 2.5 hr from HU). HP1g binding was reduced upon JMJD2A overexpression (Figure 6D, p = 0.009). HP1g depletion at Chr1 sat2 correlated with the increased chromatin accessibility observed in early S phase (Figure 5F). We also observed JMJD2A-dependent reduction in HP1g binding at X cen in early-mid S phase (Figure S6E; p = 0.007). Consistent with the loss of HP1g, replication of X cen was increased in early-mid S phase in JMJD2A-overexpressing cells (Figure S1G). These results demonstrate an isoform-specic antagonism between HP1g and JMJD2A during DNA replication.
To determine if HP1g depletion was a direct consequence of JMJD2A, we examined JMJD2A occupancy and histone methylation levels at Chr1 sat2 and X cen. In JMJD2A overexpressing cells, JMJD2A occupancy at sat2 was 2-fold higher than in control cells within the rst hour of release form HU (Figure 6E) and decreased in occupancy prior to a concomitant decrease in H3K9me3 (Figure 6F; p = 0.004) and H3K36me3 (Figure S6H; p = 0.003) that coincided with the decreased binding of HP1g. Similarly, ectopic JMJD2A bound X cen 2.5 hr after HU release (Figure S6F) and resulted in a modest decrease in H3K9me3 (Figure S6G; p = 0.04) without changes in H3K36me3 (Figure S6H). We did not observe signicant changes in H1.4 or H1.4K26me3 at either locus 2.5 hr after HU treatment (data not shown). We next sought to address whether this antagonism occurred in vivo. We determined whether the C. elegans HP1 homologs (HPL-1 and HPL-2; Schott et al., 2006) would rescue the increased DNA damage, DNA damage-induced apoptosis, and impaired DNA replication in jmjd-2/ worms. We used both RNAi and mutant alleles for the two HP1 homologs in C. elegans because the hpl-2(tm1489) allele had pleiotropic phenotypes,
Molecular Cell
Figure 6. Functional Antagonism between JMJD2A/JMJD-2 and HP1g/HPL-2

(A) HP1 protein levels change during cell cycle in whole-cell extracts. (B) HP1g antagonizes JMJD2A-dependent S phase progression. The graph depicts the percentage of cells in early (E) and late (L) S 12 hr after nocodazole release of four independent transfections and ow analyses. DN removes the N terminus of HP1g before the chromodomain, DCS removes the chromo shadow domain, and DCD removes the chromodomain. (C) HP1g overexpression antagonizes the JMJD2A-dependent increased replication at Chr1 sat2 observed using BrdU ChIP. Graph depicts averages from at least six independent transfections. (D) JMJD2A overexpression displaces HP1g from chromosome 1 sat 2 in early-mid S. Cells were arrested in HU and released into fresh media for 2.5 hr prior to xation and analysis by ChIP. (E) JMJD2A is enriched at Chr1 sat2 within 1 hr of HU release in GFP-JMJD2A cells. Data are presented as a ratio of percent IP JMJD2A to percent IP control. (F) H3K9me3 is depleted at Chr1 sat2 2.5 hr following HU release (p = 0.009). Data are presented as the ratio of percent IP of H3K9me3/percent IP of H3 in JMJD2A cells to percent IP of H3K9me3/percent IP of H3 in control cells. For all graphs, error bars represent the SEM and R represents no signicant difference from control (p > 0.05).
including a smaller germline at 20 C (A.A. and J.R.W., unpublished data; Couteau et al., 2002). We obtained the exact same data with both RNAi and mutant alleles. Depletion of HPL-2 had similar phenotypes to the jmjd-2/ worms, including increased RAD-51 in the mitotic zone and decreased number of mitotic nuclei (Figures 7A and 7B). Surprisingly, HPL-2 depletion in jmjd-2/ worms resulted in complete rescue of the increased RAD-51 foci in the mitotic and meiotic zones (Figure 7A and Figure S6I), blocked the accumulation of HU-like nuclei (Figure S6J), restored nuclei number in the mitotic zone to wild-type (Figure 7B), and rescued germline apoptosis (Figure 7C). HPL-1 depletion [hpl-1(tm1624) allele or hpl-1 (RNAi)] did not rescue the increased DNA damage (Figure 7A and Figure S6I), HU-like nuclei (Figure S6J), decreased nuclei number (Figure 7B), or DNA damage-induced apoptosis in jmjd-2/ worms (Figure 7D). In order to better resolve the relationship between HPL-2 and JMJD-2 in DNA replication, we depleted HPL-2 from wildtype and jmjd-2/ worms by RNAi and injected Cy3-dUTP into adult germlines. As expected, we observed decreased replication in the jmjd-2/ mutants and wild-type worms depleted
of HPL-2 (Figures 7E and 7F). However, HPL-2 depletion in the jmjd-2/ worms rescued the impaired DNA replication (Figure 7F).
DISCUSSION Here we demonstrate a conserved role for the JMJD2A/JMJD-2 demethylases in DNA replication. Ectopic expression of human JMJD2A promotes passage through S phase and alters replication timing, while loss of C. elegans JMJD-2 slowed DNA replication in the mitotic zone of the adult germline. Consistent with these observations, JMJD2A protein levels are regulated during cell cycle and are important in protecting cells from HU-induced replication stress, a role shared by JMJD-2. The increased DNA damage-induced apoptosis observed upon JMJD-2 depletion was shown to be p53 and ATR dependent but not ATM dependent, providing additional genetic data that support the role of JMJD-2 in DNA replication. Taken together, our human cell culture and C. elegans observations illustrate an important role for the JMJD2 family in DNA replication.
Molecular Cell
Figure 7. In Vivo Antagonism between JMJD-2 and HPL-2

(A) Depletion of HPL-2, but not HPL-1, rescues the increased RAD-51 foci observed in the mitotic zone of jmjd-2/ animals. (B) Depletion of HPL-2, but not HPL-1, by RNAi rescues the decreased mitotic index observed in jmjd-2/ worms. (C) HPL-2 depletion by either hpl-2 (RNAi) or hpl-2(tm1489) rescues the increased apoptosis observed in either jmjd-2/ or jmjd-2 (RNAi)-treated worms. (D) HPL-1 depletion by either hpl-1 (RNAi) or hpl-1(tm1624) does not rescue the increased apoptosis observed in either jmjd-2/ or jmjd-2 (RNAi)-treated worms. ** indicates p < 0.0001. For all graphs, error bars represent the SEM of at least ten animals of each genotype. R represents no statistical difference from wild-type by two-tailed Students t test (p > 0.05). (E and F) Depletion of HPL-2 rescues the decreased replication observed in jmjd-2/ worms. Germlines were injected with Cy3-dUTP and dissected and visualized after 4 hr (see Figure 3). (E) and (F) are graphical representations of the number of germlines with different percentages of nuclei labeled in the rst ten nuclear diameters.
Interestingly, the changes in S phase progression that we observe with JMJD2A overexpression are directly paralleled by modest increases in chromatin accessibility in both asynchronous and cells progressing through S phase. Consistent with JMJD2A altering DNA replication through chromatin structure, we observed increased DNA replication at sat2 and X cen, which coincided with a decrease in HP1g binding and increased chromatin accessibility. The increased chromatin accessibility at sat2 coincided with a change to earlier replication during S phase. We further fortied the importance of the relationship between JMJD2A/JMJD-2, chromatin structure, and DNA synthesis by demonstrating that HP1g/HPL-2 functionally antagonized the DNA replication phenotypes observed from worm to human. Importantly, we did not observe a strong role for the other human HP1 proteins or HPL-1 in antagonizing JMJD2A/JMJD-2 function. The altered chromatin structure and antagonism observed with HP1 proteins across species suggest a highly conserved role for JMJD2A in regulating chromatin structure so that DNA synthesis has the correct spatiotemporal control. Crosstalk between JMJD2A/JMJD-2 and HP1g/HPL-2 Our ndings describe a highly conserved model whereby JMJD2A antagonizes HP1g to control chromatin structure and
allow proper spatiotemporal timing of DNA replication (model in Figure S7). HP1 proteins could be seen as a barrier to replication initiation, or alternatively, a steric blockade to replication elongation. JMJD2A could alleviate this by removal of H3K9/ 36me3, allowing proper spatiotemporal control of replication. Consistent with this model, JMJD2A overexpression resulted in a modestly more open chromatin conformation, decreased HP1g occupancy, decreased H3K9me3, and increased replication of Chr1 sat2 and a region near the X centromere. HP1g required both the chromodomain and chromo shadow domain to antagonize JMJD2A-dependent S phase progression, supporting the model that HP1g requires the ability to bind methyl modications and interact with other chromatin regulators to antagonize JMJD2A (Kwon and Workman, 2008). While HP1g binding could be controlled independently of histone modications, the changes to H3K9me3 at sat2 are likely responsible for the reduction in HP1g occupancy. While it is possible nonhistone substrates may be important, it seems likely given the importance of the methyltransferases in yeast, mouse, and human systems that at least part of the role of JMJD2A/JMJD-2 in S phase progression is through demethylation of H3K9me3. Loss of JMJD-2 in C. elegans slowed DNA replication. In this scenario, JMJD-2 loss would result in an inability to disrupt the
Molecular Cell
HP1 steric blockade so that replication initiation and/or elongation are impaired (model in Figure S7). As predicted from the model, HPL-2 depletion rescued the defects in jmjd-2/ worms. Alternatively, loss of jmjd-2 could result in increased histone methylation, resulting in spread and dilution of HPL-2 to additional regions not normally controlled by HPL-2. Interestingly, loss of HPL-2 by itself led to increased RAD-51 foci in the mitotic zone, fewer mitotic nuclei, and defects in replication. Thus, it was surprising that HPL-2 depletion rescued the same defects in jmjd-2/ animals. Our results suggest that controlling the levels and distribution of HPL-2 is crucial for germline nuclei proliferation. HPL-2 loss could cause replication defects through at least two mechanisms: (1) HPL-2 loss removes constraints on chromatin, resulting in inappropriate early replication, which could trigger ATR checkpoint resulting in overall loss of replication efciency. (2) Loss of HPL-2 could result in spreading and mislocalization of HPL-1, resulting in inappropriate regulation of replication. These observations suggest that there is an intricate antagonistic relationship between the demethylase and specic HP1 homologs. Relationship between HP1g and HPL-2 We observed less DNA replication in young adults treated with both hpl-2 (RNAi) or JMJD-2 depletion; however, codepletion restored the worms to wild-type levels. The specic role of HPL-2 in the mitotic zone could reect changes in chromatin structure, as well as transcriptional changes. Both HPL-2 and HP1g are able to rescue the JMJD-2 and JMJD2A phenotypes, respectively. These data suggest that HPL-2 may in fact be a functional homolog to HP1g. Although sequence homology does not resolve this relationship, our functional observations across species raise this interesting possibility. JMJD2A, Chromatin Structure, and DNA Replication Our results highlight the importance of controlling chromatin structure during replication and implicate JMJD2 histone demethylases as contributors to this regulation. MacAlpine et al. recently demonstrated that increasing the number of potential origins within a region correlates with earlier replication timing (MacAlpine et al., 2010). Consistent with changes in initiation, we observed increased production of ssDNA, indicative of increased replication forks in JMJD2A-overexpressing cells. This suggests that JMJD2A might be creating more favorable chromatin conditions for initiation of replication. Furthermore, JMJD2A overexpression resulted in earlier replication of Chr1 sat2, demonstrating the ability of JMJD2A to impact replication timing. Our results do not exclude the possibility that JMJD2A is also contributing to changes in replication by altering elongation rates. The ability of a chromatin modier like JMJD2A to alter the modication state and the underlying structure could have a direct impact on the plasticity observed within specic genomic regions. JMJD2A and HP1g could be important regulators of epigenetic states required for maintaining replication timing or specifying borders of replication domains. We would predict that additional chromatin modiers (e.g., histone demethylases and histone methyltransferases) will be able to alter both chromatin state and cell cycle progression.
Indeed, the methyltransferase Dot1 regulates cell cycle genes through H3K79me2 and is required for efcient entry into S phase (Schulze et al., 2009). Our data strongly suggest that chromatin modiers can also impact cell cycle through chromatin structure. However, these two effects do not have to be mutually exclusive. There is a strong possibility that chromatin regulators will inuence both aspects, allowing for a tighter level of control of cell cycle progression. These results suggest that cells and perhaps individual origins must proceed through a chromatin checkpoint. Understanding what factors establish, read, maintain, and overcome this checkpoint will be critical to advance our knowledge of proper spatial and temporal control of replication.
EXPERIMENTAL PROCEDURES Cell Culture and Synchronization HEK293T and HeLa S cells were grown in DMEM supplemented with 10% heat-inactivated fetal bovine serum, 50 I.U. per ml penicillin, 50 mg per ml streptomycin, and 2 mM L-glutamine. Stable cell lines were periodically reselected in 3 ug/ml puromycin or by FACS to select GFP-positive cells. All experiments were performed with cells at least one passage postpuromycin treatment. Cells were synchronized by treatment with 1 mM HU (Sigma) for 24 hr or 50 ng/ml nocodazole (Sigma) for 10 hr. G2/M-arrested cells were collecting by mitotic shakeoff. To release arrested cells, cells were washed once with media and supplied fresh media. Generation of Stable Cell Lines Retrovirus was generated in Phoenix A cells 48 hr posttransfection. Cells were infected twice and allowed to recover 2448 hr prior to selection. Cells were selected twice with 4 ug/ml puromycin for 4872 hr with a 48 hr recovery period between puromycin treatments. Stable lines were veried by RT-qPCR and western analysis. Each cell line was created at least two or more times and used in the experiments described above. Subsequent selections were performed by sorting for GFP-positive cells. Cell Growth Curves Cells were plated in triplicate in six-well dishes at 4 3 104 cells per well. For growth curves, the cells were counted 24, 48, and 72 hr after plating. Cells were harvested and stained with trypan blue, and live cells were manually counted using a hemocytometer. The number of live cells counted at 24 hr was set as a relative population of 1. The cell number at 48 and 72 hr was normalized to this value to generate relative population. For HU release analysis, cells were treated with 1 mM HU on the day after plating. Twenty-four hours after addition of HU (day 0), cells were released and counted at 0, 24, 48, and 72 hr. The relative populations were calculated by normalizing to the number of live cells at day zero. Analysis of Single-Stranded DNA during S Phase 1 3 106 cells were seeded in a 10 cm dish for 24 hr prior to labeling with Cell Proliferation and Labeling Reagent (GE Healthcare) for 24 hr. Cells were supplied fresh media and arrested in nocodazole for 10 hr. Cells were released and harvested in RIPA buffer at the indicated time points. Experiment was performed with minor modications from (Jasencakova et al., 2010). DNA was slot blotted and analyzed for BrdU (BD Biosciences). Film was scanned and analyzed using ImageJ V1.43 (NIH). For each time point the ratio of native/ denatured BrdU signal was calculated. BrdU DNA Immunoprecipitation 1 3 106 cells were seeded in a 10 cm dish. Thirty-six to forty-eight hours after seeding, cells were treated for 1 hour with Cell Proliferation and Labeling Reagent (GE Healthcare). Cells were collected and BrdU ChIP was performed as described in Hansen et al. (2010). The complete method can be found in the Supplemental Information. Chr1 sat2 primers are identical to those used in
Molecular Cell
(Zeng et al., 2009). X cen primers amplify a 96 bp amplicon at 61,569 KB from the left arm of the chromosome in the region designated as the centromere on the UCSC genome browser using the NCBI36.3/hg18 build. This region lies within a 2 KB repeat encompassing approximately 36 KB. Chromatin Immunoprecipitation 1.5 3 106 cells were seeded in four 15 cm dishes. Thirty-six hours after seeding, cells were treated with 1 mM HU. Twenty-four hours after treatment, cells were released for analysis of S phase. Chromatin immunoprecipitation (ChIP) experiments were done as described earlier with slight modications (Shi et al., 2003). A complete protocol can be found in the Supplemental Information. For JMJD2A, all IPs were above 0.2% of input. For histones and histone modications, all IPs were greater than 3% of input. Flow Cytometry Complete methods can be found in the Supplemental Information. Sorting of live cells for replication timing was performed as in Hansen et al. (2010). Western Blots Western blots were performed according to Whetstine et al. (2006). A complete antibody list is in the Supplemental Information. Micrococcal Nuclease Assay 2 3 106 cells were seeded in a 15 cm dish 24 hr prior to the experiment. Complete methods can be found in the Supplemental Information. Images were captured on a GelDoc XR (Bio-Rad) using Quantity One software. Exported JPEGs were quantitated using ImageJ v1.43 (NIH). C. elegans Strain Maintenance and RNAi Knockdown Worms were cultured at 20 C on NGM plates seeded with OP50 as previously described (Brenner, 1974). For all experiments, L1 larvae were synchronized by hypochlorite treatment of gravid adults and allowed to grow for 62 hr before analysis. A complete list of strains is in the Supplemental Information. RNAi knockdown of C. elegans genes was as previously described (Whetstine et al., 2006). Dissection, DAPI Staining, and Antibody Staining of the C. elegans Germline Dissection, DAPI staining, and antibody staining of C. elegans germlines were carried out largely as previously described (Colaiacovo et al., 2003; Whetstine et al., 2006). Optical sections were collected at 0.2 or 0.5 mm intervals using an ix81 spinning disc confocal microscope (Olympus). Images were acquired and analyzed using Slidebook 5.0.0.1. Cy3- and Alexa 488-dUTP Analysis of the C. elegans Germline Injections were done on animals 62 hr post-L1 release grown on either NGM plates seeded with OP50 or RNAi plates seeded with the appropriate RNAi food source at 20 C. Cy3-dUTP (GE) was injected at a concentration of 1 pmol/ml into young adult C. elegans germ lines as previously described (Jaramillo-Lambert et al., 2007). Complete methods can be found in the Supplemental Information. Analysis of Hydroxyurea Sensitivity of C. elegans Embryos Fifteen synchronized young adult animals were seeded to ve NGM plates containing the indicated concentration of HU (Sigma-Aldrich) and 10 ml of concentrated OP50 or RNAi food for 13 hr at 20 C. Eggs were quantied and embryonic lethality was scored 20 hr later. All experiments were done in triplicate. Quantitative Analysis of C. elegans Germline Apoptosis Scoring and analysis of germ line apoptosis was done as previously described (Whetstine et al., 2006).
SUPPLEMENTAL INFORMATION Supplemental Information includes seven gures, Supplemental Experimental Procedures, and Supplemental References and can be found with this article at doi:10.1016/j.molcel.2010.11.008. ACKNOWLEDGMENTS We are grateful to Ravi Mylvaganam and the MGH Flow Cytometry core for assistance with the FACS analysis. We would like to thank Sigrid Jacobshagen for technical assistance and Othon Iliopoulos for helpful discussions and comments on the manuscript. We thank Jane Hubbard for suggestions and technical advice regarding mitotic germline nuclei. We thank Steve Smale for providing antibodies to HP1a and HP1b, and Adriana La Volpe for providing antibody to RAD-51. We thank Dr. Shohei Mitani of the National Bioresource Project for the tm2966 deletion allele and the CGC for additional C. elegans mutant strains. This work was supported by funding to J.R.W. from the Ellison Medical Foundation, Milton Fund, Astra Zeneca MGH Alliance, and the Howard Goodman Fellowship. J.C.B. is a Fellow of The Jane Cofn Childs Memorial Fund for Medical Research. This investigation has been aided by a grant from The Jane Cofn Childs Memorial Fund for Medical Research. Received: March 6, 2010 Revised: June 8, 2010 Accepted: September 10, 2010 Published: December 9, 2010 REFERENCES Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thomas, J.O., Allshire, R.C., and Kouzarides, T. (2001). Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo domain. Nature 410, 120124. Bernstein, B.E., Meissner, A., and Lander, E.S. (2007). The mammalian epigenome. Cell 128, 669681. Biswas, D., Takahata, S., Xin, H., Dutta-Biswas, R., Yu, Y., Formosa, T., and Stillman, D.J. (2008). A role for Chd1 and Set2 in negatively regulating DNA replication in Saccharomyces cerevisiae. Genetics 178, 649659. Brenner, S. (1974). The genetics of Caenorhabditis elegans. Genetics 77, 7194. Chen, E.S., Zhang, K., Nicolas, E., Cam, H.P., Zofall, M., and Grewal, S.I. (2008). Cell cycle control of centromeric repeat transcription and heterochromatin assembly. Nature 451, 734737. Cimbora, D.M., Schubeler, D., Reik, A., Hamilton, J., Francastel, C., Epner, E.M., and Groudine, M. (2000). Long-distance control of origin choice and replication timing in the human beta-globin locus are independent of the locus control region. Mol. Cell. Biol. 20, 55815591. Cloos, P.A., Christensen, J., Agger, K., and Helin, K. (2008). Erasing the methyl mark: histone demethylases at the center of cellular differentiation and disease. Genes Dev. 22, 11151140. Colaiacovo, M.P., MacQueen, A.J., Martinez-Perez, E., McDonald, K., Adamo, A., La Volpe, A., and Villeneuve, A.M. (2003). Synaptonemal complex assembly in C. elegans is dispensable for loading strand-exchange proteins but critical for proper completion of recombination. Dev. Cell 5, 463474. Couteau, F., Guerry, F., Muller, F., and Palladino, F. (2002). A heterochromatin protein 1 homologue in Caenorhabditis elegans acts in germline and vulval development. EMBO Rep. 3, 235241. Czvitkovich, S., Sauer, S., Peters, A.H., Deiner, E., Wolf, A., Laible, G., Opravil, S., Beug, H., and Jenuwein, T. (2001). Over-expression of the SUV39H1 histone methyltransferase induces altered proliferation and differentiation in transgenic mice. Mech. Dev. 107, 141153. Daujat, S., Zeissler, U., Waldmann, T., Happel, N., and Schneider, R. (2005). HP1 binds specically to Lys26-methylated histone H1.4, whereas simultaneous Ser27 phosphorylation blocks HP1 binding. J. Biol. Chem. 280, 3809038095.
Molecular Cell
Dohke, K., Miyazaki, S., Tanaka, K., Urano, T., Grewal, S.I., and Murakami, Y. (2008). Fission yeast chromatin assembly factor 1 assists in the replicationcoupled maintenance of heterochromatin. Genes Cells 13, 10271043. Fodor, B.D., Kubicek, S., Yonezawa, M., OSullivan, R.J., Sengupta, R., PerezBurgos, L., Opravil, S., Mechtler, K., Schotta, G., and Jenuwein, T. (2006). Jmjd2b antagonizes H3K9 trimethylation at pericentric heterochromatin in mammalian cells. Genes Dev. 20, 15571562. Fraga, M.F., and Esteller, M. (2005). Towards the human cancer epigenome: a rst draft of histone modications. Cell Cycle 4, 13771381. Garcia-Muse, T., and Boulton, S.J. (2005). Distinct modes of ATR activation after replication stress and DNA double-strand breaks in Caenorhabditis elegans. EMBO J. 24, 43454355. Grewal, S.I., and Jia, S. (2007). Heterochromatin revisited. Nat. Rev. Genet. 8, 3546. Hansen, R.S., Thomas, S., Sandstrom, R., Caneld, T.K., Thurman, R.E., Weaver, M., Dorschner, M.O., Gartler, S.M., and Stamatoyannopoulos, J.A. (2010). Sequencing newly replicated DNA reveals widespread plasticity in human replication timing. Proc. Natl. Acad. Sci. USA 107, 139144. Hayashi, M.T., Takahashi, T.S., Nakagawa, T., Nakayama, J., and Masukata, H. (2009). The heterochromatin protein Swi6/HP1 activates replication origins at the pericentromeric region and silent mating-type locus. Nat. Cell Biol. 11, 357362. Huang, D.W., Fanti, L., Pak, D.T., Botchan, M.R., Pimpinelli, S., and Kellum, R. (1998). Distinct cytoplasmic and nuclear fractions of Drosophila heterochromatin protein 1: their phosphorylation levels and associations with origin recognition complex proteins. J. Cell Biol. 142, 307318. Jaramillo-Lambert, A., Ellefson, M., Villeneuve, A.M., and Engebrecht, J. (2007). Differential timing of S phases, X chromosome replication, and meiotic prophase in the C. elegans germ line. Dev. Biol. 308, 206221. Jasencakova, Z., Scharf, A.N., Ask, K., Corpet, A., Imhof, A., Almouzni, G., and Groth, A. (2010). Replication stress interferes with histone recycling and predeposition marking of new histones. Mol. Cell 37, 736743. Karnani, N., Taylor, C.M., Malhotra, A., and Dutta, A. (2010). Genomic study of replication initiation in human chromosomes reveals the inuence of transcription regulation and chromatin structure on origin selection. Mol. Biol. Cell 21, 393404. Kim, H.S., Rhee, D.K., and Jang, Y.K. (2008). Methylations of histone H3 lysine 9 and lysine 36 are functionally linked to DNA replication checkpoint control in ssion yeast. Biochem. Biophys. Res. Commun. 368, 419425. Klose, R.J., Yamane, K., Bae, Y., Zhang, D., Erdjument-Bromage, H., Tempst, P., Wong, J., and Zhang, Y. (2006). The transcriptional repressor JHDM3A demethylates trimethyl histone H3 lysine 9 and lysine 36. Nature 442, 312316. Kwon, S.H., and Workman, J.L. (2008). The heterochromatin protein 1 (HP1) family: put away a bias toward HP1. Mol. Cells 26, 217227. Lachner, M., OCarroll, D., Rea, S., Mechtler, K., and Jenuwein, T. (2001). Methylation of histone H3 lysine 9 creates a binding site for HP1 proteins. Nature 410, 116120. Lan, F., Nottke, A.C., and Shi, Y. (2008). Mechanisms involved in the regulation of histone lysine demethylases. Curr. Opin. Cell Biol. 20, 316325. MacAlpine, D.M., Rodriguez, H.K., and Bell, S.P. (2004). Coordination of replication and transcription along a Drosophila chromosome. Genes Dev. 18, 30943105. MacAlpine, H.K., Gordan, R., Powell, S.K., Hartemink, A.J., and MacAlpine, D.M. (2010). Drosophila ORC localizes to open chromatin and marks sites of cohesin complex loading. Genome Res. 20, 201211. Maison, C., Bailly, D., Peters, A.H., Quivy, J.P., Roche, D., Taddei, A., Lachner, M., Jenuwein, T., and Almouzni, G. (2002). Higher-order structure in pericentric
heterochromatin involves a distinct pattern of histone modication and an RNA component. Nat. Genet. 30, 329334. Mendez, J. (2009). Temporal regulation of DNA replication in mammalian cells. Crit. Rev. Biochem. Mol. Biol. 44, 343351. Moss, T.J., and Wallrath, L.L. (2007). Connections between epigenetic gene silencing and human disease. Mutat. Res. 618, 163174. Nakayama, J., Rice, J.C., Strahl, B.D., Allis, C.D., and Grewal, S.I. (2001). Role of histone H3 lysine 9 methylation in epigenetic control of heterochromatin assembly. Science 292, 110113. Quivy, J.P., Gerard, A., Cook, A.J., Roche, D., and Almouzni, G. (2008). The HP1-p150/CAF-1 interaction is required for pericentric heterochromatin replication and S-phase progression in mouse cells. Nat. Struct. Mol. Biol. 15, 972979. Schott, S., Coustham, V., Simonet, T., Bedet, C., and Palladino, F. (2006). Unique and redundant functions of C. elegans HP1 proteins in post-embryonic development. Dev. Biol. 298, 176187. Schubeler, D., Francastel, C., Cimbora, D.M., Reik, A., Martin, D.I., and Groudine, M. (2000). Nuclear localization and histone acetylation: a pathway for chromatin opening and transcriptional activation of the human beta-globin locus. Genes Dev. 14, 940950. Schulze, J.M., Jackson, J., Nakanishi, S., Gardner, J.M., Hentrich, T., Haug, J., Johnston, M., Jaspersen, S.L., Kobor, M.S., and Shilatifard, A. (2009). Linking cell cycle to histone modications: SBF and H2B monoubiquitination machinery and cell-cycle regulation of H3K79 dimethylation. Mol. Cell 35, 626641. Shi, Y., Sawada, J., Sui, G., Affar el, B., Whetstine, J.R., Lan, F., Ogawa, H., Luke, M.P., Nakatani, Y., and Shi, Y. (2003). Coordinated histone modications mediated by a CtBP co-repressor complex. Nature 422, 735738. Stergiou, L., Doukoumetzidis, K., Sendoel, A., and Hengartner, M.O. (2007). The nucleotide excision repair pathway is required for UV-C-induced apoptosis in Caenorhabditis elegans. Cell Death Differ. 14, 11291138. Trojer, P., Zhang, J., Yonezawa, M., Schmidt, A., Zheng, H., Jenuwein, T., and Reinberg, D. (2009). Dynamic histone H1 isotype 4 methylation and demethylation by histone lysine methyltransferase G9a/KMT1C and the Jumonji domain-containing JMJD2/KDM4 proteins. J. Biol. Chem. 284, 83958405. Vakoc, C.R., Mandat, S.A., Olenchock, B.A., and Blobel, G.A. (2005). Histone H3 lysine 9 methylation and HP1gamma are associated with transcription elongation through mammalian chromatin. Mol. Cell 19, 381391. Weidtkamp-Peters, S., Rahn, H.P., Cardoso, M.C., and Hemmerich, P. (2006). Replication of centromeric heterochromatin in mouse broblasts takes place in early, middle, and late S phase. Histochem. Cell Biol. 125, 91102. Whetstine, J.R., Nottke, A., Lan, F., Huarte, M., Smolikov, S., Chen, Z., Spooner, E., Li, E., Zhang, G., Colaiacovo, M., and Shi, Y. (2006). Reversal of histone lysine trimethylation by the JMJD2 family of histone demethylases. Cell 125, 467481. Wu, R., Terry, A.V., Singh, P.B., and Gilbert, D.M. (2005). Differential subnuclear localization and replication timing of histone H3 lysine 9 methylation states. Mol. Biol. Cell 16, 28722881. Wu, R., Singh, P.B., and Gilbert, D.M. (2006). Uncoupling global and netuning replication timing determinants for mouse pericentric heterochromatin. J. Cell Biol. 174, 185194. Zeng, W., de Greef, J.C., Chen, Y.Y., Chien, R., Kong, X., Gregson, H.C., Winokur, S.T., Pyle, A., Robertson, K.D., Schmiesing, J.A., et al. (2009). Specic loss of histone H3 lysine 9 trimethylation and HP1gamma/cohesin binding at D4Z4 repeats is associated with facioscapulohumeral dystrophy (FSHD). PLoS Genet. 5, e1000559. 10.1371/journal.pgen.1000559.
Molecular Cell
Article
Taf1 Regulates Pax3 Protein by Monoubiquitination in Skeletal Muscle Progenitors
Stephane C. Boutet,1 Stefano Biressi,1 Kevin Iori,1 Vanita Natu,1 and Thomas A. Rando1,2,*
of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94305, USA Service, Veterans Affairs Palo Alto Health Care System, Palo Alto, CA 94304, USA *Correspondence: rando@stanford.edu DOI 10.1016/j.molcel.2010.09.029
2Neurology 1Department
SUMMARY
Pax3 plays critical roles during developmental and postnatal myogenesis. We have previously shown that levels of Pax3 protein are regulated by monoubiquitination and proteasomal degradation during postnatal myogenesis, but none of the key regulators of the monoubiquitination process were known. Here we show that Pax3 monoubiquitination is mediated by the ubiquitin-activating/conjugating activity of Taf1, a component of the core transcriptional machinery that was recently reported to be downregulated during myogenic differentiation. We show that Taf1 binds directly to Pax3 and overexpression of Taf1 increases the level of monoubiquitinated Pax3 and its degradation by the proteasome. A decrease of Taf1 results in a decrease in Pax3 monoubiquitination, an increase in the levels of Pax3 protein, and a concomitant increase in Pax3-mediated inhibition of myogenic differentiation and myoblast migration. These results suggest that Taf1 regulates Pax3 protein levels through its ability to mediate monoubiquitination, revealing a critical interaction between two proteins that are involved in distinct aspects of myogenic differentiation. Finally, these results suggest that the components of the core transcriptional are integrally involved in the process of myogenic differentiation, acting as nodal regulators of the differentiation program.
INTRODUCTION Pax3 is a key regulator of myogenesis during development (Buckingham et al., 2003). In splotch (Sp) mice, which carry spontaneous mutations in the Pax3 locus, limb muscles are absent (Goulding et al., 1994; Bober et al., 1994). The formation of these muscles requires Pax3 for the induction of expression of c-Met, a tyrosine kinase receptor essential for the delamination and migration of muscle progenitor cells (Bladt et al., 1995; Epstein et al., 1996; Yang et al., 1996). Similar to what occurs in the process of melanocyte stem cell differentiation (Lang et al., 2005), Pax3 protein expression seems to be associated with an intermediate precursor cell of the myogenic lineage.
While Pax3 appears to maintain an uncommitted state, it also directly regulates Myf5, which plays a major role in the determination of the myogenic cell fate (Bajard et al., 2006). The embryonic progenitors that express Pax3, and its close homolog Pax7, give rise to a population of adult muscle stem cells (Relaix et al., 2005; Gros et al., 2005; Kassar-Duchossoy et al., 2005). Both before and after expression of the myogenic regulatory factors (MRFs) Myf5 and MyoD, muscle precursor cells undergo extensive proliferation in the limb. Pax3 is most likely involved in maintaining this proliferative phase, directly or indirectly, through the activation of c-Met (Delni et al., 2000; Buckingham et al., 2003). The regulation of the transition from proliferative progenitor cell to differentiating myoblast is poorly understood but is associated with a marked downregulation of Pax3. In postnatal myogenesis, Pax3 is transiently expressed during muscle stem cell (satellite cell) activation in a highly proliferative intermediate progenitor cell population (Conboy and Rando, 2002). We have shown that when the cells are transitioning from intermediate progenitors to myoblasts, Pax3 levels decline due to protein ubiquitination and proteasomal degradation and that Pax3 degradation is a necessary step for the terminal differentiation to occur (Boutet et al., 2007). Surprisingly, Pax3 is degraded through monoubiquitination, not polyubiquitination, and shuttled to the proteasome by Rad23B (Boutet et al., 2007). Therefore, to understand the transition from a Pax3+ immature progenitor to a Pax3 mature myoblast, both during satellite cell activation, we sought to identify the protein(s) responsible for the monoubiquitination of Pax3. Monoubiquitination of nuclear proteins is important in the regulation of replication and transcription through histone monoubiquitination (Hicke, 2001). Polyubiquitination requires the concerted action of members of the E1 ubiquitin-activating, E2 ubiquitin-converting, and E3 ubiquitin ligase families, whereas monoubiquitination requires only E1 or E2 enzymes (Ciechanover et al., 2000). Taf1 (previously TafII250) is a major subunit of the TFIID transcriptional initiation complex and is an unusual multifunctional protein that possesses, in addition to a protein kinase activity (Dikstein et al., 1996) and a histone acetyltransferase activity (Mizzen et al., 1996), both E1 ubiquitin-activating and E2 ubiquitin-conjugating (UBAC) activities (Pham and Sauer, 2000). It is responsible for the monoubiquitination of Histone H1 (Pham and Sauer, 2000), a linker histone that binds DNA between two nucleosomes. In the Drosophila embryo, Taf1mediated monoubiquitination of Histone H1 appears to be important for the proper regulation of transcriptional activity (Pham and Sauer, 2000). Inactivation of Taf1 in yeast (Walker
Molecular Cell
Taf1 Monoubiquitinates Pax3
et al., 1997) and in hamster cell lines (Nishimoto et al., 1982) results in cell-cycle arrest at the G1 phase. Null mutations in Drosophila result in lethality in early larval development (Wassarman et al., 2000). These knockout studies in different organisms suggest a very broad role of Taf1 in cell proliferation and/or cell survival (Wassarman and Sauer, 2001). In addition, Taf1 has been recently demonstrated to be differentially expressed during myogenic differentiation (Deato and Tjian, 2007). Whereas Taf1 and core subunits of the TFIID complex are dramatically downregulated during myogenic terminal differentiation, Taf1 appears to be upregulated during the transition from quiescence (reserve cell) to proliferation (myoblast) of myogenic progenitors in vitro (Deato and Tjian, 2007). This result suggests that Taf1 could regulate the transition from myogenic progenitor to myoblasts during postnatal myogenesis. Overall, the nuclear localization of both Pax3 and Taf1, the ability of Taf1 to mediate protein monoubiquitination through its E1 and E2 activities, and the expression pattern of Taf1 protein during the myogenic differentiation all pointed to Taf1 as a candidate for the UBAC activity mediating the monoubiquitination of Pax3 during myogenesis. In the present study, we investigated the mechanism by which Pax3 protein levels are controlled by monoubiquitination and found that Taf1 is indeed a primary mediator of Pax3 monoubiquitination. Using in vitro and cell-free systems, we show that Taf1 is both necessary and sufcient for Pax3 monoubiquitination. These studies reveal the mechanism of Pax3 monoubiquitination. In addition, these results suggest a regulation of the myogenic differentiation program that is integrated with the core transcriptional machinery. RESULTS Regulation of Pax3 by Taf1 in Myogenic Progenitors Based on the ability of Taf1 to catalyze nuclear protein monoubiquitination, we hypothesized that Taf1 might be responsible for Pax3 monoubiquitination. To test this, we transfected satellite cell-derived (primary) myoblasts with either Taf1 siRNA or control siRNA. The siRNA oligonucleotides were transfected with extremely high efciency (>99%) and were highly effective in reducing Taf1 transcript and protein levels without affecting Pax3 levels (Figure 1A and see Figures S1A and S1B available online). Since the monoubiquitination of Pax3 leads to its rapid degradation by the proteasome, we rst analyzed the steadystate levels of Pax3 which increase when ubiquitination is inhibited (Boutet et al., 2007). We previously demonstrated that Pax3 protein is undetectable in myoblasts but becomes detectable when the cells are treated with the proteasome inhibitor, MG132, to block proteasome-mediated degradation (Boutet et al., 2007). Likewise, in myoblasts in which Taf1 protein was knocked down, Pax3 protein was clearly detectable (Figure 1A). As members of the core transcriptional complex have recently been shown to vary depending on the state of cellular differentiation (Deato and Tjian, 2007), we sought to determine if Taf1 is expressed throughout the stages of satellite cell activation and lineage progression when Pax3 protein levels initially increase, due to transcriptional upregulation (Lagha et al., 2008), and then subsequently decline due to monoubiquitination and pro-
teasomal degradation (Boutet et al., 2007). Taf1 is expressed at the transcript and protein level in quiescent cells and increases as the cells begin to proliferate and progress along the myogenic lineage, prior to the onset of differentiation (Figures 1B and 1C). During this time, Pax3 protein increases transiently, peaking during the transit amplifying stage of muscle stem cell activation (Figure 1C and Boutet et al. [2007]). Thus, during the critical transition when Pax3 protein levels are regulated by monoubiquitination and proteasomal degradation, Taf1 is clearly expressed in myogenic progenitors. In fact, Taf1 protein increases signicantly as Pax3 protein levels decrease to almost undetectable levels even though Pax3 transcript levels remain persistently elevated (Figures 1B and 1C; Figure S1C). Using a reserve cell model of satellite cell quiescence and activation (Kitzmann et al., 1998), we also found that Taf1 is expressed in the quiescent state and the expression is increased at both the transcript and protein levels following activation (Figures S1D and S1E). Taf1 Physically Interacts with Pax3 In Vitro and In Vivo Based on the hypothesis that Taf1 regulates Pax3 protein levels by monoubiquitination, the nding that the reduction of Taf1 leads to an increase in Pax3, and the fact that Taf1 and Pax3 are coexpressed in myogenic progenitors, we next determined whether the two proteins interact in cells. We transfected C2C12 myoblasts with plasmids expressing epitope-tagged Pax3 alone or with epitope-tagged Taf1 and performed coimmunoprecipitation studies. Using an antibody against either tag, we found that endogenous Taf1 could be pulled down with Pax3 (Figure 1D) and that, conversely, Pax3 could be pulled down with Taf1 (Figure 1E). Interestingly, Taf4, another member of the TFIID complex, could be pulled down with an antibody to the Pax3 tag (Figure 1D). Therefore, Pax3 binds to Taf1 in the context of the TFIID complex, although it is possible that Pax3 also binds to free Taf1. To test whether endogenous Taf1 and Pax3 proteins interact, we treated primary myoblasts for 24 hr with MG132 to prevent the degradation of Pax3 and performed coimmunoprecipitation studies. Using an antibody against endogenous Taf1, we could pull down endogenous Pax3 (Figure 1F). We previously demonstrated that only the monoubiquitinated form of Pax3 accumulates in myoblasts when degradation is inhibited (Boutet et al., 2007). Indeed, the form of Pax3 that was pulled down in these studies was the monoubiquitinated form (Figure 1F). These results suggest that there is a physical association between Taf1 and Pax3 in myoblasts. To test whether this interaction was direct, we performed pull-down experiments using puried recombinant Taf1 and Pax3 proteins. We found that Pax3 could be pulled down with Taf1 in a cell-free system (Figure 1G), suggesting that the interaction between the two proteins is direct. Taf1 Is Sufcient and Necessary to Promote Pax3 Monoubiquitination To assess whether Taf1 could regulate the level of Pax3 protein by ubiquitination and proteasomal degradation, we compared the stability of Pax3 protein when Taf1 was overexpressed in C2C12 myoblasts in pulse-chase experiments. Taf1 overexpression dramatically increased the rate of degradation of Pax3
Molecular Cell
A
siRNA Taf1: Taf1 Pax3 GAPDH + Cont
B
1.4 Relative mRNA Levels 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Day0 Day1 Day2 Day3 Day4 Day5 Taf1 Pax3
C
Time (days) Taf1 Pax3 GAPDH 0 1 2 3 4 5
D
Pax3-GFP IP: -GFP IB: T f1 IB -Taf1 Cont + Taf1
E
HA-hTaf1 Pax3-GFP + Cont + +
IP: -HA IB: -Taf4 Input: IB: -Taf1 Taf1 IB: -HA Taf4 IB: -GFP
Pax3
Taf1
F
IP: -Taf1 MG132 IB: -Taf1 IB: -Pax3 + Taf1 Pax3
G
Pulldown:
HA HA-hTaf1
protein (Figure 2A). As we had demonstrated that it is the monoubiquitinated form of Pax3 that is subject to protein degradation (Boutet et al., 2007), we tested directly for the ability of Taf1 to mediate Pax3 monoubiquitination using a cell-free assay of protein ubiquitination (Boutet et al., 2007). To test the reliability of the reconstituted ubiquitination assay, we used Histone H2B as a negative control and Histone H1 as a positive control for Taf1 UBAC activity (Pham and Sauer, 2000; Belz et al., 2002). In this assay, Histone H2B was not ubiquitinated by Taf1 in the presence or absence of ATP, whereas, as previously described (Pham and Sauer, 2000), Histone H1 was monoubiquitinated by Taf1 in the presence of ATP, but not when ATP was not present in the reaction (Figure S2). When tested with puried Pax3 in the reaction mixture, Taf1 could monoubiquitinate Pax3 protein in the presence of ATP, but not in its absence (Figure 2B). Puried Taf1 proteins also demonstrated a strong ubiquitin signal, suggesting that it activates ubiquitin efciently in the presence of ATP (Figure 2B). This assay of protein ubiquitination in a cell-free system clearly demonstrates that Taf1 alone is sufcient to monoubiquitinate Pax3 protein. Conversely, we assessed whether Taf1 is necessary for the monoubiquitination of Pax3. We rst examined whether the level of endogenous Pax3 protein could be altered by inhibiting the expression of Taf1 using siRNA. Treatment with increasing amounts of siRNA against Taf1 revealed a dose-dependent increase of endogenous Pax3 protein (Figure 2C). Control siRNA treatment had no effect on Pax3 levels. Taf1 siRNA treatment had no effect on Pax3 mRNA levels (Figure S1B). These data strongly implicate Taf1 as an essential regulator of Pax3 protein stability. As an independent test of the necessity of Taf1 for Pax3 ubiquitination, we inhibited Taf1 expression and analyzed the extent of Pax3 protein monoubiquitination. Specically, in primary myoblasts treated with MG132 to allow the accumulation of ubiquitinated Pax3, we tested whether the inhibition of Taf1 by siRNA could promote the accumulation of nonubiquitinated Pax3 protein. Compared to control myoblasts which accumulate predominantly monoubiquitinated Pax3, Taf1 siRNA-treated
IP: -HA IB: -Taf1 IB: -Pax3
Taf1 Pax3
Input:
Pax3
Figure 1. Taf1 Regulates Pax3 Protein Levels and Binds to Pax3

(A) Western blot analysis of the expression level of Taf1 and Pax3 after treatment with Taf1 siRNA and before induction of differentiation. Primary myoblasts were transfected with Taf1 or control siRNA (60 nM each). Cells were harvested and analyzed by immunoblots with anti-Taf1, anti-Pax3, and anti-GAPDH antibodies. (B) Quantitative analysis of Taf1 and Pax3 mRNA levels in satellite cells from day 0 through day 5 of activation. Each is normalized to GAPDH; the level at the peak of expression was arbitrarily set at 1. Error bars represent SD. (C) Western blot analysis of Taf1 and Pax3 levels at day 0 through day 5 of activation.
(D) Analysis of interaction between Taf1 and exogenous Pax3. C2C12 myoblasts transfected with Pax3-GFP or control (empty vector) constructs were treated for 3 hr with MG132 and lysed. Proteins were immunoprecipitated using an anti-GFP antibody and analyzed by immunoblotting using anti-Taf1 and anti-Taf4 antibodies. (E) Analysis of interaction between exogenous Pax3 and Taf1 in cells. C2C12 myoblasts were transfected with HA-tagged hTaf1 and either Pax3-GFP or GFP expression constructs, treated for 3 hr with MG132, and lysed. Proteins were immunoprecipitated using an anti-HA antibody and analyzed by immunoblotting using anti-HA and anti-GFP antibodies. (F) Analysis of endogenous Pax3 and Taf1 interaction. Primary myoblasts were treated for 24 hr with MG132 (5 mM) to allow accumulation of ubiquitinated Pax3, and lysed. Proteins were immunoprecipitated using an anti-Taf1 antibody and analyzed by immunoblotting using an anti-Pax3 antibody. The position of the band corresponds to the molecular weight of monoubiquitinated Pax3. (G) Analysis of direct interactions between Pax3 and Taf1. Puried HA-tagged hTaf1 was immunoprecipitated using an anti-HA afnity matrix and mixed with puried recombinant Pax3 proteins. Pax3 proteins were analyzed by immunoblotting using an anti-Pax3 antibody. See also Figure S1.
Molecular Cell
1.2 Relative Protein Level
+ Taf1 Control
Control Time (hrs) 0 Taf1 Pax3 DsRed 6 12 24 0
+ Taf1 6 12 24
1.0 0.8 08 0.6 0.4 0.2 0.0 0 6 12
+ATP Taf1-Ub Taf1 Ub1 / Pax3 Ub1 / Pax3 Pax3
-Ub -Taf1 -Ub
IB:
18
24
-Pax3
Chase Time (Hours)
C
Taf1 (nM) 15 30 45
siRNA Control 60 15 30 45 60 Taf1 Pax3 Pax7 GAPDH
Control siRNA Relative Pax3 Protein Level e Relative Taf1 Protein Level e 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 15 30 45 60 siRNA (nM) Taf1 siRNA 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 15 30
Control siRNA Taf1 siRNA
45
60
siRNA (nM)
Non-ub biquitinated Pax3/Total Pax3 3
D
Cont siRNA Taf1 -Taf1 IB: -Pax3 GAPDH -GAPDH + Taf1 Ub1 / Pax3 Pax3
1.0 0.8 0.6 0.4 0.2 0.0 Cont siRNA Taf1
1.0 0.8 0.6 0.4 0.2 0.0
Figure 2. Taf1 Is Both Sufcient and Necessary for Pax3 Monoubiquitination

(A) Analysis of Pax3 protein stability by pulse-chase experiments in C2C12 myoblasts transfected with constructs expressing Pax3 and DsRed, and with either a control vector or a Taf1-expression vector. The cells were treated with cycloheximide for the indicated times. In each case, the level of DsRed was used as an internal control. Protein levels were assessed by immunoblot analysis (left panel). Quantitative analysis of replicate experiments (right panel) shows increased degradation of Pax3 in Taf1-expressing cells. (B) Analysis of Pax3 ubiquitination by Taf1 in a cell-free ubiquitination assay. Reactions containing puried ubiquitin, HA-hTaf1, and Pax3 in the presence or absence of ATP were analyzed by immunoblots with anti-Pax3, anti-Taf1, and anti-ubiquitin antibodies. Arrows show the positions of Pax3 and monoubiquitinated Pax3 (Pax3-Ub1).
Re elative Taf1 protein level
1.2
1.2
-ATP
Molecular Cell
myoblasts accumulate both monoubiquitinated and nonubiquitinated Pax3 (Figures 2D and 1E). The data suggest that Taf1 is the primary UBAC for Pax3, since the reduction of Pax3 monoubiquitination parallels the reduction of Taf1 levels (Figures 2D and 2E), and that the residual Taf1 could certainly account for the residual UBAC activity. However, it cannot be excluded that the presence of another UBAC might contribute to Pax3 monoubiquitination if Taf1 could be completely eliminated (which it cannot, because that is lethal to the cell). Nevertheless, these results suggest that Taf1 is necessary for normal and full Pax3 monoubiquitination. Mutation in the UBAC Domain of Taf1 Reduces Its Ability to Drive the Monoubiquitination of Pax3 In Drosophila, Taf1 protein carrying V1072D or R1096P mutations display a reduced UBAC activity in vitro and in vivo (Pham and Sauer, 2000). We took advantage of the homology between the mammalian and Drosophila Taf1 proteins and generated the homologous mutations, Taf1V1049D and Taf1R1070P, in the mammalian Taf1. Consistent with a reduced UBAC activity, neither mutant, when overexpressed in C2C12 myoblasts, was as effective as wild-type Taf1 in increasing the degradation rate of Pax3 (Figure 3A). Furthermore, we analyzed the level of monoubiquitinated Pax3 in control myoblasts or in myoblasts overexpressing Taf1, Taf1V1049D, or Taf1R1070P. After treating Taf1-expressing cells with MG132 for 3 hr, monoubiquitinated Pax3 was clearly detectable (Figure 3B). There was less monoubiquitinated Pax3 protein when the Taf1 mutants were expressed (Figure 3B). MG132 treatment of control transfected cells for the same amount of time resulted in an even lower level of monoubiquitinated Pax3. Longer treatments (6 hr) with MG132 resulted in even higher levels of Pax3 in these cells, as shown previously (Boutet et al., 2007). Therefore, mutation in the Taf1 UBAC activity domain reduced markedly but not completely the UBAC activity of Taf1. The reduced monoubiquitination of Pax3 paralleled the increased stability of Pax3 proteins. To assess quantitatively the level of monoubiquitination of Pax3, lysates were subjected to nickel-agarose pull-down to purify monoubiquitinated proteins with histidine-tagged ubiquitin. Quantitation of the level of monoubiquitination of Pax3 with the expression of wild-type and mutant Taf1 proteins showed that there was less Pax3 monoubiquitination in the presence of Taf1 mutants than in the presence of wild-type Taf1 (Figure 3C). These data further implicate the role of Taf1 as a key regulator Pax3 monoubiquitination through its UBAC activity.
Functional Role of Taf1 in Regulating Pax3-Mediated Processes in Myogenic Progenitors The preceding studies clearly demonstrate that Taf1 is capable of monoubiquitinating Pax3 and that inhibition of Taf1 leads to increases in Pax3 protein levels. In order to test directly for the functional role of Taf1 in Pax3-mediated processes, we examined two well-established roles of Pax3 in myogenesisthe inhibition of differentiation and the promotion of myogenic progenitor migration (Epstein et al., 1995, 1996; Boutet et al., 2007). We therefore hypothesized that the inhibition of Taf1 expression with siRNA would inhibit myogenic differentiation because of the resulting increase in Pax3. To test this, we transfected primary myoblasts with either Taf1 siRNA or control siRNA and then induced those cells to undergo differentiation. In Taf1 siRNA-treated cultures, markers of terminal differentiation were repressed compared to control siRNA-treated cultures (Figure 4A). To address whether this effect was specically due to an increased level of Pax3, we transfected Taf1-siRNAtreated cells with either control or Pax3 siRNA. Pax3 siRNA efciently downregulated Pax3 transcript and protein levels (Figures S3A and S3B). Compared to control cultures (cells treated with a control siRNA and Taf1 siRNA), cultures transfected with both Taf1 and Pax3 siRNA resulted in a signicant increase in the expression of markers of terminal differentiation (Figure 4B), demonstrating a function role of Taf1 in regulating myogenic differentiation by the regulation of Pax3 levels. It should be noted that the effects of Taf1 siRNA treatment on differentiation are similar to those seen following the expression of mutant forms of Pax3 that are resistant to monoubiquitination (Boutet et al., 2007), supporting the hypothesis that Taf1 regulates myogenic differentiation by its role as a UBAC for Pax3. These data further support the hypothesis that Taf1 regulates myogenic differentiation by regulating Pax3 monoubiquitination. Pax3 has been shown to be essential during limb development for the migration of the muscle progenitor cells (Epstein et al., 1996; Yang et al., 1996). We tested whether the inhibition of Taf1 expression and the subsequent increase of Pax3 expression would have an effect on the migration of primary myoblasts in addition to the effects observed on differentiation. We used time-lapse microscopy to assess myoblast mobility and found that myoblasts treated with Taf1 siRNA migrated twice as fast as myoblasts treated with control siRNA (Figures 5A and 5B). To test whether this enhanced mobility was due to the maintenance of higher Pax3 protein levels, we transfected primary myoblast cultures with Taf1 siRNA together with either Pax3 or control siRNA. Compared to cultures treated with Taf1 siRNA
(C) Western blot analysis of Pax3 levels in myoblasts with either normal or reduced levels of Taf1. Primary myoblasts were transfected with Taf1 siRNA or control siRNA at the indicated concentration. Pax3 and Taf1 protein levels were assessed by immunoblot analysis, and in each case, the level of GAPDH was used as an internal control. Representative blots are shown above, and quantitation of replicate experiments is shown below. (D) Accumulation of nonubiquitinated Pax3 after Taf1 downregulation. Primary myoblasts were treated with 60 nM Taf1 or control siRNA. After 24 hr, cells were incubated with MG132 for 3 hr to allow accumulation of monoubiquitinated Pax3. Pax3 levels were analyzed by immunoblots with an anti-Pax3 antibody. Arrows show the positions of monoubiquitinated and unmodied Pax3. Under the conditions of reduced Taf1, there is a marked accumulation of the nonubiquitinated form, whereas in the absence of Taf1 siRNA nearly all of the protein is monoubiquitinated (and detectable only because of the addition of MG132 to the cultures). (E) Quantitative analysis of replicate immunoblot studies as shown in (D) showing the ratio of nonubiquitinated Pax3 to total Pax3. The relative level of Taf1 protein (gray squares) in cells treated with Taf1 or control siRNA oligonucleotides is superimposed (scale to the right). The ratio of nonubiquitinated Pax3/total Pax3 increased markedly when Taf1 was downregulated, even though there was a concomitant increase in monoubiquitinated Pax3 because of the marked increase of the nonubiquitinated form. The value with the control siRNA treatment was arbitrarily set at 1.0. All data are normalized to GAPDH. Error bars represent SD.
Molecular Cell
+ Taf1 Control + Taf1V1049D +Taf1R1070P
Figure 3. Mutation in the Taf1 UBAC Domain Reduces Pax3 Monoubiquitination and Degradation
(A) Analysis of Pax3 protein stability by pulse-chase experiments in C2C12 myoblasts transfected with constructs expressing Pax3 and DsRed, and with constructs expressing either Taf1V1046D or Taf1R1070P. The cells were treated with cycloheximide for the indicated times. In each case, the level of DsRed was used as an internal control. Protein levels were assessed by immunoblot analysis (left panel). Quantitative analysis of replicate experiments (right panel) shows levels of Pax3 in mutant Taf1-expressing cells that are intermediate between those in control cells and those in Taf1-expressing cells. The data from Figure 2A showing Pax3 levels in control cells or cells expressing wild-type Taf1 are presented in light gray as reference (the studies were done concurrently). (B) Analysis of the level of ubiquitination of Pax3 in the presence of overexpressed wild-type and mutant Taf1. C2C12 myoblasts transfected with Pax3-GFP and either Taf1, Taf1V1046D, Taf1R1070P, or control expression constructs were treated for 3 hr with MG132 to allow accumulation of ubiquitinated Pax3 and lysed. Proteins were denatured by boiling and analyzed by immunoblotting using an anti-GFP antibody (for Pax3 proteins) or FK-2 antibody (recognizes both polyubiquitinated and monoubiquitinated proteins) antibodies. Arrows show the positions of Pax3, monoubiquitinated Pax3 (Pax3-Ub1), and where any polyubiquitinated Pax3 (Pax3-Ubn) would migrate if present. (C) Analysis of the level of ubiquitination of Pax3 in the presence of overexpressed wild-type or mutant Taf1 in vivo. C2C12 myoblasts transfected with Pax3-GFP, His6-tagged ubiquitin, and either Taf1, Taf1V1046D, Taf1R1070P, or control expression constructs were treated for 3 hr with MG132 to allow accumulation of ubiquitinated Pax3 and lysed. His6-tagged ubiquitin conjugates were puried on nickel-agarose gels and levels of Pax3 proteins were analyzed by immunoblotting with an anti-GFP antibody. Again, arrows show the positions of monoubiquitinated Pax3 (Pax3-Ub1) and where any polyubiquitinated Pax3 (Pax3-Ubn) would migrate if present. Monoubiquitination levels (shown quantitatively on the right) were determined as the ratio of monoubiquitinated Pax3 protein to Pax3 protein input and then normalized to the value obtained for wild-type Taf1. Error bars represent SD. See also Figure S2.
Relative Protein Leve el
+ Taf1V1049D Time (hrs) Taf1 Pax3 DsRed 0 6 12 24 0
+ Taf1R1070P 6 12 24
1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 6 12
18
24
Chase Time (Hours)
B
Taf1V1049D Taf1R1070P Contro ol Pax3 Taf1 Ub1 / Pax3 Pax3 Ubn / Pax3 Ub1 / Pax3
Pax3 Monoubiquitination Level n
IP: Pax3 (-GFP) Taf1 Input: Pax3 (-GFP) -Taf1 Pax3 (-GFP) IB: Poly-Ub and IB: Mono-Ub (FK-2)
Pax3 +
C
T Taf1V1049D T Taf1R1070P Nickel Agarose Pull-down: T Taf1 Pax3 + C Control
1.0 0.8 08 0.6 0.4 0.2
1049D Taf1 V1
Input:
Pax3 (-GFP) -Taf1
Pax3 Taf1 Ubn / Pax3 Ub1 / Pax3
Taf1
1070P Taf1 R1
alone, cultures treated with both Taf1 and Pax3 siRNA demonstrated a reduced motility, similar to the motility of cells treated with no siRNA (Figures 5C and 5D). These results demonstrate that Taf1 plays a functional role in regulating this Pax3-mediated process, analogous to its regulation of myogenic differentiation. In order to complement the functional studies of the regulation of Pax3 by Taf1 in satellite cell-derived myogenic progenitors, we tested the functional effects of downregulation of Taf1 in embryonic progenitors. Using lineage tracing (see the Experi-
mental Procedures), we were able to monitor myogenic progenitors in limb explants (Figures S4A and S4B). FACS-puried cells from embryonic limbs were highly enriched for myogenic cells (Figures S4C and S4D). Around E10.5, myogenic progenitors, which are Pax3 positive, delaminate from the somitic dermomyotome and migrate into the limb buds (Bober et al., 1994). In the following 2 days, as the cells reach their destination in the limb, most of the myogenic progenitors downregulate Pax3, express MRFs to become myoblasts, and differentiate (Buckingham et al., 2006). The delamination, migration, and proliferation of skeletal muscle progenitors are all dependent on Pax3 function (Relaix et al., 2004). We chose to examine embryonic progenitors at E11.5,
Co ontrol
IB:
Pax3 (-GFP)
0.0
Molecular Cell
A
1.2 12 Relative Protein Level
Figure 4. Functional Regulation of Myogenesis by Taf1
(A) Western blot and quantitative analysis of the level of expression of myogenic differentiation MyoG 0.8 markers in cells with normal or reduced Taf1. Cont siRNA Primary myoblasts were transfected with Taf1 or 0.6 -Actinin Control siRNA (60 nM each) and switched to differTaf1 siRNA 0.4 entiation medium 24 hr after transfection for an MyHC additional 24 hr. Cells were then harvested and 0.2 GAPDH analyzed by immunoblots with anti-Myogenin 0.0 (MyoG), anti-Sarcomeric a-Actinin, and antiMyoG MyHC -Actinin Myosin heavy-chain (MyHC) antibodies. Each is normalized to GAPDH. The levels in control siRNA-treated cells were arbitrarily set to 1. Representative blots are to the left, and quantitation of replicate experiments is shown to the B right. (B) Enhancement of differentiation by Taf1 regulasiRNA Taf1: + + 1.2 tion is mediated by Pax3. Western blot and quansiRNA Pax3: + Cont 1.0 titative analysis of the level of expression of myogenic differentiation markers in myogenic MyoG 0.8 progenitors with reduced Taf1 alone or with Taf1/Pax3 siRNA 0.6 -Actinin reduced Pax3. Primary myoblasts were transTaf1/Cont siRNA 0.4 fected with Taf1 siRNA and either Pax3 or MyHC control siRNA (60 nM each) and switched to 0.2 differentiation medium 24 hr after transfection GAPDH 0.0 for an additional 24 hr. Cells were then harvested MyoG MyHC -Actinin and analyzed by immunoblots with anti-Myogenin (MyoG), anti-Sarcomeric a-Actinin, and MyHC antibodies. Each is normalized to GAPDH protein level, and the levels in Taf1 and Pax3 siRNA-treated cells were arbitrarily set to 1. Representative blots are shown to the left, and quantitation of replicate experiments is shown to the right. Error bars represent SD. See also Figure S3.
siRNA Taf1: + Cont
1.0
R Relative Protein Level
when a majority are still migrating and express a high level of Pax3, and at E12.5, when most embryonic myoblasts have ceased migrating and express a low level of Pax3 and high levels of MRFs (Buckingham et al., 2006). Indeed, in the transition from E11.5 to E12.5, Taf1 levels increased while Pax3 levels declined in puried embryonic myoblasts (Figure 6A). Pax3 protein levels declined much more dramatically than did Pax3 transcript levels (Figures 6A6C). The absence of Pax3 protein in cells with intermediate levels of the transcript is similar to the pattern seen in postnatal progenitors as they progress along the myogenic lineage when Pax3 protein levels are regulated posttranscriptionally by protein degradation (Boutet et al., 2007). Indeed, treatment of embryonic myoblasts with MG132 resulted in much higher steady-state levels of Pax3 protein (Figure 6D). Corresponding to the decline in Pax3 protein levels from E11.5 to E12.5, E12.5 myoblasts expressed higher levels of Myogenin and had lower motility rates than E11.5 myoblasts (Figures 7A and 7B and Figures S5A and S5B). To test whether Taf1 regulates these key Pax3-mediated functions of embryonic myoblasts, we knocked down Taf1 in E12.5 myoblasts when Taf1 is high and when Pax3 is low. Indeed, as in postnatal myogenic progenitors, the reduction of Taf1 in embryonic myoblasts resulted in an increase in the percentage of Pax3+ cells and a decrease in the percentage of Myogenin+ cells (Figure S6A). As with postnatal progenitors, reduction of Taf1 levels and the associated increase in Pax3 levels resulted in enhanced motility of embryonic myoblasts (Figures 7C7E).
These data further support the importance of Taf1 in the regulation of Pax3 as a key regulator of myogenesis. DISCUSSION The results of the present study demonstrate that the UBAC involved in the monoubiquitination of Pax3 is Taf1, a major subunit of the initiation complex TFIID. We show that Taf1 directly interacts with Pax3, is sufcient for the monoubiquitination of Pax3, and regulates Pax3 protein levels. Reduction of Taf1 levels in myogenic progenitors results in increased Pax3 protein. Maintenance of Pax3 protein levels in myogenic progenitors inhibits differentiation, as previously shown (Boutet et al., 2007). It is remarkable that the control of myogenic differentiation is deeply rooted in the regulation of the composition of the core promoter recognition complex with Taf1 as one of its major subunits (Ruppert et al., 1993). Discovery of cell-type-specic TATA-binding protein-associated factors (TAFs) suggest that modied TFIID complexes may be involved in mechanisms that regulate tissue-specic program of gene expression (Guermah et al., 2003; Hochheimer and Tjian, 2003; Hiller et al., 2004; Indra et al., 2005; Chen et al., 2005). While tissue-specic TAFs play an active role by trapping repression complexes (Chen et al., 2005) or by regulating a subset of specic differentiation genes (Indra et al., 2005; Fadloun et al., 2007), all these TAFs coexist with the canonical TFIID complex and seem to add specicity to the TFIID complex. Taf1 is especially important, as
Molecular Cell
A
14 12 Number of Cell ls 10
Control siRNA Taf1 siRNA Pax3 siRNA
B
2.0 1.8 1.6 1.4 1.2 12 1.0 0.8 0.6 0.4 0.2 0.0
Pax3
8 6 4 2 0
No transfection
m/min
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 00 02 04 06 08 10 12 14 16 18 20
m/min
siRNA:
Control Taf1
Taf1 siRNA + Control siRNA
C
14 12 Number of Cells 10 8 6 4 2 0
Taf1 siRNA + Pax3 siRNA Control siRNA + Pax3 siRNA No transfection
D
2.0 1.8 1.6 16 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 00
Control + +
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 m/min
Taf1 siRNA: Pax3 siRNA:
m/min
+ +
Control
Figure 5. Distribution of Cell Motility of Primary Myoblasts as Regulated by Pax3 and Taf1
(A) Primary myoblasts were transfected with Taf1 siRNA, Pax3 siRNA, control siRNA (60 nM each), or lipofectamine alone. Twenty-four hours after transfection, cell motility was assessed by recording cell positions every 5 min for 3 hr. Forty-six cells over three independent experiments were used for analysis. Each point represents the number of cells in the population that exhibited the average motility indicated on the abscissa. Cell populations treated with Taf1 siRNA were signicantly more motile than cell populations treated with control siRNA, Pax3 siRNA, or lipofectamine alone. (B) Scatter plot representation of cell motility by time-lapse microscopy shown in (A). Taf1 siRNA-treated primary myoblasts were signicantly faster than control siRNA-treated myoblasts (*p < 0.0001). Median is indicated as a horizontal bar. (C) Enhancement of migration by Taf1 regulation is mediated by Pax3. Primary myoblasts were transfected with Taf1 siRNA and either Pax3 or control siRNA (60 nM each) or lipofectamine alone. Twenty-four hours after transfection, cell motility was recorded for 3 hr every 5 min. Analysis was identical to that in (A). The knockdown of Pax3 abrogated the enhanced motility by Taf1 knockdown. (D) Scatter plot representation of cell motility by time-lapse microscopy shown in (C). Taf1 siRNA-treated primary myoblasts were signicantly faster than Taf1 and Pax3 siRNA-treated myoblasts (*p < 0.0001). Median is indicated as a horizontal bar. See also Figure S4.
alteration of its activities such as its kinase activity (Siegert and Robbins, 1999) or its HAT activity (Weissman et al., 1998) can alter the formation of the initiation complex. Thus, regulation of Taf1 activities by interactions with activators and transcription factors is integral to the process of transcriptional activation. Upon terminal myogenic differentiation, the TFIID complex is replaced by a novel core promoter recognition apparatus in which Taf1 and other core TFIID subunits disappear and are replaced with other subunits such as Taf3 and TRF3 (Deato and Tjian, 2007). In order for myogenic differentiation to proceed, Taf1, through its UBAC activity, must directly regulate the degradation of the transcription activator Pax3. As such, Taf1 appears to regulate the transition of progenitors to myoblasts and their subsequent differentiation. The regulation of the stability of transcription factors, particularly with regard to transcription factors with acidic rich domains, appears to be important for transcriptional activity. Regions of
the proteins where ubiquitination occur often overlap with transcription activation domains (Salghetti et al., 2000). Although polyubiquitination may be the canonical signal for proteasomal degradation, monoubiquitination appears to enhance transcriptional activity in some cases (Salghetti et al., 2001). In the case of Pax3, it is possible that the process of monoubiquitination enhances the transactivation potential of Pax3 by recruiting proteasome-associated proteins such as the 19S proteasome, which itself appears to be capable of potentiating transcription (Gonzalez et al., 2002). Alternatively, monoubiquitination of Pax3 and subsequent proteasomal degradation could initiate transcription, following the model of activation by destruction proposed when the destruction of the transcription factor is a requirement for the initiation of transcription to occur (Tansey, 2001; Lipford et al., 2005). It remains to be determined where the interaction of Taf1 and Pax3 occurs. If the monoubiquitination of Pax3 is a requirement
Molecular Cell
E E11.5
E E12.5
Figure 6. Pax3 and Taf1 Expression in E11.5 and E12.5 Embryonic Myogenic Progenitors
(A) Expression of Pax3 and Taf1 in Pax3DsRed and Myf5DsRed sorted myogenic cells at E11.5 and E12.5. Cells were harvested and analyzed by immunoblots with antiPax3, anti-Taf1, and anti-GAPDH antibodies. (B) Quantitative analysis of replicate immunoblot studies shown in (A). Protein levels were normalized to GAPDH and then to the values at E11.5. (C) Quantitative analysis by qRT-PCR of Pax3 and Taf1 mRNA levels in progenitor cells isolated from embryos at E11.5 (Pax3DsRed) and at E12.5 (Pax3DsRed and Myf5DsRed). Transcript levels were normalized to GAPDH and then to the values at E11.5. (D) Analysis of Pax3 protein levels in embryonic progenitors after treatment with MG132. E11.5 limbs were harvested and treated for 24 hr with DMSO or MG132 (10 mM). After treatment, limbs were harvested and lysed. Proteins were detected by western blotting with anti-Pax3, anti-Taf1, and anti-GAPDH antibodies. Under these conditions, nearly all of the protein is in the monoubiquitinated form. Error bars represent SD. See also Figure S5.
Pax3DsRed
Pax3DsRed
Myf5DsRed Taf1 Pax3 GAPDH
B
R Relative Pax3 Protein Leve el
R Relative Taf1 Protein Leve el
7 6 5 4 3 2 1 0
1.2 1.0 0.8 0.6 0.4 0.2 0.0 E11.5 (Pax3DsRed) E12.5 (Pax3DsRed) E12.5 (Myf5DsRed)
Pax3 to S5a via Rad23B (Boutet et al., 2007). To this model we now add Taf1 as the UBAC that mediates the monoubiquitination of Pax3 C D (Figure S6B). Due to the pleiotropic activities of Taf1, it is difcult to selectively knock out the UBAC MG132: + activity to establish whether Taf1 is the only enzyme responsible for the monoubiquitination Taf1 of Pax3. To attest to this difculty, the ts13 or 3.0 1.2 Pax3 tsBN462 cell lines, which lack the acetyltrans2.5 1.0 GAPDH ferase activity of Taf1, display cell-cycle arrest 2.0 0.8 and apoptosis (Sekiguchi et al., 1995). Based 1.5 0.6 on a mutation in Drosophila with reduced 1.0 0.4 0.5 UBAC activity (Pham and Sauer, 2000), we 0.2 0.0 0.0 generated homologous mutations in the mammalian Taf1 protein. Like the Drosophila mutant, UBAC activity of the mammalian Taf1 mutants was reduced but not eliminated (Figures for its activity, it is possible that Pax3 monoubiquitination occurs 3A3C). Therefore, it is possible that Pax3 monoubiquitination on the DNA during initiation of transcription of target genes. Taf1 is mediated by proteins with UBAC activity other than Taf1. In summary, our results demonstrate a novel role of Taf1 as a is also known to be found TBP-free on the chromatin but not part of the preinitiation complex in the nucleus (Bertolotti et al., 1996; key regulator of Pax3 monoubiquitination during myogenic Saurin et al., 2001; Lin et al., 2002), suggesting that Taf1 might be differentiation. The molecular mechanisms coordinating myoa direct negative regulator of Pax3 independent of any transcrip- genic lineage progression and differentiation appear to be deeply rooted in the regulation of the composition of the core tional context. Taf1 is responsible for the monoubiquitination of Histone H1 promoter recognition complex with Taf1. These results provide (Pham and Sauer, 2000), a linker histone that binds DNA a better understanding of the regulation of Pax3 and how between two nucleosomes. In this study, we have identied myogenic lineage progression and differentiation are controlled a second substrate, Pax3, for Taf1-mediated monoubiquitina- posttranscriptionally. tion. In the case of Pax3, unlike that of Histone H1, Taf1 mediates monoubiquitination for the degradation of its substrate. As our EXPERIMENTAL PROCEDURES reconstituted ubiquitination assay shows (Figure 2B), Taf1 alone Constructs is sufcient to monoubiquitinate Pax3. Previously we have The HA-hTaf1 mammalian expression construct was provided by Dr. R. Tjian shown that monoubiquitinated Pax3 is recognized by Rad23B (UC Berkeley), and the pEGFP-N3-Luciferase construct was provided by and shuttled to the proteasome by binding of monoubiquitinated Dr. C. Bertoni (UCLA). The pEGFP-N3-Pax3 construct was previously
E11.5 Pax3DsRed Pax3DsRed E12.5
D Myf5DsRed D Pax3DsRed
E11.5
D Pax3DsRed
E12.5
Relative P Pax3 mRNA Levels
Relative Taf1 mRNA Levels T
D Myf5DsRed
Molecular Cell
A
14 12 Num mber of Cells 10 8 6 4 2 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 m/min E11.5 (Pax3DsRed) E12.5 (Pax3DsRed) E12.5 (Myf5DsRed)
B
2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0
* *
m m/min
Pax3DsRed
Pax3DsRed
E11.5
E12.5
C
E12.5 (Pax3DsRed) 15 Taf1 siRNA Number of Cells
D
E12.5 (Myf5DsRed) 15 Taf1 siRNA Number of Cells 10 Control siRNA Control siRNA
10
0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 m/min
0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 m/min
E
2.0 1.8 1.6 1.4 1.2 1.0 0.8 08 0.6 0.4 0.2 0.0
Pax3DsRed
Myf5DsRed
Taf1 siRNA:
m/min
Figure 7. Analysis of the Motility of Embryonic Myogenic Progenitors by Time-Lapse Microscopy

(A) The motility of myogenic cells isolated from limbs of embryos at E11.5 (Pax3DsRed) and E12.5 (Pax3DsRed or Myf5DsRed) was determined by recording cell positions every 5 min for 3 hr. (B) These scatter plots present the distribution of motilities for the graphs shown in (A). Myogenic progenitors at E11.5 were signicantly more motile than those at E12.5 (*p < 0.0001). Median is indicated as a horizontal bar. (C and D) Sorted cells at E12.5 either from Pax3DsRed (C) or Myf5DsRed (D) mice were plated on laminin/collagen, transfected with Taf1 or control siRNA (60 nM each), and assessed for motility by recording cell position every 5 min for 3 hr. Cells treated with Taf1 siRNA were signicantly more motile than cells treated with control siRNA. (E) Scatter plot representation of data in (C) and (D) of myogenic progenitors isolated from embryos at E12.5 (Pax3DsRed or Myf5DsRed) transfected with Taf1 or control siRNA. Taf1 siRNA-treated myoblasts were signicantly more motile than control siRNA-treated myoblasts (*p < 0.0001). Median is indicated as a horizontal bar. See also Figure S6.
Myf5DsRed
Molecular Cell
described (Boutet et al., 2007). To mutate HA-hTaf1 at valine 1049 to generate Taf1V1049D, we used the forward primer 50 -CGCTGGGAAGTGATTGATGATGT GCGCACAATGTCAAC-30 and its reverse complement. To mutate HA-hTaf1 at arginine 1070 to generate Taf1R1070P, we used the forward primer 50 -GCCC GTGGATCACCGTTTTCTGTGGCTGAGCATC-30 and its reverse complement. Both mutations were generated using the QuikChange PCR-Directed Mutagenesis Kit (Stratagene) according to manufacturers instructions. Satellite Cell Isolation, Primary Myoblast Cultures, and Reserve Cell Preparation Satellite cells were puried from bulk bers and activated in vitro as described previously (Conboy and Rando, 2002; Brack et al., 2007). Primary myoblast cultures were maintained as proliferating mononucleated cells or were induced to differentiate as previously described (Quach and Rando, 2006). To prepare reserve cells, primary or C2C12 myoblasts were placed into differentiation medium for 5 days and reserve cells were prepared, according to the method of Kitzmann et al. (Kitzmann et al., 1998), except that cells were fully trypsinized and plated on a new dish. We took advantage of the differential adhesion between bers and reserve cells. After the reserve cells adhered, the cultures were washed twice in PBS and the differentiation medium was added. Quantitative RT-PCR Cells were harvested and washed twice in phosphate buffered saline (PBS). Cells were lysed and total RNA was extracted using Triazol (Invitrogen) according to the manufacturers instructions. Two micrograms (for siRNA-treated myoblasts) of total RNA was reverse transcribed using Superscript II kit (Invitrogen), and quantitative RT-PCR was carried out on a MyiQ real-time PCR (BioRad) using Pax3, Taf1, and GAPDH TaqMan probes (Applied Biosystems). Relative quantication of gene expression normalized to GAPDH was carried using the comparative CT method (Pfaf, 2001). Each measurement was performed in triplicate in three independent experiments. siRNA Transfection of Myogenic Cells Primary myoblasts were transfected with Taf1 or control siRNA (Invitrogen) at 15, 30, 45, or 60 nM. For analysis of differentiation, Taf1 or control siRNA oligonucleotides were used at the highest concentration of 60 nM. In the case of double transfections for differentiation, Taf1 and Pax3 siRNA (Invitrogen) or Taf1 and control siRNA were used at 60 nM each. Twenty-four hours after transfection, cells were harvested or placed in differentiation medium for 24 hr and then harvested. Lysates were analyzed by western blots. In specic experiments as noted, cells were transfected with Taf1 or control siRNA at a concentration of 60 mM and, 24 hr posttransfection, were then treated with MG132 (10 mM) for 4 hr. In Vivo Protein Stability Assay C2C12 myoblasts were transfected with plasmids and treated as indicated. Cycloheximide (5 mM) was added 24 hr after transfection, cells were harvested at different time points, and lysates were analyzed by western blots. The lms were scanned and quantied with ImageJ (http://rsb.info.nih.gov/ij/). In Vivo Ubiquitination C2C12 myoblasts were transfected with pEGFP-N3-Pax3 and with either Taf1 constructs (wild-type (HA-hTaf1) Taf1V1049D, or Taf1R1070P) or with empty vector constructs. Transfected cells were pretreated with 10 mM MG132 for 3 hr to block Pax3 degradation, and extracts were prepared in lysis buffer containing 100 mM MG132, 20 mM ubiquitin aldehyde, and 100 mM N-ethylmaleimide (NEM, Sigma). Lysates were denatured by boiling for 5 min in the presence of 0.1% SDS. Immunoprecipitation was carried out by adding 10 mg of anti-GFP monoclonal antibody (Santa Cruz Biotechnology). For cotransfection of the Pax3, His6-tagged ubiquitin, and HA-hTaf1 wild-type and mutant (Taf1V1049D or Taf1R1070P) expression vectors, cells were treated with 10 mM MG132 for 6 hr and extracts were prepared in CelLytic M reagent (Sigma) lysis buffer containing 100 mM MG132 and 20 mM ubiquitin aldehyde. Histagged proteins were puried on nickel afnity gel using His-Select M afnity Capture Kit (Sigma) according to the manufacturers instructions. Immunoprecipitated and puried proteins were analyzed by western blotting using an
anti-GFP monoclonal antibody and anti-ubiquitin antibodies (FK-2; Biomol International). Recombinant Proteins and Cell-free Ubiquitination Assay Recombinant Pax3 was produced as GST-fusion proteins in E. coli BL21, extracted in MT-PBS (10% glycerol, 1 mM DTT, 0.5 mM PMSF) containing 1% Triton X-100, and puried on glutathione Sepharose resin (GE). GST tags were excised with Thrombin (GE) (5 U/ml for 6 hr at room temperature). Recombinant human Taf1 protein was produced in Sf21 cells with a baculovirus expressing hemagglutinin (HA)-tagged human Taf1 (Orbigen) and extracted using anti-HA antibody resin (Roche). Beads were washed and bound proteins were eluted with 20 ml of SDS loading buffer. Ubiquitination assays were performed after purication of HA-hTaf1 and carried out directly on the agarose bound Taf1 proteins in 50 mM Tris-Cl (pH 7.6), 5 mM MgCl2 at 25 C for 2 hr in the presence of ubiquitin (5 mg) (Biomol International) and Histone H1 (Roche), Histone H2B (Roche), or Pax3 proteins with or without 2 mM ATP. Pull-down assays were performed after purication of HA tag and HA-hTaf1 and carried out in the same buffer as the ubiquitination assay at 4 C for 4 hr in the presence of Pax3 proteins. Reactions were stopped by the addition of sample buffer and subjected to SDS-PAGE followed by transfer and western blot analysis. Coimmunoprecipitation C2C12 myoblasts were transfected with expression vectors for either GFP (pEGFP-N3-Pax3), Taf1 (HA-hTaf1), or respective control vectors. Extracts were prepared in lysis buffer containing 100 mM MG132 (Sigma). Coimmunoprecipitation was carried out by either adding 10 mg of anti-GFP monoclonal antibody (Santa Cruz Biotechnology) or anti-HA antibody resin (Roche). To study endogenous proteins, primary myoblasts were treated with 10 mM MG132 for 4 hr. Coimmunoprecipitation was carried out using 10 mg of antiTaf1 antibody (Santa Cruz Biotechnology). After incubation at 4 C for at least 4 hr, beads were washed three times in lysis buffer and proteins were eluted by the addition of sample buffer and subjected to SDS-PAGE followed by transfer and western blot analysis. Western Blotting Immunoprecipitated and puried proteins were analyzed by western blotting using the following antibodies: anti-GFP monoclonal antibody (1:1000, Clontech), anti-DsRed2 polyclonal antibody (1:1000, Clontech), anti-ubiquitin monoclonal antibody (FK-2; 1:1000), anti-ubiquitin polyclonal antibody (1:1000, Sigma), anti-Pax3 monoclonal antibody (1:100, DSHB), anti-Taf1 polyclonal antibody (1:100, Santa Cruz Biotechnologies), anti-HA monoclonal antibody (1:1000, Roche), anti-Sarcomeric a-Actinin monoclonal antibody (1:500, Sigma), anti-Myosin heavy-chain monoclonal antibody (1:500, Sigma), anti-Myogenin monoclonal antibody (1:500, BD-PharMingen), and antiGAPDH monoclonal antibody (1:5000, Ambion). Time-Lapse Microscopy Primary myoblasts and sorted embryonic myogenic cells were transfected overnight in growth medium with either siRNA oligonucleotides (60 nM each) as indicated. Cells were plated on laminin/collagen (Sigma)-coated dishes at the concentration of approximately 3000 cells/cm2. Cultures were analyzed using a Zeiss Axiovert 200M inverted microscope (Carl Zeiss) tted with an incubation chamber to provide a controlled environment (CTI Controller, Tempcontrol; Carl Zeiss; humidied 5% CO2). Phase contrast images were acquired every 5 min for 3 hr with a Zeiss camera MRm (Carl Zeiss) integrated in the Axiovision system (Carl Zeiss). Tracking of cells and measurements of distances were done with ImageJ (http://rsb.info.nih.gov/ij/) with a manual cell tracker plug-in. A minimum of 46 different cells from at least three independent transfections or at least three different sorts were collected for nal analysis. Embryonic Myogenic Progenitor Isolation To purify myogenic populations that represent both migrating progenitors and differentiating myoblasts, we generated strains of mice in which the reporter gene dsred was expressed specically in those cells. This was accomplished by crossing Z/Red mice, in which DsRed expression is induced in cells (and all
Molecular Cell
their progeny) expressing Cre recombinase (Vintersten et al., 2004), with transgenic lines in which Cre is expressed in the myogenic lineage. Two different Cre-expressing strains were used. The M-Cre line expresses Cre in Pax3expressing progenitors of somitic hypaxial origin (Brown et al., 2005). At E11.5, an enriched population of such progenitors migrates to the limbs (Brown et al., 2005). The Myf5-Cre-NN line expresses Cre in Myf5-expressing progeny of the Pax3-expressing cells, as Pax3 directly activates Myf5 in muscle progenitors of the limb (Haldar et al., 2007). Therefore, at E12.5, the population of Cre-expressing cells in the M-Cre and the Myf5-Cre-NN mice overlap, since previously Pax3-expressing cells are then expressing Myf5. At this stage, these myogenic cell populations are enriched in myoblasts that are no longer migratory and are Pax3 negative and MRF positive (Buckingham et al., 2006). Crossing each Cre driver with the Z/Red line (resulting in two strains that we will refer to as Pax3DsRed and Myf5DsRed) thus yields DsRedexpressing myogenic cells that can be puried at different embryonic stages by FACS and studied in vitro. M-Cre and Myf5-NN-Cre drivers were crossed with heterozygous Z/RED reporter mice. Myogenic cells were isolated from E11.5 or E12.5 embryos. Forelimb and hindlimb buds were dissociated as previously described (Biressi et al., 2007) before sorting using a Vantage Sorter SE (Becton Dickinson). Forward scatter and side scatter parameters were used to gate out cell clumps and debris. Cells dissociated from DsRed-negative littermates were used to set the gating to exclude autouorescence. Sorted myogenic cells were resuspended in Opti-MEM (Invitrogen) supplemented with 20% FBS, 20 mM HEPES, and 5 ng/ml FGF (Peprotech) and plated on laminin/collagen (Sigma) coated dishes. The purity of the sorted cells was evaluated by immunouorescence after 2 days of culture in Dulbeccos modied Eagle medium, 2% horse serum (Invitrogen), 20 mM HEPES. For protein content analysis, sorted cells from 8 E11.5, 9 E12.5 Pax3DsRed, and 15 E12.5 Myf5DsRed embryos were pooled. Statistical Analysis For comparisons of two groups, Students t tests (unpaired, nonparametric, and two-tailed p values) were used. In all gures, error bars represent SD. SUPPLEMENTAL INFORMATION Supplemental Information includes six gures and can be found with this article at doi:10.1016/j.molcel.2010.09.029. ACKNOWLEDGMENTS We thank all the members of the Rando laboratory for help, comments, and discussion. We thank Dr. R. Tjian for the HA-hTaf1 mammalian expression construct. This work was supported by a Development Grant from the Muscular Dystrophy Association to S.C.B. and by grants from the National Institutes of Health (NIH) (AG23806, AR056849, and an NIH Directors Pioneer Award) and the Department of Veterans Affairs (Merit Review) to T.A.R. Received: November 20, 2009 Revised: June 22, 2010 Accepted: September 15, 2010 Published: December 9, 2010 REFERENCES Bajard, L., Relaix, F., Lagha, M., Rocancourt, D., Daubas, P., and Buckingham, M.E. (2006). A novel genetic hierarchy functions during hypaxial myogenesis: Pax3 directly activates Myf5 in muscle progenitor cells in the limb. Genes Dev. 20, 24502464. Belz, T., Pham, A.D., Beisel, C., Anders, N., Bogin, J., Kwozynski, S., and Sauer, F. (2002). In vitro assays to study protein ubiquitination in transcription. Methods 26, 233244. Bertolotti, A., Lutz, Y., Heard, D.J., Chambon, P., and Tora, L. (1996). hTAF(II) 68, a novel RNA/ssDNA-binding protein with homology to the pro-oncopro-
teins TLS/FUS and EWS is associated with both TFIID and RNA polymerase II. EMBO J. 15, 50225031. Biressi, S., Molinaro, M., and Cossu, G. (2007). Cellular heterogeneity during vertebrate skeletal muscle development. Dev. Biol. 308, 281293. Bladt, F., Riethmacher, D., Isenmann, S., Aguzzi, A., and Birchmeier, C. (1995). Essential role for the c-met receptor in the migration of myogenic precursor cells into the limb bud. Nature 376, 768771. Bober, E., Franz, T., Arnold, H.H., Gruss, P., and Tremblay, P. (1994). Pax-3 is required for the development of limb muscles: a possible role for the migration of dermomyotomal muscle progenitor cells. Development 120, 603612. Boutet, S.C., Disatnik, M.H., Chan, L.S., Iori, K., and Rando, T.A. (2007). Regulation of Pax3 by proteasomal degradation of monoubiquitinated protein in skeletal muscle progenitors. Cell 130, 349362. Brack, A.S., Conboy, M.J., Roy, S., Lee, M., Kuo, C.J., Keller, C., and Rando, T.A. (2007). Increased Wnt signaling during aging alters muscle stem cell fate and increases brosis. Science 317, 807810. Brown, C.B., Engleka, K.A., Wenning, J., Min Lu, M., and Epstein, J.A. (2005). Identication of a hypaxial somite enhancer element regulating Pax3 expression in migrating myoblasts and characterization of hypaxial muscle Cre transgenic mice. Genesis 41, 202209. Buckingham, M., Bajard, L., Chang, T., Daubas, P., Hadchouel, J., Meilhac, S., Montarras, D., Rocancourt, D., and Relaix, F. (2003). The formation of skeletal muscle: from somite to limb. J. Anat. 202, 5968. Buckingham, M., Bajard, L., Daubas, P., Esner, M., Lagha, M., Relaix, F., and Rocancourt, D. (2006). Myogenic progenitor cells in the mouse embryo are marked by the expression of Pax3/7 genes that regulate their survival and myogenic potential. Anat. Embryol. (Berl.) 211 (Suppl 1 ), 5156. Chen, X., Hiller, M., Sancak, Y., and Fuller, M.T. (2005). Tissue-specic TAFs counteract Polycomb to turn on terminal differentiation. Science 310, 869872. Ciechanover, A., Orian, A., and Schwartz, A.L. (2000). Ubiquitin-mediated proteolysis: biological regulation via destruction. Bioessays 22, 442451. Conboy, I.M., and Rando, T.A. (2002). The regulation of Notch signaling controls satellite cell activation and cell fate determination in postnatal myogenesis. Dev. Cell 3, 397409. Deato, M.D., and Tjian, R. (2007). Switching of the core transcription machinery during myogenesis. Genes Dev. 21, 21372149. Delni, M., Hirsinger, E., Pourquie, O., and Duprez, D. (2000). Delta 1-activated notch inhibits muscle differentiation without affecting Myf5 and Pax3 expression in chick limb myogenesis. Development 127, 52135224. Dikstein, R., Zhou, S., and Tjian, R. (1996). Human TAFII 105 is a cell typespecic TFIID subunit related to hTAFII130. Cell 87, 137146. Epstein, J.A., Lam, P., Jepeal, L., Maas, R.L., and Shapiro, D.N. (1995). Pax3 inhibits myogenic differentiation of cultured myoblast cells. J. Biol. Chem. 270, 1171911722. Epstein, J.A., Shapiro, D.N., Cheng, J., Lam, P.Y., and Maas, R.L. (1996). Pax3 modulates expression of the c-Met receptor during limb muscle development. Proc. Natl. Acad. Sci. USA 93, 42134218. Fadloun, A., Kobi, D., Pointud, J.C., Indra, A.K., Teletin, M., Bole-Feysot, C., Testoni, B., Mantovani, R., Metzger, D., Mengus, G., and Davidson, I. (2007). The TFIID subunit TAF4 regulates keratinocyte proliferation and has cellautonomous and non-cell-autonomous tumour suppressor activity in mouse epidermis. Development 134, 29472958. Gonzalez, F., Delahodde, A., Kodadek, T., and Johnston, S.A. (2002). Recruitment of a 19S proteasome subcomplex to an activated promoter. Science 296, 548550. Goulding, M., Lumsden, A., and Paquette, A.J. (1994). Regulation of Pax-3 expression in the dermomyotome and its role in muscle development. Development 120, 957971. Gros, J., Manceau, M., Thome, V., and Marcelle, C. (2005). A common somitic origin for embryonic muscle progenitors and satellite cells. Nature 435, 954958.
Molecular Cell
Guermah, M., Ge, K., Chiang, C.M., and Roeder, R.G. (2003). The TBN protein, which is essential for early embryonic mouse development, is an inducible TAFII implicated in adipogenesis. Mol. Cell 12, 9911001. Haldar, M., Hancock, J.D., Cofn, C.M., Lessnick, S.L., and Capecchi, M.R. (2007). A conditional mouse model of synovial sarcoma: insights into a myogenic origin. Cancer Cell 11, 375388. Hicke, L. (2001). Protein regulation by monoubiquitin. Nat. Rev. Mol. Cell Biol. 2, 195201. Hiller, M., Chen, X., Pringle, M.J., Suchorolski, M., Sancak, Y., Viswanathan, S., Bolival, B., Lin, T.Y., Marino, S., and Fuller, M.T. (2004). Testis-specic TAF homologs collaborate to control a tissue-specic transcription program. Development 131, 52975308. Hochheimer, A., and Tjian, R. (2003). Diversied transcription initiation complexes expand promoter selectivity and tissue-specic gene expression. Genes Dev. 17, 13091320. Indra, A.K., Mohan, W.S., Frontini, M., Scheer, E., Messaddeq, N., Metzger, D., and Tora, L. (2005). TAF10 is required for the establishment of skin barrier function in foetal, but not in adult mouse epidermis. Dev. Biol. 285, 2837. Kassar-Duchossoy, L., Giacone, E., Gayraud-Morel, B., Jory, A., Gomes, D., and Tajbakhsh, S. (2005). Pax3/Pax7 mark a novel population of primitive myogenic cells during development. Genes Dev. 19, 14261431. Kitzmann, M., Carnac, G., Vandromme, M., Primig, M., Lamb, N.J., and Fernandez, A. (1998). The muscle regulatory factors MyoD and myf-5 undergo distinct cell cycle-specic expression in muscle cells. J. Cell Biol. 142, 14471459. Lagha, M., Sato, T., Bajard, L., Daubas, P., Esner, M., Montarras, D., Relaix, F., and Buckingham, M. (2008). Regulation of skeletal muscle stem cell behavior by Pax3 and Pax7. Cold Spring Harb. Symp. Quant. Biol. 73, 307315. Lang, D., Lu, M.M., Huang, L., Engleka, K.A., Zhang, M., Chu, E.Y., Lipner, S., Skoultchi, A., Millar, S.E., and Epstein, J.A. (2005). Pax3 functions at a nodal point in melanocyte stem cell differentiation. Nature 433, 884887. Lin, C.Y., Tuan, J., Scalia, P., Bui, T., and Comai, L. (2002). The cell cycle regulatory factor TAF1 stimulates ribosomal DNA transcription by binding to the activator UBF. Curr. Biol. 12, 21422146. Lipford, J.R., Smith, G.T., Chi, Y., and Deshaies, R.J. (2005). A putative stimulatory role for activator turnover in gene expression. Nature 438, 113116. Mizzen, C.A., Yang, X.J., Kokubo, T., Brownell, J.E., Bannister, A.J., OwenHughes, T., Workman, J., Wang, L., Berger, S.L., Kouzarides, T., Nakatani, Y., and Allis, C.D. (1996). The TAF(II)250 subunit of TFIID has histone acetyltransferase activity. Cell 87, 12611270. Nishimoto, T., Sekiguchi, T., Kai, R., Yamashita, K., Takahashi, T., and Sekiguchi, M. (1982). Large-scale selection and analysis of temperature-sensitive mutants for cell reproduction from BHK cells. Somatic Cell Genet. 8, 811824. Pfaf, M.W. (2001). A new mathematical model for relative quantication in real-time RT-PCR. Nucleic Acids Res. 29, e45. Pham, A.D., and Sauer, F. (2000). Ubiquitin-activating/conjugating activity of TAFII250, a mediator of activation of gene expression in Drosophila. Science 289, 23572360.
Quach, N.L., and Rando, T.A. (2006). Focal adhesion kinase is essential for costamerogenesis in cultured skeletal muscle cells. Dev. Biol. 293, 3852. Relaix, F., Rocancourt, D., Mansouri, A., and Buckingham, M. (2004). Divergent functions of murine Pax3 and Pax7 in limb muscle development. Genes Dev. 18, 10881105. Relaix, F., Rocancourt, D., Mansouri, A., and Buckingham, M. (2005). A Pax3/ Pax7-dependent population of skeletal muscle progenitor cells. Nature 435, 948953. Ruppert, S., Wang, E.H., and Tjian, R. (1993). Cloning and expression of human TAFII250: a TBP-associated factor implicated in cell-cycle regulation. Nature 362, 175179. Salghetti, S.E., Muratani, M., Wijnen, H., Futcher, B., and Tansey, W.P. (2000). Functional overlap of sequences that activate transcription and signal ubiquitin-mediated proteolysis. Proc. Natl. Acad. Sci. USA 97, 31183123. Salghetti, S.E., Caudy, A.A., Chenoweth, J.G., and Tansey, W.P. (2001). Regulation of transcriptional activation domain function by ubiquitin. Science 293, 16511653. Saurin, A.J., Shao, Z., Erdjument-Bromage, H., Tempst, P., and Kingston, R.E. (2001). A Drosophila Polycomb group complex includes Zeste and dTAFII proteins. Nature 412, 655660. Sekiguchi, T., Nakashima, T., Hayashida, T., Kuraoka, A., Hashimoto, S., Tsuchida, N., Shibata, Y., Hunter, T., and Nishimoto, T. (1995). Apoptosis is induced in BHK cells by the tsBN462/13 mutation in the CCG1/TAFII250 subunit of the TFIID basal transcription factor. Exp. Cell Res. 218, 490498. Siegert, J.L., and Robbins, P.D. (1999). Rb inhibits the intrinsic kinase activity of TATA-binding protein-associated factor TAFII250. Mol. Cell. Biol. 19, 846854. Tansey, W.P. (2001). Transcriptional activation: risky business. Genes Dev. 15, 10451050. Vintersten, K., Monetti, C., Gertsenstein, M., Zhang, P., Laszlo, L., Biechele, S., and Nagy, A. (2004). Mouse in red: red uorescent protein expression in mouse ES cells, embryos, and adult animals. Genesis 40, 241246. Walker, S.S., Shen, W.C., Reese, J.C., Apone, L.M., and Green, M.R. (1997). Yeast TAF(II)145 required for transcription of G1/S cyclin genes and regulated by the cellular growth state. Cell 90, 607614. Wassarman, D.A., and Sauer, F. (2001). TAF(II)250: a transcription toolbox. J. Cell Sci. 114, 28952902. Wassarman, D.A., Aoyagi, N., Pile, L.A., and Schlag, E.M. (2000). TAF250 is required for multiple developmental events in Drosophila. Proc. Natl. Acad. Sci. USA 97, 11541159. Weissman, J.D., Brown, J.A., Howcroft, T.K., Hwang, J., Chawla, A., Roche, P.A., Schlitz, L., Nakatani, Y., and Singer, D.S. (1998). HIV-1 tat binds TAFII250 and represses TAFII250-dependent transcription of major histocompatibility class I genes. Proc. Natl. Acad. Sci. USA 95, 1160111606. Yang, X.M., Vogan, K., Gros, P., and Park, M. (1996). Expression of the met receptor tyrosine kinase in muscle progenitor cells in somites and limbs is absent in Splotch mice. Development 122, 21632171.
Article
The miR-17-92 MicroRNA Cluster Regulates Multiple Components of the TGF-b Pathway in Neuroblastoma
Pieter Mestdagh,1,11 Anna-Karin Bostrom,2,11 Francis Impens,3,4 Erik Fredlund,1,5,6 Gert Van Peer,1 ` Pasqualino De Antonellis,7 Kristoffer von Stedingk,2 Bart Ghesquiere,3,4 Stefanie Schulte,8 Michael Dews,9 Andrei Thomas-Tikhonenko,9 Johannes H. Schulte,8 Massimo Zollo,7,10 Alexander Schramm,8 Kris Gevaert,3,4 Hakan Axelson,2 Frank Speleman,1,12 and Jo Vandesompele1,12,*
Molecular Cell
for Medical Genetics, Ghent University Hospital, B-9000 Ghent, Belgium of Laboratory Medicine, Center for Molecular Pathology, Lund University, SE-205 02 Malmo, Sweden 3Department of Medical Protein Research, VIB, B-9000 Ghent, Belgium 4Department of Biochemistry, Ghent University, B-9000 Ghent, Belgium 5Department of Oncology, Clinical Sciences, Lund University, SE-205 02 Malmo, Sweden 6CREATE Health, Strategic Centre for Translational Cancer Research, Lund University, SE-205 02 Malmo, Sweden 7Centro di Ingegneria Genetica e Biotecnologia Avanzate (CEINGE), 80131 Naples, Italy 8University Hospital of Essen, 45147 Essen, Germany 9Division of Cancer Pathobiology, Department of Pathology & Laboratory Medicine, The Childrens Hospital of Philadelphia Research Institute and University of Pennsylvania School of Medicine, Philadelphia, PA 19104, USA 10Dipartimento di Biochimica e Biotecnologie Mediche (DBBM), Universita di Napoli, 80131 Naples, Italy ` 11These authors contributed equally to this work 12These authors contributed equally to this work *Correspondence: joke.vandesompele@ugent.be DOI 10.1016/j.molcel.2010.11.038
2Department
1Center
SUMMARY
The miR-17-92 microRNA cluster is often activated in cancer cells, but the identity of its targets remains elusive. Using SILAC and quantitative mass spectrometry, we examined the effects of activation of the miR-17-92 cluster on global protein expression in neuroblastoma (NB) cells. Our results reveal cooperation between individual miR-17-92 miRNAs and implicate miR-17-92 in multiple hallmarks of cancer, including proliferation and cell adhesion. Most importantly, we show that miR-17-92 is a potent inhibitor of TGF-b signaling. By functioning both upstream and downstream of pSMAD2, miR-17-92 activation triggers downregulation of multiple key effectors along the TGF-b signaling cascade as well as direct inhibition of TGF-b-responsive genes.
INTRODUCTION MicroRNAs (miRNAs) belong to a regulatory class of small noncoding RNAs with a fundamental role in numerous aspects of cell biology, such as cell-cycle regulation, apoptosis, differentiation, and maintaining stemness (reviewed in Bartel [2004]). Only 2025 nucleotides (nt) in length, miRNAs function as key molecules in the posttranscriptional repression of gene expression. Upon miRNA assembly in the RNA-induced silencing complex (RISC), binding between the miRNA seed (nt 27 counted from the 50 end of the miRNA) and complementary sites in the
30 untranslated region (30 UTR) of target mRNAs results in degradation of the mRNA or inhibition of translation (reviewed in Bartel [2009]). Based on the 30 UTR site context, algorithms predict that up to 60% of all coding genes are under the control of one or more miRNAs (Friedman et al., 2009). However, these predictions suffer from a high degree of false positives, and to date, only a fraction of miRNA-mRNA interactions have been experimentally validated. In cancer, miRNAs function both as oncogenes or tumor suppressors (reviewed in Calin and Croce [2006]; Esquela-Kerscher and Slack [2006]). Some of these miRNAs were identied as essential components of known cancer pathways, such as the p53-induced miR-34 family (He et al., 2007; Raver-Shapira et al., 2007) or the c-MYC/MYCN-induced miR-17-92 cluster (ODonnell et al., 2005). The oncogenic miR-17-92 cluster consists of six individual miRNAs (miR-17, miR-18a, miR-19a, miR-19b, miR-20a, and miR-92a) located within a polycistronic transcript on human chromosome 13. Gene duplications and deletions eventually resulted in two miR-17-92 paralogs, the miR-106b-25 cluster on chromosome 7 and the miR-106a-363 cluster on chromosome X. Of these clusters, miR-17-92 is the most frequently activated one in cancer. miRNA expression proling studies revealed miR-17-92 overexpression, both in hematopoietic malignancies (such as B cell lymphomas [He et al., 2005]) and solid tumors (including breast, colon, and lung cancer [Castellano et al., 2009; Hayashita et al., 2005; Lanza et al., 2007]) and neuroblastoma (NB) (Mestdagh et al., 2009a). Overexpression can result from amplication of the miR-17-92 locus (He et al., 2005) or direct miR-17-92 transactivation by c-MYC/MYCN (Dews et al., 2010; Fontana et al., 2008; Mestdagh et al., 2009a; ODonnell et al., 2005). The oncogenic nature of miR-17-92 activation is supported by the identication of
Molecular Cell
miR-17-92 Dampens TGF-b Signaling
A
miR-17-92 pathway activity score
600
100
80
100 80
EFS (%)
400
OS (%)
60 40 20
60
200
40
20
0 SL SH MNA
0
0 2
p < 0.001 4 6 8 10 follow-up time (years) 12
0
0 2
p < 0.001 4 6 8 10 follow-up time (years) 75% - 100%
0% - 25%
25% - 50%
50% - 75%
Figure 1. miR-17-92 Cluster Activation Is a Marker for Poor Prognosis

(A) miR-17-92 pathway activity is scored in three clinicogenetic subsets of NB tumors (data set D1, Table S1), MYCN amplied tumors (MNA), MYCN single copy high-risk tumors (SH), and MYCN single-copy low risk tumors (SL) (whiskers: Tukey). miR-17-92 pathway activity score is signicantly higher in MNA versus SH (Mann Whitney, p < 0.05), MNA versus SL (p < 0.0001), and SH versus SL (p < 0.01). (B) Kaplan Meier plots for overall (OS) and event free survival (EFS) based on the pathway activity score of miR-17-92, represented as quartiles. Increased activity of miR-17-92 is proportionally correlated to both poor overall and event-free survival.
miR-17-92 targets with key roles in cell-cycle control and cell death. In particular, miR-17 and miR-20a target the cyclindependent kinase inhibitor CDKN1A (p21), a negative regulator of the G1-S transition (Fontana et al., 2008), and miR-17 targets the proapoptotic BCL2L11 (Bim) (Fontana et al., 2008). In gastric cancer, downregulation of p21 by the miR-17 and miR-20a paralogs miR-106b and miR-93 renders the cells insensitive to TGF-b-induced cell-cycle arrest whereas miR-25 (a miR-92a paralog) inhibits TGF-b-dependent apoptosis through the repression of BCL2L11 (Petrocca et al., 2008). Thus far, the number of identied miR-17-92 targets remains relatively limited, thus precluding a comprehensive understanding of the full oncogenic potential of this miRNA cluster. In a rst step toward this goal, we examined the effects of miR-17-92 cluster activation on the proteome of NB cancer cells. Using quantitative mass spectrometry, we analyzed the response of thousands of proteins upon miR-17-92 activation in NB cells. NB is an excellent model to study the effects of miR-17-92 activation because high-risk NB tumors are characterized by increased MYCN/c-MYC activity either through MYCN amplication or increased c-MYC expression, both resulting in elevated miR-17-92 levels (Mestdagh et al., 2009a). Our results demonstrate that miR-17-92 is implicated in multiple hallmarks of the tumorigenic program, including proliferation and cell adhesion. Most importantly, we dissect the role of miR-17-92 as a potent inhibitor of TGF-b-signaling acting on multiple levels along the signaling cascade. RESULTS miR-17-92 Cluster Activation Is a Marker for Poor Survival In NB, miR-17-92 expression is activated through direct MYCN/ c-MYC promoter binding (Fontana et al., 2008; Mestdagh et al., 2009b). We quantied miR-17-92 expression on a cohort of 95 primary untreated NB tumor samples (data set D1, Table S1;
GEO accession number GSE21713) (Mestdagh et al., 2009a). The activation of the entire miR-17-92 cluster was evaluated by means of a pathway activity score (Fredlund et al., 2008; Mestdagh et al., 2009a). NB tumors were divided into three cohorts, MYCN single copy low-risk tumors (SL), MYCN single copy high-risk tumors (SH), and MYCN amplied tumors (MNA). The miR-17-92 pathway activity was highest in the MNA tumors, followed by the SH tumors and the SL tumors (Figure 1A). Each individual miRNA is upregulated in the MNA samples suggesting that the entire miR-17-92 cluster, rather than a subset of miRNAs, is of potential relevance (Mann Whitney, p < 0.05) (Figure S1A). We next evaluated miR-17-92 pathway activation with respect to NB patient survival. Kaplan-Meier analysis demonstrated that miR-17-92 activity was proportional to overall and eventfree survival (log rank, p < 0.001), underscoring the importance of miR-17-92 activation in NB tumor biology (Figure 1B). Except for miR-19b, expression of the other miRNAs within the miR-17-92 cluster showed similar correlations (Figure S1B). Impact of miR-17-92 Activation on Protein Output To study the regulatory effects of miR-17-92 activation, quantitative mass spectrometry was applied to measure protein response in a cellular model (SHEP-TR-miR-17-92) with tetracycline-inducible miR-17-92 expression (Mestdagh et al., 2009a). This approach provides the most relevant readout as it directly measures the impact of a miRNA on protein output (Baek et al., 2008; Selbach et al., 2008). Average miR-17-92 induction upon tetracycline treatment was in the range of miR-17-92 fold changes between MNA and SL tumors (Figure S2A) (data not shown). Proling of 430 miRNAs revealed no signicant effects on global miRNA expression suggesting that miR-17-92 induction does not affect the processing of other miRNAs (data not shown). SHEP-TR-miR-17-92 cells were differentially labeled using SILAC (stable isotope labeling with amino acids in cell culture) (Ong et al., 2002) and then either treated with tetracycline for 72 hr or left untreated, followed by methionine COFRADIC
Molecular Cell
12C 6
SILAC Arg/Lys
13C 6
SILAC Arg/Lys
B
MESGFTSK from NNMT Ratio L/H : 0.43
- TET
+ TET
lyse cells Mix
lyse cells
Primary RP-HPLC run
COFRADIC
H2O2 oxidation Secondary RP-HPLC runs
EFLFNAIETMPCVK from RRM2 Ratio L/H : 1.038
H N
H C CH CH S
O C 2 2
H N
H C CH CH S CH
O C 2 2 O 3 SDGSTVSVPMMAQTNK from SERPINE1 Ratio L/H : 4.179
CH 3 LC-MS/MS-analysis
peptide identification (MASCOT) peptide & protein quantification (MASCOT Distiller)
Figure 2. Analysis of Global Protein Expression upon miR-17-92 Activation

(A) Tetracycline treated (+TET) and untreated (TET) SHEP-TR-miR-17-92 cells were metabolically labeled using SILAC. Methionine-containing peptides were isolated using COFRADIC technology and subsequently analyzed using LC-MS/MS. (B) Representative LC-MS/MS spectra for an upregulated protein (NNMT), unchanged protein (RRM2), and downregulated protein (SERPINE1).
isolation of methionyl peptides (Gevaert et al., 2002) and identication of these peptides by LC-MS/MS (Figure 2A). Only proteins that were quantied by at least two different peptides over two different proteome analyses (n = 3249) were selected for further analysis (Colaert et al., 2010). Most proteins were in fact quantied by more than two peptides (Figure S3B). Differential protein expression was determined as the average protein ratio of the differentially labeled fractions across the biological replicates (Table S2; Figures 2B and S3C). Based on a foldchange expression cutoff of 0.5 log2 units (see Supplemental Information for cutoff denition), 144 proteins were downregulated upon miR-17-92 activation. To assess whether the measured protein response reects regulatory miR-17-92 effects, we performed an unbiased search for all possible 7-mer motifs (n = 16,384) in the 30 UTR of the downregulated proteins (15th percentile) and compared these to motif occurrence in the 30 UTR of the remaining proteins. We found seven motifs to be overrepresented in the 30 UTR of the downregulated proteins, with the ve most signicant motifs belonging to the miR-17-92 miRNAs: miR-17, miR-19a, miR-19b, miR-20a, and miR-92a (Fisher Exact, p < 0.05, Bonferoni multiple testing correction) (Figure 3A). Strikingly, there was no enrichment for miR-18a seeds, suggesting that miR-18a does not substantially contribute to protein repression upon miR-17-92 activation. Analyses using the 20th percentile gave similar results
(data not shown). Analyses for the 50 UTR and coding sequence (CDS) did not reveal signicant enrichments for miR-17-92 miRNA seed sequences. However, we did observe an enrichment for the 7-mer-m8 seed of miR-17* in the CDS of the downregulated proteins, suggesting that miR-17*-mediated protein repression might depend on CDS binding. To evaluate miR-17-92 seed efciency with respect to protein repression, we plotted the cumulative distribution of protein fold changes for proteins with at least one miR-17-92 30 UTR 6-mer, 7-mer-A1, 7-mer-m8, or 8-mer seed and compared these to proteins without miR-17-92 seeds (Figure 3B). As expected, protein repression was highest in the presence of an 8-mer seed (Kolmogorov-Smirnov, p = 2.20 3 1016) followed by 7-mer-m8 (p = 1.11 3 106), 7-mer-A1 (p = 0.00011) and 6-mer seeds (p = 0.0027). When evaluating each miR-17-92 miRNA separately, we observed similar results for miR-17/miR-20a, miR-19a/miR-19b, and miR-92a (miR-17/mir-20a and miR-19a/ miR-19b were analyzed together as they share identical seeds) (Figure 3C). For miR-18a, the relation between seed occurrence and protein fold change was less pronounced, further supporting our observation that the contribution of miR-18a to miR-17-92mediated protein repression is limited. The fraction of proteins containing at least one miR-17-92 7-8mer seed was highest for proteins that were downregulated at least 2-fold (82%) and decreased to background levels (45%)
Molecular Cell
C
100 90 cummulative fraction (%) 80 70 60 50 40 30 20
miR-17 - miR-20a 100
miR-18a
90
cummulative fraction (%) 80 70 60
50 40
30 20 10 0
10
0 -1 miR-17-92 100 90 cummulative fraction (%) 80 70 no site 6mer site 7mer-A1 site 7mer-m8 site 8mer site cummulative fraction (%) 100 90 cummulative fraction (%) 80 70 60 50 40 30 20 10 0 -1 -0.5 0 0.5 1 protein fold change (log2) -1 1 -0.5 0 0.5 protein fold change (log2) -0.5 0 0.5 1 protein fold change (log2)
-1
-0.5
0.5
protein fold change (log2) miR-92a 100 90 80 70 60 50 40 30 20 10 0 -1 1 -0.5 0 0.5 protein fold change (log2)
miR-19a - miR-19b
60
50
40
30
20
10 0
Figure 3. miR-17-92 Activation Induces Widespread Repression of Targeted Proteins

(A) Overview of all signicantly enriched heptamer motifs in the 30 UTR of transcripts from repressed proteins. The top ve signicantly enriched motifs correspond to miR-17-92 target sites. One motif corresponds to the 7-mer-m8 seed of the miR-302/miR-372 family, which differs in only one base with the 7-mer-m8 seed of miR-17/miR-20a. The last motif did not correspond to any known miRNA nor did it show any overlap with miR-17-92 seeds. (B) The cumulative distribution of protein fold changes upon miR-17-92 activation, calculated for ve different protein subsets: proteins with at least one miR-1792 8-mer site (red), 7-mer-m8 site (blue), 7-mer-A1 site (yellow), 6-mer site (green), and no site (black). (C) Identical analysis as in (B) but for each individual miRNA from the miR-17-92 cluster. The miR-17/miR-20a and the miR-19a/miR-19b were analyzed together as they share identical seeds.
for unchanged proteins (Figure S3A). Robust protein repression was also characterized by the presence of multiple miR-17-92 30 UTR sites per protein (Figure S3B), suggesting that individual miR-17-92 miRNAs cooperate to achieve target repression. This correlation was only observed for 30 UTR sites and not for 50 UTR or CDS sites (Figures S3C and S3D). To further evaluate miRNA cooperation, we analyzed co-occurrence of individual miR-17-92 sites in the 30 UTR of downregulated proteins and compared this to co-occurrence in the 30 UTR of upregulated proteins (used as a reference control set). We identied signicant co-occurrence for miR-17/miR-20a sites and miR-19a/ miR-19b sites conrming cooperation between individual miRNAs (Figure S3E). miR-18a sites almost never occurred in the absence of other miR-17-92 sites (8.33%) and were signicantly associated with miR-17/miR-20a sites (Figure S3E). miR-17-92 Affects Multiple Cancer Pathways To gain insight into the pathways affected by oncogenic miR-17-92 activation, we performed gene set enrichment analysis (GSEA) (Subramanian et al., 2005) using all measured proteins, ranked according to their fold change. Thirty-six gene sets were signicantly enriched in the positive phenotype
(i.e., downregulated proteins) while nine were enriched in the negative phenotype (i.e., upregulated proteins). Of the latter, six were related to increased metabolic activity of the mitochondrial oxidative phosphorylation energy production pathway (Figure S4). In NB, miR-17-92 expression is activated by MYCN/c-MYC transcription factors that have been shown to regulate genes involved in the biogenesis of mitochondria and metabolism (Zhang et al., 2007). Our results now provide evidence that this, at least in part, is mediated through miR-17-92 activation. The contribution of each individual miRNA to the signicant gene lists in the positive phenotype was calculated and visualized as a heatmap (Figure 4A). Among the gene lists enriched in the positive phenotype, which reect direct miR-17-92-regulated pathways, we identied multiple cancer-related processes such as cell proliferation, cell adhesion, TGF-b signaling, estrogen-signaling, and RAS signaling (Figure 4A). Hierarchical clustering reveals a close association between miR-17/ miR-20a- and miR-19a/miR-19b-regulated pathways, reecting the previously observed co-occurrence of these sites. Again, miR-18a clusters further away from the remaining miR-17-92 miRNAs and is characterized by weak gene list associations.
Molecular Cell
- TET + TET
-TET
+TET
T0
T7
T14
T21
E
11.5
SHEP-TR-miR-17-92 SHEP-TR
SHEP-TR
BLI (log10, photons/s)
10.5 9.5 8.5 7.5 6.5
** * *
SHEP-TR-miR- 17-92
0 7 14 21 time of imaging (days)
Figure 4. miR-17-92 Activation Regulates Multiple Cancer Pathways

(A) Heatmap of signicant miRNA-pathway associations, identied through gene set enrichment analysis. The intensity of the association is based on the fraction of genes with at least one 7-mer or 8-mer 30 UTR site. (B) Normalized cell index (mean standard deviation) as a measure for proliferation of tetracycline treated (+TET) and untreated (TET) SHEP-TR-miR-17-92 cells. Treatment was initiated 20 hr post seeding. (C) Evaluation of the cell-cell adhesion of tetracycline treated and untreated SHEP-TR-miR-17-92 cells. Measurements of the relative cluster area for three independent experiments using ImageJ are displayed as bar plots. Upon miR-17-92 activation, the area of the clusters dropped by >50% resulting in more but smaller aggregates. (D) Representative analyses of bioluminescence imaging (BLI) of luciferase positive SHEP-TR-miR-17-92 and SHEP-TR cells injected etherotopically and subcutaneously in the right and left ank of nude athymic mice. Bar scale color indicates the number of photons/s measured by IVIS 3D imaging instrumentation. SHEP miR17-92 cells in vivo are still alive after 21 days post cell subcutaneous implantation. (E) Bioluminescence imaging (BLI) of luciferase positive SHEP-TR-miR-17-92 and SHEP-TR cells injected etherotopically and subcutaneously in the right and left ank of nude athymic mice. Luciferase signals were measured at 0, 7, 14, and 21 days postengraftment and are shown as the mean SEM of ve mice. Signicant differences between SHEP-TR and SHEP-TR-miR-17-92 cells are indicated by * (Student t test, p < 0.05) and ** (Student t test, p < 0.01).
In NB, the oncogenic nature of miR-17-92 has been ascribed to its ability to promote cell proliferation through the regulation of CDKN1A and BCL2L11 (Fontana et al., 2008). GSEA results indicate that miR-17-92 has a much broader inuence and targets different oncogenic pathways. As a proof of concept, we tried to validate the association with increased proliferation
and decreased cell adhesion in the SHEP-TR-miR-17-92 cells. Cell proliferation was evaluated in real-time using the xCELLigence system. Upon miR-17-92 activation, proliferation of SHEP-TR-miR-17-92 cells increased (Figure 4B) and intercellular cell adhesion signicantly decreased (Figure 4C). To evaluate the effect of miR-17-92 activation in vivo we performed
Molecular Cell
TGFB_ALL_UP
TGFB_EARLY_UP
PADUA_TGFB_UP
high miR-17-92
low miR-17-92
high miR-17-92
low miR-17-92
high miR-17-92
low miR-17-92
B
3000 TGF pathway activity score 2500 2000 1500 1000 500 MNSC MNA p < 0.001
C
100 Q1 (0% - 25%) 80 EFS(%) 60 40 20 0 0 5 10 p < 0.0001 15 Q2 (25% - 50%)
Q3 (50% - 75%)
Q4 (75% - 100%)
follow-up time (years)
Figure 5. miR-17-92 Activation Represses the TGF-b Pathway

(A) Gene set enrichment analysis plots for three different TGF-b gene sets showing signicant enrichment among the miR-17-92 repressed proteins. (B) TGF-b pathway activity score in MYCN amplied NB tumors (MNA) and MYCN single-copy NB tumors (MNSC) (data set D2, Table S1). MNA tumors show signicantly lower TGF-b pathway activity (Mann Whitney, p < 0.001) (whiskers: Tukey). (C) Kaplan Meier plot for event free survival (EFS) based on the TGF-b pathway activity score, represented as quartiles (dataset D2, Table S1). Increased activity of miR-17-92 is proportionally correlated to event-free survival.
etherotopic injection of SHEP-TR-miR-17-92 and SHEP-TR (control) cells in the right and left anking site, respectively, of atymic nude mice that were given tetracyclin and visualized tumor cells using bioluminescence imaging. For SHEP-TR cells, the luciferase signal dropped to background levels after 7 days of engraftment, which is in line with previous ndings demonstarting that SHEP cells are not tumorigenic in vivo (Schweigerer et al., 1990) (Figures 4D and 4E). In contrast, SHEP-TR-miR-1792 cells persisted much longer and showed statistically higer luciferase signals at 7, 14, and 21 days, indicating that, although tumorigenesis decreases, miR-17-92 activation signicantly prolongs the engraftment of SHEP cells, probably through increased proliferation and decreased apoptosis, activities previously ascribed to miR-17-92 overexpression (Fontana et al., 2008). Together, these results conrm the relation between miR-17-92 activation and cell proliferation and reveal a role for miR-17-92 in the regulation of cell adhesion, hereby conrming the GSEA results. miR-17-92 Impairs TGF-b Activity GSEA analysis identied three TGF-b-responsive gene sets (Padua et al., 2008; Verrecchia et al., 2001) among the proteins downregulated upon miR-17-92 activation in the SHEP-TRmiR-17-92 cells (Figure 5A). To exclude the possibility that
repression of TGF-b-responsive genes is an artifact of miRNA overexpression, we analyzed eight published protein expression data sets of miRNA overexpression (Baek et al., 2008; Selbach et al., 2008) using GSEA. None of the TGF-b gene lists were signicantly enriched in any of the data sets, suggesting the observed effect to be related to miR-17-92. For a subset of the TGF-b-responsive genes, the measured protein repression was conrmed on the mRNA level using RT-qPCR (Figure S5). We next evaluated this TGF-b signature in NB tumor samples using the pathway activity score of all genes that signicantly contributed to the GSEA results (n = 21). For this purpose, we used the larger Oberthuer data set (Oberthuer et al., 2006) (data set D2, Table S1) to increase the power of our analysis. TGF-b pathway activity was signicantly downregulated in MNA NB tumors that are characterized by high miR-17-92 expression (Mann Whitney, p < 0.001) (Figure 5B), and showed a negative correlation to MYC pathway activity (Spearmans Rank p < 0.01, rho = 0.460). In addition, Kaplan-Meier survival analysis indicates that tumors with low TGF-b pathway activity are characterized by poor event-free survival (log-rank, p < 0.0001) (Figure 5C). To further substantiate the inverse relation between TGF-b target gene expression and miR-17-92 expression, we performed an expression correlation analysis in a subset of 40 of the 95 NB tumors for which also mRNA expression was
Molecular Cell
Figure 6. miR-17-92 Levels and Activity
Inhibits
pSMAD2
-TET
C
pSMAD2
relative protein expression
+TET
D
1.2 1 0.8 0.6 0.4 0.2 0 TET + TET TGF ligand TGF inhibitor + + + + + + +
(A and B) Immunohistochemical detection of phosphorylated SMAD2 protein (pSMAD2) in tetracycline treated (+TET) and untreated (TET) SHEP-TR-miR-17-92 cells. The cell intensity measurement (mean SEM) reveals a signicant decrease in pSMAD2 levels in tetracycline treated cells (Mann Whitney, p < 0.0001). (C) Western blot analysis indicates a strong decrease (2.3-fold) in pSMAD2 levels upon miR-17-92 induction (+TET). (D) The relative luciferase activity of a pSMAD2 reporter construct (mean SEM). Activation of miR-17-92 expression through tetracycline treatment (+TET) results in a signicant (*) decrease in reporter activity after stimulation of the TGF-b pathway with TGF-b1 (TGF-b ligand). TGF-b-inhibitor treatment completely abrogates the reporter activity.
pSMAD2 (60 kD) ACTB (42 kD)
available (data set D3, Table S1) (Mestdagh et al., 2009a). Hierarchical clustering of the correlation coefcients revealed that, indeed, miR-17-92 expression inversely correlates to TGF-b target gene expression (Figure S6A). These results conrm that TGF-b signaling is downregulated in aggressive NB tumors with high miR-17-92 expression and underscore the potential importance of TGF-b activity in NB tumor biology. We next evaluated which components of the TGF-b signaling cascade are controlled by miR-17-92 miRNAs. One important effector of active TGF-b signaling is phosphorylated SMAD2 protein (pSMAD2) that translocates to the nucleus to induce gene transcription. Upon tetracycline treatment of SHEP-TRmiR-17-92 cells, we observed a signicant decrease in nuclear pSMAD2 levels (Mann Whitney, p < 0.0001) (Figures 6A6C). A similar decrease was observed for pSMAD3 levels (data not shown). When SHEP-TR-miR-17-92 cells were transfected with a plasmid containing a SMAD-regulated luciferase reporter ([CAGA]12-Luc) and treated with TGF-b1, a strong activation of the reporter gene was observed (Figure 6D). However, when miR-17-92 expression was activated through tetracycline treatment, reporter gene activation was substantially attenuated (Mann Whitney, p < 0.001) (Figure 6D). When the SHEP-TRmiR-17-92 cells were cultured in the presence of the potent TGFBR1 inhibitor SB431542 (Laping et al., 2002), the SMAD reporter gene activity was completely abrogated (Figure 6D). These results suggest that miR-17-92 activation impairs the TGF-b signaling cascade by acting upstream of pSMAD2. miR-17-92 Affects Multiple Levels of the TGF-b Pathway As decreased pSMAD2 levels are either caused by reduced receptor activity or reduced SMAD2 expression, we quantied TGFBR2 and SMAD2 mRNA expression in the SHEP-TRmiR-17-92 cells. Both TGFBR2 and SMAD2 expression levels
decreased by at least 1.5-fold upon miR-17-92 activation (Figure 7A). SMAD4, the binding partner of pSMAD2, also displayed a decrease in expression upon miR-17-92 activation (Figure 7A). This negative correlation with miR-17-92 expression could be conrmed in primary NB tumor samples for SMAD2 and TGFBR2 (Spearmans Rank, p < 0.01) (Figure 7B), suggesting that miR-1792 regulates their expression. Indeed, both genes contain miR-17-92 binding sites in their 30 UTR and a direct interaction between TGFBR2 and miR-20a has been established (Volinia et al., 2006). This miR-17-92 mediated silencing of TGFBR2 ultimately results in decreased pSMAD2 levels and decreased transcription of the TGF-b-target genes. In total, we identied 13 TGF-b-target genes to be downregulated on the protein level with a log2 fold change < 0.5 (7 out of 20 proteins in the PADUA_TGFB_UP gene set, 3 out of 16 proteins from the TGFB_EARLY_UP gene set, and 7 out of 28 proteins from the TGFB_ALL_UP gene set) (Table S3). As ten of these genes harbor miR-17-92 binding sites in their 30 UTR (Table S3), we wondered whether they might also be targeted directly by miR-17-92. To exclude the effects of miR-17-92 directed inactivation of TGF-b signaling on the expression of TGF-b-responsive genes, we rst treated SHEP-TR-miR-17-92 cells for 4 hr with the TGFBR1 inhibitor SB431542, which completely abrogates TGF-b signaling (Figure 6C). Cells were subsequently treated with tetracycline to activate miR-17-92 expression and harvested at 24 hr and 48 hr after tetracycline treatment. From the six genes that were evaluated, three (CDKN1A, ITGA4, and SERPINE1) were downregulated after 24 hr of TGF-b-inhibitor treatment (t test, p < 0.05), conrming that they are regulated by TGF-b (Figure 7C). The remaining three genes (FNDC3B, ICAM1, and THBS1) did not show any differential expression after 24 hr; however, FNDC3B and THBS1 did respond to TGF-b-inhibitor treatment after 48 hr (data not shown). This suggests that, in NB, these are either not or indirectly responsive to TGF-b signaling (Figure 7D). Upon miR-17-92 activation, the TGF-b-responsive genes were further downregulated (t test, p < 0.001) (Figure 7C),
Molecular Cell
supporting our hypothesis that miR-17-92 also inuences the expression of these genes, independent of its ability to inactivate TGF-b signaling. As expected, the genes that were not responsive to TGF-b inhibition did show decreased expression upon miR-1792 activation (t test, p < 0.001) (Figure 7D). To investigate which specic miRNAs contribute to the repression of the TGF-b-pathway, we overexpressed each miRNA from the miR-17-92 cluster separately and measured the expression of TGF-b-pathway components and target genes. Interestingly, we found that each miRNA contributes to the repression of one or more genes from the TGF-b pathway, suggesting that the entire miR-17-92 cluster, rather than a subset of miRNAs, mediates the repression of TGF-b signaling in NB cells (Figure S6B). Downregulation (log2 fold change < 0.5) upon miRNA transfectection was almost exclusively observed for those genes harboring a 30 UTR seed site for the respective miRNA (Fisher Exact, p < 0.001). We next evaluated whether the miR-17-92-induced downregulation of TGF-b-pathway components is caused by direct binding between miR-17-92 miRNAs and miR-17-92 seed sites in the 30 UTR of TGFBR2, SMAD2, and SMAD4. To this purpose, DLD1DICERhypo cells were transfected with 30 UTR luciferase reporter plasmids in combination with a pre-miR negative control or a miR-17-92 pre-miR for which one or multiple seed sites were present in the 30 UTR of the respective genes. We identied a direct interaction between TGFBR2 and miR-17/20, SMAD2 and miR-18a, and SMAD4 and miR-18a, as evidenced by the signicant decrease in luciferase activity compared to the premiR negative control (t test, p < 0.01, Figure 7E). Other putative miR-17-92 sites in the 30 UTR of TGFBR2 (miR-19a/miR-19b), SMAD2 (miR-19a/miR-19b, miR-92a), and SMAD4 (miR-19a/ miR-19b) did not affect luciferase signals (data not shown). Mutagenesis of the active miRNA seed sites resulted in a signicant rescue of the luciferase signal (t test, p < 0.01), suggesting that the observed effects depend on the presence of the 30 UTR seed site. These results conrm TGFBR2 as a direct miR-17-92 target gene and identify two additional TGF-b-pathway components, SMAD2 and SMAD4, as miR-17-92 target genes. To assess the importance of TGF-b-pathway inhibition in the proliferation phenotype observed upon miR-17-92 activation, we overexpressed SMAD2 and SMAD4 in the presence of activated miR-17-92. SMAD2/SMAD4 overexpression resulted in a 25% decrease in cell growth (t test, p < 0.05), indicating that miR-17-92 accelerated proliferation is, at least in part, depending on the downregulation of the TGF-b pathway. The relatively modest decrease in cell growth is probably explained by the fact that miR-17-92 directly regulates TGF-b target genes in a SMAD2/SMAD4 independent manner. In conclusion, our data demonstrate that miR-17-92 activation triggers a targeted clampdown of TGF-b signaling by acting on multiple key effectors along the signaling cascade, as well as through the direct inhibition of TGF-b-responsive genes, hereby repressing the cytostatic effects of active TGF-b signaling (Figure S7). DISCUSSION Transcriptional activation of the miR-17-92 miRNA cluster by MYC/MYCN transcription factors occurs in multiple tumor enti-
ties, including NB (Hayashita et al., 2005; Mestdagh et al., 2009a; ODonnell et al., 2005). Although the oncogenic nature of miR-17-92 activation is well established, the underlying targets and signaling cascades that are deregulated remain largely elusive. In addition, studies aimed at determining miR-17-92 targets have focused on individual members of the cluster, despite the observation that the entire cluster is activated (Mestdagh et al., 2009b; ODonnell et al., 2005). Here, we have used an unbiased proteomics approach to identify miR-17-92 targeted pathways in a NB tumor model. Direct quantitative measurement of protein expression is preferred over the more straightforward mRNA proling as a high-throughput method for miRNA target identication (Baek et al., 2008; Selbach et al., 2008). Computational analysis of miR-17-92 seeds in the 30 UTR of transcripts from proteins supported the expected enrichment of direct miR-17-92 targets within the list of downregulated proteins detected using mass spectrometry. Moreover, a proportional relationship between seed frequency and fold downregulation was noted. This relationship not only holds for multiple seeds from an individual miR-17-92 miRNA but also for multiple seeds from different miR-17-92 miRNAs, suggesting cooperation between individual miRNAs from the cluster toward target protein repression. miR-17-92 miRNAs have indeed been shown to function in a cooperative and additive manner among others in the regulation of PTEN by miR-17 and miR-19 (Xiao et al., 2008). Our results further indicate that miR-19a/miR-19b and miR-17/ miR-20a sites signicantly co-occur in the 30 UTR of transcripts from several downregulated proteins. As these co-occurring sites were not observed for every possible combination of individual miR-17-92 miRNAs, we hypothesize that in NB, the miRNA components of the miR-17-92 cluster can regulate target expression either individually or in certain combinations with additive effects. However, miR-17-92 function might be highly context and cell-type specic as miR-19 was shown to be both necessary and sufcient to promote MYC-induced lymphomagenesis in the Em-myc mouse B cell lymphoma model (Olive et al., 2009). While the fraction of downregulated proteins was enriched for seeds of miR-17/miR-20a, miR-19a/miR-19b, and miR-92a, enrichment for the miR-18a seed was not detected. Strikingly, miR-18a seeds rarely occur as the only seed(s) in the 30 UTR of a downregulated target and showed little or no correlation to protein fold change. Although this suggests that miR-18a is not substantially contributing to target deregulation, it does not imply that miR-18a lacks functionality, as miR-18a has been shown to regulate important cancer genes such as CTGF in colon cancer and estrogen receptor-a (ESR1) in NB (Dews et al., 2006; Loven et al., 2010). Interestingly, we found miR-18a to regulate both SMAD2 and SMAD4, two key components of the TGF-b-signaling cascade, suggesting that miR-18a substantially contributes to pathway deregulation by regulating a selected set of target genes. When all cluster components were combined, we identied a large number of targeted proteins belonging to diverse cancerrelated pathways. Notably, estrogen receptor signaling was also among the targeted pathways. The fact that we identied such a wide variety of functions in NB cells suggests that miR-17-92 pleiotropy is not only related to different targets in different cell
Molecular Cell
Figure 7. miR-17-92 Targets Multiple Components of the TGF-b Pathway

(A) The relative mRNA expression of TGF-b-responsive genes in tetracycline treated (+TET) and untreated SHEP-TR-miR-17-92 cells (+TET) (mean SEM). (B) Signicant negative correlation between TGFBR2 mRNA expression and miR-17-92 expression and SMAD2 mRNA expression and miR-17-92 expression in primary NB tumors. Spearmans rank rho-values and p values are listed. (C and D) The relative mRNA expression (mean SEM) of a representative set of genes responsive to TGF-b (C) and genes not or (indirectly) responsive to TGF-b (D) in SHEP-TR-miR-17-92 cells that were either untreated, treated with TGF-b inhibitor, or treated with TGF-b inhibitor followed by miR-17-92 activation with tetracycline (TET) for 24 hr. In (C), genes respond to TGF-b-inhibitor treatment (t test, p < 0.05, indicated by *) and show an additional decrease in expression upon combined TGF-b-inhibitor treatment and miR-17-92 activation (t test, p < 0.001, indicated by *). In (D), genes only respond to mR-17-92 treatment (t test, p < 0.001, indicated by **).
Molecular Cell
types but also occurs within cell types. The molecular basis for this observation likely lies within the multiple components of the cluster and the complex interplay between them. mir-17-92-directed regulation of the TGF-b-responsive genes CDKN1A and BCL2L11 in NB cells has been described previously (Fontana et al., 2008). In gastric cancer, members of the miR-106b-25 cluster have also been shown to target CDKN1A and BCL2L11 (Petrocca et al., 2008). Here we comprehensively demonstrate that miR-17-92 dampens TGF-b signaling in a multifaceted way by acting both upstream and downstream of pSMAD2/SMAD4, further underscoring its ability to regulate multiple components of the same pathway. This ability to simultaneously target the components of the signaling cascade, as well as the downstream effectors through multiple miRNAs, allows for tight control of the TGF-b-transcriptional program. Moreover, it offers the cells enormous exibility and plasticity for regulation of different subsets of TGF-b target genes. In NB, enhanced TGF-b signaling, through increased TGFBR2 expression, results in reduced cell growth in vitro and disables the ability of the cells to form tumors in vivo (Turco et al., 2000). Instead, cells assume a terminally differentiated neuronal phenotype and display increased expression of axonal growth-associated protein (GAP43) and neurolaments (Turco et al., 2000). Treatment of NB cells with TGF-b1 induces a similar phenotype (Scarpa et al., 1996). In addition, retinoic acid (RA) induces differentiation of NB cells, known to downregulate MYCN, accompanied by the increased expression of TGF-b1, TGFBR1, TGFBR2, and TGFBR3, resulting in the induction of a negative autocrine TGF-b1 growth regulatory loop (Cohen et al., 1995). We have shown that aggressive NB tumors evade the cytostatic TGF-b pathway through miR-17-92 directed targeting of key components of the pathway as well as downstream effectors. Reactivation of TGF-b signaling through miR-17-92 inhibition could be a promising therapeutic approach, as it would not only result in reactivation of TGFBR2 expression but also relieve the direct miR-17-92-mediated repression of TGF-b-responsive genes.
EXPERIMENTAL PROCEDURES Cell Culture SHEP-TR-miR-17-92 cells (Mestdagh et al., 2009a) were cultured in RPMI (Invitrogen) supplemented with 10% fetal calf serum unless stated otherwise. SHEP-TR-miR-17-92 cells were treated with 2 mg/ml tetracycline (Sigma-Aldrich) to induce miR-17-92 expression (Figure S2A). TGF-b1 (PeproTech) and TGFBR1 inhibitor (SB431542, Sigma-Aldrich) were used at a concentration of 0.25 ng/ml and 2 mM, respectively, unless stated otherwise. COFRADIC Analysis SHEP-TR-miR-17-92 cells were metabolically labeled by growing them in DMEM medium supplemented with dialyzed fetal calf serum and with either heavy lysine and arginine (both 13C6) or with natural, light lysine and arginine (12C6). This stable isotope labeling (SILAC [Ong et al., 2002]) ensures that following trypsin digestion, all generated peptides can be quantied by mass spectrometry (MS, see Supplemental Information). Mass spectrometry data for the forward en reverse experiment are available in Tables S4 and S5
and in the PRIDE database (http://www.ebi.ac.uk/pride; Accession number 14860). mRNA and miRNA Expression Quantication See Supplemental Information for details on mRNA and miRNA quantication and data normalization. miRNA expression data are available in rdml format (Document S2) (Lefever et al., 2009). Immunohistochemistry and Western Blot Briey, SHEP-TR-miR-17-92 cells, tetracycline treated or untreated, were stimulated with TGF-b1 for 4 hr. pSMAD2 activity was evaluated by immunochemistry on cytopreparations or by western blot. See Supplemental Information for detailed experimental procedures. Cell Adhesion and Proliferation Assays Details on cell adhesion and proliferation assays are described in the Supplemental Information. Xenografts SHEP-TR-miR-17-92 and SHEP-TR (control) cells were transfected with a luciferase expressing mammalian vector. Etherotopic xenografts were established in atymic nude mice (n = 5) by injection of 106 SHEP-TR cells subcutaneosly in the left anking site and 106 SHEP-TR-miR-17-92 cells in the right anking site of each individual animal. See Supplemental Information for detailed experimental procedures. CAGA-Luciferase Reporter Assay For luciferase experiments, tetracycline or control treated SHEP-TR-miR-1792 cells were transfected with the (CAGA)12-Luc luciferase reporter vector and assayed for luciferase and renilla activity. See Supplemental Information for detailed experimental procedures. 30 UTR Reporter Assay DLD1Dicerhypo cells were seeded in DMEM (Invitrogen) supplemented with fetal calf serum (10%) at a density of 10,000 cells per well in an opaque 96-well plate. Twenty-four hours after seeding, using DharmaFECT Duo (Dharmacon), cells were cotransfected, either with a combination of a 30 UTR containing pGL4.11[luc2p] vector (Switchgear Genomics), a pRL-TK vector (Promega) for normalization, and a miR-17-92 pre-miR (Ambion) (10 nM) or with a combination of a psi-check2 vector (Promega) containing only part of the 30 UTR and a miR-17-92 pre-miR. Forty-eight hours after transfection, luciferase reporter gene activity was measured using the Dual-Glo Luciferase Assay System (Promega) and a FLUOstar OPTIMA microplate reader (BMG LABTECH). See Supplemental Information for details on plasmid construction and miRNA binding site mutation. Statistics See Supplemental Information for details on all statistical procedures and gene set enrichment analysis. ACCESSION NUMBERS The Gene Expression Omnibus accession number for the mRNA expression data reported in this paper is GSE21713. The PRIDE accession number for the protein expression data reported in this paper is 14860. SUPPLEMENTAL INFORMATION Supplemental Information includes seven gures, ve tables, Supplemental Experimental Procedures, Supplemental References, and an RDML le for
(E) Relative 30 UTR luciferase reporter activity for TGFBR2, SMAD2, and SMAD4, measured in DLD1DICERhypo cells (mean SEM). Plasmids with a wild-type seed site for the active miRNA were introduced in DLD1DICERhypo cells in combination with a pre-miR negative control (NC) or miR-17-92 pre-miR. Luciferase activity is decreased signicantly in the presence of the active miRNA (*) (t test, p < 0.01) and increases signicantly when the seed for the active miRNA is mutated (MUT) (t test, p < 0.01).
Molecular Cell
RT-qPCR proling and can be found with this article online at doi:10.1016/j. molcel.2010.11.038. ACKNOWLEDGMENTS This research was funded by the Fund for Scientic Research (grant number: G.0198.08 and 31511809), the Belgian Kids Fund, and the Stichting tegen Kanker. P.M. is supported by the Ghent University Research Fund (BOF 01D31406). A.-K. B., K.S., and H.A. are supported by grants from the Swedish Childhood Cancer Foundation and the Swedish Cancer Society. E.F. is supported by The Royal Swedish Physiographic Society and the American Cancer Society. B.G. is a Postdoctoral Research Fellow for the Fund for Scientic Research- Flanders (Belgium). The VIB/UGent lab further acknowledges support by a research grant from the Fund for Scientic Research- Flanders (Belgium) (project numbers G.0077.06), the Concerted Research Actions (project BOF07/GOA/012) from the Ghent University, and the Inter University Attraction Poles (IUAP06). NCI grants R01 CA122334 and P30 CA016520 were to A.T.-T., and FP7-Tumic HEALTH-F2-2008-201662, Associazione Italiana contro la lotta al Neuroblastoma Progetto Pensiero and AIRC Tumori Pediatrici 2007-2010 to M.Z. G.V.P. was supported by a BOF research grant (01D35609). This article represents research results of the Belgian program of Interuniversity Poles of Attraction, initiated by the Belgian State, Prime Ministers Ofce, Science Policy Programming. The study was sponsored by the GOA (01G01910). Received: April 22, 2010 Revised: October 6, 2010 Accepted: November 22, 2010 Published: December 9, 2010 REFERENCES Baek, D., Villen, J., Shin, C., Camargo, F.D., Gygi, S.P., and Bartel, D.P. (2008). The impact of microRNAs on protein output. Nature 455, 6471. Bartel, D.P. (2004). MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281297. Bartel, D.P. (2009). MicroRNAs: target recognition and regulatory functions. Cell 136, 215233. Calin, G.A., and Croce, C.M. (2006). MicroRNA signatures in human cancers. Nat. Rev. Cancer 6, 857866. Castellano, L., Giamas, G., Jacob, J., Coombes, R.C., Lucchesi, W., Thiruchelvam, P., Barton, G., Jiao, L.R., Wait, R., Waxman, J., et al. (2009). The estrogen receptor-alpha-induced microRNA signature regulates itself and its transcriptional response. Proc. Natl. Acad. Sci. USA 106, 15732 15737. Cohen, P.S., Letterio, J.J., Gaetano, C., Chan, J., Matsumoto, K., Sporn, M.B., and Thiele, C.J. (1995). Induction of transforming growth factor beta 1 and its receptors during all-trans-retinoic acid (RA) treatment of RA-responsive human neuroblastoma cell lines. Cancer Res. 55, 23802386. Colaert, N., Helsens, K., Impens, F., Vandekerckhove, J., and Gevaert, K. (2010). Rover: a tool to visualize and validate quantitative proteomics data from different sources. Proteomics 10, 12261229. Dews, M., Fox, J.L., Hultine, S., Sundaram, P., Wang, W., Liu, Y.Y., Furth, E., Enders, G.H., El-Deiry, W., Schelter, J.M., et al. (2010). The myc-miR-17$92 axis blunts TGFbeta signaling and production of multiple TGFbeta-dependent antiangiogenic factors. Cancer Res. 70, 82338246. Dews, M., Homayouni, A., Yu, D., Murphy, D., Sevignani, C., Wentzel, E., Furth, E.E., Lee, W.M., Enders, G.H., Mendell, J.T., and Thomas-Tikhonenko, A. (2006). Augmentation of tumor angiogenesis by a Myc-activated microRNA cluster. Nat. Genet. 38, 10601065. Esquela-Kerscher, A., and Slack, F.J. (2006). Oncomirs - microRNAs with a role in cancer. Nat. Rev. Cancer 6, 259269. Fontana, L., Fiori, M.E., Albini, S., Cifaldi, L., Giovinazzi, S., Forloni, M., Boldrini, R., Donfrancesco, A., Federici, V., Giacomini, P., et al. (2008).
Antagomir-17-5p abolishes the growth of therapy-resistant neuroblastoma through p21 and BIM. PLoS ONE 3, e2236. Fredlund, E., Ringner, M., Maris, J.M., and Pahlman, S. (2008). High Myc pathway activity and low stage of neuronal differentiation associate with poor outcome in neuroblastoma. Proc. Natl. Acad. Sci. USA 105, 14094 14099. Friedman, R.C., Farh, K.K., Burge, C.B., and Bartel, D.P. (2009). Most mammalian mRNAs are conserved targets of microRNAs. Genome Res. 19, 92105. Gevaert, K., Van Damme, J., Goethals, M., Thomas, G.R., Hoorelbeke, B., Demol, H., Martens, L., Puype, M., Staes, A., and Vandekerckhove, J. (2002). Chromatographic isolation of methionine-containing peptides for gelfree proteome analysis: identication of more than 800 Escherichia coli proteins. Mol. Cell. Proteomics 1, 896903. Hayashita, Y., Osada, H., Tatematsu, Y., Yamada, H., Yanagisawa, K., Tomida, S., Yatabe, Y., Kawahara, K., Sekido, Y., and Takahashi, T. (2005). A polycistronic microRNA cluster, miR-17-92, is overexpressed in human lung cancers and enhances cell proliferation. Cancer Res. 65, 96289632. He, L., Thomson, J.M., Hemann, M.T., Hernando-Monge, E., Mu, D., Goodson, S., Powers, S., Cordon-Cardo, C., Lowe, S.W., Hannon, G.J., and Hammond, S.M. (2005). A microRNA polycistron as a potential human oncogene. Nature 435, 828833. He, L., He, X., Lim, L.P., de Stanchina, E., Xuan, Z., Liang, Y., Xue, W., Zender, L., Magnus, J., Ridzon, D., et al. (2007). A microRNA component of the p53 tumour suppressor network. Nature 447, 11301134. ` Lanza, G., Ferracin, M., Gafa, R., Veronese, A., Spizzo, R., Pichiorri, F., Liu, C.G., Calin, G.A., Croce, C.M., and Negrini, M. (2007). mRNA/microRNA gene expression prole in microsatellite unstable colorectal cancer. Mol. Cancer 6, 54. Laping, N.J., Grygielko, E., Mathur, A., Butter, S., Bomberger, J., Tweed, C., Martin, W., Fornwald, J., Lehr, R., Harling, J., et al. (2002). Inhibition of transforming growth factor (TGF)-beta1-induced extracellular matrix with a novel inhibitor of the TGF-beta type I receptor kinase activity: SB-431542. Mol. Pharmacol. 62, 5864. Lefever, S., Hellemans, J., Pattyn, F., Przybylski, D.R., Taylor, C., Geurts, R., Untergasser, A., and Vandesompele, J.; RDML consortium. (2009). RDML: structured language and reporting guidelines for real-time quantitative PCR data. Nucleic Acids Res. 37, 20652069. Loven, J., Zinin, N., Wahlstrom, T., Muller, I., Brodin, P., Fredlund, E., Ribacke, U., Pivarcsi, A., Pahlman, S., and Henriksson, M. (2010). MYCN-regulated microRNAs repress estrogen receptor-alpha (ESR1) expression and neuronal differentiation in human neuroblastoma. Proc. Natl. Acad. Sci. USA 107, 15531558. Mestdagh, P., Fredlund, E., Pattyn, F., Schulte, J.H., Muth, D., Vermeulen, J., Kumps, C., Schlierf, S., De Preter, K., Van Roy, N., et al. (2009a). MYCN/cMYC-induced microRNAs repress coding gene networks associated with poor outcome in MYCN/c-MYC-activated tumors. Oncogene 29, 13941404. Mestdagh, P., Van Vlierberghe, P., De Weer, A., Muth, D., Westermann, F., Speleman, F., and Vandesompele, J. (2009b). A novel and universal method for microRNA RT-qPCR data normalization. Genome Biol. 10, R64. ODonnell, K.A., Wentzel, E.A., Zeller, K.I., Dang, C.V., and Mendell, J.T. (2005). c-Myc-regulated microRNAs modulate E2F1 expression. Nature 435, 839843. Oberthuer, A., Berthold, F., Warnat, P., Hero, B., Kahlert, Y., Spitz, R., Ernestus, K., Konig, R., Haas, S., Eils, R., et al. (2006). Customized oligonucleotide microarray gene expression-based classication of neuroblastoma patients outperforms current clinical risk stratication. J. Clin. Oncol. 24, 50705078. Olive, V., Bennett, M.J., Walker, J.C., Ma, C., Jiang, I., Cordon-Cardo, C., Li, Q.J., Lowe, S.W., Hannon, G.J., and He, L. (2009). miR-19 is a key oncogenic component of mir-17-92. Genes Dev. 23, 28392849. Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H., Pandey, A., and Mann, M. (2002). Stable isotope labeling by amino acids in cell culture,
Molecular Cell
SILAC, as a simple and accurate approach to expression proteomics. Mol. Cell. Proteomics 1, 376386. Padua, D., Zhang, X.H., Wang, Q., Nadal, C., Gerald, W.L., Gomis, R.R., and Massague, J. (2008). TGFbeta primes breast tumors for lung metastasis seeding through angiopoietin-like 4. Cell 133, 6677. Petrocca, F., Visone, R., Onelli, M.R., Shah, M.H., Nicoloso, M.S., de Martino, I., Iliopoulos, D., Pilozzi, E., Liu, C.G., Negrini, M., et al. (2008). E2F1-regulated microRNAs impair TGFbeta-dependent cell-cycle arrest and apoptosis in gastric cancer. Cancer Cell 13, 272286. Raver-Shapira, N., Marciano, E., Meiri, E., Spector, Y., Rosenfeld, N., Moskovits, N., Bentwich, Z., and Oren, M. (2007). Transcriptional activation of miR-34a contributes to p53-mediated apoptosis. Mol. Cell 26, 731743. Scarpa, S., Coppa, A., Ragano-Caracciolo, M., Mincione, G., Giuffrida, A., Modesti, A., and Colletta, G. (1996). Transforming growth factor beta regulates differentiation and proliferation of human neuroblastoma. Exp. Cell Res. 229, 147154. Schweigerer, L., Breit, S., Wenzel, A., Tsunamoto, K., Ludwig, R., and Schwab, M. (1990). Augmented MYCN expression advances the malignant phenotype of human neuroblastoma cells: evidence for induction of autocrine growth factor activity. Cancer Res. 50, 44114416. Selbach, M., Schwanhausser, B., Thierfelder, N., Fang, Z., Khanin, R., and Rajewsky, N. (2008). Widespread changes in protein synthesis induced by microRNAs. Nature 455, 5863. Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and
Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression proles. Proc. Natl. Acad. Sci. USA 102, 1554515550. Turco, A., Scarpa, S., Coppa, A., Baccheschi, G., Palumbo, C., Leonetti, C., Zupi, G., and Colletta, G. (2000). Increased TGFbeta type II receptor expression suppresses the malignant phenotype and induces differentiation of human neuroblastoma cells. Exp. Cell Res. 255, 7785. Verrecchia, F., Chu, M.L., and Mauviel, A. (2001). Identication of novel TGF-beta /Smad gene targets in dermal broblasts using a combined cDNA microarray/promoter transactivation approach. J. Biol. Chem. 276, 1705817062. Volinia, S., Calin, G.A., Liu, C.G., Ambs, S., Cimmino, A., Petrocca, F., Visone, R., Iorio, M., Roldo, C., Ferracin, M., et al. (2006). A microRNA expression signature of human solid tumors denes cancer gene targets. Proc. Natl. Acad. Sci. USA 103, 22572261. Xiao, C., Srinivasan, L., Calado, D.P., Patterson, H.C., Zhang, B., Wang, J., Henderson, J.M., Kutok, J.L., and Rajewsky, K. (2008). Lymphoproliferative disease and autoimmunity in mice with increased miR-17-92 expression in lymphocytes. Nat. Immunol. 9, 405414. Zhang, H., Gao, P., Fukuda, R., Kumar, G., Krishnamachary, B., Zeller, K.I., Dang, C.V., and Semenza, G.L. (2007). HIF-1 inhibits mitochondrial biogenesis and cellular respiration in VHL-decient renal cell carcinoma by repression of C-MYC activity. Cancer Cell 11, 407420.
Article
Structural Insights into Ligand Recognition by a Sensing Domain of the Cooperative Glycine Riboswitch
Lili Huang,1,2 Alexander Serganov,1,2 and Dinshaw J. Patel1,*
Biology Program, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA authors contributed equally to this work *Correspondence: pateld@mskcc.org DOI 10.1016/j.molcel.2010.11.026
2These 1Structural
Molecular Cell
SUMMARY
Glycine riboswitches regulate gene expression by feedback modulation in response to cooperative binding to glycine. Here, we report on crystal structures of the second glycine-sensing domain from the Vibrio cholerae riboswitch in the ligand-bound and unbound states. This domain adopts a threehelical fold that centers on a three-way junction and accommodates glycine within a bulge-containing binding pocket above the junction. Glycine recognition is facilitated by a pair of bound Mg2+ cations and governed by specic interactions and shape complementarity with the pocket. A conserved adenine extrudes from the binding pocket and intercalates into the junction implying that glycine binding in the context of the complete riboswitch could impact on gene expression by stabilizing the riboswitch junction and regulatory P1 helix. Analysis of riboswitch interactions in the crystal and footprinting experiments indicates that adjacent glycine-sensing modules of the riboswitch could form specic interdomain interactions, thereby potentially contributing to the cooperative response.
INTRODUCTION The ten-atom amino acid glycine molecule is one of the smallest organic compounds, which apart from direct use in protein synthesis serves as a precursor for other molecules and plays one of the central roles in cellular carbon ow. In bacteria, glycine is predominantly synthesized from serine by the product of the glyA gene and is cleaved by proteins encoded in the gcv operon. In Escherichia coli, glycine metabolism is tightly controlled at the level of transcription initiation by global regulators, a specic transcriptional factor, GcvA, and its glycine-binding partner protein GcvR (Heil et al., 2002). The control of the glyA gene is partially preserved (Dartois et al., 1997; Saxild et al., 2001), while Gcv-like regulators of catabolism are missing in Bacillus subtilis (Abreu-Goodger et al., 2004). It came as a surprise that RNA possesses sufcient structural sophistication to specically
recognize glycine and to employ this recognition for riboswitch-based feedback control of glycine cleavage genes in B. subtilis and other bacteria (Mandal et al., 2004). Riboswitches have emerged as one of the most widespread and important gene regulatory systems in bacteria with representatives found in archaea and eukaryotes (Nudler and Mironov, 2004; Winkler and Breaker, 2005). Riboswitches are cis-acting structured RNA segments typically located in the 50 -untranslated regions (UTRs) of bacterial mRNAs, where they function as both sensors of cellular metabolites and effectors of the regulatory response (Serganov and Patel, 2007). Glycine riboswitches directly bind glycine present in cells at overthreshold concentrations by using evolutionarily conserved sensing domains. Most glycine riboswitches function as rare gene activators that turn on expression of glycine cleavage genes (Mandal et al., 2004). The second frequently observed glycine riboswitch controls carbon metabolism through activation of the malate synthase gene glcB in a marine bacterium (Tripp et al., 2009). Glycine riboswitches are also colocalized with other glycine-associated genes in some bacteria, and along with thiamine pyrophosphate (TPP) and cobalamin riboswitches, constitute one of the most abundant riboswitches (Barrick and Breaker, 2007). Unexpectedly, the majority of glycine riboswitches are composed of two moderately similar adjacently positioned sensing domains connected by a short linker and followed by a single expression platform (Mandal et al., 2004). Although each sensing domain is capable of specic binding to a separate glycine molecule, tandem sensors in B. subtilis and V. cholerae gcvT riboswitches display a concerted response to glycine, characterized by a sigmoidal ligand-binding curve. Such binding is best described in terms of positive cooperativity between riboswitch-sensing domains that in turn ensures a strong riboswitch response to small changes in glycine concentration. Cooperativity was rst validated in the interaction of oxygen with hemoglobin and was shown later to contribute to ligand binding to multisubunit proteins and the assembly of protein complexes. In nucleic acids, cooperativity is involved in the formation of paired regions (Siegfried and Bevilacqua, 2009), RNA folding (Sattin et al., 2008), and assembly of RNA- and DNA-protein complexes (Recht and Williamson, 2004). Despite frequent utilization by proteins, biologically relevant cooperative binding of small organic molecules to natural RNAs has not been observed prior to the discovery of glycine riboswitches, although its feasibility was demonstrated by a rationally
Molecular Cell
Glycine Riboswitch Structure
Figure 1. Sequence and Structure of the V. cholerae Riboswitch

(A) Secondary structure schematics of the glycine riboswitch with tandem sensing domain arrangement. Nucleotides conserved in R 95% and R 75% sequences are in red and blue, respectively. (B) Crystal structure-based schematic of the VCII RNA fold. The bound glycine is in red. Dashes and circles indicate Watson-Crick and noncanonical base pairs. Key tertiary stacking interactions are shown as blue dashed lines. (C) Overall crystal structure of VCII RNA in a ribbon representation. (D) Zoomed-in view of the three-way junction. Green spheres depict Mg2+ cations. (E) Intercalation of A33 into the junctional region. The RNA is shown in stick representation with color scheme of atoms (nitrogen in blue, oxygen in red, phosphorus in yellow, and carbon in arbitrary colors). Putative hydrogen bonds are shown with black dashed lines. (F) Superposition of the three-way and four-way junctions of the glycine (light pink) and SAM-I (light blue) riboswitches, respectively. The rmsd is 1.48 A.
designed ribozyme that required sequential binding of two effector molecules for its activity (Jose et al., 2001). It should also be noted that some tandem glycine riboswitches do not display strong cooperativity (Tripp et al., 2009). Much effort has already been directed toward the dissection of the ligand-binding properties and the molecular mechanism of cooperative binding in glycine-sensing riboswitches (Serganov and Patel, 2009). Inline probing experiments revealed similar regions of glycine-modulated cleavages, despite different junctional architectures of the sensing domains (Figure 1A) (Kwon and Strobel, 2008; Mandal et al., 2004). The majority of cleavage reductions are clustered within the evolutionarily conserved J1/2, J3/1, J3/3a, and J3a/3 regions, which constitute the core of the riboswitch fold. Nucleotides from these and adjacent regions were also implicated in glycine binding and/or cooperativity by the nucleotide analog interference mapping (NAIM) approach (Kwon and Strobel, 2008). Specically, NAIM and mutagenesis
studies suggested that the minor groove of P1 helix of sensing domain I and the major groove of P3a helix from both sensing domains appear to participate in cooperative tertiary interactions. Small-angle X-ray scattering (SAXS) and hydroxyl radical footprinting showed that glycine binding in the presence of Mg2+ cations signicantly changed the shape of the tandem riboswitch (Lipfert et al., 2007). After ligand binding, the riboswitch became more compact, thus suggesting that sensing domains might be positioned close to each other in the ligand-bound state. However, none of these experiments were able to identify the glycine-binding sites and their relation to riboswitch structure, nor were they able to determine directly whether the pair of sensing domains could interact with each other. Therefore, the principles underlying riboswitch cooperativity are currently not understood at the molecular level. To clarify outstanding issues concerning the glycine riboswitch structure and ligand-binding cooperativity, we determined crystal
Molecular Cell
Table 1. Data Collection and Renement Statistics, Glycine-Bound VCII RNA Crystal PDB ID Data Collection Space group Cell dimensions (a,b,c [A]) Wavelength (A) Resolutiona Rmerge (%)a <I> /s(I)
a
Soaked in 2 mM [Ir(NH3)6]Cl3 3OWZ P3221 82.7, 82.7, 149.5 Peak 1.1053 20.002.95 (3.002.95) 9.0 (38.5) 38.0 (6.1) 99.9 (100.0) 12,921 (620) 6.9 (7.1) 23 0.56/0.36 20.02.95 11,610 1,280 21.0 / 23.1 3,703 10 171 58.5 44.5 37.4 0.005 0.976 0.272 Inection 1.1056 20.002.95 (3.002.95) 7.5 (47.7) 35.0 (4.6) 99.9 (100.0) 12,945 (632) 6.9 (7.0) Remote (L) 1.1256 20.003.00 (3.053.00) 7.0 (52.5) 33.0 (4.6) 99.9 (99.8) 12,318 (608) 6.9 (7.0) Remote (H) 1.0859 20.003.05 (3.103.05) 8.4 (42.3) 33.0 (6.0) 99.9 (99.8) 11,724 (548) 6.9 (7.1)
Native Cocrystals 3OWI P3221 83.8, 83.8, 198.7 1.0809 20.002.85 (2.952.85) 5.9 (57.2) 58.2 (5.5) 98.6 (98.5) 19,420 (1,893) 14.9 (15.4)
Native Soaking 3OWW P3221 83.5, 83.5, 198.5 1.0750 20.002.80 (2.902.80) 5.5 (45.1) 41.2 (4.1) 99.3 (99.4) 20,267 (1,983) 6.0 (6.1)
Completeness (%)a Unique reectionsa Redundancya Phasing Number of [Ir(NH3)6] 3+ sites Figure of merit (acentric/centric) Structure Renement Resolution (A) Working set reections Test set reections Rfactor / Rfree (%) Number of Nonhydrogen Atoms RNA Glycine Cations Average B-Factors (A2) RNA Glycine Cations Rmsd Values Bond lengths (A) Bond angles ( ) Estimated coordinate errorb
a b
20.002.85 18,400 996 20.8 / 23.6 3,807 10 33 36.8 34.1 74.6 0.005 1.024 0.253
20.002.80 18,182 2,065 20.4 / 23.6 3,785 10 33 33.6 28.5 54.2 0.005 1.003 0.248
Values for the highest resolution shell are in parentheses. Estimated coordinate error based on maximum likelihood was calculated with REFMAC (Murshudov et al., 1997).
structures of the second sensing domain from the V. cholerae riboswitch in the free and glycine-bound states. These structures coupled to biochemical and biophysical data provide insights into glycine recognition and the mechanism of riboswitch function and suggest the existence of interdomain interactions that could contribute to cooperative glycine binding in the context of the complete riboswitch. RESULTS AND DISCUSSION Riboswitch Structure Conforms to a Three-Way Junction-Based Architecture We have determined the 2.85 A crystal structure of domain II (designated as VCII RNA) from the gcvTtype glycine riboswitch identied in the 50 -UTR of the V. cholerae VC1422 gene encoding
a putative amino acid transporter (Mandal et al., 2004) (see Experimental Procedures and Table 1). The architecture of the glycine-bound VCII RNA conforms to a three-way junctional fold that comprises three helical stems, P1, P2, and P3/3a/3b, connected by joining regions J1/2 and J3/1 (Figures 1B and 1C), with the glycine-binding pocket located above the junction. The most notable feature of the scaffold, not predicted by phylogeny (Figure 1A), is that P2 helix, not P1, stacks under P3. This continuous stacking alignment is supported by stacking of G29 from P3 and G12-C28 base pair from P2 (Figure 1D). P1 is positioned at a roughly 55 angle relative to P3 and connects with the P2/P3 stem through J1/2 and J3/1. J1/2 provides a smooth transition between P1 and P2, while J3/1 connects P1 and P3 via a sharp turn that is facilitated by splayed-apart nucleotides (nt) C74 and A75 (Figure 1D) and
Molecular Cell
stabilized by Mg2+ cations M4 and M3 (Figure 1D and Figure S1A, available online). The junctional region is tied to P2 via a segment of J3/1 and to P3 and J3/3a via J1/2. These contacts contribute to the juxtaposition of the P2/P3 stack and joining segments and help constrain the relative orientation of P1 and P2/P3. The junction adopts an intricate extended fold that combines nucleotides from J1/2, J3/1, P2, and P3 and is based on two stacks formed with participation of cross-strand stacking interactions (Figure 1D). The rst base stack joins residues from three regions, J1/2 (A11, G10, and A9), J3/3a (A33), and J3/1 (A81), and continues into P1. The key feature of this alignment is highly conserved A33, which extrudes from J3/3a and intercalates between two junctional segments. The second stacking group begins at A13 in P2, spans J3/1 from A75 to G80, and extends toward G8 in J1/2 and P1. In proximity to P2, the two stacks spread out and are brought together by three noncanonical purine base pairs, G10,G79, A9,G80, and G8,A81, near P1 (Figures 1B and 1D). The junction is reinforced by tertiary pairing interactions that include packing of the A11-G10 segment with the minor groove of P3 (Figures S1BS1D) and a set of hydrogen bonds formed by intercalated A33 with bases and sugars of J1/2 and J3/1 (Figure 1E). Nucleotides involved in tertiary contacts are evolutionarily conserved, while the A75-A76-A77 region of the J3/1 stack is nonconserved and can accommodate an additional helix as predicted in domain I. Although architectures based on three-way junctions (Lescoute and Westhof, 2006) are a common feature of riboswitches, the VCII fold is distinct from the three-way junctions of other widespread riboswitches. Unlike the glycine riboswitch, where P2 helix stacks under P3, in purine (Batey et al., 2004; Serganov et al., 2004) and TPP riboswitches (Edwards and Ferre-DAmare, 2006; Serganov et al., 2006; Thore et al., 2006), P2 and P3 are aligned in parallel and anchored to each other through longrange tertiary contacts, while the regulatory P1 helix is colinearly stacked with one of these helices. Unexpectedly, the three-way junction of the glycine-sensing domain strikingly resembles the core of the four-way junction in the S-adenosyl methionine (SAM-I) riboswitch (Montange and Batey, 2006) (Figure 1F and Figure S1E). Interestingly, despite placement of the ligands above the central portion of the junction in both the SAM-I and glycine-sensing riboswitches, SAM-I directly stabilizes the junction through extensive contacts, while glycine is extruded from the interior of the junction and contributes indirectly to its organization. Glycine-Binding Pocket Is Adjacent to the Junction The experimental electron-density map assigned to bound glycine was located in the conserved internal loop J3/3a-J3a/ 3, a segment whose inline-probing cleavage pattern was modulated by bound glycine (Kwon and Strobel, 2008; Mandal et al., 2004) (Figure S2A). The helical structure of adjacent regions is distorted to create a nested pocket within the internal loop for the bound ligand through formation of two stacked major-groove A34,A71,G32 and (G35-C70),U69 base triples (Figures 2A and 2B). The oor of the pocket is built by the rst triple centered on A71, which uses its sugar edge for interaction with A34 and N6-amino group for pairing with the Hoogsteen edge of G32 (Figure 2B and Figure S2B). The second triple is positioned above
the rst triple and together with the backbone of the A33-A34G35 segment encircles the bound glycine (Figures 2A and 2B). The roof of the pocket is formed by the Watson-Crick G36-C68 base pair (Figure 2A and Figure S2C) that nearly seals the ligand-binding site and by the backbone of the A65-C66 segment that additionally contributes to formation of the pocket wall. The bases of A33 and A65 are ipped out of the pocket, and while A33 participates in formation of the three-way junction, A65 makes stacking contacts with the second molecule within the asymmetric unit. The glycine-binding pocket contains two metals, M1 and M2, assigned to hydrated Mg2+ cations based on the octahedral coordination geometry and the anomalous signal from Mn2+ cations that mimic Mg2+. Both cations are positioned in the vicinity of the backbone of the A33-A34-G35 segment (Figures 2A and 2B). M1 buttresses the compact conformation of the pocket through inner-sphere coordination with the nonbridging phosphate oxygens of A34 and G35 and outer-sphere contacts with A34, G35, and G36. M2 locks up the pocket by direct coordination to the 20 -OH group of G32 and the nonbridging phosphate oxygens of A33 and C66. Water-mediated interactions with bases of U67, C68, and U69 further constrain the dimensions of the pocket. Two more metals, M5 and M7, assigned to hydrated Mg2+ cations, reinforce the RNA structure around the binding pocket (Figure 2A). To conrm the location of the bound glycine, we determined the 3.05 A structure of VCII RNA in the unbound state (Table 2). As reported earlier for the lysine and SAM-I riboswitches (Garst et al., 2008; Serganov et al., 2008; Stoddard et al., 2010), the ligand-free VCII RNA structure superpositioned well with the glycine-bound structure (overall root-mean-square deviation [rmsd] is 0.96 A). However, the map lacked the electron density corresponding to glycine and high-occupancy cations M2 and M7 (Figures 2C2E). The subtle shifts of pocket nucleotides on strands held by M2 resulted in a slight opening of the pocket, as indicated by an increase in the distance from $4.7 to $6.2 A between phosphorus atoms of C66 and A33 (Figure 2E). It should be emphasized that the ligand-free structure most likely represents one conformation that crystallizes from an ensemble of conformations present in solution prior to glycine binding. Crystallization of the ligand-free glycine riboswitch was facilitated by high (80 mM) Mg2+ concentration in the crystallization solution and by the choice of a stable RNA construct, which lacked the expression platform and had a long P1 helix that closes the junction. The unbound VCII RNA can be converted to the glycine-bound form by soaking the ligand-free crystals in glycine-containing solution. The structure of the glycine-soaked VCII RNA was practically identical to the glycine-bound structure crystallized under the same conditions, and the 2.8 A omit jFoj-jFcj map (Table 1) revealed three electron-density peaks that correspond to glycine and cations M2 and M7 not present in the free structure. This experiment validates the position of the bound glycine and establishes that M2 does not critically contribute to overall scaffold formation. Instead, M2 most likely shields the negative charge of the glycine carboxylate, blocks the entrance to the pocket after glycine binding, and prevents dissociation of glycine by locking up the RNA fold. Thus, the bound glycine is
Molecular Cell
Figure 2. Glycine-Binding Pocket of VCII RNA and Recognition of Glycine

(A) Overall view of the glycine-binding pocket. The C31G38 and A64G72 segments are in blue and light blue, respectively. Hydrated Mg2+ cations are in green with coordination bonds shown in stick representation. (B) Base triples in the binding pocket and putative hydrogen bonds (dashed lines) contributing to glycine recognition. (C) Rened 2jFoj-jFcj electron-density map contoured at 1s level (pink) and superposed with the rened model of the binding pocket of the glycine-bound structure. Green map shows omit jFoj-jFcj map (3s level) calculated prior to the addition of glycine. (D) The same view as (C) for the binding pocket in the unbound state. (E) Superposition of the binding pockets of glycine-bound (salmon) and unbound (blue) states. Arrow shows the distance between phosphorus atoms of C66 and A33 in the glycine-bound structure. (F) Surface view inside of the glycine-bound pocket, with bound glycine in red and a pair of Mg2+ cations in green. The helix above J3/3aJ3a/3 is omitted for clarity.
completely buried inside of the pocket formed by the RNA scaffold and bound cations and has virtually no solvent-accessible surface area. A striking parallel is observed with the sensing domain of the lysine riboswitch, which also recognizes the carboxylate moiety of the amino acid with the assistance of a positively charged cation (Serganov et al., 2008). However, instead of divalent Mg2+, lysine is directly bound by monovalent K+ cation, which in effect replaces cation M2 of the glycine riboswitch. Glycine Binds in a Ligand-Shaped Pocket Assisted by a Pair of Mg2+ Cations The experimental density map of bound glycine extends from the Hoogsteen edge of G35 along the Watson-Crick edge of U69 in the direction of Mg2+ cation M2 (Figure 2C and Figures S2D and S2E). The elliptical shape of this map and the omit jFoj-jFcj map (Figure 2C), calculated prior to the addition of glycine to the
structure, orients glycine termini toward G35 and M2. Analysis of the surface electrostatic potential within the pocket suggests that the glycine should be positioned with the negatively charged carboxylate moiety oriented toward the positively charged cation M2, while the positively charged ammonium group should be directed away from cation M1 (Figure 2B). In this orientation, the ammonium group is placed within hydrogen bond distance (2.7 A) from O6 of G35 (Figure 2B) and close (distances less than 3.5 A) to N1 of A71 (Figure S2B) and O6 of G36 (Figure S2C). Therefore, the ammonium group has the potential for connecting bases from all three layers of the binding pocket, although rotational exibility could limit interactions to two layers (Figures S2D and S2E, insets). At the other glycine terminus (Figure 2B), one carboxylate oxygen atom makes a hydrogen bond with the N3 proton of U69, while the second oxygen interacts with 20 -OH group of G32 and also coordinates to the outer sphere of M2. By contrast, positioning of the glycine in an opposite orientation within the pocket would place both glycine termini in an unfavorable electrostatic environment; namely, the carboxylate oxygen would face oxygens of G35 and U69 and the ammonium group would be positioned next to the cation M2. In attempts to validate the orientation of glycine in the pocket, we mutated nucleotides around glycine and analyzed their effects by using isothermal titration calorimetry (ITC). We measured a Kd = 3.5 1.5 mM for wild-type VCII RNA with
Molecular Cell
Table 2. Data Collection and Renement Statistics, Glycine-Free VCII RNA, Mutants, and Cation Identication Crystal PDB ID Data Collection Space group Cell dimensions (a,b,c [A]) Wavelength (A) Resolutiona Rmerge (%)a <I> /s(I)a Completeness (%)a Unique reectionsa Redundancya Structure Renement Resolution (A) Working set reections Test set reections Rfactor / Rfree (%)a RNA Glycine Cations Average B-Factors (A2) RNA Glycine Cations Rmsd Values Bond lengths (A) Bond angles ( ) Estimated coordinate errorb
a
Free 3OX0 P3221 84.3, 84.3, 199.9 1.0809 20.003.05 (3.163.05) 5.2 (48.7) 34.8 (3.9) 99.8 (99.9) 16,320 (1,600) 7.1 (7.3) 20.003.05 15,445 824 20.7 / 24.0 3,807 0 28 56.5 65.8 0.005 1.069 0.290
Mutation U69C 3OXB P3221 84.0, 84.0, 197.8 1.0809 20.002.95 (3.062.95) 6.2 (54.2) 55.5 (4.0) 99.7(99.8) 17,666 (1,719) 16.5 (12.2) 20.002.95 16,755 899 19.8 / 22.1 3,785 0 33 58.2 79.3 0.005 1.082 0.304
Mutation G35C/C70G 3OXD P3221 84.0, 84.0, 200.0 1.0750 20.003.00 (3.113.00) 7.4 (52.4) 41.5 (5.2) 99.8 (99.9) 16,953 (1,624) 12.1 (12.5) 20.003.00 16,063 857 19.6 / 21.4 3,785 0 34 46.9 69.0 0.005 1.064 0.253
Soaked in 10 mM MnCl2 3OXE P3221 83.4, 83.4, 199.0 1.5000 20.002.90 (3.002.90) 6.3 (56.4) 31.3 (3.0) 99.9 (100.0) 18,565(1,814) 5.3 (5.4) 20.002.90 17,579 947 20.1 / 21.9 3,807 10 42 40.9 40.3 84.0 0.005 1.011 0.236
Soaked in 10 mM Tl-acetate 3OXM P3221 83.7, 83.7, 200.1 0.9763 20.002.95 (3.002.95) 6.7 (51.2) 47.7 (5.5) 100.0 (100.0) 17,793 (881) 11.2 (11.5) 20.002.95 16,839 904 20.0 / 22.5 3,807 10 37 33.9 35.0 73.0 0.005 1.021 0.255
Soaked in 10 mM BaCl2 3OXJ P3221 84.2, 84.2, 201.4 1.5000 30.003.20 (3.313.20) 8.1 (54.3) 29.2 (3.1) 99.9 (100.0) 14,230 (1,390) 5.2 (5.4) 20.003.20 13,482 714 19.4 / 22.4 3,790 10 45 69.9 66.6 123.5 0.005 1.110 0.302
Number of Nonhydrogen Atoms
Values for the highest resolution shell are in parentheses. b Estimated coordinate error based on maximum likelihood was calculated with REFMAC (Murshudov et al., 1997).
a long P1 helix as compared to a Kd = 24.0 mM determined by inline probing for domain II anked by single-stranded regions at the same 20 mM Mg2+ concentration (Mandal et al., 2004). For the rst mutant, we replaced a proton donor at N3 of U69 with a proton acceptor after U69C substitution (Figure S2F). This mutation disrupted a single hydrogen bond between U69 and glycine and caused a drop of afnity to a negligible level (Figures 3A and 3B). Next, the G35-C70 base pair was substituted with C-G and A-U base pairs (Figure S2F). These mutations, which disrupt the hydrogen bonds between glycine and G35, also resulted in a total loss in glycine binding. To conrm the integrity of the binding pocket in the mutant VCII RNAs, we crystallized the U69C and G35C/C70G mutants in the presence of glycine (Table 2). The $3.0 A resolution structures (Figure S2G) revealed the formation of well-organized glycine-binding pockets that lack bound glycine and M2 and M7 cations. The G35A/C70U mutant did not crystallize, apparently due to signicant perturbations of the RNA structure.
The glycine-bound structure of VCII RNA readily explains the detrimental effects of functional group substitutions on glycine binding reported previously in NAIM experiments (Kwon and Strobel, 2008). For instance, the replacement of G32 by deoxyribonucleotide eliminates the hydrogen bond with the carboxylate of glycine and the direct coordination bond with M2, while methylation of N6 of A34 or N7 of G32 prevents formation of the A34,A71,G32 triple (Figure S3A, top inset). Shape Complementarity and Heteroatom Recognition Contribute to Glycine Selectivity Glycine riboswitches selectively recognize glycine and efciently discriminate against similar compounds. Like purine, lysine, and other riboswitches (Montange and Batey, 2008), the sensing domain of the glycine riboswitch forms a ligand-shaped pocket, which snugly ts bound glycine. The surface views (Figure 2F and Figure S2H) illustrate that the binding pocket (volume $72 A3) of VCII RNA is almost completely lled by the bound
Molecular Cell
Figure 3. Glycine Binding to RNAs Determined by ITC

(A) Summary of ITC-based parameters for single and tandem RNAs. All values are mean SD. (B) Integrated tted heat plots of glycine binding to VCII RNA and mutants. The control is the FMN riboswitch sensing domain. (C) Effect of magnesium concentration on glycine binding to VCII RNA. Experimental data for 40 mM MgCl2 and integrated plots for all MgCl2 concentrations are shown in top and bottom panels, respectively. (D) Cation effects on glycine binding to VCII RNA. (E) Glycine binding to VCI RNA. Values in parentheses depict RNA refolding temperature. Representative raw data at 20 mM MgCl2 and 75 C refolding temperature are shown in top panel. (F) Glycine binding to VCI-II RNA.
2004) and abolished ligand recognition in ITC experiments (Figure S4A). In agreement with our biochemical data, the crystal structure of VCII RNA obtained in the presence of hydroxamate lacked both the ligand and M2 cation (Figure S4B). Modications of the carboxyl moiety that preserved both oxygens, such as methyl, ethyl, and tert-butyl esters, were reported to be tolerated with progressively reduced binding afnity (Kwon and Strobel, 2008; Mandal et al., 2004) (Figure S4C). Though the methyl ester could theoretically be accommodated in a slightly widened pocket, the larger ethyl and tert-butyl moieties would require more dramatic adjustments, such as the loss of M2. Unfortunately, electron-density maps calculated from the crystals grown with glycine esters did not reveal the density for additional groups (Figure S4D). Moreover, the pocket volumes, ranging from $70 to 81 A3, are incompatible with large glycine analogs. Possibly, these crystals either contained glycine as a contaminant of the original ester samples or, alternately, reected the release of glycine as a product of ester decomposition during crystallization (Figure S4E). Mg2+ Cations Participate in Both Ligand Recognition and RNA Folding The function of many cellular RNAs, including glycine riboswitches (Lipfert et al., 2007, 2010), depends critically on the presence of Mg2+ cations. The crystal structure of VCII RNA revealed 15 Mg2+ cations distributed along P1, P2, J1/2, and J3/1, where they participate in the formation of the intrinsic junctional fold. Five more Mg2+ cations are bound around and within the glycine-binding pocket. Remarkably, six out of eight NAIM sites (Kwon and Strobel, 2008), located in the vicinity of the glycine-binding pocket, can affect coordination of Mg2+ cations (Figure S3A). Our ITC experiments showed that
glycine (volume $68 A3). Consistent with published data (Kwon and Strobel, 2008; Mandal et al., 2004), the ligand-shaped pocket cannot accommodate methyl groups attached either to the amino group of glycine, as in sarcosine, or to the Ca atom, as in alanine. At the carboxyl end, both oxygens are critical for interactions with the riboswitch. Replacement of a single oxygen atom by amine or hydroxylamine decreased binding afnity in inline probing assays (Kwon and Strobel, 2008; Mandal et al.,
Molecular Cell
glycine binding to VCII RNA decreased $13-fold (Kd = 45.9 6.6 mM) and $300-fold (Kd greater than 1 mM) when Mg2+ concentration was reduced from 20 mM to 10 mM and 2 mM, respectively (Figures 3A and 3C). Increase of Mg2+ concentration from 20 to 40 mM did not signicantly affect binding afnity, although it resulted in a reduction in favorable entropy and a substantial increase in favorable enthalpy (Figure 3A). Nevertheless, ligand binding to VCII is enthalpy driven at both 20 and 40 mM Mg2+ (Figures 3A and 3C). Notably, Mg2+ could not be replaced by 1 M K+ (Figure 3D), as has been reported for the avin mononucleotide (FMN) riboswitch (Serganov et al., 2009), suggesting that monovalent cations cannot fulll the function of divalent Mg2+. In agreement with the structures and ITC results, Tl+ soaks identied ve anomalous peaks (Figures S5B), mostly coincident with divalent cation sites, further conrming that VCII RNA does not require specic monovalent cations for glycine binding. Large divalent cations, such as Ba2+, abolished glycine binding, while smaller Mn2+ cations supported ligand recognition (Figure 3D). These data are in line with earlier SAXS measurements that revealed the critical role of Mg2+ and similar cations in ligand-dependent conformational changes of the tandem glycine riboswitch (Lipfert et al., 2007; Lipfert et al., 2010). Both Mn2+ and Ba2+ cations were located in the structure based on their anomalous signals. Mn2+ cations were found in 17 sites occupied by Mg2+ cations in the native structure, including M1 and M2 (Figures S5A and S5C). M1 was substituted by Mn2+ to a smaller extent, suggesting that this partially hydrated cation is tightly bound to RNA in the glycine-bound structure (Figure 2C and Figure S2I). Ba2+ cations replaced only several Mg2+ cations (Figure S5D); M1 and M2 were not substituted. Therefore, the VCII RNA is more restrictive to the nature of divalent cations than earlier-studied sensing domains of TPP (Edwards and Ferre DAmare, 2006) or FMN (Serganov et al., 2009) riboswitches. Interestingly, while binding to VCII RNA is exothermic and enthalpy driven, binding to the isolated domain I of V. cholerae riboswitch (VCI RNA) is approximately ten times weaker, endothermic, entropy driven, and requires high refolding temperatures (Figures 3A and 3E). On the other hand, glycine binding to the RNA that contains both sensing domains of the natural riboswitch (VCI-II) is highly dependent on Mg2+ concentration (data not shown) and is characterized by Kd = 2.3 0.7 mM, a value comparable to VCII RNA-binding afnity (Figures 3A and 3F). Contribution of Junctional Folding to Riboswitch Mechanism In the majority of riboswitches, modulation of P1 helix formation is the key conformational rearrangement affecting gene expression. In the presence of glycine, P1 helix of the glycine riboswitch sequesters the regulatory sequence in the 30 -region of the sensing domain so that the downstream region folds into an antiterminator hairpin and allows the RNA polymerase to complete gene transcription (Mandal et al., 2004). Unlike other riboswitches, gene expression control in the glycine riboswitch could rely heavily on a single conserved nucleotide, A33, which is looped out of the ligand-binding pocket (Figure 1E) and organizes the junctional region through continuous stacking interac-
tions and noncanonical base-pairing that facilitate formation of P1 (Figures 1D). It is conceivable that glycine binding facilitates extrusion of A33 from the pocket and synergizes with the A33dependent stabilization of the three-way junction. Although A33 is not exible in the glycine-free VCII structure, solutionbinding assays of A33 mutants are in line with this suggestion. Disruption of the junction is the likely cause of glycine-binding loss in the A33G mutant (Figure 3B), in which the NH2 group of the introduced guanine clashes with sugar atoms of G80. On the other hand, substitution of A33 with the small uridine residue, found in 1% of glycine riboswitches, preserved glycine binding, albeit to a smaller extent (Kd = 16.2 5.2 mM). In addition, less than half of the A33U RNA was found capable of glycine binding (n = 0.350.55), most likely because of difculties in junction formation due to the loss of two hydrogen bonds with the G8 sugar. Therefore, without intercalation of A33, it is likely that the junction and P1 would not be stably folded and the regulatory sequence would engage in base-pairing with the downstream region, thereby forming a transcription terminator that aborts gene transcription. The formation of the stable junction and P1 is also precluded by the previously tested G8C mutation (Mandal et al., 2004), which disrupts G8,A81 base pair formation (Figure 1B); by G10C mutation, which prevents interactions of G10 with G79 and nucleotides of P3 helix (Figure 1B and Figure S1D); and by six functional group substitutions identied in the junctional core by NAIM experiments (Kwon and Strobel, 2008) (Figure S3A, bottom inset). Contacts between RNAs in Crystallographic Asymmetric Unit The asymmetric unit contains two similarly folded VCII RNA molecules that pack against each other and exhibit good surface complementarity and a surprisingly extensive buried surface area of 1262 A2 per molecule (Figures 4A and 4B). This area is larger than the 900 A2 value characteristic of potentially biologically relevant contact areas in protein dimers (Ponstingl et al., 2000). Although crystal contacts do not necessarily correspond to physiological interactions (Bahadur et al., 2004), there are examples in which such correspondence has been proven for proteins (Shi et al., 1997). The same analysis cannot be performed on RNA because only a few structures of small biological RNA dimers have been determined. The contacting surface of VCII molecules involves three RNA regions and three Mg2+ cations, M6, M7, and M8. The rst contact features p-stacking interaction between A65 residues from each VCII RNA that are looped out of the glycine-binding pocket (Figure 4A). Practically all glycine riboswitches have a residue at this position, with adenine observed in $56% of cases. The other two contact regions are symmetrical and formed between P1 of one molecule and internal loop J3a/3b-J3b/3a of the second molecule. To make these interactions, J3a/3b-J3b/3a adopts a twisted helical conformation constrained by the noncanonical A41,G62 and A64,C39 base pairs at each end with four adenines, A40, A41, A63, and A64, involved in crossstrand stacking interactions (Figure 4C). The stacked adenines are inserted into the minor groove of P1 where they comprise four A-minor interactions (Nissen et al., 2001): two type I, A64, (G7-C82) and A41,(G84-C5) triples, and two type II,
Molecular Cell
Figure 4. Intermolecular Interactions in the Asymmetric Unit of the VCII RNA Structure
(A) Two VCII RNA molecules (light blue and orange) in the asymmetric unit. Contact areas are indicated. (B) Split-up view of the interacting VCII RNAs in surface representation. Green spheres indicate Mg2+ cations located in the riboswitch interface. Nucleotide analog interference sites (Kwon and Strobel, 2008) that cannot be explained by the structure of the individual VCII RNA are shown in magenta. (C) Zoomed-in view of the J3b/3a-J3a/3b region. (D) A-minor interactions between J3b/3a-J3a/3b region and P1.
A63,(U6-A83) and A40,(U6-A83) triples (Figures 4D and Figure S6A). The consecutive adenine bases are inclined with respect to the P1 base pairs to form double A-minor interactions (Gilbert et al., 2008; Klein and Ferre-DAmare, 2006; Serganov et al., 2006), such that A40 forms tertiary contacts with both the A83-U6 base pair and the adjacent G84. The folding of J3a/3bJ3b/3a apparently depends on glycine binding as suggested by inline probing of the individual VCII RNA (Mandal et al., 2004). Therefore, it is conceivable that glycine binding could induce conformational changes required for allosteric communication between tandem domains in the context of the complete riboswitch. Interdomain Interactions in Engineered Tandem Riboswitch in Solution To test whether interdomain interactions are not solely the property of the crystal lattice but could also take place in solution, we conducted nuclease footprinting experiments with the individual VCII RNA and its tandem VCII-II variant, composed of a pair of VCII RNAs connected by a 7 nt linker. The linker can easily span the $35 A distance between the 50 - and 30 -ends of the packed VCII RNA molecules observed in the crystallographic asymmetric unit (Figure 4A). Addition of 200-fold excess of glycine to the VCII RNA did not alter V1 and T2 nuclease cleavages, suggesting that the individual VCII RNA, stabilized by long P1 helix, did not signicantly change its conformation upon glycine binding, at least in the context of tracking by large nucleases (Figure 5A). It should be noted that other probing techniques that are less dependent on the local
structural context, such as inline probing, detected changes in the individual domain II upon glycine binding (Mandal et al., 2004). Because VCII RNA did not stably dimerize in gel-shift experiments at similar concentrations (data not shown), the nuclease cleavage patterns of VCII RNA most likely correspond to a monomeric conformation. The digestion patterns of individual VCII and tandem VCII-II RNAs in the absence of glycine are similar (Figures 5A and 5B and Figures S7A and S7B), suggesting that VCII-II RNA does not form stable interdomain contacts prior to glycine binding. Addition of glycine to VCII-II RNA reduced many V1 cleavages in P1 (nt 8485, 100101) and P3b (nt 4547, 58, 141, 142, 153) (Figure 5B and Figures S7B and S7C). These regions are adjacent to the intermolecular contacts in the crystallographic dimer of the VCII RNA, and hence one can infer that the cleavage reductions are most likely associated with the formation of interdomain interactions in the tandem VCII-II RNA in the glycine-bound state. To reinforce our conclusion, we conducted footprinting on mutated tandem VCII-II RNAs containing a short 2 nt linker (designated VCII-IIs RNA), which cannot connect the ends of the crystallographic dimer in the VCII structure. Glycine did not modulate nuclease cleavages in this RNA (Figure 5C and Figure S7D) and, as anticipated, prevented formation of interdomain interactions. To test putative A-minor interactions in VCII-II RNA more directly, we mutated the C5-G84 base pair to a U,G pair in P1 of the rst domain (designated VCII-II-U5 RNA). This mutation changes the base alignment in the A41,(G84-C5) A-minor triple and should prevent the formation of stable tertiary interactions between P1 of the rst domain and J3a/3b-J3b/3a of the second
Molecular Cell
Figure 5. Ribonuclease V1 and T2 Probing of Interdomain Interactions

(AF) Projections of weak and strong nuclease cleavage reductions (light and dark red) and enhancements (green) on the secondary RNA structures are shown on top and the corresponding gels are on the bottom. Cleavage reduction values (mean SD from at least three gels) in the presence of 200-fold excess of glycine are shown on the secondary RNA structures at the top of (B) and (D). Note that the 50 and 30 regions of RNAs have not been analyzed. T1 and OH designate RNase T1 and alkaline ladders, respectively. NR, no reaction.
domain. In the presence of glycine, the reduction of average V1 cleavage (mean value 1.3) in this contacting area for the VCII-IIU5 mutant is 1.8-fold smaller than the corresponding footprint in unaltered RNA (mean value 2.3) (Figure 5D and Figure S7E). Although the contact between P1 of the second domain and J3a/3b-J3b/3a of the rst domain was not mutated, the V1 footprint in this contacting area was also reduced (mean value 1.4), but to a lesser extent ($1.4-fold) relative to unaltered RNA (mean value 1.9). These results are indicative of synergistic formation of two A-minor-mediated tertiary interdomain contacts in the glycine-bound VCII-II RNA. The binding afnities of the mutated tandem RNAs were preserved and their apparent Kd values, 3.9 0.03 mM for VCII-IIs mutant and 7.4 2.3 mM for VCII-II-U5 mutant, were in range of Kd values of 7.0 1.6 mM for VCII-II and 3.5 1.5 mM for VCII. Our footprinting and glycine-binding data are consistent with the engineered tandem VCII-II RNA forming interdomain interactions in solution, and we suggest that these interactions could reect the intermolecular contacts observed for the crystallographic VCII dimer.
Interdomain Interactions in Natural Tandem Riboswitch in Solution To test for interdomain contacts in the natural tandem riboswitch in solution, we performed nuclease footprinting with tandem VCI-II and individual VCI RNAs. As in VCII RNA, the cleavage of VCI RNA was not affected by glycine (Figure 5E), while VCI-II RNA showed modulated cleavage patterns (Figure 5F). Most footprints observed in VCII-II RNA were also found in VCI-II RNA, with the exception of V1 protections in P3b helix of domain I in VCII-II RNA (Figure 5B). In domain I of VCI-II RNA, the P3b/L3 hairpin was reduced to the apical loop L3, which was strongly protected against nuclease T2 cleavage (Figure 5F). These data suggest that both tandem natural VCI-II RNA and engineered VCII-II RNA probably experience similar conformation changes upon glycine binding in solution and that corresponding regions of both RNAs might form analogous interdomain interactions. In addition to protections common to both tandem RNAs, VCI-II RNA displayed strong cleavage reductions in J3/3a region, and P3a, P2-P2a, and P4-P4a helices (Figure 5F). These
Molecular Cell
Figure 6. Likely Interdomain Interaction in V. cholerae Glycine Riboswitch

(A) Projection of intermolecular interactions found in the VCII crystallographic dimer and NAIM sites (Mandal and Breaker, 2004) on the secondary structure of natural tandem riboswitch. Domain I is shown in cyan and domain II is in black. Green circles indicate NAIM sites that cannot be explained by the structure of the individual VCII RNA. Orange shading shows areas that correspond to the regions involved in interdomain interactions in the crystallographic dimer of VCII RNA. (B and C) Proposed interdomain interactions in the tandem natural riboswitch.
footprints indicate that VCI-II RNA forms a compact structure that prevents access of nucleases to these regions (Figures S7F and S7G). Correlation of Interdomain Interactions with Nucleotide Analog Interference Mapping Data on Natural Tandem Riboswitch The NAIM data on VCI-II RNA (Kwon and Strobel, 2008) include several interference sites within domain II that cannot be explained by the structure of VCII RNA alone but could be rationalized if these sites interfere with interdomain interactions. Projection of these NAIM sites on the VCII structure shows that they are located at the interface of the riboswitch dimer in the asymmetric unit (Figure 4B), where they should interfere with the formation of intermolecular A-minor interactions (Figure S3B). Because VCI and VCII RNAs resemble each other, some of these interactions could be involved in formation of biologically relevant interdomain interactions in the natural tandem riboswitch and could be responsible for riboswitch cooperativity. Indeed, the C5U mutation that disrupted the interdomain A-minor interactions in VCIIII RNA in solution (Figure 5) also abrogated cooperative binding in the natural tandem VCI-II RNA (Kwon and Strobel, 2008). Therefore, combined structural and NAIM data indicate that in the natural tandem VCI-II RNA the J3b/3a-J3a/3b region of domain II may interact with the upper part of P1 of domain I (Figures 6A and 6B and Figure S6B), which resembles the corresponding region in domain II. Although in domain I the A-rich apical loop L3 replaces the A-rich internal loop J3b/3a-J3a/3b of domain II, it is possible that the L3 loop participates in tertiary A-minor interactions with the top part of P1 from domain II (Figures 6A and 6B), as suggested by the NAIM sites at positions 6768, 140, and 216 (Figure S7G numbering) (Kwon and Strobel, 2008). The tertiary interaction scheme (Figure 6), based on the crystal contacts observed for the VCII RNA dimer, also allows prediction of interdomain contacts in the cooperative Fusobacterium nucleatum and B. subtilis glycine riboswitches (Figures S6CS6E), with the interdomain contact areas proposed by us in agreement with the NAIM data by Kwon and Strobel (2008) on the F. nucleatum glycine riboswitch. The regions of interdomain contacts, proposed by us, also coincide with the areas of glycine-induced reductions of hydroxyl radical cleavage in the
tandem V. cholerae riboswitch (Lipfert et al., 2007). Note that tandem RNAs, such as VCII-II, do not necessarily have to display a strong cooperativity because interdomain interactions may not improve glycine-binding afnity if both domains are capable of strong and identical glycine binding. Although crystallization facilitates formation of the intermolecular RNA dimer in the absence of glycine in the crystal lattice, the solution nuclease footprinting data indicate that stable interdomain interactions can be observed in the presence of glycine and when domains are connected by a linker. Lack of stable interdomain interactions prior to glycine binding suggests that glycine probably interacts with domains of the tandem riboswitch sequentially. It is conceivable that glycine binds initially to domain I, which is transcribed rst. On the other hand, domain II may initially interact with the ligand, since this domain binds glycine tighter than domain I, according to our ITC (Figures 3C and 3E) and published mutagenesis (Mandal et al., 2004) data. In addition, domain I had to be refolded at 75 C to bind glycine efciently (Figure 3E), and, therefore, it may require assistance from domain II to adopt a conformation competent for glycine binding in the tandem riboswitch. Because other factors, such as transcriptional pausing, are probably involved in the riboswitch folding, the order of glycine binding to domains of the tandem riboswitch cannot be deduced at this time. In conclusion, the VCII RNA crystal structures provide a comprehensive picture of glycine recognition by RNA, governed by electrostatics, shape complementarity, and specic direct and Mg2+-mediated interactions. Although the exact mechanism of the cooperative response remains an open question for future studies, our structural and footprinting data, combined with the results of inline probing (Kwon and Strobel, 2008; Mandal et al., 2004), NAIM (Kwon and Strobel, 2008; Mandal et al., 2004), and SAXS (Lipfert et al., 2007) experiments, have identied specic interdomain interactions between tandem sensing domains and are supportive of their participation in the cooperative binding of glycine by the tandem glycine riboswitch.
EXPERIMENTAL PROCEDURES RNA Preparation, Crystallization, and Data Collection RNAs encoding the V. cholerae riboswitch sequences were transcribed in vitro and puried by denaturing polyacrylamide gel electrophoresis (PAGE)
Molecular Cell
followed by anion-exchange chromatography and ethanol precipitation. Complexes of glycine riboswitch with glycine were prepared by mixing 0.5 mM RNA with 2 mM glycine in a buffer containing 100 mM potassium acetate (pH 6.8) and 4 mM MgCl2. Crystals were grown at 20 C for 1 week by using the hanging-drop vapor diffusion method after mixing the complex at an equimolar ratio with the reservoir solution containing 0.05 M Na-cacodylate (pH 5.1), 0.2 M KCl, 8% (w/v) PEG8000, and 80 mM magnesium acetate. Crystallization of the mutants and the riboswitch in the unbound state were carried out in a similar manner. For data collection, crystals were quickly transferred into a cryoprotectant solution containing 0.05 M 2-(N-morpholino)ethanesulfonic acid (MES)-NaOH buffer (pH 5.5), 150 mM magnesium acetate, 25% MPD, and 25% ethanol and ash-frozen in liquid nitrogen. For heavy atom and cation soaking, crystals were transferred into the cryoprotectant solution supplemented with 2 mM [Ir(NH3)6]Cl3, 10 mM thallium acetate, 10 mM BaCl2, or 10 mM MnCl2 and soaked for 8, 8, 3.5, and 4 hr, respectively. Data were collected at 100 K at the X29a beamline of the National Synchrotron Light Source (NSLS, Brookhaven) and processed with HKL2000 (HKL Research). Crystals with glycine analogs were obtained by cocrystallization or by soaking glycine-unbound crystals in the cryoprotectant solution supplemented with 2 mM analogs for 68 hr. Structure Determination, Renement, and Analysis The structure of glycine-bound VCII RNA was determined by multiwavelength anomalous dispersion (MAD) method with 2.95 A iridium hexamine data and autoSharp (Table 1). Crystals were improved by the replacement of the L3 loop with a GAAA tetraloop. The RNA model was built in Coot (Emsley and Cowtan, 2004) and rened in Phenix (Adams et al., 2002) and REFMAC (Mur shudov et al., 1997) with a 2.85 A native data set. Glycine was added to the model at a later stage based on the experimental and rened maps, electrostatic analysis, and mutational data. Cations and their coordinated waters were added based on the coordination geometry, coordination distances, and omit and anomalous maps. All other riboswitch structures were rened by using native riboswitch structure as a starting model (Table 2). The contact surface area and the solvent accessibility were calculated by using PISA webserver (http://www.ebi.ac.uk/msd-srv/prot_int/cgi-bin/piserver). The volumes of the binding pocket and ligand were calculated by using VOIDOO package (Uppsala Software Factory) and the molinspiration server (http://www.molinspiration.com). Figures were prepared with PyMOL (http:// www.pymol.org). Nuclease Probing and Footprinting For footprinting experiments, the riboswitch RNAs were radioactively labeled at the 50 end. Twenty microliter samples of the radiolabeled RNA (100,000 counts/min) with a nal RNA concentration of 0.5 3 107 M were preheated at 75 C for 13 min in 50 mM HEPES-KOH (pH 7.5), 100 mM KCl, and 20 mM MgCl2 in the presence, if necessary, of 20- and 200-fold excess of glycine, incubated at room temperature for 1 min, chilled on ice, and additionally incubated at 37 C for 15 min. Cleavage reactions were performed with 0.13 U RNase V1 (Pierce) or 0.05 U RNase T2 (Sigma) at 37 C for 10 min. Reactions were quenched by the addition of 80 ml cold buffer, immediately extracted with phenol-chloroform, and precipitated by ethanol. RNA pellets were dissolved and analyzed by PAGE. Nuclease effects were quantied after normalization and background subtraction by using ImageGauge (Fuji). Isothermal Titration Calorimetry-Binding Assays All experiments were performed at least in duplicate on Microcal calorimeter ITC200 at 35 C. Prior to titration, 0.10.3 mM RNA samples were dialyzed overnight at 4 C against experimental buffer containing 50 mM HEPES-KOH (pH 7.5), 100 mM potassium acetate, and 040 mM MgCl2 or other divalent cations. If necessary, RNAs were refolded by heating at 60 C 75 C for 310 min and cooling on ice. For measurements, the ligands, dissolved in the dialysis buffer at $13 mM concentration, were typically titrated into RNA in the sample cell (V = 207 mL) by 26 serial injections of 1.5 ml each, with a 3 ml/s rate, 180 s intervals between injections, and a reference power of 6 mcal/s. Values for c were between 15 and 25. The thermograms were integrated and analyzed by using Origin 7.0 software (Microcal, Inc.) and
SEDPHAT (Houtman et al., 2007) as described in Supplemental Experimental Procedures. All values are mean standard deviation (SD). ACCESSION NUMBERS The atomic coordinates and structure factors were deposited in the Protein Data Bank (IDs 3OWZ, 3OWI, 3OWW, 3OX0, 3OXB, 3OXD, 3OXE, 3OXM, and 3OXJ) (Tables 1 and 2). SUPPLEMENTAL INFORMATION Supplemental Information includes seven gures and Supplemental Experimental Procedures and can be found with this article online at doi:10.1016/ j.molcel.2010.11.026. ACKNOWLEDGMENTS We thank the personnel of beamline X29 at NSLS for assistance with data collection. We thank O. Ouerfelli (Memorial Sloan-Kettering Cancer Center, New York) for the synthesis of iridium hexamine. D.J.P. was supported by funds from the National Institutes of Health. L.H. grew the crystals, determined the structures, and performed ITC-binding assays. A.S. conducted footprinting experiments. D.J.P. supervised the project. All authors were involved in the planning of experiments, interpretation of the results, and writing of the manuscript. Received: April 12, 2010 Revised: August 5, 2010 Accepted: September 24, 2010 Published: December 9, 2010 REFERENCES Abreu-Goodger, C., Ontiveros-Palacios, N., Ciria, R., and Merino, E. (2004). Conserved regulatory motifs in bacteria: riboswitches and beyond. Trends Genet. 20, 475479. Adams, P.D., Grosse-Kunstleve, R.W., Hung, L.W., Ioerger, T.R., McCoy, A.J., Moriarty, N.W., Read, R.J., Sacchettini, J.C., Sauter, N.K., and Terwilliger, T.C. (2002). PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr. D Biol. Crystallogr. 58, 19481954. Bahadur, R.P., Chakrabarti, P., Rodier, F., and Janin, J. (2004). A dissection of specic and non-specic protein-protein interfaces. J. Mol. Biol. 336, 943955. Barrick, J.E., and Breaker, R.R. (2007). The distributions, mechanisms, and structures of metabolite-binding riboswitches. Genome Biol. 8, R239. Batey, R.T., Gilbert, S.D., and Montange, R.K. (2004). Structure of a natural guanine-responsive riboswitch complexed with the metabolite hypoxanthine. Nature 432, 411415. Dartois, V., Liu, J., and Hoch, J.A. (1997). Alterations in the ow of one-carbon units affect KinB-dependent sporulation in Bacillus subtilis. Mol. Microbiol. 25, 3951. Edwards, T.E., and Ferre-DAmare, A.R. (2006). Crystal structures of the thibox riboswitch bound to thiamine pyrophosphate analogs reveal adaptive RNA-small molecule recognition. Structure 14, 14591468. Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 21262132. Garst, A.D., Heroux, A., Rambo, R.P., and Batey, R.T. (2008). Crystal structure of the lysine riboswitch regulatory mRNA element. J. Biol. Chem. 283, 22347 22351. Gilbert, S.D., Rambo, R.P., Van Tyne, D., and Batey, R.T. (2008). Structure of the SAM-II riboswitch bound to S-adenosylmethionine. Nat. Struct. Mol. Biol. 15, 177182. Heil, G., Stauffer, L.T., and Stauffer, G.V. (2002). Glycine binds the transcriptional accessory protein GcvR to disrupt a GcvA/GcvR interaction and allow
Molecular Cell
GcvA-mediated activation of Microbiology 148, 22032214.
the
Escherichia
coli
gcvTHP operon.
Recht, M.I., and Williamson, J.R. (2004). RNA tertiary structure and cooperative assembly of a large ribonucleoprotein complex. J. Mol. Biol. 344, 395407. Sattin, B.D., Zhao, W., Travers, K., Chu, S., and Herschlag, D. (2008). Direct measurement of tertiary contact cooperativity in RNA folding. J. Am. Chem. Soc. 130, 60856087. Saxild, H.H., Brunstedt, K., Nielsen, K.I., Jarmer, H., and Nygaard, P. (2001). Denition of the Bacillus subtilis PurR operator using genetic and bioinformatic tools and expansion of the PurR regulon with glyA, guaC, pbuG, xpt-pbuX, yqhZ-folD, and pbuO. J. Bacteriol. 183, 61756183. Serganov, A., and Patel, D.J. (2007). Ribozymes, riboswitches and beyond: regulation of gene expression without proteins. Nat. Rev. Genet. 8, 776790. Serganov, A., and Patel, D.J. (2009). Amino acid recognition and gene regulation by riboswitches. Biochim. Biophys. Acta 1789, 592611. Serganov, A., Yuan, Y.R., Pikovskaya, O., Polonskaia, A., Malinina, L., Phan, A.T., Hobartner, C., Micura, R., Breaker, R.R., and Patel, D.J. (2004). Structural basis for discriminative regulation of gene expression by adenineand guanine-sensing mRNAs. Chem. Biol. 11, 17291741. Serganov, A., Polonskaia, A., Phan, A.T., Breaker, R.R., and Patel, D.J. (2006). Structural basis for gene regulation by a thiamine pyrophosphate-sensing riboswitch. Nature 441, 11671171. Serganov, A., Huang, L., and Patel, D.J. (2008). Structural insights into amino acid binding and gene control by a lysine riboswitch. Nature 455, 12631267. Serganov, A., Huang, L., and Patel, D.J. (2009). Coenzyme recognition and gene regulation by a avin mononucleotide riboswitch. Nature 458, 233237. Shi, Y., Hata, A., Lo, R.S., Massague, J., and Pavletich, N.P. (1997). A structural basis for mutational inactivation of the tumour suppressor Smad4. Nature 388, 8793. Siegfried, N.A., and Bevilacqua, P.C. (2009). Thinking inside the box: designing, implementing, and interpreting thermodynamic cycles to dissect cooperativity in RNA and DNA folding. Methods Enzymol. 455, 365393. Stoddard, C.D., Montange, R.K., Hennelly, S.P., Rambo, R.P., Sanbonmatsu, K.Y., and Batey, R.T. (2010). Free state conformational sampling of the SAM-I riboswitch aptamer domain. Structure 18, 787797. Thore, S., Leibundgut, M., and Ban, N. (2006). Structure of the eukaryotic thiamine pyrophosphate riboswitch with its regulatory ligand. Science 312, 1208 1211. Tripp, H.J., Schwalbach, M.S., Meyer, M.M., Kitner, J.B., Breaker, R.R., and Giovannoni, S.J. (2009). Unique glycine-activated riboswitch linked to glycine-serine auxotrophy in SAR11. Environ. Microbiol. 11, 230238. Winkler, W.C., and Breaker, R.R. (2005). Regulation of bacterial gene expression by riboswitches. Annu. Rev. Microbiol. 59, 487517.
Houtman, J.C., Brown, P.H., Bowden, B., Yamaguchi, H., Appella, E., Samelson, L.E., and Schuck, P. (2007). Studying multisite binary and ternary protein interactions by global analysis of isothermal titration calorimetry data in SEDPHAT: application to adaptor protein complexes in cell signaling. Protein Sci. 16, 3042. Jose, A.M., Soukup, G.A., and Breaker, R.R. (2001). Cooperative binding of effectors by an allosteric ribozyme. Nucleic Acids Res. 29, 16311637. Klein, D.J., and Ferre-DAmare, A.R. (2006). Structural basis of glmS ribozyme activation by glucosamine-6-phosphate. Science 313, 17521756. Kwon, M., and Strobel, S.A. (2008). Chemical basis of glycine riboswitch cooperativity. RNA 14, 2534. Lescoute, A., and Westhof, E. (2006). Topology of three-way junctions in folded RNAs. RNA 12, 8393. Lipfert, J., Das, R., Chu, V.B., Kudaravalli, M., Boyd, N., Herschlag, D., and Doniach, S. (2007). Structural transitions and thermodynamics of a glycinedependent riboswitch from Vibrio cholerae. J. Mol. Biol. 365, 13931406. Lipfert, J., Sim, A.Y., Herschlag, D., and Doniach, S. (2010). Dissecting electrostatic screening, specic ion binding, and ligand binding in an energetic model for glycine riboswitch folding. RNA 16, 708719. Mandal, M., and Breaker, R.R. (2004). Adenine riboswitches and gene activation by disruption of a transcription terminator. Nat. Struct. Mol. Biol. 11, 2935. Mandal, M., Lee, M., Barrick, J.E., Weinberg, Z., Emilsson, G.M., Ruzzo, W.L., and Breaker, R.R. (2004). A glycine-dependent riboswitch that uses cooperative binding to control gene expression. Science 306, 275279. Montange, R.K., and Batey, R.T. (2006). Structure of the S-adenosylmethionine riboswitch regulatory mRNA element. Nature 441, 11721175. Montange, R.K., and Batey, R.T. (2008). Riboswitches: emerging themes in RNA structure and function. Annu. Rev. Biophys. 37, 117133. Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Renement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240255. Nissen, P., Ippolito, J.A., Ban, N., Moore, P.B., and Steitz, T.A. (2001). RNA tertiary interactions in the large ribosomal subunit: the A-minor motif. Proc. Natl. Acad. Sci. USA 98, 48994903. Nudler, E., and Mironov, A.S. (2004). The riboswitch control of bacterial metabolism. Trends Biochem. Sci. 29, 1117. Ponstingl, H., Henrick, K., and Thornton, J.M. (2000). Discriminating between homodimeric and monomeric proteins in the crystalline state. Proteins 41, 4757.
Molecular Cell
Article
Oxidative Protein Folding by an Endoplasmic Reticulum-Localized Peroxiredoxin
Ester Zito,1 Eduardo Pinho Melo,1,5 Yun Yang,1 Asa Wahlander,1 Thomas A. Neubert,1,4 and David Ron1,2,3,6,*
Center for Biology and Medicine at the Skirball Institute of Cell Biology 3Department of Medicine 4Department of Pharmacology New York University School of Medicine, New York, NY 10016, USA 5Institute for Biotechnology and Bioengineering, Centre for Molecular and Structural Biomedicine, Universidade do Algarve, Portugal 6Institute of Metabolic Sciences, University of Cambridge, Cambridge CB2 0QQ, UK *Correspondence: dr360@medschl.cam.ac.uk DOI 10.1016/j.molcel.2010.11.010
2Department 1Kimmel
SUMMARY
Endoplasmic reticulum (ER) oxidation 1 (ERO1) transfers disuldes to protein disulde isomerase (PDI) and is essential for oxidative protein folding in simple eukaryotes such as yeast and worms. Surprisingly, ERO1-decient mammalian cells exhibit only a modest delay in disulde bond formation. To identify ERO1-independent pathways to disulde bond formation, we puried PDI oxidants with a trapping mutant of PDI. Peroxiredoxin IV (PRDX4) stood out in this list, as the related cytosolic peroxiredoxins are known to form disuldes in the presence of hydroperoxides. Mouse embryo broblasts lacking ERO1 were intolerant of PRDX4 knockdown. Introduction of wild-type mammalian PRDX4 into the ER rescued the temperature-sensitive phenotype of an ero1 yeast mutation. In the presence of an H2O2-generating system, puried PRDX4 oxidized PDI and reconstituted oxidative folding of RNase A. These observations implicate ER-localized PRDX4 in a previously unanticipated, parallel, ERO1-independent pathway that couples hydroperoxide production to oxidative protein folding in mammalian cells.
INTRODUCTION Disulde bond formation in the endoplasmic reticulum involves the transfer of electrons from the reduced cysteines of unfolded proteins to oxidoreductases of the protein disulde isomerase (PDI) family. This catalytic cycle is completed by the reoxidation of the reduced PDIs. Genetic studies in yeast and biochemical analysis in vitro have implicated the lumenal oxidase ERO1 in PDI reoxidation and disulde bond formation (reviewed in Tu and Weissman, 2004; Sevier and Kaiser, 2008). Yeast lacking ERO1 are nonviable, indicating that ERO1 is required for the essential process of disulde bond formation (Frand and Kaiser, 1998; Pollard et al., 1998). Mammals have two genes encoding proteins homologous to yeast ERO1,
known as ERO1a (or ERO1L) (Cabibbo et al., 2000) and ERO1b (or ERO1Lb) (Pagani et al., 2000). Biochemical studies in vitro and ability to complement the yeast mutation in vivo prove that the mammalian proteins also function as ER oxidases, accepting electrons from reduced PDI (Cabibbo et al., 2000; Mezghrani et al., 2001; Appenzeller-Herzog et al., 2008; Baker et al., 2008; Blais et al., 2010). The two isoforms have different tissue distribution: ERO1a is broadly expressed, whereas ERO1b expression is greatly enriched in the endocrine pancreas and in immunoglobulinsecreting cells (Pagani et al., 2000; Dias-Gunasekara et al., 2005; Zito et al., 2010). Compromise of ERO1b expression in mice delays oxidative folding of proinsulin (Zito et al., 2010). Surprisingly, compounding the loss of ERO1b by loss of ERO1a does not exacerbate the mild diabetic phenotype or the impaired insulin secretion of the single ERO1b mutant (Zito et al., 2010). Given the critical role of disulde bond formation in insulin biogenesis (Liu et al., 2007), this observation suggests the existence of pathways to mammalian ER oxidation that function in parallel to ERO1. Fly genetics also provides evidence for the existence of such parallel pathways. While the single ERO1 gene in that species (known as kiga) is essential to animal development, clones of homozygous mutant kiga cells are presumably capable of oxidative protein folding, as they expand normally in the wing disc and pupal notum (Tien et al., 2008). Various processes emerge as candidates for ERO1-independent disulde bond formation. Early studies had emphasized the potential for selective import of disulde-bonded glutathione into the ER as a driver of protein oxidation (Hwang et al., 1992). More recently Ruddock and colleagues have called attention to the potential role of direct oxidation of cysteine residues by H2O2 or dehydroascorbate as mechanisms for disulde bond formation (Karala et al., 2009; Saaranen et al., 2010). The secreted enzymes, QSOX1 and QSOX2, can catalyze the oxidation of cysteines on unfolded proteins, and an ER-localized fraction of QSOX could play a role in ER oxidation (Kodali and Thorpe, 2010). Finally, the vitamin K epoxide reductase can utilize oxidized vitamin K to drive PDI oxidation and thereby promote disulde bond formation (Li et al., 2010). Here we report on the application of an unbiased method for uncovering candidates in the parallel ERO1-independent
Molecular Cell
PRDX4 Promotes ER Disulde Bond Formation
pathway to protein oxidation and in the experimental indictment of the ER-localized peroxiredoxin IV (PRDX4) in that pathway. RESULTS PRDX4 Forms Stable Complexes with an Active-Site Trapping Mutant of PDI The minimal effect of ERO1 loss of function on disulde bond formation in higher eukaryotes was previously inferred from indirect measurements (Tien et al., 2008; Zito et al., 2010). To assess this process directly we monitored the disulde bond-dependent changes in mobility on nonreducing sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) as newly synthesized immunoglobulin M (IgM) monomers progress to dimers and decamers in pulse-chase-labeled lipopolysaccharide-induced blasts cultured from the spleens of wild-type or compound Ero1L;Ero1Lb mutant mice (lacking ERO1a and ERO1b protein). The rate of progression is nearly indistinguishable in the two genotypes (Figure 1A), conrming the existence of parallel pathways for disulde bond formation in the ER of ERO1-decient cells. ERO1 accepts electrons directly from the reduced C-terminal active site of PDI (Tsai and Rapoport, 2002; Wang et al., 2009). We reasoned that if a parallel pathway were to exploit a similar step in electron transfer, the acceptor protein (the oxidant of PDI) could be trapped in complex with a mutant PDI that is missing the resolving cysteine of its C-terminal active site. To establish this system, we expressed wild-type-, single (C400S)-, and double (C56S; C400S)-trapping mutant FLAG-M1 epitope-tagged human PDI proteins by transient transfection of HEK293T cells. Cell lysates were prepared in the presence of N-Ethylmaleimide (NEM) to quench free thiols, and protein complexes were immunopuried by the FLAG-M1 tag. Immunoblot of a reducing gel showed that the C-terminal trapping mutant PDIC400S associated with more ERO1a than the wildtype or the double-mutant PDIC56S,C400S (Figure 1B, upper panel). Immunoblot of the nonreducing gel showed that all the detectable ERO1a that copuried with PDIC400S was associated in a high-molecular-disulde-bonded complex (Figure 1B, lower panel). A species of similar mobility was also detected with the anti-FLAG antibody (Figure 1B, lower panel). These observations conrmed the utility of the trapping mutant PDIC400S in stably associating with a known ER oxidase. To identify other proteins that may associate with the trapping mutant PDIC400S in a disulde-bonded complex, we excised the region of the nonreducing SDS-PAGE containing the trapped complexes. After in-gel reduction, alkylation, and digestion with trypsin, the peptides eluted from the gel slices were analyzed by liquid chromatography (LC-MS/MS) (Figure 2A). The corresponding human proteins were sorted based on their exponentially modied protein abundance index (emPAI) (Figure 2B). Among the high emPAI scoring proteins discovered in complex with PDI, PRDX4 seemed the most likely to play a role in ER oxidation: like other members of the family of 2-cysteine peroxiredoxins, PRDX4 has been reported to reduce H2O2 to water by forming an intersubunit disulde bond (Hall et al., 2009; Ikeda et al., 2010; Tavender and Bulleid, 2010).
mut/mut
mut/mut
Figure 1. A Trapping Mutant of PDI Engages the Known Downstream Electron Acceptor, ERO1a, in Mammalian Cells
(A) Autoradiograph of IgM from metabolically labeled wild-type (a+/+;b+/+) and ERO1 mutant (amut/mut;bmut/mut) lipopolysaccharide blasts. Cells were labeled with [35S]methionine-cysteine for 15 min and chased for the indicated time with unlabeled media before lysis and immunoprecipitation. The upper panel is a radiograph of a nonreducing gel and the lower panel is a reducing gel. The migration of IgM monomers (m), dimers (2m), and pentamers of dimers (5m) is indicated. (B) Immunoblots of FLAG-tagged proteins (left) or endogenous ERO1a (right) immunopuried with the FLAG-M1 antibody from lysates of HEK293T cells that were untransfected or transfected with expression plasmids of the indicated proteins. The upper panels are of reducing and the lower panels are of nonreducing gels. The low-mobility complex containing ERO1aimmunoreactive material in complex with the FLAG-tagged PDI C-terminal active-site trapping mutant (FLAG-PDIC400S) is noted by an asterisk on both nonreducing gels.
Importantly, PRDX4 has been shown to reside in the ER lumen as a soluble protein (Tavender et al., 2008). Substantial amounts of endogenous PRDX4 immunoreactivity were recovered in complex with the tagged trapping mutant
Molecular Cell
PDIC400S in HEK293T cells (Figure 2C). On nonreducing gels the complex migrated slowly, consistent with disulde bonding (Figure 2D). The FLAG-M1 epitope of the tagged PDIC400S is exposed only after cleavage of the protrypsinogen signal sequence by the ER-localized leader sequence peptidase (Brizzard et al., 1994); thus the trapping of endogenous PRDX4 probably reects an interaction that took place in the lumen of the ER. Based on these hints, we hypothesized that PRDX4 may utilize lumenal H2O2 to form disuldes that it then passes on to PDI to oxidize ER client proteins. PRDX4 Buffers the Consequences of ERO1 Deciency in Mammalian Cells PRDX4 is not an essential ER enzyme, as mice with a deletion of the rst exon, encoding the signal sequence, are viable and fertile (Iuchi et al., 2009). However, we considered that if PRDX4 contributed signicantly to disulde bond formation, its function might become essential in cells lacking ERO1. Therefore, we compared the effect of PRDX4 knockdown on survival and growth of mouse embryo broblasts derived from wildtype and Ero1L;Ero1Lb mutant mice. Lentiviral vectors carrying a dominant selection marker imparting puromycin resistance were used to transduce small hairpin RNAs targeting mouse PRDX4. In both wild-type and double-mutant cells, transduction led to marked decline in endogenous PRDX4 signal (Figure 3A). In the ERO1 wild-type background, selection with puromycin resulted in a thick growth of PRDX4 knockdown cells, consistent with the mild phenotype of the PRDX4 mutation in mice. However, transduction of the shRNA to PRDX4 elicited growth of fewer puromycin-resistant colonies of Ero1L;Ero1Lb double-mutant mouse embryonic broblasts (MEFs), indicating that cells lacking ERO1 require PRDX4 for normal growth (Figures 3B and 3C). Cellular ultrastructure was impaired by PRDX4 shRNA. The mild ER dilation noted basally in the Ero1L;Ero1Lb double-mutant MEFs was conspicuously enhanced by further compromise of PRDX4 expression (Figure 3D). Together, these observations indicate that PRDX4 has an important role in cells lacking ERO1 and suggest that this essential role is involved in ER homeostasis. To further characterize the impact of PRDX4 loss in MEFs lacking ERO1, we examined their sensitivity to exogenously added dithiothreitol (DTT), a disulde-bond-reducing agent. Cells were exposed to a brief 30 min pulse of 5 mM DTT followed by washout with fresh media. The cell mass quantied 3 hr later was markedly lower in the ERO1 mutant cells compounded by PRDX4 knockdown (Figure 4A). ERO1-decient cells had baseline impairment in collagen secretion (another reection of ERO1s importance to mammalian physiology) that was further (3.8-fold) compromised by PRDX4 shRNA. A less conspicuous (1.8-fold) impairment of collagen secretion was also noted in ERO1 wild-type PRDX4 shRNA cells (Figure 4B). Both the secretory defect and the hypersensitivity of cells with a combined deciency of ERO1 and PRDX4 to DTT mirror the phenotype of yeast decient in ERO1 (Frand and Kaiser, 1998) and suggest a defect in disulde bond formation. To directly assess the impact of PRDX4 knockdown on the reduction-oxidation (redox) equilibrium in the ER of ERO1 mutant
Figure 2. A PDI Active-Site Trapping Mutant Engages PRDX4 in Mammalian Cells

(A) Coomassie-stained nonreducing SDS-PAGE of proteins immunopuried in complex with FLAG-tagged PDI trapping mutant from untransfected and transfected cells. The boxes demarcate the binning for the mass spectrometric protein analysis. (B) List of proteins identied by LC-MS/MS sequencing of tryptic peptides of endogenous proteins captured in a disulde-linked complex by a FLAG-M1tagged trapping mutant PDIC400S expressed in HEK293T cells (shown in A). Known cytosolic, nuclear, and mitochondrial proteins were removed from the list and the remaining proteins were sorted by descending exponentially modied protein abundance index (emPAI) (Ishihama et al., 2005). The proteins are identied by their International Protein Index accession number (IPI) and their common name. (C) Immunoblot of endogenous PRDX4 and FLAG-tagged wild-type PDI and trapping mutant PDIC400S immunopuried with the FLAG-M1 antibody from lysates of HEK293T cells that were untransfected or transfected with expression plasmids for the indicated proteins. The lower two panels are of the same proteins in the lysates before the IP (Input). The proteins shown were resolved on reducing SDS-PAGE. (D) FLAG and endogenous PRDX4 immunoblot of a nonreducing gel with samples as in (B). The low-mobility complex containing PRDX4 immunoreactive material in complex with FLAG-PDIC400S is noted by an asterisk on both nonreducing gels.
Molecular Cell
cells, we transduced cells with FLAG-M1-tagged, ER-localized reduction-oxidation sensitive green uorescent protein (roGFP-iE). This altered version of the uorescent protein is engineered to contain two exposed cysteines that form a relatively unstable disulde with a redox potential of 235 7 mV. roGFP-iE is thus suited to report on changes in the redox poise of oxidizing environments such as the ER (Lohman and Remington, 2008). The FLAG-M1-roGFP-iE immunopuried from the ER of cells with a combined deciency of ERO1 and PRDX4 was predominantly reduced (89%), whereas that puried from the ER of cells with intact PRDX4 was a mixture of reduced (55%) and oxidized protein (44%) (Figures 4C and 4D). Because the FLAG-M1 antibody only detects FLAG-roGFP-iE after signal peptide cleavage, this assay is not inuenced by variation in the efciency with which the protein translocates into the ER, and our observations strongly support an oxidative defect in ERO1 mutant cells lacking PRDX4. ER-Localized, Enzymatically Active PRDX4 Rescues a Lethal Mutation of Yeast ERO1 Yeast lack a known ER-localized peroxiredoxin. Therefore, we asked whether introducing mammalian PRDX4 into the yeast ER could rescue the lethal phenotype of a strong ERO1 mutation. The FLAG-M1-tagged coding sequence of mature mouse PRDX4 was fused in frame to the cleavable yeast proalpha mating factor signal sequence, and the chimeric protein was expressed from the yeast CUP1 promoter as a plasmid episome. Upon transduction of this plasmid, haploid yeast with the highly penetrant temperature-sensitive ero1-1 mutation (Frand and Kaiser, 1998) acquired the ability to grow at the nonpermissive temperature (Figure 5A). Expression of PRDX4 had no deleterious effect on growth at the permissive temperature, whereas the growth of the rescued ero1-1 yeast at the nonpermissive temperature approached their growth rate at the permissive temperature (Figure 5B). Basal activity of the CUP1 promoter was sufcient; addition of copper did not enhance the rescue. To conrm that the robust rescue noted above was imparted by persistent expression of the PRDX4 plasmid, we exploited the fact that the URA3(+) auxotrophic marker can be used for both positive and negative selection for the presence of the plasmid. URA3-decient parental and ero1-1 yeast that had been transformed by the empty pRS316- or PRDX4-expressing plasmid were selected for presence of the plasmid in media lacking uracil at the permissive temperature. They were then spotted onto synthetic, minimal medium (SD) plates that either did or did not contain 5-uoroorotic acid (FOA), which is converted to the toxin 50 uorouracil by orotidine-50 -phosphate decarboxylase (encoded by URA3). At the permissive temperature both the wild-type and the ero1-1 mutant yeast rid themselves of the URA3+ PRDX4 expression plasmid and survived the FOA selection. However, at the nonpermissive temperature only the wild-type yeast grew on
Figure 3. PRDX4 Buffers the Consequences of ERO1 Deciency in MEFs

(A) Immunoblot of endogenous PRDX4, ERO1a, and actin in wild-type (a+/+;b+/+) and ERO1 mutant (amut/mut;bmut/mut) MEFs 4 days after transduction with a puror-marked lentivirus carrying a irrelevant insert (mock) or three different short hairpin RNAs directed to mouse PRDX4. (B) Crystal violet-stained plates following 2 weeks of puromycin selection after seeding with 8 3 103 (low density) or 4 3 104 (high density) MEFs of the indicated genotype and transduction with puror-tagged lentivirus carrying the indicated shRNA. (C) Bar diagram of the ratio of cell mass accrued in the ERO1-double-mutant versus wild-type MEFs from the experiment shown in (B). The ratio was normalized to 1 in the cells transduced with the irrelevant shRNA. Shown are means SEM of a typical experiment reproduced three times (n = 3, *p < 0.05).
(D) Transmission electron micrographs of MEFs of the indicated genotype after transduction with a lentivirus containing an irrelevant shRNA or an shRNA directed to PRDX4. Note the dilation of the ER in the mutant cells transduced with the shRNA to PRDX4.
Molecular Cell
FOA (Figure 5C). These ndings indicate that rescue of ero1-1 is due to persistent expression of the PRDX4- expression plasmid. In addition to their thiol-mediated peroxidase activity, members of the peroxiredoxin family also have chaperone activity (Hall et al., 2009). To determine whether the rescue of the growth defect of ero1-1 mutant is mediated by the peroxidase activity or by the chaperone activity, we mutated the two active-site cysteines of PRDX4 in the yeast expression plasmid. To test whether localization to the ER is required for rescue, we removed the signal sequence from the protein. All three mutants were well expressed in the ero1-1 yeast, but only the ER-localized wild-type PRDX4 rescued the temperature-sensitive phenotype of the ero1-1 mutant (Figure 5D). We were unable to come to a rm conclusion in regard to the source of H2O2 driving oxidation of PRDX4 in yeast or in mammalian cells. Respiratory-decient ero1-1 yeast compromised by deletion of coq3 were nonetheless rescued by PRDX4, indicating that an intact respiratory chain is not a requisite for PRDX4-mediated oxidation in yeast (Figure S1A, available online). Similarly, ERO1-decient MEFs are not conspicuously hypersensitive to compromise of mitochondrial gene expression by culture in media containing ethidium bromide, nor are they hypersensitive to diphenylene iodonium (DPI), an inhibitor of NADPH oxidase (data not shown). Finally, this theme of redundancy in H2O2 source(s) is also reected by the apparent lack of effect of ERO1 genotype on the rate of endogenous PRDX4 reoxidation after a pulse of DTT (Figure S1B). PRDX4 Catalyzes H2O2 and PDI-Dependent Oxidative Refolding of RNase A In Vitro Ruddock and colleagues have discovered that H2O2 can directly serve as an electron acceptor in PDI-mediated oxidative refolding of RNase A in vitro (Karala et al., 2009). The ability of PRDX4 to accelerate this thermodynamically favored process would serve as a strong test of the hypothesis that PRDX4 accelerates disulde bond formation by converting hydroperoxides to disuldes in the ER lumen. First we tested PDIs ability to be oxidized by PRDX4 in vitro. Bacterially expressed PDI (150 mM) was reduced by DTT, desalted by gel ltration, and reacted with bacterially expressed wild-type PRDX4 (5 mM) or the active-site mutants PRDX4C127S and PRDX4C248S in the absence or presence of an H2O2-generating system consisting of glucose (2.5 mM) and glucose oxidase (10 milliunits/ml) (Figure 6A). The time-dependent oxidation of free thiols was assessed by loss of their reactivity with the Ellmans reagent [5,50 -dithiobis(2-nitrobenzoate, DTNB)]. Oxidation was dependent on a source of H2O2 and was accelerated by wild-type PRDX4 but not by the two active-site mutants (Figure 6B). In the absence of PDI, PRDX4 was unable to oxidize
proteins were resolved by SDS-PAGE under nonreducing or reducing conditions. The position of the reduced and oxidized FLAG-M1-roGFP-iE in a typical experiment reproduced three times is shown. (D) Quantitative representation of the data from the experiment represented in (C). The bar diagram shows the ratio of reduced and oxidized roGFP in the Ero1amut/mut;Ero1bmut/mut compound mutant cells transduced with mock or PRDX4 RNAi lentivirus (KD1) (n = 3, *p < 0.05, comparing oxidized fraction in the two genotypes).
Figure 4. Hypersensitivity to Reducing Agents, Defective Collagen Secretion, and a More Reduced ER Redox Poise in ERO1-Decient Cells Lacking PRDX4
(A) Bar diagram of cell mass of wild-type (WT) or Ero1amut/mut;Ero1bmut/mut compound mutant cells (MUT) transduced with mock or PRDX4 RNAi lentivirus (KD1) remaining 3 hr after a 30 min pulse with DTT (5 mM) followed by recovery in normal media, normalized to the cell mass of a parallel culture of the same cells that had not been exposed to DTT. Shown are aggregate means SEM from a typical experiment conducted in triplicate (n = 3; *p < 0.05) and reproduced three times. (B) Bar diagram of soluble collagen secreted into the conditioned media from the cells described in (A) normalized to the number of live cells in the culture. Shown are means SEM from three experiments conducted on different occasions in triplicate (n = 9; *p < 0.01). (C) Immunoblot of FLAG-M1-roGFP-iE immunopuried from the ER of ERO1decient MEFs with normal levels of PRDX4 or PRDX4 knockdown. The puried
Molecular Cell
A
pRS316 PRDX4
24C
37C ero1-1
C
24C pRS316 PRDX4 pRS316 PRDX4 pRS316 PRDX4 pRS316 PRDX4
Time(h) 37C ERO1
ero1-1
ERO1 FOA ero1-1
reduced glutathione, indicating selectivity for the source of electrons in this reaction (Figure 6C). To determine whether the accelerated oxidation of PDI would give rise to accelerated oxidation of reduced and denatured RNase A, we monitored the latters mobility on nonreduced SDS-PAGE at different time points in an in vitro refolding reaction. In reactions containing an H2O2-generating system with PRDX4 and PDI, the low-mobility, reduced RNase A forms were rapidly converted to high-mobility forms that comigrated with native RNase A. Omitting any of the three components markedly attenuated this conversion, and active-site mutants of PRDX4 were ineffective (Figure 6D). These observations indicate that efcient PRDX4-mediated oxidative folding of RNase A requires active peroxiredoxin, an electron acceptor (H2O2, in this case), and an oxidoreductase (PDI). To determine whether the refolded RNase A was functional, or whether PRDX4-mediated refolding favored scrambled RNase A, we monitored RNase A enzymatic activity by using cyclic cytidine 50 -monophosphate (cCMP) as a substrate (Lyles and Gilbert, 1991). The time-dependent acquisition of RNase A activity correlated well with the rate of refolding on the SDS-PAGE, indicating that the disulde bonds formed in the presence of PRDX4 and PDI are native (Figure 6E). Together these observations establish that PRDX4 possesses the enzymatic activity required to substitute for the lack of ERO1. DISCUSSION We have discovered that PRDX4 buffers the growth-inhibitory consequences of ERO1 deciency in settings ranging from yeast to cultured mouse cells. Such buffering requires expression of PRDX4 in the lumen of the ER and is dependent on the two cysteines of PRDX4 that are involved in its peroxidase activity. In vitro, PRDX4 is able to accelerate the H2O2-dependent oxidative folding of reduced and denatured RNase A in the presence of PDI. Together, these observations point to an unanticipated role for PRDX4 in converting ER lumenal H2O2 to disuldes and in handing these disuldes over to PDI for oxidation of newly synthesized secretory proteins (Figure 7). Neither yeasts nor worms, species in which ERO1 is essential (Frand and Kaiser, 1998; Pollard et al., 1998), have genes encoding predicted ER-localized peroxiredoxins, whereas more complex eukaryotes, such as ies and mice, in which ERO1 is not essential to cell viability, have PRDX4 genes. This correlation, together with the relatively broad expression of PRDX4 in mouse tissues (Iuchi et al., 2009) and the ability of PRDX4 to rescue ERO1 deciency in the heterologous yeast system, suggests that PRDX4 may be broadly implicated in the observed redundancy of ERO1 in higher eukaryotes.
D
pRS316 PRDX4 PRDX4C127S PRDX4C248S PRDX4SP ero1-1
O.D. 600nm
PRDX4 IB: antiFLAG
Figure 5. ER-Localized, Enzymatically Active PRDX4 Rescues a Lethal Mutation of Yeast ERO1
(A) Photomicrographs of serial dilutions of untransformed ero1-1 mutant yeast or transformed with yeast expression plasmids lacking (pRS316) or containing the PRDX4 coding sequence cultured at the permissive temperature of 24 C or the nonpermissive temperature of 37 C. (B) Plot of absorbance at 600 nm of cultures inoculated with ero1-1 yeast transformed with the expression plasmids lacking (pRS316) or containing the PRDX4 coding sequence and cultured at the permissive temperature of 24 C or the nonpermissive temperature of 37 C. Shown are means SEM of a typical experiment conducted in triplicate (n = 3, *the last four time points, p < 0.001 by two-way ANOVA). (C) Photomicrographs of serial dilutions of ero1-1 yeast transformed with pRS316 or the PRDX4 expressing plasmids and plated in the absence or presence of the toxin FOA at 24 C or 37 C. Note the inability of the ero1-1 yeast
D X PR 4 D X4 C 12 PR 7S D X4 C PR 24 8S D X pR 4 SP S3 16
75 25 75 25 ponceau
PR
transformed with the PRDX4 expression plasmid to survive on FOA at the nonpermissive temperature. (D) Photomicrographs of serial dilutions of ero1-1 yeast transformed with pRS316 or plasmids expressing wild-type or the indicated mutations in PRDX4 (DSP: lacks the signal peptide required for ER import of PRDX4). The lower panel is an immunoblot of FLAG-tagged PRDX4 from the four transformed strains and the parental ero1-1 yeast.
Molecular Cell
Peroxiredoxins are pentamers of homodimers. Their catalytic mechanism entails a nucleophilic attack by a peroxidic cysteine (C127, in mouse PRDX4) on a hydroperoxide (probably H2O2), leading to the formation of a sulfenic acid intermediate that is resolved by a second active-site cysteine (C248) from an adjoining protomer. The resulting intersubunit disulde is later reduced by a thioredoxin-like protein, completing the catalytic cycle (Hall et al., 2009). Our in vitro studies and the crystal structure of human PRDX4 in the reduced state (PDB: 2PN8) are both consistent with this catalytic mechanism and indicate that unlike ERO1, PRDX4 cannot utilize molecular oxygen as the electron acceptor and must instead rely on a source of hydroperoxides. ERO1 generates one molecule of H2O2 for every disulde bond formed (Gross et al., 2006). Recent measurements suggest that despite being a nonpolar molecule and thus freely diffusible across membranes, H2O2 produced in the ER by ERO1 activity is consumed in that compartment (Enyedi et al., 2010). PRDX4 would be well positioned to use that peroxide to generate another pair of disuldes, enhancing the efciency of oxidative protein folding (Figure 7). Such an arrangement would have the additional benet of reducing reactive H2O2 to water, as suggested by Bulleid and colleagues (Tavender and Bulleid, 2010). In this vein, it is interesting to note that in worms and yeast, which lack a PRDX4-mediated pathway to eliminate ERO1-generated H2O2, attenuation of ERO1 activity provides some protection against the lethality of severe unfolded protein stress in the ER (Haynes et al., 2004; Marciniak et al., 2004). However, no such benet was observed in the insulin-producing b cells of the mouse pancreas, which have ample PRDX4 (Zito et al., 2010, and unpublished data). This suggests that the high levels of PRDX4 in the pancreas (Iuchi et al., 2009) and b cells efciently scavenge and utilize the H2O2 produced by ERO1. Clearly a different source of H2O2 is driving PRDX4-dependent disulde bond formation in cells lacking ERO1. Mitochondria are candidates, and their involvement may explain previous reports of mitochondrial respiration-dependent production of disuldes in mammalian cells (Yang et al., 2007). Junctions known to exist between mitochondria and ER membranes (Kornmann et al., 2009) could serve to channel H2O2 into the ER (and away from
E
(C) Time-dependent change in absorbance of Ellmans reagent in reaction mixes containing 150 mM reduced glutathione in the presence of PDI (10 mM), PRDX4 (5 mM), or both enzymes. Glucose (2.5 mM) and glucose oxidase (GO, 10 mU/ml) were included as a source of H2O2 in all reactions. The absorbance of each reaction mixture at t = 0 is set at 100%. Shown are means SEM of a typical assay conducted in triplicate and reproduced three times (n = 3, **p < 0.001). (D) Coomassie-stained SDS-PAGE of reduced and denatured RNase A (25 mM) after reacting for the indicated time with reduced PDI (7 mM) and PRDX4 (5 mM) in the presence of glucose (2.5 mM) and glucose oxidase (10 mU/ml) as a source of H2O2. Lane 13 in the top panel or lane 9 in the two lower panels contains a sample of native oxidized RNase A to serve as a reference. (E) Time-dependent change in RNase activity measured spectrophotometrically by absorbance at 296 nm with cCMP (4.5 mM) as substrate. The refolding of RNase A was initiated at t = 0 in reactions as in (D). Enzymatic activity of an equal amount of native RNase A is set as 100%. Shown are means SEM of a typical experiment reproduced four times (n = 3, *p < 0.05, **p < 0.001 by two-way ANOVA).
Figure 6. PRDX4 Catalyzes H2O2- and PDI-Dependent Oxidative Refolding of RNase A In Vitro
(A) Coomassie-stained reducing SDS-PAGE of puried PDI, wild-type, and mutant PRDX4-GST tagged proteins used in the experiments shown below. (B) Time-dependent change in absorbance of Ellmans reagent reacted with reduced PDI (150 mM) after introduction of glucose (2.5 mM) and glucose oxidase (GO, 10 mU/ml) as a source of H2O2 and the indicated wild-type or mutant PRDX4 proteins shown in (A) (5 mM). The absorbance of each reaction mixture at t = 0 is set at 100%. Shown are means SEM of a typical assay conducted in triplicate and reproduced four times (n = 3, *p < 0.05, **p < 0.001).
Molecular Cell
Figure 7. Model for PRDX4-Mediated Oxidative Protein Folding in the ER

Disulde bond formation (step I) leaves PDI in a reduced state. PDI is reoxidized by reducing the active-site disulde (formed between two PRDX4 protomers) in step II. PRDX4 is reoxidized by interaction of its peroxidic cysteine (C127) with H2O2, releasing one molecule of water. The sulfenic acid is resolved by C287, regenerating the disulde and releasing the second molecule of water (step III). PRDX4 may utilize H2O2 produced by ERO1 as well as H2O2 produced by mitochondrial respiration or NADPH oxidases.
cytosolic peroxiredoxins) to be used for PRDX4-mediated oxidative protein folding (Figure 7). In b cells, the tight coupling of mitochondrial respiration to nutrient availability may provide a mechanism for accelerated disulde bond formation during proinsulin synthesis, as the latter is translationally activated by nutrients (Itoh and Okamoto, 1980). Such an arrangement may also clarify an interesting peculiarity of b cells in that they contain superoxide dismutases (that convert oxygen radicals to H2O2) but lack glutathione peroxidase or catalase activity to break down H2O2 (Tiedge et al., 1997). This feature may have evolved to enable b cells to efciently harness H2O2 for disulde bond formation required during proinsulin folding. H2O2 may also be produced by NADPH oxidases. The latter are integral membrane proteins activated posttranslationally by a variety of signals, including growth factor receptors (Goldstein et al., 2005). Cytosolic peroxiredoxins channel the H2O2 produced by NADPH oxidases to regulate the activity of signaling molecules such as oxidation-sensitive phosphatases (Woo et al., 2010). It is tempting to speculate that a conceptually similar diffusion of H2O2 into the ER from ER-localized NADPH oxidases may couple extracellular signaling to ER protein oxidation. Two-cysteine peroxiredoxins are subject to inactivation by formation of a sulnic acid on the peroxidic cysteine. In the cytosol this reaction is reversed by a sulredoxin that reduces the sulfenic acid in an ATP-dependent reaction (Biteau et al., 2003). There are no known sulredoxins in the ER, suggesting that inactivation of PRDX4 by formation of a cysteine-sulnic acid might be irreversible (Tavender and Bulleid, 2010). Such irreversible inactivation of PRDX4 by excess H2O2 may account for the seemingly paradoxical induction of unfolded protein stress in the ER under hyperoxidizing conditions and its relief by antioxidants (Malhotra et al., 2008). Alternatively, inactivation of PRDX4 by elevated levels of H2O2 may regulate the disuldebased redox poise of the ER, akin to the allosteric inhibition of ERO1 by oxidized PDI (Sevier et al., 2007). Our discovery of a PRDX4-dependent pathway operating in parallel to ERO1-mediated protein oxidation in the ER does not preclude the existence of other parallel pathways that might be revealed by the phenotype of compound ERO1 and PRDX4 deciency in other systems. Regardless of whether further redundancy exists, this study points to an unanticipated level of
complexity of oxidative protein folding in higher eukaryotes and suggests a mechanism for regulating this conserved process by the availability of H2O2.
EXPERIMENTAL PROCEDURES Isolation, Culture, and Metabolic Labeling of LPS Blasts Splenic cells from wild-type and compound homozygous Ero1a;Ero1b mutant mice (Zito et al., 2010) were cultured in RPMI-1640 at a density of 106 cells/ml and exposed to 50 mg/ml lipopolysaccaride from Escherichia coli (Sigma, L2755) for 23 days. After a 15 min labeling pulse with 60 mCi/ml of [35S]methionine/cysteine (Perkin Elmer; specic activity, > 1,000 Ci/mmol), a cold chase was performed in complete media followed by lysis and immunopurication of the radiolabeled IgM with 5 ml of rabbit anti-mouse IgM (Rockland). Immunoprecipitated proteins were resolved on reducing and nonreducing 10% SDS-PAGE and revealed by autoradiography with a Typhoon Phosphoimager (GE Healthcare). Immunopurication of Complexes Trapped by FLAG-PDI Expression plasmids encoding ER-localized, N-terminally FLAG-tagged human PDI (PDIA1, 18508) were constructed in the pFLAG-CMV1 vector (Sigma). The C-terminal FLAG-PDIC400S and double-trapping mutant FLAGPDIC56S;C400S were made by QuikChange Site-Directed Mutagenesis (Stratagene). Transfected HEK293T cells from four conuent 100 mm plates were washed in phosphate-buffered saline (PBS) with 20 mM NEM and lysed in 0.3% Triton X-100, 150 mM NaCl, 20 mM HEPES (pH 7.4), 10 mM CaCl2, 20 mM NEM, and protease inhibitors. The FLAG-tagged proteins were immunopuried with FLAG-M1 afnity gel (Sigma) in an overnight incubation and eluted in 10 mM EDTA, 20 mM Tris (pH 7.5), and 0.002% Tween 20. Ten percent of the eluted material was immunoblotted with anti-FLAG M2 (Sigma), rabbit anti-ERO1a antiserum (Zito et al., 2010), or rabbit anti PRDX4 (a gift of Junichi Fujii) (Matsumoto et al., 1999) after reducing and nonreducing SDS-PAGE. The residual 90% was resolved on nonreducing SDS-PAGE stained lightly with Coomassie, and the region of the gel containing complexes larger than the FLAG-PDIC400S bait was excised, reduced in 10 mM DTT, alkylated with 20 mM NEM, and analyzed by mass spectrometry. Mass Spectrometry A linear ion trap (LTQ)-Orbitrap (ThermoFisher, Bremen, Germany) equipped with a nano-ESI source was used for all analyses. Samples were directly infused via a Eksigent 2DLC system (Eksigent Technologies, Dublin, CA) equipped with a 12 cm PicoFrit self-packed-column PicoTip emitter (75 mm ID, 10 mm tip diameter from New Objective, Woburn, MA) packed in house with ReproSilPur C18-AQ beads (3 mm from Dr. Maisch GmbH, Ammerbuch, Germany). Samples were applied with direct loading onto the analytical column (0.8 ml/ min for 25 min with mobile phase A; 0.1% formic acid). The ow rate was then changed to 0.3 ml/min and peptides were eluted with a gradient of 1%76% acetonitrile in 0.1% formic acid over 120 min. Mass spectra were acquired in data-dependent mode with one 60,000 resolution MS survey
Molecular Cell
scan in the Orbitrap, followed by up to nine subsequent MS/MS scans in the LTQ from the nine most intense peaks detected in the survey scan. MS survey scans were acquired in prole mode and the MS/MS scans were acquired in centroid mode. Generic MASCOT format les were generated from raw data with DTA SuperCharge (version 2.0a7) and searched by using MASCOT software (version 2.3, Matrix Science, London, UK). The human IPI database was searched with peptide mass tolerance of 20 ppm and fragment mass tolerance of 0.6 Da. Trypsin specicity was set to a maximum of one missed cleavage and NEM, NEM + water, and oxidation were set as variable modications. Lentiviral Transduction and Analysis of Wild-type and Ero1a;Ero1b Mutant MEFs Wild-type and compound homozygous Ero1a;Ero1b mutant MEFs, isolated at embryonic day 13.5, were immortalized with SV-40 large T antigen and cultured in DMEM supplemented with 25 mM glucose, 10% FCS, and nonessential amino acids. Prdx4 knockdown was achieved by using Mission shRNA-encoding lentiviruses directed to mouse Prdx4 mRNA (Sigma SHCLNG-NM_016764) following the manufacturers instructions. Knockdown clone KD1 was targeted with shRNA TRCN0000055339, KD2 with TRCN0000055341, and KD3 TRCN0000055339. After transduction and selection with puromycin at 4 mg/ml for 2 weeks, the cells (in triplicate plates) were xed, stained with crystal violet, and photographed. KD2 had off-pathway toxicity and was not used in the phenotypic analysis. Cell mass was quantied by solubilizing the dye in 0.2% Triton X-100 and measuring the absorbance at 590 nm. To assess the extent of knockdown achieved, we transduced 5 3 106 cells, selected them for 4 days with puromycin (a period sufcient to kill more than 95% of the nontransduced cells), and immunoblotted the cell lysate for PRDX4. Secretion of soluble collagen was measured in the 24 hr conditioned media of MEFs by using the Sircol colorimetric assay (Biocolor, Ireland) according to the manufacturers instruction. MEFs were cultured in media with 10% FCS for 48 hr before the assay and switched to media with 0.5% FCS to collect the conditioned media. The sensitivity of cells to reductive stress was measured by crystal violet staining of the cell mass that remained on the plate 3 hr after a 30 min challenge with 5 mM DTT. The roGFP-iE cDNA (Lohman and Remington, 2008) was ligated in frame with the FLAG-M1 tag in a lentiviral vector to transduce cells of the indicated genotype. Cells were lysed in the presence of 20 mM NEM, and the FLAGM1-roGFP-iE was immunoprecipitated with the FLAG-M1 monoclonal antibody, resolved on 15% SDS-PAGE in the absence or presence of DTT, and blotted with FLAG M2. Cell pellets were xed by immersion in 2% glutaraldehyde and 2.0% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.2), postxed in 1% osmium tetroxide, and en bloc-stained with 1% uranyl acetate. The sample was dehydrated in ethanol embedded in Epon (Electron Microscopy Sciences, Hateld, PA). Ultrathin sections were poststained with uranyl acetate and lead citrate and examined with an electron microscope (CM12, FEI, Eindhoven, The Netherlands) at 120 kV. Images were recorded digitally with a camera system (Gatan, 4k 3 2.7k) with Gatan DigitalMicrograph software (Gatan Inc., Pleasanton, CA). Yeast Strains and Plasmids The temperature-sensitive yeast strain CKY598 bearing the ero1-1 mutation (C. Kaiser Lab at MIT) and the wild-type strain were transformed with PRDX4-pRS316, in which the cleavable yeast a-mating-factor signal sequence is fused in frame with FLAG-M1-tagged mature wild-type or active-site mutant mouse PRDX4 (37274) and expression driven by the yeast CUP1 promoter and termination sequences from the CPY1 gene. A cytosolic version was made by replacing the yeast amating factor signal sequence with an initiator methionine (yielding a protein nearly identical in size to mature ER version). Transformed yeast were selected on minimal media lacking uracil or SD media with and without 0.1% FOA (Sigma) and incubated at the permissive (24 C) or nonpermissive temperature (37 C). Protein Oxidation In Vitro Mouse PRDX4WT and the mutants PRDX4C127S and PRDX4C248S were expressed as glutathione S-transferase (GST) fusion proteins in E. coli Rosetta
(D3) strain followed by glutathione afnity chromatography and gel ltration on Superdex 200. Human PDI (PDIA1 18-508), a gift of C. Thorpe, University of Delaware, was expressed in E. coli BL21 (D3) strain, puried with Ni-NTA afnity chromatography, and dialyzed into the reaction buffer. PDI was reduced by incubation with 50 mM of DTT, 1 hr at room temperature, and then desalted on PD-10 gel ltration column (GE Healthcare). Oxidation in vitro was performed in the presence of glucose (2.5 mM) and glucose oxidase (10 mu/ml) type II from Aspergillus niger (Sigma) in 80 mM sodium phosphate, pH = 7.0. Where indicated reduced glutathione was added to the reaction. Free thiols were quantied at different time points by combining 100 ml of 0.5 mM Ellmans reagent [(5,50 -dithiobis-(2-nitrobenzoic acid)] with 30 ml of sample and absorbance at 405 nm measured in a Tecan Innite 500 plate reader. In some assays reduced glutathione was added as an upstream electron donor. RNase A (from bovine pancreas, Roche) was reduced and denaturated overnight in the presence of 6 M guanidinium hydrochloride and 140 mM DTT in 50 mM phosphate buffer, pH 7, and desalted into the assay buffer on PD-10 column before the refolding assays were run. Reduced RNase A (25 mM) was then incubated with PDI (7 mM) and PRDX4 (5 mM) in the presence of glucose (2.5 mM) and glucose oxidase (10 mu/ml) in 80 mM phosphate buffer, pH 7. Aliquots of 20 ml were withdrawn at different time points and reacted with 2 ml of 0.5 M NEM (NEM) for 15 min on ice. Refolding of RNaseA was then analyzed on an 18% Tris-tricine gel under nonreducing conditions. The functional state of RNase A was measured by adding cytidine 20 ,30 cyclic monophosphate monosodium salt (cCMP from Sigma) and reading the development of absorbance from its hydrolytic product at 296 nm, in a temperature-controlled Agilent diode array spectrophotometer at 25 C, as described (Lyles and Gilbert, 1991). Statistical Analysis All results are expressed as means SEM. Two-tailed Students t tests were performed to determine p values for paired samples, and two-way analysis of variance (ANOVA) was performed for the experiments that employed more than one independent variable. SUPPLEMENTAL INFORMATION Supplemental information includes one gure and can be found with this article online at doi:10.1016/j.molcel.2010.11.010. ACKNOWLEDGMENTS We thank C. Sevier and C. Kaiser (Massachusetts Institute of Technology) for the CKY598 ero1-1 mutant yeast and advice on its use, A. Liang and E. Roth from the imaging core facility of NYU School of Medicine for processing and acquiring the EM images, J. Fujii (Yamagata University) for the antiserum to PRDX4, C. Thorpe (University of Delaware) for the gift of the PDI expression plasmid, J. Remington (University of Oregon) for the roGFP-iE, A. Tzagaloff (Columbia University) for the coq3 deleted yeast and advice on its use, M. Costanzo (University of Toronto) for the sec72D;ero1-1 compound haploid, and H. Klein and A. Epshtein (NYU) for advice and help with yeast genetics. This work was supported by an EMBO long-term fellowship ALTF649-2008 to E.Z., Fundacao para a Ciencia e Tecnologia, Portugal (SFRH/BSAB/922/ 2009, PTDC/QUI/73027/2006 and IBB/CBME LA) to E.P.M., by NIH grants DK47119, DK075311, and ES08681 to D.R. and NS050276, CA016087, and a grant from the 100 Women In Hedge Funds Foundation to T.A.N. D.R. is a Wellcome Trust Principal Research Fellow. Received: June 17, 2010 Revised: August 11, 2010 Accepted: September 13, 2010 Published: December 9, 2010 REFERENCES Appenzeller-Herzog, C., Riemer, J., Christensen, B., Srensen, E.S., and Ellgaard, L. (2008). A novel disulphide switch mechanism in Ero1alpha balances ER oxidation in human cells. EMBO J. 27, 29772987.
Molecular Cell
Baker, K.M., Chakravarthi, S., Langton, K.P., Sheppard, A.M., Lu, H., and Bulleid, N.J. (2008). Low reduction potential of Ero1alpha regulatory disulphides ensures tight control of substrate oxidation. EMBO J. 27, 29882997. Biteau, B., Labarre, J., and Toledano, M.B. (2003). ATP-dependent reduction of cysteine-sulphinic acid by S. cerevisiae sulphiredoxin. Nature 425, 980984. Blais, J.D., Chin, K.T., Zito, E., Zhang, Y., Heldman, N., Harding, H.P., Fass, D., Thorpe, C., and Ron, D. (2010). A small molecule inhibitor of endoplasmic reticulum oxidation 1 (ERO1) with selectively reversible thiol reactivity. J. Biol. Chem. 285, 2099321003. Brizzard, B.L., Chubet, R.G., and Vizard, D.L. (1994). Immunoafnity purication of FLAG epitope-tagged bacterial alkaline phosphatase using a novel monoclonal antibody and peptide elution. Biotechniques 16, 730735. Cabibbo, A., Pagani, M., Fabbri, M., Rocchi, M., Farmery, M.R., Bulleid, N.J., and Sitia, R. (2000). ERO1-L, a human protein that favors disulde bond formation in the endoplasmic reticulum. J. Biol. Chem. 275, 48274833. Dias-Gunasekara, S., Gubbens, J., van Lith, M., Dunne, C., Williams, J.A., Kataky, R., Scoones, D., Lapthorn, A., Bulleid, N.J., and Benham, A.M. (2005). Tissue-specic expression and dimerization of the endoplasmic reticulum oxidoreductase Ero1beta. J. Biol. Chem. 280, 3306633075. Enyedi, B., Varnai, P., and Geiszt, M. (2010). Redox state of the endoplasmic reticulum is controlled by Ero1L-alpha and intraluminal calcium. Antioxid. Redox Signal. 13, 721729. Frand, A.R., and Kaiser, C.A. (1998). The ERO1 gene of yeast is required for oxidation of protein dithiols in the endoplasmic reticulum. Mol. Cell 1, 161170. Goldstein, B.J., Mahadev, K., Wu, X., Zhu, L., and Motoshima, H. (2005). Role of insulin-induced reactive oxygen species in the insulin signaling pathway. Antioxid. Redox Signal. 7, 10211031. Gross, E., Sevier, C.S., Heldman, N., Vitu, E., Bentzur, M., Kaiser, C.A., Thorpe, C., and Fass, D. (2006). Generating disuldes enzymatically: reaction products and electron acceptors of the endoplasmic reticulum thiol oxidase Ero1p. Proc. Natl. Acad. Sci. USA 103, 299304. Hall, A., Karplus, P.A., and Poole, L.B. (2009). Typical 2-Cys peroxiredoxins structures, mechanisms and functions. FEBS J. 276, 24692477. Haynes, C.M., Titus, E.A., and Cooper, A.A. (2004). Degradation of misfolded proteins prevents ER-derived oxidative stress and cell death. Mol. Cell 15, 767776. Hwang, C., Sinskey, A.J., and Lodish, H.F. (1992). Oxidized redox state of glutathione in the endoplasmic reticulum. Science 257, 14961502. Ikeda, Y., Ito, R., Ihara, H., Okada, T., and Fujii, J. (2010). Expression of N-terminally truncated forms of rat peroxiredoxin-4 in insect cells. Protein Expr. Purif. 72, 17. Ishihama, Y., Oda, Y., Tabata, T., Sato, T., Nagasu, T., Rappsilber, J., and Mann, M. (2005). Exponentially modied protein abundance index (emPAI) for estimation of absolute protein amount in proteomics by the number of sequenced peptides per protein. Mol. Cell. Proteomics 4, 12651272. Itoh, N., and Okamoto, H. (1980). Translational control of proinsulin synthesis by glucose. Nature 283, 100102. Iuchi, Y., Okada, F., Tsunoda, S., Kibe, N., Shirasawa, N., Ikawa, M., Okabe, M., Ikeda, Y., and Fujii, J. (2009). Peroxiredoxin 4 knockout results in elevated spermatogenic cell death via oxidative stress. Biochem. J. 419, 149158. Karala, A.R., Lappi, A.K., Saaranen, M.J., and Ruddock, L.W. (2009). Efcient peroxide-mediated oxidative refolding of a protein at physiological pH and implications for oxidative folding in the endoplasmic reticulum. Antioxid. Redox Signal. 11, 963970. Kodali, V.K., and Thorpe, C. (2010). Oxidative protein folding and the quiescinsulfhydryl oxidase family of avoproteins. Antioxid. Redox. Signal. 13, 12171230.
Kornmann, B., Currie, E., Collins, S.R., Schuldiner, M., Nunnari, J., Weissman, J.S., and Walter, P. (2009). An ER-mitochondria tethering complex revealed by a synthetic biology screen. Science 325, 477481. Li, W., Schulman, S., Dutton, R.J., Boyd, D., Beckwith, J., and Rapoport, T.A. (2010). Structure of a bacterial homologue of vitamin K epoxide reductase. Nature 463, 507512. Liu, M., Hodish, I., Rhodes, C.J., and Arvan, P. (2007). Proinsulin maturation, misfolding, and proteotoxicity. Proc. Natl. Acad. Sci. USA 104, 1584115846. Lohman, J.R., and Remington, S.J. (2008). Development of a family of redoxsensitive green uorescent protein indicators for use in relatively oxidizing subcellular environments. Biochemistry 47, 86788688. Lyles, M.M., and Gilbert, H.F. (1991). Catalysis of the oxidative folding of ribonuclease A by protein disulde isomerase: pre-steady-state kinetics and the utilization of the oxidizing equivalents of the isomerase. Biochemistry 30, 619625. Malhotra, J.D., Miao, H., Zhang, K., Wolfson, A., Pennathur, S., Pipe, S.W., and Kaufman, R.J. (2008). Antioxidants reduce endoplasmic reticulum stress and improve protein secretion. Proc. Natl. Acad. Sci. USA 105, 1852518530. Marciniak, S.J., Yun, C.Y., Oyadomari, S., Novoa, I., Zhang, Y., Jungreis, R., Nagata, K., Harding, H.P., and Ron, D. (2004). CHOP induces death by promoting protein synthesis and oxidation in the stressed endoplasmic reticulum. Genes Dev. 18, 30663077. Matsumoto, A., Okado, A., Fujii, T., Fujii, J., Egashira, M., Niikawa, N., and Taniguchi, N. (1999). Cloning of the peroxiredoxin gene family in rats and characterization of the fourth member. FEBS Lett. 443, 246250. Mezghrani, A., Fassio, A., Benham, A., Simmen, T., Braakman, I., and Sitia, R. (2001). Manipulation of oxidative protein folding and PDI redox state in mammalian cells. EMBO J. 20, 62886296. Pagani, M., Fabbri, M., Benedetti, C., Fassio, A., Pilati, S., Bulleid, N.J., Cabibbo, A., and Sitia, R. (2000). Endoplasmic reticulum oxidoreductin 1-lbeta (ERO1-Lbeta), a human gene induced in the course of the unfolded protein response. J. Biol. Chem. 275, 2368523692. Pollard, M.G., Travers, K.J., and Weissman, J.S. (1998). Ero1p: a novel and ubiquitous protein with an essential role in oxidative protein folding in the endoplasmic reticulum. Mol. Cell 1, 171182. Saaranen, M.J., Karala, A.R., Lappi, A.K., and Ruddock, L.W. (2010). The role of dehydroascorbate in disulde bond formation. Antioxid. Redox Signal. 12, 1525. Sevier, C.S., and Kaiser, C.A. (2008). Ero1 and redox homeostasis in the endoplasmic reticulum. Biochim. Biophys. Acta 1783, 549556. Sevier, C.S., Qu, H., Heldman, N., Gross, E., Fass, D., and Kaiser, C.A. (2007). Modulation of cellular disulde-bond formation and the ER redox environment by feedback regulation of Ero1. Cell 129, 333344. Tavender, T.J., and Bulleid, N.J. (2010). Peroxiredoxin IV protects cells from oxidative stress by removing H2O2 produced during disulphide formation. J. Cell Sci. 123, 26722679. Tavender, T.J., Sheppard, A.M., and Bulleid, N.J. (2008). Peroxiredoxin IV is an endoplasmic reticulum-localized enzyme forming oligomeric complexes in human cells. Biochem. J. 411, 191199. Tiedge, M., Lortz, S., Drinkgern, J., and Lenzen, S. (1997). Relation between antioxidant enzyme gene expression and antioxidative defense status of insulin-producing cells. Diabetes 46, 17331742. Tien, A.C., Rajan, A., Schulze, K.L., Ryoo, H.D., Acar, M., Steller, H., and Bellen, H.J. (2008). Ero1L, a thiol oxidase, is required for Notch signaling through cysteine bridge formation of the Lin12-Notch repeats in Drosophila melanogaster. J. Cell Biol. 182, 11131125. Tsai, B., and Rapoport, T.A. (2002). Unfolded cholera toxin is transferred to the ER membrane and released from protein disulde isomerase upon oxidation by Ero1. J. Cell Biol. 159, 207216. Tu, B.P., and Weissman, J.S. (2004). Oxidative protein folding in eukaryotes: mechanisms and consequences. J. Cell Biol. 164, 341346.
Molecular Cell
Wang, L., Li, S.J., Sidhu, A., Zhu, L., Liang, Y., Freedman, R.B., and Wang, C.C. (2009). Reconstitution of human Ero1-Lalpha/protein-disulde isomerase oxidative folding pathway in vitro. Position-dependent differences in role between the a and a domains of protein-disulde isomerase. J. Biol. Chem. 284, 199206. Woo, H.A., Yim, S.H., Shin, D.H., Kang, D., Yu, D.Y., and Rhee, S.G. (2010). Inactivation of peroxiredoxin I by phosphorylation allows localized H(2)O(2) accumulation for cell signaling. Cell 140, 517528.
Yang, Y., Song, Y., and Loscalzo, J. (2007). Regulation of the protein disulde proteome by mitochondria in mammalian cells. Proc. Natl. Acad. Sci. USA 104, 1081310817. Zito, E., Chin, K.T., Blais, J., Harding, H.P., and Ron, D. (2010). ERO1-b, a pancreas-specic disulde oxidase, promotes insulin biogenesis and glucose homeostasis. J. Cell Biol. 188, 821832.
Article
The Dynamic Distribution of CARD11 at the Immunological Synapse Is Regulated by the Inhibitory Kinesin GAKIN
Rebecca L. Lamason,1,3 Abraham Kupfer,2,3 and Joel L. Pomerantz1,3,*
of Biological Chemistry of Cell Biology 3Institute for Cell Engineering The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA *Correspondence: jpomera6@jhmi.edu DOI 10.1016/j.molcel.2010.11.007
Molecular Cell
SUMMARY
T cell receptor (TCR) signaling to NF-kB is required for antigen-induced T cell activation. We conducted an expression-cloning screen for modiers of CARD11, a critical adaptor in antigen receptor signaling, and identied the kinesin-3 family member GAKIN as a CARD11 inhibitor. GAKIN negatively regulates TCR signaling to NF-kB, associates with CARD11 in a signal-dependent manner and can compete with the required signaling protein, Bcl10, for association. In addition, GAKIN dynamically localizes to the immunological synapse and regulates the redistribution of CARD11 from the central region of the synapse to a distal region. We propose that CARD11 scaffold function and occupancy at the center of the synapse are negatively regulated by GAKIN to tune the output of antigen-receptor signaling.
INTRODUCTION During the adaptive immune response, antigen receptor signaling in B and T lymphocytes must be nely tuned so that the immune system can recognize foreign pathogens and respond efciently without harming the host. Signaling pathways that emanate from T cell receptor (TCR) and B cell receptor (BCR) complexes activate programs of gene expression that determine whether a lymphocyte proliferates, becomes activated or anergic, or dies as a result of an encounter with a putative antigen (Cancro, 2009; Smith-Garvin et al., 2009). One of the key transcription factors activated by antigen receptor signaling is NF-kB, which is required in both T and B cells for antigeninduced lymphocyte proliferation, survival, and effector functions (Vallabhapurapu and Karin, 2009). Optimal antigen-induced activation of NF-kB occurs in T cells in response to concurrent TCR engagement and costimulatory receptor ligation. Antigen recognition initiates receptor clustering and reorganization of signaling components at the cellcell contact between an antigen presenting cell (APC) and a T cell, forming the immunological synapse (IS) (Lin et al., 2005). The
mature IS is segregated into discrete zones termed supramolecular activation clusters (SMAC). The TCR and PKCq are found in the central SMAC (cSMAC), while adhesion receptors and cytoskeletal proteins, including LFA-1 and Talin, localize to the peripheral SMAC (pSMAC) (Monks et al., 1998). The distal SMAC (dSMAC) is the outermost zone and is dened by the presence of the inhibitory receptor, CD45 (Freiberg et al., 2002). CARD11 (CARMA1) is a multidomain scaffold protein that is required for TCR signaling to NF-kB and controls the recruitment of other signaling proteins to the IS (Blonska and Lin, 2009). Prior to TCR engagement, CARD11 is kept in an inactive state by an inhibitory domain (ID) that prevents the binding of multiple proteins (McCully and Pomerantz, 2008). TCR engagement causes the phosphorylation of the ID, mediated in part by PKCq (Matsumoto et al., 2005; Sommer et al., 2005), which neutralizes its inhibitory effect and allows CARD11 to associate with a group of factors that contribute to the activation of the IkB kinase (IKK) complex, including Bcl10, TAK1, TRAF6, Caspase-8, and IKKg (McCully and Pomerantz, 2008). Activated IKK phosphorylates IkB, leading to its ubiquitination and degradation, and the appearance of active NF-kB in the nucleus. CARD11 activity must be tightly regulated to prevent the hyperactivation of downstream pathways that could lead to dysregulated immune responses or the unwarranted lymphocyte proliferation that is associated with certain NF-kB-dependent types of lymphoma (Jost and Ruland, 2007). Cell lines derived from the ABC subtype of diffuse large B cell lymphoma (DLBCL) have been shown to require CARD11 signaling for their unchecked ability to proliferate in culture (Ngo et al., 2006). In addition, several mutations in CARD11 have been identied in patient samples of DLBCL that endow the protein with a hyperactive ability to activate NF-kB and confer dysregulated growth (Lenz et al., 2008). To identify components of the antigen receptor signaling pathway that regulate CARD11, we conducted an expressioncloning screen for enhancers and suppressors of CARD11 signaling activity. We describe the identication of the motor protein GAKIN, a guanylate kinase-associated kinesin, as a CARD11 inhibitor that attenuates the extent of NF-kB activation following TCR engagement. We show that GAKIN interacts with CARD11 in an inducible manner during signaling, can compete with Bcl10 for association with CARD11, and regulates a previously
Molecular Cell
Regulation of TCR Signaling by the Kinesin GAKIN
unrecognized redistribution of CARD11 from the PKCq-rich center of the IS to a distal region of the IS. RESULTS GAKIN Is an Inhibitor of CARD11 Signaling We adapted the expression-cloning strategy of Pomerantz et al. (2002) to identify cellular factors that could modulate CARD11 signaling activity. Pools of cDNAs from a human spleen expression library were screened for the ability to enhance or suppress the CARD11-mediated activation of the NF-kB-responsive luciferase reporter, Igk2-IFN-LUC, in HEK293T cells. The individual cDNA responsible for each pools activity was then puried by sib selection. Using this approach, we identied the motor protein GAKIN (Hanada et al., 2000) as an inhibitor of CARD11. GAKIN inhibited CARD11-mediated activation of the Igk2-IFNLUC reporter in HEK293T cells in a dose-dependent manner (Figure 1A). The effect of GAKIN on CARD11 was specic; GAKIN did not inhibit NF-kB activation by Bcl10, CARD9, CARD10, or CARD14 (Figures S1AS1D, available online). Furthermore, overexpression of GAKIN in Jurkat T cells inhibited NF-kB activation mediated by superantigen staphylococcal enterotoxin E (SEE) pulsed-Raji B cells (Figure 1B), indicating an ability to inuence CARD11-dependent TCR signaling. In contrast, GAKIN did not inhibit the activation of NF-kB by TNFa, which signals through a CARD11-independent pathway (Figure 1B). To assess whether endogenous GAKIN regulated CARD11dependent signaling, we reduced GAKIN expression in Jurkat T cells by lentiviral-mediated RNA interference (RNAi). We expressed two short hairpin RNAs, shRNA2b and shRNA5a, which stably reduced endogenous protein levels of GAKIN by >90% and $60%, respectively (Figures 1C and 1D). Each hairpin resulted in the enhanced activation of NF-kB achieved by anti-CD3/anti-CD28 crosslinking, as compared to that observed with the control hairpin, siGFP (Figure 1C), consistent with an inhibitory role for GAKIN in TCR signaling. The effect of the shRNA2b hairpin was reversed by transient expression of a GAKIN cDNA that contained silent mutations in the shRNA2b target sequence, conrming the specicity of the RNAi (Figure 1E). GAKIN knockdown resulted in enhanced TCR-induced IKK activity (Figures S1ES1H), consistent with an effect on a CARD11-dependent step in the pathway, and also resulted in the enhanced anti-CD3/anti-CD28-mediated production of IL-2, an endogenous NF-kB-dependent target of TCR signaling (Figure 1F). Importantly, knockdown of GAKIN in primary human CD4+ T cells by >70% also resulted in enhanced IL-2 production following anti-CD3/anti-CD28 stimulation (Figures 1G and 1H). The observed hyperresponsiveness of GAKIN-decient T cells to TCR triggering demonstrates that GAKIN functions as an inhibitor at endogenous levels in both Jurkat T cells and primary human CD4+ T cells during antigen receptor signaling. GAKIN Inhibits CARD11 Subsequent to Signal-Induced Conversion of CARD11 to an Active Scaffold Prior to TCR engagement, the ID represses the ability of CARD11 to associate with multiple signaling proteins (McCully and Pomerantz, 2008; Sommer et al., 2005). TCR signaling results in the phosphorylation of the ID, which neutralizes its inhibitory
effect and activates CARD11 scaffold activity (Matsumoto et al., 2005; Sommer et al., 2005). To test whether GAKIN could inhibit CARD11 signaling after the conversion of CARD11 to an active scaffold, we tested the effect of GAKIN on CARD11DID. This mutant, in which the ID has been deleted, is hyperactive, capable of recruiting signaling proteins in a signal-independent manner, and thus behaves as a CARD11 in which the ID has been neutralized by TCR signaling (McCully and Pomerantz, 2008; Sommer et al., 2005). GAKIN potently inhibited the activation of NF-kB achieved by CARD11DID in Jurkat T cells (Figure 2A), suggesting that GAKIN inhibits the pathway by targeting a step subsequent to ID neutralization. To test whether GAKIN and CARD11 could physically associate, we rst coexpressed GAKIN with either wild-type CARD11 or CARD11DID in HEK293T cells, and assessed association by coimmunoprecipitation. We observed a weak association between myc-tagged CARD11 and HA-tagged GAKIN but an enhanced association of GAKIN-HA with myc-CARD11DID (Figure 2B). These results indicated that the ID of CARD11 interferes with GAKIN binding and predicted that the endogenous proteins would associate only during TCR signaling. Indeed, CARD11 associated with GAKIN in Jurkat T cells in a stimulation-dependent manner; the association of CARD11 with GAKIN was apparent after 5 min of phorbol-12-myristate-13-acetate (PMA)/ionomycin treatment, with increased association after 15 min and a subsequent decrease after 45 min (Figure 2C). Coimmunoprecipitation of CARD11 was not observed with the anti-FLAG control antibody or with the GAKIN-specic antibody in the unstimulated sample. This time course of association of GAKIN with CARD11 was similar to that observed for Bcl10, TRAF6, Caspase-8, and the IKK complex (McCully and Pomerantz, 2008). These results demonstrate that CARD11 and GAKIN can associate at endogenous levels in T cells and that this association is regulated in a signal-inducible manner. Multiple Domains of GAKIN Mediate CARD11 Association To identify which domains of GAKIN could associate with CARD11, we tested ve regions of GAKIN for the ability to coimmunoprecipitate CARD11DID in HEK293T cells (Figures 3A and 3B). The motor domain, MAGUK binding stalk (MBS), and two regions from the tail (TR1 and TR2) were each independently capable of associating with CARD11DID, while the L1-FHA-L2 segment was not. Each region that could associate with CARD11DID was also able to inhibit the ability of CARD11DID to activate NF-kB when overexpressed in Jurkat T cells (Figure 3C) at comparable levels of expression (Figure 3D). GAKIN, like other members of the kinesin family, is regulated by intramolecular interactions that keep the protein in a closed, autoinhibited state prior to the binding of activating ligands or cargo (Yamada et al., 2007). Autoinhibition appears to be mediated by intramolecular interactions between the motor and MBS domains, and between the tail and a region containing the MBS (Yamada et al., 2007). We tested whether the disruption of intramolecular interactions within GAKIN would affect the ability of the protein to associate with CARD11. The deletion of the motor domain resulted in an enhanced ability of GAKIN to associate with CARD11DID (Figure 3E). The deletion of the
Molecular Cell
A
Fold Stimulation
100 90 80 70 60 50 40 30 20 10
+
B
Fold Stimulation
30 25 20 15 10 5 0 Stimulus pEBB-GAKIN
0 pEBB-GAKIN pcCARD11 -
12 -
12 +
36 -
36 +
108 108 200 200 + +
Raji/SEE Raji/SEE +
TNF -
TNF +
C
Fold Stimulation
25 Unstimulated 20 15 10 5 0 siGFP shRNA2b shRNA5a CD3/CD28
D
siGFP
20 10 5
shRNA2b
20 10 5
shRNA5a
20 10 5 ug Protein GAKIN CARD11
E
Fold Stimulation
35 30 25
F
120 100
siGFP shRNA2b
G
70 60
H
NTsi siGAKIN-A
NTsi siGAKIN-A
GAKIN Tubulin
IL-2 (pg/ml)
80 60 40 20 0 ND ND
15 10 5 0
IL-2 (pg/ml)
20
50 40 30 20 10 0 ND ND
2bRGAKIN-HA CD3/CD28
200 200
400 400
800
800
siGFP
shRNA2b
none
CD3/CD28
none
CD3/CD28
Figure 1. GAKIN Inhibits CARD11 and Attenuates TCR Signaling to NF-kB

(A) HEK293T cells were transfected with 20 ng Igk2-IFN-LUC, 6 ng pCSK-LacZ, and the indicated nanogram amounts of pEBB-GAKIN-HA in the absence or presence of 50 ng pc-mycCARD11. (B) Jurkat T cells were transfected with 1500 ng Igk2-IFN-LUC and 200 ng pCSK-LacZ in the absence or presence of 1000 ng of pEBB-GAKIN-HA and stimulated with SEE-pulsed Raji B cells or TNFa for 4.5 hr. (C) Stable Jurkat T cell lines expressing the control (siGFP) or GAKIN-specic (shRNA2b or 5a) hairpins were transfected with 1500 ng Igk2-IFN-LUC and 200 ng pCSK-LacZ and stimulated with anti-CD3/anti-CD28 for 4.5 hr. (D) Western blot analysis of samples analyzed in (C). (E) The shRNA2b stable Jurkat T cell line was transfected with 2000 ng Igk2-IFN-LUC, 200 ng pCSK-LacZ, and the indicated nanogram amounts of a hairpinresistant expression vector for GAKIN (2bRGAKIN-HA), then stimulated with anti-CD3/anti-CD28 for 5.5 hr. The control siGFP Jurkat T cell line was assayed in parallel. (F) ELISA analysis of IL-2 production from the stable RNAi Jurkat lines siGFP and shRNA2b after anti-CD3/anti-CD28 coligation for 24 hr. (G) ELISA analysis of IL-2 production from human CD4+ T cells that were incubated for 72 hr with either an Accell siRNA duplex that targets GAKIN (siGAKIN-A) or a nontarget control siRNA (NTsi) and then stimulated with anti-CD3/antiCD28 coligation for 24 hr. (H) Western blot analysis of samples analyzed in (G). The bars indicate fold stimulation (AC and E) or IL-2 concentration (F and G) and represent the mean values of triplicate samples; the error bars indicate the standard deviation. The following abbreviations are used: a, anti; ND, not detectable.
MBS also resulted in an enhanced interaction, while the deletion of the FHA and the CAP-Gly domains had no signicant effect (Figure 3E). These results suggest that the association of GAKIN with CARD11 is regulated by intramolecular interactions involving the motor and MBS domains.
The CARD and Coiled-Coil Domains of CARD11 Are Required for GAKIN Association To identify which CARD11 domains are required for GAKIN association, a panel of CARD11 deletion constructs made in the context of CARD11DID (McCully and Pomerantz, 2008)
Molecular Cell
A
900 800
WT
GAKIN-HA ID WT
IP: RFLAG 0 ID
myc-CARD11 constructs
RGAKIN 0 5 15 45 min PMA/Iono CARD11 IP IB: CARD11 1.4% Input IB: CARD11 IP IB: GAKIN 1.4% Input IB: GAKIN
+
IP: HA IB: myc 10% Input IB: myc IP: HA IB: HA
Fold Stimulation
700 600 500 400 300 200 100 0
pEBB-GAKIN pcCARD11ID
- + + + - +
10% Input IB: HA
Figure 2. GAKIN Inhibits CARD11 after ID Neutralization and Associates with CARD11 in a Signal-Inducible Manner
(A) Jurkat T cells were cotransfected with 1500 ng Igk2-IFN-LUC, 200 ng pCSK-LacZ, and 100 ng of pc-myc-CARD11DID in the absence and presence of 1000 ng of pEBB-GAKIN-HA. The bars indicate fold stimulation and represent the mean values of triplicate samples; error bars indicate the standard deviation. (B) HEK293T cells were transfected with either 300 ng WT pc-myc-CARD11 or 400 ng pc-myc-CARD11DID in the absence or presence of 500 ng of pc-GAKIN-HA and anti-HA IPs were performed. (C) Jurkat T cells were stimulated with 50 ng/ml PMA and 1 mM ionomycin, lysed, and immunoprecipitated with either anti-FLAG or anti-GAKIN antibodies. The arrow indicates the specic CARD11 band.
was coexpressed with GAKIN in HEK293T cells and assayed for coimmunoprecipitation. The deletion of either the CARD or coiled-coil domains reduced the ability of CARD11DID to associate with GAKIN (Figure 3F), while other domain deletions had no signicant effect. The requirement of the CARD and coiled-coil domains for GAKIN association is consistent with the fact that the CARD11:GAKIN association is regulated by the ID and that TCR signaling is required for maximal association (Figures 2B and 2C). GAKIN Can Compete with Bcl10 for CARD11 After the signal-induced neutralization of the ID, the CARD and coiled-coil domains are required for the recruitment of several signaling proteins to CARD11, including Bcl10, Caspase-8, IKKg, TRAF6, and TAK1 (McCully and Pomerantz, 2008). Since GAKIN association also requires the CARD and coiled-coil domains, we hypothesized that GAKIN might inhibit CARD11 scaffold function by competing with one or more signaling proteins for the association with CARD11. We tested whether the binding of these proteins to CARD11 could be competed by coexpression of DMotor-GAKIN, which displayed an enhanced interaction with CARD11DID as compared to fulllength GAKIN and contained most of the domains sufcient for association with CARD11DID (Figure 3E). Intriguingly, DMotor-GAKIN could inhibit the ability of Bcl10 to coimmunoprecipitate CARD11DID in a dose-dependent manner (Figure 3G). Several other proteins tested, including TAK1, TRAF6, IKKg, and Caspase-8 did not appear to be competed under these conditions (Figure 3H), suggesting that DMotor-GAKIN can specically inuence the association of CARD11 with Bcl10. Consistent with an inhibitory effect of GAKIN on CARD11:Bcl10 association, GAKIN knockdown in Jurkat T cells resulted in a modest but reproducible enhancement of the amount of CARD11 that dynamically associated with Bcl10 during signaling (Figure 3I). Quantitation of the amount of CARD11 associating
with Bcl10 in the GAKIN knockdown cells, as compared to control cells, suggested that GAKIN inhibits formation of at least 26%44% of the Bcl10:CARD11 complex that would otherwise form in the absence of GAKIN after 10 to 15 min of PMA/ionomycin stimulation (Figure 3J). GAKIN Translocates to the IS CARD11 has been observed at the IS (Matsumoto et al., 2005; Tanner et al., 2007; Wang et al., 2004) and regulates the recruitment of the IKK complex to the cSMAC (Hara et al., 2004). Since our results indicated that GAKIN inhibited CARD11 and associated with it upon signaling, we hypothesized that GAKIN would be recruited to the synapse. As GAKIN is a kinesin-3 family member and might exhibit dynamic localization reective of motor activity, we used live-cell imaging with eGFP-tagged GAKIN to assess localization with high temporal resolution. We stably coexpressed GAKIN-eGFP and mCherry-a-tubulin in Jurkat T cells and formed conjugates with SEE-loaded Raji B cells. mCherry-a-tubulin served as a marker of the microtubule organizing complex (MTOC), which undergoes a characteristic signal-dependent relocalization to the area of cell-cell contact in T cells engaged in a productive synapse (Smith-Garvin et al., 2009). Prior to conjugate formation, GAKIN-eGFP was present diffusely in the cytoplasm with some concentrated localization at the MTOC and at the uropod opposite the MTOC (Figure 4A, top). At early time points after rst contact between T cell and APC, GAKIN-eGFP was enriched at the area of cell-cell contact and also colocalized with the MTOC (Figure 4A, t = 60 s; see Movie S1 for an entire representative time course). This early enrichment of GAKIN-eGFP at the IS was observed in 77% of conjugates (n = 26) and occurred within 37.7 7.4 s of initial T cell-APC contact (Figure 4C). At later time points, after MTOC relocalization to the IS, GAKIN-eGFP maintained colocalization with the MTOC at the IS but interestingly, also appeared to concentrate at the distal pole in 69.2% of conjugates (n = 26)
Molecular Cell
A
1
L1-FHA-L2
361 607 832
TR1
1242
TR2
1826
B
Empty
+ myc-CARD11ID
L1-FHA-L2 Motor Flag tagged GAKIN domains
TR1
TR2
Motor
FHA
MBS
CAP-Gly
MBS
180 160 140 120 100 80 60 40 20 0
L1-FHA-L2
C
Fold Stimulation
D
Motor
IP: FLAG IB: myc

MBS
TR1
TR2
10% Input IB: myc IP: FLAG IB: FLAG
CARD11ID -
MBS
TR1
TR2
IB: FLAG
Empty
Motor
L1-FHA-L2
+ CARD11ID-FLAG
CAP-Gly Motor MBS FHA
F
IDCARD
+ GAKIN-HA
IDGUK IDPDZ IDSH3 IDCC IDL1 IDL3 IDL4
G
myc-CARD11 constructs IP: HA IB: myc
FLAG-Bcl10 Motor-GAKIN-HA myc-CARD11ID IP: FLAG IB: myc 10% Input IB: myc IP: FLAG IB: HA 10% Input IB: HA IP: FLAG IB: Flag
- + + + + - - + ++ ++ + + + + -
WT
IP: FLAG IB: HA 10% Input IB: HA IP: FLAG IB: FLAG 10% Input IB: FLAG
ID
WT
GAKIN-HA constructs
10% Input IB: myc IP: HA IB: HA
FLAGCofactor: - Bcl10
FLAGTak1 - + + + + -
FLAGTraf6 - + + + + -
FLAGFLAGIKK Caspase 8 - + + + + - + + + + -
I
0 5
siGFP
10 15 20 0 5
sh2b
10 15 20 min PMA/Iono IP: Bcl10 IB: CARD11 1.5% Input IB: CARD11 IP: Bcl10 IB: Bcl10 1.5% Input IB: Bcl10
Motor-GAKIN-HA - - + + myc-CARD11ID + + + IP: FLAG IB: myc 5.7% Input IB: myc 5.7% Input IB: HA IP: FLAG IB: Flag
J
Relative amount of CARD11 associated with Bcl10
25 20 15 10 5 0 0
siGFP sh2b
10 min
15
20
Figure 3. GAKIN Associates with CARD11 through Multiple Domains, Requires the CARD and Coiled-Coil of CARD11 for Association, and Can Compete with Bcl10
(A) Schematic of GAKIN. Numbers indicate amino acid positions, and bracketed segments show the domain constructs used in (B). (B) HEK293T were transfected with 150 ng of pc-myc-CARD11DID, and the indicated FLAG-tagged GAKIN domain expression constructs and anti-FLAG IPs were performed. The following DNA amounts (ng) were used for GAKIN constructs: Motor, 800; L1-FHA-L2, 300; MBS, 1000; TR1, 400; and TR2, 1850.
Molecular Cell
(Figure 4A, t = 600 s). These results indicate that GAKIN dynamically localizes to the IS during TCR signaling and that GAKIN is appropriately localized to exert an inhibitory effect on CARD11 after signaling has been initiated. The motor domain of GAKIN has been shown to mediate plusend-directed movement of GAKIN along microtubules (Horiguchi et al., 2006; Yamada et al., 2007). To assess the contribution of the motor domain to GAKIN localization, we stably expressed a motor-deleted variant, DMotor-GAKIN-eGFP, in Jurkat T cells with mCherry-a-tubulin. DMotor-GAKIN-eGFP displayed a very different pattern of localization (Figure 4B) as compared to GAKIN-eGFP. Prior to and throughout IS formation, DMotorGAKIN-eGFP appeared to concentrate and colocalize with the MTOC and relocalized to the IS simultaneously with the MTOC (see Movie S2 for an entire representative time course). Unlike wild-type GAKIN-eGFP, DMotor-GAKIN-eGFP did not concentrate at the IS at early time points prior to MTOC relocalization and was recruited to the IS 202.5 40.8 s after initial T cell-APC contact (Figure 4C). After MTOC relocalization, DMotor-GAKIN-eGFP colocalized with the MTOC at the IS and did not appear at the distal pole (93.3% of conjugates, n = 15). Thus, the motor domain was required for the concentration of GAKIN at the IS prior to MTOC relocalization and for the concentration of GAKIN at the distal pole after MTOC relocalization to the IS. The absence of the motor domain appeared to conne DMotor-GAKIN-eGFP to the MTOC even as the MTOC relocalized to the area of APC-T cell contact. Dynamic Localization of CARD11 at the IS To investigate a potential effect of GAKIN on CARD11 localization, we rst characterized CARD11 localization during TCR signaling using live-cell imaging and PKCq as a marker of the cSMAC. We coinfected Jurkat T cells with viruses expressing CARD11-mCherry and PKCq-eGFP and formed conjugates with Raji B cells preloaded with SEE. Prior to APC-T cell contact, both CARD11-mCherry and PKCq-eGFP exhibited a diffuse localization in the cytoplasm (Figure 5A; top row, Pre). Upon conjugation, PKCq-eGFP concentrated in the center of the IS during the rst 60 s after cell-cell contact, as expected (Figure 5A; t = 60 s, 80 s) and remained localized in this pattern throughout
the observed time course. CARD11-mCherry displayed an unexpected dynamic pattern of localization within the contact region. Initially, CARD11-mCherry was enriched in the area of cytoplasm juxtaposed to the APC-T cell contact and partially colocalized with PKCq-eGFP in the center of the contact (Figure 5A; t = 60 s). Later in the time course, however, CARD11-mCherry concentrated at a distal region of the synapse and no longer signicantly colocalized with PKCq-eGFP (Figure 5A; t = 80 s). To characterize this transition in CARD11 localization, we determined the uorescence intensity of CARD11-mCherry and PKCq-eGFP along a line passing through the APC-T cell contact (in the x-y plane from B to E) at 60 and 80 s after initial cell-cell contact for the conjugate displayed in Figure 5A. The peak of PKCq-eGFP uorescence occupied the center of the cell-cell contact at both time points (Figure 5B). In contrast, the CARD11-mCherry uorescence was initially evenly distributed across the cell-cell contact at t = 60 s (Figure 5B, left), but then resolved into two distal peaks of uorescence at t = 80 s that did not overlap with the central peak of PKCq-eGFP uorescence (Figure 5B, right). The en face view (x-z plane) of CARD11mCherry and PKCq-eGFP uorescence (Figure 5C), representing the contact region from the perspective of the APC, also demonstrated the transition in CARD11 localization from a broad pattern throughout the contact at t = 60 s that partially colocalized with PKCq, to a distributed distal pattern at t = 80 s that, for the most part, did not colocalize with PKCq. These results indicate that early in APC-T cell conjugation, CARD11 populates the area of cell-cell contact, but its occupancy at the center of the IS is reduced shortly thereafter by a redistribution to distal areas that limits the duration of its colocalization with PKCq, a key kinase that activates CARD11 scaffold function during TCR signaling. GAKIN Regulates the Redistribution of CARD11 at the IS We investigated whether GAKIN might regulate CARD11 localization by analyzing the dynamic localization of CARD11mCherry in Jurkat T cells in which GAKIN levels were stably reduced by lentiviral-mediated expression of the shRNA2b hairpin. These cells were compared to control cells expressing
(C) Jurkat T cells were transfected with 750 ng Igk2-IFN-LUC, 200 ng pCSK-LacZ, and expression vectors for the indicated FLAG-tagged GAKIN domains in the absence or presence of 100 ng of pc-myc-CARD11DID, as indicated. The following DNA amounts (ng) were used for FLAG-tagged GAKIN constructs: Motor, 600; L1-FHA-L2, 50; MBS, 1000; TR1, 600; and TR2, 1000. The bars indicate fold stimulation and represent the mean values of triplicate samples; the error bars indicate the standard deviation. (D) Western analysis of HEK293T cells transfected with the same DNA concentrations of constructs transfected in (C). Equivalent numbers of transfected cells, as determined by b-gal activity, were evaluated for each construct. (E) HEK293T cells were transfected with 500 ng of pc-CARD11DID-FLAG, and the indicated HA-tagged domain deletion constructs of GAKIN and anti-FLAG IPs were performed. The following DNA amounts (ng) were used for the GAKIN constructs: DMotor, 800; DFHA, 300; DMBS, 1000; and DCAP-Gly, 400. (F) HEK293T cells were cotransfected with 500 ng of pcDNA3-GAKIN-HA expression vector and the following amounts (ng) of each CARD11 variant and anti-HA IPs were performed: WT (150), DID (150), DIDDCARD (1500), DIDDL1 (175), DIDDCC (125), DIDDPDZ (225), DIDDL3 (260), DIDDSH3 (250), DIDDL4 (150), DIDDGUK (150). (G) HEK293T cells were transfected with expression vectors for myc-CARD11DID (10 ng), FLAG-Bcl10 (80200 ng), and a titration of DMotor-GAKIN-HA (900 or 1800 ng), as indicated, and anti-FLAG IPs were performed. (H) HEK293T cells were transfected with expression vectors for myc-CARD11DID (1012 ng), DMotor-GAKIN-HA (1600 ng), and FLAG-Bcl10 (80140 ng), FLAG-TAK1 (200 ng), FLAG-TRAF6 (400 ng), FLAG-IKKg (400 ng), or FLAG-Caspase-8 C360S (400 ng), as indicated, and anti-FLAG IPs were performed. (I) GAKIN-decient (sh2b) or control (siGFP) Jurkat T cells were stimulated with 50 ng/ml PMA and 1 mM ionomycin, lysed, and immunoprecipitated with anti-Bcl10 antibodies. (J) Quantitation of the amount of CARD11 present in the immunoprecipitates in (I), normalized to the amount of Bcl10 in the immunoprecipitates for each sample. The following abbreviations are used: CC, coiled-coil; L1 and L2, linker 1 and 2; FHA, forkhead associated; MBS, MAGUK-binding stalk; TR, tail region; CAP-Gly, cytoskeletal-associated protein-glycine-rich.
Molecular Cell
A
GAKIN-eGFP
mCherry-Tubulin
B
DIC Merge Pre
MotorGAKIN-eGFP
mCherry-Tubulin
DIC
Merge Pre
t=0s
t=0s
t=60s
t=60s
t=600s
t=600s
C
Timepoint of GAKIN recruitment to contact (sec)
600 500 400 300 200 100 0
p < 0.0001
GAKIN-eGFP
MotorGAKIN-eGFP
Figure 4. GAKIN Dynamically Localizes to the IS

(A) Representative live-cell imaging of conjugate formation between a Jurkat T cell stably expressing GAKIN-eGFP and mCherry-atubulin and a Raji B cell in the presence of SEE. Top: The observed uorescence prior to conjugate formation (Pre). Second row: Capture of the rst observed contact between T cell and B cell (t = 0 s); (third and fourth rows) images captured at 60 s and 600 s after rst contact, respectively. The white arrowhead indicates the GAKIN-eGFP signal that colocalizes with the mCherry-a-tubulin signal at the MTOC, indicated by the white arrow. The yellow arrowhead indicates GAKIN-eGFP signal at the distal pole. (B) Representative live-cell imaging of conjugate formation between a Jurkat T cell stably expressing DMotor-GAKIN-eGFP and mCherry-a-tubulin and a Raji B cell in the presence of SEE. Timestamps are shown in the upper right corner of each panel. (C) The time point of initial recruitment to the APC-T cell contact was determined for conjugates expressing either GAKIN-eGFP or DMotor-GAKIN-eGFP. Each dot represents an individual conjugate. Lines indicate the average time of recruitment of 37.7 7.4 s (standard error of the mean [SEM], n = 20) for GAKIN-eGFP and 202.5 40.8 s (SEM, n = 15) for DMotor-GAKIN-eGFP. The scale bars in (A) and (B) represent 10 mm. Pre indicates before cell conjugation. See also Movies S1 and S2.
either no hairpin or the control nontarget hairpin (NTsh). We observed that GAKIN deciency had no apparent effect on the timing of initial CARD11 occupancy at the center of the APC-T cell contact (Figure 6A), suggesting that GAKIN does not regulate the recruitment of CARD11 to the synapse. However, we observed a signicant difference between GAKIN-decient and control cells in the time point at which CARD11 redistributed to the distal region of the IS. While CARD11 redistributed at 62.0 7.6 and 58.0 10.8 s after rst
contact in the NTsh and no-hairpin control cells, respectively, this redistribution occurred at 93.4 9.4 s after rst contact in the sh2b-expressing cells (Figure 6B). For each conjugate, we measured the interval between initial occupancy in the contact center and distal redistribution and found that while duration of CARD11 occupancy at the center was 46.7 10 and 50.0 8.0 s in the no-hairpin and NTsh cells, respectively, this interval increased to 80.1 10.4 s in the sh2b cells (Figure 6C). Figures 6E and 6F show representative examples of
Molecular Cell
A
X
CARD11-mCherry
PKC-eGFP
DIC
Merge Pre
t=60s
E B
t=80s
B
1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
t=60s Relative Intensity
B 0 10 20 30 40 50
1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
t=80s CARD11-mCherry PKC-eGFP
Relative Intensity
B 0 10 20 30 40 50
Distance (pixels)
Distance (pixels)
C
X
CARD11-mCherry
PKC-eGFP
Merge t=60s
t=80s
Z
Figure 5. Two Phases of CARD11 Localization at the IS
(A) Representative live-cell imaging of conjugate formation between a Jurkat T cell stably expressing CARD11-mCherry and PKCq-eGFP and a Raji B cell in the presence of SEE. Top: The observed uorescence before conjugation (Pre). Middle and bottom: The observed uorescence at 60 and 80 s after initial B cell-T cell contact, respectively. The scale bar represents 10 mm. (B) Line intensity graphs representing the average pixel intensity along a 5 pixel wide line drawn from B to E as shown in (A). (C) En face view (x-z plane) of the IS for the conjugates shown in (A) at 60 and 80 s after initial B cell-T cell contact.
control and GAKIN-decient T cells, respectively, at the time points of initial APC-T cell contact (t = 0 s) and distal redistribution (t = 60 s for NTsh; t = 141 s for sh2b). We also reduced GAKIN expression in PKCq-mCherry-expressing cells and, importantly, did not observe any difference in the time point of
PKCq recruitment to the cSMAC (Figure 6D) or in the extended occupancy of PKCq there (data not shown). These data demonstrate that GAKIN regulates the duration of CARD11 localization in the contact center but does not alter the recruitment of another important synapse component, PKCq.
Molecular Cell
A
p = 0.6917
Timepoint of initial CARD11 occupancy at contact center (sec)
B
p = 0.7730
Timepoint of CARD11 concentration at distal region (sec)
E
p = 0.0227 p = 0.0196
CARD11mCherry GFPpLKO.1+ NTsh DIC Merge
60 50 40 30 20 10 0
CARD11-mCh CARD11-mCh + sh2b CARD11-mCh + NTsh
150
t=0s
100
t=60s
50
0
C
p = 0.0339 p = 0.0342
D
p = 1.0000
Timepoint of PKC recruitment to cSMAC (sec)
F
p = 0.7800
CARD11mCherry GFPpLKO.1+ sh2b DIC Merge
150
Duration of CARD11 in contact center (sec)
80 70 60 50 40 30 20 10 0
PKC-mCh PKC-mCh + sh2b PKC-mCh + NTsh
t=0s
100
t=141s
50
0
Figure 6. GAKIN Regulates the Duration of CARD11 Occupancy in the IS Center

The kinetics of CARD11-mCherry and PKCq-mCherry localization were quantitated for Jurkat T cells stably expressing either a GAKIN-specic (sh2b) hairpin, or a nontarget control (NTsh) hairpin in the context of the GFPpLKO.1 lentivirus. A Jurkat T cell line expressing no hairpin was also analyzed as a control. All conjugates were formed with Raji B cells in the presence of SEE. The time point of initial CARD11-mCherry recruitment (A), the time point of CARD11 concentration at the distal synapse region (B), and the duration of CARD11 occupancy in the contact center (C) are shown for all three lines. (D) The time point of PKCq-eGFP recruitment to the cSMAC region is shown for all three lines. (E) Representative live-cell imaging of a conjugated Jurkat T cell coexpressing CARD11-mCherry and the control virus, GFPpLKO.1+NTsh. Top and bottom: The time point of initial T cell-B cell contact and the time point of CARD11 concentration at the distal region, respectively. (F) Representative live-cell imaging of a conjugated T cell coexpressing CARD11-mCherry and the GAKIN-specic hairpin-expressing virus, GFPpLKO.1+sh2b. Panels selected as in (E). The scale bars in (E) and (F) represent 10 mm.
DISCUSSION Antigen-induced signaling through the TCR complex triggers the orchestrated, dynamic movement of signaling machinery to dened regions of the IS. However, the mechanisms that couple cellular localization to signaling activity for most pathway components remain poorly understood. In this report, we identify the kinesin GAKIN as an inhibitory component of the TCR pathway that regulates the signaling output and localization of the multidomain scaffold CARD11 to control the extent of TCR-mediated activation of NF-kB. It has been previously established that after TCR engagement, CARD11 undergoes a signal-dependent transition from an inactive to an active signaling scaffold (Figures 7A and 7B) that requires the PKCq-mediated phosphorylation of the ID. ID phosphorylation causes it to disengage from intramolecular interactions with the CARD and coiled-coil domains and allows the recruitment of signaling proteins to CARD11 to induce IKK activation (Figure 7B). This transition in CARD11 likely takes place in the central portion of the IS where PKCq resides during
signaling and where the IKK complex is recruited in a CARD11dependent manner (Hara et al., 2004). Our data reveal a subsequent step in TCR signaling at which GAKIN acts to limit CARD11 activity (Figure 7C). During this step, GAKIN associates with CARD11 in a manner that can reduce Bcl10 binding and promote the redistribution of CARD11 away from the PKCq-rich central portion of the synapse. The redistribution of CARD11 to distal regions of the APC-T cell contact limits the interval during which CARD11 and PKCq co-occupy the center of the synapse and very likely determines the duration of active CARD11 signaling in that region. Several mechanisms could explain how occupancy of CARD11 in the center of the synapse would determine CARD11 signaling activity. First, PKCq must phosphorylate the ID on at least two serines to convert CARD11 into an active scaffold. It is possible that the dwell time of CARD11 in the central region of the contact determines the fraction of CARD11 molecules that become fully phosphorylated by PKCq, especially if these phosphorylation events are reversible by the action of a phosphatase. Second, active signaling may
Molecular Cell
A
CD28 T cell TCR complex
B
pMHC
APC
C
TRAF6
PKC PKC
IKK
6 TRAF IKK
CC CC
PKC IKK TRAF6 Bcl10
Bcl10 Bcl10 TRAF6 IKK

ID
P
Bcl10
Closed, inactive CARD11
ID
P
Bcl10 TRAF6 IKK

ID
P
Bcl10 TRAF6 IKK

ID
P
ID
P
ID
P
Bcl10 TRAF6 IKK

ID
P
Closed, inactive GAKIN
ID
P
Open, active CARD11
Figure 7. Schematic of GAKIN Action during TCR Signaling

(A) In unstimulated T cells, CARD11 is kept in a closed, inactive conformation in the cytoplasm through intramolecular interactions between the ID and the CARD and coiled-coil domains. PKCq, GAKIN, and the depicted pathway components are cytoplasmic and unbound to CARD11. (B) Upon TCR engagement, PKCq clusters at the membrane and phosphorylates the ID of CARD11, leading to the conversion of CARD11 to an active conformation that can associate with several signaling proteins including TRAF6, IKKg, and Bcl10. (C) Subsequent to the activation of CARD11, GAKIN and CARD11 interact in a manner that can compete Bcl10 off of a fraction of CARD11 molecules and that results in the redistribution of CARD11 away from the synapse center to a distal region. GAKIN is depicted here as directly transporting CARD11 along microtubules in a plus-end-directed fashion, but other mechanisms by which GAKIN can inuence CARD11 localization are possible. Since intramolecular interactions within GAKIN can regulate the association with CARD11, GAKIN is depicted as undergoing a conformational change that is inuenced by TCR signaling.
be restricted to the center of the contact by the CARD11-independent PKCq-mediated activation of other signaling components that must work in concert with CARD11 in the pathway. For example, Shambharkar et al. (2007) demonstrated that during TCR signaling, IKK kinase activation requires both the CARD11-dependent ubiquitination of IKKg and the CARD11independent phosphorylation of IKKa/b by TAK1, which was shown to be activated and associate with PKCq during TCR signaling in a CARD11-independent manner. The dwell time of CARD11 in the center of the synapse may thus determine the extent to which CARD11-dependent signaling events can couple to CARD11-independent steps. It is possible that GAKIN transports CARD11 as cargo away from the center of the APC-T cell contact (Figure 7C), although our data do not address this directly. IS formation results in MTOC polarization to the contact (Smith-Garvin et al., 2009), which orients microtubules such that they emanate away from the center of the IS with their plus ends distal to the MTOC. GAKIN is capable of plus-end-directed motility on microtubules (Horiguchi et al., 2006; Yamada et al., 2007), and the orientation of microtubules at the IS could allow directional, GAKIN-mediated transport of CARD11 to a distal region. It is important to note that in T cells depleted of GAKIN by RNAi, CARD11 still redistributes to the distal region of the contact, albeit with altered kinetics. This delayed redistribution might be due to the action of the residual GAKIN in these cells or to a distinct mechanism that functions with GAKIN in CARD11 redistribution. Our data suggest several important roles for the motor domain of GAKIN during TCR signaling. The motor is required for the dynamic localization of GAKIN at the IS at early time points (Figure 4), where CARD11-dependent signaling to the IKK complex likely occurs (Hara et al., 2004). During signaling, the motor appears to override determinants in the other domains of GAKIN that appear to specify localization to the MTOC, because deletion of the motor results in connement of GAKIN to the MTOC. In addition, the motor participates in an autoinhibitory
intramolecular interaction with the MBS that can regulate motor activity (Yamada et al., 2007) and that appears to regulate the GAKIN:CARD11 association since deletion of either of these domains enhances the ability of GAKIN to associate with CARD11DID (Figure 3E). Since GAKIN associates with CARD11 in a signal-dependent manner, it is likely that TCR signaling inuences GAKIN conformation. It will be important to investigate this possibility, and the possibility that TCR signaling regulates a transition from an inactive to an active motor. Interestingly, in overexpression studies that bypass the constraints imposed by signal-regulated cellular localization and protein conformation (Figures 3B and 3C), the motor itself is sufcient for GAKIN-mediated CARD11DID association and inhibition but not necessary since other GAKIN fragments, including the MBS, TR1, and TR2, are also capable of associating with and inhibiting CARD11DID. We note that these studies do not address whether these domains bind CARD11 directly or indirectly since the association assays were done in the presence of other cellular proteins. Further studies will be required to understand how the multiple domains of GAKIN integrate function prior to and during antigen receptor signaling. Bcl10 is an obligate component of the TCR pathway (Ruland et al., 2001), and its recruitment to CARD11 is an essential step in TCR signaling (Blonska and Lin, 2009). The dynamic association of GAKIN with CARD11 appears to reduce the fraction of CARD11 molecules that associate with Bcl10 during signaling (Figures 3I and 3J). Bcl10 is also targeted by other mechanisms implicated in the termination of TCR signaling to NF-kB, including its signal-induced phosphorylation and degradation (Lobry et al., 2007; Scharschmidt et al., 2004; Zeng et al., 2007). The ability of GAKIN to compete with Bcl10 for CARD11 binding likely works in concert with these degradative mechanisms to accomplish the negative feedback that limits Bcl10 activity in TCR signaling. The action of GAKIN on CARD11 activity represents a previously unrecognized mechanism used by lymphocytes to limit
C ID
C ID
Molecular Cell
the extent of antigen receptor signaling to NF-kB. GAKIN is a signaling inhibitor that is poised to act after signaling has been initiated to tune signaling output. GAKIN does not act to terminate signaling, or to prevent signaling in the absence of receptor engagement. This is in contrast to other key inhibitory players in the pathway, including the E3 ligase Cbl-b, the deubiquitinases A20 and CYLD, and the kinase CK1a. Cbl-b appears to act upstream of CARD11 in the pathway, by inhibiting the extent of Akt and PKCq activation following receptor engagement (Qiao et al., 2008). Downstream of CARD11, A20 appears to dampen pathway output after signaling has begun by removing the K63-linked ubiquitin chains on MALT1 that contribute to the association between MALT1 and the IKK complex (Duwel et al., 2009). CYLD has been shown to prevent spontaneous NF-kB activation in T cells in the absence of receptor engagement by inhibiting the ubiquitination and autoactivation of TAK1 (Reiley et al., 2007). CK1a plays both positive and negative roles in this pathway and appears to attenuate signaling by phosphorylating residues in the CARD11 ID, although the mechanistic outcome of these phosphorylation events remains unclear ` (Bidere et al., 2009). Multiple checkpoints appear to be independently required to tune the output of antigen receptor signaling, to prevent the unwarranted expansion and transformation of lymphocytes observed in lymphoma, and to ensure an appropriate adaptive immune response that does not harm the host. The role of GAKIN in TCR signaling may offer opportunities for the directed manipulation of lymphocyte activation and proliferation that might prove useful in the treatment of autoimmunity, immunodeciencies, or lymphocytic cancers.
EXPERIMENTAL PROCEDURES Expression-Cloning Screen Screening was done with pools of 100 cDNAs/pool isolated from a human spleen expression library (Origene Technologies Inc). The primary screen was carried out as previously described (Pomerantz et al., 2002) except that 50 ng pcCARD11 was cotransfected along with 20 ng Igk2-IFN-LUC, 6 ng of pCSK-LacZ, and 300 ng pool cDNA into HEK293T cells by the calcium phosphate method. Pools altering CARD11 activity at least 3-fold were further puried to a single cDNA by sib selection and sequenced for identication. Secondary screens to conrm NF-kB specicity were conducted in a similar fashion except that 50 ng pCMV6-C/EBPd was used instead of pcCARD11 to activate 20 ng of a mutant NF-kB reporter (MUT-IFN-LUC). Samples were harvested 40 hr after transfection in 100 ml Promega Lysis buffer and lysates were used to measure Luciferase (Promega) and b-gal activity (Roche) as previously described. Results were calculated as the fold luciferase change, normalized to b-gal activity, compared to the empty vector sample. Transient Transfections of Jurkat T Cells Jurkat T cells were plated in 6-well plates at 2.5 3 105 cells/ml and 2 ml/well. In some assays, Fugene-6 (Roche) or LT-1 (Mirus) was used with 3 mg total DNA per manufacturers instructions. Other assays used lipofectamine LTX (Invitrogen) and 2.5 mg total DNA. Transfections included 200 ng pCSK-LacZ and 750-2000ng Igk2-IFN-LUC. In each experiment, each sample was supplemented with empty parental expression vector to keep the total amount of expression vector constant. Approximately 40 hr after transfection, cells were stimulated in 1 ml media alone or with 75 ng/ml TNFa (Sigma T6674), 1 mg/ml each of mouse anti-human CD3 (BD PharMingen 555329), mouse anti-human CD28 (BD PharMingen 555725), anti-mouse IgG1 (BD PharMingen 02231D), or
Raji B cells pulsed with superantigen. To stimulate with the Raji B cells, we rst resuspended the B cells in fresh media containing 0.4 ng/ml SEE (Toxin Technology) and incubated for 30 min at 37 C, 5% CO2, prior to being added to Jurkats at a B:T ratio of 8:1. Stimulations were carried out for 46 hr before harvesting samples in 150 ml Promega lysis buffer and assayed for luciferase and b-gal activity. Results are shown as the fold luciferase change, normalized to b-gal activity, compared to the empty vector sample. In Figures 1B, 1E, 2A, and 3C, the average fold stimulation under each condition was normalized to that observed with the corresponding unstimulated sample and the unstimulated sample value was set to 1. 293T Immunoprecipitations Approximately 40 hr after transfection of HEK293T cells via calcium phosphate, cells were harvested in 500 ml immunoprecipitation (IP) lysis buffer, incubated 10 min on ice, and debris cleared by centrifugation at 18,300 3 g at 4 C. Lysates were precleared by incubating with 7 ml bed volume protein G sepharose (Amersham) twice for 30 min at 4 C with rotation. An aliquot was removed for input analysis and the remaining lysate was incubated with 1 mg IP antibody for 1.52 hr at 4 C with rotation. Protein G sepharose (7 ml bed volume) preblocked with 1% human insulin was added to samples and incubated 1 hr at 4 C with rotation. The beads were then washed with rotation 4 3 5 min at 4 C with IP lysis buffer before being boiled in the presence of SDS loading buffer. IPs were analyzed by western blot with anti-myc (Santa Cruz SC-40), anti-HA (Santa Cruz SC-7392), and M2 anti-FLAG (Eastman Kodak IB13026). For the co-IPs in Figure 3G and 3H, the procedure was carried out as described above except that the lysates were incubated with 2 mg rabbit anti-FLAG and 14 ml bed volume protein G sepharose. Jurkat T Cell Immunoprecipitations To assess the GAKIN:CARD11 association, 108 Jurkat T cells/sample were resuspended in media alone or with 50 ng/ml PMA (Sigma) and 1 mM ionomycin (Sigma) and stimulated for the indicated times. Cells were then plunged in an ice water bath for 10 min then spun for 10 min at 423 3 g. Cell pellets were resuspended in 1.5 ml IP lysis buffer and processed as above except that the lysates were incubated with rotation overnight at 4 C with 2 mg rabbit anti-FLAG (Sigma F7425) or 10 ml rabbit anti-GAKIN serum (Covance Research Products Inc.) after the preclear step. CARD11 was detected with goat anti-CARD11 (Imgenex IMG-3653) and GAKIN was detected with rabbit anti-GAKIN serum. Live-Cell Imaging To observe conjugates, 1 3 106 c/ml of Raji B cells were incubated with 2 mg/ml SEE for 1 hr then washed twice before being resuspended in fresh media at a concentration of 0.51 3 106 cells/ml. Jurkat T cells stably expressing the indicated proteins were resuspended in fresh media to 0.250.5 3 106 cells/ml. Approximately 100 ml of T cells were added to the center of a glass bottom culture dish (14 mm Microwell, No. 0 Coverglass, MatTek, #P35G-014-C), and approximately 100 ml SEE-coated Raji B cells were added dropwise directly to the T cells. The 4D imaging was done by collecting 15 3 1.5 mm planes every 2030 s for a total of 60 time points. All images were taken on a motorized Marianas Imaging system (Intelligent-Imaging Innovations) in a humidied, 7.5% CO2 chamber maintained at 37 C and were collected and processed with Slidebook 4.2 or 5.0 (Intelligent-Imaging Innovations). An intensity-based segment mask that indicated pixels within the top 40% of the intensity range was used to determine localization of GAKIN-GFP, DMotorGAKIN-GFP, and CARD11-mCherry. The segment mask for PKCq-eGFP and PKCq-mCherry was set to include the top 30% of the intensity range. Line intensity analysis was done with ImageJ; all statistical analysis was carried out in Prism and p values were calculated from a two-tailed, unpaired Students t test. Additional experimental procedures are detailed in the Supplemental Information. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, one gure, and two movies and can be found with this article online at doi:10.1016/j.molcel.2010.11.007.
Molecular Cell
ACKNOWLEDGMENTS We thank M. Meffert, J. Yang, R. Tsien, N. Hacohen, C. Drake, and J. Sisk for reagents; S. Lew, E. Eyler, R. Jattani, A. Kim, R. McCully, d. Mackie, and M. Meffert for critical reading of the manuscript; and S. Desiderio, H. Ike, H. Kupfer, M. Spitaler, M. Meffert, T. Bruno, and D. Getnet for advice and discussions. This work was supported by National Institutes of Health Grants RO1AI078980 and PO1AI072677, American Cancer Society Grant RSG-06-17201-LIB, and funds from the Johns Hopkins University Institute for Cell Engineering. R.L. was supported by Ruth L. Kirschstein National Research Service Award F31AG031689. J.L.P is a recipient of a Kimmel Scholar Award from the Sidney Kimmel Foundation for Cancer Research and a Rita Allen Foundation Scholar. Received: April 17, 2010 Revised: July 23, 2010 Accepted: September 10, 2010 Published: December 9, 2010 REFERENCES ` Bidere, N., Ngo, V.N., Lee, J., Collins, C., Zheng, L., Wan, F., Davis, R.E., Lenz, G., Anderson, D.E., Arnoult, D., et al. (2009). Casein kinase 1alpha governs antigen-receptor-induced NF-kappaB activation and human lymphoma cell survival. Nature 458, 9296. Blonska, M., and Lin, X. (2009). CARMA1-mediated NF-kappaB and JNK activation in lymphocytes. Immunol. Rev. 228, 199211. Cancro, M.P. (2009). Signalling crosstalk in B cells: managing worth and need. Nat. Rev. Immunol. 9, 657661. Duwel, M., Welteke, V., Oeckinghaus, A., Baens, M., Kloo, B., Ferch, U., Darnay, B.G., Ruland, J., Marynen, P., and Krappmann, D. (2009). A20 negatively regulates T cell receptor signaling to NF-kappaB by cleaving Malt1 ubiquitin chains. J. Immunol. 182, 77187728. Freiberg, B.A., Kupfer, H., Maslanik, W., Delli, J., Kappler, J., Zaller, D.M., and Kupfer, A. (2002). Staging and resetting T cell activation in SMACs. Nat. Immunol. 3, 911917. Hanada, T., Lin, L., Tibaldi, E.V., Reinherz, E.L., and Chishti, A.H. (2000). GAKIN, a novel kinesin-like protein associates with the human homologue of the Drosophila discs large tumor suppressor in T lymphocytes. J. Biol. Chem. 275, 2877428784. Hara, H., Bakal, C., Wada, T., Bouchard, D., Rottapel, R., Saito, T., and Penninger, J.M. (2004). The molecular adapter Carma1 controls entry of IkappaB kinase into the central immune synapse. J. Exp. Med. 200, 1167 1177. Horiguchi, K., Hanada, T., Fukui, Y., and Chishti, A.H. (2006). Transport of PIP3 by GAKIN, a kinesin-3 family protein, regulates neuronal cell polarity. J. Cell Biol. 174, 425436. Jost, P.J., and Ruland, J. (2007). Aberrant NF-kappaB signaling in lymphoma: mechanisms, consequences, and therapeutic implications. Blood 109, 2700 2707. Lenz, G., Davis, R.E., Ngo, V.N., Lam, L., George, T.C., Wright, G.W., Dave, S.S., Zhao, H., Xu, W., Rosenwald, A., et al. (2008). Oncogenic CARD11 mutations in human diffuse large B cell lymphoma. Science 319, 16761679. Lin, J., Miller, M.J., and Shaw, A.S. (2005). The c-SMAC: sorting it all out (or in). J. Cell Biol. 170, 177182. Lobry, C., Lopez, T., Israel, A., and Weil, R. (2007). Negative feedback loop in T cell activation through IkappaB kinase-induced phosphorylation and degradation of Bcl10. Proc. Natl. Acad. Sci. USA 104, 908913.
Matsumoto, R., Wang, D., Blonska, M., Li, H., Kobayashi, M., Pappu, B., Chen, Y., Wang, D., and Lin, X. (2005). Phosphorylation of CARMA1 plays a critical role in T Cell receptor-mediated NF-kappaB activation. Immunity 23, 575585. McCully, R.R., and Pomerantz, J.L. (2008). The protein kinase C-responsive inhibitory domain of CARD11 functions in NF-kappaB activation to regulate the association of multiple signaling cofactors that differentially depend on Bcl10 and MALT1 for association. Mol. Cell. Biol. 28, 56685686. Monks, C.R., Freiberg, B.A., Kupfer, H., Sciaky, N., and Kupfer, A. (1998). Three-dimensional segregation of supramolecular activation clusters in T cells. Nature 395, 8286. Ngo, V.N., Davis, R.E., Lamy, L., Yu, X., Zhao, H., Lenz, G., Lam, L.T., Dave, S., Yang, L., Powell, J., and Staudt, L.M. (2006). A loss-of-function RNA interference screen for molecular targets in cancer. Nature 441, 106110. Pomerantz, J.L., Denny, E.M., and Baltimore, D. (2002). CARD11 mediates factor-specic activation of NF-kappaB by the T cell receptor complex. EMBO J. 21, 51845194. Qiao, G., Li, Z., Molinero, L., Alegre, M.L., Ying, H., Sun, Z., Penninger, J.M., and Zhang, J. (2008). T-cell receptor-induced NF-kappaB activation is negatively regulated by E3 ubiquitin ligase Cbl-b. Mol. Cell. Biol. 28, 24702480. Reiley, W.W., Jin, W., Lee, A.J., Wright, A., Wu, X., Tewalt, E.F., Leonard, T.O., Norbury, C.C., Fitzpatrick, L., Zhang, M., and Sun, S.-C. (2007). Deubiquitinating enzyme CYLD negatively regulates the ubiquitin-dependent kinase Tak1 and prevents abnormal T cell responses. J. Exp. Med. 204, 14751485. Ruland, J., Duncan, G.S., Elia, A., del Barco Barrantes, I., Nguyen, L., Plyte, S., Millar, D.G., Bouchard, D., Wakeham, A., Ohashi, P.S., and Mak, T.W. (2001). Bcl10 is a positive regulator of antigen receptor-induced activation of NF-kappaB and neural tube closure. Cell 104, 3342. Scharschmidt, E., Wegener, E., Heissmeyer, V., Rao, A., and Krappmann, D. (2004). Degradation of Bcl10 induced by T-cell activation negatively regulates NF-kappa B signaling. Mol. Cell. Biol. 24, 38603873. Shambharkar, P.B., Blonska, M., Pappu, B.P., Li, H., You, Y., Sakurai, H., Darnay, B.G., Hara, H., Penninger, J., and Lin, X. (2007). Phosphorylation and ubiquitination of the IkappaB kinase complex by two distinct signaling pathways. EMBO J. 26, 17941805. Smith-Garvin, J.E., Koretzky, G.A., and Jordan, M.S. (2009). T cell activation. Annu. Rev. Immunol. 27, 591619. Sommer, K., Guo, B., Pomerantz, J.L., Bandaranayake, A.D., Moreno-Garca, M.E., Ovechkina, Y.L., and Rawlings, D.J. (2005). Phosphorylation of the CARMA1 linker controls NF-kappaB activation. Immunity 23, 561574. Tanner, M.J., Hanel, W., Gaffen, S.L., and Lin, X. (2007). CARMA1 coiled-coil domain is involved in the oligomerization and subcellular localization of CARMA1 and is required for T cell receptor-induced NF-kappaB activation. J. Biol. Chem. 282, 1714117147. Vallabhapurapu, S., and Karin, M. (2009). Regulation and function of NF-kappaB transcription factors in the immune system. Annu. Rev. Immunol. 27, 693733. Wang, D., Matsumoto, R., You, Y., Che, T., Lin, X.Y., Gaffen, S.L., and Lin, X. (2004). CD3/CD28 costimulation-induced NF-kappaB activation is mediated by recruitment of protein kinase C-theta, Bcl10, and IkappaB kinase beta to the immunological synapse through CARMA1. Mol. Cell. Biol. 24, 164171. Yamada, K.H., Hanada, T., and Chishti, A.H. (2007). The effector domain of human Dlg tumor suppressor acts as a switch that relieves autoinhibition of kinesin-3 motor GAKIN/KIF13B. Biochemistry 46, 1003910045. Zeng, H., Di, L., Fu, G., Chen, Y., Gao, X., Xu, L., Lin, X., and Wen, R. (2007). Phosphorylation of Bcl10 negatively regulates T-cell receptor-mediated NF-kappaB activation. Mol. Cell. Biol. 27, 52355245.
Article
Systematic In Vivo RNAi Analysis Identies IAPs as NEDD8-E3 Ligases
Meike Broemer,1,* Tencho Tenev,1 Kristoffer T.G. Rigbolt,3 Sophie Hempel,2 Blagoy Blagoev,3 John Silke,4 Mark Ditzel,1,2,5 and Pascal Meier1,5,*
1The Breakthrough Toby Robins Breast Cancer Research Centre, Institute of Cancer Research, Mary-Jean Mitchell Green Building, Chester Beatty Laboratories, Fulham Road, London SW3 6JB, UK 2Institute of Genetics and Molecular Medicine, Edinburgh Cancer Research Centre, Crewe Road South, Edinburgh EH4 2XR, UK 3Department Biochemistry & Molecular Biology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark 4Department of Biochemistry, Level 4 RL Reid Building, La Trobe University, Victoria 3086, Australia 5These authors contributed equally to this work *Correspondence: meike.broemer@icr.ac.uk (M.B.), pmeier@icr.ac.uk (P.M.) DOI 10.1016/j.molcel.2010.11.011
Molecular Cell
SUMMARY
The intimate relationship between mediators of the ubiquitin (Ub)-signaling system and human diseases has sparked profound interest in how Ub inuences cell death and survival. While the consequence of Ub attachment is intensely studied, little is known with regards to the effects of other Ub-like proteins (UBLs), and deconjugating enzymes that remove the Ub or UBL adduct. Systematic in vivo RNAi analysis identied three NEDD8-specic isopeptidases that, when knocked down, suppress apoptosis. Consistent with the notion that attachment of NEDD8 prevents cell death, genetic ablation of deneddylase 1 (DEN1) suppresses apoptosis. Unexpectedly, we nd that Drosophila and human inhibitor of apoptosis (IAP) proteins can function as E3 ligases of the NEDD8 conjugation pathway, targeting effector caspases for neddylation and inactivation. Finally, we demonstrate that DEN1 reverses this effect by removing the NEDD8 modication. Altogether, our ndings indicate that IAPs not only modulate cellular processes via ubiquitylation but also through attachment of NEDD8, thereby extending the complexity of IAP-mediated signaling.
INTRODUCTION A major mechanism for regulating protein function involves the covalent attachment of Ubiquitin (Ub) and Ubiquitin-like proteins (UBLs). Bound Ub and UBLs thereby inuence protein function either directly via conformational changes or indirectly through mediating interactions with other proteins (Dikic et al., 2009). Ub/UBL modications regulate a multitude of cellular processes, including cell survival and apoptosis (Haglund and Dikic, 2005). Over the recent years, several Ubiquitin (Ub)-E3 ligases have emerged as key regulators of the apoptosis program (Broemer and Meier, 2009). Protein levels and activity of many pro- and
antiapoptotic molecules are controlled by E3-mediated conjugation of Ub. However, Ub-mediated regulation of cell survival is not just a mere consequence of Ub-directed proteasomal degradation. Nondegradative ubiquitylation events play also important roles (Ditzel et al., 2008; Jin et al., 2009; Schile et al., 2008), and, in combination with deubiquitylation, may allow a further level of apoptotic regulation. Ubiquitylation of caspases by inhibitor of apoptosis (IAP) proteins has been shown to play an important part in regulating caspase activity (Choi et al., 2009; Ditzel et al., 2008; Jin et al., 2009; Schile et al., 2008; Shapiro et al., 2008). In Drosophila, DIAP1 acts as RING Ub-E3 ligase that promotes the ubiquitylation of the initiator caspase DRONC (Chai et al., 2003; Herman-Bachinsky et al., 2007; Wilson et al., 2002) and the effector caspases drICE and DCP-1 (Ditzel et al., 2008). Although DIAP1 reportedly inhibits caspases in vitro (Kaiser et al., 1998; Yan et al., 2004), physical interaction alone seems does not seem to be sufcient to maintain cell viability in vivo. Mutations of DIAP1s RING-nger domain, which abrogate its E3 activity but not caspase binding, cause a loss-of-function phenotype (Lisi et al., 2000; Wilson et al., 2002). In addition, the protein levels of DIAP1 are also regulated in an Ub-dependent manner (Ditzel and Meier, 2002). Specialized IAP-antagonist proteins, such as Reaper (Rpr) and Head involution defective (Hid) (Grether et al., 1995; White et al., 1994), adjust DIAP1 protein levels by inducing its autoubiquitylation and degradation (Holley et al., 2002; Ryoo et al., 2002; Yoo et al., 2002), leading to activation of apoptotic caspases and cell death. While the consequence of Ub conjugation is intensely studied (Hoeller and Dikic, 2009), much less is known with regards to the effects of UBL proteins such as SUMO, NEDD8, ATG12, ATG8, URM1, ISG15, and FAT10. To date, 17 different human UBLs have been identied (Schulman and Harper, 2009). Besides adopting a similar overall structure (Vijay-Kumar et al., 1987), UBLs share surprisingly limited amino acid sequence similarity with Ub. The most studied among the UBLs are SUMO (small Ub-related modier) and NEDD8 (neural precursor cell expressed developmentally downregulated protein 8) (Kerscher et al., 2006). While SUMO modications are most frequently associated with transcriptional suppression (Garcia-Dominguez and Reyes, 2009), conjugation of NEDD8 is best known for its
Molecular Cell
IAPs as NEDD8 E3 Ligases
role in regulating cullin-type E3 ligases (Merlet et al., 2009). Although cullins are the best-studied substrates, they are not the only class of proteins modied by NEDD8 (Xirodimas, 2008). Recent studies have uncovered p53 (Xirodimas et al., 2004) and EGFR (Oved et al., 2006) as targets of the neddylation pathway. Ribosomal proteins reportedly are also modied by NEDD8, which seems to protect them from destabilization (Xirodimas et al., 2008). pVHL (von Hippel-Lindau tumor suppressor protein), BCA (breast cancer-associated protein) and APP intracellular domain (AICD) are further proteins that are modied by neddylation (Gao et al., 2006; Lee et al., 2008; Stickle et al., 2004). In these cases, neddylation seems to affect their interaction with binding partners. Ub/UBLs are attached to target proteins either as a single moiety or as polymeric chains of variable length (Komander, 2009). Ub/UBLs are transferred to lysine (K) residues of substrates in a stepwise process that involves activating enzymes (E1), conjugating enzymes (E2), and protein ligases (E3) (Hochstrasser, 2009). E3s thereby determine which substrate is modied, as they bind to both E2 and substrate, bringing the E2 in position for Ub/UBL transfer. While the E3 provides substrate specicity, the E2 determines which type of modication is formed. Importantly, each UBL employs its own specic E1/E2 cascade (Hochstrasser, 2009). While UBLs tend to use UBL-specic E1-E2 cascades, it seems that the last step, selection of the E3, is more exible (Brzovic and Klevit, 2006). E2s bind to a wide range of different E3s, predominantly via an interaction with an E3s HECT (homologous to E6-associated protein C terminus), RING (really interesting new gene), or U-box domain (Deshaies and Joazeiro, 2009; Hatakeyama and Nakayama, 2003; Rotin and Kumar, 2009). The covalent attachment of Ub/UBLs to target proteins is a reversible process. Specialized deconjugating enzymes (referred to as deubiquitylating enzymes [DUBs]) remove the Ub/UBL message (Reyes-Turcu et al., 2009). The human genome encodes approximately 100 predicted DUBs capable of deconjugating Ub, NEDD8, and SUMO from target proteins (Mukhopadhyay and Dasso, 2007; Nijman et al., 2005). DUBs are involved in the maturation of Ub/UBLs from precursor peptides, as well as in constitutive or regulated removal of the Ub/UBL modication from target proteins. Together with E3 ligases, DUBs are crucial regulators of many cellular processes, determining stability of proteins and inuencing signaling events (Reyes-Turcu et al., 2009). Here, we have identied three NEDD8-specic isopeptidases that, when knocked down, suppress Rpr- and Hid-induced cell death. This suggests that the NEDD8 modication prevents apoptosis signaling. Genetic validation, using null mutant animals, conrmed the involvement of deneddylase 1 (DEN1) in regulating apoptosis. Consistent with the notion that conjugation of NEDD8 protects from apoptosis, we found that endogenous drICE is neddylated in healthy cells and that the NEDD8 modication reduces the proteolytic activity of drICE. Surprisingly, conjugation of NEDD8 to effector caspases was mediated by IAPs, in both Drosophila and mammals. Therefore, IAPs not only function as E3 ligases of the Ub conjugation system, but also take part in the NEDD8-specic cascade of protein modication. Consequently, IAPs can inuence cellular
processes by conjugating Ub as well as NEDD8 to target substrates. RESULTS A Systematic In Vivo RNAi Screen Identies NEDD8-Specic Proteases Involved in Apoptosis To identify DUBs and deconjugating enzymes for UBL that regulate programmed cell death, we conducted a systematic in vivo RNA interference (RNAi)-modier screen in which we selectively knocked down individual DUBs in the developing eye of ies ectopically expressing the IAP antagonists Reaper (Rpr) and Hid (M.B. and P.M., unpublished data). This identied three NEDD8-specic proteases that, when knocked down, suppressed Rpr and Hid killing (Figure 1A): DEN1, CSN5, and CG1503, a predicted deneddylase. The observation that knockdown of these three deneddylases suppress Rpr and Hid killing suggests that conjugation of NEDD8 suppresses the activity of proapoptotic proteins, or, alternatively, enhances the antiapoptotic potential of cell death inhibitors. Of particular interest was Deneddylase 1 (DEN1), which reportedly removes NEDD8 from non-Cullin proteins in vivo (Chan et al., 2008). To corroborate the involvement of DEN1 in the regulation of apoptosis, we assessed genetically whether reduction of DEN1 suppressed cell death induced by IAP antagonists. Consistently, a null allele of DEN1 (DEN1EX9) (Chan et al., 2008) suppressed both Rpr- and Hid-induced cell death, establishing DEN1 as a potential proapoptotic regulator of cell death pathways (Figure 1B). The Effector Caspase drICE Is Neddylated In Vivo To decipher how NEDD8 and DEN1 might regulate apoptosis, we rst examined whether proteins of the cell death machinery are conjugated with NEDD8. To test whether the Drosophila initiator caspase DRONC and the effector caspases drICE and DCP1 are subject to NEDD8 conjugation, we immunoprecipitated caspases from cellular extracts and analyzed the eluates for the presence of NEDD8 conjugates using a NEDD8-specic antibody. In agreement with endogenous drICE being targeted for neddylation, we detected several distinct bands corresponding to potentially mononeddylated (46 kDa) and polyneddylated drICE, or drICE that is conjugated with NEDD8 and Ub chains (Figure 2A). The detected bands were specic for neddylated drICE since immunoprecipitation using preimmune serum of the anti-drICE antibody, or cell extracts in which drICE was knocked down via RNAi, reduced the amount of neddylated drICE detected. To calculate the proportion of drICE that is modied, we quantied the different areas of the anti-drICE immunoblot (Figure 2A) using LI-COR Odyssey technology (Figure S1B available online). This indicated that approximately 23% of total drICE is modied. The effector caspase DCP-1 is also subject to neddylation; however, we found no evidence for robust neddylation of DRONC (Figure S2; M.B. and P.M., unpublished data). Next, we knocked down the E1/E2s for Ub and NEDD8, respectively. RNAi-mediated knockdown of the NEDD8-selective E1/E2 APPBP1/Ubc12(CG7375) almost completely abrogated neddylation of drICE (Figure 2B). Intriguingly, knockdown of the Ub-E1/E2 Uba1/UbcD1 also affected the NEDD8-specic
Molecular Cell
A
Control CSN5 DEN1 CG1503
Figure 1. Knockdown or Mutation of Deneddylating Enzymes that Modify Rpr- and Hid-Induced Cell Death
(A) dsRNA of the indicated deneddylases was expressed with the eye-specic driver GMR-GAL4, and its effect on GMR-rpr and GMR-hid-mediated eye phenotypes was analyzed by light microscopy of whole mounts. Unmodied phenotypes of Rpr and Hid are shown in A0 and A00 . Genotypes in B0 D00 are GMR-GAL4,GMR-rpr or GMR-hid/ UAS-dsRNA. (B) Genetic validation with DEN1EX9 mutant ies. DEN1EX9 mutant ies (B and D) exhibit a reduced Rpr (compare A with B) and Hid (compare C with D) eye phenotype. The graphs depict the eye size of (GMR-GAL4,GMR-rpr,DEN1EX9/DEN1EX9, left graph) and (GMR-GAL4,GMR-hid/DEN1EX9, right graph) ies. Shown is the average eye size of at least ten animals standard error (SE).
Rpr
Hid
B
Rpr control den1EX9/den1EX9 control Hid den1EX9/+
proteins were afnity puried under denaturing conditions using nickel columns to avoid isolation of protein complexes. The presence of modied drICE was assessed by immunoblot analysis of the eluate. Under these conditions, DIAP1 readily ubiquitylated drICE, as previously reported (Ditzel et al., 2008) (Figure 3A). Interestingly, DIAP1 also neddylated 80000 30000 Rpr Hid drICE but was unable to promote the 60000 25000 conjugation of SUMO to drICE. The notion that DIAP1 functions as NEDD840000 20000 E3, but not SUMO-E3, is further corrobo20000 15000 rated by the observation that DIAP1 physically interacted with the NEDD8-E2 0 10000 den1EX9/+ control den1EX9/den1EX9 control conjugating enzyme UbcD12/CG7375 (Figure 3B) but not the SUMO-E2 lesswright/UbcD9 (data not shown). Under smearing pattern of drICE, albeit less prominently than was the same conditions, DIAP1 also interacted with the Ub-E2 achieved by knockdown of APPBP1/UbcD12, suggesting that UbcD1, as previously reported (data not shown) (Ryoo et al., endogenous drICE is both neddylated as well as ubiquity- 2002). latedeither by mixed chains or via separate chains on To determine whether endogenous DIAP1 is responsible for individual K residues. This is also consistent with a correspond- neddylating drICE, we depleted DIAP1 protein levels using ultraing change in the overall smearing pattern of total modied drICE violet (UV) treatment. Exposure to UV causes rapid proteasomal (Figure 2B). Of note, the weaker reduction in the neddylation degradation of DIAP1 (Ditzel et al., 2003). While under nontreated pattern following knockdown of Uba1/UbcD1 might be due to conditions drICE was readily neddylated, UV-mediated deplea less efcient knockdown of Uba1/UbcD1 (Figure S1A). Taken tion of DIAP1 abrogated the appearance of neddylated forms together, these data indicate that endogenous drICE is neddy- of drICE (Figure 3C). Since UV induces apoptosis under these lated in living cells and that the conjugation of NEDD8 to drICE conditions, these results also indicate that the neddylation status is catalyzed by the sequential action of the bona de NEDD8 of drICE drastically changes during cell death. This may be achieved not only by DIAP1 depletion (removal of the NEDD8activation and conjugation cascade. E3) but also through UV-mediated activation of the deneddylase DIAP1 Can Function as NEDD8-E3 Ligase DEN1. Consistent with this view, a recent publication reports that To identify the E3 ligase that promotes drICE neddylation, we genotoxic stress induces NEDP1 (Watson et al., 2010), the rst focused on DIAP1. To examine whether DIAP1 targets drICE mammalian homolog of DEN1. Therefore, drICE deneddylation for neddylation, we coexpressed drICE and DIAP1 [DIAP1(21438)] upon cell death insult might be the result of coordinated DIAP1 (Ditzel et al., 2008) in the presence of His-tagged NEDD8. As depletion and DEN1 activation. Taken together, these data indicontrols, we also included His-Ub and His-SUMO. Conjugated cate that endogenous DIAP1 functions as an NEDD8-E3 ligase
pixels
pixels
Molecular Cell
pre-immune
B
-drICE - + neddylated drICE IP: Uba1/UbcD1 dsRNA: APPBP1/UbcD12 dsRNA: (kDa) 150 100 75 Blot: -N8 50 37 IP: -drICE
pre-immune
Figure 2. Endogenous drICE Is Conjugated with the Ubiquitin-like Modier NEDD8

-drICE + + neddylated drICE
IP: drICE dsRNA:

(kDa)
150 100 75 Blot: -N8 50 37 25 20 100 75 50 Blot: -drICE 37
20
drICE
Blot: -drICE 50 37 25 1 2 3 4
20
Lysate
Blot: -drICE 37 Blot: -Actin 37
drICE Actin 1 2 3
for drICE in vivo and that drICE becomes deneddylated during cell death. To compare the extent of drICE ubiquitylation versus neddylation, we made use of His-tagged Ub and NEDD8 (Figure 3D). Note that anti-Ub- and anti-NEDD8-specic antibodies cannot be used to determine the proportion of drICE that is modied with either adduct because these antibodies harbor different afnities for their respective antigens, which precludes a direct comparison. Purication of total neddylated and/or ubiquitylated drICE under denaturing conditions indicated that drICE was modied by both Ub and NEDD8, although longer and more prominent chains were formed in the presence of His-Ub (Figure 3D). The same could be observed for DIAP1 automodication (Figure 3D). The ability of DIAP1 to promote neddylation of drICE was dependent on a functional RING nger, as the E3 mutants DIAP1(21438/C406Y) and DIAP1(21438/CD6) (Ditzel et al., 2008) failed to neddylate drICE (Figure 4A). Previous work has indicated that DIAP1 also requires (1) N-terminal cleavage and (2) binding to UBR domain-bearing N-end-rule E3s (UBR-E3s) to maximally ubiquitylate drICE (Ditzel et al., 2008). Therefore, we established the determinants for DIAP1-mediated neddylation of drICE. As shown in Figure 4A, N-terminal cleavage was necessary for DIAP1 to act as a NEDD8-E3 for drICE. DIAP1(D20A), which carries a point mutation in the caspase cleavage site and hence resides in its full-length form, failed to neddylate drICE efciently. This is most likely due to the observation that cleaved DIAP1(21438) interacts with drICE far better than full-length DIAP1, which binds caspases only weakly (Ditzel et al., 2008). While N-terminal cleavage was required, binding to UBR-E3s was dispensable for neddylation of drICE. M-DIAP1(21438), in
which N(21) was replaced by methionine (M), an amino acid that does not allow UBR binding (Tasaki et al., 2005), was as efcient in neddylating drICE as wildtype DIAP1(21438). Therefore, DIAP1mediated neddylation of drICE seems to occur independently of the N-end-rule machinery. We also tested whether neddylation of drICE required direct binding of drICE to DIAP1. Accordingly, ALG-drICE, which exposes an IAP-binding motif (IBM) at its neo-N terminus and binds to DIAP1 (Tenev et al., 2005), was neddylated by DIAP1(21438). In contrast, LG-drICE, which lacks a functional IBM and fails to associate with DIAP1 (Tenev et al., 2005), also failed to be neddylated by DIAP1(21438) (Figure 4B). This indicates that a physical interaction between drICE and DIAP1 is required for DIAP1-mediated neddylation of drICE. DIAP1 not only promoted neddylation of drICE, but also stimulated the conjugation of NEDD8 to itself (autoneddylation). This is evident as the RING mutants DIAP1(21438/C406Y) and DIAP1(21438/CD6) failed to become neddylated (Figure 4A), indicating that DIAP1s own RING nger is required for the conjugation of NEDD8 to itself. Moreover, cleavage at position D20 of DIAP1, which is known to enhance its E3 ligase activity (Ditzel et al., 2008), seemed to boost DIAP1s ability to autoneddylate. Accordingly, cleaved DIAP1 (N-DIAP121438) was signicantly more effective in autoneddylation than the noncleavable DIAP1(D20A) mutant (Figure 4A). Given that neddylation of cullin-type E3 complexes activates their ligase activity (Duda et al., 2008; Furukawa et al., 2000; Morimoto et al., 2000; Podust et al., 2000; Read et al., 2000; Saha and Deshaies, 2008; Wu et al., 2000), we next examined whether neddylation of DIAP1 similarly stimulates its E3 ligase activity. To test this unambiguously, we devised an in vitro assay (Figure 4C) in which we rst incubated DIAP1 with a conjugation mixture in the presence or absence of NEDD8 (rst step, top panel, lanes A and B) or Ub (lane C). Unmodied, neddylated, and ubiquitylated forms of DIAP1 were puried and
modified drICE
150 100 75
drICE
(A) drICE was immunoprecipitated from S2 cell lysates with preimmune serum (lane 1) or a-drICE antibodies (lanes 2 and 3). Immunoprecipitates were analyzed by immunoblotting with the indicated antibodies. Bands correspond to potentially mononeddylated (46 kDa) and polyneddylated forms of drICE, or drICE that is conjugated with NEDD8 and Ub chains. RNAi-mediated knockdown of drICE also reduced the amount of neddylated drICE detected (lane 3). (B) dsRNA-mediated knockdown of the NEDD8E1 dAPPBP1 and NEDD8-E2 UbcD12 (lane 4) abrogates drICE neddylation. Similarly, knockdown of Ub-E1 Uba1 and Ub-E2 UbcD1 (lane 3) reduced the NEDD8 signal, indicating that drICE is modied with both NEDD8 and Ub. Endogenous drICE was immunoprecipitated and analyzed as in (A). The knockdown was veried by RT-PCR (Figure S1A). See also Figures S1 and S2.
IP: -drICE
modified drICE
Molecular Cell
V5-tagged : His-tagged : DIAP121-438 : (kDa) 100 75 Ub
B
drICE SUMO N8
Figure 3. DIAP1 Functions as NEDD8-E3 Ligase for drICE

V5-tagged : UbcD12 control
His-purification
modified drICE
GST-tagged: 45 Blot: -V5 30 GST-purification 97 66 45 Blot: -GST 30 Lysate Blot: -V5 30

(kDa)
50 Blot: -V5 37
DIAP1
UbcD12
GST-DIAP1
25 20 Blot: -V5 37 Blot: -HA 1 2 3 4 5 6
Lysate
drICE
GST UbcD12 1 2
DHFR-HA-Ub (DIAP1 reference protein)
C
IP: UV:
(kDa)
D
-drICE neddylated drICE His-N8: His-Ub:
(kDa)
drICE-V5 DIAP1
- - +
- + - + - + + modified drICE
250 150 IP: -drICE Blot: -N8 100 75 50 37 Blot: -drICE
His-purification
Blot: -V5
150 102 76 52 38 31
drICE Lysate
24 17 150 102 Blot: -DIAP1 76 52 Blot: -V5
37 Lysate Blot: -drICE
drICE processed drICE DIAP1 1 2 3
38 Blot: -DIAP1 52 1 2 3 4
drICE DIAP1
(A) drICE is ubiquitylated and neddylated in a DIAP1-dependent manner. However, DIAP1 does not SUMOylate drICE. drICE-V5 was expressed in S2 cells in the presence or absence of DIAP1(21438), and His-Ub, -SUMO, or -NEDD8, respectively. Note that DIAP1 was expressed as HA-DHFR/Ub-DIAP1 fusion in which the reference protein DHFR/Ub is cotranslationally cleaved off (Varshavsky, 2000). Expression of the reference protein HA-DHFR/Ub indirectly indicates the expression level of the protein of interest (DIAP1). Purication of His-tagged proteins was performed under denaturing conditions, and the presence of modied drICE was detected with a-V5 antibody. An asterisk marks unmodied drICE, which is detected in all lanes as a result of nonspecic drICE:matrix interaction. (B) DIAP1 binds to the NEDD8-E2 UbcD12. UbcD12-V5 was expressed together with GST (lane 1) or GST-DIAP1 (lane 2) in S2 cells. GSTand DIAP1-bound protein complexes were puried and analyzed by immunoblotting. (C) Endogenous drICE is neddylated in a DIAP1dependent manner. S2 cells were either left untreated or exposed to UV. The presence of neddylated forms of drICE was determined by immunoblotting. Note that UV treatment causes depletion of DIAP1 protein levels (bottom panel, compare lanes 1 and 2 with lane 3) (Ditzel et al., 2008). The top panel is a reblot with an a-NEDD8 antibody of the experiment shown in Ditzel et al. (2008), while the input controls (bottom panels) are the same as in Ditzel et al. (2008). (D) Comparison between drICE neddylation and ubiquitylation. drICE was coexpressed with the indicated constructs and analyzed as in A.
pre-immune
Blot: -DIAP1
subsequently assayed for their ability to promote ubiquitylation and neddylation of drICE, respectively (second step). Incubation of DIAP1 with drICE in the presence of E1, E2, and NEDD8 leads to the appearance of several slow migrating species of drICE (bottom panel, lanes 4 and 5), demonstrating that DIAP1 is able to act as NEDD8-E3 ligase for drICE in vitro. DIAP1 appeared to preferentially modify drICE with mono- or polymono-NEDD8 conjugates. In contrast, DIAP1 promoted the conjugation of poly-Ub chain to drICE. Comparison between unmodied and neddylated DIAP1 indicates that neddylated DIAP1 is no more active than unmodied DIAP1 in ubiquitylating drICE (4C). Likewise, ubiquitylation of DIAP1 did not signicantly affect the ability of DIAP1 to neddylate drICE under these conditions. This demonstrates that neddylation or ubiquitylation of DIAP1 itself does not modulate its E3 ligase activity. DEN1 but Not CSN5 Removes NEDD8 from drICE Since the deneddylases DEN1 and CSN5 were both identied as suppressors of IAP antagonist-induced eye phenotypes
(Figure 1), we examined whether DEN1 and CSN5 removed NEDD8 conjugates from drICE. Expression of DEN1 efciently removed NEDD8 conjugates from drICE (Figure 5A). In contrast, CSN5 failed to cleave NEDD8 from drICE, despite being expressed to similar levels. DEN1-mediated deconjugation of NEDD8 required a functional protease domain since the catalytically inactive mutant DEN1(C165A) failed to remove NEDD8 from drICE (Figure 5B). Of note, DEN1-mediated deconjugation of NEDD8 is unlikely to be due to an effect of DEN1 on DIAP1s E3 ligase activity, since DIAP1-mediated ubiquitylation of drICE still occurred under these conditions (data not shown). This indicates that DEN1, but not CSN5, acts as deneddylase for drICE. Conjugation of NEDD8 Inhibits Active drICE To test whether neddylation directly affects the proteolytic activity of drICE, we devised an in vitro neddylation assay followed by a cleavage reaction using caspase substrates (Figure 6). Several slow-migrating drICE species were detected when recombinant, active drICE was incubated with E1, E2, DIAP1, and increasing amounts of NEDD8 (Figure 6A), which
modified DIAP1
Molecular Cell
His-tagged : V5-tagged :
N8 drICE N21-438/C406Y N21-438/C6
B
His-tagged :
21-438
C
N8 + ALG + LG N-DIAP121-438: -
D20A
21-438
step 1:
(DIAP1 auto-Ub/N8ylation)
DIAP1 :
(kDa)
V5-tagged drICE :
(kDa)
ALG
neddylated drICE
(kDa)
N8 Ub
neddylated drICE
His-purification
His-purification
100 75 50 37 Blot: -V5 25 20
Blot: -V5
100 75 50 37 25
250 130 95 72 Blot: -DIAP1 55 36
DIAP1
neddylated DIAP1
His-purification
Blot: -DIAP1
25 20
DIAP1: step2: (kDa)
un
37 25 Blot: -V5 20 15
drICE
* *
Lysate
36 drICE p10 28 17 N8-drICE drICE p20 1 2 3 4 5
Blot: 50 -DIAP1 1 2 3 4 5 6
DIAP1
Figure 4. DIAP1 Neddylates drICE in a RING- and Binding-Dependent Manner

(A) DIAP1 requires N-terminal cleavage and a functional RING nger for its ability to conjugate NEDD8 to drICE and itself. drICE was coexpressed with a noncleavable DIAP1 mutant (D20A, lane 2), active DIAP(21438) (lane 3) starting with N(21) (UBR-E3-binding procient) (Ditzel et al., 2008), RING mutant DIAP(21438/C406Y) (lane 4), DIAP(21438/CD6) (lane 5), or cleaved DIAP1(21438) (lane 6) that carries an M instead of N and therefore is defective in its ability to recruit UBR-E3s. Purication of His-tagged neddylated proteins was performed under denaturing conditions, and the presence of neddylated drICE and DIAP1 was detected by immunoblotting. (B) Physical interaction between DIAP1 and drICE is required for drICE neddylation. ALG-drICE but not LG-drICE, which lacks an IBM and is impaired in DIAP1 binding, is neddylated by active DIAP(21438). The experiment was carried out as in (A). Note that DIAP1 was expressed as HA-DHFR/Ub-DIAP1 fusion. (A and B) An asterisk marks unmodied drICE which is detected in all lanes due to nonspecic drICE:matrix interaction. (C) Sequential in vitro neddylation and ubiquitylation assay. DIAP1 autoneddylation is not required for DIAP1s ability to function as an Ub-E3. Step 1 (top): autoneddylation (lane B) or autoubiquitylation (lane C) of DIAP1. Step 2 (bottom): nonmodied DIAP1 (from A) or neddylated DIAP1 (from B) was used to ubiquitylate drICE (lanes 2 and 3). Vice versa, nonmodied DIAP1 (from A) or ubiquitylated DIAP1 (from C) was used to neddylate drICE (lanes 4 and 5). A purication step was performed after step 1 to remove any free NEDD8 and Ub from the reaction mix. Asterisks indicate nonspecic background bands from the recombinant drICE preparation.
were not detected in the presence of the E3 mutant DIAP1(F437A), indicating that they are neddylated products. To assess the effect of neddylation on the activity of drICE, we incubated in vitro neddylated and/or ubiquitylated forms of drICE with recombinant PARP1 protein or DEVD-AMC as caspase substrates (Figures 6A6C and Figure S3). The proteolytic activity of modied drICE was compared with the one of nonmodied drICE that was incubated with the DIAP1 RING nger mutant DIAP1(F437A). drICE protein that had been modied with increasing amounts of NEDD8 was strongly impaired in its ability to cleave PARP1 (Figure 6A). NEDD8-mediated inhibition of drICE required the conjugation of NEDD8 to drICE and was not the result of free NEDD8 poisoning the caspase cleavage assay nonspecically. This is evident because increasing amounts of
free NEDD8, in the presence of the RING mutant DIAP1(F437A), did not reduce the catalytic activity of drICE (Figure 6A). Quantication of PARP1 cleavage showed that while unmodied drICE cleaved 36% of PARP1, neddylated drICE was signicantly less active in processing PARP1 (Figure 6B). Only 6% of PARP1 was cleaved in the mixture that carries the highest levels of neddylated drICE (Figure 6B, compare columns 1 and 3; Figure S3). Ubiquitylation of drICE similarly impaired its catalytic activity (Figure 6B; Figure S3) (Ditzel et al., 2008), but not as efciently as neddylation (Figure 6B, compare lanes 2 and 4 and lanes 3 and 6). The observation that DIAP1 suppresses the catalytic activity of drICE in a Ub- and NEDD8-dependent manner is further supported by the nding that coexpression of DIAP1 and drICE prevents appearance of the p10 subunit of
modified drICE
250 130 95 72 55 Blot: -drICE
-m un od -m . ( ne od A) dd . (A un yl ) -m ate ub od d ( iq . (A B) ui ) ty la te d (C
Ub Ub N8 N8
100 75 50 37
Blot: -V5 37 36 Blot: -HA 28 1 2 3
Lysate
drICE DHFR-HA-Ub (DIAP1 reference protein)
28 A B C
step 2:
)
(drICE Ub/neddylation)
Molecular Cell
His-N8 drICE-V5 DIAP1 HA-CSN5: HA-DEN1:

(kDa)
His-N8 drICE-V5 DIAP1 C165A
Figure 5. DEN1 Reverses DIAP1-Mediated Neddylation of drICE

(A) DEN1 acts as deneddylase for drICE. drICE-V5 was coexpressed with DIAP1(21-438) and HisNEDD8 in the absence (lane 1) or presence of HA-DEN1 (lane 2) or HA-CSN5 (lane 3). Coexpression of DEN1 but not CSN5 abolishes neddylation of drICE. (B) The catalytic mutant of DEN1 [DEN1(C165A)] fails to deneddylate. An asterisk marks unmodied drICE, which is detected in all lanes as a result of nonspecic drICE:matrix interaction.
- + + neddylated drICE
HA-DEN1:
(kDa)
WT
His-purification
Blot: -V5
His-purification
150 100 75 50 37
150 100 75 Blot: -V5 50 37
25 150 100 75 Blot: -HA Lysate 50 HA-CSN5 37 HA-DEN1 25 20 Blot: -DIAP1 Blot: -V5 37 37 Blot: -HA 25 1 2 3 HA-DEN1 HA-DHFR-Ub (DIAP1 reference protein)
Lysate
50 1 2 3
DIAP1
drICE in vivo (Figure 4A, lane 3). It is important to note that cleavage of drICE in this system requires an input from active drICE itself (Ditzel et al., 2008). Accordingly, z-VAD-fmk also suppresses the appearance of the small subunit of drICE (Ditzel et al., 2008). Therefore, after DIAP1-mediated modication of drICE, drICEs catalytic potential is inhibited, and hence no further drICE processing is possible. Concentration-dependent cleavage assays with DEVD-AMC further corroborate the notion that nonneddylated and neddylated drICE signicantly differ in their processivity (Figure 6C). Since DIAP1 and DIAP1(F437A) bind to drICE equally well, but DIAP1(F437A) fails to neddylate drICE, this result indicates that neddylation reduces the catalytic potential of drICE. Nonlinear regression using the Michaelis-Menten equation showed a reduction in Vmax (62195 6845 RFU/min for nonmodied drICE to 40669 4536 RFU/min for neddylated drICE) but not KM. This suggests that conjugation of NEDD8 functions as noncompetitive inhibitor that suppresses caspase activity via a conformational change of the caspase, reducing its catalytic processivity. Importantly, the mechanism through which NEDD8 suppresses caspase activity is distinct from the one of Ub, which affects both Vmax and KM of drICE and, therefore, acts as mixed inhibitor (Ditzel et al., 2008). Taken together, these results suggest that drICE is regulated by the conjugation of both NEDD8 and Ub, which together cooperate to lower the activity of drICE. Note that immunoprecipitated drICE is modied by both NEDD8 and Ub chains (Figure 2B). Structural prediction indicates that drICE carries nine surface exposed lysine (K) residues (Ditzel et al., 2008) (Figure S4). Using mass spectrometric analysis of neddylated drICE, we identied that K142 (Figure S4) was neddylated. As a result of limited coverage, no data were obtained on the neddylation status of other surface exposed K residues. Although K142 can function
as acceptor K for NEDD8, it seems not to be the only K being neddylated, or else other Ks are used when K142 is mutated. This is evident because mutation of K142 did not abrogate DIAP1-mediated neddylation of drICE (Figure S4, compare lanes 4 and 5), and mutation of all nine surface exposed K residues in the p20 [drICE(9K > R)] was required to fully abrogate drICE neddylation (Figure S4). This is reminiscent to the conjugation of Ub, where also all nine surface-exposed residues of drICE needed to be mutated to abrogate its modication by Ub. Therefore, drICE mutants that retain a single K at various positions are ubiquitylated and neddylated as efciently as WT drICE (data not shown) (Ditzel et al., 2008). This indicates that any of the nine K residues can serve as acceptor site for Ub and NEDD8, and that conjugation of these UBLs to drICE does not occur on a specic K residue at a xed position, a phenomena that is frequently observed (also seen in p53, IkBa, c-jun, and cyclinB1), and contributes to the tremendous plasticity and exibility of the Ub/UBL-system (Kirkpatrick et al., 2006; Rodriguez et al., 1996, 2000; Scherer et al., 1995; Treier et al., 1994; Xirodimas et al., 2004). Structural studies show that K residues positioned less than 50 A away from the active site cysteine of the E2 can serve as acceptor sites (Duda et al., 2008). In this regard, it is interesting to note that the majority of the surface exposed K residues of drICE are positioned in a ring-like orientation (Figure S4D). Since they all can serve as acceptor Ks, they all must be in close proximity to the E2s, making it difcult to generate Ub- and NEDD8-selective drICE mutants. IAP-Mediated Neddylation Is Evolutionarily Conserved Next, we wished to establish whether mammalian IAPs can also function as NEDD8-E3 ligases. To this end, we studied mammalian XIAP, which has previously been shown to function as Ub-E3 ligase for itself and caspases (Morizane et al., 2005; Schile et al., 2008; Suzuki et al., 2001). XIAP readily promoted autoneddylation and neddylation of caspase-7 (Figure 7A). Similar to DIAP1, XIAP-mediated neddylation was RING dependent since XIAP(DC8), which lacks the last eight amino acids essential for its E3 activity (Silke et al., 2005), failed to promote conjugation of NEDD8 to itself and caspase-7. Further, Smac
neddylated drICE
Molecular Cell
1. Step:
in vitro neddylation assay

F>A 0 1 2 4 modified drICE
Figure 6. Neddylation of drICE Suppresses Its Catalytic Potential through Noncompetitive Inhibition
(A) In vitro neddylation of recombinant drICE suppresses its catalytic ability to cleave PARP1. A schematic representation of the assay procedure is shown in the leftmost panels. In vitro neddylation assay of active drICE with DIAP1 or DIAP1(F437A) and the indicated amount of NEDD8. Modication of drICE (upper panel) and automodication of DIAP1 (middle panel) was assessed by immunoblot analysis. Bottom panel, 2 step depicting in vitro PARP1 cleavage assay: the neddylation reactions from step 1 were incubated with recombinant PARP1. Shown is immunoblot analysis with a-PARP1 antibodies. (B) In vitro ubiquitylation/neddylation was performed as in (A). The percentage of PARP1 cleavage was determined by immunoblotting and quantication with LI-COR-Odyssey. The mean SE of three independent experiments is shown. Values and immunoblot analysis of a representative experiment are shown in Figure S3. (C) Neddylation of drICE affects kinetic parameters of substrate cleavage. In vitro neddylation assay of active drICE with DIAP1 and DIAP1(F437A), respectively. Reactions were subsequently incubated with increasing concentrations of the caspase substrate DEVD-AMC. Curves were tted with nonlinear regression with the Michaelis-Menten equation. Shown is the mean of three independent experiments SE. See also Figure S3.
1. Step:
DIAP1: WT N8 (g): 0 1 2 4
(kDa)
in vitro Ubylation and N8ylation assay
DIAP1 drICE
130 95 72 55 36 28 Blot: -drICE 17 250 130 95 Blot: -DIAP1 72 55
modified drICE drICE p20 (unmodified)
modified DIAP1
36 28 1 2 3 4 5 6 7 8
2. Step:
2. Step:
PARP cleavage assay
PARP1 Blot: -PARP1 1 2 3 4 5 6 7 8 full length PARP1 cleaved PARP1
drICE
PARP1
cleaved PARP1
B
50
% PARP processed
1. step: DIAP1+drICE+Ubl 2.step: +PARP

Odyssey protein quantification (LI-COR)
DEVDase assay non-modified drICE N8ylated drICE
40 30 20 10 0
l
We also examined whether cIAP1 is capable of promoting autoneddylation and the conjugation of NEDD8 to RIP1, a bona de target of cIAP1-mediated ubiquitylation (Broemer and Meier, 2009). cIAP1 readily targeted itself and RIP1 for neddylation (Figure 7D). While the physiological role of XIAP and cIAP1-mediated neddylation remains to be explored, our data clearly indicate that IAPs not only function as E3 ligases of the Ub conjugation system, but also are capable of promoting the conjugation of NEDD8 to target substrates, thereby extending the complexity of IAP-mediated signaling. DISCUSSION
N8
N8
Ub
Ub g N8 + 2
Ub
no
Ub
mimetic treatment, which targets XIAP as well as other IAPs (Gaither et al., 2007), resulted in a signicant reduction of caspase-7 neddylation (Figure 7B). Since XIAP strongly binds and inactivates caspase-7 and is the only physiological caspase inhibitor in mammals (Eckelman et al., 2006), it is likely that neddylation of caspase-7 is mediated by XIAP. However, since Smac mimetics also target other IAPs, such as cIAP1 and cIAP2, it is formally possible that these IAPs also contribute to neddylation of caspase-7. Consistent with the notion that neddylation suppresses caspase activity, treatment with the compound MLN4924, an inhibitor of the NEDD8 conjugation pathway (Soucy et al., 2009), resulted in increased caspase-3 and -7 dependent DEVDase activity (Figure 7C). Taken together, these results suggest that also mammalian IAPs act as NEDD8-E3 ligases for effector caspases.
Using an unbiased modier screen, we identied three NEDD8-specic isopeptidases that, when knocked down, suppress Rpr- and Hid-induced cell death. Genetic validation conrmed the involvement of DEN1, and therefore NEDD8, in regulating apoptosis. Importantly, NEDD8 maturation still occurs in DEN1 mutant animals (Chan et al., 2008), indicating that DEN1 is not the only NEDD8-processing enzyme and that in the absence of DEN1, the conjugation of NEDD8 to target substrate occurs normally. This is in contrast to APPBP1 mutants that lack the NEDD8-E1 enzyme and in which no neddylation is seen (Chan et al., 2008). Consistent with the notion that conjugation of NEDD8 protects from apoptosis, we found that endogenous drICE is neddylated in healthy cells and that the conjugation of NEDD8 directly reduces the proteolytic activity of drICE. Neddylated drICE is less active toward caspase substrates than nonmodied drICE. NEDD8-mediated suppression of drICE occurs via a mechanism
Molecular Cell
A
His-tagged : HA-XIAP :
(kDa)
B
N8 N-Casp-7 His-tagged : Smac mimetic: neddylated Casp-7
(kDa)
N-Casp-7
Figure 7. Evolutionarily Conservation of IAP-Mediated Conjugation of NEDD8

(A) The mammalian IAP XIAP targets caspase-7 and itself for neddylation in a RING nger-dependent manner. DN-caspase-7 was coexpressed with His-NEDD8 in 293T cells in the absence (lane 1) or presence of HA-XIAP(wt) (lane 2) or the RING mutant HA-XIAP(DC8) (lane 3). The experiment was performed as described in Figure 3A. (B) Inhibition of endogenous IAPs impairs caspase-7 neddylation. DN-caspase-7 was coexpressed with His-NEDD8 in 293T cells and incubated for 6 hr in the presence or absence of 1 mM Smac-mimetic (Gaither et al., 2007) before lysis. Neddylation of caspase-7 was determined by immunodetection with caspase-7 antibodies after His-purication under denaturing conditions (as above). Note: neddylation of caspase-7 is, in contrast to (A), also visible in the absence of XIAP overexpression as a result of stronger exposure of the immunoblot. (C) Inhibiting the NEDD8 conjugation pathway by the NEDD8-E1 inhibitor MLN4924 (Soucy et al., 2009) increases caspase activity. 293T cells were transfected with active caspase-3 or caspase-7 or control vector and treated with MLN4924 for 16 hr. Caspase activity in cell lysates was determined by cleavage of DEVD-AMC. The experiment was performed in triplicates and mean standard deviation is shown. (D) cIAP1 can target RIP1 and itself for neddylation. RIP1 and cIAP1 were expressed together with His-tagged N8 in 293T cells and the experiment performed as in (A).
N8
wt C8
+
neddylated Casp-7 Casp-7 XIAP
100 75 50 Blot: Casp-7 His-purification 37
100 75 His-purification 50 Blot: -Casp-7 37 25 20
25 150 100 Blot: HA 75 50 37 neddylated XIAP
37 Lysate Blot: -Casp-7 Blot: -XIAP 25 50 1 2 3
end. Casp-7 Blot: Casp-7 25 Lysate Blot: HA 50 1 2 3 N-Casp-7 XIAP His-tagged : cIAP1 :
(kDa)
N8 RIP1 neddylated RIP1 neddylated cIAP1
WT
Blot: RIP1 His-purification
250 150 100 75 50 37
C
6x105
DEVDase assay
RFU
4x105
150 100 Blot: cIAP1 75 50 37 Blot: RIP1 75 75 Blot: cIAP1
2x105
MLN4924:
- +
- +
- +
control Casp-3 Casp-7
that relies on noncompetitive inhibition, most likely through a NEDD8-induced conformational change of the caspase. Similar to the effect of NEDD8 on drICE, allosteric regulation is also seen in caspase-1 (Scheer et al., 2006), indicating that caspases can be regulated through modication of regulatory residues outside the catalytic pocket. Although conjugation of Ub and/or NEDD8 reduces the catalytic activity of drICE from 100% (unmodied drICE) to 16% (modied drICE), it is important to note that neddylation or ubiquitylation of drICE does not completely inhibit the caspase in vitro. Moreover, combined modication of NEDD8 and Ub also does not further inhibit drICE under these conditions. However, it is likely that in vivo additional mechanisms are in place to limit drICE-mediated proteolysis. A common mode of regulation involves the recognition of the Ub and NEDD8-adducts by proteins with Ub- or NEDD8-binding domains (UBDs) (Haglund and Dikic, 2005). Therefore, Ub- and NEDD8-binding proteins may contribute to the regulation of modied caspases in vivo, perhaps by spatial sequestration.
Our data indicate that DIAP1 can promote the covalent attachment of NEDD8 RIP1 to the effector caspase drICE and cIAP1 DCP-1. This is surprising, since DIAP1 1 2 reportedly functions as an Ub-specic E3 (Vaux and Silke, 2005). Although NEDD8 is homologous to ubiquitin (57% identity), it has unique E1 and E2 enzymes that specically handle NEDD8 (Lee et al., 2008; Souphron et al., 2008; Whitby et al., 1998). The NEDD8 E1 is a heterodimer composed of the amyloid precursor protein-binding protein (APPBP1) and Uba3, while the NEDD8-E2 is Ubc12 (Gong and Yeh, 1999; Lammer et al., 1998; Liakopoulos et al., 1998; Osaka et al., 1998; Pozo et al., 1998). Moreover, until recently, the only known substrates for NEDD8 modication were the six members of the cullin family of proteins that are components of SCF (SKP-Cullin-F-Box) Ub-E3 ligase complexes. Lately, three other substrates of neddylation have been discovered (Oved et al., 2006; Xirodimas et al., 2004; Xirodimas et al., 2008). This study now identies IAPs as versatile E3 ligases that bind to E2s of the Ub and NEDD8 conjugation pathways and can stimulate the conjugation of Ub and NEDD8 to target substrates. Several lines of evidence indicate that IAPs can function as NEDD8-E3 in the NEDD8 conjugation cascade. First, in vivo DIAP1 readily promotes the covalent attachment of NEDD8 to
Lysate
Molecular Cell
itself and drICE. This is mediated by DIAP1 itself because neddylation of DIAP1 and drICE is dependent on DIAP1s RING nger domain. The RING mutants DIAP1(21438/C406Y) and DIAP1(21438/CD6) (Ditzel et al., 2008) fail to promote the conjugation of NEDD8 to DIAP1 and drICE. Likewise, in mammals XIAP is also capable of neddylating caspase-7 in a RING nger-dependent manner, indicating that IAP-mediated conjugation of NEDD8 to target substrates is evolutionarily conserved. Second, DIAP1 physically interacts with the NEDD8-E2 conjugating enzyme UbcD12. Under the same conditions, DIAP1 did not bind to the SUMO conjugating enzyme lesswright/UbcD9 and promoted neither SUMOylation of itself or target substrates, indicating that DIAP1 selectively interacts with the Ub and NEDD8 conjugation cascades. Third, DIAP1-mediated neddylation is diminished in cells in which components of the NEDD8 activation and conjugation cascade are knocked down by RNAi. This indicates that DIAP1 not only physically interacts with UbcD12 but also uses UbcD12 for the conjugation of NEDD8. Finally, neddylation of drICE appears to be DIAP1dependent since depletion of DIAP1 reduces the amount of neddylated forms of drICE. Therefore, DIAP1 seems to function as a versatile E3 ligase that is capable of interacting with Ub-specic E2s, such as UbcD1 (Ryoo et al., 2002) and Ubc13/Uev1a (Herman-Bachinsky et al., 2007) (M.B. and P.M., unpublished data), and the NEDD8-selective E2 UbcD12/ CG7375. Consistent with the notion that DIAP1 can handle both these conjugation systems, we nd that endogenous drICE is modied with both Ub and NEDD8. Accordingly, knockdown of the respective E1 and E2 enzymes results in a corresponding reduction in the overall smearing pattern of total modied drICE. In healthy cells, only a relatively small fraction of total drICE is modied. This is not entirely surprising because healthy cells only harbor a small, yet detectable, amount of processed caspases. Importantly, caspase cleavage is a prerequisite for DIAP1 binding and DIAP1-mediated ubiquitylation and neddylation. After their proteolytic cleavage, effector caspases expose an evolutionarily conserved IAP-binding motif that is required for their recruitment to DIAP1 (Zachariou et al., 2003; Ditzel et al., 2008). Hence, DIAP1 does not bind or modify full-length drICE. Consistent with the view that deneddylation is required to execute apoptosis, we nd that DEN1 efciently removes NEDD8 conjugates from drICE. Since NEDD8 suppresses the catalytic processivity of drICE, DEN1 seems to reverse NEDD8-mediated inhibition of drICE. Consistently, the neddylation status of drICE drastically changes upon UV-induced cell death. While the loss of NEDD8 modications on drICE may be achieved by UV-mediated depletion of DIAP1 alone, it is highly likely that UV also activates the deneddylase DEN1. In agreement with this view, a recent publication reports that genotoxic stress induces NEDP1 (Watson et al., 2010), the mammalian homolog of DEN1. Therefore, drICE deneddylation might be the result of coordinated action of DIAP1 depletion and DEN1 activation. While DEN1 is highly effective in removing NEDD8 from drICE, CSN5 failed to deneddylate drICE efciently. CSN5 acts as part of the COP9 signalosome that regulates the activity of SCF E3 ligases (Cope et al., 2002; Lyapina et al., 2001; Schwechheimer et al., 2001). Therefore, it is likely that CSN5
regulates cell death indirectly via modulating the activity of SCF E3 ligases. It is interesting to note that the F-box protein Morgue and SkpA, which are likely to form a SCF complex, have previously been reported to regulate apoptosis via modulating DIAP1 levels (Hays et al., 2002; Wing et al., 2002), leaving open the possibility that CSN5 might mediate its proapoptotic effect through Morgue/SkpA-SCF. Taken together, our data identify Drosophila and mammalian IAPs as versatile E3 ligases that promote the conjugation of Ub and NEDD8 to target substrates. Since IAPs act as multifunctional signaling devices that also exert caspase-independent functions, it is likely that IAPs modulate such processes via conjugation of both Ub and NEDD8. Clearly, both modications occur in vivo and appear to be required for proper regulation of cell death. Our data indicate that neddylation and ubiquitylation affect effector caspases through different mechanisms. While attachment of Ub acts as a mixed type inhibitor for caspase activity, acting through a competitive and a noncompetitive mode, neddylation inhibits caspases in a noncompetitive manner, most likely through conformational changes. Differential neddylation/deneddylation and/or ubiquitylation/deubiquitylation may well be used to ne-tune caspase activity to allow them to take part in nonapoptotic signaling processes. Furthermore, Ub- and NEDD8-specic binding proteins (Di Fiore et al., 2003) may also contribute to the suppression of modied caspases. Ultimately, a deeper understanding of how conjugation and deconjugation of Ub and UBLs determine cellular phenotypes will be key since aberrant Ub/UBL and apoptosis signaling sits at the heart of many human pathologies and may have signicant impact on the development of new therapeutic strategies.
EXPERIMENTAL PROCEDURES Fly Work GMR-GAL4,GMR-rpr, GMR-GAL4,GMR-hid, and DEN1EX9 ies, lacking the complete coding region of DEN1, were previously described (Chan et al., 2008; Leulier et al., 2006). UAS-DUB-dsRNA (RNAi) lines were obtained from the Vienna Drosophila RNAi Center (VDRC) and the National Institute of Genetics (NIG-y), Japan. All crosses were performed at 25 C. Eye sizes were captured with the Quick Selection Tool of Adobe Photoshop, and the numbers of pixels were measured. Constructs, Antibodies, and Recombinant Proteins Constructs were cloned into pAc, pMT, pMT-GTC (Zachariou et al., 2003), or pcDNA3 (Invitrogen) and veried by sequencing. DHFR-HA-Ub-based diap1 constructs, full-length DRONC-V5/FLAG, and Ub-ALG-drICE(WT)V5/FLAG constructs were described previously (Ditzel et al., 2003, 2008; Tenev et al., 2007). Recombinant proteins were produced as described previously (Ditzel et al., 2008). Antibodies were as follows: a-V5 (AbD Serotec), a-GST (GE Healthcare), a-HA (Roche), a-PARP1 (F-2, Santa Cruz), a-Actin (C-11, Santa Cruz), a-NEDD8 (Alexis, 210-194 and Cell Signaling, #2745), a-Caspase-7 (BD PharMingen), a-cIAP1 (Alexis), and a-RIP1 (BD PharMingen). a-DIAP1 and a-drICE antibodies were described previously (Zachariou et al., 2003). RNAi in S2 Cells Production of double-stranded DNA (dsRNA) and transfection were performed as described previously (Wilson et al., 2002). The following coding regions were used to produce dsRNA: APPBP1, nt 94-760; UbcD12 (CG7375), nt 15525; Uba1, nt 40546; UbcD1, nt 297808; and drICE, nt 84763.
Molecular Cell
Immunoprecipitations, Ubiquitylation, and Neddylation Assays Coimmunoprecipitation and immunoblot assays were performed as described previously (Tenev et al., 2005). Endogenous drICE was immunoprecipitated with guinea pig-derived a-drICE antibody that was crosslinked to protein A agarose beads (Af-Gel, Biorad). For detection of neddylated drICE, 50 mM Chloroacetamide was added to the lysis buffer. Where indicated, S2 cells were irradiated with 150K mJ UV (UV Stratalinker, Stratagene) and lysed 4 hr after treatment. For the ubiquitylation and neddylation assays, cells were cotransfected with the indicated constructs and harvested as described previously (Ditzel et al., 2008; Rodriguez et al., 1999). For in vitro neddylation, PARP1 cleavage, and DEVDase, 4 mg recombinant drICE was incubated in the presence or absence of 1 mg DIAP1, 1 mg E2 (UbcD1), 80 ng E1, and 2 mg Ub or 4 mg NEDD8 (or as indicated) in a nal volume of 30 ml reaction buffer. Reactions were carried out at 37 C for 90 min. Note that in vitro neddylation was carried out in the presence of Ub-E1 and Ub-E2, which have been shown to handle NEDD8 in vitro (Whitby et al., 1998). PARP1 cleavage and DEVDase assays were carried out as previously described (Ditzel et al., 2008). For the two-step in vitro neddylation/ubiquitylation assay, DIAP1 was incubated with E1, E2, and either Ub or N8 for 1 hr. Unincorporated Ub and NEDD8 was removed with Amicon Centrifugal Filter Units (30K cutoff, Millipore). For the second step, new reaction mixture, E1, E2, recombinant drICE, and either Ub or N8 were added and incubated for 90 min at 37C. Cell-Based Caspase Activity Assay 293T cells were transfected with pcDNA3-based C-terminally, V5-tagged caspase-3 and -7 constructs, or control vector. Twenty-four hours after transfection, cells were incubated with 5 mM MLN4924 (Millennium Pharmaceuticals) for 16 hr or left untreated. Cells were lysed, and 20 ml supernatant was added to 350 ml DEVDase assay mixture (Ditzel et al., 2008). The reaction was incubated at RT and analyzed at 380 nM excitation/460 nM emission. LI-COR Odyssey Analysis and Enzyme Kinetics For quantitation of expression levels, immunoblots were incubated with Odyssey-compatible secondary antibodies (mouse IR680 [Invitrogen], rat IR680 [Invitrogen], and guinea pig IR800 [Rockland]) and analyzed at 700 nm and 800 nm excitation, respectively. Odyssey software was used to determine the integrated intensities. Experiments for determining KM and Vmax were performed in triplicates and analyzed with Prism software. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures and four gures and can be found with this article online at doi:10.1016/ j.molcel.2010.11.011. ACKNOWLEDGMENTS We would like to thank Millenium Pharmaceuticals for MLN4924 compound. We thank members of the Meier Lab for support and critical reading of the manuscript. In particular, we would like to thank Mariam Orme for advice with y genetics, Marketa Zvelebil for help with bioinformatics, and Helen Walden for support in structural information. M.B. is supported by a fellowship of the Deutsche Forschungsgemeinschaft and Breakthrough. We acknowledge National Health Service funding to the National Institute for Health Research Biomedical Research Centre. Received: January 25, 2010 Revised: August 2, 2010 Accepted: September 13, 2010 Published: December 9, 2010 REFERENCES Broemer, M., and Meier, P. (2009). Ubiquitin-mediated regulation of apoptosis. Trends Cell Biol. 19, 130140.
Brzovic, P.S., and Klevit, R.E. (2006). Ubiquitin transfer from the E2 perspective: why is UbcH5 so promiscuous? Cell Cycle 5, 28672873. Chai, J., Yan, N., Huh, J.R., Wu, J.W., Li, W., Hay, B.A., and Shi, Y. (2003). Molecular mechanism of Reaper-Grim-Hid-mediated suppression of DIAP1-dependent Dronc ubiquitination. Nat. Struct. Biol. 10, 892898. Chan, Y., Yoon, J., Wu, J.T., Kim, H.J., Pan, K.T., Yim, J., and Chien, C.T. (2008). DEN1 deneddylates non-cullin proteins in vivo. J. Cell Sci. 121, 32183223. Choi, Y.E., Butterworth, M., Malladi, S., Duckett, C.S., Cohen, G.M., and Bratton, S.B. (2009). The E3 ubiquitin ligase cIAP1 binds and ubiquitinates caspase-3 and -7 via unique mechanisms at distinct steps in their processing. J. Biol. Chem. 284, 1277212782. Cope, G.A., Suh, G.S., Aravind, L., Schwarz, S.E., Zipursky, S.L., Koonin, E.V., and Deshaies, R.J. (2002). Role of predicted metalloprotease motif of Jab1/ Csn5 in cleavage of Nedd8 from Cul1. Science 298, 608611. Deshaies, R.J., and Joazeiro, C.A. (2009). RING domain E3 ubiquitin ligases. Annu. Rev. Biochem. 78, 399434. Di Fiore, P.P., Polo, S., and Hofmann, K. (2003). When ubiquitin meets ubiquitin receptors: a signalling connection. Nat. Rev. Mol. Cell Biol. 4, 491497. Dikic, I., Wakatsuki, S., and Walters, K.J. (2009). Ubiquitin-binding domains from structures to functions. Nat. Rev. Mol. Cell Biol. 10, 659671. Ditzel, M., and Meier, P. (2002). IAP degradation: decisive blow or altruistic sacrice? Trends Cell Biol. 12, 449452. Ditzel, M., Wilson, R., Tenev, T., Zachariou, A., Paul, A., Deas, E., and Meier, P. (2003). Degradation of DIAP1 by the N-end rule pathway is essential for regulating apoptosis. Nat. Cell Biol. 5, 467473. Ditzel, M., Broemer, M., Tenev, T., Bolduc, C., Lee, T.V., Rigbolt, K.T., Elliott, R., Zvelebil, M., Blagoev, B., Bergmann, A., and Meier, P. (2008). Inactivation of effector caspases through nondegradative polyubiquitylation. Mol. Cell 32, 540553. Duda, D.M., Borg, L.A., Scott, D.C., Hunt, H.W., Hammel, M., and Schulman, B.A. (2008). Structural insights into NEDD8 activation of cullin-RING ligases: conformational control of conjugation. Cell 134, 9951006. Eckelman, B.P., Salvesen, G.S., and Scott, F.L. (2006). Human inhibitor of apoptosis proteins: why XIAP is the black sheep of the family. EMBO Rep. 7, 988994. Furukawa, M., Zhang, Y., McCarville, J., Ohta, T., and Xiong, Y. (2000). The CUL1 C-terminal sequence and ROC1 are required for efcient nuclear accumulation, NEDD8 modication, and ubiquitin ligase activity of CUL1. Mol. Cell. Biol. 20, 81858197. Gaither, A., Porter, D., Yao, Y., Borawski, J., Yang, G., Donovan, J., Sage, D., Slisz, J., Tran, M., Straub, C., et al. (2007). A Smac mimetic rescue screen reveals roles for inhibitor of apoptosis proteins in tumor necrosis factor-alpha signaling. Cancer Res. 67, 1149311498. Gao, F., Cheng, J., Shi, T., and Yeh, E.T. (2006). Neddylation of a breast cancer-associated protein recruits a class III histone deacetylase that represses NFkappaB-dependent transcription. Nat. Cell Biol. 8, 11711177. Garcia-Dominguez, M., and Reyes, J.C. (2009). SUMO association with repressor complexes, emerging routes for transcriptional control. Biochim. Biophys. Acta 1789, 451459. Gong, L., and Yeh, E.T. (1999). Identication of the activating and conjugating enzymes of the NEDD8 conjugation pathway. J. Biol. Chem. 274, 12036 12042. Grether, M.E., Abrams, J.M., Agapite, J., White, K., and Steller, H. (1995). The head involution defective gene of Drosophila melanogaster functions in programmed cell death. Genes Dev. 9, 16941708. Haglund, K., and Dikic, I. (2005). Ubiquitylation and cell signaling. EMBO J. 24, 33533359. Hatakeyama, S., and Nakayama, K.I. (2003). U-box proteins as a new family of ubiquitin ligases. Biochem. Biophys. Res. Commun. 302, 635645.
Molecular Cell
Hays, R., Wickline, L., and Cagan, R. (2002). Morgue mediates apoptosis in the Drosophila melanogaster retina by promoting degradation of DIAP1. Nat. Cell Biol. 4, 425431. Herman-Bachinsky, Y., Ryoo, H.D., Ciechanover, A., and Gonen, H. (2007). Regulation of the Drosophila ubiquitin ligase DIAP1 is mediated via several distinct ubiquitin system pathways. Cell Death Differ. 14, 861871. Hochstrasser, M. (2009). Origin and function of ubiquitin-like proteins. Nature 458, 422429. Hoeller, D., and Dikic, I. (2009). Targeting the ubiquitin system in cancer therapy. Nature 458, 438444. Holley, C.L., Olson, M.R., Colon-Ramos, D.A., and Kornbluth, S. (2002). Reaper eliminates IAP proteins through stimulated IAP degradation and generalized translational inhibition. Nat. Cell Biol. 4, 439444. Jin, Z., Li, Y., Pitti, R., Lawrence, D., Pham, V.C., Lill, J.R., and Ashkenazi, A. (2009). Cullin3-based polyubiquitination and p62-dependent aggregation of caspase-8 mediate extrinsic apoptosis signaling. Cell 137, 721735. Kaiser, W.J., Vucic, D., and Miller, L.K. (1998). The Drosophila inhibitor of apoptosis D-IAP1 suppresses cell death induced by the caspase drICE. FEBS Lett. 440, 243248. Kerscher, O., Felberbaum, R., and Hochstrasser, M. (2006). Modication of proteins by ubiquitin and ubiquitin-like proteins. Annu. Rev. Cell Dev. Biol. 22, 159180. Kirkpatrick, D.S., Hathaway, N.A., Hanna, J., Elsasser, S., Rush, J., Finley, D., King, R.W., and Gygi, S.P. (2006). Quantitative analysis of in vitro ubiquitinated cyclin B1 reveals complex chain topology. Nat. Cell Biol. 8, 700710. Komander, D. (2009). The emerging complexity of protein ubiquitination. Biochem. Soc. Trans. 37, 937953. Lammer, D., Mathias, N., Laplaza, J.M., Jiang, W., Liu, Y., Callis, J., Goebl, M., and Estelle, M. (1998). Modication of yeast Cdc53p by the ubiquitin-related protein rub1p affects function of the SCFCdc4 complex. Genes Dev. 12, 914926. Lee, M.R., Lee, D., Shin, S.K., Kim, Y.H., and Choi, C.Y. (2008). Inhibition of APP intracellular domain (AICD) transcriptional activity via covalent conjugation with Nedd8. Biochem. Biophys. Res. Commun. 366, 976981. Leulier, F., Ribeiro, P.S., Palmer, E., Tenev, T., Takahashi, K., Robertson, D., Zachariou, A., Pichaud, F., Ueda, R., and Meier, P. (2006). Systematic in vivo RNAi analysis of putative components of the Drosophila cell death machinery. Cell Death Differ. 13, 16631674. Liakopoulos, D., Doenges, G., Matuschewski, K., and Jentsch, S. (1998). A novel protein modication pathway related to the ubiquitin system. EMBO J. 17, 22082214. Lisi, S., Mazzon, I., and White, K. (2000). Diverse domains of THREAD/DIAP1 are required to inhibit apoptosis induced by REAPER and HID in Drosophila. Genetics 154, 669678. Lyapina, S., Cope, G., Shevchenko, A., Serino, G., Tsuge, T., Zhou, C., Wolf, D.A., Wei, N., Shevchenko, A., and Deshaies, R.J. (2001). Promotion of NEDD-CUL1 conjugate cleavage by COP9 signalosome. Science 292, 13821385. Merlet, J., Burger, J., Gomes, J.E., and Pintard, L. (2009). Regulation of cullin-RING E3 ubiquitin-ligases by neddylation and dimerization. Cell. Mol. Life Sci. 66, 19241938. Morimoto, M., Nishida, T., Honda, R., and Yasuda, H. (2000). Modication of cullin-1 by ubiquitin-like protein Nedd8 enhances the activity of SCF(skp2) toward p27(kip1). Biochem. Biophys. Res. Commun. 270, 10931096. Morizane, Y., Honda, R., Fukami, K., and Yasuda, H. (2005). X-linked inhibitor of apoptosis functions as ubiquitin ligase toward mature caspase-9 and cytosolic Smac/DIABLO. J. Biochem. 137, 125132. Mukhopadhyay, D., and Dasso, M. (2007). Modication in reverse: the SUMO proteases. Trends Biochem. Sci. 32, 286295. Nijman, S.M., Luna-Vargas, M.P., Velds, A., Brummelkamp, T.R., Dirac, A.M., Sixma, T.K., and Bernards, R. (2005). A genomic and functional inventory of deubiquitinating enzymes. Cell 123, 773786.
Osaka, F., Kawasaki, H., Aida, N., Saeki, M., Chiba, T., Kawashima, S., Tanaka, K., and Kato, S. (1998). A new NEDD8-ligating system for cullin-4A. Genes Dev. 12, 22632268. Oved, S., Mosesson, Y., Zwang, Y., Santonico, E., Shtiegman, K., Marmor, M.D., Kochupurakkal, B.S., Katz, M., Lavi, S., Cesareni, G., and Yarden, Y. (2006). Conjugation to Nedd8 instigates ubiquitylation and down-regulation of activated receptor tyrosine kinases. J. Biol. Chem. 281, 2164021651. Podust, V.N., Brownell, J.E., Gladysheva, T.B., Luo, R.S., Wang, C., Coggins, M.B., Pierce, J.W., Lightcap, E.S., and Chau, V. (2000). A Nedd8 conjugation pathway is essential for proteolytic targeting of p27Kip1 by ubiquitination. Proc. Natl. Acad. Sci. USA 97, 45794584. Pozo, J.C., Timpte, C., Tan, S., Callis, J., and Estelle, M. (1998). The ubiquitinrelated protein RUB1 and auxin response in Arabidopsis. Science 280, 1760 1763. Read, M.A., Brownell, J.E., Gladysheva, T.B., Hottelet, M., Parent, L.A., Coggins, M.B., Pierce, J.W., Podust, V.N., Luo, R.S., Chau, V., and Palombella, V.J. (2000). Nedd8 modication of cul-1 activates SCF(beta(TrCP))-dependent ubiquitination of IkappaBalpha. Mol. Cell. Biol. 20, 23262333. Reyes-Turcu, F.E., Ventii, K.H., and Wilkinson, K.D. (2009). Regulation and cellular roles of ubiquitin-specic deubiquitinating enzymes. Annu. Rev. Biochem. 78, 363397. Rodriguez, M.S., Wright, J., Thompson, J., Thomas, D., Baleux, F., Virelizier, J.L., Hay, R.T., and Arenzana-Seisdedos, F. (1996). Identication of lysine residues required for signal-induced ubiquitination and degradation of I kappa B-alpha in vivo. Oncogene 12, 24252435. Rodriguez, M.S., Desterro, J.M., Lain, S., Midgley, C.A., Lane, D.P., and Hay, R.T. (1999). SUMO-1 modication activates the transcriptional response of p53. EMBO J. 18, 64556461. Rodriguez, M.S., Desterro, J.M., Lain, S., Lane, D.P., and Hay, R.T. (2000). Multiple C-terminal lysine residues target p53 for ubiquitin-proteasomemediated degradation. Mol. Cell. Biol. 20, 84588467. Rotin, D., and Kumar, S. (2009). Physiological functions of the HECT family of ubiquitin ligases. Nat. Rev. Mol. Cell Biol. 10, 398409. Ryoo, H.D., Bergmann, A., Gonen, H., Ciechanover, A., and Steller, H. (2002). Regulation of Drosophila IAP1 degradation and apoptosis by reaper and ubcD1. Nat. Cell Biol. 4, 432438. Saha, A., and Deshaies, R.J. (2008). Multimodal activation of the ubiquitin ligase SCF by Nedd8 conjugation. Mol. Cell 32, 2131. Scheer, J.M., Romanowski, M.J., and Wells, J.A. (2006). A common allosteric site and mechanism in caspases. Proc. Natl. Acad. Sci. USA 103, 75957600. Scherer, D.C., Brockman, J.A., Chen, Z., Maniatis, T., and Ballard, D.W. (1995). Signal-induced degradation of I kappa B alpha requires site-specic ubiquitination. Proc. Natl. Acad. Sci. USA 92, 1125911263. Schile, A.J., Garca-Fernandez, M., and Steller, H. (2008). Regulation of apoptosis by XIAP ubiquitin-ligase activity. Genes Dev. 22, 22562266. Schulman, B.A., and Harper, J.W. (2009). Ubiquitin-like protein activation by E1 enzymes: the apex for downstream signalling pathways. Nat. Rev. Mol. Cell Biol. 10, 319331. Schwechheimer, C., Serino, G., Callis, J., Crosby, W.L., Lyapina, S., Deshaies, R.J., Gray, W.M., Estelle, M., and Deng, X.W. (2001). Interactions of the COP9 signalosome with the E3 ubiquitin ligase SCFTIRI in mediating auxin response. Science 292, 13791382. Shapiro, P.J., Hsu, H.H., Jung, H., Robbins, E.S., and Ryoo, H.D. (2008). Regulation of the Drosophila apoptosome through feedback inhibition. Nat. Cell Biol. 10, 14401446. Silke, J., Kratina, T., Chu, D., Ekert, P.G., Day, C.L., Pakusch, M., Huang, D.C., and Vaux, D.L. (2005). Determination of cell survival by RING-mediated regulation of inhibitor of apoptosis (IAP) protein abundance. Proc. Natl. Acad. Sci. USA 102, 1618216187. Soucy, T.A., Smith, P.G., Milhollen, M.A., Berger, A.J., Gavin, J.M., Adhikari, S., Brownell, J.E., Burke, K.E., Cardin, D.P., Critchley, S., et al. (2009). An inhibitor of NEDD8-activating enzyme as a new approach to treat cancer. Nature 458, 732736.
Molecular Cell
Souphron, J., Waddell, M.B., Paydar, A., Tokgoz-Gromley, Z., Roussel, M.F., and Schulman, B.A. (2008). Structural dissection of a gating mechanism preventing misactivation of ubiquitin by NEDD8s E1. Biochemistry 47, 89618969. Stickle, N.H., Chung, J., Klco, J.M., Hill, R.P., Kaelin, W.G., Jr., and Ohh, M. (2004). pVHL modication by NEDD8 is required for bronectin matrix assembly and suppression of tumor development. Mol. Cell. Biol. 24, 3251 3261. Suzuki, Y., Nakabayashi, Y., Nakata, K., Reed, J.C., and Takahashi, R. (2001). X-linked inhibitor of apoptosis protein (XIAP) inhibits caspase-3 and -7 in distinct modes. J. Biol. Chem. 276, 2705827063. Tasaki, T., Mulder, L.C., Iwamatsu, A., Lee, M.J., Davydov, I.V., Varshavsky, A., Muesing, M., and Kwon, Y.T. (2005). A family of mammalian E3 ubiquitin ligases that contain the UBR box motif and recognize N-degrons. Mol. Cell. Biol. 25, 71207136. Tenev, T., Zachariou, A., Wilson, R., Ditzel, M., and Meier, P. (2005). IAPs are functionally non-equivalent and regulate effector caspases through distinct mechanisms. Nat. Cell Biol. 7, 7077. Tenev, T., Ditzel, M., Zachariou, A., and Meier, P. (2007). The antiapoptotic activity of insect IAPs requires activation by an evolutionarily conserved mechanism. Cell Death Differ. 14, 11911201. Treier, M., Staszewski, L.M., and Bohmann, D. (1994). Ubiquitin-dependent c-Jun degradation in vivo is mediated by the delta domain. Cell 78, 787798. Varshavsky, A. (2000). Ubiquitin fusion technique and its descendants. Methods Enzymol. 327, 578593. Vaux, D.L., and Silke, J. (2005). IAPs, RINGs and ubiquitylation. Nat. Rev. Mol. Cell Biol. 6, 287297. Vijay-Kumar, S., Bugg, C.E., and Cook, W.J. (1987). Structure of ubiquitin rened at 1.8 A resolution. J. Mol. Biol. 194, 531544. Watson, I.R., Li, B.K., Roche, O., Blanch, A., Ohh, M., and Irwin, M.S. (2010). Chemotherapy induces NEDP1-mediated destabilization of MDM2. Oncogene 29, 297304.
Whitby, F.G., Xia, G., Pickart, C.M., and Hill, C.P. (1998). Crystal structure of the human ubiquitin-like protein NEDD8 and interactions with ubiquitin pathway enzymes. J. Biol. Chem. 273, 3498334991. White, K., Grether, M.E., Abrams, J.M., Young, L., Farrell, K., and Steller, H. (1994). Genetic control of programmed cell death in Drosophila. Science 264, 677683. Wilson, R., Goyal, L., Ditzel, M., Zachariou, A., Baker, D.A., Agapite, J., Steller, H., and Meier, P. (2002). The DIAP1 RING nger mediates ubiquitination of Dronc and is indispensable for regulating apoptosis. Nat. Cell Biol. 4, 445450. Wing, J.P., Schreader, B.A., Yokokura, T., Wang, Y., Andrews, P.S., Huseinovic, N., Dong, C.K., Ogdahl, J.L., Schwartz, L.M., White, K., and Nambu, J.R. (2002). Drosophila Morgue is an F box/ubiquitin conjugase domain protein important for grim-reaper mediated apoptosis. Nat. Cell Biol. 4, 451456. Wu, K., Chen, A., and Pan, Z.Q. (2000). Conjugation of Nedd8 to CUL1 enhances the ability of the ROC1-CUL1 complex to promote ubiquitin polymerization. J. Biol. Chem. 275, 3231732324. Xirodimas, D.P. (2008). Novel substrates and functions for the ubiquitin-like molecule NEDD8. Biochem. Soc. Trans. 36, 802806. Xirodimas, D.P., Saville, M.K., Bourdon, J.C., Hay, R.T., and Lane, D.P. (2004). Mdm2-mediated NEDD8 conjugation of p53 inhibits its transcriptional activity. Cell 118, 8397. Xirodimas, D.P., Sundqvist, A., Nakamura, A., Shen, L., Botting, C., and Hay, R.T. (2008). Ribosomal proteins are targets for the NEDD8 pathway. EMBO Rep. 9, 280286. Yan, N., Wu, J.W., Chai, J., Li, W., and Shi, Y. (2004). Molecular mechanisms of DrICE inhibition by DIAP1 and removal of inhibition by Reaper, Hid and Grim. Nat. Struct. Mol. Biol. 11, 420428. Yoo, S.J., Huh, J.R., Muro, I., Yu, H., Wang, L., Wang, S.L., Feldman, R.M., Clem, R.J., Muller, H.A., and Hay, B.A. (2002). Hid, Rpr and Grim negatively regulate DIAP1 levels through distinct mechanisms. Nat. Cell Biol. 4, 416424. Zachariou, A., Tenev, T., Goyal, L., Agapite, J., Steller, H., and Meier, P. (2003). IAP-antagonists exhibit non-redundant modes of action through differential DIAP1 binding. EMBO J. 22, 66426652.
Molecular Cell
Article
The Proapoptotic Function of Noxa in Human Leukemia Cells Is Regulated by the Kinase Cdk5 and by Glucose
Xazmin H. Lowman,1,2 Maureen A. McDonnell,1 Ashley Kosloske,6 Oludare A. Odumade,5,7 Christopher Jenness,8 Christine B. Karim,3 Ronald Jemmerson,2,4,7 and Ameeta Kelekar1,2,*
of Laboratory Medicine and Pathology Cancer Center 3Department of Biochemistry, Molecular Biology and Biophysics 4Department of Microbiology 5Combined MD/PhD Training Program University of Minnesota, Minneapolis, MN 55455, USA 6College of Biological Sciences, University of Minnesota, St. Paul, MN 55108, USA 7Center for Immunology, University of Minnesota, Minneapolis, MN 55414, USA 8Lineld College, McMinnville, OR 97128, USA *Correspondence: ameeta@umn.edu DOI 10.1016/j.molcel.2010.11.035
2Masonic 1Department
SUMMARY
The BH3-only protein, Noxa, is induced in response to apoptotic stimuli, such as DNA damage, hypoxia, and proteasome inhibition in most human cells. Noxa is constitutively expressed in proliferating cells of hematopoietic lineage and required for apoptosis in response to glucose stress. We show that Noxa is phosphorylated on a serine residue (S13) in the presence of glucose. Phosphorylation promotes its cytosolic sequestration and suppresses its apoptotic function. We identify Cdk5 as the Noxa kinase and show that Cdk5 knockdown or expression of a Noxa S13 to A mutant increases sensitivity to glucose starvation, conrming that the phosphorylation is protective. Both glucose deprivation and Cdk5 inhibition promote apoptosis by dephosphorylating Noxa. Paradoxically, Noxa stimulates glucose consumption and may enhance glucose turnover via the pentose phosphate pathway rather than through glycolysis. We propose that Noxa plays both growthpromoting and proapoptotic roles in hematopoietic cancers with phospho-S13 as the glucose-sensitive toggle switch controlling these opposing functions.
INTRODUCTION Interactions between multidomain Bcl-2 proteins and BH3-only proteins of the family are pivotal in regulating apoptosis. Activated BH3-only proteins bring proapoptotic signals to the mitochondria where they interact with specic binding partners to promote cytochrome c release. Noxa interacts with the prosurvival family member Mcl-1L in response to stimuli such as DNA
damage and oxygen deprivation (Kim et al., 2004; Willis et al., 2005). Both murine and human proteins are activated by transcriptional mechanisms in a cell type- and stimulus-dependent manner (Oda et al., 2000; Shibue et al., 2003; Villunger et al., 2003); however, there are signicant structural and regulatory differences between the two. Human (h)Noxa has 54 residues and one BH3 domain; mouse Noxa comprises 103 residues and has two BH3 domains. Most studies to date have identied upregulation via transcription factors, such as E2F, (HIF)-1a, c-myc, and ATF3, as the underlying mechanism of activation of the human protein (Erster et al., 2004; Hershko and Ginsberg, 2004; Nikiforov et al., 2007; Perez-Galan et al., 2006; Wang et al., 2009). However, hNoxa is constitutively expressed in certain cell types, particularly in primary and malignant cells of hematopoietic lineage (Alves et al., 2006), raising the possibility of its posttranslational regulation. Mcl-1L is required for survival and maintenance of cells in the developing human immune system (Opferman et al., 2003, 2005), and Noxa may play an important part in regulating this survival function (Alves et al., 2006; Ploner et al., 2008). The Noxa/Mcl-1L axis also acts to restrain lymphocyte expansion after an immune response (Alves et al., 2007). In keeping with its contribution to the survival of hematopoietic cells, increased expression of Mcl-1L is associated with a majority of hematological malignancies (Warr and Shore, 2008). Paradoxically, many of these cancers also tolerate high, constitutive levels of Noxa (Alves et al., 2006; Hallaert et al., 2007). Both Mcl-1L and Noxa are markedly upregulated in activated human primary T cell populations in the presence of adequate nutrients and glucose. The dramatic increase in aerobic glycolysis in activated T cells is essential for their proliferation, survival, and effector function, and the requirement for Noxa and Mcl-1L during this period of expansion and high energy utilization (Maurer et al., 2006; Zhao et al., 2007) suggests a role for the Noxa/Mcl-1L axis in glucose metabolism. The induction of Noxa also ensures that the cells are poised to mount a robust apoptotic response to
Molecular Cell
Glucose and Cdk5 Regulate Human Noxa Function
glucose and nutrient stress. Although BH3-only proteins Bim, Bid, and Puma all interact with Mcl-1L, only the Noxa/Mcl-1L interaction targets the antiapoptotic protein for proteasomal degradation (Czabotar et al., 2007). Exposure of B-CLL cells to r-Roscovitine results in apoptosis, preceded by Mcl-1L degradation, in a Noxa-dependent manner (Hallaert et al., 2007). Roscovitine (rosco) is a potent inhibitor of the atypical cyclin-dependent kinase, Cdk5 (Goodyear and Sharma, 2007; Mapelli et al., 2005; Muruais et al., 2009), and its dependence on Noxa for its apoptotic effect suggested a role for Cdk5 in regulating Noxa. Cdk5 is a proline-directed serine/threonine kinase highly homologous to members of the Cdk family, yet not activated by cyclins (Dhavan and Tsai, 2001). It has a diverse array of substrates, including pRb, ErbB, src, b-catenin, STAT3, and its own binding partners and activators, p35 and p39 (Dhavan and Tsai, 2001). Until recently, Cdk5 activity was primarily associated with neuronal development and survival. However, growing evidence is attributing diverse functions, including differentiation, senescence, and apoptosis, to the kinase in nonneuronal hematopoietic cells, adipocytes, and pancreatic b cells. Of particular relevance is the contribution of active Cdk5/p35 complexes to glucosedependent insulin secretion and regulation of glucose metabolism and uptake in the latter (Lalioti et al., 2009; Rosales and Lee, 2006; Ubeda et al., 2004). The requirement for Noxa in glucose deprivation-induced apoptosis of proliferating immune cells, coupled with its high constitutive expression in hematopoietic cancers, prompted us to investigate regulatory mechanisms that control the proapoptotic function of this protein. Our studies reveal that stably expressed hNoxa is phosphorylated and sequestered in the cytosol in actively proliferating leukemia cell lines and primary T cells. We identify Cdk5 as the Noxa kinase and show that it modies the BH3-only protein in a glucose-dependent manner. Our studies also reveal a growth-promoting role for phosphorylated and sequestered Noxa in glucose metabolism.
Human Noxa Is Phosphorylated on a Serine Residue Since constitutively expressed BH3-only proteins are often posttranslationally modied to keep their proapoptotic activity in check, we scanned the Noxa amino acid sequence for potential regulatory sites. The CBS Prediction Server NetPhos (Blom et al., 1999) identied serine 13 (S13) on the protein (Figure 1E) as a potential kinase target. The motif QPSPARA is a strong consensus for the cyclin-dependent kinase Cdk5, and a weaker target site for Cdk1, GSK-3 a/b, and p38MAP kinases. We rst determined whether Noxa could be phosphorylated in vitro using extracts from a variety of cell lines and recombinant (r) Noxa-FLAG as substrate (Figure 1F). All the cell extracts tested labeled rNoxa-FLAG in the assay. An S13 to A mutant of Noxa (SA-FLAG), however, remained unlabeled in a similar assay (Figure 1G). To determine whether Noxa could be phosphorylated intracellularly and to identify the phosphorylated residue(s), HEK293 cells, transiently transfected with FLAG-tagged wildtype (WT) Noxa and DA (D34), SA and SE (S13) mutants, were labeled with [32P]-orthophosphate. Although the transfected proteins were comparably expressed, only WT Noxa-FLAG and BH3 domain mutant, DA, were labeled (Figure 1H). Next, we generated phosphospecic monoclonal and polyclonal antibodies (mAbs and pAbs) against a synthetic peptide encompassing the S13 motif (production and purication described in the Supplemental Experimental Procedures, available online; Abs listed in Table S1). The rabbit pAbs (rAK2), afnity puried against the nonphosphopeptide to remove any cross-reactivity, specically detected ectopically expressed WT Noxa, but not the SA or SE mutants, in western blots (Figure 1I). Phosphorylated Noxa Localizes Primarily to the Cytosol Fluorescent labeling of HL-60, Jurkat (Figures 2A and 2B), K562, and B cells (not shown) with the phosphospecic 2B3 mAb showed pS13 Noxa primarily in the cytosol. Confocal microscopy indicated minimal overlap with the Mitotracker Red (MTR)stained mitochondria. We also examined 2B3-, MTR-, and DAPI-labeled cells by imaging ow cytometry (IFC), a technique that combines ow cytometry and confocal microscopy. The ImageStream Multispectral Imaging Flow Cytometer (MIFC), which allows quantication of uorescence intensity as well as measurement of uorescent signal distribution and numerous morphological and absorbance characteristics of each cell (Figure 2C), conrmed the cytosolic localization of the pS13Noxa. We were also able to determine that although 82% of pS13 Noxa was in the same (cytosolic and nonnuclear) compartment as mitochondria, only 13% directly colocalized with these organelles. Noxa Is Phosphorylated by Cdk5, an Atypical Cyclin-Dependent Kinase The motif QPSPARA on Noxa is a strong consensus motif for Cdk5, a proline-directed S/T kinase (Dhavan and Tsai, 2001). Using common substrate histone H1 as positive control, we determined that puried active Cdk5, not Cdk1, could phosphorylate Noxa-FLAG in vitro (Figure 3A). Using SA-FLAG as substrate, we conrmed S13 as the target residue for this kinase (Figure 3B). Neither active rGSK-3a nor -b could phosphorylate Noxa in kinase assays (data not shown). The small size of Noxa increased the potential for an epitope tag to interfere
RESULTS Human Noxa Is Constitutively Expressed in Myeloid and Lymphoid Cancer Cells Noxa is induced via p53-dependent and -independent transcriptional mechanisms in response to stress stimuli such as DNA damage and hypoxia, but the human protein shows high steady-state levels in proliferating hematopoietic cells. A western blot of selected established lines is shown in Figure 1A. Fresh lymphoid and myeloid cancer patient samples also show constitutive expression of Noxa (Figures 1B and 1C). Moreover, Noxa protein levels do not increase in response to DNA damaging agents, camptothecin (CPT), or etoposide in the hematopoietic cell lines shown (Figure 1A), although the protein translocates to mitochondria within a few hours of treatment. A representative experiment with CPT-treated HL-60 cells is shown in Figure 1D. Consistent with published studies, levels of Mcl-1L, Noxas antiapoptotic binding partner are reduced within hours of CPT exposure in both whole-cell lysates and mitochondrial fractions.
Molecular Cell
Figure 1. Human Noxa Is Constitutively Expressed in Hematopoietic Cancer Cell Lines and Modied on a Serine Residue Near Its Amino Terminus
(AC) Western blots (WB) of Noxa and Actin expression in 50 mg cell protein. (A) Established lines derived from cancers. Cell lines from left to right are as follows: Molt-4 (T-ALL), K562 (CML), Jurkat (T-ALL), HL-60 (AML), LN-CAP, C4-2 (prostate cancer), MCF-7 (breast cancer), H522 (small cell lung cancer), A549 (alveolar basal epithelial carcinoma), and DLD1 (colon cancer). (B) T-ALL cells. From left to right, established lines (2), patient-derived (p) T-ALL cells (3), and fresh thymocytes from isolated normal human thymus. (C) AML cells. From left to right, established CML cell line and four patient-derived (p) AML cells. (D) Noxa translocates to mitochondria in response to CPT. WBs of whole-cell lysate protein (left panel) or mitochondrially enriched fractions (right panel) of HL-60 cells that were treated with 2 mM CPT for the times indicated. Blots were sequentially stripped and reprobed with relevant Abs. Mitochondrial cytochrome oxidase subunit IV (COX IV) is a loading control. (E) Amino acid sequence of hNoxa. The BH3 domain (LRRFGDK) and putative kinase substrate motif (QPSPARA) are underlined. (F) Recombinant (r) FLAG-tagged Noxa is phosphorylated by endogenous kinases. Cell lysates (10 mg) from HL-60, K562, Jurkat, and murine FL5.12 cells were incubated for 30 min with 1 mg Noxa-FLAG at 30 C in a total volume of 50 ml in kinase buffer (KB) with 5 mCi [g-32P]-ATP per reaction. The top panel is an autoradiograph of labeled Noxa-FLAG immunoprecipitated with anti-FLAG mAb, and the lower panel is a WB showing total Noxa levels. (G) HL-60 cell lysates (10 mg) were incubated with 1 mg rNoxa-FLAG or SA-FLAG in KB and [g-32P]ATP, the reactions were processed and resolved by SDS-PAGE; gel was dried and autoradiographed. (H) Noxa is phosphorylated on S13 in transiently transfected HEK293 cells. Cells transfected with Noxa-FLAG plasmids and point mutants were labeled for 6 hr with 0.25 mCi [32P]-orthophosphate 36 hr posttransfection in 6-well plates. FLAG-tagged proteins were immunoprecipitated from cell lysates; autoradiograph of [32P]-labeled Noxa is shown in the top panel, WB of total immunoprecipitated Noxa in the lower panel. (I) Phospho-S13-specic rabbit pAbs detect pS13 Noxa expression in transiently transfected 293 cells. The production and purication of polyclonal antibodies (rAK2) against a pS13-Noxa peptide is described in the Supplemental Experimental Procedures. Data show an IP/western of lysates from 293 cells transfected with WT Noxa-FLAG, or SA/SE mutants immunoblotted with the phospho-specic rAK2 pAb (top), then stripped and reblotted for total Noxa detection (Figure 1 linked to Table S1).
with its secondary structure and intracellular interactions. To eliminate any potential inuence of the tag on in vitro assays, we synthesized the protein de novo (Figure S1) and conrmed that phosphorylation of the full-length peptide was sensitive to concentrations of rosco inhibitory to Cdk5 (Figure 3C). We then tested the ability of lysates from Jurkat cells, pretreated with specic inhibitors of Cdk5, GSK-3 a/b, and MAP38K, to phosphorylate the peptide in a kinase assay (Figure 3D). These results again identied Cdk5 as the Noxa-modifying endogenous kinase. We then generated stable clonal populations of Jurkat cells knocked down for Cdk5 (Figure 3E). Extracts from knockdowns demonstrated a reduced ability to phosphorylate Noxa compared to parental and vector control extracts (Figure 3F), further establishing Cdk5 as the Noxa kinase. In this experiment,
pS13 Noxa was detected using the phosphospecic pAb, rAK2 (Table S1). Active Cdk5 is known to stably associate with its targets (Luo et al., 2005); data in Figure 3G show that endogenous Cdk5 could be immunoprecipitated from Jurkat cells using a Noxa antibody. Noxa Accelerates Glucose Deprivation-Induced Death in Leukemia Cell Lines To identify death pathways in the leukemia cell lines that might be regulated by Noxa, we generated stable Jurkat clones (Figure 4A) that either overexpressed Noxa or were knocked down for expression and determined their viability in response to DNA damage, hypoxia (data not shown), and glucose deprivation. In epithelial cells, Noxa is induced in a p53-dependent
Molecular Cell
Figure 2. Subcellular Localization of Phosphorylated Noxa

(A) HL-60 cells labeled with MTR were xed, permeabilized, and incubated with anti-pS13 Noxa mAb 2B3 (see the Supplemental Experimental Procedures) or isotype-matched Abs. Secondary Abs were conjugated to AF488. Cells were mounted on slides using Vectastain with DAPI and observed under a 603 objective on a Fluoview 1000 multiphoton confocal microscope. (B) Jurkat cells labeled with 2B3 mAb and AF488 conjugated secondary Abs and mounted with DAPI Vectastain. (C) pS13 Noxa is localized to the cytosolic compartment, but not to mitochondria. HL-60 cells were stained with DAPI, MTR, and 2B3 (AF488, green). A 20,000-event image le was obtained using the ImageStream MIFC. Five-channel images of a single cell are shown in the top panel. To limit analysis to intact single nonapoptotic cells (lower left), events with intermediate DAPI intensity and high brighteld aspect ratio (R1) were gated on, rather than events with low DAPI intensity (enucleate cells) and high DAPI intensity/low aspect ratio (multiple cells). Condensed, fragmented nuclei from apoptotic cells (R2) have high DAPI texture intensity (Spot Small Total) and small DAPI area values compared to the larger and less textured nuclear images of live cells (R3), as shown in the lower center panel. R3 region cell cycle prole is shown in the inset. To determine relative colocalization of 2B3 to mitochondria, the correlation between the bright details of the indicated image pair for each was measured using the Similarity Bright Detail feature (lower left). As two molecules colocalize, their similarity increases. Values greater than 1.5 indicate that the two probes reside in the same cellular compartment, while values greater than 2.5 indicate molecular colocalization (Figure 2 linked to Table S1).
manner following exposure to DNA damaging drugs and is required for the DNA damage response (Abedin et al., 2007; Alves et al., 2006; Ploner et al., 2008). In (p53 null) Jurkat cells, Noxa levels did not signicantly inuence the rate of response to CPT or etoposide (Figures 4B and 4C) but was essential for their response to glucose deprivation (Figure 4D). Noxa-overexpressing Jurkat clones mounted an accelerated response to 36 hr of glucose deprivation, while loss of Noxa imparted partial protection. Glucose deprivation also promoted apoptosis in K562 (CML) cells, albeit at a slower rate (Figure S3A) that was delayed in Noxa shRNA knockdowns (data not shown). Despite the robust apoptosis displayed in glucose-free (G-F) medium, Noxa overexpressing Jurkat clones N5 and N11 showed little to no induction in protein levels compared to cells grown in glucose-rich (G-R) medium (Figure 4E, left). Hindered immunoprecipitation with anti-Noxa Abs (Figure 4E, right) suggested that, in cells grown in G-R medium, Noxa was either complexed with other proteins or in an altered conformation that rendered the epitopes allowing for immunoprecipitation inaccessible. Indeed, these epitopes became exposed following glucose deprivation. This release of Noxa from sequestration could be reversed if cells were transferred back to G-R medium after 12 hr of deprivation (Figure S2A). Levels of Mcl-1L were also restored in the latter. These data suggested that, in the presence of glucose, Noxa was sequestered in a stable complex that inhibited its apoptotic function. A variety of cell lysis reagents,
including mild NETI- and CHAPS-based buffers, as well as harsher RIPA- (with 0.1% SDS) and TritonX-based buffers, failed to allow immunoprecipitation of Noxa in proliferating cells (data not shown). Gel exclusion chromatography of extracts from Jurkat cells grown in G-R medium showed Noxa consistently cofractionating with two high molecular weight complexes and detectable both as a free protein and as a component of larger complexes following glucose withdrawal (X.H.L. and A.K., unpublished data). Representative Superose 6 proles of Noxa from Jurkat cells grown in the presence and absence of glucose are shown in Figure S2B. Phosphorylation of Noxa by Cdk5 Controls Its Apoptotic Function and Is Regulated by Glucose We next determined whether S13 modication of Noxa regulated its function in a glucose-dependent manner. In Figure 5Aa and Figure S3B we show that Cdk5 from Jurkat cells phosphorylated Noxa in vitro only if it was derived from cells grown in G-R medium, indicating that its activity was glucose dependent. Data from the same experiment (Figure 5Ab) showed similar levels of immunoprecipitated Cdk5 in glucose-starved cells and controls. These data suggested that modication of Noxa by Cdk5 in the presence of glucose caused its sequestration and masked its apoptotic function. Consistent with previous observations, glucose deprivation did not upregulate endogenous Noxa but promoted its release and/or accessibility as
Molecular Cell
Figure 3. Human Noxa Is Phosphorylated by Cdk5, an Atypical Cyclin-Dependent Kinase

(A) In vitro kinase assays were carried out for 30 min in KB containing 3 mM cold ATP and 5 mCi [g32P]-ATP, and either 5 mg of histone H1 or rNoxa-FLAG as substrate, using rCdk1 and rCdk5 (Upstate). Radioactivity in terminated, processed reactions was measured by liquid scintillation counting (mean and SE, n = 3). (B) Kinase assays were carried out in the presence of Cdk1 and Cdk5, with rFLAG-Noxa or the SA mutant as substrates. Reactions were terminated, Noxa was immunoprecipitated with anti-FLAG antibodies and resolved by SDS-PAGE; gel was dried and autoradiographed. (C) rCdk5 phosphorylates de novo synthesized untagged full-length Noxa peptide (described in Figure S1 and the Supplemental Experimental Procedures). Kinase reactions included the inhibitor rosco (25 and 10 mM) in the two left lanes. (D) Cdk5 inhibition reduces in vitro phosphorylation of the Noxa peptide by Jurkat cell lysates. Jurkat cells were pretreated with various kinase inhibitors for 6 hr before harvesting. Lysates (10 mg) were incubated with 1 mg Noxa in KB + [32P]-ATP for 30 min at 30 C. Noxa was immunoprecipitated with the NP pAbs (Table S1), resolved by SDS-PAGE, transferred to membrane, and autoradiographed (top panel). The blot was then processed for total Noxa detection (lower panel). Left to right: untreated, 25 mM Rosco (Cdk5 inhibitor), 10 mM SB216763 (GSK-3a/b inhibitor), 10 mM SB203580 (MAP38K inhibitor). (E) WB showing Cdk5 levels in clonal lines of Jurkat knocked down for Cdk5 expression. (F) Cdk5 knockdowns show a reduced ability to phosphorylate Noxa. Jurkat parental, Cdk5 sh8, and vector control lysates (10 mg) were incubated with KB, Noxa (250 ng), and cold ATP. Phosphorylation was detected using the rAK2 pAbs (top panel). The membrane was later immunoblotted for total Noxa. The lower two panels show Cdk5 and Actin levels in the lysates.
indicated by the dramatic increase in immunoprecipitable protein. A similar observation was made in actively proliferating T cells (Figure 5B). Noxa had previously been shown to be transcriptionally induced in activated primary human T cells in G-R medium (Alves et al., 2006). Consistent with data from Jurkat cells, we observed that, following 2438 hr of glucose deprivation, Noxa was not further induced in these cells but was more accessible to immunoprecipitating Abs. If Noxa phosphorylation masks its apoptotic function, then a nonphosphorylatable S13A mutant should be more proapoptotic than WT. Our inability to generate stable Jurkat clones expressing SA-Noxa suggested that the phosphorylated S13 was protective. Furthermore, Jurkat and Molt-4 cells transiently transfected with IRES-EGFP vectors expressing SA-Noxa showed increased apoptosis in the absence of a stress stimulus and an accelerated response to glucose deprivation compared to cells expressing WT Noxa or the empty vector (Figure 5C, Figure S3C). Additionally, Jurkat cells knocked down for Cdk5 demonstrated increased sensitivity to glucose deprivation compared to controls (Figure 5D). Finally, Cdk5-specic inhibitor, rosco, promoted apoptosis and increased levels of immunoprecipitable Noxa in Jurkat, activated primary T, and HL-60 cells (Figures 5E and 5Fa, Figure S3D). The mAb 2B3 detected reduced levels of endogenous pS13Noxa in HL-60 cells within 6 hr of exposure to rosco (Figure S3E). In Figure 5Fa, 15 hr of rosco exposure, but not 15 hr of glucose deprivation, was adequate to promote both Noxa release and cell death in primary T cells. Taken together, these results indicate that phosphorylation of Noxa by Cdk5 leads to its sequestration in a survival complex effectively rendering it inactive as a proapoptotic protein. Cdk5 control of Noxa function appears to be tissue specic, since high Noxa levels are associated primarily with proliferating cells of hematopoietic lineage. Noxa is a known binding partner of Mcl-1L and promotes its degradation during apoptosis. However, a number of studies have suggested that the two proteins may interact even in proliferating cells. Alves et al. (2006) showed that Noxa coprecipitated with Mcl-1L in activated T cells. We were also able to observe both the induction of Mcl-1L in anti-CD3/CD28-activated T cells as well as the coimmunoprecipitation of Noxa with antiMcl-1 Abs (Figure 5Fb, top two panels). Interestingly, only Noxa that coimmunoprecipitated with Mcl-1L in proliferating cells was phosphorylated. Figure 5Fb (bottom two panels) shows that the protein brought down from rosco-treated cells by Noxa Abs was, as predicted, not phosphorylated. These data suggest that Mcl-1L is also a component of the complex that harbors phosphorylated Noxa, although it is unclear whether the two proteins directly interact with each other. Binding studies with in vitro-translated Mcl-1L showed no interaction with Noxa phosphorylated in vitro (Figure S3F). Paradoxically, a Noxa BH3 mutant (DA) could be immunoprecipitated from stably transfected cells (Figure S3G), suggesting that the BH3 domain is required for Noxa to be sequestered.
(G) Noxa and Cdk5 interact endogenously. IP/western showing that endogenous Cdk5 is immunoprecipitated with anti-Noxa IP antibody from lysates of untreated Jurkat parental and vector controls (Figure 3 linked to Figure S1 and Table S1).
Molecular Cell
Figure 4. Noxa Is Required for the Response of Hematopoietic Cells to Glucose Deprivation
(A) Western blot of cell lysates from transfected, cloned Jurkat cell lines overexpressing Noxa (Noxa 5 and 11) or Noxa shRNA (Noxa sh 1, 4, 6) and scrambled control (scr6). Selected cell lines, including parentals (PAR), were treated for 6 hr with either 2 mM CPT (B) or 100 mM etoposide (C), or grown in RPMI with 25 mM glucose or no glucose for 36 hr. (D) Viability (mean and SE, n = 3) was determined by ow cytometric analysis of Annexin (Ann) V and PI uptake (*p < 0.01). (E) Noxa is posttranslationally regulated following glucose starvation. Noxa overexpressing N5 and N11 cells were cultured in high glucose (+) or glucose-free () RPMI for 24 hr. The left panel shows a WB of lysates (50 mg) immunoblotted for Noxa expression, and an IP/western of Noxa immunoprecipitated from 200 mg of the same lysate is shown on the right (Figure 4 linked to Figures S2A, S2B, and S3A).
Glucose Metabolism in Noxa-Overexpressing Jurkat Cells Most cancer cells preferentially break glucose down via the relatively inefcient aerobic glycolytic pathway, which causes them to consume more glucose and produce more lactate than normal cells (Vander Heiden et al., 2009; Warburg, 1956). Jurkat parental and Noxa-overexpressing cells were tested in culture for their cell accumulation, survival, glucose consumption, and lactate production properties. Although all three lines were >95% viable at initial seeding, cell viability in culture without passage became compromised after 4 days in Noxa overexpressers (Figure 6A). Both N5 and N11 cells exhausted their media of glucose more rapidly than the controls, which may account for the lower viability (Figure 6C). The levels of secreted lactate were also higher in N5 and N11 culture media (Figure 6B). Finally, glucose readdition almost completely abrogated the decline in viability of N5 cells (Figure 6D). This suggested that glucose was being metabolized at a greater rate in Noxa-overexpressing cells than in controls. However, the glycolytic rate as measured by the conversion of [3H] glucose to [3H] water per hour was signicantly lower in the N5, N11 (p < 0.0003), and the SE-expressing lines, compared to controls (Figure 6E). The DA cells behaved similar to controls (data not shown). Thus, although overexpression of Noxa was increasing the dependence on glucose in T-ALL cells and suggesting a stimulatory role for Noxa in glucose metabolism, not all the glucose was being converted to H2O via aerobic glycol-
ysis. We also determined the rate of uptake of [3H]-2-deoxyglucose (2-DG), a common indicator of glucose consumption, and found, contrary to prediction, that uptake was inhibited in the N5, N11 (p < 0.0001), and SE lines (Figure 6G).
Noxa Overexpression Imparts Resistance to 2-Deoxyglucose Although 2-DG can be taken up by cells, it cannot be metabolized beyond its conversion to 2-DG-6-phosphate by hexokinase (HK). Jurkat cells and indicated clonal lines were cultured in G-F medium with or without 2-DG for 48 hr (Figure 7A). While controls, DA cells and Noxa knockdowns exhibited 5%10% higher viability in 2-DG than in G-F medium, Noxa overexpressers, N5 and N11, which respond poorly to the absence of glucose (Figures 4C and 7A), survived signicantly better in the presence of 2-DG than did control cells. The SE-expressing line also demonstrated higher tolerance to medium with 2-DG than without; however, the generally poor response of SE cells to glucose deprivation suggested that regulatory phosphorylation/dephosphorylation at S13 may contribute to Noxa function. Although 2-DG cannot be utilized in glycolysis, the compound can be catabolized via the pentose phosphate pathway (PPP) under certain conditions (Zabos et al., 1978). Taken together, data in Figures 6 and 7 point to a role for Noxa in diverting glucose through the pentose phosphate shunt and will be discussed in the following section.
Molecular Cell
Figure 5. Glucose Deprivation and Cdk5 Inhibition Lead to Noxa Dephosphorylation and Activation in Jurkat and Activated Primary T Cells
(A) Cdk5 was immunoprecipitated from lysates of Jurkat cells grown in G-R or G-F medium for 36 hr and used in an in vitro kinase assay with Noxa peptide as substrate and a WB of total Noxa (upper panel set). Lower panel set: an IP/western (top) and WB of Cdk5 (second from top) from control and glucose-deprived Jurkat lysates (50 mg). The lower two panels are IP/westerns and WBs of endogenous Noxa with 200 and 50 mg of lysate, respectively, from the same experiment. (B) A mixture of Cd4+ and Cd8+ human T cells were left unstimulated or were stimulated with anti-CD3 and CD28 Abs for 24 hr and then cultured in G-F medium for the indicated times. An IP/western of Noxa from 200 mg cell lysate protein and WBs of total Noxa and Actin levels are shown. Cell viability at the time of harvesting is indicated under each lane. (C) Jurkat cells electroporated with pIRES2-EGFP vector alone, or pIRES2-EGFP plasmids expressing Noxa WT or Noxa SA, were Ficoll puried within 12 hr to eliminate dead cells and cultured in G-R or G-F medium. Viability of the GFP positive population (mean and SE, n = 3) was measured after 16 hr by ow cytometric analysis of Ann V-PE-Cy5 uptake (*p < 0.03). (D) Loss of Cdk5 increases sensitivity of Jurkat cells to glucose deprivation. Cells were cultured in G-F medium and viability was assessed by ow cytometric analysis of Ann V-PE-Cy5 uptake (*p = 0.02, **p = 0.01). (E and F) Cdk5 inhibitor, rosco, increases accessibility of Noxa to IP in Jurkat and primary T cells. Jurkat cells (E) were cultured in the presence of 25 mM rosco or DMSO for 20 hr and lysates processed as described earlier. Viability of treated cells was assessed prior to harvesting and conrmed in WB of cleaved caspase-3. Stimulated Cd4+/Cd8+ human T cells (Supplemental Experimental Procedures) were cultured either in G-F medium or G-R medium in the presence of 25 mM rosco for 15 hr (Fa). Viability was assessed prior to harvesting; 200 mg of cell protein were used for the Noxa IP/western and 40 mg for the WB. Fb. Mcl-1L is complexed with phospho-Noxa in dividing T cells. Lysates from the T cell activation experiment in Fa were immunoprecipitated either with anti-Mcl-1L (Fb) or antiNoxa pAbs and immunoblotted for Noxa or phospho-Noxa. Note: top panel from Fa is shown again in Fb (Figure 5 linked to Figures S3BS3G, Table S1).
DISCUSSION Noxa was originally identied as a phorbol 12-myristate-13acetate inducible protein (PMAIP) in a T cell leukemia line (Hijikata et al., 1990). Although it is known to be activated primarily via transcriptional mechanisms (reviewed in Ploner et al., 2008), the protein is stably expressed in proliferating primary and malignant human hematopoietic cells (Alves et al., 2006; and Figure 1), suggesting that it is also subject to posttranslational regulatory control. Here we report that constitutively expressed hNoxa is phosphorylated on S13 by the kinase Cdk5 in dividing hematopoietic cells in the presence of glucose. This phosphorylation serves two functions: it suppresses Noxas proapoptotic function by targeting it to stable protein complexes, and it regulates a novel role of the protein in glucose
metabolism. Dephosphorylation of S13 in response to glucose stress acts as the switch that activates its proapoptotic function. Phosphorylation, as a mechanism for keeping BH3-only proteins in check, has been observed previously (Datta et al., 2000), but the 54 aa hNoxa would be the smallest known BH3only protein to be thus regulated. Phosphospecic antibodies localized pS13-Noxa to the cytoplasm in dividing cells. The punctate nature of the uorescence suggested that Noxa was a component of large particles. The inability of two pAbs to immunoprecipitate Noxa from growing cells may be due to a conformational change in the phosphoprotein or masking of the epitope by other proteins. We have recently conrmed that pS13-Noxa copuries with two large multiprotein (survival) complexes in cells cultured in G-R medium and is either free or detected in a larger multiprotein (apoptotic)
Molecular Cell
Figure 6. Noxa Regulates Glucose Metabolism

Cells were seeded in triplicate at 5 3 104/mL and >95% viability (values shown are the mean with SE, n = 3). Cell counts and automated trypan blue viability measurements were carried out using a Vi-CELL Series Analyzer. (A) Viable cell counts were carried out over 7 days in culture without passage. (B and C) Lactate and glucose in the culture media (normalized to cell numbers) were quantied using colorimetric reaction kits (BioAssay Systems). (D) Cultures of Par and N5 cells were either supplemented or not with 5 mM glucose after 4 days. (E) Noxa overexpressers exhibit lower glycolytic rates than control cells. Glycolysis was measured by monitoring the conversion of 5-[3H]-glucose to [3H]2O, as described in the Experimental Procedures. The rates were determined in pmoles [3H]2O/h/106 cells, and values were plotted as a fraction of the rate-determined Jurkat Par cells (*p < 0.0003). (F) Western blots of lysates from stable lines used in experiments shown in (E) and (G) and in Figure 7A. Con, lentiviral control; sh6 and sh7, clonal lines infected with lentiviral constructs of Noxa shRNA; SE and DA, clones expressing S13 and BH3 domain point mutants, respectively. (G) 2-deoxy-D-[3H]glucose uptake is inhibited in N5 and N11 lines. Incorporated [3H] 2-DG (dpm/hr/1 3 106cells) is shown in the bar graph (*p < 0.0001).
complex following glucose deprivation (Figure S2B). Studies to characterize these complexes and identify their components are currently underway. Cdk5, an atypical cyclin-dependent kinase, was identied as the Noxa kinase. Although Cdk5 has been extensively investigated for its role in neurodegenerative disease (Dhavan and Tsai, 2001), the protein is ubiquitous in its expression and active in human myeloid and lymphoid cells (Rosales and Lee, 2006; and Figure 4E, Figure S3A). Cdk5 phosphorylates hNoxa in a glucose-dependent manner. The kinase was shown to have a glucose-sensing function in pancreatic b cells and adipocytes (Lalioti et al., 2009; Ubeda et al., 2004) but had not previously been associated with a glucose-dependent prosurvival role in hematological malignancies. The stable expression of Noxa in proliferating hematopoietic cells suggests that Cdk5 control of Noxa function may be tissue specic. Most cancer cells use the low-efciency aerobic glycolytic pathway, which requires them to consume more glucose than normal cells (Warburg, 1956). Preferential recruitment of a special function of Noxa (and possibly Cdk5) to the metabolic machinery may help manage the increased glucose requirement of actively proliferating hematopoietic cells and cancers. This could partially
explain the increased glucose consumption observed in Noxaoverexpressing Jurkat cells. Noxa interacts almost exclusively with prosurvival Bcl-2 protein Mcl-1L, inactivating its survival function and targeting it for degradation (Chen et al., 2005; Czabotar et al., 2007). As a sensitizing BH3-only protein, it is attributed with weak proapoptotic potential. However, the toxicity of the transiently transfected SA mutant as well as its ability to accelerate apoptosis following glucose stress underscores the protective role of the S13 modication and suggests that Noxa may have greater proapoptotic potential than previously recognized. The high toxicity of unphosphorylated Noxa, coupled with its constitutive expression in hematological cancers, makes it an attractive therapeutic target, and intervention strategies aimed at dephosphorylating Noxa may offer a viable alternative to the BH3 mimetics-based approach. Noxa/Mcl-1L interactions play a major role in glucose deprivation-induced death in hematopoietic cells, but they may also play major roles in glucose-dependent survival. It is becoming increasingly clear that the two interacting partners of the Bcl-2 family are coordinately upregulated and remain closely associated during periods of increased metabolic and biosynthetic
Molecular Cell
Figure 7. Noxa Diverts Glucose to the Pentose Phosphate Pathway

(A) Jurkat controls and selected clonal lines (see 6F) were cultured in glucosefree medium with or without 2-DG (50 mM) for 48 hr. Viability was determined by Ann V/PI uptake (mean and SE, n = 3). *p < 0.005, **p < 0.0005. (B) A simplied model for the dual function of Noxa in human hematopoietic cells. The BH3-only protein toggles between a metabolic and an apoptotic role in a glucose-dependent manner. In the presence of glucose, Noxa is phosphorylated on S13 by Cdk5 and sequestered in a cytosolic complex that imparts a metabolic (growth and survival) function to the protein.
activity in actively proliferating cells (Alves et al., 2006; and Figure 5F). We show that the Mcl-1L associates with phosphorylated Noxa in activated T cells. However, evidence that it interacts directly with phospho-Noxa is, as yet, inconclusive (Figure 5F, Figures S3F and S3G). Our studies point to a possible role for Noxa in regulating glucose turnover via the PPP. The role of the PPP is to produce NADPH as reducing equivalents and pentoses as essential parts of nucleotides. Glucose-6-phosphate (G-6-P) generated through HK activity is either processed via glycolysis to pyruvate and thence to lactate, or oxidized by G-6-P dehydrogenase and eventually converted to ribose-5-phosphate (R-5-P) via the PPP. The increased glucose consumption/lactate production by the N5 and N11 lines during growth in G-R medium (Figures 6A
6D) initially suggested increased aerobic glycolysis, but glycolytic rate measurements showing low rates of glucose conversion to water (Figure 6E) indicated otherwise, suggesting that the high levels of lactate in the media were likely to be the result of glutaminolysis or glutamine breakdown (Medina and Nunez de Castro, 1990). The resistance of Noxa overexpressers to 2-DG (Figure 7A) offered an important clue. Although HK can phosphorylate 2-DG, the glucose analog cannot be further metabolized in the glycolysis pathway. It can, however, be turned over via the PPP under certain conditions (Zabos et al., 1978). Tumorpromoting phorbol esters have been shown to impart resistance to 2-DG in hematopoietic cells. The Zabos et al. study had demonstrated that PMA (or TPA) enhanced the conversion of 2-DG-6-P into an analog of R-5-P with release of carbon1 (C-1) as CO2, but did not identify a mechanism. Stabilized hypoxia-inducible factor 1 (HIF-1) also confers resistance to 2-DG under hypoxic conditions (Maher et al., 2007). In the latter study, the authors had implicated increased HK levels as the underlying cause. Although the mechanism still remains to be determined, Noxa may be the mediator of both PMA- and HIF1-induced resistance to 2-DG. The BH3-only protein had been initially identied as an adult T cell leukemia-derived PMAresponsive gene (Hijikata et al., 1990). It is also consistently upregulated in hypoxic cells as one of the targets of the stabilized transcription factor, HIF-1a (Kim et al., 2004). If Noxa is, indeed, enhancing the turnover of 2-DG-6-P via the PPP, the inconsistency between the higher glucose consumption and apparent inhibition of [3H]-2-DG uptake in N5 and N11 cells (Figures 6C, 6D, and 6G) may be simply explained by the fact that 90% of the [3H] in [3H]-2-DG is attached to C-1, which is lost during the second step of the oxidative phase of PPP (Zabos et al., 1978). As a BH3-only protein, Noxa was believed to be a canonical tumor suppressor and had not previously been directly associated with a growth function. One other BH3-only protein, BAD, has been shown to regulate glucose metabolism (Danial et al., 2008). Phosphorylation of S155 in its BH3 domain controls this function by directly regulating BADs binding properties. Modied S13 on Noxa is not located within the BH3 domain, but the domain seems to be essential for inclusion of Noxa in a functional protein complex. Moreover, the metabolic function of BAD involves mitochondria and glycolysis in insulinsecreting pancreatic and liver cells, rather than cancer cells. Our studies suggest that Noxa plays dual roles in proliferating hematopoietic cells and cancersit enhances glucose turnover via the PPP in the presence of the sugar and promotes apoptosis in its absence, engaging Mcl-1L in both these roles. Studies directed at understanding the mechanism underlying Noxas contribution to the pentose phosphate shunt are underway.
EXPERIMENTAL PROCEDURES Cell Lines, Chemicals, Antibodies, Plasmid Constructs, RNA Interference, and Expression Cell lines, chemicals, antibodies, plasmid constructs, RNA interference, and expression are described in the Supplemental Information.
Molecular Cell
Western Blotting and Immunoprecipitation For western blotting, cells were lysed in RIPA buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% v/v sodium deoxycholate, 1% v/v Nonidet P-40, 0.1% SDS) supplemented with protease and phosphatase inhibitor cocktails. Lysates were resolved by SDS-PAGE and transferred to nitrocellulose membrane. Membrane blocking, washing, Ab incubations, and chemiluminescence reactions carried out using the ECL Plus kit (Amersham) or Super Signal West Femto kit (Pierce Biotech). Blots were stripped for reuse by washing for 30 min in TBS-T buffer (pH 2.53.0). For immunoprecipitation, cells were lysed in NETI buffer (100 mM NaCl, 1 mM EDTA, 20 mM Tris [pH 8.0], 0.5% IGEPAL) supplemented with protease and phosphatase inhibitors. Cell lysate was incubated with 15 mg/mL of antibody, and complexes were captured with protein G agarose beads (Invitrogen); immunoprecipitates were resolved by SDSPAGE and processed for detection as described above. Confocal Microscopy, Flow, and Imaging Flow Cytometry Cells were xed in 3.7% paraformaldehyde at 37 C for 30 min and permeabilized with ice-cold methanol:acetone (1:1) at 20 C for 10 min. For mitochondria labeling, live cells were incubated with 20 nM MTR for 15 min prior to xation. After blocking in 1% BSA, permeabilized cells were incubated with primary Abs for 1 hr, washed and incubated with uorescently tagged secondary Abs for 1 hr, washed, and either mounted on coverslips with Vectastain containing DAPI or resuspended in PBS for ow cytometric analyses. Cells were visualized using FluoView 1000 Olympus multiphoton inverted or upright microscopes and analyzed with Photoshop and ImageJ software. For IFC, cells were resuspended at 5 3 107 cells/mL in PBS plus DAPI prior to analysis on the ImageStream. Cell Viability Assays Exponentially growing hematopoietic cells were cultured in fresh medium before exposure to CPT (2 mM), etoposide (100 mM), glucose-deprived medium, or hypoxia (2% O2) for indicated time periods. Cell viability was measured by ow cytometric analysis of cells stained with Annexin V (Ann V)-FITC and propidium iodide (PI) and analyzed using FlowJo Software (Tree Star, Inc.). Ann V-PE-CY5 was used to detect apoptosis of EGFP-expressing cells. Automated viable cell counts were also carried out by trypan blue dye uptake using a ViCell counter (Beckman Coulter). Kinase Assays In vitro kinase assays with recombinant Cdk1 and Cdk5 (Upstate) were carried out according to the manufacturers instructions. For in vitro assays of endogenous kinases, lysate protein (10 mg) harvested in Tonks lysis buffer (50 mM Tris-HCl, 10 mM EDTA, 0.5% IGEPAL, 25 mM b-glycerophosphate), supplemented with protease and phosphatase inhibitors, was incubated for 30 min at 30 C with kinase buffer (KB) (50 mM Tris-HCl, 10 mM MgCl2, 1 mM DTT, 10 mM b -glycerophosphate), 10 mM cold ATP, 5 mCi [g-32P]-ATP, and 250 ng1 mg Noxa full-length peptide per reaction. Assays were attenuated with either 2.5 mM EDTA and used for immunoprecipitation, or with loading buffer for direct SDS-PAGE resolution. For cold kinase reactions the concentration of ATP was 0.1 mM. For in vitro assays of endogenous Cdk5, the kinase was captured with protein G agarose beads from precleared lysates using an anti-Cdk5 antibody, washed in NETI and KB, and resuspended in KB before being added to the reaction mix. Measurement of Glycolysis Glycolysis was measured by monitoring the conversion of 5-[3H]-glucose to [3H]2O, as described previously (Liang et al., 1997; Vander Heiden et al., 2001). Briey, 1 3 106 cells were washed and incubated in G-F medium for 30 min at 37 C. Cells were then pelleted, resuspended in 0.5 ml of Krebs buffer containing 10 mM glucose, and spiked with 10 mCi of 5-[3H]-glucose. Following incubation for 1 hr at 37 C, triplicate 50 ml aliquots were transferred to uncapped thin-walled PCR tubes containing 50 ml of 0.2 N HCl, and each tube was transferred to a scintillation vial containing 0.5 ml of H2O such that the liquids in the vial and the tube did not mix. Vials were sealed and diffusion was allowed to occur for 36 hr. The rate of glycolysis was calculated as described previously (Ashcroft et al., 1972) after determining the total amount of [3H]2O in
each sample, taking into account appropriate 5-[3H]-glucose- and [3H]2Oonly controls. 2-Deoxy-D-[3H]Glucose Uptake Measurement Jurkat cells were washed once in G-F RPMI and maintained for 1 hr in the G-F medium containing 160 mM 2-deoxy-D-[3H]glucose (0.5 mCi for 1 3 106 cells) (Perkin Elmer). The reaction was stopped by addition of one volume of an icecold phloretin solution (0.3 mM ICN in PBS). Cells were then washed once with ice-cold PBS, and the 2-deoxy-D-[3H]glucose incorporated by cells was measured using a liquid scintillation counter. SUPPLEMENTAL INFORMATION Supplemental Information includes three gures, one table, Supplemental Experimental Procedures, and Supplemental References and can be found with this article at doi:10.1016/j.molcel.2010.11.035. ACKNOWLEDGMENTS The authors thank Dr. Jeffrey Miller as well as the Tissue Procurement Facility in the Masonic Cancer Center at the University of Minnesota for providing deidentied patient leukemia cells, the Flow Cytometry core facilities for help with data collection and analysis, John Oja from the Biomedical Image Processing Laboratory (BIPL) for help with confocal and uorescence microscopy, and Tad George and AMNIS Corporation for help with Imaging Flow Cytometry data collection and analysis. We also thank Dr. Kristin Hogquist for helpful advice on the primary T cell experiments, Dr. M.J. Abedin for assistance with protein purication, and Christopher Valley and Lamya Yamani for helpful discussions. X.H.L. and O.A.O. were supported by National Institutes of Health Grants T32-CA009138 (to X.H.L. and O.A.O.) and F31-AI084524 (to O.A.O.). C.B.K. was supported in part by National Institutes of Health Grant GM27906. This research was primarily funded through grants from the Elsa U. Pardee Foundation, the Leukemia Research Foundation, and the Minnesota Medical Foundation (to A.K.). Received: April 19, 2010 Revised: August 20, 2010 Accepted: October 5, 2010 Published: December 9, 2010 REFERENCES Abedin, M.J., Wang, D., McDonnell, M.A., Lehmann, U., and Kelekar, A. (2007). Autophagy delays apoptotic death in breast cancer cells following DNA damage. Cell Death Differ. 14, 500510. Alves, N.L., Derks, I.A., Berk, E., Spijker, R., van Lier, R.A., and Eldering, E. (2006). The Noxa/Mcl-1 axis regulates susceptibility to apoptosis under glucose limitation in dividing T cells. Immunity 24, 703716. Alves, N.L., van Lier, R.A., and Eldering, E. (2007). Withdrawal symptoms on display: Bcl-2 members under investigation. Trends Immunol. 28, 2632. Ashcroft, S.J., Weerasinghe, L.C., Bassett, J.M., and Randle, P.J. (1972). The pentose cycle and insulin release in mouse pancreatic islets. Biochem. J. 126, 525532. Blom, N., Gammeltoft, S., and Brunak, S. (1999). Sequence and structurebased prediction of eukaryotic protein phosphorylation sites. J. Mol. Biol. 294, 13511362. Chen, L., Willis, S.N., Wei, A., Smith, B.J., Fletcher, J.I., Hinds, M.G., Colman, P.M., Day, C.L., Adams, J.M., and Huang, D.C. (2005). Differential targeting of prosurvival Bcl-2 proteins by their BH3-only ligands allows complementary apoptotic function. Mol. Cell 17, 393403. Czabotar, P.E., Lee, E.F., van Delft, M.F., Day, C.L., Smith, B.J., Huang, D.C., Fairlie, W.D., Hinds, M.G., and Colman, P.M. (2007). Structural insights into the degradation of Mcl-1 induced by BH3 domains. Proc. Natl. Acad. Sci. USA 104, 62176222.
Molecular Cell
Danial, N.N., Walensky, L.D., Zhang, C.Y., Choi, C.S., Fisher, J.K., Molina, A.J., Datta, S.R., Pitter, K.L., Bird, G.H., Wikstrom, J.D., et al. (2008). Dual role of proapoptotic BAD in insulin secretion and beta cell survival. Nat. Med. 14, 144153. Datta, S.R., Katsov, A., Hu, L., Petros, A., Fesik, S.W., Yaffe, M.B., and Greenberg, M.E. (2000). 14-3-3 proteins and survival kinases cooperate to inactivate BAD by BH3 domain phosphorylation. Mol. Cell 6, 4151. Dhavan, R., and Tsai, L.H. (2001). A decade of CDK5. Nat. Rev. Mol. Cell Biol. 2, 749759. Erster, S., Mihara, M., Kim, R.H., Petrenko, O., and Moll, U.M. (2004). In vivo mitochondrial p53 translocation triggers a rapid rst wave of cell death in response to DNA damage that can precede p53 target gene activation. Mol. Cell. Biol. 24, 67286741. Goodyear, S., and Sharma, M.C. (2007). Roscovitine regulates invasive breast cancer cell (MDA-MB231) proliferation and survival through cell cycle regulatory protein cdk5. Exp. Mol. Pathol. 82, 2532. Hallaert, D.Y., Spijker, R., Jak, M., Derks, I.A., Alves, N.L., Wensveen, F.M., de Boer, J.P., de Jong, D., Green, S.R., van Oers, M.H., and Eldering, E. (2007). Crosstalk among Bcl-2 family members in B-CLL: seliciclib acts via the Mcl1/Noxa axis and gradual exhaustion of Bcl-2 protection. Cell Death Differ. 14, 19581967. Hershko, T., and Ginsberg, D. (2004). Up-regulation of Bcl-2 homology 3 (BH3)-only proteins by E2F1 mediates apoptosis. J. Biol. Chem. 279, 8627 8634. Hijikata, M., Kato, N., Sato, T., Kagami, Y., and Shimotohno, K. (1990). Molecular cloning and characterization of a cDNA for a novel phorbol-12myristate-13-acetate-responsive gene that is highly expressed in an adult Tcell leukemia cell line. J. Virol. 64, 46324639. Kim, J.Y., Ahn, H.J., Ryu, J.H., Suk, K., and Park, J.H. (2004). BH3-only protein Noxa is a mediator of hypoxic cell death induced by hypoxia-inducible factor 1alpha. J. Exp. Med. 199, 113124. Lalioti, V., Muruais, G., Dinarina, A., van Damme, J., Vandekerckhove, J., and Sandoval, I.V. (2009). The atypical kinase Cdk5 is activated by insulin, regulates the association between GLUT4 and E-Syt1, and modulates glucose transport in 3T3-L1 adipocytes. Proc. Natl. Acad. Sci. USA 106, 42494253. Liang, Y., Buettger, C., Berner, D.K., and Matschinsky, F.M. (1997). Chronic effect of fatty acids on insulin release is not through the alteration of glucose metabolism in a pancreatic beta-cell line (beta HC9). Diabetologia 40, 1018 1027. Luo, S., Vacher, C., Davies, J.E., and Rubinsztein, D.C. (2005). Cdk5 phosphorylation of huntingtin reduces its cleavage by caspases: implications for mutant huntingtin toxicity. J. Cell Biol. 169, 647656. Maher, J.C., Wangpaichitr, M., Savaraj, N., Kurtoglu, M., and Lampidis, T.J. (2007). Hypoxia-inducible factor-1 confers resistance to the glycolytic inhibitor 2-deoxy-D-glucose. Mol. Cancer Ther. 6, 732741. Mapelli, M., Massimiliano, L., Crovace, C., Seeliger, M.A., Tsai, L.H., Meijer, L., and Musacchio, A. (2005). Mechanism of CDK5/p25 binding by CDK inhibitors. J. Med. Chem. 48, 671679. Maurer, U., Charvet, C., Wagman, A.S., Dejardin, E., and Green, D.R. (2006). Glycogen synthase kinase-3 regulates mitochondrial outer membrane permeabilization and apoptosis by destabilization of MCL-1. Mol. Cell 21, 749760. Medina, M.A., and Nunez de Castro, I. (1990). Glutaminolysis and glycolysis interactions in proliferant cells. Int. J. Biochem. 22, 681683. Muruais, G., Lalioti, V., and Sandoval, I.V. (2009). The Cdk5 inhibitor roscovitine strongly inhibits glucose uptake in 3T3-L1 adipocytes without altering GLUT4 translocation from internal pools to the cell surface. J. Cell. Physiol. 220, 238244. Nikiforov, M.A., Riblett, M., Tang, W.H., Gratchouck, V., Zhuang, D., Fernandez, Y., Verhaegen, M., Varambally, S., Chinnaiyan, A.M., Jakubowiak, A.J., and
Soengas, M.S. (2007). Tumor cell-selective regulation of NOXA by c-MYC in response to proteasome inhibition. Proc. Natl. Acad. Sci. USA 104, 19488 19493. Oda, E., Ohki, R., Murasawa, H., Nemoto, J., Shibue, T., Yamashita, T., Tokino, T., Taniguchi, T., and Tanaka, N. (2000). Noxa, a BH3-only member of the Bcl-2 family and candidate mediator of p53-induced apoptosis. Science 288, 1053 1058. Opferman, J.T., Letai, A., Beard, C., Sorcinelli, M.D., Ong, C.C., and Korsmeyer, S.J. (2003). Development and maintenance of B and T lymphocytes requires antiapoptotic MCL-1. Nature 426, 671676. Opferman, J.T., Iwasaki, H., Ong, C.C., Suh, H., Mizuno, S., Akashi, K., and Korsmeyer, S.J. (2005). Obligate role of anti-apoptotic MCL-1 in the survival of hematopoietic stem cells. Science 307, 11011104. Perez-Galan, P., Roue, G., Villamor, N., Montserrat, E., Campo, E., and Colomer, D. (2006). The proteasome inhibitor bortezomib induces apoptosis in mantle-cell lymphoma through generation of ROS and Noxa activation independent of p53 status. Blood 107, 257264. Ploner, C., Koer, R., and Villunger, A. (2008). Noxa: at the tip of the balance between life and death. Oncogene 27 (Suppl 1 ), S84S92. Rosales, J.L., and Lee, K.Y. (2006). Extraneuronal roles of cyclin-dependent kinase 5. Bioessays 28, 10231034. Shibue, T., Takeda, K., Oda, E., Tanaka, H., Murasawa, H., Takaoka, A., Morishita, Y., Akira, S., Taniguchi, T., and Tanaka, N. (2003). Integral role of Noxa in p53-mediated apoptotic response. Genes Dev. 17, 22332238. Ubeda, M., Kemp, D.M., and Habener, J.F. (2004). Glucose-induced expression of the cyclin-dependent protein kinase 5 activator p35 involved in Alzheimers disease regulates insulin gene transcription in pancreatic betacells. Endocrinology 145, 30233031. Vander Heiden, M.G., Plas, D.R., Rathmell, J.C., Fox, C.J., Harris, M.H., and Thompson, C.B. (2001). Growth factors can inuence cell growth and survival through effects on glucose metabolism. Mol. Cell. Biol. 21, 58995912. Vander Heiden, M.G., Cantley, L.C., and Thompson, C.B. (2009). Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 10291033. Villunger, A., Michalak, E.M., Coultas, L., Mullauer, F., Bock, G., Ausserlechner, M.J., Adams, J.M., and Strasser, A. (2003). p53- and druginduced apoptotic responses mediated by BH3-only proteins puma and noxa. Science 302, 10361038. Wang, Q., Mora-Jensen, H., Weniger, M.A., Perez-Galan, P., Wolford, C., Hai, T., Ron, D., Chen, W., Trenkle, W., Wiestner, A., and Ye, Y. (2009). ERAD inhibitors integrate ER stress with an epigenetic mechanism to activate BH3-only protein NOXA in cancer cells. Proc. Natl. Acad. Sci. USA 106, 22002205. Warburg, O. (1956). On the origin of cancer cells. Science 123, 309314. Warr, M.R., and Shore, G.C. (2008). Unique biology of Mcl-1: therapeutic opportunities in cancer. Curr. Mol. Med. 8, 138147. Willis, S.N., Chen, L., Dewson, G., Wei, A., Naik, E., Fletcher, J.I., Adams, J.M., and Huang, D.C. (2005). Proapoptotic Bak is sequestered by Mcl-1 and BclxL, but not Bcl-2, until displaced by BH3-only proteins. Genes Dev. 19, 12941305. Zabos, P., Kyner, D., Mendelsohn, N., Schreiber, C., Waxman, S., Christman, J., and Acs, G. (1978). Catabolism of 2-deoxyglucose by phagocytic leukocytes in the presence of 12-O-tetradecanoyl phorbol-13-acetate. Proc. Natl. Acad. Sci. USA 75, 54225426. Zhao, Y., Altman, B.J., Coloff, J.L., Herman, C.E., Jacobs, S.R., Wieman, H.L., Wofford, J.A., Dimascio, L.N., Ilkayeva, O., Kelekar, A., et al. (2007). Glycogen synthase kinase 3alpha and 3beta mediate a glucose-sensitive antiapoptotic signaling pathway to stabilize Mcl-1. Mol. Cell. Biol. 27, 43284339.
Short Article
Uncoupling of Sister Replisomes during Eukaryotic DNA Replication
Hasan Yardimci,1 Anna B. Loveland,1 Satoshi Habuchi,2 Antoine M. van Oijen,1,3,4,* and Johannes C. Walter1,3,*
Molecular Cell
of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02115, USA of Organic and Polymeric Materials, Tokyo Institute of Technology, O-okayama 2-12-1, Meguro-ku, Tokyo 152-8552, Japan 3These authors contributed equally to this work 4Present address: The Zernike Institute for Advanced Materials, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands *Correspondence: a.m.van.oijen@rug.nl (A.M.v.O.), johannes_walter@hms.harvard.edu (J.C.W.) DOI 10.1016/j.molcel.2010.11.027
2Department
1Department
SUMMARY
The duplication of eukaryotic genomes involves the replication of DNA from multiple origins of replication. In S phase, two sister replisomes assemble at each active origin, and they replicate DNA in opposite directions. Little is known about the functional relationship between sister replisomes. Some data imply that they travel away from one another and thus function independently. Alternatively, sister replisomes may form a stationary, functional unit that draws parental DNA toward itself. If this double replisome model is correct, a constrained DNA molecule should not undergo replication. To test this prediction, lambda DNA was stretched and immobilized at both ends within a microuidic ow cell. Upon exposure to Xenopus egg extracts, this DNA underwent extensive replication by a single pair of diverging replisomes. The data show that there is no obligatory coupling between sister replisomes and, together with other studies, imply that genome duplication involves autonomously functioning replisomes.
INTRODUCTION The spatial and functional relationship between the two sister replisomes (Figure 1Ai) that emanate from each origin of DNA replication is not understood (Bochman and Schwacha, 2009; Takahashi et al., 2005). In one scenario, sister replisomes move apart after initiation and function independently (Figure 1Aii). In other models, sister replisomes must remain physically coupled after origin ring to allow unwinding by the replicative helicase (Falaschi, 2000; Sclafani et al., 2004; Weisshart et al., 1999; Wessel et al., 1992) (Figure 1Aiii). In bacteria, highresolution imaging demonstrated that there is no physical interaction between sister replisomes (Reyes-Lamothe et al., 2008). In archaea and eukaryotes, the picture is less clear. Live-cell imaging in Saccharomyces cerevisiae (Kitamura et al., 2006) and pulse-chase experiments in HeLa cells (Ligasova et al., 2009) indicate that sister replisomes reside close to one another within the nucleus, consistent with a physical association. The
most prominent example of physical coupling involves the simian virus 40 (SV40), in which the replicative DNA helicase large T-antigen (T-ag) is proposed to form obligatory double hexamers (Alexandrov et al., 2002; Weisshart et al., 1999; Wessel et al., 1992). In archaea, the MCM helicase forms a complex of two hexamers in solution (Fletcher et al., 2003), and in Saccharomyces cerevisiae, the MCM2-7 helicase is loaded onto origins as stable, double hexamers (Evrin et al., 2009; Remus et al., 2009). However, analysis of endogenous (Gambus et al., 2006; Moyer et al., 2006) and recombinant (Ilves et al., 2010) MCM27 complexes suggests that the active MCM2-7 complex might be a single hexamer. In summary, it is not known whether eukaryotic sister replisomes function independently or as dimeric complexes. This distinction is crucial to elucidate the architecture and mechanism of eukaryotic replisomes, and to understand the spatiotemporal coordination of replication involving a large number of origins. The physical coupling model envisions that DNA is pumped toward the associated sister replisomes, and that newly replicated DNA is extruded laterally from the replisome-replisome interface (Figure 1Aiii). If this model is correct, a constrained DNA template whose ends are xed should not undergo efcient DNA replication due to tension that accumulates on the unreplicated portions of the molecule (Figure 1Av). In contrast, independently functioning replisomes should travel apart and copy a constrained DNA template (Figure 1Aiv). Thus, to differentiate between these two models, we used Xenopus egg extracts to replicate DNA that was constrained at one or both ends, and replication was visualized at the single molecule level. The data show that in this vertebrate model system, efcient replication is independent of physical coupling between sister replisomes. RESULTS Replication of Immobilized DNA in Xenopus Egg Extracts Biotinylated l phage DNA (48.5 kb) was coupled at one or both 30 ends to the streptavidin-coated, bottom surface of a microuidic ow cell (see Figure S1A available online). To replicate these DNA molecules, we used a soluble cell-free system derived from Xenopus egg extracts (Walter et al., 1998). DNA is rst exposed to a high-speed supernatant (HSS) of egg cytoplasm that supports ORC-dependent but sequence-independent recruitment of MCM2-7 complexes to DNA. A second, nucleoplasmic
Molecular Cell
Eukaryotic Replisomes Can Function Autonomously
Figure 1. Single-Molecule Visualization of Eukaryotic Replication

(A) Possible congurations of sister replisomes. The sister replisomes assembled at each origin (i) travel away from each other (ii) or remain physically coupled (iii). Doubly tethered DNA is replicated efciently by independently functioning replisomes (iv) but not physically coupled replisomes, which stall after available slack in the DNA is consumed (v). (B) Protocol to induce replication of l DNA immobilized at one end in a ow cell. (C) Visualization of replicated DNA by TIRF microscopy. l DNA was incubated with extracts lacking (i and ii) or containing (iii and iv) Geminin and stained with anti-digoxigenin or SYTOX, as indicated.
ii
iii
iv
Immobilize DNA
Replication
HSS
NPE + dig-dUTP Remove Extracts Label DNA with SYTOX & Anti-dig
ing revealed that the immobilized l phage DNA contained alternating tracts of weak and strong uorescence intensity (Figure 1Cii). The strong tracts were FLOW dsDNA twice as intense as the weak tracts (Figdig-dUTP labeled nascent strand ure S1Bi, orange trace), and they coloPEG ii calized with anti-dig staining (Figure 1C, biotin compare i and ii; Figure S1Bi), suggeststreptavidin ing the strong tracts were due to the presence of two daughter duplexes C (schematically depicted in Figure 1B). -geminin iii +geminin i Both the anti-dig tracts and intense SYTOX tracts disappeared in the presAnti-Dig ence of Geminin, an inhibitor of MCM2iv ii 7 loading (Figures 1Ciii and 1Civ and SYTOX Figure S1Bii), indicating that these two signals reect chromosomal DNA repli5 m cation. Our results demonstrate that Xenopus egg extracts can efciently replicate DNA templates immobilized extract (NPE) is then added, which supports Cdk2-dependent within a microuidic ow cell, and that this process is readily activation of the MCM2-7 helicase, origin unwinding, replisome detected by two independent means (see Figure S2A for additional examples). assembly, and replication of the DNA. We rst examined replication kinetics of l phage molecules immobilized at only one end, leaving the DNA template uncon- Replication Kinetics of Singly Tethered DNA Molecules strained. After coupling l phage DNA to the surface, HSS was To characterize replication of singly tethered molecules, we drawn into the ow cell and allowed to incubate for 10 min (Fig- quantied several properties of the replication products ure 1Bi; for details, see the Experimental Procedures). Subse- (Figures 2A2C, black bars). As shown in Figure 2A, the quently, HSS was exchanged with NPE containing digoxigenin average number of bubbles per l DNA was 4.54 1.82, suglabeled dUTP (dig-dUTP). After a further 15 min, proteins were gesting an average interorigin distance of 10.7 kb, which agrees removed by washing the ow cell with SDS-containing buffer, well with previous measurements in Xenopus egg extracts uorescein-conjugated anti-digoxigenin antibody (anti-dig) (Blow et al., 2001). The lengths of the replication bubbles was added, and the DNA was stretched by buffer ow (Fig- showed an exponential distribution (Figure 2B), implying that ure 1Bii). Using total internal reection uorescence (TIRF) initiation events occurred randomly in time (Herrick et al., microscopy, we observed colinear tracts of anti-dig, indicating 2000). Most l DNA molecules were more than 60% replicated that replication of the immobilized l phage DNA had occurred after 15 min in NPE (Figure 2C). These observations suggest (Figure 1Ci). We also imaged the DNA using SYTOX Orange that replication of singly tethered DNA molecules is similar to (SYTOX), a uorescent DNA intercalating dye that labels what was previously observed in Xenopus egg extracts (see duplex DNA in a sequence-independent fashion. SYTOX stain- also below).
Molecular Cell
Singly-tethered
Doubly-tethered
A
Number of DNAs
B
40 30 20 10 0 2 4 6 8 Number of bubbles 10
Number of bubbles
150 100 50 0 10 20 30 Bubble size (kb) 40
Number of DNAs
C 60
40 20
20 40 60 80 100 Replication (%)
Figure 2. Replication Kinetics of Immobilized l DNA

Quantication of replication on singly tethered (black) and doubly tethered (gray) DNA after 15 min incubation in NPE. (A) Number of replication bubbles (anti-dig tracts) per immobilized l DNA. (B) Length distributions of replication bubbles. (C) Percent replication of individual l DNA molecules. Error bars indicate standard deviations.
Stretched l DNA Is Efciently Replicated in Extracts from Multiple Origins To test whether physical coupling between sister replisomes is required for their function, we repeated the experiment on DNA that was stretched and doubly tethered. To achieve this condition, DNA molecules biotinylated at both 30 ends were introduced into the ow cell at high ow rates. Under these conditions, DNA molecules attached to the surface of the ow cell at one end, whereupon they were instantaneously stretched by buffer ow before binding to the surface at the other end. Using this procedure, we achieved end-to-end distances corresponding to $90% of the expected contour length of B-form l-DNA (Figure S2C, see the Experimental Procedures for details). Importantly, such stretched (doubly tethered) DNA molecules replicated efciently from multiple origins (Figures 2A2C, gray bars; Figure S2B). When doubly tethered DNA molecules were incubated in extracts supplemented with Geminin, there was no DNA replication, demonstrating that replication of constrained DNA molecules is also MCM2-7 dependent (data not shown). Importantly, there was no signicant difference in the replication of singly tethered and doubly tethered molecules (Figures 2A2C, compare black and gray bars), suggesting that no physical contact between sister replisomes is required for replication in our system. A Single Pair of Diverging Forks Replicates Stretched l DNA Given that there was on average about 10% slack present in the doubly tethered l DNA, the replication observed above could have involved many short, neighboring replicons synthesized by physically coupled replisomes. To address this caveat, we examined whether a single pair of diverging sister forks can replicate stretched l DNA to an extent larger than the slack
originally present. To ensure that only a single pair of replisomes was activated on each DNA molecule, we used p27Kip, a Cdk2 inhibitor that blocks new initiations but does not affect elongation (Walter and Newport, 2000). Thus, 25 min after replication was initiated with NPE, we owed in fresh NPE containing p27Kip (Figure 3A). Under these conditions, the majority of DNA molecules exhibited one or no replication bubbles (Figure 3C). To verify that the observed bubbles were produced by two diverging replication forks, we supplied dig-dUTP 15 min after the initial NPE addition and allowed replication to proceed for a further 25 min (Figure 3A). In the replicated molecules, two tracts of dig-dUTP were visible whose outer edges coincided with the boundaries of the intense SYTOX tract, as expected for bidirectional replication from a single origin (Figure 3B and Figure S3A). Figure 3D shows that the extent of replication on every l molecule examined was much greater than the slack present in the substrate. For a single pair of physically coupled replisomes to produce such large replication bubbles, the l molecule would have to be stretched well beyond the contour length of B-form DNA. Because the force required to overstretch DNA to such an extent (van Mameren et al., 2009) is larger than any reported for individual DNA motors (Smith et al., 2001), and because such overstretching would almost entirely denature the DNA substrate (van Mameren et al., 2009), it is very unlikely that sister replisomes remained connected during replication in this system. We conclude that sister replisomes can function independently on immobilized l DNA molecules. Uncoupling of Sister Replisomes Does Not Affect Fork Rates To test whether the uncoupling of sister replisomes affects their replication dynamics, we compared fork rates on stretched,
Molecular Cell
A
0 min HSS 10 min NPE 12 min NPE + p27KIP 25 min NPE + p27KIP +dig-dUTP 50 min SDS
Figure 3. Replication of Stretched DNA by a Single Pair of Diverging Forks

(A) Scheme used to limit replication initiation to a single event on each l DNA molecule and to visualize bidirectional replication. (B) SYTOX (top), anti-dig (middle), and merged (bottom) images of three mechanically stretched l DNA molecules containing a single replication bubble. Extent of slack and replication are indicated. (C) Number of replication bubbles per monomeric l DNAs (n = 39). (D) Extent of replication versus the amount of slack on individual l DNA molecules that underwent single initiations. The amount of slack was calculated by comparing the end-to-end distance of doubly tethered DNA molecule to the B-form contour length of l DNA (16.5 mm). The solid line depicts the extent of replication expected if replication stops when the slack initially present in the l DNA is used up, as expected for physically coupled forks.
B
SYTOX Anti-Dig Merged Slack (%) Replicated (%) 8 41 10 45 10 50
C 1.0
Frequency
60 50 40
weak, positive correlation between the rates at which the two sister replisomes moved (Figure 4B, black squares, n = 30 0.5 48; R = 0.26, p = 0.07). Since uncoupling 20 of sister replisomes did not affect fork 10 progression in our system, the correlation between sister forks in relaxed DNA is 0.0 0 0 1 2 3 0 10 20 unlikely to be related to a functional interNumber of replication bubbles action between sisters. Consistent with Slack (%) per DNA this, termination of one replication fork by a double-strand DNA break in yeast does not affect progression of the sister doubly tethered and relaxed, singly tethered DNAs that had fork (Doksani et al., 2009). Therefore, correlations that we undergone a single initiation event (Figure S3). Dividing the and others observed likely represent chromatin microenvironlengths of the anti-dig tracts by the duration of the dig-dUTP ments that result in similar activity of nearby replisomes pulse yielded a mean fork rate of 267 160 bp/min for stretched (Conti et al., 2007), perhaps due to similar concentrations of DNA (n = 91; Figure 4A, gray bars) and 268 161 bp/min for key replication factors. Consistent with this idea, sister replirelaxed DNA (n = 98; Figure 4A, black bars). The presence of somes moving on stretched DNA, which are separated in space, dig-dUTP did not affect the rate of DNA synthesis (Figure S3C). showed no correlation (Figure 4B, gray squares, n = 45; R = 0.1, The measured rates were close to the lower estimates of fork p = 0.47). rates in conventional, nuclear assembly Xenopus egg extracts (Lu et al., 1998; Mahbubani et al., 1992). Thus, replisomes on DISCUSSION constrained and unconstrained molecules move at the same rates, demonstrating that replisome uncoupling does not It has been proposed that sister replisomes function as an obligatory dimeric complex (Falaschi, 2000; Kitamura et al., adversely affect replication fork progression. 2006; Ligasova et al., 2009; Sclafani et al., 2004). However, our Correlation between Rates of Sister Forks data demonstrate that no physical association is required between sister replisomes on l DNA replicating in Xenopus on Singly Tethered but Not Doubly Tethered DNA Previous studies in different experimental systems showed egg extracts, suggesting that replisomes can function indepenvarying but signicant degrees of correlation between the rates dently during vertebrate DNA replication. Together with previous at which sister replication forks progress (Conti et al., 2007; Du- results which failed to nd evidence of MCM2-7 double hexambey and Raman, 1987; Tapper and Depamphilis, 1980). We ers in S phase using coIP approaches (Gambus et al., 2006), and looked for a correlation between progression of sister forks in recent experiments using puried MCM2-7 holocomplexes (Ilves our system. Figure 4B plots the length of the left versus right et al., 2010; Moyer et al., 2006), our data suggest that sister anti-dig tracts within single replication bubbles on singly and MCM2-7 helicases (and replisomes) normally uncouple upon doubly tethered DNA. On singly tethered DNA, there was a activation, as seen in bacteria (Reyes-Lamothe et al., 2008),
Replicated (%)
Molecular Cell
40 Counts 30 20 10 0
Singly-tethered Doubly-tethered
0 2 4 6 8 10 12 14 16 Length of anti-dig tract (kb)
The independent action of sister replisomes has signicant advantages for eukaryotic cells. First, the multireplicon model, in which sister replisomes complete DNA synthesis at different times, would be difcult to reconcile with obligatory physical coupling between sisters. Second, single replisomes, having smaller dimensions than double replisomes, may be more adept at bypassing certain roadblocks and navigating a highly populated nuclear environment (Takahashi et al., 2004). The single-molecule techniques described here should be suitable to investigate the requirement for physical association between sister replisomes in other systems.
EXPERIMENTAL PROCEDURES DNA Immobilization in the Microuidic Flow Cell Sample ow cells were prepared as described previously (Lee et al., 2006). Briey, coverslips were functionalized with partially biotinylated high-molecular-weight poly(ethylene glycol) and incubated with 1 mg/ml streptavidin. Flow cells were assembled using these coverslips, double-adhesive tape, and glass slides with predrilled holes (Figure S1A). Outlet polyethylene tubing (0.03 inch inner diameter) was attached to an automated syringe pump (Harvard Instruments) to provide constant ow. Inlet tubing with 5 cm length and 0.015 inch inner diameter reduced dead volume. To prevent nonspecic DNA sticking to the surface, the ow cell was incubated with blocking buffer (20 mM Tris [pH 7.5], 50 mM NaCl, 2 mM EDTA, 0.2 mg/ml BSA) for at least 15 min. To attach l DNA (New England Biolabs) to the streptavidin coated surface at one end, the single-stranded 50 tails of l DNA were annealed and ligated to complementary oligonucleotides 50 -AGGTCGCCGCCC-Biotin-30 and 50 -GGGCGGCGACCT-30 (Integrated DNA Technologies). For double tethering, both oligos contained biotin at the 30 end. Biotinylated l DNA (1550 pM) in blocking buffer was injected into the ow cell at a constant rate of 20100 ml/min. At a ow rate of 50100 ml/min, l DNA biotinylated at both ends was stretched to 70%80% of its contour length (16.5 mm) (Figure S2CiS2iii). To stretch DNA further, we used chloroquine, which intercalates into and extends the pitch of dsDNA (Cohen and Yielding, 1965). When injected at 100 ml/min in the presence of 100 mM chloroquine in blocking buffer, end-to-end distance of doubly tethered l DNAs was 85%95% of its contour length (Figure S2Civ). After DNA injection and before addition of extract, chloroquine was removed by extensive washing of the ow cell with blocking buffer (5 min at 100 ml/min). To limit our analysis to DNA molecules that remained doubly tethered during the entire replication reaction, we used a reduced ow rate (25 ml/min) for all buffer exchanges following replication. In this way, even if a DNA molecule that detached from one end in extracts became doubly tethered during subsequent washes, it would be stretched to a much lesser extent than molecules that remained stretched throughout the experiment. Thus, l DNA stretched to 85%95% of its contour length at the end of the experiment must have stayed doubly tethered during replication. Therefore, we analyzed only those molecules that were stretched to 85% or more as doubly tethered. Replication of Immobilized DNA In a separate line of investigation, we recently discovered that DNA replication in Xenopus egg extracts requires a minimum threshold concentration of DNA ($1 ng/ml) in HSS and NPE (Lebofsky et al., 2010). Since the effective concentration of l DNA immobilized in the ow cell was extremely low, we supplemented HSS (Walter et al., 1998) and NPE (Walter et al., 1998) with carrier plasmid to raise the overall DNA concentration to levels that are compatible with DNA replication. Thus, after immobilizing l DNA on the functionalized surface, HSS containing carrier plasmid (510 ng/ml of pBluescript II KS []) was injected at 10 ml/min for 2 min and further incubated for 8 min without ow. Next, a 2:1 mixture of NPE and HSS supplemented with 510 ng/ml of pBS (replication extract) was owed in at 10 ml/min for 80 s, followed by incubation for different lengths of time without ow, as indicated. dig-dUTP (Roche Inc.) (7 mM) was also included in the replication extract for labeling
15 10 5 0
Length of anti-dig on left (kb)
10
15
Length of anti-dig on right (kb)

Figure 4. Analysis of Fork Rates
(A) Length of anti-dig tracts under single-initiation conditions on singly tethered (black) and doubly tethered (gray) l DNA molecules. Error bars indicate standard deviations. (B) Lengths of sister anti-dig tracts of the rightward fork versus the leftward fork on singly (black) and doubly tethered (gray) DNA molecules. The dashed line represents perfectly correlated sister forks.
even though they may colocalize within replication foci (Kitamura et al., 2006; Ligasova et al., 2009). The double hexamer model for MCM2-7 was largely inspired by the analysis of SV40 T-ag. Electron microscopy showed that T-ag loads onto the SV40 origin as two hexamers, which associate through their N termini (Valle et al., 2000). In addition, during T-ag-mediated DNA unwinding, a fraction of DNAs adopt a rabbit ear conformation, in which two loops of singlestranded DNA emanate from the T-ag complex, implying an association of two hexamers (Wessel et al., 1992). Finally, mutations in T-ag that compromise double-hexamer formation inhibit DNA unwinding, and double hexamers of T-ag possess higher unwinding activity than single hexamers (Alexandrov et al., 2002; Weisshart et al., 1999). Taken together, these studies strongly suggest that double-hexamer formation is crucial for SV40 replication. However, it is not clear whether the interactions between two hexamers are essential only for replication initiation or also during fork elongation.
Molecular Cell
of replicated regions. All reactions were carried out at room temperature (22 C). To observe bidirectional replication involving single initiation events, immobilized l DNA was incubated with HSS/carrier plasmid and subsequently replaced with replication extract (lacking dig-dUTP). After the time specied, a second replication extract containing 66 mg/ml p27Kip was injected and incubated further. Finally, a replication extract containing 66 mg/ml p27Kip and 7 mM dig-dUTP was introduced. In each case, replication extract was injected at 10 ml/min for 80 s. To stop the replication reaction, the ow cell was washed with SDS buffer (20 mM Tris [pH 7.5], 50 mM NaCl, 12 mM EDTA, 0.1% SDS) for 10 min at 25 ml/min. To label dig-dUTP, anti-digoxigenin-uorescein antibody (Roche Inc.) diluted to 0.4 mg/ml with buffer (10 mM HEPES [pH 7.7], 2.5 mM MgCl2, 50 mM KCl, 0.2 mg/ml BSA) was drawn into the ow cell for 20 min at 25 ml/min. Excess antibody was removed by washing the ow cell with blocking buffer. Finally, blocking buffer containing 15 nM SYTOX Orange (Invitrogen) was introduced to uorescently label dsDNA. Use of Oligonucleotides as Carrier DNA As described above, DNA replication of tethered l DNA in the ow cell was carried out using egg extracts (HSS and NPE) that were supplemented with high concentrations of a carrier plasmid. Like the immobilized l DNA, the free carrier plasmid is also expected to undergo replication in the ow cell. To ensure that the activity of replisomes on the immobilized l DNA was not dependent on interactions with replisomes on the carrier plasmid, we replaced the carrier plasmid with a short, double-stranded oligonucleotide (29 bp; Lebofsky et al., 2010) that does not support loading of MCM2-7 helicase due to its small size and therefore cannot undergo replication (Edwards et al., 2002). We have shown that such a nucleotide is able to replace carrier plasmid to promote licensing in HSS (Lebofsky et al., 2010). Doubly tethered l DNA was licensed with HSS containing 10 ng/ml of the oligo duplex. Subsequently, standard replication extract (NPE/HSS/carrier plasmid) was injected. Importantly, the carrier plasmid supplied with the replication extracts did not undergo DNA replication because it was added to a mixture of HSS and NPE, and NPE contains very high concentrations of Geminin, which block DNA replication (Figure S4A) (Walter et al., 1998). Thus, in this sequence, none of the added carrier DNA (oligo or plasmid) underwent DNA replication, yet single initiations on many stretched l DNA molecules still produced bubbles larger than the slack present on stretched l DNA (Figure S4B). These observations conrm that replication of carrier plasmid does not mediate replication of immobilized l DNA molecules. TIRF Microscopy Immobilized l DNA molecules were imaged on an objective-type TIRF conguration using an inverted microscope (IX-71; Olympus) equipped with a 603 oil objective (PlanApo, N.A. = 1.45; Olympus) and a 1.63 magnication unit. A multiwavelength Ar-Kr ion laser (Innova 70C-Spectrum, Coherent Inc.) was used for illumination. SYTOX Orange and uorescein were excited with 568 and 488 nm laser light, respectively, using varying intensities and 100 ms exposures per frame. Images were acquired using an Andor iXon backilluminated electron-multiplying CCD camera (Andor Technology) at 2 Hz. Singly tethered molecules were imaged at a ow rate of 100125 ml/min (for stretching), while doubly tethered DNAs were imaged in the absence of ow since they were already stretched. Image Processing To improve the signal-to-noise ratio on uorescence images of replicated l DNA, multiple (530) consecutive images of SYTOX and uorescein were averaged separately using ImageJ and merged using Adobe Photoshop. The end-to-end distance of each DNA molecule was measured via the SYTOX image, and the size of a replication bubble was determined using the SYTOX or the uorescein signal. SUPPLEMENTAL INFORMATION Supplemental Information includes four gures and Supplemental References and can be found with this article at doi:10.1016/j.molcel.2010.11.027.
ACKNOWLEDGMENTS We thank Ronald Lebofsky for communicating unpublished results and Charles C. Richardson, Ronald Lebofsky, and Jerard Hurwitz for critical reading of the manuscript. This work was supported by National Institutes of Health (NIH) grant GM62267 and a Leukemia and Lymphoma Scholar Award (to J.C.W.). A.M.v.O. acknowledges support from American Cancer Society grant RSG-08-234-01 and Searle Scholarship 05-L-104. A.B.L. was supported by NIH/NIGMS Molecular Biophysics Training Grant T32 GM008313. Received: April 20, 2010 Revised: September 13, 2010 Accepted: September 24, 2010 Published: December 9, 2010 REFERENCES Alexandrov, A.I., Botchan, M.R., and Cozzarelli, N.R. (2002). Characterization of simian virus 40 T-antigen double hexamers bound to a replication fork. J. Biol. Chem. 277, 4488644897. Blow, J.J., Gillespie, P.J., Francis, D., and Jackson, D.A. (2001). Replication origins in Xenopus egg extract are 5-15 kilobases apart and are activated in clusters that re at different times. J. Cell Biol. 152, 1526. Bochman, M.L., and Schwacha, A. (2009). The Mcm complex: unwinding the mechanism of a replicative helicase. Microbiol. Mol. Biol. Rev. 73, 652683. Cohen, S.N., and Yielding, K.L. (1965). Spectrophotometric studies of the interaction of chloroquine with deoxyribonucleic acid. J. Biol. Chem. 240, 31233131. Conti, C., Sacca, B., Herrick, J., Lalou, C., Pommier, Y., and Bensimon, A. (2007). Replication fork velocities at adjacent replication origins are coordinately modied during DNA replication in human cells. Mol. Biol. Cell 18, 30593067. Doksani, Y., Bermejo, R., Fiorani, S., Haber, J.E., and Foiani, M. (2009). Replicon dynamics, dormant origin ring, and terminal fork integrity after double-strand break formation. Cell 137, 247258. Dubey, D.D., and Raman, R. (1987). Do sister forks of bidirectionally growing replicons proceed at unequal rates? Exp. Cell Res. 168, 555560. Edwards, M.C., Tutter, A.V., Cvetic, C., Gilbert, C.H., Prokhorova, T.A., and Walter, J.C. (2002). MCM2-7 complexes bind chromatin in a distributed pattern surrounding the origin recognition complex in Xenopus egg extracts. J. Biol. Chem. 277, 3304933057. Evrin, C., Clarke, P., Zech, J., Lurz, R., Sun, J., Uhle, S., Li, H., Stillman, B., and Speck, C. (2009). A double-hexameric MCM2-7 complex is loaded onto origin DNA during licensing of eukaryotic DNA replication. Proc. Natl. Acad. Sci. USA 106, 2024020245. Falaschi, A. (2000). Eukaryotic DNA replication: a model for a xed double replisome. Trends Genet. 16, 8892. Fletcher, R.J., Bishop, B.E., Leon, R.P., Sclafani, R.A., Ogata, C.M., and Chen, X.S. (2003). The structure and function of MCM from archaeal M. Thermoautotrophicum. Nat. Struct. Biol. 10, 160167. Gambus, A., Jones, R.C., Sanchez-Diaz, A., Kanemaki, M., van Deursen, F., Edmondson, R.D., and Labib, K. (2006). GINS maintains association of Cdc45 with MCM in replisome progression complexes at eukaryotic DNA replication forks. Nat. Cell Biol. 8, 358366. Herrick, J., Stanislawski, P., Hyrien, O., and Bensimon, A. (2000). Replication fork density increases during DNA synthesis in X. laevis egg extracts. J. Mol. Biol. 300, 11331142. Ilves, I., Petojevic, T., Pesavento, J.J., and Botchan, M.R. (2010). Activation of the MCM2-7 helicase by association with Cdc45 and GINS proteins. Mol. Cell 37, 247258. Kitamura, E., Blow, J.J., and Tanaka, T.U. (2006). Live-cell imaging reveals replication of individual replicons in eukaryotic replication factories. Cell 125, 12971308.
Molecular Cell
Lebofsky, R., van Oijen, A.M., and Walter, J.C. (2010). DNA is a co-factor for its own replication in Xenopus egg extracts. Nucleic Acids Res. Published online September 21, 2010. 10.1093/nar/gkq739. Lee, J.-B., Hite, R.K., Hamdan, S.M., Sunney Xie, X., Richardson, C.C., and van Oijen, A.M. (2006). DNA primase acts as a molecular brake in DNA replication. Nature 439, 621624. Ligasova, A., Raska, I., and Koberna, K. (2009). Organization of human replicon: singles or zipping couples? J. Struct. Biol. 165, 204213. Lu, Z.H., Sittman, D.B., Romanowski, P., and Leno, G.H. (1998). Histone H1 reduces the frequency of initiation in Xenopus egg extract by limiting the assembly of prereplication complexes on sperm chromatin. Mol. Biol. Cell 9, 11631176. Mahbubani, H.M., Paull, T., Elder, J.K., and Blow, J.J. (1992). DNA replication initiates at multiple sites on plasmid DNA in Xenopus egg extracts. Nucleic Acids Res. 20, 14571462. Moyer, S.E., Lewis, P.W., and Botchan, M.R. (2006). Isolation of the Cdc45/ Mcm2-7/GINS (CMG) complex, a candidate for the eukaryotic DNA replication fork helicase. Proc. Natl. Acad. Sci. USA 103, 1023610241. Remus, D., Beuron, F., Tolun, G., Grifth, J.D., Morris, E.P., and Difey, J.F.X. (2009). Concerted loading of Mcm2-7 double hexamers around DNA during DNA replication origin licensing. Cell 139, 719730. Reyes-Lamothe, R., Possoz, C., Danilova, O., and Sherratt, D.J. (2008). Independent positioning and action of Escherichia coli replisomes in live cells. Cell 133, 90102. Sclafani, R.A., Fletcher, R.J., and Chen, X.S. (2004). Two heads are better than one: regulation of DNA replication by hexameric helicases. Genes Dev. 18, 20392045. Smith, D.E., Tans, S.J., Smith, S.B., Grimes, S., Anderson, D.L., and Bustamante, C. (2001). The bacteriophage phi29 portal motor can package DNA against a large internal force. Nature 413, 748752.
Takahashi, T.S., Yiu, P., Chou, M.F., Gygi, S., and Walter, J.C. (2004). Recruitment of Xenopus Scc2 and cohesin to chromatin requires the pre-replication complex. Nat. Cell Biol. 6, 991996. Takahashi, T.S., Wigley, D.B., and Walter, J.C. (2005). Pumps, paradoxes and ploughshares: mechanism of the MCM2-7 DNA helicase. Trends Biochem. Sci. 30, 437444. Tapper, D.P., and Depamphilis, M.L. (1980). Preferred DNA sites are involved in the arrest and initiation of DNA synthesis during replication of SV40 DNA. Cell 22, 97108. Valle, M., Gruss, C., Halmer, L., Carazo, J.M., and Donate, L.E. (2000). Large T-antigen double hexamers imaged at the simian virus 40 origin of replication. Mol. Cell. Biol. 20, 3441. van Mameren, J., Gross, P., Farge, G., Hooijman, P., Modesti, M., Falkenberg, M., Wuite, G.J.L., and Peterman, E.J.G. (2009). Unraveling the structure of DNA during overstretching by using multicolor, single-molecule uorescence imaging. Proc. Natl. Acad. Sci. USA 106, 1823118236. Walter, J., and Newport, J. (2000). Initiation of eukaryotic DNA replication: origin unwinding and sequential chromatin association of Cdc45, RPA, and DNA polymerase alpha. Mol. Cell 5, 617627. Walter, J., Sun, L., and Newport, J. (1998). Regulated chromosomal DNA replication in the absence of a nucleus. Mol. Cell 1, 519529. Weisshart, K., Taneja, P., Jenne, A., Herbig, U., Simmons, D.T., and Fanning, E. (1999). Two regions of simian virus 40 T antigen determine cooperativity of double-hexamer assembly on the viral origin of DNA replication and promote hexamer interactions during bidirectional origin DNA unwinding. J. Virol. 73, 22012211. Wessel, R., Schweizer, J., and Stahl, H. (1992). Simian virus 40 T-antigen DNA helicase is a hexamer which forms a binary complex during bidirectional unwinding from the viral origin of DNA replication. J. Virol. 66, 804815.
Molecular Cell
Short Article
Intronic miR-211 Assumes the Tumor Suppressive Function of Its Host Gene in Melanoma
Carmit Levy,1,10 Mehdi Khaled,1,2,7,9,10 Dimitrios Iliopoulos,5,12 Maja M. Janas,2,7,9 Steffen Schubert,2,7,9,11 Sophie Pinner,2,7 Po-Hao Chen,2,7,9 Shuqiang Li,2,7,9 Anne L. Fletcher,2,7 Satoru Yokoyama,1 Kenneth L. Scott,3 Levi A. Garraway,3,4,9 Jun S. Song,8 Scott R. Granter,6 Shannon J. Turley,2,7 David E. Fisher,1,7,* and Carl D. Novina2,7,9,*
of Dermatology, Massachusetts General Hospital of Cancer Immunology and AIDS 3Medical Oncology 4Center for Cancer Genome Discovery, Dana-Farber Cancer Institute 5Department of Biological Chemistry and Molecular Pharmacology 6Pathology Department, Brigham and Womens Hospital 7Department of Pathology Harvard Medical School, Boston, MA 02115, USA 8Institute for Human Genetics, University of California, San Francisco, San Francisco, CA 94143, USA 9Broad Institute of Harvard and MIT, Cambridge, MA 02141, USA 10These authors contributed equally to this work 11Present address: Cenix Bioscience, 01307 Dresden, Germany 12Present address: Department of Cancer Immunology and AIDS, and Department of Pathology, Harvard Medical School, Boston, MA 02115, USA *Correspondence: dsher3@partners.org (D.E.F.), carl_novina@dfci.harvard.edu (C.D.N.) DOI 10.1016/j.molcel.2010.11.020
SUMMARY
When it escapes early detection, malignant melanoma becomes a highly lethal and treatment-refractory cancer. Melastatin is greatly downregulated in metastatic melanomas and is widely believed to function as a melanoma tumor suppressor. Here we report that tumor suppressive activity is not mediated by melastatin but instead by a microRNA (miR-211) hosted within an intron of melastatin. Increasing expression of miR-211 but not melastatin reduced migration and invasion of malignant and highly invasive human melanomas characterized by low levels of melastatin and miR-211. An unbiased network analysis of melanoma-expressed genes ltered for their roles in metastasis identied three central node genes: IGF2R, TGFBR2, and NFAT5. Expression of these genes was reduced by miR-211, and knockdown of each gene phenocopied the effects of increased miR-211 on melanoma invasiveness. These data implicate miR-211 as a suppressor of melanoma invasion whose expression is silenced or selected against via suppression of the entire melastatin locus during human melanoma progression.
INTRODUCTION A large proportion of advanced melanomas show downregulation of melastatin, a founding member of the transient receptor potential (TRPM) cation channel family that is highly expressed in melanocytes and retinal pigmented epithelium. Whereas mel-
astatin (TRPM1) is robustly expressed in benign and dysplastic nevi and in melanomas in situ, it is only variably expressed in invasive melanomas, shows widespread downregulation in melanoma metastases, and is inversely correlated with metastatic potential and prognosis of melanomas (Duncan et al., 1998, 2001). However, the mechanism(s) by which melastatin might suppress melanoma metastasis remains unknown. Interestingly, intron 6 of the TRPM1 gene (Hunter et al., 1998) hosts the gene for miR-211, a microRNA (miRNA) whose expression is restricted to the melanocyte lineage (Gaur et al., 2007). Numerous miRNAs have been implicated in tumorigenesis (e.g., miR-17$92 as an oncogenic cluster [He et al., 2005; ODonnell et al., 2005]; let-7 as a tumor suppressor [Takamizawa et al., 2004; Yanaihara et al., 2006]; and miR-10b [Ma et al., 2007], miR-373 and miR-520c [Huang et al., 2008], miR-335 [Tavazoie et al., 2008], and miR-9 [Ma et al., 2010] in tumor invasion and metastasis). These observations, together with the lack of mechanistic data linking melastatin to melanoma tumor suppression, led us to hypothesize that miR-211 activity might regulate melanoma invasion and metastasis independently of its host gene function. Melanoma short-term cultures (MSTCs) represent a physiologically relevant and an experimentally tractable cancer model because they have relatively low passage numbers since biopsy, are readily annotated with respect to genetic alterations, and are easily grown in culture. Through genetic and functional studies performed on a collection of MSTCs, we identied two populations of malignant melanomas: one with low miR-211 expression and high invasive activity, and the other with high miR-211 expression and low invasive activity. Here we report that manipulation of miR-211 levels, but not melastatin levels, alters the invasive potential of malignant melanomas, and we demonstrate a network of miR-211-responsive genes affecting melanoma invasive activity.
Molecular Cell
miR-211 Reduces Melanoma Invasiveness
MicroRNA library
Invasive cells/field
200 160 120 0 40 80 40 120 160 200
transfection A375M cells
16h incubation Invasive cells/field
0 miR-211 Invasive cells/field

Replicate 2
250 Invasive cells/field 200 150 100 50 hsa-miR-509 miR-NC Untreated hsa-miR-211 hsa-miR-204 hsa-miR-330 hsa-let-7b hsa-miR-210 hsa-miR-489 hsa-miR-20 hsa-miR-378 hsa-miR-19a hsa-miR-532 hsa-miR-515-3p hsa-miR-222 hsa-miR-520f hsa-miR-221 hsa-miR-181b hsa-miR-133a hsa-miR-133b hsa-miR-335 hsa-miR-199b hsa-miR-140 hsa-miR-449b hsa-miR-182 hsa-miR-302c hsa-miR-21 hsa-miR-34a hsa-miR-219 hsa-let-7d hsa-let-7a 0
miR-211
Replicate 1 Replicate 2
untreated
Replicate 1
miR-21
Figure 1. A Genome-wide miRNA Screen Identies miR-211 as a Suppressor of Melanoma Invasion

(A) Screening strategy to identify miRNAs regulating A375M melanoma cell invasiveness. (B) Number of invasive cells (average of ve high-power elds) after transfection with the miRNA library in A375M cells. The experiment was performed in duplicate (replicates 1, 2). (C) Number of invasive cells after transfection with positive hits derived from the miRNA library screen experiment. The experiment was performed in duplicate; means SEM.
RESULTS AND DISCUSSION To determine whether miR-211 or other miRNAs can regulate melanoma invasiveness independently from melastatin, we performed an unbiased miRNA library screen in a highly invasive melanoma cell line, A375M (Figures 1A and 1B and see Figure S1 available online). We found that expression of the melanomaspecic miR-211 signicantly decreased invasiveness of A375M cells. Interestingly, the second highest hit in this screen was the miR-211 paralog, miR-204, which is not expressed in melanocytes but shares the same seed region as miR-211 and therefore targets similar genes. Each of the miRNAs identied in the primary screen was validated and ranked by invasiveness in
a secondary invasive activity screen (Figure 1C). Importantly, increased expression of tumor suppressor miRNAs miR-335 (Tavazoie et al., 2008), let-7a, let-7b, and let-7d (Takamizawa et al., 2004; Yanaihara et al., 2006) led to decreased melanoma invasiveness (Figure 1C). Conversely, increases in the oncogenic miRNAs miR-19a, miR-20 (He et al., 2005; ODonnell et al., 2005), and miR-21 (Medina et al., 2010) led to increased melanoma invasiveness (Figure 1C). To determine whether the expression levels of miR-211 or its host gene melastatin (Figure 2A) affect invasive activity of human melanomas, we compared mature miR-211 and its completely processed host gene mRNA levels across human melanocytes and malignant melanomas by qRT-PCR (Figure 2B).
Molecular Cell
B
10
Relative TRPM1 & miR-211 expression
mature miR-211 TRPM1 exons 6-7
miR211 TRPM1
0.1
27
0.01
0.001
0.0001
0.00001
C
Anti-miR -Ctrl AntimiR-211
WM3682
WM3526
451LU Ctrl mimic
Cell speed (m/hr)
miR-211 mimic
siCtrl
Renilla
siTRPM1
TRPM1 20 18 16 14 12 10 8 6 4 2 0
Before scraping Invasive cells /field
Before scraping Invasive cells /field
12 10 8 6 4 2 0
6 4
*
1 Anti-miR-Ctrl Anti-miR-211
2 0
*
siCtrl siTRPM1
12 10 8 6 4 2 0
30 20 10 0
30
40 30 20 10 0
*
0 Ctrl mimic 2 miR-211 mimic
20 10 0
*
Renilla TRPM1
Cell speed (m/hr)
Figure 2. Perturbing miR-211 but Not Melastatin Affects Melanoma Invasion and Motility
(A) Schematic presentation of the miR-211 gene in intron 6 of the melastatin (TRPM1) gene. (B) qRT-PCR of TRPM1 (gray bars) normalized to actin, and mature miR-211 (black bars) normalized to RNU48 in multiple human primary melanocytes (HP) and melanoma short-term cultures (MSTCs). Y axis is logarithmic scale (n = 5; means SEM). (C) Matrigel assays for WM3682, WM3526, and 451LU (panels at left) transfected with control antagomir (Anti-miR-Ctrl, row 1), a miR-211-specic antagomir (Anti-miR-211, row 2), a control siRNA (siCtrl, row 3), or a melastatin-specic siRNA (siTRPM1, row 4). Membranes prior to scraping noninvasive cells from the top are shown (row 5). Matrigel assays for WM1745, WM1716, or WM3314 (panels at right) transfected with control miRNA mimic (Ctrl mimic, row 1), a miR-211 mimic (miR-211-mimic, row 2), a control Renilla cDNA (Renilla), or TRPM1 cDNA (TRPM1, row 3). The membranes prior to scraping cells from the top of the membrane are shown (row 4). Graphical presentation of the data quanties the number of invasive cells in ten high-power elds (means SEM; *p < 0.005). (D) Modulating miR-211 but not melastatin affects melanoma motility. Real-time video microscopy of WM1716 (top panel) expressing Renilla cDNA, melastatin cDNA (TRPM1), a control miRNA mimic (Ctrl mimic), or miR-211 mimic. Real-time video microscopy of 451LU (bottom panel) expressing a control antagomir (Anti-miR-Ctrl), a miR-211 antagomir (Anti-miR-211), a siRNA control (siCtrl), or a melastatin siRNA (siTRPM1). Results quantify cell migration in mm/hr (>25 representative cells analyzed for each category from two movies, means SEM; *p < 0.005). Real-time video images are added as Web-browsable links.
Consistent with possible tumor suppressor roles in melanomas, both melastatin and miR-211 were reduced in almost all malignant melanomas compared to melanocytes (Figure 2B). Linear regression analysis identied a strong direct correlation between melastatin mRNA levels and mature miR-211 levels (R2 = 0.73, Figure S2A). Moreover, our data indicate that miR-211 and melastatin share a common promoter (Figures S2BS2D), which is consistent with previous observations (Marson et al., 2008; Ozsolak et al., 2008). In addition, the expression of both genes is MITF dependent (Figures S2ES2H). Together, these data
H P1 H P2 H P3 H P4 H P5 H W P6 M 1 W 716 M 17 W 19 M 1 W 720 M 17 W 45 M 1 W 963 M 20 W 29 M 30 W 61 M 3 W 066 M 32 W 28 M 3 W 246 M 32 W 82 M 3 W 314 M 3 W 457 M 3 W 506 M 35 W 26 M 3 W 619 M 3 W 627 M 36 8 45 2 1L U
Primary melanocyte
MSTC
WM1745
WM1716
WM3314
D
16 14 12 10 8 6 4 2 0 WM1716
*
Ctrl miR-211 Renilla TRPM1 mimic mimic
451LU
Anti- AntisiCtrl siTRPM1 miR miR-211 -Ctrl
demonstrate that the levels of melastatin and miR-211 are coordinately increased or decreased in melanomas and indicate coupled transcriptional regulation by MITF. Next, we examined invasive activity and proliferation rates as a function of melastatin and miR-211 expression across several representative melanomas with mildly ($10-fold) and greatly ($1000-fold) reduced miR-211 levels (Figure 2B). More than 20-fold higher invasive activity was observed in melanomas with greatly reduced miR-211 levels (WM1745, WM1716, and WM3314) compared to melanomas with mildly reduced
Molecular Cell
miR-211 (WM3682, WM3526, and 451LU; Figure 2C, top panel, and Figure S2I). Consistent with reduced melastatin expression in slowly proliferating but highly invasive melanomas (Hoek et al., 2006), the doubling times of melanomas with higher melastatin and miR-211 (1.2 days and 2 days for 451LU and WM3526, respectively) were considerably shorter than the doubling times of melanomas with lower melastatin and miR-211 (more than 8 days for WM1745; Figure S2J). To distinguish effects of miR-211 from effects of its host gene, melanomas with no reduction (WM3682) or with 10-fold reduction (WM3526 and 451LU) in melastatin and miR-211 relative to human primary melanocytes were transfected with miR-211-specic antagomirs (antisense oligonucleotides that inhibit miRNAs in a sequence-specic fashion [Krutzfeldt et al., 2005]), and the effect on melanoma invasiveness was assessed in matrigel assays (Figure 2C and Figure S2K). Inhibition of miR-211 increased melanoma invasiveness by at least 10-fold compared to a control antagomir. Importantly, knockdown of melastatin had no effect on melanoma invasiveness (Figure 2C and Figure S2K). These data indicate that melanoma invasiveness is suppressed by miR-211 but not melastatin. The effects of increasing the expression of miR-211 were also examined. Melanomas with $1000-fold reductions in miR-211 (WM1745, WM1716, and WM3314) relative to human primary melanocytes were transfected with a plasmid expressing a miR-211 precursor (Figure S2N). Increasing miR-211 levels in these melanomas had no effect on growth rates (Figure S2O) but reduced the invasive potential by 2- to 3-fold compared to scrambled controls (Figure 2C). In contrast, modulation of melastatin expression in these melanomas had no effect on melanoma invasiveness (Figure 2C and Figures S2LS2N). However, increasing MITF levels reduced melanoma invasiveness, and decreasing MITF levels increased melanoma invasiveness (Figures S2G, S2H, and S2P), likely due to its effect on miR-211 expression. Modulating miR-211 had no effect on melastatin expression (Figure S2M). Similarly, modulating melastatin had no effect on miR-211 expression (Figure S2M). We conclude that changes in expression of miR-211 affect melanoma invasiveness independent of melastatin and do not affect melanoma proliferation rates. To determine the effect of miR-211 expression levels on melanoma cell migration, we performed real-time video microscopy (Figure 2D and Supplemental Movies). WM1716 cells, expressing reduced miR-211 (Figure S2N), exhibited spindle-like morphology when cultured on a deformable collagen-matrigel matrix. When transfected with a plasmid expressing melastatin cDNA or controls (renilla luciferase cDNA or a scrambled miRNA), WM1716 demonstrated an average motility speed of approximately 5 mm/hr and maintained the spindle-like morphology. Strikingly, expression of miR-211 signicantly reduced cell migration to only 1.5 mm/hr and induced rounded cell morphology. The effects of reducing miR-211 or melastatin on cell migration were also examined in 451LU cells expressing elevated miR-211. In agreement with the results in WM1716 cells, reduction of miR-211 but not melastatin increased cell motility (Figure 2D). Thus miR-211, but not melastatin, affects melanoma cell motility and invasiveness.
To discover the genetic network mediating the effects of miR-211 on melanoma invasiveness, we interrogated a literature-based set of genes restricted for melanoma expression and ltered for annotated functions in metastasis (Figure 3A and Table S1). A computationally predicted, melanoma-specic metastasis network was generated using IPA gene network software (described in the Experimental Procedures). This analysis revealed two statistically signicant (p < 105) gene networks, with the most statistically signicant network (p = 1037) containing three central nodes (dened in the Experimental Procedures): IGF2R, TGFBR2, and NFAT5 (Figure 3B and Figure S3). To identify any function of miR-211 in the bioinformatically predicted melanoma metastasis networks, miR-211 expression data was compared to sample-matched mRNA expression data across a large collection of melanomas (Lin et al., 2008). Translational repression mediated by miRNAs is usually accompanied by reduction of target mRNA levels (Filipowicz et al., 2008). Thus, to a rst approximation, potential target mRNAs are expected to demonstrate inverse correlation with miR-211. Kendalls tau (t) rank coefcient was calculated for all mRNA/miR-211 pairs to discern concentration-dependent relationships between miR-211 and mRNAs expressed in melanomas (Figure 4A). As an internal measure of the accuracy of this analysis, a strong direct correlation was observed between the abundance of miR-211 and important target genes of MITF, including TYRP1, SILV, MLANA, and CDK2 (Figure 4A) (Levy et al., 2006). A similar ranked list of genes correlated with miR-211 expression was obtained using class-dependent identication of marker genes with the ComparativeMarkerSelection module in GenePattern (data not shown). The mRNAs with signicant Kendalls tau coefcients were compared with predicted targets of miR-211 using TargetScan. Confounding effects of genomic amplication were eliminated by removing mRNAs that showed increased gene copy numbers in our cell lines based on matching single nucleotide polymorphism data for those genes (Lin et al., 2008) (Table S2). Six genes were found to be represented in both the list of plausible miR-211 targets as well as the independently generated melanoma metastasis network, including IGF2R (TargetScan context score = 0.42), NFAT5 (TargetScan context score = 0.06), TGFBR2 (TargetScan context score = 0.49), FBXW7, ANGPT1, and VHL. Interestingly, several genes demonstrating inversely correlated expression with miR-211 levels and with lowestrank coefcients had established roles in melanomagenesis including the central nodes IGF2R, TGFBR2, and NFAT5. To determine whether these central node genes are biological targets of miR-211, the levels of miR-211 were perturbed in melanomas, and IGF2R, TGFBR2, and NFAT5 expression levels were examined (Figure 4B and Figures S4A and S4B). Reducing miR-211 in WM3526 cells increased the expression of all three mRNAs. Conversely, increasing miR-211 in WM1716 and in A375M cells reduced levels of all three mRNAs, indicating that endogenous IGF2R, TGFBR2, and NFAT5 are biological targets of miR-211 in melanomas. To determine whether these genes are direct targets of miR-211, we transfected luciferase-expressing constructs containing the wild-type and predicted miR-211 binding site mutant 30 untranslated regions (30 UTRs) of these
Molecular Cell
Figure 3. Gene Network Analysis Identies Three Central Nodes Regulating Melanoma Metastasis Potentially Targeted by miR-211
(A) Literature-based melanoma and metastasis gene lists as demonstrated in the Venn diagram were used to construct the melanoma metastasis gene network. (B) Bioinformatically predicted melanoma metastasis gene network (p = 1037). Solid lines indicate direct interactions, and dashed lines indicate indirect interactions. Arrows indicate activation. Lines ending in short perpendicular lines indicate repression. TargetScan-predicted miR-211 targets genes are shaded in gray (IGFR = 0.42, TGFBR2 = 0.49, NFAT = 0.06, DPP4 = 0.14, 0.08, AKT2 = 0.01, MMP3 = 0.09, ADAM19 = 0.11, 0.08, MCAM = 0.12). Knockdown of each central nodes gene is predicted to inhibit the entire metastasis network except for the genes indicated by hatch marks.
Melanoma genes
Network genes
Metastasis genes
genes into HeLa cells (which lack endogenous miR-211) with or without a miR-211 mimic. Unexpectedly, TGFBR2 and NFAT5, but not IGF2R, were shown to be direct targets of miR-211 (Figure 4C). Because miR-211 affects endogenous expression of the central node genes, we tested the effect of knockdown of these genes on melanoma invasive activity. Interestingly, knockdown of any central node gene phenocopied the effect of increasing miR-211 levels in melanomas with high levels of IGF2R, TGFBR2, and NFAT5 (Figure 4D). Consistent with a lack of direct interaction between these genes as predicted bioinformatically (Figure 3B), knockdown of each central node gene did not affect the expression of the other central node genes (Figure S4C). However, we noted that one direct input to IGF2R is DPP4, which has two predicted miR-211 binding sites (TargetScan context
score = 0.14 and 0.08). Knockdown of the peripheral node gene DPP4, but not ADAM19 (another predicted miR-211 target gene), phenocopied the effect of increased miR-211 on melanoma invasiveness (Figure 4D and Figures S4D and S4E). Interestingly, knockdown of DPP4 also led to reduced expression of IGF2R, TGFBR2, and NFAT5 (Figure S4D). These data demonstrate the high-predictive index of the melanoma metastasis network and provide a mechanistic basis for reduced IGF2R expression upon increased miR-211 expression. Together, our data indicate that increasing miR-211 directly represses NFAT5 and TGFBR2, and indirectly represses IGF2R, leading to the reduction of the invasive potential of malignant melanomas. Here we report an intronic miRNA that assumes a tumor suppressive function previously ascribed to its host gene. Reducing miR-211 but not melastatin levels is sufcient to convert a noninvasive melanoma to an invasive melanoma. Conversely, increasing miR-211 but not melastatin is sufcient to convert an invasive melanoma to a noninvasive melanoma. Previous observations comparing primary melanomas identied reduced expression of melastatin in distant metastasis (Duncan et al., 1998, 2001) and an inverse correlation between melastatin expression and tumor thickness (Deeds et al., 2000). Our data raise the possibility, however, that these observations may have been due to altered expression levels of miR-211. Because miRNA genes are frequently hosted by protein-coding genes, phenotypes attributed to genetic deletion of protein-coding genes may actually be attributable to abrogated expression of the hosted miRNAs (Moffett and Novina, 2007).
Molecular Cell
A
Kendalls tau coefficient
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1
CDK2
TRPM1 MLANA MITF SILV TYRP1
mRNA Features
IGF2R
NFAT5 TGFBR2
B
3.5 Relative mRNA expression 3 2.5 2 1.5 1 0.5 0
Anti-miR-Ctrl Anti-miR-211 WM3526 1.2 1 0.8 0.6 0.4 0.2 0
Ctrl mimic miR-211 mimic WM1716 1.2 1 0.8 0.6 0.4 0.2 IGF2R TGFBR2 TGFBR2 NFAT5 IGF2R NFAT5 0 A375M
C
Position 114-120 of IGF2R 3' UTR 5' ...UAACAGAAACUUUCAAAAGGGAA... 3' UCCGCUUCCUACUGUUUCCCUU
Ctrl mimic miR-211 mimic
Position 1846-1852 of NFAT5 3' UTR 5' ... CUUCUUGCCGAUAUGAAGGGAAC... 3' UCCGUAUCCUACUGUUUCCCUU
IGF2R
NFAT5
TGFBR2
Position 2324-2330 of TGFBR2 3' UTR 5'...UUCACUACUAUACAUAAAGGGAA... 3' UCCGCUUCCUACUGUUUCCCUU
1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
Relative luciferase activity
wt mut wt mut wt mut IGF2R NFAT5 TGFBR2 3UTR
D
siCtrl siIGF2R siCtrl Invasive cells siIGF2R siNFAT5 siTGFBR2 siADAM19 siDPP4 Invasive cells/field 35 30 25 20 15 10 5 0 siNFAT5 siTGFBR2 siADAM19 siDPP4
Before scraping
* * * *
Figure 4. miR-211 Targets Multiple Genes Regulating Melanoma Metastasis

(A) Statistical association analysis of previously published mRNA proling data (13,211 transcripts) versus miR-211 expression using Kendalls tau. mRNA names are sorted by Kendalls tau coefcient and periodically listed on the horizontal axis. Magnitude of coefcient measures the degree of association, and a negative coefcient denotes an inverse association. Representative melanoma-relevant genes directly correlated with miR-211 are highlighted in red: MLANA (t = 0.831), MITF (t = 0.821), CDK2 (t = 0.776), SILV (t = 0.738), and TYRP1 (t = 0.627). Predicted miR-211 target genes are inversely correlated with miR-211 and are highlighted in green: IGF2R (t = 0.297), NFAT5 (t = 0.357), and TGFB2R (t = 0.620). (B) miR-211 represses endogenous IGF2R, NFAT, and TGFBR2 expression in melanoma cells. Transfection of a miR-211 antagomir into WM3526 derepresses IGF2R, NFAT5, and TGFBR2 expression compared to a control antagomir (Anti-miR-Ctrl). Transfection of a miR-211 mimic into WM1716 or A375M represses IGF2R, NFAT5, and TGFBR2 expression compared to a control miRNA mimic (Ctrl mimic). n = 3; means SEM. (C) NFAT5 and TGFBR2 but not IGF2R are direct miR-211 target genes. Schematic presentation of predicted miR-211 target sites identied in the IGF2R, NFAT5, or TGFBR2. Numbers indicate positions of miR-211 binding sites on target mRNA 30 UTRs. HeLa cells were transfected with luciferase constructs possessing wild-type (WT) or miR-211 binding site mutant (mut) 30 UTRs of IGF2R, NFAT5, or TGFBR2 and with a miR-211 mimic or a control miRNA mimic (Ctrl Mimic). n = 5; means SEM. (D) Knockdown of IGF2R, NFAT5, or TGFBR2 phenocopies the effect of increasing miR-211 on melanoma invasive activity. WM1716 cells were transfected with a control siRNA (siCtrl), or siRNAs against indicated genes and subjected to matrigel assays. Invasive cells were counted (top panels) and are quantitated (graphs, right). The bottom panels demonstrate the membranes prior to scraping (n = 3; means SEM).
TGFB has been implicated in the metastatic spread of melanoma and other cancers (Giampieri et al., 2009; Pinner et al., 2009; Van Belle et al., 1996). TGFB signaling has also been shown to promote hypopigmentation (Martinez-Esparza et al.,
1997; Pinner et al., 2009), downregulation of MITF (which controls the melastatin/miR-211 locus [Miller et al., 2004]), disruption of cell-cell adhesion, single-cell invasion to escape the primary tumor, and invasion of blood vessels (Giampieri
Molecular Cell
et al., 2009). Therefore, increased TGFB signaling through increased TGFBR2 expression might partially explain increased melanoma invasion and motility when miR-211 levels are low. To our knowledge, NFAT5 and IGF2R have not been previously linked to melanoma metastasis. Several other genes represented in the melanoma metastasis network are predicted targets of miR-211 and demonstrate reduced expression upon increased miR-211 (Figure 4 and data not shown). These genes may participate in miR-211-regulated melanoma invasion or subsequent steps of melanoma metastasis. Complex phenotypes such as changes in cell morphology, motility, and invasiveness are rarely regulated by a single gene. Because a single miRNA may target thousands of genes, it is possible that altered expression of a single miRNA can regulate complex phenotypes. It has been suggested that therapeutic manipulation of a single miRNA may result in the regulation of many genes and may affect the normalization of a disease phenotype (Wurdinger and Costa, 2007). Our data suggest that modulating miR-211 or its target gene levels may pave the way toward new therapies for metastatic melanoma.
EXPERIMENTAL PROCEDURES Cell Culture HeLa cells and A375M cells were cultured in DMEM medium supplemented with 10% FBS and 1% penicillin-streptomycin. MSTCs were selected from cryopreserved collections at the Wistar Institute (62 lines), the University Hospital of Zurich (24 lines), and the Dana-Farber Cancer Institute (DFCI, 37 lines). All MSTCs were cultured in RPMI medium (MediaTech) supplemented with 10% fetal bovine serum (MediaTech) and 1% penicillin/streptomycin/glutamine (Invitrogen), except for DFCI MSTCs, which were cultivated in DMEM (MediaTech) supplemented with 10% serum. For some Wistar lines, to enhance cell adherence, tissue culture dishes were rst coated with 1% porcine gel solution (Sigma). Primary human melanocytes were isolated and grown from neotal foreskins as described (McGill et al., 2002). Primary melanocytes between passages 2 and 5 were stimulated with 20 uM forskolin (Sigma Aldrich) for indicated times after an overnight starvation in F10 medium supplemented only with penicillin/streptomycin/ glutamine. For microarrays, primary melanocytes were stimulated for 16 hr with forskolin. 501mel human melanoma cells (gift from Dr. Ruth Halaban, Yale University) were maintained in Hams F-10 media (Mediatech) supplemented with 10% fetal bovine serum (FBS, Sigma Aldrich) and 1% penicillin/ streptomycin/glutamine (PSQ). UACC62 human melanoma cell line was obtained from NCI and maintained in RPMI-1640 media (Mediatech) supplemented with 10% fetal bovine serum (FBS) (Sigma Aldrich) and 1% penicillin/streptomycin/glutamine (PSQ). Genome-wide miRNA Screen A miRNA library of 300 precursor miRNAs (Ambion, Inc.) was transfected in triplicate (100 nM) using siPORT NeoFX into A375M cells, seeded in 24-well plates. SiPORT NeoFX is a lipid transfection agent consisting of a mixture of lipids that spontaneously complex small interfering RNAs and facilitates its transfer to the cells. After 24 hr, cells were serum starved overnight. 1 3 105 cells were added to invasion chambers coated with Matrigel (BD Biosciences), and cells were allowed to invade for 16 hr toward media containing 10% FBS. Cells remaining on the top side of the membrane were removed using a cotton swab, and invading cells were counted in ve elds per insert. The library experiment was performed in duplicate. No cell toxicity from the transfection agent was detected. Transfection conditions were optimized with a miRNA negative control (AM17110, Ambion, Inc.) and miR-182 as a positive control. For validation, the 29 precursor miRNAs identied from the primary screen were transfected into A375M cells and seeded in 24-well plates, and the invasion assay was repeated. Each miRNA was transfected in triplicate.
RNA Purication and qRT-PCR Total RNA was puried using TRIzol reagent (Invitrogen) according to the manufacturers instructions, followed by treatment with RNase-free DNase (QIAGEN). Dried RNA pellets were resuspended in appropriate volumes of DEPC/ddH2O. RNA was quantied by measuring OD260/280. For qRT-PCR analysis of mature miR-211 or miR-107, 10 ng total RNA was used in a TaqMan miRNA assay according to the manufacturers protocols (Applied Biosystems). Mature miR-211 expression was normalized to expression of RNU48, also detected by TaqMan assay (Applied Biosystems). For qRT-PCR analysis of TRPM1, NFAT5, IGF2R, TGFBR2, and other genes, RNA was subjected to one-step qRT-PCR using a QuantiTect RT-PCR kit (QIAGEN) and iQ SYBRgreen Supermix (Bio-Rad). Primer sequences and manufacturers are listed in Table S3. All experiments are n R 3. Standard error of the mean (SEM) is presented. Oligonucleotide Transfection MiRNA mimics, antagomirs, or siRNAs were transfected into MSTCs using HiPerFect according to the manufacturers protocols (QIAGEN). The sequences and manufacturers of the miR-211 mimic, miR-211 antagomir, control miRNA mimic, control antagomir, TRPM1, NFAT5, IGF2R, TGFBR2, DPP4 and ADAM19 siRNA, and control siRNA are listed in Table S4. Cells were transfected twice with 100 pmol of oligonucleotide per well (0.5 3 106 cells) at 24 hr intervals. Transfected MSTCs were assayed 48 hr after the second transfection. Invasion and Migration Assays MSTCs were transfected with oligonucleotides as described above. Fortyeight hours after the second transfection, cells were serum starved overnight. Fifty thousand cells were added in duplicate to invasion chambers coated with Matrigel (BD Biosciences). Cells were allowed to invade for 9 hr toward media containing 10% FBS. Cells remaining on the top side of the membrane were removed using a cotton swab, and invading cells were xed and stained with 40 -6-diamidino-2-phenylindole (DAPI, Vector Laboratories Inc.). In all assays, ten elds per insert were photographed, and scored SEM was measured. Cell Growth and Metabolic Rate Assay Cells (104/well) were plated in multiple 24-well plates. The following day was considered day 0. For analysis of cell growth/number, cells were xed using 10% ethanol/10% acetic acid and stained with 0.2% crystal violet xative. Following washes with water, the plates were dried. Each 24-well plate was quantied by crystal violet staining. The crystal violet dye was redissolved in xation solution. One hundred microliters of the redissolved dye was used for color measurement, in duplicate. Color intensity was measured at OD595 in a 96-well plate reader and normalized to day 0 as 100%. For metabolic rate tests, 5 3 104 cells were cultured in 96-well plates, and cell proliferation assays were performed 48 hr after siRNA, miRNA, or control nucleic acid transfection. The cell proliferation reagent WST-1 (Roche) was added and cell proliferation was measured according to the manufacturers instructions (Roche). All experiments are n R 3, and SEM is presented. Microscopy Real-Time Cell Mobility A ZeissLSM510META confocal microscope was used. Analysis of cell migration speed was performed as described (Pinner et al., 2009) by tracking phasecontrast movies of primary melanoma cell lines plated on collagen-matrigel gels (Hooper et al., 2006) spread on 24-well glass-bottomed plates (MatTek corporation). Multiposition imaging was conducted over 10 hr, and images were acquired at 10 min intervals. Mobility of Cherry/GFP-positive cells (>25 cells) was measured from each movie. Tracks were calculated using ImageJ and Excel software. Gene Network Analysis Gene networks were constructed and identied important hubs using Ingenuity Gene Network Analysis. Specically, we created a melanomametastasis gene list including common genes derived from a literature-based melanoma gene list and a literature-based metastasis gene list. These literature-based gene lists were created using IPA software analysis (Ingenuity,
Molecular Cell
Inc.). Using this melanoma metastasis gene list, we performed gene network analysis. Pathways of highly interconnected genes were identied by statistical likelihood using the equation ! f1 X CG; iCN G; s i Score = log10 1 CN; s i=0 where N is the number of genes in the network of which G are central node genes, for a pathway of s genes of which f are central node genes. C(n,k) is the binomial coefcient. Central nodes are considered genes with very high ux capacity (total number of incoming and outgoing links higher than 9). The ux capacity score for TGFBR2 is 10, for NFAT5 is 9, and for IGF2R is 12. The nodes with ux capacity score lower than 9 were considered peripheral nodes. We considered statistically signicant gene networks those with a p value < 1010. Gray color indicates genes included in the melanomametastasis gene list. Solid lines indicate direct gene interactions, while dashed lines indicate indirect gene interactions. Hierarchical Clustering of Genomic Proles An input matrix was generated by calculating the base-2 logorithm of normalized mRNA expression proles. The input matrix was visualized in TIGR MeV v4.1 (http://www.tm4.org/mev.html). Both the gene tree and the sample tree were generated using Euclidean distance and average linkage. Unsupervised Rank Correlation of Microarrays To identify concentration-dependent relationships involving miR-211, rigid nonparametric association statistical analyses were performed. Computations were done using scripts written in Python (http://www.python.org/) and the SciPy module (http://www.scipy.org/). A matrix was generated with each column of the input representing mRNA expression values from a single culture and each row of the input representing expression values for each gene. A copy of the miR-211 expression array was then paired with each row of the input matrix, and the association between each pair was evaluated by calculating the Kendalls tau rank coefcient. Class-Dependent Identication of Marker Genes MSTCs were separated into three classes based on miR-211 expression level: high, medium, and low. Only the high and low classes are used in the actual calculation. For each gene, the degree of difference in mRNA expression between the high and low classes were measured using pairwise, two-sided t test and median values in each class. The calculations were done with 1000 iterations and smooth p values through the ComparativeMarkerSelection module in the GenePattern software suite (http:// genepattern.broad.mit.edu/; version 3.1.1). Genes were ranked by t test score. A negative score denotes an inverse relationship between a genes mRNA expression and miR-211 expression, whereas a positive score denotes a direct relationship. Combination Analysis of Correlation with miRNA and Target Prediction Computationally predicted mRNA targets for Homo sapiens miR-211 (hsamiR-211) were identied using a database from Memorial Sloan-Kettering Cancer Center (http://www.microrna.org/) as the rst input for this analysis. The second input comprised a list of genes whose differential mRNA expressions had strong negative correlation with miR-211 expression from unsupervised rank correlation analysis. The two inputs were combined to identify overlaps. Plasmid Construction and Stable Cell Line Establishment PLKO.1 vector was modied by replacing the puromycin resistance cassette with a cassette coding for luciferase, mCherry, and puromycin using BamHI and KpnI restriction sites. This vector was used for miRNA overexpression. For cDNA overexpression, the U6 promoter of the vector described above was replaced by a HPGK promoter using SanDI and AgeI. For establishement of stable cell lines, MSTCs were seeded in a 6-well dish and transfected the modied PLKO.1 vectors using FugeneHD (Roche). After
2 days, the cells were grown with 1 mg/ml puromicyn. After 46 weeks, the cells were used for experiments. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, four gures, four tables, six movies, and Supplemental References and can be found with this article online at doi:10.1016/j.molcel.2010.11.020. ACKNOWLEDGMENTS We would like to gratefully acknowledge Dr. Meenhard Herlyn for supplying several melanoma short-term cultures for these studies. We also wish to thank Dr. Andrew Kung and Dr. Hans Wildund for insightful discussions. This work was funded by National Institutes of Health (NIH) grant AR043369 (D.E.F.), the Melanoma Research Alliance (D.E.F), the Adelson Medical Research Foundation (D.E.F.), The US-Israel Binational Science Foundation (D.E.F.), the Doris Duke Medical Foundation (D.E.F.), the Doris Duke Medical Foundation (C.D.N.), and by American Cancer Society (C.D.N.). S.L. was supported by NRSA Fellowship (NIH/NCI T32 CA070083). Received: July 20, 2010 Revised: September 27, 2010 Accepted: October 21, 2010 Published online: November 24, 2010 REFERENCES Deeds, J., Cronin, F., and Duncan, L.M. (2000). Patterns of melastatin mRNA expression in melanocytic tumors. Hum. Pathol. 31, 13461356. Duncan, L.M., Deeds, J., Hunter, J., Shao, J., Holmgren, L.M., Woolf, E.A., Tepper, R.I., and Shyjan, A.W. (1998). Down-regulation of the novel gene melastatin correlates with potential for melanoma metastasis. Cancer Res. 58, 15151520. Duncan, L.M., Deeds, J., Cronin, F.E., Donovan, M., Sober, A.J., Kauffman, M., and McCarthy, J.J. (2001). Melastatin expression and prognosis in cutaneous malignant melanoma. J. Clin. Oncol. 19, 568576. Filipowicz, W., Bhattacharyya, S.N., and Sonenberg, N. (2008). Mechanisms of post-transcriptional regulation by microRNAs: are the answers in sight? Nat. Rev. Genet. 9, 102114. Gaur, A., Jewell, D.A., Liang, Y., Ridzon, D., Moore, J.H., Chen, C., Ambros, V.R., and Israel, M.A. (2007). Characterization of microRNA expression levels and their biological correlates in human cancer cell lines. Cancer Res. 67, 24562468. Giampieri, S., Manning, C., Hooper, S., Jones, L., Hill, C.S., and Sahai, E. (2009). Localized and reversible TGFbeta signalling switches breast cancer cells from cohesive to single cell motility. Nat. Cell Biol. 11, 12871296. He, L., Thomson, J.M., Hemann, M.T., Hernando-Monge, E., Mu, D., Goodson, S., Powers, S., Cordon-Cardo, C., Lowe, S.W., Hannon, G.J., and Hammond, S.M. (2005). A microRNA polycistron as a potential human oncogene. Nature 435, 828833. Hoek, K.S., Schlegel, N.C., Brafford, P., Sucker, A., Ugurel, S., Kumar, R., Weber, B.L., Nathanson, K.L., Phillips, D.J., Herlyn, M., et al. (2006). Metastatic potential of melanomas dened by specic gene expression proles with no BRAF signature. Pigment Cell Res. 19, 290302. Hooper, S., Marshall, J.F., and Sahai, E. (2006). Tumor cell migration in three dimensions. Methods Enzymol. 406, 625643. Huang, Q., Gumireddy, K., Schrier, M., le Sage, C., Nagel, R., Nair, S., Egan, D.A., Li, A., Huang, G., Klein-Szanto, A.J., et al. (2008). The microRNAs miR-373 and miR-520c promote tumour invasion and metastasis. Nat. Cell Biol. 10, 202210. Hunter, J.J., Shao, J., Smutko, J.S., Dussault, B.J., Nagle, D.L., Woolf, E.A., Holmgren, L.M., Moore, K.J., and Shyjan, A.W. (1998). Chromosomal
Molecular Cell
localization and genomic characterization of the mouse melastatin gene (Mlsn1). Genomics 54, 116123. Krutzfeldt, J., Rajewsky, N., Braich, R., Rajeev, K.G., Tuschl, T., Manoharan, M., and Stoffel, M. (2005). Silencing of microRNAs in vivo with antagomirs. Nature 438, 685689. Levy, C., Khaled, M., and Fisher, D.E. (2006). MITF: master regulator of melanocyte development and melanoma oncogene. Trends Mol. Med. 12, 406414. Lin, W.M., Baker, A.C., Beroukhim, R., Winckler, W., Feng, W., Marmion, J.M., Laine, E., Greulich, H., Tseng, H., Gates, C., et al. (2008). Modeling genomic diversity and tumor dependency in malignant melanoma. Cancer Res. 68, 664673. Ma, L., Teruya-Feldstein, J., and Weinberg, R.A. (2007). Tumour invasion and metastasis initiated by microRNA-10b in breast cancer. Nature 449, 682688. Ma, L., Young, J., Prabhala, H., Pan, E., Mestdagh, P., Muth, D., TeruyaFeldstein, J., Reinhardt, F., Onder, T.T., Valastyan, S., et al. (2010). miR-9, a MYC/MYCN-activated microRNA, regulates E-cadherin and cancer metastasis. Nat. Cell Biol. 12, 247256. Marson, A., Levine, S.S., Cole, M.F., Frampton, G.M., Brambrink, T., Johnstone, S., Guenther, M.G., Johnston, W.K., Wernig, M., Newman, J., et al. (2008). Connecting microRNA genes to the core transcriptional regulatory circuitry of embryonic stem cells. Cell 134, 521533. Martinez-Esparza, M., Jimenez-Cervantes, C., Beermann, F., Aparicio, P., Lozano, J.A., and Garcia-Borron, J.C. (1997). Transforming growth factorbeta1 inhibits basal melanogenesis in B16/F10 mouse melanoma cells by increasing the rate of degradation of tyrosinase and tyrosinase-related protein-1. J. Biol. Chem. 272, 39673972. McGill, G.G., Horstmann, M., Widlund, H.R., Du, J., Motyckova, G., Nishimura, E.K., Lin, Y.L., Ramaswamy, S., Avery, W., Ding, H.F., et al. (2002). Bcl2 regulation by the melanocyte master regulator Mitf modulates lineage survival and melanoma cell viability. Cell 109, 707718. Medina, P.P., Nolde, M., and Slack, F.J. (2010). OncomiR addiction in an in vivo model of microRNA-21-induced pre-B-cell lymphoma. Nature 467, 8690.
Miller, A.J., Du, J., Rowan, S., Hershey, C.L., Widlund, H.R., and Fisher, D.E. (2004). Transcriptional regulation of the melanoma prognostic marker melastatin (TRPM1) by MITF in melanocytes and melanoma. Cancer Res. 64, 509516. Moffett, H.F., and Novina, C.D. (2007). A small RNA makes a Bic difference. Genome Biol. 8, 221. ODonnell, K.A., Wentzel, E.A., Zeller, K.I., Dang, C.V., and Mendell, J.T. (2005). c-Myc-regulated microRNAs modulate E2F1 expression. Nature 435, 839843. Ozsolak, F., Poling, L.L., Wang, Z., Liu, H., Liu, X.S., Roeder, R.G., Zhang, X., Song, J.S., and Fisher, D.E. (2008). Chromatin structure analyses identify miRNA promoters. Genes Dev. 22, 31723183. Pinner, S., Jordan, P., Sharrock, K., Bazley, L., Collinson, L., Marais, R., Bonvin, E., Goding, C., and Sahai, E. (2009). Intravital imaging reveals transient changes in pigment production and Brn2 expression during metastatic melanoma dissemination. Cancer Res. 69, 79697977. Takamizawa, J., Konishi, H., Yanagisawa, K., Tomida, S., Osada, H., Endoh, H., Harano, T., Yatabe, Y., Nagino, M., Nimura, Y., et al. (2004). Reduced expression of the let-7 microRNAs in human lung cancers in association with shortened postoperative survival. Cancer Res. 64, 37533756. Tavazoie, S.F., Alarcon, C., Oskarsson, T., Padua, D., Wang, Q., Bos, P.D., Gerald, W.L., and Massague, J. (2008). Endogenous human microRNAs that suppress breast cancer metastasis. Nature 451, 147152. Van Belle, P., Rodeck, U., Nuamah, I., Halpern, A.C., and Elder, D.E. (1996). Melanoma-associated expression of transforming growth factor-beta isoforms. Am. J. Pathol. 148, 18871894. Wurdinger, T., and Costa, F.F. (2007). Molecular therapy in the microRNA era. Pharmacogenomics J. 7, 297304. Yanaihara, N., Caplen, N., Bowman, E., Seike, M., Kumamoto, K., Yi, M., Stephens, R.M., Okamoto, A., Yokota, J., Tanaka, T., et al. (2006). Unique microRNA molecular proles in lung cancer diagnosis and prognosis. Cancer Cell 9, 189198.

CELL

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

CELL

Enviado por

Direitos autorais:

Formatos disponíveis

Molecular Cell

Figure 1. Coactivators and Corepressors Are Histone-Modifying Enzymes

constitutively condensed chromatin

-globin domain HS4

Figure 1. Epigenetic Dysfunction in the b-Globin Locus of rev1 DT40

rev1 Cambridge rev1 Munich

H3K9/K14ac Fold enrichment relative to HS4

1.0 0.8 0.6 0.4 0.2 0

H3K4me3 Fold enrichment relative to HS4

Relative HpaII protection

600 500 400 300 200 100 0

Fold increase in -globin mRNA relative to WT

Figure 2. Derepression of r-Globin Expression in rev1 Cells

Fold increase in -globin mRNA relative to WT

Gap Length (bp)

Length of nucleosome tract in which marks are lost

Gap length (nucleosomes)

0.2 0.4 0.6 0.8 probability of stall at fixed site

G4 on leading strand template ggggagtaaaagggagcggggtgctgggg

G4 on lagging strand template

Figure 5. Derepression of Silenced Loci in rev1 Cells Is Associated with G4 DNA

LYSC expression (fold change relative to wild type)

kbp >20 5.3

5 MAR CpG island

lys rev c +/G 1

G4 unwinding + REV1 at the fork

G4 unwinding + PCNA-ubiquitin or HR-dependent gap filling

copying of preexisting marks to adjacent nucleosomes

Figure 1. Mfd and UvrA Proteins

Figure 3. Effect of Mfd Derivatives on TCR In Vitro

Figure 4. Effect of Mfd Derivatives In Vivo

Figure 5. Interactions with UvrA

Figure 6. Effect of UvrA Derivatives on TCR In Vitro

Figure 7. Activity of UvrA D290-400 in Incision and DNA Binding Assays

Figure 3. Details of the Spn1-Spt6 Interface

Figure 4. Mutations at the Spn1-Spt6 Interface Disrupt Binding In Vitro

Figure 6. Spt6 Binds Nucleosomes and Is Competed by Spn1

K1 = KD and RTlnKA = DH TDS: A

Figure 1. Overexpression of JMJD2A Allows Faster Progression through S Phase

Figure 2. JMJD2A/JMJD-2 Expression Levels Alter Sensitivity to the Replication Inhibitor HU

Figure 3. In Vivo Labeling of DNA Synthesis in Wild-Type and jmjd-2/ Worms

Figure 5. JMJD2A Overexpression Increases Chromatin Accessibility

Figure 6. Functional Antagonism between JMJD2A/JMJD-2 and HP1g/HPL-2

Figure 7. In Vivo Antagonism between JMJD-2 and HPL-2

IP: -HA IB: -Taf1 IB: -Pax3

Figure 1. Taf1 Regulates Pax3 Protein Levels and Binds to Pax3

1.2 Relative Protein Level

Control Time (hrs) 0 Taf1 Pax3 DsRed 6 12 24 0

1.0 0.8 08 0.6 0.4 0.2 0.0 0 6 12

+ATP Taf1-Ub Taf1 Ub1 / Pax3 Ub1 / Pax3 Pax3

-Ub -Taf1 -Ub

Chase Time (Hours)

siRNA Control 60 15 30 45 60 Taf1 Pax3 Pax7 GAPDH

Control siRNA Taf1 siRNA

Non-ub biquitinated Pax3/Total Pax3 3

1.0 0.8 0.6 0.4 0.2 0.0 Cont siRNA Taf1

1.0 0.8 0.6 0.4 0.2 0.0

Figure 2. Taf1 Is Both Sufcient and Necessary for Pax3 Monoubiquitination

Re elative Taf1 protein level

+ Taf1 Control + Taf1V1049D +Taf1R1070P

Relative Protein Leve el

+ Taf1V1049D Time (hrs) Taf1 Pax3 DsRed 0 6 12 24 0

1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 6 12

Chase Time (Hours)