doi:10.1016/j.jmb.2006.03.

065

J. Mol. Biol. (2006) 359, 840847

Cortactin Binding to F-actin Revealed by Electron Microscopy and 3D Reconstruction

Kiran Pant1, David Chereau2, Victoria Hatch1, Roberto Dominguez2 and William Lehman1*
Department of Physiology & Biophysics, Boston University School of Medicine, 715 Albany Street, Boston, MA 02118 USA Boston Biomedical Research Institute, Watertown, MA 02472, USA
2 1

Cortactin and WASP activate Arp2/3-mediated actin lament nucleation and branching. However, different mechanisms underlie activation by the two proteins, which rely on distinct actin-binding modules and modes of binding to actin laments. It is generally thought that cortactin binds to mother actin laments, while WASP donates actin monomers to Arp2/3-generated daughter lament branches. Interestingly, cortactin also binds WASP in addition to F-actin and the Arp2/3 complex. However, the structural basis for the role of cortactin in lament branching remains unknown, making interpretation difcult. Here, electron microscopy and 3D reconstruction were carried out on F-actin decorated with the actin-binding repeating domain of cortactin, revealing conspicuous density on F-actin attributable to cortactin that is located on a consensus-binding site on subdomain-1 of actin subunits. Strikingly, the binding of cortactin widens the gap between the two long-pitch lament strands. Although other proteins have been found to alter the structure of the lament, the cortactin-induced conformational change appears unique. The results are consistent with a mechanism whereby alterations of the F-actin structure may facilitate recruitment of the Arp2/3 complex to the mother lament in the cortex of cells. In addition, cortactin may act as a structural adapter protein, stabilizing nascent lament branches while mediating the simultaneous recruitment of Arp2/3 and WASP.
q 2006 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: actin; Arp2/3; cortactin; WASP; electron microscopy

Introduction
The cortical actin cytoskeleton is a dynamic structure that is essential for vital cellular functions as diverse as cell motility, adhesion, endocytosis and cytokinesis.1 Different types of cytoskeletal assemblies, including lamellipodia, lopodia, stress bers and focal adhesions, form by recruiting actin laments in response to various receptor-mediated signaling cascades, typically involving Rho family GTPases and receptor tyrosine-kinases. These pathways modulate the activities of numerous actin-binding proteins, which control lament assembly and inter-lament architecture. Thus, actin-binding proteins connect extra- and intra K.P. & D.C. contributed equally to the work. Abbreviations used: EM, electron microscopy; WASP, WiskottAldrich syndrome protein; IHRSR, iterative helical real space reconstruction. E-mail address of the corresponding author: wlehman@bu.edu
0022-2836/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.

cellular stimuli to cytoskeleton modeling and reorganization. A key participant in this scheme is the Arp2/3 complex, which mediates the nucleation and branching of actin laments at the leading edge of the cell. The activity of the Arp2/3 complex is stimulated by lament nucleation promoting factors (NPFs), including members of the extensively investigated WiskottAldrich syndrome protein (WASP) family.13 These proteins share a C-terminal WASP-homology 2 (WH2) domain adjacent to a central-acidic region (C-A); WH2-CA represents the minimal sequence needed for activation of the Arp2/3 complex by WASP.4 Cortactin, an additional activator of the Arp2/3 complex, also has attracted considerable attention.5,6 Cortactin is an F-actin-binding protein that appears to bind to the Arp2/3 complex and specically activate Arp2/3 lamentogenesis at the cell cortex, but by a mechanism distinct from that of WH2-C-A-containing NPFs. Like most cytoskeletal proteins, cortactin is built from several distinct domains.710 Its N-terminal

3D-EM of Cortactin on F-actin

841

acidic (NTA) domain binds to the Arp2/3 complex and contributes to its activation.5,6 The 19DDWE23 sequence, within this region, resembles 498DEWD501 in the acidic region of WASP and mutations in the sequence5,11,12 perturb the activation of the Arp2/3 complex. However, further analogy between cortactin and WASP is limited, since cortactin lacks a WH2, G-actin-binding domain. Instead, a series of 6.5 tandem repeats (37 amino acid residues per full repeat) located C-terminal to the NTA region, form an F-actin binding module. Because the protein sequence is unique to cortactin, the entire modular ensemble is referred to as the cortactin-repeating domain.13 However, it appears that only the fourth in the series of six repeats and its associated anking sequences are needed for binding to F-actin in vitro.14 Nevertheless, the NTA region, as well as the cortactin-repeating domain, is required for cortactin-activation of the Arp2/3 complex5, and cortactin variants that lack either of these distinct domains fail to localize at the cell periphery.14 The cortactin molecule also contains three additional domains, which may be needed to bind to other protein partners or respond to tyrosine phosphorylation. These consist of an a-helical spacer, followed by a proline-rich motif and a Src homology (SH3) domain at the proteins C-terminus. Pro-rich sequences, which are abundant among cytoskeletal proteins, often mediate the binding of proteins with SH3 or WW signaling adapter domains and prolin; Ser and Tyr phosphorylation sites within this region may regulate the activity of cortactin. In fact, cortactin was initially characterized as a substrate of Src kinase, possibly connecting tyrosine kinase signaling and cytoskeleton organization.79 Interestingly, the C-terminal SH3 domain of cortactin binds to several targets,7 including N-WASP,15,16 hence linking cortactin and WASP function.12 While the importance of cortactin as determinant in cytoskeleton dynamics is emerging, little is known about the structural interactions between cortactin and actin laments; thus the behavior and action of cortactin cannot be adequately described. Here, we addressed this deciency by carrying out electron microscopy and 3D reconstruction of F-actin complexed with a construct representing the actin-binding repeating domain of cortactin. In addition to determining its binding site on F-actin, we found that the six-module cortactin-repeating domain modied the global structure of the actin lament. Such a change in structure might mark a lament as a target for Arp2/3-binding or stabilize preformed laments.

Figure 1. Assessing the purity and F-actin-binding activity of the cortactin six-repeat region. The cortactinrepeating domain runs as a single band on SDS-PAGE and co-sediments with F-actin after ultracentrifugation. As indicated, 0, 2.5, 5 or 10 mM puried cortactinrepeating domain construct (residues 83306) incubated with 0 or 10 mM F-actin (in 10 mM TrisHCl (pH 7.5), 1 mM MgCl2, 50 mM KCl, 1 mM EGTA, 0.2 mM ATP, 0.2 mM DTT) for 30 min at room temperature was centrifuged at 400,000g for 30 min. (To preclude measuring non-specic binding, higher ratios of cortactin to actin were not used.) The pellet was gently washed and resuspended in a volume of fresh buffer equal to that of the supernatant. Equal aliquots of supernatant (S) and pellet (P) were analyzed by SDS-PAGE followed by Coomassie blue staining. Binding was quantied by densitometric scanning and plotted (see inset).

that should maximize binding of the construct on laments (see Materials and Methods and Figure 1). Electron micrographs of negatively stained preparations showed that the decorated laments were well dispersed and approximately 10% wider than undecorated control F-actin (Figure 2). When compared to control F-actin, the actin subunit structure of the decorated laments appeared less well dened, due to the binding of additional protein, yet the laments showed no obvious visual evidence of the shape and orientation of cortactin. To detect the binding of cortactin and its impact on F-actin structure with greater clarity, image processing and 3D reconstruction were carried out. 3D Reconstruction of F-actincortactin Electron micrographs of F-actin complexed with two different preparations of the cortactin construct were analyzed independently. Since each data set yielded virtually the same results, the information was combined. Density maps of reconstituted laments were calculated from the averages of the Fourier transform layer-line data (not shown). All the maps obtained showed typical actin subdomain structure (Figure 3(a) and (d)). When compared with maps of F-actin controls (Figure 3(b) and (e)),

Results and Discussion

Electron microscopy of F-actincortactin complexes F-actin was complexed with a construct containing the 224 amino acid residue long repeating domain of cortactin under conditions

842

3D-EM of Cortactin on F-actin

Figure 2. Electron micrographs of F-actincortactin. Negative staining of (a) F-actin complexed with the sixrepeat cortactin actin-binding domain and (b) control Factin laments (no cortactin); note the wider diameter of the decorated laments. The scale bar represents 50 nm.

extra density attributable to cortactin (not present in controls) was observed on the surface of actin subdomain-1 (Figure 3(a) and (d)). The boundaries of the cortactin density were not easily outlined because the extra density in averaged data localized very close to the surface of actin. However, difference density maps, calculated by subtracting reconstructions of F-actin controls from those of decorated laments, clearly outlined cortactin connected to actin subdomain-1 near the junction of subdomains-1 and 3 (Figure 3(c) and (f)). The attached density extended over subdomain-1 toward subdomain-2 of the same actin monomer (i.e. in the direction of the pointed end of the lament). The difference density was statistically signicant at greater than the 99.9% condence level. However, without further information on the 3D volume of a protein such as the cortactin construct studied here, it is difcult to delineate the exact molecular boundaries of corresponding densities in moderate (21 to 23 A) resolution reconstructions, particularly for density masses that show no characteristic asymmetries as in those generated in this study. No crystal structures

of the full-length cortactin-repeating domain (or ones for individual repeat modules) are available as standards to dene the edges of cortactin in our maps. Thus while the region of actin that is in contact with cortactin has been localized, the exact dimensions and shape of the cortactin construct itself is uncertain. The cortactin density detected in the reconstruction was weaker than that observed for most other actin-binding proteins. It is unlikely that this simply resulted from cortactin assuming a narrow, extended conformation along F-actin, since any one of the six cortactin repeats should have contributed more density than was observed. In fact, prior to averaging, reconstructions of single laments showed variably shaped, strong cortactin density centered on subdomain-1 of actin, and no evidence of an extended conformation in contact with multiple actin monomers along laments (Figure 4). Reconstructions of single laments of control F-actin (no cortactin) never showed spurious extra densities on subdomain-1. Given the well-dened localization but inconsistent shape of the cortactin density in single laments decorated with cortactin, it is likely that only a small portion of the cortactin construct associated specically with actin (presumably at its site of binding) and that the remainder of the construct assumed different conformations and lacked binding partners. This conclusion is in agreement with the suggestion that only repeat 4 and adjacent sequences represent the primary binding patch.9 Hence it follows that only the consistently localized densities were retained after helical averaging while the disordered domains were averaged out. In fact, maps plotting the variance associated with the F-actincortactin reconstructions showed that the greatest amount of variability occurs at the site of cortactin binding and in the cleft between azimuthally neighboring actin monomers (Figure 3(h)). This conclusion is supported by IHRSR reconstruction of F-actincortactin laments divided into short segments (2842 nm), where again a reconstruction of all the data showed weak cortactin density, but individual segments could be separated into classes yielding reconstructions with variably shaped but stronger cortactin density (Figure 5). Conformational exibility of the cortactin-repeat domain may explain our inability to crystallize it for X-ray analysis. Fitting cortactin on the atomic model of F-actin To further characterize the cortactin-binding site on F-actin, reconstructions were aligned to the atomic model of F-actin.17 Fitting the atomic model within the envelope formed by the actin part of the reconstruction of F-actincortactin dened the cortactinactin binding interface and the location of the cortactin difference densities (Figure 3(i)). The procedure indicated that cortactin densities closely approached amino acid residues 338 to 348 of actin (highlighted in yellow in Figure 3(i)). This cluster of amino acids represents

3D-EM of Cortactin on F-actin

843

Figure 3. Helical reconstructions of F-actincortactin complexes. (a)(c) Surface views of reconstructions (displayed at 9s above the mean) representing data averaged from (a) 33 cortactin-decorated F-actin laments, (b) 31 control F-actin laments (actin subdomains numbered on one monomer) and (c) difference densities (rose) attributable to cortactin (displayed at 2s above the mean), superimposed on the control F-actin map shown in (b). Note the cortactin density associated with actin subdomain-1 in (a) indicated by the white arrow, and the deep cleft between azimuthally adjacent actin monomers in the cortactin-decorated actin (compare the cleft, indicated by open arrows, in (a) and (b)). (d)(h) Transverse sections (z-sections) through 3D reconstructions of laments: z-sections of (d) cortactin-decorated actin and (e) control F-actin; the extra density attributable to cortactin is indicated by a white arrow and the prominent cleft between actin monomers noted by an open arrow; actin subdomains numbered in (e). (Since azimuthally opposite actin monomers along F-actin are staggered, transverse sections through subdomains-1 and 3 of one actin monomer slice through subdomains-2 and 4 of the adjacent monomer). (f) Difference densities (rose) attributable to cortactin, superimposed on the control F-actin map. (g) Negative difference densities (gold) associated with the cortactininduced deepening of the cleft between adjacent actin monomers (computed by subtracting densities in the map of Factincortactin in (d) from those in control F-actin (e)) and superimposed on control F-actin. Both the positive and

844 the front of a hydrophobic pocket on actin that is a hot-spot for binding of many proteins to actin laments.18,19 Effects of cortactin on the structure of F-actin The cleft located between azimuthally adjacent actin monomers along thin laments is noticeably deeper in the cortactin-decorated actin laments than it is in control F-actin (Figure 3(a), (b), (d) and (e)). This change in lament conformation is consistent with the negative difference density associated with maps of F-actincortactin subtracted from ones of F-actin (Figure 3(g)), which was found to be signicant at greater than the 99.9% condence level. This change in the overall structure of F-actin induced by cortactin is not accompanied by a change in the 1678 average angular rotation between adjacent actin monomers, a parameter that denes the twist of the actin helix (Table 1), although the variance associated with the mean rotation appears to be greater than for F-actin controls. Interestingly, variance mapping shows that densities contributing to the edges of the cleft between azimuthally adjacent actin monomers are especially variable (Figure 3 (h)). We presume that any perturbation in the azimuthal connectivity between actin subunits (such as those caused by cortactin binding) may promote deviation from the standard helical twist of laments. Thus, a cortactininduced local or transient instability in the actin helix together with a more open cleft between neighboring actin monomers may facilitate the binding of other proteins such as Arp2/3 to actin (see below). Signicantly, the effect on F-actin structure differs from that brought about by colin, which changes the lament twist angle, and where laments decorated with colin, unlike those with cortactin, have shorter actin crossovers.20

3D-EM of Cortactin on F-actin

Conclusions
Compared to WASP, cortactin is a relatively weak activator of the Arp2/3 complex,6 yet it still could have an impact on actin-lament branching in at

least three different ways: by strictly serving as an activator of Arp2/3, by recruiting N-WASP to actin laments, and by stabilizing pre-existing lament branches. Information on the structural impact of the cortactin-repeating domain on F-actin determined here, together with biochemical data, are consistent with these mechanisms and provide a framework for addressing the role of cortactin in branching and nucleation. WASP-family proteins stabilize the active conformation of the Arp2/3 complex and, via their WH2 domain, contribute actin monomers to Arp2/3generated daughter lament branches growing off pre-existing mother actin laments. However, in contrast to cortactin, WASP family proteins lack an F-actin binding domain. Whether adapter proteins link WASP to F-actin is not certain. Interestingly, N-WASP is one of several targets of the SH3 domain of cortactin.15 Hence an important role of cortactin in Arp2/3 complex activation might be to position WASP on mother laments. Thus, cortactin at the cell cortex could rst bind to actin laments and recruit the Arp2/3 complex by means of its NTA domain, but then in addition, also recruit WASP-family proteins. In fact, under certain circumstances, cortactin and N-WASP appear to bind to the Arp2/3 complex simultaneously, contributing synergistically to actin assembly.12 By changing the local structure of the mother lament, cortactin may initially mark actin laments found at the cell cortex for the binding of the Arp2/3 complex at specic locations and then latch onto WASP. Alternatively, the cortactin-stimulated structural effects themselves may explain the ability of cortactin to stabilize WASP/Arp2/3-induced lament branches and to inhibit lament-debranching.6,21 Indeed, after binding to the mother lament and drawing in Arp2/3 via its NTA domain, cortactin may serve as a major stabilizer of actinlament branches at the cell cortex. Although considerable information has emerged recently about the structure of F-actin at the point of contact with the Arp2/3 complex,21,22 our current knowledge about the 3D arrangement of this interaction is still insufcient to help to rigorously discriminate between the possible mechanisms proposed here.

negative difference densities shown were statistically signicant at condence levels greater than 99.9%. (h) Maps of variance (magenta) associated with contributing density points in the F-actincortactin reconstruction were computed and superimposed on a map of F-actincortactin. Regions of relatively high variance depicted were present in parts of the reconstruction where cortactin bound (arrow), as well as at the edges of actin subunits bordering the cleft between neighboring actin monomers (asterisks) and in the center of the lament where azimuthally adjacent actin subunits bind to each other. (i) Mapping the binding site of cortactin on the atomic structure of F-actin. Here the atomic model of F-actin17 was tted to the EM reconstruction and the relative position of the cortactin difference density superimposed over one actin monomer. The actin is shown as an a-carbon ribbon (cyan) and the cortactin density is highlighted in red. Residues 338 to 348, which represent the front of a hydrophobic pocket on actin that is a hot spot for the binding of many actin-binding proteins19 is highlighted in yellow. Reconstructions in (a)(c) were aligned to each other and sections in (d)(f) are at the same axial position and have the same orientation. The average phase residual (JGSD), a measurement of the geometrical agreement among laments generating reconstructions, was 50.3(G6.5)8 for cortactin-decorated F-actin (33 laments) and 46.9(G6.7)8 for control F-actin control (31 laments). The average up-down phase residuals (DJGSD), a measure of lament polarity, were 27.5(G6.9)8 and 26.0(G5.6)8 for decorated and undecorated F-actin, respectively. These values are comparable to those found previously for other reconstructions of F-actin27 and among the best for any reconstructions of decorated F-actin that we have generated.

3D-EM of Cortactin on F-actin

845

Figure 4. Surface maps of reconstructions of single (unaveraged) F-actincortactin laments. Examples of reconstructions of individual laments decorated with the cortactin actin-binding domain showing prominent extra density consistently bound over actin subdomain-1 (indicated by asterisk). The extra density is stronger than that of cortactin in maps of averaged data, but its shape and orientation is not uniform. In some cases, the cortactin density is suggestive of a wedge (black arrows), prying-open the cleft between neighboring actin monomers. In order to determine if any one of the arrangements depicted above best characterizes the binding of cortactin, a data set, consisting of 393 cortactin-decorated laments cut into 20,699, 28 nm long segments, was tted against each of the individual reconstructions shown. Cross-correlation values were used to sort the segments according to their closest match to one or another reference reconstruction.29,33 (An undecorated F-actin reference was also used to lter out poorly decorated segments.) No one reference model dominated the selection process, and data split as follows: 2287, 2676, 2192, 1028, 1534, 727, 1100, 4403 segments tted best to reconstructions shown in (a) to (h), respectively, and 2314 tted best to the F-actin reconstruction; the polarities of 2438 segments were ambiguous and were discarded from the analysis. The sorted data were separated and IHRSR reconstruction carried out on each, which generated reconstructions reecting cortactin density on actin characteristic of the respective references (e.g. see Figure 5). Iteration of the IHRSR process (20 rounds), using reconstructions of the sorted data as references for the next round of tting, did not yield any particular solution that differed from those in the original maps or result in any obvious enhancement of the cortactin signal.

Figure 5. IHRSR single particle reconstructions of F-actincortactin laments. IHRSR reconstruction was carried out on 393 cortactin-decorated laments cut into 20,699, 28 nm long segments. (a) Reconstructions generated from the average of the full data set showed weak extra density attributable to cortactin and associated with subdomain-1 of actin (arrow). EM segments were also sorted to various reference models representing helical reconstructions of individual laments (see the legend to Figure 4) and to a reconstruction of F-actin. Reconstructions of sorted data showed variably shaped cortactin density centered on actin subdomain-1. (b) Cortactin density was strongest for EM segments that tted best to models corresponding to reconstructions shown in Figure 4(a) and (c) above. (c) Cortactin density was not detected in EM segments (only 13% of the data) that tted best to the model of F-actin (no cortactin). Thus, a small percentage of F-actin apparently was unoccupied after decoration with the cortactin-repeating domain.

Materials and Methods

Protein expression and purication A fragment of murine cortactin corresponding to amino acid residues 83 to 306 (actin-binding repeats 1 to 6) was generated by polymerase chain reaction (PCR) using a complementary DNA (pXZ112) as a template and primers 5 0 -GGGAATTCCATATGGGCTATGGCGGGAAGTTCG and 5 0 -GCTCATGAATTCTCAGAATCCTTTGGCATAG Table 1. Twist of the F-actin helix
Angular rotation between adjacent actin monomers (8) 166.85G2.29 166.91G3.35 Number of actin monomers per 3608 turn 2.158 2.157

Further investigation is required to understand the impact of cortactin-induced changes on actin lament structure and the role of these perturbations in the multifaceted activation of Arp2/3.

Sample F-actin control F-actincortactin

846
TCTTGCTGG. The PCR product was inserted in between NdeI and EcoRI cleavage sites in plasmid pTYB12 (New England Biolabs, MA), downstream of an inteinchitinbinding domain purication tag. The vector was then introduced for expression into Escherichia coli strain BL21DE3 (Invitrogen, CA), and the cells were grown at 37 8C in LB medium. Protein expression was induced when the culture reached an absorbance (A600 nm) of 0.60.8 by addition of 0.5 mM isopropyl-1-thio-b-D-galactosidase (IPTG), and carried out for 5 h at 20 8C. Bacteria were then harvested by centrifugation and resuspended in 20 mM Hepes (pH 7.65), 0.5 M NaCl, 1 mM EDTA, 20 mM PMSF. After sonication, the lysate was claried by centrifugation at 50,000g for 30 min at 4 8C, and the supernatant was loaded onto a chitin afnity column (New England Biolabs, MA). The column was then extensively washed using the buffer described above, with addition of 1 M NaCl, followed by 0.3% Triton X-100, and nally 0.3 M NaCl. Self-cleavage of the intein was induced by overnight incubation with 50 mM dithiothreitol at room temperature. After elution from the chitin column, the protein was loaded onto a Mono S column (Amersham Biosciences, NJ), equilibrated with 20 mM Hepes (pH 7.0), 25 mM NaCl, 1 mM EDTA, 20 mM PMSF. The protein was eluted using a 25 mM500 mM NaCl gradient and concentrated to approximately 1 mg/ml using a Centricon device (Millipore). The puried cortactin construct ran as a single band on SDS-PAGE (Figure 1), indicating that the protein was homogeneous. Co-sedimentation with F-actin showed an approximate binding ratio of one of the 224 residue long cortactinrepeating domains for every two actin molecules at close to saturation (10 mM actin:10 mM cortactin construct). This is consistent with previous results indicating that only the fourth repeat and closely neighboring anking segments of the repeating-domain construct bind to F-actin,14 and that it is unlikely that the other ve repeat modules of the full-length fragment bind to neighboring actin subunits along laments. Electron microscopy and image reconstruction Negative staining Samples of F-actin alone (10 mM) (prepared as described23) or ones decorated with preparations of the cortactin construct (20 mM) were suspended in solutions consisting of 100 mM NaCl, 2 mM MgCl2, 1 mM EGTA, 1 mM NaN3, 5 mM Pipes, 5 mM Na-phosphate buffer (pH 7.1) at room temperature (w25 8C) and then diluted tenfold in the same buffer containing 4 mM cytochrome c (adding a larger 5 to 10:1 molar excess of cortactin to actin did not change the amount of binding detected, indicating that laments were close to saturated under the conditions used). Samples were immediately applied to carbon-coated electron microscope grids and negatively stained as described.24 Grids were dried at 80% relative humidity, which, in addition to the inclusion of cytochrome c,25 promoted even spreading of the stain. Electron micrographs of decorated laments were recorded on a Philips CM120 electron microscope at 60,000! magnication under low dose conditions (w12 eK/A) at a defocus of 0.5 mm. 3D reconstruction Micrographs were digitized using a Zeiss SCAI scanner at a pixel size corresponding to 0.7 nm in the laments.

3D-EM of Cortactin on F-actin

Well-preserved regions of the laments were selected and straightened as described.26,27 Helical reconstruction was carried out by standard methods.28 Iterative Helical Real Space Reconstruction (IHRSR29) of lament segments divided into 28 nm lengths was used to sort the data into various possible conformational modes and to determine the variance in the twist of thin lament segments (here laments were divided into 42 nm lengths). The statistical signicance associated with densities in reconstructions was evaluated from the standard deviations associated with contributing points.30,31 The coordinates of the atomic model of F-actin17 were modeled to the decorated F-actin and F-actin control reconstructions using the program O.32

Acknowledgements
We thank Dr Roger Craig for discussions and advice on electron microscopy and we are grateful to Dr Xi Zhan for his gift of the murine cortactin cDNA. This work was supported by NIH grants (HL36153 to W.L. and GM073791 to R.D.) and a grant from the Muscular Dystrophy Association (to W.L.). EM was carried out in the Core Electron Microscopy facility at the University of Massachusetts Medical School, supported by NIH grant RR08426 (to R.C.)

References
1. Pollard, T. D. & Borisy, G. G. (2003). Cellular motility driven by assembly and disassembly of actin laments. Cell, 112, 453465. 2. Bompard, G. & Caron, E. (2004). Regulation of WASP/WAVE proteins: making a long story short. J. Cell Biol. 166, 957962. 3. Symons, M., Derry, J. M., Karlak, B., Jiang, S., Lemahieu, V., McCormick, F. et al. (1996). Wiskott Aldrich syndrome protein, a novel effector for the GTPase CDC42Hs, is implicated in actin polymerization. Cell, 84, 723734. 4. Machesky, L. M. & Insall, R. H. (1998). Scar1 and the related WiskottAldrich syndrome protein, WASP, regulate the actin cytoskeleton through the Arp2/3 complex. Curr. Biol. 8, 13471356. 5. Uruno, T., Liu, J., Zhang, P., Fan, Y., Egile, C., Li, R. et al. (2001). Activation of Arp2/3 complex-mediated actin polymerization by cortactin. Nature Cell Biol. 3, 259266. 6. Weaver, A. M., Karginov, A. V., Kinley, A. W., Weed, S. A., Li, Y., Parsons, J. T. & Cooper, J. A. (2001). Cortactin promotes and stabilizes Arp2/3induced actin lament network formation. Curr. Biol. 11, 370374. 7. Daly, R. J. (2004). Cortactin signaling and dynamic actin networks. Biochem. J. 382, 1325. 8. Selbach, M. & Backert, S. (2005). Cortactin: an Achilles heel of the actin cytoskeleton targeted by pathogens. Trends Microbiol. 13, 181189. 9. Weed, S. A. & Parsons, J. T. (2001). Cortactin: coupling membrane dynamics to cortical actin assembly. Oncogene, 20, 64186434.

3D-EM of Cortactin on F-actin

847
22. Volkmann, N., Amann, K. J., Stoilova-McPhie, S., Egile, C., Winter, D. C., Hazelwood, L. et al. (2001). Structure of Arp2/3 complex in its activated state and in actin lament branch junctions. Science, 293, 24562459. 23. Spudich, J. A. & Watt, S. (1971). The regulation of rabbit skeletal muscle contraction. I. Biochemical studies of the interaction of the tropomyosintroponin complex with actin and the proteolytic fragments of myosin. J. Biol. Chem. 242, 48664871. 24. Moody, C., Lehman, W. & Craig, R. (1990). Caldesmon and the structure of smooth muscle thin laments: electron microscopy of isolated thin laments. J. Muscle Res. Cell Motil. 11, 176185. 25. Moore, P. B., Huxley, H. E. & DeRosier, D. J. (1970). Three-dimensional reconstruction of F-actin, thin laments and decorated thin laments. J. Mol. Biol. 50, 279295. 26. Egelman, E. H. (1986). An algorithm for straightening images of curved lamentous structures. Ultramicroscopy, 19, 367374. 27. Hodgkinson, J. L., Marston, S. B., Craig, R., Vibert, P. & Lehman, W. (1997). Three-dimensional image reconstruction of reconstituted smooth muscle thin laments: effects of caldesmon. Biophys. J. 72, 23982404. 28. Owen, C., Morgan, D. G. & DeRosier, D. J. (1996). Image analysis of helical objects: the Brandeis helical package. J. Struct. Biol. 116, 167175. 29. Egelman, E. H. (2000). A robust algorithm for the reconstruction of helical laments using singleparticle methods. Ultramicroscopy, 85, 225234. 30. Milligan, R. A. & Flicker, P. F. (1987). Structural relationships of actin, myosin, and tropomyosin revealed by cryo-electron microscopy. J. Cell. Biol. 105, 2939. 31. Trachtenberg, S. & DeRosier, D. J. (1987). Threedimensional structure of the frozen hydrated agellar lament: the left-handed lament of Salmonella typhimurium. J. Mol. Biol. 195, 581601. 32. Jones, T. A., Zou, J.-Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and location of errors in these models. Acta Crystallog. sect. A, 47, 110119. 33. Pirani, A., Xu, C., Hatch, V., Craig, R. & Lehman, W. (2005). Single particle analysis of relaxed and activated thin laments. J. Mol. Biol. 346, 761772.

10. Wu, H. & Parsons, J. T. (1993). Cortactin, an 80/85kilodalton pp60src substrate, is a lamentous actinbinding protein enriched in the cell cortex. J. Cell Biol. 120, 14171426. 11. Marchand, J. B., Kaiser, D. A., Pollard, T. D. & Higgs, H. N. (2001). Interaction of WASP/Scar proteins with actin and vertebrate Arp2/3 complex. Nature Cell Biol. 3, 7682. 12. Weaver, A. M., Heuser, J. E., Karginov, A. V., Lee, W. L., Parsons, J. T. & Cooper, J. A. (2002). Interaction of cortactin and N-WASp with Arp2/3 complex. Curr. Biol. 12, 12701278. 13. Sparks, A. B., Hoffman, N. G., McConnell, S. J., Fowlkes, D. M. & Kay, B. K. (1996). Cloning of ligand targets: systematic isolation of SH3 domaincontaining proteins. Nature Biotech. 14, 741744. 14. Weed, S. A., Karginov, A. V., Schafer, D. A., Weaver, A. M., Kinley, A. W., Cooper, J. A. & Parsons, J. T. (2000). Cortactin localization to sites of actin assembly in lamellipodia requires interactions with F-actin and the Arp2/3 complex. J. Cell Biol. 151, 2940. 15. Mizutani, K., Miki, H., He, H., Maruta, H. & Takenawa, T. (2002). Essential role of neural WiskottAldrich syndrome protein in podosome formation and degradation of extracellular matrix in srctransformed broblasts. Cancer Res. 62, 669674. 16. Martinez-Quiles, N., Ho, H. Y., Kirschner, M. W., Ramesh, N. & Geha, R. S. (2004). Erk/Src phosphorylation of cortactin acts as a switch onswitch off mechanism that controls its ability to activate N-WASP. Mol. Cell. Biol. 24, 52695280. 17. Lorenz, M., Popp, D. & Holmes, K. C. (1993). Renement of the F-actin model against X-ray ber diffraction data by use of a directed mutation algorithm. J. Mol. Biol. 234, 826836. 18. McGough, A. (1998). F-actin binding proteins. Curr. Opin. Struct. Biol. 8, 166178. 19. Dominguez, R. (2004). Actin-binding proteinsa unifying hypothesis. Trends Biochem. Sci. 29, 572578. 20. McGough, A., Pope, B., Chui, W. & Weeds, A. (1997). Colin changes the twist of F-actin: implications for actin lament dynamics and cellular function. J. Cell Biol. 138, 771781. 21. Egile, C., Rouiller, I., Xu, X. P., Volkmann, N., Li, R. & Hanein, D. (2005). Mechanism of lament nucleation and branch stability revealed by the structure of the Arp2/3 complex at actin branch junctions. PLoS Biol. 3, 19021909.

Edited by M. Moody (Received 18 January 2006; received in revised form 17 March 2006; accepted 31 March 2006) Available online 4 May 2006