Você está na página 1de 10

Review TRENDS in Biochemical Sciences Vol.30 No.

10 October 2005

Structural proteomics of
macromolecular assemblies using
oxidative footprinting and mass
spectrometry
Jing-Qu Guan1,2 and Mark R. Chance1
1
Case Center for Proteomics and Mass Spectrometry, 930 BRB, Case Western Reserve University, 10900 Euclid Avenue, Cleveland,
OH 44106, USA
2
Current address: Phytoceutica, 5 Science Park, Suite 13, New Haven, CT 06511, USA

Understanding the composition, structure and dynamics oligomerization, disorder, or conformational flexibility
of macromolecules and their assemblies is at the fore- [4,5]. Oxidative protein-footprinting approaches using
front of biological science today. Hydroxyl-radical- mass spectrometry (MS) have been specifically developed
mediated protein footprinting using mass spectrometry to analyze macromolecules, and their complexes and their
can define macromolecular structure, macromolecular conformational changes [6–10]. Hydroxyl radicals can
assembly and conformational changes of macromol- react with proteins to yield stable oxidative modifications
ecules in solution based on measurements of reactivity of solvent-accessible amino acid side chains. Following
of amino acid side-chain groups with covalent-modifi- proteolysis, liquid-chromatography-coupled MS (LC-MS)
cation reagents. Subsequent to oxidation by reactive analysis is performed to quantify side-chain reactivity,
oxygen species, proteins are digested by specific pro- and tandem-MS (MS/MS) analysis is used to identify
teases to generate peptides for analysis by mass spectro- the modification site(s). The side-chain-reactivity data,
metry. Accurate measurements of side-chain reactivity which are informative with respect to side-chain solvent
are achieved using quantitative liquid-chromatography- accessibility, can be combined with computational and
coupled mass spectrometry, whereas the side-chain bioinformatics methods in a ‘hybrid’ approach for macro-
sites within the macromolecular probes are identified molecular-structure analysis [5,11,12]. Here, we summar-
using tandem mass spectrometry. In addition, the use of ize the overall development to date of oxidative protein
footprinting data in conjunction with computational footprinting using MS and provide some insight into likely
modeling approaches is a powerful new method for future directions for research in this area.
testing and refining structural models of macromol-
ecules and their complexes.
Classical footprinting approaches
The term ‘footprinting’ refers to assays that examine
macromolecular structural changes by determining the
Introduction solvent accessibility of the backbone, bases or side chains
Understanding the molecular structure and dynamics of of macromolecules using their sensitivity to chemical or
macromolecules at high resolution and with high through- enzymatic cleavage or modification reactions [6,13,14].
put is a topic of great importance in biology. Nuclear mag- The technique of footprinting was originally developed for
netic resonance (NMR) and crystallographic approaches the analysis of the binding of protein to DNA; it provides a
are the foundation of rapid progress in this area. Access to sensitive and specific structure probe that uses a non-
genome sequences and cloning resources from an ever- specific enzymatic nuclease to cleave the DNA backbone,
increasing number of organisms and allied high-through- the labeled cleavage fragments are analyzed using gel
put structure studies are likely to enable resolution of the electrophoresis [15,16]. Segments of the DNA made
structure of most protein domains in the near future [1]. inaccessible to solvent by protein binding preclude
However, the machinery of eukaryotic cell biology involves cleavage, interaction sites are thus mapped by examin-
multi-domain proteins that interact in large complexes as ation of the ‘missing’ gel bands. Hydroxyl radicals have
molecular machines [2,3]. Understanding how these proven to provide high-resolution probes for examining
domains interact is crucial in understanding their the structure and conformational changes of both DNA
function. However, many biochemical systems of interest and RNA, yielding quantitative thermodynamic and
are not amenable to structural analysis by crystallo- kinetic data [17,18], even for reactions on millisecond
graphy or NMR because of factors such as size, timescales [19].
Corresponding author: Chance, M.R. (mark.chance@case.edu). This approach was extended to proteins; end-labeled
Available online 29 August 2005 macromolecules were subjected to cleavage and the
www.sciencedirect.com 0968-0004/$ - see front matter Q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tibs.2005.08.007
584 Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005

cleavage products examined by sodium dodecyl sulfate of ligand binding or macromolecular interactions can be
(SDS)-gel methods to reveal ligand-dependent changes in inferred.
backbone conformation [20–23]. Tethering the chemical- The overall footprinting experiment is summarized
cleavage reagents simplifies data interpretation because Figure 1. Proteins are exposed to hydroxyl radicals that
the cleavage events are limited to solvent-accessible sites react with side-chain sites that are both reactive and
that are proximal to the tethered reagent; this method has solvent accessible (Figure 1b). Subsequent to oxidation,
been used to map sites that are proximal to protein proteolysis using site-specific proteases is used to generate
interfaces [24,25]. However, the inefficiency of protein well-defined peptide species (Figure 1c). Coupled high-
cleavage provided by hydroxyl radicals, the limited performance liquid chromatography (HPLC)-MS is used
coverage of protease-cleavage reagents and the low to separate the peptide mixture and to determine the
resolution of protein gel technologies have limited masses of peptides; this approach is similar to methods
progress. developed for deuterium-exchange MS studies [31–33].
However, in contrast to deuterium exchange, the pro-
duction of stable modifications by hydroxyl-radical
MS approaches to structure determination exposure enables a wide range of sample sizes, proteases
Gentle ionization methods for MS, including electrospray and solution conditions to be examined because compli-
ionization (ESI) and matrix-assisted laser desorption cations of back exchange are alleviated. In addition, the
ionization (MALDI), have been coupled with sensitive stable modification of side chains enables a specific probe
detection methods to provide high-resolution tools for site to be easily identified using well-established MS/MS
mass analysis of peptides and proteins at the femtomole sequencing methods [26,27,29,30]. Within the HPLC chro-
level [26,27]. Protein-footprinting approaches have been matogram, the peak areas that correspond to the mass-to-
developed that use oxidation of proteins by hydroxyl charge values of the modified and unmodified peptides in
radicals, followed by proteolysis and mass-spectroscopic the total ion chromatogram can be extracted and
analysis [7,28–30]. Changes in reactivity of side chains integrated (Figure 1d). The ratio of the peak area of the
with †OH are revealed by this analysis such that unmodified peptide compared with the sum of the peak
conformational changes in protein structure as a function areas from the modified and unmodified peptides provides

(a) (b) (c)

(f) (e) (d)

50k
DA FLEK LPEN 100
y6
40k
Relative abundance

% Unmodified

Relative abundance

b7
30k y3
y5 y7 90
20k
b3 b4
b9
10k y4 b5 b6
y8
y2 b8 80 14 15 16 17 18 19 20 21
0 0 50 100 150 200
200 400 600 800 1000 Elution time (min)
X-ray exposure time (ms)
m/z

Figure 1. Protein footprinting using hydroxyl radicals and MS. An unoxidized protein (a), which can be in any of several conformational states, is exposed to a hydroxyl radical
source; the modifiable and solvent-accessible amino acids are oxidized (b). Proteolysis of the protein yields peptide fragments (c), which are subject to HPLC-MS for
separation and identification. (d) The extracted ion chromatograms for oxidized and unoxidized peptides 118–125 of actin are shown; the unoxidized, singly oxidized, doubly
oxidized and triply oxidized species have successively reduced retention times. Comparison of the integrated areas of these peaks gives the extent of oxidation; the fraction
(or percentage) modified is plotted versus the exposure time to provide the dose-response curves. A first-order function is fit to the dose-response curve to provide rate data
(e). The variation of the oxidation rates of a peptide reflects the change of solvent accessibility of its probe residues. The resolution of this approach is maximized when
MS/MS sequencing is performed to identify the specific oxidation sites (f).

www.sciencedirect.com
Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005 585

a quantitative measure of oxidation extent expressed as


Box 1. Chemical basis of hydroxyl-radical production using
fraction (or %) unmodified, where the value for the
Fenton and radiolysis
unexposed protein sample is assigned to 1.0. Multiple
experiments that progressively increase the dose of FeðIIÞ  ðEDTAÞ C H2 O2 / FeðIIIÞ  ðEDTAÞ C †OH C OHK (Eqn 1)

radicals are carried out and a plot of the unmodified hv


H2 O 
/ H2 OC
C eK
H2 O
dry 
/  H3 OC C †OH C eK
aq (Eqn 2)
fraction versus time of exposure gives a dose-response
curve. This curve is fit to a first-order function to provide Equation 1 shows the reaction of peroxide with Fe (II). The result is
the modification rate (in Figure 1e). The protein is then oxidation of iron and formation of hydroxyl radical. To maximize the
production of hydroxyl radical, sodium ascorbate is also added to
examined under an alternative condition in which the reaction, this reduces the iron (III) product and regenerates iron
changes in conformation are expected. Peptides that are (II). The concentrations of the three species [e.g. Fe–ethylenedi-
seen to experience changes in the observed oxidation rate aminetetraacetic acid (EDTA), hydrogen peroxide (H2O2) and
have side-chain residues that change conformation in ascorbate] can be varied to optimize the generation of hydroxyl
response to the change in solution conditions or ligand. radicals, providing appropriate radical doses for either protein-
backbone cleavage or side-chain modifications [9,21]. A clever
The specific oxidation sites, which represent the struc- modification of the approach has used the intrinsic metal ion of a
tural probes in the experiment, are further identified by metalloprotein as the radical source; addition of peroxide and
MS/MS sequencing of the oxidized peptides (Figure 1f). ascorbate in defined quantities has been shown to specifically target
Several aspects of the experimental protocols designed ligands of the metal and residues within the metalloprotein active
site [64,65]. In addition to Fe–EDTA, other aminocarboxylic chelate–
for protein footprinting assure that reliable structure data
iron complexes can be used. Chelates might not be necessary if the
relevant to function are provided. First, the extent of metal ions are Cu(II) instead of Fe(II). For instance, Cu(II) with
oxidation is kept to the minimum required for adequate ascorbate and H2O2 has been shown to successfully oxidize peptides
detection of oxidized products. The oxidation events might bound to Cu(II) [64,65]. Specific conjugation of the chelate system to
be expected to induce protein unfolding [7,9,34] and this a protein via the reaction of Fe(S)-(p-bromoacetamidobenzyl) EDTA
(Fe–BABE) with cysteine residues can provide a site source of
has, in fact, been observed at sufficiently high doses of
radicals. Upon activation with hydrogen peroxide and ascorbate, the
radicals [9]. Second, for quantitative determination of Fe–BABE reagent generates hydroxyl radicals that modify and/or
peptide rate constants, measurement of the relative cleave adjacent sites [24,25]. Radiolysis using X-ray or g-ray sources
amounts of modified and unmodified peptide products in generates †OH (Equation 2), whereby photons interact with H2O and
the same experiment provides consistent quantitation generate a water ion (H3OC) and an electron [47,66]. H3OC reacts
with another water molecule within several nanoseconds, and the
[28–30]. Third, dose-response curves generated for quan- products are †OH, a hydronium ion, and a ‘hydrated’ electron. The
titative footprinting must indicate a (pseudo) first-order electron rapidly reacts with dissolved oxygen in the solution to form
process extrapolated to zero fraction modified, this assures superoxide. Low-flux sources of g-rays, such as cobalt sources
that the reactivity of particular sites are not changing [8,46,47], require timescales of exposure similar to Fenton and
owing to the oxidation process itself [28–30]. In addition, photolysis of peroxide (i.e. minutes). Synchrotron sources, owing to
their brightness, permit exposure times of milliseconds. The
quantitative footprinting methods that examine loss of the advantages include the possibility of time-resolved studies [19,67]
unmodified fraction emphasize the interrogation of an and a time frame of exposure that is potentially faster than
intact ensemble [28–30]. Although these particular conformational changes due to oxidative damage. Photolysis of
refinements are specifically adapted to protein footprint- peroxide could provide accelerated cleavage, given sufficient light
intensity using pulsed lasers.
ing as described here, they are derived from a detailed
knowledge and respect for safeguards that have evolved
throughout the long history of development of ‘foot-
derivatives. Cleavage of the protein results from hydroxyl-
printing’ research [6,14–18]; such safeguards ensure
radical abstraction of backbone Ca–H bonds. A Ca centered
that the structural and biochemical information provided
radical is generated that reacts with oxygen and leads to
by footprinting methods are reliable.
main-chain cleavage. Because the rate of this initial
abstraction (at w107 MK1sK1) is much slower than the
Chemistry of hydroxyl radical production and protein
rate of reaction with most side chains (at 108–109 MK1sK1)
modification
[44,45], reactions at side-chains sites are preferred and
Hydroxyl radicals that are suitable for footprinting
experiments have been generated using Fenton reagents cleavage reactions are not observed except at high doses of
[9,21], from photo-oxidation of peroxide [12,35], using hydroxyl radicals [7,9,21,24]. Cross-linking, because it
electrical discharge [36–39] and from radiolysis of water requires the properly oriented bimolecular association of
[8,29,30]. In Fenton chemistry, hydroxyl radicals are radicals on separate macromolecules, is also not usually
generated by the reaction of a redox-active metal ion observed owing to the low micromolar protein concen-
complex with hydrogen peroxide, whereas, in radiolysis, trations typically used.
the radicals are generated by fragmentation of water The amino acid side-chain products typically observed
(see Box 1). in protein-footprinting experiments are listed in Table 1
The modification of peptides and proteins by oxidative [46–50]. The most frequent oxidation events result in
species has been examined in the pioneering studies of formation of alcohol or aldehyde (or ketone) groups with
Garrison, Stadtman and others [40–43]. These studies associated mass increases of C16 and C14 Da, respec-
show that the oxidation of peptides and proteins by tively. Figure 2 shows oxidation mechanisms for selected
reactive oxygen species can lead to oxidation of amino residues [40,41,46–48]. For example, the abstraction of
acid residue side chains, cleavage of peptide bonds or hydrogen from aliphatic side-chain carbon atoms by †OH,
formation of covalent protein–protein cross-linked and subsequent reactions with oxygen and hydroperoxy
www.sciencedirect.com
586 Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005

Table 1. Primary products and corresponding mass changes for radical results in hydroxylated side chains in addition to
various amino acid side chains when subjected to radiolysisa,b production of aldehydes or ketones. Aromatic residues
Amino Acid Side-chain modification and mass changes also suffer hydroxylation, but by a quite different
Cysteine Sulfonic acid (C48), sulfonic acid (C32), hydroxy mechanism. For acidic residues, initial attack at the
(K16)
Cystine Sulfonic acid, sulfinic acid
aliphatic side chain can lead to elimination of CO2 and the
Methionine Sulfoxide (C16), aldehydes (K32), sulfone (C32) generation of a product with a mass change of K30 Da
Tryptophan Hydroxy (C16, C32, C48 and so on), pyrrole-ring [46], whereas ring opening at histidine provides multiple
opening (C32) products [47]. Multiple oxidations at the same side-chain
Tyrosine Hydroxy (C16, C32 and so on)
Phenylalanine Hydroxy (C16, C32 and so on)
position are frequently seen for highly reactive aromatic
Histidine Oxo (C16), pyrrole-ring opening (K22, K10, C5) and sulfur-containing residues [48,50,51].
Leucine Hydroxy (C16), carbonyl (C14) The relative reactivity of the 20 amino acids and cystine
Isoleucine Hydroxy (C16), carbonyl (C14) approximately parallels their rate of initial reactivity with
Valine Hydroxy (C16), carbonyl (C14)
Proline Hydroxy (C16), carbonyl (C14)
hydroxyl radical [12,47,48,51]. A reactivity order of CysO
Arginine Deguanidination (K43), hydroxy (C16), carbonyl MetOTrpOTyrOPheOcystineOHisOLeu, IleOArg, Lys,
(C14) ValOSer, Thr, ProOGln, GluOAsp, AsnOAlaOGly in
Lysine Hydroxy (C16), carbonyl (C14) MS-based experiments, when the oxidation source is
Glutamic acid Decarboxylation (K30), hydroxy (C16), carbonyl
(C14)
generated by ionizing radiation has been reported [51].
Glutamine Hydroxy (C16), carbonyl Based on these data, alanine and glycine are far too
(C14) unreactive to be useful probes in typical experiments, and
Aspartic acid Decarboxylation (K30), hydroxy (C16) aspartic acid and asparagine are also unlikely to be useful.
Asparagine Hydroxy (C16)
Serine Hydroxy (C16), K2(C16 and loss of H2O)
Although serine and threonine are more reactive than
Threonine Hydroxy (C16), K2(C16 and loss of H2O) proline, which is known to be a useful probe, their
Alanine Hydroxy (C16) oxidation products are not easily detectible [51]. Thus, it
a
For all aliphatic side chains, C14-Da products normally occur much less frequently seems that only 14 of the 20 side chains (plus cystine) are
than C16-Da products.
b
For serine and threonine, only trivial amounts of C16-Da product and K2-Da
likely to be useful in typical experiments. Because these
product are observed, possibly owing to further loss of water. residues comprise w65% of the sequence of a typical

(a)
O O O2 H O O
H HO• H H
N N N N

OH OH
• HOO •
OO H OH
H •
H H
Phe (and other sites of the ring) +16 Da

(b)
O O O
H H H
NH
N α N N H O
N
β HO • O2 H2N NH2
γ H2O
δ • O O O
H O
NH H2O NH HOO . NH
HN H
HN HN
NH2 NH2
NH2 -43 Da

Arg

Ti BS

Figure 2. Mechanisms of side-chain oxidation by hydroxyl radicals. (a) In the case of phenylalanine, addition of †OH to the aromatic side chain creates a hydroxylated radical
species that lacks aromaticity. Reaction with oxygen and re-aromatization (with liberation of hydroperoxy radical) leaves 2-, 3-, or 4-hydroxyphenylalanine. (b) The oxidation
of residues such as arginine initially follows the hydroxylation seen for aliphatic residues, if the initial abstraction is on the b or g carbon, aC16-Da product is generated. If the
(more) reactive g carbon is attacked an elimination reaction ensues, generating a species with a mass change of K43 Da [47].

www.sciencedirect.com
Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005 587

protein, the footprinting approach can provide excellent retention time increases or decreases based on changes in
coverage of solvent-accessible side chains. Methionine is hydrophobicity of the peptide [52]. The peptide ions are
one of the most reactive residues and has been observed to fragmented and analyzed. From a comparison of the two
be oxidized even in cases where it was seen to be solvent MS/MS spectra, the position(s) of the modification product
inaccessible [7,9,30]. Recent data confirm that methionine can be identified. For example, fragment ions retaining
and cysteine (but not the other side chains) suffer the modification will show mass shifts relative to that
significant oxidation from peroxide that is used in Fenton observed for the unmodified peptide.
and photo-oxidation protocols and is a by-product of
radiolysis [49]. However, in footprinting experiments Analysis of protein structure and protein–ligand
using ionizing radiation addition of quenching agents or interactions by footprinting
catalase subsequent to radiolysis eliminates this second- Protein footprinting coupled with mass spectrometry is an
ary oxidation. Thus, methionine can be used as a reliable especially powerful technique when coupled with existing
probe of solvent accessibility under certain protocols [49]. structural data, either experimentally or computationally
Overall, these stable oxidized species are markers of generated to understand ligand-induced conformational
solvent accessibility of the modified side chains. MS changes or to help decide among competing computational
methods provide an approach suitable for readout of the models of structure. Often, structures of individual pro-
data. teins or domains are available or can be reliably modeled
[3], however, the interacting surfaces of their complexes
Detecting oxidation sites by MS/MS are not known. In other cases, owing to lack of structural
The structural resolution of the footprinting approach data and/or unavailability of appropriate templates for
depends on the ability to specifically localize the site of structural modeling, experimental data to assist the
oxidation within a target peptide. To localize the oxidation modeling process are required [12]. In these cases, foot-
site, an oxidized peptide of interest is selected and printing provides a sensitive and accurate probe of
dissociated into fragments using a process known as surface-accessible residues and their changes induced by
collision-induced dissociation (CID). The masses of the complex formation. The following examples illustrate how
fragments are then analyzed to identify the sites of footprinting could reliably answer important structural
oxidation, which are characteristic for the different side questions where other techniques have failed.
chain types (Table 1). Technically, CID in MS is a process
whereby a projectile ion selected by its ratio of mass to Footprinting data defines the binding interface of
charge is dissociated as a result of interaction with a transferrin C-lobe on its receptor
target-neutral species, typically a gas in a collision cell. Although crystallographic structures of isolated transferrin
The dissociation is brought about by conversion of part of (Tf) and its receptor (TfR; Figure 3a,b) have been available
the translational energy of the ion to internal energy in for some time [53,54], the nature of the domain–domain
the ion during the collision. The conditions of dissociation interactions that drive complex formation have been
determine the fragmentation products. Of interest here controversial. Tf is presumed to interact with a highly
are b-type fragment ions, which retain the original N conserved Arg-Gly-Asp (RGD) sequence within the
terminus of the peptide, and y-type fragment ions, which C-terminal helical domain of TfR [55] (Figure 3c). To
retain the original C terminus. A b-type ion that retains probe the interactions of the tight binding Tf C-lobe
only the N-terminal residue would be noted as b1 and its (residues 334–679) with the TfR ectodomain (residues
sister ion with two residues would be b2 and so on. The 121–760), their biologically active 2:2 complex was
mass of b1 provides information on the identity of the examined by footprinting [56]. As outlined in the protocol
amino acid at the N terminus (where the mass is unique of Figure 1, the isolated C-lobe of Tf, TfR and their com-
the amino acid is uniquely identified) and the mass plex were separately subjected to radiolysis and digested
difference between b2 and b1 provides information on the with trypsin. Table 2 shows data for seven of the 63
identity of the second residue from the N terminus. Thus, peptides detected after proteolysis of TfR. These data focus
the sequence of the fragmented peptide can be inferred to on the observed reactivity of side chains in the helical
the extent that sufficient fragment ions are observed. domain of the receptor and show that the binding of the
An example for an actin peptide with the sequence Asp- C-lobe of Tf to TfR protects the residues of several peptides
Ala-Phe-Leu-Glu-Lys-Leu-Pro-Glu-Asn (DAFLEKLPEN) from hydroxyl radical attack. For two peptides that
is shown in Figure 1f. This MS/MS experiment performed include the essential RGD-binding sequence (peptide
on the unmodified peptide can reveal the sequence of 634–646 ends in Arg and peptide 647–651 begins in
the peptides generated by the specific protease used Gly-Asp), the oxidation rates are reduced w95% in the
(Figure 1c) and, thus, identify the location in the chro- complex compared with the oxidation rate observed for
matogram (e.g. the retention time) where peptides of free TfR. Figure 3 illustrates the footprinting data for TfR.
specific sequence are seen to elute (Figure 1d). To Several peptides in the receptor outside the helical domain
determine the site of modification of an oxidized peptide, exhibit enhancements in the rate of modification that are
ions corresponding to a specific oxidation product with consistent with allosteric effects of complex formation.
altered mass relative to the unmodified peptide are Using MS/MS, the sites of modification with altered
selected. These modified peptides will run at a similar reactivity in the complex were identified (Figure 3d).
retention time relative to the unmodified peptide Footprinting, in combination with genetic, biochemical
(Figure 1d), the oxidation mechanism determines if the and other structural data, confirms that the C-lobe of Tf
www.sciencedirect.com
588 Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005

(a) Apical (b)


domain
Helical
domain

Protease-like
domain

(c) Apical
domain (d)
Helical
domain

Met635

Leu619
RGD Phe760 Tyr503 Pro581

Trp740 Phe650

C-terminal
tail
Trp754
Tyr611
Leu505

Protease-like
domain

Figure 3. Footprinting data illustrated on the structure of ectodomain of TfR dimer. Front view (a) and rotation of 908 around the horizontal axis (b). Within the footprinting
experiment, the peptides that are modified and protected upon binding by Tf C-lobe are shown in green; the modified but not protected peptides are red; the detected and not
modified peptides are yellow; peptides not detected are white. The ectodomain of TfR monomer with the crucial RGD sequence labeled (c) and a close-up of the helical and
part of protease-like domain with key residues probed by footprinting labeled (d). Reproduced, with permission, from Ref. [16]. q (1979) American Chemical Society.

binds to the helical domain of the transferrin receptor, covalently linked to AVP via a disulfide bond in addition to
including residues from Leu619 to Phe650 [56]. Confir- charge–charge and hydrogen-bonding interactions [59].
mation of this exact binding mode was subsequently The second co-factor is the viral DNA. The location of the
provided by cryo-EM data [57]. pVIc-binding site on AVP is known [59], but the mode of
DNA binding is not. The AVP structure has two major
Footprinting data provides a model for the adenovirus domains with the active site sandwiched between the
protease–DNA complex domains. The pVIc cofactor interacts with both domains to
The human adenovirus serotype 2 proteinase (AVP) is form an anti-parallel b sheet with strand 5 of domain 1,
required for the synthesis of the infectious virus. and its N terminus interacts with strand 7 from domain 2
Biochemical studies with the recombinant form of AVP (Figure 4).
show that AVP requires two co-factors for maximal Hydroxyl-radical footprinting has been performed for
proteinase activity [58]. One co-factor is the 11-amino AVP, its binary complexes with DNA and the pVIc peptide,
acid residue peptide pVIc, which originates from the and the ternary complex [11,60]. Significant protections
C terminus of the precursor protein pVI. This peptide is that suggest buried surface contacts between DNA and
www.sciencedirect.com
Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005 589

Table 2. Oxidation rate date for selected peptides derived from the transferrin receptor
Domain Receptor Peptidea Rate constant (sK1)b
sequence
Receptor Complex
Protease-like domain and 601–613 LTHDVELNLDYER 0.61G0.04 0.37G0.06
helical domain 2 37 1 2 16
Helical domain 614–623 YNSQLLSFVR 0.69G0.1 0.04G0.02
0 0 65 13
Helical domain 634–646 EMGLSLQW L YSAR 2.13G0.12 0.17G0.04
10 28 6 135 2 76
Helical domain 647–651 GDF FR 1.69G0.1 0.13G0.03
1 103
Helical domain 681–693 VEY H F LSP YVSPK 0.49G0.05 0.42G0.04
38 54 0 1 66 88 23
Helical domain 699–717 HVFWGSGSHTL PALLENLK 0.56G0.05 0.50G0.06
0 7 43 18 3 67 1 16 0
Helical domain 733–760 NQLALATWTIQGAANALSGDVWDIDNEF 0.90G0.06 0.11G0.02
18 70 5 1 211 84
a
For each peptide, the most reactive residues are outlined within the peptide sequence in bold, and the solvent accessibility of the potentially modifiable side chains based on
an analysis of the crystal structure of the isolated domain (e.g. the surface area, in units of Å2) is shown below the one-letter codes.
b
The rate constants shown in bold are significantly different in the cases of Tf or TfR versus their complex (within error).

AVP were observed on each domain at some distance from satisfaction of potential favorable charge and base-
each other. Consistent with these protections, positively stacking interactions; use of conserved residues to provide
charged surface patches were also located in both the trace of the DNA–AVP interaction; and reasonable
domains, providing a facile potential surface for DNA constraints on the bond lengths and angles of the
binding. A hypothesis was formed that DNA could macromolecules. The final model (Figure 4a,b) satisfied
stretch across the surface of AVP to satisfy these these constraints. In particular, the coordinates of the
protections. The DNA–AVP interface was modeled using ternary complex compared with that of the AVP–pVIc
a 12-mer single-stranded DNA (ssDNA) with the sequence binary complex showed DNA-dependent protections from
GACGACTAGGAT. The constraints in the modeling solvent for Phe86, Pro101, Pro183 and Tyr175. These
included: DNA-dependent protection of the probe resi- residues were entirely protected upon complex formation
dues clearly buried in the footprinting experiments; in the footprinting experiments. The model also satisfied

Figure 4. A model for DNA binding to AVP–pVIc complex is shown in two different orientations. The AVP–pVIc complex is shown in ribbon format, where AVP and pVIc are
colored in slate and light pink, respectively. The ssDNA is in stick representation (off-white). The amino acid residues proposed to interact with DNA are shown in dark pink for
basic (arginine, histidine and lysine); cyan is for tyrosine, magenta for phenylalanine from AVP and light pink for phenylalanine from pVIc. The probe residues identified by
MS/MS are also indicated and highlighted with the same colors of the AVP ribbon diagram. The active-site residues and Trp55 are also colored yellow. Reproduced, with
permission, from Ref. [11].

www.sciencedirect.com
590 Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005

multiple base-stacking interactions between highly con- analyzed by LC-MS and LC-MS/MS to examine oxidation
served aromatic residues in the AVP sequence and bases extent and residues modified, respectively (Figures 1d,f).
within the DNA. These results demonstrate how foot- At least 18 oxidation sites were identified in the denatured
printing data can explicitly drive modeling efforts when protein. By contrast, Pro22, Phe33, Phe87 and Phe106
combined with structural data and bioinformatics were not oxidized in the native protein, whereas the other
analysis. residues were still reactive, suggesting that these four
residues are buried in the native structure. A satisfactory,
Computational model of the Sml1 protein is consistent low-energy model was built based on a refinement of the
with footprinting data NMR structure that satisfied the solvent-accessibility
The Sml1 protein from yeast binds to and regulates the data derived from both footprinting and fluorescence
large subunit of the ribonucleotide reductase complex [61]. data. This study provides an excellent example of how
Although NMR studies indicate that SMl1 has two large a footprinting data can be integrated with existing struc-
helices that are oriented in an anti-parallel manner, no tural information to provide more accurate models of
well-defined tertiary structure has been demonstrated for protein structure.
the protein [62]. In the absence of appropriate compara-
tive templates, de novo methods to determine the struc- Concluding remarks
ture of Sml1 were carried out. Hettich and co-workers [12] Quantitative monitoring of side-chain reactivity changes
used oxidative footprinting coupled to MS to examine the of peptides using oxidative footprinting provides a sensi-
validity of both the NMR structure and the models tive probe of solvent accessibility and, therefore, of struc-
generated by de novo methods. ture. Because of its ability to provide structural data in
This work has two innovative features. First, guanidine solution with femtomoles to picomoles of material and its
HCl-denatured Sml1 was used as a control molecule to ability to monitor multiple sites within multi-protein
determine the side chains that are expected to be reactive complexes simultaneously, its use is increasing in several
in a generally solvent-accessible environment. These data laboratories, and creative approaches to extend its use-
were then compared with reactivity data for the native fulness are rapidly being developed. The size limits of the
protein. The residues that were observed to be susceptible technique are currently unknown, but footprinting
to hydroxyl-radical attack in the denatured state but approaches do not have intrinsic limitations with respect
unreactive in the native state were considered to be buried to the size of the macromolecules under study [6,14]. For
in the native protein. Second, the residues determined to example, the transferrin–receptor complex, which has a
be either accessible or buried were analyzed for consist- total molecular weight of w330K (and is effectively a
ency with models of the protein sequence generated by the dimer) was examined with 75% peptide coverage using
Hidden Markov model/Rosetta server [63]. standard electrospray LC-MS instruments [56]. In terms
Sml1 was exposed to hydroxyl radicals generated by photo- of structural resolution, the residues in Table 1 represent
lysis of peroxide and digested with trypsin (Figures 1a–c), 65% of the sequence coverage for proteins; in practice, the
and the oxidation products of specific peptides were observed probes must be both reactive and surface

Box 2. Future directions for oxidative footprinting coupled to mass spectrometry


Time-resolved methods on observed side-chain reactivity. A union of these two methods,
Similar to the developments that led from static nucleic-acid where sites are engineered in proteins (typically by introducing single
footprinting to millisecond time-resolved approaches [18,19], protein cysteine residues) and coupled to Fenton-type reagents, could provide
footprinting needs to be extended such that seconds to milliseconds sources of radicals that could map intra-atomic distances within a
timescale reactions could be routinely monitored. It has been protein and inter-atomic distances in a complex. A challenge of such
demonstrated that protein footprinting, using either cleavage approaches would be to develop calibration experiments so that the
methods [21] or modification [10], can provide reliable biophysical reactivity could be converted into soft distance constraints. If
isotherms in equilibrium experiments. The next step would involve successful, such technologies could be competitive with cross-linking
delivering the radical dose in a short time using an appropriate mixing approaches.
device [18,67]. Fenton approaches are suitable for monitoring
reactions that are on the timescale of minutes; peroxynitrite as a Computational approaches
radical source could reduce the timescales to seconds, and synchro- This review highlights some of the first published attempts to
tron experiments can clearly push the timescale to milliseconds explicitly use solution-footprinting data with computational methods
[18,67]. Alternatively, photolysis of peroxide – if developed with to refine structure or evaluate competing structural models [11,12,60].
modern UV laser technology – could also attain seconds to To make these approaches more valuable, progress in two separate
millisecond timescales. areas is needed. First, although the relationship of solvent accessibility
and side-chain reactivity is demonstrable, its quantitative basis is far
Hydroxyl-radical site sources from well understood. Progress in relating measured changes in
One disadvantage of ‘bulk’ footprinting experiments is that the reactivity to explicit changes in solvent accessibility would provide
footprinting data do not have a 3D context. Tethered footprinting specific constraints for protein modeling. In addition, modeling
approaches developed by Meares and co-workers [24,25] were programs could be improved to enable a more flexible inclusion of
designed to generate radicals from a specific site, their migration to surface accessibility and/or distance constraints, particularly with
adjacent sites provided proximity information and helped to map respect to providing seamless user control over the weighting factors
protein–protein interfaces. However, because these methods used applied to specific constraint terms in the modeling process [12]. As
gels, the mapping was imprecise. A more recent set of experiments by these improvements accrue, oxidative-footprinting approaches using
the Vachet group [64,65] has used the intrinsic metal atom of proteins MS will have an increasingly important role in understanding the
with oxidation and MS to map residues in the metal-binding site based structure and dynamics of macromolecules and their assemblies.

www.sciencedirect.com
Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005 591

accessible. Using state-of-the art instrumentation and binding of Chironomus high mobility group protein 1a (cHMG1a).
chromatography approaches, it is likely that much larger Regions flanking an HMG1 box domain do not influence the bend
angle of the DNA. J. Biol. Chem. 272, 19763–19770
and more complex samples than described here, with 21 Heyduk, E. and Heyduk, T. (1994) Mapping protein domains involved
moderate structural resolution, could be examined with in macromolecular interactions: a novel protein footprinting
the technique. However, there are several areas where approach. Biochemistry 33, 9643–9650
future progress is needed to make protein footprinting a 22 Zhong, M. et al. (1995) A method for probing the topography and
more valuable tool for biophysics research (see Box 2). interactions of proteins: footprinting of myoglobin. Proc. Natl. Acad.
Sci. U. S. A. 92, 2111–2115
23 Shea, M.A. et al. (2000) Proteolytic footprinting titrations for
Acknowledgements estimating ligand-binding constants and detecting pathways of
We thank our colleagues for support and insights, particularly Steve conformational switching of calmodulin. Methods Enzymol. 323,
Almo, Mike Brenowitz, Mike Sullivan, Janna Kiselar, Rutao Liu, Qin He, 254–301
Keiji Takamoto, Guozhong Zu and Sayan Gupta. This research is 24 Datwyler, S.A. and Meares, C.F. (2000) Protein–protein interactions
supported by the Biomedical Technology Program of the National mapped by artificial proteases: where s factors bind to RNA
Institute for Biomedical Imaging and Bioengineering under P41-EB- polymerase. Trends Biochem. Sci. 25, 408–414
01979. 25 Datwyler, S.A. and Meares, C.F. (2001) Artificial iron-dependent
proteases. Met. Ions Biol. Syst. 38, 213–254
26 Ferguson, P.L. and Smith, R.D. (2003) Proteome analysis by mass
References
spectrometry. Annu. Rev. Biophys. Biomol. Struct. 32, 399–424
1 Chance, M.R. et al. (2004) High-throughput computational and
27 Aebersold, R. and Mann, M. (2003) Mass spectrometry-based
experimental techniques in structural genomics. Genome Res. 14,
proteomics. Nature 422, 198–207
2145–2154
28 Goldsmith, S.C. et al. (2001) Synchrotron protein footprinting: a
2 Russell, R.B. et al. (2004) A structural perspective on protein–protein
technique to investigate protein-protein interactions. J. Biomol.
interactions. Curr. Opin. Struct. Biol. 14, 313–324
Struct. Dyn. 19, 405–418
3 Sali, A. et al. (2003) From words to literature in structural proteomics.
29 Guan, J.Q. et al. (2002) Mapping the G-actin binding surface of cofilin
Nature 422, 216–225
using synchrotron protein footprinting. Biochemistry 41, 5765–5775
4 Chance, M.R. et al. (1997) Examining the conformational dynamics of
30 Kiselar, J.G. et al. (2002) Hydroxyl radical probe of protein surfaces
macromolecules with time-resolved synchrotron X-ray ‘footprinting’.
using synchrotron X-ray radiolysis and mass spectrometry. Int.
Structure 5, 865–869
J. Radiat. Biol. 78, 101–114
5 Guan, J.Q. et al. (2004) Synchrotron radiolysis and mass spectro-
31 Katta, V. and Chait, B.T. (1993) Hydrogen/deuterium exchange
metry: a new approach to research on the actin cytoskeleton. Acc.
electrospray ionization mass spectrometry: a method for probing
Chem. Res. 37, 221–229
protein conformational changes in solution. J. Am. Chem. Soc. 115,
6 Guan, J.Q. and Chance, M.R. (2004) Footprinting methods to examine
6317–6321
the structure and dynamics of proteins. In Encyclopedia of Molecular
32 Zhang, Z. and Smith, D.L. (1993) Determination of amide hydrogen
Cell Biology and Molecular Medicine (Meyers, R.A., ed.), pp. 549–568,
exchange by mass spectrometry: a new tool for protein structure
Wiley-VCH
elucidation. Protein Sci. 2, 522–531
7 Maleknia, S.D. et al. (2001) Determination of macromolecular folding
and structure by synchrotron X-ray radiolysis techniques. Anal. 33 Hoofnagle, A.N. et al. (2003) Protein analysis by hydrogen exchange
Biochem. 289, 103–115 mass spectrometry. Annu. Rev. Biophys. Biomol. Struct. 32, 1–16
8 Nukuna, B.N. et al. (2004) Hydroxyl radical oxidation of cytochrome c 34 Chance, M.R. (2001) Unfolding of apomyoglobin examined by
by aerobic radiolysis. Free Radic. Biol. Med. 37, 1203–1213 synchrotron footprinting. Biochem. Biophys. Res. Commun. 287,
9 Sharp, J.S. et al. (2003) Protein surface mapping by chemical 614–621
oxidation: structural analysis by mass spectrometry. Anal. Biochem. 35 Sharp, J.S. et al. (2004) Analysis of protein solvent accessible surfaces
313, 216–225 by photochemical oxidation and mass spectrometry. Anal. Chem. 76,
10 Kiselar, J.G. et al. (2003) Visualizing the Ca2C-dependent activation 672–683
of gelsolin by using synchrotron footprinting. Proc. Natl. Acad. Sci. 36 Maleknia, S.D. et al. (1999) Electrospray-assisted modification of
U. S. A. 100, 3942–3947 proteins: a radical probe of protein structure. Rapid Commun. Mass
11 Gupta, S. et al. (2004) DNA binding provides a molecular strap Spectrom. 13, 2352–2358
activating the adenovirus proteinase. Mol. Cell. Proteomics 3, 950–959 37 Maleknia, S.D. et al. (2004) Photochemical and electrophysical
12 Sharp, J.S. et al. (2005) Photochemical surface mapping of production of radicals on millisecond timescales to probe the
C14S–Sml1p for constrained computational modeling of protein structure, dynamics and interactions of proteins. Photochem. Photo-
structure. Anal. Biochem. 340, 201–212 biol. Sci. 3, 741–748
13 Revzin, A. (1993) Footprinting Techniques for Studying Nucleic Acid– 38 Wong, J.W. et al. (2003) Study of the ribonuclease-S-protein–peptide
Protein Complexes (A Volume of Separation, Detection, and Charac- complex using a radical probe and electrospray ionization mass
terization of Biological Macromolecules), Academic Press spectrometry. Anal. Chem. 75, 1557–1563
14 Takamoto, K. and Chance, M. (2004) Footprinting methods to examine 39 Wong, J.W. et al. (2005) Hydroxyl radical probe of the calmodulin–
the structure and dynamics of nucleic Acids. In Encyclopedia of melittin complex interface by electrospray ionization mass spectro-
Molecular Cell Biology and Molecular Medicine (Myers, R.A., ed.), metry. J. Am. Soc. Mass Spectrom. 16, 225–233
pp. 569–578, Wiley-VCH 40 Garrison, W.M. (1987) Reaction mechanisms in the radiolysis of
15 Galas, D.J. and Schmitz, A. (1978) DNAse footprinting: a simple peptides, polypeptides, and proteins. Chem. Rev. 87, 381–398
method for the detection of protein-DNA binding specificity. Nucleic 41 Garrison, W.M. et al. (1962) Radiation induced oxidation of protein in
Acids Res. 5, 3157–3170 aqueous solution. Radiat. Res. 16, 483–502
16 Schmitz, A. and Galas, D.J. (1979) The interaction of RNA polymerase 42 Swallow, A.J. (1980) Effect of ionizing radiation on proteins RCO
and lac repressor with the lac control region. Nucleic Acids Res. 6, groups, peptide bond cleavage, inactivation, SH oxidation. In
111–137 Radiation Chemistry of Organic Compounds (Swallow, A.J., ed.),
17 Brenowitz, M. et al. (1986) Footprint titrations yield valid thermo- pp. 211–224, Pergamon Press
dynamic isotherms. Proc. Natl. Acad. Sci. U. S. A. 83, 8462–8466 43 Stadtman, E.R. and Berlett, B.S. (1997) Reactive oxygen-mediated
18 Brenowitz, M. et al. (2002) Probing the structural dynamics of nucleic protein oxidation in aging and disease. Chem. Res. Toxicol. 10, 485–494
acids by quantitative time-resolved and equilibrium hydroxyl radical 44 Buxton, G.V. et al. (1988) Critical review of rate constants for reactions
‘footprinting’. Curr. Opin. Struct. Biol. 12, 648–653 of hydrated electrons, hydrogen atoms and hydroxyl radicals in
19 Sclavi, B. et al. (1998) RNA folding at millisecond intervals by aqueous solution. J. Phys. Chem. Ref. Data 17, 513–886
synchrotron hydroxyl radical footprinting. Science 279, 1940–1943 45 Hawkins, C.L. and Davies, M.J. (2001) Generation and propagation of
20 Heyduk, E. et al. (1997) Conformational changes of DNA induced by radical reactions on proteins. Biochim. Biophys. Acta 1504, 196–219
www.sciencedirect.com
592 Review TRENDS in Biochemical Sciences Vol.30 No.10 October 2005

46 Xu, G. and Chance, M.R. (2004) Radiolytic modification of acidic 58 Baniecki, M.L. et al. (2001) Interaction of the human adenovirus
amino acid residues in peptides: probes for examining protein–protein proteinase with its 11-amino acid cofactor pVIc. Biochemistry 40,
interactions. Anal. Chem. 76, 1213–1221 12349–12356
47 Xu, G. et al. (2003) Radiolytic modification of basic amino acid residues 59 McGrath, W.J. et al. (2003) Crystallographic structure at 1.6-Å
in peptides: probes for examining protein–protein interactions. Anal. resolution of the human adenovirus proteinase in a covalent complex
Chem. 75, 6995–7007 with its 11-amino-acid peptide cofactor: insights on a new fold.
48 Maleknia, S.D. et al. (1999) Millisecond radiolytic modification of Biochim. Biophys. Acta 1648, 1–11
peptides by synchrotron X-rays identified by mass spectrometry. Anal. 60 Gupta, S. et al. (2005) Mapping a functional viral protein in solution
Chem. 71, 3965–3973 using synchrotron X-ray footprinting technology. Synch. Rad. News
49 Xu, G. et al. (2005) Secondary reactions and strategies to improve 18, 25–34
quantitative protein footprinting. Anal. Chem. 77, 3029–3037 61 Chabes, A. et al. (1999) Yeast Sml1, a protein inhibitor of ribo-
50 Xu, G. and Chance, M.R. (2005) Radiolytic modification of sulfur-
nucleotide reductase. J. Biol. Chem. 274, 36679–36683
containing amino acid residues in model peptides: fundamental
62 Zhao, X. et al. (2000) Mutational and structural analyses of the
studies for protein footprinting. Anal. Chem. 77, 2437–2449
ribonucleotide reductase inhibitor Sml1 define its Rnr1 interaction
51 Xu, G. and Chance, M.R. (2005) Radiolytic modification of sulfur
domain whose inactivation allows suppression of mec1 and rad53
containing acidic amino residues in model peptides: fundamental
lethality. Mol. Cell. Biol. 20, 9076–9083
studies for protein footprinting. Anal. Chem. 77, 4549–4555
63 Bystroff, C. et al. (2000) HMMSTR: a hidden Markov model for local
52 Kiselar, J.G. et al. (2003) Structural analysis of gelsolin using
synchrotron protein footprinting. Mol. Cell. Proteomics 2, 1120–1132 sequence-structure correlations in proteins. J. Mol. Biol. 301, 173–190
53 Hall, D.R. et al. (2002) The crystal and molecular structures of diferric 64 Lim, J. and Vachet, R.W. (2003) Development of a methodology based
porcine and rabbit serum transferrins at resolutions of 2.15 and on metal-catalyzed oxidation reactions and mass spectrometry to
2.60 Å, respectively. Acta Crystallogr. D Biol. Crystallogr. 58, 70–80 determine the metal binding sites in copper metalloproteins. Anal.
54 Lawrence, C.M. et al. (1999) Crystal structure of the ectodomain of Chem. 75, 1164–1172
human transferrin receptor. Science 286, 779–782 65 Lim, J. and Vachet, R.W. (2004) Using mass spectrometry to study
55 Dubljevic, V. et al. (1999) A conserved RGD (Arg-Gly-Asp) motif in the copper-protein binding under native and non-native conditions:
transferrin receptor is required for binding to transferrin. Biochem. J. b2-microglobulin. Anal. Chem. 76, 3498–3504
341, 11–14 66 Klassen, N.V. (1987) Primary products in radiation chemistry. In
56 Liu, R. et al. (2003) Structural reorganization of the transferrin C-lobe Radiation Chemistry Principles & Applications (Farhataziz and
and transferrin receptor upon complex formation: the C-lobe binds to Rodgers, M.A.J., ed.), pp. 29–64, VCH Publishers, Weinheim
the receptor helical domain. Biochemistry 42, 12447–12454 67 Sclavi, B. et al. (1998) Following the folding of RNA with time-resolved
57 Cheng, Y. et al. (2004) Structure of the human transferrin receptor– synchrotron X-ray footprinting. Methods Enzymol. 295, 379–402
transferrin complex. Cell 116, 565–576

AGORA initiative provides free agriculture journals to developing countries

The Health Internetwork Access to Research Initiative (HINARI) of the WHO has launched a new community scheme with the UN Food
and Agriculture Organization.

As part of this enterprise, Elsevier has given 185 journals to Access to Global Online Research in Agriculture (AGORA). More than 100
institutions are now registered for the scheme, which aims to provide developing countries with free access to vital research that will
ultimately help increase crop yields and encourage agricultural self-sufficiency.

According to the Africa University in Zimbabwe, AGORA has been welcomed by both students and staff. ‘It has brought a wealth of
information to our fingertips’ says Vimbai Hungwe. ‘The information made available goes a long way in helping the learning, teaching
and research activities within the University. Given the economic hardships we are going through, it couldn’t have come at a better time.’

For more information visit:


http://www.healthinternetwork.net

www.sciencedirect.com

Você também pode gostar