Identification of Cross-Linked Peptides for Protein Interaction Studies

Anal. Chem. 2002, 74, 4417-4422

Identification of Cross-Linked Peptides for Protein Interaction Studies Using Mass Spectrometry and 18O Labeling Jaap Willem Back,*,† Valerie Notenboom,‡ Leo J. de Koning,† Anton O. Muijsers,† Titia K. Sixma,‡ Chris G. de Koster,† and Luitzen de Jong†

Swammerdam Institute for Life Sciences (SILS), Mass Spectrometry Group, University of Amsterdam, Amsterdam, The Netherlands, and Department of Molecular Carcinogenesis, Netherlands Cancer Institute (NKI), Amsterdam, The Netherlands

A new method is presented to screen proteolytic mass maps of cross-linked protein complexes for the presence of cross-linked peptides and for the verification of proposed structures. On the basis of the incorporation of 18O from isotopically enriched water into the C-termini of proteolytic peptides, cross-linked peptides are readily distinguished in mass spectra by a characteristic 8 amu shift. This is due to the incorporation of two 18O atoms in each C-terminus, so that normal and surface-labeled peptides shift 4 amu and cross-linked peptides containing two C-termini will shift 8 amu compared with their unlabeled counterparts. The method is fast, sensitive, and reliable and can be combined with any available crosslinking reagent and a wide range of proteolytic agents. As proof of principle, we successfully applied the method to a complex of two DNA repair proteins (Rad18-Rad6) and identified the interaction domain. Chemical cross-linking of proteins is an established method for gathering structural information of protein complexes.1;2 Depending on the depth of analysis of the cross-link data, different conclusions can be drawn from the data. From the identity of the cross-linked proteins, one may find new structural relationships in multi-protein complexes and build a rough topological model. The real bonus, however, comes from pinpointing the cross-linked residues, a task not easily accomplished. As shown by Young et al.,3 cross-links within the same polypeptide chain represent restrictions in 3D space and can be used to delimit the possible fold families to which the protein under study may belong. Crosslinks between different peptide chains are particularly valuable since they reveal contact sites of the complexes. * Corresponding author. E-mail: [email protected]. Tel: +31(20)5256922. Fax: +31(20)5256568. † University of Amsterdam. ‡ Netherlands Cancer Institute. (1) Rappsilber, J.; Siniossoglou, S.; Hurt, E. C.; Mann, M. Anal. Chem. 2000, 72, 267-75. (2) Wong, S. S. Chemistry of protein conjugation and cross-linking; CRC Press: Boca Raton, FL, 1991. (3) Young, M. M.; Tang, N.; Hempel, J. C.; Oshiro, C. M.; Taylor, E. W.; Kuntz, I. D.; Gibson, B. W.; Dollinger, G. Proc. Natl. Acad. Sci. U.S.A 2000, 97, 5802-6. 10.1021/ac0257492 CCC: $22.00 Published on Web 07/30/2002

© 2002 American Chemical Society

A first approach usually involves peptide mass fingerprinting experiments followed by computational methods in which the masses of peptides found in the spectrum of the cross-linked sample are computed to match a combination of peptide masses with the added mass of the cross-linker within a certain experimental error.4;5 However, unequivocal discrimination of the possibilities generated by this approach requires additional evidence, which will be discussed below. A first choice would be MSMS fragmentation of the putative cross-linked peptides, but these studies often require more material than just a peptide mass fingerprint of the digested and cross-linked material, which may be limiting. Up to date, several strategies have been brought forward to help the investigator find the cross-linked peptides in mixtures of unmodified, surface-labeled (i.e., modified by the cross-linker but not actually cross-linked), and cross-linked peptides. Our own previous work in the field resulted in a prototype of a cross-linker that can be specifically traced by parent-scanning methods on mass spectrometers.6 Roepstorff et al. have pioneered the mass mapping of cross-links containing disulfide-bridged spacers, followed by control experiments with on-target reduction and alkylation of the cross-linkers.7 The observation of the correct masses after derivatization can serve as additional proof for the correct assignment of the cross-link. However, surface-labeled peptides present in the preparation will yield the same reduced and alkylated products, and the contribution of cross-linked product cannot be appreciated. An approach launched by Muller and co-workers8 is susceptible to a similar flaw. They have synthesized deuterated analogues of existing cross-linkers. The observation of peak pairs separated by the amount of deuterium incorporated draws the researchers’ attention to the peaks of peptides that have reacted with cross-linker, without addressing the question of whether they are true cross-links or surface-labeled peptides. (4) Chen, T.; Jaffe, J. D.; Church, G. M. J. Comput. Biol. 2001, 8, 571-83. (5) Wallon, G.; Rappsilber, J.; Mann, M.; Serrano, L. EMBO J. 2000, 19, 21322. (6) Back, J. W.; Hartog, A. F.; Dekker, H. L.; Muijsers, A. O.; de Koning, L. J.; de Jong, L. J. Am. Soc. Mass Spectrom. 2001, 12, 222-7. (7) Bennett, K. L.; Kussmann, M.; Bjork, P.; Godzwon, M.; Mikkelsen, M.; Sorensen, P.; Roepstorff, P. Protein Sci. 2000, 9, 1503-18. (8) Muller, D. R.; Schindler, P.; Towbin, H.; Wirth, U.; Voshol, H.; Hoving, S.; Steinmetz, M. O. Anal. Chem. 2001, 73, 1927-34.

Analytical Chemistry, Vol. 74, No. 17, September 1, 2002 4417

The approach of Chen et al.9 involves the notion that a crosslinked peptide usually has two N-termini. After blocking of lysine side chains prior to digestion, they specifically derivatized the N-termini after digestion with a group that results in an altered RP-LC retention. Derivatization with 50% isotopically labeled reagent results in a characteristic pattern for true cross-linked peptides, for which a 1:2:1 (unlabeled/singly labeled/doubly labeled) abundance distribution is found. However, this complex multistep derivatization relies on complete methylation of all lysine side chains and cannot detect links involving the N-terminal amino group. Recently, an elegant method for quantitative proteomics has been described, in which proteins are digested in 18O water.10-12 It has been noted that trypsin after hydrolysis subsequently exchanges the second oxygen in the newly formed carboxy terminus.13 This results in peptides differing 4 amu in mass with the corresponding peptides digested with trypsin in normal water. Our method is based on the notion that a two C-terminicontaining cross-linked peptide digested in H218O will shift 8 amu compared with the peptide formed upon proteolysis in normalabundance water, as has been observed by Reynolds et al. in a GluC digest of a synthetic HSP peptide.14 Using fully enriched 18O water, full advantage is taken of complete incorporation. This is in contrast to studies performed by Wallis et al. in a recent effort to locate disulfide bonds with the use of 50% 18O isotopically enriched water.15 Also, these authors used the enzyme pepsin, an enzyme for which the exchange characteristics for the second oxygen have not yet been established. Our methodology is readily applicable in combination with any available cross-linking reagent and a range of proteolytic enzymes14 and does not require synthesis of custom-labeled chemicals. Another advantage is that 16O- and 18O-labeled analytes will not display altered reversed-phase HPLC retention times, while deuterated compounds may cause significant retention time shifts.16;17 In this study, we cross-linked the Rad18-Rad6 heterodimer and demonstrate the usefulness of the technique by identifying several cross-links allowing the identification of the interaction domain. The Rad6-Rad18 protein complex is an E2/E3 ubiquitin ligase pair that is involved in postreplicative DNA repair.18;19 EXPERIMENTAL SECTION Chemicals. 18O-Labeled water (95+% atom 18O) was purchased from Isotec (Miamisburg, OH). The cross-linker bis(sulfosuccinimidyl) suberate (BS3) was purchased from Pierce (Rockford, IL). (9) Chen, X.; Chen, Y. H.; Anderson, V. E. Anal. Biochem. 1999, 273, 192203. (10) Mirgorodskaya, O. A.; Kozmin, Y. P.; Titov, M. I.; Korner, R.; Sonksen, C. P.; Roepstorff, P. Rapid Commun. Mass Spectrom. 2000, 14, 1226-32. (11) Wang, Y. K.; Ma, Z.; Quinn, D. F.; Fu, E. W. Anal. Chem. 2001, 73, 374250. (12) Yao, X.; Freas, A.; Ramirez, J.; Demirev, P. A.; Fenselau, C. Anal. Chem. 2001, 73, 2836-42. (13) Rose, K.; Savoy, L. A.; Simona, M. G.; Offord, R. E.; Wingfield, P. Biochem. J 1988, 250, 253-9. (14) Reynolds, K. J.; Yao, X.; Fenselau, C. J. Proteome Res. 2002, 1, 27-33. (15) Wallis, T. P.; Pitt, J. J.; Gorman, J. J. Protein Sci. 2001, 10, 2251-71. (16) Zhang, R.; Sioma, C. S.; Wang, S.; Regnier, F. E. Anal. Chem. 2001, 73, 5142-9. (17) Conrads, T. P.; Alving, K.; Veenstra, T. D.; Belov, M. E.; Anderson, G. A.; Anderson, D. J.; Lipton, M. S.; Pasa-Tolic, L.; Udseth, H. R.; Chrisler, W. B.; Thrall, B. D.; Smith, R. D. Anal. Chem. 2001, 73, 2132-9.

4418

Analytical Chemistry, Vol. 74, No. 17, September 1, 2002

Sulfo-N-benzyliminodiacetoyloxysuccinimide (sBID), a watersoluble analogue of N-benzyliminodiacetoyloxysuccinimide (BID), was synthesized as previously described for BID,6 with the replacement of N-hydroxysuccinimide by sulfo-N-hydroxysuccinimide (Pierce). The sulfonate groups introduced by this procedure render the cross-linker water-soluble. Identity and purity were assessed by NMR and mass spectrometry. Protein Expression and Isolation. cDNA encoding Rad6 and Rad18 were cloned into a pET25 vector and expressed in bacterial cell cultures upon IPTG induction. The complex was purified from clarified cell extract on Talon resin (Clontech, Palo Alto, CA) using a polyhistidine tag on the Rad6 N-terminus. The eluate was further purified on a size exclusion matrix (Sepharose 200, Pharmacia, Uppsala, Sweden), yielding the final purified complex. For amino terminal sequence analysis, proteins separated by gel electrophoresis and electroblotted on PVDF membranes were sequenced by automatic Edman degradation on a Procise model 494 sequencer (Applied Biosystems, Foster City, CA). Cross-Linking and Digestion. The protein complex at a concentration of 0.5 mg/mL in a 20 mM sodium phosphate buffer (pH 7.9) was incubated with 1 mM cross-linker at room temperature for 30 min. SDS-PAGE was performed according to Laemmli.20 Protein bands were cut out of the gel and digested as described by Shevchenko.21 For direct digestion in solution, after cross-linking, cysteines were reduced with dithiothreitol and S-alkylated with iodoacetamide. The sample was subsequently diluted 10-fold in water and spun in a Ultrafree-0.5 centrifugal filter (Millipore, Bedford, MA). The dilution and centrifugation step was repeated twice, resulting in a 1000-fold reduction of buffer salts and alkylating agent. The sample was then split in two equal aliquots, which were dried in a vacuum centrifuge to complete dryness. Sample was reconstituted in 29 µL of either naturalabundance water or 18O-labeled water. To this solution, 1.5 µL of acetonitrile and 1 µL of a trypsin solution (sequencing grade, Roche, Basel, Switzerland) in 1 M ammonium bicarbonate was added, so that the trypsin: substrate ratio was approximately 1:25 (w:w). The digestion mixture was incubated at 37 °C overnight. MALDI-TOF Peptide Analysis. Peptides were collected on ZipTip µC18 pipet tips (Millipore), washed with 0.1% formic acid and eluted with 60% acetonitrile/0.1% formic acid. Subsequently 0.5 µL of the eluate was mixed with 0.5 µL of a 10 mg/mL R-hydroxycinnaminic acid solution in 1:1 (v:v) acetonitrile/ethanol. The mixture was spotted on to a MALDI target plate and allowed to dry. Reflectron MALDI-TOF spectra were acquired on a Tofspec 2EC mass spectrometer (Micromass, Wythenshawe, U.K.). LC-ESI-MS and MSMS Peptide Analysis. The digested peptide mixture was loaded onto the precolumn of an Ultimate nano-HPLC system (LC Packings, Amsterdam, The Netherlands) and separated on a PepMap C18 nano-reversed-phase column (75µm i.d.). Elution was performed using a gradient of 5-40% (18) Roest, H. P.; van Klaveren, J.; de Wit, J.; van Gurp, C. G.; Koken, M. H.; Vermey, M.; van Roijen, J. H.; Hoogerbrugge, J. W.; Vreeburg, J. T.; Baarends, W. M.; Bootsma, D.; Grootegoed, J. A.; Hoeijmakers, J. H. Cell 1996, 86, 799-810. (19) van der, Laan R.; Roest, H. P.; Hoogerbrugge, J. W.; Smit, E. M.; Slater, R.; Baarends, W. M.; Hoeijmakers, J. H.; Grootegoed, J. A. Genomics 2000, 69, 86-94. (20) Laemmli, U. K. Nature 1970, 227, 680-5. (21) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-8.

acetonitrile with 0.1% formic acid. The flow was infused directly into an ESI-QTOF mass spectrometer (Micromass) via a modified nanoelectrospray device (New Objective, Woburn, MA). MS/MS experiments were conducted with argon as collision gas at a pressure of 4 × 10-5 bar measured on the quadrupole pressure gauge. Analysis of Mass Spectra with the FindLink Algorithm. After internal mass calibration (better than 40 ppm), the MALDI and electrospray MS spectra were charge deconvoluted using the MaxENT 3 algoritm (MaxENT solutions, Cambridge, U.K.). For each experiment, the above-obtained MS data were scanned for digest fragments that were modified with the used chemical cross-linker. A custom-made software tool called FindLink supported these analyses. The FindLink program generates a mass/fragment database, based on the input of the residue sequence of the proteins in the complex, the selectivity of the digest cleavages, and the chemical selectivity of the chemical cross-linker. Database entries include all fragment candidates for surface label modifications, for intramolecular cross-linking within a fragment, and all fragment combinations for cross-linking within and between proteins in the complex. Each database entry is automatically matched within a definable mass tolerance with the experimentally obtained mass lists. The matches for surface labeling, intramolecular cross-linking within a single digest fragment, and intermolecular cross-linking between digest fragments within and between proteins in the complex are systematically documented as output of the analyses. RESULTS AND DISCUSSION Bacterial expression of the His-tagged Rad6 protein yielded the sequence as expected, including the polyhistidine tag, and is referred to as Rad6H. Expression of Rad18 yielded the full-length protein as well as a truncated form, most likely due to the presence of an alternative start codon in the sequence recognized by Escherichia coli. This was confirmed by N-terminal sequencing using Edman degradation (results not shown). The truncated form of Rad18 is referred to as Rad18S and starts at Met312. It purifies with Rad6H as a stable complex. Numbering of the residues in both proteins follows the order of the TrEMBL database (accession numbers: Rad6, Q9D0J6; Rad18, Q9QXK2). The polyhistidine tag is numbered in negative. The full-length and the truncated form of the Rad6-Rad18 complex could be easily separated by size exclusion chromatography. After purification, no impurities could be detected on coomassie-stained polyacrylamide gels. Stable cross-links were formed in the Rad18S-Rad6H complex with both sBID and BS3, as shown in Figure 1. The size of the newly formed band upon incubation with cross-linker corresponds to a heterodimer, consisting of one copy of each of the subunits. Using in-gel digestion and peptide mass fingerprinting,1 the presence of both polypeptide chains in the new band was verified. After reductive alkylation of the cysteine residues and removal of the excess alkylating agent, the cross-linked and non-crosslinked control complexes were subjected to tryptic digestion in either normal-abundance water or 18O-labeled water as described in the Experimental Section. The sample was desalted and concentrated prior to MALDI-MS analysis. Peaks that corresponded to unmodified tryptic peptides displayed a nearcomplete 4 amu shift in the corresponding 16O and 18O traces (data

Figure 1. SDS-PAGE of the cross-linked Rad18S-Rad6H complex. Upon incubation with either sBID or BS3, a new band appears, which consists of a covalently bound heterodimer, containing one of each subunits. The cross-link efficiency is estimated to be ∼40%.

not shown), as already demonstrated by others.10-13 The Cterminal peptide of Rad6H at m/z 1490.72 did not shift, in accordance with the proposed mechanism, in which trypsin will only exchange oxygen atoms of peptides ending in Arg or Lys. The C-terminus of Rad18S yields fragments too small to be detected in this approach. Peaks shifting 8 amu in the corresponding 16O and 18O traces were already evident at a first glance, as well as their absence from the control spectra. An example is shown in Figure 2, where the presence of a peak pair with an 8 amu increase is present in the traces of both sBID and BS3 cross-linked complex. As expected, the peak pairs for the different cross-linkers are separated 49 amu, the mass difference of the cross-linker spacer chain. Calculations of possible cross-links with our FindLink algorithm showed that the peak at m/z 1008.5 could be attributed to a cross-linked peptide with an error of only 2 ppm (m/zmeasured ) 1008.5310; m/zcalculated ) 1008.5295, data for BS3). The nearest alternative explanation requires an error of 971 ppm (including single and double surface labels, or a combination of surface labels and cross-links). This renders the evidence convincing. However, this is also due to the fact that the peptide mass is small, limiting the number of combinations that can be fitted. For larger masses, the number of possible combinations within the experimental error margins tends to increase. As a final piece of evidence, LC-MSMS analyses were performed on this cross-linked peptide, as shown in Figure 3. The peptide was selected as a doubly charged ion of m/z 504.7 for the unlabeled sample and m/z 508.7 for the 18O-labeled sample, respectively. As expected, fragment ions containing one of the C-termini of the cross-linked peptide displayed a shift of 4 amu in the corresponding MS/MS spectra. The N-terminal a2-ion at m/z 253.1 did not shift. Taken together, this is consistent with the localization of all four 18O atoms at the carboxy termini of the arginine residues. It follows that de novo sequencing of crosslinked peptides is simplified by the presence of an 18O-labeled fragmentation spectrum. For a more comprehensive analysis, MALDI data were deconvoluted with the MaxEnt 3 algorithm. The output was read into the FindLink software, an in-house-written utility program. This program calculates a matrix of all possible peptide combinations including cross-links within the same polypeptide chain. It also filters for the availability of reactive groups (in this example, amino groups present at N-termini or lysine side chains) and adds the mass of the cross-linker or its corresponding surface label. It is Analytical Chemistry, Vol. 74, No. 17, September 1, 2002

4419

Figure 2. MALDI-TOF MS analysis of a cross-linked tryptic peptide. (A) The amino terminus of Rad18S (MR) is linked to Lys14 of Rad6H. Asterisks indicate the positions of 18O atoms. (B) Trace 1: cross-link with BS3, digested in normal abundance water. The measured m/z 1008.531 deviates only 2 ppm from the calculated mass (m/z 1008.529). Trace 2: the same BS3 cross-linked peptide after digestion in 18O water; the mass is shifted 8 amu due to the incorporation of four oxygens into the peptide C termini. Trace 3: same peptide cross-linked with sBID in normal-abundance water. The mass difference of sBID and BS3 is 49 amu. Trace 4: same peptide cross-linked with sBID and digested in 18O water.

Figure 3. LC-QTOF-MSMS analysis of the BS3 cross-linked peptide depicted in Figure 2. (A) The structure of the cross-linked peptide is drawn, with the cross-linker and the amino acids involved in the cross-link drawn fully, the other residues in standard one-letter abbreviation. (B) Trace 1: low-energy CID MS/MS of the doubly charged ion at m/z 504.7. Trace 2: low-energy CID MS/MS of the corresponding 18O-labeled peptide (doubly charged ion at m/z 508.7). It can be seen that after fragmentation some ions that include one labeled arginine shift 4 amu with respect to the unlabeled peptide. Fragment ions from the N-terminus (a- and b-type ions) do not shift.

capable of handling various different combinations and has the possibility of subtracting reference spectra. From digest spectra of cross-linked peptide mixtures, FindLink advanced several putative cross-links and surface-labeled peptides that were subsequently checked for correct label incorporation. Table 1 and Figure 4 illustrate an example where two possible candidates are listed for a cross-linked peptide peak at m/z 2510.23. The choice for the cross-link spanning from the Nterminus of Rad18S to Lys66 of Rad6H (see Figure 5 for residue numbering) was made because this peptide incorporates up to 4420 Analytical Chemistry, Vol. 74, No. 17, September 1, 2002

four 18O atoms. The alternative explanation, with a lower error margin was ruled out, since this peptide with an internal crosslink contains only one C-terminus and hence cannot incorporate more than two 18O atoms. The corresponding BS3 cross-linked peptide at m/z 2461.24 displayed the same labeling characteristics (data not shown). Another illustrative example is seen when the peptide crosslink spans a linear stretch of amino acids. This is seen for the newly formed peak at m/z 2166.10 in the BS3 cross-linked (and m/z 2215.14 in the sBID cross-linked) sample (data not shown).

Table 1. FindLink Analysis of m/zmeasured 2510.23, Considering SBID Cross-Links in the Rad18S-Rad6H Complex with Trypsin as a Cleaving Reagenta mass 2510.231 2510.231

int 2 2

calculated 2510.259488 2510.274746

ppm 11 17

linked fragments LEASKLNENVMVFTKNQTEK MR

LVIEFSEEYPNKPPTVR

a

Residues involved in the cross-links are in boldface type. FindLink generates two possible solutions within 20 ppm. The possibility with an error of 11 ppm (an internal cross-link in Rad18S from K318 to K328) is ruled out, as this product would have only one free C-terminus and can hence incorporate only two 18O atoms. The correct interpretation is a cross-link spanning from the amino terminus of Rad18S to K66 of Rad6H. (See also Figure 4.).

Figure 4. MALDI-TOF MS analysis of an sBID cross-linked peptide at m/z 2510.23. Trace 1: digestion in normal-abundance water. Trace 2: digestion in 18O-labeled water. The peptide incorporates up to four 18O atoms and must hence contain two free C-termini.

The computational analysis cannot distinguish between the possibility of a cross-link between the N-terminus and Lys318 of the peptide Met312-Lys328 (Rad18S) and a subsequent hydrolysis of the bond between Arg313 and Leu314 or, alternatively, a hydrolyzed surface label at either the N-terminus or Lys318 and a missed cleavage at Arg313. This is because both outcomes result in the same gross formula and hence are identical in exact mass. In this case, the mass shift of 8 amu reflecting the incorporation of up to four 18O atoms immediately identified the first mentioned alternative (i.e., the cross-link). A special case is cross-linking of the C-terminus. As trypsin does not exchange oxygens of C-terminal residues other than Arg or Lys, a cross-linked peptide containing the C-terminus will display the shifting behavior of the cross-linked peptide it is attached to. In this case, the BS3-cross-linked peptide at m/z 3350.62 (error 14 ppm) and the corresponding sBID-cross-linked peptide at m/z 3399.634 (error 8 ppm) were identified as a crosslink spanning from Lys508 of Rad18S to Lys75 of Rad6H. Both peptides incorporated up to two 18O atoms, consistent with the fact that trypsin will not exchange the oxygens of the carboxy terminal Asn509 of Rad18S. Another cross-link was identified along the same lines. The internal cross-link Lys359-Lys363 in the peptide K359GYK362K363TGR of Rad18S was found as an 8 amu shifting peptide pair at m/z 1093.631 (error from calculated mass of 5 ppm and a mass difference of 49 amu with the corresponding sBID cross-linked peptide, which also displayed a complete 8 amu shift) and as a

peptide pair shifting 4 amu at m/z 1231.70, differing 98 amu with the corresponding sBID cross-linked peptide. This latter peptide contains two bound cross-linker molecules, of which one is in the form of a surface label. Because in the two-cross-linker peptide of m/z 1231.70 no internal cleavage site for trypsin is left, it cannot incorporate more than two 18O atoms, consistent with the 4 amu shift. From the data available, it is not possible to discriminate whether the cross-link runs also from Lys359 to Lys362 and the surface label is at Lys363 or any other combination or a mixture of all possible combinations. However, the cross-linked product of m/z 1093.631 can only represent a cross-link from Lys359 to Lys363 subsequently cleaved by trypsin at Lys362, which therefore is the only cross-link we have assigned. Still, the peptide of m/z 1231.7 is successfully identified. Using this technique, we finally identified a number of crosslinks in the Rad18S-Rad6H dimer, some of which are within the same polypeptide chain, others linking the two chains together. The cross-links found for the two cross-linking agents used in this study are shown in the diagram of Figure 5. Most cross-links are found for both agents, contributing to the body of evidence gathered. In a recent study, the average amine-amine distance for BS3 was estimated to amount 9 Å ((1 Å).22 Such quantitative data do not exist for sBID, but based on our simulations, we estimate the cross-link distance for this linker to be 7.5 Å ((1 Å). This is well compatible with distances bridged by the linkers and especially useful to appreciate the flexibility of the N-terminus of the truncated Rad18S. This amino terminus has been found to be involved in four distinct cross-links, three of which link to residues on Rad6H. Of these, one links to the N-terminus of the HisTag on Rad6H, which is also thought to be flexible (V.N.: unpublished observations), the other two being on the top side of Rad6. This pinpoints the truncated N-terminus of Rad18S (and thus the corresponding stretch in the full-length Rad18 protein) to be in this region. It is noteworthy that residues M13, K14, and K66 of Rad6H are on the apical side of the molecule. This region is homologuous to the region of UbcH7 that interfaces a Ring-type E3 in the c-CblUbcH7 crystal structure (PDB accession: 1fbv),23 implying that the cross-links are in the interface region. Also, the deletion constructs made for the yeast Rad6-Rad18 complex are indicative that interaction of the two proteins relies on the N-terminus of Rad6H and the N-terminal region of Rad18S.24A more detailed account of a possible structure for the Rad18S-Rad6H complex (22) Green, N. S.; Reisler, E.; Houk, K. N. Protein Sci. 2001, 10, 1293-304. (23) Zheng, N.; Wang, P.; Jeffrey, P. D.; Pavletich, N. P. Cell 2000, 102, 533-9. (24) Bailly, V.; Prakash, S.; Prakash, L. Mol. Cell Biol. 1997, 17, 4536-43.

Analytical Chemistry, Vol. 74, No. 17, September 1, 2002

4421

Figure 5. (A) Amino acid sequence of the proteins used in this study. Residues involved in cross-links are boxed. (B) Summary of the crosslinks identified using BS3. (C) Summary of the cross-links identified using sBID.

based on the data obtained is outside the scope of this paper and will be reported elsewhere. CONCLUSION After chemical cross-linking of polypeptide chains, retrieval of the positions of the introduced cross-links is the next step. This starts with a comparison of a digest of cross-linked and control untreated sample to find modified peptides in the sample. Computational techniques such as our program FindLink can suggest possible explanations for the differential peaks and should be used as a guideline in the interpretation of the spectra. However, additional proof of the proposed cross-linked peptides remains necessary, and often distinction between alternative solutions has to be made. For these purposes, several strategies may be employed. In this study, we have demonstrated the usefulness of digestion of cross-linked protein complexes in H218O. The shift of 8 amu in mass spectra of cross-linked peptides digested in H218O relative to the same peptides digested in normal-abundance water demonstrates the presence of two C-termini. It also provides additional analytical advantages, i.e., the assignment of b- and y-fragment ions in MS/MS spectra. Furthermore, it is important that masses are measured as accurately as possible, to limit the number of cross-link candidates. Cross-linking with an alternative cross-linking agent can also

4422

Analytical Chemistry, Vol. 74, No. 17, September 1, 2002

contribute to the elucidation.8 This is especially useful in the case where more than one cross-linker is expected to be bound to the peptide of interest, in which case the observed difference in the masses from the alternate experiments will be a multiple of the mass difference in the cross-linker spacer chains. The ease of incorporation of 18O and the flexibility to combine this technique with all available cross-linkers seem to hold great promise both for swiftly retrieving cross-links in a digest spectrum and for confirmation and assignment of proposed linkages. This will open new avenues for identification of interaction sites in protein complexes in a rapid and sensitive way. ACKNOWLEDGMENT The Q-TOF and MALDI-TOF mass spectrometers and the protein sequencer were largely funded by grants from the Council for Medical Sciences, and the nano HPLC was funded by a grant from the Council for Chemical Sciences of The Netherlands Organization for Scientific Research (NWO). V.N. thanks the Leukemia and Lymphoma Society for financial support. The authors thank Roald van der Laan, H.P. Roest, and J.H. Hoeijmakers for their contributions. Louis Hartog is thanked for synthesis of sBID. Received for review May 3, 2002. Accepted July 3, 2002. AC0257492