Isotope-Tagged Cross-Linking Reagents. A New Tool in Mass

In protein interaction analysis, one promising method to identify the involved proteins and to characterize interacting sites at the same time is the ...
1 downloads 0 Views 239KB Size
Download PDF
Anal. Chem. 2001, 73, 1927-1934

Isotope-Tagged Cross-Linking Reagents. A New Tool in Mass Spectrometric Protein Interaction Analysis D. R. Mu 1 ller,*,† P. Schindler,† H. Towbin,† U. Wirth,† H. Voshol,† S. Hoving,† and M. O. Steinmetz†,‡

Functional Genomics, Novartis Pharma AG, CH-4002 Basel, Switzerland

In protein interaction analysis, one promising method to identify the involved proteins and to characterize interacting sites at the same time is the mass spectrometric analysis of enzymatic hydrolysates of covalently crosslinked complexes. While protein identification can be accomplished by the methodology developed for proteome analysis, the unequivocal detection and characterization of cross-linked sites remained involved without selection criteria for linked peptides in addition to mass. To provide such criteria, we incorporated cross-links with a distinct isotope pattern into the microtubule-destabilizing protein Op18/stathmin (Op18) and into complexes formed by Op18 with tubulin. The deuterium-labeled cross-linking reagents bis(sulfosuccinimidyl)-glutarate-d4, -pimelate-d4, and -sebacate-d4 were prepared together with their undeuterated counterparts and applied as a 1:1 mixture of the respective d0 and d4 isotopomers. The resulting d0/ d4 isotope tags allowed a straightforward mass spectrometric detection of peptides carrying the linker even in complex enzymatic protein hydrolysates. In the structure elucidation of the linked peptides by MS/MS, the assignment of the linked amino acids was again greatly facilitated by the d0/d4 tag. By applying two cross-linkers with similar reactivity but different spacer length in parallel, even doublets with very low intensity could be assigned with high confidence in MS and MS/MS spectra. Since in the Op18-tubulin complexes only a limited number of peptides carried the linker, the identification of the involved proteins per se was not impeded, thus accomplishing both protein identification and characterization of interacting sites in the same experiment. This novel methodology allowed us to significantly refine the current view of the complex between Op18 and tubulin corroborating the tubulin “capping” activity of the N-terminal domain of Op18.

Proteins as the active gene products are the direct carriers of most of the essential biological functions within a cell. An important aspect of their function is the interaction with binding * Corresponding author: (e-mail) [email protected]. † Novartis Pharma AG. ‡ Present address: Paul Scherrer Institute, Life Sciences, CH-5232 Villigen PSI, Switzerland. 10.1021/ac001379a CCC: $20.00 Published on Web 03/29/2001

© 2001 American Chemical Society

partners as is the case, for example, in signal transduction pathways. Therefore, a major step toward a first understanding of protein functionality is to identify interacting partners and thus to establish the position of individual proteins within the network of cellular pathways. Subsequent detailed analysis of an identified protein complex should then provide a more mechanistic insight. At the molecular level, interaction analysis has therefore to comprise the identification of interacting proteins as well as the characterization of interacting sites. Of all currently available methods, the most promising approach to tackle both problems simultaneously is the analysis of covalently cross-linked complexes by modern mass spectrometry (MS).1 Identification of Interacting Proteins. Besides the yeast twohybrid system and conventional sequencing techniques, the MS methodology developed for proteome analysis has proven highly successful for the identification of interacting proteins.2 A combination of established affinity purification techniques with mass spectrometric protein identification by the MALDI peptide map and the nano-ESMS/MS sequence tag approach allowed the identification of the components of the yeast and the human U1snRNP complexes,3,4 the anaphase promoting complex,5 the yeast nuclear pore complex,6 and the NMDA receptor-adhesion protein signaling complex.7 Characterization of Interacting Areas. For the detailed characterization of interacting areas, X-ray crystallography or nuclear magnetic resonance spectroscopy of protein complexes are the methods of choice provided that the demanding requirements, such as adequate complex size, enough suitable material, or high solution concentration, can be met. In all other cases, alternative techniques that indicate complex-related changes of protein properties have to be applied. The structure-dependent accessibility of reactive groups is one parameter that is influenced by complex formation. The resulting (1) Farmer, T. B.; Caprioli, R. M. J. Mass Spectrom. 1998, 33, 697-704. (2) Rappsilber, J.; Siniossoglou, S.; Hurt, E. C.; Mann, M. Anal. Chem. 2000, 72, 267-275. (3) Neubauer, G.; Gottschalk, A.; Fabrizio, P.; Seraphin, B.; Lu ¨ hrmann, R.; Mann, M. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 385-390. (4) Neubauer, G.; King, A.; Rappsilber, J.; Calvio, C.; Watson, M.; Ajuh, P.; Sleeman, J.; Lamond, A.; Mann, M. Nat. Genet. 1998, 20, 46-50. (5) Zachariae, W.; Shevchenko, A.; Andrews, P. D.; Ciosk, R.; Galova, M.; Stark, M. J. R.; Mann, M.; Nasmyth, K. Science 1998, 279, 1216-1219. (6) Zhao, Y.; Qin, J.; Chait, B. T. Proceedings of the 46th ASMS Conference on Mass Spectrometry and Allied Topics, Orlando, FL, 1998. (7) Husi, H.; Ward, M. A.; Choudhary, J. S.; Blackstock, W. P.; Grant, S. G. N. Nat. Neurosci. 2000, 3, 661-669.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001 1927

different reactivity of these groups was successfully probed by chemical derivatization followed by comparative mass mapping of the respective enzymatic hydrolysates.8 This method depends, however, on the solubility of the derivatized proteins and the comparative analysis still needs sample amounts in the low nanomolar range. Limited proteolysis, which is the basis of mass spectrometric epitope mapping,9 is another way to elucidate complex-induced changes in reactivity. A disadvantage of this method is that careful enzyme selection and extended series of time course experiments are necessary to obtain reliable results.10 H/D exchange rates of backbone amide hydrogens are another probe for protein structure,11,12 and sensitive LC/MS techniques allow determination of the deuterium content of specific regions by the analysis of the respective peptic peptides.13 In comparative H/D exchange of free and complex bound species, distinct binding areas in protein complexes could thus be localized.14,15 However, this experimentally delicate methodology is only applicable to kinetically stable complexes with subunits changing the orientation or exchange rate of the respective backbone amide hydrogens significantly upon complexation. An alternative method to detect interacting regions, which relies directly on complex structure and not on changes of topological properties, is chemical cross-linking. Within this traditional protein chemistry field (for reviews, see ref 16-19), a variety of homo- and heterobifunctional as well as zero-length cross-linking reagents have been developed for the introduction of covalent links between most of the reactive functional groups in proteins (for selected commercially available compounds cf. ref 20). Subsequent characterization of intramolecularly linked residues was a first means to probe three-dimensional protein structures21,22 and to validate atomic models.23 Only very recently, however, protein fold identification could be achieved on this basis. By a combination of modern sequence threading programs with distance constraints derived from highly sensitive MS and MS/ MS characterization of hydrolysates of cross-linked proteins, FGF-2 could be identified as a member of the β-trefoil fold family.24 (8) Steiner, R. F.; Albaugh, S.; Fenselau, C.; Murphy, C.; Vestling, M. Anal. Biochem. 1991, 196, 120-125. (9) Suckau, D.; Ko ¨hl, J.; Karwath, G.; Schneider, K.; Casaretto, M.; BitterSuermann, D.; Przybylski, M. Proc. Natl. Acad. Sci. U.S.A 1990, 87, 98489852. (10) Scaloni, A.; Miraglia, N.; Orru, S.; Amodeo, P.; Motta, A.; Marino, G.; Pucci, P. J. Mol. Biol. 1998, 277, 945-958. (11) Wagner, G.; Wu ¨ thrich, K. J. Mol. Biol. 1982, 160, 343-361. (12) Molday, R. S.; Englander, S. W.; Kallen, R. G. Biochemistry 1972, 11, 150158. (13) Zhang, Z.; Smith, D. L. Protein Sci. 1993, 2, 522-531. (14) He, F.; Li, W.; Emmett, M. R.; Marshall, A. G.; Zhang, W.; Laue, E. D.; Domaille, P. Proceedings of the 48th ASMS Conference on Mass Spectrometry and Allied Topics, Palm Beach, CA, 2000. (15) Yu, J.; Hasan, A. S.; Smith, J. B.; Smith, D. L. Proceedings of the 48th ASMS Conference on Mass Spectrometry and Allied Topics, Palm Beach, CA, 2000. (16) Mattson, G.; Conklin, E.; Desai, S.; Nielander, G.; Savage, M. D.; Morgensen, S. Mol. Biol. Rep. 1993, 17, 167-183. (17) Wong, S. S.; Wong, L. J. Enzyme Microb. Technol. 1992, 14, 866-874. (18) Ji, H. Methods Enzymol. 1983, 91, 580-609. (19) Brunner, J. Annu. Rev. Biochem. 1993, 62, 483-514. (20) Pierce Products Catalog, Rockford, IL. (21) Zahn, H.; Meyenhofer, J. Makromol. Chem. 1958, 26, 153-166. (22) Fasold, H.; Klappenberger J., Meyer, C.; Remold, H. Angew. Chem., Int. Ed. Engl. 1971, 10, 795-801. (23) Lutter, L. C.; Kurland, C. G. Mol. Cell Biochem. 1975, 30, 105-116. (24) 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, 5802-5806.

1928

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

In a similar way, changes in the orientation of domains in HIV-1 integrase upon binding to DNA could be assessed.25 The limited reproducibility of mass spectrometric mixture analysis did not, however, allow unequivocal detection of linkagespecific peptides just by comparing the tryptic mass maps of the cross-linked proteins with that of the untreated compounds.24,25 Basically, cross-linked peptides were detected by matching their molecular masses with peptide masses calculated for all possible links. With increasing protein size, however, not all matches remain unique and already for the verification of the assignments more involved techniques such as MS/MS sequencing have to be applied. Therefore, for multiple protein complexes,2 the much higher redundancy of matching masses makes this approach less suited. Furthermore, linkage analysis of complexes with an unknown constituent is obviously not feasible on this basis. Apart from in-depth analysis of all mixture constituents10 or selective enrichment of cross-linked entities,26 only additional, specific selection criteria for linked peptides irrespective of mass would secure a straightforward assignment even in difficult cases. In the only approach reported so far,27 all primary amino groups of cross-linked ubiquitin were reductively methylated, and after enzymatic hydrolysis, the newly formed N-termini were derivatized with a 1:1 mixture of 2,4-dinitrofluorobenzene-d0/d3. Only crosslinked peptides reacted twice and they could be discriminated by their 1:2:1 isotope pattern from the 1:1 pattern of the nonlinked peptides by LC/MS analysis. The success of this method relies heavily on the complete methylation of the cross-linked products, and together with sensitivity concerns regarding the multistep chemistry, applications to higher mass complexes still have to be demonstrated. In addition, linkages incorporating the N-terminus of a protein cannot be detected by this method. In this study, we report a straightforward method for mass spectrometric detection of peptides carrying a linker. Instead of tagging the cross-linked peptides, the linker itself was labeled a priori. We synthesized three deuterium-labeled homobifunctional cross-linking reagents with different spacer lengths, which provided, as a 1:1 mixture with their undeuterated counterparts, a distinctive linker-specific isotopic pattern. This novel methodology in the field of cross-linkers, which is based on the long tradition of isotope tagging in mass spectrometry,28-31 was successfully applied to the characterization of one specific interaction site of Op18 with tubulin. EXPERIMENTAL SECTION Synthesis of Cross-Linking Reagents. Bis(sulfosuccinimidyl) glutarate (BSG), bis(sulfosuccinimidyl) pimelate, bis(sulfosuccin(25) Tang, N.; Young, M.; Brameld, K.; Tang, A.; Dollinger, G.; Kuntz, I. D.; Leavitt, A.; Gibson, B. W. Proceedings of the 48th ASMS Conference on Mass Spectrometry and Allied Topics, Palm Beach, CA, 2000. (26) Trester, M. L.; Holford, M.; Muir, T. W.; Chait, B. T. Proceedings of the 47th ASMS Conference on Mass Spectrometry and Allied Topics, Dallas, TX, 1999. (27) Chen, X.; Chen, Y. H.; Anderson, V. E. Anal. Biochem. 1999, 273, 192203. (28) Hunt, D. F.; Buko, A. M.; Ballard, J. M.; Shabanowitz, J.; Giordani, A. B. In Soft Ionization Biological Mass Spectrometry; Morris, H., Ed.; Heyden: London, 1980; pp 85-109. (29) Rose, K.; Simona, M. G.; Offord, R. E.; Prior, C. P.; Otto, B.; Thatcher, D. R. Biochem. J. 1983, 215, 273-277. (30) Sechi, S.; Chait, B. T. Anal. Chem. 1998, 70, 5150-5158. (31) Goodlett, D. R.; Bruce, J. E.; Anderson, G. A.; Rist, B.; Pasa-Tolic, L.; Fiehn, O.; Smith, R. D.; Aebersold, R. Anal. Chem. 2000, 72, 1112-1118.

imidyl) sebacate, and their deuterated analogues were prepared in analogy to the synthesis of bis(sulfosuccinimidyl) suberate.32 N-Hydroxysulfosuccinimide sodium salt, 1.6 mmol (Fluka, Chemie AG, Buchs, Switzerland) was reacted with 0.8 mmol of the respective dicarboxylic acid in the presence of 1.76 mmol of N,N′dicyclohexylcarbodiimide (Fluka, puriss., ∼99%). Glutaric acid (purum, 98%), pimelic acid (>99%), and sebacic acid (purum, 97%) were purchased from Fluka, glutaric acid-2,2,4,4-d4 was from Cambridge Isotope Laboratories (Andover, MA), and 1,7-heptanedioic acid-2,2,6,6-d4 and 1,10-decanedioic acid-2,2,9,9-d4 were from C/D/N Isotopes Inc. (representative for Switzerland, ABCR GmbH, Karlsruhe, Germany). The educts were dissolved in 4 mL of dimethylformamide (Fluka, puriss., over molecular sieve) and stirred overnight at room temperature. N,N′-Dicyclohexyl urea was precipitated by cooling the reaction mixture (2 h at 4 °C) and removed by filtration. The reaction products were precipitated by addition of ∼30 mL of ethyl acetate and recovered by filtration. All manipulations were carried out under argon. Remaining solvents were removed in a vacuum desiccator. The purity of the products was between 95 and 99% as assessed by 1H NMR. Preparation and Separation of Cross-Linked Products. Recombinant human Op18/stathmin (Op18) was produced as described in ref 33. It should be noted, that the protein contained an additional N-terminal dipeptide, His-Met, derived from the pET16b vector (Novagen) which is not part of the Op18 sequence. For cross-linking studies, 1.5 µmol of Op18 was incubated with 1.5 µmol of bovine brain R/β-tubulin (Cytoskeleton, Inc., Denver, CO) in 100 µL of M-buffer (50 mM MES-KOH pH 6.8, 0.5 mM MgCl2, 1 mM EGTA). After 15 min at room temperature, a 1:1 mixture of the d0/d4 isotopomers of cross-linking reagent bis(sulfosuccinimidyl)-glutarate-d0/d4 (BSG- d0/d4), -pimelate-d0/d4 (BSP-d0/d4), or -sebacate-d0/d4 (BSS-d0/d4) was added (final concentration, 4 mM), and the samples were incubated at room temperature for 30 min. The reaction was quenched by the addition of 2-mercaptoethanol (5% final concentration) and glycine or alanine (100 mM final concentration). Products were subsequently precipitated with 66% acetone, and cross-linked material was analyzed by 4-20% SDS-PAGE. In-Gel Trypsin Digestion. SDS-PAGE protein bands were excised, transferred to 96-well plates and subjected to in-gel digestion, essentially according to ref 34. After reduction, alkylation, and overnight digestion with trypsin, peptides were finally eluted with 5% formic acid. MALDI Mass Spectrometry. R-Cyano-4-hydroxycinnamic acid/nitrocellulose matrices were prepared by a modified version of the fast-evaporation technique of Jensen et al.35 A solution of nitrocellulose (Trans-Blot Transfer Medium, Bio-Rad, Glattbrugg, Switzerland) in acetone (10 mg in 0.5 mL) was admixed to a suspension of R-cyano-4-hydroxycinnamic acid (Sigma, Buchs, Switzerland) in 2-propanol (20 mg in 0.5 mL,), and 0.5 µL of this solution was applied to the sample plate followed by 0.8 µL of sample solution in 15:83:2 (v/v/v) CH3CN/H2O/HCOOH corresponding to ∼2% of total sample. Reflectron positive MALDI mass (32) Staros, J. V. Biochemistry 1982, 21, 3950-3955. (33) Steinmetz, M. O.; Kammerer, R. A.; Jahnke, W.; Goldie, K. N.; A. Lustig, A.; van Oostrum, J. EMBO J. 2000, 19, 572-580. (34) Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M. Anal. Chem. 1996, 68, 850-858. (35) Jensen, O. N.; Podtelejnikov, A.; Mann, M. Rapid Commun. Mass Spectrom. 1996, 10, 1371-1378.

spectra were recorded on a PerSeptive Voyager Elite mass spectrometer (Framingham, MA) at 20 kV accelerating potential in the delayed extraction mode using standard settings for delay times and grid voltages. Samples were irradiated by a nitrogen laser pulse at 337 nm, and 256 laser shots were summed into a single mass spectrum. Spectra were calibrated internally on known background signals. Nanoelectrospray Tandem Mass Spectrometry. An aliquot (12 µL out of a total of ∼40 µL) of the sample solution was concentrated/desalted on a needle which was packed with ∼100 nL of POROS R3 sorbent (PE Biosystems, Framingham, MA). After a washing step with 16 µL of 2% formic acid in aqueous solution, the peptide mixture was directly eluted into a gold-coated borosilicate needle (Protana, Odense, Denmark) by the addition of 1-2 µL of 2:50:48 (v/v/v) HCOOH/CH3OH/H2O. The spraying process was started by applying a voltage difference (900 V) between the needle tip and the orifice (1.5-mm distance) of the PE-Sciex QSTAR mass spectrometer (Toronto, Canada). A full mass spectrum was acquired over a wide mass range (m/z 3501800), and parent ions (mass window of 4.5 Da) of interest were mass-selected and fragmented by collision-induced dissociation with nitrogen (collision energy 141 eV; collision gas pressure 6 arbitrary units). RESULTS AND DISCUSSION Selection and Synthesis of Labeled Cross-Linking Reagents. For mass spectrometric linkage analysis based on tagged linking elements, several requirements have to be met. From the wide variety of cross-linkers currently known,16-20 in principle only reagents that introduce additional chemical entities upon linkage formation are suited for isotopic labeling. This prerequisite is met by all bifunctional reagents and modified amino acids such as trifluoromethyldiazirinylphenylalanine36 ((Tmd)Phe) but obviously excludes zero-length linkers such as the carbodiimide derivative 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC). As a second criterion, an ideal tag for mass spectrometric detection should provide a 1:1 doublet with at least a 4 Da spacing of the respective isotopomers (e.g., d0/d4). The 1:1 ratio allows detection of both signals with optimal signal-to-noise ratio, and the 4 Da spacing provides sufficient separation of the isotope distributions of linked peptides even at higher mass. Mass differences bigger than 4 would certainly further improve separation, but simultaneous selection of all isotopomers as precursor ions in subsequent MS/ MS structure elucidation would include an increased amount of chemical background. Spacing by 4 Da is therefore a reasonable compromise, and this condition can be best realized by the use of derivatives of dicarboxylic acids, containing two hydrogen atoms in each R-position, which can be exchanged with good yield.37 Since a number of dicarboxylic acids R,R,R′,R′-d4 are even commercially available, we concentrated our efforts on the synthesis of selected water-soluble bis(sulfosuccinimidyl) derivatives. By reacting glutaric, pimelic, and sebacic acid and their R,R,R′,R′-d4 analogues with N-hydroxysulfosuccinimide and N,N′dicyclohexylcarbodiimide (DCC) according to ref 32 (Figure 1), (36) High, S.; Martoglio, B.; Go¨rlich, D.; Andersen, S. S. L.; Ashford, A. J.; Giner, A.; Hartmann, E.; Prehn, S.; Rapoport, T. A.; Dobberstein, B.; Brunner, J. J. Biol. Chem. 1993, 268, 26745-26751. (37) Atkinson, J. G.; Csakvary, J. J.; Herbert, G. T.; Stuart, R. S. J. Am. Chem. Soc. 1968, 90, 498-499.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

1929

Figure 1. Synthesis of cross-linking reagents. Bis(sulfosuccinimidyl) glutarate-d0 (BSG-d0) or -d4 (BSG-d4) is formed by reaction of glutaric acid-d0 or 2,2,4,4-glutaric acid-d4 (n ) 1), bis(sulfosuccinimidyl) pimelate-d0 (BSP-d0) or -d4 (BSP-d4) by reaction of pimelic acid-d0 or 2,2,6,6-pimelic acid-d4 (n ) 3), bis(sulfosuccinimidyl) sebacate-d0 (BSS-d0) or -d4 (BSS-d4) by reacting sebacic acid-d0 or 2,2,9,9sebacic acid-d4 (n ) 6) with N-hdroxysulfosuccinimid.

the cross-linking reagents bis(sulfosuccinimidyl) glutarate, pimelate and sebacate with spacer lengths of about 7.5, 10, and 14 Å, respectively, were prepared in high purity (95-99% by 1H NMR). These activated carbonyl compounds are highly reactive toward the primary amino group of the side chains of lysine and the N-terminus of proteins (between pH 7 and pH 8 aminolysis is ∼5 times faster than hydrolysis38). The imidazole side chain of histidine forms only unstable reactive intermediates.38 MALDI Mass Maps of Cross-Linked Op18. To test the behavior of the cross-linking reagents BSG, BSP, and BSS, a recombinant form of the 149-amino acid-long microtubuledestabilizing protein Op1839,40 was incubated with a 1:1 mixture of the respective d0/d4 derivatives. The reactions were quenched by addition of either glycine or alanine, and the products linked by BSG (G-link), BSP (P-link), and BSS (S-link) were purified by SDS-PAGE. The excised bands were subsequently subjected to the standard protein identification procedure34 and the resulting tryptic hydrolysates analyzed by MALDI MS in a first step. Thinlayer sample preparation including nitrocellulose was applied to suppress the formation of metal cationized species and to enhance peptide response.35 For the reaction products obtained with BSG, the MALDI mass map (Figure 2) revealed, apart from unlabeled tryptic peptides, 24 easily detectable doublets with the expected 4 Da spacing. Detailed analysis suggested the presence of nine single-chain peptides with an intramolecular link (labeled G-H2O). From the remaining 15 peptides carrying the linker, 8 could contain either two intermolecularly linked chains or a single chain with a derivatized lysine residue. The remaining seven peptides contained only one lysine residue in addition to the one at the C-terminus. Since trypsin cleavage requires an underivatized C-terminal lysine, the additional lysine must carry a linker with a hydrolyzed end. This hydrolysis apparently takes place before quenching, since tryptic maps of G-linked Op18 after alanine or glycine quenching did not show additional signals shifted by 57 (38) Anjaneyulu, P. S. R.; Staros, J. V. Int. J. Pept. Protein Res. 1987, 30, 117124. (39) Sobel, A. Trends Biochem. Sci. 1991, 16, 301-305. (40) Belmont, L. D.; Mitchison, T. J. Cell 1996, 84, 623-631.

1930 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

or 71 Da, respectively. The quenching reagents, which were indispensable for removing the excess of reactive cross-linker before precipitating the reaction mixtures, were apparently not incorporated. Therefore, no additional diversity had to be taken into account in further data evaluation. With respect to the proportion of intramolecularly linked singlechain peptides, similar reactivities were found in the treatment of Op18 with BSP (5 single-chain peptides with an internal bridge out of 17) and BSS (4 single-chain peptides with an internal bridge out of 11). As judged from protein band intensities after Coomassie blue staining, the yield of cross-linked products was comparable for BSG and BSP and slightly reduced for BSS. The size of the cross-linking reagents seems to influence their reactivity. However, a major effect of concomitant differences in hydrophobicity on the recovery of the corresponding linked peptides during extraction with 5% formic acid was not observed. It should be noted that in cases where enough material is available, the application of all three linker types can be a great advantage, not only for probing different spacer lengths but also for improving the reliability of doublet assignment for one specific linkage. Op18-Tubulin Interaction Analysis. The new methodology was subsequently applied to analyze the interaction of Op18 with tubulin in the complex, which has been previously studied by a variety of methods41 including electron microscopy,33 EDC linkage,42 limited proteolysis/MALDI mapping combined with comparative size exclusion chromatography,43 and very recently X-ray crystallography at 4-Å resolution.44 Op18 is a potent regulator of microtubule dynamics, and by electron microscopy, we could attribute one of its destabilizing effects on straight microtubules to the stabilization of a kinked and asymmetric protofilament-like tubulin tetramer (Figure 3A and B).33 Deletion mapping of Op18 further demonstrated that the full length of the C-terminal R-helical domain of Op18 (aa 41-140) is necessary and sufficient for stable ternary complex formation (Figure 3C). In addition, the results indicated that this helix allows Op18 to bind along the entire length of two longitudinally head-to-tail aligned R/β-tubulin heterodimers. This conclusion is unambiguously supported by the recently solved X-ray structure of a tubulin-stathmin-like domain complex.44 However, in contrast to full-length Op18, the C-terminal helix does not prevent the formation of protofilament-based double-ring oligomers in the presence of GDP (Figure 3D). Hence, the function of the N-terminal part of the molecule, which most probably accounts for the observed increase in density (see arrows in Figure 3, A and B) in electron micrographs of Op18-tubulin complexes, is to “cap” tubulin subunits at a site engaged in the formation of longitudinal contacts along protofilaments within the microtubule wall.33 Unfortunately, neither our electron microscopybased study nor the 4-Å resolution X-ray structure of the tubulinstathmin-like domain complex44 allowed us to unambiguously define the orientation of the Op18 molecule relative to the tubulin subunits within the complex. To probe for interacting areas between Op18 and tubulin, equimolar amounts of the proteins were reacted with an excess (41) Lawler, S. Cur. Biol. 1998, 8, R212-R214. (42) Wallon, G.; Rappsilber, J.; Mann, M.; Serrano, L. EMBO J. 2000, 19, 213222. (43) Redeker, V.; Lachkar, S.; Siavoshian, S.; Charbaut, E.; Rossier, J.; Sobel, A.; Curmi, P. A. J. Biol. Chem. 2000, 275, 6841-6849. (44) Gigant, B.; Curmi, P. A.; Martin-Barbey, C.; Charbaut, E.; Lachkar, S.; Lebau, L.; Siavoshian, S.; Sobel, A.; Knossow, M. Cell 2000, 102, 809-816.

Figure 2. MALDI mass map of BSG-linked Op18. Peptides carrying the cross-linker are indicated by connected circles.

Figure 3. Electron microscopic analysis (A-C) (data from ref 33) and chemical cross-linking (E) of the Op18-tubulin complex. (A) Scanning transmission electron micrographs of negatively stained Op18-tubulin complexes. (B) Average projection image of the kinked and asymmetric Op18-tubulin complex obtained by multivariate statistical analysis. Arrows point to the slightly thicker appearing tips of the complexes. (C) Transmission electron micrograph of glycerolsprayed/rotary metal-shadowed Op18-[41-140]-tubulin complexes obtained under microtubule-forming conditions (i.e., 6 mM Mg2+, 0.5 mM GTP, 3.4 M glycerol; 1 h at 37 °C). For comparison with Op18[41-140]-tubulin complexes, unbound tubulin dimers are highlighted by arrowheads. (D) Transmission electron micrograph of negatively stained protofilament-based Op18-[41-140]-tubulin ring oligomers obtained under ring-forming conditions (i.e., 16 mM Mg2+, 1 mM GDP; 1 h at 4 °C). (E) SDS-PAGE analysis of BSG cross-linked Op18tubulin: tubulin control (lane 1), full-length Op18 control (lane 2), and Op18-tubulin (lane 3). The asterisk marks the specific cross-linked protein band obtained with Op18-tubulin. The migration of marker proteins is given on the right in kilodaltons.

of BSG-d0/d4 and BSP-d0/d4, respectively. For controls, tubulin and Op18 were incubated separately with the cross-linking reagents. As illustrated in Figure 3E, a distinct protein band with an apparent molecular mass of ∼90 kDa was observed by SDS-

PAGE analysis of cross-linked Op18-tubulin (lane 3) which was absent in the tubulin (lane 1) and Op18 (lane 2) control. For linkage analysis, this complex-specific band was excised and subjected to the same identification procedure applied already for cross-linked Op18. For this medium-size complex, direct MALDI mapping proved still feasible. However, for more complex mixtures derived from higher-mass complexes, some form of prefractionation, e.g., by HPLC, will become indispensable. As shown in Figure 4, four doublets can be detected in the tryptic MALDI map of the G-linked complex. Their assignment is substantiated by the presence of the corresponding signals shifted by 28 Da (mass difference between P- and G-linkage) in the spectrum of the P-linked complex (Figure 5). The value of this parallel procedure is clearly evident for the detection of signals of low intensity or for the deconvolution of overlapping isotope patterns (cf. m/z 1630/1658). From the four doublets observed, m/z 1630 and 2837 are unique to the complex. Not surprisingly, the high number of possible lysine-linked peptide combinations for a complex with constituents of this size (50 kDa for R-tubulin, 50 kDa for β-tubulin, and 18 kDa for Op18) does not allow an unequivocal assignment based on mass only. Even when considering only candidates with one additional lysine residue in each of the respective peptides (boldface type in Table 1), a reliable assignment is not possible because the exact sequences of R- and β-tubulin from bovine brain are still unknown. This uncertainty can, however, immediately be resolved by the application of highresolution nanoelectrospray MS/MS. As illustrated in Figure 6, the MS/MS spectrum of the quadruply charged precursor ions (m/z 710.11), corresponding to m/z 2837 in the MALDI map, displays a complete series of y-ions for Op18 (aa 1-10) and a nearly complete y-series for Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

1931

Figure 4. MALDI mass map obtained from G-linked Op18-tubulin complexes with doublet regions expanded in insets.

Figure 5. Selected mass regions of MALDI mass maps representing complex-specific G-linked (A, B) and corresponding P-linked (C, D) peptides.

R-tubulin (aa 321-336; see also Figure 7). This result unequivocally identifies the linked peptide as the second candidate from the list of possible combinations for m/z 2837 (Table 1). The presence or absence of the isotopic label in selected fragments (specific d0/d4 pattern) allows, in addition, the exact localization of the linked sites. Labeled y12- (proline cleavage) and y13-ions derived from the R-tubulin part together with unlabeled b3-b4 and y1-y10 from the R-tubulin part and unlabeled y*1-y*10 from the Op18 part clearly establish linkage between the Nterminus of Op18 and Lys 326 of R-tubulin (Figure 7). All assignments were again substantiated by the observation of the 28 Da increment in the related fragments in the MS/MS spectrum of the corresponding P-linked peptides (data not shown). The overall fragmentation pattern did not change by switching from G- to P-linkage. As expected, the fragmentation behavior of the linked peptides is mainly determined by the proline residue and 1932

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

the C-terminal lysines and thus follows the rules for low-energy CID of peptides. As it is evident from Figure 6, only the combination of the high sensitivity and the high resolution of the quadrupole time-of-flight instrument allowed the detection of the characteristic isotopic pattern for high charge states (3+) as well as for low signal intensities. In a similar way, the cross-linked peptides at m/z 1630 (MALDI map) were assigned to Op18 (aa 1-10) linked to R-tubulin (aa 337-339). The two linked tubulin sequences (aa 321-336 and aa 337339) both encompass helix 10 (aa 325-337) of R-tubulin (Figure 8; according to ref 45). Notably, helix 10 and the loop connecting helix 10 with β-strand 9 (aa 338-348) represent critical secondary structural elements that are involved in establishing both longitudinal and lateral protofilament contacts of tubulin subunits within (45) Nogales, E.; Wolf, S. G.; Downing, K. H. Nature 1998, 391, 199-203.

Table 1. Calculated Massesa for [M + H]+ Ions of G-Linked Peptides Derived from the Op18-Tubulin Complex Fitting the Experimental Masses m/z 1630.798 and 2837.445 (MALDI MS) with an Error of