MALDI In-Source Decay of High Mass Protein Isoforms - American

Jun 16, 2010 - Technologique Timone, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 5, France, Aix-Marseille Université,. Faculté de Pharmacie, 27 ...
0 downloads 0 Views 3MB Size
Download PDF
Anal. Chem. 2010, 82, 6176–6184

MALDI In-Source Decay of High Mass Protein Isoforms: Application to r- and β-Tubulin Variants David Calligaris,†,§ Claude Villard,† Lionel Terras,‡ Diane Braguer,†,§ Pascal Verdier-Pinard,†,§ and Daniel Lafitte*,†,§ INSERM UMR 911, Centre de Recherche en Oncologie biologique et en Oncopharmacologie, Plateforme d’Innovation Technologique Timone, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 5, France, Aix-Marseille Universite´, Faculte´ de Pharmacie, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 5, France, and Socie´te´ Synprosis, Hoˆtel Technologique-BP 100, Technopoˆle de Chaˆteau Gombert, 13382 Marseille Cedex 13, France Tubulin is one of the major targets in cancer chemotherapy and the target of more than twenty percent of the cancer chemotherapic agents. The modulation of isoform content has been hypothesized as being a cause of resistance to treatment. Isoform differences lie mostly in the C-terminus part of the protein. Extensive characterization of this polypeptide region is therefore of critical importance. MALDI-TOF fragmentation of tubulin C-terminal domains was tested using synthetic peptides. Then, isotypes from HeLa cells were successfully characterized for the first time by in-source decay (ISD) fragmentation of their C-terminus coupled to a pseudo MS3 technique named T3-sequencing. The fragmentation occurred in-source, preferentially generating yn-series ions. This approach required guanidination for the characterization of the βIII-tubulin C-terminus peptide. This study is, to our knowledge, the first example of reflectron in-source decay (reISD) of the C-terminus of a 50 kDa protein. This potentially occurs via a CID-like mechanism occurring in the MALDI plume. There are now new avenues for top-down characterization of important clinical biomarkers such as βIII-tubulin isotypes, a potential marker of drug resistance and tumor progression. This paper raises the challenge of protein isotypes characterization for early cancer detection and treatment monitoring. Sensitive and reliable measurement of protein isotypes for early cancer detection and treatment monitoring is a significant challenge. Detection of multiple isoforms within a protein family in a sample necessitates extensive coverage of the amino acid sequence. Characterization of proteins is routinely performed using bottom-up approaches based on tandem mass spectrometry by collision-induced dissociation or post-source decay.1 Nevertheless, associated protocols require preliminary protein digestion that lengthens the sample processing time. One rarely * Corresponding author: Phone: (+) 33-49-183-5680, E-mail: daniel.lafitte@ univmed.fr. † INSERM UMR 911. § Aix-Marseille Universite´. ‡ Socie´te´ Synprosis. (1) Kaufmann, R.; Chaurand, P.; Kirsch, D.; Spengler, B. Rapid Commun. Mass Spectrom. 1996, 10, 1199–1208.

6176

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

achieves extensive coverage of the protein sequence, in particular at the N- and C-termini, which often harbor most of the divergence in a given protein family.2 Thus, there is an increasing need for alternative modes of protein analysis by mass spectrometry that complement more widespread bottomup approaches. To circumvent such difficulties in the characterization of N- and C-terminus part of a protein, top-down strategies that involve intact protein fragmentation in the mass spectrometer are an interesting alternative to classical peptide mass fingerprinting.3 The method of choice is electrospray ionization followed by electron transfer or electron capture dissociation on Fourier Transform mass spectrometers.4,5 MALDI ISD is also a powerful tool.6,7 ISD was recently presented as the new Edman sequencing method, since this technique produces tags mostly from the N-terminus end of proteins whether modified or not.8 C-terminus characterization of myoglobin and of a 35-kDa protein (ABRF-ESRG 2009) was also obtained.9,10 However, for heavier proteins such as bovine serum albumin (BSA), only the cn-fragments were deciphered.8 Tubulins are major targets in cancer chemotherapy, and modulation of isotype expressions is a potential factor of resistance to treatment by antimicrotubule agents. More specifically, an increase in βIII-tubulin expression is a clinical marker of poor prognosis in various cancers.11-19 Tubulin is a 110-kDa heterodimer made of R- and β-tubulin subunits assembled head to tail into protofilaments that associate laterally to build a hollow fiber called the microtubule. In humans, eight R-isotypes (R1A, R1B, R1C, R3C, R3E, R4A, R8, and R-like3) and seven β-isotypes (βI, βII, βIII, βIVa, βIVb, βV, and βVI) have been discovered to date.20 Each R- or β-tubulin monomer can roughly be divided in three domains: The N-terminal domain (residues 1-206) binds a guanosine (2) Luduen ˜a, R. F.; Banerjee, A. Cancer Drug Discovery and Development; Humana Press, 2008. (3) McLafferty, F. W.; Breuker, K.; Jin, M.; Han, X.; Infusini, G.; Jiang, H.; Kong, X.; Begley, T. P. FEBS J. 2007, 274, 6256–6268. (4) Syka, J. E.; Coon, J. J.; Schroeder, M. J.; Shabanowitz, J.; Hunt, D. F. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 9528–9533. (5) Bakhtiar, R.; Guan, Z. Biotechnol. Lett. 2006, 28, 1047–1059. (6) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995, 67, 3990–3999. (7) Hardouin, J. Mass Spectrom. Rev. 2007, 26, 672–682. (8) Suckau, D.; Resemann, A. Anal. Chem. 2003, 75, 5817–5824. (9) Demeure, K.; Quinton, L.; Gabelica, V.; De Pauw, E. Anal. Chem. 2007, 79, 8678–8685. (10) Suckau, D.; Resemann, A. J. Biomol. Technol. 2009, 20, 258–262. 10.1021/ac100996v  2010 American Chemical Society Published on Web 06/16/2010

triphosphate (GTP) molecule, the central domain (207-384) is involved in lateral/longitudinal contacts, and the Cterminal domain starting at residue 385 is mainly involved in the binding of microtubule-associated proteins (MAPs).21 The C-terminus (the last 20 amino acids) is disordered and forms a cloud of negative charge around the microtubule.21,22 The amino acid sequence of the tubulin C-terminus represents an isotype-specific signature21 and is essential for MAPs interaction and modulation of self-assembly into microtubules. The characterization of tubulin isoforms in human cell lines has been under intense investigation by classical proteomic approaches,12,13 but certain isoforms such as βIII have been difficult to characterize probably due to the specificity of their C-terminus sequence and their low expression level. In this work, we first studied MALDI-TOF fragmentation of a tubulin C-terminal synthetic peptide. Then we performed direct identification of intact tubulin isotypes by a reISD and pseudo MS3 strategy also called T3-sequencing.8 The tubulin C-terminus was thus fragmented, generating principally yn- and bn-ion series. All tubulin isotypes previously described in HeLa cells12,23 were successfully characterized. This paper describes, to our knowledge, the first example of ISD of the C-terminus of a 50-kDa protein and the first detection of the C-terminus of βIII-tubulin, a tumor progression biomarker, by MALDI-MS. EXPERIMENTAL SECTION Reagents. Purified matrices (R-cyano-4-hydroxycinnamic acid, CHCA; 3,5 dimethoxy-4-hydroxycinnamic acid, SA; and 2,5dihydroxybenzoic acid, DHB) for MALDI-TOF MS were purchased from LaserBio Laboratories, Sophia-Antipolis, France. 1,5diaminonaphthalene (1,5-DAN) was purchased from Fluka. Peptides and Proteins. Peptide choice corresponds to tyrosinated and detyrosinated forms of R1B tubulin isotype C-terminal peptide: C-VDSVEGEGEEEGEE(-Y), respectively, noted CTR1BE and CTR1BY. They were produced by solid-phase peptide synthesis using Fmoc chemistry and purified by HPLC. Each peptide was extended by addition of an N-terminal cysteine for future studies, not described in this paper. Theoretical monoisotopic masses are M ) 1595.6 Da for CTR1BE and M ) 1758.6 Da for CTR1BY, respectively. These peptides (11) Orr, G. A.; Verdier-Pinard, P.; McDaid, H.; Horwitz, S. B. Oncogene 2003, 22, 7280–7295. (12) Verdier-Pinard, P.; Wang, F.; Burd, B.; Angeletti, R. H.; Horwitz, S. B.; Orr, G. A. Biochemistry 2003, 42, 12019–12027. (13) Rao, S.; Aberg, F.; Nieves, E.; Band Horwitz, S.; Orr, G. A. Biochemistry 2001, 40, 2096–2103. (14) Verdier-Pinard, P.; Wang, F.; Martello, L.; Burd, B.; Orr, G. A.; Horwitz, S. B. Biochemistry 2003, 42, 5349–5357. (15) Martello, L. A.; Verdier-Pinard, P.; Shen, H. J.; He, L.; Torres, K.; Orr, G. A.; Horwitz, S. B. Cancer Res. 2003, 63, 1207–1213. (16) Kavallaris, M.; Kuo, D. Y.; Burkhart, C. A.; Regl, D. L.; Norris, M. D.; Haber, M.; Horwitz, S. B. J. Clin. Invest. 1997, 100, 1282–1293. (17) Ranganathan, S.; Benetatos, C. A.; Colarusso, P. J.; Dexter, D. W.; Hudes, G. R. Br. J. Cancer 1998, 77, 562–566. (18) Ranganathan, S.; Dexter, D. W.; Benetatos, C. A.; Hudes, G. R. Biochim. Biophys. Acta 1998, 1395, 237–245. (19) Nicoletti, M. I.; Valoti, G.; Giannakakou, P.; Zhan, Z.; Kim, J. H.; Lucchini, V.; Landoni, F.; Mayo, J. G.; Giavazzi, R.; Fojo, T. Clin. Cancer Res. 2001, 7, 2912–2922. (20) Luduena, R. F. Mol. Biol. Cell 1993, 4, 445–457. (21) Nogales, E.; Wolf, S. G.; Downing, K. H. Nature 1998, 391, 199–203. (22) Minoura, I.; Muto, E. Biophys. J. 2006, 90, 3739–3748. (23) Winefield, R. D.; Williams, T. D.; Himes, R. H. Anal. Biochem. 2009, 395, 217–223.

were found to be more than 95% pure by HPLC. Using positiveand negative-ion electrospray analysis of CTR1BE and CTR1BY, we detected only the singly and doubly charged ions of the peptide. No fragments were present, confirming the purity of the peptides (data not shown). Freeze-dried tubulin purified from HeLa cell, a human epithelial carcinoma cell line, was purchased from Cytoskeleton, Denver, CO. Tubulin Guanidination. A solution of O-methylisourea was prepared by dissolving 0.54 g in water to make a total volume of 5 mL after the pH was adjusted to 10.5 by addition of 5 M NaOH. We then added 20 µL of this solution to 1 µg of HeLa cell tubulin. The final solution was incubated 4 days at 4 °C and then tubulin was analyzed. SDS-PAGE and Western Blot. HeLa tubulin (130 ng) was mixed with 30 µL of Laemmli 2× buffer (4% SDS, 20% glycerol, 4% 2-mercaptoethanol, 0.004% bromophenol blue, 0.125 M Tris HCl) with 8 M urea and incubated for 30 min at room temperature prior to SDS-PAGE. HeLa tubulin migration was done using a Criterion TBE Gel, 4-20% (Bio-Rad) in Tris/ Glycine/SDS buffer 1× at 200 V constant voltage. The sample migration was allowed to proceed until the blue dye reached the bottom of the gel. Then, samples were transferred onto nitrocellulose membranes at 100 mA constant intensity, overnight. Membranes were blocked with 0.1% Tween-20 TBS containing 4% nonfat milk for 30 min. Membranes were then incubated in diluted primary antibodies. Monoclonal antibodies against βI-, βIII-, and βIV-tubulin isotypes produced in mouse were obtained from Sigma-Aldrich, Saint-Louis, MO (ref T7816, T8660 and T7941 respectively). After being washed three times for 5 min in 0.1% Tween-20 TBS, membranes were incubated for 1 h with Peroxidase AffiniPure Goat AntiMouse IgG + IgM (H+L) (Jackson ImmunoResearch Laboratories, INC.). After extensive washing, bands were visualized by ECL (GE Healthcare) according to the manufacturer’s instructions. MALDI Mass Spectrometry. MALDI mass spectra were obtained on an Ultraflex III TOF/TOF instrument (Bruker Daltonics, Wissembourg, France) equipped with LIFT.24 All the spectra were acquired in positive mode. One µL of CTR1BE or CTR1BY solutions (1 µM) was purified on ZipTipC18 microcolumns (Millipore) and mixed with 9 µL of a 10 mg/mL CHCA solution (water/acetonitrile/TFA, 50/50/0.1) for MS and LIFT analyses, or 4 µL of a 10 mg/mL SA solution (water/ acetonitrile/TFA, 30/70/0.1) for ISD. Two µL of each solution were deposited on a plain steel MALDI plate. One µL of HeLa cell tubulin solution (1 mg/mL) was purified on ZipTipC4 microcolumns (Millipore) and mixed with 4 µL of SA or DHB (40 mg/mL in water/acetonitrile/TFA, 30/70/0.1) or 1,5-DAN (at saturation in water/acetonitrile/TFA, 50/50/0.1) for reISD and T3-sequencing analyses. Two µL were deposited on the MALDI target. Linear mode MALDI spectra were acquired with an accelerating potential of 20 kV and laser fluence was kept just above threshold. For reISD analyses, laser power was increased up to 20% above threshold to increase fragmentation without losing resolution (laser power at 40%). Ion extraction delay was set between 40 and 200 ns. Spectra were externally (24) Suckau, D.; Resemann, A.; Schuerenberg, M.; Hufnagel, P.; Franzen, J.; Holle, A. Anal Bioanal Chem 2003, 376, 952–965.

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

6177

calibrated using peptide calibration standard (Peptide calibration standard, part no. 206195, Bruker Daltonics) for CTR1B peptide analyses. For HeLa cell tubulin analyses, spectra were externally calibrated using peptide calibration standard for reISD analysis. After calibration, the mass accuracy of the molecular mass of somatostatin-28 (3147.471 Da) was within 10 ppm. reISD mode MALDI spectra were acquired with laser power at 60%. With this fluence, y28-ion of R1B-tubulin isotype (m/z 3107.27 Da) was observed with a mass resolution of 11932. For T3-sequencing of CTR1BE/Y peptides and HeLa cell tubulin, the reISD fragment ions were further fragmented for sequence analysis using the LIFT method.24 The acceleration voltage was 8 kV in the MALDI ion source and 19 kV for fragment postacceleration in the LIFT cell. The precursor ion selector resolution was set to 1%. Standard T3-spectra (ISDLIFT) were obtained at a laser fluence increased by 30%. A total of 600 shots were accumulated for each ISD-LIFT spectrum. Monoisotopic masses were labeled using the BioTools 3.1 software with SNAP (Sophisticated Numerical Annotation Procedure), a pick picking algorithm, with a signal-to-noise (S/N) threshold of 0.1. For reISD analyses, spectra were permanently assigned as “ISD-type”. A rather simple approach was used to assigning peaks with a mass tolerance of 0.5 Da (Supporting Information (SI) S-1, S-2). Lysine modified or not tubulin Cterminus sequences (SI Table S-1) were created and edited with sequence editor (Bruker Daltonics BioTools 3.1) and y-ions were automatically annotated. Peak lists derived from T3-spectra from reISD y-ions were matched with tubulin databases on a local intranet Mascot (Matrix Science, London) server. Guanidized βIII-tubulin isotype identification was based on y-ion requiring setting “Guanidinyl (K)” as global modification. RESULTS Fragmentation of the C-Terminal Peptide of Tubulin. ISD and T3-sequencing were combined to identify tubulin isotypes by exhaustive characterization of their C-terminal part15 (SI S-3). Peptides corresponding to the last 15 or 16 amino acids of the protein in its detyrosinated or tyrosinated forms, CTR1BE and CTR1BY, respectively, were synthesized and analyzed by MALDI mass spectrometry and subsequently submitted to ISD and pseudo MS3 to determine fragmentation patterns. As previously shown for similar synthetic acidic peptides,25 these polypeptides rich in acidic amino acid residues could be easy to sequence by this strategy and produce information for the fragmentation of the intact tubulin. Figure 1 and SI Figure S-6 show MALDI mass spectrometry spectra of CTR1BE and CTR1BY obtained in reISD and linear modes. On these spectra, fragmentation patterns were similar for CTR1BE and CTR1BY peptides. In reISD mode, resolved peaks were measured at m/z 1092.740 (y10), 963.481 (y9) and 906.587 Da (y8) for CTR1BE and m/z 1255.965 (y11), 1126.889 (y10) and 1069.923 Da (y9) for CTR1BY and corresponded to yn-series ions in reISD mode (Figure 1A and SI Figure S-4C) and linear mode (SI Figure S-4A and B). Intact peptide fragmentation products occurred on the N-terminal (25) Sachon, E.; Clodic, G.; Blasco, T.; Jacquot, Y.; Bolbach, G. Anal. Chem. 2009, 81, 8986–8992.

6178

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

sides of Glu6, Glu8, and Gly7 for both peptides. Many poorly resolved peaks (mass resolution below 1000) were observed that corresponded to metastable products. Peptides were then analyzed by MS/MS. LIFT analysis produced an informative fragmentation pattern of CTR1BE and CTR1BY (Figure 1B and SI S-5). Exhaustive fragmentation was obtained and y9- and y10-ion intensity evidenced a preferential arrest of fragmentation at the N-terminus of CTR1BE and CTR1BY Gly7, respectively. T3-sequencing preformed on CTR1BE y10-ion confirmed C-terminus tubulin peptide sequencing by this approach (Figure 1C). The same result was observed for the CTR1BY y11-ion (data not shown). Intact Tubulin Fragmentation. Intact HeLa cell tubulin was analyzed using the strategy applied to the C-terminus peptides. To eliminate the possibility that the presence of proteolytic fragments would interfere with interpretation of data, LC-ESI MS was performed after protein trichloroacetic acid precipitation. No proteolytic fragment was detected. Increasing laser fluence produced protein fragmentation. Peaks in the mass region between 2000 and 4000 Da were measured in reflectron mode (Figure 2). Fragmentation efficiency was not enhanced replacing SA by DHB (SI Figure S-6). Increasing the delayed extraction settings from 40 to 200 ns did not influence the ion time-of-flight. Masses were matched to potential tubulin isotypes fragments using an accuracy of 10 ppm. All the ions in the m/z 2000-4000 Da range are C-terminus fragments and could be assigned to ynseries ions of tubulin isotypes (Figure 2 and Table 1). The absence of any bn-serie ions makes proteolysis of the protein impossible. All considered tubulin isoforms (R1B-, R1C-, βI-, βIVb-, and βIII) were tentatively assigned from at least three reISD fragments (Table 1). To confirm isotype identity, pseudo MS3 analyses using T3sequencing were performed. All reISD fragments were submitted to LIFT analysis. However, MS3 fragmentation was obtained only for the most intense reISD fragments. The sequences of the three most abundant tubulin isotypes, that is, R1B (Figure 3A), βI (Figure 3B), and βIVb (Figure 3C), were then further confirmed using yn and bn fragments, (Table 1). βIII-tubulin described as a biomarker for resistance to chemotherapy was not T3-sequenced. Therefore, it was critical to confirm the presence of this isotype since moreover it could be to our knowledge the first detection of the C-terminus of this species by MALDI-TOF mass spectrometry. Western blot analysis using three anti-β isotype specific antibodies were performed. The presence of a large concentration of βI- and of detectable amounts of βIV- and βIII-tubulin isotypes was confirmed (SI Figure S-7). To allow T3-sequencing of the βIII-tubulin isotype, tubulin was guanidinated by O-methylisourea to substitute lysine for homoarginine that improves sensitivity and favors fragmentation. The T3-sequencing of the m/z 3368.556 Da fragment confirmed the presence of βIII-tubulin isotype in HeLa cell (Figure 4B and SI Table S-2). Peaks with an m/z shift of +42 Da for each modified lysine, were present on the guanidination-condition spectrum (Figure 4A). Besides all the previously described R-tubulin isotypes, additional peaks corresponding to yn ions of R3C/E-, R4Aisotypes were detected due to the presence of a lysine in the C-terminus (Table 1).

Figure 1. CTR1BE and CTR1BY peptide MALDI mass spectra. MALDI mass spectrometry analyses of the C-VDSVEGEGEEEGEE CTR1BE peptide in positive reISD (A), LIFT (B) and T3-sequencing modes (C). T3-spectrum correspond to fragmentation of the y10-ion (m/z ) 1093.546 Da) of CTR1BE peptide.

Absence of fragmentation of the protein N-terminus is intriguing. This could be due to the presence of disulfide bridges

between cysteine present at N-terminus parts of R- and β-tubulin.26 The use of 1,5-DAN to reduce disulfide bridges of the protein Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

6179

Figure 2. MALDI-reISD spectrum of HeLa cell tubulin. Monoisotopically peaks were assigned using BioTools 3.1 software. Insets show isotopic patterns of y28-ion (m/z 3107.27 Da) and y28-ion (m/z 3253.24 Da) of R1B- and βI-tubulin isotype, respectively. Observed mass resolutions are 11932 and 9852 for m/z 3107.27 Da and m/z 3253.24 Da, respectively.

and promote ISD fragmentation of the N-terminus has been attempted without success (data not shown). DISCUSSION Based on the work of Takayama and co-workers on chromogranin A,27 we hypothesized that the C-terminus of tubulin, which contains an isotype-specific amino acid sequence, could be fragmented by ISD using MALDI-TOF MS. In this process, an increase in laser fluence generates a peptide radical by an intermolecular hydrogen transfer from the matrix to the peptide backbone carbonyl-oxygen that favors fragmentation. ISD polypeptides are principally cleaved at the NHsCR bond of the peptide backbone, giving cn- and (zn + 2)-series ion fragments according to the standard nomenclature.28-30 However, bn-, yn-, an-, and xn-series ions can also be formed depending on the matrix, the laser fluence, and the sequence of the polypeptide.31 The presence of an acidic cluster of glutamic acids induces special fragmentation upon ISD.31 The observation of only yn fragments from the tubulin peptides in reflectron mode is consistent with the report of Takayama and co-workers on chromogranin A that showed that the primary structure strongly influences ion series obtained by ISD.27 In our case C-terminal fragmentation is favored and therefore cn-series ions are not detected due to the molecular weight of the protein. A recent report explained the presence of such yn fragments by a nonclassical ISD fragmentation of fragile peptides similar to CID.25 Although proline is the most sensitive amino acid responsible for this phenomenon, one could think that this type of fragmentation also occurs at a lower level on glutamic acid, an amino acid as basic as proline in the gas phase.32 This notion is also supported by the fact that the (26) Chaudhuri, A. R.; Khan, I. A.; Luduena, R. F. Biochemistry 2001, 40, 8834– 8841. (27) Takayama, M. J. Am. Soc. Mass Spectrom. 2001, 12, 420–427. (28) Kocher, T.; Engstrom, A.; Zubarev, R. A. Anal. Chem. 2005, 77, 172–177. (29) Takayama, M. J. Am. Soc. Mass Spectrom. 2001, 12, 1044–1049. (30) Takayama, M. J Mass Spectrom Soc Japon 2002, 50, 304–310. (31) Takayama, M.; Tsugita, A. Int. J. Mass Spectrom. 1998, 181, L1–L6. (32) Harrison, A. G. Mass Spectrom. Rev. 1997, 16, 201–217.

6180

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

fragmentation pattern is similar whatever the matrix tested and not enhanced by DHB; if classical ISD occurred, the DHB matrix should increase the fragmentation level. On heavy proteins such as BSA, only N-terminus cn-series ions are generated by reISD.8 As above for synthetic peptide ISD fragmentation, we can also hypothesize that the primary structure of the tubulin C-terminus is responsible for such peculiar reISD fragmentation. Another hypothesis comes from the tridimensional structure of the protein. The protein core is tightly folded and the C-terminus is a rather flexible tail accessible to proteases for instance. Protein crystallization on the MALDI target disrupts the protein structure, but potential aggregation of the protein core will be more intense than that of its C-terminus part. This region could therefore be more accessible to ion molecule reaction occurring in the plasma during the reISD process. However, the mechanistic cause of the absence of fragmentation on the extreme C-terminus of the whole tubulin protein is unknown. It has been suggested that the C-terminus interacts with the core of tubulin.33 In this way, C-terminus could be protected from reISD-induced fragmentation. That the fragmentation occurs in both R- and β-tubulin on R helix H12 is also intriguing (Figure 5). This suggests that this helix is sticking out on the surface and is mobile. Interestingly, in R-tubulin, the fragmentation occurred at the start of H12, a region that becomes exposed to solvent upon stabilization of microtubules by stabilizing agents, whereas the rest of H12 is protected34 and β-tubulin H12 becomes more protected.35 Computer-driven modeling of tubulin flexibility has predicted that both R- and β-tubulin C-termini including H12 would increase in mobility in presence of paclitaxel.36 Indeed, for β-tubulin, the fragmentation occurred more (33) Pal, D.; Mahapatra, P.; Manna, T.; Chakrabarti, P.; Bhattacharyya, B.; Banerjee, A.; Basu, G.; Roy, S. Biochemistry 2001, 40, 15512–15519. (34) Khrapunovich-Baine, M.; Menon, V.; Verdier-Pinard, P.; Smith, A. B., 3rd.; Angeletti, R. H.; Fiser, A.; Horwitz, S. B.; Xiao, H. Biochemistry 2009, 48, 11664–11677. (35) Xiao, H.; Verdier-Pinard, P.; Fernandez-Fuentes, N.; Burd, B.; Angeletti, R.; Fiser, A.; Horwitz, S. B.; Orr, G. A. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 10166–10173. (36) Keskin, O.; Durell, S. R.; Bahar, I.; Jernigan, R. L.; Covell, D. G. Biophys. J. 2002, 83, 663–680.

Table 1. MALDI-reISD Fragmentation of Native and Guanidized HeLa Cell Tubulina

a * yn-Ions observed during reISD analysis of guanidinized hela cell tubulin. Grey highlighted yn-ions correspond to T3-sequenced ions (Figures 3 and 4B).

in the middle and the end of H12, suggesting that H12 on R- and β-tubulin do not have the same exposure at the surface of microtubules. That exposures differ is supported by the fact that none of the residues at the start of H12, which are conserved in R- and β-tubulin, were involved in fragmentation of β-tubulin. Preclinical studies have shown that modulation of tubulin isoform expression has a wide impact on the response to treatments. As an example, high levels of expression of class βIIItubulin are associated with drug resistance in human cancer cell lines.37 For this reason, the isotypic composition of tubulin in cancer cell lines and tumor tissues has been the subject of

extensive research. Various proteomic approaches have been used such as LC-MS analysis of intact tubulin, or cyanogen bromide treatment followed by negative MALDI characterization of the C-terminus.13 Recently, label-free quantitative strategies using Fourier transform instruments were also applied to tubulins.23 The latter strategy was able to characterize the βIII isotype in a mixture by sequence variation not located in the C-terminus. This technique was applied to HeLa cells tubulin and served to quantify βIII-tubulin isotype concentration. For (37) Seve, P.; Dumontet, C. Lancet Oncol. 2008, 9, 168–175.

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

6181

Figure 3. T3-Spectra of the y28-ion (m/z 3107.27 Da), y28-ion (m/z 3253.24 Da), and y29-ion (m/z 3366.398 Da) of R1B(A), βI(B), and βIVb(C) tubulin isotypes from HeLa cell line, respectively.

putative in situ biomarker characterization and drug development,38 a major step foward would be to find a MALDI strategy compatible with the top-down characterization of this isotype and potentially all the others. 6182

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

Therefore it was a breakthrough to obtain yn-ions from tubulin C-terminus and identify the seven major tubulin isotypes of HeLa cells. Guanidination is of crucial help for low abundant species such as βIII. Ions with C-terminus arginine have greater

Figure 4. MALDI-reISD spectrum of guanidinized HeLa cell tubulin and T3-spectrum of the y33-ion (m/z 3368.556 Da) of βIII isotype. (A) MALDIreISD spectra of guanidinized HeLa cell tubulin. (*) C-terminus isotype fragments with modified Lysine. In inset, mass shift due to guanidination of the y33-ion of the βIII-tubulin isotype. ((Gua) refers to guanidinized versus nonguanidized species (B) T3-spectrum of the y33-ion (m/z 3368.556 Da) of βIII isotype. * is a nonassigned peak corresponding to a fragment from a parent ion with a mass within the selection gate of y33-ion.

stability, which generates a better signal intensity. Noteworthy, no other β-tubulin C-terminus fragments than βIII, are guanidinated (Table 1). No post-translational modification has been detected, which is consistent with the limited existence of modification in tubulin from HeLa cells.4 The preferential fragmentation observed may be used to sequence peptides or protein fragments present in body fluids from healthy individuals and patients. Such markers should be

easily characterized by LC-MALDI-ISD MS. Indeed the increase in strongly immunogenic tubulin peptides, such as C-terminal peptides, has been evidenced via the detection of antibodies against tubulin in serum from neuroblastoma cancer patients.39 Tubulin is one of the main proteins of the cell (10% in brain and 3-4% in other tissues) and our study show that the C-terminus of this protein appear to be very labile upon laser irradiation. If tubulin can produce C-terminal peptide upon laser scanning of Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

6183

Figure 5. Tubulin C terminus amino acids involved in tubulin fragmentation by ISD. Molecular modeling was obtained using PyMOL(TM) (DeLano Scientific LLC, Palo Alto, CA). Proteins originate from tubulin R/β heterodimer structure obtained by electron crystallography (encoded as 1TUB in PDB). Panels A and B show R- and β-tubulin monomers, respectively. Fragmentation occurred on both R- and β-tubulin on R helix H12 (in white). H12 is partially exposed to solvent. Moreover, the start of H12 is more protected in β-tubulin. These particularities explain that this part is involved in reISD fragmentation.

tissue sections, MALDI imaging may be envisioned. β-tubulin isotypes are differentially expressed in tissues and blood vessels (βVI-tubulin in platelets), nerves (βII- and βIII-tubulin), and other cells (βI-, βIVb-tubulin). They could be mapped by detection of C-terminal ion fragments by MALDI-imaging.40 Moreover, in tissues affected by a solid tumor, the level of βIII-tubulin expression is often enhanced in cancer cells, which could be morphologically distinguished from nerves and surrounding stroma and from invaded normal tissue. The major advantages in these imaging approaches compared to antibody-based detection is the direct relation between the signal and the entity to be analyzed, which affords quantification and intrinsic multiplexing. (38) Trim, P. J.; Henson, C. M.; Avery, J. L.; McEwen, A.; Snel, M. F.; Claude, E.; Marshall, P. S.; West, A.; Princivalle, A. P.; Clench, M. R. Anal. Chem. 2008, 80, 8628–8634. (39) Prasannan, L.; Misek, D. E.; Hinderer, R.; Michon, J.; Geiger, J. D.; Hanash, S. M. Clin. Cancer Res. 2000, 6, 3949–3956. (40) Debois, D.; Bertrand, V.; Quinton, L.; De Pauw-Gillet, M. C.; De Pauw, E. Anal. Chem. , 82, 4036–4045.

6184

Analytical Chemistry, Vol. 82, No. 14, July 15, 2010

ACKNOWLEDGMENT We are extremely thankful to Pr. Ole Jensen for discussions and critical reading of the manuscript. We thank Dr. Maya Belghazi and Dr. Pierre Edouard Bougis for preliminary results; we also thank Dr. Patrick Fourquet for technical assistance and Didier Esquieu and Sylvie Annappa (Synprosis, Marseille) for peptide synthesis. This work was supported by cance´ropoˆle PACA, David Calligaris fellowship was awarded by the PACA council and partially funded by the Synprosis Company. Abbreviations

ISD GTP MAPs

in-source decay guanosine triphosphate microtubule-associated proteins

SUPPORTING INFORMATION AVAILABLE Tables S-1 and S-2, Figures S1-S7. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review April 15, 2010. Accepted June 7, 2010. AC100996V