Mapping of Phosphorylation Sites by a Multi-Protease Approach with

Jul 12, 2005 - Phosphopeptide-enriched fractions are analyzed with a highly sensitive nanoLC−MS/MS system using a 25-μm-i.d. reversed-phase column,...
0 downloads 5 Views 200KB Size
Anal. Chem. 2005, 77, 5243-5250

Mapping of Phosphorylation Sites by a Multi-Protease Approach with Specific Phosphopeptide Enrichment and NanoLC-MS/MS Analysis Andreas Schlosser,* Jens T. Vanselow, and Achim Kramer

Institute of Medical Immunology, Charite´ , Hessische Strasse 3-4, 10115 Berlin, Germany

We have developed a multi-protease approach that allows sensitive and comprehensive mapping of protein phosphorylation sites. The combined application of the lowspecificity proteases elastase, proteinase K, and thermolysin in addition to trypsin results in high sequence coverage, a prerequisite for comprehensive phosphorylation site mapping. Phosphopeptide enrichment is performed with the recently introduced phosphopeptide affinity material titansphere. We have optimized the selectivity of the phosphopeptide enrichment with titansphere, without compromising the high recovery rate of ∼90%. Phosphopeptide-enriched fractions are analyzed with a highly sensitive nanoLC-MS/MS system using a 25-µm-i.d. reversedphase column, operated at a flow rate of 25 nL/min. The new approach was applied to the murine circadian protein period 2 (mPER2). A total of 21 phosphorylation sites of mPER2 have been detected by the multi-protease approach, whereas only 6 phosphorylation sites were identified using solely trypsin. Titansphere proved to be well suited for the enrichment of a large variety of phosphopeptides, including peptides carrying two, three, or four phosphorylated residues, as well as phosphopeptides containing more basic than acidic amino acids. Reversible phosphorylation of serine, threonine, and tyrosine is a posttranslational modification of particular biological significance in eucaryotic cells. It is involved in the regulation of a large variety of cellular functions, such as cell cycle regulation, cell differentiation, apoptosis, or cytoskeletal regulation.1,2 Many mass spectrometry-based methods for the analysis of phosphorylation sites have emerged over the past decade.3,4 Among these are methods that utilize selective tandem MS scan techniques, such as neutral loss or precursor ion scanning.5,6 Other strategies apply phosphopeptide-specific enrichment, mostly used are immobilized metal affinity chromatography (IMAC)7-11 or chemical tagging of * To whom correspondence should be addressed. E-mail: [email protected]. Fax: (+49) 30 450 524 942. (1) Hunter, T. Cell 1995, 80, 225-236. (2) Hunter, T. Cell 2000, 100, 113-127. (3) McLachlin, D.; Chait, B. T. Curr. Opin. Chem. Biol. 2001, 5, 591-602. (4) Loyet, K. M.; Stults, J. T.; Arnott, D. Mol. Cell. Proteomics, in press. (5) Schlosser, A.; Pipkorn, R.; Bossemeyer, D.; Lehmann, W. D. Anal. Chem. 2001, 73, 170-176. (6) Steen, H.; Ku ¨ ster, B.; Fernandez, M.; Pandey, A.; Mann, M. Anal. Chem. 2001, 73, 1440-1448. 10.1021/ac050232m CCC: $30.25 Published on Web 07/12/2005

© 2005 American Chemical Society

phosphorylation sites,12-14 prior to mass spectrometric analysis. Another approach is based on inductively coupled plasma mass spectrometry coupled to capillary LC for the specific detection of phosphopeptides.15 The vast majority of these strategies is applied to tryptic phosphoprotein digests, although it is well known that the sequence coverage obtained with trypsin is far from being complete.16 Therefore, phosphorylation site maps obtained with a single specific protease cannot be expected to be comprehensive, regardless of the strategy used for phosphopeptide detection. For improving protein sequence coverage and phosphorylation site coverage, the use of additional proteases turned out to be advisable.5,17-19 More than 15 years ago, the chemoaffinity of titania (TiO2) for organic phosphates was discovered,20,21 but only quite recently, this chromatographic material (titansphere) was applied for the selective enrichment of phosphopeptides.22-24 Pinkse et al. recently reported very promising results with this new type of phosphopeptide-affinity material.23,24 (7) Posewitz, M. C.; Tempst, P. Anal. Chem. 1999, 71, 2883-2892. (8) Stensballe, A.; Andersen, S.; Jensen, O. N. Proteomics 2001, 1, 207-222. (9) Schlosser, A.; Bodem, J.; Bossemeyer, D.; Grummt, I.; Lehmann, W. D. Proteomics 2002, 2, 911-918. (10) Ficarro, S. B.; McCleland, M. L.; Stukenberg, P. T.; Burke, D. J.; Ross, M. M.; Shabanowitz, J.; Hunt, D. F.; White, F. M. Nat. Biotechnol. 2002, 20, 301-305. (11) Brill, L. M.; Salomon, A. R.; Ficarro, S. B.; Mukherji, M.; Stettler-Gill, M.; Peters, E. C. Anal. Chem. 2004, 76, 2763-2772. (12) Oda, Y.; Nagasu, T.; Chait, B. T. Nat. Biotechnol. 2001, 19, 379-382. (13) Zhou, H.; Watts, J. D.; Aebersold, R. Nat. Biotechnol. 2001, 19, 375-378. (14) McLachlin, D. T.; Chait, B. Anal. Chem. 2003, 75, 6826-6836. (15) Wind, M.; Edler, M.; Jakubowski, N.; Linscheid, M.; Wesch, H.; Lehmann, W. D. Anal. Chem. 2001, 73, 29-35. (16) MacCoss, M. J.; McDonald, W. H.; Saraf, A.; Sadygov, R.; Clark, J. M.; Tasto, J. J.; Gould, K. L.; Wolters, D.; Washburn, M.; Weiss, A.; Clark, J. I.; Yates III, J. R. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 7900-7905. (17) Gatlin, C. L.; Eng, J. K.; Cross, S. T.; Detter, J. C.; Yates, J. R., III. Anal. Chem. 2000, 72, 757-763. (18) Choudhary, G.; Wu, S. L.; Shieh, P.; Hancock, W. S. J. Proteome Res. 2003, 2, 59-67. (19) Liu, H.; Stupak, J.; Zheng, J.; Keller, B. O.; Brix, B. J.; Fliegel, L.; Li, L. Anal. Chem. 2004, 76, 4223-4232. (20) Kawahara, M.; Nakamura, H.; Nakajima, T. Anal. Sci. 1989, 5, 763-764. (21) Matsuda, H.; Nakamura, H.; Nakajima, T. Anal. Sci. 1990, 6, 911-912. (22) Sano, A.; Nakamura, H. Anal. Sci. 2004, 20, 565-566. (23) Pinkse, M. W. H.; Uitto, P. M.; Hilhorst,.J.; Ooms, B.; Heck, A. J. R. Proc. 52nd ASMS Conf. Mass Spectrom. Allied Topics, Nashville, TN, 2004; WPU 400. (24) Pinkse, M. W. H.; Uitto, P. M.; Hilhorst,.J.; Ooms, B.; Heck, A. J. R. Anal. Chem. 2004, 76, 3935-3943.

Analytical Chemistry, Vol. 77, No. 16, August 15, 2005 5243

We have optimized the selectivity of the phosphopeptide enrichment with titansphere and combined this technique with a multi-protease approach and ultralow flow nanoLC-MS/MS analysis. We have applied this new approach for phosphorylation site mapping of the murine circadian protein period 2 (mPER2), one of the key components constituting the mammalian circadian clock, an endogenous timing system that adapts physiological processes to the varying demands of a day. EXPERIMENTAL SECTION Materials. Solvent blends (water with 0.1% formic acid and acetonitrile with 0.1% formic acid) for LC-MS (Chromasolv) were purchased from Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany). Trypsin (modified, sequencing grade), elastase, and recombinant proteinase K (PCR grade) were from Roche (Mannheim, Germany), thermolysin was from Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany). The synthetic phosphopeptides EAIpSAAPFAK-NH2 and EAIpTAAPFAK-NH2 were kindly provided by Eberhard Krause (Forschungsinstitut fu¨r Molekulare Pharmakololgie, Berlin, Germany). Fused-silica capillaries were ordered from Polymicro Technologies (Phoenix, AZ). PEEK capillaries, PEEK sleeves, ferrules, nuts, inline filter, and microtees were obtained from Upchurch Scientific (Oak Harbor, WA). Plugs and mixing tee were from Valco (Houston, TX). Trap column (C18 reversed phase, 5-µm particle size, 120-Å pore size, 50-µm i.d., 2-cm length) and analytical column (C18 reversed phase, 5-µm particle size, 120-Å pore size, 25-µm i.d., 20-cm length) were purchased from NanoSeparations (Nieuwkoop, The Netherlands). ESI emitter (360-µm o.d., 20-µm i.d., 5-µm tip i.d., distal coated) were obtained from NewObjective (Woburn, MA). PTFE tubing (300-µm i.d.) for connecting ESI emitter and analytical column were ordered from Dionex (Sunnyvale, CA). Phosphoprotein Preparation. C-terminal V5-tagged murine period 2 protein (mPER2; SwissProt accession number O54943), expressed in stably transfected HEK293 cells, was immunoprecipitated with anti-V5-antibody (Invitrogen, Karlsruhe, Germany) and protein A/G-PLUS agarose beads (Santa Cruz, Heidelberg, Germany). The protein was dephosphorylated in vitro with λ phosphatase (NEB, Frankfurt a.M., Germany) and subsequently rephosphorylated with casein kinase 1δ (NEB, Frankfurt a.M., Germany) according to manufacturer’s instructions. The immunocomplexes were dissolved and proteins were reduced and alkylated prior to SDS-PAGE by boiling for 5 min in 3× sample buffer (150 mM Tris-HCl pH 6.8, 6 mM EDTA, 3% SDS, 30% glycerin) supplemented with 10 mM reducing agent TCEP (Calbiochem, Schwalbach/Ts., Germany). The agarose beads were pelleted, and 50 mM iodoacetamide (Fluka, Buchs, Switzerland) was added to the supernatant. Proteins were separated by SDS-PAGE and stained with Coomassie R250 (Merck, Darmstadt, Germany). Proteolytic Cleavage. In-gel digests were perfomed as described in standard protocols. Briefly, the excised gel bands were destained with 30% ACN, shrunk with 100% ACN, and dried in a vacuum concentrator (Concentrator 5301, Eppendorf, Hamburg, Germany). Digests with trypsin, elastase, proteinase K, and thermolysin were performed overnight at 30 °C in 0.1 M NH4HCO3 (pH 8). About 0.1 µg of protease was used for one gel band. Peptides were extracted from the gel slices with 5% formic acid. For the multi-protease approach, all digests were combined and 5244

Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

spiked with 20 fmol of the synthetic phosphopeptides EAIpSAAPFAK-NH2 and EAIpTAAPFAK-NH2. The combined supernatants and extracts were dried in a vacuum concentrator, redissolved in 10 µL of 30% ACN, 2% FA, and applied to a titansphere nanocolumn. Titansphere Nanocolumns. Integra Frit columns with an inner diameter of 50 µm, obtained from NewObjective (Woburn, MA), were used to prepare the titansphere nanocolumns. The columns were packed by applying 50 bar He pressure to a 2-propanol slurry of 5-µm titansphere particles. The slurry was placed in a high-pressure cell and stirred during the packing procedure. The columns were packed to a length of ∼15 mm. The titansphere nanocolumn was connected to a manually operated six-port valve (Valco), equipped with a 10-µL sample loop. The solvent (0.1% FA) was supplied via CapLC pump C (Micromass, Manchester, U.K.). Before initial use, the column was conditioned with two times 10 µL of 0.1 M NH4HCO3 (pH 9) supplied via the sample loop. Samples (10 µL in 30% ACN, 2% FA) were manually injected with a syringe (Hamilton, Reno, NV) and loaded on the column at a flow rate of 1 µL/min. After washing with additional 10 µL of 30% ACN, 2% FA, elution was performed with 10 µL of 0.1 M NH4HCO3 (pH 9) at a flow rate of 0.25 µL/ min. Eluates were made up with formic acid before nanoLC-MS/ MS analysis. NanoLC. All LC-MS analyses were performed with a nanoLC system, type CapLC (Micromass) equipped with a programmable 10-port 2-position valve (stream select module). Solvent A (water with 0.1% formic acid) and solvent B (acetonitrile with 0.1% formic acid) were degassed by applying a continuous stream of helium. The column switching system, consisting of trap column (C18 reversed phase, 5-µm particle size, 120-Å pore size, 50-µm i.d., 2-cm length) and analytical column (C18 reversed phase, 5-µm particle size, 120-Å pore size, 25-µm i.d., 20-cm length), was assembled as previously described.25 The spray voltage (1100 V) was applied via a piece of copper wire to the ESI emitter, which was hold in place by an in-house fabricated assembly of two metal blocks with chamfers. The restrictor (fused-silica capillary, 25µm i.d.) was adjusted to a proper length, so that a flow of 10 µL/ min before the trap column resulted in a flow rate of 25 nL/min (at a pressure of 900 psi with 100% solvent A) during analytical separation (split ratio 1:400). Sample injection was done manually using a syringe (Hamilton). A fused-silica capillary (254-µm i.d., 10-µL volume) was used as sample loop. After sample injection, the 10-port valve was switched and a flow rate of 2 µL/min (at a pressure of 2000 psi) was applied to load the sample with 100% solvent A onto the trap column. After 20 min, the 10-port valve was switched back and a linear gradient from 0 to 60% acetonitrile (3%/min) was used for analytical separation with a flow rate of 25 nL/min. Mass Spectrometry. All mass spectra were recorded on a quadrupole time-of-flight tandem mass spectrometer, type Q-TOF micro (Micromass), equipped with a nanoESI source. The instrument was calibrated over a mass range from m/z 100 to 2000 using a 10 mM solution of phosphoric acid in 50% 2-propanol. Calibration was performed before each LC-MS run. The tune parameters of the instrument were optimized in MS and MS/MS mode by (25) Schlosser, A. Proc. 52nd ASMS Conf. Mass Spectrom. Allied Topics, Nashville, TN, 2004; WPJ 163.

spraying a solution of [Glu1]-fibrinopeptide B (0.5 pmol/µL) from a gold-coated borosilicate capillary prepared in-house. The applied capillary voltage was 1100 V for all nanoLC experiments. No cone gas was used. Data-Dependent MS/MS. Mass Lynx 3.5 was used as control software for the mass spectrometer. The parameters for datadependent MS/MS were set to fragment up to three precursors at a time (charge states +1 to +4). The intensity threshold for precursor selection was set to 20 counts/s. Scan time for MS/ MS spectra was set to 2 s using two different collision offsets. Data Analysis. Mascot Distiller (Matrix Science, London, U.K.) was used to create peak lists from MS and MS/MS raw data. Charge states +1 to +4 were taken into account for precursor and fragment ions. Mascot Server (Matrix Science, London, U.K.) was used for database searching. Mass tolerance was set to (0.1 Da for precursor mass and fragment ion mass, respectively. Searches were performed in SwissProt without any taxonomic restrictions, considering the following variable modifications: carbamidomethyl (C), oxidation (M), phosphorylation (ST), phosphorylation (Y), phosphorylation-NL (S), phosphorylation-NL (T), pyro-Glu (N-term Q), and deamidation (NQ). RESULTS AND DISCUSSION It was our aim to develop an approach that should facilitate comprehensive and sensitive phosphorylation site mapping of proteins, usually separated by one-dimensional SDS-PAGE. To obtain a phosphorylation site map that is as complete as possible, it is a basic requirement to have a roughly complete sequence coverage of the protein. This is in most instances not achievable by digestion with a single sequence-specific protease. For the circadian protein mPER2, a 140-kDa phosphoprotein, we have obtained a sequence coverage of ∼25% applying in-gel digestion with trypsin and nanoLC-MS/MS. Thus, it appears that analysis of mPER2 using exclusively trypsin will result in a quite incomplete phosphorylation site map, regardless of the methods used for phosphopeptide detection. To increase the sequence coverage, several low-specificity proteases were tested regarding their applicability for in-gel protein digestion and subsequent nanoLC-MS/MS analysis. From these proteases, elastase, proteinase K, and thermolysin turned out to be particularly suited in terms of sequence coverage and digest efficiency. A sequence coverage of about 30-40% was achieved with a single of these low-specificity proteases for mPER2. The application of elevated temperatures for proteinase K and thermolysin did not increase the sequence coverage but substantially increased the number of autoproteolytic peptides. NanoLC-MS/ MS analyses of mPER2 digests with trypsin, elastase, proteinase K, and thermolysin without phosphopeptide enrichment resulted in an overall sequence coverage above 90%, indicating that it should be possible to obtain a largely comprehensive phosphorylation site map by combining these proteases. Elastase, proteinase K, and thermolysin generate peptides with an average length of ∼10 amino acids under the applied conditions. All three proteases tend to make sets of overlapping peptides. From our data, we expect ∼500 different peptides from a digest of mPER2 with a single low-specificity protease, whereas only ∼80 peptides are generated with trypsin. More than 1500 peptides are expected for the multi-protease approach. It turned out, that specific enrichment of the phosphopeptides from such

Figure 1. Schematic diagram of the multi-protease approach for phosphorylation site mapping. The phosphoprotein is digested separately with trypsin, elastase, proteinase K, and thermolysin. Phosphopeptides are enriched from the pooled peptide mixture with a titansphere nanocolumn and subsequently analyzed with nanoLCMS/MS.

complex mixtures drastically inreases the number of detected phosphorylation sites and is an important step in the multi-protease approach, even if single proteins are analyzed. Principle of the Multi-Protease Approach. The principle of the multi-protease approach is outlined in Figure 1. The approach generally starts from phosphoproteins separated by oneor two-dimensional gel electrophoresis. The phosphoprotein is separately digested in-gel with trypsin, elastase, proteinase K, and thermolysin. The individual peptide mixtures are pooled, and the phosphopeptides are selectively enriched with a titansphere nanocolumn from the resulting complex peptide mixture, composed of at least several hundred different peptides. The phosphopeptide-enriched fraction is finally analyzed with nanoLCMS/MS. NanoLC-MS/MS Analysis. The nanoLC system was set up similar to the system described by Meiring et al.26 and comprises a 25-µm-i.d. analytical column.25 The small column inner diameter allows the application of flow rates in the range of 25 nL/min. The sensitivity of the nanoLC-MS/MS analysis strongly benefits from this ultralow flow rate. Decreasing column inner diameter from 75 to 25 µm and flow rate from 200 to 25 nL/min resulted in an increased sensitivity of at least 1 order of magnitude. The increased sensitivity of the nanoLC system more than compensates the higher protein demand of the multi-protease approach. Comparison of Proteases. To explore the contribution of the single proteases to the generated phosphopeptide pool, mPER2 was separately digested in-gel with the four different proteases. Instead of pooling the individual peptide mixtures, the phosphopeptides from each digest were separately enriched with the titansphere nanocolumn and analyzed with nanoLC-MS/MS. To point out the effect of phosphopeptide enrichment Figure 2A shows the extracted ion chromatograms for the phosphopeptide (26) Meiring, H. D.; van der Heeft, E.; ten Hove, G. J.; de Jong, A. P. J. M. J. Sep. Sci. 2002, 25, 557-568.

Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

5245

Figure 2. NanoLC-MS analysis of mPER2 digested with elastase. (A) Extracted ion chromatograms for the phosphopeptide NLLQLEEAPEGpSTGAAGT (m/z 919.4) with and without phosphopeptide enrichment. (B) MS spectra (single survey scan) corresponding to the elution of the phosphopeptide NLLQLEEAPEGpSTGAAGT with and without phosphopeptide enrichment. The peak intensity of m/z 919.4 is ∼20 counts/s in both spectra.

NLLQLEEAPEGpSTGAAGT (m/z 919.4) with and without using titansphere. Without phosphopeptide enrichment, two other peptides appear in the extracted ion chromatogram, although a narrow mass window of 0.1 Da was selected. These peptides disappeared after phosphopeptide enrichment with titansphere. The MS spectra (single survey scans) corresponding to the elution of NLLQLEEAPEGpSTGAAGT (m/z 919.4) are shown in Figure 2B. About seven coeluting peptides with higher abundance are detected without using titansphere, whereas NLLQLEEAPEGpSTGAAGT is the most abundant peptide after phosphopeptide enrichment. The result of the individual digest analyses is summarized in Table 1. A total of 21 phosphorylation sites could be identified with the four different proteases. Only six of these sites were detected in the trypsin digest. The digests with elastase, proteinase K, and thermolysin resulted in 15, 14, and 13 assigned phosphorylation sites, respectively. Inspection of the in silico tryptic digest of mPER2 reveals possible reasons for the missing phosphopeptides in the trypsin digest. The phosphorylation sites T-858, T-866, or T-868 and S-939 are located within a region of 152 amino acids without any tryptic cleavage site. These phosphorylation sites are part of a tryptic peptide of ∼17 kDa, which was not detectable under the applied conditions. Other sequence-specific proteases would generate a peptide of the same size (ArgC), a marginal 5246 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

Table 1. Phosphorylation Sites of mPER2 Using Single-Protease Approaches phosphorylation sitea [S-68 S-69 T-71] [S-91 T-92] S-525 S-528 S-531 S-538 S-544 T-554 [S-693 S-697] S-706 S-758 S-763 T-858 [T-866 T-868] S-939 T-964 S-971 [S-977 S-980] [S-997 T-998] S-1126 [T-1237 S-1238]

trypsinb x

x x

elastaseb xx x x x x xx x x x

x x

xx xx xx x

proteinase Kb x

x x x x x x xx xx xxx x xx

thermolysinb x x x x x

x x x x x x x

x

x x

xx x

x

a Brackets: one of the sites is phosphorylated. Boldface type: exact defined sites. b x: each identified phosphopeptide is represented by one x.

Figure 3. MS/MS spectrum of the doubly charged precursor ion at m/z 722.31 observed in the phosphopeptide-enriched fraction of mPER2 digested with elastase. The peptide is identified as the phosphopeptide SEITPApSQAEFPS. The phosphorylation site is unambiguously identified at the serine on position 7 (S-933).

smaller peptide (V8), or even larger peptides (LysC, AspN). Phosphorylation nearby a tryptic cleavage site can interfere with proteolysis at this position, possibly resulting in improperly large phosphopeptides.5 This complication may explain the absence of phosphopeptides for S-977 or S-980 and S-997 or S-998. If cleavage on the C-terminal side of R-979 is inhibited by nearby phosphorylation, the resulting phosphopeptide would be ∼6 kDa and hardly be detectable with LC-MS. Since the low-specificity proteases inherently have much more cleavage sites than trypsin, the risk of creating improperly large peptides is significantly lower with these proteases. This may explain the higher phosphorylation site coverage observed with elastase, proteinase K, and thermolysin. MS/MS Spectra of Nontryptic Phosphopeptides. Figure 3 shows an example of a MS/MS spectrum of a phosphopeptide generated by digestion of mPER2 with elastase. The doubly charged phosphopeptide SEITPApSQAEFPS was observed at m/z 722.31 and shows an easily interpretable MS/MS spectrum. Due to the absence of any basic residue, the ratio between b and y ions is rather balanced. The peptide contains four potential phosphorylation sites, three serines and one threonine. The observed fragment ions, especially the proline-directed fragments, unambiguously identify the serine on position 7 (S-933) as the phosphorylated amino acid. Mascot scores this phosphopeptide with 57 and just as well identifies the correct phosphorylation site. A similar situation was found for the majority of the peptides, and there were no indications of any adverse effects for peptides generated by a low-specificity protease in comparison to tryptic peptides. The smaller phosphopeptides produced by low-specificity proteases seem to be even advantageous for manual and automatic interpretation, since they contain less potential phosphorylation sites and in many cases show more clearly fragmentation patterns. Application of the Multi-Protease Approach. After having analyzed the individual digests separately, now equivalent mPER2derived samples generated with the four proteases were pooled and analyzed in a single run according to the procedure depicted in Figure 1. To ensure adequate sensitivity for phosphopeptide detection, the sample was spiked with 20 fmol of a synthetic phosphopeptide as control for phosphopeptide enrichment and proper nanoLC-MS/MS conditions. The corresponding nanoLCMS/MS analysis provided 123 MS/MS spectra. Interpretation of the spectra with Mascot together with a detailed manual evaluation

revealed a total of 58 phosphopeptides comprising 29 phosphorylation sites of mPER2. The complete list of identified phosphopeptides is shown in Table 2. In the single-run analysis, 29 phosphorylation sites have been identified, compared to 21 phosphorylation sites detected in the separately analyzed digests. Ten phosphorylation sites were exclusively detected in the singlerun analysis, and two were exclusively identified in the separately analyzed samples, indicating that the complexity of the phosphopeptide-enriched fraction, including the phosphopeptides from four different digests, seems to be adequately low for a one-dimensional nanoLC-MS/MS run. The two phosphorylation sites missed in the single-run analysis may be due to the inherent run-to-run variability of the data-dependent LC-MS/MS analysis; the additional 10 phosphorylation sites emerging in the analysis of the pooled digests may for the most part be due to differences in phosphoprotein amount and sample preparation. Characterization of Identified Phosphopeptides. The peptides summarized in Table 2 exhibit an average length of 13 amino acids with only two peptides being larger than 20 amino acids. This is an appropriate peptide size for fragmentation by collisioninduced dissociation, resulting largely in spectra that allow the accurate localization of the phosphorylation site. Only if the potentially phosphorylated Ser and Thr residues are adjacent, the exact localization is not always feasible. The software for data-dependent MS/MS was set to initiate fragmentation of molecular ions with the charge states 1+ to 4+. The interpretation of the MS/MS data shows that 53 of the phosphopeptides selected for MS/MS have the charge state 2+, 6 have 3+, 5 have 1+, and none shows the charge state 4+. From these peptides, three appear exclusively in the charge state 3+, and four appear exclusively in the charge state 1+. Although the great majority of phosphopeptides generated by low-specificity proteases appears to be doubly charged, excluding other charge states may result in the loss of several phosphorylations sites. In the case of mPER2, six phosphorylation sites would have been missed completely by excluding the charge states 1+ and 3+. For example, the singly phosphorylated peptides with the sequences GSSPLQL and GSSPLQLN, which are free of any basic residues and appear exclusively in the charge state 1+, are the only peptides identifying the phosphorylation site at S-980/981. The phosphopeptides identified for mPER2 carry two, three, and four phosphorylated residues. A total of 52 out of 58 phosphopeptides carry 1 phosphate, 4 carry 2 phosphates, and 1 each carries 3 and 4 phosphates, demonstrating the possibility to enrich even highly phosphorylated peptides with titansphere. To elucidate whether basic phosphopeptides can be enriched with titansphere as well, the 58 phosphopeptides shown in Table 2 were analyzed in terms of their ratio between acidic (pS, pT, D, E) and basic (R, K, H) amino acids. Nine of the peptides show an excess of more than 3 acidic residues over basic residues, 29 a slight excess of 1 or 2 acidic residues, 17 have an equal number of acidic and basic residues, and 3 peptides have an excess of 1 basic amino acid. Thus, the enrichment of basic phosphopeptides is feasible with titansphere as well. Up to 6 phosphopeptides are identified for a single phosphorylation site; only 8 of the 29 phosphorylation sites are represented by a single peptide. This redundancy makes the multi-protease approach more robust compared to a single-protease approach, Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

5247

Table 2. Phosphorylation Sites of mPER2 Identified Using the Multi-Protease Approach

a Brackets: one of the sites is phosphorylated. Boldface type: exact defined sites. b n: total number of phosphorylated residues in a given peptide. c Underlined: phosphorylation sites in accordance with the MS/MS data. Boldface type: exact defined sites.

5248 Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

Figure 4. Coenrichment of acidic non-phosphopeptides with different concentrations of formic acid. (A) Phosphopeptides and nonphosphopeptides specified in the Mascot Peptide Summary Report for a multi-protease digest of mPER2 after phosphopeptide enrichment with titansphere using 0.1 and 2% formic acid, respectively. (B) Observed non-phosphopeptides from panel A ordered by their number of acidic residues.

since missing a phosphopeptide does not necessarily cause a missed phosphorylation site. In addition, overlapping peptides are sometimes useful for the exact localization of a phosphorylation site, since there is MS/MS information from different peptides for the same phosphorylation site. Phosphopeptide Enrichment. Recovery. To optimize the conditions for phosphopeptide enrichment, the titansphere nanocolumn was initially tested with a mixture of synthetic peptides containing the two phosphopeptides EAIpSAAPFAK-NH2 and EAIpTAAPFAK-NH2. Flow-through and eluate were analyzed for a set of different conditions with static nanoESI. It turned out that binding of the phosphopeptides to titansphere is quite fast, whereas elution is slow and has to be performed at a relatively low flow rate in order to obtain high recovery rates. Proper conditions for the applied 50-µm-i.d. column (1.5-cm length) used for all experiments described in this publication proved to be the following: Phosphopeptide binding was carried out in 30% ACN, 2% FA (10-µL volume) at a flow rate of 1 µL/min. After washing with an additional 10 µL of 30% ACN, 2% FA at 1 µL/min, elution was performed with 10 µL of 0.1 M NH4HCO3 (pH 9) at a flow rate of 0.25 µL/min. The recovery under these conditions for 300 fmol of the phosphopeptide EAIpSAAPFAK-NH2, determined by using an internal standard and static nanoESI, was 90 ( 10%. Pinkse et al. reported a very similar recovery for the peptide RKIpSASEF24 under comparable conditions. Phosphopeptide Enrichment. Selectivity. Coenrichment of peptides with multiple acidic residues is a well-known problem in IMAC experiments.10 Pinkse et al. observed coenrichment of acidic peptides with titansphere as well.24 To enhance the selectivity of the phosphopeptide enrichment, various conditions have been tested. A key parameter governing the selectivity of the phosphopeptide enrichment turned out to be the concentration of formic acid. Figure 4A shows a comparison of a multi-protease digest of mPER2 using either 30% ACN, 0.1% FA or 30% ACN, 2% FA for phosphopeptide binding and washing. With 0.1% FA, 28 phosphopeptides are specified in the Mascot Peptide Summary Report, accompanied by 58 mostly acidic non-phosphopeptides. With 2% FA, Mascot identifies 46 phosphopeptides and 19 non-phosphopeptides. Thus, not only was the relative amount of phosphopeptides increased but also the absolute number of phosphopeptides, indicating that phosphopeptides are missed in the nanoLC-MS/ MS analysis, if the background of non-phosphopeptides is too high. All 28 phosphopeptides identified with 0.1% FA are still present in the phosphopeptide-enriched fraction treated with 2%

FA. The majority of the acidic peptides still present with 2% FA are derived from a constricted acidic region of mPER2 and contain five or more glutamic or aspartic acid residues. Peptides with less than five acidic residues can be eliminated from the phosphopeptide-enriched fraction almost completely by using 30% ACN, 2% FA (see Figure 4B). Elevating the concentration of formic acid, thus lowering the pH and increasing the concentration of formate anions, seems to be a good remedy against coenrichment of acidic peptides with titansphere. Further increasing the concentration of formic acid to 5 or 10% results in even better selectivities, but also slightly reduced recovery rates (data not shown). Besides the concentration of formic acid, other parameters of the washing procedure, such as flow rate and duration, additionally affect the selectivity of the phosphopeptide enrichment and have to be adjusted to the size of the titansphere column and to the complexity of the applied sample. CONCLUSIONS The multi-protease approach outlined here has proved to be superior to the widely used trypsin digest in terms of phosphorylation site coverage. The low-specificity proteases elastase, proteinase K, and thermolysin, applied for the multi-protease approach in addition to trypsin, generate peptides well suited for the subsequent nanoLC-MS/MS analysis. These proteases are able to excise peptides from protein regions that are, due to the absence of tryptic cleavage sites, not accessible with trypsin. This advantage may be even more pronounced for other proteins, such as basic proteins (e.g., histones) or membrane proteins. Since the multi-protease approach generates a very large number of peptides, a highly specific phosphopeptide enrichment is indispensable, if applied for phosphorylation site mapping. The recently introduced phosphopeptide affinity material titansphere turned out to be well suited for this requirement. Compared to IMAC materials, titansphere seems to be superior in terms of selectivity and recovery. Titansphere handling is easier, since there is no requirement for metal loading, and less problematic, since there is no bleeding of metal ions that can cause severe troubles with subsequent LC-MS analysis. High recovery rates are obtained with titansphere by using the completely volatile buffer ammonium bicarbonate, whereas phosphate buffer is mostly used in IMAC experiments. Phosphate buffer requires a much more extensive desalting prior to ESI-MS. We detected a large variety of phosphopeptides, including multiple phosphorylAnalytical Chemistry, Vol. 77, No. 16, August 15, 2005

5249

ated peptides, as well as basic phosphopeptides. In addition, it turned out that titansphere is applicable for the enrichment of tyrosine-phosphorylated peptides as well (data not shown). In view of these excellent properties, this new type of phosphopeptide affinity material seems to be promising for various future applications. The combination of multi-protease digests, highly specific phosphopeptide enrichment with titansphere, and ultralow flow nanoLC-MS/MS analysis proved to be a successful combination for comprehensive phosphorylation site mapping in the lowfemtomole range.

5250

Analytical Chemistry, Vol. 77, No. 16, August 15, 2005

ACKNOWLEDGMENT We thank Anna Czypionka for performing initial tests with the low-specificity proteases, Ad de Jong (NanoSeparations) for his valuable support during installation of the nanoLC system, Wolf D. Lehmann for supporting with a high pressure cell and for critical reading of the manuscript, and the Deutsche Forschungsgemeinschaft (DFG) for financial support.

Received for review February 7, 2005. Accepted June 14, 2005. AC050232M