A Neutral Loss Activation Method for Improved Phosphopeptide

May 18, 2004 - Department of Chemistry, University of Virginia, McCormick Road, ... of Pathology, University of Virginia, Charlottesville, Virginia 22...
3 downloads 0 Views 198KB Size
Download PDF
Anal. Chem. 2004, 76, 3590-3598

A Neutral Loss Activation Method for Improved Phosphopeptide Sequence Analysis by Quadrupole Ion Trap Mass Spectrometry Melanie J. Schroeder,† Jeffrey Shabanowitz,† Jae C. Schwartz,§ Donald F. Hunt,†,‡ and Joshua J. Coon*,†

Department of Chemistry, University of Virginia, McCormick Road, Charlottesville, Virginia 22901, Department of Pathology, University of Virginia, Charlottesville, Virginia 22901, and Thermo Electron Corporation, 355 River Oaks Parkway, San Jose, California Recent advances in phosphopeptide enrichment prior to mass spectrometric analysis show genuine promise for characterization of phosphoproteomes. Tandem mass spectrometry of phosphopeptide ions, using collisionactivated dissociation (CAD), often produces product ions dominated by the neutral loss of phosphoric acid. Here we describe a novel method, termed Pseudo MSn, for phosphopeptide ion dissociation in quadrupole ion trap mass spectrometers. The method induces collisional activation of product ions, those resulting from neutral loss(es) of phosphoric acid, following activation of the precursor ion. Thus, the principal neutral loss product ions are converted into a variety of structurally informative species. Since product ions from both the original precursor activation and all subsequent neutral loss product activations are simulataneously stored, the method generates a “composite” spectrum containing fragments derived from multiple precursors. In comparison to analysis by conventional MS/MS (CAD), Pseudo MSn shows improved phosphopeptide ion dissociation for 7 out of 10 synthetic phosphopeptides, as judged by an automated search algorithm (TurboSEQUEST). A similar overall improvement was observed upon application of Pseudo MSn to peptides generated by enzymatic digestion of a single phosphoprotein. Finally, when applied to a complex phosphopeptide mixture, several phosphopeptides misassigned by TurboSEQUEST under the conventional CAD approach were successfully identified after analysis by Pseudo MSn. Phosphorylation is one of the most important reversible, covalent protein posttranslational modifications (PTM). In cells, phosphorylation is responsible for events such as initiating signal transduction,1 altering protein-protein affinity,2 and modulating kinase activity,3 among others. That said, phosphorylation site identification is a critical step toward the eventual understanding of these complex biological processes. * To whom correspondence should be addressed. E-mail: [email protected]. † Department of Chemistry, University of Virginia. ‡ Department of Pathology, University of Virginia. § ThermoElectron Corporation. (1) Hunter, T. Cell 2000, 100, 113-127. (2) Ichimura, T.; Wakamiya-Tsuruta, A.; Itagaki, C.; Taoka, M.; Hayano, T.; Natsume, T.; Isobe, T. Biochemistry 2002, 41, 5566-5572.

3590 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

A traditional analytical approach to monitor protein phosphorylation includes 32P-radiolabeling, separation by polyacrylamide gel electrophoresis (PAGE, one or two dimensions), and autoradiographic imaging. Next, phosphorylation sites are located by (1) chemical or enzymatic digestion, (2) chromatographic separation,4 and (3) sequence analysis with Edman degradation.5 Besides imposing large labor demands, the method typically requires picomole levels of sample. In contrast, antibodies present a sensitive, nonradioactive option for studying phosphorylated proteins. Unfortunately, generation of specific, high-affinity phosphoserine (pSer) or phosphothreonine (pThr) antibodies typically requires a priori knowledge of phosphorylated amino acid residues, with a few exceptions.6-8 This is an obvious limitation for large-scale phosphorylation site identification, especially since pSer and pThr account for the majority of phosphorylated amino acids in eukaryotic cells. Owing to its reliability, speed, and sensitivity, tandem mass spectrometry has become a valuable tool for peptide sequence analysis and PTM characterization.9-12 A typical mass spectrometric method includes (1) enzymatic protein digestion and (2) liquid chromatographic separation of the resulting peptides coupled online to (3) data-dependent MS/MS analysis. A limitation of this approach for phosphopeptide analysis is the large dynamic range between the nonphosphorylated and phosphorylated peptides, since phosphoproteins are typically present at low concentrations in comparison to their unmodified forms in the cell. Further, the dynamic range of three-dimensional ion trap mass (3) Chen, Z.; Gibson, T. B.; Robinson, F.; Silvestro, L.; Pearson, G.; Xu, B. E.; Wright, A.; Vanderbilt, C.; Cobb, M. H. Chem. Rev. 2001, 101, 2449-2476. (4) Boyle, W. J.; van der Geer, P.; Hunter, T. Methods Enzymol. 1991, 201, 110-149. (5) Roepstorff, P.; Kristiansen, K. Biomed. Mass Spectrom. 1974, 1, 231-236. (6) Gronborg, M.; Kristiansen, T. Z.; Stensballe, A.; Andersen, J. S.; Ohara, O.; Mann, M.; Jensen, O. N.; Pandey, A. Mol. Cell Proteomics 2002, 1, 517527. (7) Kane, S.; Sano, H.; Liu, S. C.; Asara, J. M.; Lane, W. S.; Garner, C. C.; Lienhard, G. E. J. Biol. Chem. 2002, 277, 22115-22118. (8) McLachlin, D. T.; Chait, B. T. Curr. Opin. Chem. Biol. 2001, 5, 591-602. (9) Mann, M.; Hendrickson, R. C.; Pandey, A. Annu. Rev. Biochem. 2001, 70, 437-473. (10) Hunt, D. F.; Yates, J. R.; Shabanowitz, J.; Winston, S.; Hauer, C. R. PNAS 1986, 83, 6233-6237. (11) Mann, M.; Jensen, O. N. Nat. Biotechnol. 2003, 21, 255-261. (12) Salomon, A. R.; Ficarro, S. B.; Brill, L. M.; Brinker, A.; Phung, Q. T.; Ericson, C.; Sauer, K.; Brock, A.; Horn, D. M.; Schultz, P. G.; Peters, E. C. PNAS 2003, 100, 443-448. 10.1021/ac0497104 CCC: $27.50

© 2004 American Chemical Society Published on Web 05/18/2004

spectrometers is low, ∼102. This combination often limits the applicability of mass spectrometry to phosphopeptide sequence analysis. Recently, our laboratory reported a novel strategy for enrichment of phosphopeptides capable of reducing this problem.13 This method selectively binds phosphorylated peptide methyl esters to an Fe3+-activated immobilized metal affinity (IMAC) column. After IMAC enrichment, phosphopeptide separation by nanoflow reversed-phase high-pressure liquid chromatography (RP-nHPLC) with online microelectrospray ionization for subsequent tandem mass spectrometry (RP-nHPLC/µESI-MS/MS) can result in thousands of acquired phosphopeptide tandem mass spectra.13 Notwithstanding, certain problems still remain in the application of tandem mass spectrometry to phosphoproteomics. Successful peptide identification requires random fragmentation along the peptide backbone to generate product ions (amide cleavage with collisional activation (amide cleavage with collisionally activated dissociation (CAD), b/y-type).14,15 However, exposure of phosphopeptide ions to CAD often results in extensive neutral loss(es) of phosphoric acid and generates tandem mass spectra containing few sequence-informative ions.16 This deficiency in peptide amide backbone cleavage can inhibit successful identification (up to 80%),13 either by automated search algorithms (SEQUEST, Mascot, etc.) or manual de novo peptide sequencing. Alternative strategies to enrich phosphopeptides and avoid the phosphopeptide fragmentation issue, discussed above, have been reported. These approaches share a common theme: chemical modification of pSer/pThr residue to remove the phosphate group.17 Modifications reported to date include sulfur derivatives,18 enzyme-cleavable sites,19 replacement of the phosphate moiety with affinity20 or isotope tags,21 and phosphopeptide mapping after phosphatase treatment.22 Though some of these approaches show promise, each requires additional sample manipulation to effect chemical modification. Besides increasing preanalysis labor demands, incomplete conversion in one or more of the steps can result in undesired side reactions,23 loss of analyte, or both. Furthermore, in some cases, the modification can still result in strong neutral losses (e.g., biotin). A preferred approach to phosphopeptide analysis would manipulate the process of ion dissociation to produce sequenceinformative product species and, thus, eliminate the need for chemical modification. Recently, a neutral loss-driven MS3 scan function was reported to improve phosphopeptide fragmentation (via CAD) in a linear ion trap mass spectrometer.24 Here a full(13) 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. (14) Roepstorff, P.; Fohlman, J. Biomed. Mass Spectrom. 1984, 11, 601. (15) Johnson, R. S.; Martin, S. A.; Biemann, K.; Stults, J. T.; Watson, J. T. Anal. Chem. 1987, 59, 2621-2625. (16) Tholey, A.; Reed, J.; Lehmann, W. D. J. Mass Spectrom. 1999, 34, 117123. (17) Molloy, M. P.; Andrews, P. C. Anal. Chem. 2001, 73, 5387-5394. (18) Li, W.; Boykins, R. A.; Backlund, P. S.; Wang, G.; Chen, H. Anal. Chem. 2002, 74, 5701-5710. (19) Knight, Z. A. S. B.; Row, R. H.; Kenski, D. M.; Gibson, B. W.; Shokat, K. M. Nat. Biotechnol. 2003, 21, 1047-1054. (20) Oda, Y. N. T.; Chait, B. T. Nat. Biotechnol. 2001, 19, 379-382. (21) Qian, W. J.; Gosche, M. B.; Camp, D. G.; Yu, L. R.; Tang, K. Q.; Smith, R. D. Anal. Chem. 2003, 75, 5441-5450. (22) Larsen, M. R.; Sorensen, G. L.; Fey, S. J.; Larsen, P. M.; Roepstorff, P. Proteomics 2001, 1, 223-238. (23) McLachlin, D. T.; Chait, B. T. Anal. Chem. 2003, 75, 6826-6836.

scan MS was followed by three data-dependent MS/MS events. A subsequent MS3 event (CAD of one MS/MS product ion) was triggered upon detection of -98, -49, or -32.7 from the precursor (corresponding to the neutral loss of phosphoric acid from 1+, 2+, and 3+ charge states, respectively) during any of the MS/ MS scans. Unfortunately, acquisition of the subsequent MS3 scan requires additional analysis time per peptide (∼double), an obvious concern for analysis of phosphopeptide enriched complex mixtures. Taking a different approach, our laboratory recently described a new strategy for ion dissociation of phosphopeptides in an ion trap mass spectrometer.25 The method, termed Pseudo MSn, induces constitutive collisional-activation of common neutral loss fragments following activation of the precursor ion, converting the dominant neutral loss product ions into a variety of structurally informative fragments. The Pseudo MSn approach is distinguished from the MS3 method, since product ions produced during the initial activation of isolated precursor ion and all subsequent neutral loss activations are simultaneously stored. Thus, the resulting Pseudo MSn mass spectrum comprises fragments derived from multiple precursor activations (a “composite” spectrum). In this report, we further characterize the method and its utility as applied to phosphopeptide ion dissociation. MATERIALS AND METHODS Instrumentation. All mass spectra were collected using electrospray ionization (ESI) on a quadrupole ion trap mass spectrometer (LCQ Deca XP, ThermoElectron, San Jose, CA). Data-dependent scanning was incorporated to select abundant precursor ions for both CAD-based fragmentation methods. In the conventional CAD mode, the instrument cycled through acquisition of a full-scan mass spectrum and five MS/MS spectra (five most abundant MS ions, two microscans per spectra) every ∼10 s, (3-Da window; precursor m/z (1.5 Da, 30% collision energy, 30-ms ion activation, 45-s dynamic exclusion, repeat count 2). For Pseudo MSn, the instrument control software was modified to allow up to 10 sequential, user-defined ion activations (each activation targeted ∼3 m/z wide window) following precursor activation. In these experiments, masses corresponding to single and multiple losses of [H3PO4] and [H3PO4 + H2O] (per desired charge state), were activated for a period of 10 ms in succession, followed by mass analysis. Synthetic Phosphopeptide Preparation and Analysis. Phosphopeptide standards were selected from a variety of ongoing, in-house projects on the basis of the presence of the characteristic phosphoric acid loss upon collisional activation: RLPIFNRIpSVSE, pSRpSFDYNYR, RpSpSGLpSRHR, RSMpSLLGYR, GpSPHYFSPFRPY, DRpSPIRGpSPR, LPISASHpSpSKTR, RRpSPpSPYYSR, SRVpSVpSPGR, APVApSPRPAApTPNLSK. Synthesis was performed using standard Fmoc solid-phase chemistry (AMS 422 multiple peptide synthesizer, Gilson, Middleton, WI). Sample was introduced by loading 200 fmol of each peptide onto a polyimide-coated, fused-silica microcapillary column (360-µm o.d. × 50-µm i.d., Polymicro Technologies, Phoenix, AZ) with an (24) Zumwalt, A.; Choudhary, G.; Cho, D.; Hemenway, E.; Mylchreest, I. Proc. 51st ASMS Confer. Mass Spectrom. Allied Top. 2003. (25) Coon, J. J.; Syka, J. E. P.; Schwartz, J. C.; Patrie, S. M.; Kelleher, N. L.; Shabanowitz, J.; Hunt, D. F. Proc. 51st ASMS Confer. Mass Spectrom. Allied Top. 2003.

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3591

integrated, laser-pulled, (P-2000 Sutter Instruments, Novato, CA) electrospray emitter. Microcapillary columns were packed with 5 cm of reversed-phase (RP) packing material (5-µm diameter, C18, YMC, Kyoto, Japan) with subsequent elution (Agilent 1100 series binary LC pump, Palo Alto, CA) at a flow rate of ∼60 nL/min, coupled to online analysis by tandem mass spectrometry (nHPLC/ µESI-MS/MS). Synthetic phosphopeptides were eluted with the following gradient: 0-60% B in 60 min, 60-100% B in 70 min; A ) 100 mM acetic acid (Sigma-Aldrich, St. Louis, MO) in water, B ) 70% aqueous acetonitrile (Mallinckrodt, Paris, KY) in 100 mM acetic acid. Conventional CAD and Pseudo MSn were performed, in triplicate, separately. In this case, Pseudo MSn targeted neutral losses derived from doubly charged phosphopeptide molecular ions (loss of 49, 58, 98, and 107, corresponding to [H3PO4], [H3PO4 + H2O], 2[H3PO4], 2[H3PO4 + H2O], respectively). Single Protein Analysis. A FLAG-tagged protein was purified by immunoprecipitation from HEK293 cells following treatment with the phosphatase inhibitors calyculin A and peroxovanadate. This highly phosphorylated protein was reduced for 1 h at 51 °C with dithiothreitol (DTT), (Amersham Pharmacia Biotech, Piscataway, NJ) (0.1 mM final concentration) and alkylated for 45 min at room temperature (RT) with iodoacetamide (Sigma-Aldrich, St. Louis, MO) (0.2 mM final concentration) in 100 mM ammonium bicarbonate buffer, pH 8.5. The protein was enzymatically digested with trypsin (Promega, Madison, WI; 1:20, enz/subs) for 4 h at 37 °C, followed by 4 h at RT. The resulting peptide mixture was acidified to pH 3.5 with acetic acid and loaded onto a microcapillary “precolumn” (360-µm o.d. × 75-µm i.d.) with 5 cm of 5-20 µm C18 packing material (YMC, Kyoto, Japan). Next, the precolumn was butt-connected with PTFE tubing (0.06-in. o.d. × 0.012-in. i.d., Zeus Industrial Products, Orangeburg, SC) to an analytical column (described above). Peptides were eluted over a 2 h gradient (0-60% B in 120 min, 60-100% B in 130 min) and analyzed first with the conventional CAD mode and second using Pseudo MSn (2+ charge state condition), Then, the sample was analyzed a third time under modified Pseudo MSn conditions to targeting neutral loss m/z values from both 2+ (49, 58, 98; [H3PO4], [H3PO4 + H2O], 2[H3PO4]) and 3+ (32.6, 38.6, 65; [H3PO4], [H3PO4 + H2O], 2[H3PO4]) charge states. Note activation of -98 targets either the neutral loss of two phosphoric acid residues from a doubly charged phosphopeptide or one phosphoric acid loss from a singly charged phosphopeptide. Here we limit the number of neutral loss activations to six, although more could be added at the expense of longer activation times. Complex Mixture Analysis. A complex mixture of human nuclear proteins containing unidentified phosphorylation sites was purified from a nuclear pellet of HEK293 cells using the Trizol reagent. The precipitated proteins were dissolved with 1% SDS to a final concentration of 5 µg/µL. An aliquot corresponding to ∼300 µg was diluted 10-fold with 100 mM ammonium bicarbonate and digested with trypsin (1:20) at 37 °C overnight. Peptides were acidified to pH 3.5 with acetic acid. Peptide carboxyl groups were converted to methyl esters as previously described.13 The reagents were lyophilized, and the sample was reconstituted in 1:1:1 solution of acetonitrile/methanol/0.1% aqueous acetic acid. The mixture was loaded onto an Fe3+-activated IMAC column (360µm o.d. × 100-µm i.d.) packed with 6 cm of POROS 20 MC metal chelate affinity packing material (Perceptive Biosystems, Framing3592

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

ham, MA). The phosphopeptides were then eluted onto a C18 microcapillary precolumn using 20 µL of 250 mM Na2HPO4 (Sigma, St. Louis, MO) (adjusted to pH 6 with 95% formic acid, Sigma, St. Louis, MO). Finally, the phosphopeptides were eluted with a 1-h gradient (0-60% B in 60 min, 60-100% B in 70 min) and analyzed once using the conventional CAD mode and a second time with Pseudo MSn under the simultaneous 2+ and 3+ charge state activation conditions described above. Peptide Identification. TurboSEQUEST26 [v.27 (rev. 11), (c) 1999-2002] supplied by Thermo Electron (San Jose, CA) was used for automated peptide identification. Presented sequences were validated by manual interpretation of the corresponding spectra. RESULTS AND DISCUSSION In this study, we investigate the potential of a new neutralloss, ion-activation method, Pseudo MSn, as applied to phosphopeptide ion dissociation. A schematic illustrating the various CADbased approaches for phosphopeptide ion dissociation, including Pseudo MSn, is presented in Figure 1. Figure 1A displays a fullscan MS spectrum where, in our case, the five most abundant ions will be selected for further interrogation - MS/MS. For this demonstration, let us assume the base peak in Figure 1A represents a doubly charged ion derived from a doubly phosphorylated peptide. Application of the conventional CAD approach typically results in product ion spectra dominated by the neutral loss of phosphoric acid (Figure 1B). For multiply phosphorylated peptides, multiple neutral loss products are common. And though the efficiency of ion dissociation is high, the bulk of the precursor ion signal is simply converted to fragments bearing little or no additional sequence information (e.g., [M - nH3PO4]2+). From this point, as described by Zumwalt et al.,24 one of the neutral loss products can be activated, following isolation, to generate a MS3 spectrum ([M - H3PO4]2+ in this case), characterized by Figure 1C. For multiply phosphorylated peptides, such as this one, the neutral loss-driven MS3 approach will likely generate a second product spectrum dominated by loss of another phosphoric acid. Meanwhile, the low-abundance, sequenceinformative product ions, produced during the initial activation, are lost during isolation of the major neutral loss product. In contrast, Pseudo MSn does not incorporate an additional step of product ion isolation prior to neutral loss activation; by doing so, Pseudo MSn offers several unique attributes for phosphopeptide ion dissociation. First, the method allows for multiple neutral loss activation events and, thus, can target multiple neutral losses. Second, product ions derived from these multiple precursor activations are simultaneously stored, resulting in a single mass spectrum comprising product ion species derived from multiple ion activation events (Figure 1D, dashed/solid lines). Finally, short activation times (10 ms) allow multiple neutral loss activations on a time scale comparable to the conventional CAD approach. Table 1 highlights potential neutral loss values observed following CAD of phopshopeptides. Synthetic Phosphopeptide Analysis. To determine the effect of the Pseudo MSn method, a series of phosphopeptides were synthesized and analyzed under data-dependent conditions with both conventional CAD and Pseudo MSn conditions. A comparison (26) Eng, J. K. M. A. L.; Yates, J. R., III J. Am. Soc. Mass Spectrom. 1994, 5, 976-989.

Figure 1. Schematic illustration of CAD-based methods for phosphopeptide ion dissociation: (A) full-scan mass spectrum containing a doubly charged, doubly phosphorylated peptide; (B) MS/MS spectrum containing typical fragment ions following conventional phosphopeptide ion dissociation of precursor ion by CAD (note the majority of fragment ion signal is represented by loss of two phosphoric acid residues, denoted by P); (C) MS3 spectrum following isolation and activation of the most abundant neutral loss product ion (dashed lines indicate fragment ions produced as a result of neutral loss activation); note that fragment ions generated from original MS/MS event are not retained; (D) Pseudo MSn spectrum, a composite containing fragment ions generated from both initial MS/MS event (solid lines) and subsequent activations of the neutral loss product ions (dashed lines).

Table 1. Potential Neutral Loss Values Following Phosphopeptide Fragmentation with Collisionally Activated Dissociationa neutral loss moeity

+1

H3PO4 H3PO4 + H2O 2H3PO4 2H3PO4 + H2O 2H3PO4 + 2H2O 3H3PO4 + H2O 3H3PO4 + 2H2O

98 116 196 214 232 312 330

neutral loss values +2 +3 49 x 58 x 98 x 107 x 116 156 165

32.7 x 38.7 65.4 x 71.4 77.4 104 110

+4 24.5 29 49 53.5 58 78 82.5

a Checked losses indicate those activated in this study. Note some values overlap (e.g., H3PO4 from a singly charged peptide with 2H3PO4 from a doubly charged precursor) so that one activation targets either neutral loss product. Here, only losses from the [M + 2H]2+ and [M + 3H]3+ precursors were targeted, but overlapping values from other charge states are also activated (nonchecked m/z values shown in bold).

of the two ion dissociation methods for the doubly charged, singly phosphorylated peptide RLPIFNRIpSVSE, is presented in Figure 2. Under conventional CAD conditions, the ion dissociates mainly via phosphoric acid loss with relatively few b/y product ions

(Figure 2A). The Pseudo MSn spectrum, however, is devoid of the major neutral loss fragment ion (Figure 2B). Instead, increases in abundance of nearly every b/y-type ion is observed, along with several new cleavages (e.g., y7, y8, y9, y11, y7 - H3PO4, y10 - H3PO4, y11 - H3PO4, b5, b7, b112+ - H3PO4). The amino acid sequence of the phosphopeptide and the predicted fragment m/z values for b/y-type ions are shown; underlined masses indicate the detection of a b/y product ion, whereas low-abundance product ions are indicated by a dashed underline. When present, neutral loss(es) of phosphoric acid from b/y ions are indicated in italics (where P denotes phosphoric acid). Note the intensity scales on both spectra are identical. The improvements highlighted in Figure 2 were typical for most of the synthetic phosphopeptides analyzed in this study (data not shown). To assess the magnitude of enhancement, mass spectra resulting from each method were matched to the correct peptide sequence and scored by a computer algorithm, TurboSEQUEST. A summary of these results is presented in Figure 3. From these data, the cross correlation scores (Xcorr) of the synthetic phosphopeptides are improved by use of Pseudo MSn for 7 of the 10 peptides studied, translating to higher confidence phosphopeptide sequence analysis. Single Protein Analysis. Although the synthetic peptide study presented above demonstrates the ability of Pseudo MSn to Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3593

Figure 2. Product ion spectra of the [M + 2H]2+ ion from RLPIFNRIpSVSE (m/z ) 756) acquired by (A) conventional CAD, and (B) Pseudo MSn. Underlined m/z values indicate the presence of b/y-type fragment ions, whereas low-abundance product ions are indicated by a dashed underline. Loss of phosphoric acid (-P) is indicated in italics. Note the spectra are normalized to each other.

improve phosphopeptide ion dissociation, it does not represent a real-world application, nor does it provide means for a negative control. To address this, we applied the method to the analysis of a single phosphoprotein digest containing both non-phosphorylated and phosphorylated peptides. This allowed us to determine the effect of Pseudo MSn on the most abundant phosphopeptides 3594 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

and their nonphosphorylated forms. Figure 4A demonstrates the average improvement in phosphopeptide Xcorr upon application Pseudo MSn to a representative set of phosphopeptides contained in the sample (∼2 to 6 spectra were recorded for each peptide, depending on its abundance and chromatographic peak width). Twenty-two pSer-containing peptides were observed. Of those, 16

Figure 3. Comparison of cross-correlation scores (Xcorr) assigned to spectra acquired from analysis of synthetic phosphopeptides by either conventional CAD or Pseudo MSn. Error bars represent one standard deviation of the mean.

had increased Xcorr values (∆ > +0.3 units) after analysis with Psuedo MSn. Figure 4B demonstrates little, if any, effect is observed when nonphosphorylated peptides (characterized during the same analysis) were exposed to Pseudo MSn conditions, whereas 16 of the 22 phosphopeptides showed improved Xcorrs. Note that the Xcorr for ASEEEHVpYSFPNK does not improve (see starred pair in Figure 4A). Although neutral loss of 98 (HPO3 + H2O) from phosphotyrosine-containing peptides has been reported,27,28 no detectable neutral losses were observed for the phosphotyrosine-containing peptides analyzed here (data not shown). This is supported by the minimal change in Xcorr between the two dissociation methods. On the other hand, the corresponding ApSEEEHVYSFPNK peptide, which contains a pSer residue, undergoes extensive neutral loss of phosphoric acid (Figure 5A), and the Pseudo MSn method increases its Xcorr by one unit (Figure 4A). Figure 5A and B illustrate the enhancement of phosphopeptide spectra when Pseudo MSn is applied. Again, analysis with the conventional CAD produces a product spectrum dominated by neutral loss of phosphoric acid ([M + 2H - P]2+, Figure 5A), whereas the Pseudo MSn spectrum lacks this ion (Figure 5B). Analysis with the Pseudo MSn method results in substantial increases in b/y-type product ions (solid underline). Further, because Pseudo MSn activates product ions after loss of phosphoric acid, additional information is often gained in the form of b/y ion - H3PO4, a benefit for de novo sequencing. Note the enhancement presented in Figure 5 is not due to an increase in precursor signal intensity (similar precursor intensity for both methods), as evidenced by a 2.0, and after manual interrogation, only three peptide sequences could be confirmed. Application of Pseudo MSn improved scores for those phosphopeptides and correctly assigned another five phosphopeptides (manually validated) previously misassigned following conventional CAD analysis. From this analysis and those previously described (see above), we conclude that Pseudo MSn is most effective for singly and doubly phosphorylated peptides. Further, peptides containing multiple highly basic residues (missed tryptic cleavages), which prohibit random protonation along the backbone, are difficult to sequence following collisional activation in general. Automated Search Algorithms. Two important parameters evaluated during SEQUEST peptide scoring include the presence and signal intensity of b and y fragment ions. Thus, the improvement in phosphopeptide Xcorr using Pseudo MSn is most likely due to increased production of b/y-type ions and improved relative product ion signal intensities, which are no longer normalized to the abundant neutral loss. To our knowledge, no presently available software considers fragment ions in combination with and without their potential neutral loss(es) of phosphoric acid (e.g., b7 and b7 - H3PO4, etc.) in either the identification of the peptide sequence or the assignment of the phosphorylation site. MacCoss and co-workers report improved phosphopeptide identification with a new search algorithm, SEQUEST-PHOS, though details and availability of this program were not described.29 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3595

Figure 4. Comparison of cross-correlation scores (Xcorr) assigned to spectra acquired from analysis of peptides generated from a single phosphoprotein digest by either conventional CAD or Pseudo MSn for (A) phosphorylated peptides and (B) nonphosphorylated peptides. (f) Note that ApSEEEHVYSFPNK shows improved Xcorr with Pseudo MSn, but ASEEEHVpYSFPNK does not (see Results and Discussion for details).

Compared to conventional CAD phosphopeptide ion dissociation, Pseudo MSn increases the number and intensity of fragment ion signals; however, many of these increased or newly formed products have already experienced phosphoric acid loss (Figures 2 and 5B). Hence, we expect further improvements in the scoring and automated identification of phosphopeptides as software is (29) 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, J. R. I. PNAS 2002, 99, 7900-7905.

3596 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

developed to include these additional product ions. We recognize that the characteristic neutral loss signature ion(s) generated under conventional CAD conditions are often useful for determination of both the charge state and the number of phosphorylation sites present; however, neutral losses alone do not facilitate peptide sequence identification. In Pseudo MSn, phosphoric acid neutral losses are often observed within the product ions (e.g., b7 and b7 - H3PO4). The presence or absence of these ions can be utilized to distinguish phosphopeptide spectra from their nonphosphory-

Figure 5. Product ion spectra of the [M + 2H]2+ ion for ApSEEEHVYSFPNK (m/z ) 808.5) resulting from (A) conventional CAD (inset is the expanded view), and (B) Pseudo MSn. Underlined m/z values indicate the presence of b/y-type fragment ions, whereas low-abundance product ions are indicated by a dashed underline. Loss of phosphoric acid (- P) is indicated in italics. Note the spectra are normalized to each other.

lated counterparts contained in a data set (as in the case of our single protein analysis). At present, we are developing software to selectively distinguish phosphopeptide from nonphosphorylated peptide spectra resulting from Pseudo MSn analysis. Alternatively, the mass spectrometer could be operated to acquire a conventional CAD tandem mass spectrum, followed by Pseudo MSn analysis. That mode of operation would offer the diagnostic neutral loss fragment ions but would increase the analysis time per peptide.

CONCLUSIONS Pseudo MSn is a useful alternative to conventional collisionalactivated phosphopeptide ion dissociation. Activation of multiple neutral losses and simultaneous storage of fragment ions, derived from multiple activations, allows one to generate higher-quality spectra than by MS3 or CAD alone. Pseudo MSn increases both the quantity and intensity of fragment ion signal; thus, the scores generated by automated search algorithms increase, leading to Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3597

higher confidence in phosphopeptide sequence identification. When the Pseudo MSn method was applied to a series of synthetic phosphopeptides, improved cross-correlation scores (Xcorrs) for 7 of the 10 peptides studied were observed. Application of the method to a single protein digest yielded an improvement in Xcorr for 16 of the 22 observed phosphopeptides. In addition, the impact on nonphosphorylated peptides was negligible, allowing Pseudo MSn to be applied to the analysis of peptide mixtures independent of enrichment. Finally, this method was employed to successfully identify phosphopeptides whose spectra were uninterpretable with CAD alone. In most cases, the Xcorrs of the phosphopeptides studied were improved using Pseudo MSn, whereas a minority were unaffected or slightly lower. The most substantial impact was observed with singly or doubly phosphorylated peptides at lower charge (charge e3). Inclusion of enhanced neutral loss product species, formed

3598

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

during Pseudo MSn, by automatic search algorithms is expected to further improve automatic sequence identification. Finally, the method has attendant possibilities for use on other compound classes where neutral losses are typical. ACKNOWLEDGMENT The authors gratefully acknowledge the National Institutes of Health (D.F.H., GM 37537; J.J.C., F32 RR 018688-01, postdoctoral fellowship) for generous support. We thank Donna J. Webb, Alan Rick Horwitz, Rob A. Figler, and Joel Linden for providing the protein samples and Mark D. Platt and John E.P. Syka for helpful discussions. Received for review February 20, 2004. Accepted April 9, 2004. AC0497104