Neutral Loss-Based Phosphopeptide Recognition: A Collection of Caveats Wolf D. Lehmann,* Ralf Kru1 ger,† Mogjiborahman Salek,‡ and Chien-Wen Hung Central Spectroscopy, German Cancer Research Center, Heidelberg, Germany
Florian Wolschin§ and Wolfram Weckwerth Max-Planck-Institute for Molecular Plant Physiology, Golm, Germany Received October 31, 2006
Abstract: The standard strategy for analysis by tandem mass spectrometry of protein phosphorylation at serine or threonine utilizes the neutral loss of H3PO4 () 97.977/ z) from proteolytic peptide molecular ions as marker fragmentation. Manual control of automatically performed neutral loss-based phosphopeptide identifications is strongly recommended, since these data may contain false-positive results. These are connected to the experimental neutral loss m/z error, to competing peptide fragmentation pathways, to limitations in data interpretation software, and to the general growth of protein sequence databases. The fragmentation-related limitations of the neutral loss approach cover (i) the occurrence of abundant ‘close-to-98/z’ neutral loss fragmentations, (ii) the erroneous assignment of a neutral loss other than loss of H3PO4 due to charge state mix-up, and (iii) the accidental occurrence of any fragment ion in the m/z windows of interest in combination with a charge-state mix-up. The ‘close-to-98/z’ losses comprise loss of proline (97.053/z), valine (99.068/z), threonine (101.048/z), or cysteine (103.009/z) preferably from peptides with N-terminal sequences PP, VP, TP, or CP, and loss of 105.025/z from alkylated methionine. Confusion with other neutral losses may occur, when their m/z window coincides with a 98/z window as result of a charge state mix-up. Neutral loss of sulfenic acid from oxidized methionine originating from a doubly charged precursor (63.998/2 ) 31.999) may thus mimic the loss of phosphoric acid from a triply charged phosphopeptide (97.977/3 ) 32.659). As a consequence of the large complexity of proteomes, peptide sequence ions may occur in one of the mass windows of H3PO4 loss around 97.977/z. Practical examples for false-positive annotations of phosphopeptides are given for the first two groups of error. The majority of these can be readily recognized using the guidelines presented in this study. * Wolf D. Lehmann, Prof. Dr., Central Spectroscopy, German Cancer Research Center, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. Tel.: ++49-6221-424563. Fax: ++49-6221-424554. E-mail: wolf.lehmann@ dkfz.de. † Current address: Institute of Clinical Chemistry and Laboratory Medicine, University Hospital Mainz, Germany. ‡ Current address: Sir William Dunn School of Pathology, Oxford, UK. § Current address: School of Life Sciences, Arizona State University, Tempe, AZ, U.S.A.
2866
Journal of Proteome Research 2007, 6, 2866-2873
Published on Web 06/15/2007
Keywords: neutral loss • parent ion scan • phosphopeptide • phosphoserine • phosphothreonine • tandem mass spectrometry • MS3 • database searching • CID • mass accuracy • methionine oxidation
Introduction Reversible protein phosphorylation is crucial for many regulatory and functional features of living organisms including, but not restricted to, enzyme activity, metabolic control, protein localization, cell division, and signal transduction.1,2 Mass spectrometry has developed into a key technology for the recognition of protein phosphorylation and the pinpointing of phosphorylation sites. An established mass spectrometric strategy for detection of peptides phosphorylated at serine or threonine is based on collision-induced dissociation (CID) which induces the loss of phosphoric acid.3 Neutral loss of H3PO4 is specific for pSer/pThr-phosphopeptides and often proceeds more readily than the fragmentation of the peptide backbone. Therefore, the product ion spectra of pSer/pThr phosphopeptides typically show the fragment ion generated by neutral loss of H3PO4 as the most prominent signal accompanied by a few sequence-specific fragment ions of moderate to low abundance. In case of collision cell MS/MS, this is particularly observed for spectra recorded at low offset. Increasing the collision offset induces multistep fragmentations so that peptide backbone cleavages occur. In ion trap MS/MS, the preference of neutral loss of H3PO4 over formation of backbone fragments is often enhanced compared to collision cell fragmentation. In this case, fragmentation of the M-H3PO4 ion using an MS3 experiment may provide the desired sequence information.4-7 With the often abundant loss of H3PO4, pSer/ pThr phosphopetides exhibit a unique fragmentation reaction. Its usage for automated detection of phosphopeptides in parent ion scans or LC-MSn runs, however, is limited by the complex and sequence-dependent fragmentation behavior of peptides, by instrumental features, and by shortcomings of softwaresupported data evaluation. Currently, phosphoproteome studies report the identification of large numbers of phosphorylation sites.8,9 It is generally recognized that manual control of search engine-assigned phosphorylation sites is mandatory, since these currently may contain a substantial portion of falsepositive hits. Careful quality control in phosphorylation site pinpointing is also of utmost importance, given that it is very tedious and time-consuming to proof the false-positive nature of a published phosphorylation site. This is due to the fact that the lack of detection of a phosphorylation site, for example, 10.1021/pr060573w CCC: $37.00
2007 American Chemical Society
technical notes
Lehmann et al.
Figure 1. Partial nanoESI-MS/MS spectrum of the tryptic R-casein peptide VPQLEIVPN-pS-AEER. The peptide shows two consecutive neutral loss reactions, loss of the N-terminal valine residue and loss of H3PO4. Considering only the two neutral loss events with an m/z tolerance >0.4, the analyte may be assigned as a doubly phosphorylated instead of a monophosphorylated species. (nl ) neutral loss)
by another investigator, does not proof the absence of this site, since MS data in general do not justify a solid statement about some nondetected analyte. This study aims at an improved quality control of mass spectrometric phosphopeptide recognition based on neutral loss of H3PO4 by highlighting the most likely pitfalls.
Materials and Methods Protein Sample Preparation and Liquid Chromatography. Protein samples were purified by standard 1D-polyacrylamide gel electrophoresis and stained by Coomassie. Protein bands were cut and processed for MS analysis using trypsin following standard procedures.10 The observed oxidation of methionine is probably a side reaction during sample work up.11,12 The sample of golgin-84 was kindly supplied by M. Lowe, University of Manchester. Q-TOF nanoESI-MS/MS. Borosilicate spray capillaries were manufactured in-house using a micropipette puller type P-87 (Sutter Instruments, Novato, CA) and coated with a semitransparent film of gold in a sputter unit type SCD 005 (BAL-TEC AG, Balzers, Liechtenstein). A Q-TOF instrument type Q-Tof2 (Waters Micromass, Manchester, U.K.) was used. Q-TOF parent ion scanning was performed with a precursor ion isolation window of 3.5 Da and a step width of 2 Da over a mass range from m/z 400 to m/z 1300. The data acquisition time was 10 s/spectrum. The collision offset was set to 25 V. The MS/MS spectra were acquired over a range from m/z 50 to m/z 2000. Evaluation of Q-TOF parent ion scan data for neutral loss was performed with a data window equivalent to the step width for data acquisition (2 Da). The software version used was MassLynx 3.5. Linear Ion Trap µLC-ESI-MS/MS. Peptides were eluted from the reversed-phase µLC column directly into an LTQ mass spectrometer (Thermo, San Diego, CA). MS2 was triggered for the three most abundant peaks in each spectrum. If a MS2 spectrum showed a signal with a mass difference of 98, 49 () 98/2), or 32.7 () 98/3) on the low-mass side of the precursor ion, an MS3 experiment was triggered automatically for this fragment. The isolation window was set to 3 m/z, the collision energy to 35, and the activation time to 30 ms. With the use of the search tool Sequest (Thermo, San Diego, CA), the spectra
were matched against an Arabidopsis thaliana sequence database (http://www.arabidopsis.org/) containing additional trypsin and keratin sequences. The MS/MS data analysis was performed including the modifications + 80 Da for S, T, and Y, -18 Da for S and T, and + 16 Da for M. For refined searches, a modification of -48 Da for M was additionally included. The Xcorr threshold was set to 2.0, 2.5, and 3.5 for singly, doubly, and triply charged peptides, respectively.
Results and Discussion General. Neutral loss of H3PO4 from pS/pT phosphopeptides is a specific event, since it is directly connected with phosphorylation and thus cannot occur from unmodified peptides or from peptides carrying other common covalent modifications. Noncovalent adduction of H3PO4 to proteins has been described,13 which, upon CID, results in its neutral loss. However, for peptides analyzed under standard conditions of acidic pH, this phenomenon is not of importance. The region on the low-mass side of the precursor ion is the region with the highest density of fragment ions in an MS/MS spectrum of peptides, since there both the ions from diverse neutral loss reactions14,15 and from sequence-specific backbone fragmentations are observed. Therefore, an enhanced probability of overlap of fragment ions exists in this region, in particular since a majority of MS and MS/MS data are currently acquired at low to medium resolution. Hence, ambiguities in H3PO4 lossdirected phosphorylation analysis arise mainly from problems encountered with its detection and assignment. This is particularly critical in case data acquisition and evaluation is performed under software control, which often does not make use of the full information content of the available data. Fragmentation-related limitations of the neutral loss approach are due to (i) the occurrence of abundant ‘close-to-98/z’ neutral loss fragmentations, (ii) the erroneous assignment of a neutral loss other than loss of H3PO4 due to charge state mix-up, and (iii) the accidental occurrence of any fragment ion in the neutral loss windows of interest ()97.977/z ( experimental error) in combination with a charge-state mix-up. For the last category, it is currently difficult to present rules. However, for the first two classes of interference, the most likely sources of error are presented and discussed below. Journal of Proteome Research • Vol. 6, No. 7, 2007 2867
Neutral Loss-Based Phosphopeptide Assignment
technical notes
Figure 2. Positive parent ion scan of a tryptic digest of recombinant golgin-84, evaluated for neutral loss of 24.5 ()98/4) with the aim to detect phosphopeptides via neutral loss of H3PO4 from quadruply charged precursor ions. Two peak groups with an m/z difference of 20 () 80/4) are observed. The pair at m/z 626/646 represents a peptide/phosphopeptide pair, the peaks at 746/766 are caused by a singly and doubly phosphorylated species of another peptide. For the corresponding MS/MS spectra, see Figure 3.
Ser/Thr- Phosphorylation Mimicry by a ‘Close-to-98/z’ Loss. Figure 1 shows the partial MS/MS spectrum of a triply charged tryptic peptide of R-casein, which upon preliminary inspection was assumed to be a doubly phosphorylated peptide, because of the two abundant triply charged fragment ions with a mass difference of about 33 m/z to the molecular ion. However, exact mass measurement of the mass differences clearly showed the presence of two different fragmentation processes. First, a loss of 99.09 Da (valine, C5H9NO, 99.068, calc.) occurred followed by loss of 97.98 Da () H3PO4, 97.977, calc.). Interpretation of the complete MS/MS spectrum showed the presence of the monophosphorylated R-casein peptide VPQLEIVPN-pS-AEER. This identification fits to the abundant neutral loss of valine from the N-terminus, which is a typical feature of the N-terminal peptide sequence motif VP.14 Manual determination of the exact value of a neutral loss is straightforward from an MS/MS spectrum as displayed in Figure 1, where the monoisotopic species of both the molecular ion and of the fragment ion can be clearly identified. Nevertheless, software-supported neutral loss evaluations may interpret the spectrum in Figure 2 as exhibiting two H3PO4 loss events. This is due to the broad precursor ion isolation window (23.5 Da), where the precursor ion m/z value is represented by the so-called setmass16 which is located within the precursor ion isolation window. A related false-positive indication of a pS/pT-phosphorylation was observed in the evaluation of a Q-TOF parent ion scan of a tryptic digest of recombinant golgin-84. Figure 2 shows the evaluation of this parent ion scan for the occurrence of a neutral loss of 24.5, which was performed to detect pS/pT phosphopeptides with the charge state 4+. Two sets of peaks with a mass difference of 20 were observed (at m/z 626 and 646, and at m/z 746 and 766, respectively), where the mass difference suggests the presence of two phosphopeptide pairs, consisting each of a mono- and doubly phosphorylated species. However, manual inspection of the MS/MS spectra showed that this is only valid for the pair of signals found at m/z 746 and 766. In contrast, the signal pair at m/z 626 and 646 was found to originate from a peptide/phosphopeptide pair (TPVEASHPVENASVPRPSSHFR, optional phosphorylation at the underscored S). The partial MS/MS spectra of these quadruply charged peptides are given in Figure 3. The nonphosphorylated peptide (Figure 3a) shows an abundant neutral loss of the N-terminal 2868
Journal of Proteome Research • Vol. 6, No. 7, 2007
threonine residue (loss of 101), which due to the noisy molecular ion pattern is difficult to discriminate from a possible loss of H3PO4, even with manual evaluation. Similar to the example given in Figure 1, the N-terminal motif XP (X: any amino acid) is responsible for the abundant neutral loss of threonine. For a quadruply charged precursor ion, the m/z difference between loss of H3PO4 and loss of Thr is only 0.75 m/z units. The corresponding MS/MS spectrum of the phosphorylated analogue (Figure 3b) exhibits two abundant consecutive losses, loss of H3PO4 and loss of Thr. Again, the monoisotopic ion of the precursor is difficult to recognize; however, the mass difference between the two consecutive neutral loss reactions can be clearly assigned to loss of Thr. The reverse order of the two consecutive losses in Figure 1 (first loss, Val; second loss, H3PO4) and Figure 3b (first loss, H3PO4; second loss, Thr) points to an influence of the peptide sequence which currently cannot be explained. Table 1 presents a summary of ‘close-to-98/z’ neutral loss reactions, which may give rise to a false-positive assignment of a peptide as phosphopeptide, when the neutral loss criterion (identical precursor/product charge state) is fulfilled. Only charge states from +1 to +4 are considered in this Table, since in our experience higher charge states are rarely observed for proteolytic peptides generated by common proteases such as trypsin, chymotrypsin, AspN, pepsin, or elastase. The charge state 4+ may occur occasionally when proteases other than trypsin are used, due to the presence of several basic residues in one peptide. The risk of charge state mix-up increases with increasing charge state of the precursor ion and with decreasing mass spectrometric resolution. Drawbacks of the evaluation software with respect to accurate mass determination of the precursor or product ion are another source of error. In manual evaluation, difficulties may persist, when the mass of the precursor ion cannot be unambiguously determined in the survey spectrum (MS1 spectrum) due to low signal-to-noise ratio or due to overlapping peaks or background ions. The possible interferences listed in Table 1 are equally relevant for Q-TOF and ion trap instruments, since, with respect to the predominant occurrence of neutral loss events of peptides, the MS/MS spectra observed with both types of instruments are very similar. Figure 4 shows the abundances of N-terminal sequence motifs in tryptic peptides from human proteins, which will give rise to abundant close-to-98/z neutral loss signals caused by loss of P, V, or T.
technical notes
Lehmann et al.
Figure 3. Partial positive ion nanoESI-MS/MS spectra giving rise to the peaks at m/z 626 and m/z 646 in the Q-TOF neutral loss analysis (Figure 2). (a) precursor ion setmass 626, the spectrum shows neutral loss of Thr from TPVEASHPVENASVPRPSSHFR; (b) precursor ion setmass 646, the spectrum shows consecutive loss of H3PO4 and Thr from TPVEASHPVENA-pS-VPRPSSHFR. Incorrect assignments are printed in italics. Table 1. Peptide Modifications Showing an Abundant ‘Close-to-98/z’ Neutral Loss Compared to the Specific Loss of H3PO4 from pSer/pThr Phosphopeptidesa structure
neutral loss
1+
2+
3+
4+
phosphoS/phosphoT P -P... (N-term) C-sulfo V -P... (N-term) T -P... (N-term) ...-dhBA (C-term) C -P... (N-term) ...- S (C-term) M-acetamido
H3PO4 Pro SO3 + H2O Val Thr dhBA Cys Ser + H2O MTA
97.9769 97.0528 97.9674 99.0684 101.0477 101.0477 103.0092 105.0426 105.0248
48.9884 48.5264 48.9837 49.5342 50.5238 50.5238 51.5046 52.5213 52.5124
32.6590 32.3509 32.6558 33.0228 33.6826 33.6862 34.3364 35.0142 35.0083
24.4942 24.2632 24.4918 24.7671 25.2619 25.2619 25.7523 26.2606 26.2562
a Neutral loss m/z values for the charge states +1 to +4 are listed (dhBA, dehydrobutyric acid; MTA, 2-(methylthio)acetamide; all amino acid symbols refer to the amino acid mass minus water).
N-terminal neutral loss reactions may also be induced when basic residues are present in the fourth position from the N-terminus (not included in Figure 4). H3PO4 Loss Mimicry by Precursor/Product Charge State Mix-Up. An efficient method for data-directed analysis of phosphopeptides consists of an LC-MS method, where the instrument screens all MS2 spectra for a dominant loss of 98/ z. For instance, in a screening procedure for triply charged peptides, an MS3 scan is triggered automatically, in case one of the most abundant peaks in the MS2 spectrum exhibits a neutral loss of 98/3 ) 32.7. In the M-H3PO4 fragment ion selected as precursor for the MS3 experiment, the ex-pSer- or ex-pThr-site is represented by a dehydroalanine or dehydrobutyric acid residue, respectively. Both residues can be included in database searches by search engines like Sequest or Mascot by allowing a variable modification of -18 Da for Ser and Thr. This strategy was developed, since in particular ion trap MS2 spectra of phosphopeptides often exhibit an abundant loss of
H3PO4 accompanied by only a few sequence-specific ions of low abundance. In contrast, the MS3 spectra of the M-H3PO4 ions then show the normal variety of sequence-specific ions as observed for unmodified peptides. A complex mixture of tryptic peptides obtained by digestion of an enriched phosphoprotein fraction of A. thaliana was subjected to LC-MS/MS with data-directed initiation of MS3 as described above. Recently, we observed that, when the instrument was set up for detection of loss of H3PO4 from triply charged peptides (98/3 ) 32.7), the triggering of a neutral lossdirected MS3 scan could also be initiated by a loss of methanesulfenic acid () 64/z) from a doubly charged peptide (64/2 ) 32) containing oxidized methionine.17 The neutral loss of methanesulfenic acid from oxidized methionine18-20 is highly efficient, and the residue left within the peptide chain is exactly isobaric to that obtained after loss of H3PO4 from a phosphothreonine residue. Consequently, there is a certain probability that the evaluation of the automatically acquired MS3 spectrum may lead to a false-positive phosphopeptide annotation. As described recently,17 a false-positive identification occurred due to the presence of two sequences in the database with a single (M f T) amino acid substitution. However, false-positive phosphopeptide identifications resulting from methionine oxidation can also occur for completely unrelated sequences. As an example, Figure 5 displays an MS2 spectrum with an abundant loss of 63.998/2 ) 31.999. The correct identification as a trypsin-derived tryptic peptide is only achieved by allowing the modification of +16 Da for methionine and, of course, by inclusion of trypsin into the sequence database. Figure 6 shows two different automatic interpretations of the MS3 spectrum of the ion m/z 1113 in Figure 5. Figure 6a shows the correct assignment of the peptide sequence, which is only found after allowing a variable modification of -48 Da for methionine. An abundant series of doubly charged y ions, that Journal of Proteome Research • Vol. 6, No. 7, 2007 2869
Neutral Loss-Based Phosphopeptide Assignment
technical notes
Figure 4. Abundances of N-terminal sequence motifs in tryptic peptides from human proteins (NCBInr database), which favor an N-terminal neutral loss with close-to-98 m/z value. The overall abundance of the sequences shown adds up to about 2.5% of all tryptic peptides (no uncleaved sites considered). (X ) any amino acid.)
Figure 5. Selected LC-ESI-MS2 spectrum from an analysis of cytosolic tryptic peptides of A. thaliana. An abundant neutral loss of 32.6 from the precursor ion at m/z 1146 is observed, which initiates the recording of an MS3 spectrum of the ion at m/z 1113.4 (see Figure 6), since this neutral loss may indicate the loss of H3PO4 from a triply charged phosphopeptide. As indicated, a tryptic peptide of trypsin with oxidized methionine is assigned by Sequest (Xcorr: 5.26).
represent stepwise neutral loss of the four N-terminal residues, supports this identification. However, if only the variable modifications Ser-18 and Thr-18 are allowed, the identification shown in Figure 6b is obtained. At first glance, the fragment ion assignment does look moderately correct. Weak points, which come up with manual control, are that in the center of the spectrum two abundant doubly charged ions remain unassigned and that the prominent cleavage connected with the complementary b15/y4 ions is located between Phe and Met, which is not known as preferential cleavage motif. However, the other prominent cleavage leading to the pair b4/y15 is 2870
Journal of Proteome Research • Vol. 6, No. 7, 2007
inconspicuous, since it is connected with a cleavage at the N-terminal site of Pro. The example shown in Figure 6 reveals how an accidental fit may lead to partially convincing false-positive phosphopeptide identification. On the basis of high-resolution data, the hit shown in Figure 6b could be excluded, since the molecular weights of the two precursor peptides (Figure 6) differ by 1 Da. However, a large portion of current data in the field is still generated using low to medium resolution mass spectrometers. In addition, the size of protein sequence databases steadily increases so that the chances of false-positive identifications will probably increase. Table 2 summarizes several cases of neutral loss mimicry caused by a precursor/product charge state mix-up. The practically most relevant case for this type of error is probably the loss of methanesulfenic acid from [M + 2H]2+ peptide ions carrying one or more oxidized methionine residues. In the evaluation of ion trap data from complex samples such as the proteome of A. thaliana, we observed that even though conservative filtering criteria were applied in the automated evaluation of ion trap MS/MS data (see section Materials and Methods) several false-positive assignments remained, which amounted to 10-30% of the positive hits using MS3 spectra alone. To overcome the most prominent of the described pitfalls, we suggest to use both MS2 and MS3 spectra and to include the variable modification of -48 Da for methionine in database searching. In addition, one might add methionine or thiourea21 that can act as a scavenger at an early state of sample preparation to minimize methionine oxidation. Influence of Abundant Concomitant Neutral Loss Reactions. In case a phosphopeptide exhibits an additional loss simultaneous with loss of H3PO4, the latter is not detectable in one of the 98/z mass windows. For example, this may occur for phosphopeptides with N-terminal Q or E, since upon CID such peptides may efficiently form N-terminal pyroglutamate
technical notes
Lehmann et al.
net loss of H2SO4 is difficult to discriminate from loss of H3PO4, since the mass difference is only 10 mDa. In case a loss of NH3 accompanies the loss of SO3, a cumulative loss of NH3SO3 () 96.983/z) is observed. In view of the difficulties of determining the exact value of a neutral loss in routine analyses as discussed above, the cumulative loss of NH3SO3 may interfere with loss of H3PO4 or loss of a proline residue located at the N-terminus of a peptide. Mimicry of the Phosphorylation-Induced Increase of + 80/z. In case a ‘close-to-98/z’ neutral loss is occurring in combination with a +80/z (or close to 80/z) increase of the molecular weight not caused by phosphorylation, a particularly high chance for false-positive phosphopeptide annotation exists. Currently, we consider the combination of a nonspecific alkylation by iodoacetamide in combination with sodium adduct formation as the most important source for such an error. Alkylation by iodoacetamide at Lys, His, or at the N-terminal amino group results in an unrecognized mass shift of +57 Da. Additional sodium adduction results in an overall mass increase of +79 Da. For alkylation of methionine, a mass increase of +58 Da is observed,16 resulting in an almost perfect mimicry of phosphorylation of +80 Da, when sodium adduction is occurring in addition. Since the signals caused by a combined nonspecific alkylation and sodium adduction are expectecd to be of low abundance, the interpretation of candidate phosphopeptide signals of low abundance requires particular attention. In general the probability for the occurrence of metal ion adduction appears to be increased using static nanoESI compared to LC-ESI, and it increases with increasing pH value of the solvent system. Figure 6. Two interpretations (Sequest) of the same LC-ESIMS3 spectrum triggered by an abundant loss of 32.5 in the corresponding MS2 spectrum at the precursor ion setmass of 1146. The residues T-18 and M-48 are exactly isobaric and correspond to dehydroaminobutyric acid. (a) Correct assignment (Xcorr 4.91); (b) false-positive assignment including a phosphorylation at threonine (Xcorr 2.55). Table 2. Peptide Modifications and Sequence Motifs which in Combination with a Charge State Mix-Up May Lead to False-Positive pS/pT Phosphopeptide Hitsa structure
neutral loss
1+
2+
3+
4+
phosphoS/phosphoT M-ox M -P...(N-term) ...- I/L (C-term) ...- N (C-term) ...- D (C-term) F -P...(N-term) Mox -P...(N-term) ...- E (C-term)
H3PO4 msa M I/L + H2O N + H2O D + H2O F Mox E + H2O
97.9769 63.9983 131.0405 131.0946 132.0535 133.0375 147.0684 147.0354 147.0532
48.9884 31.9991 65.5202 65.5473 66.0267 66.5188 73.5342 73.5177 73.5266
32.6590 21.3328 43.6802 43.6982 44.0178 44.3458 49.0228 49.0118 49.0177
24.4942 15.9996 32.7601 32.7737 33.0134 33.2594 36.7671 36.7588 36.7633
a For comparison, the m/z values for loss of H3PO4 are listed (first line). Possible interferences are given in bold print. Neutral loss m/z values for the charge states +1 to +4 are listed (msa, methanesulfenic acid).
by loss of NH3 or H2O, respectively. As a result, only an abundant neutral loss of 115/z or 116/z from the molecular ion is observed, which is the sum of the two losses of (98 + 17) or (98 + 16).22 A simultaneous loss of H2O may also lead to false-positive recognition of a peptide with a residue of tyrosine-O-sulfate,23 serine-O-sulfate,24 or S-sulfo-cysteine25,26 as putative pS or pT phosphopeptide. This is because these three modified residues are all characterized by an abundant loss of SO3 ()79.957/z). In case a neutral loss of H2O is added, the observed neutral loss is shifted to H2SO4 () 97.967). This
Rules for Identification of Loss of H3PO4. The starting point for reliable identification of a neutral loss of H3PO4 is determination of the monoisotopic molecular weight of the precursor and of its charge state. In case the survey MS data do not allow a clear statement, a complementary b/y ion pair in the MS/MS spectrum may allow this. The observation of two or more complementary b/y ion pairs strongly increases the reliability of such result. The next step is the determination of the neutral loss m/z difference of the presumed H3PO4 neutral loss, which includes both identification of the monoisotopic [M-H3PO4] peak and determination of its charge state. The experimental mass error for neutral loss determination is about 1.4 times (square root of 2) the mass error, since it is calculated as difference of two mass determinations. The determination of the experimental neutral loss error will allow a decision, if some of the above-mentioned ‘close-to-98/z’ loss reactions have to be considered in the interpretation. In case no clear result for the neutral loss from the molecular ion can be obtained, for example, due to low-abundance or overlapping sequence ions, neutral loss peaks originating from sequence ions are extremely helpful. These also provide the information for phosphorylation site pinpointing. Knowledge of the mass error and combined consideration of neutral losses from both molecular and fragment ions will strongly reduce the chances for false-positive identification of phosphopeptides. The pitfalls summarized refer both to Q-TOF and ion trap analyses. Highly accurate MS and MS/MS data such as generated by FT-ICR instruments or orbitraps would exclude many of the problems discussed, including metal adduction. However, currently, only a minor fraction of the data generated in protein phosphorylation analysis is acquired with this type of instrumentation. Thus, for ion trap data, a joint evaluation of both MS2 and MS3 spectra is recommended as recently demonstrated.4,6,27 Journal of Proteome Research • Vol. 6, No. 7, 2007 2871
Neutral Loss-Based Phosphopeptide Assignment
For Q-TOF and ion trap data, additional manual control of search-engine annotated phosphorylation sites is strongly recommended. Negative ion MS/MS,28 phosphatase treatment,29 and synthesis of suitable reference phosphopeptides can finally be used as additional, highly reliable tools, in case a dependable phosphopeptide assignment cannot be obtained by positive ion MS/MS data evaluation alone. With the use of the emerging fragmentation techniques electron capture dissociation (ECD)30 or electron-transfer dissociation (ETD),31,32 pS and pT residues in phosphopeptides exhibit a high stability, since radical-induced peptide backbone fragmentations prevail. As a consequence, ECD- and ETD-induced MS/MS spectra of phosphopetides are less complex than corresponding collisioncell MS/MS spectra at the expense that the highly diagnostic loss of H3PO4 is not observed. Future experience with ECD and ETD will show the value of these novel fragmentation techniques for the practice of protein phosphorylation analysis.
Summary and Perspective C-terminal and N-terminal peptide sequence motifs and covalent modifications, sometimes in combination with metal ion adduction, can mimic peptide phosphorylation, when analyzed using the neutral loss of H3PO4 as marker fragmentation. Taking into account the hints and rules discussed above, the major sources for false-positive assignment of phosphopeptides by tandem mass spectrometry based on neutral loss of H3PO4 may be recognized. The correction of published falsepositive phosphorylation site identifications is very difficult, since it is highly elaborate to verify the absence of a phosphorylation site by mass spectrometric techniques. Therefore, critical evaluation of MS/MS spectra in protein phosphorylation analysis is strongly recommended to prevent the accumulation of false-positive annotations of protein phosphorylation sites in the scientific literature and databases.
Acknowledgment. We thank R. Pikorn, DKFZ Heidelberg, for peptide synthesis and D. Borsotti, IFOM, Milano, for expert computational support. References (1) Rubin, C. S.; Rosen, O. M. Protein phosphorylation. Annu. Rev. Biochem. 1975, 44, 831-887. (2) Ma, H. Protein phosphorylation in plants: enzymes, substrates and regulators. Trends Genet. 1993, 9, 228-230. (3) DeGnore, J. P.; Qin, J. Fragmentation of phosphopeptides in an ion trap mass spectrometer. J. Am. Soc. Mass Spectrom. 1998, 9, 1175-1188. (4) Giorgianni, F.; Beranova-Giorgianni, S.; Desiderio, D. M. Identification and characterization of phosphorylated proteins in the human pituitary. Proteomics 2004, 4, 587-598. (5) Beausoleil, S. A.; Jedrychowski, M.; Schwartz, D.; Elias, J. E.; Villen, J.; Li, J.; Cohn, M. A.; Cantley, L. C.; Gygi, S. P. Large-scale characterization of HeLa cell nuclear phosphoproteins. Proc. Natl. Acad. Sci. U.S.A. 2004, 10, 12130-12135. (6) Wolschin, F.; Weckwerth, W. Combining metal oxide affinity chromatography (MOAC) and selective mass spectrometry for robust identification of in vivo protein phosphorylation sites. Plant Methods 2005, 1, 9. (7) Gruhler, A.; Olsen, J. V.; Mohammed, S.; Mortensen, P.; Faergeman, N. J.; Mann, M.; Jensen, O. N. Quantitative phosphoproteomics applied to the yeast pheromone signaling pathway. Mol. Cell. Proteomics 2005, 4, 310-327. (8) Yang, F.; Stenoien, D. L.; Strittmatter, E. F.; Wang, J.; Ding, L.; Lipton, M. S.; Monroe, M. E.; Nicora, C. D.; Gristenko, M. A.; Tang, K.; Fang, R.; Adkins, J. N.; Camp, II, D. G.; Chen, D. J.; Smith, R. D. Phosphoproteome profiling of human skin fibroblast cells in response to low- and high-dose irradiation. J. Proteome Res. 2006, 5, 1252 -1260.
2872
Journal of Proteome Research • Vol. 6, No. 7, 2007
technical notes (9) Nousiainen, M.; Sillje´, H. W. H.; Sauer, G.; Nigg, E. A.; Ko¨rner, R. Phosphoproteome analysis of the human mitotic spindle. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 5391-5396. (10) Kinter, M. and Sherman, N. E. Protein Sequencing and Identification Using Tandem Mass Spectrometry; Wiley-Interscience: New York, 2000. (11) Fantes, K .H; Furminger, I. G. Polyacrylamide gel electrophoresis of highly purified chick interferon. Nature 1967, 216, 71-72. (12) Patterson, S. D.; Aebersold, R. Mass spectrometric approaches for the identification of gel-separated proteins. Electrophoresis 1995, 16, 1791-1814. (13) Chowdhury, S. K.; Katta. V.; Beavis, R. C.; Chait, B. T. Origin and removal of adducts (molecular mass ) 98 u) attached to peptide and protein ions in electrospray ionization mass spectra. J. Am. Soc. Mass Spectrom. 1990, 1, 382-388. (14) Salek, M.; Lehmann, W. D. Neutral loss of amino acid residues from protonated peptides in collision-induced dissociation generates N- or C-terminal sequence ladders. J. Mass Spectrom. 2003, 38, 1143-1149. (15) Martin, D. B.; Eng, J. K.; Nesvizhskii, A. I.; Gemmill, A.; Aebersold, R. Investigation of neutral loss during collision-induced dissociation of peptide ions. Anal. Chem. 2005, 77, 4870-4882. (16) Kruger, R.; Hung, C. W.; Edelson-Averbukh, M.; Lehmann, W. D. Iodoacetamide-alkylated methionine can mimic neutral loss of phosphoric acid from phosphopeptides as exemplified by nanoelectrospray ionization quadrupole time-of-flight parent ion scanning. Rapid Commun. Mass Spectrom. 2005, 19, 17091716. (17) Wolschin, F.; Weckwerth, W. Methionine oxidation in peptides a source for false positive phosphopeptide identification in neutral loss driven MS3. Rapid Commun. Mass Spectrom. 2006, 20, 2516-2518. (18) Lagerwerf, F. M.; van de Weert, M.; Heerma, W.; Haverkamp, J. Identification of oxidized methionine in peptides. Rapid Commun. Mass Spectrom. 1996, 10, 1905-1910. (19) Guan, Z.; Yates, N. A.; Bakhtiar, R. Detection and characterization of methionine oxidation in peptides by collision-induced dissociation and electron capture dissociation. J. Am. Soc. Mass Spectrom. 2003, 14, 605-613. (20) Reid, G. E.; Roberts, K. D.; Kapp, E. A, Simpson, R. I. Statistical and mechanistic approaches to understanding the gas-phase fragmentation behavior of methionine sulfoxide containing peptides. J. Proteome Res. 2004, 3, 751-759. (21) Korlimbinis, A.; Hains, P. G.; Truscott, R. J.; Aquilina, J. A. 3-Hydroxykynurenine oxidizes alpha-crystallin: potential role in cataractogenesis. Biochemistry 2006, 45, 1852-1860. (22) Ko¨cher, T.; Savitski, M. M.; Nielsen, M. L.; Zubarev, R. A. PhosTShunter: A fast and reliable tool to detect phosphorylated peptides in liquid chromatography Fourier transform tandem mass spectrometry data sets. J. Proteome Res. 2006, 5, 659668. (23) Severs, J. C.; Carnine, M.; Equizabal, H.; Mock, K. K. Characterization of tyrosine sulfate residues in antihemophilic recombinant factor VIII by liquid chromatography electrospray ionization tandem mass spectrometry and amino acid analysis. Rapid Commun. Mass Spectrom. 1999, 13, 1016-1023. (24) Medzihradszky, K. F.; Darula, Z.; Perlson, E.; Fainzilber, M.; Chalkley, R. J.; Ball, H.; Greenbaum, D.; Bogyo, M.; Tyson, D. R.; Bradshaw, R. A.; Burlingame, A. L. O-sulfonation of serine and threonine: mass spectrometric detection and characterization of a new posttranslational modification in diverse proteins throughout the eukaryotes. Mol. Cell. Proteomics 2004, 3, 429440. (25) Nakanishi, T.; Sato, T.; Sakoda, S.; Yoshioka, M.; Shimizu, A. Modification of cysteine residue in transthyretin and a synthetic peptide: analyses by electrospray ionization mass spectrometry. Biochim. Biophys. Acta 2004, 1698, 45-53. (26) Raftery, M. J. Selective detection of thiosulfate-containing peptides using tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2005, 19, 674-682. (27) Jin, W. H.; Dai, J.; Zhou, H.; Xia, Q. C.; Zou, H. F.; Zeng, R. Phosphoproteome analysis of mouse liver using immobilized metal affinity purification and linear ion trap mass spectrometry. Rapid Commun. Mass Spectrom. 2004, 18, 2169-2176. (28) Edelson-Averbukh, M.; Pipkorn, R.; Lehmann, W. D. Phosphate group-driven fragmentation of multiply charged phosphopeptide anions - improved recognition of peptides phosphorylated at serine, threonine or tyrosine by negative ion electrospray tandem mass spectrometry. Anal. Chem. 2006, 78, 1249-1256.
technical notes (29) Torres, M. P.; Thapar, R.; Marzluff, W. F.; Borchers, C. H. Phosphatase-directed phosphorylation-site determination: A synthesis of methods for the detection and identification of phosphopeptides. J. Proteome Res. 2005, 4, 1628-1635. (30) Cooper, H. J.; Hakansson, K.; Marshall, A. G. The role of electron capture dissociation in biomolecular analysis. Mass Spectrom. Rev. 2005, 24, 201-222. (31) Chia, A.; Huttenhower, C.; Geer, L. Y.; Coon, J. J.; Syka, J. E. P.; Bai, D. L.; Shabanowitz, J.; Burke, D. J.; Troyanskaya, O. G.; Hunt, D. F. Analysis of phosphorylation sites on proteins from
Lehmann et al. Saccharomyces cerevisiae by electron transfer dissociation (ETD) mass spectrometry. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 21932198. (32) Molina, H.; Horn, D. M.; Tang, N.; Mathivanan, S.; Pandey, A. Global proteomic profiling of phosphopeptides using electron transfer dissociation tandem mass spectrometry. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2199-2204.
PR060573W
Journal of Proteome Research • Vol. 6, No. 7, 2007 2873
