Identification of Phosphopeptides by MALDI Q-TOF MS in Positive and Negative Ion Modes after Methyl...

Identification of Phosphopeptides by MALDIQ-TOF MS in Positive and Negative IonModes after Methyl Esterification*□S

Chong-Feng Xu‡, Yun Lu‡, Jinghong Ma§, Moosa Mohammadi§,and Thomas A. Neubert‡§¶

We have developed an efficient, sensitive, and specificmethod for the detection of phosphopeptides present inpeptide mixtures by MALDI Q-TOF mass spectrometry.Use of the MALDI Q-TOF enables selection of phos-phopeptides and characterization by CID of the phos-phopeptides performed on the same sample spot. How-ever, this type of experiment has been limited by lowionization efficiency of phosphopeptides in positive ionmode while selecting precursor ions of phosphopeptides.Our method entails neutralizing negative charges onacidic groups of nonphosphorylated peptides by methylesterification before mass spectrometry in positive andnegative ion modes. Methyl esterification significantly in-creases the relative signal intensity generated by phos-phopeptides in negative ion mode compared with positiveion mode and greatly increases selectivity for phos-phopeptides by suppressing the signal intensity gener-ated by acidic peptides in negative ion mode. We used themethod to identify 12 phosphopeptides containing 22phosphorylation sites from low femtomolar amounts of atryptic digest of �-casein and �-s-casein. We also identi-fied 10 phosphopeptides containing five phosphorylationsites from an in-gel tryptic digest of 100 fmol of an in vitroautophosphorylated fibroblast growth factor receptor ki-nase domain and an additional phosphopeptide contain-ing another phosphorylation site when 500 fmol of thedigest was examined. The results demonstrate that themethod is a fast, robust, and sensitive means of charac-terizing phosphopeptides present in low abundance mix-tures of phosphorylated and nonphosphorylated pep-tides. Molecular & Cellular Proteomics 4:809–818,2005.

Reversible phosphorylation of serine, threonine, and tyro-sine residues is one of the most common and importantregulatory modifications of proteins and often is a key event incellular signal transduction (1, 2). In recent years, mass spec-trometry has become a key technology for characterization of

protein phosphorylation and phosphoproteome analysis. Twocomplementary ionization techniques, MALDI and ESI, incombination with a variety of mass analyzers, have been usedto identify phosphopeptides and determine the phosphoryl-ated amino acids on the peptides (3–16). In most cases,characterization of phosphopeptides by MS requires selec-tion of phosphorylated peptides from complex peptide mix-tures resulting from proteolysis of phosphorylated proteinsfollowed by MS/MS to confirm the phosphorylation and toidentify the phosphorylated amino acid residues on phos-phopeptides containing more than a single serine, threonine,or tyrosine residue. Despite many recent advances in meth-odology, identification of phosphorylation sites on proteinsremains a difficult challenge (17).

High sensitivity, resolution, and mass accuracy make Q-TOF MS a powerful tool for the characterization of phos-phopeptides. The MALDI Q-TOF allows for increased effi-ciency and sample throughput because identification ofphosphopeptides by MS and characterization by CID MS/MScan be performed on a single sample spot (18). However, alimitation of this type of experiment is the low ionizationefficiency of phosphopeptides in positive ion mode, resultingin low sensitivity of phosphopeptide detection and consump-tion of a large portion of the sample during the search forphosphorylated precursor peptides before MS/MS can beperformed (18). Phosphorylated precursor ions can be de-tected by comparing MALDI spectra of a single sample takenin positive and negative ion modes, with phosphopeptidesdemonstrating greater relative ion intensities in negative ionmode (19, 20). However, this approach suffers from poorspecificity for phosphopeptides because of the high back-ground of nonphosphorylated acidic peptides in negative ionmode caused by the ability of carboxylate groups on glutamateor aspartate residues to develop negative charges in a mannersimilar to that of phosphate groups. In this article, we show thatremoval of these acidic groups by methyl esterification (21, 22)can greatly diminish the ion intensity of these acidic nonphos-phorylated peptides in negative ion mode and therefore greatlyincrease the selectivity of the method for phosphopeptides inpeptide mixtures. We used the method to identify 12 phos-phopeptides containing 22 phosphorylation sites from lowfemtomolar amounts of a tryptic digest of a model phospho-

From the ‡Skirball Institute of Biomolecular Medicine, §Departmentof Pharmacology, New York University School of Medicine, NewYork, New York 10016

November 24, 2004, and in revised form, February 4, 2005Published, MCP Papers in Press, March 7, 2005, DOI

10.1074/mcp.T400019-MCP200

Technology

© 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Molecular & Cellular Proteomics 4.6 809This paper is available on line at http://www.mcponline.org

protein, �-casein, and its minor contaminant �-s-casein.The fibroblast growth factor receptors (FGFRs)1 are a family

of tyrosine kinase receptors that play critical roles in humanskeletal development. Gain of function mutations in the tyro-sine kinase domain of FGFRs are responsible for a number ofhuman skeletal disorders. The degree of clinical severity as-sociated with the mutations correlates with the level of con-stitutive kinase activity in these mutants (23–25). A method forrapidly comparing phosphorylation sites on various mutantsof FGFRs and then correlating phosphorylation status withreceptor activity would be very useful for understanding themolecular basis for receptor gain of function and potentiallyfacilitate development of therapeutic interventions. We usedour method to characterize phosphopeptides on the in vitrophosphorylated kinase domain of the N549H mutant of theFGFR2, which is responsible for the severe craniosynostosisdisorder known as Crouzon syndrome (26). After in vitro phos-phorylation by incubation with ATP, isolation by SDS-PAGE,and in-gel tryptic digestion, we identified 10 phosphopeptidescontaining five phosphorylation sites from 100 fmol of themutant kinase domain and an additional phosphopeptide con-taining another phosphorylation site when 500 fmol of the digestwas examined. These identified tyrosine phosphorylation sitescorrespond to those previously found to be phosphorylated onthe kinase domain of FGFR1 (27), demonstrating that ourmethod can be used to rapidly characterize phosphorylationsites on low levels of receptors or other proteins such as canbe obtained in biological experiments.

EXPERIMENTAL PROCEDURES

Materials—2,6-Dihydroxyacetophenone (DHAP), �-cyano-4-hy-droxycinnamic acid (CHCA), dihydroxybenzoic acid (DHB), diammo-nium hydrogen citrate, and acetyl chloride were purchased fromSigma-Aldrich (St. Louis, MO). Sequencing grade modified trypsinwas purchased from Promega Co. (Madison, WI). Ammonium bicar-bonate, TFA, [Glu1]-Fibrinopeptide B, BSA, �-casein (from bovinemilk, purity �90% by electrophoresis,) and the monophosphopeptideand tetraphosphopeptide from �-casein were from Sigma ChemicalCo. HPLC grade water, acetonitrile, and methanol were purchasedfrom Fisher Scientific (Hanover Park, IL). Coomassie Blue R-250 andprecast 12% SDS-PAGE gels were purchased from Bio-Rad.

Preparation of in Vitro Phosphorylated FGFR Kinase Domain—Affinity purified kinase domain of FGFR2 harboring the N549H muta-tion was concentrated using a Centricon filtration device to a con-centration of 10 mg/ml as measured by A280. To generate thephosphorylated form, the kinase domain was incubated with 5 mM

ATP and 10 mM MgCl2. The course of phosphorylation was monitoredby native gel analysis until the phosphorylation was complete (�10min). The phosphorylation reaction was stopped by adding 20 mM

EDTA. For preparation of in-gel digests for mass spectrometry anal-ysis, the sample buffer was prepared by mixing Bio-Rad Laemmlisample buffer with mercaptoethanol at a ratio of 20:1 (v/v). Twopicomoles of the phosphorylated kinase domain was diluted in thissample buffer at a ratio of 1:1 (v/v), denatured at 95 °C for 5 min. The

sample was separated on a 12% SDS-PAGE gel at 100 V. Proteinswere visualized by staining with Coomassie Blue R-250.

In-solution Digestion and in-Gel Digestion of Proteins—For in-so-lution digestion, proteins were dissolved in 25 mM ammonium bicar-bonate and denatured by heating in 95 °C water for 5 min and thendigested in 25 mM ammonium bicarbonate by sequence-grade tryp-sin at a ratio of 1:50 (enzyme/protein) for 5 h at 37 °C. The in-geldigestion of the phosphorylated FGFR2 kinase domain was carriedout according to the protocol of Shevchenko et al. (28), except thatthe reduction and alkylation step was omitted.

Formation of Methyl Esters—The methanolic HCl solution wasprepared by the dropwise addition of 160 �l of acetyl chloride to 1 mlof dry methanol (22). The protein tryptic digests up to 250 pmol inamount were lyophilized and redissolved in 50 �l of 2 M methanolicHCl regent. When more than 250 pmol of peptides were methylated,the volume was 200 �l of methanolic HCl. Methyl esterification wasallowed to proceed for 2–3 h at room temperature. Solvent wasremoved by lyophilization, and the resulting samples were redis-solved in 30% acetonitrile in 0.2% TFA.

Preparation of Matrix for MALDI Q-TOF Mass Spectrometry—A 50mM solution of the DHAP matrix was prepared by dissolving 15.2 mgof 2,6-dihydroxyacetophenone in 1 ml water/methanol (10:90, v/v)followed by the addition of 100 mM diammonium hydrogen citrate inwater at a ratio of 1:1 (v/v).2 CHCA matrix was prepared by dissolving2 mg of �-cyano-4-hydroxycinnamic acid in 1 ml of water/acetonitrile(50:50, v/v) containing 0.1% TFA.

MALDI Q-TOF Mass Spectrometry—Sample and matrix weremixed at a ratio of 1:1 (v/v), and 1.0 �l of this mixture was spottedonto the MALDI sample stage. Positive and negative ion MALDIQ-TOF mass spectra were acquired with a Micromass Q-TOF UltimaMALDI mass spectrometer (Waters, MA). The instrument was oper-ated in V mode with a mass resolution of �10,000, which enabled thediscrimination of carboxyl methylation (�14 atomic mass units) andasparagine/glutamine methylation (�15 atomic mass units). Laserpulses were generated by a nitrogen laser (337 nm) with laser energyof 350 �J per pulse. Mass spectra were acquired and processedby Masslynx 4.0 software (Micromass Ltd., Manchester, UnitedKingdom). A total of 200–800 laser shots were averaged per massspectrum, the background was subtracted, and the spectrum wassmoothed using a mass window appropriate for the significant peakwidths. Known peptide masses were used as internal mass standards.

RESULTS

Identification of Phosphorylation Sites on �-Casein—�-Ca-sein is a model phosphoprotein used to develop and testmethods to characterize phosphopeptides. Fig. 1 shows pos-itive and negative mode MALDI Q-TOF mass spectra of 140fmol of a tryptic digest of �-casein before and after methyla-tion. As shown in Fig. 1, a and b, and previously by us andothers (19, 20), the phosphopeptides (indicated by asterisks)have higher relative abundance in the negative ion MALDIspectrum compared with the positive ion spectrum. However,the relative difference between the positive and negative ionintensities of the phosphopeptides is subtle and difficult toobserve in the case of the monophosphorylated peptides ofm/z 1949.94 from (�-casein) and 2059.81 (FQpSEEQQQT-EDELQDK, calculated monoisotopic MW � 2060.82) from�-casein (Fig 1b). In addition to the tetraphosphopeptide

1 The abbreviations used are: FGFR, fibroblast growth factor re-ceptor; DHAP, 2,6-dihydroxyacetophenone; DHB, dihydroxybenzoicacid; CHCA, �-cyano-4-hydroxycinnamic acid. 2 S. R. Weinberger, personal communication.

Phosphopeptide Methylation and Detection by MALDI Q-TOF MS

810 Molecular & Cellular Proteomics 4.6

FIG. 1. MALDI Q-TOF MS of methyl-ated and nonmethylated �-caseintryptic digest in positive and negativeion modes. a, spectrum of 140 fmol of�-casein digest acquired in positive ionmode; b, spectrum of 140 fmol of �-ca-sein digest acquired in negative ionmode; c, spectrum of 140 fmol of meth-ylated �-casein digest acquired in posi-tive ion mode; d, spectrum of 140 fmol ofmethylated �-casein digest acquired innegative ion mode; e, spectrum of 7 fmolof methylated �-casein digest acquiredin positive ion mode; f, spectrum of 7fmol of methylated �-casein digest ac-quired in negative ion mode. Predomi-nant phosphopeptides species are la-beled (*), undermethylated (one or morecarboxylate groups not methylated) arelabeled (�), and methyl ester side prod-ucts of deamidated Gln or Asn residuesare labeled (�). Each spectrum is thesum of 740 laser shots. Base peak (max-imum) ion counts are shown in the upperright corner of each spectrum. The 12phosphopeptides identified in d, con-taining five phosphorylation sites from�-casein and 17 phosphorylation sitesfrom the minor contaminants �-s1 and�-s2-casein, are listed in Table I.


Molecular & Cellular Proteomics 4.6 811

(RELEELNVPGEIVEpSLpSpSpSEESITR, calculated MW �

3121.26) at m/z 3120.25 and several resulting neutral lossfragments resulting from the phosphopeptides, numerousnonphosphorylated peptides also can be observed in thenegative ion mode spectrum. Fig. 1, c and d, show positiveand negative ion spectra of 140 fmol of the same tryptic digestof �-casein, but after methylation. In this case, the phospho-rylated peptides (designated by asterisks and listed in Table I)produce the predominant peaks in the negative ion spectrumand demonstrate a much greater relative increase in negativeion mode compared with positive ion mode. Indeed, 17 phos-phorylated peptides from the minor contaminants �-s1-ca-sein and �-s2-casein (present at less than 14 fmol based onthe manufacturer’s analysis that the �-casein is more than90% pure) can be observed in Fig 1d. These peptides containall 5 previously described phosphorylation sites from �-ca-sein, all 8 phosphorylation sites from the contaminant �-s1-casein, and 8 of the 10 previously described phosphorylationsites on �-s2-casein. We also identified one phosphorylationsite on �-s2-casein (Ser-143) not reported in the ExPaSydatabase. The sensitivity of the method is further illustrated inFig. 1, e and f, in which both of the phosphopeptides, con-taining all five phosphorylation sites from the �-casein digest,can be seen at the 7 fmol level in negative ion mode. Althoughthe positive ion mode signal intensity of many nonphospho-rylated peptides increased after methylation (e.g. peptideLLYQEPVLGPVR of m/z 1383.79 in Fig. 1a compared with m/z1411.75 in Fig. 1c), the intensity of the major methylatedpeptide species decreased in some cases because of partialmethylation of glutamine and asparagine residues (see be-low), which leads to additional peak complexity for thesepeptides (e.g. peptide SPAQILQWQVLSNTVPAK of m/z1980.10 in Fig. 1a compared with m/z 1994.00, 2008.98,2023.95, etc. in Fig. 1c).

Identification of Phosphorylation Sites in a Mixture of BSAand �-Casein Tryptic Digests—To demonstrate that themethod can be used to study more complex peptide mixtures,a similar analysis was carried out on a tryptic digest of amixture of 140 fmol of BSA and 140 fmol of �-casein (Fig. 2).The presence of the BSA peptides increased the complexityof the spectra of the unmethylated peptides in both positiveand negative modes (Fig. 2, a and b) and the methylatedpeptides in positive ion mode (Fig 2c). However, only phos-phopeptides produced peaks of significant ion intensity in thenegative ion spectra of the methylated peptide digest (Fig.2d). In Fig. 2f, three phosphopeptides (including one missedcleavage peptide) containing all five phosphorylation sitesfrom �-casein and four of the �-casein phosphopeptides canbe observed in negative ion mode spectra of 14 fmol of themethylated peptide mixture (containing less than 2 fmol of the�-casein contaminants). These results indicate that ourmethod is capable of identifying small amounts of phos-phopeptides directly from simple protein mixtures without theneed for further purification or enrichment.

In each negative ion spectrum of the methylated peptides,the major phosphopeptide peaks, designated by asterisksin the figures, correspond to the peptides with completemethylation of carboxylate groups. For example, in Figs. 1, dand f, and 2, d and f, the peak at m/z 2157 corresponds toFQpSmEmEQQQTmEmDmELQmDmK, where m indicates themethyl esterified acidic or C-terminal amino acid residue,whereas 3232 corresponds to the tetraphosphorylated pep-tide with complete methylation of carboxylate groups. Minorpeaks, designated by (�), correspond to phosphopeptideswith incomplete methyl esterification of acidic groups. Somepeptides, indicated by (�) in Figs. 1–3, undergo deamidationfollowed by methyl esterification on side chains of asparagineand/or glutamine, resulting in a mass increase of 15 Da be-

TABLE IPhosphopeptides detected in a methylated tryptic digest of 140 fmol of �-casein

MW (without methylation) is the calculated molecular weight of the phosphopeptides before methylation; MW-H (after methylation) is theobserved m/z of the singly charged methylated phosphopeptides measured in the negative ion spectra shown in Figs. 1 and 2.

Protein andseq. no.

SequenceMW (withoutmethylation)

MW-H (aftermethylation)

Phosphorylationsites

�-Casein33–48 FQpSEEQQQTEDELQDK 2060.82 2157.82 Ser-3530–48 IEKFQpSEEQQQTEDELQDK 2431.04 2542.07 Ser-352–25 ELEELNVPGEIVEpSLpSpSpSEESITR 2965.16 3076.24 Ser-15, Ser-17, Ser-18, Ser-191–25 RELEELNVPGEIVEpSLpSpSpSEESITR 3121.26 3232.35 Ser-15, Ser-17, Ser-18, Ser-19

�-s1-Casein106–119 VPQLEIVPNpSAEER 1659.79 1714.84 Ser-11543–58 DIGpSEpSTEDQAMEDIK 1926.68 2023.79 Ser-46, Ser-48104–119 YKVPQLEIVPNpSAEER 1950.95 2006.00 Ser-11559–79 QMEAEpSIpSpSpSEEIVPNpSVEQK 2719.90 2802.98 Ser-64, Ser-66, Ser-67, Ser-68, Ser-75

�-s2-Casein138–149 TVDMEpSTEVFTK 1465.60 1520.62 Ser-143138–150 TVDMEpSTEVFTKK 1593.69 1648.72 Ser-14346–70 NANEEEYSIGpSpSpSEEpSAEVATEEVK 3007.02 3132.10 Ser-56, Ser-57, Ser-58, Ser-611–24 KNTMEHVpSpSpSEESIIpSQETYKQEK 3131.19 3214.27 Ser-8, Ser-9, Ser-10, Ser-16



FIG. 2. MALDI Q-TOF MS of methyl-ated and nonmethylated mixtures of�-casein and BSA tryptic digests inpositive and negative ion modes. a,spectrum of 140 fmol of nonmethylateddigest mixture acquired in positive ionmode; b, spectrum of 140 fmol of non-methylated digest mixture acquired innegative ion mode; c, spectrum of 140fmol of methylated digest mixture ac-quired in positive ion mode; d, spectrumof 140 fmol of methylated digest mixtureacquired in negative ion mode; e, spec-trum of 14 fmol of methylated digestmixture acquired in positive ion mode;and f, spectrum of 14 fmol of methylateddigest mixture acquired in negative ionmode. c–f, signal intensity of m/z above2500 is increased 10-fold. Predominantphosphopeptides species are labeled (*),undermethylated (one or more carboxylgroups not methylated) are labeled (�),and methyl ester side products of de-amidated Gln or Asn residues are la-beled (�). The labeled phosphopeptidesare listed in Table I.



cause of substitution of methyl-oxide (-OCH3) for the aminegroup (-NH2) (29). Consistent with the results of He et al. (29),we have found that this side reaction is unavoidable evenwhen using dehydrated methanol and extensively drying thepeptides, given that the reaction must proceed for a minimumof 2–3 h to ensure nearly complete methyl esterification ofacidic groups (21). We evaluated the methylation of Glu1-Fibrinopeptide B peptide (sequence EGVNDNEEGFFSAR)and a tryptic digest of �-s-casein tryptic digest in a timecourse experiment. Methyl esterification of acidic groups ofmost peptides was more than 95% complete after 2 h withoutside product (amine) methylation in 60 and 80% of peptidescontaining 2 to 4 and mono Gln/Asn, respectively. However,those peptides with multiple acidic amino acids and Gln/Asn,such as the tetraphosphopeptide in �-casein and the penta-phosphopeptide in �-s1-casein, required 3 h to achieve com-plete methylation of acidic groups on 95% of the peptides.

Identification of FGFR2 Autophosphorylation Sites—Weused the method to identify phosphorylation sites on theN549H mutant FGF receptor kinase domain after incubationwith ATP to induce autophosphorylation. The minimumamount of autophosphorylated kinase domain (2 pmol, 76 ng)required to visualize by Coomassie Blue staining was sub-jected to SDS-PAGE. After electrophoresis, the correspond-ing gel band (37 kDa) was excised for in-gel digestion withtrypsin, and half of the resulting tryptic peptides were methyl

esterified as described under “Experimental Procedures.” Theresults of the analysis of 10% of the methylated peptides (100fmol before electrophoresis and digestion) by MALDI Q-TOFin positive and negative modes are shown in Fig. 3 andTable II.

Fig. 3b shows 10 phosphopeptides containing four tyrosineand one serine phosphorylation sites (Table II). All of the majorpeaks in the spectrum acquired in negative ion mode (Fig. 3b)are caused by FGFR2 phosphopeptides with the exception ofthe peak of m/z 1342.58. Tandem mass spectra in positive ionmode of the phosphopeptides were acquired from the samespot on the MALDI sample plate to confirm the identity of thephosphopeptides, and in most cases to determine the phos-phorylated amino acids. The sequences of these phos-phopeptides are shown in Table II. Fig. 3a (inset) shows theMS/MS spectrum for the precursor ion of m/z 1487.02 ofsequence mDINNImDpYpYKmK. This peptide is derived fromthe activation loop of the kinase domain. Autophosphorylationof both corresponding tyrosine residues in FGFR1 is requiredfor up-regulation of kinase activity (27). The MS/MS spectrumsupports phosphorylation on the two tyrosine residues. Theimmonium ion of phosphotyrosine is observed at m/z 216.04,peaks at m/z 1407.65 and 1327.72 indicate the loss of oneand two HPO3 from the molecular ion (MH� � 1487.62),respectively, and the y4 to y7 fragment ions are clearly seen.

Interpretation of the previous spectrum was straightforward

FIG. 3. MALDI Q-TOF MS of 10% ofthe methylated tryptic peptides fromthe in-gel digestion of 1 pmol of auto-phosphorylated FGFR2 kinase domainoperated in positive (a) and negative(b) modes. The peaks with mass differ-ence of 80 in b indicate different levels ofphosphorylation (e.g. peaks of m/z1717.74, 1797.71, and 1877.71 are themono-, di-, and triphosphorylated formsof the peptide 580–592). a, inset, showsthe MS/MS spectrum of the diphospho-rylated species of peptide 650–659

mDINNImDpYpYKmK, M�H� � 1487.63.c, MS/MS spectrum of the nonmethyl-ated monophosphorylated peptide ofsequence RPPGMEYpSYDINR from theactivation loop of the FGFR2 kinase in-sert region. Labeled phosphopeptidesare listed in Table II.



because both potential sites on the peptide were phospho-rylated. In many cases MALDI Q-TOF MS/MS can also beused to identify phosphorylated amino acids on peptides withseveral potential phosphorylation sites (18). Fig. 3c shows thepositive ion MS/MS spectrum of the monophosphorylatedpeptide RPPGMEYpSYDINR (MH� � 1677.71). Lack of im-monium ion at 216.04, the abundant b7 (831.38) and c7(848.41) ions, as well as a minor b5 ion, and lack of b7 � 80,c7 � 80, and b5 � 80 ions indicated that the tyrosine residueswere not phosphorylated or were phosphorylated at very lowstoichiometry. The loss of H3PO4 from the b8 ion indicatedthat the serine was phosphorylated, which is also consistentwith the loss of H3PO4 from c9, b10, c11, b12, and themolecular ion. We concluded that most or all of the mono-phosphorylated peptides are phosphorylated on serine, whichhas not been previously reported. For comparison, we haveshown the MS/MS spectrum of both methylated and non-methylated forms of this peptide in supplemental Fig. S1. Wefind no systematic advantage or disadvantage from fragment-ing either form of the phosphopeptides but often performMS/MS analyses of both forms to provide more completeinformation. The doubly and triply phosphorylated forms ofthis peptide were also observed (Table II). The peptide isderived from the kinase insert region, and the equivalenttyrosines on the highly similar FGFR1 have been shown to bephosphorylated in cells using 32PO4 labeling and Edman deg-radation (27). An additional N-terminal phosphopeptide wasdetected by analysis of 500 fmol of the kinase domain digest(Table II). This peptide is derived from the juxtamembraneregion preceding the core kinase domain and contains atyrosine residue previously shown to be phosphorylated uponreceptor activation (27). Positive and negative ion mode spec-tra of methylated and nonmethylated tryptic digests of 500fmol of the FGFR2 kinase domain are shown in supplementalFig. S2, and sequence coverage of the FGFR2 with and

without methyl esterification is shown in supplemental Fig. S3.In summary, we demonstrated that the FGFR2 kinase do-

main contains a total of six phosphorylation sites, includingfour identified tyrosine residues and an identified serine resi-due as well as an additional phosphopeptide containing anundetermined phosphorylation site. This result is consistentwith a native gel showing the time course of phosphorylationof the FGFR2 kinase domain, which shows nearly completephosphorylation of five or six sites after 10 min of incubationwith ATP (supplemental Fig. S4). The result is also consistentwith a mass of the intact phosphoprotein as determined byMALDI-TOF MS of 37,458 Da (data not shown). This mass is505 Da higher than the predicted average mass of the non-phosphorylated protein, 36,953 Da, suggesting that six phos-phate groups modified the FGFR kinase domain at high stoi-chiometry. However, the mass of the intact protein alone isnot definitive because of variable and uncertain stoichiometryof phosphorylation as well as the possibility of additionalposttranslational modifications. Although we detected somepotentially multiphosphorylated peptides that contained par-tially phosphorylated sites (Table II and supplemental table),we did not detect any nonphosphorylated versions of thephosphopeptides in any of our experiments.

Comparison of DHAP, DHB, and CHCA Matrices—We usedDHAP as the matrix because it is a “cooler” matrix than CHCA(30) and resulted in less PSD at the relatively high fixed laserenergy (350 �J/pulse) of our MALDI Q-TOF instrument. Fig. 4shows positive ion MALDI Q-TOF mass spectra of 500 fmol oftwo synthetic standard phosphopeptides from �-casein usingCHCA (Fig. 4a) and DHAP (Fig. 4b) as matrix. The loss ofH3PO4 is significantly greater in Fig. 4a than in 4b for both themono- and tetraphosphorylated peptides, which decreasesthe ion intensity generated by the intact phosphopeptides aswell as increases spectrum complexity. The increased sensi-tivity is also demonstrated by the clear presence of an addi-

TABLE IIPhosphorylation sites identified in autophosphorylated kinase domain of the N549H mutant of the FGF receptor 2

MW (without methylation) is the calculated molecular weight of the phosphopeptides before methylation, and MW-H (after methylation) is theobserved m/z of the singly charged methylated phosphopeptides measured in the negative ion spectrum of 100 fmol of a tryptic digest ofFGFR2 kinase domain shown in Fig 3. The peptide his tag � 459–473 (sequence GSSHHHHHHSQDPMLAGVSEYELPEDPK) was onlyobserved by analysis of the tryptic digest of 500 fmol of the FGFR2 kinase domain. For comparison with results published previously, aminoacid numbers refer to the sequence of full-length FGFR2, not the kinase domain construct used in our experiments.

Sequence no. Primary sequenceMW (beforemethylation)

M-H (aftermethylation)

Phosphorylation sites

650–659 DINNIDYYKK 1236.51 1405.60 Tyr-656/Tyr-6571316.48 1485.58 Tyr-656, Tyr-657

650–668 DINNIDYYKKTTNGRLPVK 2411.14 2452.25 Tyr-656, Tyr-657580–592 RPPGMEYSYDINR 1676.74 1717.74 Ser-587/Tyr-588

1756.71 1797.71 Ser-587, Tyr-5881836.68 1877.71 Tyr-586, Ser-587, Tyr-588

578–592 ARRPPGMEYSYDINR 1983.81 2024.88 Ser-587, Tyr-5882081.78 2104.86 Tyr-586, Ser-587, Tyr-588

580–601 RPPGMEYSYDINRVPEEQMTFK 2846.19 2915.46 Ser-587, Tyr-5882926.16 2995.44 Tyr-586, Ser-587, Tyr-588

His tag � 459–473 GSSHHHHHHSQDPMLAGVSEYELPEDPK 3237.37 3320.46 Not determined



tional peak at m/z 2967.23 in Fig. 4b but not Fig. 4a, whichcorresponds to a contaminant in the peptide mixture becauseof a missing N-terminal arginine residue. A comparison ofpositive and negative ion mode spectra of methyl esterified�-casein digest using CHCA and DHAP matrices is shown insupplemental Fig. S5, also demonstrating less � elimination ofphosphate in the DHAP spectra. Whereas we found that useof DHB as matrix improved the ion intensity for nonmethylatedphosphopeptides in both positive and negative ion modes,and inclusion of phosphoric acid improved sensitivity in neg-ative ion mode (31), lack of improvement of the signal formethylated phosphopeptides makes DHB suboptimal for ourmethod (data not shown).

DISCUSSION

Previous studies have shown that comparing MALDI-TOFspectra of mixtures of phosphorylated and nonphosphoryl-ated peptides acquired in positive and negative modes canbe useful for identifying the phosphopeptides based on therelatively higher ion intensities generated by the phosphoryl-ated peptides in negative ion mode. However, a limitation ofthis method is its poor specificity based on the tendency ofacidic peptides to exhibit behavior similar to that of phos-phopeptides. We have greatly improved the selectivity of themethod by methylation of carboxylate groups on the peptideswith methanolic HCl (22) to suppress the contribution of thesegroups to formation of negative charges on the peptides in

negative ion mode. We then analyze the phosphopeptides onthe same sample spot by CID in the MALDI Q-TOF instrumentto confirm phosphopeptide identification and locate the phos-phorylated amino acid(s).

We have demonstrated the utility of the method by identi-fying all of the known phosphorylation sites on low femtomo-lar amounts of a simple protein mixture, including �-caseinand its contaminant �-s-casein. We have also shown that themethod can be used to rapidly monitor all previously identifiedphosphorylation sites on subpicomolar amounts of the med-ically important N549H mutant FGFR2 tyrosine kinase do-main. This study is the first characterization of the autophos-phorylation sites on the FGFR2. Based on a previous study ofthe highly homologous (more than 90% identical) FGFR1 ki-nase domain, we expected to find six tyrosine phosphoryla-tion sites on the kinase domain of FGFR2. Indeed, four of ouridentified tyrosine phosphorylation sites correspond to thoseon FGFR1, and we found a fifth site on a tryptic peptide thatcontains tyrosine 466, which corresponds to phosphorylatedtyrosine 463 on FGFR1. We did not find phosphorylation ontyrosine 733 of FGFR2, which corresponds to tyrosine 730 onFGFR1, which was shown by Mohammadi et al. (27) to bephosphorylated. However, the crystal structure of FGFR1 ki-nase domain shows that tyrosine 730 is poorly solvent-ex-posed and therefore would be expected to be a poor sub-strate for phosphorylation (32). We found an additional

FIG. 4. Positive ion mode MALDI Q-TOF MS of 500 fmol of two �-casein-derived standard phosphopeptides(mono- and tetraphosphorylated) us-ing 2 mg/ml CHCA as matrix (a) and7.6 mg/ml DHAP as matrix (b). Pep-tides are listed in Table I, and the peak atm/z 2967.23 corresponds to the tetra-phosphorylated peptide missing a singleArg residue.



phosphorylation on serine 587, which we believe was phos-phorylated by a heterologous serine kinase during proteinexpression (data not shown).

Although each of the phosphorylation sites on these twoproteins is phosphorylated to very high stoichiometry, we alsobelieve the method is capable of detecting phosphorylationevents at low stoichiometry. We detected 16 of 18 previouslyreported phosphorylation sites as well as an additional sitenot previously reported on �-casein that was present at lessthan 10% of the �-casein in the same sample. After methyl-ation, the ratio of intensity (ion counts) of phosphorylatedpeptides in negative ion mode compared with positive ionmode was more than 25 greater than that of nonphosphoryl-ated peptides (see supplemental table), further suggestingthe method should efficiently detect phosphopeptides of lowstoichiometry. However, it is possible for the method to fail todetect phosphorylated peptides when enzymatic digestion ofa phosphoprotein fails to yield a sufficient quantity of phos-phopeptides for MALDI-TOF MS analysis, when the resultingphosphopeptides are too small or too large to be efficientlydetected by MALDI-TOF MS, or when a highly abundantnonphosphorylated peptide has nearly the exact same mo-lecular weight as the phosphorylated peptide.

The identification of multiphosphorylated peptides by massspectrometry has been especially challenging because of lowionization efficiency in positive ion mode and a high propen-sity for nonspecific adsorption to metallic and hydrophilicsurfaces (33). For example, without methyl esterification, thetetraphosphopeptide of �-casein (containing eight acidicamino acids and four phosphates) and the pentaphosphopep-tide from �-s1-casein (amino acids 59–79, containing sixacidic amino acids and five phosphates) were difficult todetect by MALDI-TOF or nanoflow LC-MS (Q-TOF) in positiveion mode at the 50 pmol level, or several pmol in negative ionmode for the �-s1-casein pentaphosphorylated peptide (datanot shown and Kim et al. (33)). However, after methyl esteri-fication, the limit of detection by MALDI Q-TOF in negative ionmode for these peptides decreased dramatically to the lowfemtomolar level. The most likely explanation is that methylesterification achieves this effect by neutralizing the carbox-ylate groups on the aspartate and glutamate side chains aswell as the carboxyl terminus of each peptide. The phosphategroups of phosphopeptides are then the only remaining acidicgroups, and only phosphopeptides ionize efficiently in nega-tive ion mode. In this case, the ionization of phosphopeptidesis less likely to be suppressed by non-phosphopeptides, sothat the negative ion mode spectrum of a mixture of methyl-esterified phosphorylated and nonphosphorylated peptides issimilar in ion detection efficiency to that of purified phos-phopeptides. By comparing negative and positive ion spectraof methylated peptides, low abundance phosphorylated pep-tides can be discriminated from highly abundant nonphos-phorylated peptides that produce detectable signals in neg-ative ion mode because the nonphosphorylated peptides

produce much stronger relative signals in positive ion mode.The results demonstrate that the method is a fast, robust,

and sensitive means of characterizing phosphopeptides pres-ent in low abundance mixtures of phosphorylated and non-phosphorylated peptides. An advantage of the method is thatit can be used to characterize peptides phosphorylated onserine, threonine, or tyrosine residues. Another advantage isthat the method also can be used with other MALDI instru-mentation that can be operated in positive and negative ionmodes. Because the method relies on methylating carboxylgroups on peptide mixtures before mass spectrometry, itcould be used for relative quantification experiments usingstable isotopic labeling of the methyl groups (29), thoughpartial methylation of glutamine and asparagine residueswould complicate analyses of such experiments. In addition,our method is an ideal complement to enrichment of phos-phopeptides from complex mixtures using immobilized metalaffinity chromatography IMAC (22). Such pre-enrichmentwould enable the method to be used in large scale analyses ofthe phosphoproteome.

Acknowledgments—We thank Dr. Vivek Shetty and Dr. IgnatiusKass (Waters Corp.) for helpful advice and Dr. Scot R. Weinberger(Ciphergen, Inc.) for helpful advice about the DHAP/diammoniumhydrogen citrate MALDI matrix.

* This work was supported by National Institutes of Health SharedInstrumentation Grants 1-S10-RR14662 and 1-S10-RR017990 (toT. A. N.) and National Institutes of Health grants 1-R21-NS44184 (toT. A. N.) and R01-DE013686 (to M. M.). The costs of publication ofthis article were defrayed in part by the payment of page charges.This article must therefore be hereby marked “advertisement” inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

□S The on-line version of this article (available at www.mcponline.org) contains supplemental material.

¶ To whom correspondence should be addressed: Skirball Insti-tute, Lab 5-18, New York University School of Medicine, 540 FirstAve., New York, NY 10016. Tel.: 212-263-7265; Fax: 212-263-8214;E-mail: [email protected].

REFERENCES

1. Posada, J., and Cooper, J. A. (1992) Molecular signal integration. Interplaybetween serine, threonine, and tyrosine phosphorylation. Mol. Biol. Cell3, 583–592

2. Hunter, T. (1995) Protein kinases and phosphatases: the yin and yang ofprotein phosphorylation and signaling. Cell 80, 225–236

3. Neubauer, G., and Mann, M. (1999) Mapping of phosphorylation sites ofgel-isolated proteins by nanoelectrospray tandem mass spectrometry:potentials and limitations. Anal. Chem. 71, 235–242

4. Annan, R. S., and Carr, S. A. (1996) Phosphopeptide analysis by matrix-assisted laser desorption time-of-flight mass spectrometry. Anal. Chem.68, 3413–3421

5. Annan, R. S., and Carr, S. A. (1997) The essential role of mass spectrometryin characterizing protein structure: mapping posttranslational modifica-tions. J. Protein Chem. 16, 391–402

6. Liao, P. C., Leykam, J., Andrews, P. C., Gage, D. A., and Allison, J. (1994)An approach to locate phosphorylation sites in a phosphoprotein: massmapping by combining specific enzymatic degradation with matrix-as-sisted laser desorption/ionization mass spectrometry. Anal. Biochem.219, 9–20

7. Carr, S. A., Huddleston, M. J., and Annan, R. S. (1996) Selective detectionand sequencing of phosphopeptides at the femtomole level by massspectrometry. Anal. Biochem. 239, 180–192



8. Zhang, X., Herring, C. J., Romano, P. R., Szczepanowska, J., Brzeska, H.,Hinnebusch, A. G., and Qin, J. (1998) Identification of phosphorylationsites in proteins separated by polyacrylamide gel electrophoresis. Anal.Chem. 70, 2050–2059

9. Asara, J. M., and Allison, J. (1999) Enhanced detection of phosphopeptidesin matrix-assisted laser desorption/ionization mass spectrometry usingammonium salts. J. Am. Soc. Mass. Spectrom. 10, 35–44

10. Ogueta, S., Rogado, R., Marina, A., Moreno, F., Redondo, J. M., andVazquez, J. (2000) Identification of phosphorylation sites in proteins bynanospray quadrupole ion trap mass spectrometry. J. Mass Spectrom.35, 556–565

11. Zhou, H. L., Watts, J. D., and Aebersold, R. (2001) A systematic approachto the analysis of protein phosphorylation. Nat. Biotechnol. 19, 375–378

12. McLachlin, D. T., and Chait, B. T. (2001) Analysis of phosphorylated pro-teins and peptides by mass spectrometry. Curr. Opin. Chem. Biol. 5,591–602

13. Mann, M., Ong, S. E., Gronborg, M., Steen, H., Jensen, O. N., and Pandey,A. (2002) Analysis of protein phosphorylation using mass spectrometry:deciphering the phosphoproteome. Trends Biotechnol. 20, 261–268

14. Knight, Z. A., Schilling, B., Row, R. H., Kenski, D. M., Gibson, B. W., andShokat, K. M. (2003) Phosphospecific proteolysis for mapping sites ofprotein phosphorylation. Nat. Biotechnol. 21, 1047–1054

15. Chalmers, M. J., Kolch, W., Emmett, M. R., Marshall, A. G., and Mischak, H.(2004) Identification and analysis of phosphopeptides. J. Chromatogr. B803, 111–120

16. Annan, R. S., Huddleston, M. J., Verma, R., Deshaies, R. J., and Carr, S. A.(2001) A multidimensional electrospray MS-based approach to phos-phopeptide mapping. Anal. Chem. 73, 393–404

17. Arnott. D., Gawinowicz. M. A., Grant. R. A., Neubert. T. A., Packman. L. C.,Speicher. K. D., Stone. K., and Turck, C. W. (2003) ABRF-PRG03: phos-phorylation site determination. J. Biomol. Tech. 14, 205–215

18. Bennett, K. L., Stensballe, A., Podtelejnikov, A. V., Moniatte, M., andJensen, O. N. (2002) Phosphopeptide detection and sequencing bymatrix-assisted laser desorption/ionization quadrupole time-of-flight tan-dem mass spectrometry. J. Mass Spectrom. 37, 179–190

19. Ma, Y. L., Lu, Y., Zeng, H. Q., Ron, D., Mo, W. J., and Neubert, T. A. (2001)Characterization of phosphopeptides from protein digests using matrix-assisted laser desorption/ionization time-of-flight mass spectrometryand nanoelectrospray quadrupole time-of-flight mass spectrometry.Rapid Commun. Mass Spectrom. 15, 1693–1700

20. Janek, K., Wenschuh, H., Bienert, M., and Krause, E. (2001) Phosphopep-tide analysis by positive and negative ion matrix-assisted laser desorp-tion/ionization mass spectrometry. Rapid Commun. Mass Spectrom. 15,1593–1599

21. Hunt, D. F., Yates, J. R. d., Shabanowitz, J., Winston, S., and Hauer, C. R.(1986) Protein sequencing by tandem mass spectrometry. Proc. Natl.Acad. Sci. U. S. A. 83, 6233–6237

22. Ficarro, S. B., McCleland, M. L., Stukenberg, P. T., Burke, D. J., Ross,M. M., Shabanowitz, J., Hunt, D. F., and White, F. M. (2002) Phospho-proteome analysis by mass spectrometry and its application to Saccha-romyces cerevisiae. Nat. Biotechnol. 20, 301–305

23. Webster, M. K., and Donoghue, D. J. (1997) FGFR activation in skeletaldisorders: too much of a good thing. Trends Genet. 13, 178–182

24. Raffioni, S., Zhu, Y. Z., Bradshaw, R. A., and Thompson, L. M. (1998) Effectof transmembrane and kinase domain mutations on fibroblast growthfactor receptor 3 chimera signaling in PC12 cells. A model for the controlof receptor tyrosine kinase activation. J. Biol. Chem. 273, 35250–35259

25. Naski, M. C., Wang, Q., Xu, J., and Ornitz, D. M. (1996) Graded activationof fibroblast growth factor receptor 3 by mutations causing achondro-plasia and thanatophoric dysplasia. Nat. Genet. 13, 233–237

26. Kan, S. H., Elanko, N., Johnson, D., Cornejo-Roldan, L., Cook, J., Reich,E. W., Tomkins, S., Verloes, A., Twigg, S. R., Rannan-Eliya, S., Mc-Donald-McGinn, D. M., Zackai, E. H., Wall, S. A., Muenke, M., and Wilkie,A. O. (2002) Genomic screening of fibroblast growth-factor receptor 2reveals a wide spectrum of mutations in patients with syndromic cranio-synostosis. Am. J. Hum. Genet. 70, 472–486

27. Mohammadi, M., Dikic, I., Sorokin, A., Burgess, W. H., Jaye, M., andSchlessinger, J. (1996) Identification of six novel autophosphorylationsites on fibroblast growth factor receptor 1 and elucidation of theirimportance in receptor activation and signal transduction. Mol. Cell. Biol.16, 977–989

28. Shevchenko, A., Jensen, O. N., Podtelejnikov, A. V., Sagliocco, F., Wilm,M., Vorm, O., Mortensen, P., Boucherie, H., and Mann, M. (1996) Linkinggenome and proteome by mass spectrometry: large-scale identificationof yeast proteins from two dimensional gels. Proc. Natl. Acad. Sci.U. S. A. 93, 14440–14445

29. He, T., Alving, K., Feild, B., Norton, J., Joseloff, E. G., Patterson, S. D., andDomon, B. (2004) Quantitation of phosphopeptides using affinity chro-matography and stable isotope labeling. J. Am. Soc. Mass Spectrom. 15,363–373

30. Gorman, J. J., Ferguson, B. L., and Nguyen, T. B. (1996) Use of 2,6-dihydroxyacetophenone for analysis of fragile peptides, disulphidebonding and small proteins by matrix-assisted laser desorption ioniza-tion. Rapid Commun. Mass Spectrom. 10, 529–536

31. Kjellstrom, S., and Jensen, O. N. (2004) Phosphoric acid as a matrixadditive for MALDI MS analysis of phosphopeptides and phosphopro-teins. Anal. Chem. 76, 5109–5117

32. Mohammadi, M., Schlessinger, J., and Hubbard, S. R. (1996) Structure ofthe FGF receptor tyrosine kinase domain reveals a novel autoinhibitorymechanism. Cell 86, 577–587

33. Kim, J., Camp, D. G., and Smith, R. D. (2004) Improved detection ofmulti-phosphorylated peptides in the presence of phosphoric acid inliquid chromatography/mass spectrometry. J. Mass Spectrom. 39,208–215



Identification of Phosphopeptides by MALDI Q-TOF MS in Positive and Negative Ion Modes after Methyl...

Documents

Transcript of Identification of Phosphopeptides by MALDI Q-TOF MS in Positive and Negative Ion Modes after Methyl...