Detection of Multiphosphorylated Peptides inLC-MS/MS Analysis under Low pH Conditions

Hyunwoo Choi, Hye-suk Lee, and Zee-Yong Park*

Department of Life Sciences, Gwangju Institute of Science & Technology 1 Oryong-Dong, Buk-Gu, Gwangju, Korea 500-712

An improved method of detection of multiphosphorylatedpeptides by RPLC-MS/MS analysis under low pH condi-tions (pH ∼1.7, 3% formic acid) is demonstrated for themodel phosphoproteins, bovine r- and â-casein. Changesin the pH conditions from normal (pH ∼3.0, 0.1% formicacid) to low (pH ∼1.7, 3% formic acid) significantlyimproved the detection limit of multiphosphorylated pep-tides carrying negative (-) solution charge states. Inparticular, bovine â-casein tetraphosphorylated peptide,was detected with a loading amount of only 50 fmol oftrypsin-digested bovine â-casein under low pH conditions,which is 200 times lower than necessary to detect thepeptide under normal pH conditions. In order to under-stand the low pH effect, various loading amounts oftrypsin-digested bovine r- and â-caseins were analyzedby RPLC-MS/MS analyses under two different pH condi-tions. The question of whether the low pH conditionimproves the detection of multiphosphorylated peptidesby increasing ionization efficiencies could not be provenin this study because synthetic multiphosphorylated pep-tides could not be easily obtained by peptide synthesis.Interestingly, increased hydrophilicity resulting from mul-tiple phosphorylation events is shown to negatively affectthe peptide retention on reversed-phase column material.It was also demonstrated that the low pH condition couldeffectively enhance the retention of multiphosphorylatedpeptides on reversed-phase column material. The useful-ness of low pH RPLC analysis was tested using an actualphosphopeptide-enriched sample prepared from mousebrain tissues. Previously, low pH solvents have been usedin SCX fractionation and TiO2 enrichment processes toselectively enrich phosphopeptides during the phospho-peptide enrichment procedure, but the improved detec-tion of multiphosphorylated peptides in RPLC-MS/MSanalysis under low pH conditions has not been reportedbefore (Ballif, B. A.; Villen, J.; Beausoleil, S. A.; Schwartz,D.; Gygi, S. P. Mol. Cell. Proteomics 2004, 3, 1093-1101. Villen, J.; Beausoleil, S. A.; Gerber, S. A.; Gygi, S.P. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 1488-1493. Schlosser, A.; Vanselow, J. T.; Kramer, A. Anal.Chem. 2005, 77, 5243-5250.)

Mass spectrometry became a method of choice for proteinphosphorylation analysis because it can provide both qualitative

and quantitative information about protein phosphorylation withoutthe need for phosphorylation site-specific antibodies.4-11 Manybiological systems have been analyzed by mass spectrometry toreveal the role of protein phosphorylation in important biologicalprocesses.1,12-16 Mass spectrometric analysis of protein phospho-rylation is often not successful, however, because there are certainlimitations.4 For example, a low stoichiometry of phosphorylatedproteins to unmodified proteins and extensive numbers of phos-phorylation sites are known to be the two characteristics of proteinphosphorylation that hamper routine analysis.17 In addition,current mass spectrometric techniques usually analyze peptidesin positive ion mode to increase the detection of both unmodifiedand modified peptides. The acidic property, i.e., negative (-)charge of phosphorylated amino acids could lower the degree ofprotonation; therefore, the ionization efficiencies of phosphorylatedpeptides in positive ion mode could be adversely affected.4,18

A number of mass spectrometric techniques have beendeveloped to increase the detection limit of phosphorylated

* To whom correspondence should be addressed. E-mail: zeeyong@gist.ac.kr.

(1) Ballif, B. A.; Villen, J.; Beausoleil, S. A.; Schwartz, D.; Gygi, S. P. Mol. Cell.Proteomics 2004, 3, 1093-1101.

(2) Villen, J.; Beausoleil, S. A.; Gerber, S. A.; Gygi, S. P. Proc. Natl. Acad. Sci.U.S.A. 2007, 104, 1488-1493.

(3) Schlosser, A.; Vanselow, J. T.; Kramer, A. Anal. Chem. 2005, 77, 5243-5250.

(4) Garcia, B. A.; Shabanowitz, J.; Hunt, D. F. Methods 2005, 35, 256-264.(5) Pandey, A.; Podtelejnikov, A. V.; Blagoev, B.; Bustelo, X. R.; Mann, M.;

Lodish, H. F. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 179-184.(6) Yan, J. X.; Packer, N. H.; Gooley, A. A.; Williams, K. L. J. Chromatogr., A

1998, 808, 23-41.(7) Annan, R. S.; Huddleston, M. J.; Verma, R.; Deshaies, R. J.; Carr, S. A. Anal.

Chem. 2001, 73, 393-404.(8) McLachlin, D. T.; Chait, B. T. Curr. Opin. Chem. Biol. 2001, 5, 591-602.(9) Mann, M.; Ong, S. E.; Gronborg, M.; Steen, H.; Jensen, O. N.; Pandey, A.

Trends. Biotechnol. 2002, 20, 261-268.(10) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc.

Natl. Acad. Sci. U.S.A. 2003, 100, 6940-6945.(11) Goshe, M. B.; Conrads, T. P.; Panisko, E. A.; Angell, N. H.; Veenstra, T.

D.; Smith, R. D. Anal. Chem. 2001, 73, 2578-2586.(12) Beausoleil, S. A.; Jedrychowski, M.; Schwartz, D.; Elias, J. E.; Villen, J.; Li,

J.; Cohn, M. A.; Cantley, L. C.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A.2004, 101, 12130-12135.

(13) Blagoev, B.; Ong, S. E.; Kratchmarova, I.; Mann, M. Nat. Biotechnol. 2004,22, 1139-1145.

(14) 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. 3rd Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 7900-7905.

(15) Olsen, J. V.; Blagoev, B.; Gnad, F.; Macek, B.; Kumar, C.; Mortensen, P.;Mann, M. Cell 2006, 127, 635-648.

(16) Li, X.; Gerber, S. A.; Rudner, A. D.; Beausoleil, S. A.; Haas, W.; Villen, J.;Elias, J. E.; Gygi, S. P. J. Proteome Res. 2007, 6, 1190-1197.

(17) Steen, H.; Jebanathirajah, J. A.; Rush, J.; Morrice, N.; Kirschner, M. W.Mol. Cell. Proteomics 2006, 5, 172-181.

(18) Liu, S.; Zhang, C.; Campbell, J. L.; Zhang, H.; Yeung, K. K.; Han, V. K.;Lajoie, G. A. Rapid. Commun. Mass Spectrom. 2005, 19, 2747-2756.

Anal. Chem. 2008, 80, 3007-3015

10.1021/ac7023393 CCC: $40.75 © 2008 American Chemical Society Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 3007Published on Web 03/18/2008

peptides and the coverage of phosphorylation sites.11,14,19-26 Mostof these techniques employ metal-chelate affinity chromatogra-phy or ion-exchange chromatography to selectively enrichthe substoichiometric phosphorylated peptides or phosphopro-teins.1,19,24,25 Chemical modification type approaches have also beenintroduced. For example, phosphorylated peptides/proteins canbe chemically modified to remove the negatively charged stateof the phosphate moiety in order to increase ionization efficiencies.A second approach involves chemically changing phosphorylatedamino acids into lysine homologues to allow distinct traces to bemade after trypsin digestion.11,21-23,27 Other approaches usingadditives such as phosphoric acid or EDTA have also beendeveloped; these techniques are designed to minimize unwantedinteractions between phosphorylated peptides and reversed-phasecolumn material or metal ions in an LC system.18,28

Among the mass spectrometric techniques that have beendeveloped, only a few of them have focused on the improveddetection of multiphosphorylated peptides in LC-MS analy-sis.18,28,29 The ionization behavior of multiphosphorylated peptidesis reported to be quite different from that of mono- or diphospho-rylated peptides.18,28,29 For example, trypsin digestion of R- andâ-casein generates multiphosphorylated peptides (3, 4, and 5phosphorylations) in addition to mono- or diphosphorylatedpeptides. Some of these multiphosphorylated peptides exhibitsignificantly decreased ion intensities compared to the lessphosphorylated peptides by conventional RPLC-MS/MS analysis,and a significantly larger amount of enzymatic digests is requiredin order to detect them.18,28,29 We have recently determined thatsub-stoichiometric yields of multiphosphorylated bovine â-caseinpeptides were not due to inefficient trypsin digestion near themultiphosphorylation sites.29 The tryptic digestion pattern ofbovine caseins is relatively simple; therefore, the ionization ofmultiphosphorylated peptides might be governed by three previ-ously reported causes for the unsuccessful mass spectrometricidentification of phosphorylation sites: (1) their decreased reten-tion on reversed-phase column material due to increased hydro-philicity, (2) selectively suppressed ionization by more ionizablecoeluting peptides, and (3) lower ionization efficiency thanunmodified counterparts.17 One way of investigating the effectsof multiple phosphorylation on the ionization behavior of phos-phopeptides would be to use synthetic phosphopeptides; however,it is difficult to obtain synthetic phosphopeptides with more thanthree modification sites by peptide synthesis. For this reason,researchers have used bovine R- and â-casein tryptic digest

peptides as model phosphopeptides with multiple phosphorylationsites. Although such multiphosphorylated peptides may not bepredominant in LC-MS analyses, the presence of phosphopep-tides with more than three modifications has often been observedin actual proteome samples. Indeed, a recent study by Gygi et al.identified more than 900 multiphosphorylated peptides from alarge-scale phosphorylation analysis of primary animal tissuesample (liver).2

In this study, we demonstrate the significantly improveddetection of multiphosphorylated peptides of bovine R- andâ-casein by RPLC-MS/MS analysis under low pH conditions. Wereasoned that lowering the pH of the elution solvents during RPLCseparation could decrease the negatively charged phosphopeptidespecies (deprotonated form) and thereby increase the degree ofprotonation or the ionization efficiency of multiphosphorylatedpeptides. We also investigated whether the increased hydrophi-licity or negative (-) solution charge state resulting from multiplephosphorylation affects peptide retention on reversed-phasecolumn material. To validate the utility of low pH phosphopeptideanalysis in actual proteome samples, we analyzed a phosphopep-tide enriched sample of mouse brain tissue lysates and comparedthe number of phosphopeptides identified and the number ofmodification sites per peptide under two different pH conditions.

METHODSBovine â-casein, bovine R-casein, phosphoric acid, formic acid,

ammonium bicarbonate, ammonium hydroxide, calcium chloride,dithiothreitol (DTT), trifluoroacetic acid (TFA), and iodoacetamide(IAA) were purchased from Sigma. Acetonitrile, methanol, andHPLC grade water were all purchased from Fisher Science.Sequencing grade modified trypsin was obtained from Promega.Aqua C18 (particle size 5 µm) reversed-phase column materialwas purchased from Phenomenex. Titanium oxide (Titansphere,particle size 5 µm) was purchased from GL Sciences(GL SciencesInc., Tokyo, Japan).

In-Solution Enzymatic Digestion of r- and â-Casein.R-Casein. Bovine R-casein that had been dissolved in 50 mM NH4-HCO3 (pH 8.0) was reduced by adding 5 mM DTT at roomtemperature for 1 h and alkylated with 25 mM IAA at roomtemperature for 30 min in the dark. In-solution digestion wasperformed using sequencing grade modified trypsin in 50 mMNH4HCO3 (pH 8.0) at a 1:50 (w/w) trypsin-to-protein ratio for 16h at 37 °C. The digested peptides were adjusted with 5% formicacid (v/v) to quench the enzyme activity.

â-Casein. Bovine â-casein in 5 mM CaCl2/100 mM Tris buffer(pH 8.0) was digested using sequencing grade modified trypsinin 5 mM CaCl2/50 mM NH4HCO3 (pH 8.0) at a 1:50 (w/w) trypsin-to-protein ratio for 12-14 h at 37 °C. Bovine â-casein does nothave a disulfide bond, and thus, reduction and alkylation processeswere not carried out. The digested peptides were adjusted with5% formic acid (v/v) to quench the enzyme activity and wereimmediately stored at -80 °C.

Mouse Brain Tissue Collection. At 8 weeks of age, ICR malemice were obtained from Samtako Bio Korea and were sacrificedby vertebral fracture without anesthesia. The whole brain wasimmediately dissected and washed with ice-cold 1× PBS. Thesamples were immediately frozen in liquid nitrogen and trans-ferred to a -80 °C freezer.

(19) Andersson, L.; Porath, J. Anal. Biochem. 1986, 154, 250-254.(20) 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.

(21) Knight, Z. A.; Schilling, B.; Row, R. H.; Kenski, D. M.; Gibson, B. W.; Shokat,K. M. Nat. Biotechnol. 2003, 21, 1047-1054.

(22) Li, W.; Boykins, R. A.; Backlund, P. S.; Wang, G.; Chen, H. C. Anal. Chem.2002, 74, 5701-5710.

(23) McLachlin, D. T.; Chait, B. T. Anal. Chem. 2003, 75, 6826-6836.(24) Oda, Y.; Nagasu, T.; Chait, B. T. Nat. Biotechnol. 2001, 19, 379-382.(25) Pinkse, M. W.; Uitto, P. M.; Hilhorst, M. J.; Ooms, B.; Heck, A. J. Anal.

Chem. 2004, 76, 3935-3943.(26) Zhou, H.; Watts, J. D.; Aebersold, R. Nat. Biotechnol. 2001, 19, 375-378.(27) Vosseller, K.; Hansen, K. C.; Chalkley, R. J.; Trinidad, J. C.; Wells, L.; Hart,

G. W.; Burlingame, A. L. Proteomics 2005, 5, 388-398.(28) Kim, J.; Camp, D. G., 2nd; Smith, R. D. J. Mass Spectrom. 2004, 39, 208-

215.(29) Kim, S.; Choi, H.; Park, Z. Y. Mol. Cells 2007, 23, 340-348.

3008 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008

Protein Extraction and Enzymatic Digestions of MouseBrain. After the whole mouse brain was ground using a mortarand pestle, the powder was collected in Eppendorf tubes that hadbeen previously weighed, and the final weight of the tissue wascalculated. To make protein extracts, the powder was dissolvedin an appropriate volume of extraction buffer containing 50 mMTris-HCl (pH 8.0), 8 M urea, and 1 mM DTT. The sample wasincubated with rotation at 4 °C for 2 h and sonicated five times at30% amplitude for 3 s on ice. In order to remove insoluble tissuedebris from the suspensions, the samples were centrifuged at100 000g for 1 h. The supernatant was transferred into separatetubes, and the protein concentration was determined using a BCAassay kit (Bio-Rad). After reduction (5 mM DTT, 1 h) andalkylation (25 mM IAA, 30 min in the dark), endoproteinase Lys-C(Roche) was added at a 1:100 enzyme/substrate ratio anddigestion was allowed to proceed at 37 °C for 14 h. After 4 ×dilution with 100 mM Tris (pH 8.0), lysate proteins were furtherdigested with sequencing grade trypsin at a 1:50 enzyme/substrateratio in the presence of calcium chloride (2 mM), at 37 °C,overnight.

Isolation of Phosphopeptides Using Titansphere. Boden-miller’s method was employed to enrich phosphopeptides usingTitansphere.30 A 250-µm-i.d., 360-µm-o.d. fused-silica capillary wasfilled with 6-7 cm of TiO2 using a MicroFilter. Before initial use,the column was washed twice with 100 µL of 0.3 M NH4HCO3

(pH 10.5) and equilibrated with 80% CH3CN, 0.1% TFA. Samplesin 80% CH3CN, 2.5% TFA were loaded on to the column at a flowrate of 1 µL/min, and then the column was washed twice with80% CH3CN, 0.1% TFA, and finally twice with 0.1% TFA. Elutionwas performed with 0.3 M NH4HCO3 (pH 10.5). Eluted peptideswere dried and reconstituted in 0.1% TFA before LC-MS/MSanalysis

RPLC Analysis under Normal pH Buffer Conditions andLow pH Buffer Conditions. Normal pH Buffer Conditions. Fornormal pH mobile-phase solutions, buffer A (5% CH3CN and 0.1%formic acid) and buffer B (80% CH3CN and 0.1% formic acid) wereused.

Low pH Buffer Conditions. For low pH mobile-phase solutions,buffer A (5% CH3CN and 3% formic acid) and buffer B (80% CH3-CN and 3% formic acid) were used. Peptides were loaded onto afused-silica capillary column (100-µm i.d., 360-µm o.d.) containing7.5 cm of Aqua C18 reversed-phase column material (Phenom-enex). The following gradient method was used for the separationof peptide mixtures over a period of 85 min: 0-10% B in 1 min;10-60% B in 50 min; 60-100% B in 10 min; then 100% B hold for15 min. Peptides were eluted at a flow rate of 250 nL/min providedacross a flow splitter by the HPLC pumps.

Mass Spectrometry. Analyses of peptide samples wereperformed using an Agilent 1100 Series HPLC pump (AgilentTechnologies) coupled with a quadrupole ion trap mass spec-trometer (LCQ Deca XP Plus, Thermo Finnigan, San Jose, CA)or a linear quadrupole ion trap mass spectrometer (LTQ, ThermoFinnigan) using an in-house-built nano ESI interface. In order toidentify the eluting peptides, the ion trap mass spectrometer wasoperated in a data-dependent MS/MS mode (m/z 400-1400), inwhich a full MS scan was followed by 10 MS/MS scans (LTQ) or

3 MS/MS scans (LCQ). The temperature of the heated capillarywas 200 °C, and 2.3 kV of electrospray voltage was applied via agold electrode. The dynamic exclusion function of the Xcalibursoftware was employed to minimize redundant MS/MS dataacquisition for identical peptides. Normalized collision energy of35% was used to generate MS/MS spectra.

Mass Spectrometric Data Analysis. MS/MS spectra thatwere obtained in the LC-MS/MS analyses were searched againstan in-house protein database containing the bovine R- and â-caseinsequence or a composite protein database containing the mouseIPI protein database (v.3.28) and its reversed complement usingTurboSequest or Sequest Sorcerer Solo (SageN). The differentialmodification search option for phosphorylation modification (+80on Ser, Thr, Tyr) and oxidation (+16 on Met) was considered inthe search, and the maximum number of modifications that wereallowed per peptide was seven (the maximum number of modi-fication per type was 5). Scaffold (Proteomesoftware, version 1.6)was used to filter the search results, and the following Xcorr valueswere applied to fully tryptic peptides based on the charge state:2.0 for singly charged peptides, 2.0 for doubly charged peptides,3.0 for triply charged peptides, and 0.0 for ∆ Cn. Somewhat lessstringent criteria were used in the Sequest score filtering processsince phosphopeptides tend to show low Xcorr scores due toextensive neutral loss of phosphate group. A false positive rate ofthese filtering criteria was estimated to be over 10%, thus manualvalidation of all MS/MS spectra assigned as phosphopeptides wasfollowed to confirm the search results and to remove incorrectassignments. The presence of neutral loss of phosphoric acid (M+ H - 98 for singly charged, M + 2H - 49 Da for doubly charged,and M + 3H - 32.6 for triply charged peptides) was used first asa way of confirming Ser- and Thr-phosphorylated peptides andmajor fragment ions (b and y series ions, N-Terminal sidefragment ions of Pro-containing peptides, neutral loss ion ofoxidized Met-containing peptides) of the phosphorylated peptidesequences were then manually assigned and checked for theirintensities. The requirement of neutral loss of phosphoric acidwas not applied to Tyr phosphorylated peptides. Manual valida-tions appeared to be necessary in the phosphopeptide analysisbecause the number of MS/MS spectra removed after manualvalidations was always higher than those calculated from estimatedfalse positive rates. For example, 26% of filtered MS/MS spectrawere actually removed in a 12.5% false positive rate phosphopep-tide search case (10 reverse sequence phosphopeptide matchesout of 159 total phosphopeptide matches).

Construction of Selected Ion Chromatograms. XCaliburv.2.0 was used to construct selected ion chromatograms ofphosphopeptides. Peptide ions in a range of m/z 400-1400 werescanned, and mass window was set at 500 ppm. A peak smoothingfeature using Gaussian function was employed prior to the peakintegration process. The peak area of each peptide was calculatedwith the built-in peak area integration feature of Xcalibur v.2.0.Each peak in the selected ion chromatogram was manuallyconfirmed by the presence of corresponding tandem mass spectraand residual peaks resulting from the background noises in theMS scan were removed for clarity.

RESULTSPreviously, various mass spectrometric phosphopeptide analy-

sis techniques have been tested using one of the model phos-(30) Bodenmiller, B.; Mueller, L. N.; Mueller, M.; Domon, B.; Aebersold, R.

Nat. Methods 2007, 4, 231-237.

Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 3009

phoproteins, bovine R- and â-casein, to measure detection limits.In this study, we compared two different pH conditions (0.1%formic acid and 3% formic acid containing elution solvents) forRPLC-MS/MS analysis to identify any changes in the level ofdetection of phosphorylated â-casein peptides. We varied theamounts of trypsin-digested â-casein that were loaded, and thetandem mass spectrometric database search score was monitoredfor phosphorylated peptides (mono- and tetraphosphorylated)under two different pH conditions.(Table 1) As previously re-ported, the detection limit of â-casein tetraphosphorylated peptideunder normal pH conditions (conventional method) was ∼10 pmol.Changes in the pH condition from 2.9 to 1.7 significantly improvedthe detection limit of the multiphosphorylated peptide. Indeed,

the tetraphosphorylated peptide was detected when only 50 fmolwas loaded under low pH conditions, which is 200 times less thanthe amount of peptide that is necessary for detection under normalpH conditions. The detection limit of monophosphorylated peptidewas also slightly enhanced under low pH conditions. pH conditionlower than pH 1.7 were tested using higher concentrations offormic acid; however, under these conditions, the detection limitof â-casein phosphopeptides was not improved.(data not shown)Figure 1 shows a base peak chromatogram of 50 fmol of bovineâ-casein tryptic digests that were obtained by RPLC-MS/MSanalysis under low pH conditions (inset) and a manually assignedtandem mass spectrum of â-casein tetraphosphorylated peptide.Characteristic neutral losses of phosphoric acid from fragmentions and the parent ion were evident in the MS/MS spectra.

Since the improved detection limit of â-casein tetraphospho-rylated peptide in low pH conditions can be interpreted asincreased ionization efficiencies, the ion intensities of phospho-rylated and unmodified peptides under two different pH conditionswere examined. Instead of a direct ion intensity comparison,selected ion chromatograms of individual peptide ions wereconstructed for the elution profiles and the peak areas werecalculated.(Table 2) Tryptic digest peptides of R-casein, a majorcontaminant of â-casein, were also analyzed and included in thecomparison of ionization efficiencies. While tetraphosphorylatedâ-casein peptides exhibited greater peak areas in the low pHanalysis, nearly all of the monophosphorylated and unmodifiedâ-casein peptides showed similar or greater peak areas in thenormal pH analysis. In order to explain the improved ionizationefficiencies of â-casein tetraphosphorylated peptides under lowpH conditions, elution profiles of â-casein tryptic digest peptideswere compared using selected ion chromatograms. Interestingly,tetraphosphorylated peptides under normal pH conditions showedan unusual elution profile compared to tetraphosphorylatedpeptides under low pH conditions.(Figure 2) The elution of

Table 1. Comparisons of Tandem Mass SpectrometryDatabase Search Score of Bovine â-Casein Mono- andTetraphosphorylated Peptides with Various LoadingAmounts under Normal (0.1% Formic Acid) and Low pH(3% Formic Acid) Elution Solvent Conditions(Instrument: LCQ Deca XP)

tandem mass spectrometric databasesearch score (SEQUEST Xcorr)

monophosphorylatedpeptide

FQpSEEQQQTEDELQDK

tetraphosphorylated peptideRELEELNVPGEIVEpS-

LpSpSpSEESITR

bovineâ-casein 2+ 3+ 3+

Low pH Condition9.6 pmol 4.5 3.6 51 pmol 4.6 - 4.8100 fmol 5.0 - 4.750 fmol 5.3 - 3.9

Normal pH Condition9.6 pmol 4.4 - 4.71 pmol 4.9 - nd100 fmol 4.3 - nd

Figure 1. MS/MS spectrum of tetraphosphorylated peptide obtained from a loading amount of 50 fmol of bovine â-casein tryptic digests in theRPLC-MS/MS analysis. Fragment ions were manually assigned with a mass tolerance of (0.5 Da. Inset shows the base peak ion chromatogramobtained in the LC-MS/MS analysis and the retention time of the â-casein tetraphosphorylated peptide.

3010 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008

tetraphosphorylated peptide under normal pH conditions wasextended over a broad range of retention time, which is verydifferent from the conventional peptide elution that was observedin the RPLC separation. Tetraphosphorylated peptide under lowpH conditions and other tryptic digest peptides from â-casein allshowed typical peptide elution profiles. The elution profiles ofmonophosphorylated â-casein peptides under normal and low pHconditions are shown in Figure 3 for comparison purposes.

The tendency for the improved ionization efficiency of mul-tiphosphorylated peptides under low pH conditions was alsoconfirmed in a separate RPLC-MS/MS analysis of bovineR-casein, another model phosphoprotein.(Table 3) A trypticdigestion of bovine R-casein generates phosphorylated peptidesof various modification degrees: two monophosphorylated pep-tides, two diphosphorylated peptides, one triphosphorylated pep-tide, two tetraphosphorylated peptides, and one pentaphospho-rylated peptide. Various amounts of bovine R-casein tryptic digestswere analyzed under normal and low pH conditions. The tandemmass spectrometric database search score was monitored foridentified R-casein phosphopeptides. We identified a total of sixphosphopeptides (out of eight phosphopeptides) from the trypticdigestion of R-casein; phosphopeptides containing more than twophosphorylation sites exhibited better ionization efficiencies in lowpH conditions. When the elution profiles of R-casein tetraphos-phorylated peptides were compared in the two pH conditions, onlyNANEEEYSIGpSpSpSEEpSAEVATEEVK exhibited extended elu-tion under normal pH conditions.(Figure 4)

In order to validate the use of low pH phosphopeptide analysisin an actual phosphoproteome analysis, a complex phosphopeptidemixture sample that was prepared from mouse brain whole tissuelysate was analyzed. A total of 50 µg of mouse brain protein wastrypsinized to generate peptide mixtures, and then a one-stepphosphopeptide enrichment procedure was performed using thephosphopeptide affinity column material, Titansphere (TiO2). Inorder to increase the phosphopeptide coverage of the sample,aliquots from the phosphopeptide-enriched sample were analyzedby RPLC-MS/MS at least twice under each pH condition, andthe two best tandem mass spectrometric database search resultsof each pH condition were combined to make phosphopeptide/protein lists. We applied simple and stringent criteria whencompleting the phosphopeptide lists in order to quickly comparethe results between the two pH conditions. SEQUEST score-basedfiltering and manual validation was used. The SEQUEST scorefiltering was only applied to phosphopeptides with fully trypticdigest ends and the following Xcorr values for different chargestate ions (singly charged, 1.8; doubly charged, 2.0; triply charged,3.0; ∆ Cn, 0) were employed. Somewhat less stringent criteriawere used in the filtering process since phosphorylated peptidesusually exhibit lower Xcorr scores than unmodified counterparts.All the filtered MS/MS spectra were then manually validated withthe requirements of having phosphate group neutral loss ions asthe major ions (within 10% of base peak ion intensity) andmatching major fragment ions (b or y series ions, N-terminal sidepreferential cleavage of Pro-containing case, and neural loss ofoxidized Met containing case). The number of phosphopeptides/proteins that were identified in the normal pH analysis was only63/43; however, the number of phosphopeptides/proteins thatwere identified in the low pH analysis was 137/89.(Table 4) Acomplete list of the phosphopeptides that were identified in thetwo pH analyses is provided as Supporting Information.(SI; Table1) More than 62% of the phosphopeptides that were identified inthe normal pH analysis were redundantly identified in the lowpH analysis. Most of the phosphopeptides that were identifiedcontained only one modification site per peptide.(84.1% for normalpH and 84.7% for low pH analysis) The number of multiphospho-rylated peptides with more than 2 modification sites was 10 (15.9%of total number of phosphopeptides) for normal pH analysis and21 (15.3% of a total number of phosphopeptides) for low pHanalysis. No more than three phosphorylation sites per peptidewere observed in the normal pH analysis, but phosphopeptideswith four or five modification sites were observed in the low pHanalysis.

DISCUSSIONThe mass spectrometric identification of multiphosphorylated

peptides in positive ion mode has been a challenging task becausethe ionization efficiencies of these peptides are much lower thantheir unmodified counterparts.18,28,29 In this study, we havedemonstrated that a simple pH change (from pH 2.9 to pH 1.7)in the elution solvents used in the RPLC-MS/MS analysis caneffectively improve the detection of multiphosphorylated peptidesof model phosphoproteins. Conventional RPLC separation uses alow concentration (0.1-0.4%) of formic acid to adjust the pHconditions of elution solvents; in this case, the pH ranges between2.5 and 3. Most unmodified tryptic digest peptides maintainsolution charge states of +2 at this pH range; however, the

Table 2. Comparisons of Integrated Peak Areas (× 106)of Phosphopeptides and Unmodified Peptides underLow pH and Normal pH Conditionsa

low pH normal pHPhosphopeptide

FQpSEEQQQTEDELQDK9.6 pmol 176.9 (53.99) 537.0 (18.27)1 pmol 22.0 (3.20) 63.0 (10.42)100 fmol 2.3 (0.27) 9.5 (0.96)RELEELNVPGEIVEpSLpSpSpSEESITR9.6 pmol 435.6 (124.17) 55.1 (16.98)1 pmol 48.4 (5.24) 1.6 (0.37)100 fmol 2.2 (0.96) not detected50 fmol 0.59 (0.033) not detectedVPQLEIVPNpSAEER (alpha casein S1)9.6 pmol 17.2 (8.51) 33.7 (4.88)1 pmol 2.5 (0.30) 3.4 (0.39)100 fmol 0.25 (0.042) 0.45 (0.063)TVDMEpSTEVFTK (alpha casein S2)9.6 pmol 25.0 (6.15) 60.3 (10.91)1 pmol 1.6 (3.20) 3.5 (1.50)100 fmol 0.16 (0.001) not detected

Non-PhosphopeptideMHQPHQPLPPTVMFPPQSVLSLSQSK+3

9.6 pmol 15.4 (2.70) 65.1 (3.06)1 pmol 1.8 (0.46) 2.0 (1.10)100 fmol 0.14 (0.003) Not detectedDMPIQAFLLYQEPVLGPVR+2

9.6 pmol 50.2 (2.74) 100.6 (9.42)1 pmol 3.2 (0.87) 4.7 (1.55)100 fmol 0.24 (0.007) 0.57 (0.144)HQGLPQEVLNENLLR+2 (R-casein S1)9.6 pmol 16.3 (13.91) 23.9 (2.37)1 pmol 2.5 (0.25) 2.5 (0.70)100 fmol 0.12 (0.019) 0.64 (0.176)FPQYLQYLYQGPIVLNPWDQVK+2

(R-casein S2)9.6 pmol 15.8 (3.77) 21.0 (3.02)1 pmol 1.2 (0.32) 1.1 (0.33)

a Standard deviation is shown in parentheses. [Three replicateexperiments on LTQ MS)

Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 3011

solution charge state of the phosphorylated peptides is lower than+2 depending on the number of phosphorylation sites and basicamino acid residues.2 Some multiphosphorylated peptides withvery low pI values (pI <2), can even have negative (-) solutioncharge states under conventional RPLC separation conditions. Forexample, the two multiphosphorylated peptides tested in this studyhave negative (-) solution charge states at pH 2.9: -1 for thebovine â-casein tetraphosphorylated peptide, RELEELNVPGEIVE-pSLpSpSpSEESITR and -2 for the bovine R-casein tetraphospho-rylated peptide, NANEEEYSIGpSpSpSEEpSAEVATEEVK. Inter-estingly, these multiphosphorylated peptides required significantlyhigher amounts to be loaded than other multiphosphorylatedpeptides in conventional RPLC-MS/MS analysis.(∼pmol vs∼fmol) It can be speculated that negative (-) solution chargestates lower the ionization efficiency of multiphosphorylatedpeptides. It was previously reported that protein gas-phase charge-state distributions are closely related to their conformation; thegas-phase charge-state distribution is determined, to some degree,by the solution charge-state distribution or the degree of proto-nation in solution.31,32 Unlike proteins, peptides are not expected

to have high-order structure and their gas-phase charge statepredominantly reflects the degree of protonation or the solutioncharge state. Based on these reports, we reasoned that loweringthe pH of the RPLC elution solvents could render the solutioncharge state of phosphorylated peptides less negative by decreas-ing the population of negatively charged phosphopeptide species.In addition, shifting the solution charge states of multiphospho-rylated peptides from negative (-) to positive (+) could help theelectrospray ionizations in positive ion mode. Multiphosphorylatedpeptides and phosphorylated peptides with many acidic aminoacids would have a higher solution charge state in low pHconditions, and their chances of ionization would be increased.One method of validating the low pH effects on increasedionization efficiencies would be to use synthetic phosphopeptides;however, it is quite difficult to obtain synthetic phosphopeptideswith more than three modification sites by peptide synthesis. Forthis reason, the question of whether the increased solution chargestate of multiphosphorylated peptides leads directly to increasedionization efficiencies could not be proven at this point.

In addition to the increased solution charge states of mul-tiphosphorylated peptides, there appears to be another factoraffecting the detection of multiphosphorylated peptides in RPLC-

(31) Mirza, U. A.; Chait, B. T. Anal. Chem. 1994, 66, 2898-2904.(32) Wang, F.; Tang, X. Biochemistry 1996, 35, 4069-4078.

Figure 2. Elution profiles of bovine â-casein tetraphosphorylated peptide under two different pH conditions. Three different loading amountsof bovine â-casein tryptic digests were analyzed by RPLC-MS/MS. (A) Selected ion chromatograms of m/z 1042.3 under low pH conditions,(B) Selected ion chromatograms of m/z 1042.3 under normal pH conditions.

3012 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008

MS/MS analysis. The tetraphosphorylated peptide of bovineâ-casein exhibited a very unusual RPLC elution profile undernormal pH conditions. Unlike other phosphorylated peptides orunmodified peptides, the RPLC elution of the bovine â-caseintetraphosphorylated peptide continued for almost 10 min. Mul-tiphosphorylated peptides tend to be far more hydrophilic than

their unmodified cognates, and greater hydrophilicity couldinterfere with phosphopeptide separation in the RPLC analysisby weakening the interaction between peptide and reversed-phasecolumn material. Previously, Steen and his co-workers conducteda systematic study using synthetic phosphopeptides to see theeffects of phosphorylation on RPLC elution time.17 When the

Figure 3. Elution profiles of bovine â-casein monophosphorylated peptide under two different pH conditions. Three different loading amountsof bovine â-casein tryptic digests were analyzed by RPLC-MS/MS. (A) Selected ion chromatograms of m/z 1032.1 under low pH conditions,(B) Selected ion chromatograms of m/z 1032.1 under normal pH conditions.

Table 3. Comparison of Tandem Mass Spectrometric Database Search Score of Bovine r-Casein Tryptic DigestPhosphopeptides under Two Different pH Conditions

normal pH low pH

10 pmol 1 pmol 100 fmol 10 fmol 10 pmol 1 pmol 100 fmol 10 fmol

monophosphorylated peptideVPQLEIVPNpSAEER+2 (S1) 4.42 4.34 4.36 4.45 4.81 3.96 4.65 4.34TVDMEpSTEVFTK+2 (S2) 3.23 2.87 3.80 3.76 3.94 3.37 4.05 3.57diphosphorylated peptideDIGpSEpSTEDQAMEDIK+2 (S1) 3.57 6.72 6.85 n/d 6.35 6.49 5.85 4.75triphosphorylated peptideVNELpSKDIGpSEpSTEDQAMEDIK+2 (S1) 3.43 2.33 n/d n/d 2.38 2.51 3.01 n/dVNELpSKDIGpSEpSTEDQAMEDIK+3 (S1) 4.81 4.54 n/d n/d 5.11 4.67 5.94 n/dtetraphosphorylated peptideKNTMEHVpSpSpSEESIIpSQETYK+2 (S2) 4.66 3.83 5.94 n/d 4.02 4.98 6.20 n/dNANEEEYSIGpSpSpSEEpSAEVATEEVK+2 (S2) 3.08 n/d n/d n/d 3.08 2.32 n/d n/dNANEEEYSIGpSpSpSEEpSAEVATEEVK+3 (S2) 3.92 n/d n/d n/d 4.13 3.66 n/d n/d

Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 3013

number of phosphorylation modifications exceeds the number ofbasic amino acids, the elution of a phosphorylated peptide tendsto precede that of an unmodified counterpart. Based on thesefindings, they proposed that the retention time of a phosphorylatedpeptide is governed by the overall hydrophilicity and the overallhydrophilicity of the phosphopeptide is defined by subtracting thenumber of basic amino acid from the number of phosphorylationsites. Steen and co-workers did not observe any phosphopeptideswith an unusual elution profile, probably because all of thephosphopeptides they tested had only minor modification (1-2phosphorylations per peptide), and all but one phosphopeptidehad a positive (+) solution charge state. Nonetheless, Steen etal. speculated that multiple phosphorylation without additionalbasic amino acid residues could cause serious retention problems.The unusual RPLC elution profile of the bovine â-casein tetra-phosphorylated peptide has also been observed by Lajoie et al.18

In the Lajoie et al. study, it was shown that metal ions on thesurface of the reversed-phase column material can cause seriousretention problems by forming phosphopeptide-metal ion com-plexes; this problem was resolved by the addition of EDTA tothe elution solvent. We tested the effects of EDTA on the analysis

of bovine â-casein in our LC-MS system (Agilent 1100 SeriesHPLC pump, Aqua C18 reversed-phase column material-Phenom-enex), but it did not improve the detection of the bovine â-caseintetraphosphorylated peptide.(data not shown) The level of metalions in our LC-MS system might be below the limit of causingserious detection or retention problems. Among the phosphoryl-ated peptides identified in our bovine R-casein analysis, NANEEEY-SIGpSpSpSEEpSAEVATEEVK (solution charge state, -2 at pH2.9) showed the same elution profile as the â-casein tetraphos-phorylated peptide under normal pH elution conditions. Incontrast, another tetraphosphorylated peptide of bovine R-casein,KNTMEHVpSpSpSEESIIpSQETYK (solution charge state, 0 at pH2.9) did not show the same extended elution profile as bovineâ-casein tetraphosphorylated peptides. This is probably becausethe presence of two basic amino acids (lysine and histidine) dropsthe overall hydrophilicity or increases the solution charge stateof the phosphopeptide. Notably, the two tetraphosphorylatedpeptides that exhibited an extended elution profile in normal pHconditions had a normal peptide elution profile in the low pH RPLCseparation. It can be speculated that lowering the pH of the elutionsolvent must have affected the overall hydrophilicity or solutioncharge state of the multiphosphorylated peptides and therebyraised the interaction between the phosphopeptides and thereversed-phase column material during RPLC separations.

Our low pH phosphopeptide analysis was applied to an actualphosphopeptide-enriched sample that was prepared from mousewhole brain tissue lysate. Although our low pH phosphopeptideanalysis technique demonstrated greater detection sensitivity formultiphosphorylated peptides in the model phosphoprotein samples,the number of multiphosphorylated peptides identified from thelow pH analysis of actual phosphopeptide-enriched samples was

Figure 4. Elution profiles of bovine R-casein tetraphosphorylated peptide obtained from the two different pH RPLC-MS/MS analyses. (A)Selected ion chromatogram of m/z 1505.1 under low pH conditions, (B) Selected ion chromatogram of 1505.1 under normal pH conditions.

Table 4. Tandem Mass Spectrometric DatabaseSearch Results of a Phosphopeptide-Enriched SamplePrepared from Mouse Brain Whole Tissue Lysates

no. ofphospho-peptides

no. ofphospho-proteins

no.of multiphos-

phorylatedpeptidesa

max no. of phos-phorylation sites

identified per peptide

normal pH analysis 63 43 10 3low pH analysis 137 89 21 5

a Only di-, tri-, tetra-, and pentaphosphorylated peptides are included.

3014 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008

not significantly greater than that of the normal pH analysis. Onepossible explanation could be the very high sample complexityin the actual phosphopeptide-enriched samples. Since the ioniza-tion efficiencies of multiphosphorylated peptides are lower thanthose of phosphopeptides with minor modifications in the low pHanalysis, suppressed ionization by many coeluting phosphopep-tides and unmodified peptides could hamper the successfuldetection of multiphosphorylated peptides. Recently, Gygi et al.conducted a phosphorylation analysis using a combined approachof low pH SCX phosphopeptide fractionation and an IMACphosphopeptide enrichment technique to investigate the phos-phorylation events in a primary animal tissue sample.2 Low pHSCX fractionation was carried out first to separate peptides into15 different fractions. An IMAC-based phosphopeptide enrichmentprocedure was then used to selectively enrich the phosphopep-tides in individual fractions. Conventional RPLC-MS/MS analysison the 15 phosphopeptide-enriched fractions revealed more than900 multiphosphorylated peptides. The sample complexity of theirsample is estimated to be at least 1 order of magnitude lower thanthat of our sample, and contamination from unmodified peptideswould be a far less severe (single-step versus two-step phospho-peptide enrichment procedure). These two factors (samplecomplexity and sample heterogeneity) appear to be the mainreason for the differences between the two phosphopeptideanalysis results. In order to prove the effectiveness of our low pHRPLC-MS/MS analysis in actual proteome samples, it would becrucial to employ more elaborate phosphopeptide enrichmentprocesses than the single-step phosphopeptide enrichment processthat we used.

Low pH elution solvents in SCX peptide separations and TiO2

phosphopeptide enrichment processes have been previouslyreported by several research groups.1-3 All of these attempts werenot intended to improve the detection sensitivity of phosphopep-tides in the RPLC-MS/MS analysis, but to selectively enrichphosphopeptides during the phosphopeptide enrichment process.

CONCLUSIONSThe use of low pH elution solvents in RPLC-MS/MS analysis

is shown to effectively improve the detection of bovine R- and

â-casein multiphosphorylated peptides. In particular, multiphos-phorylated peptides with low pI values (below 3) showed improveddetection under low pH conditions. Direct evidence for enhancingthe ionization efficiency of multiphosphorylated peptides usinglow pH solvents could not be provided by this study; however,negative (-) solution charge states or increased overall hydro-philicity of the phosphopeptides resulting from multiple phospho-rylation is shown to interfere with peptide retention on RPLCcolumn material. Clearly, the enhanced detection of multiphos-phorylated peptides under low pH conditions can be attributedto the reduced overall hydrophilicity resulting from neutralizingnegative charges.

One of the main benefits of RPLC-MS/MS analysis under lowpH conditions is that it can increase the detection sensitivity ofmultiphosphorylated peptides without an additional enrichmentprocess. Hence, this technique can be easily applied to any large-scale phosphorylation analysis. Our RPLC-MS/MS analysisunder low pH conditions is expected to find more novel mul-tiphosphorylation sites or motifs that are involved in serialphosphorylation events that cannot be easily discovered byRPLC-MS/MS analysis under normal pH conditions. Furtheranalysis combining phosphopeptide fractionation and enrichmentprocesses with low pH RPLC-MS/MS analysis is currentlyunderway.

ACKNOWLEDGMENTThis work has been funded by the 21C Frontier Functional

Proteomics Project from Korean Ministry of Science & Technol-ogy, FPR 05 A1-350 (Z.-Y.P.). The authors thank Shilpa Rameshat the Gwangju Institute of Science & Technology for experimentalhelp.

SUPPORTING INFORMATION AVAILABLEAdditional information as noted in text. This material is

available free of charge via the Internet at http://pubs.acs.org.

Received for review November 14, 2007. AcceptedJanuary 30, 2008.

AC7023393

Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 3015