www.elsevier.com/locate/ymeth

Methods 42 (2007) 289–297

Soluble nanopolymer-based phosphoproteomics for studyingprotein phosphatase

Minjie Guo, Jacob Galan, W. Andy Tao *

Departments of Biochemistry, Chemistry, and Medicinal Chemistry & Molecular Pharmacology, Purdue University, West Lafayette, IN 47907, USA

Accepted 24 February 2007

Abstract

Protein phosphorylation is a vital reversible post-translational modification that regulates protein–protein interactions, enzymaticactivity, subcellular localization, complex formation and protein stability. The emerging field of mass spectrometry-based proteomicsallows us to investigate phosphorylation and dephosphorylation on a global scale. In this review, we describe a new strategy basedon soluble nanopolymers that have been used to selectively isolate phosphopeptides for mass spectrometric analysis. Functionalized sol-uble nanopolymers provide a homogeneous environment and linear reaction kinetics for chemical derivatization to isolate phosphopep-tides with high specificity. Combined with phosphatase inhibitors and stable isotopic labeling, the approach has the capability ofquantitatively measuring phosphorylation and dephosphorylation on individual sites. We provide experimental details for the approachand describe some other complementary techniques that can be used.� 2007 Elsevier Inc. All rights reserved.

Keywords: Quantitative proteomics; Mass spectrometry; Nanopolymer; Phosphatase inhibitor; Database search; Isotopic labeling

1. Introduction

Reversible phosphorylation of proteins catalyzed bykinases and phosphatases plays a pivotal role in the regula-tion of important cellular functions such as growth, metab-olism, division, and signaling. Under normal conditions,the coordinated and proper balance of kinase and phos-phatase activity is tightly controlled. Any factor that obvi-ates the balance and results in either an excessive ordiminished substrate phosphorylation can cause a varietyof human diseases such as cancer, immune diseases, anddiabetes. It is estimated that there are hundreds of proteinkinases/phosphatases differing in their substrate specifici-ties, stoichiometry, cellular localization, and associationwith regulatory pathways. For instance, the complete geno-mic DNA sequence of Saccharomyces cerevisiae predicted123 different protein kinases and 40 protein phosphatases

1046-2023/$ - see front matter � 2007 Elsevier Inc. All rights reserved.

doi:10.1016/j.ymeth.2007.02.019

* Corresponding author. Fax: +1 765 494 7897.E-mail address: watao@purdue.edu (W.A. Tao).

that could be expressed. Thus, approximately 2% of theproteins expressed in yeast are involved in protein phos-phorylation reactions, and presumably a much larger num-ber of proteins are phosphorylated under specificphysiological conditions [1]. The completion of genome-sequencing projects has also advanced the developmentof proteomics techniques, providing new tools for the glo-bal analysis of protein phosphorylation. A comprehensivestudy of protein phosphorylation involves: (i) the identifi-cation of phosphoproteins and sites of phosphorylation;(ii) the identification of the kinase(s) and phosphatase(s)responsible for reversible phosphorylation and dephospho-rylation; (iii) the understanding of the biological conse-quence of the observed phosphorylation events.

Mass spectrometry (MS) has emerged as the method ofchoice for phosphoproteomics [2,3]. However, multiple fac-tors can complicate the analysis of protein phosphorylationin complex mixtures by MS-based methods. First, the stoi-chiometry of phosphorylation is frequently low, and onlyfractions of expressed proteins may be phosphorylated atany given time. Second, a specific phosphoprotein may

290 M. Guo et al. / Methods 42 (2007) 289–297

exist in several differentially phosphorylated forms and thestate of phosphorylation may be dynamically changingwith changing states of the cell. In particular, many signal-ing molecules are present at extremely low abundance.Third, with the exception of tyrosine phosphate, the phos-phate bonds range from labile to very labile. Therefore,specific precautions have to be taken to prevent the elimi-nation of these phosphates during sample preparation.Specifically, N-phosphorylation and acyl-phosphorylationis extremely acid labile, while O-phosphorylation is rela-tively base labile. Forth and finally, chemical lability ofthe phosphate group on amino acid residues induced bycollisions in the gas phase also has negative influence ondatabase-searching results for protein/peptideidentification.

While specific proteomic strategies for the investigationof protein phosphorylation keep evolving, there is a con-sensus on the general approach to study protein phosphor-ylation or dephosphorylation in complex samples. Fig. 1

Fig. 1. A generic protocol for phosphoproteomic analysis. See text fordetails.

schematically illustrates the proteomic strategy to studyprotein phosphatases. In the first step, cells are treated withphosphatase inhibitor before lysis. In the second step,phosphoproteins are enriched. This step is useful but notabsolutely necessary unless proteins of very low abundanceare being analyzed. In the third step, peptides are generatedby chemical or enzymatic cleavage from the protein mix-ture. In the forth step, phosphopeptides are selectively iso-lated to identify sites of phosphorylation from complexsamples. Finally, the phosphopeptides are analyzed byMS using specific data acquisition and database searchprotocols optimized for sequence and phosphorylation sitedetermination. For a parallel comparison, the control willbe processed similarly except that there is no treatment ofphosphatase inhibitor in the first step. Differential labelingwith stable isotopes can be performed on either the proteinor peptide stage. The step to efficiently isolate phosphopep-tides is critical and therefore it is the main focus in thisreview. We will briefly describe methods for other stepsas well.

2. Isolation of phosphopeptides

Since phosphorylation is frequently a low stoichiometryevent in a complex peptide mixture, nonphosphopeptidesare the dominant species. In order to identify the sites ofphosphorylation, it is essential to have an efficient strategyfor the isolation of actual phosphopeptides. Severalapproaches have been explored for the selective isolationof phosphopeptides, the most notable of these being eitheraffinity- or chemical derivatization-based. Antibodies havebeen used to enrich phosphopeptides but the efficiency ofthis affinity steps is relatively low [4,5]. Therefore, immobi-lized metal ion affinity chromatography (IMAC) has devel-oped into the leading method to affinity enrichphosphopeptides [5–11]. Steady steps have been taken toimprove its specificity [7,12]. However, the method appearsto be highly dependent on the type of resin and pH condi-tion for binding and elution, and prefers peptides with mul-tiple phosphorylation sites [13]. More recent applicationsemployed TiO2 as the stationary phase and the results indi-cated increased specificity for phosphopeptides [14]. Thenegative charge carried by phosphate groups on phospho-peptides has also been adopted to enrich phosphopeptidesusing cation exchange chromatography [15]. Because aphosphate group maintains a negative charge at acidicpH values, the net charge state of a tryptic phosphopeptideis generally only 1+. Therefore on the ion exchange chro-matography many phosphopeptides elute before non-phos-phopeptides which are multiply charged. It is obvious thatthis strategy does not have high specificity and cannotrecover all phosphopeptides either. However, the attractionof the method is that it can be fully automated and is par-ticularly suitable for large scale experiments. Chemicalmodifications of phosphate groups are more favorablefor the specificity but typically involve multiple derivatiza-tion steps, resulting in low yield for the analysis of phos-

M. Guo et al. / Methods 42 (2007) 289–297 291

phopeptides within complex mixtures [16,17]. It is thereforeapparent that a new method for phosphorylation analysis,in particular for quantitative measurements, fulfills anurgent need.

2.1. Rational for the use of soluble nanopolymers forproteomics

Quantitative approaches commonly apply MS with var-ious stable isotope-labeling strategies. The isotope-codedaffinity tagging (ICAT) method is perhaps the best charac-terized method for quantitative proteomics that combinesstable isotope labeling with affinity purification [18]. Theadaptation of solid-phase capture and release process isan important improvement for isotope tagging and selec-tive peptide isolation [19]. This procedure is simple and effi-cient by eliminating extra purification steps and has thepotential for automation and high throughput experi-ments. However, the most notable liability of solid-phaseapproach is the heterogeneous reaction conditions, whichcan exhibit several of the following problems: nonlinearkinetic behavior, unequal distribution and/or access tothe chemical reaction, and solvation problems. As proteo-mics handles complex mixture of proteins in low-abun-dance, the heterogeneous nature of solid-phase reactionpresents a serious issue for sample recovery.

To address this issue, we devised a new strategy basedon soluble polymers for quantitative proteomics, termedsoluble polymer-based isotopic labeling (SoPIL) [20]. Thecornerstone of SoPIL is a soluble, globular nanopolymer(e.g., dendrimer) which was functionalized with reactivegroups for site-specific, stable isotopic labeling of a subsetof proteome, and with an optional ‘handle’ which can facil-itate the isolation of tagged peptides through a highly effi-cient bio-conjugation (Fig. 2). Such design was based onthe assumption that slow reactions with limited amountof samples would be carried out in the solution phase formaximum yield and then the attached samples could be iso-lated on a solid-phase through a highly efficient reactionbetween functional groups on the soluble polymer and onthe solid-phase. High concentration ratio of the reactivegroup to the ‘handle’ group facilitates the completion ofthe solution reaction to capture samples on the solublepolymer, while eliminating the extra step to remove excessreagents.

Fig. 2. Modular composition of proteomics strategy which consists of twostep reactions: a homogeneous reaction between proteins/peptides and theSoPIL reagent and a heterogeneous reaction between the solid-phase andthe SoPIL reagent.

As a first step, thiol-reactive SoPIL reagents were syn-thesized to quantitatively analyze Cys-containing peptides.We demonstrated that the capture and release of cysteine-containing peptides using the SoPIL reagent were specificand more efficient than the one-step solid-phase isotopiclabeling reagent. In the experiments analyzing proteins insnake venom the SoPIL reagents also efficiently labeledmultiple close-spaced cysteine residues, a feature thesolid-phase method cannot achieve due to steric hindrance.Furthermore, soluble nanopolymers show the potential todirectly tag and label proteins in living cells and in vivo [20].

2.2. Isolation of phosphopeptides using soluble nanopolymers

Reagents based on soluble nanopolymers to isolatephosphopeptides are based on the derivatization of phos-phate to phosphoramidate group [21], a reaction com-monly-used for the immobilization of oligo-nucleotides[22]. Two types of soluble nanopolymers, unmodified pol-yamindoamine dendrimer generation 5 (PAMAM G5)and bi-functionalized dendrimer, have been employed,respectively. A simple scheme comparing the protocolsusing these two reagents is shown in Fig. 3. The first proto-col uses underivatized PAMAM G5 and molecular weight

Fig. 3. Schematic illustration of two methods based on dendrimercoupling to isolate phosphopeptides.

292 M. Guo et al. / Methods 42 (2007) 289–297

cut off device to isolate phosphopeptides. In the secondprotocol when the bifunctionalized dendrimer is used,phosphopeptides are attached to polymer and then purifiedon the solid-phase in the following step.

2.2.1. Isolation of phosphopeptides using unmodified

polyamidoamine dendrimerA protein mixture (cell extract or immuno-purified com-

plex) was digested to generate peptides. Peptides includingboth phosphopeptides and nonphosphopeptides were thenmethylated in the next step. The resulting methylated phos-phopeptides were captured directly on the dendrimer via acarbodiimide-activated reaction between amine and phos-phate groups. Phosphopeptides covalently bound to den-drimer (the molecular weight of PAMAM G5 is 28 kDa)were readily isolated from unbound non-phosphopeptidesusing size selective methods such as a simple membranebased filter device [21]. Phosphopeptides were detachedfrom the dendrimer through acid hydrolysis of the phos-phoramidate bonds and isolated using the same membranebased filter device. The advantage of this strategy is thatthe utility of soluble polyamines such as dendrimers allowsfor a homogeneous reaction with adequate amino groupsin the solution. Large reagent excesses can still be used todrive reactions to completion. Noncovalently associatedmolecules and excess reagents can be removed by size selec-tive methods such as size exclusion chromatography (SEC)or a molecular weight cut off filter device. Quantitativemeasurement can be conveniently achieved with thismethod when peptides are d0- or d3-methylated.

2.2.1.1. Protein digestion and peptide methylation. Proteinmixtures were dissolved in 20 mM ammonium bicarbonate(pH 8.0) containing 0.1% RapiGest (Waters Co.), andheated at 95 �C for 5 min. Samples were reduced with5 mM dithiothreitol (DTT) for 30 min at 37 �C, and thenalkylated with 15 mM iodoacetamide in the darkness(30 min, room temperature). Proteins were digested withtrypsin (proteins:trypsin ratio at 50:1) overnight at 37 �C.The resulting peptides were desalted by MCX column(Waters Co.), lyophilized and reconstituted in 75 lL ofmethanolic HCl which was prepared by adding 100 lL ofacetyl chloride to 500 lL of anhydrous methanol-d0 or d4

(Cambridge Isotope Laboratories). The methyl esterifica-tion was allowed to proceed at 12 �C for 90 min. Solventwas removed in a Speedvac, and peptides were reconsti-tuted in appropriate solvent for the next step reaction.

2.2.1.2. Purification of phosphopeptides using PAMAM G5.

Peptide methyl esters were dissolved in 40 lL reaction solu-tion containing 50 mM EDC, 100 mM imidazole, 200 mMMES (pH 5.5), and 9 mg of PAMAM dendrimer Genera-tion 5 (final concentration of amine group is 1M; dendri-mer was supplied as a 10 wt% solution in methanol(Sigma–Aldrich) and methanol was removed in vacuo priorto use). The reaction was allowed to stand at room temper-ature with vigorous shaking for 15 h and the solution was

transferred to a Biomax filter device (5 kDa cutoff; Milli-pore Co.). Dendrimer-bound phosphopeptides werewashed with 500 lL of 2 M NaCl in 40% methanol andwater a few times and the filtrates were discarded to removenonspecifically bound non-phosphopeptides. Finally, 10%TFA was added to incubate at room temperature for30 min to recover phosphopeptides. The polymer waswashed twice with 30% methanol and the filtrates werecombined and dried for LC–MS analysis.

2.2.2. Isolation of phosphopeptides using bifunctionalized

polyamidoamine dendrimer

In the second approach, polyamidoamine dendrimerwas partially derivatized with a ‘handle’ which facilitatesthe isolation of attached phosphopeptides on the solid-phase on the following step through a highly efficient bio-conjugation. We sought to engineer a pair of biologicallyinert coupling partners on dendrimers and on the solid-phase, respectively, that would react at practical rates atsubmillimolar concentrations and at high yield. Examplesinclude the Staudinger ligation of azides with triaryl phos-phines, the formation of a Schiff’s base between ketone/aldehyde-hydrazine groups, and copper(I) catalyzedazide–alkyne cycloaddition (the click chemistry). We haveexamined several types of bio-orthogonal coupling reac-tions. The click chemistry [23,24], a coupling between ter-minal alkyne group and azide group, was found to begenerally biocompatible and highly efficient (the reactioncan be completed in minutes). Hence, the dendrimer sur-face was also functionalized with alkyne group as the ‘han-dle’ (Fig. 2).

The N-(3-dimethylaminopropyl)-N 0-ethylcarbodiimide(EDC)-catalyzed reaction between the phosphate andamine groups in aqueous solution is a rather slow reaction.High concentration of amine or amino group and over-night reaction is therefore required. Potential side reactionscan occur during the overnight reaction, most notably thehydrolysis of peptide methyl esters. In the second protocol,N,N-diisopropylcarbodimide (DIPCI), instead of EDC,was used to catalyze the coupling reaction between thephosphate and amine groups. The reaction was allowedto proceed in DMSO for less than 5 h. The solid-phaseresin functionalized with azide groups is then added to cap-ture dendrimers through click chemistry, and the resin isextensively washed to remove unbound peptides and con-taminants. The beads are then treated with acid, resultingin the cleavage of phosphoramidate bond. Finally, therecovered phosphopeptides are analyzed by lLC–MS/MSto determine the peptide sequence and correspondingphosphoprotein.

2.2.2.1. Synthesis of PAMAM G4 partially functionalized

with terminal alkyne groups (Scheme 1). A mixture of1.6 mg 4-pentynoic acid (16 lmol), 5 lL DIPCI (32 lmol),and 2.16 mg N-hydroxybenzotriazole (HOBt) was dis-solved in 70 lL DMSO for activation for 30 min at roomtemperature. Then the activated pentynoic acid solution

Scheme 1.

M. Guo et al. / Methods 42 (2007) 289–297 293

was added to 2 lmol PAMAM G4 (28 mg) dissolved in400 lL of DMSO. The reaction was allowed to proceedat room temperature for 12 h. The reaction was quenchedwith 3 mL of water and the product was purified via exten-sive dialysis to remove small molecule impurities. The finalproduct was lyophilized and redissolved in DMSO forfuture application.

2.2.2.2. Synthesis of azide beads. Two hundred milligramsof aminopropyl controlled pore glass (CPG) beads (NH2:400 lmol/g) were mixed with 80 mg of succinic anhydridein 400 lL DMF and 200 lL pyridine. After incubationovernight at room temperature, the beads were washed suc-cessively with 0.6 mL of DMF three times, 0.6 mL of 1NHCl twice, 0.6 mL of water three times, and 0.6 mL ofMeOH three times, followed by complete dryness in speed-vac. 109 mg of 1-amino-11-azido-3,6,9-trioxaundecane(x-amino(PEG)4azide, prepared as reported in the litera-ture [25]) (500 lmol) in 500 lL DMF was added to thebeads followed by the addition of 63 mg HOBt (500 lmol)and 80 lL DIPCI (500 lmol) in 200 lL DMF. After over-night incubation at room temperature, the beads wereextensively washed with DMF and CH2Cl2. Beads weredried in speedvac and ready for use.

2.2.2.3. Purification of phosphopeptides using bifunctional-

ized dendrimer. Peptide methyl esters were incubated in100 lL of reaction solution containing 10 nmol of G4-PAMAM-(alkyne)8, 0.5 mg of imidazole, 0.5 mg of HOBt,and 20 lL DIPCI in DMSO (pH 4.5) at 37 �C for 5 h. Itwas followed by the addition of 230 lL H2O, 20 mMtris(triazolyl)amine 50 lL and 50 mM CuSO4 20 lL. Finalvolume was 400 lL and concentration of TCEP, CuSO4,tris(triazolyl)amine, respectively, was 2.5 mM, 2.5 mM,0.25 mM. The whole mixture was added to 10 mg azidebeads and incubated at room temperature for 1 h. Thebeads was washed extensively by methanol and water,and then incubated with methanol/water/trifluoroaceticacid (1/1/1) 100 lL for 1 h. The cleaved peptides were col-lected and the beads were washed once with 0.1% TFA.The elutes were combined, dried by speedvac and readyfor MS analysis.

3. MS data acquisition, phosphopeptides identification and

determination of sites of phosphorylation

Peptides are commonly identified via the generation oftandem mass (MS/MS) spectra of individual peptides andsearching the fragment ion species against protein sequencedatabases [26]. For phosphopeptide identification, variablemodifications on the amino acid residues Ser, Thr, and Tyr

(+80) are applied prior to database searching. This stepgenerates a database in which proteins with possible phos-phorylation on Ser, Thr, and Tyr residues are included. Aphosphopeptide is identified when a match is foundbetween a MS/MS spectrum and an in silico spectrum ofa phosphopeptide from the database. In addition, the anal-ysis of phosphopeptides by MS can be facilitated by spe-cific data acquisition methods. These mass spectrometricmethods take advantage of chemical lability of the phos-phoester bonds in phospho-serine, -threonine and, in lessdegree, -tyrosine. The phosphoester bonds can be inducedto fragment in a collision cell or the ion source of an MSinstrument, resulting in a loss of phosphoric acid fromthe peptide. This characteristic fragmentation pattern canbe used to detect phosphopeptides among non-phosphory-lated peptides.

However, the predominant loss of phosphoric acid inthe collision cell of the mass spectrometer also creates aproblem for sequencing serine/threonine phosphate-containing peptides because the resulting spectra that aredominated by the eliminated phosphate are the dephospho-rylated peptide [7]. The lack of informative fragmentationat the peptide backbone severely reduces the ability ofdatabase searching algorithms to unambiguously identifythe phosphopeptide. Furthermore, when a phosphopeptideis identified, it is often not possible to assign the site ofphosphorylation to a particular Ser or Thr residue.

To obtain conclusive sequence information of S/Tphosphorylated peptide, the dominant fragment ionresulting from the neutral loss can be subjected to anMS/MS/MS (MS3) experiment. The systematic use of thisstrategy in an ion trap mass spectrometer has resulted inthe identification of a number of phosphopeptides whichwere not identified by MS2 experiments [15]. The disad-vantage of MS3 experiments is the significant loss of trap-ping ions over several stages of MS. As a result, MS3

experiments were mainly successful for phosphopeptideswith relatively high signal intensity. The introduction ofMS instruments with high trapping capacity such as quad-rupole linear ion traps facilitates MS3 experiments onphosphopeptides. Correspondingly, a specific data acquisi-tion method, named data-dependant MS3-NL scan, wasdeveloped. Using the MS3-NL scan, an MS3 spectrum willbe automatically collected by isolating and fragmentingthe neutral loss fragment ion from the MS/MS spectrum,if a significant loss of phosphoric acid upon fragmentationis one of the most intense ions detected in the MS2

spectrum.

4. Quantitative phosphoproteomics

Quantitative determination of phosphorylation basedon MS-based proteomics is commonly achieved by com-paring the intensity of peptides of the same sequence butwith differential stable isotope composition [27]. The label-ing step can be made in vitro by chemical derivatization [18]or enzymatic reaction [28], or in vivo by the incorporation

294 M. Guo et al. / Methods 42 (2007) 289–297

of isotopes by metabolic labeling [29]. In theory, quantita-tive phosphoproteomics can be achieved by the combina-tion of any existing stable isotope labeling method withphosphoproteins/phosphopeptide enrichment. However,special caution has to be taken for quantitative phosphor-ylation analysis. Since only a small fraction of proteins arephosphorylated, quantitative information at the level ofproteins does not necessarily reflect the phosphorylationstatus. Instead, quantitative measurement has to be madeat the level of the phosphopeptide. In addition, stable iso-tope labeling, in particular chemical derivatization, usuallyintroduces extra steps into the sample preparation whichcould have negative effect on the analysis of already lowabundant phosphoproteins. Finally, many enrichmentmethods in phosphoproteome analysis are affinity basedand therefore may not be quantitative.

A few years ago Aebersold group introduced an in vitro

labeling method to quantitate proteins by labeling cysteineresidues, a strategy designated isotope-coded affinity tag-ging (ICAT) [18]. The same commercially available(ICAT�) reagents were used to introduce a biotin tag intophosphoserine and phosphothreonine residues by b-elimi-nation and Michael addition, enabling enrichment andsimultaneous quantitation of phosphoproteins [30].Several global labeling methods, e.g., N-terminal derivati-zation and methylation of carboxylic groups in peptides,combined with the enrichment of phosphopeptides, havealso been reported for quantitative phosphorylation[12,21].

All proteins in the whole cell can be labeled in vivo bygrowing them in 15N-labeled media without any furthermanipulation [31]. This can also be used to quantify theextent of phosphorylation. An alternative strategy to labelproteins in vivo, designated stable isotope labeling byamino acid in cell culture (SILAC) [32–34] which usesamino acids containing a stable isotope, has been increas-ingly used. For example, a human cell line was grown inmedia containing normal or 13C6-containing lysine andarginine (this increases the mass of every lysine- or argi-nine-containing peptide by 6 Da) [33]. The resulting massspectrum clearly showed the differences in intensitybetween two triply charged peptide ions separated by6 Da. The advantage of using stable-isotope-containingamino acids over media containing 15N is that it can beused in cases where the sequence is not known. Neverthe-less the in vivo labeling strategies are mostly restricted tocells which can be grown in culture.

5. Quantitative phosphorylation analysis with protein

tyrosine phosphatase (PTP) inhibitor

Dynamic serine, threonine, and tyrosine phosphoryla-tion and dephosphorylation of signaling intermediates iscritical for the regulation of the T cell signaling network[35]. The phosphorylation status of signaling moleculesmodulates dynamic protein-protein interactions allowingfor the integration of extracellular signals into various T

cell responses. In order to fully understand T cell responsesit is necessary to pinpoint temporal phosphorylation anddephosphorylation events in the course of the response ofT cells to various stimuli. We measured phosphorylationchanges with respect to the treatment with a strong tyrosinephosphatase inhibitor, pervanadate. Jurkat T cells weretreated with pervanadate for 2 and 10 min, respectively,before lysis. In order to detect tyrosine phosphorylatedpeptides and tyrosine phosphorylation sites we coupledan immuno-affinity step to our dendrimer-based phospho-peptide enrichment method. Anti-phosphotyrosine anti-bodies were employed to first immuno-purify tyrosinephosphoproteins and reduce the background of serineand threonine phosphorylation before chemical enrichmentof trypsinized phosphopeptides. Methyl-d0/d3 esters oftryptic peptides were combined and phosphopeptides wereenriched using the dendrimer-based derivatization methodand followed by micro-capillary reverse-phase LC–MS/MSanalysis.

This tandem purification method allowed for the identi-fication and quantification of 52 tyrosine phosphoproteinswith 75 tyrosine phosphorylation sites, as well as a numberof serine and threonine sites in these proteins. A partial listof 24 proteins is shown in Table 1.

Our two step approach led to the MS detection of allknown tyrosine phosphorylated residues in the immunore-ceptor tyrosine-based activation motifs (ITAM)[36,37] ofthe TCR’s CD3d, CD3e, CD3c, and CD3f chains uponpervandate treatment in a single experiment. The identifiedproteins also indicate that by using our strategy a range ofwell and less well characterized signaling molecules are nowdetectable in an MS based approach. These observationsopen up new avenues of investigation, for instance theMS based study of signaling pathways upon cell perturba-tion with selective drugs.

The quantitative nature of the approach is apparentfrom the ability to obtain accurate measurements of smallchanges in the extent of phosphorylation. Fig. 4 illustratesthe identification of a phosphopeptide and the quantifica-tion of its relative abundance in two cell states, with thecase of a CD3f drived doubly phosphorylated peptide,SADAPAY*QQGQNQLY*NELNLGR. Tandem massspectrometry enables the unambiguous assignment of thepeptide sequence and its tyrosine phosphorylation site.Quantitative information was obtained by reconstructingindividual ion chromatograms of these two species andintegrating the contour of the two respective peaks usingthe ASAPRatio program [38]. In this case, the ratio(light:heavy) was 2.1, indicating that after a 10-min treat-ment with pervanadate, phosphorylation on two tyrosineresidues on this peptide was increased twofold comparedto the 2-min treatment.

Overall, tyrosine phosphorylation was increased uponlonger treatment of pervanadate, as indicated by manyabundance ratios larger than 1.0 of phosphopeptidesisolated from T cells treated 10 min and 2 min withpervanadate. For some phosphoproteins, the extent of

Table 1Selected phosphorylated peptides from immuno-purified Jurkat lysates

IPI number Protein name Peptide sequence Ratioa Ref.

IPI00000861 LASP1 H.HIPTSAPVY*QQPQQQPVAQSYGGYK.E 1.62 ± 0.34 [40]IPI00003479 ERK1 R.VADPDHDHTGFLTEY*VATR.W 0.72 ± 0.10 [41]IPI00004407 SIT K.Y*SEVVLDSEPK.S 1.09 ± 0.16 [6]IPI00004407 SIT R.LS*QDPEPDQQDPTLGGPAR.A 0.74 ± 0.06IPI00004407 SIT R.SGES*VEEVPLYGNLHYLQT*GR.L 2.61 ± 0.77IPI00015287 DOC1 K.SHNSALY*SQVQK-.S 0.36 ± 0.10 [10]IPI00015287 DOC1 R.VKEEGY*ELPYNPATDDY*AVPPPR.S 1.01 ± 0.05 [42]IPI00015287 DOC1 K.EDPIY*DEPEGLAPVPPQGLY*DLPR.E 1.25 ± 0.31 [42]IPI00015287 DOC1 R.ADS*HEGEVAEGK.L 1.20 ± 0.16IPI00022602 DOC2 R.GQEGEY*AVPFDAVAR-.S 0.47 ± 0.04 [10]IPI00022934 CD3d R.DDAQY*SHLGGNWAR.N 1.99 ± 0.36 [10]IPI00022934 CD3d R.DRDDAQY*SHLGGNWAR-.N 1.55 ± 0.09IPI00022934 CD3d R.NDQVY*QPLR.D 1.83 ± 0.38 [10]IPI00012923 CD3e R.DLY*SGLNQR.R 1.74 ± 0.28 [43]IPI00012923 CD3e K.ERPPPVPNPDY*EPIR.K 1.56 ± 0.13IPI00016020 CD3c R.EDDQY*SHLQGNQLR.R 1.23 ± 0.12IPI00016020 CD3c K.QTLLPNDQLY*QPLK.D 1.70 ± 0.23IPI00218634 CD3f K.DTY*DALHMQALPPR.- 0.92 ± 0.15 [44]IPI00218634 CD3f K.GHDGLY*QGLSTATK.D 1.05 ± 0.32 [45]IPI00218634 CD3f R.KNPQEGLY*NELQK-.D 0.99 ± 0.11 [45]IPI00218634 CD3f K.MAEAY*SEIGMK.G 1.40 ± 0.08 [45]IPI00218634 CD3f K.NPQEGLY*NELQK.D 1.08 ± 0.12 [45]IPI00218634 CD3f R.REEY*DVLDK.R 1.24 ± 0.29 [45]IPI00218634 CD3f R.SADAPAYQQGQNQLY*NELNLGR-.R 1.59 ± 0.15 [6]IPI00218634 CD3f R.SADAPAY*QQGQNQLY*NELNLGR.R 2.10 ± 0.23 [6]IPI00021076 Splice isoform of Plakophilin 4 R.SAVSPDLHITPIY*EGR.T 1.00 ± 0.10IPI00021076 Splice isoform of Plakophilin 4 R.SSY*ASQHSQLGQDLR.S 4.13 ± 1.14IPI00022339 CD28 R.LLHSDY*MNMTPR.R 2.53 ± 0.41 [46]IPI00022339 CD28 K.HYQPY*APPR.D 1.08 ± 0.81 [46]IPI00023704 LPP R.NDSDPTY*GQQGHPNTWK.R 3.02 ± 0.70IPI00023704 LPP R.YYEGYY*AAGPGYGGR.N 1.85 ± 0.38 [6]IPI00026156 HS1 K.SAVGHEY*VAEVEK.H 1.63 ± 0.19IPI00026156 HS1 R.S*PEAPQPVIAMEEPAVPAPLPK.K 0.50 ± 0.21IPI00030851 Sec24B N.TVNQQPGAQQLY*SR.G 1.12 ± 0.21IPI00030851 Sec24B R.DSRPLS*PILHIVK.D 1.23 ± 0.13IPI00032003 Emerin R.TY*GEPESAGPSR.A 4.72 ± 0.60IPI00045486 Erbin R.AQIPEGDY*LSYR.E 1.69 ± 0.16 [6]IPI00045486 Erbin K.DFNLPEY*DLNVEER.L 4.07 ± 0.63IPI00045486 Erbin R.TY*SIDGPNASR.P >20IPI00045486 Erbin N.Y*SQIHHPPQASVAR.H 1.66 ± 0.12IPI00215637 DDX3 K.DKDAY*SSFGSR.S 0.74 ± 0.08 [6]IPI00101968 Hypothetical protein R.FQDVGPQAPVGSVY*QK.T 2.42 ± 0.19IPI00233255 SHP-2 R.VY*ENVGLMQQQK.H 0.68 ± 0.20 [47]IPI00297169 SLP-76 N.SNSMY*IDRPPSGK.T 0.66 ± 0.15IPI00329789 ZAP-70 K.ALGADDSY*Y*TAR.S 1.47 ± 0.27 [48]IPI00329789 ZAP-70 R.IDTLNSDGY*TPEPAR-.I 1.22 ± 0.12 [48]IPI00329789 ZAP-70 R.PMPMDTSVY*ESPY*SDPEELKDK-.K 1.68 ± 0.14 [48,49]IPI00003406 Drebrin R.SPS*DSSTASTPVAEQIER.A 1.74 ± 0.11IPI00012442 G3BP-1 K.SSS*PAPADIAQTVQEDLR-.T 0.77 ± 0.07 [15]IPI00015029 Telomerase-binding protein p23 K.DWEDDS*DEDMSNFDR.F 1.05 ± 0.06 [50]

From Ref. [21].a Ratio of intensities of phosphopeptides after 10 and 2 min treatment with pervanadate.

M. Guo et al. / Methods 42 (2007) 289–297 295

phosphorylation is different on individual tyrosine residuesin the same protein (Table 1). Some phosphorylation sitescan be fully induced within 2 min of stimulation (indicatedby the ratio of phosphopeptides close to 1.0), althoughoverall phosphorylation of the protein is increased overextended treatment. Only a few proteins have shown thedecrease of tyrosine phosphorylation over extended stimu-lation (Table 1). In contrast to tyrosine phosphorylation

where longer treatment of pervanadate increases phos-phorylation, many serine/threonine phosphorylation sitesdecrease over the same time of treatment with pervana-date, although the biological meaning is unclear. Forexample, haematopoietic lineage cell-specific protein,HS1 protein, a known tyrosine phosphoprotein, hasrecently been described to undergo Ser/Thr phosphoryla-tion as well [39].

Fig. 4. Quantitative phosphopeptides analysis. (a) Tandem mass spectrumof the CD3f drived peptide SADAPAY*QQGQNQLY*NELNLGR. Itscharacteristic peptide bond fragment ions, type b and type y ions, arelabeled. (b) Reconstructed ion chromatogram of the precursor ion (m/z778.3) and its heavy version (m/z 782.8) using ASAPRatio program. Theprogram draws a smoothed chromatogram based on the ion signal andthen calculates the ratio of the peak areas. From Ref. [21].

296 M. Guo et al. / Methods 42 (2007) 289–297

6. Conclusion

Global phosphorylation analysis is one of the mostimportant research fields in proteomics. The existing proto-cols are immature but evolving rapidly. Methods based onsoluble nanopolymers represent one important step towarda general and reproducible procedure for phosphoproteo-mics. Effective application of phosphoproteomics willaccelerate the functional characterization of phosphatases,thereby facilitating our understanding of phosphatases incell signaling and in diseases.

Acknowledgments

This work was supported in part by Purdue University,the National Scientific Foundation CAREER award(CHE-0645020) and American Society for Mass Spectrom-etry (ASMS).

References

[1] T. Hunter, Cell 100 (2000) 113–127.[2] R. Aebersold, D.R. Goodlett, Chem. Rev. 101 (2001) 269–295.

[3] M. Mann, S.E. Ong, M. Gronborg, H. Steen, O.N. Jensen, A.Pandey, Trends Biotechnol. 20 (2002) 261–268.

[4] J. Rush, A. Moritz, K.A. Lee, A. Guo, V.L. Goss, E.J. Spek, H.Zhang, X.M. Zha, R.D. Polakiewicz, M.J. Comb, Nat. Biotechnol. 23(2005) 94–101.

[5] Y. Zhang, A. Wolf-Yadlin, P.L. Ross, D.J. Pappin, J. Rush, D.A.Lauffenburger, F.M. White, Mol. Cell. Proteomics 4 (2005) 1240–1250.

[6] L.M. Brill, A.R. Salomon, S.B. Ficarro, M. Mukherji, M. Stettler-Gill, E.C. Peters, Anal. Chem. 76 (2004) 2763–2772.

[7] S.B. Ficarro, M.L. McCleland, P.T. Stukenberg, D.J. Burke, M.M.Ross, J. Shabanowitz, D.F. Hunt, F.M. White, Nat. Biotechnol. 20(2002) 301–305.

[8] M.C. Posewitz, P. Tempst, Anal. Chem. 71 (1999) 2883–2892.[9] A. Pandey, J.S. Andersen, M. Mann, Sci. STKE 2000 (2000)

PL1.[10] A.R. Salomon, S.B. Ficarro, L.M. Brill, A. Brinker, Q.T. Phung, C.

Ericson, K. Sauer, A. Brock, D.M. Horn, P.G. Schultz, E.C. Peters,Proc. Natl. Acad. Sci. USA 100 (2003) 443–448.

[11] H. Shu, S. Chen, Q. Bi, M. Mumby, D.L. Brekken, Mol. Cell.Proteomics 3 (2004) 279–286.

[12] L. Riggs, E.H. Seeley, F.E. Regnier, J. Chromatogr. B. Analyt.Technol. Biomed. Life Sci. 817 (2005) 89–96.

[13] C.E. Haydon, P.A. Eyers, L.D. Aveline-Wolf, K.A. Resing, J.L.Maller, N.G. Ahn, Mol. Cell. Proteomics 2 (2003) 1055–1067.

[14] M.R. Larsen, T.E. Thingholm, O.N. Jensen, P. Roepstorff, T.J.Jorgensen, Mol. Cell. Proteomics 4 (2005) 873–886.

[15] S.A. Beausoleil, M. Jedrychowski, D. Schwartz, J.E. Elias, J. Villen, J.Li, M.A. Cohn, L.C. Cantley, S.P. Gygi, Proc. Natl. Acad. Sci. USA101 (2004) 12130–12135.

[16] H. Zhou, J.D. Watts, R. Aebersold, Nat. Biotechnol. 19 (2001) 375–378.

[17] Y. Oda, T. Nagasu, B.T. Chait, Nat. Biotechnol. 19 (2001) 379–382.[18] S.P. Gygi, B. Rist, S.A. Gerber, F. Turecek, M.H. Gelb, R.

Aebersold, Nat. Biotechnol. 17 (1999) 994–999.[19] H. Zhou, J.A. Ranish, J.D. Watts, R. Aebersold, Nat. Biotechnol. 20

(2002) 512–515.[20] M. Guo, J. Galan, W.A. Tao, Chem. Commun. (2007) 1251–1253.[21] W.A. Tao, B. Wollscheid, R. O’Brien, J. Eng, X. Li, B. Bodenmiller,

J. Watts, L. Hood, R. Aebersold, Nat. Methods 2 (2005) 591–598.[22] B. Chu, G.M. Wahl, L.E. Orgel, Nucleic Acids Res. 11 (1983) 6513–

6529.[23] H.C. Kolb, M.G. Finn, K.B. Sharpless, Angew. Chem. Int. Ed. Engl.

40 (2001) 2004–2021.[24] D.D. Diaz, K. Rajagopal, E. Strable, J. Schneider, M.G. Finn, J. Am.

Chem. Soc. 128 (2006) 6056–6057.[25] A.W. Schwabacher, J.W. Lane, M.W. Schiesher, K.M. Leigh, C.W.

Johnson, J. Org. Chem. 63 (1998) 1727–1729.[26] E.A. Kapp, F. Schutz, L.M. Connolly, J.A. Chakel, J.E. Meza, C.A.

Miller, D. Fenyo, J.K. Eng, J.N. Adkins, G.S. Omenn, R.J. Simpson,Proteomics 5 (2005) 3475–3490.

[27] W.A. Tao, R. Aebersold, Curr. Opin. Biotechnol. 14 (2003) 110–118.[28] X. Yao, A. Freas, J. Ramirez, P.A. Demirev, C. Fenselau, Anal.

Chem. 73 (2001) 2836–2842.[29] S.E. Ong, B. Blagoev, I. Kratchmarova, D.B. Kristensen, H. Steen, A.

Pandey, M. Mann, Mol. Cell. Proteomics 1 (2002) 376–386.[30] M.B. Goshe, T.D. Veenstra, E.A. Panisko, T.P. Conrads, N.H.

Angell, R.D. Smith, Anal. Chem. 74 (2002) 607–616.[31] T.P. Conrads, K. Alving, T.D. Veenstra, M.E. Belov, G.A. Anderson,

D.J. Anderson, M.S. Lipton, L. Pasa-Tolic, H.R. Udseth, W.B.Chrisler, B.D. Thrall, R.D. Smith, Anal. Chem. 73 (2001) 2132–2139.

[32] N. Ibarrola, D.E. Kalume, M. Gronborg, A. Iwahori, A. Pandey,Anal. Chem. 75 (2003) 6043–6049.

[33] A. Gruhler, J.V. Olsen, S. Mohammed, P. Mortensen, N.J. Faerg-eman, M. Mann, O.N. Jensen, Mol. Cell. Proteomics 4 (2005) 310–327.

[34] Y. Oda, K. Huang, F.R. Cross, D. Cowburn, B.T. Chait, Proc. Natl.Acad. Sci. USA 96 (1999) 6591–6596.

M. Guo et al. / Methods 42 (2007) 289–297 297

[35] A. Marie-Cardine, B. Schraven, Cell. Signal. 11 (1999) 705–712.[36] A. Weiss, Cell 73 (1993) 209–212.[37] P.E. Love, E.W. Shores, Immunity 12 (2000) 591–597.[38] X. Li, H. Zhang, J.A. Ranish, R. Aebersold, Anal. Chem. 75 (2003)

6648–6657.[39] M. Ruzzene, D. Penzo, L.A. Pinna, Biochem. J. 364 (2002) 41–47.[40] Y.H. Lin, Z.Y. Park, D. Lin, A.A. Brahmbhatt, M.C. Rio, J.R. Yates

3rd, R.L. Klemke, J. Cell Biol. 165 (2004) 421–432.[41] E.R. Butch, K.L. Guan, J. Biol. Chem. 271 (1996) 4230–4235.[42] N. Kashige, N. Carpino, R. Kobayashi, Proc. Natl. Acad. Sci. USA

97 (2000) 2093–2098.[43] I. de Aos, M.H. Metzger, M. Exley, C.E. Dahl, S. Misra, D. Zheng,

L. Varticovski, C. Terhorst, J. Sancho, J. Biol. Chem. 272 (1997)25310–25318.

[44] H. Zheng, P. Hu, D.F. Quinn, Y.K. Wang, Mol. Cell. Proteomics(2005).

[45] E.N. Kersh, A.S. Shaw, P.M. Allen, Science 281 (1998) 572–575.[46] J.M. Teng, P.D. King, A. Sadra, X. Liu, A. Han, A. Selvakumar, A.

August, B. Dupont, Tissue Antigens 48 (1996) 255–264.[47] M. MacGillivray, M.T. Herrera-Abreu, C.W. Chow, C. Shek, Q.

Wang, E. Vachon, G.S. Feng, K.A. Siminovitch, C.A. McCulloch,G.P. Downey, J. Biol. Chem. 278 (2003) 27190–27198.

[48] J.D. Watts, M. Affolter, D.L. Krebs, R.L. Wange, L.E. Samelson, R.Aebersold, J. Biol. Chem. 269 (1994) 29520–29529.

[49] M. Pelosi, V. Di Bartolo, V. Mounier, D. Mege, J.M. Pascussi, E.Dufour, A. Blondel, O. Acuto, J. Biol. Chem. 274 (1999) 14229–14237.

[50] T. Kobayashi, Y. Nakatani, T. Tanioka, M. Tsujimoto, S. Nakajo, K.Nakaya, M. Murakami, I. Kudo, Biochem. J. 381 (2004) 59–69.