Selective Enrichment of Cysteine-Containing Phosphopeptides forSubphosphoproteome AnalysisMingming Dong,†,‡ Yangyang Bian,†,‡ Jing Dong,† Keyun Wang,†,‡ Zheyi Liu,†,‡ Hongqiang Qin,†

Mingliang Ye,*,† and Hanfa Zou*,†

†Key Laboratory of Separation Sciences for Analytical Chemistry, National Chromatographic Research and Analysis Center, DalianInstitute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China‡University of Chinese Academy of Sciences, Beijing 100049, China

*S Supporting Information

ABSTRACT: Among the natural amino acids, cysteine isunique since it can form a disulfide bond through oxidationand reduction of sulfhydryl and thus plays a pervasive role inmodulation of proteins activities and structures. Crosstalkbetween phosphorylation and other post-translational mod-ifications has become a recurrent theme in cell signalingregulation. However, the crosstalk between the phosphor-ylation and the formation and reductive cleavage of disulfidebond has not been investigated so far. To facilitate the study ofthis crosstalk, it is important to explore the subset ofphosphoproteome where phosphorylations are occurred nearto cysteine in the protein sequences. In this study, wedeveloped a straightforward sequential enrichment method bycombining the thiol affinity chromatography with the immobilized titanium ion affinity chromatography to selectively enrichcysteine-containing phosphopeptides. The high specificity and high sensitivity of this method were demonstrated by analyzingthe samples of Jurkat cells. This “divide and conquer” strategy by specific analysis of a subphosphoproteome enablesidentification of more low abundant phosphosites than the conventional global phosphoproteome approach. Interestingly, aminoacid residues surrounding the identified phosphosites were enriched with buried residues (L, V, A, C) while depleted withexposed residues (D, E, R, K). Also, the phosphosites identified by this approach showed a dramatic decrease in locating indisorder regions compared to that identified by conventional global phosphoproteome. Further analysis showed that moreproline directed kinases and fewer acidophilic kinases were responsible for the phosphorylation sites of thissubphosphoproteome.

KEYWORDS: subphosphoproteome analysis, cysteine-containing phosphopeptide, crosstalk

■ INTRODUCTION

The biological functions and structures of almost all eukaryoticproteins are fine-tuned by various post-translational modifica-tions (PTMs).1,2 The interplay and coregulation of differentPTMs have drawn much more attention than before.3,4 Amongthe 20 natural amino acids, cysteine is quite unique since it isthe active site of redox and can form disulfide bond throughoxidation and reduction of sulfhydryl; thus, it plays a pervasiverole in modulation of protein activities and structures.5

Reversible protein phosphorylation, on the other hand, isamong the most important and best explored PTMs and plays apivotal role in almost all biological processes including celldivision, differentiation, polarization, and apoptosis.2 Moreover,it functions as molecular switches in cellular signal transduction.Crosstalk between phosphorylation and other PTMs hasbecome a recurrent theme in cell signaling regulation. Forexample, Swaney et al. performed the large-scale analysis of thecrosstalk between phosphorylation and ubiquitylation during

protein degradation6 and demonstrated that distinct phosphor-ylation sites were often used in conjunction with ubiquitylationand that these sites were highly conserved. However, thecrosstalk between phosphorylation and the formation andreductive cleavage of disulfide bond has not been investigatedso far. As the active site of redox, cysteine often plays criticalroles in key functional protein such as protein kinase andphosphatase. To facilitate the study of the crosstalk betweenphosphorylation and disulfide bond, and also to characterizethe phosphorylation events of the key functional cysteine-containing proteins, it is important to explore the subset ofphosphoproteome where phosphorylations occurred nearbycysteine in the protein sequences.Recent advances in mass spectrometry (MS)-based proteo-

mics have made it possible to identify tens of thousands of

Received: September 6, 2015Published: November 10, 2015

Article

pubs.acs.org/jpr

© 2015 American Chemical Society 5341 DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

phosphosites from a cell or tissue sample per study.7,8 However,identification of low abundant phosphosites is still a bigchallenge due to the low abundance of phosphoproteins andthe low stoichiometry of the phosphorylation. The problemwith shotgun proteomics is that the peptides produced in theproteolysis step overwhelm the analytical capacity of currentliquid chromatography (LC)−MS systems both in number andin dynamic range. By accepting the fact that complete proteomeanalysis will remain difficult using currently available methods, a“divide and conquer” strategy has emerged to comprehensivelyanalyze specific subsets of the proteome that are selectivelyisolated.9 Similarly, the “divide and conquer” strategy could alsobe applied in phosphoproteome analysis to identify lowabundant phosphosites. The subphosphoproteome approachby selective enrichment of pTry (phosphorylated tyrosine)peptides was proved to identify more pTry sites than the globalphosphoproteome approach where all phosphopeptides wereenriched.7,10,11 A negative enrichment strategy by depletion of asubset of phosphopeptides was used, that is, acidicphosphopeptides were reported to enhance the identificationof basophilic kinase substrates.12 Selective enrichment ofmonophosphopeptides or multiphosphopeptides was provento improve the coverage for that specific type of subphospho-proteome.13 Therefore, as a subphosphoproteome approach,selective enrichment of the phosphopeptides containingcysteine residue also has the potential to identify low abundantphosphosites. We term this subphosphoproteome as Cys-subphosphoproteome since these phosphosites neighbor withcysteine.For analyzing the Cys-subphosphoproteome, the key is to

selectively enrich cysteine-containing phosphopeptides. In thisstudy, we developed a straightforward method to do this. First,proteins were reduced by dithiothreitol (DTT) and digested bytrypsin. Then the acquired peptides were incubated with thiol-specific affinity resins, and the cysteine-containing peptidecould be enriched with high specificity. The cysteine-containingpeptides were further incubated with immobilized titanium(IV)ion affinity chromatography (Ti4+-IMAC) beads to enrichphosphopeptides. This subphosphoproteome approach identi-fied three-times more cysteine-containing phosphopeptidesthan that achieved by the conventional global phosphopro-teome approach. It was found that amino acid residuessurrounding the phosphosites identified by Cys-subphospho-proteome approach were enriched with buried residues whiledepleted with exposed residues. On the basis of the kinasesubstrates predicted by bioinformatics tool, we found moreproline-directed kinases, and fewer acidophilic kinases wereresponsible for the phosphorylation of residues neighboringcysteine.

■ MATERIALS AND METHODS

Materials and Chemical Reagents

All water used in this experiment was prepared using a Milli-Qsystem (Millipore, Bedford, MA). DTT, iodoacetamide (IAA),ammonium bicarbonate (NH4HCO3), trifluoroacetic acid(TFA), sodium orthovanadate (Na3VO4), sodium fluoride(NaF), and trypsin were all obtained from Sigma-Aldrich (St.Louis, MO, USA). Formic acid (FA) was bought from Fluka(Buches, Germany), and acetonitrile (ACN, HPLC grade) waspurchased from Merck (Darmstadt, Germany). Fused-silicacapillaries with 200 μm i.d. and 75 μm i.d. were purchased fromPolymicro Technologies (Phoenix, AZ, USA). Daisogel ODS-

AQ (5 μm, 12 nm pore) was purchased from DAISO ChemicalCo., Ltd. (Osaka, Japan). The Thiopropyl Sepharose 6B waspurchased from GE healthcare (Shanghai, China).

Cell Culture and Digest Preparation

The Jurkat cells were grown in RPMI-1640 (Roswell ParkMemorial Institute 1640), supplemented with 10% bovineserum, 100 U/mL of streptomycin and penicillin. The cellswere harvested at about 80% density. Half of the cells weretreated with 1 mM freshly prepared pervanadate for 15 min at37 °C, and the other ones were left untreated. After that, thecell pellets were softly homogenized in an ice-cold lysis buffercontaining 8 M urea, 50 mM Tris-HCl (pH = 7.4), 2% proteaseinhibitor cocktail (v/v), 1% Triton X-100 (v/v), 1 mM NaF,and 1 mM Na3VO4, sonicated at 400 W for 120 s, andcentrifuged at 25 000g for 1 h. The proteins were precipitatedand purified with ice-cold acetone/ethanol/acetic acid, asdescribed in our previous study.14 Then the proteins weredissolved in a buffer containing 100 mM NH4HCO3 (pH =8.2) and 8 M urea. After being reduced by DTT at 37 °C for 2h, the samples were diluted to 1 M urea and digested by trypsinwith an enzyme-to-protein ratio of 1:25 (w/w). The resultedpeptide mixture was desalted by OASIS HLB column (Waters,Milford, MA, USA).

Cysteine-Containing Peptide Enrichment andPhosphopeptide Enrichment

The enrichment of cysteinyl containing peptide was performedas described by Liu15 et al. Briefly, tryptic digest from 1 mgprotein was dissolved in 100 μL of 50 mM Tris buffer, pH 7.5,1 mM EDTA (coupling buffer) and reduced by 5 mM DTT at37 °C for 30 min. Then the sample was diluted to 500 μL byadding coupling buffer and was incubated with 100 μL ofThiopropyl Sepharose 6B thiol-affinity resin for 2 h at 25 °C.After that, the resin was washed in turn with 0.5 mL × 5 of 50mM Tris buffer, pH 8.0 (washing buffer), 1 M NaCl, 80%ACN, and coupling buffer. The resin was incubated triple with100 μL of 20 mM freshly prepared DTT to release the capturedcysteinyl-peptides. Then the sample was alkylated with 80 mMiodoacetamide for 30 min at room temperature (RT) in thedark. Part of the sample was cleaned by C18 SPE column andanalyzed by LC−MS/MS. The other was further enriched withimmobilized Ti4+-IMAC for phosphopeptide enrichment.The enrichment of phosphopeptides was performed as

described previously.16 Briefly, peptide mixtures were firstincubated with Ti-IMAC beads at a ratio of 1:10 w/w inloading buffer (80% ACN and 6% TFA). After centrifugation,the supernatant was removed. The Ti-IMAC beads withadsorbed phosphopeptides were then washed in turn by twowashing buffers (50% ACN, 6% TFA containing 200 mM NaClas washing buffer 1; 30% ACN containing 0.1% TFA aswashing buffer 2) to remove nonspecific adsorbed peptides.The bound phosphopeptides were then eluted by 10% NH3·H2O. After centrifugation at 20 000g for 5 min, the supernatantwas collected and lyophilized to dryness.

Nano LC−MS/MS Analysis

The nano RPLC−MS/MS-experiments were performed on anUltiMate 3000 RSLCnano systems (Thermo Scientific, USA)connected to a Q Exactive (Thermo Scientific, USA). Thesamples were dissolved in 0.1% FA, and 6 μL of each samplewas automatically loaded onto the C18 trap column (3 cm ×200 μm i.d.) at a flow rate of 5 μL/min. The 75 μm i.d.analytical column was packed with C18 AQ particles (5 μm, 12

Journal of Proteome Research Article

DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

5342

nm) to 15 cm length. The mobile phase A was 99.9% water/0.1% FA, and mobile phase B was 80% ACN/0.1% FA. Theelution gradient executed was 5% to 35% mobile phase B lastedfor 78 min.The Q Exactive instrument was operated in the data-

dependent mode. Survey scan MS spectra (m/z 400−2000)were acquired by the Orbitrap with 70 000 resolution (m/z200), and the AGC target was set to 1 × 106 with a maxinjection time of 120 ms. Dynamic exclusion was set to 30 s.The 12 most intense multiply charged ions (z ≥ 2) werefragmented by higher-energy collisional dissociation (HCD).The MS/MS scans were also acquired by the Orbitrap with35 000 resolution (m/z 200), and the AGC target was set to 1×105 with a max injection time of 120 ms. Typical massspectrometric settings were as follows: spray voltage, 2 kV;heated capillary temperature, 250 °C; normalized HCDcollision energy, 27%.

Database Search and Data Analysis

The raw data files generated by the Q Exactive were searchedwith software MaxQuant version 1.3.0.517 against the Uniprothuman database (released on December 11, 2013 andcontaining 88 473 protein sequences), supplemented byfrequently observed contaminants, and reversed versions ofall sequences were contained. Enzyme specificity was set totrypsin (KR/P), and up to two missed cleavage sites wereallowed. Phospho (STY), oxidation (M), loss of ammonia andwater were chosen for variable modifications; carbamidomethylwas set as fixed modifications. The maximum false-discoveryrate (FDR) was set to 1% for both the peptides and proteins.The minimum required peptide length was set at six aminoacids. A web-based application Two Sample Logo18 was used tocalculate and visualize differences between two sets of alignedsamples of amino acids. Motif-X algorithm19 (http://motif-x.med.harvard.edu) was also used to generate phosphorylationmotifs for the identified phosphorylation sites. The web serverIUPred20 was used to predict the disorder tendency for theidentified phosphorylation sites with the prediction type set asshort disorder.

■ RESULTS AND DISCUSSION

Since the main goal of this study was to analyze the Cys-subphosphoproteome, where phosphosites neighboring withcysteine residue, it is of interest to know how many cysteine-containing phosphopeptides present in the data set acquired byconventional phosphoproteomics approach. Therefore, weperformed a global phosphoproteome analysis of Jurkat cellsby conventional phosphopeptide enrichment approach where asingle-step Ti4+-IMAC enrichment was performed. The threereplicate LC−MS/MS analyses of the enriched phosphopep-tides resulted in the identification of 7346 phosphopeptides. Itwas found that only 573 of the 7346 phosphopeptides (lessthan 8%) contained cysteine (details provided in Tables S-1and S-3), indicating the low abundance of cysteine-containingphosphopeptides. Thus, the selective enrichment of cysteine-containing phosphopeptides is indispensable for comprehensiveanalysis of the Cys-subphosphoproteome. Theoretically, thiscould be achieved by combining the thiol affinity chromatog-raphy with the Ti4+-IMAC enrichment, and these twoenrichment steps should be proceeded in sequence. Phospho-peptides captured on the Ti4+-IMAC resins could be eluted byvolatile ammonium solution and then lyophilized for directRPLC−MS/MS analysis and thus had good compatibility with

MS analysis. Considering this, we adopted the workflow shownin Figure 1 where the thiol affinity chromatography wasperformed before Ti4+-IMAC for the selective enrichment ofcysteine-containing phosphopeptides.

This sequential enrichment method was applied to analyzethe Cys-subphosphoproteome of Jurkat cells. We firstinvestigate the performance of the thiol-affinity chromatog-raphy for the enrichment of cysteine-containing peptides.Proteins derived from pervanadate treated Jurkat cells werereduced with DTT and digested by trypsin, as shown in Figure1. The resulting peptides were reduced by 5 mM DTT again tobreak down the scrambling disulfide bonds that may haveformed by the free sulfydryl groups on cysteine residues atalkaline pH during protein digestion. Then the peptidemixtures were incubated with a commercially available thiol-affinity resin (thiopropyl sepharose 6B) for the enrichment ofcysteine-containing peptides. The eluted peptides weresubjected to LC−MS/MS analysis by Q-Exactive in triplet.The three raw data files were combined together and searchedby MaxQuant. After the FDR was controlled to <1%, weidentified 2435 peptides in total, of which 99% (2412)contained at least one cysteine in the peptide sequence. Forcomparison, the same tryptic digest sample without thiol-affinity enrichment was directly analyzed by LC−MS/MS intriplet. Of the 2202 peptides identified, only 19% of themcontaining cysteine. Details of the identified peptides wereprovided in Table S-4. The above data indicated the highspecificity of thiol-affinity enrichment.The eluted peptide mixture from the first step thiol-affinity

enrichment was then subjected to Ti4+-IMAC enrichment(Figure 1). The acquired phosphopeptides by this sequentialenrichment method were analyzed in triplet by LC−MS/MS,and the corresponding raw files were combined for databasesearching by MaxQuant. In total, we identified 2089 uniquephosphopeptides, and among them, 1919 (92%) had at leastone cysteine in their sequence. Compared to the globalphosphoproteome, the percentage of cysteine-containingphosphopeptide was increased by around 12-fold (Figure2A,B). The high-quality result we obtained benefits from thehigh specificity of both the thiol-affinity enrichment and thephosphopeptide enrichment.Next, we compared the phosphopeptides identified by the

global phosphoproteome approach (single step Ti-IMACenrichment) with that identified by the Cys-subphosphopro-

Figure 1. Workflow for the analysis of Cys-subphosphoproteome by asequential enrichment method.

Journal of Proteome Research Article

DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

5343

teome approach (sequential enrichment). It was found that 352phosphopeptides overlapped in both data sets account for only4% of the total phosphopeptides. Additionally, 1734phosphopeptides (of which 1615 were cysteine-containingphosphopeptides) were newly identified by the sequentialenrichment method, which illustrates that the “divide andconquer” strategy allowed the identification of phosphopeptideswith lower abundance. Though the Cys subphosphoproteomeapproach identified many fewer phosphopeptides than theglobal phosphoproteome approach did (7346 vs 2089), itidentified three-times more cysteine-containing phosphopep-tides (573 vs 1919), and two-thirds of the cysteine-containingphosphopeptides identified by the globe phosphoproteomeapproach can also be identified by the Cys-subphosphopro-teome approach as shown in Figure 2, panel C. Clearly, thissequential enrichment approach significantly improved thecoverage for Cys-subphosphoproteome analysis.This approach was applied to analyze the Cys-subphospho-

proteomes of two other Jurkat cell samples. One (Sample 2)was treated with pervanadate as the sample analyzed above(Sample 1), and the other one (Sample 3) was not treated withpervanadate. The enriched phosphopeptides were analyzed intriplet by LC−MS/MS as above, which led to the identificationof 2106 and 2046 unique cysteine-containing phosphopeptides,respectively, with the specificity higher than 80% (Table S-1).The above data confirmed that this approach could consistentlyidentify cysteine-containing phosphopeptides. In terms ofphosphosites, 1658, 2165, and 1823 sites were identified fromSamples 1, 2 and 3, with the percentage of pTyr(phosphorylated tyrosine) sites accounting for 9%, 5%, and1%, respectively. Detailed information was provided in TablesS-2 and S-5. Higher percentages of pTyr sites were identifiedfrom the first two samples because sodium pervanadate was theinhibitors of tyrosine phosphatase. From these three samples,we identified 3195 phosphosites in total; among them, 1826were high confident sites (with localization probability >0.75and score difference >5). Among the high confident sites, 1463sites were found to be with at least one cysteine residue locatedwithin ±15 amino acid residues surrounding the phosphosite(set as 0). This high confident data set was then used tocharacterize the Cys-subphosphoproteome.

To get a straightforward visualization of the differencebetween phosphosites identified by the two approaches, that is,the Cys-subphosphoproteome approach and the globalphosphoproteome approach, the online software, Two SampleLogo,18 was applied to analyze the aligned sequences of theidentified phosphosites. The 1463 phosphosites neighboringwith Cysteine identified by the Cys-subphosphoproteomeapproach were set as positive data set, and phosphositesidentified by the global phosphoproteome approach were set asnegative data set. From the results shown in Figure 3, we can

see that the distribution of Cysteine around the phosphositeswas not even. For pSer/pThr sites, cysteine was enriched onthe positions of −2 and −1. The pTyr sites showed a quitedifferent pattern, where Cysteine was largely enriched on the+1 position. Cysteine is typically considered as one type of theburied residues in globular proteins.21 It is not a surprise thatcysteine (C) is enriched on sites surrounding the phosphositesin the Cys-phosphoproteome since the cysteine-containingphosphopeptides were enriched for identification. Interestingly,other buried residues such as L, V, and A were enriched in theCys-subphosphoproteome, while the exposed residues such asD, E, R, and K were depleted. This is especially true for pSersites. This hints that there may be a subtle difference in thepreference of the kinases to recognize these phosphosites.Motif-X (http://motif-x.med.harvard.edu) is a software tool

designed to extract overrepresent patterns from any sequencedata set. In this study, it was used to extract the overrepresentmotifs from the Cys-phosphoproteome and the globalphosphoproteome using the human fasta database as back-ground (Figures S-1 and S-2), respectively. As we know, theSer/Thr protein kinases fall into three major subgroups, pro-directed, basophilic, and acidophilic, on the basis of the types ofsubstrate sequences that they preferred. After we comparedFigures S-1 and S-2, we found that proline-directed kinasesubstrates (with a proline shown on the +1 position) were

Figure 2. Cys-subphosphoproteome approach identified more Cys-containing phosphopeptides. Pie graphs and for phosphopeptidesidentified by the (A) global and the (B) Cys-subphosphoproteomeapproaches. (C) Overlap of Cys-containing phosphopeptides identi-fied by the two methods.

Figure 3. Two sample logos showed the differences betweenphosphosites identified by the Cys-subphosphoproteome approachand the global phosphoproteome approach. The aligned sequencescentered with phosphosites were used to generate logos, withphosphosites identified in the Cys-subphosphoproteome and theglobal phosphoproteome set as positive and negative data set,respectively.

Journal of Proteome Research Article

DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

5344

largely enriched by the sequential cysteine-containing phos-phopeptide enrichment method, with a cysteine simultaneouslyshown in the motif. Further analysis showed that the top ninemotifs generated by such phosphosites were pSP with acysteine shown within the −7 to +7 position, as shown inFigure S-1. There were two motifs with the presence ofbasophilic residue arginine (R) and cysteine, RXXpSXXC andRCXpS. Phosphopeptides originating from acidic kinasesubstrates always made up a large proportion in theconventional phosphopeptide enrichment method, and manymotifs with D/E were observed for global phosphoproteome(Figure S-2). However, after we performed the cysteine-containing phosphopeptide enrichment, it was hard to find amotif with both acidic residues D/E and C in a single motif.This is consistent with the fact that acidic residues D/E aredepleted in the Cys-subphosphoproteome as discussed earlier.Identification of phosphorylation sites with their cognate

protein kinases is of great importance in understanding signaltransduction in complex biological systems. To obtain theputative kinase information on the phosphorylation sitesidentified, GPS22 (Group-based Prediction System) was usedto predict kinase-specific phosphorylation sites at a highstringency level. The bar chart in Figure 4, panel A showedthe relative distribution of predicted kinase-specific substratesfound in global phosphoproteome (blue bar) and Cys-subphosphoproteome (red bar). From the result, we can seethat the CMGC kinase group substrates identified in the Cys-subphosphoproteome increased notably, from 27% to 38%, ascompared to that identified in the global phosphoproteome. Onthe contrary, phosphopeptides originating from acidic kinasesubstrates such as Casein kinase 1 (CK1) and basophilic AGCkinase group (including protein kinase A, protein kinase G,protein kinase C, etc.) showed a decrease in the Cys-subphosphoproteome. The CMGC kinase group included aseries of key kinases such as Mitogen-activated protein kinases(MAPK), cell cycle cyclin dependent kinases (CDK), andkinases involved in splicing and metabolic control. Theenrichment of CMGC kinase group substrates in the Cys-subphosphoproteome indicated the crosstalk between phos-phorylation, and disulfide bond formation and dissociation mayplay an important role in regulating biological processes.Previously, it was found that phosphorylated Ser, Thr, and

Tyr residues exhibited notable differences in structuralclassification compared to their unmodified counterparts; theywere predominantly predicted to locate in intrinsicallydisordered regions rather than in ordered secondary

structures.23 Recent research shows that tyrosine phosphor-ylation is kind of different from phosphorylated Ser/Thr and isoften observed in ordered interface regions, which are notpredicted to be disordered in the unbound state.24,25 Intrinsi-cally disordered regions are flexible and extend proteinsegments that have no ordered secondary structure underphysiological conditions; they featured their sequence by rich indisorder-promoting residues (D, K, R, S, Q, P, and E) and aredevoid of order-promoting residues (W, Y, F, I, V, L, and T).Despite a large number of phosphorylation sites that show apreference to locating in disordered regions, there are alsophosphosites that undertake important functions located in thestructured regions. For example, Gygi et al. found in their workthat most of the 120 observed activation loop phosphosites ofkinase were ordered with elevated levels classified as strands.26

For our result, the buried residues (L, V, A, C) wereenriched, while the exposed residues (D, E, R, K) were depletedin the Cys-subphosphoproteome, which also implies that thesephosphosites may have some connection with protein structure.We used a web-based software IUPred to predict the disorder/order tendency for the identified phosphosites, and the resultswere shown in Figure 4, panel B. It was found that 85.8% ofpSer/pThr sites in global phosphoproteome were predicted tolocate in disorder regions, while only 46.1% phosphorylatedTyr sites were located in disorder regions. This was consistentwith the knowledge that pSer/pThr sites were more frequentlyobserved in disordered regions compared with pTyr sites. Afterperforming the cysteine-containing phosphopeptides enrich-ment, the identified phosphorylated Ser/Thr/Tyr sites allshowed a dramatic decrease in locating in disorder regions. ThepSer/pThr sites decreased from 85.8% to 59.5%, and the pTyrsites decreased from 46.1% to 18%. Collectively, the order-promoting cysteine affects the microenvironment of thephosphorylation site and a subclass of phosphorylation sites,which tend to be located in ordered regions of the protein, wereidentified.The above analysis indicated that there are some differences

between global phosphoproteome and Cys-subphosphopro-teome. However, we do not know if there is a biologicalsignificance for the phosphorylation that occurred close tocysteine residues yet. It is well-known that both Cys andphosphorylation play critical roles in key functional proteinssuch as protein kinase and phosphatase. A bioinformaticanalysis of the human kinome27 revealed that there are 46kinases that have a conserved Cys residue located in the ATP-binding site, and there are approximately 200 different kinases

Figure 4. Differences of the phosphosites in Cys-subphosphoproteome and global phosphoproteome. (A) Bar chart shows the relative distribution ofpredicted kinase-specific substrates found in global phosphoproteome (blue bar) and Cys-subphosphoproteome (red bar). (B) Ratio of phosphositesidentified in the global phosphoproteome (blue bar) and Cys-subphosphoproteome (red bar) locating in disorder regions.

Journal of Proteome Research Article

DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

5345

that have a cysteine located nearby the ATP pocket, whichsuggests that cysteine may play important roles in regulation ofenzyme activities. Indeed, an important class of kinaseinhibitors is capable of forming an irreversible, covalent bondto the kinase active site by reacting with a cysteine residue,28

thus blocking the binding of ATP to the kinase, therebyrendering the kinase inactive. The activities of kinases are alsoregulated by the phosphorylation of the activation loop.29 Takethe 90-kDa ribosomal S6 kinase 1 (Q15418) for example; theS221 phosphorylation site was identified in our data set with anadjacent C223. Those two residues are located in N-terminalkinase domain. It was reported that this kinase can be activatedby phosphorylation at S221 by PDK1 (3-phosphoinositide-dependent protein kinase 1).30,31 It was also reported that theC223 is the active-site cysteine residue, and the S-glutathionylation of C223 is both necessary and sufficient forthe inhibition of this kinase during oxidative stress.32 Clearly,the phosphorylation of S221 and oxidative state of C223 bothhave close relationships with the activity of this kinase. Protein-tyrosine phosphatases (PTPs) are another kind of importantfunctional protein that can dephosphorylate phosphotyrosylresidues of proteins and peptides. PTPs are also phosphopro-teins. Intradomain phosphorylation modulates the catalyticactivity of the PTP domain.33,34 On the other hand, thehallmark of PTPs is an essential cysteine residue at the catalyticsite, which forms a thiol−phosphate intermediate duringcatalysis. The oxidation of the catalytic Cys residue causesthe inactivation of the enzyme.35 In our data set, we identifiedthe Y50 sites with an adjacent C46 in the protein tyrosinephosphatase type IVA 2 (Q12974).36 The active site of thisPTP was C101, which was confirmed to form disulfide bondwith C46. The formation of disulfide bond may lead to theinactivity of the phosphatase. Considering the important rolesof Cys and phosphorylation in regulating enzyme activities, webelieve there is some type of crosstalk between these twomodifications. The identification of the peptides that containboth Cys and phosphorylated groups will facilitate the study ofthis crosstalk.

■ CONCLUSIONIn this study, we developed a sequential enrichment method toselectively enrich cysteine-containing phosphopeptides. Thehigh specificity and high sensitivity of this method weredemonstrated by analyzing the samples of Jurkat cells. This“divide and conquer” strategy by specific analysis of Cys-subphosphoproteome enables the identification of lowabundant phosphosites neighboring cysteine residues. Theobtained data sets allows the characterization of thissubphosphoproteome for the first time. Interestingly, aminoacid residues surrounding the phosphosites in this subphos-phoproteome were enriched with buried residues whiledepleted with exposed residues. It was also found that moreproline-directed kinases and less acidophilic kinases areresponsible for the phosphorylation sites. This is the firststudy to explore the relationship between the cysteine andphosphorylated S/T/Y and will shed some light on thecrosstalk between the form of disulfide bond and phosphor-ylation; this study may also supply a tool for characterizing thephosphorylation events of key functional cysteine-containingproteins. We believe that the true relationship between disulfidebond and phosphorylation is quite complex, and its studydepends not only the combination of bottom-up proteomicsand top-down proteomics, but also the application of

bioinformatics tools. Also, our method could be one of theimportant tools in the toolbox to decipher this importantcrosstalk.

■ ASSOCIATED CONTENT*S Supporting Information

The Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jproteo-me.5b00830.

Number of peptides identified by the Cys-subphospho-proteome approach or the Global phosphoproteomeapproach; number of peptides identified by threeindependent sequential enrichment experiments; motifsgenerated by the Cys-subphosphoproteome; motifsgenerated by the Global phosphoproteome (PDF)Phosphopeptide identified by the single-step Ti4+-IMAC(XLS)Peptides identified from tryptic digests of Jurkat cells orafter performing thiol affinity chromatography enrich-ment (XLS)Phosphopeptides identified by three independentsequential enrichment experiments (XLS)

■ AUTHOR INFORMATIONCorresponding Authors

*Phone: 86-411-84379610. Fax: 86-411-84379620. E-mail:hanfazou@dicp.ac.cn.*Phone: 86-411-84379620. Fax: 86-411-84379620. E-mail:mingliang@dicp.ac.cn.Notes

The authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the China State Key BasicResearch Program Grant (2013CB911204, 2012CB910101,and 2012CB910604), the Creative Research Group Project ofNSFC (21321064), and the National Natural ScienceFoundation of China (21235006, 21275142, 81361128015,21535008, and 21525524).

■ REFERENCES(1) Hunter, T. Signaling - 2000 and beyond. Cell 2000, 100, 113−127.(2) Pawson, T.; Scott, J. D. Protein phosphorylation in signaling - 50years and counting. Trends Biochem. Sci. 2005, 30, 286−290.(3) Beltrao, P.; Albanese, V.; Kenner, L. R.; Swaney, D. L.;Burlingame, A.; Villen, J.; Lim, W. A.; Fraser, J. S.; Frydman, J.;Krogan, N. J. Systematic functional prioritization of proteinposttranslational modifications. Cell 2012, 150, 413−25.(4) Hunter, T. The age of crosstalk: phosphorylation, ubiquitination,and beyond. Mol. Cell 2007, 28, 730−8.(5) Xu, H.; Zhang, L.; Freitas, M. A. Identification and character-ization of disulfide bonds in proteins and peptides from tandem MSdata by use of the MassMatrix MS/MS search engine. J. Proteome Res.2008, 7, 138−144.(6) Swaney, D. L.; Beltrao, P.; Starita, L.; Guo, A.; Rush, J.; Fields, S.;Krogan, N. J.; Villen, J. Global analysis of phosphorylation andubiquitylation cross-talk in protein degradation. Nat. Methods 2013,10, 676−682.(7) Sharma, K.; D’Souza, R. C. J.; Tyanova, S.; Schaab, C.;Wisniewski, J. R.; Cox, J.; Mann, M. Ultradeep Human Phosphopro-teome Reveals a Distinct Regulatory Nature of Tyr and Ser/Thr-BasedSignaling. Cell Rep. 2014, 8, 1583−1594.

Journal of Proteome Research Article

DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

5346

(8) Villen, J.; Beausoleil, S. A.; Gerber, S. A.; Gygi, S. P. Large-scalephosphorylation analysis of mouse liver. Proc. Natl. Acad. Sci. U. S. A.2007, 104, 1488−93.(9) Zhang, H.; Yan, W.; Aebersold, R. Chemical probes and tandemmass spectrometry: a strategy for the quantitative analysis ofproteomes and subproteomes. Curr. Opin. Chem. Biol. 2004, 8, 66−75.(10) Kettenbach, A. N.; Gerber, S. A. Rapid and Reproducible Single-Stage Phosphopeptide Enrichment of Complex Peptide Mixtures:Application to General and Phosphotyrosine-Specific Phosphoproteo-mics Experiments. Anal. Chem. 2011, 83, 7635−7644.(11) Kwon, O. K.; Sim, J.; Yun, K. N.; Kim, J. Y.; Lee, S. Globalphosphoproteomic analysis of Daphnia pulex reveals evolutionaryconservation of Ser/Thr/Tyr phosphorylation. J. Proteome Res. 2014,13, 1327−35.(12) Dong, M.; Ye, M.; Cheng, K.; Song, C.; Pan, Y.; Wang, C.; Bian,Y.; Zou, H. Depletion of Acidic Phosphopeptides by SAX To Improvethe Coverage for the Detection of Basophilic Kinase Substrates. J.Proteome Res. 2012, 11, 4673−4681.(13) Zhong, H.; Xiao, X.; Zheng, S.; Zhang, W.; Ding, M.; Jiang, H.;Huang, L.; Kang, J. Mass spectrometric analysis of mono- and multi-phosphopeptides by selective binding with NiZnFe2O4 magneticnanoparticles. Nat. Commun. 2013, 4, 1656.(14) Wang, C.; Ye, M.; Bian, Y.; Liu, F.; Cheng, K.; Dong, M.; Dong,J.; Zou, H. Determination of CK2 specificity and substrates byproteome-derived peptide libraries. J. Proteome Res. 2013, 12, 3813−21.(15) Liu, T.; Qian, W. J.; Strittmatter, E. F.; Camp, D. G.; Anderson,G. A.; Thrall, B. D.; Smith, R. D. High-throughput comparativeproteome analysis using a quantitative cysteinyl-peptide enrichmenttechnology. Anal. Chem. 2004, 76, 5345−5353.(16) Zhou, H. J.; Ye, M. L.; Dong, J.; Han, G. H.; Jiang, X. N.; Wu, R.A.; Zou, H. F. Specific phosphopeptide enrichment with immobilizedtitanium ion affinity chromatography adsorbent for phosphoproteomeanalysis. J. Proteome Res. 2008, 7, 3957−3967.(17) Cox, J.; Mann, M. MaxQuant enables high peptide identificationrates, individualized ppb-range mass accuracies and proteome-wideprotein quantification. Nat. Biotechnol. 2008, 26, 1367−1372.(18) Vacic, V.; Iakoucheva, L. M.; Radivojac, P. Two Sample Logo: agraphical representation of the differences between two sets ofsequence alignments. Bioinformatics 2006, 22, 1536−1537.(19) Schwartz, D.; Gygi, S. P. An iterative statistical approach to theidentification of protein phosphorylation motifs from large-scale datasets. Nat. Biotechnol. 2005, 23, 1391−1398.(20) Dosztanyi, Z.; Csizmok, V.; Tompa, P.; Simon, I. IUPred: webserver for the prediction of intrinsically unstructured regions ofproteins based on estimated energy content. Bioinformatics 2005, 21,3433−3434.(21) Janin, J. Surface and inside volumes in globular proteins. Nature1979, 277, 491−492.(22) Xue, Y.; Ren, J.; Gao, X.; Jin, C.; Wen, L.; Yao, X. GPS 2.0, atool to predict kinase-specific phosphorylation sites in hierarchy. Mol.Cell. Proteomics 2008, 7, 1598−1608.(23) Iakoucheva, L. M.; Radivojac, P.; Brown, C. J.; O’Connor, T. R.;Sikes, J. G.; Obradovic, Z.; Dunker, A. K. The importance of intrinsicdisorder for protein phosphorylation. Nucleic Acids Res. 2004, 32,1037−1049.(24) Nishi, H.; Fong, J. H.; Chang, C.; Teichmann, S. A.; Panchenko,A. R. Regulation of protein-protein binding by coupling betweenphosphorylation and intrinsic disorder: analysis of human proteincomplexes. Mol. BioSyst. 2013, 9, 1620−6.(25) Nishi, H.; Hashimoto, K.; Panchenko, A. R. Phosphorylation inprotein-protein binding: effect on stability and function. Structure2011, 19, 1807−15.(26) Huttlin, E. L.; Jedrychowski, M. P.; Elias, J. E.; Goswami, T.;Rad, R.; Beausoleil, S. A.; Villen, J.; Haas, W.; Sowa, M. E.; Gygi, S. P.A tissue-specific atlas of mouse protein phosphorylation andexpression. Cell 2010, 143, 1174−89.(27) Schirmer, A.; Kennedy, J.; Murli, S.; Reid, R.; Santi, D. V.Targeted covalent inactivation of protein kinases by resorcylic acid

lactone polyketides. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 4234−4239.(28) Zhang, J.; Yang, P. L.; Gray, N. S. Targeting cancer with smallmolecule kinase inhibitors. Nat. Rev. Cancer 2009, 9, 28−39.(29) Huse, M.; Kuriyan, J. The conformational plasticity of proteinkinases. Cell 2002, 109, 275−282.(30) Frodin, M.; Gammeltoft, S. Role and regulation of 90 kDaribosomal S6 kinase (RSK) in signal transduction. Mol. Cell.Endocrinol. 1999, 151, 65−77.(31) Chen, R. H.; Sarnecki, C.; Blenis, J. Nuclear localization andregulation of erkand rsk-encoded protein kinases. Mol. Cell. Biol. 1992,12, 915−927.(32) Takata, T.; Tsuchiya, Y.; Watanabe, Y. 90-kDa ribosomal S6kinase 1 is inhibited by S-glutathionylation of its active-site cysteineresidue during oxidative stress. FEBS Lett. 2013, 587, 1681−1686.(33) Andersen, J. N.; Mortensen, O. H.; Peters, G. H.; Drake, P. G.;Iversen, L. F.; Olsen, O. H.; Jansen, P. G.; Andersen, H. S.; Tonks, N.K.; Moller, N. P. H. Structural and evolutionary relationships amongprotein tyrosine phosphatase domains. Mol. Cell. Biol. 2001, 21, 7117−7136.(34) Mitra, S.; Beach, C.; Feng, G.-S.; Plattner, R. SHP-2 is a noveltarget of Abl kinases during cell proliferation. J. Cell Sci. 2008, 121,3335−3346.(35) Huyer, G.; Liu, S.; Kelly, J.; Moffat, J.; Payette, P.; Kennedy, B.;Tsaprailis, G.; Gresser, M. J.; Ramachandran, C. Mechanism ofinhibition of protein-tyrosine phosphatases by vanadate andpervanadate. J. Biol. Chem. 1997, 272, 843−851.(36) Zhao, Z. Y.; Lee, C. C.; Monckton, D. G.; Yazdani, A.;Coolbaugh, M. I.; Li, X. R.; Bailey, J.; Shen, Y.; Caskey, C. T.Characterization and genomic mapping of genes and pseudogenes of anew human protein tyrosine phosphatase. Genomics 1996, 35, 172−181.

Journal of Proteome Research Article

DOI: 10.1021/acs.jproteome.5b00830J. Proteome Res. 2015, 14, 5341−5347

5347