Thr

Feb 27, 2012 - and Henrik Daub*. ,†,∥. †. Department of Molecular Biology, Max Planck Institute of Biochemistry, Am Klopferspitz 18, 82152 Marti...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/jpr

Global Analysis of Phosphoproteome Regulation by the Ser/Thr Phosphatase Ppt1 in Saccharomyces cerevisiae Thiemo B. Schreiber,†,⊥ Nina Maü sbacher,†,§,⊥ Joanna Soroka,‡,⊥ Sebastian K. Wandinger,‡,∥ Johannes Buchner,‡ and Henrik Daub*,†,∥ †

Department of Molecular Biology, Max Planck Institute of Biochemistry, Am Klopferspitz 18, 82152 Martinsried, Germany Center for Integrated Protein Science, Department of Chemistry, Technische Universität München, Lichtenbergstraße 4, 85747 Garching, Germany



S Supporting Information *

ABSTRACT: Even though protein phosphatases are key regulators of signal transduction, their cellular mechanisms of action are poorly understood. Here, we undertook a large-scale proteomics survey to identify cellular protein targets of a serine/threonine phosphatase. We used SILAC-based quantitative MS to measure differences in protein expression and phosphorylation upon ablation of the serine/threonine phosphatase Ppt1 in Saccharomyces cerevisiae. Phosphopeptide fractionation by strong cation exchange chromatography combined with immobilized metal affinity chromatography (IMAC) enrichment enabled quantification of more than 8000 distinct phosphorylation sites in Ppt1 wild-type versus Ppt1-deficient yeast cells. We further quantified the relative expression of 1897 yeast proteins and detected no major protein changes accompanying Ppt1 deficiency. Notably, we found 33 phosphorylation sites to be significantly and reproducibly upregulated while no phosphorylation events were repressed in cells lacking Ppt1. Ppt1 acted on its cellular target proteins in a sequence- and site-specific fashion. Several of the regulated phosphoproteins were involved in the response to heat stress in agreement with known Ppt1 functions. Additionally, biosynthetic enzymes were particularly prominent among Ppt1-regulated phosphoproteins, pointing to unappreciated roles of Ppt1 in the control of various metabolic functions. These results demonstrate the utility of large-scale and quantitative phosphoproteomics to identify cellular sites of serine/threonine phosphatase action in an unbiased manner. KEYWORDS: SILAC, yeast, phosphoproteomics, Ser/Thr phosphatase, Ppt1



INTRODUCTION Phosphorylation-based signaling is regulated by the antagonizing catalytic activities of protein kinases and protein phosphatases in all eukaryotic cells. Phosphatases ensure reversibility of kinase-mediated, site-specific phosphorylations on serine, threonine, and tyrosine residues and are therefore involved in the same wide range of biological functions as members of the protein kinase superfamily of enzymes. Similar to protein kinases, phosphatases exhibit either specificity for tyrosine or for serine/threonine residues.1,2 Ser/Thr phosphatases are divided into three major families: phosphoprotein phosphatases (PPPs) such as PP1, PP2A, PP2B, PP4, and PP5, metaldependent protein phosphatases such as PP2C, and aspartatebased phosphatases.3−8 Many Ser/Thr phosphatases are highly conserved from yeast to man, such as Ppt1 which is the Saccharomyces cerevisiae homologue of the human PPP PP5 sharing over 60% sequence identity.4 Unlike other PPPs, PP5 and Ppt1 have their substrate binding, regulatory, and catalytic domains clustered on a single polypeptide chain. A second © 2012 American Chemical Society

distinguishing feature shared by PP5 and Ppt1 is the presence of a tetratricopeptide repeat (TPR) domain which serves both regulatory and targeting functions.3,9−11 Structural studies of PP5 revealed that the TPR domain (in concert with the C-terminal αJ domain) folds over and blocks the catalytic cavity, thus maintaining the phosphatase in an autoinhibited state. 9,12−14 Biochemical studies of Ppt1 revealed that binding of ligand or deletion of the TPR domain strongly stimulates its low basal phosphatase activity.9,15−17 Generally, TPR domains mediate the interaction with TPR acceptor sites in proteins.18−21 Interestingly, in the case of Ppt1, the TPR acceptor is the molecular chaperone Hsp90 in vivo and in vitro, suggesting a role in the chaperone complex.11,16,18,22−26 In this context, Ppt1 was shown to be involved in dephosphorylating Hsp90 and the cochaperone Cdc37.23,27 No reports, however, have yet demonstrated how Ppt1 regulates further clients in vivo. Received: November 15, 2011 Published: February 27, 2012 2397

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

phosphatase in vivo regulating a limited number of substrates in a phosphorylation-site specific manner.

Ppt1 is not an essential protein in yeast. Cells lacking ppt1 do not exhibit significant growth defects or any obvious phenotype;4 however, its absence leads to decreased activation of Hsp90-specific clients.23 Ppt1 is localized both in the cytoplasm and the nucleus, implying that it may function in several subcellular compartments.17 Genetic studies in yeast extended our knowledge about the function of Ppt1. Extensive microarray studies showed that ppt1 expression is strongly down-regulated in response to different stress stimuli including metabolic, oxidative, osmotic, and thermal stress.28 Ppt1 expression levels were induced by growth in zinc-deficient medium or by treating cells with alpha factor29,30 and during early log phase growth.17,31,32 Though several physiological substrates for PP5 were identified to date;25,33 still little is known about its specific client proteins, therefore determining its biological functions is still a challenging task. All currently known Ppt1 substrates have been identified by candidate-based approaches. To date, no unbiased and large-scale analysis of cellular Ppt1 substrates has been reported. This requires both quantitative and comprehensive analysis of the cellular phosphoproteome, which has become feasible due to recent advances in MS-based proteomics. In particular, isotope encoding of entire proteomes for comparative analysis, such as the stable isotope labeling by amino acids in cell culture (SILAC) approach pioneered by Mann and colleagues, efficient phosphopeptide fractionation and enrichment from complex mixtures, and sensitive highaccuracy liquid chromatography−tandem mass spectrometry (LC−MS/MS) on linear ion trap/orbitrap hybrid mass spectrometers have made it feasible to identify and quantify many thousand site-specific phosphorylations in highly complex extracts.34−43 Moreover, breakthroughs in MS data processing have further increased mass accuracy and identification rates in addition to enabling automated quantification of SILAC proteome and phosphoproteome data.34,38,44 SILAC-based quantitative MS has been successfully applied to study the proteome and phosphoproteome regulation by the mammalian protein-tyrosine phosphatase 1B and its Drosophila orthologue Ptp61F in mouse embryonic fibroblasts and Drosophila Schneider cells, respectively.45,46 Apart from these SILAC studies, a small number of label-free phosphopeptide quantifications upon PP5 overexpression have been performed by Ham et al.47 Also, very recently, a large-scale phosphoproteomics survey based on label-free quantification of 97 kinase and 27 phosphatase yeast knockout strains has been reported, including a comparison of wild-type and ppt1 deletion yeast strains.48 In this study, a total of 158 168 phosphorylation site identifications were made across 124 different experiments, with an average of 1240 phospho-events captured per individual wild-type/knockout strain comparison. As such, a coverage falls short of the more than 5000 phosphosites analyzed upon extensive prefractionation by strong cation exchange (SCX) or hydrophilic interaction chromatography prior to phosphopeptide enrichment,35,43,49 and further in-depth phosphoproteome analyses of the in vivo substrate specificity of individual Ser/Thr phosphatases appears warranted. Such knowledge would likely promote additional insights into cellular mechanisms of Ser/ Thr phosphatase-mediated dephosphorylations. Here, we used SILAC to enable quantitative comparisons of wild-type yeast cells expressing Ppt1 and those in which the gene for the Ser/Thr phosphatase Ppt1 was deleted. We performed large-scale proteomics and phosphoproteomics analyses that established Ppt1 as a selective Ser/Thr



EXPERIMENTAL SECTION

Cell Culture and Lysis

Yeast cells in the YAL6B background (with a deletion of arg4 and lys1 and therefore auxotrophic for arginine and lysine) with or without ppt1 deletion were grown in yeast nitrogen base (YNB) liquid medium supplemented with 100 μg/mL of arginine and lysine.23,50 For SILAC labeling, Arg0/Lys0 were added to the medium of Ppt1 wild-type and the medium for Ppt1-deficient yeast cells was supplemented with the heavier isotopic variants Arg10/Lys8. To assess biological reproducibility irrespective of labeling conditions, the SILAC scheme was inverted in a second replicate experiment. Cells were grown for 18 h at 30 °C to ensure quantitative metabolic incorporation of added amino acids through de novo protein biosynthesis. After this growth period both Ppt1 wild-type and deficient yeast cells had reached an optical density of 5.5, as determined by A600 measurements. Equal amounts of cells were combined and harvested by centrifugation for 5 min at 4500g at 4 °C, washed once with a washing buffer (40 mM HEPES/KCl, pH 7.5, 150 mM KCl), resuspended in a lysis buffer [40 mM HEPES/KOH pH 7.5, 150 mM KCl, 2 mM EDTA, 2 mM EGTA, 20 mM NaF, 5 mM Na3VO4, protease inhibitor cocktail with AEBSF, aprotinin, leupeptin, E-64/EDTA, phosphatase inhibitor cocktail 1 (Sigma, St. Louis, MO), phosphatase inhibitor cocktail 2 (Sigma, St. Louis, MO)], and lysed using glass beads. The total protein concentration in the cellular extracts was determined by a Bradford assay.51 Sample Preparation for Mass Spectrometry

In both replicate experiments, lysates were subjected to protein precipitation according to the method by Wessel and Flügge.52 Protein pellets were resolved in 20 mM HEPES-NaOH buffer (pH 7.5) containing 7 M urea and 2 M thiourea. Proteins were reduced, alkylated, and digested in solution with endoproteinase Lys-C and modified trypsin as described previously.36 The resulting peptide mixtures were desalted on C18 SepPak cartridges (Waters). Briefly, samples were adjusted to 0.5% TFA and loaded onto cartridges with 500 mg maximum capacity, which were pre-equilibrated with 0.1% TFA. After thorough washing with 0.1% TFA/0.5% acetic acid, bound peptides were eluted with 50% acetonitrile (ACN)/0.5% acetic acid. Organic solvent was evaporated by lyophilization. In total, 98% of the total peptide sample corresponding to approximately 4.5 mg of material was fractionated by strong cation exchange (SCX) chromatography.53 Peptide samples were dissolved in 500 μL of 7 mM KH2PO4, pH 2.65, 30% ACN and loaded on a 250 mm × 9.4 mm PolySULFOETHYL A column (PolyLC). Peptides were fractionated by a gradient ranging from 0% to 30% elution buffer (7 mM KH2PO4, pH of 2.65, 30% ACN, 350 mM KCl). The resulting fractions were pooled to 10 samples according to the UV absorption at 215 nm to ensure even peptide distribution. Afterward, samples were desalted, lyophilized, and phosphopeptides enriched by immobilized metal affinity chromatography (IMAC). Therefore, each fraction was dissolved in 200 μL of binding buffer (40% ACN, 25 mM formic acid) and incubated for 1 h with 5 μL of equilibrated IMAC beads (Sigma). Beads were transferred onto C18 StageTips, washed twice with binding buffer, once with 1% formic acid, and then eluted from the IMAC beads by three incubations with 500 mM K2HPO4, pH 7. 2398

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

of less than 1% and an individual posterior error probability of peptide assignments of less than 0.1.58 Peptides identified as SILAC pairs were automatically quantified, and position(s) of phosphorylated residues were determined by post-translational modification (PTM) scorebased localization in MaxQuant.36 Only phosphorylation sites assigned with a localization probability of at least 0.75 and a localization score difference of at least five were considered (class I phosphosites). In the case of singly and multiply phosphorylated peptides harboring the same site, ratios were calculated for singly phosphorylated species only. To identify phosphosites exhibiting significant quantitative differences between Ppt1-expressing and -deficient yeast cells, biological and technical reproducibility was assessed. For all phosphosites, the ratios of the individual ratios from both replicate experiments were calculated. These ratios-of-ratios were log2transformed and plotted as a histogram followed by Gaussian regression analysis using Sigmaplot (version 10.0, Systat Software Inc.). The obtained standard deviation across the whole quantitative replicate data was used to determine thresholds for significant changes of at least ±2.58 σ, corresponding to a chance of less than 1% that a given ratio is wrongly identified as regulated due to biological and/or technical variability. Furthermore, only phosphorylation sites matching these criteria in both biological replicates were considered as significantly different for further analysis. The same assessment was done for all protein ratios determined in both replicate analyses to define criteria for significant protein changes. For significantly regulated class I sites, annotated phosphopeptide spectra have been up-loaded to the Tranche file-sharing system (ProteomeCommons.org, Hash: BZG/sx0z/ lp34JUxgVUdD0FkL0HO90vmb+sBcM+NUMjvt+hiZe9viWXs0pDO4PNTjQoxcncBNf7zzmpQow3r2wIHKVQAAAAAAAAhbw==). Analysis of overrepresented Gene Ontology (GO) annotation terms was performed with the functional annotation tool DAVID.59,60 Phosphoproteins harboring up-regulated phosphorylation sites were compared to all S. cerevisiae entries within DAVID requiring an EASE score of less than 0.05 as criterion for significant enrichment of GO biological process terms.

Remaining salts were removed from C 18 StageTip-bound peptides by washing twice with 1% formic acid. Bound peptides were eluted with 50% ACN, 0.5% acetic acid, organic phase was removed, and samples were adjusted to 2% ACN, 0.2% TFA prior to analysis by mass spectrometry. For proteome analysis, the remaining 2% of the peptide mixtures derived from in-solution digests were separated by hydrophilic interaction chromatography (HILIC).54 Peptides were dissolved in buffer A (90% ACN, 0.1% TFA), loaded onto a PVA-Sil 150 mm × 2.1 mm column (YMC Europe), and fractionated over a gradient from 5% to 40% buffer B (5% ACN, 0.1% TFA). The resulting fractions were pooled to 12 samples according to the UV absorption at 215 nm and subjected to MS analysis. Mass Spectrometry Analysis

Tryptic peptides were separated by online reversed-phase nanoscale capillary liquid chromatography (nanoLC) coupled to a nanoelectrospray ion source (Proxeon Biosystems). Using the nanoLC system, samples were injected into a 15 cm reversed-phase, fused-silica capillary column (inner diameter 75 μm, packed in-house with 3 μm ReproSil-Pur C18-AQ media, Dr. Maisch GmbH) kept at 31 °C. Loaded peptides were eluted in 140 min LC−MS runs with the gradient ranging from 5 to 40% ACN in 0.5% acetic acid and a flow rate of 250 nL per min. Eluted peptides were directly electrosprayed into a LTQOrbitrap mass spectrometer (Thermo Fisher Scientific). Datadependent acquisition was performed on the LTQ-Orbitrap using the Xcalibur 2.0 software in the positive ion mode as described.37,55 Briefly, the instrument was recalibrated in realtime by coinjection of an internal standard from ambient air into the C-trap (“lock mass option”).37 Survey spectra were acquired in the orbitrap with a resolution of 60 000 at m/z 400. Up to five of the most intense multiply charged ions were sequentially isolated, fragmented, and analyzed in the LTQ part of the instrument. To improve phosphopeptide analysis, multistage activation was enabled and the neutral loss species at 97.97, 48.99, or 32.66 m/z below the precursor ion were activated for 30 ms during fragmentation (pseudo-MS3).56 All raw data files from this study have been up-loaded to the Tranche file-sharing system (ProteomeCommons.org, hash: f5TzHKgUCeBcvwIiDb5olQvbheTWYGGUdZoDK8l6o8lwNskeN8kdEBRty5hLcFXpG0CIkFjLaY1D1aWfvCr5K4rR3zwAAAAAAAAmeg==).



RESULTS AND DISCUSSION

Experimental Approach

Peptide Identification, Quantification, and Data Analysis

To create an experimental system for the unbiased in vivo analysis of Ppt1-mediated phosphoregulation, we first deleted the ppt1 gene in the previously described YAL6B strain of S. cerevisiae.50 YAL6B cells are auxotrophic for arginine and lysine due to disruptions of their arg4 and lys1 genes, which allows quantitative incorporation of these amino acids when supplied with the growth medium. We used SILAC61 to encode Ppt1 wild-type yeast cells with normal arginine and lysine (Arg0, Lys0) and Ppt1-deficient cells with the heavier isotopic variants Arg10 and Lys8. After labeling to completion cells were pooled in equal numbers and lysed. Extracted proteins were digested in-solution with the proteases Lys-C and trypsin. For phosphoproteome analysis, we performed SCX chromatography with 98% of the total peptide mixture. We collected 10 fractions from which phosphopeptides were further enriched by IMAC prior to their analysis by online LC−MS on a LTQOrbitrap hybrid mass spectrometer. We resolved the remaining part of the SILAC-encoded sample into 12 fractions by HILIC. The resulting total peptide samples were also analyzed by

Raw MS data were processed using the in-house software MaxQuant (version 1.0.13.12).44 MaxQuant-generated peak lists were searched with the Mascot search engine (version 2.2.04, Matrix Science, London, U.K.) against a curated yeast ORF SGD protein sequence database containing 6719 sequences and frequent contaminants such as human keratins, porcine trypsin, and endoproteinase Lys-C (175 sequences) to which reversed versions of all sequences had been added.57 The maximum mass deviations allowed for MS and MS2 peaks were 7 ppm and 0.5 Da, respectively. Carbamidomethylcysteine was set as fixed modification, and oxidized methionine, phosphorylation on serine, threonine, or tyrosine, and protein Nacetylation were searched as variable modifications. In addition, the SILAC labels Lys 8 and Arg 10 were searched as modifications. Full tryptic specificity was required and up to two missed cleavages were allowed. Searches against a concatenated target/decoy database allows to filter peptide and protein identifications for an estimated false-discovery rate 2399

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

LC−MS for in-depth proteome quantification in addition to quantitative phosphoproteome analysis. Furthermore, we employed the same combination of workflows in a biological replicate SILAC experiment with inverse labeling conditions (Figure 1). We then used the MaxQuant software to process

Figure 2. Quantitative proteome analysis. (A) Comparison of protein quantifications from the two independent SILAC experiments performed in this study. (B) Combined histogram and scatter plot analysis of protein quantifications in replicate experiments. Ratios within the gray boxes would represent reproducibly found and significant protein expression differences (p < 0.01).

influences of biological variability and experimental noise. For statistical evaluation, we considered all proteins quantified in both experiments and divided protein-specific ratios determined in the first by those measured in the second experiment followed by log2 transformation of the resulting quotients (Supplemental Figure 1A in the Supporting Information). As the overall distribution of these “ratios-of-ratios” reflects the interexperimental variability of our quantifications, we could then define thresholds of ±2.58 × σ to filter for proteins that were either significantly up- or down-regulated upon Ppt1 deficiency with greater than 99% certainty. Notably, none of the repeatedly quantified proteins satisfied these criteria of either more than 1.41-fold induction or 0.71-fold down-regulation in both replicate experiments (Figure 2B). Therefore, our data indicate no major role of Ppt1 in the regulation of transcriptional processes or the control of protein homeostasis under normal growth conditions. Ppt1 was quantified in just one of the two experiments, which was likely due to its rather low expression. Consequently, Ppt1 was not included in our overall assessment of quantitative protein data. However, the very low knockout versus wild-type cell ratio determined in our first experiment verified the absence of the phosphatase in ppt1Δ yeast cells in accordance with our genomic and immunoblot analyses (data not shown).

Figure 1. Experimental workflow. YAL6B cells expressing physiological levels of Ppt1 and phosphatase-deficient YAL6B ppt1Δ cells were SILAC-encoded with normal or isotopically labeled arginine and lysine. Two biological replicate experiments were performed with reciprocal labeling schemes. Upon lysis of equal numbers of pooled cells, proteins were digested in-solution with Lys-C followed by trypsin. After desalting, 98% of the peptide mixture was fractionated by SCX chromatography, followed by StageTip extraction and IMAC phosphopeptide enrichment prior to MS analysis. In parallel, 2% of the total peptides were fractionated by HILIC, desalted, and then directly subjected to MS analysis.

the 44 MS raw files acquired in the two replicate experiments.44 MaxQuant filtered all peptide and protein identifications according to a false-discovery rate of 1% and performed automated quantification of all identified peptides and proteins for which SILAC ratios could be determined. Protein Quantification

In total, we detected 3274 distinct yeast proteins in this study based on significant peptide and/or phosphopeptide identifications (Supplementary Table 1 in the Supporting Information). About 58% of all identified proteins could be quantified in their relative abundance in Ppt1-deficient versus wild-type cells in at least one of the two experiments due to quantified SILAC ratios of nonphosphorylated peptide species. Moreover, for 1524 proteins we quantified protein ratios in both biological replicate analyses (Figure 2A), and this quantitative data was used for further analysis as it allowed distinguishing reproducible differences in protein expression from potential

Quantitative Phosphoproteome Analysis

Because of phosphopeptide enrichment from SCX fractions and high-sensitivity high-accuracy LC−MS analysis, we could identify as many as 9747 distinct phosphopeptides that could be assigned to 2209 distinct S. cerevisiae proteins (Supplementary Table 2 in the Supporting Information). In the 2400

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

Figure 3. Quantitative phosphoproteome analysis. (A) Numbers of class I phosphorylation sites quantified in only one or both replicate experiments. (B) Numbers of proteins for which protein ratios, phosphorylation site ratios, or both were determined in this study. (C) Combined histogram and scatter plot representation of phosphorylation sites quantified in both replicate experiments. Consistently up-regulated sites are highlighted in red. Gray boxes indicate area of significant regulation (p < 0.01). (D) SILAC MS spectra of the phosphopeptides NVGDNGpSHELIVSSEDMK (upper spectra) and pSPILHNTGYEILGLPHKFDK (lower spectra) derived from the nucleotide exchange factor SIL1 and the pentafunctional AROM polypeptide, respectively. Switching of SILAC labels resulted in reciprocal ratios obtained in biological replicate experiments. Peptide species originating from either Ppt1-expressing or Ppt1-deficient cells are indicated.

identified and quantified phosphopeptides, 8100 distinct phosphorylation events could be localized to a specific residue

with high confidence (class I sites with localization probability ≥0.75) (Figure 3A and Supplementary Table 3 in the 2401

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

Table 1. Ppt1-Regulated Phosphorylation Sites ratio Ppt1 ko/wta protein

gene

protein ID

site

26S protease regulatory subunit 6A actin-related protein 5 asparagine synthetase [glutamine-hydrolyzing] 2 aspartyl-tRNA synthetase, cytoplasmic calcium/calmodulin-dependent protein kinase II DNA-directed RNA polymerase II subunit RPB2 E3 ubiquitin-protein ligase RSP5 eukaryotic translation initiation factor 5 fatty acid synthase subunit alpha fatty acid synthase subunit alpha fructose-bisphosphate aldolase glutamate synthase [NADH] glutamate synthase [NADH] glycogen debranching enzyme glycogen phosphorylase heat shock factor protein histone deacetylase RPD3 Hsp90 cochaperone Cdc37 inositol-3-phosphate synthase nucleotide exchange factor SIL1 pentafunctional AROM polypeptide PKHD-type hydroxylase TPA1 proteasome chaperone 1 protein ECM21 protein transport protein SEC24 ribonucleoside-diphosphate reductase large chain 1 trehalose synthase complex regulatory subunit TSL1 uncharacterized protein YPL071C V-type proton ATPase catalytic subunit A V-type proton ATPase catalytic subunit A V-type proton ATPase catalytic subunit A V-type proton ATPase catalytic subunit A V-type proton ATPase subunit B

RPT5 ARP5 ASN2 DPS1 CMK2 RPB2 RSP5 TIF5 FAS2 FAS2 FBA1 GLT1 GLT1 GDB1 GPH1 HSF1 RPD3 CDC37 INO1 SIL1 ARO1 TPA1 PBA1 ECM21 SEC24 RNR1 TSL1 YPL071C TFP1 TFP1 TFP1 TFP1 VMA2

YOR117W YNL059C YGR124W YLL018C YOL016C YOR151C YER125W YPR041W YPL231W YPL231W YKL060C YDL171C YDL171C YPR184W YPR160W YGL073W YNL330C YDR168W YJL153C YOL031C YDR127W YER049W YLR199C YBL101C YIL109C YER070W YML100W YPL071C YDL185W YDL185W YDL185W YDL185W YBR127C

S181 S38 T205 S502 S68 S156 T623 S268 S50 S920 S56 S612 S1267 T1265 T308 S478 T365 T198 T48 S107 S1321 S276 S82 S672 S78 S635 S1093 S21 T89 S94 T131 S858 S137

sequence KPTETY SAYNPQ EKRIPS MKAHGL RQARKL ELIAEE RFGEVV DKENLP VVEIGP VELCQK AARDSK DRLLLK AINIDL SVGIPG LWQARP PNNDDN NMFNVN AKPLEA SGRFDV NVGDNG PIGHSR TLAQIE TTLTVK RKERSV SELSNA YSRRVL ELFTII IRKRNH GDPVLR RTGKPL IPRGID KAVALG YLDING

S S T S S S T S S S S S S T T S T T T S S S S S S S S S T S T S S

DVGGLD PIAIDF PVDYHA PEDPGL TNEDVA EDDSES VDLKPD DVELYK PTLAGM PVMADL PIILQT PILHWN PILTPA PRDGAA TEFDFA IDTAST PEYLDK PSEALS PTVQDY HELIVS PILHNT NVLEDF PSLPQS LFEIRT PDYIRS GEFQVV RIIED_ PDPIGI GKPLSV VELGPG PALDRT PDRTGS PINPYA

exp. 1

exp. 2

2.46 (0.99) 3.45 (0.95) 2.26 (1.03) 1.99 (1.04) 2.49 (1.15) 3.11 (0.97) 2.68 (1.00) 2.24 (1.01) 1.94 (1.05) 2.91 (1.05) 2.04 (1.03) 6.73 (1.17) 2.66 (1.17) 2.53 (0.93) 2.26 (1.08) 8.13 2.89 (0.87) 5.4 (1.00) 2.25 (1.03) 6.84 (0.89) 3.27 (1.02) 2.23 (0.98) 7.34 (1.03) 2.53 5.56 (0.98) 2.53 (0.98) 2.42 (0.97) 2.08 6.61 (1.10) 7.22 (1.10) 3.51 (1.10) 2.73 (1.10) 2.94 (1.10)

2.13 (0.95) 3.37 (0.98) 2.08 (1.05) 2.3 (1.02) 1.93 (1.14) 2.64 (1.04) 2.81 (1.12) 1.93 (1.03) 2.62 (0.97) 3.2 (0.97) 2.33 (1.01) 9.09 (1.10) 3.75 (1.10) 3.13 (1.38) 2.01 (1.44) 11.28 13.83 (1.06) 5.43 (1.01) 2.42 (1.16) 8.01 (0.95) 4.52 (1.03) 3.8 (0.93) 7.76 (1.14) 4.21 (1.21) 7.95 (1.03) 3.31 (1.10) 2.53 (1.26) 2.22 5.61 (1.05) 5.9 (1.05) 3.28 (1.05) 2.29 (1.05) 3.16 (1.03)

a

Measured phosphorylation site ratios are indicated for both replicate experiments. The corresponding protein ratios are shown in parentheses when determined in the parallel quantitative proteome analysis.

Supporting Information).36,55 Class I phosphorylation site ratios could be quantified on 1947 distinct phosphoproteins, about half of which were quantified in their relative protein abundance in Ppt1 wild-type versus knockout yeast (Figure 3B). Moreover, 5215 distinct class I phosphorylation sites were quantified in both replicate experiments (Figure 3A). This large overlap fraction indicates high comprehensiveness of the individual experiments and, importantly, allowed for assessing the reproducibility of SILAC-based quantifications for the major part of the detected yeast phosphoproteome. To identify significantly regulated phosphorylation sites, we applied the same strategy as for protein expression analysis. We considered phosphorylation site ratios without correction for relative protein abundance as no significant, reproducible changes were found on the protein level (Figure 2B), resulting in very similar ratios with or without normalization for protein abundance (Supplementary Figure 2 in the Supporting Information). Statistical assessment of biological reproducibility provided ratio thresholds of 1.87 for up- and 0.53 for down-regulated phosphorylation sites in our analysis (Supplementary Figure 1B in the Supporting Information). We then examined which site-specific phosphorylations were significantly different according to these criteria based on consistent Ppt1

knockout versus wild-type cell ratios in both replicate analyses. As seen in Figure 3C, the vast majority of quantified sites exhibited rather similar phosphorylation in the compared yeast strains. Strikingly, against this vast background of similar phosphorylations we found 33 sites that were reproducibly up-regulated in Ppt1deficient cells while not a single site showed consistent downregulation (Table 1 and Supplementary Table 3 in the Supporting Information). Biological reproducibility is exemplified by MS spectra of the phosphopeptides NVGDNGpSHELIVSSEDMK and pSPILHNTGYEILGLPHKFDK derived from the protein nucleotide exchange factor SIL1 and pentafunctional AROM polypeptide, respectively (Figure 3D). The exclusive detection of sites that were induced in Ppt1 knockout cells is consistent with preferential detection of direct cellular Ppt1 substrate proteins because cellular absence of Ppt1 is expected to induce site-specific phosphorylations that are directly targeted of its phosphatase activity. Furthermore, since less than 1% of all quantified sites were induced in ppt1Δ yeast cells, our results characterize Ppt1 as a rather selective phosphatase in vivo. These results contrast data from a proteomics study by Ham et al., in which the human Ppt1 homologue PP5 was overexpressed as either wild-type enzyme or catalytically inactive variant in human cells exposed to DNA-damaging agent.47 In this study, 122 and 227 distinct 2402

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

Figure 4. Site-specific regulation of phosphoproteins. (A) Enrichment of phosphorylation motifs within regulated phosphorylation sites. Lack of cellular Ppt1 expression preferentially induces phosphorylation at pSP and pTP motifs compared to all quantified phosphorylation sites. Statistical significance according to the χ2 test is indicated by asterisks (* p < 0.05, *** p < 0.001). (B) Diagram showing the fraction of consistently upregulated phosphorylation sites among all sites quantified in both replicate analyses. Phosphoproteins containing more than one regulated phosphorylation site are highlighted in red. Protein identities are indicated by the corresponding gene names.

kinase/phosphatase knockout analyses by Bodenmiller et al. and for none of these regulation in ppt1Δ yeast was reported.48 This indicates that our experimental approach including SCX fractionation prior to phosphopeptide affinity purification allows for more sensitive and comprehensive analysis of individual phosphoproteomes.

phosphopeptides were detected in two biological replicates indicating rather modest coverage of the phosphoproteome. Label-free quantification revealed differential regulation of 15 phosphopeptides. Such a far larger proportion of regulated sites in this previous data might be due to phosphatase overexpression instead of ablation of endogenous expression as performed in our study. Moreover, both up- and downregulated phosphorylations were reported in the PP5 study while we found exclusively up-regulated sites in cells devoid of Ppt1 activity.47 We compared our results with data from the recently published screen of yeast kinase and phosphatase perturbations by Bodenmiller et al.48 In this study, 16 up- and 11 downregulated phosphopeptides were reported based on label-free quantification of ppt1Δ versus wild-type yeast. For all of them we identified and quantified phosphopeptides in our study that encompassed the reported modification sites (Supplementary Table 4 in the Supporting Information). However, for none of them did we detect reproducible regulation in our SILAC-based study. This might be in part due to biological reasons as we performed our analyses on yeast cells in the late-log phase, whereas Bodenmiller et al. harvested in the midlog phase.48 Interestingly, 12 out of the 16 phosphopeptides reported as induced in the previous study were derived from the protein HBT1. We measured a 2-fold increased level of HBT1 in ppt1Δ cells in our second replicate while no such change was detected in the first experiment. Although this protein did not meet our stringent criteria for biologically reproducible regulation, our data points to some biological variability of its cellular expression, which might underlie the previously observed phosphopeptide changes.48 More generally, the discrepant quantitative results might warrant further independent evaluation of both data sets, for which we not only provide a complete list of all peptide identifications (Supplementary Table 2 in the Supporting Information) but have further up-loaded all our MS raw data files to the Tranche file-sharing platform for future analyses. Furthermore, only 5 out of the 33 phosphorylation sites we found reproducibly induced in Ppt1-deficient cells were matched by phosphopeptides recorded in any of the

Analysis of Ppt1-Regulated Proteins

Initial inspection revealed that 19 out of the 33 Ppt1-regulated sites harbored a proline residue in the +1 position. We therefore compared all regulated phosphorylation sites with all that were quantified in both experiments. Notably, statistical evaluation confirmed significant enrichment of Ser-Pro and Thr-Pro motifs in the regulated data set (Figure 4A). We conclude that proline next to a phosphorylated serine or threonine residue might favor Ppt1-mediated dephosphorylation in vivo. However, this sequence feature likely represents one among various other factors that determine Ppt1 substrate recognition, considering that the vast majority of cellular SerPro and Thr-Pro sites did not change despite the enrichment of these motifs among Ppt1-regulated phosphorylation sites. Further analysis of Ppt1-regulated phosphoproteins revealed regulation of individual sites in the context of varying numbers of phosphorylation sites that were not affected (Figure 4B). Importantly, this identifies Ppt1 as a phosphorylation sitespecific Ser/Thr phosphatase in vivo. To gain insights into possible cellular Ppt1 functions, we performed GO analysis to identify over-represented biological processes in the group of Ppt1-modified proteins. Interestingly, despite the rather small number of Ppt1 targets, GO analysis revealed statistically relevant (p < 0.05) GO terms evident of Ppt1 functions in the heat shock response and biosynthetic processes (Figure 5A and Supplementary Table 5 in the Supporting Information). Preferential regulation of components of the heat shock response is concordant with described Ppt1 functions as a modulator of the Hsp90 chaperone machinery.23,62 Previous studies of yeast Ppt123 and its C. elegans (PPh5)63 and human (PP5) homologues18,26,27 revealed that this phosphatase is part of a large Hsp90-based chaperone complex that may function to assist the conformational maturation of many substrate proteins including protein 2403

dx.doi.org/10.1021/pr201134p | J. Proteome Res. 2012, 11, 2397−2408

Journal of Proteome Research

Article

Figure 5. (A) GO analysis of Ppt1-regulated phosphoproteins. Proteins identified with up-regulated phosphosites in ppt1Δ cells were compared to all entries for S. cerevisiae using the DAVID functional classification tool. Fold enrichment is shown for significantly overrepresented GO biological process terms (EASE score