Global Proteomic Assessment of the Classical Protein-Tyrosine Phosphatome and ``Redoxome''

Resource

Global Proteomic Assessment of the Classical Protein-Tyrosine Phosphatome and ‘‘Redoxome’’

Robert Karisch,1,2,* Minerva Fernandez,2 Paul Taylor,3 Carl Virtanen,2 Jonathan R. St-Germain,3 Lily L. Jin,3 Isaac S. Harris,2 Jun Mori,4 Tak W. Mak,2 Yotis A. Senis,4 Arne O¨ stman,5 Michael F. Moran,3 and Benjamin G. Neel1,2 1Department of Medical Biophysics, University of Toronto, Toronto M5G 2M9, ON, Canada 2Campbell Family Cancer Research Institute, Ontario Cancer Institute and Princess Margaret Hospital, University Health Network, Toronto, ON M5G 1L7, Canada 3Program in Molecular Structure and Function, Hospital For Sick Children, and Department of Molecular Genetics and McLaughlin Centre for Molecular Medicine and Banting and Best Department of Medical Research, University of Toronto, Toronto, ON M5G 1L7, Canada 4Centre for Cardiovascular Sciences, Institute of Biomedical Research, School of Clinical and Experimental Medicine, College of Medical and Dental Sciences, University of Birmingham, Birmingham B15 2TT, UK 5Cancer Center Karolinska, Department of Pathology and Oncology, Karolinska Institute, Stockholm 17176, Sweden *Correspondence: [email protected] DOI 10.1016/j.cell.2011.07.020

SUMMARY their regulation. Most activated PTKs are phosphorylated on tyrosine and can be identified by phosphotyrosine (pY)-proteo- Protein-tyrosine phosphatases (PTPs), along with mics (Rikova et al., 2007; Zhang et al., 2005). By contrast, pro- protein-tyrosine kinases, play key roles in cellular teomic approaches to assess PTP expression and regulation signaling. All Class I PTPs contain an essential active have lagged behind. site cysteinyl residue, which executes a nucleophilic The PTP superfamily comprises 107 genes, divided into four attack on substrate phosphotyrosyl residues. The families (Alonso et al., 2004). The largest, the class I cysteine- high reactivity of the catalytic cysteine also predis- based PTPs, can be subdivided into classical and dual-specificity PTPs (DUSPs). The classical PTP subfamily consists of poses PTPs to oxidation by reactive oxygen species, 21 receptor and 17 nonreceptor PTPs that exclusively hydrolyze such as H2O2. Reversible PTP oxidation is emerging pY residues. Classical PTPs contain a conserved ‘‘signature as an important cellular regulatory mechanism and motif,’’ [I/V]HCSXGXGR[S/T]G, within their active sites, wherein might contribute to diseases such as cancer. We ex- the invariant cysteine is essential for catalysis (Andersen et al., ploited these unique features of PTP enzymology to 2001). Owing to its low pKa (4.5–5.5), the catalytic cysteine exists develop proteomic methods, broadly applicable to in the thiolate (S-) state at physiological pH (Zhang et al., 1994). cell and tissue samples, that enable the comprehen- This property facilitates nucleophilic attack on substrate phos- sive identification and quantification of expressed photyrosines (Denu and Dixon, 1998), but also renders PTPs classical PTPs (PTPome) and the oxidized subset highly susceptible to oxidation by reactive oxygen species of the PTPome (oxPTPome). We find that mouse (ROS) (Salmeen and Barford, 2005; Tanner et al., 2010). and human cells and tissues, including cancer cells, Although high levels of ROS damage cellular components, ROS, particularly H O , also function as intracellular second display distinctive PTPomes and oxPTPomes, 2 2 messengers [reviewed in (den Hertog et al., 2005; Rhee et al., revealing additional levels of complexity in the regu- 2000; Tonks, 2005)]. Early work showed that receptor tyrosine lation of protein-tyrosine phosphorylation in normal kinase (RTK) activation leads to transient H2O2 production, and malignant cells. which is required for full RTK phosphorylation and downstream signaling (Sundaresan et al., 1995). PTPs were proposed as INTRODUCTION important ROS targets (Finkel, 1998), and soon PTPN1 (PTP1B) was found to be reversibly oxidized and inactivated in The phosphorylation of proteins on tyrosyl residues is controlled response to EGFR activation (Lee et al., 1998). Subsequently, by protein-tyrosine kinases (PTKs) (Lemmon and Schlessinger, studies have reported that oxidation of specific PTPs, including 2010) and protein-tyrosine phosphatases (PTPs) (Alonso et al., PTPN11 (SHP2) in PDGF signaling (Meng et al., 2002), PTPN1 2004), which direct diverse cellular processes, such as cell and PTPN2 (TC-PTP) in insulin signaling (Meng et al., 2004), growth, proliferation, and migration (Tonks, 2006). Aberrations and PTPN6 (SHP1) in B cell receptor signaling (Singh et al., in PTKs or PTPs contribute to several human diseases, including 2005), is required for optimal responses to various stimuli. cancer (Julien et al., 2011; Lahiry et al., 2010). A systems level Cancer cells often produce high levels of ROS (for review, see understanding of cell signaling requires methods to identify Cairns et al., 2011; Liou and Storz, 2010), which could decrease PTKs and PTPs, determine their expression levels, and monitor basal PTP activity and enhance tyrosyl phosphorylation (Ostman

826 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. et al., 2006). Indeed, total PTP activity is decreased in BCR/ABL presence of DTT to reduce PTPs to the active (PTP-S-) state.

(Sattler et al., 2000) and Src (Gianni et al., 2008) transformed Then, PTPs are ‘‘hyperoxidized’’ to the SO3H state and detected cells and can be restored by antioxidants or PTK inhibitors. by immunoblotting with the oxPTP Ab or processed for mass Furthermore, overexpression of the ROS-producing protein spectrometry (MS). For MS, PV-treated lysates are digested

Nox1 can cause cellular transformation and tumor formation with trypsin, and PTP-SO3H active site peptides are purified (Arnold et al., 2001), implicating excess ROS production in with the oxPTP Ab and identified (‘‘discovery’’ proteomics) by carcinogenesis. LC-MS/MS. Quantification can be achieved by various methods, The identification of ROS-inactivated PTPs might provide including labeling with stable isotopes and/or chemical tags (for important clues to key PTPs that regulate normal and oncogenic review, see Gstaiger and Aebersold, 2009; Walther and Mann, signaling pathways. Yet while approaches to identify specific 2010; Yates et al., 2009) or, as follows, by selected reaction oxidized PTPs are available, it has been difficult to develop general monitoring (SRM, also known as multiple reaction monitoring methods that detect and quantify any reversibly oxidized PTP in or MRM) (for review, see Elschenbroich and Kislinger, 2011). different physiological and pathological contexts. A modified in- gel phosphatase assay is currently the best global assay for PTP qPTPome Reliably Identifies Classical PTPs oxidation (Meng et al., 2002), but this approach is biased toward For qPTPome, NIH 3T3 cells were lysed in the presence of DTT, nonreceptor PTPs (Burridge and Nelson, 1995), is not quantitative, followed by a buffer exchange and treatment with PV. Multiple and requires the oxidized PTP(s) to be identified by immunodeple- bands were detected in immunoblots with the oxPTP Ab in PV- tion. We exploited the biochemistry of PTP catalysis and oxidation treated, compared with untreated, lysates (Figure 1B). Next, and an available monoclonal antibody against ‘‘hyperoxidized’’ native or denatured PV-treated lysates were immunoprecipi- PTP active sites to develop global quantitative proteomic ap- tated and immunoblotted with the oxPTP Ab (Figure 1C). Again, proaches that monitor classical PTP expression and oxidation. new bands were detected in the PV-treated lysates (Figure 1C, compare lanes 5 and 6). Denaturing the lysate prior to immuno- RESULTS precipitation increased recovery of oxidized PTPs (Figure 1C, compare lanes 4 and 6), suggesting that the epitope for the

Design of qPTPome and q-oxPTPome Assays oxPTP Ab is not very accessible in the native PTP-SO3H confor- H2O2 promotes reversible oxidation of PTPs to the sulfenic acid mation. As expected, NEM treatment prior to reduction markedly (PTP-SOH) state (Rhee et al., 2000), which is labile and, in decreased oxPTP Ab-immunoreactive bands (Figure 1C, lanes 3 different family members, rapidly rearranges to form a sulfenyla- and 5). We also immunoblotted oxPTP and anti-Ptpn11 immuno- mide with the adjacent main-chain nitrogen (Salmeen et al., precipitates from denatured lysates with anti-Ptpn11 antibodies. 2003; van Montfort et al., 2003) or a disulfide bond to a nearby Similar levels of Ptpn11 were recovered in each immunoprecip- cysteine (Chen et al., 2009). The sulfenylamide and disulfide itate, suggesting that immunoprecipitation by the oxPTP Ab was states help limit ‘‘hyperoxidation’’ to the sulfinic (PTP-SO2H) quantitative (Figure 1C, compare lanes 2 and 6). and sulfonic (PTP-SO3H) acid states, which are biologically irre- To identify the PV-inducible species, cell extracts were di- versible (Tonks, 2005). gested with trypsin, and peptides immunopurified with the We exploited these biochemical properties to develop oxPTP Ab were analyzed by LC-MS/MS. These experiments methods for quantifying classical PTP expression (qPTPome) revealed active site PTP-SO3H peptides from 17 classical and oxidation (q-oxPTPome), respectively (Figure 1A and Fig- PTPs (Figure 1D). Importantly, unlike the in-gel PTPase approach ure S1 available online). Both methods use a monoclonal anti- (see Introduction), qPTPome detected the D1 and D2 domains body (oxPTP Ab) raised against the signature motif of PTPN1 from several RPTPs (Ptpra, Ptpre, Ptprf, and Ptprs). Near- oxidized to the sulfonic acid (VHCSO3HSAG) (Persson et al., complete overlap in PTP expression was observed in biological 2005) to isolate ‘‘hyperoxidized’’ PTPs (PTP-SO3H). This anti- replicates using NIH 3T3 cells (Figure 1E) and other cell lines and body has been used for immunoblotting of cell lysates and to tissues (see Figure S2A). monitor oxidation of specific PTPs (Groen et al., 2005; Persson To determine the range of classical PTPs that could be et al., 2004; Weibrecht et al., 2007). Its precise reactivity had detected, we performed qPTPome on tissues from C57BL/6J not been defined, but because the signature motif is highly mice (Figure 2A Table S1, and Table S2), several human cell lines conserved, we suspected (and found; see below) that the oxPTP (Figure 2B, and Table S1, and Table S2) and the rat basophilic Ab would react with most, if not all, classical PTPs. leukemia line RBL2H3 (Figure S2B, Table S1, and Table S2). Basal or ligand-induced PTP oxidation results in two pools of Nearly all classical PTPs (17/17 nonreceptor and 19/21 RPTPs) PTPs: oxidized (PTP-SOH; inactive) and reduced (PTP-S-; were identified, and PTPs accounted for 70% of the Cys- active). For q-oxPTPome, active PTPs are alkylated with N-ethyl- SO3H-containing proteins recovered in these experiments (Table maleimide (NEM; PTP-S-NEM), rendering them resistant to S1). Notably, members of other phosphatase families (e.g., further modification, while oxidized PTPs remain unaffected (Fig- DUSPs) were not detected, indicating that the oxPTP Ab is highly ure 1A). Excess NEM is removed by gel filtration, and reversibly specific for the classical PTP signature motif. Other recovered oxidized PTPs are reduced with dithiothreitol (DTT). Following peptides had sequence similarity to this motif (see below) or a buffer exchange, the reduced PTPs (representing PTPs that were from abundant proteins (Table S1). initially were reversibly oxidized), are oxidized to the sulfonic Each cell line and tissue had a unique PTP profile. For acid (PTP-SO3H) using pervanadate (PV) (Huyer et al., 1997). example, PTPN1 and PTPN2 were expressed ubiquitously, For qPTPome, alkylation is omitted, and cells are lysed in the whereas PTPRC and PTPRD were present more selectively.

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 827 A

B CD Protein MW (kDa) Oxidized Active Site Ptpn1 50 Yes Ptpn2 47 Yes Ptpn3 104 Yes Ptpn4 106 Yes Ptpn6 68 Yes Ptpn9 68 Yes Ptpn11 68 Yes Ptpn12 87 Yes Ptpn13 270 Yes Ptpn14 135 Yes E Ptpn21 133 Yes Ptpn23 185 Yes Ptpra 94 Yes (D1 & D2) Ptpre 83 Yes (D1 & D2) Ptprf 211 Yes (D1 & D2) Ptprk 165 Yes Ptprs 168 Yes (D1 & D2)

Figure 1. Development of qPTPome and q-oxPTPome (A) Scheme for each assay. (B) NIH 3T3 cells were lysed in presence of DTT, applied to a gel ﬁltration column, treated with PV or left untreated, and then analyzed by immunoblotting with the oxPTP Ab. (C) Native and denatured PV-treated lysates from (B) were immunoprecipitated with oxPTP or Ptpn11 antibodies, as indicated, and analyzed by immunoblotting. (D) PTPs in NIH 3T3 cells, as identiﬁed by qPTPome. (E) Venn diagram demonstrating reproducibility of the qPTPome assay (n = 3). See Figure S1 for additional variations of these assays.

There was near-complete overlap between PTPs identified by that posttranscriptional mechanisms contribute significantly to qPTPome and in published microarray analyses (Figure 2C and determining PTP protein levels. Table S1), providing further evidence of the reliability of our We also applied our method to a cell type with an unknown assay. This correlation was not absolute, though, indicating PTP expression profile. Platelets are anuclear and express only

828 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. Figure 2. qPTPome Broadly Identifies Classical PTPs in Cell Lines and Tissues (A and B) The indicated mouse (A) and human (B) cell lines and tissues were profiled by qPTPome. Classical PTPs identified are shown as black boxes. (C) Comparison of qPTPome results (blue boxes) with published microarray data (black boxes). Correlations for each PTP were evaluated by Spearman r coefficient, as represented by the black-red scale. (D) PTPs in human platelets, as identified by qPTPome. (E) Multiple sequence alignment showing sequence of tryptic fragments containing the catalytic cysteines of each classical PTP. PTPs identified by qPTPome are bolded. Also, see Figure S2 and Tables S1 and S2. small amounts of residual megakaryocyte mRNA, making it including PTPN2, PTPRA, and PTPRE, which had not been difficult to define their PTPome by RNA-based approaches. identified previously. When antibodies were available, these Using qPTPome, we identified 16 PTPs in platelets (Figure 2D), identifications were confirmed by immunoblotting (Figure S2C).

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 829 The above data set allowed us to define the specificity of the immunoblot using the oxPTP Ab (Figure 3F) and with RT-PCR oxPTP Ab. We aligned the sequences of the tryptic peptides measurements of Ptprd mRNA (Figure 3G). Also consistent from the active sites of all classical PTPs, highlighting those de- with the oxPTP immunoblot, there were no significant changes tected by qPTPome (Figure 2E and Figure S2D). Although the D1 in classical PTP expression aside from Ptprd in these samples domains of Ptprc, Ptprm, Ptprk, Ptprt, and Ptpru were identified, (Figure 3H and Table S3). their D2 domains, which cluster at the top of the multiple As implemented above, qPTPome displays the relative level of sequence alignment, were not. The active site sequences of each classical PTP. We asked if, by combining qPTPome with these D2 domains (VHCRDG and [I/V]HCLNG) diverge substan- absolute quantification (AQUA) (Gerber et al., 2003), the absolute tially from the sequence used to generate the oxPTP Ab. To ask if amount of a PTP could be determined (Figure S1). We synthe- the oxPTP Ab could recognize these sequences, we performed sized a Ptpn11-SO3H active site tryptic peptide containing a 13 15 immunoprecipitations with synthetic PTP-SO3H peptides. As ex- ‘‘heavy’’ arginine (six C and four N). Using a standard curve pected, the oxPTP Ab isolated the Ptpn1 peptide, but not the D2 generated with this peptide (Figure 3I), we found that our SRM domains of Ptprk, Ptprm, Ptprt, and Ptpru (data not shown). assay is linear over at least three orders of magnitude. By com- These experiments suggested that the oxPTP Ab recognizes paring the intensities of coeluting ‘‘heavy’’ (AQUA) and ‘‘light’’

[I/V]HCSO3HS[A/S]G, and that, given their divergent active sites, (endogenous) Ptpn11-SO3H peaks recovered by the oxPTP Ab Ptprn (VHCSDG) and Ptprn2 (VHCSDG) also would not be (Figure 3J) and assuming complete digestion with trypsin, we detectable. Notably, Ptprn and Ptprn2 are catalytically inactive determined that there are 32,000 copies of Ptpn11 in each (Gross et al., 2002); hence, qPTPome can identify all catalytically NIH 3T3 cell. This measurement compares well (within 4-fold) active classical PTPs in a single MS experiment. with immunoblot determinations (Figure S3D).

Quantification of the PTPome Quantification of the PTP ‘‘Redoxome’’ Selected reaction monitoring (SRM) is a label-free method that Next, we attempted to quantify PTP oxidation (Figure 4A). NIH determines the relative abundance of peptide ions by measuring 3T3 cells were stimulated with increasing concentrations of signals associated with specifically defined, distinctive, coelut- H2O2 for 5 min and analyzed by q-oxPTPome. In parallel, ing parent ion-to-product ion transitions (Picotti et al., 2009). qPTPome measurements were obtained to assess the stoichi- We developed a multiplexed SRM (mSRM) protocol to quantify ometry of PTP oxidation. Under these conditions, the Ptpn6 the classical PTPome or oxPTPome from human, mouse, or rat and Ptpn11 SRM profiles showed dose-dependent, but differen- (Figure S3A). Following enrichment by the oxPTP Ab, PTP tial increases in oxidation (Figure 4A), which were confirmed by peptide ions and their product ion fragments first were observed immunoblotting (Figure 4B). We used a heat map to display the in ‘‘discovery mode’’ LC-MS/MS, and then SRM transitions for oxidized proportion of each PTP relative to its total amount each PTP were designed, optimized, and combined into a single (q-oxPTPome/qPTPome; Figure 4C and Table S4). As expected, mSRM protocol (Table S3). The optimized mSRM transitions under normal growth conditions, most PTPs had very low levels were validated for several (50%) mouse and human PTPs using of basal oxidation to the PTP-SOH or SO3H states. By contrast, synthetic peptides (Table S3). Furthermore, all active site Ptpn23, Ptpra D2, Ptprf D2, and Ptprk D1 had high basal oxida- peptides from classical PTPs in NIH 3T3 (Figure S3B) and tion. With the exception of Ptprk D1, these PTP domains report- A431 (Figure S3C) cells were immunodepleted by the oxPTP edly are catalytically inactive (Gingras et al., 2009; Persson et al., Ab, demonstrating that the assay is quantitative. 2004). The D1 and D2 domains of specific RPTPs showed clear Next, we assayed Ptpn11fl/KO fibroblasts (Figure 3A), in differences in oxidation sensitivity. For example, Ptpra D1 had which 4-hydroxytamoxifen (4-OHT) addition activates a Cre-ER a low level of oxidation, while its D2 domain was highly oxidized. fusion protein and evokes deletion of a floxed Ptpn11 allele By contrast, Ptprs D1 and D2 both showed low basal oxidation. KO/KO (Ptpn11 ). Notably, the Ptpn11 signal from 4-OHT-treated As expected, increasing the H2O2 concentration increased the cells was much lower than from controls, whereas their Ptpn1 oxidation of multiple PTPs, including Ptpn1, Ptpn3, Ptpn9, and signals were similar (Figure 3B). Moreover, the SRM signals Ptpra D1 (Figure 4C). Surprisingly, though, Ptpn23 and Ptpra paralleled the respective Ptpn11 and Ptpn1 levels detected by D2 oxidation appeared to decrease. immunoblotting (Figure 3C). To display the mSRM signals for We suspected that this might reflect oxidation to the SO2H all PTPs simultaneously, we developed a heat map representa- state, which should not be converted to the SO3H state by PV tion (Figure 3D and Table S3), in which the relative abundance (Huyer et al., 1997) and hence would not be recognized by of each PTP is shown in shades of red and the magnitude of SRM. To test this possibility, we modified the q-oxPTPome pro- the SRM signal for each PTP (from control cells) is indicated by tocol (Figure 4D; see Experimental Procedures for details) to shades of blue (black denotes no signal/undetectable expres- enable identification and quantification of all PTP oxidation - sion). With the exception of Ptpn11, PTP expression was largely states (S , SOH, SO2H, SO3H). Not surprisingly, most PTPs unaffected by 4-OHT treatment. were oxidized by 10 mM H2O2 (Figure 4E and Table S4). Yet while To see if qPTPome could be used for tissues, we assayed several PTPs were converted to the SO2H state, including +/+ +/À À/À brains from Ptprd , Ptprd , and Ptprd mice (Mizuno Ptpn23 and Ptpra D2, very few were converted to the SO3H state. et al., 1993). As expected, Ptprd levels, obtained by SRM of Hence, extremely high levels of oxidative stress are required to the Ptprd D1 domain, were progressively lower in Ptprd+/+, completely oxidize PTPs inside cells. Most other PTPs, including Ptprd+/À, and PtprdÀ/À brains (Figure 3E); by contrast, Ptpn1 ex- Ptpn11 and Ptprs D1, showed high levels of reversible oxidation pression remained similar. These findings agreed with a global (PTP-SOH), whereas a few remained reduced.

830 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. Ptpn11 Ptpn1 A B (-)4-OHT

(+)4-OHT

C D E Ptprd Ptpn1

Ptprd+/+

Ptprd+/-

Ptprd-/-

I H

J Ptpn11 G Overlap

NIH3T3

AQUA Peptide

Figure 3. PTP Quantification Using qPTPome (A) Ptpn11fl/KO cells stably expressing CreER were treated with 4-OHT or left untreated, then assessed by qPTPome. (B) Representative SRM profiles for Ptpn11 and Ptpn1. Quantification of each PTP is displayed next to its elution profile (***p < 0.002, paired Student’s t test; two biological replicates, each with two technical replicates). (C) Immunoblot showing PTP expression levels. Arrow shows position of Ptpn11. (D) Heat map showing relative changes (compared with untreated cells) in PTP expression (measured by qPTPome) following 4-OHT treatment. The intensity value is the average signal for the SRM profiles for each PTP in untreated cells. (E) Ptprd+/+, Ptprd+/À, and PtprdÀ/À brains were profiled by qPTPome (n = 3; three technical replicates each). Representative SRM profiles are shown for Ptprd and Ptpn1, with quantification of each displayed next to the profiles. Data represent mean ± SEM (*p < 0.05; **p < 0.01; ***p < 0.001, ANOVA with Bonferroni post-test). (F) Immunoblot with oxPTP Ab showing PTP expression in Ptprd mutant brains. Arrow shows position of Ptprd. (G) Ptprd mRNA levels (assessed by qPCR; n = 3). Data represent mean ± SEM (***p < 0.001, ANOVA with Bonferroni post-test). (H) Relative levels of PTP expression (measured by qPTPome) in Ptprd mutant (compared with Ptprd+/+) brains. (I) Standard curve of MS peak area for Ptpn11 AQUA peptide (n = 3). (J) SRM peak of the AQUA peptide mixed with native Ptpn11 peptide from NIH 3T3 cells. Also, see Figure S3 and Table S3.

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 831 Figure 4. Monitoring Multiple PTP Oxidation States by q-oxPTPome

(A) Experimental scheme (left). NIH 3T3 cells were stimulated with the indicated concentrations of H2O2 for 5 min and analyzed by q-oxPTPome (n = 2; four technical replicates each). Representative SRM proﬁles for Ptpn6 and Ptpn11 (right).

(B) Detection of Ptpn6 and Ptpn11 oxidation. Following H2O2 stimulation, cells were processed by q-oxPTPome, immunoprecipitated with Ptpn6 or Ptpn11 antibodies and analyzed by immunoblotting, as indicated.

832 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. We next assessed the changes in PTP oxidation caused by We then compared PTP oxidation in six HER2+ breast cancer exposure of NIH 3T3 cells to buthionine sulphoximine (BSO), cell lines. Although some PTPs were differentially oxidized which inhibits glutathione synthesis. As expected, intracellular (Figures 5E and 5F and Table S5), there was less heterogeneity ROS levels increased (Figure 4F), and concomitantly, several between these lines than between lines derived from different PTPs showed increased oxidation (Figure 4G and Table S4). types of cancer (compare Figures 5C and 5E). Also, while there Interestingly, although PTP oxidation levels were similar in BSO- were some general trends in relative oxidation levels, each cancer treated and H2O2 (1 mM)-stimulated cells, the profile of oxidized line contained a unique profile of oxidized PTPs. Notably, the ROS PTPs differed significantly (compare Figures 4C and 4G). levels in a given cell line and the extent of PTP oxidation did not correlate strongly (compare Figure 5 and Figures S4B and S4D). Quantifying PTP Nitrosylation To ask if PTP oxidation was controlled by a mutant PTK, we PTPs also can be regulated by reversible nitrosylation of their treated SrcY527F cells with the Src kinase inhibitor, PP2. Global active site cysteines. Targeted indirect approaches have been tyrosyl phosphorylation (Figure 6A) and ROS (Figure 6B) levels used to detect nitrosylation of a few PTPs, including PTPN1 decreased following PP2 treatment, accompanied by an overall and PTPN11 (Li and Whorton, 2003). To see if our method could decrease in PTP oxidation (Figure 6C and Table S6). Neverthe- monitor PTP nitrosylation, NIH 3T3 cells were exposed to three less, some PTPs, (e.g., Ptpn3 and Ptpra D2) remained oxidized different nitrosylating agents, processed and analyzed by following PP2 treatment. q-oxPTPome. We detected nitrosylation of several PTPs (Fig- ure 4H and Table S4), including Ptpn1 and Ptpn11 (Hsu and Combining qPTPome with Phosphotyrosine Proteomics Meng, 2010; Li and Whorton, 2003). Consistent with the q-oxPT- to Suggest PTP Substrates Pome signals arising from nitrosylation, not oxidation, SRM Finally, we asked if combining qPTPome with pY-proteomics signals for Ptpn23 and Ptpra D2 did not decrease (Figure 4H), could suggest PTP/substrate relationships. We used qPTPome and ROS levels remained unchanged (data not shown) in cells to quantify PTP expression in several multiple myeloma (MM) exposed to these agents. cell lines (Figure 7A and Table S7) and measured 93 unique pY sites in the same cells by pY-proteomics (Figure 7B and Application of qPTPome and q-oxPTPome Table S7). Using partial least-squares regression (PLSR) anal- to Cancer Cells ysis, we sought relationships between the levels of various ROS levels often increase in cancer (Liou and Storz, 2010). pY-peptides and the expression of specific PTPs (see Experi- Although increased oxidation of individual PTPs has been re- mental Procedures for details). The resultant model predicted ported in selected cancer cell lines (Lou et al., 2008), the global relative pY-peptide levels well (Figure 7C and Table S7), with effects of excess ROS on PTP oxidation have not been as- Component 1 accounting for 97% of the variance and Compo- sessed. As a proof of principle, we analyzed A431 (Figure 5A nent 2 explaining another 2%. PTPN1 levels correlated most and Table S5) and NIH 3T3 cells transformed with constitutively strongly with Component 1, with PTPN6 and PTPN11 also active Src (Src Y527F; Figure 5B and Table S5). Many PTPs were contributing significantly. PTPN7 explained most of the variation reversibly oxidized in both lines, with generally higher levels of in Component 2, and contributed to Component 1. oxidation in A431 cells. Notably, we detected 20% oxidation Overall, PTPN1, PTPN6, PTPN7, and PTPN11 levels could of PTPN1 in A431 cells, remarkably similar to the 25% oxidation explain the majority of the variation for many of the pY sites ana- determined using a targeted MS method (Lou et al., 2008). lyzed. Interestingly, the site best explained by the first two We compared PTP oxidation in A431 and Src Y527F cells with components was from TYK2 (Table S7), a known PTP1B sub- that in cell lines derived from several types of cancer (Figures 5C strate (Myers et al., 2001). The relative pY levels of the FGFR3 and 5D and Table S5). Global PTP oxidation was higher in Src activation loop phosphotyrosyl residue also were highly corre- Y527F cells, compared with nontransformed NIH 3T3 cells. lated with Components 1 and 2 (Table S7). Surprisingly, these A431 and H1975 cells showed the highest levels of PTP oxida- correlations were positive: peptide phosphorylation increased tion and the highest levels of intracellular ROS (Figures S4B as the PTP expression level increased (Table S7). Nevertheless, and S4C). Some PTP domains (e.g., PTPRA D2, PTPN23) were given the prominent role of PTPN1 in regulating tyrosyl phos- highly oxidized in all cancer cells tested; others (e.g., PTPN1, phorylation of several other RTKs (Bourdeau et al., 2005), we PTPN11) had lower and variable levels of basal oxidation in these tested the effects of a PTPN1 shRNA on one of these lines cell lines. (KMS11). Remarkably, PTPN1 depletion led to increased tyrosyl

(C) Heat map displaying levels of PTP oxidation following H2O2 stimulation, measured by q-oxPTPome. A black box indicates low levels of PTP oxidation; red boxes denote high levels. The intensity value is the average signal for the SRM proﬁles for each PTP measured by qPTPome. (D) Scheme showing modiﬁcations of qPTPome and q-oxPTPome to assess all potential PTP oxidation states (n = 2; each with three technical replicates); the asterisk indicates a buffer exchange. - (E) Fraction of each PTP in each potential oxidation state (i.e., S , SOH, SO2H, SO3H) following 10 mM H2O2 stimulation. (F) ROS levels in NIH 3T3 cells treated with 100 mM BSO (n = 6 ***p < 0.001, Student’s t test). Data represent mean ± SEM. (G) Levels of PTP oxidation in cells treated with 100 mM BSO or left untreated (n = 2; two technical replicates each). Asterisks indicate PTPs with increased oxidation following BSO treatment. (H) Levels of nitrosylation measured by q-oxPTPome following stimulation with S-nitroso-N-penicillamine (SNAP), S-nitrosoglutathione (GSNO), or S-nitro- socysteine (CSNO). Asterisks indicate PTPs that were nitrosylated. See also Table S4.

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 833 Figure 5. Cancer Cells Have Unique PTP Oxidation Proﬁles (A and B) Heat map showing basal PTP oxidation in A431 (A) and Src Y527F-transformed NIH 3T3 (B) cells, measured by q-oxPTPome (n = 2; two technical replicates each). ‘‘Negative control’’ represents cells that were lysed in the presence of DTT, treated with NEM followed by PV, and analyzed by LC-MS/SRM. (C) Basal PTP oxidation in cancer lines determined by q-oxPTPome (n = 2; two technical replicates each). (D) Same as (C), but with scale altered to show PTPs that were oxidized to lower extents.

834 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. Figure 6. Reversibility of PTP Oxidation in Cancer Cells following Oncogene Inhibition (A and B) (A) Antiphosphotyrosine immunoblot and (B) ROS levels of untransformed and Src Y527F cells treated with or without PP2, respectively. Data represent mean ± SEM (n = 3; ***p < 0.001, ANOVA with Bonferroni post-test). (C) Heat map showing effects of PP2 on levels of PTP oxidation in Src Y527F cells (n = 2; two technical replicates each). Asterisks indicate PTPs with decreased oxidation following PP2 treatment. See also Table S6. phosphorylation of FGFR3 and several other proteins, including approaches could monitor oxidation (and expression) of these one with the molecular weight of TYK2 (Figure 7D). These data (and other) cysteinyl proteins (Figure S1). suggest that, like TYK2, FGFR3 is probably a bona fide in vivo The lack of global methods to identify oxidized PTPs has target of PTPN1. hampered progress on redox regulation. Several other indirect approaches use an overall strategy similar to q-oxPTPome. DISCUSSION The modified in-gel PTPase assay (Meng et al., 2002) has been the most successful, but it relies on immunodepletion for PTP By using proteomic methods that enable comprehensive anal- identification, rarely detects RPTPs and is not quantitative. The ysis of the classical PTPome (Figure 1A), we identified the modified cysteinyl-labeling assay isolates oxidized PTPs using PTPs in multiple human, mouse, and rat cell lines and tissues iodoacetylpolyethylene oxide biotin (IAP-Biotin) (Boivin et al., (Figure 2), including platelets (Figure 2D), cells in which RNA- 2010) or activity-based probes, such as a-bromobenzylphosph- based approaches have been of limited benefit. Our studies onate biotin (BBP-Biotin) (Kumar et al., 2004). Direct approaches uncover significant posttranslational regulation of PTP levels using dimedone, which reacts selectively with SOH groups, (Figure 2C) and unanticipated complexity in PTP oxidation represent a promising alternative (Poole and Nelson, 2008). evoked by exogenous and endogenous ROS in normal and Biotin-coupled dimedone probes (Nelson et al., 2010) and dime- neoplastic cells (Figures 4–6). Such differential PTP oxidation done-specific antibodies (Seo and Carroll, 2009) are available. represents a new layer of complexity in signal transduction. However, none of these approaches have been developed Furthermore, combining our assay with pY-proteomics can into a global MS assay for PTP oxidation. Recently, a modified help to identify PTP substrates (Figure 7). IAP-Biotin probe was used to identify reactive cysteinyl residues We used SRM to quantify PTP expression because it is easily by MS (Weerapana et al., 2010), but only three PTP-derived applied to tissues and cell lines, but our assay also accommodates peptides were detected (from >1000 cysteine-containing other quantification methods (Yates et al., 2009). Additional varia- species), and none corresponded to active site peptides. The tions could permit simultaneous identification of posttranslational high reactivity of IAP-Biotin with any reactive S- probably in- modifications (Figure S1): e.g., the oxPTP Ab could be used to creases the background of abundant (non-PTP) proteins, which isolate full-length (denatured) PTPs, and phosphorylated tryptic also could be problematic for dimedone-based approaches. peptides could be recovered (with TiO2 or anti-pY antibodies), fol- These limitations are surmounted by q-oxPTPome, which uses lowed by MS. The oxPTP Ab does not detect DUSPs, but by using a highly specific antibody to purify PTPs. Notably, q-oxPTPome more degenerate anti-Cys-SO3H hapten antibodies, analogous cannot yet identify the PTPs oxidized upon growth factor

(E) Basal PTP oxidation in HER2+ breast cancer lines measured by q-oxPTPome (n = 2; two technical replicates each). (F) Same as (E), but with scale altered to show PTPs that were oxidized to lower extents. Also, see Figure S4 and Table S5.

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 835 Figure 7. Combining qPTPome and Phosphotyrosine Proteomics (A) Heat map showing relative PTP expression (based on z-score across all lines) measured by qPTPome in a panel of MM cell lines (n = 2, two technical replicates each). Black box indicates PTPs that were not identiﬁed in any line. (B) Relative phosphotyrosine peptide levels (based on z-score across all MM lines) in the same cells (n = 2). (C) Correlation loadings plot displaying results of the PLSR analysis. (D) Lysates from KMS11 cells stably infected with scrambled or PTPN1 shRNAs were immunoprecipitated with FGFR3 antibodies and immunoblotted with antiphosphotyrosine or -FGFR3 antibodies (top). Immunoblots of total cell lystate (TCL) with the indicated antibodies (bottom). See also Table S7.

836 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. stimulation, probably because only a small subset of PTPs is in- was induced by exposing cells for 30 min to 0.5 mM SNAP (Calbiochem) and activated (under steady state conditions) by growth factor- L-cysteine (Sigma), GSNO (Alexis Biochemicals) and L-cysteine or CSNO evoked ROS (e.g., see den Hertog et al., 2005; Rhee et al., (Sigma), each prepared as described (Li and Whorton, 2003). Src Y527F cells were provided by Dr. S. Courtneidge (Sanford-Burnham Medical Research 2000; Woo et al., 2010). Similar obstacles confronted initial phos- Institute, USA) and treated with 10 mM PP2 (Calbiochem) for 4 hr, when indi- phoproteomic (including pY-proteomic) methods, prior to cated. Generation of Ptpn11fl/KO cells and deletion of the floxed Ptpn11 allele secondary enrichment strategies (Ahn et al., 2007; Pandey are described in Supplemental Information. Lentiviruses expressing PTPN1 et al., 2002). shRNA or a nonsilencing control hairpin were generated by standard proce- Our results add to growing evidence that ROS production and dures and used to infect KMS11 cells. Human platelets were prepared by stan- PTP oxidation are highly controlled in cells (Li et al., 2006; Woo dard techniques (Pearce et al., 2004). For further details, see Supplemental Information. et al., 2010). First, because our assay is quantitative, we can provide an upper bound on growth factor-induced increases in Immunoblotting PTP oxidation (<5%). For such a small change to impact Lysates were prepared in RIPA buffer containing a protease and phosphatase signaling (see Introduction), local control would be essential. inhibitor cocktail, clarified, resolved by SDS-PAGE and analyzed by immuno- Consistent with in vitro studies (Groen et al., 2005; Weibrecht blotting. For details, see Supplemental Information. et al., 2007), we find that PTPs also differ in relative oxidation sensitivity in cells. Surprisingly, though, the pattern of oxidized Quantitative Real-Time PCR RNA from brains of 10-week-old Ptprd+/+, Ptprd+/À, and PtprdÀ/À mice PTPs in cells stimulated by exogenous H O differs from that 2 2 (C57BL/6J) was reverse transcribed and analyzed by Sybr green-based evoked by increasing endogenous H2O2 levels (Figure 4). qPCR (Sigma), according to the manufacturer’s instructions. For details, see Furthermore, different cancer cells have distinct PTP oxidation Supplemental Information. profiles (Figure 5) that do not simply correlate with global ROS levels (Figure S4), again consistent with the idea that local ROS ROS Levels alterations are critical for altering tyrosyl phosphorylation events. Cells (80% confluence) were serum-starved overnight and treated as indicated. Plates were washed with PBS (23), trypsinized, incubated for Interestingly, transformation can have fairly durable effects on 5 min with 300 nM DCF-DA (Invitrogen) at 37C and washed immediately redox homeostasis, as oxidation of some PTPs cannot be with ice-cold PBS (33). Mean fluorescence intensity (MFI) was quantified by reversed readily by oncogene inhibition (Figure 6). Further anal- flow cytometry (FACs Calibur). yses of this phenomenon might shed light on key regulators of local ROS production/redox regulation. qPTPome Finally, qPTPome can be combined with pY-proteomics to Cells or tissues were homogenized in 50 mM HEPES (pH 7.4), 150 mM NaCl, suggest specific PTP substrates (Figure 7). We identified four 10% glycerol, 1% NP40, containing 10 mM DTT and a protease and phosphatase inhibitor cocktail. Lysates were clarified at 16,100 3 g for 15 min at 4C. PTPs that appeared to play a key role in regulating phosphotyro- Supernatants were rotated in the dark at room temperature (RT) for 30 min., syl protein homeostasis in MM cells. Two of these PTPs have desalted by gel filtration into 20 mM HEPES (pH 7.4), treated with 100 mMPV been implicated in MM: PTPN6 is reportedly a tumor suppressor, (Huyer et al., 1997), and rotated at 4C for 1 hr in the dark. Proteins were dena- whereas PTPN7 is overexpressed (Julien et al., 2011). PTPN11 is tured in SDS (0.5%) and EDTA (5 mM) at 95C for 5 min, diluted 5-fold into NP40 a known positive regulator of the RAS-MAPK pathway and is a buffer (20 mM Tris-HCl [pH 8.0], 137 mM NaCl, 10% glycerol, 1% NP-40, 2 mM bona fide oncogene (Chan et al., 2008). Although PTPN11 has EDTA) and immunoprecipitated overnight at 4 C with the oxPTP Ab (mouse monoclonal anti-oxidized PTP active site antibody, clone 335636 [R&D not been implicated specifically in MM, it is required for the trans- Systems]; 5 mg Ab/mg protein) and 40 ml of packed protein G Sepharose beads. formation of NIH 3T3 cells expressing oncogenic FGFR3 (Agazie For MS, samples were incubated in 9M urea/4.5 mM DTT at 60C for 30 min, et al., 2003), a common MM oncogene. Moreover, PTPN1 deple- cooled to RT, treated with 10 mM iodoacetamide (IAM) for 15 min in the dark, tion in KMS11 cells leads to FGFR3 hyperphosphorylation and diluted to a final concentration of 2 M urea, 20 mM HEPES (pH 8.0), and increased global phosphotyrosine levels (Figure 7D), suggesting digested overnight at RT in 10 mg/ml TPCK-trypsin (Pierce). IAM prevents that PTPN1 may be a key negative regulator of FGFR3. Unex- incomplete digestion due to disulfide bonds and ensures alkylation of other pectedly, expression of PTPN1 (and the other three PTPs) corre- cysteines (besides the catalytic cysteine) found in some PTP active site peptides. Peptides were purified on a C18 column (Waters) by following the lates positively with tyrosyl phosphorylation of several peptides, manufacturer’s instructions, lyophilized, resuspended in 1.4 ml peptide IP yet knockdown causes increased phosphorylation of the same buffer (50 mM HEPES [pH 7.4], 50 mM NaCl) and incubated for 30 min on species. Conceivably, these apparently paradoxical findings a shaker at RT. Insoluble matter was removed by centrifugation at 2000 3 g. reflect an attempt by MM cells to limit the effects of excess Supernatants were collected, and PTP-SO3H active site peptides were FGFR3 signaling by upregulating its negative regulators. isolated by immunoprecipitation with the oxPTP Ab. Immunoprecipitates 3 In any case, our results argue strongly that altered redox regu- were washed in 1 ml peptide IP buffer (3 ), followed by 1 ml HPLC grade H2O(33), eluted by incubating with 80 ml 0.15% trifluoroacetic acid for lation provides an additional level of complexity to the cancer 10 min at RT (33), dried, and analyzed by MS. The Ptpn11 AQUA peptide phenotype. Further refinement and application of qPTPome 13 15 (QESIVDAGPVVVHCSO3HSAGIGR; six C and four N) was purchased from and q-oxPTPome should help delineate the contribution of Thermo Scientific. ROS regulation of PTPs to normal and pathological cell signaling. q-oxPTPome q-oxPTPome buffer (50 mM HEPES [pH 6.5], 150 mM NaCl, 10% glycerol, 1% EXPERIMENTAL PROCEDURES NP40) was degassed overnight at 30–35 mmHg, transferred into a sealed

round-bottom ﬂask equilibrated with N2, purged with N2 for 1 hr, transferred

Cells and Viral Infection to an anaerobic workstation (ESBE Invivo2 400), and treated with 10 mM

Cell lines were cultured under standard conditions and treated with H2O2 NEM (Sigma), 100 mg/ml catalase (Calbiochem) and a protease and phospha- (Bioshop) for 5 min or 100 mM BSO (Sigma) for 24 hr, as indicated. Nitrosylation tase inhibitor cocktail. Cells were washed with ice-cold degassed PBS,

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 837 transferred to the workstation and lysed in q-oxPTPome buffer. Lysates were the EU-funded PTPNET Research Training Network (A.O.). Additional support incubated for 1 hr at 4C in the dark, and homogenates were clariﬁed at 16,100 was provided by the Ontario Ministry of Health and Long Term Care and the 3 g for 15 min at 4C. Supernatants were collected in the anaerobic worksta- Princess Margaret Hospital Foundation. B.G.N. and M.F.M. are Canada tion, and the buffer exchanged by gel ﬁltration into degassed 20 mM HEPES Research Chairs, Tier 1. Y.A.S. is a British Heart Foundation (BHF) Interme- (pH 7.4). Desalted lysates were treated with 10 mM DTT and rotated in the diate Research Fellow (FS/08/034/25085) and J.M. is a BHF Post Doctoral dark for 30 min at RT. After exchanging the buffer to remove DTT, samples Fellow (PG/07/034/22775). R.K. and I.S.H. are the recipients of graduate were treated with 100 mM PV, rotated at 4C for 1 hr in the dark and analyzed fellowships from the CIHR and the Natural Sciences and Engineering Council by immunoprecipitation or MS. The DTT step is included to reduce all revers- of Canada, respectively. A.O. is a coinventor of the monoclonal oxidized PTP ibly oxidized PTPs; omitting this step prevents PV from accessing the PTP active site antibody and receives royalties for sales of the antibody. active site and limits PTP oxidation to the SO3H state. Received: November 6, 2010 Measurement of all PTP Oxidation States Revised: May 6, 2011 The total signal for each PTP is provided by qPTPome, while q-oxPTPome Accepted: July 1, 2011 determines the cumulative fraction of each PTP in the SOH and SO3H states. Published: September 1, 2011

To measure the proportion of each PTP in the SO3H state, H2O2-treated cells were lysed in the presence of DTT and analyzed by SRM. The difference REFERENCES between the q-oxPTPome and SO3H signal provides the fraction of PTPs in the SOH state. Because the SO2H state is not converted to the SO3H state Agazie, Y.M., Movilla, N., Ischenko, I., and Hayman, M.J. (2003). The phospho- by PV, the difference in signal between cells subjected to qPTPome in unsti- tyrosine phosphatase SHP2 is a critical mediator of transformation induced mulated and H2O2-stimulated cells provides the fraction of each PTP in the by the oncogenic ﬁbroblast growth factor receptor 3. Oncogene 22, 6909–

SO2H state. If PTPs are not converted to the SO2H state, the intensities from 6918. these two procedures should be the same; any difference corresponds to Ahn, N.G., Shabb, J.B., Old, W.M., and Resing, K.A. (2007). Achieving in-depth the fraction of each PTP converted to the SO2H state. Finally, the fraction of proteomics profiling by mass spectrometry. ACS Chem. Biol. 2, 39–52. active PTPs is the difference between the qPTPome and q-oxPTPome plus Alonso, A., Sasin, J., Bottini, N., Friedberg, I., Friedberg, I., Osterman, A., SO H signals, respectively (see Figure 4D). 2 Godzik, A., Hunter, T., Dixon, J., and Mustelin, T. (2004). Protein tyrosine phosphatases in the human genome. Cell 117, 699–711. LC-MS/MS Andersen, J.N., Mortensen, O.H., Peters, G.H., Drake, P.G., Iversen, L.F., LC-MS, peptide identification, and SRM were performed by using standard Olsen, O.H., Jansen, P.G., Andersen, H.S., Tonks, N.K., and Møller, N.P. techniques and procedures. Samples were prepared for pY-proteomics as (2001). Structural and evolutionary relationships among protein tyrosine phos- described (St-Germain et al., 2009). See Supplemental Information for details. phatase domains. Mol. Cell. Biol. 21, 7117–7136. Arnold, R.S., Shi, J., Murad, E., Whalen, A.M., Sun, C.Q., Polavarapu, R., Statistical Analysis Parthasarathy, S., Petros, J.A., and Lambeth, J.D. (2001). Hydrogen peroxide Data are presented as the mean ± SEM. Statistical significance was deter- mediates the cell growth and transformation caused by the mitogenic oxidase mined using Student’s t test (paired or unpaired) or one-way ANOVA, as Nox1. Proc. Natl. Acad. Sci. USA 98, 5550–5555. appropriate. If ANOVA was significant, individual differences were evaluated using Bonferroni post-test. Statistical analyses were performed using Graph- Boivin, B., Yang, M., and Tonks, N.K. (2010). Targeting the reversibly oxidized Pad Prism 5, with p < 0.05 considered significant. protein tyrosine phosphatase superfamily. Sci. Signal. 3, pl2. Given the multivariate multiple linear regression problem: Y = XB+ε (1), Bourdeau, A., Dube´ , N., and Tremblay, M.L. (2005). Cytoplasmic protein where X is an IxKpredictor matrix of PTPs and Y is an IxJresponse matrix tyrosine phosphatases, regulation and function: the roles of PTP1B and TC- of pY sites with both X and Y measured in five (I) different MM cell lines, PLSR PTP. Curr. Opin. Cell Biol. 17, 203–209. was implemented in the R (version 1.40) package PLS (Mevik and Wehrens, Burridge, K., and Nelson, A. (1995). An in-gel assay for protein tyrosine phos- 2007) to fit a solution to (1) using the NIPALS algorithm. PLSR attempts to phatase activity: detection of widespread distribution in cells and tissues. Anal. predict Y from X and describe their common structure by simultaneously de- Biochem. 232, 56–64. composing the predictor and response variables into a common set of orthog- Cairns, R.A., Harris, I.S., and Mak, T.W. (2011). Regulation of cancer cell onal factors and specific loadings: X = TPT;Y=UCT = TBCT, where T is the metabolism. Nat. Rev. Cancer 11, 85–95. score matrix, P is the loading matrix, B is a diagonal matrix of regression weights, and C is a weight matrix. T and U are chosen such that their covari- Chan, G., Kalaitzidis, D., and Neel, B.G. (2008). The tyrosine phosphatase ance is maximal. ‘‘Leave-one-out’’ cross-validation was used to determine Shp2 (PTPN11) in cancer. Cancer Metastasis Rev. 27, 179–192. the optimal number of components to use in the regression. Prior to fitting Chen, C.Y., Willard, D., and Rudolph, J. (2009). Redox regulation of SH2- the model, we excluded those PTPs and pY sites where there were two or domain-containing protein tyrosine phosphatases by two backdoor cysteines. fewer measurements across the five cell lines. Biochemistry 48, 1399–1409. den Hertog, J., Groen, A., and van der Wijk, T. (2005). Redox regulation of SUPPLEMENTAL INFORMATION protein-tyrosine phosphatases. Arch. Biochem. Biophys. 434, 11–15. Denu, J.M., and Dixon, J.E. (1998). Protein tyrosine phosphatases: mecha- Supplemental Information includes Extended Experimental Procedures, nisms of catalysis and regulation. Curr. Opin. Chem. Biol. 2, 633–641. four figures, and seven tables and can be found with this article online at Elschenbroich, S., and Kislinger, T. (2011). Targeted proteomics by selected doi:10.1016/j.cell.2011.07.020. reaction monitoring mass spectrometry: applications to systems biology and biomarker discovery. Mol. Biosyst. 7, 292–303. ACKNOWLEDGMENTS Finkel, T. (1998). Oxygen radicals and signaling. Curr. Opin. Cell Biol. 10, 248–253. We thank Dr. Petr Heneberg for qPCR and Dr. Suzanne Trudel (Ontario Cancer Institute/Princess Margaret Hospital) for the MM cell lines. This work was Gerber, S.A., Rush, J., Stemman, O., Kirschner, M.W., and Gygi, S.P. (2003). supported by National Institutes of Health Grant R37CA49152 (B.G.N.), the Absolute quantification of proteins and phosphoproteins from cell lysates by 100 Human Frontiers of Science Program Grant RGP0039/2009-C (B.G.N.), the tandem MS. Proc. Natl. Acad. Sci. USA , 6940–6945. Canadian Institutes of Health Research (CIHR; M.F.M.), the Canadian Cancer Gianni, D., Bohl, B., Courtneidge, S.A., and Bokoch, G.M. (2008). The Society Research Institute (M.F.M.), the Swedish Research Council (A.O.), and involvement of the tyrosine kinase c-Src in the regulation of reactive oxygen

838 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. species generation mediated by NADPH oxidase-1. Mol. Biol. Cell 19, 2984– Myers, M.P., Andersen, J.N., Cheng, A., Tremblay, M.L., Horvath, C.M., 2994. Parisien, J.P., Salmeen, A., Barford, D., and Tonks, N.K. (2001). TYK2 and 276 Gingras, M.C., Zhang, Y.L., Kharitidi, D., Barr, A.J., Knapp, S., Tremblay, M.L., JAK2 are substrates of protein-tyrosine phosphatase 1B. J. Biol. Chem. , and Pause, A. (2009). HD-PTP is a catalytically inactive tyrosine phosphatase 47771–47774. due to a conserved divergence in its phosphatase domain. PLoS ONE 4, Nelson, K.J., Klomsiri, C., Codreanu, S.G., Soito, L., Liebler, D.C., Rogers, e5105. L.C., Daniel, L.W., and Poole, L.B. (2010). Use of dimedone-based chemical Groen, A., Lemeer, S., van der Wijk, T., Overvoorde, J., Heck, A.J., Ostman, A., probes for sulfenic acid detection methods to visualize and identify labeled 473 Barford, D., Slijper, M., and den Hertog, J. (2005). Differential oxidation of proteins. Methods Enzymol. , 95–115. protein-tyrosine phosphatases. J. Biol. Chem. 280, 10298–10304. Ostman, A., Hellberg, C., and Bo¨ hmer, F.D. (2006). Protein-tyrosine phospha- 6 Gross, S., Blanchetot, C., Schepens, J., Albet, S., Lammers, R., den Hertog, J., tases and cancer. Nat. Rev. Cancer , 307–320. and Hendriks, W. (2002). Multimerization of the protein-tyrosine phosphatase Pandey, A., Blagoev, B., Kratchmarova, I., Fernandez, M., Nielsen, M., (PTP)-like insulin-dependent diabetes mellitus autoantigens IA-2 and IA-2beta Kristiansen, T.Z., Ohara, O., Podtelejnikov, A.V., Roche, S., Lodish, H.F., with receptor PTPs (RPTPs). Inhibition of RPTPalpha enzymatic activity. J. and Mann, M. (2002). Cloning of a novel phosphotyrosine binding domain Biol. Chem. 277, 48139–48145. containing molecule, Odin, involved in signaling by receptor tyrosine kinases. Oncogene 21, 8029–8036. Gstaiger, M., and Aebersold, R. (2009). Applying mass spectrometry-based proteomics to genetics, genomics and network biology. Nat. Rev. Genet. 10, Pearce, A.C., Senis, Y.A., Billadeau, D.D., Turner, M., Watson, S.P., and 617–627. Vigorito, E. (2004). Vav1 and vav3 have critical but redundant roles in mediating platelet activation by collagen. J. Biol. Chem. 279, 53955–53962. Hsu, M.F., and Meng, T.C. (2010). Enhancement of insulin responsiveness by nitric oxide-mediated inactivation of protein-tyrosine phosphatases. J. Biol. Persson, C., Sjo¨ blom, T., Groen, A., Kappert, K., Engstro¨ m, U., Hellman, U., Chem. 285, 7919–7928. Heldin, C.H., den Hertog, J., and Ostman, A. (2004). Preferential oxidation of the second phosphatase domain of receptor-like PTP-alpha revealed by an Huyer, G., Liu, S., Kelly, J., Moffat, J., Payette, P., Kennedy, B., Tsaprailis, G., antibody against oxidized protein tyrosine phosphatases. Proc. Natl. Acad. Gresser, M.J., and Ramachandran, C. (1997). Mechanism of inhibition of Sci. USA 101, 1886–1891. protein-tyrosine phosphatases by vanadate and pervanadate. J. Biol. Chem. 272, 843–851. Persson, C., Kappert, K., Engstro¨ m, U., Ostman, A., and Sjo¨ blom, T. (2005). An antibody-based method for monitoring in vivo oxidation of protein tyrosine Julien, S.G., Dube´ , N., Hardy, S., and Tremblay, M.L. (2011). Inside the human phosphatases. Methods 35, 37–43. cancer tyrosine phosphatome. Nat. Rev. Cancer 11, 35–49. Picotti, P., Bodenmiller, B., Mueller, L.N., Domon, B., and Aebersold, R. (2009). Kumar, S., Zhou, B., Liang, F., Wang, W.Q., Huang, Z., and Zhang, Z.Y. (2004). Full dynamic range proteome analysis of S. cerevisiae by targeted proteomics. Activity-based probes for protein tyrosine phosphatases. Proc. Natl. Acad. Cell 138, 795–806. Sci. USA 101, 7943–7948. Poole, L.B., and Nelson, K.J. (2008). Discovering mechanisms of signaling- Lahiry, P., Torkamani, A., Schork, N.J., and Hegele, R.A. (2010). Kinase muta- mediated cysteine oxidation. Curr. Opin. Chem. Biol. 12, 18–24. tions in human disease: interpreting genotype-phenotype relationships. Nat. Rev. Genet. 11, 60–74. Rhee, S.G., Bae, Y.S., Lee, S.R., and Kwon, J. (2000). Hydrogen peroxide: a key messenger that modulates protein phosphorylation through cysteine Lee, S.R., Kwon, K.S., Kim, S.R., and Rhee, S.G. (1998). Reversible inactiva- oxidation. Sci. STKE 2000, pe1. tion of protein-tyrosine phosphatase 1B in A431 cells stimulated with epidermal growth factor. J. Biol. Chem. 273, 15366–15372. Rikova, K., Guo, A., Zeng, Q., Possemato, A., Yu, J., Haack, H., Nardone, J., Lee, K., Reeves, C., Li, Y., et al. (2007). Global survey of phosphotyrosine Lemmon, M.A., and Schlessinger, J. (2010). Cell signaling by receptor tyrosine signaling identifies oncogenic kinases in lung cancer. Cell 131, 1190–1203. kinases. Cell 141, 1117–1134. Salmeen, A., Andersen, J.N., Myers, M.P., Meng, T.C., Hinks, J.A., Li, Q., Harraz, M.M., Zhou, W., Zhang, L.N., Ding, W., Zhang, Y., Eggleston, T., Tonks, N.K., and Barford, D. (2003). Redox regulation of protein tyrosine Yeaman, C., Banfi, B., and Engelhardt, J.F. (2006). Nox2 and Rac1 regulate phosphatase 1B involves a sulphenyl-amide intermediate. Nature 423, H2O2-dependent recruitment of TRAF6 to endosomal interleukin-1 receptor 769–773. complexes. Mol. Cell. Biol. 26, 140–154. Salmeen, A., and Barford, D. (2005). Functions and mechanisms of redox Li, S., and Whorton, A.R. (2003). Regulation of protein tyrosine phosphatase regulation of cysteine-based phosphatases. Antioxid. Redox Signal. 7, 1B in intact cells by S-nitrosothiols. Arch. Biochem. Biophys. 410, 269–279. 560–577. Liou, G.Y., and Storz, P. (2010). Reactive oxygen species in cancer. Free Sattler, M., Verma, S., Shrikhande, G., Byrne, C.H., Pride, Y.B., Winkler, T., Radic. Res. 44, 479–496. Greenfield, E.A., Salgia, R., and Griffin, J.D. (2000). The BCR/ABL tyrosine Lou, Y.W., Chen, Y.Y., Hsu, S.F., Chen, R.K., Lee, C.L., Khoo, K.H., Tonks, kinase induces production of reactive oxygen species in hematopoietic cells. N.K., and Meng, T.C. (2008). Redox regulation of the protein tyrosine phospha- J. Biol. Chem. 275, 24273–24278. tase PTP1B in cancer cells. FEBS J. 275, 69–88. Seo, Y.H., and Carroll, K.S. (2009). Profiling protein thiol oxidation in tumor Meng, T.C., Fukada, T., and Tonks, N.K. (2002). Reversible oxidation cells using sulfenic acid-specific antibodies. Proc. Natl. Acad. Sci. USA 106, and inactivation of protein tyrosine phosphatases in vivo. Mol. Cell 9, 16163–16168. 387–399. Singh, D.K., Kumar, D., Siddiqui, Z., Basu, S.K., Kumar, V., and Rao, K.V. Meng, T.C., Buckley, D.A., Galic, S., Tiganis, T., and Tonks, N.K. (2004). (2005). The strength of receptor signaling is centrally controlled through Regulation of insulin signaling through reversible oxidation of the protein- a cooperative loop between Ca2+ and an oxidant signal. Cell 121, 281–293. 279 tyrosine phosphatases TC45 and PTP1B. J. Biol. Chem. , 37716– St-Germain, J.R., Taylor, P., Tong, J., Jin, L.L., Nikolic, A., Stewart, I.I., 37725. Ewing, R.M., Dharsee, M., Li, Z., Trudel, S., and Moran, M.F. (2009). Multiple Mevik, B.H., and Wehrens, R. (2007). The pls Package: Principal Component myeloma phosphotyrosine proteomic profile associated with FGFR3 expres- and Partial. Least Squares Regression in R. J. Stat. Software 18, 1–24. sion, ligand activation, and drug inhibition. Proc. Natl. Acad. Sci. USA 106, Mizuno, K., Hasegawa, K., Katagiri, T., Ogimoto, M., Ichikawa, T., and Yakura, 20127–20132. H. (1993). MPTP delta, a putative murine homolog of HPTP delta, is expressed Sundaresan, M., Yu, Z.X., Ferrans, V.J., Irani, K., and Finkel, T. (1995). Require- in specialized regions of the brain and in the B-cell lineage. Mol. Cell. Biol. 13, ment for generation of H2O2 for platelet-derived growth factor signal 5513–5523. transduction. Science 270, 296–299.

Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc. 839 Tanner, J.J., Parsons, Z.D., Cummings, A.H., Zhou, H., and Gates, K.S. (2010). Weibrecht, I., Bo¨ hmer, S.A., Dagnell, M., Kappert, K., Ostman, A., and Redox regulation of protein tyrosine phosphatases: structural and chemical Bo¨ hmer, F.D. (2007). Oxidation sensitivity of the catalytic cysteine of the aspects. Antioxid. Redox Signal. 15, 77–97. protein-tyrosine phosphatases SHP-1 and SHP-2. Free Radic. Biol. Med. 43, 100–110. Tonks, N.K. (2005). Redox redux: revisiting PTPs and the control of cell signaling. Cell 121, 667–670. Woo, H.A., Yim, S.H., Shin, D.H., Kang, D., Yu, D.Y., and Rhee, S.G. (2010). Inactivation of peroxiredoxin I by phosphorylation allows localized H(2)O(2) Tonks, N.K. (2006). Protein tyrosine phosphatases: from genes, to function, to accumulation for cell signaling. Cell 140, 517–528. disease. Nat. Rev. Mol. Cell Biol. 7, 833–846. Yates, J.R., Ruse, C.I., and Nakorchevsky, A. (2009). Proteomics by mass van Montfort, R.L., Congreve, M., Tisi, D., Carr, R., and Jhoti, H. (2003). spectrometry: approaches, advances, and applications. Annu. Rev. Biomed. Oxidation state of the active-site cysteine in protein tyrosine phosphatase Eng. 11, 49–79. 1B. Nature 423, 773–777. Zhang, Z.Y., Wang, Y., and Dixon, J.E. (1994). Dissecting the catalytic Walther, T.C., and Mann, M. (2010). Mass spectrometry-based proteomics in mechanism of protein-tyrosine phosphatases. Proc. Natl. Acad. Sci. USA cell biology. J. Cell Biol. 190, 491–500. 91, 1624–1627. Weerapana, E., Wang, C., Simon, G.M., Richter, F., Khare, S., Dillon, M.B., Zhang, Y., Wolf-Yadlin, A., Ross, P.L., Pappin, D.J., Rush, J., Lauffenburger, Bachovchin, D.A., Mowen, K., Baker, D., and Cravatt, B.F. (2010). Quantitative D.A., and White, F.M. (2005). Time-resolved mass spectrometry of tyrosine reactivity proﬁling predicts functional cysteines in proteomes. Nature 468, phosphorylation sites in the epidermal growth factor receptor signaling 790–795. network reveals dynamic modules. Mol. Cell. Proteomics 4, 1240–1250.

840 Cell 146, 826–840, September 2, 2011 ª2011 Elsevier Inc.