Differential Oxidation of Protein-tyrosine Phosphatases*

Oxidation is emerging as an important regulatory mechanism of protein-tyrosine phosphatases (PTPs). Here we report that PTPs are differentially oxidized, and we provide evidence for the underlying mechanism. The membrane-proximal RPTPα-D1 was catalytically active but not readily oxidized as assessed by immunoprobing with an antibody that recognized oxidized catalytic site cysteines in PTPs (oxPTPs). In contrast, the membrane-distal RPTPα-D2, a poor PTP, was readily oxidized. Oxidized catalytic site cysteines in PTP immunoprobing and mass spectrometry demonstrated that mutation of two residues in the Tyr(P) loop and the WPD loop that reverse catalytic activity of RPTPα-D1 and RPTPα-D2 also reversed oxidizability, suggesting that oxidizability and catalytic activity are coupled. However, catalytically active PTP1B and LAR-D1 were readily oxidized. Oxidizability was strongly dependent on pH, indicating that the microenvironment of the catalytic cysteine has an important role. Crystal structures of PTP domains demonstrated that the orientation of the absolutely conserved PTP loop arginine correlates with oxidizability of PTPs, and consistently, RPTPμ-D1, with a similar conformation as RPTPα-D1, was not readily oxidized. In conclusion, PTPs are differentially oxidized at physiological pH and H2O2 concentrations, and the PTP loop arginine is an important determinant for susceptibility to oxidation.

response to H 2 O 2 treatment, because RPTP␣-D2 undergoes a conformational change in response to oxidation, leading to stabilization of an inactive dimeric conformation (8,29). Mutation of the catalytic cysteine in RPTP␣-D2 renders RPTP␣ less sensitive to oxidation, in that H 2 O 2 and UV treatment of living cells do not lead to complete inactivation of RPTP␣-C723S, whereas wild type RPTP␣ is inactivated completely (8,28).
In the present study, we investigated differential oxidation of PTPs. oxPTP immunoprobing and matrix-assisted laser desorption ionization time-of-flight (MALDI-ToF) mass spectrometry indicated that the mutations in the Tyr(P) loop and the WPD loop that are responsible for the difference in catalytic activity between RPTP␣-D1 and RPTP␣-D2 were also involved in the difference in oxidizability. Furthermore, we demonstrate that PTP1B and LAR-D1 were highly oxidizable. The susceptibility of PTPs to oxidation was strongly dependent on pH. Comparison of the crystal structures suggested that the orientation of the PTP loop arginine is important for the oxidizability of PTPs. Taken together, the subtle differences in the microenvironment of the catalytic cysteines determine the oxidizability of PTPs.
Analysis of Oxidized PTPs Using the oxPTP Antibody-Reversible oxidation was assayed using an antibody, oxPTP, that specifically recognizes the sulfonic acid form of catalytic site cysteines of PTPs (28,32). As outlined in Fig. 1A, following treatment, the sample is alkylated to protect reduced cysteines. Subsequently, the sample is treated with dithiothreitol (DTT) to reduce reversibly oxidized cysteines and treated with pervanadate to convert reduced cysteines to the sulfonic acid form. Finally, the sulfonic acid form is detected using the oxPTP antibody by immunoblotting. The procedure has been described in detail (28). Briefly, GST fusion proteins, purified using standard procedures, were bound to glutathione beads. All of the incubations and washes were done with 20 mM Tris, pH 7.5, unless otherwise stated. The proteins were reduced for 30 min in 10 mM DTT. The beads were washed twice and were incubated with varying concentrations of H 2 O 2 for 20 min. Immediately after oxidation proteins were alkylated, using 40 mM iodoacetic acid for 30 min. The proteins were washed extensively, reduced with 10 mM DTT, washed twice, and incubated for 1 h with 0.1 mM pervanadate (1 mM pervanadate, 1 mM Na 3 VO 4 mixed with 5 mM H 2 O 2 , final concentrations, and left for at least 5 min at room temperature) to convert reduced cysteines irreversibly into the sulfonic acid form. Laemmli sample buffer was added, and the samples were boiled for 5 min and run on a 10% SDS-polyacrylamide gel (5 g of protein/lane). The samples were transferred to polyvinylidene difluoride membrane by semi-dry blotting. For detection with oxPTP antibody, the blots were blocked in 0.1% bovine serum albumin, 0.1% Tween in 50 mM Tris, pH 7.5, 150 mM NaCl for 1 h or overnight and incubated with oxPTP antibody (1.5 h at room temperature or overnight at 4°C). Subsequently, horseradish peroxidase-coupled goat-anti-rabbit (1:10,000; BD Pharmigen) was used, and the antibody signals were visualized by ECL. In some cases, the membranes were stripped and reprobed with anti-RPTP␣ antibody (30) and subsequently developed using ECL.
Phosphatase Assays-Fusion proteins were bound to GST beads in 20 mM Tris-HCl, pH 7.5, 10 mM DTT. The beads were incubated with varying concentrations of H 2 O 2 for 20 min at pH 7.5 and then directly incubated in 20 mM MES buffer, pH 6.0, 1 mM EDTA, 150 mM NaCl, 10 mM p-nitrophenylphosphate for 45 min at 30°C. The reactions were quenched with 0.4 M NaOH, and optical density was measured with a spectrophotometer at 440 nm (wavelength).
MALDI-ToF Mass Spectrometry-All of the reactions and washing steps were performed in 20 mM Tris-HCl, pH 7.5. GST fusion proteins of RPTP␣-D2 and mutated RPTP␣-D2 (V555Y, E690D, and V555Y/ E690D) were immobilized on glutathione-Sepharose beads and incubated with 10 mM DTT. The beads were washed and incubated with H 2 O 2 for 20 min at room temperature. Subsequently, the beads were treated with 100 mM iodoacetamide for 30 min at room temperature to derivatize the unaffected cysteines. Finally, the beads were washed and incubated with 10 mM DTT for 10 min to reduce singly oxidized cysteines. The fusion protein was cleaved off by thrombin protease (4 h at room temperature), and the protein of interest was subjected to overnight tryptic digestion. For desalting, the peptides were adsorbed on C 18 ZipTips (Millipore) and washed with 0.1% trifluoroacetic acid. The peptides retained on the columns were eluted using ␣-cyano-4-hydroxycinnamic acid (10 mg/ml) in 10% acetonitrile, 0.1% trifluoroacetic acid. Peptide mass fingerprint spectra were recorded on a MALDI-ToF Voyager DE-STR (Applied Biosystems) mass spectrometer operated in positive ion Reflectron mode. After time-delayed extraction, the ions were accelerated to 20 kV for ToF mass spectrometric analysis. A total of 150 shots were acquired, and the signal was averaged per spectrum. MALDI-ToF spectra were analyzed using the Applied Biosystems Data Explorer software.

RESULTS
Differential Oxidation of RPTP␣-D1 and RPTP␣-D2-We investigated oxidation of PTPs using the oxPTP antibody (Fig.  1A). Consistent with a previous report (28), RPTP␣-D1 was oxidized at 250 M H 2 O 2 but not at lower concentrations ( Fig.  1B), and RPTP␣-D2 was already detectably oxidized at the lowest concentration used (31 M H 2 O 2 ) (Fig. 1C). The negative control in our oxPTP experiments represents the fully reduced PTP, achieved by treatment with 10 mM DTT for 20 min. For the positive control, the alkylation step was omitted, leading to complete triple oxidation of the catalytic cysteine upon pervanadate treatment (Fig. 1).
In parallel with the oxPTP immunoprobing experiments, we performed PTP activity assays using p-nitrophenylphosphate as a substrate. RPTP␣-D1 retained much of its activity at low concentrations of H 2 O 2 (31 M) (Fig. 1D). RPTP␣-D2, although much less active than D1, is more susceptible to H 2 O 2 than RPTP␣-D1 and is already inactivated at low concentrations of H 2 O 2 (31 M; Fig. 1D). Together, these results show that RPTP␣-D2 is more susceptible to oxidation than RPTP␣-D1 as detected by the oxPTP antibody and by PTP activity assays.
To test whether the Tyr(P) loop and the WPD motif were responsible for low oxidizability of RPTP␣-D1, Tyr 262 was mutated to valine, and Asp 401 was mutated to glutamate. RPTP␣-D1-Y262V and RPTP␣-D1-D401E were much more sensitive to oxidation than wild type RPTP␣-D1, in that appreciable oxidation was already detected at 31-62 M H 2 O 2 (Fig. 2B), similar to wild type RPTP␣-D2. These results support the hypothesis that the Tyr(P) loop and WPD loop are important for susceptibility to oxidation of RPTP␣-D1 and RPTP␣-D2.
Mass spectrometry experiments were performed to further investigate oxidation of RPTP␣-D1 and RPTP␣-D2. Unfortunately, the Cys 433 containing peptide of RPTP␣-D1 did not resolve well in the MALDI-ToF spectra, and detection of Cys 433 oxidation failed. The peptide containing Cys 723 of RPTP␣-D2 and its oxidized derivatives were readily detectable by MALDI-ToF ( Fig. 3). At 1 mM H 2 O 2 , four peaks were evident in the spectra. As a result of the alkylation/reduction procedure, the reversibly oxidized Cys 723 peptide was detected in its reduced form (m/z 2075.98), whereas alkylated Cys 723 peptide corre-sponds to reduced Cys 723 peptide (m/z 2133.00). The other two peaks, m/z 2107.98 and 2123.98, represent irreversibly oxidized RPTP␣-D2, with relative mass shifts of ϩ32 and ϩ48 corresponding to the formation of the sulfinic and sulfonic acid species, respectively. Reversibly oxidized Cys 723 peptide was already detected at the lowest concentrations of H 2 O 2 (31 M), and the fraction of reversibly oxidized Cys 723 peptide increased gradually with increasing H 2 O 2 concentrations (Fig. 3) Based on the MALDI-ToF spectra, we calculated the fraction of reversibly oxidized Cys 723 peptide, relative to the total amount of Cys 723 peptide. Wild type RPTP␣-D2 was highly sensitive to oxidation, whereas RPTP␣-D2-E690D was moder- ately sensitive, showing low levels of reversible oxidation at low H 2 O 2 concentrations (Fig. 3). RPTP␣-D2-V555Y and RPTP␣-E690D/V555Y were the least sensitive to oxidation, because for both mutants significant oxidation was detected only at high concentrations of H 2 O 2 (250 M and higher; Fig. 3). The sulfinic and sulfonic acid form of Cys 723 peptides in the MALDI-ToF spectra of RPTP␣-D2 were not detected in any of the mutants (data not shown), again pointing toward a decreased sensitivity to oxidation for all mutants. The MALDI-ToF results are consistent with the oxPTP antibody data.
The oxidation and PTP activity results (Figs. 2 and 3) demonstrate that the two residues that determine the difference in catalytic activity between RPTP␣-D1 and RPTP␣-D2 are also responsible for the difference in oxidizability, which may suggest that PTP catalytic activity and oxidizability are linked.
Oxidizability of PTP1B and LAR-Our results suggest an inverse relation between catalytic activity and oxidizability. However, other PTPs that are catalytically active have been reported to be oxidized. Therefore, we investigated oxidation of PTP1B in parallel to RPTP␣-D1 and RPTP␣-D2. Oxidation of zebrafish PTP1B in which the oxPTP epitope (VHCSAG) is fully conserved (31) was readily detected at very low levels of H 2 O 2 (31 M), and PTP1B oxidation was maximal at 125-250 M H 2 O 2 (Fig. 4), similar to RPTP␣-D2 (Fig. 1C). PTP1B is catalytically active (data not shown), arguing against an inverse correlation between catalytic activity and oxidation.
Next, we investigated oxidizability of the two catalytic domains of the RPTP, LAR. Both LAR-D1 and LAR-D2 were readily oxidized at low H 2 O 2 concentrations (31 M) with maximal oxidation levels at 125-250 M, like RPTP␣-D2 and PTP1B (Fig. 4). PTP activity assays that were done in parallel demonstrate that LAR-D1 is catalytically active and LAR-D2 is not (data not shown), consistent with previous reports (6). Taken together, the finding that the catalytically active PTP1B and LAR-D1 are readily oxidized indicates that there is not a strict inverse correlation between catalytic activity and oxidizability.
Because the Tyr(P) loop and WPD loop are conserved in PTP1B and LAR-D1, other features of the PTPs must play a role in determining the sensitivity to oxidation. Gly 259 in PTP1B (Gln 474 in RPTP␣-D1 and His 764 in RPTP␣-D2) is an important determinant of substrate specificity (34) and determines the flexibility of Gln 262 in the Q loop of the catalytic pocket (35). Gly 259 might therefore be involved in the difference in oxidizability between PTP1B and RPTP␣-D1. We mutated Gln 474 in RPTP␣-D1 and His 764 in RPTP␣-D2 to glycine, the residue at the corresponding position in PTP1B. RPTP␣-D1-Q474G was poorly oxidized (Fig. 4), like wild type RPTP␣-D1 (Fig. 1B). RPTP␣-D2-H764G was readily oxidized (Fig. 4), like wild type RPTP␣-D2 (Fig. 1C). Taken together, these results indicate that Gly 259 , a determinant of substrate specificity in PTP1B and RPTP␣, is not involved in differential oxidation of PTPs.
Extensive comparison of the primary sequences of the PTPs involved did not lead to the identification of residues that might be responsible for the difference in oxidizability, except for three residues that are conserved in RPTP␣-D1 ( 331 TNL 333 ) and RPTP␣-D2 ( 624 TEL 626 ) but not in PTP1B ( 109 NRV 111 ). Thr 331 is buried in the catalytic pocket of RPTP␣-D1 close to the catalytic Cys 433 and therefore might have a role in oxidizability of Cys 433 . We mutated Thr 331 and the two following residues to the corresponding residues in PTP1B (T 3 N, TN 3 NR, and TNL 3 NRV) and investigated oxidizability and catalytic activity of the mutants. However, these mutants were not catalytically active, and oxidation was not detectable using the oxPTP antibody (data not shown), suggesting that these mutations severely disrupted the conformation of the catalytic pocket.
pH-dependent PTP Oxidation-Cysteine residues are most vulnerable to oxidation in the thiolate anion form. Therefore, the pH may be an important determinant for oxidation of PTPs. Purified RPTP␣-D2 was treated with H 2 O 2 , and MALDI-ToF spectra were acquired as described under "Materials and Methods." The m/z range that contains the active site Cys 723 peptide (residues 710 -729) is depicted. Reversibly oxidized Cys 723 peptide was detected in reduced form (m/z 2075.98, *), the sulfinic acid containing peptide shifted ϩ32.00 to m/z 2107.98 (**), and the sulfonic acid containing peptide ϩ48.00 to m/z 2123.98 (***). The reduced Cys 723 containing peptide was detected in alkylated form with a shift in relative molecular mass of ϩ 57.02 (m/z 2133.00, ****). Peak intensities of reduced and reversibly oxidized Cys 723 peptides were determined from the MALDI-ToF spectra (panels on the right). Relative oxidation (y axes run from 0 to 100%) is plotted against increasing H 2 O 2 concentrations (0, 31, 62, 125, 250, 500, and 1000 M, left to right) for wild type (WT) and mutant RPTP␣-D2 (as indicated).
It is well known that the catalytic activity of PTPs is strongly dependent on pH (36,37). We assayed the PTP activity of RPTP␣-D1, RPTP␣-D2, RPTP␣-D2-V555Y/E690D, and PTP1B at different pH. PTPs are highly active at low pH (optimum activity at pH 5.5-6.0 for all PTPs tested). PTP activity decreased rapidly with increasing pH to 10% of the maximum activity at pH 8.0. It is noteworthy that although the absolute activity of the four PTPs tested is different (PTP1B Ͼ RPTP␣-D1 Ͼ RPTP␣-D2-V555Y/E690D Ͼ Ͼ RPTP␣-D2), the pH-dependent decrease in activity is very similar (Fig. 5A).
To investigate whether oxidation is a pH-dependent process, H 2 O 2 treatment was done at a range of different pHs. Three H 2 O 2 concentrations were used that are indicative of oxidizability, 31, 62, and 125 M. Strikingly, the oxidizability of RPTP␣-D1 was clearly dependent on pH in that RPTP␣-D1 was oxidized in response to low levels of H 2 O 2 (31 M) at pH 8.5 and 9.0 (Fig. 5B). RPTP␣-D2 oxidation was not affected by elevating the pH, but lowering the pH to 7.0 and 6.5 gradually reduced oxidation of RPTP␣-D2 (Fig. 5B). PTP1B was sensitive to oxidation at pH 7.5 and, like RPTP␣-D2, lost its sensitivity when the pH was lowered to 6.5. RPTP␣-D2-V555Y/E690D responded to different pHs much like RPTP␣-D1, in that RPTP␣-D2-V555Y/E690D was oxidized at high pH (pH 8.5 and 9.0). Taken together, these results demonstrate that oxidation of PTPs is strongly dependent on the pH, suggesting that the microenvironment of the catalytic cysteines has an important role.
PTP loop Arginine Is Important for Oxidizability-The crystal structures of the PTPs involved have all been solved: RPTP␣-D1 (38), RPTP␣-D2 (11), PTP1B (9), and LAR-D1 and LAR-D2 (10). Comparison of the microenvironment of the catalytic cysteine of these five PTPs indicated that the absolutely conserved PTP loop arginine is oriented differently in RPTP␣-D1 than in RPTP␣-D2, PTP1B, or LAR (Fig. 6). We determined the distance between the S␥ atom of the catalytic cysteine and the three N atoms of the guanidinium group of the arginine (Table I). Strikingly, the N 1 atom was much closer to the S␥ atom of the catalytic cysteine in RPTP␣-D1 than in the other PTPs (4.9 Å versus 7.1-8.0 Å; Table I). The N⑀ atoms were located at similar distances in all five PTPs (5.2-6.1 Å), whereas the N 2 atoms were partially buried in most structures at similar distances to the S␥ atom (6.1-7.3 Å). The proximity of the N 1 atom to the S␥ atom may account for the difference in oxidizability between RPTP␣-D1 and other PTPs, because the guanidinium group of Arg 439 may sterically hinder the reaction of H 2 O 2 with the thiolate group of Cys 433 .
We analyzed the crystal structures of other classical PTPs, RPTP-D1, TC-PTP, Shp-1, Shp-2, PTP-SL, and YOP51. We found that only in RPTP-D1, was the guanidinium group of the PTP loop arginine oriented similarly as in RPTP␣-D1 (Fig.  6), and the N 1 atom was relatively close to the S␥ atom (5.5 Å; Table I). Indeed, RPTP-D1 was only oxidized at high H 2 O 2 concentrations (Fig. 4), similar to RPTP␣-D1. Our results are consistent with a crucial role for the PTP loop arginine in the oxidizability of PTPs.

DISCUSSION
Oxidation is an attractive regulatory mechanism for PTPs, because ROS are readily formed in response to various stimuli, and oxidation is reversible. Here, we report that PTPs were oxidized differentially, which may be important for fine-tuning of the cellular responses to stimuli. We demonstrate that oxidizability and catalytic activity of RPTP␣-D1 and RPTP␣-D2 were reversed upon mutation of two residues in the Tyr(P) loop and WPD loop that are absolutely conserved in catalytically active PTPs but not in RPTP-D2s. However, these two residues cannot explain the difference in oxidizability between RPTP␣-D1 on the one hand and the catalytically active PTPs, PTP1B, and LAR-D1 on the other, because these residues are conserved in all active PTPs. We found that oxidation of PTPs was highly dependent on pH, and we identified the PTP loop arginine as an important determinant in oxidizability.
oxPTP immunoprobing, PTP activity assays, and MALDI-ToF mass spectrometry consistently showed differential oxidation of purified RPTP␣-D1 and RPTP␣-D2 in vitro. We have previously shown that Cys 723 , the catalytic cysteine of RPTP␣-D2, is preferentially oxidized in living cells in response to UV treatment (28). Mutation of Cys 723 , but not Cys 433 , abolished oxidation of full-length RPTP␣ in living cells. Moreover, differential oxidizability of RPTP␣-D1 and RPTP␣-D2 is consistent with previous data that RPTP␣-D2 rather than RPTP␣-D1 is the redox sensor. Mutation of Cys 723 renders full-length RPTP␣ at least in part insensitive to oxidation, in that H 2 O 2 treatment or UV treatment of living cells completely inactivates wild type RPTP␣, but not RPTP␣-C723S (8,28). H 2 O 2induced inactivation of RPTP␣ is due to a conformational change in RPTP␣-D2, which leads to stabilization of an inactive RPTP␣ dimer conformation (8,29). How RPTP␣-D2 changes conformation in response to H 2 O 2 remains elusive and will require elucidation of the crystal structure of oxidized RPTP␣-D2. The catalytic Cys 723 is required for the conformational change (8), and recently we found evidence that none of the other cysteines in RPTP␣ is involved in H 2 O 2 -induced stable dimer formation (39), ruling out the possibility of intramolecular disulfide bond formation, which has been observed in nonclassical Cys-based PTPs, including LMW-PTP (40), PTEN (41), and CDC25 (42). It is likely that the conformational change in RPTP␣-D2 is initiated by cyclic sulfenamide formation, which leads to extensive reorganization of the catalytic site of PTP1B (19,20).
Although oxidizability of RPTP␣-D1 and RPTP␣-D2 was reversed upon mutation of only two residues in the Tyr(P) loop and WPD loop (Figs. 2 and 3), this cannot explain the difference in oxidizability between PTP1B, LAR-D1, and RPTP␣-D1 (Fig.  4), because the Tyr(P) loop and WPD loop are conserved in all three active PTPs. Oxidation of PTPs is strongly dependent on pH (Fig. 5), suggesting that the microenvironment of the cat-alytic cysteine is an important determinant for oxidizability. Comparison of the crystal structures of the classical PTPs suggested that the orientation of the PTP loop arginine was important. The orientation of the PTP loop arginine in RPTP-D1 suggested that RPTP-D1 would not be highly susceptible to oxidation. Indeed, like RPTP␣-D1, RPTP-D1 was relatively insensitive to oxidation (Fig. 4). The proximity of the PTP loop arginine to the catalytic cysteine may sterically hinder the reaction of H 2 O 2 with the thiolate anion group. To directly test the role of the PTP loop arginine in oxidizability, we mutated Arg 439 to lysine in RPTP␣-D1. Unfortunately, GST-RPTP␣-D1-R439K was not stable, and we did not obtain sufficient material to determine the oxidation of this fusion protein.
It is not clear why the guanidinium group of Arg 439 is oriented differently in RPTP␣-D1 than in most other PTPs. Unlike the other crystal structures, RPTP␣-D1 formed dimers in the crystals in which a helix loop-helix wedge-like structure inserts into the catalytic site of the dyad-related monomer (38), which may affect the orientation of the guanidinium group. However, the position of the PTP loop arginine and particularly its guanidinium group in RPTP-D1, which did not form dimers like RPTP␣-D1 in crystals (43), is similar to that in RPTP␣-D1, indicating that this conformation is not driven by dimerization.
It is noteworthy that differential oxidation of PTPs is especially obvious at relatively low H 2 O 2 concentrations. Cellular  stimuli induce H 2 O 2 concentrations in a range, similar to the H 2 O 2 concentrations we have used for our experiments. For instance, platelet-derived growth factor induces ROS levels in cells that are comparable with the levels in cells resulting from exogenously added H 2 O 2 concentrations of 0.1-1.0 mM (21). Differential oxidation of PTPs is important for fine-tuning of the cellular response to ROS production. Apparently, different inactivating mechanisms have evolved in the RPTPs: direct oxidation of the catalytic cysteine in LAR and indirect inactivation through oxidation of the catalytic cysteine in RPTP␣-D2, resulting in stabilization of an inactive dimer conformation (8,29). LAR-D2 does undergo a conformational change in response to H 2 O 2 , like RPTP␣-D2, and H 2 O 2 induces heterodimer formation between full-length RPTP␣ and LAR (44). Whether oxidation of LAR-D2 is required for full inactivation of LAR remains to be determined.
In conclusion, we demonstrate that PTPs are differentially oxidized, and we provide evidence for the underlying mechanism. It will be interesting to see whether other PTP family members, including the nonclassical Cys-based PTPs, are differentially oxidized as well.