Interconversion of the Kinetic Identities of the Tandem Catalytic Domains of Receptor-like Protein-tyrosine Phosphatase PTPα by Two Point Mutations Is Synergistic and Substrate-dependent*

The two tandem homologous catalytic domains of PTPα possess different kinetic properties, with the membrane proximal domain (D1) exhibiting much higher activity than the membrane distal (D2) domain. Sequence alignment of PTPα-D1 and -D2 with the D1 domains of other receptor-like PTPs, and modeling of the PTPα-D1 and -D2 structures, identified two non-conserved amino acids in PTPα-D2 that may account for its low activity. Mutation of each residue (Val-536 or Glu-671) to conform to its invariant counterpart in PTPα-D1 positively affected the catalytic efficiency of PTPα-D2 toward the in vitro substratespara-nitrophenylphosphate and the phosphotyrosyl-peptide RR-src. Together, they synergistically transformed PTPα-D2 into a phosphatase with catalytic efficiency forpara-nitrophenylphosphate equal to PTPα-D1 but not approaching that of PTPα-D1 for the more complex substrate RR-src.In vivo, no gain in D2 activity toward p59 fyn was effected by the double mutation. Alteration of the two corresponding invariant residues in PTPα-D1 to those in D2 conferred D2-like kinetics toward all substrates. Thus, these two amino acids are critical for interaction with phosphotyrosine but not sufficient to supply PTPα-D2 with a D1-like substrate specificity for elements of the phosphotyrosine microenvironment present in RR-src and p59 fyn . Whether the structural features of D2 can uniquely accommodate a specific phosphoprotein substrate or whether D2 has an alternate function in PTPα remains an open question.

Protein tyrosine phosphorylation status is a key determinant of nearly all eukaryotic cell processes and is controlled by the protein-tyrosine kinases and phosphatases (PTPs). 1 Phosphotyrosine hydrolysis is catalyzed by members of the large and diverse PTP superfamily, and although the specific roles of most of these enzymes have yet to be determined, they can positively or negatively regulate cellular signaling pathways (1,2). The PTPs include enzymes with absolute specificity for phosphotyrosine as well as dual specificity enzymes that can also dephosphorylate serine and threonine residues. Most do not share significant sequence identity, with the exception of the phosphotyrosine-specific receptor and non-receptor-like PTPs, which have stretches of amino acid identity throughout their catalytic domains. Nevertheless, despite dissimilarities in primary sequence, PTPs from diverse subgroups have a remarkably conserved tertiary structure and predicted catalytic mechanism (3,4).
Enzymological and mutational studies have elucidated several features of the catalytic mechanism of PTPs (5). All have an absolutely conserved CX 5 R motif in the active site and nucleophilic attack on the phosphate ester by this essential cysteine results in the formation of a covalent thiophosphate intermediate (6 -8). This phosphate transfer is facilitated by proton donation to the phenolic oxygen of phosphotyrosine from a general acid, identified in the tyrosine-specific PTP1 and Yop51 from Yersinia as the aspartate residue within a conserved WPD motif about 30 -40 residues N-terminal to the active site sequence (9,10). Subsequent transfer of the phosphate to water is likely aided by the same aspartate residue acting as a general base (10). The crystal structures of several PTPs support this mechanism and have further shown that the CX 5 R lies at the base of a phosphotyrosine-binding pocket, with substrate binding inducing the movement of a loop containing the WPD sequence so that the aspartate residue is brought into the catalytic site and in proximity to the leaving group oxygen (11)(12)(13).
Many receptor-like PTPs (RPTPs) have the structural distinction of possessing two homologous tandem catalytic domains in the intracellular region, raising the intriguing question of the functional roles of each repeat. The first, or membrane proximal domains (D1) of LAR, CD45, PTP, and PTP⑀ are catalytically active, whereas the second or membrane distal domains (D2) have either no detectable or extremely low in vitro activity, usually less than 0.1% of the activity of D1 (14 -19). This points to a non-redundant function of the tandem domains, with D2 possibly playing a regulatory role. Where studied, there is also no evidence for an in vivo catalytic action of D2, as the inactivation of D2 of CD45 does not detectably affect the action of CD45 in T cell activation (20). The RPTPs CD45, LAR, PTP, and PTP␣ all display altered D1 activity or substrate specificity in vitro in the absence of D2 (14 -16, 19, 21). A novel interaction of D2 of PTP␦ with D1 of PTP inhibits PTP-D1 activity in in vitro assays, suggesting a role for D2 in regulating the formation and activity of receptor heterodimers (22).
On the other hand, the D2 domain of PTP␣ is exceptional in exhibiting in vitro catalytic activity of up to about 10% of D1 activity toward certain substrates (19,23,24) and cannot be ruled out as a direct contributor to cellular PTP␣ activity. The relatively high activity of PTP␣-D2 is due to two factors, the higher intrinsic activity of PTP␣-D2 compared with other D2 domains and the lower activity of PTP␣-D1 compared with other D1 domains (19). Structure-function studies of PTP␣-D2 present a unique opportunity to assess the minimal sequence requirements that might distinguish the characteristic catalytic properties of the homologous D1 and D2 domains. Most of the highly conserved and invariant residues among the tyrosine-specific PTPs are found within or near the catalytic cleft and are involved in interaction with phosphotyrosine or in actual hydrolysis (12,13). Mutation of many of these residues impairs PTP activity (8,14,25,26). The lack of certain of these apparently critical residues in PTP␣-D2 suggests that its low activity may simply be due to defective substrate binding or catalysis. If so, this would imply a non-catalytic role for D2 rather than an enzymatic function. To investigate this possibility, we have mutated two atypical residues in PTP␣-D2 to conform to the corresponding amino acid found in all other tyrosine-specific PTPs with activity. One such residue is the putative general acid of D2 necessary for formation of the thiophosphate intermediate. The other is a residue that, in PTP1B, is located at the top of the catalytic cleft where it interacts with phosphotyrosine of the substrate (13). The in vitro and in vivo activities of the D2 single and double mutants, as well as those of D1 single and double mutants possessing wild-type D2 residues in these positions, have been analyzed.

EXPERIMENTAL PROCEDURES
Molecular Modeling-Modeling of structures was performed using LOOK (Molecular Applications Group). Sequences of D1 and D2 were initially aligned to the target sequence of PTP1B (sequence identities of 47 and 41%, respectively), and the structure was subsequently modeled based on the algorithm of Lee and Subbiah (the algorithm uses selfconsistent ensemble optimization to determine the global minimum structure resulting in the location of side chains with high accuracy) (27). The target structure was the complexed form of PTP1B(C215S) with phosphotyrosyl-hexapeptide (13).
Phosphatase Assays-The expression, purification, quantitation, and storage of GST-PTP␣ fusion proteins have been previously described (19). Phosphatase activity toward RR-src was measured in 30-l reactions containing 50 mM Mes (pH 6.0), 0.5 mg/ml bovine serum albumin, and 0.5 mM dithiothreitol. Dephosphorylation of pNPP was measured in 450-l reactions containing 50 mM sodium acetate (pH 5.5), 0.5 mg/ml bovine serum albumin, and 0.5 mM dithiothreitol. For K m and V max determinations, RR-src concentrations generally ranged from 2.0 to 25 M, and pNPP concentrations ranged from 0.5 to 10 mM. All reactions were carried out at 30°C and terminated during the linear portion of the reaction. Released 32 P or p-nitrophenol was quantitated as described previously (23). Phosphatase activity was plotted against substrate concentration in the form of a Lineweaver-Burk plot and manually extrapolated to determine K m and V max values.
Cell Culture and Transient Transfections-COS-1 cells were obtained from American Type Culture Collection (Rockville, MD). Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and penicillin/streptomycin in an atmosphere of 5% CO 2 at 37°C. Prior to transfection, confluent monolayers of COS-1 cells were trypsinized and replated in 60-or 100-mm tissue culture dishes and incubated for 16 h until 50 -70% confluency. Cells were transfected with 2-4 g of plasmid DNA by liposome-mediated transfection with 10 l (1 mg/ml) (60-mm dishes) or 30 l (1 mg/ml) (100-mm dishes) of Lipofectin or LipofectAMINE reagent (Life Technologies, Inc.) for 6 h as described by the manufacturer and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum for an additional 18 h prior to harvesting. The empty expression plasmid pXJ41neo was used to normalize the amount of DNA in each transfection. Equivalent amount of the various forms of PTP␣ were expressed.
Western Blots, Immunoprecipitations, and Kinase Assays-The preparation of cell extracts and subsequent Western blot procedures have been described (30). Membranes were immunoblotted with anti-PTP␣-D1 antiserum (no. 2205, raised against a GST-PTP␣ fusion protein containing the first catalytic domain of PTP␣) (1:1000) and followed by goat anti-rabbit IgG conjugated to peroxidase (Sigma) (1:2500), anti-p59 fyn monoclonal antibody (Transduction Laboratories) (1:300) followed by goat anti-mouse IgG conjugated to peroxidase (1:2500), or peroxidase-conjugated anti-phosphotyrosine antibody (Transduction Laboratories) (1:2000). Immunoblots were developed using the ECL system (Amersham Pharmacia Biotech). For immunoprecipitation of p59 fyn , anti-p59 fyn (FYN3, Santa Cruz) was added to the cell lysates (1 l per 100 g of protein) and incubated for 2 h at 4°C. Protein A cell suspension (Sigma) was then added and mixed at 4°C for 2 h. After low speed centrifugation, the immunoprecipitates were washed twice each with lysis buffer and once in 2ϫ kinase assay buffer containing 10 mM Pipes (pH 7.0), 5 mM MnCl 2 , and 0.5 mM dithiothreitol. Part of the immunoprecipitates were used in kinase assays to measure p59 fyn autophosphorylation as described previously (30). Other portions of the immunoprecipitates were probed for p59 fyn as described above. The p59 fyn level, phosphotyrosine content, and kinase activity were quantitated using a GS700 Bio-Rad densitometer.

RESULTS
Modeling of the Catalytic Domains of PTP␣-Alignment of the amino acid sequences of the catalytic D1 domains of 16 active mammalian receptor-like PTPs shows that they possess 42 invariant residues, highlighted in the amino acid sequence of PTP␣-D1 (Fig. 1A). No D2 domain of these RPTPs possesses all 42 invariant residues, although PTP␣-D2 is only lacking 3 of them: a tyrosine at position 536, a leucine at position 549, and an aspartate at position 671 (Fig. 1A). In fact, all of the D2 domains examined are lacking the corresponding tyrosine and aspartate residues, suggesting that the substitution of these residues may be a common denonimator that, in the absence of other obvious defects (for example, the substitution of the essential cysteine residue in the active site of PTP␥, PTP, and PTP-OST), accounts for low D2 activity. Furthermore, the counterpart Tyr in the KNRY motif and the Asp in the WPD motif of non-receptor PTP1B are involved in interactions with the substrate. The crystal structures of PTP1B complexed with phosphopeptide shows that the corresponding invariant tyrosine (Tyr-46) interacts with the phenyl ring of phosphotyrosine of the substrate (13). This tyrosine is one of several hydrophobic, conserved residues that form the recognition site for phosphotyrosine. In the PTP1B structure, the invariant aspartate (Asp-181) is found in the movable WPD loop and is brought into the catalytic site upon substrate binding, where it acts as a general acid to facilitate phosphoester hydrolysis. The involve-ment of these residues in substrate binding and catalysis suggests that their altered nature in PTP␣-D2 could have profound effects on phosphatase activity.
To see if the substitution of these two invariant residues in PTP␣-D2 affected D2 structure, we modeled PTP␣-D1 and -D2 (Fig. 1B). Although the x-ray structure of PTP␣-D1 has been reported (31), it is not complexed with substrate, so models were prepared based on the structure of PTP1B complexed with phosphopeptide (13). Inspection of all three superimposed structures showed that the molecules were largely similar. One area of interest and notable difference among the structures was the WPD loop region. The carboxylate group of Asp-382 in PTP␣-D1 sits close to the phosphate (approximately 4 Å), whereas the carboxylate group of Glu-671 in the WPE loop of PTP␣-D2 sits about 7.45 Å from the phosphate, likely due to the larger glutamate side chain having steric hindrance as well as repulsive forces from adjacent negative charges within the active site pocket (Fig. 1B, inset). The distances between these carboxylate groups and the phenolic oxygen are 3.7 and 6.94 Å, respectively (not shown). The aspartate thus fits within the active site, whereas the glutamate is forced to remain at a distance from active site residues. Additional models were made of mutants where the loop sequence of PTP␣-D1 was changed to WPE and that of PTP␣-D2 to WPD (data not shown). The resulting models had the aspartate and glutamate residues at the expected locations so that glutamate was outside the active site pocket and aspartate was nestled inside the pocket of D2. The conserved KNRY region containing the invariant tyrosine (recognition region) was also analyzed. The model reveals that, as in PTP1B, Tyr-243 of PTP␣-D1 supplies a hydrophobic environment to the phosphotyrosine-binding pocket. The Val-536 in the analogous sequence position in PTP␣-D2 could result in reduced tenacity of binding due to the absence of a complementary surface for phosphotyrosines.
PTP␣-D1 and -D2 Mutants-The roles of Val-536 and Glu-671 in PTP␣-D2 catalysis were tested by mutating these residues, singly and in combination, to the corresponding residues in PTP␣-D1. Likewise, the counterpart invariant residues Tyr-243 and Asp-382 of PTP␣-D1 were mutated to those present in PTP␣-D2. A schematic representation of these mutants is shown in Fig. 2A. Recombinant wild-type and mutant forms of the PTP␣ catalytic domains were expressed as GST-fusion proteins and cleaved with thrombin. The integrity, purity, and amounts of the released PTP␣ proteins were evaluated by SDS-polyacrylamide gel electrophoresis (Fig. 2B) and densitometric scanning prior to the assays described below.
Mutation of Glutamate 671 of PTP␣-D2 to Aspartate or Ala-nine-To test the theory that proton donation to the phenolic oxygen of phosphotyrosine by the more distant hydroxyl moiety of glutamate in wild-type PTP␣-D2 would be less catalytically favorable than from a closer hydroxyl moiety of aspartate, Glu-671 was mutated to either Asp or Ala (E671D and E671A, respectively). The kinetic parameters of activity of these PTP␣-D2 mutants were assayed toward pNPP and the RR-src phosphotyrosyl peptide. As predicted, the E671D substitution positively affected PTP␣-D2 activity toward pNPP, resulting in an enzyme with a 10-fold increased turnover number (k cat ) intermediate to those of wild-type PTP␣-D1 and -D2 and with an overall 4-fold increase in catalytic efficiency ratio (k cat /K m ) ( Table I). In contrast, the E671D substitution had surprisingly little effect on the kinetics of RR-src dephosphorylation by PTP␣-D2 (Table I), indicating that Glu-671 in wild-type D2 is not responsible for the very low activity toward this substrate. Consistent with a role for Glu-671 as a general acid in catalysis, the PTP␣-D2 E671A mutant exhibited less favorable K m and V max values than wild-type D2, with a 60-fold reduction in the catalytic efficiency ratio of pNPP dephosphorylation and such low activity toward RR-src that it could not be reliably measured (Table I).
Mutation of Aspartate 382 of PTP␣-D1 to Glutamate or Alanine-The mutation of Asp-382 to Glu or Ala in PTP␣-D1 (D382E or D382A, respectively) had pronounced effects on the kinetics of dephosphorylation of pNPP and RR-src peptide (Table I). Toward pNPP, the PTP␣-D1 D382E mutant exhibited a 7-fold increase in K m and an 8-fold decrease in V max relative to wild-type PTP␣-D1, resulting in a turnover number and a catalytic efficiency ratio only 2-3-fold higher than wild-type PTP␣-D2. However, toward RR-src, the PTP␣-D1 D382E mutant had an essentially unchanged K m value relative to wildtype PTP␣-D1, whereas the V max decreased 200-fold. Despite this, k cat and the catalytic efficiency ratio of this mutant D1 were still 54-and 136-fold higher than those of wild-type D2. The D1 mutant containing alanine rather than an acidic residue in position 382 (D382A) exhibited K m values similar to those of the D382E mutant for both substrates but had a further reduced rate of activity (Table I). The D1(D382A) mutant and wild-type D1 showed no essential difference in K m values for RR-src, although the K m of the mutant for pNPP was about 5-fold higher than that of wild-type D1 (Table I). The latter contrasts with a report that this same mutation in PTP␣-D1 has virtually no effect on the K m for pNPP (24). We do not know the reason for this difference, but in our experiments the mutation of Glu-671 to Ala in D2 had a similar effect on the K m value for pNPP (Table I). The catalytic efficiency ratios of PTP␣-D1 D382A were about 560-and 6200-fold lower than wild-type D1 for pNPP and RR-src, respectively, consistent with a role of Asp-382 as a general acid in catalysis.
The above results suggest that in the case of pNPP, PTP␣-D1 and -D2 can be induced to behave more, but not entirely, like the other catalytic domain (i.e., D1 like D2 and vice versa) by mutation of their different putative general acids to that present in the counterpart domain, but in the case of the peptide substrate RR-src, the D1 and D2 activities are still quite distinct even after mutation. This is especially true for D2, where the kinetics of RR-src dephosphorylation by the wild-type and E671D forms are virtually the same.
Mutation of Valine 536 of PTP␣-D2 to Tyrosine-Most active PTPs have a conserved tyrosine near the N terminus of the catalytic domain, which in PTP1B (Tyr-46) interacts with the phenyl ring of the substrate phosphotyrosine moiety (13). To test if the presence of valine in PTP␣-D2 accounted for the reduced activity of D2 relative to that of D1, Val-536 was altered to Tyr. This enhanced the kinetics of dephosphorylation of both pNPP and RR-src, with a 3-4-fold increase in turnover numbers (k cat ) and reduced K m values intermediate to that of wild-type D2 and D1 (Table I). Overall, this mutation resulted in a 16-and 5-fold increase in the catalytic efficiency ratio of PTP␣-D2 toward pNPP and RR-src, respectively.
Mutation of Tyrosine 243 of PTP␣-D1 to Valine-To confirm that this conserved tyrosine was important in D1-mediated catalysis, it was mutated to valine, as found in wild-type PTP␣-D2. The PTP␣-D1 Y243V mutant exhibited an increased K m for both pNPP and RR-src, which was equal to or higher than that of wild-type D2 (Table I). Furthermore, the k cat was reduced by about 3-and 490-fold toward pNPP and RR-src, respectively. The resulting catalytic efficiency ratios of the D1 mutant were thus reduced compared with wild-type D1, very significantly so (2300-fold) in the case of RR-src dephosphorylation. Nevertheless, they were still higher than the corresponding catalytic efficiency ratios of wild-type D2.
Kinetic Analyses of PTP␣-D1 and -D2 Double Mutants-Double mutants PTP␣-D1 and PTP␣-D2 were created to examine the combined effects of these point mutations on the kinetic parameters. Toward both pNPP and RR-src, the double mutant PTP␣-D1 (Y243V/D382E) displayed higher K m and lower V max values than those of either of the D1 single mutants (Table I). Compared with wild-type D1, the catalytic efficiency ratios of the double mutant D1 were reduced about 460-fold for pNPP and 150,000-fold for RR-src. Compared with wild-type D2, the double mutant D1 was a 3-fold less efficient phosphatase toward both substrates. This demonstrates that the N-terminal Tyr-243 and the nature of the putative general acid are both critical factors for substrate dephosphorylation by D1. Alteration of these two residues to those found in D2 results in a corresponding alteration in the kinetic behavior of D1 to closely resemble that of D2.
Can D2 be altered to display D1-like kinetics? Analysis of the double mutant D2 (V536Y/E671D) shows a dramatic increase in activity toward pNPP, which is higher than either of the single D2 mutants and which results in a 3-fold higher turnover number (k cat ) and a slightly higher catalytic efficiency ratio than even wild-type D1 (Table I). In contrast, although the kinetic parameters of RR-src dephosphorylation by double mutant D2 are significantly improved relative to the D2 single mutants, they do not approach those of wild-type D1, with a 160-fold lower k cat and a 560-fold lower catalytic efficiency ratio than wild-type D1 (Table I).
In Vivo Substrate Specificity of PTP␣-D1 and -D2-We have previously identified p59 fyn as an in vivo substrate of PTP␣ (30). To determine whether one or both catalytic domains of PTP␣ are involved in the cellular dephosphorylation and activation of p59 fyn , we employed several forms of full-length mutant PTP␣, which contained both tandem catalytic domains but with one of these inactivated by point mutation of the essential cysteine residue to a serine residue (PTP␣-D1D2 S and -D1 S D2), or which contained only one catalytic domain due to deletion of the other (PTP␣-D1 and -D2). Co-expression of wild-type or mutant PTP␣ together with p59 fyn revealed that PTP␣-D1D2 S and PTP␣-D1, which have an active D1 and an inactive or absent D2, dephosphorylated p59 fyn to a similar extent as wildtype PTP␣ (Fig. 3A). Neither PTP␣-D1 S D2 nor PTP␣-D2, having an active D2 and an inactive or absent D1, effected p59 fyn dephosphorylation (Fig. 3A). In addition, only forms of PTP␣ with an active D1 were able to activate p59 fyn , as evaluated by measuring the in vitro autophosphorylation activity of p59 fyn immunoprecipitates (Fig. 3, B and C, lanes 2, 3, and 6). Thus, D1 is responsible for the majority, if not all, of the PTP␣catalyzed dephosphorylation and consequent activation of p59 fyn .

Activity of PTP␣-D1 and -D2 Y/V and D/E Mutants Toward p59 fyn in Vivo-
The clearly different abilities of D1 and D2 to dephosphorylate p59 fyn make this an ideal substrate for testing the in vivo effects of the Tyr/Val and Asp/Glu mutations in each domain. Furthermore, the catalytic ability of mutant D1 or D2 can be examined within the context of the PTP␣ holoenzyme and not as an isolated entity. The activation of p59 fyn was assessed as a measure of co-expressed PTP␣ activity. The single mutation of Asp to Glu in D1 (PTP␣-D1(D382E)D2 S ) abolished p59 fyn activation, as did the double mutation of Tyr and Asp (PTP␣-D1(Y243V/D382E)D2 S ) (Fig. 4, lanes 4 and 6). Although D1 (D382E) exhibited reduced but detectable activity in the in vitro assays described above, the absence of in vivo activity effected by this mutation likely reflects the lower sensitivity of the cellular assay. Nevertheless, this demonstrates that Tyr-243 and Asp-382 are important, if not essential, for D1 catalysis in vivo. No activation of p59 fyn was observed with the corresponding single or double mutant D2 (PTP␣-D1 S D2(E671D) and PTP␣-D1 S D2(V536Y/E671D)) ( Fig. 4, lanes 5 and 7). Thus, the in vivo activity of the D2 mutants parallels the lack of in vitro activity seen toward the phosphopeptide RR-src rather than the increased in vitro D2-phosphatase activity observed with pNPP.

DISCUSSION
To address the basis of the kinetic differences between PTP␣-D1 and PTP␣-D2, we have mutated two residues that are predicted to play key roles in catalysis and which differ between the two domains. One of these amino acids is the acidic residue in the putative movable loops, which is an aspartate in the motif WPD of loop 13 (31) of PTP␣-D1 and a glutamate in the corresponding WPE sequence of PTP␣-D2. Mutation to a non-acidic alanine in D1 or D2 resulted in a pronounced reduction in k cat , particularly so with the peptide substrate RR-src, indicating that Asp-382 and Glu-671 act as a general acid in D1 and D2 catalysis, respectively. This, together with the threedimensional structural similarity between the substrate-free forms of PTP␣-D1 (31) and PTP1B (11) and also between the PTP␣-D1 and -D2 models (Fig. 1B), which were prepared based on the crystal structure of PTP1B complexed with phosphopeptide (13), suggests that both domains share a catalytic mechanism in common with each other and with other PTPs. The other amino acid we mutated is a hydrophobic residue that is a tyrosine (Tyr-243) in the KNRY motif of loop 1 (31) of PTP␣-D1 and a valine (Val-536) in the corresponding position of PTP␣-D2. In PTP1B, the equivalent tyrosine residue is located at the top of the catalytic cleft and interacts with the main-chain atoms and phenyl ring of phosphotyrosine in the substrate (13). We report here that both the hydrophobic residue and the general acid are key determinants of the catalytic activities of FIG. 3. Dephosphorylation and activation of p59 fyn by various forms of PTP␣. A, COS-1 cells were transfected with 2 g of p59 fyn cDNA and 0.5 g of empty plasmid (neo) or 0.5 g of wild-type PTP␣ or mutant PTP␣ cDNA. Cell lysates were resolved by SDS-polyacrylamide gel electrophoresis and probed with anti-phosphotyrosine and anti-p59 fyn antibodies, and the signals were quantitated by densitometry to determine the phosphotyrosine content per unit of p59 fyn . The p59 fyn dephosphorylation in the presence of the various PTP␣ species (where the subscript S denotes a mutation of the essential Cys to Ser in that catalytic domain, see "Experimental Procedures") was calculated by taking the phosphotyrosine content of p59 fyn from COS-1 cells expressing p59 fyn alone as 100%. Bars shown are the means from at least three independent experiments, and the error bars indicate the mean Ϯ S.E. B, cells were transfected as in A. The p59 fyn was immunoprecipitated from 500 g of cell lysate, and a portion was used in an immunocomplex kinase assay while the rest was probed for p59 fyn . Following densitometric quantitation, the extent of p59 fyn autophosphorylation (equalized per unit of p59 fyn protein in each immunoprecipitate) was calculated by taking the kinase activity of p59 fyn from COS-1 cells expressing p59 fyn alone as 100%. In the designation of the mutant forms of PTP␣, the subscript S denotes a mutation of the essential Cys to Ser in that catalytic domain (see "Experimental Procedures"). The p59 fyn was immunoprecipitated from 500 g of lysate, and a portion was used in an immunocomplex kinase assay while the rest was probed for p59 fyn . Following densitometric quantitation, the extent of p59 fyn autophosphorylation (equalized per unit of p59 fyn protein in each immunoprecipitate) was calculated by taking the kinase activity of p59 fyn from COS-1 cells expressing p59 fyn alone as 100%. Bars shown are the means from three independent experiments, and the error bars indicate the mean Ϯ S.E. B, representative results from one of the three experiments described in A. The p59 fyn immunoprecipitates were assayed for p59 fyn autophosphorylation (top panel) and p59 fyn amount (bottom panel). Numbering of the lanes corresponds to that described in A.
PTP␣-D1 and -D2. However, the contribution of these residues to the efficiency of dephosphorylation is determined in a large part by the nature of the substrate and highlights major differences between D1 and D2, as discussed further below.
Mutation of either the hydrophobic residue (Tyr-243) or of the general acid (Asp-382) in D1 to the corresponding residue present in D2 (Val or Glu, respectively) reduced the catalytic efficiency of D1 significantly toward the in vitro substrates pNPP and RR-src. Neither mutation alone was sufficient to bring this parameter of activity down to the level observed with wild-type D2, and the D1(Y243V) mutant phosphatase was less efficient than the D1(D382E) mutant in dephosphorylating RR-src, whereas the converse was true for pNPP. The dramatic effect of the D1(Y243V) mutation in reducing the efficiency of RR-src dephosphorylation (4 orders of magnitude from wildtype D1), compared with its much lesser effect in reducing the efficiency of pNPP dephosphorylation (1 order of magnitude from wild-type D1), suggests that Tyr-243 is involved not only in interactions with the phenyl ring of the substrate but also with other elements of the phosphotyrosine microenvironment such as adjacent residues of the substrate. This would be consistent with its positioning near the top of the catalytic cleft. The double mutant D1 (Y243V/D382E) was catalytically very comparable to wild-type D2 with both of these substrates, supporting the view that the observed deviation from invariant residues in these positions of D2 could alone be responsible for the relatively poor enzymatic activity of D2. The synergistic reduction in catalytic activity observed with the double mutant D1 indicates that the orientation of phosphotyrosine in the binding pocket through its interaction with the hydrophobic residue at position 243 may position phosphotyrosine for protonation by the general acid, be it aspartate or, less optimally, glutamate. The substitution of valine for Tyr-243, where valine lacks the bulkier aromatic ring of tyrosine, may eliminate or reduce the interaction with phosphotyrosine, which positions the latter for ready protonation. In accord with the mutant D1 activities in vitro, the in vivo dephosphorylation of p59 fyn by single (D382E) or double (Y243V/D382E) mutant D1 within the context of the PTP␣ holoenzyme was likewise substantially reduced.
The ability of two point mutations to alter PTP␣-D2 to an enzyme with catalytic efficiency comparable with that of D1 is consistent with the reverse effects observed with the D1 double mutant. Mutation of Glu-671 and Val-536 in D2 to the Asp and Tyr, respectively, found in these positions in wild-type D1 converts D2 to a D1-like pNPP phosphatase. Each single mutation in D2 improves its catalytic efficiency toward pNPP by 1 order of magnitude, whereas the double mutant displays a synergistic improvement in this measure of catalysis. Thus, it appears that, at least with pNPP as a substrate, the different natures of these critical amino acids at positions 243/382 of D1 and 536/671 of D2 can account for the kinetic differences between D1 and D2.
However, a very limited effect on D2 catalytic efficiency toward the phosphopeptide substrate RR-src was found upon mutation of Val-536 and Glu-671 to Tyr and Asp, respectively. With the single mutants, the D2(V536Y) mutant showed a 5-fold increase in k cat /K m , but essentially no change in k cat /K m was observed with the D2(E671D) mutant. In another study that examined the effect of the single Glu-671-to-Asp mutation, there were modest changes in the kinetic properties (3-4-fold increase in k cat /K m ) of PTP␣-D2 toward three other phosphotyrosyl peptides (24). More importantly, in stark contrast to the D1-like behavior of double mutant D2 on pNPP and the 150,000-fold reduction in catalytic efficiency toward RR-src effected by mutation of these residues in D1, our combined mutation of both these amino acids in D2 effected only a 90-fold gain in D2 catalytic efficiency for RR-src. This did not approach the catalytic efficiency of the wild-type D1 enzyme. Thus, the potentially optimal positioning and protonation of phosphotyrosine bestowed by double mutation of Val and Glu are important but not sufficient to confer a D1-like activity upon D2. This is not a unique feature of the RR-src peptide since mutant forms of D2 (D671E and V536Y/E671D) are also unable to dephosphorylate p59 fyn in vivo. The more complex microenvironment of phosphotyrosine within the RR-src peptide or p59 fyn , compared with the free phosphotyrosine-like structure of pNPP, must account for the differences in double mutant D2 activity toward these substrates and for the difference in double mutant D2 and wild-type D1 activities toward RR-src and p59 fyn . Other regions of difference in D2 and D1, which likely lie outside the catalytic cleft, are thus key to overall substrate recognition and specificity. To date, little is known of the features that determine PTP substrate specificity. Site-directed mutagenesis of PTP␣-D2(V536Y/E671D) or regional replacement with PTP␣-D1 sequences could identify residues or parts of these catalytic domains involved in substrate recognition. This would be optimally carried out in conjunction with information from an x-ray structure of D1 complexed with substrate.
These studies make the important point that there is an absolute difference in substrate specificity between D1 and D2. In the absence of specificity determinants on the substrate (i.e., with pNPP), it is clear that D2 is catalytically suboptimal due to the nature of amino acids 536 and 671. However, this statement cannot be made respecting substrates with more complex structure as those examined to date (i.e., RR-src, p59 fyn ) do not appear to be recognized or bound properly by D2. This raises the questions of whether there exists a cellular phosphoprotein for which D2 exhibits specificity and, if so, whether interaction with such a substrate effects a novel orientation of the phosphate in the catalytic cleft so as to permit efficient dephosphorylation.