Comparative kinetic analysis and substrate specificity of the tandem catalytic domains of the receptor-like protein-tyrosine phosphatase alpha.

The catalytic activity and substrate specificity of protein-tyrosine phosphatase α (PTPα) is primarily controlled by the membrane proximal catalytic domain (D1). The membrane distal (D2) domain of PTPα by itself is a genuine PTPase, possessing catalytic activity comparable to that of D1 using aryl phosphates as substrates. Surprisingly, kcat and kcat/Km for the D2-catalyzed hydrolysis of phosphotyrosine-containing peptides are several orders of magnitude reduced in comparison with those of D1. Substitution of the putative general acid/base Glu-690 in D2 by an Asp, which is invariably found in the WPD motifs in all cytoplasmic PTPases and all the D1 domains of receptor-like PTPases, only increases the kcat for D2 by 4-fold. Thus the much reduced D2 activity toward peptide substrates may be due to structural differences in the active sites other than the general acid/base. Alternatively, the D2 domain may have a functional active site with a highly stringent substrate specificity. PTPα display modest peptide substrate selectivity and are sensitive to charges adjacent to phosphotyrosine. In the sequence context of DADEpYLIPQQG (where pY stands for phosphotyrosine), the minimal sizes recognized by PTPα are either ADEpYLI or DADEpY-NH2.

Protein-tyrosine phosphatases (PTPases) 1 catalyze the hydrolysis of phosphoryl groups on tyrosine residues in proteins that are introduced by protein-tyrosine kinases. A tightly balanced phosphatase and kinase activity is required for proper cellular functions (1,2). PTPases constitute a growing family (Ͼ70 members) of enzymes that can be structurally categorized into two major groups: receptor-like (transmembrane) and nonreceptor (intracellular) PTPases (3). Although many PTPases are proteins of greater than 400 amino acids, their catalytic domains are usually contained within a span of 250 residues referred to as the PTPase domain. This domain is the only structural element that has amino acid sequence identity among all PTPases from bacteria to mammals (4). The hallmark that defines the PTPase family is the active site sequence (H/V)C(X) 5 R(S/T), called the PTPase signature motif in the catalytic domain (5,6). The receptor-like PTPases, exemplified by the leukocyte phosphatase CD45, generally have an extracellular domain, a single transmembrane region, and cytoplasmic PTPase domain(s). The intracellular PTPases, exemplified by PTP1B, contain a single catalytic domain and various amino-or carboxyl-terminal extensions including SH2 domains that may have targeting or regulatory functions.
The cytoplasmic segment of many receptor-like PTPases have two tandem PTPase domains: D1, which is proximal to the membrane, and D2, which is distal to the membrane. The existence of homologous PTPase domains in the receptor-like PTPases raises the interesting possibility of differential functions or regulations of the two domains. However, the significance of the repeated PTPase domain in the receptor-like PTPases is not clear. One important question is whether both PTPase domains in the receptor-like phosphatases are catalytically active? There is evidence that suggest that D2 is catalytically inactive and may only play, if any, a regulatory role. This is supported by the following observations. First, there are receptor-like PTPases that lack either the essential Cys residue within the PTPase signature motif in D2 (HPTP␥ and RPTP␤) (7,8) or the entire D2 domain (PTP␤ and DPTP10D) (7, 9 -11). Second, replacement of the Cys residue in the signature motif of D1 in LAR and CD45 abolishes the phosphatase activity while the same substitution in the signature motif in D2 has little or no effect on PTPase activity (12,13). Third, recombinant D1 of LAR or LCA (13)(14)(15) exhibits phosphatase activity that are identical to the wild-type double domain PTPase, whereas constructs for D2 alone show no measurable activity (13,14). Last, inactivation of D1 in PTP␣ (16) or CD45 (17) results in loss of biological activity of these two PTPases, indicating that catalytic activity of D1 is essential for the function of these PTPases.
On the other hand, there are reports that indicate that in some receptor-like PTPases the second domain may display catalytic activity. For example, elimination of the critical Cys in D1 of LAR or PTP results in Ͼ99% loss of activity (18,19). The residual activity (usually less than 0.1%) is concluded to be associated with D2. Recent evidence suggest that under some conditions D2 of CD45 could be a viable phosphatase (20), and that certain structural elements in D2 may be required for full expression of CD45 PTPase activity in D1 (21,22). The most compelling data that identify intrinsic phosphatase activity with D2 come from work on PTP␣. Working with purified recombinant individual PTPase domains, Wang and Pallen (23) showed that D2 of PTP␣ could hydrolyze p-nitrophenyl phosphate and dephosphorylate a phosphotyrosine-containing synthetic peptide, albeit at a very low rate relative to D1. Furthermore, recombinant RPTP␣ containing an inactive D1 still retained a low but measurable activity against the Tyr-789-phosphorylated catalytically inactive RPTP␣ (24).
PTP␣ is a receptor-like PTPase that has a small, highly glycosylated extracellular segment, a transmembrane region, and an intracellular segment possessing two PTPase domains (7,25). PTP␣ is expressed ubiquitously, although the levels of expression vary. During mouse development, enhanced expression was detected in derivatives of the neural crest (26). Evidence suggest that it may be involved in the activation of c-src by dephosphorylation of Tyr-527 (16,27) and the down-regulation of insulin receptor signaling (28). To better understand the relative importance of the two PTPase domain in PTP␣ catalysis, we have compared their intrinsic phosphatase activities using a variety of low molecular weight aryl phosphates and Tyr(P)-containing peptides.
Recombinant Enzymes-The PTP␣-glutathione S-transferase fusion protein constructs were transformed into Escherichia coli BL21(DE3). A 10-ml overnight culture of BL21(DE3) cells containing pGEX recombinant plasmids for PTP␣ were diluted 100-fold into 1 liter of 2ϫYT medium with 100 g/ml ampicillin. The culture was grown at 37°C until the absorbance at 600 nm reached 0.6. Expression of glutathione S-transferase fusion proteins were then induced by the addition of 100 M isopropyl-1-thio-␤-D-galactopyranoside. Following the induction, the cells were allowed to grow overnight at room temperature. Purification of the PTP␣-glutathione S-transferase fusion proteins by glutathioneagarose beads and the subsequent thrombin cleavage of the glutathione S-transferase fusion protein were accomplished as described (31). The recombinant PTP␣ proteins were at least 90% in purity as judged by SDS electrophoresis.
pH and Leaving Group Dependences-The PTPase activity was assayed at 30°C in a reaction mixture (0.2 ml) containing appropriate concentrations of pNPP as substrate. Buffers used were as follow: pH 4.0 -5.5, 100 mM acetate; pH 5.5-6.5, 50 mM succinate; pH 6.6 -7.3, 50 mM 3,3-dimethylglutarate, and pH 7.5-9.0, 50 mM Tris. All of the buffer systems contained 1 mM EDTA and the ionic strength of the solutions were kept at 0.15 M using NaCl. The reaction was initiated by addition of enzyme and quenched after 2-3 min by addition of 1 ml of 1 N NaOH.
The nonenzymatic hydrolysis of the substrate was corrected by measuring the control without the addition of enzyme. The amount of product p-nitrophenol was determined from the absorbance at 405 nm using a molar extinction coefficient of 18,000 M Ϫ1 cm Ϫ1 . Michaelis-Menten kinetic parameters were determined from a direct fit of the v versus [S] data to the Michaelis-Menten equation (Equation 1) using the nonlinear regression program GraFit (Erithacus Software).
v ϭ k cat ͓E͔͓S͔ K m ϩ ͓S͔ (Eq. 1) pH dependence of k cat data were fitted to Equation 2 using a nonlinear least-squares regression program (KaleidaGraph, Synergy Software). In Equation 2, k cat lim , the pH independent turnover number, and K 1 app and K 2 app , are the apparent ionization constants of the enzyme-substrate complex (32). Leaving group dependence was conducted at pH 6.0 and 30°C with aryl phosphates having leaving group pK a values from 7.14 to 10.26 (33).
Determination of Kinetic Constants Using Tyr(P)-containing Peptides as Substrates-Initial rates for the PTPase catalyzed hydrolysis of Tyr(P)-containing peptide substrates were determined by following the production of inorganic phosphate (30). k cat and K m values were calculated from a direct fit of the v versus [S] data to Equation 1 using the nonlinear regression program GraFit (Erithacus Software). The PTPase-catalyzed hydrolysis of Ty(P)-containing peptides can also be monitored continuously at 305 nm for the increase in tyrosine fluorescence with excitation at 280 nm (30).
Under this condition, the reaction is first-order with respect to [S]. At a given enzyme concentration, the observed apparent first-order rate constant is equal to (k cat /K m )[E]. The substrate specificity constant k cat /K m value is calculated by dividing the apparent first-order rate constant by the enzyme concentration. Fluorometric determinations were performed on a Perkin-Elmer LS50B fluorometer. The instrument was equipped with a water-jacketed cell holder, permitting maintenance of the reaction mixture at the desired temperature (30°C). Inhibition Studies-The following procedure for the preparation of stock solution of sodium orthovanadate was adopted from Dr. Mike Gresser's laboratory: vanadium (V) oxide was dissolved in 1 mol eq per vanadium atom of 1.0 M aqueous NaOH. The resulting orange solution (mainly decavanadate) was boiled, allowed to stand overnight, and pH adjusted to 10. The final solution was colorless containing mainly orthovanadate (34). The inhibition constants for inorganic phosphate, arsenate, vanadate, and p-nitrophenol were determined for the phosphatases in the following manner. When oxyanions were used as inhibitors, the initial rate at various pNPP concentrations was measured by following the production of p-nitrophenol (35). When p-nitrophenol was used as an inhibitor, the initial rate at various pNPP concentrations was measured by following the production of inorganic phosphate (33). The inhibition constant and inhibition pattern were evaluated using a direct curve-fitting program KINETASYST (IntelliKinetics, State College, PA).
We next compared the ability of the individual PTPase domains to dephosphorylate Tyr(P)-containing peptides corresponding to Tyr autophosphorylation sites in platelet-derived growth factor receptor and EGF receptor, or the Tyr phosphorylation site (Tyr-789) in the COOH-terminal of RPTP␣ (24). The dephosphorylation reaction was followed by monitoring the production of inorganic phosphate with a colorimetric procedure (30). Kinetic constants for the hydrolysis of peptide substrates are listed in Table II. In comparison with the double PTPase domain-containing bPTP␣, bPTP␣-D1 displays slightly reduced k cat values toward the peptide substrates, but the K m values are indistinguishable. However, in contrast to small nonpeptidic substrate pNPP, the k cat values for the bPTP␣-D2catalyzed hydrolysis of Tyr(P)-containing peptides decrease by 2 to 3 orders of magnitude compared with bPTP␣-D1, while the K m values increase by 1 to 2 orders of magnitude. Thus, the combined effects translate into a difference in substrate specificity constant (k cat /K m ) in the range of 8,000 to 40,000-fold in favor of bPTP␣-D1 over bPTP␣-D2.
The Conserved General Acid/Base in the PTPase Domains-To begin to gain insight into the structural basis for the differential activity exhibited by D1 and D2, we have performed an amino acid sequence alignment for both of the PT-Pase domains in several receptor-like PTPases including RPTP␣ (Fig. 2). Invariant residues in D1 and D2 are listed below the sequence alignment. It is clear that D1 is more conserved than D2. Previous biochemical studies have demonstrated that the Cys residue in the PTPase signature motif (H/V)C(X) 5 R(S/T) is essential for phosphatase activity and formation of a covalent cysteinyl phosphoenzyme intermediate (39 -41). Mutations at the Cys residue result in complete loss of phosphatase activity. The invariant Arg residue in the signature motif has been shown to play an important role in substrate binding and transition state stabilization (5). Substitutions of Arg-409 in the Yersinia PTPase by either a Lys or an Ala cause a decrease in k cat by nearly 4 orders of magnitude. In addition to nucleophilic catalysis and transition state stabilization, PTPases also utilize an invariant Asp residue in the WPD motif ( Fig. 2) to facilitate the catalytic turnover: the Asp residue acts as a general acid to donate a proton to the phenolate leaving group in the phosphoenzyme formation step, while it acts as a general base to activate a nucleophilic water molecule in the phosphoenzyme hydrolysis step (4,42). Replacement of Asp-356 by an Asn in Yersinia PTPase reduces the turnover number by 3 orders of magnitude. As can be seen in Fig. 2, the PTPase signature motif and the WPD motif are strictly conserved among the D1 domains. These two motifs are not conserved among the D2 domains.
A closer examination of the sequence alignment shows that in the D2 domain of PTP␥, the essential Cys and Arg residues in the active site are replaced by an Asp and a Ser, respectively. Furthermore, the Asp residue in the WPD motif is replaced by an Asn. Thus, it is unlikely that the second PTPase domain of PTP␥ will possess any catalytic activity. Although the active site Cys is retained in the D2 domain of CD45, it lacks the critical Arg residue while the Asp residue is substituted by a Val. It is thus questionable whether the second PTPase domain of CD45 will exhibit any significant phosphatase activity. 5 Interestingly, the D2 domains of PTP␣, PTP⑀, and LAR preserve all of the essential features of the PTPase signature motif, but the Asp residue in the WPD motif has been switched to a Glu. Based on the sequence alignment, Asp-401 is the putative general acid/base in the WPD motif in bPTP␣-D1 whereas Glu-690 is the putative general acid/base in the WPE motif in bPTP␣-D2. It is found that replacement of Asp with Glu or vice versa generally leads to a reduction of up to 3 orders of magnitude in catalytic efficiency in enzymes that require a carboxylate group as a general acid/base (43)(44)(45)(46). Since an Asp is always found to be the general acid/base in the catalytic domains of cytoplasmic PTPases and the D1 domains of receptorlike PTPases, and since the D2 domains of receptor-like PTPases have much reduced catalytic activity, we asked the following question: can substitution of Glu-690 by an Asp in bPTP␣-D2 increase its catalytic activity to a level comparable to bPTP␣-D1?
To answer this question, we have converted Glu-690 in bPTP␣-D2 to an Asp by site-directed mutagenesis. We have also made the Asp-401 to Ala mutation in bPTP␣-D1 to test the importance of the Asp residue in the WPD motif. Although 5 A previous study suggested that the D2 domain of CD45 was a functional PTPase (20). This was based on the following observations: 1) phosphatase activity could be detected when CD45 lacking the PTPase signature region in D1 was expressed in mammalian cells, and 2) a proteolytically cleaved fragment of CD45 containing most of D2 and the COOH-terminal part of D1 (including the PTPase signature sequence) displayed phosphatase activity. Although it is possible that the D2 domain of CD45 may employ a different mechanism from the one used by all the cytoplasmic PTPases and the D1 domains of receptor PTPases for phosphate monoester hydrolysis (see above), this needs to be further evaluated and verified.    pNPP as a substrate, the k cat for bPTP␣-D1/D401A is 900-fold slower than that of wild-type bPTP␣-D1, while the K m for bPTP␣-D1/D401A is identical to that of bPTP␣-D1 (Table I). These results are similar to those observed with cytoplasmic PTPases and suggest that Asp-401 is indeed important for the activity of D1 and that a common catalytic strategy is shared by both the receptor-like and soluble PTPases.
Asp substitution for Glu at 690 in bPTP␣-D2 has modest effects on the kinetic parameters using pNPP as a substrate (Table I). A 4-fold increase in k cat and nearly identical K m were observed for bPTP␣-D2/E690D when compared with the wildtype D2. In fact, bPTP␣-D2/E690D has a k cat value that is only 2-fold lower than that of bPTP␣-D1 using pNPP as a substrate. Similar results were obtained with the Tyr(P)-containing peptides (Table II), namely D2/E690D displays 3-4-fold higher k cat value and identical K m value in comparison with the native D2 domain. However, the k cat and k cat /K m values of bPTP␣-D2/ E690D toward Tyr(P)-containing peptides are still 3 and 4 orders of magnitude lower when compared with bPTP␣-D1 (Table II). These results suggest that substitution of Glu-690 for an Asp in D2 may not be the main cause that D2 is a much less efficient phosphatase than D1 toward phosphopeptide substrates.
Inhibition by Inorganic Phosphate, Arsenate, Vanadate, and p-Nitrophenol-Inhibition constants were determined for inorganic phosphate, arsenate, vanadate, and p-nitrophenol at pH 6.0 (Table III). In general, effects of inhibitors to bPTP␣-D1catalyzed reaction are similar to that catalyzed by the double domain construct bPTP␣. Substitution of Glu-690 by an Asp does not enhance significantly the affinity of bPTP␣-D2 toward the oxyanions. Oxyanions were competitive inhibitors with respect to pNPP, whereas p-nitrophenol was noncompetitive. The pattern of product (phosphate and p-nitrophenol) inhibition suggests that D1 and D2 utilize a similar kinetic scheme (Scheme 1), particularly in terms of the order of addition of substrates and release of products (33). Inorganic phosphate usually shows K i values around 1-5 mM against PTPases (33,42,48). Compared with other PTPases, bPTP␣-D1 binds phosphate with an unusually weak affinity (K i ϭ 120 mM) which is even 2-fold lower than bPTP␣-D2. Arsenate binds to bPTP␣-D1 and bPTP␣-D2 more tightly than phosphate, however, the in-crease in affinity is seen greater with bPTP␣-D1 than bPTP␣-D2. Thus, arsenate inhibits the bPTP␣-D1-catalyzed reaction (K i ϭ 0.25 mM) 30-fold more potently than the bPTP␣-D2catalyzed reaction.
Vanadate is by far the most potent small molecule inhibitor for PTPases. Because vanadate can adopt five-coordinate structures readily, it was expected that a covalent bond can be formed between the active site thiol and the vanadium atom. The structure of Yersinia PTPase complexed with vanadate revealed continuous electron density between the bound anion and the active site Cys-403, consistent with a covalent bond (49). Indeed, vanadate in the active site adopts a slightly distorted trigonal bipyramidal geometry that resembles the transition state for the hydrolysis of the thiophosphate enzyme intermediate in the PTPase-catalyzed reaction (6,50). Previous studies with vanadate have often included EDTA in the assay buffers. Since EDTA forms a very stable 1:1 complex with vanadate even at micromolar concentrations of both EDTA and vanadate (51), the presence of EDTA in the assay solution will reduce the free vanadate concentration. Here we performed all of our inhibition experiments with vanadate in the absence of EDTA. Vanadate shows an apparent K i of 5.5 M toward bPTP␣-D1 whereas the K i for bPTP␣-D2 is 1.9 mM. Thus, bPTP␣-D1 binds vanadate 350-fold tighter than bPTP␣-D2. Furthermore, bPTP␣-D1 binds vanadate (a crude transition state analog) 22,000-fold more tightly than phosphate (a crude substrate analog), while bPTP␣-D2 binds vanadate only 40-fold more tightly than phosphate. Since one of the major means an enzyme utilizes to catalyze a reaction is via transition state stabilization, it is understandable that bPTP␣-D1 is a better PTPase than bPTP␣-D2. Because vanadate can take on different forms in solution, and because the enzyme-bound vanadate has a different structure from the tetrahedral vanadate in solution, it is very likely that the kinetically determined K i may underestimate how good a transition state analog the PTPasebound vanadate really is (52).
pH Dependence-The pH dependence of the steady-state kinetic parameters for the various forms of PTP␣ were studied in detail using pNPP as a substrate. Within the pH range investigated, K m values only exhibited moderate variations from those reported in Table I. The pH-k cat profile is shown in Fig. 3. The line that is drawn through the experimental data is based on the nonlinear least-squares fit of the data to Equation 2. pK 1 app , pK 2 app , and the pH-independent maximum turnover number k cat lim are listed in Table IV. It appears that bPTP␣-D1 displays a similar pH rate profile to that of bPTP␣, except k cat lim for bPTP␣ is twice that of bPTP␣-D1. Although substitution of an Asp in place of Glu-690 in bPTP␣-D2 increase the k cat lim by 4-fold, it does not alter the pH rate profile. It is also clear that bPTP␣-D1 and bPTP␣-D2 display different pH rate profile: the pH optimum for bPTP␣-D1-catalyzed hydrolysis of pNPP is 0.5 pH unit lower than that of bPTP␣-D2.
Leaving Group Dependence-We have shown above that when compared with bPTP␣-D1, the k cat and k cat /K m values of bPTP␣-D2 toward pNPP are reduced by 10-and 50-fold, respectively (Table I). On the other hand, when Tyr(P)-containing peptides are used as substrates, up to 3 and 4 orders of magnitude reduction are observed with k cat and k cat /K m , respectively (Table II). One possible explanation for this differential effects may be that Glu-690 in bPTP␣-D2 is a weaker general acid than an Asp in bPTP␣-D1, so that substrates with Tyr (pK a ϭ 10.07) as a leaving group becomes more difficult to expel because the negative charge developed on the phenolate oxygen in the transition state (47) cannot be efficiently stabilized by Glu-690. Leaving groups in substrates such as pNPP (pK a ϭ 7.14) are much easier to expel since there is much less demand for general acid catalysis.
To test this hypothesis, we investigated the leaving group dependence of the PTPase-catalyzed hydrolysis of aryl phosphates. Since Tyr(P)-containing peptides are more complex and different from pNPP in terms of structure, we chose several low molecular weight aryl phosphates to minimize the influence on the reaction by steric effects. As shown in Fig. 4, both bPTP␣ and bPTP␣-D1 exhibit no leaving group dependence and show very similar k cat values for hydrolysis of aryl phosphates that differ markedly in their leaving group pK a (for example, 7.14 for p-nitrophenol and 10.26 for 4-methylphenol). Interestingly, bPTP␣-D2 and bPTP␣-D2/E690D display only moderate leaving group dependences, with ␤ lg values of Ϫ0.092 Ϯ 0.041 and Ϫ0.11 Ϯ 0.051, respectively. Thus although bPTP␣-D2/E690D is four times more active than bPTP␣-D2, it does not alter the leaving group sensitivity. From these observations, it appears that the 3-4 orders of magnitude decrease in reactivity toward Tyr(P)-containing peptides by bPTP␣-D2 and bPTP␣-D2/ E690D is not due to the intrinsic lower stability of the Tyr leaving group. This is because, in comparison with pNPP, one would only predict a 2-fold decrease in rate for phosphotyrosine based on the leaving group dependences of bPTP␣-D2 and bPTP␣-D2/E690D catalyzed aryl phosphate hydrolysis reaction.
Peptide Substrate Amino Acid Sequence Specificity-Substrate specificity in terms of peptide size and amino acid se-quence preference were investigated using Tyr(P)-containing peptides. The PTPase-catalyzed dephosphorylation reaction was continuously monitored spectrofluorimetrically (excitation at 280 nm and emission at 305 nm) at pH 6.0 and 30°C. This assay takes advantage of the increase in fluorescence at 305 nm when the phosphate is hydrolyzed from Tyr(P) (30). At substrate concentrations Ͻ Ͻ K m , the Michaelis-Menten equation (Equation 1) reduces to Equation 3, and the reaction is first order with respect to [S]. With automatic data collection and nonlinear least square fit analysis, the first-order rate constant can be determined with a precision of less than 4%. The observed apparent first-order rate constant is equal to (k cat /K m )[E], from which the substrate specificity constant k cat /K m value can be calculated by dividing the apparent firstorder rate constant with the enzyme concentration. Additional advantages of this assay include that very small amount of peptide substrate is used per reaction and that accurate substrate concentration determination is not necessary. In general, substrate concentrations were at least 15-fold lower than the K m values (Table II) and ranged from 2 to 10 M. To ensure first-order kinetics, reactions were conducted at several substrate concentrations. Since the reaction is first order in nature, the same rate constant is obtained regardless the substrate concentration used. Because the k cat /K m values for bPTP␣-D2 and bPTP␣-D2/E690D are 5 orders of magnitude slower than those of bPTP␣, they cannot be determined by the fluorescence assay.
The peptide substrate amino acid sequence specificity was probed with Tyr(P)-containing peptides corresponding to Tyr autophosphorylation sites in platelet-derived growth factor receptor, EGF receptor, Neu oncogene, Src kinase, or RPTP␣ ( Table V). The substrate specificity constants (k cat /K m values) determined by the fluorescence assay are very similar to those calculated from the ratio of k cat and K m , which are determined by the initial rate method (Table II). In agreement with results described above, bPTP␣ and bPTP␣-D1 display similar substrate specificity toward peptide substrates, but the k cat /K m values for bPTP␣ are slightly higher that those of bPTP␣-D1, which is likely due to the slightly higher k cat for bPTP␣. In general, bPTP␣ and bPTP␣-D1 only exhibit a moderate selectivity toward the various peptides. Only a 8-fold difference in k cat /K m was observed among the peptides. Although the enzyme does appear to prefer peptides with two acidic residues  immediately NH 2 -terminal to Tyr(P) (DADEpYLIPQQG and DAEEpYLVPQQG), there is also exception to this (DTSSVLpYTAVQ).
The Minimum Length of a Peptide Substrate-To determine the minimum length of a peptide substrate for bPTP␣ and bPTP␣-D1, we selected a series of various sized phosphopeptides modeled after the autophosphorylation site in EGF receptor (EGFR) at Tyr-992 (Table VI). For the sake of discussion, we will focus the results on bPTP␣, since identical conclusions can be reached from results on bPTP␣-D1. Phosphotyrosine (Tyr(P)) by itself is not a good substrate. In fact, its k cat /K m value cannot be determined by the fluorescence assay because of the slow rate. We therefore determined its k cat and K m by the initial rate method using a molar absorption coefficient of 2,400 for tyrosine (29). The calculated k cat /K m values for Tyr(P) are listed in Table VI. It seems that the low k cat /K m for Tyr(P) is primarily due to its high K m : 22.1 mM for bPTP␣ (19.5 mM for bPTP␣-D1). As shown in Table VI, the k cat /K m value for the parent peptide DADEpYLIPQQG (EGFR 988 -998 ) is 1090-fold higher than Tyr(P). This indicates that structural elements surrounding Tyr(P) in the peptide contribute to specificity.
Strikingly, Ac-Tyr(P)-NH 2 , which is obtained by acetylating the free amino group and amidating the free carboxyl group of Tyr(P), exhibit a k cat /K m that is 76-fold higher than Tyr(P). Interestingly, the tripeptide EpYL is only 4.7-fold better than Tyr(P). Since acetyl group is smaller than a Glu residue, the dramatic enhancement in substrate efficacy for Ac-Tyr(P)-NH 2 seems to mainly arise from the elimination of charges on the free amino and carboxyl groups, rather than the increase in size. The addition of an Ile to the COOH-terminal side of EpYL (generating EpYLI) results in only a 1.9-fold improvement, while addition of an Asp to NH 2 -terminal side of EpYL (generating DEpYL) leads to a 1.5-fold increase in activity. Although the acetyl group is smaller than an Asp residue, the kinetic parameter for Ac-EpYL is 17-fold higher than that of DEpYL. This may again be due to the elimination of the positive charge on the NH 2 -terminal amino group of EpYL.
The incorporation of an additional residue, Ile, to the COOHterminal side of DEpYL produces a pentapeptide (DEpYLI) that is 56 times more efficient than DEpYL. Thus, the extension of DEpYL beyond the ϩ1 position by one residue is more than enough to offset the undesirable charge effect on the NH 2 -terminal side. Further extension at the NH 2 -terminal of DEpYLI by an Ala (to give ADEpYLI) improves its efficacy by 1.8-fold. Attachment of a Pro residue to the COOH-terminal of peptide ADEpYLI does not produce significant increase in substrate specificity, whereas attachment of an acetyl group to the NH 2 -terminal of peptide ADEpYLI actually reduces its efficiency slightly (by 1.3-fold). This suggests that the hexapeptide ADEpYLI may represent a minimal peptide substrate for bPTP␣ and bPTP␣-D1. In fact, the k cat /K m value for ADEpYLI is within 1.5-fold of the parent peptide DADEpYLIPQQG. We are surprised by the observation that in the absence of additional residues beyond the ϩ2 position in ADEpYLI, elongation at the NH 2 -terminal side (e.g. Ac-DADEpYLI) interferes with bPTP␣ activity. However Ac-DADEpYLI is still about twice as efficient as Ac-DADEpYL, consistent with the importance of residues at the ϩ2 position noted above.
We also investigated whether residues on the NH 2 -or COOH-terminal side of Tyr(P) are more important in determining bPTP␣ substrate specificity. When all of the residues NH 2terminal to the Tyr(P) of EGFR 988 -998 are removed, the resulting peptide pYLIPQQG is such a poor substrate that its k cat /K m value cannot be determined by the fluorescence method. Due to the limited amount of material, we could not measure its k cat and K m by the initial rate method. Since the lowest k cat /K m value determined by the fluorescence assay is 690 M Ϫ1 s Ϫ1 (for the bPTP␣-D1-catalyzed hydrolysis of EpYL), we assume it as an upper limit for pYLIPQQG. Thus, pYLIPQQG behaves kinetically similarly to phosphotyrosine. This could either indicate that positive charge on the amino group of Tyr(P) is deleterious or that residues COOH-terminal to Tyr(P) are not important for binding. Judging from the results described above, the former possibility is most likely. Further experiments are required to resolve this. Interestingly, when all of the residues COOH-terminal to the Tyr(P) are removed from EGFR 988 -998 and the negative charge on the free carboxylate eliminated by amidation, the k cat /K m value of the pentapeptide DADEpY-NH 2 approaches that of the parent peptide. Surprisingly, the hexapeptide DADEpYL-NH 2 , which is obtained by incorporating just one more Leu residue to the COOH-terminal side of Tyr(P) in DADEpYL-NH 2 , shows a 4-fold decrease in substrate specificity in comparison with the pentapeptide.
Thus, it appears that in the absence of charge and additional residues on the COOH-terminal side of Tyr(P), four residues NH 2 -terminal to Tyr(P) can serve as an effective substrate. Collectively, from a systematic analysis of different fragments of EGFR 988 -998 peptide, we conclude that efficient binding and  a The k cat /K m value for phosphotyrosine cannot be determined by the fluorescence assay. The k cat and K m values were determined by following the production of tyrosine (using an absorption coefficient of 2400 at 293nm) at several phosphotyrosine concentrations (29). b ND, the k cat /K m value cannot be determined by the fluorescence assay.
catalysis by bPTP␣ and bPTP␣-D1 require minimally either six amino acids (including Tyr(P), three residues NH 2 -and two residues COOH-terminal to Tyr(P)) or five amino acids (including Tyr(P)-NH 2 and four residues NH 2 -terminal to Tyr(P)). DISCUSSION PTP␣ is a representative of the receptor-like PTPases that contains two PTPase domains. In this paper we try to address whether both of the PTPase domains in PTP␣ are active, and if both are active, what are the relative contributions for each domain to the double domain molecule. We have compared the kinetic properties of bPTP␣, bPTP␣-D1, and bPTP␣-D2 using near homogeneous preparations of recombinant proteins. We demonstrate that bPTP␣-D1 and bPTP␣ (containing two tandem PTPase domains) display similar steady-state kinetic parameters, pH and leaving group dependences, sensitivity to oxyanion and p-nitrophenol inhibitors, and substrate specificity. This suggests that the catalytic activity of PTP␣ is dominated by the D1 domain and that the D2 domain does not contribute significantly to the overall property of the double domain construct. This is likely due to the fact that the k cat for D2 is much reduced and the K m for D2 is much elevated when compared with those of D1, especially in cases of phosphopeptide substrates.
The three-dimensional structure of RPTP␣-D1 (containing residues 202-503) has been determined (53). This particular D1 construct crystallized as a homodimer. It was found that a segment of the NH 2 -terminal sequence (residues 208 -242) of one monomer formed a helix-turn-helix structural wedge which tucked into the active site of the dyad-related monomer. This restrained the WPD loop of each monomer in the open (inactive) conformation and would block potential substrate binding. The D1 construct used in this study contains amino acids from 167 to 503 (Fig. 1). We find that under our assay conditions both bPTP␣-D1 and bPTP␣ are active. Furthermore, at bPTP␣-D1 concentration between 10 and 100 nM, the apparent first-order rate constant for the hydrolysis of Tyr(P)-containing peptides is strictly first-order with respect to the enzyme concentration with a zero y intercept when the rate constant is plotted against enzyme concentration (data not shown), consistent with bPTP␣-D1 exists as an active monomer in solution.
We also demonstrate that the D2 domain of PTP␣ is a genuine PTPase in its own right. This is especially true when low molecular weight aryl phosphates are used as substrates. In fact, the k cat value for D2 is only 10-fold lower than that of D1 while the K m is 5-fold elevated with pNPP as a substrate. However, the D2-catalyzed reaction shows different pH and leaving group dependences from that of D1. It also possesses a much lower affinity to the transition state analog vanadate and greatly reduced k cat and k cat /K m values for peptide substrates. In a previous study, the V max and K m values for the D2 domain of human PTP␣-catalyzed hydrolysis of a Tyr(P)-containing peptide (RR-Src) were shown to be 200-fold lower and 47-fold higher, respectively, when compared with the D1 domain (23). These differences in k cat and K m between D1 and D2 fall within the range that we observed with peptide substrates. However, from the same study, the V max value for the D2-catalyzed hydrolysis of pNPP was shown to be 1.6-fold higher than that of D1. We do not know what causes the discrepancy regarding the pNPP data, except to note that in the previous study, D1 domain included residues from 167 to 542 whereas D2 domain included residues from 487 to 774 (23), which are different from the constructs we made in this work (Fig. 1).
We hypothesized that the large differences in catalytic properties between D1 and D2 might be caused by structural variations between the active sites of the two PTPase domains. Amino acid sequence alignment of both D1 and D2 from a number of receptor-like PTPases reveals that the D2 domain of PTP␣ contains all the essential features found in the PTPase signature motifs of cytoplasmic PTPases and the D1 domains of receptor PTPases. In addition to the fact that D1 domains are more conserved and homologous to the catalytic domains of cytoplasmic PTPases than D2 domains, the most apparent difference is that the putative general acid (Asp residue) in the WPD motif observed in all D1 domains and all cytoplasmic PTPases has been changed to Glu-690 in bPTP␣-D2. We illustrate that mutation of Asp-401 in the WPD motif of bPTP␣-D1 to an Ala decreases the k cat by 900-fold, confirming the importance of general acid/base in the D1 domain of receptor PT-Pases. Replacement of Asp with Glu or vice versa can have dramatic effects in enzymes that require a carboxylate group as a general acid/base (43)(44)(45)(46). Interestingly, the restoration of the general acid/base to an Asp at position 690 in bPTP␣-D2 only raises the activity by 4-fold. This is consistent with our previous mutational studies of the general acid Asp-128 in the low molecular weight PTPase: D128E retained 15% of the native catalytic activity (42). Furthermore, substitution of Glu-690 by an Asp in bPTP␣-D2 does not alter its pH and leaving group dependences, its sensitivity to oxyanion inhibitors, and its substrate specificity. Collectively, these results indicate that the active site of PTPase is fairly flexible and can tolerate a 1 methylene unit lengthened side chain as long as the carboxylate functionality is preserved.
It is interestingly to note that the phosphatase activity of D2 toward aryl phosphates are comparable to that of D1, but its activity toward Tyr(P)-containing peptides/proteins is much reduced (this work and Refs. 23 and 24). Since the k cat for bPTP␣-D2 catalyzed hydrolysis of 4-methylphenyl phosphate (pK a of 4-methylphenol ϭ 10.26) is only 2-3-fold slower than that of pNPP (pK a of p-nitrophenol ϭ 7.14), the very low activity toward peptide substrates is unlikely due to the higher pK a of tyrosine (10.07). This together with the fact that Glu-690 to Asp substitution did not restore D2's activity to D1 suggest structural features other than the general acid/base residue may be responsible for the low activity toward peptide substrates. A three-dimensional structure for the D2 domain may shed more light on this. On the other hand, the reduced activity toward peptide substrates may also suggest that bPTP␣-D2 has a functional active site with a highly stringent substrate selectivity. Thus, the configuration of active site residues in D2 responsible for binding and/or catalysis may not be optimal for nonspecific peptide substrates that require extensive intermolecular interactions. The possibility that there exists physiological substrate(s) for the D2 domain of PTP␣ remains to be explored.
The k cat values for the bPTP␣ and bPTP␣-D1 catalyzed hydrolysis of aryl phosphates and Tyr(P)-containing peptides are similar. This is consistent with the rate-determining step being the hydrolysis of the phosphoenzyme intermediate (E-P in Scheme 1) (6). The rate-limiting step for the D2-catalyzed hydrolysis of low molecular weight aryl phosphates may be predominately determined by E-P hydrolysis as well, given the small ␤ lg value observed. The fact that k cat values for phosphopeptides are 40 -160-fold slower than small aryl phosphates such as pNPP in the D2-catalyzed reactions suggest that the rate-limiting step for peptide substrates may be changed to the formation of E-P. This is consistent with the notion that the D2 domain may recognize a global conformation and require a precise alignment between residues in the enzyme active site and the specific substrate. Nonspecific peptide substrates may not be capable of forming productive enzyme-substrate complexes for the forward reaction.
Detailed kinetic analysis of bPTP␣ and bPTP␣-D1 catalyzed Tyr(P)-containing peptides show moderate selectivity toward different substrates. Our data also indicate that positive charges from the free amino group at Tyr(P), Ϫ1, and Ϫ2 positions and the negative charge of the free carboxylate at Tyr(P) can be detrimental to PTP␣ catalysis. We have shown previously that in the sequence context of DADEpYLIPQQG (EGFR 988 -998 ), the Yersinia PTPase, and the mammalian PTP1 require a minimum of six amino acid residues for most efficient binding and catalysis (54). These include Tyr(P), four amino acid residues NH 2 -terminal, and one amino acid residue COOH-terminal to the Tyr(P) (DADEpYL-NH 2 ). Using the same series of peptides, we conclude that for optimal PTP␣ activity the minimal size of peptide substrates can either be ADEpYLI or DADEpY-NH 2 . Thus, different PTPases may possess different active site specificity in terms of short peptide substrates. It appears that short optimal peptides can be accommodated in the active site of the D1 domain by more than one mode. Furthermore, the recognition between PTP␣ and short peptide substrates seems sensitive to structural modifications. Further understanding of the specific functional roles of PTPases in cellular signaling requires detailed investigation of PTPase substrate specificity. Clearly, a more systematic and thorough approach, such as the use of degenerate peptide libraries containing randomized amino acid residues surrounding the phosphotyrosine, is needed for the search of consensus sequence motif for individual PTPases. Although consensus peptide motifs of many protein kinases have been elucidated (55), those of PTPases are essentially unknown. Specific consensus peptide motif for PTPases will lay the foundation for structure based rational PTPase inhibitor design. Since the identity of the physiological substrates for most PTPases is not known at present, such optimal substrate motifs may also be useful for the identification of physiological substrates for PTPases from available protein sequence data bases.
Finally, based on the kinetic data and the comparison of the amino acid sequence alignment of D1 and D2 from a number of receptor-like PTPases presented in this paper, we suggest that D2 domains of certain types of receptor-like PTPases (e.g. PTP␣, PTP⑀, and LAR) should display significant intrinsic phosphatase activity whereas D2 domains of other PTPases may possess little (e.g. CD45) or no (e.g. PTP␥) phosphatase activity. The functional role of the second domain, i.e. whether it regulates the specificity and activity of the first domain and/or exhibits independent substrate selectivity, requires further investigations.