Altering the Nucleophile Specificity of a Protein-tyrosine Phosphatase-catalyzed Reaction

Protein-tyrosine phosphatases (PTPases) catalysis involves a cysteinyl phosphate intermediate, in which the phosphoryl group cannot be transferred to nucleophiles other than water. The dual specificity phosphatases and the low molecular weight phosphatases utilize the same chemical mechanism for catalysis and contain the same (H/V)C(X)5R(S/T) signature motif present in PTPases. Interestingly, the latter two groups of phosphatases do catalyze phosphoryl transfers to alcohols in addition to water. Unique to the PTPase family are two invariant Gln residues which are located at the active site. Mutations at Gln-446 (and to a much smaller extent Gln-450) to Ala, Asn, or Met (but not Glu) residues disrupt a bifurcated hydrogen bond between the side chain of Gln-446 and the nucleophilic water and confer phosphotransferase activity to theYersinia PTPase. Thus, the conserved Gln-446 residue is responsible for maintaining PTPases’ strict hydrolytic activity and for preventing the PTPases from acting as kinases to phosphorylate undesirable substrates. This explains why phosphoryl transfer from the phosphoenzyme intermediate in PTPases can only occur to water and not to other nucleophilic acceptors. Detailed kinetic analyses also suggest roles for Gln-446 and Gln-450 in PTPase catalysis. Although Gln-446 is not essential for the phosphoenzyme formation step, it plays an important role during the hydrolysis of the intermediate by sequestering and positioning the nucleophilic water in the active site for an in-line attack on the phosphorus atom of the cysteinyl phosphate intermediate. Gln-450 interacts through a bound water molecule with the phosphoryl moiety and may play a role for the precise alignment of active site residues, which are important for substrate binding and transition state stabilization for both of the chemical steps.

Protein-tyrosine phosphatases (PTPase) 1 consist of a family of enzymes that catalyze the removal of phosphoryl groups on tyrosine residues in proteins that are introduced by proteintyrosine kinases. So far, approximately 100 PTPases have been identified, and the predicted total number of human PTPases may reach 500 based on the available genome sequencing data (1,2). Amino acid sequence alignment of PTPases from bacteria, yeast, and mammalian organisms indicates that the only structural element that has amino acid sequence identity among all PTPases corresponds to the catalytic domain that spans over 250 residues (3). The hallmark feature that defines the PTPases is the active site sequence (H/V)C(X) 5 R(S/T) in the catalytic domain also known as the PTPase signature motif (4). Work from a number of laboratories has led to the conclusion that PTPase catalysis proceeds through a double displacement mechanism involving a covalent phosphoenzyme intermediate. The side chain of the active site Cys residue serves as a nucleophile to accept the phosphoryl group from the phosphotyrosine in a substrate and form a kinetically competent cysteinyl phosphate intermediate (5,6). The active site Arg residue interacts with the phosphoryl moiety of the substrate and plays a role in both substrate binding and transition state stabilization (4,7). In addition, the conserved hydroxyl group of the Ser/Thr in the PTPase signature motif is essential for efficient hydrolysis of the phosphoenzyme intermediate (8,9). To facilitate substrate turnover, PTPases also employ an invariant Asp residue acting as a general acid/base catalyst (3,9).
Interestingly, the catalytic strategy employed by PTPases is also shared by the dual specificity phosphatases and the low molecular weight phosphatases, which share little sequence identity with the PTPases (10,11). The only similarities among these three groups of phosphatases are the relative placements of the essential cysteine, arginine, and serine/threonine residues in the active sites that constitute the PTPase signature motif (H/V)C(X) 5 R(S/T). Despite the lack of sequence identity, the spatial arrangements of the catalytic Cys, Arg, Ser/Thr, and Asp residues are conserved (12,13), consistent with the common catalytic mechanism. However, it is also apparent that even though these phosphatases utilize the same mechanism to catalyze phosphate monoester hydrolysis, they differ in active site specificity and exhibit measurably different structure of the transition state (14). To understand the structural and biochemical origins for these functional differences, we have begun to investigate the role of amino acids that are conserved only in a specific phosphatase subfamily.
The pathogenic bacteria Yersinia encodes a PTPase essential for its virulence (15). Because of its extraordinary phosphatase activity, the Yersinia PTPase has been the archetype of the PTPase family for mechanistic investigations. Site-directed mutagenesis of conserved residues in the Yersinia PTPases combined with detailed kinetic analyses have provided important insight into residues that are essential for PTPase catalysis (3,4,8). In this paper, we examined the functional role of two invariant glutamines (Gln-446 and Gln-450 in the Yersinia PTPase) located at the C terminus of the PTPase catalytic domains (Fig. 1). As shown in the crystal structures, the two Gln residues are situated at the active site (16,17). There are no corresponding Gln residues in the dual specificity phosphatases and the low molecular weight phosphatases. The crystal structures of the Yersinia PTPase complexed with oxyanion such as tungstate or sulfate have shown that the apical anion oxygen (O1) is coordinated by the side chain of Gln-446 ( Fig. 2 and Refs. 18 and 19). Both Gln-446 and Gln-450 form hydrogen bonds with a conserved structural water, WAT1 (WAT505 in PTP1B, Ref. 20), which also interacts with the apical oxygen and one of the phosphoryl oxygens (O2) in the oxyanion (Fig. 2). The apical oxygen in the oxyanion occupies the position that is equivalent to the phenolic oxygen of the leaving group in the phosphoenzyme formation step or the oxygen atom of the attacking nucleophilic water molecule in the phosphoenzyme hydrolysis step. Furthermore, the structure of the catalytic inactive C215S mutant PTP1B with phosphotyrosine bound indicates that the side chain of Gln-262 (equivalent to Gln-446 of the Yersinia PTPase) forms part of the phenyl ring binding pocket, and the N⑀2 of Gln-266 interacts with WAT505, which makes a hydrogen bond with the phenolic oxygen of phosphotyrosine (20). In this paper, we demonstrate the functional importance of the two invariant Gln residues in PTPase-catalyzed reaction.
Expression and Purification of the Recombinant Phosphatases-The wild type Yersinia PTPase and the mutants Q446A, Q446E, Q446M, Q446N, Q450A, Q450E, Q450M were expressed under the control of T7 promoter in Escherichia coli BL21 (DE3) grown at 23-25°C after induction with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside, and recombinant proteins were purified to near homogeneity as described (23). The recombinant dual specificity phosphatase VHR was expressed in E. coli and purified to homogeneity as described previously (24). The yeast low molecular weight phosphatase Stp1 was expressed in E. coli and purified to homogeneity as described previously (25).
Urea Denaturation-The urea-induced Yersinia PTPase denaturation was studied by monitoring the decrease in the intrinsic tryptophan fluorescence at 340 nm (slit width, 5 nm) with an excitation wavelength of 295 nm (slit width, 3.5 nm). The experimental details were described in Zhang et al. (23).
Steady-state Kinetics-Initial rates for the hydrolysis of pNPP and other aryl phosphate monoesters by the Yersinia PTPases was measured as described previously (8). Buffers used were as follow: pH 3.8 -5.7, 100 mM acetate; pH 5.8 -6.5, 50 mM succinate; pH 6.6 -7.3, 50 mM 3,3-dimethylglutarate; and pH 7.5-9.0, 100 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 enzyme active site concentration was treated as the protein concentration determined from the absorption at 280 nm using the coefficient A 280 1 mg/ml of 0.352 (8). The Michaelis-Menten kinetic parameters were determined from a direct fit of the v versus [S] data to the Michaelis-Menten equation using the nonlinear regression program KinetAsyst (IntelliKinetics, State College, PA). Inhibition constants for the Yersinia PTPases by arsenate were determined as described previously (4).
Pre-steady-state Kinetics-Pre-steady-state kinetic measurements of the Gln-446 and Gln-450 mutant Yersinia PTPases catalyzed hydrolysis of pNPP were conducted at pH 6.0 and 4.5°C. The reaction was monitored by the increase in absorbance at 410 nm of the p-nitrophenolate product. The enzyme concentrations were 42 M for Q446A and 49 M for Q450A, and the pNPP concentration was 20 mM. The details for data collection and analysis were as described (8). Detection of Reaction Products by 31 P NMR-Dephosphorylation of pNPP in the presence of phophoryl acceptor (1 M ethylene glycol) was monitored on a Varian VXR 500 MHz spectrometer operating at 202.3 MHz using the following parameters: acquisition time, 3.0 s; pulse width, 37 ms; delay time, 1.0 s; spectral width, 8,000 Hz. All spectra were recorded with proton broad band decoupler on. The spectrometer was locked on D 2 O resonance line. The buffer used contained 50 mM succinate (pH 6.0), 1 mM EDTA, 1 M ethylene glycol, and 20% D 2 O. The chemical shift of inorganic phosphate in the same buffer was set to zero. The concentration of pNPP was 20 mM, and the reaction was initiated with the addition of a catalytic amount of enzyme (1.5 -2.0 ϫ 10 Ϫ7 M) in 3 ml of the buffer mentioned above. NMR data were collected at two time points of the reaction (at which the sample was incubated for 5 min in boiling water to inactivate the enzyme): one at 1 min to 1.5 h (wild type enzyme at 1 min, Q446A at 30 min, Q450A at 1.5 h, etc.), the other at 12 h. Multiple scans were taken to reduce the noise.
Protein Phosphatase Assays-The phosphorylated substrate (myelin basic protein (MBP), GST-Elk, Kemptide, or GST-Rb), diluted in phosphatase buffer (20 mM HEPES, pH 7.0, 1 mM EDTA, 1 mM dithiothreitol), was added to an equal volume of phosphatase (wild type or mutant Yersinia PTPase), also in phosphatase buffer, to a final volume of 20 L. This phosphatase reaction mixture was incubated at 37°C for one hour. For MBP, GST-Elk, and GST-Rb, 20 L of SDS sample buffer was added to each reaction mixture to quench the phosphatase reaction. Half of this final mixture was resolved on 15% SDS gel and subjected to autoradiography. For Kemptide, all of the reaction mixture at the end of 1 h of incubation was blotted onto P81 filter paper (Whatmann). The filters were then washed four times with 80 mM phosphoric acid for 10 min each, washed with 95% ethanol, dried, and counted in the scintillation counter.
The amount of wild type or mutant Yersinia PTPase used was normalized to an equal amount of phosphatase activity in each of the reactions. The following amounts of each phosphatase was used for all experiments except for the second round of Rb dephosphorylation, where different amounts of the same phosphatases were incubated for two different time points to determine concentration and time dependence of possible dephosphorylation observed in the first phospho-Rb dephosphorylation experiment. The amount of phosphatase used (in g) was 0.84, wild type; 1.08, Q446A; 0.64, Q446E; 3.35, Q446M; 3.08, Q446N; 5.72, Q450A; 57.0, Q450E; and 5.5, Q450M.
Preparation of Phosphorylated Protein Substrates-1) MBP and GST-Elk (a transcription factor phosphorylated by extracellular signalregulated kinase-1). GST-ERK1 (1.5 g), GST-super mitogen-activated protein or extracellular signal-regulated kinase (0.5 g), 10 Ci of [␥-32 P]ATP and either 10 g of MBP or 10 g GST-Elk were mixed together in the kinase buffer (18 mM HEPES, pH 7.4, 10 mM magnesium acetate, 50 M ATP) to a final volume of 20 L, and was incubated at 30°C for 1 h. The kinase reaction was quenched by the addition of 80 L of phosphatase buffer. Ten L of this mixture was used for phosphatase assay. 2) Kemptide (synthetic peptide as substrate of protein kinase A, Leu-Arg-Arg-Ala-Ser-Leu-Gly). Protein kinase A (catalytic subunit) was mixed with 10 g of the synthetic peptide, Kemptide, and 20 Ci of [␥-32 P]ATP in 50 L of the protein kinase A kinase buffer (20 mM Tris, 10 mM dithiothreitol, 5 mM NaF, 10 mM MgOAc, 200 M ATP). This mixture was incubated at 30°C for 1 h to allow phosphorylation of the Kemptide. The kinase reaction was quenched by the addition of 200 L of the phosphatase buffer. Ten L of this mixture was then used for phosphatase assay. 3) GST-Rb (C-terminal domain of retinoblastoma tumor suppressor protein). GST-CDK4 and GST-cyclin D2 (1 g each) were activated by incubation in 40 L of proliferating Jurkat cell lysate (supplemented with 1 mM ATP) at room temperature for 1 h. The activated kinase was affinity purified on glutathione-agarose and eluted in 25 mM HEPES, pH 7.0, containing 10 mM glutathione. Half of the eluent (10 L) was mixed with 10 L of the cyclin-dependent kinase kinase buffer (50 mM HEPES, pH 7.0, 10 mM MgCl 2 , 5 mM MnCl 2 , 1 mM dithiothreitol, 20 M ATP) containing 10 g of GST-Rb and 10 Ci [␥-32 P]ATP. This reaction mixture was incubated at 37°C for 1 h. The kinase reaction was quenched by the addition of 80 L of the phosphatase buffer, and 10 L of this mixture was used for phosphatase assay.

Physicochemical Properties of the Gln-446 and Gln-450 Mutant Yersinia
PTPases-To evaluate the role of the conserved Gln-446 and Gln-450 in PTPase catalysis, these Gln residues were altered by site-directed mutagenesis. The carboxyamide side chain does not ionize at physiological pHs but is relatively polar, being capable of both donating and accepting hydrogen bonds. The Gln residue was changed to (a) Glu, which is isosteric with Gln but can ionize as the pH is raised from the acidic range, (b) Asn, which is the lower homolog (one methylene group shortened) with the carboxyamide group retained, (c) Met, which is similar in size to Gln but lacks the carboxyamide group, and (d) Ala, which completely eliminates the Gln side chain. All of the mutations were verified by DNA sequencing. Initially, the mutant Yersinia PTPases were expressed in E. coli as described for the wild type enzyme, i.e. 6 h growth at 37°C after 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside induction (23). Like the wild type, Q446E remained soluble in the cell lysate. However, more than half of the total recombinant Q446A, Q446N, and Q450E protein was found in the pellet, and the majority of the expressed protein for mutant Q446M, Q450M, and Q450A became insoluble at the same expression conditions. Subsequently, when the cells were grown at room temperature overnight after isopropyl-1-thio-␤-D-galactopyranoside induction, the amount of the soluble protein obtained increased markedly. All of the mutant PTPases were purified to near homogeneity as judged by SDS-polyacrylamide gel electrophoresis, using procedures identical to those described for the wild type enzyme (23).
The Gln-446 and Gln-450 mutants had chromatographic and ultraviolet absorption spectral characteristics similar to those of the wild type. The max values of the fluorescence emission spectra of the mutants were the same as the wild type. To assess the conformational stability of the mutants, urea-induced denaturation was studied (see "Experimental Procedures"). As can be seen in Table I, the free energies of unfolding for Q446E and Q450E are similar to the wild type, whereas they are significantly reduced for the Q446A and Q450A mutants. Collectively, the side chains of Gln-446 and Gln-450 may be involved in the appropriate folding process and contribute to the overall stability of the enzyme. Table II summarizes the steady-state kinetic parameters for the wild type and the mutant enzymes at 30°C and several pH values using p-nitrophenyl phosphate (pNPP) as a substrate. The competitive inhibition constants for the binding of arsenate to the Yersinia PTPases are also listed in Table II. There are several points that are worth mentioning. First, although Glu is isosteric with Gln, the Q450E mutant exhibited k cat and k cat /K m values that were 30 -60-fold and 100 -1000-fold lower than those of the wild type enzyme at every pH examined. Furthermore, the binding affinity of Q450E for arsenate decreased by 30-fold as the pH was raised from 5 to 7. Interestingly, substitution of Gln-446 by a Glu residue had minimal effect on the k cat values, although both the K m for pNPP and the K i for arsenate increased by severalfold as the pH was raised. Second, with the exception of Q450E, substitutions at Gln-446 and Gln-450 generally had rather modest effects on the kinetic parameters, although the effects were greater for the Gln-450 mutants than the Gln-446 mutants. For example, substitutions at Gln-446 generally led to less than 10-and 5-fold reduction in k cat and k cat /K m values, respectively. In contrast, substitutions at Gln-450 resulted in a decrease in k cat by 10-fold and k cat /K m by 10 -100-fold. Third, with the exception of Q446E and Q450E, the affinity of the native and mutant PTPases toward arsenate did not change appreciably with pH, and alterations at positions 446 and 450 did not lead to significant changes in the K i values.

Steady-state Kinetic Properties of the Yersinia PTPases-
Leaving Group Dependence-The leaving group dependence of the wild type Yersinia PTPase as well as the Q446A and Q450A mutants was investigated using a series of aryl phosphates. To minimize nonspecific steric effects, only para-substituted phenyl phosphates were used. Fig. 3 shows the Brønsted plots which relate the logarithms of k cat and k cat /K m , respectively, to the pK a values of the leaving groups for 4-nitrophenyl, 4-acetylphenyl, 4-cyanophenyl, 4-ethoxycarbonylphenyl, 4-fluorophenyl, 4-bromophenyl, 4-chlorophenyl, 4-methylphenyl, 4-ethyl phosphate, and 4-methoxylphenyl phosphate. Linear least-squares fitting of the Brønsted plots yielded the slopes, which correspond to the ␤ lg values. The ␤ lg values describe the leaving group dependence of the reaction and can often be regarded as an approximate measure of the extent of bond cleavage between the reaction center and the leaving group. For the wild type enzyme, the ␤ lg for k cat was 0.036 Ϯ 0.032, whereas the ␤ lg for k cat /K m was Ϫ0.19 Ϯ 0.051. Similarly, Q446A exhibited a ␤ lg for k cat of Ϫ0.011 Ϯ 0.011, and a ␤ lg for k cat /K m of Ϫ0.12 Ϯ 0.050, whereas Q450A exhibited a ␤ lg for k cat of Ϫ0.013 Ϯ 0.037, and a ␤ lg for k cat /K m of Ϫ0.13 Ϯ 0.053. Thus, elimination of the carboxyamide functional group from residues 446 or 450 had negligible effect on the ␤ values.
Pre-steady-state Analysis-The overall mechanism of hydrolysis of aryl phosphates catalyzed by the Yersinia PTPase involves a number of steps that are represented schematically in Scheme 1. The reaction proceeds through a sequence involving binding of substrate which is then cleaved with phosphoryl transfer (k 2 ) to the nucleophilic Cys residue. Subsequent basecatalyzed reaction with water cleaves the phosphoenzyme intermediate (E-P) (k 3 ) and release of phosphate completes the catalytic cycle. Burst kinetics was observed with the Yersinia PTPase at pH 6.0 and 3.5°C using pNPP as a substrate (8). It was concluded that under these conditions the rate-limiting step corresponds to E-P hydrolysis. Burst kinetics was also observed at pH 6.0 and 4.5°C with all of the Gln mutants characterized. The stopped flow traces for the Q446A and Q450A mutant reactions are shown in Fig. 4. The observed rate constant for E-P formation (k 2 ) in the Q446A and Q450A reactions was slightly higher (1.5-2-fold), whereas the rate constant for E-P hydrolysis (k 3 ) was lowered by 4 -5-fold in comparison with those of the wild type reaction reported in a previous study (8). It is not clear whether the slightly higher burst rate is significant since rate constants in the range of several hundreds per second are barely within the detection limit of a typical stopped flow spectrophotometer. Collectively, these results suggest that alteration of either Gln-446 or Gln-450 does not change the rate-limiting step of the dephosphorylation of pNPP by the enzyme.
Substitutions  reaction. For example, when the Yersinia PTPase was incubated with pNPP in the presence of 1 M ethylene glycol at pH 6.0, no phosphoryl group transfer to the alcohol was observed and only the hydrolysis product, inorganic phosphate, was detected using 31 P NMR (Fig. 5A). In contrast, when the same experiment was performed with the Q446A, Q446N, or Q446M mutant, in addition to hydrolysis, ethylene glycol phosphate was generated during the process of dephosphorylation of pNPP (data for Q446A is shown in Fig. 5B). Phosphoryl transfer to alcohol also occurred with the Q450A and Q450M mutants, although only a minute amount of ethylene glycol phosphate could be detected even after prolonged incubation with the mutant phosphatases (data for Q450A is shown in Fig. 5C). Interestingly, like the wild type enzyme, neither Q446E nor Q450E was able to produce the alkyl phosphate product (data not shown). Dual Specificity Phosphatases and Low Molecular Weight Phosphatases Possess Phosphoryl Transfer Activity-Dual specificity phosphatases and low molecular weight phosphata-  ses also proceed through a phosphocysteine intermediate and utilize a strategy similar to that of the PTPases for phosphate monoester hydrolysis. Interestingly, 31 P NMR analysis of the dual specificity phosphatase VHR and the low molecular weight phosphatase Stp1 catalyzed pNPP hydrolysis in the presence of alcohol clearly demonstrate transfer of the phosphoryl group onto alcohols (data not shown). The structural basis for this additional phosphotransferase activity may be suggested by the fact that the bound oxyanion in VHR (26) or the bovine low molecular weight phosphatase (27,28) is not shielded from bulk solvent as in the Yersinia PTPase (18) due to the absence of the corresponding Gln residues. It is likely that alcohols and water have equal access to the cysteinyl phosphate intermediate in VHR and the low molecular weight phosphatases.
Partition Experiments-The effect of ethylene glycol on the kinetics and product distribution of the Yersinia PTPase-catalyzed hydrolysis of pNPP was examined to ascertain the ratelimiting step and to provide additional evidence for phosphoryl transfer. In these experiments (Scheme 2), the overall rate of substrate turnover (hydrolysis plus phosphate transfer) was measured by the production of p-nitrophenol whereas the hydrolysis rate was measured by the production of inorganic phosphate (21). As shown in Table III, the kinetic parameters of pNPP hydrolysis determined by the p-nitrophenol assay were similar to those determined by the phosphate assay in the absence of ethylene glycol. The k cat values for the hydrolysis of pNPP as determined by the phosphate assay in the absence of ethylene glycol were also close to those determined in the presence of 1 M ethylene glycol, because the molarity of water (ϳ55.5 M) was not changed significantly by the introduction of 1 M ethylene glycol.
The overall rate of the substrate turnover (k cat ) as measured by the production of p-nitrophenol is given by Equation 1. When the rate-limiting step is the hydrolysis of E-P, Equation 1 can be reduced to Equation 2. Thus, the presence of ethylene glycol will increase the rate of E-P breakdown and thus accelerate the overall reaction rate (Scheme 2 and Equation 2), provided ethylene glycol can serve as a phosphoryl acceptor.
It turned out that the presence of ethylene glycol did not have any effect on the rate of catalysis by the wild type and the mutants Q446E and Q450E, since they were unable to catalyze the transfer of a phosphoryl group to an alcohol. However, as shown in Table III, mutant PTPases that could catalyze phosphoryl transfer to alcohols as judged by the 31 P NMR data did exhibit increased overall reaction rate in the presence of 1 M ethylene glycol. This was especially so for the Q446A mutant for which the k cat for overall reaction was accelerated 1.5-fold in the presence of 1 M ethylene glycol, consistent with the relative amount of products generated as measured by 31 P NMR (Fig.  5B). These results support that E-P decomposition is the ratelimiting step for the hydrolysis of pNPP catalyzed by Q446A, Q446N and Q446M.
To assess the significance of Gln-446 in controlling the nucleophile specificity (H 2 O versus alcohol), we have compared the rates of phosphotransfer to ethylene glycol for Q446A, the dual specificity phosphatase VHR, and the low molecular weight phosphatase Stp1. The second-order rate constant (k 4 ) for phosphotransfer can be determined from Equation 2 by measuring the dependence of k cat on the alcohol concentration (29). As shown in Fig. 6, k 4 for Q446A is 21.6 M Ϫ1 s Ϫ1 which is 3 times faster than that for Stp1 (6.6 M Ϫ1 s Ϫ1 ) and 45 times faster than that for VHR (0.48 M Ϫ1 s Ϫ1 ). Since the Yersinia PTPase, VHR, and Stp1 exhibit different intrinsic catalytic activities, we also compare the ratio of the rate of phosphotransfer (k 4 ) to the rate of hydrolysis (k 3 Ј ϭ k 3 /[H 2 O], Scheme 2 and Ref. 29). The ratio (k 4 /k 3 Ј) is 0 for the Yersinia PTPase, 19.4 for the Q446A mutant, 122 for Stp1 and 4.4 for VHR, indicating that the tendency to catalyze phosphoryl transfer to ethylene glycol is zero for the Yersinia PTPase, moderate for VHR, and great for Stp1. Furthermore, replacement of Gln-446 with an Ala has rendered the Yersinia PTPase with considerable phosphotransferase potential (k 4 /k 3 Ј) that is 4.4-fold higher than VHR and 6.3-fold lower than Stp1.

Do Mutations at the Two Gln Residues Alter the Yersinia
PTPase Substrate Specificity?-The Yersinia PTPase displays strict specificity toward Tyr(P)-containing peptides/proteins (30). This specificity for Tyr(P) has been proposed to be due to the depth of the amphipathic Tyr(P) binding pocket that exactly matches the length of Tyr(P) (20). Since the side chain of Gln-262 (equivalent to Gln-446 of the Yersinia PTPase) is positioned at the rim of the pocket and interacts with the aromatic ring of Tyr(P) (Ref. 20 and Fig. 7A), elimination of the side chain may open the binding pocket such that phosphorylated Ser/Thr-containing peptides/proteins may also be able to serve as substrates. A number of Ser-phosphorylated proteins (MBP, GST-Elk, Kemptide, or GST-Rb) were prepared and tested as potential substrates for the wild type as well as the Gln-446 and Gln-450 mutants (see "Experimental Procedures"). No significant dephosphorylation was observed for any of the substrates tested (data not shown).

DISCUSSION
The function of the two conserved glutamines in the active site of PTPases has been probed by site-directed mutagenesis. These two Gln residues are unique to the PTPase family and no corresponding residues are present in the dual specificity phosphatases and the low molecular weight phosphatases. In the structure of PTP1B/C215S bound with substrate, the side chain of Gln-262 (equivalent to Gln-446 in Yersinia PTPase) interacts with the phenyl ring of the phosphotyrosine (Tyr(P)), and defines a portion of the rim for the Tyr(P) binding pocket (20). A similar interaction between Gln-446 and Tyr(P) is observed in an energy minimized model of the Yersinia PTPase with Tyr(P) bound (19). In the structures of PTPases bound with substrates or oxyanions, a trapped structural water molecule, WAT1 in the Yersinia PTPase and WAT505 in PTP1B, is observed (19,20). WAT1 forms hydrogen bonds with the side chains of the invariant residues Gln-446 and Gln-450, O-1 and O-2 of the oxyanion, and the carboxylate of the general acid/base Asp-356 in the PTPase-oxyanion complexes (Fig. 2). In the PTPasesubstrate complexes, this water molecule makes hydrogen bonds with the side chain of Gln-450, the tyrosine leaving group oxygen (the scissile oxygen), O-2 of the phosphate, and the carboxylate of the general acid/base Asp-356 (Fig. 7A, and Ref. 20). Interestingly, a new water molecule, WAT3, is identified in the Yersinia PTPase-nitrate complex as well as in an energy minimized model of the phosphocysteine intermediate (19). Nitrate is coplanar and lacks the apical oxygen present in tetrahedral oxyanion such as tungstate. Instead, WAT3 sits directly above the nitrate and is perfectly in line with the sulfur atom of the active site Cys residue. WAT3 is proposed to be the nucleophilic attacking molecule for hydrolysis of the cysteinyl phosphate intermediate and is coordinated primarily by the side chains of Gln-446, Gln-357, Asp-356, and WAT1 through hydrogen bonds (Fig. 7B, Ref. 19).
Based on the reported structural data, Gln-446 and Gln-450 likely participate in both of the chemical steps, i.e., phosphoenzyme formation and decay. Gln-446 may function to help to align the phenyl ring of Tyr(P) in the first step and to position the attacking water in the second step. Gln-450 makes a hydrogen bond with WAT1, which in turn is in hydrogen bonding distance with the apical oxygen (O-1) and one of the phosphoryl oxygen (O-2). Thus, Gln-450 may play a role in proper stabilization of the metaphosphate-like transition state (31) as well as in leaving group stabilization and nucleophile activation. For the Yersinia PTPase, the kinetic parameter k cat /K m is primarily limited by the phosphoenzyme intermediate formation step (31), whereas the k cat term is mostly determined by intermediate breakdown. Results from kinetic analysis of the site-directed mutants of Gln-446 and Gln-450 confirm some of the structural data and provide additional insights into the mechanisms by which these two residues effect catalysis.
Replacements of Gln-446 by Ala, Asn, or Met have minimal effect on k cat /K m and only less than 10-fold reduction in k cat . In the case of PTP1B/Q262A, k cat /K m and k cat are reduced by 7and 83-fold, respectively, using the tyrosine-phosphorylated lysozyme as a substrate (7). It is not clear why a larger reduction in k cat is observed for the PTP1B/Q262A mutant. It appears that Gln-446 does not play a significant role in the phosphoenzyme formation step. As shown in Fig. 7B, the putative nucleophilic water molecule is coordinated by a bidentate hydrogen bond with the side chain of Gln-446. Such an interaction does not exist in the Q446A, Q446N, and Q446M mutants so that the positioning of the attacking water could be affected leading to a reduced rate for the phosphoenzyme hydrolysis step. Indeed, within the PTP1B/Q262A phosphoenzyme intermediate, the position of the nucleophilic water is displaced by 1.5 Å from being colinear with the active site Cys-215 S␥-P bond. 2 The moderate decrease in rate for phosphoenzyme hydrolysis observed in the Yersinia Gln-446 mutants is consistent with a dissociative transition state for which minimal activation of the nucleophilic water is required (29). In addition, a water molecule may take the place of the Gln-446 side chain and partially rescue its function in the second step. A new water is found at the position of the side chain of Gln-262 in the structure of PTP1B/Q262A phosphoenzyme complex. 2 This may also explain the fact that Gln-446 mutants possess similar oxyanion binding affinity as the wild type.
Replacements of Gln-450 by Ala or Met have larger effects on both the k cat /K m and k cat , consistent with the notion that it interacts through WAT1 with a phosphoryl oxygen, and the leaving group oxygen in the first step and a phosphoryl oxygen, and the attacking water in the second step. The results obtained with Q446E and Q450E are rather interesting. Although glutamic acid and glutamine are isosteric, the Gln to Glu substitution at residue 450 manifests the most deleterious effects with up to 1200-fold decrease in k cat /K m , and 66-fold decrease in k cat , 20-fold increase in K m for pNPP and 30-fold increase in K i for arsenate. Furthermore, these deleterious effects increase as the pH is raised. These results strongly suggest that in the native Yersinia PTPase, the carboxyamide side chain must donate a hydrogen bond to WAT1 through its amide group (Fig. 7). WAT1 may receive another hydrogen bond from Asp-356 or the backbone amide from Gln-357, and donate one hydrogen bond to a phosphoryl oxygen and another hydrogen bond to the leaving group oxygen or the attacking water (Fig. 7). Thus, deprotonation of the carboxylic acid at residue 450 in Q450E will cause detrimental effects due to charge repulsion. In contrast, it appears that Gln-446 can be largely replaced by a Glu residue with little change in k cat and only moderate increase in K m for pNPP and K i for arsenate.
Gln-446 Is the Residue Escorting the Nucleophilic Water- The ability to catalyze phosphoryl transfer to nucleophilic acceptors in addition to water is a common feature for phosphate ester hydrolases, such as nonspecific alkaline and acid phosphatases, that involve a covalent phosphoenzyme intermediate (32 (34). The increase in rate is likely caused by a glycerol induced conformational change which alleviates the autoinhibited state. Alcohols are better nucleophiles than water for accepting a phosphoryl group (29). The fact that alcohols such as methanol, ethanol, ethylene glycol and glycerol are much better nucleophilic acceptors of phosphoryl group than water and yet they fail to serve as phosphate acceptors in the PTPase reaction indicates that the phosphoryl group in the intermediate is only accessible by water. As shown in the PTPase-oxyanion complexes, the side chain of Gln-446 interacts with the apical oxygen of the oxyanion, which effectively shields the oxyanion from the aqueous environment (18). In the PTPase-nitrate complex or the energy-minimized model of the phosphoenzyme intermediate (19) the side chain of Gln-446 coordinates the attacking nucleophilic water via a bidentate hydrogen bond. Thus, nucleophiles larger than water such as alcohols cannot gain access into the active site due to steric hindrance. Alter-2 D. Barford, personal communication.  (17)(18)(19)(20), the energy-minimized atomic models of the noncovalent enzyme-substrate complex, the phosphoenzyme intermediate (19), and the kinetic data presented in this paper. natively, it is also possible that both protons of the nucleophilic water are required for proper positioning for the nucleophilic attack. For example, the nucleophilic water can donate one H-bond to O⑀1 of Gln-446 and the other to the side chain of Asp-356, which is consistent with the carboxylate functioning as a general base in the phosphoenzyme hydrolysis step (21). The nucleophilic water can also receive a hydrogen bond either from N⑀2 of Gln-446 or from WAT1 (Fig. 7). Such interactions would position the water in an optimal position for nucleophilic attack on the phosphorus atom (Fig. 7). Since alcohols are incapable of making the same interactions and they can only donate one H-bond, they cannot serve as phosphoryl acceptors.
Substitutions at Gln-446 by an Ala, an Asn, or a Met abolishes the specific bidentate hydrogen bonding interactions with the nucleophilic water and results in either smaller side chains (Q446A and Q446N) or a side chain with an increased mobility (Q446M), such that alcohols can replace the nucleophilic water and gain access to the phosphoenzyme intermediate. This notion is supported by the observations that Q446A, Q446N, and Q446M all displayed phosphoryl transfer activities. Q450A and Q450M also exhibited minor phosphotransfer activities. This may be caused by the slight misalignment in Gln-446 due to the disruption of the interactions between Gln-450 and WAT1 (Fig.  7). Collectively, these results strongly suggest that Gln-446 functions to escort and orient the attacking water molecule for the hydrolysis of the phosphoenzyme intermediate. The isosteric substitution of Gln-446 by a Glu residue introduces minimal alterations in the kinetic parameters and does not confer phosphotransfer activity to the Yersinia PTPase, suggesting that the Glu residue at 446 is protonated at pH 6.0 so that a similar bidentate hydrogen bond can be made with the nucleophilic water. Interestingly, although substitution of Gln-450 by a Glu residue leads to dramatic decrease in both catalytic efficiency and substrate/inhibitor affinity, the Q450E mutant still maintains the hydrolytic activity with no phosphotransferase activity. This implies that the Gln-450 to Glu substitution does not introduce drastic structural alterations into the enzyme active site, especially in the context of Gln-446.
Conclusions-Regulation of enzyme activity by reversible cycles of phosphorylation and dephosphorylation is now considered to be one of the most important means by which cells respond to physiological stimuli. To ensure the precision and fidelity of this regulation, it may be desirable that the PTPase only possesses hydrolytic activity and protects the highly reactive phosphoenzyme intermediate from unwanted nucleophilic acceptors (such as Ser or Thr residues in a protein) during its course of action. It is shown here that Gln-446 is responsible for keeping this unwanted "kinase" activity in check. The PTPases, the dual specificity phosphatases and the low molecular weight phosphatases employ a similar chemical mechanism for phosphate monoester hydrolysis. However, in contrast to the PTPases, which possess no phosphoryl transfer activity, both the low molecular weight phosphatases (21,29,35) and the dual specificity phosphatase VHR (this study) are able to catalyze the transfer of the phosphoryl group from the intermediate to alcohol nucleophiles as well as to water. The biological implication for this observed mechanistic difference is unclear.
Results from kinetic characterization of the Gln-446 and Gln-450 mutant PTPases combined with structural information on PTPases support the conclusion that the two invariant Gln residues constitute an integral part of the PTPase active site, serving different roles in the catalytic reaction. Gln-450 plays a role for the precise alignment of active site residues, which are important for substrate binding and transition state stabilization, whereas Gln-446 is important for the optimal positioning of the nucleophilic water molecule in the phosphoenzyme hydrolysis step.