The Catalytic Mechanism of Cdc25A Phosphatase*

Cdc25 phosphatases are dual specificity phosphatases that dephosphorylate and activate cyclin-dependent kinases (CDKs), thereby effecting the progression from one phase of the cell cycle to the next. Despite its central role in the cell cycle, relatively little is known about the catalytic mechanism of Cdc25. In order to provide insights into the catalytic mechanism of Cdc25, we have performed a detailed mechanistic analysis of the catalytic domain of human Cdc25A. Our kinetic isotope effect results, Bronsted analysis, and pH dependence studies employing a range of aryl phosphates clearly indicate a dissociative transition state for the Cdc25A reaction that does not involve a general acid for the hydrolysis of substrates with low leaving group pK a values (5.45–8.05). Interestingly, our Bronsted analysis and pH dependence studies reveal that Cdc25A employs a different mechanism for the hydrolysis of substrates with high leaving group pK a values (8.68–9.99) that appears to require the protonation of glutamic acid 431. Mutation of glutamic acid 431 into glutamine leads to a dramatic drop in the hydrolysis rate for the high leaving group pK a substrates and the disappearance of the basic limb of the pH rate profile for the substrate with a leaving group pK a of 8.05, indicating that glutamic acid 431 is essential for the efficient hydrolysis of substrates with high leaving group pK a . We suggest that hydrolysis of the high leaving group pK a substrates proceeds through an unfavored but more catalytically active form of Cdc25A, and we propose several models illustrating this. Since the activity of Cdc25A toward small molecule substrates is several orders of magnitude lower than toward the physiological substrate, cyclin-CDK, we suggest that the cyclin-CDK is able to preferentially induce this more catalytically active form of Cdc25A for efficient phosphothreonine and phosphotyrosine dephosphorylation.

Progression from one phase of the cell cycle to the next is largely mediated by the rapid phosphorylation of various cellular targets by cyclin-dependent kinases (CDKs) 1 at the phase transitions (12). This, in turn, is triggered by the rapid dephosphorylation of cyclin-CDK complexes on Thr 14 and Tyr 15 by Cdc25 proteins, which are dual specificity phosphatases (13)(14)(15)(16). Humans have three isoforms of Cdc25 termed Cdc25A, Cdc25B, and Cdc25C, and each plays a distinct role in the cell cycle (3).
Despite what is known about Cdc25's role in the cell cycle, relatively little is known about its catalytic mechanism. Cdc25 is classified as a protein-tyrosine phosphatase (PTPase) because it contains the CX 5 R PTPase signature motif, which characterizes the PTPase superfamily. Enzymes of the PTPase superfamily can be divided into three classes: tyrosine-specific, low molecular weight, and dual specific phosphatases. The dual specific phosphatases can be further split into the VH1-like phosphatases, mitogen-activated protein kinase phosphatases, kinase-associated phosphatase, and Cdc25 phosphatases based upon sequence comparisons. In previous years, much progress has been made in understanding the mechanism of PTPases. The CX 5 R motif plays a vital catalytic role in all PTPases. It has been shown that the cysteine residue in this motif functions as a nucleophile, attacking the substrate phosphate atom directly and displacing the leaving group to form a phosphocysteine intermediate (17)(18)(19)(20)(21)(22). The arginine in this motif has been shown to be particularly important for transition state stabilization (23)(24)(25) and is essential for catalysis (26 -28). For Cdc25B, the cysteine and arginine in the PTPase signature motif have been shown by site-directed mutagenesis to be required for catalysis (29). This suggests that the cysteine and arginine residues in the CX 5 R motif play the same role in Cdc25 as they do for other PTPases.
Apart from the cysteine and arginine residues in the CX 5 R motif, it is not known whether other residues participate in catalysis for Cdc25. This is in contrast to other PTPases for which a set of conserved, catalytically important residues has been defined, and functions have been determined for each of these residues (21). For example, a conserved acidic residue, the general acid, has been shown to play an important role in catalysis in all PTPases examined to date, including the Yersinia PTPase (30), PTP1B (31,32), the low molecular weight PTPases (33)(34)(35), VHR (36,37), and MKP3 (38). The general acid facilitates catalysis by protonating the leaving group oxygen, thereby neutralizing the negative charge that develops as the phosphate ester bond is broken. This stabilizes the transi-tion state and enhances the rate of catalysis by Ͼ1000-fold in the case of the Yersinia PTPase (30).
This information regarding the identities and functions of catalytically important residues in PTPases is very valuable. One group has used this information to guide structure-based inhibitor design for PTPases in which they were able to design selective PTPase inhibitors by targeting both conserved residues important for catalysis and nonconserved residues important for substrate specificity (39). In addition, knowing the identity of the general acid has been used to design more highly potent substrate-trapping mutants of PTPases (40,41) and may aid in the identification of physiological substrates of PTPases.
Although it is potentially valuable information, the identity of the general acid in Cdc25 is still unknown. As it turns out, performing a sequence alignment of Cdc25 with a template based upon the prototypical PTPases described above does not reveal the catalytically important residues of Cdc25 (42). Cdc25 is somewhat of an outlier among the PTPases. This was revealed by homology modeling (43) and x-ray crystallographic studies (44,45). These studies revealed that Cdc25 is structurally much more similar to the sulfur transfer enzyme rhodanese than it is to any other PTPases. Because it is so dissimilar to other PTPases, aside from the role of the cysteine and arginine in the CX 5 R motif, no obvious conclusions can be drawn about the mechanism of Cdc25 or the identity of Cdc25's catalytically important residues simply based upon previous studies of PTPases. For this reason, we chose to perform a detailed mechanistic analysis of Cdc25A. The main goals of this work are to provide a description of the transition state of the Cdc25A-catalyzed reaction and to determine the role of general acid catalysis and the identity of the general acid in Cdc25A. In the process, we have revealed some unique features of the Cdc25A-catalyzed reaction.

Expression and Purification of Cdc25A Catalytic Domain and Mutant Proteins-
The catalytic domain of Cdc25A was expressed and purified based upon the procedure described by Fauman et al. (44). A mammalian expression vector containing the full-length human Cdc25A was obtained from Dr. Katsumi Yamashita (Kanazawa University, Japan). The region encoding the catalytic domain of Cdc25A (residues 336 -523) was amplified by PCR and subcloned into the vector pCR2.1 using the TA cloning kit from Invitrogen, with a 5Ј primer that contained an NheI site followed by an initiating methionine codon and a 3Ј primer that contained an EcoRI site followed by a stop codon. The region encoding the catalytic domain was then subcloned into the vector pUC118. The Cdc25A catalytic domain mutants, E431Q, E435Q, F432A, and G437S, were obtained using the QuikChange site-directed mutagenesis kit from Stratagene and the appropriate mutagenic primers obtained from the Oligonucleotide Synthesis Facility of the Albert Einstein College of Medicine. The sequences of the wild type and mutant Cdc25As were confirmed by DNA sequencing. Using the NheI and EcoRI sites, the catalytic domain was subcloned into the bacterial expression vector pET21a from Novagen. Escherichia coli BL21DE3 cells were transformed with this vector, grown at 37°C in 1-liter cultures to an OD at 600 nm of 0.6 -0.8, and induced to express the catalytic domain of Cdc25A by the addition of 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside. After induction, the cultures were grown at room temperature for 16 h and then pelleted by centrifugation. Pellets from 1 liter of culture were resuspended in 30 ml of ice-cold 25 mM HEPES, pH 7.5, 2 mM DTT, 1 mM EDTA (buffer A). The cells were lysed in a French press at 12,000 p.s.i. The lysate was then centrifuged at 15,000 rpm for 40 min, and supernatant was filtered using a 0.45-m membrane syringe filter from Pall Gellman Laboratories. The supernatant was then bound at 4°C to 50 ml of Poros HS-50 resin (PerSeptive Biosystems) that was equilibrated at 4°C in buffer A. After 20 min, the resin was added to a column and washed with 250 ml of buffer A. The protein was eluted using a 500-ml, 0 -0.5 M linear, sodium chloride in buffer A gradient at 4°C. The fractions containing Cdc25A were concentrated in a Centriprep YM-10 (10,000 molecular weight cut-off) concentrator from Amicon. Gel filtration chromatography (G-50) was then performed in buffer A containing 200 mM sodium chloride, and the protein was concentrated as before. Proteins were judged to be at least 95% pure by SDS-PAGE, and the identity of the Cdc25A wild type catalytic domain was confirmed by N-terminal protein sequencing performed by the Laboratory for Macromolecular Analysis and Proteomics of the Albert Einstein College of Medicine. We routinely obtained 30 -50 mg of protein/liter of culture. The Cdc25A mutants were purified identically except that E431Q and E435Q required a 0 -0.75 M linear sodium chloride gradient for elution during the ion exchange chromatography step. Also, the yields of the E431Q mutant were significantly lower, since much of it seemed to reside in the pellet fraction.
Steady State Kinetics-All assays were performed at 30°C. The following buffers were used: 100 mM succinate (pH 5.4 -5.8), 50 mM 3,3-dimethylglutarate (pH 5.8 -7.2), and Tricine (pH 7.2-8.8). In addition, each buffer also contained 1 mM EDTA, 1 mM DTT and was adjusted to an ionic strength of 150 mM with sodium chloride. Assay mixtures of 100-l total volume were set up in 1.1-ml polypropylene tubes from Marsh Biomedical. Reactions were started by the addition of an appropriate amount of enzyme (2-10 M). For the substrates 4-acetylphenyl phosphate, 4-trifluoromethyl phenyl phosphate, 4-chlorophenyl phosphate, 4-fluorophenyl phosphate, 4-methylumbelliferyl phosphate, ␤-naphthyl phosphate, and phenyl phosphate, the reaction mixture was quenched with 50 l of 10% trichloroacetic acid, and the amount of inorganic phosphate released was quantitated based on the method of Black and Jones (47). The only differences were that 100 l of the ammonium molybdate/ascorbic acid solution was added to each quenched reaction mixture followed 2-5 min later by 200 l of the citrate/arsenite solution. 200 l of each reaction mixture was then transferred to a 96-well plate, where the absorbance at 700 nm was read using a Spectra Max 340 plate reader from Molecular Devices. Since some of these substrates are only soluble to 5-10 mM, sufficient substrate could not be added to saturate Cdc25A. Therefore, k cat /K m values were determined by employing a series of assay solutions containing a range of substrate concentrations (0.3-1.7 mM) well below the K m . The k cat /K m values were obtained from a linear least square fit of the plot of absorbance versus substrate concentration and using the appropriate form of the Michaelis-Menten equation, where v represents the initial rate and [E] is the total enzyme concentration. The substrates 2-chloro-4-nitrophenyl phosphate and pNPP were assayed similarly except that the reaction mixtures were quenched with 100 l of 2 M sodium hydroxide. 150 l of each mixture was then transferred to a 96-well plate, where the absorbance at 405 nm was read in a plate reader. In addition, both of these compounds were sufficiently soluble such that concentrations encompassing the K m value of Cdc25A could be prepared, and the Michaelis-Menten equation (Equation 2) was used to determine both k cat and K m . The initial rate versus substrate concentration could be fit directly to the equation, where v represents the initial rate, [E] is the total enzyme concentration, and [S] is the initial substrate concentration using KaleidaGraph (Synergy Software). For the substrates 3-O-methylfluorescein phosphate (OMFP) and 6,8-difluoro-4-methylumbelliferyl phosphate (DIFMUP), continuous assays were performed. In both cases, 1-ml reaction volumes were set up in 1-ml quartz cuvettes. The cuvettes were placed in the temperaturecontrolled holder of the spectrophotometer or fluorimeter and allowed to equilibrate to 30°C before an appropriate amount of enzyme was added to initiate the reaction. For OMFP, 100 M OMFP was used, and the absorbance was monitored at 400 nm over the entire reaction time course in a PerkinElmer Lambda 14 UV-visible spectrophotometer.
Both the k cat and K m values were obtained by a nonlinear least squares fit to Equation 3 as described previously (48), where [S] 0 represents the initial substrate concentration, while [P] is the product concentration at time t. For DIFMUP, 0.5 M substrate was used, and the fluorescence of the product was monitored at 450 nm using an excitation of 380 nm in a PerkinElmer Life Sciences 50 B Luminescence spectrometer. The k cat /K m value was obtained by nonlinear least square fit to Equation 4, where [S] 0 represents the initial substrate concentration, while [P] is the product concentration at time t. Leaving Group Dependence-The leaving group dependence studies were performed at pH 7.0. The log(k cat /K m ) value was plotted against leaving group pK a value for the one-ringed aryl phosphate substrates. The ␤ lg value was obtained as the slope of the linear least squares fit to the data. For the wild type Cdc25A in which there was a distinct kink in the leaving group dependence, the data from pK a 5.45 to 8.05 was fit separately from the data from pK a 8.68 to 9.99.
pH Dependence Studies-The k cat /K m values were plotted as a function of pH for several substrates. The second ionization constants (pK S2 ) of 5.00 Ϯ 0.01, 5.24 Ϯ 0.01, 5.52 Ϯ 0.01, and 5.72 Ϯ 0.02 were obtained, by titration, for the substrates pNPP, 4-acetylphenyl phosphate, 4-chlorophenyl phosphate, and phenyl phosphate, respectively. To fit the k cat /K m versus pH data, one of two equations was used, where (k cat /K m ) max is the highest theoretical value of k cat /K m , H is the proton concentration, K S2 is the second ionization constant of the substrate, and K E1 and K E2 are the ionization constants of the enzyme. For the wild-type Cdc25A using the substrates 2-chloro-4-nitrophenyl phosphate and pNPP and for the E431Q mutant in all cases,  18 O]p-nitrophenol were then mixed to reconstitute the natural abundance of 15 N, and then the mixture was phosphorylated to produce p-nitrophenyl phosphate using the same method as referred to above. This mixture was used for determination of 18 (V/K) bridge . The [ 14 N]p-nitrophenyl phosphate and [ 15 N,nonbridge-18 O 3 ]p-nitrophenyl phosphate were also mixed to reconstitute the natural abundance of 15 N. This mixture was used for determination of 18 (V/K) nonbridge . The isotopic abundance of the mixtures was determined by isotope ratio mass spectrometry.
Isotope effect determinations were made at 30°C, in either 500 mM Tricine buffer (pH 7.2) or 250 mM BisTris buffer (pH 6.1), containing 1 mM DTT. The buffer was degassed by bubbling nitrogen through it for 30 min before initiating experiments. The protocols for carrying out these experiments were the same as those previously described (25). The enzymatic reactions were performed in triplicate and allowed to proceed to fractions of reaction ranging from 33 to 65% and stopped by titration to pH 4. The reaction solutions were then extracted three times with an equivalent volume of ethyl ether to quantitatively remove the p-nitrophenol product. The aqueous fractions containing unreacted pNPP were titrated to pH 9.0 and treated with excess alkaline phosphatase overnight to hydrolyze the remaining pNPP. The nitrophenol thus released was isolated as described above. The nitrophenol samples were purified by sublimation and analyzed by isotope ratio mass spectrometry. Isotopic analyses were performed using an ANCA-NT combustion system in tandem with a Europa 20 -20 isotope ratio mass spectrometer.
Reactions were begun with 100 mol or more of the substrate and sufficient enzyme so that the background hydrolysis rate was negligible compared with the enzymatic reaction. Parallel experiments without enzyme were used to establish background hydrolysis rates under the experimental conditions. Isotope effects were calculated from the isotopic ratio in the p-nitrophenol product at partial reaction (R p ), in the residual substrate (R s ), and in the starting material (R o ). Equations 1 and 2 were used to calculate the observed isotope effect either from R p and R o or from R s and R o , respectively at fraction of reaction f (50). Thus, each experiment yields two independent determinations of the isotope effect.
R O was determined separately from unreacted substrate by isotope ratio mass spectroscopic analysis and, as a control, from p-nitrophenol isolated after complete hydrolysis of substrate using the isolation and purification procedures used in the isotope effect experiments. The agreement of these two numbers demonstrated that, within experimental error, no isotopic fractionation occurs as a result of the procedures used to isolate and purify the p-nitrophenol. The 18 O isotope effects were measured by the remote label method (51), as previously described for the solution reactions of pNPP (49). In these experiments, the nitrogen atom in the substrate is used as a reporter for the bridging oxygen atom or the nonbridging oxygen atoms. These experiments yield an observed isotope effect that is the product of the effect due to 15 N and to 18 O substitutions. The observed isotope effects from these experiments were then corrected for the 15 N effect and for incomplete levels of isotopic incorporation in the starting material as previously described (52).
The notation used to express isotope effects is that of Northrop (53), in which a leading superscript denoting the heavier isotope indicates the isotope effect on the following kinetic quantity; for example, 15 k denotes k 14 /k 15 , the nitrogen-15 isotope effect on the rate constant k. Since all isotope effects in this study were measured by the competitive method, they are isotope effects on V/K, and thus are designated as 15 (V/K), etc. Fig. 1 shows a diagram of the substrate with the nomenclature used to describe the isotope effects at each position.

RESULTS AND DISCUSSION
The goals of this study were to further define the catalytic mechanism of Cdc25A phosphatase and identify the catalytically important residues, and in particular to identify the general acid. Structural studies of human Cdc25A (44) and Cdc25B (45) phosphatases revealed enzymes with shallow, solvent-exposed active sites with relatively few protein functional groups in positions where they would be likely to participate in catalysis. This is in contrast to other PTPases such as Yersinia PTPase (54), PTP1B (55-57), the low molecular weight PTPase (58,59) and VHR (60), which have surface loops that close around the substrate, bringing a variety of functional groups, including the general acid, into the active site and virtually surrounding the substrate. In Cdc25, the only residues that appear able to participate directly in catalysis are located in the phosphate-binding loop itself, corresponding to the CX 5 R motif (residues Cys 430 -Arg 436 in Cdc25A). For example, it has been suggested that either Glu 431 (44) or Glu 435 (45) may serve as the general acid for Cdc25. From the crystal structures, it appeared that these were the only acidic residues that would be in the vicinity of the bridging oxygen in a phosphate ester substrate. It was also suggested that F432 may play a role in binding aryl phosphate substrates such as phosphotyrosine by participating in a hydrophobic interaction with the phenyl ring of the substrate (44). We, therefore, decided to determine the role of these residues in catalysis through site-directed mutagenesis and kinetic analysis.
Hallmarks of General Acid Catalysis-General acid catalysis in PTPases has several distinguishing features that can be revealed experimentally through kinetic analysis. First, general acid catalysis in PTPases is marked by characteristic kinetic isotope effects, which will be discussed below. Second, PTPases with general acids show bell-shaped pH profiles where the maximum catalytic rate usually occurs around pH 5-6 with aryl phosphates as substrates. Third, PTPases with general acids show relatively flat leaving group dependences (i.e. the rate of catalysis is affected minimally by the pK a of the leaving group). Fourth, removal of the general acid through site-directed mutagenesis results in characteristic changes in the kinetic isotope effects, a flat pH profile with a significant drop in the rate of catalysis, and a significant leaving group dependence where the rate of catalysis drops as the leaving group pK a rises.
Mutation of Acidic and Hydrophobic Residues within the Cdc25A Active Site Has Little Effect on pNPP Hydrolysis-The catalytic domain of Cdc25A (residues 336 -523) and several point mutants were constructed, overexpressed in E. coli, and purified to Ͼ95% purity based on SDS-PAGE. The kinetic parameters of these mutants using pNPP as a substrate are shown in Table I. As shown in Table I, mutation of either acidic residue in the active site loop of Cdc25A has only modest effects on the kinetic parameters. In the case of both E431Q and E435Q, the K m decreases 2-fold, while the k cat is enhanced slightly. The Phe 432 mutant, likewise, shows only modest differences from the wild type enzyme, a 2-fold drop in both k cat and K m , leaving the k cat /K m value virtually the same as wild type. These results suggest that neither of these acidic residues nor Phe 432 are required for the hydrolysis of pNPP.
The Addition of a Serine Residue after the Active Site Arginine Has a Detrimental Effect on Cdc25A PTPase Activity-All PTPases identified to date, except Cdc25, have an active site loop consisting of CX 5 R(S/T). The serine/threonine hydroxyl hydrogen-bonds with the active site cysteine sulfur atom (21,54,55,58,60,61). This interaction stabilizes the thiolate anion and assists in the hydrolysis of the phosphocysteine intermediate (21,32,(62)(63)(64). It has been known for some time that the activity of Cdc25 against pNPP is orders of magnitude below that of most other phosphatases such as Yersinia PTPase, PTP1B, or VHR (65). Unlike these enzymes, Cdc25 phosphatases do not contain a hydroxyl in this position, instead containing a glycine. One possibility is that Cdc25's relatively low pNPP activity results from the absence of this hydroxyl group. To test this possibility, we mutated this glycine into a serine residue in order to determine whether this would rescue Cdc25's pNPP activity. As Table I shows, this mutation resulted in a 5-fold reduction in k cat . Therefore, rather than rescuing Cdc25's activity, this mutation was actually detrimental, probably due to perturbation of the active site structure.
Kinetic Isotope Effects-Another possibility is that Cdc25's low pNPP activity is due to a lack of a general acid to facilitate catalysis. One of the best ways to probe the transition state of PTPase-catalyzed reactions and reveal the existence of a general acid is through the measurement of kinetic isotope effects. In the past, this approach has been highly successful in revealing the existence of general acid catalysis and identifying the residues that function as general acids in PTPases such as Yersinia PTPase and PTP1 (31), the low molecular weight PTPase Stp1 (66), VHR (37), and MKP3 (38). We decided to determine the kinetic isotope effects for the wild-type Cdc25A in order to determine whether it in fact does employ general acid catalysis. Fig. 1 shows the structure of pNPP along with the isotope effects that were measured in this study. Table II shows the resulting isotope effects for the wild-type Cdc25A-catalyzed pNPP hydrolysis reaction along with isotope effects for the reactions catalyzed by wild-type VHR and the general aciddeficient VHR determined in a previous study (37) for comparison. The kinetic isotope effect data for the VHR-catalyzed reaction are typical of those obtained with the Yersinia PTPase, PTP1, and Stp1.
Isotope effects are only expressed for an enzyme-catalyzed reaction when certain conditions are met. Namely, the chemical step must be at least partially rate-limiting, and the formation of the enzyme-substrate complex must be freely reversible. As Table II indicates, significant isotope effects were observed for Cdc25A. Because the observed isotope effects are large and similar to those of the uncatalyzed reactions, it suggests that chemistry is at least largely rate-limiting for the hydrolysis of pNPP by Cdc25A.
The observed isotope effects reveal several important details about the mechanism of Cdc25A. The primary 18 (V/K) bridge isotope effect reveals the extent of P-O bond cleavage in the transition state (49). The large 18 (V/K) bridge of 2.91-3.57% for Cdc25A indicates a substantial extent of bond cleavage between the phosphorus atom and the leaving group in the transition state. This effect is of similar magnitude to that of either the general acid-deficient PTPases (31,38,66) or the dual specific phosphatase VHR/D92N, which has a 18 (V/K) bridge isotope effect of 2.94%. For PTPases in which a general acid is functioning, the 18 (V/K) bridge isotope effect is suppressed due to protonation, as O-H bond formation partially compensates for O-P bond cleavage. This can be seen in the native VHR 18 (V/K) bridge isotope effect of 1.18%, which is 2.5-fold lower than that of the general acid-deficient VHR/D92N. The high magnitude of the 18 (V/K) bridge isotope effect of 3.57% for Cdc25A suggests that Cdc25A does not employ a general acid for pNPP hydrolysis.
The 15 (V/K) isotope effect is sensitive to the amount of negative charge that is delocalized into the phenyl ring at the transition state and therefore reflects the amount of negative charge that develops on the leaving group as the P-O bond is broken (49). For PTPases with functional general acids, this isotope effect is unity within experimental error, as typified by the value of 15 (V/K) for wild type VHR (Table II). This is attributed to protonation of the bridging oxygen by the general acid (Asp 92 for VHR), which neutralizes the negative charge on the bridging oxygen as the P-O bond is broken. When the general acids of PTPases are impaired, the 15 (V/K) isotope effect becomes more like that of the uncatalyzed hydrolysis of the pNPP dianion, which is 1.0028 (0.28%) at 95°C. For VHR/ D92N, 15 (V/K) is 0.30%. Strikingly, the 15 (V/K) isotope effect is also 0.30% for the wild-type Cdc25A at pH 7.2, and this value is unchanged when the isotope effect is measured at pH 6.1. Together, this implies that Cdc25A does not employ general acid catalysis in the hydrolysis of pNPP. This is a very rare finding that has never before been observed for native PTPases.  1. The structure of pNPP and the isotope effects determined for Cdc25A in this study. The isotope effect notation used is described under "Materials and Methods." Additionally, the 18 (V/K) nonbridge isotope effect indicates the degree to which the transition state is associative or dissociative (67). Experimental results for nonbridge isotope effects in phosphoryl transfer reactions generally follow the trend suggested by calculations that predict normal 18 k nonbridge isotope effects for associative transition states and small, inverse values for dissociative ones. For the hydrolysis of the dianion of pNPP, which proceeds by a loose transition state, 18 k nonbridge ϭ 0.9994. The tighter transition states of diesters and triesters are characterized by values for 18 k nonbridge in the range of 1.0028 -1.0254. In all previous cases examined, when the general acid of a PTPase is mutated, the 18 (V/K) nonbridge isotope effect becomes more normal, indicating that a slightly more nucleophilic participation is required to expel the leaving group. Interestingly, for Cdc25A, although the 18 (V/K) bridge and the 15 (V/K) clearly indicate that there is no general acid catalysis, the value for 18 (V/K) nonbridge is even more inverse than that of the solution dianion reaction. This indicates that Cdc25A employs a different strategy for transition state stabilization in the absence of general acid catalysis than do PTPases that have their general acids removed via site-directed mutagenesis.
Taken together, the kinetic isotope effects for Cdc25A reveal three important things about the mechanism of pNPP hydrolysis by Cdc25A. First, the presence of substantial isotope effects, some of which are at their maximum expected values, reveals that the chemical step of phosphoryl transfer from substrate to the enzymatic nucleophile is at least largely ratelimiting for V/K. Second, the substantial, normal 18 (V/K) bridge and 15 (V/K) isotope effects reveal that Cdc25A does not employ general acid catalysis for pNPP hydrolysis. And third, the 18 (V/K) bridge and the inverse 18 (V/K) nonbridge isotope effects reveal a highly dissociative transition state for the hydrolysis of pNPP by Cdc25A.
Active Site Preferences of Cdc25A-Clearly, based on the kinetic isotope effects, Cdc25A does not employ general acid catalysis for the hydrolysis of pNPP. Another way to determine the presence or absence of general acid catalysis is by performing a leaving group dependence study and determining the ␤ lg value. A ␤ lg value is obtained from the slope of a plot of the logarithm of the hydrolysis rate constant versus the pK a of the leaving group of the substrate. The ␤ lg value, like the 15 (V/K) isotope effect, reflects the amount of negative charge delocalized into the phenyl ring at the transition state. The solution dianion reaction for a series of aryl phosphate esters shows a strong leaving group dependence, with a ␤ lg value of Ϫ1.23 (68). In the uncatalyzed reaction at lower pH where the monoanion is the reactive species, the leaving group is protonated in the transition state, and ␤ lg is reduced to Ϫ0.27 (68). For PTPases with general acid catalysis, the leaving group dependence is generally similar to that observed for the uncatalyzed monoanion reaction (35, 69 -71) or near zero (20,34,(72)(73)(74). We decided to test the ability of Cdc25A to hydrolyze a variety of aryl phosphates with varying leaving group pK a values and steric properties. We did this in order to both determine the ␤ lg value and determine the steric preferences of Cdc25A's active site. Table III shows the k cat /K m values of Cdc25A and two of its site-directed mutants for a variety of substrates. The value of k cat /K m was chosen for comparison, because it is known, from the kinetic isotope effects, to reflect the rate of the first chemical step in catalysis for Cdc25A in pNPP hydrolysis. The k cat /K m value is also a measure of the substrate specificity. A comparison of some of the k cat /K m values reveals several important features of the active site preferences and mechanism of Cdc25A. First, Cdc25A shows a strong (73-fold) preference for OMFP over DIFMUP although these two substrates have very similar pK a values. Cdc25A also shows a more modest preference for two-ringed substrates over one-ringed substrates. The k cat /K m value for ␤-naphthyl phosphate is 3.5-fold higher than that for 4-chlorophenyl phosphate although these substrates share the same pK a value, and the k cat /K m value for 4-methylumbelliferyl phosphate is 2.3-fold higher than that for 4-acetylphenyl phosphate although these substrates also have similar pK a values. These results are similar to those obtained for another dual specificity phosphatase, MKP3, for which it was shown that the enzyme has a strong preference for OMFP over pNPP (280-fold) and a more modest preference for ␤-naphthyl phosphate over pNPP (5-fold) (73). This preference for bulky substrates has been noted for several dual specificity phosphatases (72)(73)(74)(75)(76)(77), which may be the result of the shallow active site pockets exhibited by dual specificity PTPases, which allow these enzymes to accommodate phosphothreonine and phosphoserine as well as phosphotyrosine.
Leaving Group Dependence of Cdc25A-Comparison of substrates with similar structures but different pK a values reveals that the pK a of the substrate can be an important factor in the rate of Cdc25A catalysis. DIFMUP and 4-methylumbelliferyl phosphate (MUP) differ structurally only by two fluorine atoms, yet their leaving group pK a values are significantly different, 4.7 for DIFMUP and 7.8 for MUP. This difference leads to a 430-fold enhancement in k cat /K m for DIFMUP versus MUP. A similar pattern is observed for pNPP and 2-chloro-4-nitrophenyl phosphate. The addition of a chlorine atom at the 2-position to pNPP decreases the pK a of the leaving group from 7.1 to 5.5 and increases the k cat /K m value for Cdc25A 16-fold. Clearly, the pK a of the leaving group can have a dramatic effect on the rate of catalysis for Cdc25A.
We further quantified this leaving group effect by constructing a Bronsted plot with the one-ringed aryl phosphate sub-  strates. This is shown in Fig. 2. Since the purpose of a Bronsted plot is to quantify the effect of the electronic properties of the substrates on the rate of catalysis, it is best to minimize the steric differences between the substrates used in such a plot.
Since we have shown that the steric properties of substrates alone can have substantial effects on the rate of catalysis, we chose to include only the one-ringed aryl phosphates in the Bronsted plot. The most dramatic feature of the Bronsted plot for Cdc25A is the kink at the pK a value of 8. In the low pK a region, from 5.45 to 8.05, the slope is substantially negative, giving a ␤ lg value of Ϫ0.8 Ϯ 0.1. This value indicates that a substantial amount of negative charge develops on the leaving group oxygen at the transition state and suggests that no general acid catalysis occurs for substrates with leaving group pK a values between 5.45 and 8.05. Since pNPP has a leaving group pK a of 7.14, this suggests that general acid catalysis does not occur for the hydrolysis of pNPP by Cdc25A, which is entirely consistent with the isotope effects for pNPP discussed above. Surprisingly, the leaving group dependence from pK a 8.05 to 9.99 is flat, giving a ␤ lg value of around zero (0.08 Ϯ 0.02). A kink in the leaving group dependence is usually attributed to a change in the mechanism or a change in the ratelimiting step. A flat leaving group dependence suggests one of two things for phosphate ester hydrolysis; either protonation of the leaving group occurs at the transition state as in general acid catalysis, or cleavage of the phosphate ester bond is not the rate-limiting step. We performed several further studies in an attempt to distinguish these possibilities.
The pH Dependence of Cdc25A-catalyzed Reaction-To ascertain the possible involvement of a general acid in hydrolyzing these high leaving group pK a substrates we performed pH dependence studies with several aryl phosphate substrates with a wide range of leaving group pK a values. Fig. 3A shows the pH dependence of wild-type Cdc25A-catalyzed hydrolysis of the aryl phosphates 2-chloro-4-nitrophenyl phosphate, pNPP, 4-acetylphenyl phosphate, 4-chlorophenyl phosphate, and phenyl phosphate. Not surprisingly, pNPP shows relatively flat pH dependence between pH 6 and 8.8, which is characteristic of a lack of general acid catalysis and consistent with the kinetic isotope effects and the leaving group dependence mentioned above. The substrate 2-chloro-4-nitrophenyl phosphate, which has a leaving group pK a of 5.45, also shows a flat pH dependence, which is again characteristic of a lack of general acid participation and consistent with the leaving group dependence for substrates in this low leaving group pK a range. The k cat /K m values below pH 5.8 could not be accurately measured because of the instability of Cdc25A at these pH values. Table  IV lists the pH-dependent parameters for the data displayed in Fig. 3 along with the equation that was used to fit the data. The second ionization constants of the substrates were fixed at the experimentally determined values. The observed pK E1 (5.47-5.57) for the pNPP and 2-chloro-4-nitrophenyl phosphate reactions probably corresponds to the active site cysteine nucleophile and is similar to that (5.6 -6.3) observed for Cdc25B (77).
Interestingly, the substrates 4-chlorophenyl phosphate and phenyl phosphate, which have leaving group pK a values of 9.38 and 9.99, respectively, show pH optima at 6.2 and steadily decreasing k cat /K m values at higher pH values, quite unlike the pH dependences of the lower pK a substrates. This behavior is consistent with a Cdc25A residue that must be protonated for optimal catalytic efficiency, possibly a general acid. The pH dependences for these two substrates are similar to the pH dependences of other general acid-containing PTPases such as Yersinia PTPase (30), PTP1 (70), or VHR (63). Since the pK E1 values (ionization constants of the active site Cys residue) were largely obscured by the ionizations giving rise to pK E2 , the pK E1 values were fixed at 5.5 (derived from the pNPP and 2-chloro-4-nitrophenyl phosphate reactions) for these two substrates, and Equation 6 was used to fit the data. As Table IV indicates, an apparent pK E2 value of 6.00 -6.06 was obtained for these substrates, similar to that obtained for Cdc25B with the physiological substrate (77) and consistent with the pK a values obtained for the general acid of other PTPases (30,63,70).
The pH dependence of 4-acetylphenyl phosphate, which has a leaving group pK a of 8.05, appears to show a pH dependence intermediate between those of the high pK a substrates and the low pK a substrates. The k cat /K m is highest at pH 5.8 and decreases as the pH is raised but not as drastically as for 4-chlorophenyl phosphate or phenyl phosphate. Both the pH dependence and leaving group dependence of Cdc25A are consistent  E431Q (filled triangles). The leaving group dependences were determined at pH 7.0, 30°C. The lines are based upon the linear least squares fit to the log(k cat /K m ) versus leaving group pK a . For wild-type Cdc25A, two separate linear regressions were done, one for the substrates with leaving group pK a values 5.45-8.05 (␤ lg ϭ Ϫ0.8 Ϯ 0.1) and one for the substrates with leaving group pK a values 8.68 -9.99 (␤ lg ϭ 0.08 Ϯ 0.02). For the E431Q mutant, the substrates with leaving group pK a values 5.45-8.05 gave a ␤ lg value of Ϫ1.3 Ϯ 0.2, while the substrates with leaving group pK a values 8.68 -9.99 had k cat /K m values that were too low to be measured accurately. with a general acid independent mechanism for substrates with leaving group pK a values below 8 and a general acidcatalyzed mechanism for substrates with leaving group pK a values above 8.
Effect of Altering the Active Site Acidic Residues on the pH Dependence and Leaving Group Dependence Studies-In order to identify the residue that appears to serve as a general acid for the hydrolysis of the high leaving group pK a substrates, we determined the leaving group dependence and the pH dependence for the two site-directed mutants of Cdc25A, E431Q, and E435Q. As stated before, neither mutant showed deficient pNPP activity, which is not surprising, since all of the results discussed previously indicate that Cdc25A does not employ general acid catalysis for the hydrolysis of pNPP. Table III shows the kinetic parameters obtained with these mutant enzymes. The E435Q mutant followed the same general trends as the wild-type Cdc25A in both leaving group dependence and pH dependence. Namely, the leaving group dependence is substantially negative below a pK a value of 8 and flat from 8 to 10, while the pH dependence for pNPP is flat, and the pH dependence for phenyl phosphate is more bell-shaped (data not shown). We therefore concluded that Glu 435 does not act as a general acid and focused on the E431Q mutant.
The leaving group dependence of E431Q is shown in Fig. 2. The ␤ lg value obtained using the substrates 2-chloro-4-nitrophenyl phosphate, pNPP, and 4-acetylphenyl phosphate was Ϫ1.3 Ϯ 0.2. This value is even greater than that obtained for the wild-type enzyme and similar to the nonenzymatic solution reaction (␤ lg ϭ Ϫ1.23 (68)) and is consistent with an absence of general acid catalysis. Further, the difference in k cat /K m values for the substrates 6,8-difluoro-4-methyl umbelliferyl phosphate and 4-methylumbelliferyl phosphate corresponds to a ␤ lg value of Ϫ1.35. Unfortunately, we were not able to obtain k cat /K m values for the substrates with leaving group pK a values higher than 8.05 for E431Q (Table III), so we do not know whether the trend in leaving group dependence would continue over this range. The signal was simply too low and not reproducible, even at high enzyme concentrations (up to 13 M) and long incubation times (4 h). We estimate that the k cat /K m values for the high pK a substrates 4-trifluoromethylphenyl phosphate, 4-chlorophenyl phosphate, 4-fluorophenyl phosphate, and phenyl phosphate are well below 0.1 M Ϫ1 s Ϫ1 and are at least 10-fold lower than that of 4-acetylphenyl phosphate. This is a significant difference from both the wild-type enzyme and E435Q, which show virtually the same k cat /K m values for all of these substrates. We therefore conclude that Glu 431 plays an essential role in the efficient hydrolysis of substrates with leaving group pK a values higher than 8.
We also examined the pH dependence of E431Q using the substrates 2-chloro-4-nitrophenyl phosphate, pNPP, and 4-acetylphenyl phosphate as shown in Fig. 3B. Just as for the wild-type enzyme, relatively flat pH dependences were ob-served for 2-chloro-4-nitrophenyl phosphate and pNPP. As Table IV shows, pK E1 values of 6.06 Ϯ 0.18 and 6.20 Ϯ 0.08 were obtained for 2-chloro-4-nitrophenyl phosphate and pNPP reactions, respectively. However, a flat pH dependence was also observed for the 4-acetylphenyl phosphate reaction with a pK E1 value of 6.51 Ϯ 0.14. The pK a value of the active site Cys residue (pK E1 ϭ 6.1-6.5) for the E431Q mutant is substantially higher than that (pK E1 ϭ 5.5) for the wild-type Cdc25A, possibly due to the elimination of the carboxyl group near the active site Cys residue. The flat pH dependence for 4-acetylphenyl phosphate is quite unlike the wild-type enzyme, which showed a steady decrease in k cat /K m value as the pH was raised. Thus, mutating residue 431 from glutamic acid into glutamine abolishes the basic limb of the pH profile for 4-acetylphenyl phosphate. Therefore, we conclude that Glu 431 is responsible for the basic limb of the pH profile seen for wild-type Cdc25A with 4-acetylphenyl phosphate (Fig. 3A) and that Glu 431 must be protonated for optimal catalysis of this substrate. Glutamic acid 431 most likely plays the same role in the hydrolysis of the higher pK a substrates 4-trifluoromethylphenyl phosphate, 4-chlorophenyl phosphate, 4-fluorophenyl phosphate, and phenyl phosphate. Based on the fact that mutation of glutamic acid 431 into glutamine both greatly diminishes Cdc25A's ability to hydrolyze high pK a substrates and eliminates the pH dependence of 4-acetylphenyl phosphate, we conclude that Glu 431 must be protonated in order for Cdc25A to efficiently hydrolyze substrates with leaving group pK a values of 8 and higher.
It should be noted that one possible criticism of this study is that the phenomena that we have observed are the result of phosphatase contaminants in our preparations of Cdc25A that copurified with it. It could be argued that the kink in the leaving group dependence for the wild-type enzyme is the result of impurities that hydrolyze the high pK a substrates very efficiently. And since these enzymes were purified by ion exchange chromatography, mutation of glutamic acid 431 into glutamine would simply eliminate a negative charge on the enzyme and cause E431Q to be purified without these impurities. Thus, it could be argued that the differences observed between the wild-type enzyme and E431Q are the result of phosphatase impurities present in the wild-type preparation that are not present in the E431Q preparation. We have taken several approaches to address this issue (data not shown). First, we performed leaving group dependence studies on different forms of Cdc25 proteins obtained under much different purification conditions, reasoning that different forms of Cdc25 should copurify with much different impurities. The same leaving group dependence pattern was observed for Cdc25A/ E435Q, glutathione S-transferase-Cdc25A, and glutathione S-transferase-Cdc25B, suggesting that this kink in the leaving group dependence was due to Cdc25A and not impurities. Next, we determined the IC 50 value of vanadate for Cdc25A using both pNPP and 4-chlorophenyl phosphate as substrates. In The pH dependence parameters were obtained by fitting the kinetic data to Equation 5 or 6 as indicated and described under "Materials and Methods." 2-Cl-4-NO 2 , 2-chloro-4-nitrophenyl phosphate; pNPP, 4-nitrophenyl phosphate; 4-Cl, 4-chlorophenyl phosphate; PhPi, phenyl phosphate; 4-Ac, 4-acetylphenyl phosphate. 4-Ac 5.24 6.50 Ϯ 0.14 1.9 Ϯ 0.1 5 each case, we obtained the same IC 50 value of 9 M, which suggests that both of these substrates are hydrolyzed via the same active site. We also performed iodoacetate inactivation of Cdc25A. Iodoacetate can specifically modify the active site Cys residue in PTPases, leading to enzyme inactivation (78). Since Cdc25A has an active site thiolate that is required for activity, iodoacetate should completely inactivate Cdc25A. And since E. coli is not known to have any thiol-dependent phosphatases, iodoacetate should not inactivate any E. coli-derived phosphatase impurities. Upon incubation of wild-type Cdc25A with iodoacetate, we observed complete inactivation within 6 min, suggesting that all phosphatase activity in the Cdc25A preparation was the result of Cdc25A and not impurities. Additionally, we observed the same rate of inactivation whether pNPP or phenyl phosphate was used to measure residual phosphatase activity, which again suggests that these substrates are hydrolyzed via the same active site. Taken together, these results show that the kinetic data presented in this study are the intrinsic properties of Cdc25A and its mutants and not phosphatase impurities.
Proposed Models for Cdc25A Catalysis-The finding that Cdc25A appears to hydrolyze low leaving group pK a substrates, like pNPP, without the aid of a general acid, and higher pK a substrates with a general acid is puzzling. We propose several possible kinetic models, shown in Figs In model A (Fig. 4), we propose that Glu 431 can act as a general acid, but it is poorly positioned to participate efficiently in hydrolysis of small molecule substrates. For the substrates with low leaving group pK a values (2-chloro-4-nitrophenyl phosphate or pNPP), the transition state for phosphoenzyme intermediate formation is reached well before any protonation of the leaving group can occur. The kinetic isotope effects are consistent with this, indicating no protonation of the leaving group, para-nitrophenol, at the transition state. Since the transition state can be reached easily for these substrates without protonation of the leaving group, the rate of catalysis does not depend upon the protonation state of the enzyme. Therefore, these substrates show flat pH dependences. The substrates with higher leaving group pK a values, 4-chlorophenyl phosphate and phenyl phosphate, are intrinsically much less reactive than 2-chloro-4-nitrophenyl phosphate or pNPP. Therefore, the attainment of the transition state will occur much more slowly for the higher leaving group pK a substrates, in the absence of a general acid, than for the lower leaving group pK a substrates. This is depicted as the set of arrowheads of descending length emanating from the enzyme-substrate complex in which Glu 431 is not protonated in the bottom half of Fig. 4. For the higher leaving group pK a substrates, the attainment of the transition state is sufficiently slow that protonation of the leaving group occurs at or before the transition state when the enzyme is in its properly protonated form. Therefore, protonation of the leaving group facilitates catalysis for the substrates 4-acetylphenyl phosphate, 4-chlorophenyl phosphate, and phenyl phosphate, but the rate of hydrolysis of these substrates is still relatively low because the general acid is not optimally FIG. 4. Model A for Cdc25A catalysis. Glu 431 (E431) is shown as the general acid, but it is not properly positioned to facilitate catalysis efficiently. Therefore, protonation of the leaving group lags behind the attainment of the transition state for the low leaving group pK a substrates, which are intrinsically more reactive than the high leaving group pK a substrates. Attainment of the transition state for the high pK a substrates is sufficiently slow to require protonation of the leaving group to occur at or before the transition state. positioned when bound with small molecule substrates. As depicted in Fig. 4, the rate of general acid-independent catalysis is much lower than the rate of general acid-dependent catalysis for the substrates 4-chlorophenyl phosphate and phenyl phosphate. Therefore, the rate of catalysis for these substrates depends upon the proportion of Cdc25A in which Glu 431 is protonated, and they produce bell-shaped pH profiles. For the substrate 4-acetylphenyl phosphate, the rate of general acid-independent catalysis is more comparable with the rate of general acid-dependent catalysis. For this reason, as the pH of the buffer is raised, the k cat /K m value decreases, but not as dramatically as for 4-chlorophenyl phosphate or phenyl phosphate.
In model B (Fig. 5), we propose that Glu 431 is a general acid but that it is poorly positioned to participate in catalysis in the initial enzyme-substrate complex. The enzyme-substrate complex can subsequently undergo a relatively slow conformational rearrangement to bring the Glu 431 into proper position to function as a general acid. At the far left of Fig. 5, Cdc25A is shown as an equilibrium between its general acid-protonated and deprotonated forms. The enzyme and substrate then come together to form the enzyme-substrate complex in which Glu 431 is not properly positioned to function as a general acid. At this point, there are two competing processes: general acid-independent catalysis and a conformational change that brings Glu 431 into proper position for catalysis. For the low leaving group pK a substrates, the general acid-independent catalysis is much faster than the conformational change, whereas for the high pK a substrates, the conformational change is much faster than the general acid-independent catalysis. Once the conformational change occurs that brings Glu 431 into proper position, general acid-catalyzed hydrolysis will presumably occur rapidly, regardless of the identity of the substrate, as long as Glu 431 is protonated. The substrates 2-chloro-4-nitrophenyl phosphate and pNPP primarily proceed through the general acid-independent route, before the conformational change has had time to occur. Since this route does not depend upon the protonation state of Glu 431 , the rate of hydrolysis of these substrates shows little pH dependence. As the pK a of the substrate leaving group is raised, the substrate becomes less and less reactive in the absence of a general acid, which is depicted in Fig. 5 as the series of arrows of decreasing size emanating from the enzyme-substrate complex structures in which Glu 431 is deprotonated or poorly positioned. Therefore, the substrates 4-chlorophenyl phosphate and phenyl phosphate primarily proceed through the acid-catalyzed route, because the rate of conformational change followed by acid catalysis is simply much faster than the rate of hydrolysis in the absence of a general acid for these substrates. The overall rate of hydrolysis for these high leaving group pK a substrates depends upon both the rate of conformational change and the proportion of Cdc25A molecules that have Glu 431 protonated. As a result, the k cat /K m values for these substrates decreases as the pH of the buffer is increased.
For the substrate 4-acetylphenyl phosphate, which has a leaving group pK a that lies between those of pNPP and 4-chlorophenyl phosphate, the reaction seems to proceed through a mechanism that lies between those of the high and low pK a substrates. It is likely that for this substrate, the rate of general acid-independent catalysis is comparable with the rate of conformational change. As a result, the hydrolysis of this substrate can proceed via the general acid-independent or the FIG. 5. Model B for Cdc25A catalysis. A slow conformational change can bring Glu (E431) into the active site to act as a general acid. The low leaving group pK a substrates are sufficiently reactive that they can be hydrolyzed more rapidly than the conformational change can occur, whereas the high leaving group pK a substrates are sufficiently unreactive that the conformational change can occur, which brings the general acid into the proper position to facilitate catalysis. acid-catalyzed route efficiently. However, as the pH of the buffer is raised, less of the protonated form of the enzyme is available, and a greater proportion of the hydrolysis takes place through the general acid-independent routes. The net result is a significant pH dependence for 4-acetylphenyl phosphate but one that is substantially flatter than the pH dependence for the higher pK a substrates.
In model B, we have shown Glu 431 as the general acid protonating the leaving group directly. It is also possible that Glu 431 is simply a residue that must be protonated in order for Cdc25A to achieve a more active conformation, perhaps through a slow conformational change involving rearrangement of the active site phosphate-binding loop. In this case, the reasoning would be essentially identical as model B, so we have not shown this possibility as a separate model. In both cases, the conformational change must be rate-limiting for substrates with more basic leaving groups, in order to account for the lack of leaving group effect.
The idea that a conformational change occurs during the mechanism was tested by determining the kinetic constants of Cdc25A in 45% glycerol. It has been suggested that glycerol can alter the conformational properties of PTPases, such as MKP3 for which glycerol appears to stabilize the more active conformation (73). In both cases in model B (shown in Fig. 5), we would expect glycerol to have differential effects on different substrates. Indeed, we found that 45% glycerol modestly activates Cdc25A toward pNPP (1.5-fold), more strongly activates Cdc25A toward 4-acetylphenyl phosphate (2-fold), and more substantially activates Cdc25A's activity toward phenyl phosphate (4.2-fold) (Table V). Further, when Glu 431 was mutated, these effects were no longer observed, and in fact glycerol had a slight inhibitory effect on this mutant. This suggests that glycerol stabilizes the more active form of the enzyme; in mechanism B, this corresponds to the form in which the general acid Glu 431 is properly positioned or in which the active site is more properly formed. In terms of mechanism A, perhaps glycerol could serve to position the general acid closer to the substrate and facilitate proton transfer. It has been known for some time and shown quantitatively (77) that Cdc25 has much higher activity toward cylcin-CDK complexes than for small molecule substrates. In terms of the mechanisms presented here, we suggest that binding of the physiological cyclin-CDK substrate may facilitate the proper positioning of the general acid and/or stabilizes the active conformation of Cdc25A, leading to an enhanced catalytic rate for the dephosphorylation of the physiological substrate. This is analogous to what has been observed in the dual specificity phosphatase MKP3 (22,73,74,79).
Biological Implications-Admittedly, the three possible mechanisms presented here cannot necessarily be distinguished by the experiments that we have presented. However, we have revealed some novel features of the Cdc25A mecha-nism, and some general conclusions can be drawn that are applicable regardless of the specific mechanism. The finding that Cdc25A may hydrolyze low leaving group pK a substrates, like pNPP, via one mechanism and higher leaving group pK a substrates by another is intriguing. In many studies of PTPases, pNPP is used as a substrate where it is regarded as a phosphotyrosine mimic. While pNPP is sterically very much like phosphotyrosine, it is very different electronically. The pK a of para-nitrophenol is 7.1, whereas the pK a of a tyrosine hydroxyl is 10.5. The large negative ␤ lg value (Ϫ1.23) for the solution dianion reaction means that pNPP is, intrinsically, 10,000-fold more reactive to hydrolysis than phosphotyrosine based on the difference in their pK a values. Because of this substantial difference in reactivity, it is possible that the mechanism that Cdc25A employs for pNPP hydrolysis may not reflect the mechanism that Cdc25A employs for the hydrolysis of phosphotyrosine or phosphothreonine in the physiological substrates. The pK a of tyrosine is actually much closer to that of phenol (pK a ϭ 9.99) than it is to para-nitrophenol; therefore, phenyl phosphate is a more accurate phosphotyrosine mimic than is pNPP. For this reason, it is likely that the mechanism that Cdc25A employs for the hydrolysis of higher pK a substrates such as phenyl phosphate more accurately reflects the mechanism that Cdc25A employs for the hydrolysis of physiological substrates, which have the high pK a leaving groups tyrosine and threonine.
It has been suggested recently that the general acid of Cdc25B does not actually reside on Cdc25B but may actually reside on the physiological cyclin-CDK substrate instead (77). This was concluded based on the facts that the small molecule substrates pNPP and OMFP showed flat pH dependences, whereas the bell-shaped pH profile was only observed in the hydrolysis of the physiological cyclin-CDK substrate. We, on the other hand, have shown that the pH dependence of Cdc25A for high leaving group pK a , small molecule substrates appears identical to that obtained by Chen et al. (77) using the cyclin-CDK substrate, indicating that this pH-dependent behavior is not a phenomenon particular to the physiological substrate. This suggests that the general acid may in fact reside on Cdc25 itself. Mutation of Glu 431 both drastically reduces the catalytic rate for the high leaving group pK a substrates, but not for the lower pK a substrates, and flattens the pH profile for a substrate with intermediate leaving group pK a value (4-acetylphenyl phosphate). This suggests that Glu 431 can possibly act as a general acid for the hydrolysis of these higher leaving group pK a substrates. Chen et al. (77) found that when the corresponding residue of Cdc25B (Glu 474 ) was mutated, the rate of hydrolysis of the cyclin-CDK substrate dropped 100-fold, while it remained the same for pNPP. This is consistent with our results in which the mutation of Glu 431 to glutamine dramatically decreased the rate of catalysis for the high leaving group pK a substrates but not the lower leaving group pK a substrates like pNPP. However, Chen et al. (77) found that when Glu 474 was mutated to glutamine, the pH profile for the CDK-cyclin substrate hydrolysis remained bell-shaped. This is not consistent with our results, in which we have shown that mutation of Glu 431 to glutamine flattens the pH dependence of the substrate 4-acetylphenyl phosphate. The source of this discrepancy is currently unknown.
In any case, this study reveals several important features of the mechanism of Cdc25A. First, Cdc25A does not employ a general acid in the hydrolysis of low leaving group pK a substrates like pNPP. Second, even in the absence of general acid catalysis, the transition state of pNPP hydrolysis is highly dissociative. Third, Cdc25A shows a strong preference for bulky, multiringed substrates over smaller one-ringed sub- strates. Fourth, Glu 431 plays an essential role in the hydrolysis of high leaving group pK a substrates, for which it appears to behave as a general acid. Whether or not Glu 431 is the true general acid in the hydrolysis of the physiological cyclin-CDK substrates remains to be seen. Future experiments will be directed toward this question.