Kinetic Analysis of Human Serine/Threonine Protein Phosphatase 2Cα*

The PPM family of Ser/Thr protein phosphatases have recently been shown to down-regulate the stress response pathways in eukaryotes. Within the stress pathway, key signaling kinases, which are activated by protein phosphorylation, have been proposed as thein vivo substrates of PP2C, the prototypical member of the PPM family. Although it is known that these phosphatases require metal cations for activity, the molecular details of these important reactions have not been established. Therefore, here we report a detailed biochemical study to elucidate the kinetic and chemical mechanism of PP2Cα. Steady-state kinetic and product inhibition studies revealed that PP2Cα employs an ordered sequential mechanism, where the metal cations bind before phosphorylated substrate, and phosphate is the last product to be released. The metal-dependent activity of PP2C (as reflected ink cat andk cat/K m ), indicated that Fe2+ was 1000-fold better than Mg2+. The pH rate profiles revealed two ionizations critical for catalytic activity. An enzyme ionization with a pK a value of 7 must be unprotonated for catalysis, and an enzyme ionization with a pK a of 9 must be protonated for substrate binding. Brönsted analysis of substrate leaving group pK a indicated that phosphomonoester hydrolysis is rate-limiting at pH 7.0, but not at pH 8.5 where a common step independent of the nature of the substrate and alcohol product limits turnover (k cat). Rapid reaction kinetics between phosphomonoester and PP2C yielded exponential “bursts” of product formation, consistent with phosphate release being the slow catalytic step at pH 8.5. Dephosphorylation of synthetic phosphopeptides corresponding to several protein kinases revealed that PP2C displays a strong preference for diphosphorylated peptides in which the phosphorylated residues are in close proximity.

Protein phosphatases (PP) 1 catalyze the dephosphorylation of proteins containing phosphoserine/phosphothreonine and are divided into two distinct gene families, designated PPP and PPM (1). Although both PP families require divalent cations for activity, the PPM family is often distinguished by its Mg 2ϩ and Mn 2ϩ dependence. PP2C is the defining member of the PPM family. PP2C homologues have been identified in bacteria, plants, yeast, and mammals and appear to have a conserved role in negatively regulating stress response. PP2C was shown to be a negative regulator of two mitogen-activated protein kinase (MAPK) pathways involved in stress response, the p38 and c-Jun N-terminal kinase pathways. Like other MAPK pathways, these consist of a MAPK, a MAPK kinase (MAPKK), and a MAPKK kinase (MAPKKK) (2). MAPK is phosphorylated on conserved threonine and tyrosine residues by the activated MAPKK. The MAPKK is activated by phosphorylation on conserved threonine and/or serine residues by the MAPKKK. The stress response pathways are activated by proinflammatory cytokines, osmotic shock, oxidative stress, UV irradiation, and heat shock (3,4). PP2C is thought to directly dephosphorylate and inactivate protein kinases at several levels in the pathway. The PP2C homologue, MP2C, functions as a negative regulator of a stress-activated pathway in plants. By a yeast genetic analysis, the molecular target of MP2C was determined to be the MAPKKK Ste11 (5). MKK6 and SEK 1, both MAPKKs important in stress response signaling pathways, have been identified as PP2C substrates (6). Additional MAPKKs have also been suggested to be PP2C substrates (7). PP2C␣ has been shown to dephosphorylate and inactivate the MAPK p38 (6). Genetic studies with PP2C yeast homologues Ptc1, Ptc2, and Ptc3 (8,9) implicated an essential role for PP2C in downregulating the stress response. Collectively, these studies demonstrate a major role for PP2C in turning off, or resetting, the stress response pathways. Unlike the PPP family of protein phosphatases, there are no known specific inhibitors of PP2C and no regulatory subunits have been identified. The catalytic domain appears to be sufficient to impart strict substrate specificity.
The crystal structure of PP2C␣ bound with Mn 2ϩ has been solved (10) and reveals a central ␤-sandwich surrounded by ␣-helices. The active site is located at one end of this ␤-sandwich and is composed of several invariant carboxylates, which serve as metal-coordinating residues. There are two hexa-coordinated metal sites, M1 and M2, which are 4 Å apart and share a water molecule and the carboxylate side chain of Asp-60. Site M1 is coordinated by 3 water molecules and two additional aspartic acids, Asp-239 and Asp-282. Site M2 is coordinated by 4 water molecules and makes only two direct contacts with the protein, at the backbone carbonyl of Gly-61 and at the bridging Asp-60. The presence of bound phosphate in the x-ray structure suggested the importance of an active site Arg-33. This arginine appears to position the phosphate near the water molecule bridged by the two metal ions. Das et al. (10) have proposed that catalysis proceeds by direct attack of an activated water molecule at the phosphorus center of substrate. Except for this crystallographic study, very little mechanistic data are available for this important class of enzymes. * This work was supported by the American Heart Association, American Cancer Society Grant RPG-97-175-01-TBE, and the Tartar Research Fund of the Oregon Health Sciences Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Here we describe a detailed kinetic investigation of human PP2C␣, the archetypal member of the PPM family of protein phosphatases. In this study, we explore the metal-dependent activation, identify critical enzyme ionizations, determine the kinetic mechanism, determine the rate-limiting step in catalysis and probe substrate specificity.
Overexpression and Purification of Human PP2C␣-The plasmid pCW-PP2C␣ was a generous gift from Dr. Patricia T. W. Cohen (Dundee University). The enzyme was purified according to the method described (11) with several modifications. The pCW-PP2C␣ plasmid was used to transform competent BL21/DE3 bacteria. The transformed bacteria were grown on 2ϫ YT plates containing 100 g/ml ampicillin. Overnight 10-ml cultures originating from isolated colonies were used to inoculate 1 liter of 2ϫ YT containing 140 mg/liter ampicillin. When the growth reached an optical density of 0.8 at 600 nm, 100 mg/liter isopropyl-1-thio-␤-D-galactopyranoside was added, and the bacteria were grown for an additional 10 h. The cells were harvested by centrifugation at 5,000 ϫ g (15 min), resuspended in 30 ml of buffer A (50 mM Tris-HCl, pH 7.5, 1 mM DTT, 1 mM EDTA, 0.1 mM EGTA, 100 mM NaCl, 2 mM MnCl 2 , 1 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride, 0.03% (v/v) Brij-35), and lysed by French press. The cell debris was pelleted by centrifugation at 48,000 ϫ g (20 min), the supernatant decanted, and the protein precipitated between 30% and 55% saturation with (NH 4 ) 2 SO 4 . The pellet was resuspended in 8 ml of buffer B (20 mM triethanolamine, pH 7.0, 1 mM EGTA, 2 mM MnCl 2 , 0.1% (v/v) ␤-mercaptoethanol, 5% (v/v) glycerol, 0.03% (v/v) Brij-35) and dialyzed against 1 liter of buffer B for at least 2 h with one change in buffer. The dialysate was loaded onto a 50-ml Q-Sepharose (Amersham Pharmacia Biotech) anion exchange column pre-equilibrated with buffer B. The column was then washed with 250 ml of buffer B, and the enzyme was eluted with a 0 -0.4 M linear NaCl gradient of 400 ml. PP2C␣ elutes at 250 mM NaCl. Fractions containing high phosphatase activity with para-nitrophenyl phosphate (pNPP) were pooled and concentrated to between 5 and 10 ml by ultrafiltration (Centricon 10 concentrator by Amicon). The concentrated solution was dialyzed against 1 liter of buffer containing 25 mM Na 2 HPO 4 , pH 7.0, 1 M (NH 4 ) 2 SO 4 , 1 mM DTT, and 2 mM EDTA for at least 1 h with one change of buffer and then loaded onto a 50-ml Phenyl-Sepharose (Amersham Pharmacia Biotech) hydrophobic column. The column was then washed with 250 ml of this buffer and eluted with a 1-0 M linear (NH 4 ) 2 SO 4 gradient of 400 ml. PP2C␣ elutes at 200 mM (NH 4 ) 2 SO 4 . Fractions with phosphatase activity were analyzed by SDS-polyacrylamide gel electrophoresis to determine purity. Fractions of highest purity were pooled and concentrated by ultrafiltration to a final concentration of 0.04 to 0.4 mM. PP2C␣ was dialyzed against 1 liter of 50 mM Tris-HCl, pH 7.0, 10% (v/v) glycerol, 1 mM DTT, 2 mM EDTA for 2 h with a change after 1 h, followed by a final dialysis in 50 mM Tris-HCl, pH 7.0, 10% (v/v) glycerol, 1 mM DTT for 2 h with one change. The EDTA was included to obtain metal free (apo)-PP2C. EDTA is effective in completely removing bound metal cations, as was evident from the apo-PP2C x-ray structure, which revealed a virtually identical structure to that of holoenzyme (10). The enzyme was stored at Ϫ20°C or Ϫ80°C until use.
Enzyme Kinetics-All assays were carried out in a reaction buffer containing 0.05 M Tris, 0.05 M Bis-Tris, and 0.1 M acetate at 25°C. To determine the kinetic parameters k cat and k cat /K m , the initial velocities were measured at various substrate concentrations, and the data were fitted to Equation 1. The computer program KinetAsyst (IntelliKinetics, State College, PA) was used for fitting kinetic data to Equations 1-7. Three types of assays were used to monitor PP2C phosphatase activity and are described below.
Because of its relative stability and high specific activity, the divalent cation Mn 2ϩ was utilized for the majority of the kinetic studies. The metal cation Fe 2ϩ was largely avoided due to difficulties with oxidation. For the phosphate inhibition and pH dependence studies, the cation Mg 2ϩ was used to avoid the precipitation that occurs with Mn 2ϩ in the presence of phosphate and at high pH.
Continuous Assay-The continuous assay monitoring the dephosphorylation of pNPP was performed using a temperature-controlled Shimadzu BioSpec-1601 UV-visible spectrophotometer and monitoring the absorbance change recorded at 410 nm, pH 7.0. Initial linear rates were determined using the molar extinction coefficient (⑀) of 9 mM Ϫ1 cm Ϫ1 for the product para-nitrophenol (pNP) at pH 7.0.
End Point Assays-Two end point assays were also used to measure PP2C activity. The phosphate detection assay was used for measuring the initial linear rates of dephosphorylation of peptides, amino acids, and aryl substrates. The release of phosphate was determined using the colorimetric method described by Brotherus et al. (12), which measures the formation of phosphomolybdate complexes at 850 nm. Briefly, reaction mixtures of 600 l were stopped with 1 ml of 0.5 N HCl containing 30 mg of ascorbic acid, 5 mg of ammonium heptamolybdate, and 10 mg of SDS. For color development, 1.5 ml containing 30 mg of sodium arsenite, 30 mg of sodium citrate, and 30 l glacial acetic acid were added and the absorbance was read at 850 nm. The other end point assay was developed to determine the rate of pNPP hydrolysis for the pH profile using the molar extinction coefficient of 18,000 M Ϫ1 cm Ϫ1 for pNP at basic pH values. This assay was developed after the method described in Ref. 13, which utilizes a stop solution containing 1 N NaOH. To prevent the precipitation of metal which occurs upon the addition of 1 N NaOH to a PP2C reaction mixture, a stop solution containing 0.5 M EDTA, pH 10, was used. A comparison of the continuous assay and the two end point assays revealed that these three assays yielded identical rates.
Metal Dependence-Various concentrations of metal ions were combined with apo-PP2C and allowed to incubate for 2 min. The phosphatase reaction was initiated by the addition of pNPP to a final concentration of 20 mM and activity was measured using the continuous assay. The inhibition constants for Ca 2ϩ and Zn 2ϩ were determined by adding various amounts of MnCl 2 at fixed concentrations of the inhibiting metal ion at saturating levels of pNPP. The data were fitted to Equation 2 to yield the inhibition constant. The inhibition was competitive with respect to Mn 2ϩ .
I is the inhibitor concentration and S is the Mn 2ϩ concentration, and K is is the inhibition constant. Fe 2ϩ Detection-A reaction mixture containing 3 mM Fe 2ϩ , 4 mM pNPP, and 0.59 M PP2C was separated into two cuvettes. One was monitored for phosphatase activity using the continuous assay, and the other was used to monitor the Fe 2ϩ concentration during the course of the reaction. The Fe 2ϩ concentration was determined using the Fe 2ϩ specific chelator, ferrozine, which has a characteristic absorbance at 562 nm. The measured Fe 2ϩ concentrations and PP2C activity were plotted versus time using the computer program KaleidaGraph (Abelbeck Software).
Steady-state Kinetics-The end point assay using the EDTA stop solution was used to measure the pH dependence of PP2C activity. For the construction of the pH profiles, k cat and k cat /K pNPP were obtained at various pH values with pNPP as the varied substrate at saturating levels of Mn 2ϩ (10 mM) or Mg 2ϩ (40 mM). The pH data for Mn 2ϩ and Mg 2ϩ were fitted to Equations 3 and 4, respectively, where C is the pH-independent value of either k cat or k cat /K pNPP ; H is the proton concentration; and K a and K b are the ionization constants of the groups involved in the reaction.
The catalytic mechanism of PP2C was determined by performing bisubstrate studies treating the metal cation as a pseudo-substrate. Continuous assays were performed by varying the amounts of pNPP at fixed levels of Mn 2ϩ and the data were fitted to Eq. 5 where K a and K b are the Michaelis constants for pNPP and Mn 2ϩ , respectively.
The P i product inhibition of PP2C was analyzed with Mg 2ϩ to avoid the precipitation of Mn 2ϩ and P i . Continuous assays were performed by varying Mg 2ϩ or pNPP in saturating levels of the other substrate at fixed concentrations of the product P i . The inhibition of P i with respect to pNPP was competitive, and the data were fitted to Equation 2; in contrast, the inhibition of P i with respect to metal was uncompetitive, and the data were fitted to Equation 6.
The leaving group dependence of PP2C was analyzed by determining the dephosphorylation rate of several artificial compounds, which differ by the pK a of the leaving group. Substrates 6,8-difluoro-4-methylumbelliferyl phosphate (DiFMUP) (pK a ϭ 4.9), pNPP (7.1), 4-methylumbelliferyl phosphate (7.8), ␤-naphthyl phosphate (9.38), phenyl phosphate (9.99), phosphoserine (14.1) (Ref. 14), and phosphothreonine in the peptide GIPIRVY(pT)HEV (15.0) (Ref. 14) were analyzed using the phosphate detection assay at pH 7.0 and 8.5. Brönsted values were obtained by linear least-squares fitting of log (k cat ) versus pK a of leaving group for the substrates using the computer program KaleidaGraph (Abelbeck Software.) Pre-steady-state Kinetics-Enzyme and substrate were rapidly mixed at pH 8.5 and 25°C in a temperature-controlled SF-61 stopped-flow spectrophotometer (Hi-Tech Scientific). The absorbance of pNP and DiFMU were monitored at 410 nm and 358 nm, respectively. The enzyme was combined with Mn 2ϩ in reaction buffer, pH 8.5, prior to rapid mixing with the substrate also diluted into reaction buffer, pH 8.5. The final concentration of Mn 2ϩ was 10 mM. The data were fitted to Equation 7 using the nonlinear least-squares fitting capability of the kinetics software (KinetAsyst, Hi-Tech Ltd, Salisbury, U.K.) where A is the amplitude of the burst, k is the first-order rate of the burst, B is the slope of the linear portion of the curve, C is the intercept of the line, and t is time.
The amplitude was converted to concentration of pNP using the ⑀ 410 of 18 mM Ϫ1 cm Ϫ1 . The correlation between the concentration of pNP burst and the final enzyme concentration was determined by linear leastsquares fitting.

RESULTS
Metal Dependence-The nature of the metal cation has been suggested to be important for regulating the activity of PP2C (18). To explore which divalent cations are effective in activating PP2C, k cat and k cat /K m values were determined for several divalent cations at saturating pNPP concentrations. The k cat constant is the first-order rate constant and reflects the ratedetermining step in the overall reaction. The k cat reveals the maximal turnover rate of substrate in saturating levels of the metal ion being analyzed. The k cat /K m is the apparent secondorder rate constant for the reaction between free enzyme and free substrate and includes the binding of substrate through the first irreversible step. The k cat /K metal parameter reveals the specificity of PP2C and suggests which metal ions are preferred for catalysis. The fact that the metal ion concentration obeys Michaelis-Menten saturation kinetics (Fig. 1) suggests that the metal ions are acting as pseudo-substrates.
Although PP2C has been defined by its Mg 2ϩ and Mn 2ϩ dependence, these two metals do not activate PP2C to the same extent. A 25-fold higher k cat value and 370-fold higher k cat /K m value were observed with Mn 2ϩ compared with Mg 2ϩ (Table I). Surprisingly, Fe 2ϩ resulted in the greatest activity of all metal ions analyzed, with a 500-fold higher k cat value and a 1,800-fold greater k cat /K metal value compared with Mg 2ϩ . A 7-fold higher k cat value and 20-fold greater k cat /K metal value were observed with Co 2ϩ , as compared with Mg 2ϩ . A k cat value was not determined with Cu 2ϩ due to precipitation at Cu 2ϩ levels below enzyme saturation; however, at low concentrations of Cu 2ϩ , the k cat /K metal was determined by calculating the slope of the initial rate versus low Cu 2ϩ concentrations. When PP2C activity was measured with Ni 2ϩ , the k cat and k cat /K metal values were 6-fold lower than those obtained with Mg 2ϩ . No activity was detected with Ca 2ϩ or Zn 2ϩ . These ions were found to competitively inhibit the PP2C Mn 2ϩ -dependent activity. By varying the Mn 2ϩ concentration at fixed levels of either Ca 2ϩ or Zn 2ϩ , a fit to Equation 2 yielded K i values of 4.45 Ϯ 0.54 mM and 12.0 Ϯ 1.8 M for Ca 2ϩ and Zn 2ϩ , respectively.
To determine the oxidation state of iron required for PP2C activity, the Fe 2ϩ detector ferrozine, was used to monitor the effective concentration of Fe 2ϩ in the phosphatase assay. PP2C activity in a parallel reaction was recorded at specific time points and related to the free Fe 2ϩ concentration (Fig. 2). A direct relationship between loss in Fe 2ϩ cation (due to oxidation to Fe 3ϩ ) and loss in activity was observed.
pH Profiles-To determine the ionizations important for catalysis and substrate binding, the k cat and k cat /K pNPP values were obtained at various pH values and saturating concentrations of either Mn 2ϩ or Mg 2ϩ . The plot of k cat versus pH obtained in Mn 2ϩ revealed one ionization that must be unprotonated for catalysis and has a pK a value of 7.00 Ϯ 0.09 (Fig. 3A). The pH-independent k cat value (C) was 2.21 Ϯ 0.65 s Ϫ1 . The k cat /K pNPP pH profile obtained in Mn 2ϩ also exhibited an ionization that must be unprotonated for catalysis, with a pK a value of 7.18 Ϯ 0.08 (Fig. 3B). The pH-independent k cat /K pNPP value was 1,280 Ϯ 314 M Ϫ1 s Ϫ1 . Due to precipitation of Mn 2ϩ at basic pH values, data could not be collected at pH values higher than 8.5. However, difficulties with precipitation were not encountered with Mg 2ϩ . This allowed us to measure the Mg 2ϩdependent rates above pH 8.5. The plot of k cat versus pH obtained with Mg 2ϩ revealed an ionization that must be unprotonated for catalysis with a pK a value of 7.42 Ϯ 0.09 (Fig.  4A)   the k cat /K pNPP pH profile revealed a second critical ionization that must be protonated for activity and has a pK a value of 8.96 Ϯ 0.12 (Fig. 4B). The pH-independent k cat /K pNPP value Steady-state Kinetic Analysis-Bisubstrate studies were performed to determine the kinetic mechanism of PP2C and to probe the order of substrate binding. The concentration of pNPP was varied at fixed concentrations of Mn 2ϩ and the double-reciprocal plot of 1/v versus 1/[pNPP] was constructed for each fixed concentration of Mn 2ϩ (Fig. 5). A series of lines that intersect to the left of the vertical axis and above the horizontal axis were obtained, suggesting a sequential mechanism. In a sequential mechanism, both substrates must bind before catalysis can occur. Thus, PP2C forms a ternary complex with metal and the phosphomonoester substrate prior to catalysis.
To determine the order of substrate binding and product release, product inhibition studies were performed. The concentration of pNPP was varied at fixed levels of P i and saturating levels of Mg 2ϩ . The double-reciprocal plot revealed a series of lines that intersect at the vertical axis, demonstrating that the inhibition is competitive (Fig. 6A). The inhibition constant K is was determined to be 0.67 Ϯ 0.05 mM. The concentration of Mg 2ϩ was varied at fixed levels of P i and saturating levels of pNPP. A series of parallel lines was obtained in the double-reciprocal plot indicating that P i behaves as an uncompetitive inhibitor with respect to Mg 2ϩ (Fig. 6B). The K ii value was determined to be 3.61 Ϯ 0.16 mM.
To determine the nature of the rate-limiting step during  PP2C catalysis, several phosphomonoester substrates with different leaving group pK a values were analyzed. A Brönsted plot of log k cat versus pK a of leaving group was obtained at pH 7.0 and at pH 8.5. At pH 7.0, a ␤ of Ϫ0.32 Ϯ 0.03 (Fig. 8) was determined, indicating that chemistry is at least partially ratelimiting. A ␤ value of 0.03 Ϯ 0.02 was calculated at pH 8.5, suggesting that chemistry does not contribute to the rate-limiting step at this pH. The similar k cat values obtained at pH 8.5 for these distinct substrates suggest that they share a common rate-limiting step, which is independent of the nature of the substrate and the released alcohol product. The rates obtained with DiFMUP are identical at pH 7.0 and 8.5, suggesting that the rate-limiting step is the same at both pH values and that chemistry does not contribute to the turnover rate with this substrate. With a leaving group pK a of 4.6, DiFMUP does not fall on the Brönsted plot at pH 7. Instead, there appears to be a break in the plot when the pK a of the leaving group is less than 7. The Brönsted analysis indicated that the turnover rate of DiFMUP will not exceed this common slow step, even when chemistry is made very fast with good leaving groups. The point at which the two Brönsted plots cross (Fig. 7) appears to represent a transition in the rate-limiting step as a function of pH and the pK a of the leaving group. Collectively, these data suggest that the common physical release of P i may limit the rate of turnover when the rate of chemistry is fast.
Pre-steady-state Kinetics-Product inhibition studies revealed that pNP is released prior to P i . Leaving group dependence studies suggested that at pH 8.5 phosphate release may be the common slow step among the various substrates. If pNP release occurs fast relative to a later step, a pre-steady-state "burst" of product should be detectable at pH 8.5. The product pNP was a poor inhibitor, making K i determination difficult; a K i value greater than 7 mM was estimated for pNP, suggesting that its release from enzyme is rapid and will not contribute to the rate-determining step. At lower pH values and with poorer leaving groups, chemistry is rate-limiting; therefore, we would predict that exponential alcohol product "bursts" would not be observed. On the other hand, if chemistry is fast relative to P i release, a burst of pNP would be expected according to this mechanism. Using a stopped-flow spectrophotometer, the rapid reaction of PP2C with substrate was followed by measuring the formation of pNP and DiFMU from the substrates pNPP and DiFMUP, respectively. These experiments revealed an exponential product burst at pH 8.5, followed by a slower linear rate (Fig. 8A). The amplitude of the observed burst was proportional to the enzyme concentration (Fig. 8B). The burst phase yielded first order rate constants between 20 -200 s Ϫ1 , while the linear phase corresponded to the steady-state initial velocity. Burst kinetics were not observed with pNPP at pH 7.0, consistent with rate-limiting chemistry.
Phosphopeptide Substrate Specificity-Several phosphorylated peptides were analyzed to explore PP2C's substrate specificity. The k cat and k cat /K m values were determined for a variety of physiologically relevant peptide substrates and are listed in Table II. The k cat values were obtained at pH 7.0 and found to range from 0.05 to 0.35 s Ϫ1 compared with 1.02 s Ϫ1 with pNPP. Given the pK a for serine/threonine residues, these k cat values are 15-100-fold faster than the Brönsted plot obtained at pH 7.0 would predict, suggesting that properties intrinsic to the peptide are important for increasing the rate of chemistry at physiological pH values. At pH 8.5, the k cat value obtained for the phosphopeptide was similar to the k cat obtained for the other substrates at this pH, indicating that dephosphorylation of the peptide is restricted by the same rate-limiting step. The k cat /K m values for the peptides approached a 90-fold greater value than that for pNPP. Thus, the specificity constants re- vealed that PP2C prefers peptide substrates over the more rapidly hydrolyzed (k cat ) artificial substrates at physiological pH values. The k cat /K m values of the peptides ranged from 3,200 to 75,000 M Ϫ1 s Ϫ1 , demonstrating that PP2C discriminates between different peptide substrates. No detectable activity was observed against the tyrosine-phosphorylated pep-tide KIGEGT(pY)GVVYK, supporting the identification of PP2C as a serine/threonine specific phosphatase).

DISCUSSION
Catalytic Mechanism of PP2C-Bisubstrate steady-state kinetics were combined with product inhibition studies to determine the kinetic mechanism of PP2C. The bisubstrate studies revealed a sequential mechanism where both metal cations and phosphorylated substrate must bind prior to catalysis. The product inhibition studies indicated the order of substrate binding and product release. Product P i is a competitive inhibitor against pNPP, suggesting that product and substrate bind the same enzyme form. Product P i is an uncompetitive inhibitor with respect to the metal ions, suggesting that P i can only bind to the metal-bound form of the enzyme. Thus, if P i and pNPP bind to the same enzyme form and P i only binds to the metal-bound form, it follows that the metal ions bind prior to pNPP and that P i is the last product to be released. In summary, our kinetic analysis of PP2C suggests an ordered sequential mechanism where PP2C binds Mg 2ϩ before pNPP. The products are released in the order pNP and then P i (Scheme 1).
The pH profiles have revealed an enzyme ionization that must be unprotonated for catalysis (pK a 7) and an additional ionization (pK a 9) that must be protonated for substrate binding (Table III). Based on the x-ray structure of PP2C (10), the identity of these important ionizable groups is somewhat ambiguous. Das et al. (10) reported the presence of 5 carboxylic acid residues (Asp-38, Asp-60, Asp-239, Asp-282, and Glu-37) involved in metal binding and an arginine (Arg-33) involved in phosphate binding. No amino acid side-chain nucleophile was apparent from the structure. Das et al. (10) proposed that a water molecule bridged between the two metals is the nucleophile that attacks the phosphoester in a single chemical step. The ionizable group that must be unprotonated with a pK a ϭ 7.0 is consistent with a metal-bound water molecule, acting as the nucleophile (10). There are several examples of metallohydrolases where the proposed nucleophile is an activated water molecule with a pK a ϳ7 (19,20). Although the x-ray structure does not suggest the identity of the protonated group with pK a value of 9, this ionization could result from a structurally important group (perhaps tyrosine, cysteine, or lysine) that must be protonated for substrate binding. It is unlikely to be a cysteine residue, as iodoacetate modification of PP2C does not affect activity (data not shown). It is worthy to note that the x-ray structure was determined at pH 5. Under these conditions, the enzyme is incorrectly protonated and therefore the structure is that of the inactive form of PP2C. It is quite possible that significant structural changes will accompany this shift from inactive enzyme at pH 5 to active enzyme above pH 7. Future mutational studies coupled to a detailed kinetic analysis should provide these answers.
The Brönsted plots indicated that increasing the pH from 7.0 to 8.5 results in a change in the rate-limiting step for substrates having pK a values greater than about 7. At pH 7.0 chemistry is rate-limiting, and at pH 8.5 a step common to all substrates becomes rate-limiting. The rate of turnover for substrates with pK a values below 7 does not significantly increase with increasing pH. For instance, DiFMUP (pK a 4.9) clearly does not fit on the Brönsted plot obtained at pH 7.0, suggesting that chemistry is not rate-limiting for this substrate at either pH value. The fact that the rate for DiFMUP hydrolysis is not accelerated by increasing pH is consistent with a mechanism where a step independent of leaving group pK a and common to all substrates is rate-limiting at pH 8.5. That same step becomes rate-limiting at pH 7.0 as the leaving group pK a decreases (i.e. chemistry is fast). Since chemistry is fast for substrates with low pK a values, the turnover rates are governed by FIG. 7. Effect of leaving group pK a value on the k cat . Substrates DiFMUP (pK a ϭ 4.9), pNPP (7.1), DiFMU (7.8), ␤-naphthyl phosphate (9.38) and phenyl phosphate (9.99), phosphoserine (14.2), and the phosphothreonine-containing peptide GIPIRVY(pT)HEV (15) were analyzed using the phosphate detection assay at pH 7.0 (circles) and 8. the same slow step at both pH 7.0 and 8.5 (k cat ϳ1.5 s Ϫ1 ). Consistent with the minimal kinetic mechanism proposed in Scheme 1, the physical release of phosphate would be a common step that is independent of the nature of the phosphomonoester and the alcohol product.
To explore the possibility that phosphate release was ratelimiting at pH 8.5, pre-steady-state kinetic experiments were performed to determine if an exponential burst of the alcohol product could be detected prior to the build-up of the steadystate rate. The Brönsted plots suggest that at pH 8.5, phosphate release may be the rate-limiting step. These data also indicated that as the leaving group pK a decreases, chemistry becomes less rate-limiting, or phosphate release becomes more rate-limiting. This suggested that product bursts could be observed at pH 8.5 with substrates that have lower leaving group pK a values, such as pNPP and DiFMUP. The detection of a pre-steady-state burst requires that the release of the chromophoric product (pNP and DiFUMB) occurs before the ratelimiting step in the reaction. The product inhibition studies suggested that phosphate release is the last kinetic step. If phosphate release is the rate-limiting step, a burst of product (pNP and DiFMU) would be detectable upon rapid mixing of enzyme and substrate. The detection of product bursts (Fig. 8) whose amplitude correlates with the enzyme concentration is in excellent agreement with this kinetic model. Although x-ray studies did not reveal a nucleophilic residue (10), we cannot rule out the possibility that PP2C forms a phosphoenzyme intermediate. Both the pre-steady-state and steady-state data are also fully consistent with this mechanism. In this model, formation of the intermediate would limit turnover at low pH, while at high pH intermediate hydrolysis would be rate-determining. The phosphoenzyme intermediate mechanism would be akin to that utilized by the protein-tyrosine phosphatases (21). The fact that we were unable to trap a phosphoenzyme intermediate with a variety of peptide substrates at pH 8.5 argues against the involvement of a sidechain nucleophile and, as we have proposed, suggests that the slow step at pH 8.5 involves the physical release of phosphate after the rapid attack of the metal bound water molecule.
Metal Dependence-The metal ions Mg 2ϩ and Mn 2ϩ were previously shown to activate putative PP2C enzymes. The results reported here reveal that PP2C is most active with Fe 2ϩ , as determined by comparing k cat and k cat /K metal values obtained with various metal cations. The metal dependence is of interest since the physiological metals ions of PP2C are not known. A comparison of k cat /K metal values suggests that Fe 2ϩ is preferred over all other metal ions analyzed. In fact, the k cat /K metal for Fe 2ϩ was determined to be 1,800-fold greater than the k cat /K metal obtained with Mg 2ϩ . This was surprising considering that PP2C was initially defined by its Mg 2ϩ dependence (22) and that there were no previous reports of PP2C Fe 2ϩ -dependent activity. It is possible that post-translational modification or effector molecule binding may alter the metal-binding properties of native PP2C␣ from those observed with recombinant enzyme.
The importance of the iron oxidation state appears to be similar to that seen with the di-iron proteins ribonucleotide reductase and purple acid phosphatase. Ribonucleotide reductase has been shown to only bind Fe 2ϩ (23). The oxidation of one of the Fe 2ϩ ions in ribonucleotide reductase is required for the activation of this enzyme. The active form of purple acid phosphatase is thought to be the mixed valence state with an Fe 3ϩ -Fe 2ϩ center. This mixed valence state can be mimicked by an Fe 3ϩ -Zn 2ϩ center (24) in oxidized purple acid phosphatase. Studies with the PPP calcineurin revealed that the active site of the native enzyme consists of one Zn 2ϩ and one Fe 3ϩ ion (25). Calcineurin can be reconstituted with iron, and the active form has been shown to consist of an Fe 3ϩ -Fe 2ϩ containing active site (26). Our experiments demonstrating the correlation between decreased activity and the oxidation of Fe 2ϩ indicate that the dinuclear site of active PP2C consists of at least one Fe 2ϩ cation. The metal-dependent activity of PP2C suggests that both the nature of the metal and its valence state will play a key role in regulating the cellular activity of PP2C.
Substrate Specificity-Although artificial substrates yielded k cat values higher than those obtained with several phosphorylated peptides (Table II), the k cat /K m values clearly demonstrated that phosphopeptides are greatly preferred over the artificial substrates examined. The k cat /K m values are more useful for analyzing the substrate specificity, since the k cat /K m constant SCHEME 1  reflects both binding and catalysis, whereas k cat does not reflect binding affinity. The k cat /K m for the Cdc2 peptide KIGEG(pT) (pY)GVVYK was 85-fold greater than the k cat /K m value for pNPP. The peptide DDENle(pT)G(pY)VATR, corresponding to the activation lip of PP2C's authentic in vivo substrate p38 (6), was also found to be a far better substrate than pNPP, with an 18-fold greater k cat /K m value. It is interesting to note that diphosphorylated peptides (Table II) appear to be much better substrates, than singly threonine-phosphorylated peptides. This is a significant observation since the proposed substrates MKK6, SEK 1, and p38 (6) are diphosphorylated on neighboring residues. Collectively, this may suggest that a large component of PP2C's substrate specificity lies within the ability to recognize and bind the diphosphorylated "active" form of these physiological substrates. By analogy, the dual-specificity protein-tyrosine phosphatases are believed to specifically recognize and dephosphorylate the diphosphorylated forms of the mitogen-activated protein kinases (13,27). With the dual-specificity protein-tyrosine phosphatases, both phosphoresidues can be hydrolyzed, although a few dual-specificity protein-tyrosine phosphatases will only catalyzed hydrolysis of phosphotyrosine in the context of the diphosphorylated species (13,27). Since dual phosphorylation is required to maintain high kinase activity, the actions of singlespecificity phosphatases are sufficient to inactivate the mitogenactivated protein kinases. The use of single-specificity phosphatases may indicate an additional level of mitogen-activated protein kinase regulation in which PP2C and a tyrosinespecific phosphatase may act in concert to fully inactivate and reset the stress signaling pathways.