Characterization of a Phosphoenzyme Intermediate in the Reaction of Phosphoglycolate Phosphatase*

When 32P-glycolate and phosphoglycolate phosphatase from spinach are mixed, 32P is incorporated into acid precipitated protein. Properties that relate the phosphorylation of the enzyme to the phosphatase are: the K,,, value for P-glycolate is similar for protein phosphorylation and substrate hydrolysis; the 32P in the phosphoenzyme is diluted by unlabeled P-glycolate or the specific alternative substrate, ethyl-P; the acti-vator C1- enhances the effectiveness of ethyl-P as a substrate and as an inhibitor of the formation of 32P-enzyme; and ”P is lost from the enzyme when 32P-glycolate is consumed. The phosphorylated protein has a molecular weight of 34,000, which is half that of the native protein and is similar in size to the labeled band that is seen on sodum dodecyl sulfate-polyacrylamide gels. The enzyme-bound phosphoryl group appears to be an acylphosphate from its pH stability, being quite stable at pH 1, less stable at pH 5, and very unstable above pH 5. The bond is readily hydrolyzed in acid molybdate and it is sensitive to cleavage by hydroxyl- amine at pH 6.8. The demonstration of enzyme phosphorylation by 32P-glycolate resolves the dilemma initial rate studies in which alternative substrates appeared have

When 32P-glycolate and phosphoglycolate phosphatase from spinach are mixed, 32P is incorporated into acid precipitated protein. Properties that relate the phosphorylation of the enzyme to the phosphatase are: the K,,, value for P-glycolate is similar for protein phosphorylation and substrate hydrolysis; the 32P in the phosphoenzyme is diluted by unlabeled P-glycolate or the specific alternative substrate, ethyl-P; the activator C1-enhances the effectiveness of ethyl-P as a substrate and as an inhibitor of the formation of 32Penzyme; and "P is lost from the enzyme when 32Pglycolate is consumed. The phosphorylated protein has a molecular weight of 34,000, which is half that of the native protein and is similar in size to the labeled band that is seen on sodum dodecyl sulfate-polyacrylamide gels.
The enzyme-bound phosphoryl group appears to be an acylphosphate from its pH stability, being quite stable at pH 1, less stable at pH 5, and very unstable above pH 5. The bond is readily hydrolyzed in acid molybdate and it i s sensitive to cleavage by hydroxylamine at pH 6.8.
The demonstration of enzyme phosphorylation by 32P-glycolate resolves the dilemma presented by initial rate studies in which alternative substrates appeared to have different mechanisms (

11002).
The fixation of carbon dioxide into organic form by ribulose bisphosphate carboxylase in the chloroplasts of leaves is always accompanied by the seemingly unproductive oxygenase reaction that forms P-glycolate. A specific phosphatase (1)(2)(3)(4)(5) prevents the accumulation of P-glycolate, which is a potent inhibitor of phosphofructokinase of spinach chloroplasts (6) and triose phosphate isomerase of pea chloroplasts and cytoplasm (7). Phosphoglycolate phosphatase also occurs in all mammalian tissues that have been examined (8)(9)(10). The occurrence of P-glycolate in mammalian cells was first shown by Rose and Salon (11) and has been confirmed by Spear and Vora (12). Data obtained with pyruvate kinase deficient red blood cells suggest that P-glycolate is synthesized by pyruvate kinase i n vivo (13). P-glycolate inhibits rabbit muscle triose phosphate isomerase (14) and it activates the breakdown of 2,3-bisphosphoglycerate (15), which is a regulator of the oxygen affinity of hemoglobin (16).
* This work was supported by United States Public Health Service Grants GM-19875 (to Z. B. R.) and CA-06927 and RR-05539 (to the Institute for Cancer Research) and by an appropriation from the Commonwealth of Pennsylvania. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The phosphoglycolate phosphatases from red blood cells and spinach have many properties in common. Both are highly specific enzymes, yet each hydrolyzes P-glycolate and ethyl-P with the same maximum velocity (5). Both enzymes are activated by millimolar concentrations of monovalent anions with either P-glycolate or ethyl-P as substrate (5, [17][18][19]. With both enzymes, initial rate studies in which Pglycolate and the activating anion are varied give parallel line double-reciprocal plots (ping-pong kinetics), whereas with ethyl-P as substrate the lines intersect. The conservation of these properties in enzymes from such diverse sources suggests that the two enzymes have a common mechanism. A mechanism that is consistent with all of the data was proposed in which phosphoryl transfer from P-glycolate does not require the activating anion but transfer from ethyl-P does so, possibly by satisfying a function performed by the carboxyl group of P-glycolate, i.e. homosteric activation (5). The kinetic pattern of C1-activation with P-glycolate as substrate implies an intermediate, a phosphorylated enzyme, the further hydrolysis of which requires C1-, again to satisfy a homosteric function. In the present study it is shown that the transfer of phosphate to the enzyme and from the enzyme to water occurs during hydrolysis of P-glycolate, consistent with covalent catalysis.

EXPERIMENTAL PROCEDURES
Materials-Phosphoglycolate was from Sigma. DEAE-Sephacel and Sephacryl S-200 were from Pharmacia Biotechnology, Inc. DE52 was from Whatman. Enzyme grade ammonium sulfate was from Schwarz/Mann. Malachite green oxalate was from Harleco. Crystalline bovine albumin was from Pentex. Coomassie Blue was from Bio-Rad. The Bio-Rad protein assay reagent was used. To concentrate protein fractions, a filtration chamber with a YM-10 membrane or Centricon Microconcentrators, both from Amicon, were used. 32Pi, carrier-free, was obtained from Amersham Corp. 32P-Glycolate was synthesized enzymatically as reported earlier (5); 99% of the labeled product is hydrolyzed by the enzyme. 3H-Deo~yglucose was from Du Pont-New England Nuclear. An Intertechnique scintillation counter was used for Cerenkov counting when possible or with Liquifluor (Du Pont-New England Nuclear) in ethanol-toluene.

Enzyme
Purification-This improved procedure produces nearly homogeneous enzyme (summarized in Table I). Spinach leaves (240 g) were deveined and washed with water. Subsequent steps were done at 0-4 "C. Portions of 100 g of spinach plus an equal weight of buffer 0.1 mM EDTA, 0.05% Triton X-100 (included for stabilization but 1 (10 mM triethanolamine-C1buffer, pH 7.2, 1 mM mercaptoethanol, possibly unnecessary)) were homogenized in a Waring Blendor with four 10-5 bursts. The homogenate was filtered through several layers of cheesecloth and then centrifuged 10 min at 12,000 X g. To the supernatant was added solid ammonium sulfate (23 g/lOO ml). After removing the precipitate by centrifugation, ammonium sulfate (19.5 g/100 ml) was added to the supernatant. The supernatant was removed by centrifugation and the precipitate was dissolved in 20 ml of buffer 2 (buffer 1 containing 1 mM MgS04). The enzyme was dialyzed overnight against two changes of buffer 2.
The enzyme (diluted to 20 mg/ml) was applied to a DEAE-Sephacel column (1.5 X 17 cm) equilibrated with buffer 2. The column was  The concentrated enzyme was applied to a column of Sephacryl S-200 (1.3 X 86 cm) equilibrated with buffer 2. The peak fractions were concentrated. After electrophoresis on polyacrylamide gels containing sodium dodecyl sulfate (20), the best fraction was found by densitometry to be 94% homogeneous with the major band at M, 31,500. The molecular weight estimated on Sephacryl S-200 is 69,000-70,000, in agreement with our earlier value (5) but not that of Christeller and Tolbert (2). Therefore, the enzyme is a dimer.
Enzyme Assays-The enzyme activity was assayed by phosphate release at 30 "C either colorimetrically with malachite green or by extraction of the 3zPi acid molybdate complex into isobutyl alcohol as reported earlier (5). For the standard colorimetric assay, the incubations were at 30 "C for 10 min in a 0.2-ml volume and contained 25 mM Hepesl-Na+ buffer, pH 7.2, 10 mM KC!, 2 mM MgSO,, 0.1 mM mercaptoethanol, 2.5 mM P-glycolate, 5 pg of serum albumin, and enzyme. The reactions were stopped by the addition of the malachite green color reagent (0.03% malachite green in 3 N HCl plus 1.5% ammonium molybdate) and the absorbance of the molybdate complex of Pi with malachite green was read at 650 nm. Under these conditions the absorbance of 1 nmol of Pi was 0.250. One unit of enzyme activity releases 1 pmol of phosphate/min under these conditions, which give 0.9 of the maximum velocity with C1-as the activating anion. Protein was determined spectrophotometrically in the cruder fractions (21) and with more purified fractions by the procedure of Bradford (22) using the Bio-Rad protein assay reagent with bovine albumin as the standard. The former method gives values that are 2.3 times higher than the latter (Table I). This discrepancy accounts for the large difference between the specific activity we reported earlier (5) and that reported by Husic and Tolbert (19).
The activity of the enzyme used to study enzyme phosphorylation was 300 units/ml of enzyme under the standard assay conditions, or a V,,, of 333 units/ml of enzyme. The specific activity was 184 units/ mg, with protein determined with the Bio-Rad assay. In the absence of C1-, the maximal rate under initial rate conditions was 25 units/ ml of enzyme, not corrected for activation by the assay components, and the K,,, for P-glycolate was 4 p~.
Phosphoenzyme Formation-The enzyme and substrate were mixed in a rapid mixing apparatus from Update Instruments Co. that allows reactions to be terminated at times of 25-300 ms after mixing.
The enzyme solutions contained 3H-deoxyglucose (2 X IO6 cpm) to estimate the fraction of the enzyme recovered after mixing. Unless otherwise indicated, the reactions were quenched with cold 10% trichloroacetic acid. After centrifuging for 1 min in an Eppendorf centrifuge, the protein precipitates were washed with 1.4 ml of cold 5% trichloroacetic acid and the precipitated protein was counted in a scintillation counter by Cerenkov radiation. The supernatants were assayed for the recovery of 3H and for 32Pi to determine extent of 32Pglycolate hydrolysis.
Stability of the Phosphoenzyme in the Presence of Hydroxylamim-The acid-precipitated phosphoprotein was suspended in 20 mM Mes-Na+ buffer, pH 5.3. Incubations were at 25 "C for 10 min and contained 0.5 M NH20H. HCl (neutralized to pH 6.8 just before use) or KC1 (control) (23). The reactions were stopped with cold trichloroacetic acid at 10% final concentration. After precipitating the protein, 32P of the precipitates and acid supernatants was determined.

RESULTS
Phosphorylation of the Enzyme-When the enzyme was mixed with 32P-glycolate and then precipitated with trichlomacetic acid, the precipitate contained much more 32P than a control with serum albumin idstead of enzyme. In order to be sure that the radioactivity was associated with P-glycolate phosphatase, the denatured phosphorylated protein was applied to a molecular sieve column of Sephacryl S-200. A single sharp peak containing 87% of the phosphoenzyme applied (12,250 cpm) was found corresponding to 34 kDa (Fig. 1). This agrees well with the subunit molecular weight of 31,500 found on sodium dodecyl sulfate gels.
Effects of Substrates on Enzyme Phosphorylation-In order to isolate the phosphorylated protein, it is necessary to use short mixing times since the hydrolytic reaction occurs even in the absence of added activating anions (see Ref. 5). In these studies the enzyme and substrate were mixed in a rapid mixing device and the reactions were usually stopped in about 30 ms. We were unable to find conditions in which the enzyme was phosphorylated but did not turn over, such as were found for alkaline phosphatase (24). With P-glycolate phosphatase, the amount of phosphorylated protein isolated after a 30-ms incubation of P-glycolate with enzyme changed very little over the pH range from 5.3 to 9.1, whereas the amount of phosphate released as Pi was 3 times greater at pH 5.3 and 7.2 than at pH 9.1. Incubations of 30-100-ms duration showed no evidence for a burst or a lag in the formation of the phosphoenzyme. As the 32P-glycolate is used up, the 32Penzyme is lost. Since the enzyme is in the steady state, the amount of protein cannot be titrated by the extent of phosphorylation.
The amount of protein phosphorylation was studied as a function of the concentration of P-glycolate in the absence of an activating anion (Table 11). There is a direct correlation between the amount of 32P-enzyme and the extent of hydrolysis. The data in Table I1 also show that nonlabeled Pglycolate dilutes the 32P compound. The apparent K, value for phosphorylation is 10 PM and that for hydrolysis is 20 MM, which is reasonable agreement in view of the occurrence of some experimental error in the timing of reactions in the rapid mixing device. This resulted in some variability in the calculated extent of hydrolysis, whereas there was very good reproducibility of the extent of protein phosphorylation in duplicate incubations and in successive experiments. The phosphorylation is a steady state value and is not affected by the incubation time, as long as the substrate concentration is essentially unchanged, whereas the calculation of the hydrolysis rate requires an accurate measurement of the reaction time. The observed K, value and hydrolysis rate are higher than the values obtained for the nonactivated reaction under initial rate conditions and are probably due to low level activation by the assay components.
In the presence of 7.5 mM C1-and increasing concentrations of P-glycolate, the maximum extent of enzyme phosphorylation is similar to that in the absence of C1- (Tables I1 and  111). Assuming full activity for the enzyme, the greatest extent of phosphorylation observed was 32% of the total protein as determined by absorbance at 260 nm (21) or 14% using the method of Bradford (22). The large activation shown with C1under Vmax conditions at low enzyme concentration (5) is not

TABLE I1
Phosphoenzyme formation: Effect of varying the P-glycolate concentration in the absence of C1-Reactions were for approximately 30 ms at 25 "C in a 40-p1 volume and contained 29 mM Hepes-Na+ buffer, pH 7.2, 2 mM MgSOI, 5 p M EDTA, 5% glycerol, 0.15 mM j3-mercaptoethanol, 0.15 mg of bovine serum albumin, and 1.2 units of enzyme from a stock of 300 units/ml determined in the standard assay (183 pmol based on the Bio-Rad protein assay). 32P-Glycolate (83-3962 cpm/pmol) was varied, the specific activity decreasing with increasing substrate concentration. Rates are calculated assuming a 30-ms incubation period. v is expressed in micromoles of Pi formed/min/ml of enzyme. See "Experimental Procedures" for details.  seen in these experiments with lo4 greater enzyme concentration. It is not clear whether this difference is attributable to different states of aggregation of the phosphatase or to a hysteretic phenomenon that requires more than 10 turnovers of the enzyme to reach a C1"dependent steady-state rate.
Ethyl-P, an alternative substrate, has been shown to inhibit P-glycolate hydrolysis competitively in the presence of C1-( 5 ) . Ethyl-P decreases both the amount of 32P-phosphoenzyme observed and the rate of hydrolysis of 32P-glycolate (Table   IV). At a given ratio of [ethyl-P]/[P-glycolate], ethyl-P is much more effective as an inhibitor in the presence of C1-. This is consistent with the results from steady state kinetics which indicate that C1-has a role in the reaction prior to the phosphoryl transfer step when ethyl-P (but not P-glycolate) is the substrate (5). The results show that this C1-effect is evident at high enzyme concentration as well.
Chemical Characterization of the Enzyme-Phosphoryl Bond (Fig. 2)"Acid-denatured enzyme loses only 5% of its 32P after 30 min at 46 "C at pH 1. At pH 5 the hydrolysis is 27% after 30 min at 46 "C, 2% after 10 min at 25 "C, and undetectable for at least 2 hours at 0 "C. The bond is very unstable above pH 7. At pH 10.6 it is completely hydrolyzed in 10 min at 25 "C. Hokin et al. (25) observed a similar partial instability in mild acid and great sensitivity to hydrolysis at alkaline pH for the acylphosphate intermediate of the Na+-K'-ATPase. The observed properties rule out phosphoserine or phosphothreonine (26), phosphohistidine (27), phospholysine (28), S- This lability is characteristic of N-P (33) and acyl-P bonds (34). The bond is also readily cleaved by hydroxylamine. When the phosphoenzyme was incubated for 10 min at 25 "C in 0.5 M hydroxylamine at pH 6.8, the phosphoryl bond was 95% hydrolyzed compared to a control containing 0.5 M KCl. Under these conditions the nucleophilic attack by hydroxylamine is fairly specific for carboxylic acid anhydrides (23).

DISCUSSION
The demonstration that 32P-enzyme can be isolated after mixing 32P-glycolate with P-glycolate phosphatase suggests that the phosphoenzyme is an intermediate in the reaction. Evidence consistent with a phosphoenzyme mechanism was reported by Christeller and Tolbert (4), who observed transphosphorylation with high concentrations of ethylene glycol, ethanol and n-propyl alcohol as well as nonphysiological levels of glycolate, glyceraldehyde, and glucose. They also observed weak inhibition by diisopropyl fluorophosphate, similar to that found with alkaline (35,36) and acid phosphatases (37). Since alkaline phosphatases are serine phosphates (38,39) and many acid phosphatases are histidine phosphates (40-42); this does not allow a conclusion as to the residue that is likely to be phosphorylated in P-glycolate phosphatase. Covalent intermediates have also been demonstrated in many other phosphatase reactions. 5"Nucleotide phosphodiesterase (43) is phosphorylated on a serine residue. Glucose-6-phosphatase (44) is phosphorylated on a histidine residue. The Na'-K'-ATPase (25), the Ca2+-Mg2'-ATPase of sarcoplasmic reticulum (45), and two plant nucleoside phosphotransferases (46-49) have acylphosphate intermediates. Fructose bisphosphatase (50) and inorganic pyrophosphatase (51) are not known to be phosphorylated by their substrates.
The mechanism in Scheme 1 is consistent with the present observations and the different kinetic patterns observed in anion activated initial rate studies with P-glycolate and ethyl-P (5) and is similar to that proposed earlier ( 5 ) . The maximum extent of phosphorylation that was observed was about 25% of the enzyme. This is consistent with a steady state rate of hydrolysis that is four times faster than the rate of formation of the phosphoenzyme.
When P-glycolate is the substrate, C1-and other activators function only after enzyme phosphorylation has occurred. With ethyl-P as substrate, C1-assists in achieving a favorable conformation for the phosphoryl transfer to the enzyme, possibly by binding to the carboxyl region of the active site. It is proposed that C1-also assists the phosphoryl transfer from the enzyme to water by binding to the carboxyl site for the substrate. Since these processes occur at the active site, we designate them as "homosteric," in contrast to allosteric. Similar effects also appear to explain the activation of the 2,3-bisphosphoglycerate phosphatase activity of 2,3-bisphosphoglycerate synthase-phosphatase by specific combinations of anions or by P-glycolate (15,52). In both of these enzymes, anions activate the reaction by substituting for portions of the substrate. Such effects may be relatively common and may be a factor in the control of reaction rates in uiuo.