The Phosphorylation of Diphosphoglycerate Mutase*

Diphosphoglycerate mutase, purified to apparent homogeneity from human erythrocytes, was found to have a molecular weight of 60,000 (gel filtration) and subunit weight of 32,000 (electrophoresis on polyacrylamide gels containing sodium dodecyl sulfate). One [32P]phosphoryl group is covalently bound per subunit upon incubation of enzyme with either the 32P-labeled substrate, 1,3-diphosphoglycerate, or the product, 2,3-diphosphoglycerate. The phosphate group is transferred to either 2or d-phosphoglycerate or to water if glycolate-2-P is added to phosphorylated enzyme. The phosphoryl group was stable at alkaline pH but was liberated from the denatured phosphoprotein in the acid range at rates consistent with a phosphoramidate linkage to histidine. Since a similar phosphorylation reaction had been shown previously with monophosphoglycerate mutase, it was necessary to achieve complete separation in the preparation of the enzyme.


SUMMARY
Diphosphoglycerate mutase, purified to apparent homogeneity from human erythrocytes, was found to have a molecular weight of 60,000 (gel filtration) and subunit weight of 32,000 (electrophoresis on polyacrylamide gels containing sodium dodecyl sulfate). One [32P]phosphoryl group is covalently bound per subunit upon incubation of enzyme with either the 32P-labeled substrate, 1,3-diphosphoglycerate, or the product, 2,3-diphosphoglycerate.
The phosphate group is transferred to either 2-or d-phosphoglycerate or to water if glycolate-2-P is added to phosphorylated enzyme. The phosphoryl group was stable at alkaline pH but was liberated from the denatured phosphoprotein in the acid range at rates consistent with a phosphoramidate linkage to histidine. Since a similar phosphorylation reaction had been shown previously with monophosphoglycerate mutase, it was necessary to achieve complete separation in the preparation of the enzyme.
Diphosphoglycerate mutase catalyzes the intermolecular transfer of the acyl phosphate of 1,3-DPGI to an acceptor molecule of monophosphoglycerate with the formation of 2,3-DPG. Either 2-PGA or 3-PGA can act as the phosphoryl acceptor (Equation 1) with higher rates observed in the presence of 2-PGA (1). Kinetic studies failed to show the participation of a phos-phor\-I enzyme intermediate (1).
1,3-DPG + 3-PG.4 (or 2-PGA) + 2,3-DPG + 3-PGA Intersecting lines were obtained in double reciprocal plots of either substrate as a function of the second, constant substrate. In addition, attempts to show the equilibration shown in Equation 2 were unsuccessful.
Although these results indicate that a ternary complex of enzyme, phosphoryl donor, and phosphoryl acceptor is an inter-* This investigation was supported by Public Health Service Research Grant CA-07819 from the National Cancer Institute, by United States Public Health Service Grants CA-06927 and RR-05539 to this Institute, and by an appropriation from the Commonwealth of Pennsylvania.
In the present study the possible participation of phosphorglated enzyme has been reinvest,igated by the method of direct isolation used previously with the monophosphoglycerate mutases of muscle (2) and yeast (3). The cofactor, 2,3-DPG, phosphorylates these enzymes forming an acid-labile linkage identified as 3-phosphohistidine (2,3). It has also been shown that 1,3-DPG, the substrate of diphosphoglycerate mutase, can phosphorylate monophosphoglycerate mutase (2,4). Therefore, to investigate the possible phosphorylation of diphosphoglycerate mutase, it was necessary to obtain a preparation free of monophosphoglycerate mutase. The best preparation obtained previously contained monophosphoglycerate mutase activity (1). In the present study an apparently homogeneous preparation of diphosphoglycerate mutase has been obtained, its phosphorylation shown, and the role and nature of the phosphoryl group examined.
The monophosphoglycerate mutase activity remaining in the diphosphoglycerate mutase appears to be an intrinsic property of the enzyme. was extracted with 2 ml of the same buffer mixture, saturated with phenol, followed by two 4-ml washes with the same buffer. When the distribution of phosphate between inorganic and organic phosphate compounds was determined, the first aqueous extracts were treated with acid molybdate and extracted with isobutyl alcohol-benzene (1: 1) (9). Diphosphoglycerate mutase was assayed spectrophotometritally.
The observed rates have been multiplied by 3 in order to obtain a true rate of formation of products (1). Monophosphoglycerate mutase was determined according to the method of Cowgill and Pizer (10) by coupling to enolase and measuring the increase in absorbance at 240 nm due to the conversion of the 2-PGA formed to P-enolpyruvate.
A change of 0.975 per min indicated the formation of 1 pmole of 2-PGA per ml of incubation.
One unit of enzyme activity allows the conversion of 1 pmole of substrates to products per min at 25' under the conditions of the assay.

E?zzyme Purrfication-For
the phosphorylation studies it was necessary to obtain enzyme free of monophosphoglycerate mutase. The purification procedure used was essentially that reported previously (1). Although there was separation of the two enzymes upon fractionation with a7nmonium sulfate as well as upon chromatography on hydrosylapatite, the best preparation obtained at that time contained 1 $ZO monophosphoglycerate mutase activity.
In the present study, in spite of efforts to improve the separations obtained at each step in the preparation, the diphosphoglycerate mutase obtained after fractionation through the hydroxylapatite step had 3 7. monophosphoglycerate mutase activity.
Second Nydroxylapatite Column-In order to determine whether the monophosphoglycerate mutase activity in the diphosphoglycerate mutase was due to trailing of the former peak ( Fig. la), the diphosphoglycerate mutase was rechromatographed on hydroxylapatite.
The enzyme was eluted with a 150-ml linear gradient in which potassium phosphate, pH 7.2, containing 0.5 InM dithiothreitol, increased from 5 to 100 mM.
The flow rate of the column was 6.7 ml per hour and 2-ml fractions were collected.
There was no monophosphoglycerate mutase detectable in the fractions preceding t.he diphosphoglyccrate mutase peak, which eluted with 41 to 67 mM phosphate (Fig. IB).
As little monophosphoglycerate as 0.002 unit per ml would have been detected.
The recovery of diphosphoglycerate mutase was 16 units or 847,. Monophosphoglycerate mutase activity was found in the fractions containing diphosphoglycerate mutase with the peak activity for both enzymes in the same fractions.
The monophosphoglycerate mutase activity was 3 $& of the diphosphoglycerate mutase level, as in t.he starting material for this step.
Isoelectric Focusing-The column contained a 1 7. ampholyte solution, pH range 4 to 6, in a sucrose gradient; the anode was the lower electrode. A cooling bath at 7" was used. The pH gradient was allowed to form overnight in order to remove any metal ions and to minimize the time the enzyme would be on the column.
The concentrated sample of enzyme from the second hydroxylapatite column (1.5 ml) was adjusted to contain 10% sucrose and applied about one-fourth of the distance from the top of the column with a Pasteur pipette extended with a piece of polyethylene tubing, 1.2 mm in dinmetcr. The current flowed again for 339 hours by which time it had reached a constant value of 1 ma. Fractions of 2 ml were collected.
The peak of diphosphoglycerate mutase activity was found in fractions of pH 5.05 to 5.40 (average value pH 5.23). Since we found in separate esperiments that red cell monophosphoglycerate mutase has an isoelectric point, p1, of 6.10 to 6.40 (average value 6.25), if traces of that enzyme had been present initially, they would be removed by this procedure.
The active fractions (10 ml) were combined and concentrated with Carbowas to 1.1 ml. In order to stabilize the activity of the still dilute protein solution, bovine serum albumin to 1 mg per ml and EDTA to 1 InM were added.
The enzyme was stored in pellets in liquid nitrogen.
Of 5 units of diphosphoglycerate mutase that had been applied to the column, 3 units (60%) were recovered.
This material was used for the initial phosphorylation studies. Monophosphoglycerate mutase activity was still detectable at 1 to 370 of the diphosphomutase level.
The void volume was 72 ml as determined by the elution of blue dextran.
The column was calibrated from the peaks of the elution volumes of the components of a mist,ure of proteins of known molecular weight: cytochrome c (12,400) ; ovalbumin (44,000) ; phosphoglycerate mutase (rabbit muscle) (57,000) ; alcohol dehydrogenase (yeast) (151,000). After reequilibration of the column, a dialyzed red cell ammonium sulfate fraction containing diphosphoglycerate mutase (15 units) and monophosphoglycerate mutase (2 units), cytochrome c, and yeast alcohol dehydrogenase was applied to the column. Both the red cell monophosphoglycerate mutase and diphosphoglyceratc mutase had an elution volume corresponding to a molecular weight of 60,000.
Enzyme that had been purified through the isoelectric focusing step was analyzed by polyacrylamide gel clectrophorexis in sodium dodecyl sulfate (13). Fig. 2 shows that it appears to contain a single component in addition to the bovine serum albumin which had been added to stabilize the activit,y.
The enzyme band corresponds to a molecular weight of 32,000. It appears from this that the diphosphoglyccratc mutase molecule is a dimer.
Phosphorylation of Diphosphoglycerate Mutase-When the substrate 1,3-[lJ*P]DPG is incubated with diphosphoglycerate mutase, radioactivity is found associated with the enzyme in covalent linkage (Fig. 3). Phosphorylation of the enzyme also occurs when the product, 2,3-[U-3ZP]DPG reacts with the enzyme. Using enzyme purified through the isoelectric focusing step with 0.035 unit of diphosphoglycerate mutase and 1.12 x lop3 unit of monophosphoglycerate mutase activity, and with 2,3-[U3*P]DPG as the phosphorylating agent, 0.06 nmole of 32P is found associated w-ith the enzyme.
The data shown in Table I  When phenol and aqueous phases are separated, the indicating that the 1,3-DPG has been converted to the product, protein remains in the phenol phase and small molecules are 2,3-DPG.
A similar incubation, designated as having had no found in the aqueous phase. Radioactivity in the phenol phase additions to indicate that no phosphoryl acceptor has been added, indicates the presence of 32P covalently bound to the protein.
was stopped with phenol instead of acid, and the phenol phase In Table I, Experiment A, the zero time control, which lacks extracted with aqueous buffer. The radioactivity (2054 cpm) enzyme, illustrates the lability of the acyl phosphate of 1 ,3-DPG in the phenol phase indicates the formation of 0.027 nmole of in the presence of the acid molybdate used in the procedure for phosphoenzpme based on the specific activity of the acyl phosthe differentiation between organic and inorganic phosphates phate of 1,3-DPG. The radioactivity in 37Pi indicates that the since, in this case, most of the V analyzes as I'i. As shown for sample of 1,3-DPG contained 8% Pi. When one of the normal the total reaction sample, when acid is added to the enzymatic phosphoryl acceptors, 3-PGA or 2-PGA, is added to an incubasystem after 2 min, most of the 3?P is found in acid-stable form tion containing phosphorplated enzyme, the 32P is transferred 1516 FIG.
from the enzyme to form 2,3-DPG as shown by the fact that now no 32P is found associated with the protein in the phenol phase and no 32Pi is generated.
When glycolate-2-P, a structural analogue of the normal acceptors, is added to phosphorylated enzyme, the 32P is also released from the enzyme, as shown by the loss of radioactivity from the phenol phase. In this case half of the radioactivity is found as Pi. Thus, glycolate-2-P induces the enzyme to behave as a phosphatase, releasing half of the phosphoryl groups of 2,3-DPG as Pi.
Reactions were stopped by the addition of phenol and phosphorylation of the enzyme was determined as under "Methods." Incubations were at 25" in a 0.2-ml volume and contained: glycylglycine-sodium buffer, pH 7.5 (2 pmoles) and 0.035 unit of diphosphoglycerate mutase purified through the isoelectric focusing step. In Experiment A, 1,3-[1-32P]DPG was present (0.4 nmole; 30,000 cpm). In Experiment B, 2,3-[U-82P]DPG was present (0.4 nmole; total counts 70,000 cpm). There was no enzyme in the zero time sample. The total reaction sample was stopped with acid after 2 min at 25". The remaining incubations were for 2 min at 25" followed by the noted additions.
After an additional 2-min period, the reactions were stopped with phenol.
The first wash from the phenol extraction was analyzed for the distribution of phosphate into organic and inorganic forms (see "Methods"). The total number of counts in each form is reported. No phosphatase activity is observed under these conditions.
In Experiment A of Table I it was shown that before the addition of the phosphoryl acceptors, the 1,3-DPG had been completely converted to 2,3-DI'G. Therefore the results ob-ser\-cd must br, :ultl indeed arc, the same no matter which diphosphate was used initially.
The apparent lower stoirlliometr: obeervcd when starting with I ,3-[ 1 m32P]DI)G is consistent with randomization of the radioactivity into the two phosphoryl group% of 2,3-Dl'G.
It is also in agreement with the observation that either 2-PGA or 3-PGh can act as phosl~l~oryl acceptor in the diphosphoglycerate mutase reaction (1).
Subunit Structure 01 Diphosphoglycerate Jlutase--After isoelec*tric focusing, the enzyme preparation appeared to be homogeneous on polyacrylnmide gel clectrophoresis in sodium dodecyl sulfate except for a band attributed to added bovine serum albumin (Fig. 2). In order to substantiate that the protein in the 32,000 molecular weight band is indeed the one that is phosphorylated by diphosphoglycerate, the migration of 32P-enzyme, phosphorylated with 2, 3-[IJ-32P]DPG, was determined under the same conditions.
[%]Ketodeosyphosphogluconate aldolase, with a subunit weight of 24,000 (14)) was included as an internal standard in both an unstained gel containing [321']diphosphoglyccratc mutnse and a stained gel containing diphosphoglycerate mutase and ribonuclease.
The 32P is lost from diphosphoglycerate mutase on destaining, making it necessary to compare two gels. It was found that relative t.o the 14C standard, the 32P ran as would a molecule of 31,200 and the stained band ran as a molecule of 31,700. From the specific activity of the 32P, if one phosphoryl group combined per subunit, the gel in Fig. 2  Hydrolysis was rapid in 0.1 N HCl with a t1/2 of about 3 min. At pH 7.5 and 11.2 1517 hydrolysis was not perceptible in 60 min. Hydrolysis at pH 3 was more rapid than at pH 4. This pattern of acid lability, increasing stability with increasing pH in the acid range, and stability at neutrality or in alkaline solution is consistent with the phosphorylation of diphosplioglycerate mutase having occurred on a histidyl residue.
This monophosphoglycerate mutase activity may be an intrinsic property of the diphosphoglgccrate mut'ase molecule.
There is evidence for kinetic differences bet.neen this activity and that shown by monophosphoglyccrate mutase of red cells. As shown in Fig. 4, the red cell monophosphoglycerate mutase is not stimulated by the addition of 2,3-DPG to the standard assay systtm which contains 30 nmoles of 2,3-DPG.
However the monophosphoglycerate mutase activity of diphosphoglycerate mutase purified through the hydrosylapatite step is enhanced by the addition of a high level of 2,3-DPG.
,I 4fold increase in the 2,3-DPG from the 0.03 111~ of the st~andard assay to 0.13 mM gave a N-fold increase in the reaction rate; with 1.03 mM 2,3-DPG there was an additional increase. cnder these conditions the monophosphoglycerate mutase act'ivity was about 15y0 of the diphosphomutase.
The reactions were stopped by the addition of H2S04 and the Pi measured (see "Methods"). ma1 activity in stimulating the release of Pi when it is present at a level up to 400 PM; at 2 rnnf and above the rate of release of the phosphoryl group was inhibited.

DISCUSSION
Previous initial rate studies (I) indicated the participation of a ternary complex of enzyme. 1,3-DPG. PGA in the diphosphoglycerate mutase reaction.
The present study has shown phosphorylation of the enzyme and the capability of the phosphoryl group to be transferred to appropriate acceptors.
In the kinetic studies, the failure to observe the parallel lines that have usually been considered indicative of a covalent enzyme-substrate intermediate (15), suggests only that no product is released before the addition of both substrates to the enzyme and is not inconsistent with phosphorylation of the enzyme in a complex containing the elements of both subst'rates.
The following sequence of reactions is proposed (the subscripts d and a refer to molecules occupying donor and acceptor sites on the enzyme).
The phosphorylation of the enzyme, Equation 7, is considered to be the step that makes t#he over-all reaction essentially irreversible. Whether the transfer of the phosphoryl group to the enzyme occurs before or after the acceptor molecule of PGA combines with the enzyme has not been established.
Although PGA was not added as such in the phosphorylation studies, 1,3-DPG solutions always contain 3.PGA (1). The 2,3-DPG used for phosphorylation could have contained a trace of PGA or the 3-PGA donor site may have some affinity for 2,3-DPG so that a 2nd molecule of 2,3-DPG could fill the 3-PGA requirement. There is precedent for this with monophosphoglycerate mutase for which 2,3-DPG is a competitive inhibitor of PGA (16,17). It is also possible that occupation of the donor site is important kinetically but not required for the partial reaction, phosphorylation (e.g. 18). Alternatively the reaction sequence could have been written with phosphoryl transfer occurring before addition of the acceptor PGA molecule and without loss of the donor 3-PGL4. It should be emphasized that either alternative provides a mechanism consistent with the intersecting pattern of kinetics and the observation of a phosphoryl enzyme.
A pattern of intersecting lines was also observed in initial rate studies of succinyl coenzyme A synthetase (19) for which evidence of a phosphoryl enzyme intermediate also exists (20). The kinetics of phosphoryl transfer from enzyme to substrate have yet to be investigated to determine whether the phosphorylated intermediate is catalytically important.
The activation by 2-PGA and glycolate-2-P of the phosphorylation of 3-PG,4 which was shown in the previous steady state studies (1) can be understood readily in terms of a mechanism involving a ternary complex.
It had been observed that 2-PGA lowered the K, of acceptor 3-PGA and increased the maximal velocity of the reaction.
The activation can be visualized as resulting from the dissociation of the 3-PGA molecule from the donor site of the ternary complex to form an inactive complex. 2-PGA may add to that site to form a more reactive complex than the original one as shown.
Enzyme-P .3-PGA, + 2-PGA ti enzyme-P. 3-PGA,.  The striking parallels between the observations of the partial reactions of diphosphoglycerate mutase and monophosphoglycerate mutase suggest that the two enzymes may have similar reaction paths.
Phosphorylation of muscle monophosphoglycerate mutase by 2,3-DPG or 1,3-DPG has been shown (2,4). Yeast monophosphoglycerate mutase was phosphorylated by 2,3-DPG (3). Other properties that are similar for the three enzymes are the properties of the phosphoryl transfer reactions, the chemistry of the phosphorylated group on the enzyme, and subunit size. Kinetic data in the literature are conflicting concerning the reaction sequences of the monophosphoglycerate mutases.
Studies of the muscle enzyme that favored a mechanism including a covalent enzyme intermediate but no ternary complex (16) were considered inconclusive when evaluated by others (21). Kinetic studies of the yeast enzyme suggested that the reaction path was sequential and the data were not consistent with the participation of phosphorylated enzyme (17). The 2,3-DPG phosphatase activity of both enzymes is activated by glycolate-2-l' (2,3,22). Studies with yeast monophosphoglycerate mutase strongly indicated the formation of a ternary complex as an intermediate in the glycolate-2-P activation of the 2,3-DPG phosphatase activity (22). In view of the many similarities observed (recently reviewed by Kay and Peck (23)), it appears likely that yeast and muscle monophosphoglycerate mutases have a common mechanism which may parallel that of diphosphoglycerate mutase in involving both a ternary complex and phosphorylation of the enzyme as indicated.
Enzyme + 2,3-DPG = enzyme-PaPGAd Enzyme-P*PGBd In view of the failure to obtain ready mixing between isotope in 2,3-DPG and PGA in the presence of muscle monophosphoglycerate mutase (IS), it appears that the phosphorylated enzyme .PGh complex generated according to Equation 16 will usually reform ternary complex instead of hydrolyzing to free enzyme and 2,3-DPG in a reversal of Reaction 13.
The conclusions reached in this study concerning the reactions of diphosphoglycerate mutase are only valid if the enzyme used was free of monophosphoglycerate mutase.
If the monophosphoglyceratc mutase activity present in the enzyme used to show the phosphorylation of diphosphoglycerate mutase ( Fig. 3 and Table I) were due to contamination by monophosphoglycrrate mutase with specific activity similar to that of t,he muscle enzyme, one can calculate the contribution it would make to the total observed phosphorylation, assuming one site of phosphorylation per subunit.
Crystalline muscle monophosphoglycerate has a specific activity of 1010 units per mg measured under the conditions of our assay. For a subunit weight of 28,500, there are 28.8 units per nmole of subunits.
In the experiments of Fig. 3 and Table I, there was 1.12 X 1OV unit of phosphoglycerate mutase per incubation or 3.9 X lo-" nmole of subunits. This would make a contribution of only O.O7Gj, to the 0.06 nmole of phosphoprotein observed. The specific activity of purified enzyme can be calculated from the number of units of activity of diphosphoglycerate mutase that combine with a micromole of phosphorylated enzyme, assuming one active site per subunit.
With the purest enzyme obtained, the value calculated was about 20 units per mg.

1519
These studies indicate that mono-and diphosphoglycerate mut.ases have striking similarities which suggest evolution from a common molecule.
Future studies will consider the relationship of 2,3-DPG phosphatase to the other enzymes of 2,3-DPG metabolism.