Yeast Inorganic Pyrophosphatase

SUMMARY The kinetics of the Mg2+ activation of crystalline yeast inorganic pyrophosphatase have been thoroughly investi-gated from pH 7.40 to pH 9.05 using a sensitive isotope assay in order to determine the role of the divalent metal ion acti-vator in the reaction. A computer program has been de-vised for calculating the concentration of each of the various components of the complex equilibrium involving Mg2+ and inorganic pyrophosphate (PPi). The reaction rate was measured over a wide range of Mg2+ and PPi concentrations, and the concentration dependence of the measured rate was kinetically analyzed by a computerized algorithm for nonlinear regression. The computer analysis included the test-ing of several plausible kinetic models for goodness of fit to the data, and determination of best values of kinetic parameters for the various models. The simplest kinetic model which provides a good fit to all of the data involves binding of free Mg2+ by the enzyme followed by binding of PPi ligands. Both MgPPi Mg,PPi substrates; pH the the PPi for the

two of its C-substituted derivatives, and the Co and C4 members of the a,w-glycol diphosphate series are not inhibitory under the conditions employed, which are near optimal for PPi hydrolysis.
Specificity for inhibitors is not markedly broader than specificity for substrate, and, like substrate specificity, is maximal in the presence of Mg*f as activating ion.
No evidence could be obtained for phosphoryl transfer to any acceptor other than water.
Using 32PPi in a variety of conditions, several types of assays failed to detect formation of a phosphorylated enzyme intermediate in the reaction. The results are interpreted as favoring a concerted mechanism for this enzyme.
The inorganic pyrophosphatase reaction (EC 3.6.1.1) provides a convenient system for studyin g enzymatic utilization of the simplest possible high energy polyphosphate.
The system is complicated, however, by the large number of equilibrium species resulting from association in solution of PPi, the required metal ion activator (l), and hydrogen ions. Interactions of crystalline )-cast pgrophosphntase with individual components of these equilibria have been analyzed in terms of equilibrium binding (2) and kinetics (3,4).
In order to obtain information about possible transition states in the reaction and about admissible variations in the chemical nature of the substrate, we have examined several structura analogs of PPi as substrates or inhibitors of the reaction. The enzyme is known to hydrolyze, at rates slower than those observed with PPi, a variety of monosuhstituted pyrophosphates in the presence of Znz+, MnQ+, and Co2+ as activating ions, but is quite specific for PPi in the presence of Mg2+ (5-S), which gives maximum activity against PPi (I).
The analogs tested here differ in the nature of the linkage between the two phosphoryl groups.
The oxygen linkage of Pl'i is replaced by nitrogen in imidodiphosphate, by carbon in the methylene diphosphonates, and by a glycol in t'he o(, w-glycol diphosphatcs. These compounds were tested in the presence of a large excess of Mg2+ in order to be sure that the effect's observed were not due to competition betweeu PPi, inhibitor, and enzyme for Mg2+, which is required to activate the enzyme (3,4). No compound other than PPi could be shown to serve as substrate for the Mg2+-activated reacbion, and with I\&"+ as the activating ion the structural requirements for inhibitor are also rather stringent. 'V\'e have considered the possibility, based on the isotope exchange studies of Cohn (9) and the occurrence of such an intermediate in other phohphntasc and phosphoryl transfer enzymes, that the mechanism of this enzyme includes a covalent phosphoryl enzyme intermediate. We have employed several direct and indirect techniques which have been successful in other systems in an attempt to detect a phosphorylated rnzyme in this reaction.
Our results provide no evidence for the postulat'cd phosphoryl enzyme intermediate, and are more consistent with a conccrtcd reaction mechanism.

MATERIALS An'D METHODS
Reagents-Sodium imidodiphosphate [ (NaZOaP)&H~ 10HzO] was synthesized (10) from diphenylimidodiphosphoric acid produccd from diphenyl phosphoramidate (1 I), which was prepared by the method of Stokes (12). Based on paper chromatography and analysis for total phosphate, the imidodiphosphate preparation contained no more than 57, phosphate impurities.
The velocitv was normalized to standard conditions (4). The average deviation from the mean is displayed as error bars.
The source or preparation of all other materials including 321'I'i and crystalline enzyme has been described (2).
Jfethods-The measurement of I'l'i hydrolysis using both a radioisotope assay (4) and a spectrophotometric assay (2) has previously been detailed.

Inhibitor
Xpec$cLcity-cnder conditions nearoptimal for Pl'i hydrolysis (2 mM metal ion, 2 inM PPi or imidophosphate, 0.1 M Tris, pH 7.2) imidodiphosphate is not appreciably hydrolyzed (rate <O.l% of that observed with PI';) by yeast inorganic pyrophosphatase in the presence of either 11g2+ or ZnZf. However, imidodiphosphate appears to be a competitive inhibitor of 1'Pi hydrolysis catalyzed by this enzyme (Fig. I) These compounds were much less inhibitory than imidodiphosphate; I, 3-propanc,diol diphosphate and 1,4butanediol diphosphate showed no iirhibition at all at concentrations 20-fold higher than that, of l'l'i.
Although the data were neither as reproducible nor as consistent as was the rase with imidodiphosphate, apparent K1 values of approximately 2 MM and 11 mM were obtained for methanediol diphosphate and 1,2-ethanediol diphosphatc, respectively. A variety of conditions was used to evaluate the effect of mcthylene diphosphonates on PPi hydrolysis (the stable P-C bonds preclude their hydrolysis by the enzyme).
Inhibition was observed only when these compounds were present in excess ovrr both PPi and Mg2'.
With 5.0 mM Mg2+, 0.5 mM PPi, and 0.1 to 1.5 m&f analog, no inhibition was observed with the substituted compounds.
Our results indicate that methylene diphosphonate and some of its C-substituted derivatives arc not inhibitory under these conditions.

Ph,osphoryl
Transfer-In preliminary experiments utilizing a variety of alcohols as potential phosyhoryl group acceptors, no evidence was obtained for phosphoryl transfer to any acceptor other than water.
In order t,o increase the likelihood of observing phosphory-1 transfer, ADP was employed as potential phosphor)-1 acceptor.
A binding site which can accommodate ADP is present on the enzyme, since it is a substrate of the Zn"+activated reaction (5), so AD1 could effectively compete with water for the phosphoryl group bring transferred.
All transfer experiments were conducted at pH 6 to 6.5, which gives nearmaximal rates of nucleotide hydrolysis, but relatively slow 1'Pi hydrolysis (5). &if was used as the metal cofactor because it produces the highest rate of nucleot,ide hydrolysis (5, 6). Equimolar concentrations of PPi and ADP were used SO that there would be a reasonable rate of PP; utilization, yet at the same time a good opportunity for ADP to interact with the enzyme.
Three different techniques were used in the attempt to demonstrate phosphor?1 transfer from PPi to ADP to form ATP: (CI) a spectrophot'ometric assay coupled through hexokinase to the reduction of X-%Dl' by glucose 6-phosphate dehydrogenase; (6) absorption onto charcoal of [32P],4TP formed from 3"PPi; and (c) formation of the acid-stable compound, [32P]glucose-6-P from 32PPi, which is acid-labile, in the hexokinnse-coupled assay. Addition of known amounts of ATP, and other controls, demonstrated that phosphoryl transfer to ADP could have been detected with these techniques at a level of 5'%, O.l%, and 0.003%, respectively, of that of the observed PPi utilization. However, significant amounts of phosphoryl transfer were not observed under any conditions with any of these techniques.
Phosphoryl Bnzynle-We have used a variety of conditions and techniques to maximize the opportunity for trapping the enzyme in its postulated phosphorylated form including incubation at pH 5.25 to 8.25 in the presence of a variety of met.al ions, and the use of EDTA, trichloroacetic acid, and phenol to stop the reaction.
3'PPi with high specific activity (0.5 Ci per mole) was used as substrate to increase the sensitivity of the detection, but. hpecial care was necessary to prevent complete hydrolysis of the PPi before the reaction could be stopped.
In some cases the reaction was slowed by carrying out the incubation with met'al ions other than Mg2+, or by incubation at temperatures as low as -20" (in the presence of 25 to 40 volume W glycerol to prevent freezing).
Alternat,ively, the reaction was carried out in a rapid mixing device which allowed a reaction time of approximately 10 ms. EDTA-stopped samples were analyzed for 32P incorporation by gel filtration to remove 32PPi and 32P;, trichloroacetic acid-stopped samples by extensive washing of the precipit'ated protein after adding carrier protein, and phenol-stopped samples by extraction into phenol followed by extensive washing with aqueous buffers containing carrier Pi and PPi (14).
Since the enzyme is a dimer, up to 2.0 moles of 32P could be incorporated per mole of enzyme. The phenol assay is especially sensitive; it is capable of detecting incorporation of 0.002 moles of 32P per mole of enzyme, three orders of magnitude lower than the above maximum.
However, in no case was significant 32P incorporation observed under any of the conditions utilized.

DISCUSSION
These studies provide additional evidence for the remarkably high degree of substrate specificity of yeast pyrophosphatase n-hen activated by Rig"+. Imidodiphosphate, which resembles PPi in structure (Is), in chemical stability (16), and in susceptibility to hydrolysis by certain phosphatases,' is not a substrate even in the presence of Zn2+, which broadens the specificity (5, 6). PPi remains t'he only known substrate of the Mg2f-activated enzyme.
The enzyme is also highly specific with respect to inhibitors. In addition to the compounds previously mentioned, in other experiments it was found that inorganic orthophosphate and orthophosphite were not inhibitory at concentrations 20-fold higher than the substrate PPi. This is in good agreement with Cooperman's (17) recent report that up to a concentration of 25 mM, Pi is not inhibitory.
In contrast, free PPi is strongly inhibitory (4). The weakness of interaction of Pi with the en-1 S. J. Kelly, F. Feldman, J. W. Sperow, and L. G. Butler, ms.nuscript in preparation. zyme suggested that Pi would not significantly phosphorylate the enzyme nor exchange back into 1'Pi; these predictions were confirmed by control experiments with 32Pi (data not shown). Cohn (9)) however, has reported significant exchange of 32P from Pi back into Pl'i.
Inhibitor specificity is apparently dependent upon subtle differences in conformation. The inhibitory nitrogen analog very closely resembles PPi but the inert methylene derivative is less similar structurally (15). All three compounds interact similarly with certain metal ions (18). Differences in geometry have been suggested as being responsible for similar observations with the analogous ATP analogs (16,19).
If this suggestion is correct, Md 02+ complexes with methylcne diphosphonate (or its substituted derivatives) cannot form t'lle correct shape to bind to the active site, possibly due to the fact that the methylene group protrudes more than the oxygen or imido linkage (15). Inhibition of this enzyme by complexes of other metals with rnethylene diphosphonate (17) suggests that the geometry of the complex depends upon the nature of metal involved.
The reported weak competitive inhibition of >fg2+activated yeast pyrophosphat.ase by methylene diphosphonate (20) was observed only at high concentrations of methylene diphosphonate (12.5 and 25 mM) and did not take into account the requirement for free Mg2+ as activator (4). We interpret, this observation, as well as that of others who have reported inhibition in the absence of a large excess of Mg*f (al), as inhibition due to competition for Mgz+ between PPi, "inhibitor" and enzyme.
If the actual substrate is a strain-free, g-membered ring structure formed by MgPPi" (22) a possible transition state is Mg(Pi)s.
The a ,w-glycol diphosphates were tested as possible stable analogs of this transition state. The phosphoryl groups in the inert propane and butane derivatives are presumably too far apart to form the proper conformation.
The ethane and methane derivatives though, may increasingly approximate the transition state. The weak inhibition (apparent KI on the order of millimolar), indicates that the approximation is very poor; this may be due to the necessity of overcoming electrostatic repulsion or steric interaction to force the two phosphates together.
The use of ADP as a potential phosphoryl acceptor permits employment of several sensitive techniques for assaying the ATP formed.
However, if the enzyme recognizes and binds only the pyrophosphoryl portion of ADP and ATP, the mode of ADP binding might not allow it to accept a phosphoryl group. Substrate saturation curves for Zn2f-activated hydrolysis of PPi, ADP, and ATP are similar (5) suggesting that the pyrophosphoryl group is the major, if not sole, site of interaction between nucleotide substrate and enzyme, but this interpretation is difficult to reconcile with the observed differences in the rate of the reaction (PPi-ATP-ADP, 100:4: 1) (5). Our failure to detect phosphoryl transfer to ADP or alcohols leaves Cohn's (9) observation of exchange of 32P from Pi into PPi as the only demonstration with this enzyme of phosphoryl transfer to any acceptor other than water.
If the mechanism includes a phosphoryl enzyme intermediate, this suggests the presence of a specific acceptor site, which can effectively accommodate water but no other potential acceptor. Such acceptor specificity is not observed in other hydrolytic enzymes which form phosphoryl intermediates (14, 23). Alternatively, a concerted mechanism, without a phosphoryl enzyme intermfxdiate, might be expected to be rather more specific for phosphoryl acceptor because of the requirement for formation of the prol)er ternary complex in order for reaction to occur at all.
We are in agreement with hvaeva that there is no O-phosphoryl serine or N-phosphoryl histidine intermediate in the mechanism of the enzyme.
Whereas hvaeva and co-workers postulate an acyl phosphate intermediate (27), we prefer a concerted mechanism.
Our preference is based on the absence of phosphoryl transfer, our inability to detect a phosphoryl intermediate, the absence of essential carboxyl groups at the active site (17), the mechanistic awkwardness of formation of a pyrophosphoryl enzyme, the thermodymic improbability of formation of an acyl phosphate from Pi, and the unusual conditions for enzyme phosphorylation reported by Avaeva et al. In one study, they utilized a 3.3.fold molar excess of enzyme over 32Pi. incubated under conditions (p1-I 5.0, absence of metal ion, 3") where the enzyme is virtually inactive, and after adding detergent allowed the resulting mixture to stand overnight before labeled protein was isolated (27). We suggest that the relationship between results obtained in this manner, and the mechanism of the native enzyme's hydrolysis of its optimal substrate, P.Pi, under conditions where the enzyme is highly active, is tenuous, at best. Detection of a phosphorylated enzyme does not constitute proof that the reaction mechanism necessarily includes such an intermediate; we must also concede that because of the possible instability of a phosphorylated enzyme, the failure to detect it also does not constitute proof that such an intermediate