The formation of enzyme-bound and medium pyrophosphate and the molecular basis of the oxygen exchange reaction of yeast inorganic pyrophosphatase.

Yeast inorganic pyrophosphatase, with 10 mM 32Pi and 10 mM Mg2+ present at pH 7.3 TO 7.6, rapidly forms enzyme-bound pyrophosphate equivalent to about 5% of the total catalytic sties on the two enzyme subunits. The enzyme thus appears to bind PPi so as to favor thermodynamically its formation from Pi. The enzyme catalyzes a measurable equilibrium formation of free PPi at a much slower rate. Under similar conditions, the enzyme catalyzes a rapid exchange of oxygen atoms between Pi and water with the relative activation by metals being Mg2+ greater than Zn2+ greater than Co2+ greater than Mn2+. Millisecond mixing and quenching experiments demonstrate that the rate of formation and cleavage of the enzyme-bound PPi is rapid enough to explain most or all of the oxygen exchange reaction.

Yeast inorganic pyrophosphatase, with 10 mM 32Pi and 10 mu MS+ present at pH 7.3 to 7.6, rapidly forms enzyme-bound pyrophosphate equivalent to about 5% of the total catalytic sites on the two enzyme subunits. The enzyme thus appears to bind PPi so as to favor thermodynamically its formation from Pi. The enzyme catalyzes a measurable equilibrium formation of free PPi at a much slower rate. Under similar conditions, the enzyme catalyzes a rapid exchange of oxygen atoms between Pi and water with the relative activation by metals being M&+ > Zn'+ > Co2+ > Mn'+. Millisecond mixing and quenching experiments demonstrate that the rate of formation and cleavage of the enzyme-bound PPi is rapid enough to explain most or all of the oxygen exchange reaction. In 1958, Cohn reported that inorganic pyrophosphatase catalyzed a rapid exchange of oxygens of inorganic orthophosphate with water oxygens, that is, a Pi + HOH exchange (1). A satisfactory explanation for this exchange reaction has not been available. Cohn established that this exchange did not result from the overall reversal of the hydrolysis of medium PPi (1). Other possibilities suggested are that the exchange may result from reversible formation of a phosphoryl enzyme from Pi and enzyme (1, 2) or a reversible formation of a pentacovalent intermediate from enzyme-bound PPi and water (1, 3). The present paper gives evidence for an exchange mechanism not previously presented.
An explanation for ATP synthesis by oxidative phosphorylation currently being studied in our laboratory is based in part on a molecular explanation for the rapid Pi + HOH exchange reactions catalyzed by mitochondria (see Ref. 4). This explanation proposes that the oxygen exchanges associated with oxidative phosphorylation accompany the reversible cleavage of ATP at the catalytic site. If such a proposal for the exchanges of oxidative phosphorylation were valid, it seemed quite possible that the rapid Pi s HOH exchange reaction catalyzed by yeast inorganic pyrophosphatase might result from a correspondingly rapid and reversible formation of enzyme-bound PPi from Pi, without release of the bound PPi to the medium.
The purpose of the present paper is to present evidence demonstrating that yeast inorganic pyrophosphatase will indeed catalyze the rapid and reversible formation of enzymebound PPi from Pi with Mg2+ or Mr?' present. The techniques used for these measurements were sufficiently sensitive to allow direct measurement of the equilibrium amounts of pyrophosphate formed. Amounts observed were somewhat less than those predicted by the data of Flodgaard and Fleron (5). In agreement with an earlier measurement of Cohn (l), the rate of formation of medium PP, is too slow to account for the "0 exchange. However, enzyme-bound PP; is formed from and cleaved to Pi sufficiently rapidly to account for most, or all, of the observed oxygen exchange. and, after addition of carrier P, and PP,, the presence of ""PP, was detected after separation from "P,, as described above.  Table  I. The observed oxygen exchange rates were linear with time (not shown). Values are given in Table I for the apparent first order velocity constants for exchange. The order of effectiveness of the ions was Mg2+ > Zn'+ > Co2+ > Mn"'. At pH 7.6, Mg'+ is some 25   had the capacity to form bound pyrophosphate from Pi, were a relatively high concentration of enzyme, use of "'P, containing little or no ""PP,, and an adequate method for separation of any small amount of ""PPi formed from the relatively large amount of "'Pi present. Although our approaches, as will be shown, sufficed for the demonstration of the formation of enzyme-bound PPi from Pi, the quantitation of the amount of formation was not as precise as desired. The amounts measured are equivalent to only a fraction of the total concentration of catalytic sites, which is necessarily much less than the 'lPL molarity. Zero-time values often approached experimental values. Data were quite sufficient, however, to establish major conclusions.

Relative
In preliminary experiments not reported in detail here, an apparent formation of ""PP, from pyrophosphatase and "'P, was observed, and the amount of "'PP, formed was roughly proportional to the enzyme concentration.
On the basis of these results, conditions were chosen for subsequent experimentation.
For measurements of the formation of the apparent "'PPi in rapid mixing experiments, a trace amount of ""P, was added to enzyme in the presence of nonlabeled Pi and Mg".
This helped avoid any possible changes in enzyme properties that might follow the initial dissolving of the lyophilized enzyme or the initial exposure of dissolved enzyme to Mg2+ and Pi. If formation of enzyme-bound PP, from Pi and its subsequent hydrolysis were the basis for the P, + HOH exchange, enzyme-bound PP, would already be present at a steady state concentration and would be rapidly labeled by  (5). Such results were reported in preliminary form (16) and justified more extensive experimentation on the rate and extent of ""PPi formation.
Prior to making the more extensive studies given later in this paper, it was deemed necessary to test for the authenticity of the "'P-labeled product as PP, and to check on the indicated binding to protein by an independent method. Such tests are given in the following sections. Also, as mentioned under "Experimental Procedures," a commercial enzyme preparation was used for most of the present experiments. It was thus desirable to check on bound PPi formation with an enzyme of known preparative procedure.
With a highly purified pyrophosphatase prepared as described by Cooperman et al. (7), PP, formation per unit of enzyme activity was identical within experimental error with that noted with the commercial enzyme.
Identification of the Labeled Product as 32PP,-For these tests, a sample of the ""P-labeled product prepared under conditions similar to those given in Table III was mixed with  authentic  nonlabeled PPi. The ""P-labeled product and authentic PP, were found to be completely converted to "'Pi by inorganic pyrophosphatase action in the presence of Mg"+ or by heating in 1 N HCl at 100°C for 10 min. In addition, in a gradient elution on Dowex-1 (0.01 M HCl + 0.2 M HCl and 0.05 M KCl), the radioactive "P product and authentic PP, appeared in maximum concentration in the same fraction (tube no. 114) with comparable specific activity in preceding and following fractions.
As a further test, the relative rate of acid hydrolysis of the 'lP-labeled product and authentic pyrophosphate at 40°C was measured (see "Experimental Procedures").
Results of these measurements are given in Fig. 1. The data of Fig. 1 show that the rate of liberation of "P, from the labeled product and of P, from authentic PPi were identical. The first order kinetic plots show strict linearity over two half-lives. These acid and enzymic hydrolysis tests thus allow identification of the "Plabeled product as "lPPi.
Direct Demonstration of Enzyme-bound Pyrophosphate-Although the preceding experiments gave evidence that most of the "'PP, formed was enzyme-bound, confirmation of this important conclusion by independent means appeared desirable. Very small traces of 32PP, would be expected in the medium, but if all or nearly all of the "'PPi present were that bound to the enzyme, appropriate gel filtration should show that most of the "PP, migrates with the protein peak and not where free PP, migrates.
Results of a gel filtration experiment are given in Fig. 2. Enzyme was exposed to ""Pi under conditions where '=PP, is formed. The enzyme solution was then placed on a Sephadex column to which sufficient "'Pi-containing medium had been added to assure that, as pyrophosphatase moved ahead of low molecular weight substances, it would always be equilibrated with "'Pi. Any "2PPi that was not protein-bound would lag behind the pyrophosphatase in the column and appear in the eluate after the protein. Shown in Fig. 2 are the fractions where protein and PP, would be expected to appear on the basis of independent trials with the columns used. Nearly all the "'PP, in the sample where pyrophosphatase was equilibrated with ""Pi appears with the protein peak. These results demonstrate that most of the "'PPi formed is indeed proteinbound. With conditions as used for Fig. 2, a small amount of the total '"PPi formed would be expected to represent medium PPi present at equilibrium concentration with the Pi (5). This may account for the apparent presence of some "'PPi migrating after the protein (Fig. 2).

Measurement of the Total Pyrophosphate Formed under Equilibrium
Conditions-These measurements were undertaken in part because the sensitivity of the methods appeared adequate to measure directly both the amount of PP, present at equilibrium and its rate of formation. Also, a direct measurement of the rate of medium PPi formation from medium Pi would give another method of checking the possibility that reversal of the overall reaction contributed to the Pi = HOH exchange.
As a means of differentiating between enzyme-bound and medium PPi at equilibrium, measurements were made of total "PPi formation from 32Pi at various enzyme concentrations. Results are presented in Table III. With 5 mM Pi and 5 IIIM Mg2+, total PPi formation drops to a constant value of 0.035 to 0.045 pM at 22°C and 0.01 to 0.015 pM at 4'C as enzyme concentration is decreased, indicating that these amounts represent equilibrium levels. Data presented later in this paper give assurance that the incubation times used were sufficient for >98'% of equilibrium PP, concentration to be reached even at the lowest enzyme concentration used. At the highest enzyme concentration (5 pM) the total PPi formed is 0.29 PM, of which only about 0.04 pM represents an equilibrium concentration and thus about 0.25 pM, equivalent to about %o of the enzyme molarity, is enzyme-bound.
With 10 mM Pi and Mg2+, the amount of total PPi present in all measurements at 0.0125 pM or less enzyme (conditions where enzyme-bound PPi would be negligible) was 0.175 + 0.03 pM. This may be compared with a value of 0.46 pM calculated from measurements of Flodgaard and Fleron under similar conditions (5).
Also shown in Table III are some measurements of oxygen exchange rates under the same conditions. The exchange rate per mol of enzyme shows little or no change as enzyme concentration is decreased. Thus, the enzyme is not changing properties as it is diluted.
Rate of 32Pz Labeling of Medium PP, at Equilibrium-As a means of checking on the possible contribution of reversal of the overall hydrolytic reaction to oxygen exchange, the rate of attainment of an equilibrium concentration of medium PP, at low enzyme concentration was measured. Results of one series of measurements with 10 mM 32Pi and 10 mM Mg'+ are given in Fig. 3. From a semilog plot of the data, a t1/2 for approach to isotopic equilibrium was calculated to be 4.6 min and the initial rate of PP, labeling was 75 ~mol/~mol of enzyme/min. This may be compared to a rate of 20,000 pmol of PPi cleaved/pmol of enzyme/min under comparable conditions but with 5 InM PPi instead of Pi present.
Under the same conditions (Table III), the rate of oxygen exchange is 4300 patoms of oxygen exchanged/pm01 of enzyme/min. Clearly, and in confirmation of Cohn (l), the rate of formation and cleavage of medium PPi does not account for the exchange. However, the reversal of overall hydrolysis is considerably faster than the lower estimate given by Cohn. The Kinetic Competency of Bound Pyrophosphate as an Intermediate in the P, + HOH Exchange-If the formation and hydrolysis of enzyme-bound pyrophosphate is responsible for the Pi = HOH exchange, it should be theoretically possible to demonstrate that the rate of formation and cleavage of the bound pyrophosphate is sufficiently rapid to account for the exchange. The difficulty of obtaining precise measurements of E. PPi formation, particularly with millisecond mixing and quenching, did not favor a critical test of kinetic competency. However, it seemed possible to assess if the rates of formation and cleavage of E. PPi fall within the range required to explain the oxygen exchange.
For this purpose, use was made of a simple millisecond mixing and quenching apparatus described elsewhere (17). Results are shown in Figs. 4 and 5. At 22°C (Fig. 4), the labeling of E .PP, appears to be completed within 20 ms indicating a tl,2 of 10 ms or less. At 4"C, the formation of enzyme-bound PP, is much slower with a t1,2 of roughly 30 ms. As noted from Table III, some of the PPi formed at 1 min is due to equilibrium formation. The experiments required high amounts of "'P and this, together with the time required for the individual separations necessary for each experimental point, the variability encountered, and the limitation of the apparatus used to cover a wider mixing time and temperature range, made attempts to get more complete and precise data unwarranted at this stage.
FIG. 3. Approach to isotopic equilibrium of "PP, and "P1. A 5.0ml reaction mixture of 50 mM Hepes, 10 mM Mg'+, 10 mM P,, 135 mM KCl, and 0.25 nM enzyme at pH 7.5 and 25°C was incubated for 2 h. Purified, pH-adjusted '*PL (10' cpm) was added and 0.5-ml aliquots were withdrawn at times indicated, quenched, and assayed as in Table  III The results obtained, as noted in the "Discussion," suffice to establish formation and cleavage of bound PPi as a prominent and quite possibly the only mechanism of exchange.

DISCUSSION
The findings presented in this paper allow the conclusion that yeast inorganic pyrophosphatase catalyzes the rapid formation of a small but highly significant amount of ""PP, from medium 32Pi. This initial labeling is much faster than the rate of formation of an equilibrium concentration of PPi. The initial burst in '"PPi formation, the approximate proportionality of the amount of "2PPi formation to enzyme concentration, and the separation of most of the "'PPi with the protein in gel filtration establish that this ""PPj is enzyme-bound. Our data further establish that the formation and hydrolysis of the enzyme-bound PPi is responsible for much and probably all of the P, + HOH exchange catalyzed by the enzyme.
Present information indicates that the bound PPi formed from P, is held by noncovalent bonds, as in a firm Michaelis complex, and is thus not covalently attached to the enzyme. The bound PP, is readily released to the medium by cold perchloric acid under conditions where known phosphoryl protein derivatives show limited or no hydrolysis.
It is apparent that pyrophosphatase has the capacity to favor formation of PP, from P, at the catalytic site considerably above that expected from the known -AC? for hydrolysis in solution. Factors that might favor such formation of PP, are a preferential tight binding of PPi, a low water activity, a high effective local concentration of Pi, and an increased proton availability.
In the presence of the higher concentrations of enzyme used in our experiments, an equilibrium concentration of free PP, in solution is soon attained. As enzyme concentration is reduced, the-amount of enzyme-bound PPi is correspondingly reduced and the remaining PPi at the lowest enzyme concentrations greatly exceeds the total enzyme concentration (Table  III). Such PP, is in equilibrium with P, under the experimental conditions. Estimates of the equilibrium concentration of PP, from data in Table III give values somewhat lower than the more extensive measurements of Flodgaard and Fleron (5 Theoretical relationships governing oxygen exchange in such a system but without the complications of equilibrium formation have been developed more fully in another paper from this laboratory on the Pi + HOH exchange catalyzed by the sarcoplasmic reticulum ATPase (18). For the present discussion, some quite simple considerations will suffice. Two limiting possibilities may be envisaged for the oxygen exchange, with either the fiit step of Equation 1 rapid compared to the second, or vice versa. The fate of E .2P,, once formed, is conveniently expressed by the partition coefficient, P,, where P, = k2/(km1 + k2). As P, approaches 0, then each medium Pi that forms E.PP, will return to the medium P, pool with one oxygen atom acquired from water. The oxygen exchange rate would equal the rate of bound PP, formation. As P, approaches 1, a medium Pi upon forming the enzymebound Pi will acquire nearly four oxygens from water before returning to the medium.' An estimate of the value for the partition coefficient has been made from measurements of the distribution of the ["O]P, species during exchange starting with fully ' (19). The results show t,hat the partition coefficient does not approach either extreme and would be expected to be in the range of 0.2 to 0.4 for our experimental conditions. From the data of Table III at 22"C, with 5 mM Mg" and 1.25 pM enzyme, E .PP, is 0.105 pM minus 0.04 or 0.065 pM, and the constant relating E. PPi concentration to the exchange rate is thus 800 s-'.
If PC were close to 0, the tlp2 for labeling of E e PPi would be 0.69/k or 0.9 ms. As P, values increase, labeling of enzymebound PPi would not need to be as rapid to account for the oxygen exchange because more than 1 atom of oxygen would ' All four oxygens of E. P, are assumed to have the same probability of exchange.
This assumption is in harmony with unpublished findings of D. D. Hackney in this laboratory on patterns of '"0 loss from fully '"O-labeled P, during oxygen exchange catalyzed by pyrophosphatase.
by guest on March 24, 2020 http://www.jbc.org/ Downloaded from be replaced for each P, bound. Thus, if P,. were close to 1, the t,,2 would be close to 4 times greater, or -3.6 ms. At intermediate values, the relation between E .PP, and time would deviate slightly from hyperbolic (18), and the apparent tlj2 would be between 1 and 4 ms. As mentioned above, P,. is somewhat greater than 0 and labeling of half of the E .PP, would be expected in no more than 2 to 3 ms. Data of Fig. 4 at 22°C are consistent with this. Results of oxygen exchange measurements and E .PPj formation at 4°C by similar approaches lead to a lower limit of 10 ms for P, approaching 0 and an upper limit of 40 ms for P, approaching 1. An estimate of the apparent half-time of labeling from data of Fig. 5 is 30 to 40 ms.
These results show that the formation and cleavage of enzyme-bound pyrophosphate from medium P, accounts for most, if not all, of the Pi G+ HOH exchange. Such a conclusion is in harmony with other related considerations.
The reasonable assumption may be made that E. PP, has similar rates of cleavage whether formed from Pi or PPi. Thus, the appreciable fraction of E .PP, present when only Pi and the enzyme are mixed would be cleaving at a rapid rate and account for the rapid net hydrolysis observed when E and PPi are mixed and form E .PPi. The rate of formation of E .PPi from E and P, obviously also must be rapid in order to maintain an appreciable fraction of the total enzyme in the form of E.PPi. Thus, a rapid Pi + HOH exchange would be expected to result from the rapidly reversible formation of E. PPi from E and P,. Our data establish that enzyme-bound pyrophosphate is rapidly and preferentially converted to medium P,. With the premise that the same type of bound PP, participates in the net hydrolysis reaction, the results show that once the productive E .PP, complex is formed, it is about 30 times more likely to be cleaved to P, than to return to medium PP,.
The high rate of P, + HOH exchange noted in Table III  For the Pi + HOH exchange, a possibility previously considered was that the exchange results from the reversible formation of a phosphoryl enzyme from P,. This would be akin to the exchange catalyzed by alkaline phosphatase (20) and by transport ATPases (21,22). Although experiments of Nazarova and Avaeva (23) have indicated a binding of "'Pi interpreted as representing a phosphorylated intermediate of the hydrolysis reaction, conditions for the detection of the binding were unusual and rapid interchange with medium P, or PP, was not established.
In the careful experiments of Sperow et al. (2) and of Rapoport et al. (24), no evidence for a phosphoryl enzyme intermediate has been obtained. It is difficult to prove that something does not exist, however, and it remains conceivable that in our experiments a phosphorylated enzyme precedes the formation of enzyme-bound PPi from Pi. Also, Baykov et al. (25)(26)(27) have reported evidence interpreted as showing a covalently bound PP, formed in the presence of enzyme, F-, PPi, and Mg"+. We thus checked the effect of F-on enzyme-bound PPi formation. At 10 mM Mg" and 10 mM "'P,, 75 mM F-decreased the amount of "'PPi formed from 0.15 to 0.01 ,UM and the oxygen exchange to 0. With 100 mM Mg"+ present, 75 mM F-did not change the amount of ""PP, formation.
Other suggestions have been that the oxygen exchange results from the reversible formation of a pentacovalent derivative from bound PP, and water (1, 3). Our data give no support to such suggestions. If a relatively rapid and reversible formation of any such derivative occurred, the oxygen exchange rate should have considerably exceeded the rate of formation of bound PP, from medium P,. Also, formation of considerable E. PPi would not be expected. As noted in the introduction, these studies with pyrophosphatase were prompted by suggested mechanisms for oxygen exchange accompanying oxidative phosphorylation. Related studies from this laboratory with myosin have demonstrated formation of bound ATP from ""Pi (28). These results, the findings of Bagshaw and Trentham (29), and the demonstration of water-oxygen incorporation into myosin-bound ATP (30) are consistent with the reversible cleavage of bound ATP being responsible for the oxygen exchanges observed with muscle. Evidence favoring this explanation comes from data showing a capacity of myosin to cause scrambling of the ,L?,vbridge oxygens of ATP (31).
With other enzymes, unlike with pyrophosphatase, a P, = HOH exchange could result from dynamic reversal of formation of a phosphorylated enzyme. With alkaline phosphatase, oxygen exchanges do appear to result from such reversible enzyme phosphorylation (20). Also, recent studies with sarcoplasmic reticulum ATPase show that the rate of formation and cleavage of the phosphoryl enzyme from Pi can account for the rapid Pi + HOH exchange observed (18). Present evidence makes it tenable to suggest that all exchanges of phosphate oxygen with water catalyzed by enzymes are associated with dynamic reversal of cleavage of covalent phosphorylated substances.