The Role of Nucleoside Triphosphate Pyrophosphohydrolase in in Vitro Nucleoside Triphosphate-dependent Matrix Vesicle Calcification*

Nucleoside triphosphate pyrophosphohydrolase (EC 3.6.1.8) activity is associated with matrix vesicles purified from collagenase digests of fetal calf epiphyseal cartilage. This enzyme hydrolyzes nucleoside triphosphates to nucleotides and PPi, the latter inducing precipitation in the presence of Ca2+ and Pi. An assay for matrix vesicle nucleoside triphosphate pyrophosphohydrolase is developed using 0,y-methylene ATP as substrate. The assay is effective in the presence of matrix vesicle-associated ATPase, pyrophosphatase, and alkaline phosphatase activities. A soluble nucleoside triphosphate pyrophosphohydrolase is obtained from matrix vesicles by treatment with 5 mM sodium deoxycholate. The solubilized enzyme induced the pre- cipitation of calcium phosphate in the presence of ATP, Ca2+, and Pi. Extraction of deoxycholate-solubilized enzymes from matrix vesicles with 1-butanol destroys nucleoside triphosphate pyrophosphohydrolase activ- ity while enhancing the specific activities of ATPase, pyrophosphatase, and alkaline phosphatase. In solutions devoid of ATP and matrix vesicles, concentrations of PPi between 10 and 100 p~ induce cal- cification in mixtures containing initial Ca2+ X P ion products of 3.5 to 7.9 mM2. This finding of and Ions on Actiu- ity-Matrix vesicles decalcified vesicle

§ TO whom requests for reprints should be sent. tracted with 1-butanol, produce a stable water-soluble enzyme which may be purified to homogeneity (4, 5). The ATPase, alkaline phosphatase, and pyrophosphatase activities are associated with a single molecular entity (4).
Matrix vesicles promote mineral deposition from metastable solutions of Ca2+ and Pi (6)(7)(8). A number of theories have been advanced to explain matrix vesicle-induced calcification. First, matrix vesicles may actively transport Ca2+ and Pi. Vesicle enzymes are believed to increase local Pi concentrations by their phosphatase activity and derive energy to pump Ca2+ from ATP cleavage (9). Secondly, matrix vesicles may initiate mineral precipitation by heterogeneous nucleation (10). Complexes of Ca2+, phospholipid, and Pi (11,12) nucleate metastable calcium phosphate solutions. Phosphatidylserine and phosphatidylinositol are present in high concentration in matrix vesicle membranes (13). It is suggested that Ca2+phospholipid-Pi complexes may invest these membranous surfaces and act as nucleation sites for mineral phase separation. Finally, matrix vesicles may initiate mineralization by removal of inhibitors (14). Pyrophosphate inhibits the transformation of amorphous calcium phosphate to crystalline hydroxyapatite (15). Hence, matrix vesicles may induce calcification through its pyrophosphatase activity.
Pyrophosphate is conspicuously present in mineralizing tissues. Calcification in tissue slices is stimulated by PPi (16).
Successive zones of cartilage in the epiphyseal growth plate show progressively increasing concentrations of PPi proceeding from the resting zone to the zone of calcification (17). The initial precipitation of amorphous calcium phosphate from solution is enhanced by the presence of pyrophosphate (18,19). Finally, elevated synovial fluid PPi concentrations are found in articular chondrocalcinosis, osteoarthritis, and gout, all conditions associated with pathological mineralization (20).
It is the purpose of this report to show that matrix vesicles are directly responsible for the production of PPi in in. vitro calcifying mixtures. Data are presented which support the presence in matrix vesicles of NTP':pyrophosphohydrolase which cleaves NTP to NMP and PPi. It is further demonstrated that the increase in PPi concentration induces in vitro calcification.

EXPERIMENTAL PROCEDURES
Materials-Alkaline phosphatase, Type 111, from Escherichia coli, alkaline phosphatase, Type VII-S from bovine intestine, p-nitrophenyl phosphate, AMP-PCP, GMP-PCP, and UMP-PCP were obtained from Sigma. dTMP-PCP was a product of P-L Laboratories, ATP was a Calbiochem product, and [6Ca]CaClz was purchased from New England Nuclear. All other chemicals were reagent grade.
Enzyme Assays-Alkaline phosphatase activity was measured using p-nitrophenyl phosphate as substrate at pH 10.5 and 25 "C by following the liberation of p-nitrophenolate ion spectrophotometrically at 405 nm (4, 21). A unit of activity is defined as the amount of enzyme required to hydrolyze 1 @mol of substrate/min. Enzymatic activity toward ATP was measured as previously described (5, 22). NTP:pyrophosphohydrolase activity was measured using AMP-PCP as substrate in the presence of excess Escherichia coli alkaline phosphatase. Pyrophosphorolytic cleavage of the substrate produces methylene diphosphonate and AMP, the latter product being cleaved by the matrix vesicle and bacterial alkaline phosphatase to adenosine and Pi. Inorganic phosphate is estimated colorimetrically by the method of Martin and Doty (22). Sufficient Escherichia coli alkaline phosphatase is added to ensure that the first step in the coupled assay is rate-determining. Typically, reaction mixtures contained 0.12 M Hepes/NaOH buffer, pH 7.6,0.25 mM AMP-PCP, 0.6 unit of Escherichia coli alkaline phosphatase, and matrix vesicles in a total volume of 1.6 ml. All ingredients except substrate were preincubated at 25 "C for 10 min and reactions were initiated by adding substrate. Duplicate aliquots of 100 pl were removed at 0, 15, 30, 60, 120, 180, and 240 min and assayed for Pi. modified Lowry method (23).
Protein Concentration-Protein concentration was measured by a Preparation of Matrix Vesicles-Intact matrix vesicles were prepared from bovine fetal epiphyseal cartilage by the method of Ali et al. (24). The specific alkaline phosphatase activities of different preparations varied between 3.3 and 5.6 units/mg of protein. Preparations were stored frozen at -20 "C in 120 mM NaCl, 10 mM KCl, 2 mM Tes/NaOH buffer, pH 7.45, at a concentration of 1 mg of protein/ ml and generally lost 10% of their activity per month upon storage. Preparations were always used within 3 months of isolation. Matrix vesicle alkaline phosphatase was isolated and purified by the method of Fortuna et al. (4). ured at 25 "C using the sedimentation method of Hsu and Anderson Calcification Assay--In vitro matrix vesicle calcification was meas- (25). Each reaction mixture contained approximately 6,000,000 cpm of ["Ca2+]CaClz. Washed pellets of precipitated calcium phosphate and matrix vesicles were mixed with 0.1 ml of Hz0 and 5 ml of Bray's solution and shaken overnight and the resuspended pellet was counted in a liquid scintillation counter. The extent of calcification is given as nanomoles of Caz+ precipitated per ml of reaction mixture and each point represents the average of three determinations. Errors are reported in the appropriate figures as the S.D.
Solubilization of Matrix Vesicle NTP:pyrophosphohydrolase-Matrix vesicles (1 mg of protein, 5.5 alkaline phosphatase units) were solubilized in 1 ml of 120 mM NaCl, 10 mM KC1, 2 mM Tes/NaOH, pH 7.45, containing 5 mM sodium deoxycholate. After 2.5 h, the solution was centrifuged for 6,000,000 X g-min and the supernatant was assayed for NTP:pyrophosphohydrolase and calcifying activity.
Effect of M$+ and Ca2+ Ions on NTP:pyrophosphohydrolase Actiuity-Matrix vesicles were decalcified by incubating 2.5 ml of vesicle suspension (13 alkaline phosphatase units) with an equal volume of 200 mM Na citrate, pH 6.0, at 0 "C for 4 h. Vesicles were harvested by centrifugation and washed twice in 120 mM NaC1, 10 mM KCl, 2 mM Tes/NaOH, pH 7.45, and finally suspended in 2 ml of the pH 7.45 buffer. Assays were performed in 120 mM NaC1,lO mM KC1,25 mM Tris. HCl, pH 7.4, at 25 "C. Reaction mixtures contained 0.3 unit of Escherichia coli alkaline phosphatase and matrix vesicles equivalent to 0.24 alkaline phosphatase unit/ml. Free Ca2+ and M%+ ion concentrations varied from 0.01 NM to 0.01 M at 10-fold intervals. The fraction of total added metal ion bound to the substrate, AMP-PCP, was estimated from the conditional binding constants of AMP-PCP for Caz+ and M%+ at pH 7.4 (26). Reactions were initiated by adding AMP-PCP to 0.25 mM. Duplicate aliquots were removed after 0,2,4, and 6 h and the Pi content was determined. All reactions were carried out in polyallomer tubes rinsed in 1 mM EDTA followed by deionized water.

RESULTS
The effect of matrix vesicle concentration on the progresstime curves for phase separation of calcium phosphate salts is shown in Fig. 1. Each curve is characterized by three distinct regions: an initial induction period in which no precipitate forms, an intermediate stage in which the rate of accumulation of insoluble calcium salts is roughly proportional to the matrix vesicle concentration, and a terminal phase in which no further precipitation occurs. The duration of the induction period is inversely proportional to the matrix vesicle concentration while the terminal phase is independent of matrix vesicle concentration. The effect of Ca2+ and Pi concentrations on the time course of calcium phosphate precipitation is shown in Fig. 2. Increasing Pi levels do not abolish or appreciably shorten the lag phase. The highest initial Pi concentration used, 4.8 mM, represents one which would result if all the ATP were completely hydrolyzed to adenosine and 3 mol of Pi. In contrast, increasing the initial Ca2+ concentration decreases and eventually removes the lag while also accelerating the precipitation rate in the intermediate phase.
Preincubation of matrix vesicles in the presence of Ca" and Pi for 2.5 h in the absence of ATP does not alter the kinetic course of calcium phosphate precipitation. When vesicles are preincubated with ATP and Pi for 2.5 h in the absence of Ca2+, the induction period is abolished (cf. Fig. 3 When an ATP-regenerating system (27) composed of pyruvate kinase and phosphoenolpyruvate is added to the calcification mixture, the induction period is extended by 1 h (225% of control) in the standard calcification assay (initial ca2+, 2.2 mM; initial Pi, 1.6 mM). When 1. that the first step is rate-determining and provides a linear release of Pi throughout the entire time course of the reaction. Table I illustrates that yeast pyrophosphatase acts as a calcification inhibitor by hydrolyzing the PPi formed in the NTP:pyrophosphohydrolase reaction. A 95% inhibition is observed when ATP is substrate but no inhibition is seen when AMP-PCP is used consistent with the fact that yeast pyrophosphatase is incapable of splitting methylene diphospho- All reactions were carried out at pH 7.6 and 25 "C and contained initially 2.2 mM Ca2+ and 1.6 mM P,. The curves are: o " 0 , matrix vesicles -0.32 alkaline phosphatase unit/ml of reaction mixture, 1 mM AMP-PCP; U , matrix vesicles -0.20 alkaline phosphatase unit/ml of reaction mixture, 1 mM AMP-CPP; X-X, matrix vesicles -0.47 alkaline phosphatase unit/ml of reaction mixture preincubated for 2 h with 1 mM ATP and the reaction was initiated by the addition of Ca2+; U , matrix vesicles -0.40 alkaline phosphatase unit/ml of reaction mixture preincubated for 2 h with 1 mM AMP-PCP and the reaction was initiated by the addition of CaZ+. ATP are replaced by 3.2 mM Pi and 1 mM ADP, conditions which represent the complete hydrolysis of ATP to ADP in the calcification mixture, no calcium phosphate precipitates.
Addition of I mM ADP + 1 mM ATP to the calcification mixture reduces the relative extent of calcification after 3 h by over 90% compared to mixtures containing ATP alone.
The above experiments suggested that increased Pi production by matrix vesicles is not responsible for calcium precipitation, that ATP is specifically required, and that ATP must be consumed during the calcification process. To test whether matrix vesicles might support calcification by specifically binding a fraction of the ATP present without cleavage, ATP was replaced with AMP-PCP. Fig. 3 (o"-o) shows that this ATP analog supports calcium phosphate precipitation, while the a,@-methylene analog, AMP-CPP (cf Fig. 3, U), does not. The analog, AMP-PCP, exhibits a lag phase similar to ATP and if preincubated with matrix vesicles in the absence of Ca2+ for 2.5 h, the induction period is destroyed (cf. Fig. 3, U).
It was next necessary to show whether the a,@-and @,ymethylene analogs of ATP are cleaved by matrix vesicle enzymes. Fig. 4 demonstrates an initial linear release of Pi from ATP ( U ) and AMP-CPP (X-X). Inorganic phosphate is released in the enzymatic hydrolysis of AMP-PCP (cf. Fig. 4, -) after a considerable lag phase. When reaction mixtures containing AMP-PCP are supplemented with Escherichia coli alkaline phosphatase, the lag phase is eliminated and the initial Pi release rises linearly in the supplemented reaction mixtures (cf. Fig. 4
The data from Fig. 4 support a two-step sequential hydrolysis of AMP-PCP by matrix vesicles in which the initial site of attack is the a,@-phosphoanhydride linkage. In a second step, AMP is hydrolyzed to adenosine and Pi. The lag associated with Pi release when AMP-PCP is incubated with matrix vesicles alone indicates that in the initial phase of hydrolysis the second step is a t least partially rate-determining. Addition of excess bacterial alkaline phosphatase ensures nate.
Support for the role of PP, in calcification comes from studies of Ca2+ precipitation in the absence of both ATP and matrix vesicles. Fig. 5 shows that after 24 h there is no Ca2+ precipitation in the absence of PPi when the initial ion product of Ca2+ X Pi varies between 3.5 and 7.9 mM2. On the other hand, addition of PPi to the same solutions at levels of 10, 50, and 100 PM induce Ca2+ precipitation. The extent of calcification is a function of both Pi and PPi concentration.
Kinetic studies using 2.2 mM Ca2+ and 3.6 mM Pi supplemented with 10 or 100 PM PP, show that maximum precipitation occurs instantaneously. The extent of calcification observed after 24 h shows a 10% decrease compared to the zero time measurement.
Solubilization of matrix vesicles in aqueous buffers containing 5 mM sodium deoxycholate gives a nonsedimentable fraction which is capable of inducing calcium phosphate precipi- TAELE I Ca2+ precipitation in the presence of added yeast pyrophosphatase All reactions were run for 24 h at 25 "C and pH 7.6 in the standard calcification mixture containing initially 2.2 mM ca2+, 1.6 mM P,, and 1 mM substrate. Each sample contained matrix vesicles at a concentration equivalent to 0.39 alkaline phosphatase unit/ml of reaction mixture. Yeast pyrophosphatase (Sigma) was added to the appropriate reactions a t a concentration of 1.4 units/ml of reaction mixture.
Results are reported as the mean of triplicate determinations and errors are reported as the S.D.   The deoxycholate-soluble extract is the 6,000,000 X g-min supernatant prepared as described under "Experimental Procedures." The final concentration of sodium deoxycholate in the calcification assay mixture is 0.75 mM.

DreciDitated
One volume of 1-butanol is shaken with 1 volume of deoxycholatesoluble extract for a total of 105 s in 15-s intervals separated by 15-s periods of nonmixing. The mixture was separated by centrifugation for 25,600 X g-min and the aqueous layer was assayed for calcifying activity. tation (cf. Table 11). This supernatant fraction possesses NTP:pyrophosphohydrolase activity. Extraction of the solubilized fraction with 1-butanol produces an aqueous phase devoid of NTP:pyrophosphohydrolase activity but enriched with respect to ATPase, pyrophosphatase, and alkaline phosphatase activities. Neither the butanol-extracted enzyme, nor the Escherichia coli, nor bovine intestinal alkaline phosphatase supports calcification. Addition of heat-inactivated vesicles to these preparations did not induce calcium phosphate precipitation.
Progress-time curves of calcification in the presence of dTMP-PCP, UMP-PCP, GMP-PCP, and AMP-PCP are shown in Fig. 6. Each curve exhibits an induction period followed by the intermediate active calcification phase. The plateau stage (not shown) is reached in 24 h for dTMP-PCP and UMP-PCP, in 48 h for GMP-PCP, and 72 h for AMP-PCP. Rates of calcification were computed from the linear response in the early portion of the intermediate phase and the induction times was estimated by extrapolating the initial linear calcification rate curve to the time axis (cf. Table 111). Table I11 correlates the induction time and calcification rate with the substrate specificity of the matrix vesicle NTP:pyro-phosphohydrolase. Residual Ca2+ present in the soluble phase at the plateau stage is independent of the structure of the substrate and is 8.00 k 0.2%.
The pH optimum for NTP:pyrophosphohydrolase using AMP-PCP as substrate is 10.0. Progress-time curves are linear from pH 6.0 to pH 8.0 and curvilinear at more alkaline pH values. All buffers used are zwitterionic in the acid form and monoanionic in the basic form. When a HC03-/C03" buffer was used at pH 10.5, inhibition was observed with increasing buffer concentration. At total HCO3-plus C03'concentrations of 24 and 48 mM, the NTP:pyrophosphohydrolase activity is 61 and 36%, respectively, of the activity when the total buffer concentration is 12 mM.
Preincubation of matrix vesicles in buffers at pH 7.5 and 10.5 for up to 120 min results in no change in NTP:pyrophosphohydrolase activity, whereas preincubation in 12 mM sodium acetate buffer, pH 4.0, leads to irreversible loss of activity.
NTP:pyrophosphohydrolase is neither activated nor inhibited by concentrations of free Ca2+ or Mg2+ in the range of   Fig. 6 Substrate specificity was determined using the 0,y-methylene analogs of ATP, GTP, UTP, and dTTP. Reactions were carried out at 25 "C and contained 0.12 M Hepes/NaOH, pH 7.6, 0.3 unit of Escherichia coli alkaline phosphatase/ml, and matrix vesicles equivalent to 0.25 alkaline phosphatase unit/ml. Reactions were initiated by addition of substrate to 0.23 mM. Duplicate aliquots were taken at zero time and at three additional times. The Pi liberation was linear over the entire progress-time curve for all substrates. Numbers in parentheses represent the value of the parameter normalized to that for dTMP-PCP taken as 1. ion is 0.0159 f 0.0020 pmol/min/mg and 0.0154 * 0.0016 pmol/min/mg for the Mg2+ ion-treated preparation. When matrix vesicles were treated with Ca2+ at 0.1 mM plus M e varying from 0.1 to 1.0 mM, the specific NTPpyrophosphohydrolase activity was 0.0150 f 0.0009 pmol/min/mg. NTP:pyrophosphohydrolase is fully saturated with the substrate, AMP-PCP, between 0.01 and 0.9 mM at pH 7.6 and 25 "C. At pH 10.0, the enzyme exhibits saturation kinetics between 0.1 and 0.5 mM. V,,, for hydrolysis of AMP-PCP at pH 7.6 and 25 "C is 0.012 pmol/min/mg.

DISCUSSION
The induction phase associated with the time course of calcium phosphate precipitation suggests a requirement for the buildup of a critical concentration of a component not present in the system initially. The rate of production of this component is proportional to matrix vesicle concentration since the induction time is inversely proportional to vesicle concentration. Elimination of the lag period by preincubation with ATP supports this concept and implies that the formation of the critical component results from an ATP-matrix vesicle interaction. The lengthy induction periods observed in Fig. 1 tend to rule out an ATP-matrix vesicle binding phenomenon. This idea is supported by the observation that AMP-CPP does not induce calcium phosphate precipitation. Fig. 1 shows that after the lag period, the rate of precipitation of calcium increases with increasing matrix vesicle concentration. This could result from the requirement for a continued production of the critical component during the precipitation phase or from the effect of a second matrix vesicle component on the rate of precipitation.
Active transport of Ca'+ and Pi appears unlikely since the final amount of Ca2+ precipitated is independent of the vesicle concentration. In addition, vesicles solubilized in deoxycholate solution support calcification. Generally, transport against a concentration gradient in biological systems requires space enclosed by a membrane. Also, since exogenously added yeast pyrophosphatase may be expected to exert its action external to the intravesicular environment, the PPi formed in the NTP:pyrophosphohydrolase reaction must be produced on and released from the outer surface of the vesicle membrane.
Initial Pi levels play no role in the induction phase; however, increasing initial Ca2+ levels ultimately eliminates the lag period suggesting that Ca'+ plays a role in the induction phase. The instantaneous precipitation observed upon ATP or AMP-PCP preincubation strongly suggests that the critical component accumulated during the induction phase is PP, and that a critical Caz+ x PPi ion product may be necessary to induce calcium precipitation. When initial calcium levels are elevated, instantaneous precipitation is not observed because a finite time is required for the enzymatic formation of PP, from ATP. I n vitro studies have shown that PP, stimulates the initial deposition of calcium phosphate salts (18,19), in agreement with the results reported in this paper. This initial lag phase may also be viewed as a time when the enzymatic hydrolysis of ATP, which has a strong affinity for Ca'+, effectively increases the concentration of free Ca2+. Supporting evidence is derived from the observation that elevated ATP levels inhibit in vitro calcium (7) and phosphate' deposition, and the addition of an ATP-regenerating system extends the induction time. Experiments measuring Pi liberation from AMP-PCP show that matrix vesicles possess NTP:pyrophosphohydrolytic ac-* R. Carty, C. Hummel, and S. Siegel, unpublished experiments.
tivity. An alternative mechanism in which ATP is cleaved to adenosine and trimetaphosphate, the latter undergoing cleavage to Pi and PPi, is ruled out by the fact that AMP-CPP is hydrolyzed by matrix vesicles with an initial linear phosphate release, and this analog does not support matrix vesicle calcification. The liberation o f Pi from AMP-CPP does indicate that matrix vesicles also possess true ATPase activity. The presence of true ATPase activity is supported by the fact that the ATP-regenerating system, which depends on cleavage of ATP to ADP, is capable of increasing the induction time of calcification.
The results in Fig. 4 indicate that the Escherichia coli alkaline phosphatase modifies only the early phase of the time course of P, liberation from AMP-PCP in the presence of matrix vesicle NTP:pyrophosphohydrolase. The implied poor affinity of the matrix vesicle alkaline phosphatase for AMP makes 5"AMPase activity rate-limiting at low AMP levels and, hence, early times. As AMP levels increase above the K , for matrix vesicle alkaline phosphatase, the NTP:pyrophosphohydrolase reaction becomes rate-limiting. The Escherichia coli alkaline phosphatase, which supposedly has a greater affinity for AMP than the matrix vesicle alkaline phosphatase (28), ensures that the NTP:pyrophosphohydrolytic reaction is rate-limiting at all AMP concentrations. Fig. 3 and Table I show that the final concentration of calcium precipitated when vesicles are incubated with AMP-PCP exceeds that when ATP is present by a factor of 1.4 to 1.6. In the presence of vesicles, PPi concentrations may never be expected to reach methylene diphosphonate levels because of the competing ATPase and pyrophosphatase reactions. Differences in the final yields of precipitable Ca'+ when ATP and AMP-PCP are the substrates suggests that the extent of Ca2+ deposition depends on the magnitude of PPi formation. Support for this concept comes from Ca2+ precipitation studies in the absence of ATP and matrix vesicles (cf. Fig. 5). Since the extent of precipitation never becomes PPi-independent throughout the range of concentrations studied, it suggests that the sediments contain stoichiometric quantities of Ca2+, Pi, and PPi. The idea of a Ca2+-PP, or Ca2+-PPi-Pi nucleation complex which would induce further calcium phosphate deposition without the need for further PPi seems unlikely.
There is some evidence that this in vitro mechanism may be important physiologically. The presence of adenine nucleotides in the matrix has been confirmed by several investigators (29,30) and the concentrations reported are in the range of ATP concentrations used in the present study. However, more recent evidence indicates that the 254 nm absorbing material in the matrix of chick epiphyseal cartilage is mainly adenosine with very little AMP and ADP and no ATP (31). Analytical studies reveal increasing PPi levels progressing from the resting zone to the zone of calcification in epiphyseal cartilage (17). In addition, added PP, increases Ca'+ deposition 4-fold in slices of embryonic chicken femur that normally calcify (16). The role of PPI as an inhibitor of amorphous calcium phosphate + hydroxyapatite transformation and subsequent crystal growth is a well-established fact (15,32). The matrix vesicles possess an active pyrophosphatase (33) and it may be responsible for removal of excess PPI after initial precipitation of an amorphous Ca'+ form. Regulatory mechanisms may exist for both NTP:pyrophosphohydrolase and pyrophosphatase activities allowing for buildup of PPi prior to precipitation and hydrolysis of PPi after phase separation occurs.
The matrix vesicle NTP:pyrophosphohydrolase hydrolyzes the &y-methylene analogs of pyrimidine nucleoside triphosphate more rapidly than the corresponding purine derivatives. NTP:pyrophosphohydrolase from the plasma membrane of the liver cell hydrolyzes ATP, CTP, GTP, UTP, and dATP at approximately equal rates (34). p-Nitrophenylphosphorylmethylenediphosphonate,2 wherein the p-nitrophenyl residue replaces the nucleoside moiety, is not a substrate for the matrix vesicle enzyme. Rate differences with different substrates are highly correlated with the reciprocal induction time of calcification as well as the ensuing intermediate stage calcification rate (cf. Table 111). The high correlation between reciprocal induction time and intermediate stage calcification, illustrated in Table 111, provides further evidence that the substance formed in each of these stages which is required for in vitro calcification is identical.
The pH optimum of 10.0 for matrix vesicle NTP:pyrophosphohydrolase is similar to that for the purified matrix vesicle alkaline phosphatase-associated ATPase ( 5 ) and the hepatocyte plasma-membrane NTP:pyrophosphohydrolase (34). The activity at pH 7.5 is 2% of the activity at the maximum, comparable to what is observed for the purified vesicle ATPase (5). Measurements of the rate of AMP-PCP hydrolysis at pH 7.5 and 10.0 at different substrate concentrations suggest that 0.234 mM is probably saturating over most of the pH range studied and that values recorded in Table IV are values of the maximum velocity. Substitution of the HC03-/ C03'-buffer system for the Caps/NaOH buffer at pH 10.5 indicates that bivalent anions may be effective inhibitors of the enzyme.
Matrix vesicle NTP:pyrophosphohydroIase is acid-labile and stable in mildly alkaline media, similar to vesicle alkaline phosphatase.
In contrast to these pH effects, the behavior of matrix vesicle NTP:pyrophosphohydrolase in the presence of Ca2+ and Mg2+ ions differs notably from that observed for the rat liver enzyme (34,35).
Matrix vesicle NTP:pyrophosphohydrolase activity is unaffected by either M$+ or Ca'+ ions. At 10 mM free Ca2+ and M P , the fraction of total nucleotide existing as a 1:l metal nucleotide complex is 0.98 and 0.99, respectively. At 1 PM free Ca'+ and Mg'+, this fraction is 0.005 and 0.013, respectively. These results suggest that either the unbound nucleotide or its 1:1 metal complex of Ca2+ or M$+ Reaction mixtures contained 120 mM NaCI, 10 mM KC1, 12 mM buffer, and 0.3 unit of Escherichia coli alkaline phosphatase/ml. Between pH 6.0 and 8.0, the matrix vesicle protein concentration was 46.6 pg/ml equivalent to 0.25 alkaline phosphatase unit/ml. At pH 8.5 and above, the concentration of matrix vesicle protein was 4.66 pg/ml. Reactions were carried out at 25 "C and were initiated by the addition of AMP-PCP to 0.234 mM. Duplicate aliquots were taken at zero time and at three additional times and analyzed for P,.  (35). Purified matrix vesicle ATPase (5) shows no effect of Ca2+ on activity between 0.1 and 10 mM, while M$+ gives 50% stimulation at 0.1 mM M$+ compared to the values observed a t 1.0 mM M e .
Matrix vesicle NTP:pyrophosphohydrolase is fully saturated a t 10 p~ AMP-PCP at pH 7.5. Kinetic considerations in the coupled assay used in these experiments prevented an accurate determination of K,. However, the low K,,, is comparable to what is observed for the purified matrix vesicle ATPase for which K,,, is 2 pM at pH 7.5 ( 5 ) .
All of the examined kinetic properties of matrix vesicle NTP:pyrophosphohydrolase are similar or identical with those of the purified ATPase, which by virtue of solubilization by deoxycholate followed by 1-butanol extraction is devoid of NTP:pyrophosphohydrolase activity. This suggests that the alkaline phosphatase may also possess, in addition to phosphatase, pyrophosphatase and ATPase activities, a pyrophosphohydrolase action. The expression of this activity may be related to the presence of an additional subunit which is sensitive to 1-butanol extraction. Attempts are under way at the present time to separate NTP:pyrophosphohydrolase activity from the associated activities of the alkaline phosphatase.
In conclusion, the present study demonstrates the existence of NTP:pyrophosphohydrolase activity in matrix vesicles isolated from bovine epiphyseal cartilage. This enzyme is directly responsible for the formation of PPi which in the presence of metastable Ca'+ x Pi solutions acts as an inducing agent for calcium phosphate precipitation.
This report affirms the calcification theories of Cartier and Picard (36) and Perkins and Walker (37) who discovered ATPpyrophosphohydrolase activity in extracts from calcifying cartilage of sheep and rachitic rats, respectively.