Quaternary Structure and Metal Ion Requirement of Family II Pyrophosphatases from Bacillus subtilis,Streptococcus gordonii, and Streptococcus mutans *

Pyrophosphatase (PPase) from Bacillus subtilis has recently been found to be the first example of a family II soluble PPase with a unique requirement for Mn2+. In the present work, we cloned and overexpressed in Escherichia coli putative genes for two more family II PPases (fromStreptococcus mutans and Streptococcus gordonii), isolated the recombinant proteins, and showed them to be highly specific and active PPases (catalytic constants of 1700–3300 s− 1 at 25 °C in comparison with 200–400 s− 1 for family I). All three family II PPases were found to be dimeric manganese metalloenzymes, dissociating into much less active monomers upon removal of Mn2+. The dimers were found to have one high affinity manganese-specific site (K d of 0.2–3 nm for Mn2+and 10–80 μm for Mg2+) and two or three moderate affinity sites (K d ∼ 1 mmfor both cations) per subunit. Mn2+ binding to the high affinity site, which occurs with a half-time of less than 10 s at 1.5 mm Mn2+, dramatically shifts the monomer ↔ dimer equilibrium in the direction of the dimer, further activates the dimer, and allows substantial activity (60–180 s− 1) against calcium pyrophosphate, a potent inhibitor of family I PPases.

Inorganic pyrophosphatase (EC 3.6.1.1; PPase) 1 catalyzes specifically interconversion of pyrophosphate and orthophosphate. Owing to the simplicity of its substrate, PPase is a convenient system to study the mechanism of phosphoryl transfer from polyphosphates, including nucleoside triphosphates, to water, an essential but still incompletely understood biochemical transformation. Soluble PPase is essential for life (1,2), since it provides a thermodynamic pull for biosynthetic reactions (3).
Until recently, only one family (family I) of soluble PPases had been known, of which the PPases of Saccharomyces cerevi-siae and Escherichia coli are the most extensively characterized representatives (4,5). Despite variability of subunit size and of quaternary structure (eukaryotic PPases are dimers of 30 -35 kDa subunits, whereas prokaryotic PPases are hexamers of ϳ 20 kDa subunits), PPases of family I have a highly conserved active site structure formed by 14 -16 amino acid residues and 3-4 Mg 2ϩ ions and very similar catalytic properties. Catalysis by these enzymes proceeds via direct attack of water on a phosphorus atom without formation of a covalent intermediate. The metal ions are the key to catalysis and mediate the major protein-PP i interactions, which serve to shield the charge on the electrophilic phosphorus, activate the nucleophilic water molecule, and increase the acidity of the leaving phosphate group (6 -8).
Recently, a long-known PPase of Bacillus subtilis (9) was found to have a completely different amino acid sequence and therefore to belong to a different family (family II) of soluble PPases (10,11). B. subtilis PPase is similar to family I PPases in requiring divalent metal ions for activity (9), but unlike bacterial family I PPases, it is formed by large subunits (34 kDa) and displays 10 -20 times greater activity (9,12). Another unique property of B. subtilis PPase is its preference for Mn 2ϩ over Mg 2ϩ as the activator. A search through GenBank TM revealed four more putative prokaryotic members of family II (two streptococcal and two archeal), showing 40 -57% identity in amino acid sequence (10,11). One of them (from Methanococcus jannaschii) has been recently cloned and expressed in E. coli (13). The catalytic mechanism employed by family II PPases and the structural basis for their remarkable activity remain to be elucidated.
In this work, we cloned and overexpressed in E. coli the putative genes for family II PPases from Streptococcus mutans, implicated together with dietary sugars as the principal cause in the development of dental caries (14), and Streptococcus gordonii, another human oral bacterium. The recombinant proteins were purified and shown, along with the B. subtilis PPase, to be highly active dimeric PPases with a unique requirement for Mn 2ϩ . umns were from Amersham Pharmacia Biotech.
E. coli XL2blue b (Stratagene) and E. coli C43(DE3) (15) were used as hosts in the cloning and expression, respectively. S. gordonii ATCC 10558 and S. mutans ATCC 25175 were obtained from the Culture Collection of the University of Göteborg (Sweden) and the Institute of Dentistry (University of Turku, Finland), respectively. The E. coli strains were grown in 2ϫ YT broth or on LA plates (16). Ampicillin (100 g/ml) was added when required.
DNA Manipulation and Protein Expression-Genes for Sg-PPase and Sm-PPase were expressed in E. coli under the inducible phage T7 promoter by making use of the pET system. Chromosomal DNA was isolated from both streptococcal strains as described by Ushiro et al. (17). The open reading frames encoding Sg-PPase and Sm-PPase were amplified by PCR using a 5Ј-sense oligonucleotide primer containing a restriction site for NcoI and a 3Ј-reverse complement primer with a BamHI site. The PCR product was purified by using the Wizard DNA Clean-up system, digested with NcoI and BamHI, and ligated into the vector pET-15b. In creating the NcoI site, a T 3 G mutation after the initiation codon (ATG) was made both in S. gordonii and S. mutans PPase genes. After ligating the PCR products into the vector, the mutations were reversed by using the QuikChange site-directed mutagenesis kit, and the produced DNA constructions were transformed into E. coli XL2blue b and E. coli C43(DE3) for DNA sequencing and expression, respectively. The expression was induced for 5-6 h by 1 mM isopropyl-1-thio-␤-D-galactopyranoside. B-PPase was expressed in E. coli as described by Shintani et al. (11).
Protein Purification-Sg-PPase and Sm-PPase were purified to homogeneity by DEAE fast flow ion exchange column chromatography in 20 mM Tris/HCl, pH 7.3, 10 mM MgCl 2 , 10 mM MnCl 2 followed by gel filtration on Superdex 200 prep grade in 0.15 M Tris/HCl, pH 7.2, 15 mM MgCl 2 , 1.5 mM MnCl 2 . B-PPase was additionally chromatographed on a phenyl-Sepharose column that was equilibrated with 1.7 M (NH 4 ) 2 SO 4 in the buffer used for the ion exchange chromatography and eluted with a downward gradient of (NH 4 ) 2 SO 4 concentration. Enzyme purity was checked by electrophoresis in 8 -25% gradient polyacrylamide gels in the presence of 0.55% SDS using the Phast System (Amersham Pharmacia Biotech).
Metal ions were removed from the enzyme stocks (30 -50 mg/ml) as follows. The enzyme solution was diluted 20-fold with 83 mM TES/KOH buffer, pH 7.2, containing 2 mM EDTA and 17 mM KCl, incubated for 2 days (B-PPase) or 4 days (Sg-PPase and Sm-PPase) at 4°C and subjected to two 40-fold dilution/reconcentration cycles in a Centricon YM-30 centrifugal filter device (Amicon) using 83 mM TES/KOH buffer, pH 7.2, containing 50 M EGTA and 17 mM KCl. Stock PPase solutions and other solutions used in experiments involving long incubations were sterilized by passing them through a 0.2-m filter.
Concentrations of B-PPase, Sg-PPase, and Sm-PPase were determined on the basis of extinction coefficients ⑀ 280 0.1% of 0.264, 0.343, and 0.307, calculated from the amino acid composition using the program ProtParam and the subunit molecular masses of 34.0, 33.5 and 33.4 kDa, respectively, calculated from the amino acid sequences (11). The above value of ⑀ 280 0.1% for B-PPase was confirmed by direct measurement of the absorbance of the solution prepared from dried and weighed enzyme. The Bradford method (18), standardized against the above method, was sometimes employed as an alternative to direct measurement of the absorbance.
Activity-Rates of PP i hydrolysis were determined from continuous recordings of P i liberation obtained using an automatic P i analyzer (19). Reactions were initiated by adding enzyme.
Sedimentation-Analytical ultracentrifugation was carried out in a Spinco E instrument (Beckman) with scanning at 280 nm. Sedimentation velocity was measured at 48,000 -60,000 rpm, and the sedimentation coefficient, s 20,w , was calculated using a standard procedure (20). Sedimentation equilibrium was attained at 16,000 rpm for 16 h or at 24,000 rpm for 10 h, and the molecular mass was calculated according to Chernyak and Magretova (21). The partial specific volume at 25°C was calculated from the amino acid composition and found to be 0.735, 0.729, and 0.730 cm 3 /g for B-PPase, Sg-PPase, and Sm-PPase, respectively.
Equilibrium Dialysis-Mg 2ϩ and Mn 2ϩ binding was assayed by equilibrium microdialysis in combination with atomic absorption spectroscopy to measure Mg 2ϩ and Mn 2ϩ content in the dialysis chambers (22).
Except where noted, all activity and binding measurements were performed at 25°C in the medium containing 83 mM TES/KOH buffer, pH 7.2, 50 M EGTA, and 17 mM KCl. For incubations with Mg 2ϩ , EGTA concentrations were increased to 0.5 mM. Bovine serum albumin was added at a concentration of 1 mg/ml to all incubation media, except for those used in sedimentation and equilibrium dialysis.
Calculations-Equations 1 and 2, derived from Scheme I, describe the activity (A) of an equilibrium mixture of dimer (D) and monomer (M) as a function of enzyme concentration at a zero or fixed concentration of divalent metal ion. A D and A M are the specific activities of dimer and monomer, respectively, [E] t is the total enzyme concentration, expressed in monomers, ␣ D is the fraction of dimeric enzyme, k d and k a are the rate constants, and K d ϭ k d /k a is the dissociation constant.
The Mn 2ϩ concentration dependence of the equilibrium activity could be described by Equations 1 and 2 in combination with  where A M , A D Ј, and A D Љ are the activities of monomer, metal-free dimer, and metal-bound dimer, respectively, K d,0 is K d at zero Mn 2ϩ concentration, and K M1 is the dissociation constant governing Mn 2ϩ binding to dimer. Equation 5 is an implicit extended mass balance equation for metal. The second and third terms on the right side of Equation 5 correspond to protein-bound and EGTA-bound metal, respectively, [EGTA] t is the total concentration of EGTA in the system (50 M), and K MEGTA is the dissociation constant for its complex with Mn 2ϩ . This treatment implies that only dimer can bind Mn 2ϩ , which has a dual effect on activity;Mn 2ϩ both increases the amount of dimer (by decreasing K d ) and further activates it (by increasing A d ). Equations 1 and 2 or 1-5 were simultaneously fit to data with the program SCIENTIST (MicroMath).
The dissociation constants for Mn-EDTA and Mn-EGTA complexes at pH 7.2 (0.013 and 6.3 nM, respectively) used to estimate the concentrations of free Mn 2ϩ in solutions containing EGTA and EDTA were calculated from the stability constants of their deprotonated Mn 2ϩ complexes taking into account the pK a values of EGTA and EDTA (23).

Cloning and Expression of the Streptococcal PPase Genes-
The putative open reading frames of Sg-PPase and Sm-PPase were amplified by PCR. By sequencing the genes we noticed Sg-PPase to have four differences from the gene-deduced amino acid sequences found from GenBank TM ; positions 109, 135, 137, and 166 are occupied by Ser, Gly, Pro, and Val (the corresponding codons are AGT, GGC, CCA, and GTC), respectively, rather than by Asn, Ser, Ser, and Ala, as found in GenBank TM . These differences were reproduced in two independent PCR amplifications, ruling out possible mutations during gene manipulations.
The PCR products expressed under phage T7 promoter in E. coli yielded transformants with about 100-fold higher PPase activity than the host strain, clearly indicating that the two open reading frames encode PPases. Because of the high expression level, the recombinant enzymes were easily purified to homogeneity by ion exchange and gel filtration chromatography, the first of these steps readily separating the chromosomeencoded E. coli PPase from the streptococcal PPases. From 50 to 100 mg of pure enzymes were obtained from 1 liter of cell culture, corresponding to about 5 g of cell paste.
Quaternary Structure-The sedimentation coefficients measured for the B. subtilis and the two streptococcal PPases in the presence of Mg 2ϩ or Mn 2ϩ by the sedimentation velocity method were within 3.5-4.1 S (Table I). In the absence of the divalent metal ions (2 mM EDTA present), s 20,w decreased markedly for B-PPase and less markedly for Sg-PPase and Sm-PPase, indicating dissociation into lower molecular mass species.
The sedimentation equilibrium method gave molecular masses of 63 Ϯ 3 kDa for B-PPase and 68 Ϯ 5 kDa for Sg-PPase in the presence of 1.5 mM MnCl 2 ( Fig. 1) and 32 Ϯ 2 kDa for B-PPase in the presence of 2 mM EDTA. The molecular masses of B-PPase and Sg-PPase dimers predicted from their amino acid sequences (11) are 68.0 and 67.1 kDa, respectively. These proteins are therefore clearly dimers in the presence of MnCl 2 , but B-PPase dissociates into monomers in the absence of the metal ions. The assumption of a monomeric or trimeric structure in the presence of MnCl 2 resulted in a poor fit (Fig. 1).
Activity Versus Enzyme Concentration Profiles-The equilibrium between different oligomeric forms of enzyme can be studied by measuring its specific activity as a function enzyme concentration if the specific activities of the oligomeric forms differ and their interconversion is slow on the time scale of enzyme assay (24). Earlier, we used this approach in studies of E. coli PPase variants with weakened quaternary structure (25)(26)(27)(28). The specific activities of the three PPases under study increased with increasing enzyme concentration in the stock solution containing 2 mM EDTA (no free divalent metal ion) or 1.5 mM Mg 2ϩ and remained unchanged if the stock solution contained 1.5 mM Mn 2ϩ (Fig. 2). The activity values shown, except for the Sm-PPase-EDTA curve (see below), correspond to equilibrium, as indicated by comparison with similar curves obtained after 1 and 2 days of incubation (not shown; see also Fig. 4 below). Three sources of evidence indicated that the profiles for B-PPase and Sg-PPase incubated with EDTA and the profiles for all three PPases incubated with Mg 2ϩ in Fig. 2 describe a monomer-dimer equilibrium. First, they obeyed Equations 1 and 2 derived for such an equilibrium. Second, the enzymes preincubated at low concentration could be completely reactivated upon the addition of 3.5 mM MnCl 2 , as illustrated in Fig. 3 for Sg-PPase. Moreover, the reactivation rate increased with increasing enzyme concentration, exhibiting half-times of 14 and 1.4 min at enzyme concentrations of 1.8 and 18 nM, respectively, as expected for the second-order reaction of dimer formation from monomers. By contrast, the half-times would be similar for a first-order reaction, involving only one molecule of the reactant. Third, these findings agree with the sedimentation data above, showing that s 20,w is greater in the presence of Mn 2ϩ or Mg 2ϩ than in the presence of EDTA at 20 -26 M enzyme concentration (Table I).
The activity of Sm-PPase preincubated with EDTA was only partially restored upon the addition of MnCl 2 , indicating irreversible inactivation during the incubation. Interestingly, the inactivation was significant only at [E] t Ͼ 50 M and increased with increasing [E] t , suggesting the occurrence of enzyme aggregation. This explanation is supported by the observation that Sm-PPase, but not B-PPase or Sg-PPase, could be precipitated by heating its solution above 30°C.
Three immediate inferences can be made from the data in Figs. 2 and 3 and the parameter values listed in Table II. First, dimers are more active than monomers. Second, Mg 2ϩ and, especially, Mn 2ϩ stabilize dimer versus monomer (compare the K d values), primarily due to a change in k a . Dimer stability is  Table II.  Fig. 4 shows the time courses of activity upon dilution of the EDTA-treated PPases into media containing EDTA, Mg 2ϩ , or Mn 2ϩ . In all cases EDTA stimulated inactivation, and Mn 2ϩ stimulated activation, whereas the effect of Mg 2ϩ was variable and much smaller. The inactivation by Mg 2ϩ and EDTA proceeded on the time scale of hours, whereas the activation by Mn 2ϩ was complete in 30 s, except for a small slower phase seen with Sm-PPase. Importantly, the enzymes were predominantly dimeric at the start of incubation (Fig. 2); therefore the major activation is not due to stimulation of dimer formation, except for the slower phase seen with Sm-PPase (Fig. 4). By contrast, the inactivation clearly results from dimer dissociation into monomers because the data in Fig. 2 indicated that the PPases equilibrated with EDTA or Mg 2ϩ at the concentrations used for Fig. 4 represent a mixture of dimer and monomer. A similar pattern of inactivation by EDTA was reported for wild-type B-PPase (12). Values for k d (Scheme I) estimated from the data in Fig. 4 are summarized in Table II.
The Mn 2ϩ concentration dependence of the equilibrium activity (Fig. 5) could be described by assuming that Mn 2ϩ selectively binds to dimer (with a dissociation constant of K M1 ), thus shifting the equilibrium in Scheme I in the direction of dimer, and further activates it. Estimates of K M1 (Table III) were obtained by fitting Equations 1-5 to the data in Fig. 5. Values of A D Љ and K M1 were treated as adjustable parameters in these fittings, and values of A M , A D Ј. and K d,0 were constrained to the values of A M , A D , and K d determined above in the presence of EDTA (Table II). For Sm-PPase, no K d,0 and A D Ј values are available (see above); therefore, K d,0 was set to zero, and A D Ј was also treated as an adjustable parameter, allowing estimation of the upper limit for K M1 . Remarkably, the values of K M1 estimated for all three PPases are extremely low (Table III), characteristic of metalloenzymes.
Direct Measurements of Mn 2ϩ and Mg 2ϩ Binding-As measured by equilibrium dialysis, dimeric Sg-PPase has one high affinity and two or three low affinity sites for Mn 2ϩ and Mg 2ϩ per subunit (Fig. 6). The binding curves for both cations exhibited saturation, not clearly seen in Fig. 6 because of the loga-rithmic scaling of the metal concentration axis. The high affinity site demonstrated a marked preference for Mn 2ϩ over Mg 2ϩ , whereas the low affinity sites bound Mn 2ϩ with only slightly greater strength than they bound Mg 2ϩ . Very similar binding curves were obtained for B-PPase and Sm-PPase (not shown). The dissociation constant K M1,obs characterizing metal binding to the high affinity site was estimated from these data using only points measured at Ͻ1 M for Mn 2ϩ and Ͻ100 M for Mg 2ϩ (Table III). The values of K M1,obs thus obtained did not vary much between the three PPases, and the values of K M1,obs for Mn 2ϩ binding to B-PPase and Sg-PPase were slightly greater than the corresponding values of K M1 determined above. This difference appears to result from the presence of appreciable amounts of monomer, which lacks the high affinity site possessed by the dimer, at low metal ion concentrations. For the same reason, values of K M1,obs for Mg 2ϩ shown in Table III may also exceed the corresponding K M1 values for this cation.
In the presence of 50 M Mn 2ϩ , no high affinity site was seen for Mg 2ϩ , and the total binding stoichiometry decreased by one in Sg-PPase (Fig. 7) and B-PPase (not shown).This is consistent with Mn 2ϩ and Mg 2ϩ competing for the same M1 site, which binds Mn 2ϩ more tightly than Mg 2ϩ by 4 orders of magnitude (Table III). Mg 2ϩ binding to the low affinity sites changed insignificantly in the presence of 50 M Mn 2ϩ (Fig. 7).
Metal Ion Requirements for Activity-No hydrolytic activity was observed when the family II PPases were assayed in the absence of divalent metal ions. The activity was maximal when Mn 2ϩ was present in the preincubation or assay medium or in both of these media (Fig. 8, curves Mn/Mn,Mg, Mg/Mn,Mg, Mn/Mn, Mn/Mg). However, if Mn 2ϩ was present only during preincubation, the activity gradually decreased during the assay, resulting in a nonlinear P i production curve (curve Mn/ Mg). The latter effect clearly resulted from displacement of Mn 2ϩ from the high affinity site by Mg 2ϩ . That the curve Mg/Mn,Mg was linear and had the maximal slope suggested that the back substitution occurs quite rapidly during catalysis.
The Michaelis-Menten parameters for the Mg 2ϩ -activated PP i hydrolysis were determined for dimeric and monomeric PPases obtained by preincubations identical to those used for Fig. 2. The preincubations with Mn 2ϩ or Mg 2ϩ at the enzyme concentrations indicated in Table IV yielded dimer, whereas the preincubations with EDTA (no divalent metal ion present) yielded a mixture of monomer and dimer, with monomer being the dominant species. Fitting of rate versus substrate concentration profiles (not shown) to the sum of two Michaelis-Menten equations, one for monomer and the other for dimer, yielded k cat and K m values for monomers and K m values for dimers. The results presented in Table IV indicate that preincubation with Mn 2ϩ renders dimeric PPases better catalysts in terms of k cat and worse catalysts in terms of K m than preincubation with Mg 2ϩ or without divalent metal ions.
Ca 2ϩ is known to strongly inhibit family I PPases by replacing the activating Mg 2ϩ in its complex with enzyme and substrate (29). Typically, activity drops by 50% at [Ca 2ϩ ]/[Mg 2ϩ ] Ϸ 1/50. By contrast, no inhibition (Յ15%) was observed when the family II PPases were assayed in the presence of 2-10 mM Ca 2ϩ (20 mM Mg 2ϩ , 57 M total PP i ). Moreover, Ca 2ϩ was able to activate family II PPases preincubated with Mn 2ϩ (compare curves Mn/Ca and Mn/none in Fig. 8). Although the activities measured with 0.5 mM Ca 2ϩ (64 s Ϫ1 with B-PPase, 135 s Ϫ1 with Sg-PPase, and 160 s Ϫ1 with Sm-PPase) were only about 10% of the activities conferred by Mn 2ϩ , they compare well with the activities of family I PPases measured with their best activator, Mg 2ϩ . However, Ca 2ϩ did not support PP i hydrolysis by Sg-PPase and Sm-PPase, from which Mn 2ϩ was removed by pre- incubation with EDTA (activities less than 0.2 and 0.004 s Ϫ1 , respectively).
Substrate Specificity-Dimeric PPases preincubated with 1.5 mM Mn 2ϩ and assayed in the presence of 0.17-2 mM Mn 2ϩ or 20 mM Mg 2ϩ plus 0.07 mM Mn 2ϩ exhibited low but measurable activity against tripolyphosphate (0.03-0.06 s Ϫ1 at 57 M substrate) but not against ATP (Ͻ0.004 s Ϫ1 ) (data not shown).

Family II of Soluble PPases-A new family of soluble PPases
(family II) has recently been observed by two research groups (10,11). The first verified member of family II PPases was B. subtilis PPase, but in addition, amino acid sequences of four putative members of this family were found in the GenBank TM , two streptococcal and two archeal (11). The results shown above indicate that the two streptococcal sequences do belong to highly efficient and specific PPases. Furthermore, the open reading frames encoding the putative family II PPases of M. jannaschii (13) and Archaeoglobus fulgidus 2 have recently been expressed in E. coli and also shown to be PPases. Table V summarizes the major properties distinguishing families I and II of the bacterial PPases. Kinetically, family II PPases are superior in terms of k cat , but their full activity can hardly be manifested in vivo because of high K m values. PP i strongly inhibits the biosynthesis pathways, which produce it as a by-product, and for this reason it is unlikely to be accumulated to the levels comparable with K m . As a result of high K m , the hydrolysis rate will be more sensitive to PP i concentration, perhaps allowing for a better control of the level of this important metabolite. Knowledge of free PP i concentration in cells possessing family II PPases is required to address this point more specifically. B-PPase was initially reported to be   Table II.

FIG. 5. Mn 2؉ concentration dependence of PPase activation.
The enzymes (20 -26 M) were pre-equilibrated with Mn 2ϩ for 1 day (B-PPase) or 4 days (Sg-PPase and Sm-PPase) before activity was assayed as in Fig. 2. The upper scale refers to free Mn 2ϩ concentration in the medium before PPase was added and, therefore, neglects Mn 2ϩ binding by PPase. The lines are drawn according to Equations 1-5 using the A M , K d , and K M1 values found in Tables II and III (see "Results").  Fig. 5. b Estimated from Fig. 6 for Sg-PPase and similar dependencies for other PPases using only the data measured at Ͻ1 and Ͻ100 M free Mn 2ϩ and Mg 2ϩ concentrations, respectively.
insensitive to fluoride (10), a well known inhibitor of family I PPases, but more recent studies (13) have shown both families to be fluoride-sensitive.
A search through GenBank TM using B. subtilis PPase as a template indicated 11 more verified and 3 unverified fulllength putative family II PPase sequences. The amino acid sequences of Thermotoga maritima, Clostridium difficile, and Geobacter sulfurreducens putative PPases include a specific 230-residue-long insertion between residues Glu-67 and Val-68 (B-PPase numbering), indicating that their subunit size is significantly larger than those of the other family II PPases (60 kDa versus 33-34 kDa; Table V). However, the conservation pattern of the amino acid residues strongly suggests that all 14 of these sequences represent family II PPases. Interestingly, Vibrio cholerae appears to be the first example of a species having genes for PPases from both families. According to subunit size and dependence on Mn 2ϩ , the PPase of Bacillus megaterium, described by Tono and Kornberg (30), may also belong to family II. So far, all family II PPase sequences belong to bacteria. Quaternary Structure and Its Role in Catalysis-Unlike the bacterial PPases of family I, which are hexamers, the three PPases studied in this work are dimers, dissociating into monomers at low enzyme concentration in the absence of Mn 2ϩ . The dimeric form, which is expected to be dominant under physiological conditions (Mn 2ϩ present), is much more active. The latter fact is at variance with the data of Kuhn et al., who report B-PPase (12) and M. jannaschii PPase (13) to exist as an inactive dimer and active trimer. It should be noted that their molecular mass estimates were obtained by gel filtration, a method more prone to error than the sedimentation analysis used in the present study. The dimeric structure of Sm-PPase has recently been confirmed by x-ray crystallography (31) The metal-free dimer is more active than the monomer due to increased k cat and decreased K m values (Table IV). With family I PPase from E. coli, dissociation of the hexamer to trimers and dimers increases K m without affecting k cat (27,28). This result was interpreted as evidence that hexamer formation and substrate binding both induce a catalytically optimal structure for the active site. With family II PPases, substrate binding is not sufficient to produce a catalytically optimal structure, as indicated by decreased k cat .
Role of Mn 2ϩ -Family I PPases can utilize both Mg 2ϩ and Mn 2ϩ as cofactors; however, Mg 2ϩ is more efficient in all cases. With family II PPases, the efficiency of these cations as cofactors is reversed (Table IV), and the available data indicate that this difference is due to a unique site, which binds Mn 2ϩ with an affinity characteristic of metalloenzymes. This site is quite specific for Mn 2ϩ versus Mg 2ϩ ; the ratio of the respective K M1,obs values is 5,000 -23,000 (Table III). Mn 2ϩ binding to this site controls activity in three ways. First, it dramatically shifts the monomer 7 dimer equilibrium in the direction of the more active dimer (K d changes more than 10 5 -fold; Table II). A similar but smaller effect is exerted by Mg 2ϩ (K d changes 1,000 -3,600-fold). Kuhn and Ward (12) also observed an effect of Mn 2ϩ and Co 2ϩ on the quaternary structure of B-PPase, consistent with metal ion binding to the high affinity site, but interpreted this effect in terms of a dimer-trimer equilibrium. Second, Mn 2ϩ changes the kinetic parameters k cat and K m for the dimer (Table IV). Interestingly, both k cat and K m are increased in the presence of Mn 2ϩ , but the effect on k cat is larger. This means that switching from Mg 2ϩ to Mn 2ϩ at the high affinity site would always activate the enzymes, but the effect would be larger at higher substrate concentration. The highaffinity site of the dimer obtained in the absence of metal ions apparently binds Mg 2ϩ from the assay medium, which explains the similarity of the kinetic parameters for this dimer and the dimer obtained in the presence of Mg 2ϩ (Table IV). Finally, Mn 2ϩ bound to the high affinity site allows substantial activity with Ca 2ϩ (Fig. 8). This means that CaPP i is a reasonably good  substrate for family II PPases. By contrast, CaPP i is a strong, nonhydrolyzable inhibitor of family I PPases (29,32,33). At the low affinity sites and in the PP i complex (true substrate), Mn 2ϩ and Mg 2ϩ appear to be equally effective in catalysis, as indicated by similar activities of Mn 2ϩ -pretreated Sg-PPase (i.e. containing Mn 2ϩ at the high affinity site) with Mn 2ϩ and Mg 2ϩ as activators (curves Mn/Mn and Mn/Mg in Fig. 8). Kuhn and Ward (12) report that Co 2ϩ can activate B-PPase 70% as much as Mn 2ϩ by binding to the high affinity site, but no activation was observed with other transition raw and alkali earth metal ions. Recent x-ray crystallographic studies (31) indicate a twodomain structure for each subunit of dimeric Sm-PPase, the C-terminal domain being quite flexible. The active site was located at the domain interface and contained two proteinbound metal ions. One had three Asp and one His, and the other had two Asp and one His side chains as ligands. The four-ligand site apparently corresponds to the high affinity site and the three-ligand site to one of the low affinity sites detected in the present study. The other low affinity site(s) appear(s) to have low occupancy in the crystals. In terms of the threedimensional structure, the effects of Mn 2ϩ and Mg 2ϩ on dimer stability may result from decreased domain flexibility caused by metal ion binding at the domain interface.
Cells of B. subtilis (34) and S. mutans (35) accumulate large amounts of Mn 2ϩ . Hence this cation appears to be a physiological ligand of family II PPases, at least at the high affinity site, and the effects of Mn 2ϩ reported above may well be involved in PPase activity regulation in vivo. S. mutans has been identified as the primary cause of dental caries (14) and manganese as a caries-promoting element (36) due to its stimulation of S. mutans growth (37). Keeping in mind that Mn 2ϩ decreases the suppressive effect of fluoride both on the activity of family II PPase (13) and on the growth of S. mutans (37), it is tempting to speculate that the action of Mn 2ϩ on S. mutans is due to its effect on Sm-PPase. In in vitro studies of family II PPases, one should keep in mind that the high affinity site may also bind other metal ions (13) with unpredictable consequences for quaternary structure and activity. Therefore parameters and fac-tors such as the concentration and nature of the added metal ion, enzyme concentration, and purity of reagents should be carefully controlled in the solutions used to store and assay these enzymes.