Novel Triphosphate Phosphohydrolase Activity of Clostridium thermocellum TTM, a Member of the Triphosphate Tunnel Metalloenzyme Superfamily*

Triphosphate tunnel metalloenzymes (TTMs) are a newly recognized superfamily of phosphotransferases defined by a unique active site residing within an eight-stranded β barrel. The prototypical members are the eukaryal metal-dependent RNA triphosphatases, which catalyze the initial step in mRNA capping. Little is known about the activities and substrate specificities of the scores of TTM homologs present in bacterial and archaeal proteomes, nearly all of which are annotated as adenylate cyclases. Here we have conducted a biochemical and structure-function analysis of a TTM protein (CthTTM) from the bacterium Clostridium thermocellum. CthTTM is a metal-dependent tripolyphosphatase and nucleoside triphosphatase; it is not an adenylate cyclase. We have identified 11 conserved amino acids in the tunnel that are critical for tripolyphosphatase and ATPase activity. The most salient findings are that (i) CthTTM is 150-fold more active in cleaving tripolyphosphate than ATP and (ii) the substrate specificity of CthTTM can be transformed by a single mutation (K8A) that abolishes tripolyphosphatase activity while strongly stimulating ATP hydrolysis. Our results underscore the plasticity of CthTTM substrate choice and suggest how novel specificities within the TTM superfamily might evolve through changes in the residues that line the tunnel walls.

cal property of this branch of the TTM superfamily is the ability to hydrolyze NTPs to nucleoside diphosphates and P i in the presence of manganese (2). Their defining primary structure features are two glutamate-containing motifs that are required for catalysis by every family member and which comprise the metal-binding site.
The crystal structure of the S. cerevisiae RNA triphosphatase Cet1 bound to manganese and sulfate (a proposed mimetic of the product complex with phosphate) revealed a then-novel tertiary structure in which the active site is situated within a topologically closed tunnel composed of eight antiparallel ␤ strands (18). Extensive mutational analyses of yeast Cet1 identified a large ensemble of side chains essential for triphosphatase activity in vitro and in vivo, many of which make direct or water-mediated contacts with the divalent cation or the sulfate anion (3). It was thought initially that the Cet1-like RNA triphosphatases arose de novo in unicellular eukarya in tandem with the emergence of caps as the defining feature of eukaryal mRNA. This notion has been eclipsed by the finding that the heretofore unique tertiary structure and active site of yeast RNA triphosphatase are recapitulated in the crystal structures of archaeal and bacterial proteins of unknown biochemical function, including proteins from Pyrococcus (Protein Data Bank (PDB) accession codes 1YEM and 2DC4), Vibrio (2ACA), and Nitrosomonas (2FBL) (1). The Cet1-like archaeal/bacterial proteins are usually annotated as belonging to the so-called CYTH family (19), which is defined by its two biochemically characterized founding members, an Aeromonas hydrophila adenylate cyclase CyaB and a mammalian thiamine triphosphatase (20,21). A crystal structure of a Cet1-like adenylate cyclase from Yersinia (2FJT) has been reported recently (22). Given that the Cet1 clade, CyaB, and thiamine triphosphatase are all metal-dependent enzymes that act on triphosphate-containing substrates, we concluded that the tunnel fold first described for Cet1 is the prototype of a larger enzyme superfamily, which we named "triphosphate tunnel metalloenzyme" (TTM) to reflect the defining structural features (1).
Although it is now clear that TTM proteins are distributed widely among bacterial, archaeal, and eukaryal taxa, virtually nothing is known about the presumptive enzymatic activities and substrate specificities of the dozens of TTM homologs present in bacterial and archaeal proteomes. Indeed, the majority of these TTM proteins are annotated as adenylate cyclases without any evidence that they have this activity.
Here we have presented a biochemical characterization and structure-function analysis of a 156-amino-acid TTM pro-tein from the bacterium Clostridium thermocellum (CthTTM). Although annotated as an adenylate cyclase (GenBank TM EAM47091), CthTTM had no detectable adenylate cyclase activity in our hands. Rather, CthTTM is a metal-dependent phosphohydrolase that specifically cleaves the ␤-␥ phosphoanhydride bond of tripolyphosphate and nucleoside triphosphates. We have determined the effects of 35 mutations of 14 amino acids within the putative triphosphate tunnel of CthTTM and thereby distinguished the functional groups required for global phosphohydrolase activity from ones selective for activity on nucleotide versus inorganic phosphoanhydride.

EXPERIMENTAL PROCEDURES
Recombinant CthTTM-The open reading frame encoding CthTTM was amplified from C. thermocellum genomic DNA (purchased from the American Type Culture Collection) with primers that introduced an NdeI site at the start codon and a BamHI site 3Ј of the stop codon. The PCR product was digested with NdeI and BamHI and inserted into pET16b to generate an expression plasmid encoding the CthTTM polypeptide fused to an N-terminal His 10 tag. Alanine and conservative substitution mutations were introduced by PCR using the two-stage overlap extension method (23). The inserts of all plasmids were sequenced to exclude the acquisition of unwanted coding changes during amplification or cloning. Wild-type and mutant pET-CthTTM plasmids were transformed into Escherichia coli BL21(DE3). Cultures (250 ml) were grown at 37°C in Luria-Bertani medium containing 0.1 mg/ml ampicillin until the A 600 reached ϳ0.6. The cultures were adjusted to 0.1 mM isopropyl-␤-D-thiogalactopyranoside and incubated at 37°C for 3 h. Cells were harvested by centrifugation, and the pellet was stored at Ϫ80°C. All subsequent procedures were performed at 4°C. Thawed bacteria were resuspended in 25 ml of buffer A (50 mM Tris-HCl, pH 7.5, 0.25 M NaCl, 10% sucrose). Lysozyme, phenylmethylsulfonyl fluoride, and Triton X-100 were added to final concentrations of 1 mg/ml, 0.5 mM, and 0.1%, respectively. The lysates were sonicated to reduce viscosity, and insoluble material was removed by centrifugation. The soluble extracts were applied to 1-ml columns of nickel-nitrilotriacetic acid-agarose (Qiagen) that had been equilibrated with buffer A. The columns were washed with 10 ml of the same buffer and then eluted stepwise with 4-ml aliquots of 50, 100, and 200 mM imidazole in buffer B (50 mM Tris-HCl, pH 8.0, 0.25 M NaCl, 10% glycerol, 0.05% Triton X-100). The polypeptide compositions of the column fractions were monitored by SDS-PAGE. The His 10 -CthTTM polypeptide was recovered predominantly in the 200 mM imidazole fraction. The 200 mM imidazole eluates were dialyzed against 50 mM NaCl in buffer C (50 mM Tris-HCl, pH 7.5, 10% glycerol, 0.05% Triton X-100, 1 mM EDTA, 1 mM dithiothreitol) and then applied to 1-ml columns of DEAE-Sephacel that had been equilibrated with 50 mM NaCl in buffer C. The columns were washed with 8 ml of the same buffer and then eluted stepwise with 4-ml aliquots of 100, 200, and 500 mM NaCl in buffer C. CthTTM was recovered in the flow-through. The proteins were stored at Ϫ80°C. Protein concentrations were determined by using the Bio-Rad dye reagent with bovine serum albumin as the standard. The yield of wild-type CthTTM was 10 mg from a 250-ml bacterial culture.
Nucleoside Triphosphatase Assay-Reaction mixtures containing (per 10 l) 50 mM Tris-HCl, pH 8.0, 2 mM MnCl 2 , and 100 M [␥-32 P]ATP or [␣-32 P]ATP (PerkinElmer Life Sciences), and CthTTM, as specified, were incubated for 30 min at 37°C. An aliquot (1.5 l) of the mixture was applied to a polyethyleneimine-cellulose TLC plate, which was developed with 0.5 M LiCl and 1 M formic acid. The radiolabeled material was visualized by autoradiography and 32 P i formation was quantified by scanning the TLC plate with a Fujix BAS2500 imager. Alternatively, reaction mixtures (50 l) containing 50 mM Tris-HCl, pH 8.0, 2 mM MnCl 2 , 100 M unlabeled NTP or dNTP, and CthTTM, as specified, were incubated for 30 min at 37°C. The reactions were quenched by adding 1 ml of malachite green reagent (BIOMOL Research Laboratories, Plymouth Meeting, PA). Phosphate release was determined by measuring A 620 and interpolating the value to a phosphate standard curve.
Tripolyphosphatase Assay-Reaction mixtures (50 l) containing 50 mM Tris-HCl, pH 9.0, 10 mM MgCl 2 or 0.5 mM MnCl 2 , 100 M inorganic tripolyphosphate (PPP i ) (Sigma), and CthTTM, as specified, were incubated for 30 min at 37°C. The reactions were quenched by adding 1 ml of malachite green reagent. Release of phosphate was determined by measuring A 620 and interpolating the value to a phosphate standard curve.
Glycerol Gradient Sedimentation-An aliquot (100 g) of CthTTM was mixed with catalase (50 g), bovine serum albumin (50 g), and cytochrome c (50 g). The mixture was applied to a 4.8-ml 15-30% glycerol gradient containing 50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 1 mM EDTA, 1 mM dithiothreitol, and 0.05% Triton X-100. The gradient was centrifuged at 50,000 revolutions/min for 17 h at 4°C in a Beckman SW55Ti rotor. Fractions (ϳ0.18 ml) were collected from the bottom of the tube. Aliquots of even-numbered fractions were analyzed by SDS-PAGE and assayed for nucleoside triphosphatase and tripolyphosphatase activities.

RESULTS
A TTM from C. thermocellum-The primary structure of the 156-amino-acid putative TTM from C. thermocellum (CthTTM) is aligned in Fig. 1A to a closely related TTM protein from Nitrosomonas europaea (NeuTTM; 151 amino acids) for which a crystal structure is available (PDB accession code 2FBL). The NeuTTM fold is distinguished by the fact that the eight-stranded antiparallel ␤ barrel is not a closed tunnel but rather a C-shaped "cup" that is prevented from closing by virtue of the insertion of a broken C-terminal helix into one end of the tunnel (Fig. 1B). The position of the C-terminal segment in the tunnel aperture appears to be stabilized by a hydrogen bond from a conserved lysine in the ␤1 strand of the tunnel (Lys-7 in NeuTTM; Lys-8 in CthTTM) to a backbone carbonyl at the break in the C-terminal helix (Fig. 1B). The primary structures of CthTTM and NeuTTM have 73 positions of side chain identity/similarity. Especially striking is the identity of the constellation of acidic and basic residues that project into the tunnel (Fig. 1B) and are the equivalents of the essential constituents of the Cet1 active site.
CthTTM Is a Nucleoside Triphosphate Phosphohydrolase-Reaction of purified CthTTM with 100 M [␣-32 P]ATP in the presence of a divalent cation resulted in quantitative conversion of the labeled substrate to ADP ( Fig. 2A). No labeled product corresponding to 3Ј,5Ј-cyclic AMP was detected. A parallel reaction of CthTTM with 100 M [␥ 32 P]ATP resulted in conversion of nearly all of the labeled material to inorganic phosphate ( Fig. 2A). The rate of release of 32 P i from [␥-32 P]ATP was identical to the rate of conversion of [␣-32 P]ATP to [␣-32 P]ADP in parallel reaction mixtures containing the same enzyme concentrations (Fig. 2B), indicating that CthTTM catalyzed the hydrolysis of ATP to ADP and P i . ATP hydrolysis by CthTTM was optimal at pH 8.0 -8.5 in Tris buffer and declined sharply as the pH was increased to 9.5 or decreased to Յ7.0 (Fig. 3A). No ATP hydrolysis was evident in the absence of a divalent cation ( Fig. 2A). ATPase activity increased with manganese concentration up to an optimum of 2.5 mM and declined as the concentration was increased to 5, 10, and 20 mM (Fig. 3B). Other divalent cations (magnesium, cobalt, nickel, cadmium, calcium, copper, and zinc) were unable to support ATPase activity at 2 mM concentration (data not shown). CthTTM activity displayed a hyperbolic dependence on ATP concentration (Fig. 3C). From a double reciprocal plot, we calculated a K m value of 26 M ATP and a k cat of 25 min Ϫ1 . NTP specificity was examined by colorimetric assay of the release of P i from unlabeled ribonucleotides ATP, GTP, CTP, or UTP and deoxynucleotides dATP, dGTP, dCTP, and dTTP. CthTTM was active with all eight nucleoside triphosphates (Fig. 3D).
Hydrolysis of Inorganic Tripolyphosphate Exceeds NTP Hydrolysis-CthTTM displayed vigorous activity in releasing P i from PPP i in the presence of a divalent cation cofactor, which could be either magnesium or manganese (Fig. 4A). The extent of P i formation was proportional to CthTTM concentration and saturated at a level corresponding to 1 mol of P i /mol of input PPP i substrate ( Fig. 4A and data not  APRIL 20, 2007 • VOLUME 282 • NUMBER 16

JOURNAL OF BIOLOGICAL CHEMISTRY 11943
shown). CthTTM failed to release P i from inorganic pyrophosphate (PP i ) in the presence of magnesium or manganese, even at enzyme concentrations sufficient for maximal P i release from tripolyphosphate (Fig. 4A). We conclude that CthTTM converts PPP i to PP i plus P i . Thus, CthTTM is dedicated to hydrolysis of the ␤-␥ phosphoanhydride linkage of triphosphate-containing substrates, whether inorganic or nucleoside-linked. The specific activity in magnesium-dependent tripolyphosphate cleavage corresponded to a turnover number of ϳ3800 min Ϫ1 . This value is 150-fold higher than CthTTM ATPase activity.
When various metals were tested at 2 mM concentration, magnesium supported the highest tripolyphosphatase activity, followed by manganese and cobalt, which were about one-half and one-forth as effective as magnesium, respectively (Fig. 4B). Nickel, copper, calcium, cadmium, and zinc were ineffective (Fig. 4B). Magnesium supported optimal activity at 10 -20 mM, whereas manganese-dependent activity was optimum at 0.31-0.62 mM and declined at higher concentrations (Fig. 4C). Tripolyphosphatase activity with magnesium or manganese was optimal at pH 9.0 -9.5 in Tris buffer (Fig. 4D). Activity was virtually nil at pH Յ6.5 (Fig. 4D).
CthTTM Is a Homodimer-The quaternary structure of CthTTM was gauged by zonal velocity sedimentation through a 15-30% glycerol gradient. Marker protein catalase (248 kDa), bovine serum albumin (66 kDa), and cytochrome c (13 kDa) were included as internal standards. CthTTM sedimented as a discrete peak in fractions 20 -22 overlapping the "light" side of the bovine serum albumin peak (Fig. 5). The ATP and tripolyphosphatase activity profiles peaked in fractions 20 -22 and coincided with an abundance of the CthTTM polypeptide (Fig. 5). A plot of the S values of the three standards versus fraction number yielded a straight line (data not shown). An S value of 3.6 was determined for CthTTM by interpolation to the internal standard curve. We surmise that CthTTM is a homodimer of the 156-amino-acid polypeptide. These native sizing results are consistent with crystallographic evidence that NeuTTM is a homodimer (2FBL), as are many other TTM proteins, including the founding TTM protein Cet1 (18, 24).
The remarkable finding was that the K8A change elicited a ϳ16-fold increase in ATPase specific activity compared with wild-type CthTTM (Fig. 6B). We observed this effect with two independent preparations of the K8A protein. A substrate titration experiment (not shown) revealed that K8A resulted in a 21-fold increase in k cat (530 min Ϫ1 ) and a 3-fold increase in K m (75 M) relative to wild-type CthTTM. The gain-of-function effect can be rationalized by reference to the corresponding ␤1 lysine in NeuTTM (Fig. 1B), which forms a hydrogen bond to the main chain carbonyl at the break in the C-terminal helix that we presume impedes the adoption of the closed tunnel architecture seen in Cet1. The twelve side chains essential for CthTTM triphosphatase activity are splayed apart in the C-shaped cup conformation of NeuTTM compared with their positions in the closed Cet1 tunnel, which implies that the CthTTM ␤ barrel must undergo a conformational change to trigger catalysis. We postulate that loss of the inhibitory hydrogen bond in the K8A protein favors the adoption of a catalytically competent tunnel conformation.
Effects of Alanine Mutations on Tripolyphosphatase Activity-All 14 of the alanine mutations abolished or severely diminished magnesium-dependent tripolyphosphatase activity (Fig. 7A). However, the tripolyphosphatase activity of mutants E100A and D102A was restored (to 46 and 96% of wild-type, respectively) when manganese replaced magnesium as the metal cofactor (Fig. 7B). The relatively benign effect of the E100A change on the manganese-dependent tripolyphosphatase was consistent with its modest effects on the manganese-dependent ATPase (Fig. 6B). The notable finding was that Asp-102 was essential for manganese-dependent ATPase, yet dispensable for manganese-dependent tripolyphosphatase. Thus, the exact constellation of functional groups required for catalysis by CthTTM depends on the nature of the metal cofactor and the triphosphate substrate. Perhaps the most striking instance of substrate dependence was the finding that the K8A change stimulated ATP hydrolysis but suppressed the hydrolysis of tripolyphosphate (Figs. 6B and 7). We speculate that Lys-8 makes a key ionic contact with the ␣ phosphate dianion of PPP i that is not applicable to ATP, because its ␣ phosphate is a monoanion.
Probing Structure-Activity Relationships via Conservative Substitutions-Twenty-one conservative substitutions were introduced at 12 of the residues defined as essential in the alanine scan: Glu-4, Glu-6, Arg-39, Arg-41, Lys-52, Glu-62, Glu-64, Lys-87, Arg-89, Asp-102, Glu-115, and Glu-117 (Fig.  8). As noted in the alanine scan, the magnesium-dependent tripolyphosphatase activity was more sensitive to inactivation by conservative mutations than were either of the manganese-dependent hydrolytic reactions (Fig. 8). Based on the crystal structure of the Cet1-Mn 2ϩ complex (18) and their clustering close together in the NeuTTM structure (Fig. 1B), Glu-4, Glu-6, Glu-115, and Glu-117 are likely metal-binding residues. These four glutamates were strictly essential for Mg 2ϩ -PPPase, insofar as their replacement with either aspartate or glutamine eliminated activity. We surmise that the carboxylate moiety is critical (i.e. the isosteric glutamine was ineffective) and that activity requires a critical distance from the carboxylate to the main chain of the tunnel strand that is not satisfied by the shorter Asp side chain. Yet, several of the conservative substitutions for these glutamates elicited significant gains of tripolyphosphatase function (relative to the Ala mutants) in the presence of manganese. For example, the E6D change restored Mn 2ϩ -PPPase activity to a near wild-type level, whereas the E6Q change had no salutary effect (Fig. 8). Thus, CthTTM tolerates the shorter Asp side chain for tripolyphosphate hydrolysis with manganese as the cofactor. A plausible interpretation of this result is that Glu-6 directly coordinates the metal ion as part of an octahedral complex (as is the case for the equivalent glutamate of Cet1) and that the larger atomic radius of manganese (0.8 Å) versus magnesium (0.65 Å) allows Asp-6 to reach and engage a manganese coordination complex but not a magnesium complex. A similar ablation of magnesium-dependent phosphohydrolase activity, although preserving manganesedriven catalysis, was noted previously for a Glu3 Asp mutation of baculovirus RNA triphosphatase (13), which is a putative member of the TTM family. The striking finding here is that the E6D change did not restore manganese-dependent ATPase activity (Fig. 8); this is another instance in which the enzymic requirements for catalysis by CthTTM differ for nucleoside versus inorganic triphosphate substrates.
The Cet1 equivalent of CthTTM Glu-4 coordinates the metal cofactor directly. Changing Glu-4 to either Asp or Gln resulted in restoration of Mn 2ϩ -PPPase activity to ϳ20% of wild-type (Fig. 8), a gain of function relative to the grossly defective E4A mutant (Fig. 7). Apparently, the role of this side chain in the Mn 2ϩ -PPP i complex can be fulfilled, in part, by the amide of Gln or the carboxylate of Asp. The Cet1 analog of CthTTM Glu-115 makes a water-mediated contact to the divalent cation (18). Replacing Glu-115 with Asp or Gln afforded a modest increase in Mn 2ϩ -PPPase activity (25 and 35% of wild-type, respectively) compared with the E115A mutant (Fig.  8). We speculate that CthTTM has residual activity when Glu-115 is mutated, because a water can still join the Mn 2ϩ -PPP i complex. The Cet1 equivalent of CthTTM Glu-117 coordinates the divalent cation directly. Here we find that the conservative E117D mutant was globally defective in triphosphate hydrolysis, implying a stringent requirement for the longer main chain-to-carboxylate linker. (We did not test the effects of a glutamine substitution for Glu-117.) The essential positively charged residues Arg-39, Arg-41, Lys-52, Lys-87, and Arg-89 are good candidates to contact the anionic phosphates of the PPP i and NTP substrates, as suggested by the interactions of several of the equivalent side chains with a sulfate in the Cet1 tunnel (18). In CthTTM, Arg-39 in strand ␤3 is strictly essential for all phosphohydrolase activities, insofar as replacement with lysine or glutamine phenocopied the global catalytic defect of the alanine mutant (Fig. 8). We speculate that Arg-39 makes an essential bidentate interaction via its guanidinium nitrogens that cannot be fulfilled by lysine. At CthTTM residue Lys-87 in strand ␤6, the glutamine mutant was globally defective (Fig. 8), indicating that positive charge is critical at this position (Fig. 8).

DISCUSSION
We found that CthTTM is a metal-dependent tripolyphosphatase and nucleoside triphosphatase. It was most assuredly not an adenylate cyclase in our hands, notwithstanding that it, and virtually every other TTM homolog in the NCBI data base, is annotated as an adenylate cyclase in the absence of corroborating functional data. Our survey of various substrates, although not exhaustive, established the chemical specificity of CthTTM for hydrolysis of the ␤-␥ phosphoanhydride linkage of triphosphate-containing substrates. CthTTM displayed little or no discrimination of the sugar or base components of nucleoside triphosphate substrates, which were hydrolyzed to form a nucleoside diphosphate and P i . The key finding was that CthTTM was at least two orders of magnitude more active in cleaving tripolyphosphate than ATP. This strong preference for an inorganic triphosphate is unprecedented among TTM proteins studied previously. For example, T. brucei Cet1 can hydrolyze PPP i but only 0.5% as well as it hydrolyzes ATP (9). Chlorella virus RNA triphosphatase cleaves PPP i 3% as well as it hydrolyzes ATP (15).
The ensemble of side chains important for CthTTM tripolyphosphatase and ATPase activity, irrespective of metal cofactor choice, consists of 11 amino acids: Glu-4, Glu-6, Arg-39, Arg-41, Lys-52, Glu-62, Glu-64, Lys-87, Arg-89, Glu-115, and Glu-117. We discussed above the likely contributions of several of these side chains to metal and triphosphate binding based on structural and functional studies of other TTM proteins and the homologous NeuTTM crystal structure. A fuller appreciation of their catalytic contributions will hinge on obtaining crystal structures of this and other TTM proteins with genuine triphosphate substrates bound in the active site.
It is most remarkable that the substrate specificity of CthTTM can be transformed by single missense mutation. The K8A change virtually abolished tripolyphosphatase activity while strongly stimulating the hydrolysis of ATP. This result highlights the plasticity of the CthTTM substrate choice and has implications for the rapid acquisition of novel specificities within the TTM superfamily through changes in the residues that line the tunnel walls. We discussed above the likely roles of tunnel conformational equilibria and substrate-specific contacts in mediating the changes in activity accompanying the K8A mutation. It is worth pointing out that the equivalent lysine3alanine change in the ␤1 strand of yeast Cet1 had no effect on ATPase or RNA triphosphatase activity (3), which is sensible, because Cet1 has no equivalent of the "inhibitory" C-terminal helix of NeuTTM that clogs one of the ends of the tunnel. Our studies of CthTTM emphasize that nothing can be taken for granted about the activity of any particular TTM protein, be it substrate specificity or structure-activity relationships, which need to be probed directly for each TTM family member of interest.