Yeast and Viral RNA 5 * Triphosphatases Comprise a New Nucleoside Triphosphatase Family*

Cet1p catalyzes the first step of mRNA capping, the hydrolysis of the g phosphate of triphosphate-terminated RNA to form a 5 * diphosphate end. The RNA triphosphatase activity of Cet1p is magnesium-dependent and has a turnover number of 1 s 2 1 . Here we show that purified recombinant Cet1p possesses a robust ATPase activity ( K m 5 2.8 m M ; V max 5 25 s 2 1 ) in the presence of manganese. Cobalt is also an effective cofactor, but magnesium, calcium, copper, and zinc are not. Cet1p displays broad specificity in convert-ing ribonucleoside triphosphates and deoxynucleoside triphosphates to their respective diphosphates. The manganese- and cobalt-dependent nucleoside triphosphatase of Cet1p resembles the nucleoside triphosphatase activities of the baculovirus LEF-4 and vaccinia virus D1 capping enzymes. Cet1p, LEF-4, and D1 share three collinear sequence motifs. Mutational analysis establishes that conserved glutamate and arginine side chains within these motifs are essential for the RNA triphosphatase and ATPase activities of Cet1p in vitro and for Cet1p function in vivo . These findings are in accord with the effects of single alanine mutations at analogous positions of vaccinia capping enzyme. We suggest that the metal-dependent RNA triphosphatases encoded by yeast and DNA viruses comprise a novel family of phosphohydrolase enzymes with a common active site. The m7GpppN by three enzymatic

Metazoan capping enzymes consist of an amino-terminal RNA triphosphatase domain and a carboxyl-terminal guanylyltransferase domain. In the mammalian enzyme, the two catalytic domains are autonomous and nonoverlapping (6,7). The metazoan RNA triphosphatase domains contain a (I/V)H-CxAGxGR(S/T)G signature motif initially described for the protein tyrosine phosphatase/dual-specificity protein phosphatase enzyme family. These enzymes catalyze phosphoryl transfer from a protein phosphomonoester substrate to the thiolate of the cysteine of the signature motif to form a covalent phosphocysteine intermediate, which is then hydrolyzed to liberate phosphate. The metazoan capping enzymes hydrolyze the phosphoanhydride bond between the ␤ and ␥ phosphates of triphosphate-terminated RNA; they are not active on nucleoside triphosphates (2,6,25). The conserved cysteine of the signature motif is essential for RNA triphosphatase function (2,14). A characteristic of the cysteine phosphatases is their lack of a requirement for a divalent cation cofactor.
The RNA triphosphatases of S. cerevisiae and DNA viruses are structurally and mechanistically unrelated to the cysteine phosphatases. The vaccinia RNA triphosphatase depends absolutely on a divalent cation cofactor. Vaccinia triphosphatase displays broad specificity in its ability to hydrolyze the ␥ phosphate of ribonucleoside triphosphates, deoxynucleoside triphosphates, and triphosphate-terminated RNAs (26). The NTPase 1 and RNA triphosphatase reactions occur at a single active site within a 545-amino acid amino-terminal domain of vaccinia capping enzyme that is distinct from the guanylyltransferase active site (17)(18)(19)(20)(21). Although the K m of vaccinia triphosphatase for NTPs is 1000-fold higher than that for triphosphate-terminated RNA, the turnover number is 10-fold higher for NTP hydrolysis than for RNA triphosphate cleavage (20,26). Also, the divalent cation cofactor specificities are distinct. The vaccinia RNA triphosphatase is optimal with magnesium, is 12% as active in manganese, and is inactive with cobalt. In contrast, the vaccinia NTPase is fully active with cobalt, manganese, or magnesium (26,27). Baculovirus LEF-4 hydrolyzes the ␥ phosphate of RNA and NTPs; the LEF-4 NTPase is activated by manganese or cobalt, but not by magnesium (22,23).
Itoh et al. (28) isolated a yeast capping enzyme containing both RNA triphosphatase and guanylyltransferase activities.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  The yeast RNA triphosphatase, which was magnesium dependent, could not hydrolyze ATP or GTP (28). A highly purified yeast capping enzyme consisted of two major polypeptides: (i) a 52-kDa guanylyltransferase identified as the 459-amino acid Ceg1 protein (11), and (ii) an RNA triphosphatase component with an apparent electrophoretic mobility of 80 kDa (29). Tsukamoto et al. (13) identified the RNA triphosphatase polypeptide as the 549-amino acid product of the CET1 gene. We independently isolated CET1 as a multicopy suppressor of a temperature-sensitive mutation in the yeast guanylyltransferase Ceg1p (14). The CET1 gene is essential for yeast cell growth (13,14).
Tsukamoto et al. (13) reported that recombinant Cet1p catalyzed the hydrolysis of the ␥ phosphate of triphosphate-terminated RNA but did not hydrolyze GTP, prompting the conclusion that Cet1p acts only on RNA substrates. This would suggest that the yeast RNA triphosphatase has metazoan-like substrate specificity for RNA triphosphate ends yet exploits a virus-like catalytic mechanism that depends on metal-assisted cleavage of the phosphoanhydride bond. Here, we report that purified recombinant Cet1p is a highly active NTPase. Thus, it is not specific for RNA substrates. Cet1p resembles the vaccinia and baculovirus triphosphatases in that its NTPase function is activated by manganese and cobalt. We discerned three collinear sequence motifs that are shared by yeast Cet1p, vaccinia D1, and baculovirus LEF-4. Mutational analysis establishes that conserved glutamate and arginine side chains within these motifs are essential for Cet1p function in vitro and in vivo.
Mutational Analysis-Alanine substitution mutations were introduced into the CET1(201-549) gene by a polymerase chain reaction using the two-stage overlap extension method. The residues targeted for amino acid substitutions were Glu-305, Glu-307, Arg-454, Glu-492, Glu-494, and Glu-496. Plasmid p358-CES5(201-549) (14) was the template for the first-stage amplifications. The DNA products of the secondstage amplification were digested with NdeI and BamHI and inserted into pET-16b (Novagen). The presence of the desired mutations was confirmed by DNA sequencing; the inserted restriction fragment was sequenced completely to exclude the acquisition of unwanted mutations during amplification and cloning. The His-tagged mutant proteins were purified from soluble bacterial lysates by nickel-agarose chromatography as described previously (14). The 0.2 M imidazole eluate fractions containing Cet1(201-549)p were dialyzed against Buffer C (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 2 mM DTT, 10% glycerol, and 0.05% Triton X-100). Protein concentration was determined by using the Bio-Rad dye reagent with bovine serum albumin as the standard.

Cet1p
Hydrolyzes the ␥ Phosphate of ATP-We have previously shown that purified recombinant Cet1p catalyzed the release of 32 P i from ␥-32 P-labeled triphosphate-terminated poly(A) with a turnover number of ϳ1 s Ϫ1 (14). RNA triphosphatase activity was strictly dependent on the inclusion of magnesium in the reaction mixture. The experiment shown in Fig. 1 demonstrates that Ceg1p also catalyzes the near-quantitative release of 32 P i from 1 mM [␥-32 P]ATP in the presence of 1 mM manganese as the divalent cation cofactor. There was no detectable ATP hydrolysis in the absence of a divalent cation, and 1 mM magnesium was extremely feeble in supporting catalysis. The failure of previous investigators to appreciate the NTPase activity of the capping enzyme isolated from yeast extracts or recombinant Cet1p is most likely attributable to the reliance on magnesium as the divalent cation cofactor.
Divalent cation specificity was tested in reaction mixtures containing 1 mM ATP and 2 mM divalent cation ( Fig. 2A). Cobalt was at least as effective as manganese in activating the ATPase, whereas magnesium was one-twentieth as effective. Calcium, copper, and zinc did not activate the ATPase. Cofactor titration experiments showed that hydrolysis of 1 mM ATP was optimal at 1-3 mM MnCl 2 and declined slightly as MnCl 2 was increased to 5 mM (Fig. 2B). Lowering the ATP concentration to 0.5 mM elicited a shift to the left in the manganese titration curve, whereas increasing ATP to 2 mM caused a shift to the right (Fig. 2B). The titration curves were sharply sigmoidal at manganese concentrations below the level of input ATP. Cobaltdependent hydrolysis of 1 mM ATP was optimal at 1-5 mM CoCl 2 (data not shown). ATP hydrolysis by an equivalent amount of Cet1p was 25% higher with cobalt than with manganese (data not shown). The activity with magnesium at the optimum concentration of 5-10 mM MgCl 2 was only 10% of the cobalt-dependent activity (data not shown).
ATP hydrolysis was measured in 50 mM Tris buffer from pH 6.0 to pH 9.5 (Fig. 2C). Activity was optimal from pH 6.5 to pH 7.0 and declined with increased alkalinity. Activity at pH 9.5 was 25% that at pH 7.0.
Kinetics of ATP Hydrolysis-The extent of 32 P i release from [␥-32 P]ATP during a 15-min reaction was proportional to the amount of input Cet1p protein (Fig. 3A). 32 P i accumulated with time over 30 min; the rate of reaction varied linearly with Cet1p concentration (Fig. 3B). From a plot of initial rate versus enzyme, we calculated a turnover number of 25 s Ϫ1 . Cet1p catalyzed the quantitative conversion of [␣-32 P]ATP to [␣-32 P]ADP. The rate of [␣-32 P]ADP formation was identical to the rate of 32 P i release from [␥-32 P]ATP assayed in a parallel reaction mixture (Fig. 3C). We detected no formation of [␣- 32  tion, by which time all of the nucleotide had been converted to ADP. Thus, we conclude that Cet1p catalyzes the hydrolysis of ATP to ADP plus P i and is unable to further hydrolyze the ADP reaction product. Kinetic parameters were determined by measuring ATPase activity as a function of input [␥-32 P]ATP concentration (Fig. 3D). From a double-reciprocal plot of the data, we calculated a K m of 2.8 M ATP and a V max of 25 s Ϫ1 (Fig. 3D) ATPase Activity of Cet1(201-549) and Cet1(246 -549)-We previously purified and characterized two amino-terminal truncation mutants, Cet1(201-549)p and Cet1(246 -549)p, that retained full activity in catalyzing the release of 32 P i from ␥-32 P-labeled triphosphate-terminated poly(A) (14). Similarly, the two truncated proteins were as active as full-length Cet1p in hydrolyzing ATP (Fig. 3A). We infer that both reactions are catalyzed by the same catalytic site within the carboxyl-terminal domain. The kinetic parameters determined for ATP hydrolysis by Cet1(201-549)p (K m ϭ 3.3 M ATP; V max ϭ 33 s Ϫ1 ) were similar to those of full-length Cet1p. These data suggest that the deleted amino-terminal 200-amino acid segment does not contribute to nucleotide binding or reaction chemistry.
Conserved Motifs in Metal-dependent RNA Triphosphatases-The yeast Cet1p, vaccinia D1, and baculovirus LEF-4 enzymes display remarkably similar biochemical characteristics in their hydrolysis of the ␤-␥ phosphoanhydride linkage of RNA and NTP substrates. The activation of NTP hydrolysis by manganese and cobalt are the signature features of these enzymes (22,27). Is there a common structural basis for metaldependent catalysis? A Blastp search of the National Center for Biotechnology Information database with the catalytic domain of Cet1p revealed no similarity to vaccinia D1 or baculovirus LEF-4 or to any other known NTPases or phosphatases. A more instructive approach was to screen "by eye" for conservation of the several residues identified by mutational analysis of vaccinia D1 as being essential for the triphosphatase activity of the viral capping enzyme (18,19). In this way, we found that the metal-dependent RNA triphosphatases share three collinear sequence motifs, designated A, B, and C (Fig. 4). These are present in yeast Cet1p, in the triphosphatase-guanylyltransferase domains of the vaccinia virus, Shope fibroma virus, molluscum contagiosum virus, and African swine fever virus capping enzymes, and in baculovirus LEF-4. The residues that are essential for the RNA triphosphatase and ATPase activities of vaccinia virus capping enzyme include two glutamates in motif A, an arginine in motif B, and two glutamates in motif C (19). Alanine substitutions at any of these positions in the vaccinia capping enzyme reduced phosphohydrolase specific activity by 2-3 orders of magnitude (51). Yu et al. (19) proposed that these 5 residues comprise part of the triphosphatase active site. All 5 residues essential for vaccinia triphosphatase activity are conserved in LEF-4 and Cet1p. Conceivably, the metaldependent RNA triphosphatase/NTPases share a common active site, in which case mutations of the equivalent residues of Cet1p should result in a loss of triphosphatase activity.
Conserved Residues in Motifs A, B, and C Are Essential for the ATPase Activity of Cet1(201-549)p-Cet1p residues Glu-305 and Glu-307 (motif A), Arg-454 (motif B), and Glu-492, Glu-494, and Glu-496 (motif C) were replaced individually by alanine. The Ala mutations were introduced into Cet1(201-549)p, which is fully active in vitro and in vivo (14). The six mutated proteins were expressed as His-tagged fusions and purified from soluble lysates by nickel-agarose column chromatography in parallel with wild type Cet1(201-549)p (Fig. 5A). Mutational effects on manganese-dependent ATP hydrolysis were gauged by enzyme titration under conditions of substrate excess (1 mM ATP). The E305A and E307A mutants of motif A and the E494A mutant of motif C were unable to hydrolyze ATP at levels of input enzyme sufficient for the quantitative release of 32 P i by wild type Cet1(201-549)p (Fig. 5B). From these data, we calculated that the specific activities of E305A, E307A, and E494A were Ͻ0.5% of the activity of the wild type enzyme. Alanine substitutions for two other glutamate residues in motif C, E492A and E496A, reduced specific activity to 2.8 and 0.6% of the wild type value, respectively. The specific activity of the R454A mutant of motif B was 15% of the wild type value. Similar mutational effects were noted when the ATPase assays were performed in the presence of 2 mM cobalt (data not shown).
Kinetic parameters for ATP hydrolysis were determined for three of the mutants (R454A, E492A, and E496A) that displayed reduced ATPase activity (Fig. 5C). The R454A mutation increased the K m for ATP to 100 M (nearly 30-fold higher than the wild type K m ), whereas the V max was 30% of the wild type value. The E492A mutation increased the K m for ATP 9-fold to 32 M and lowered the V max to 8% of the wild type value. The ratios of V max to K m for R454A and E492A were 2 orders of magnitude lower than the wild type ratio. The E496A mutation increased the K m for ATP 5-fold and decreased the V max to 0.9% of the wild type value; the ratio of V max to K m for E496A was 0.2% of the wild type ratio.
Mutational Effects on RNA Triphosphatase-The RNA triphosphatase activities of the wild type and six mutant Cet1(201-549)p proteins were assayed by the release of 32 P i from 2 M [␥-32 P]-labeled poly(A). Specific enzyme activity was determined from the slopes of the protein titration curves in the linear range of enzyme dependence (Fig. 6). The specific activity of the wild type enzyme (16.5 nmol of P i released per microgram of protein in 15 min) corresponds to a turnover number of ϳ0.8 s Ϫ1 . Analysis of RNA triphosphate cleavage by wild type Cet1(201-549)p as a function of RNA substrate concentration revealed a K m of 1 M for poly(A) triphosphate termini and a V max of 1 s Ϫ1 (data not shown). The specific activities of the alanine mutants, calculated from the data in Fig. 6 and expressed as a percentage of the wild type value, were as follows: E305A, 0.03%; E307A, 0.03%; R454A, 0.2%; E492A, 1.7%; E494A, 0.03%; and E496A, 0.8%. Kinetic parameters were not determined for these catalytically defective enzymes. The effects of the single Glu-to-Ala mutations on RNA triphosphatase specific activity were similar in magnitude to their effects on ATPase specific activity. The R454A mutation appeared to exert a more drastic effect on RNA triphosphate cleavage than on ATP hydrolysis. The overall concordance of the mutational effects suggests that both reactions are catalyzed at a single active site.
Conserved Residues in Motifs A, B, and C Are Essential for CET1 Function in Vivo-Mutant alleles of CET1(201-549) encoding triphosphatase-defective enzymes were tested for their function in vivo using the plasmid shuffle assay described by Ho et al. (14). The wild type and mutants coding sequences were cloned into a CEN TRP1 vector so as to place the CET1 (201-549) gene under the control of the natural CET1 promoter. The plasmids were transformed into the ⌬cet1 strain YBS20, in which the chromosomal CET1 locus has been deleted and replaced by LEU2. Growth of YBS20 is contingent on the maintenance of a wild type CET1 allele on a CEN URA3 plasmid. Hence, YBS20 is unable to grow in agar medium containing 0.75 mg/ml 5-fluoroorotic acid, which selects against the URA3 plasmid, unless it is transformed with a biologically active CET1 allele or a functional homologue from another organism (8,14). We found that growth on 5-fluoroorotic acid was complemented by CET1(201-549), but not by mutant alleles E305A, E307A, R454A, E492A, E494A, or E496A (data not shown). The correlation of in vitro and in vivo mutational effects suggests that the triphosphatase activity of Cet1p is essential for yeast cell growth. DISCUSSION We have shown that the yeast RNA 5Ј-triphosphatase Cet1p hydrolyzes the ␥ phosphate of nucleoside triphosphates. The NTPase of Cet1p is activated by manganese and cobalt. This is a property shared with the triphosphatase components of the vaccinia D1 and baculovirus LEF-4 capping enzymes. The turnover number of the yeast enzyme in ATP hydrolysis (25 s Ϫ1 ) is similar to the values reported for the baculovirus (30 s Ϫ1 ) and vaccinia virus (10 s Ϫ1 ) triphosphatases, although the affinity of the yeast enzyme for ATP (K m ϭ 2.8 M) is significantly higher than that of either LEF-4 (K m ϭ 43 M) or vaccinia triphosphatase (K m ϭ 800 M) (21,22).
We propose that Cet1p is the prototype of a previously unrecognized family of metal-dependent phosphohydrolases that includes the DNA virus and yeast RNA triphosphatases involved in cap formation. A common structural basis for catalysis by members of this enzyme family is suggested by the conservation of collinear motifs A, B, and C. Conserved amino acid residues within the motifs likely comprise part of the active site. As shown here and by Yu et al. (19), single alanine substitutions within the three motifs abrogate or severely diminish the triphosphatase activity of yeast Cet1p and vaccinia D1. The essential residues identified by alanine scanning include the two alternating glutamates in motif A, an arginine residue in motif B, and either two (in the poxvirus capping enzymes) or three (in Cet1p and African swine fever virus capping enzyme) alternating glutamates in motif C. The effects of alanine cluster mutations in baculovirus LEF-4 are concord- P]ATP, and wild type or mutant enzymes as specified were incubated for 15 min at 30°C. The amount of input enzyme was adjusted to attain roughly similar levels of ATP hydrolysis at saturating ATP concentrations. The extent of P i release at the lowest ATP concentrations tested was less than than 20% of the input substrate. Kinetic parameters were determined from double-reciprocal plots of the data. ant with the findings for Cet1p and D1, to wit, manganese-dependent NTPase and RNA triphosphatase activities are eliminated or drastically reduced by simultaneous replacement of both motif A glutamates by alanine, by single alanine substitution for the motif B arginine, or by simultaneous replacement of the two glutamates of motif C (EYEFD) by alanine (23).
The marked decrease in affinity for ATP elicited by the R454A mutation of Cet1(201-549)p suggests a role for this motif B residue in substrate binding. We speculate that the arginine side chain interacts with the negatively charged 5Ј triphosphate moiety. Elimination of the glutamate side chains in motifs A and C profoundly inhibited ATPase activity, even at very high substrate concentrations, which suggests a role for these residues in phosphohydrolase reaction chemistry. The glutamates in motif A and C may coordinate the essential divalent cation(s). An additional role for the proximal glutamate of motif C in substrate binding is inferred from the 9-fold increase in K m for ATP caused by the E492A mutation of Cet1(201-549)p.
The sequence similarity between Cet1p and the DNA virusencoded RNA triphosphatase/NTPases appears to be limited to motifs A, B, and C. Cet1p possesses triphosphatase activity only, whereas LEF-4 and D1 also have guanylyltransferase activities. In the baculovirus and vaccinia capping enzymes, the triphosphatase motifs are located amino-terminal to the six conserved motifs that comprise the guanylyltransferase active site (19,22,23). The vaccinia D1 protein cannot be truncated to less than 545 amino acids without loss of its triphosphatase and guanylyltransferase activities. The minimum functional domains of the 464-amino acid baculovirus LEF-4 triphosphatase/guanylyltransferase have not yet been defined. The 305-amino acid Cet1p truncation mutant, Cet1(245-549)p, which retains full triphosphatase activity in vitro, is the smallest functional domain that has been identified for the metaldependent RNA triphosphatase family. Additional candidates for membership in this family may emerge as genome sequencing uncovers new polypeptides containing motifs A, B, and C. For example, the 320-amino acid polypeptide encoded by the S. cerevisiae YMR180C open reading frame is homologous to the carboxyl-terminal domain of Cet1p and includes all six of the amino acids in motifs A, B, and C that are important for NTP hydrolysis (Fig. 4). Preliminary characterization of recombinant YM180C protein indicates that it possesses magnesiumand cobalt-dependent ATPase activity. 2 The function of YMR180C is unknown; this enzyme may well catalyze phosphohydrolase reactions unrelated to mRNA capping.
In summary, there is now clear evidence for at least two mechanistically and structurally distinct classes of RNA 5Ј triphosphatases: (i) the divalent cation-dependent RNA triphosphatase/NTPase family (exemplified by yeast Cet1p, baculovirus LEF-4, and vaccinia D1), which require motifs A, B, and C for activity, and (ii) the divalent cation-independent RNA triphosphatases, e.g. the metazoan cellular enzymes and the baculovirus enzyme BVP (30,31), which require the HCx-AGxGR(S/T)G phosphate-binding motif. The existence of additional classes of RNA 5Ј-triphosphatases is likely, given that the candidate capping enzymes of several RNA viruses and trypanosomatid protozoa lack the defining motifs of the two known RNA triphosphatase families (32,33). Hence, the triphosphatase components of the capping apparatus provide attractive targets for the identification of specific antifungal, antiviral, and antiprotozoal drugs that will block capping of pathogen mRNAs but spare the mammalian host enzyme.