Characterization of the Vaccinia Virus RNA 5`-Triphosphatase and Nucleoside Triphosphate Phosphohydrolase Activities

D1R, an active subdomain of the large subunit of vaccinia virus mRNA capping enzyme possessing ATPase, RNA 5′-triphosphatase, and guanylyltransferase activities, was expressed in Escherichia coli and shown to be functionally equivalent to the heterodimeric enzyme (Myette, J. R., and Niles, E. G. (1996) J. Biol. Chem. 271, 11936-11944). A detailed characterization of the phosphohydrolytic activities of D1R demonstrates that, in addition to ATPase and RNA 5′-triphosphatase activities, the capping enzyme also possesses a general nucleoside triphosphate phosphohydrolase activity that lacks a preference for the nucleoside base or sugar. Nucleoside triphosphate and mRNA saturation kinetics are markedly different, with RNA exhibiting a K and turnover number 100- and 10-fold less, respectively, than those values measured for any NTP. The linear competitive inhibition of RNA 5′-triphosphatase activity by ATP, and the relative manner by which both ATPase and RNA 5′-triphosphatase activities are inhibited by specific oligonucleotides, kinetically demonstrate that each activity is carried out at a common active site. Direct UV photo-cross-linking of either P-radiolabeled ATP or 23-mer triphosphorylated RNA, followed by cyanogen bromide cleavage of the photo-linked enzyme, localizes the major binding site for both ATP and RNA to a region between amino acids 1 and 221. The inability of ATP to competitively inhibit either EGMP formation or the transfer of GMP to RNA kinetically differentiates the phosphohydrolase active site from the guanylyltransferase active site.

The vaccinia virus mRNA capping enzyme catalyzes three of the four reactions required for cap I formation, including RNA 5Ј-triphosphatase (1,2), guanylyltransferase, and (guanine-7-)methyltransferase activities (3)(4)(5). The RNA 5Ј-triphosphatase, nucleoside triphosphate phosphohydrolase, and guanylyltransferase active sites have been mapped to a 60-kDa NH 2 -terminal domain of the D1R subunit (7,8,10), while the methyltransferase resides in a independent domain comprised of the carboxyl terminus of D1R together with D12L (9 -11). In an accompanying report (12), kinetic analyses demonstrated that the D1R 1-545 subdomain expressed in Escherichia coli is functionally equivalent to the full-size capping enzyme with respect to RNA 5Ј-triphosphatase, ATPase, and guanylyltransferase activities, indicating that the methyltransferase domain exerts little influence on catalysis carried out at the active sites within D1R  . In addition to triphosphorylated RNA, the mRNA capping enzyme employs ATP and GTP as substrates for phosphohydrolysis (1,6,23,24), raising the question whether both the nucleoside triphosphate phosphohydrolase activity and the RNA 5Ј-triphosphatase activity are carried out at the same active site.
In this report, we describe the kinetic properties of the RNA 5Ј-triphosphatase and nucleoside triphosphate phosphohydrolase activities of the vaccinia virus mRNA capping enzyme. Through competitive inhibition analyses, we demonstrate that both activities are carried out at a single active site. We also present UV photo-cross-linking mapping data which locates the phosphohydrolysis active site to the 25-kDa NH 2 -terminal region of D1R (amino acids 1-221). A kinetic argument is made for the separation of this active site from the guanylyltransferase active site, supporting a three-active site model for the mRNA capping enzyme.

RNA Synthesis
Triphosphorylated RNA substrate was synthesized by in vitro transcription of linearized plasmid DNA templates as described (12). Linearization of either pGEM3Zf(ϩ) or pGEM9Zf(ϩ) plasmids at different restriction sites enabled the synthesis of RNA of discrete lengths.

Preparation of Diphosphorylated RNA
Diphosphorylated RNA was generated in vitro through the RNA 5Ј-triphosphatase activity of the capping enzyme using triphosphorylated 23-mer RNA as substrate. D1R  (100 pmol, specific activity 50 pmol of P i /min/pmol of enzyme) was added to a 500-l reaction containing 50 mM Tris-HCl, pH 8.0, 2 mM MgCl 2 , 10 mM ␤-mercaptoethanol, 10% glycerol, and 10 nmol of triphosphorylated RNA 23-mer. The reaction was incubated at 37°C for 30 min. Dephosphorylation of RNA was monitored by inclusion of a small amount of high specific activity [␥-32 P]RNA 23-mer (Ͼ2000 cpm/pmol) in the reaction and assaying for the release of 32 P i by PEI-cellulose thin layer chromatography as described for the RNA 5Ј-triphosphatase assay (12). The reaction was quenched and extracted by adding an equal volume of 24:24:1 phenol: chloroform:isoamyl alcohol. RNA was precipitated with 3 volumes of 95% ethanol, washed, dried and resuspended in 300 l of diethyl pyrocarbonate-treated water. The sample was further desalted and concentrated by ultrafiltration in a Micron 3 concentrator (Amicon).
Inhibition of ATPase by Oligonucleotides-D1R 1-545 ATPase activity was measured in a standard assay containing 10 mM MgCl 2 , in the presence of varying concentrations of oligonucleotides (see below). In the measurement of triphosphorylated 23-mer RNA inhibition of ATPase activity, a 2-min reaction time was employed to ensure that less than 20% of the mRNA was dephosphorylated.
Nucleoside Triphosphate Phosphohydrolase Assay-Nucleoside triphosphate phosphohydrolase activity was measured by a colorimetric phosphate assay as described elsewhere (13). A standard assay contained 50 mM Tris-HCl, pH 8.0, 20 mM MgCl 2 , 10 mM ␤-mercaptoetha-nol, 10% glycerol, 5 mM nucleoside triphosphate (U. S. Biochemical Corp., ultrapure grade), and 100 nM enzyme, in a 50-l reaction volume. Enzyme was added last to the chilled assay mixture, and the reaction was transferred to 37°C for 5 min. Reactions were quenched by the addition of 5 l of 0.5 M EDTA and assayed for the release of phosphate using a molybdate-based colorimetric assay. Sterox was omitted from the color development reagent. Color development reagent (800 l, malachite green:ammonium molybdate 3:1) was added to the enzyme assay mix at room temperature, gently vortexed, and developed for 1 min at room temperature. Development was quenched by the addition of 100 l of 34% sodium citrate (w/v), pH 8.5. Moles of phosphate were determined from the absorbance at 660 nm by use of a KH 2 PO 4 standard curve. Absorbances were corrected for background phosphate generated by the nonenzymatic hydrolysis of nucleoside triphosphates. Absorbances were linear up to 1 mM phosphate and did not vary over the time course of the assay. The assay was linear for 15 min at an enzyme concentration between 20 nM and 150 nM.

UV Photo-cross-linking
Photo-cross-linking of ATP to D1R 1-545 -The standard photo-crosslinking reactions were carried out in a 96-well microtiter dish floating in an ice water slurry. Direct UV photo-linkage was achieved using a single GE G15T8 15-watt germicidal ultraviolet lamp held at a distance of 10 cm for 45 min. A standard reaction mixture of 30 l contained 50 picomol of D1R 1-545(K260M) , 50 mM Tris-HCl, pH 8.0, 10 mM ␤-mercaptoethanol, 10% glycerol (w/v), 50 M cold ATP, and 10 -20 Ci of [␣ 32 P]ATP (DuPont NEN, 800 Ci/mmol). The reaction was quenched by adding 10 l of 4 ϫ SDS sample buffer, and the samples were boiled for 5 min and separated by SDS-PAGE 1 on 10% acrylamide gels. Modified enzyme was visualized by autoradiography of dried gels and incorporated radiolabel quantified by Cerenkov counting. Approximately 1% direct photo-cross-linking product was achieved at saturating ATP concentrations.
Inhibition of ATP Photo-cross-linking by Oligonucleotides-ATP photo-cross-linking was carried out in the presence of varying concentrations of oligonucleotide and nucleotide competitors. Oligonucleotide competitors included triphosphorylated RNA, diphosphorylated RNA, 5Ј-OH RNA (Oligos, Etc.), and DNA 23-mers, all of the sequence GGGC-GAAUUAAUAUAUUUGAAUU (with uracil replaced by thymidine in DNA). Nucleotide competitors included ADP, AMP, adenosine, and the dinucleotides ApU and GpppA (a methyl acceptor in the methyltransferase reaction). In a parallel experiment, the inhibition of ATP hydrolysis by these same reagents was also examined. ATPase activity was measured as described previously (12) except a subsaturating concentration of 50 M ATP and 10 mM MgCl 2 were employed. UV Photo-cross-linking of RNA to D1R 1-545 -Direct UV photo-crosslinking of triphosphorylated 23-mer RNA to D1R 1-545 was conducted as described (12) except 200 pmol of D1R 1-545 were employed in a 200-l reaction volume.
Cyanogen Bromide Cleavage of D1R 1-545 -D1R 1-545 (200 pmol) were UV photo-cross-linked with 50 M ATP or 1 M 23-mer triphosphorylated RNA as described above. Cross-linking reactions were scaled up to a 200-l reaction volume and split into four microtiter wells containing 50 l each. Following cross-linking, samples were pooled and precipitated with ice-cold 10% trichloroacetic acid. Pellets were washed twice with 200 l of acetone, dried, and resuspended in 200 l of 70% formic acid containing 20 mg/ml CNBr (Pierce) (17). The digestions were carried out for 18 -20 h at room temperature while under nitrogen. Samples were then dried three times under a stream of nitrogen, each time resuspending in 200 l of water. The final wash and concentration step was by Speed-Vac desiccation, after which samples were resuspended in 20 l of 1 ϫ peptide sample buffer and boiled, and the pH was adjusted when necessary by addition of 1-2 l of 1 M Tris-HCl, pH 8.5, 0.1% SDS. Samples were resolved on Tricine gels (14), stained with Coomassie Brilliant Blue (Sigma), and dried. Radiolabeled peptides were visualized by autoradiography. To prepare cleavage products for microsequencing, 600 picomol of D1R 1-545 were directly digested with CNBr. The dried cleavage products were resuspended in 45 l of 1 ϫ peptide sample buffer, and the peptides were resolved by electrophoresis on Tricine gels pre-run at a low voltage in the presence of 100 M thioglycolic acid as recommended (Probe-Design Peptide Separation System Technical Manual, Promega, 1992). Approximately 200 picomol of protein were loaded per lane. The peptides were transferred to a Transblot (Bio-Rad) polyvinylidene difluoride membrane and the pep-tides were visualized by Coomassie Brilliant Blue staining (0.1% Coomassie Brilliant Blue R-250 in 45% methanol and 5% acetic acid). The N-terminal amino acid sequence of the cleavage products was determined by the CAMBI Peptide Sequencing Facility (SUNY-Buffalo).

RESULTS
Previous work demonstrated that the mRNA capping enzyme possesses the ability to hydrolyze ATP and GTP in addition to the 5Ј-terminal phosphate of triphosphorylated RNA (6,12,23,24). To characterize the kinetic properties of the nucleoside triphosphate phosphohydrolase activity and to determine whether the RNA 5Ј-triphosphatase and nucleoside triphosphate phosphohydrolase activities are carried out at a single active site, the following kinetic and biochemical studies were conducted.
Kinetic Properties-The magnesium ion concentration dependence of both ATPase and RNA 5Ј-triphosphatase activities is presented in Fig. 1. Although both activities require MgCl 2 for catalysis, the concentrations needed for optimal activity are substantially different. RNA 5Ј-triphosphatase shows optimal activity between 1 and 2 mM MgCl 2 , followed by a sharp inhibition at higher concentrations. ATPase requires a roughly 10-fold higher concentration of MgCl 2 and reaches a stable plateau. Both activities exhibited a modest sensitivity to NaCl concentration, each being inhibited by 25% at 50 mM (data not shown).Substrate saturation kinetics for both RNA and ATP also demonstrate marked differences (12), Fig. 2. For an RNA 23-mer substrate, the RNA 5Ј-triphosphatase K m is approxi- 1 The abbreviations used are: PAGE, polyacrylamide gel electro- mately 1 M, roughly 1000-fold lower than the K m measured for ATP. Moreover, RNA turnover is considerably slower, 50 min Ϫ1 for RNA as compared to 500 -600 min Ϫ1 for ATP. RNA substrates of defined lengths from 9 to 42 nucleotides exhibit similar kinetic properties which demonstrates that beyond 9 bases, RNA chain length has little influence on RNA 5Ј-triphosphatase activity.
To determine whether the mRNA capping enzyme is a general nucleoside triphosphate phosphohydrolase, additional nucleoside triphosphates were tested as potential substrates. A broad substrate specificity was observed, Table I. The kinetic properties were comparable for each substrate, exhibiting a range of K m values from 0.8 to 1.3 mM and turnover numbers from approximately 500 to 800 min Ϫ1 . This enzyme is able to hydrolyze equally well both purine and pyrimidine nucleoside triphosphates, containing either ribose or 2Ј deoxyribose. However, the capping enzyme does not possess a general phosphatase activity, since phosphohydrolysis is specific for the ␤-␥phosphoester bond (data not shown). In general, the D1R 1-545 subdomain exhibited comparable nucleoside triphosphate phosphohydrolase kinetics to the holoenzyme, demonstrating that this activity is not affected by the methyltransferase domain (12).
The Number of Active Sites-The fact that the mRNA capping enzyme possesses both RNA 5Ј-triphosphatase and general nucleoside triphosphate phosphohydrolase activities raises the question of whether the catalysis of each reaction is carried out at a common active site or at two independent sites. Two approaches were taken to investigate this issue. In the first, the ability of ATP to competitively inhibit RNA 5Ј-triphosphatase was assessed. ATP was shown to be a linear competitive inhibitor of RNA 5Ј-triphosphatase, Fig. 3, strongly suggesting that ATP binds to the RNA 5Ј-triphosphatase active site and prohibits the binding of the RNA 5Ј terminus. The apparent K i for this ATP inhibition was calculated from a Dixon plot to be approximately 1.4 mM. This value is comparable to the K m for ATPase activity (12) as would be predicted if both substrates are competing for the same active site. The minor variation in the K m and K i values is a result of the different MgCl 2 concentrations (Fig. 1) employed in the two measurements.
In a second approach, the ability of various oligonucleotides to inhibit ATPase activity was assessed, Fig. 4. Inhibition was measured at 5 mM ATP while varying the concentration of the oligonucleotide present in the assay. The oligonucleotides employed were all 23 nucleotides in length and possessed the same nucleotide sequence but differed in the nature of their 5Ј terminus, and in the case of single-stranded DNA, the substitution of thymidine for uracil. As expected, triphosphorylated RNA 23-mer was an effective inhibitor of ATPase, inhibiting activity by 50% at a concentration of approximately 35 M. Diphosphorylated RNA, the product of the RNA 5Ј-triphosphatase activity, also inhibited the hydrolysis of ATP, although to a lesser extent in comparison to the inhibition by triphosphorylated RNA. 5Ј-OH RNA was a relatively weak inhibitor of ATPase activity while single-stranded DNA did not significantly inhibit ATP hydrolysis. The inhibition of ATP hydrolysis by these oligonucleotides is consistent with their relative inhibition of RNA 5Ј-triphosphatase activity (12). Although RNA 5Ј-triphosphatase and ATPase differed in the relative extent to which their activities were inhibited by these reagents, both activities were inhibited in a similarly graded fashion by the same oligonucleotides. The most effective inhibition of either ATPase or RNA 5Ј-triphosphatase activity was observed with a nucleic acid possessing either a triphosphorylated or diphosphorylated 5Ј terminus. Taken together, these kinetic competition data argue for a common active site for RNA 5Ј-triphosphatase and ATPase.
Relationship of Phosphohydrolase and Guanylyltransferase  Active Sites-A close structural relationship between the RNA 5Ј-triphosphatase/nucleoside triphosphate phosphohydrolase and guanylyltransferase activities is inferred from the relative levels of each activity measured in a set of carboxyl-terminal deletions of D1R which were used to delineate the D1R 1-545 subdomain (12). To determine the degree to which the nucleotide binding regions of each active site overlap, ATP was tested as a potential competitive inhibitor of guanylyltransferase activity. ATP is known not to be a substrate for guanylyltransferase (15) since it can neither form a phosphoryl intermediate with the enzyme nor serve in the transfer of AMP to RNA. Inclusion of up to 5 mM ATP does not inhibit either EϳGMP formation (Fig. 5A) or the transfer of GMP to RNA at saturating concentrations of RNA (Fig. 5B). Under the conditions for EϳGMP formation, less than 1% of the 5 mM ATP is hydrolyzed by the ATPase activity in D1R 1-545 (data not shown), ruling out the possibility that the ATP is significantly depleted and therefore, unable to effectively inhibit guanylyltransferase activity. We conclude from this lack of inhibition that the ATP binding site in the RNA 5Ј-triphosphatase/NTPase active site does not overlap with either the GTP or the diphosphorylated RNA binding site in the guanylyltransferase active site. As a further indication of separate active sites, the catalytic lysine (lysine 260) of guanylyltransferase (16,22), required for formation of the phosphoryl GMP intermediate, is not required for ATP binding or hydrolysis since a mutation of K 260 to methionine eliminates EϳGMP formation but does not affect ATPase activity (data not shown).
Localization of the Nucleotide Binding Site-In an attempt to localize the nucleoside triphosphate binding site within D1R 1-545, we employed direct UV photo-cross-linking of ATP followed by the specific chemical cleavage of the covalently modified enzyme. The extent of ATP photo-cross-linking to D1R 1-545 was found to be dependent upon the time of the reaction and enzyme concentration, and was saturable by ATP, Fig. 6. Spurious covalent linkage of ITP (a minor deamination product of ATP present in some preparations of [␣-32 P]ATP) to the guanylyltransferase active site was prevented by utilizing the mutant D1R 1-545(K260M) (22) which lacks the catalytic lysine 260 required for EϳGMP formation, but possesses wildtype ATPase activity (data not shown). The specificity of ATP photolinkage was further assessed by testing the ability of either triphosphorylated RNA (a competing substrate) or various analogues to inhibit both cross-linking and ATPase activity at 50 M ATP, Fig. 7. As expected, triphosphorylated RNA 23-mer was a good inhibitor of both ATPase and photo-cross-linking activities. In contrast, compounds such as single-stranded DNA, ApU, and GpppA which are poor inhibitors of ATPase activity were also poor inhibitors of UV photo-linkage. In comparison to 5Ј triphosphorylated 23-mer RNA, 5Ј OH RNA was a relatively poor inhibitor of ATPase activity and also a less effective inhibitor of the UV photo-linkage of ATP to D1R 1-545 .
In an attempt to localize the site(s) of ATP photo-linkage, D1R 1-545 photo-labeled with ATP in the presence or absence of triphosphorylated RNA 23-mer, was cleaved with CNBr and the cleavage products were resolved by electrophoresis on Tricine gels, Fig. 8. The products obtained were assigned to the D1R primary sequence, Fig. 8A, according to their apparent molecular weights. This assignment was confirmed by N-terminal amino acid sequence analysis of the isolated peptides, generated by CNBr cleavage of D1R 1-545 in the absence of UV photo-linkage. Based on the location of methionine residues in At various times, the crosslinking reaction was quenched with 10 l of 4 ϫ sample buffer, the samples were boiled for 5 min, and the radiolabeled enzyme was separated from free ATP by SDS-PAGE. B, cross-linking reactions were as in the standard reaction described except UV photo-linkage was measured as a function of D1R 1-545 concentration. C, photo-linkage reactions were carried out as in A, in the presence of varying concentrations of ATP. For all three experiments, the moles of cross-linked product was quantified by Cerenkov counting of radiolabeled enzyme excised from the dried gel. D1R 1-545 primary sequence, complete digestion by CNBr should yield six fragments whose respective molecular weights and location are depicted in Fig. 8A. The typical cleavage products, including some partial digestion fragments, are represented by the Coomassie-stained Tricine gel in Fig. 6B. NH 2terminal amino acid sequence analysis of the 15-kDa fragment demonstrates that this peptide is a partial cleavage product (denoted by the P 2 in Fig. 8B) corresponding to the 10.0-and 5.2-kDa fragments. On the basis of size, we also conclude that P 1 is a partial cleavage product comprised of the 25.1 and 2.1 fragments. The persistence of these partial cleavage products indicate that the respective methionines at positions 221 and 410 in the primary sequence are less efficiently cleaved by CNBr.
A low cross-linking efficiency (less than 1% ATP cross-linked to D1R 1-545 at saturating ATP) prohibited the isolation of a specific radiolabeled peptide in sufficient amounts for sequencing. However, an analysis of the distribution of radiolabel in the CNBr cleavage products of ATP photo-linked enzyme was informative, Fig. 8C, lanes 4 and 5. Greater than 75% of the radiolabeled nucleotide typically cross-linked to the aminoterminal 25.1-kDa peptide corresponding to amino acids 1-221 (Fig. 8C, lane 4) with additional radiolabeled peptides also resolved by gel electrophoresis. These additional cleavage products always represented a minor portion of the total radioactivity. The labeling of each peptide was competed by including 10 M triphosphorylated RNA in the cross-linking reaction prior to cleavage by CNBr (Fig. 8C, lane 5). Inclusion of RNA in the reaction diminished the extent of ATP cross-linking and not the pattern of radiolabeled cleavage products obtained. From these analyses, we conclude that the major ATP binding site resides within the first 221 amino acids of D1R.
In an attempt to localize the RNA binding site in D1R 1-545 , 32 P-radiolabeled RNA was UV photo-cross-linked to D1R 1-545 , the modified enzyme was treated with RNase A and was then subject to CNBr cleavage. The resulting cleavage products were resolved by electrophoresis on Tricine gels, Fig. 9. As was the case for ATP, RNA was cross-linked primarily to the 25.1-kDa NH 2 terminus of D1R, Fig. 9B, lanes 4 and 5. Additional radiolabeled peptides of lower molecular weight (not evident in Fig. 9B) were also detected in longer autoradiographic exposures. The percentage of radioactivity present in these fragments somewhat varied between experiments. However, these bands always represented less than 25% of the total radiolabeled peptide present. Prominent photo-linkage to the 25.1-kDa region of D1R was observed whether the RNA employed as a cross-linking reagent was radiolabeled at internal guanosines located primarily in the 5Ј region (Fig. 9, A and B, lane 4) or exclusively in the ␥ phosphate at the 5Ј terminus, Fig. 9B, lane 5. This latter result demonstrates that RNA binding to this region of D1R is specific to the 5Ј terminus of RNA and presumably at the phosphohydrolysis active site. DISCUSSION We have investigated the RNA 5Ј-triphosphatase and nucleoside triphosphate phosphohydrolase activities of the mRNA capping enzyme. In the previous report (12), we demonstrated that both activities, along with the guanylyltransferase, reside entirely within the fully active D1R 1-545 subdomain. We report here that the kinetic properties of the two phosphohydrolytic activities vary markedly. First, the two activities differ in the MgCl 2 concentration at which maximal catalysis is achieved: 2 mM MgCl 2 versus 20 mM MgCl 2 for RNA 5Ј-triphosphatase and ATPase, respectively. Second, the capping enzyme exhibits an approximately 1000-fold lower K m for RNA compared to the K m for ATP and an approximately 10-fold slower turnover of RNA compared to ATP, indicating substantive binding differences for the two substrates. Catalytically relevant binding of RNA at the RNA 5Ј-triphosphatase active site must occur in a region of the RNA downstream from the 5Ј terminus. The ability of a 5Ј OH RNA oligonucleotide to inhibit RNA 5Ј-triphosphatase activity (12) reflects this fact. Based on the influence of RNA chain length on RNA 5Ј-triphosphatase kinetics, this binding site resides within 9 nucleotides from the 5Ј terminus.
The ability of the mRNA capping enzyme to employ ATP, GTP, and triphosphorylated RNA as a substrate for phosphohydrolysis raises questions related both to substrate specificity and to the presence of a single or multiple phosphohydrolysis active site(s). In regard to the question of substrate specificity, the capping enzyme exhibits comparable substrate saturation kinetics for a broad range of nucleoside triphosphates. The ability of each NTP to serve equally as substrates indicates that the enzyme exhibits little specificity based on sugar or base. In regard to the number of phosphohydrolase active sites, we conclude, based on two kinetic observations, that a common active site exists. First, ATP is a linear competitive inhibitor of RNA 5Ј-triphosphatase activity. The calculated K i of 1.4 mM is comparable to the K m measured for ATPase activity (12), consistent with ATP acting as a competing substrate at the RNA 5Ј-triphosphatase active site. Second, ATPase and RNA 5Јtriphosphatase activities (12) are inhibited in a similarly graded manner by the same set of oligonucleotides. The degree of competition of ATPase activity by these oligonucleotides reflects the need for either a tri-or diphosphorylated 5Ј RNA terminus. The concentration at which triphosphorylated 23mer RNA effectively inhibits ATPase (with 50% inhibition occurring at approximately 35 M) is substantially higher than the approximately 1 M K m measured for RNA 5Ј-triphosphatase activity (12). However, this inequivalence may be explained by an inhibition of RNA binding at the high MgCl 2 concentration present in the kinetic competition assay; at 10 mM MgCl 2 , RNA 5Ј-triphosphatase activity is inhibited approximately 70% (Fig. 1). The apparent K m for 23-mer triphosphorylated RNA under such conditions is greater than 10-fold higher than the K m measured under optimal divalent cation conditions (data not shown). Nevertheless, the inhibition measured is both valid and specific. These competition data, taken together with the differential kinetic properties of the two activities, indicate that within this active site, triphosphorylated RNA preferentially binds. The strong substrate bias toward RNA reflects the fact that the primary function of this active site is the dephosphorylation of mRNA prerequisite to mRNA cap formation.
A major goal in our study of the mRNA capping enzyme is to identify the amino acid residues present in the active sites catalyzing each of the three reactions in the cap formation pathway. The straight-forward method of direct UV photocross-linking specific substrates to the methyltransferase domain has permitted the preliminary localization of the Sadenosylmethionine and GTP binding regions of the methyltransferase active center (17,18). Using this same general approach, we attempted to localize the phosphohydrolase active site in the D1R 1-545 subdomain. Specific UV photo-crosslinking of ATP to D1R 1-545 was achieved as evidenced by several criteria. First, the extent of cross-linking was found to be dependent upon time and enzyme concentration, and was saturable by ATP. Second, the cross-linking of ATP to D1R 1-545 was effectively inhibited by triphosphorylated 23-mer RNA in a manner which paralleled the inhibition of ATP hydrolysis by this RNA. In contrast, reagents such as single-stranded DNA, and the dinucleotides ApU and GpppA (a substrate for methyltransferase) which were poor inhibitors of ATPase activity were likewise poor inhibitors of ATP cross-linking. Direct photo-linkage of ATP to D1R 1-545 , coupled with the specific chemical cleavage of this subdomain by CNBr, has identified a major ATP binding site in a 25 kDa amino terminus of D1R from amino acids 1-221. The minor but specific labeling of other peptides indicate, however, that other regions in D1R 1-545 interact with ATP. The observation of multiple radiolabeled cleavage products is not surprising given the approach taken. Direct photo-cross-linking of nucleotides to proteins is likely to result in a covalent linkage to any amino acids in juxtaposition to the nucleotide with a most efficient linkage to those residues that are most proximal to the adenine base.
Using this same experimental approach, the major RNA binding site was also localized to the same NH 2 -terminal 25-kDa region of D1R to which ATP binds. RNA cross-linking to this region represents specific binding to the RNA 5Ј-triphosphatase active site based on the following observations. First, photo-linkage to the same NH 2 -terminal 25-kDa fragment occurred when either ␣or ␥-32 P-radiolabeled 23-mer RNA was employed as a ligand. This result clearly indicates that D1R is binding to the 5Ј terminus of RNA and protecting this region of RNA from RNase digestion. Second, kinetic competition data demonstrate that both ATPase and RNA 5Ј-triphosphatase activities share a common active site. Therefore, the observation that both ATP and RNA cross-link to the same region of D1R is consistent with both substrates binding at the same phosphohydrolysis site.
The identification of specific residues involved in nucleotide binding was not possible by the photo-cross-linking method as a low cross-linking efficiency prohibited the isolation of a modified peptide in amounts sufficient for sequencing. Moreover, an evaluation of the D1R amino acid sequence failed to reveal any consensus sequences in D1R 1-545 for nucleotide binding. However a few sequences which show some homology to nucleotide binding sequences should be noted. In particular is the sequence GSGAQSKS located at position 167-174. This sequence falls within the 25-kDa amino-terminal CNBr peptide representing the major ATP (and RNA) cross-linking product and loosely fits the Walker A NTP binding motif GXXXXGK-(S/T) (19). Another sequence in D1R which partially matches this motif is at position 434 -443 and consists of the sequence GX 5 GXGK. This sequence lacks the typical S or T at the carboxyl terminus. However, following this sequence is a hydrophobic stretch of amino acids comprising a sequence motif reported for some herpesvirus proteins possessing ATPase functions (20). Other consensus sequences for nucleotide bind-ing including the Walker B type, (R/K)X 3 GX 3 L-hydrophobic-D (19), and the GTP binding motif, NKXD (25), are not present in D1R 1-545 . This is not surprising given that the preferred substrate for phosphohydrolysis is RNA and not single nucleoside triphosphates.
An RNA 5Ј-triphosphatase activity for the West Nile virus has been reported (21). The authors have suggested the amino acids ILRPRW as comprising the putative RNA 5Ј-triphosphatase active site and note its homology to the vaccinia virus mRNA capping enzyme sequence LKPR starting at position 492. This sequence is based entirely on homology to other flaviviruses in a region of the viral genome to which no functional predictions had been previously made. No structurefunction correspondence has actually been demonstrated.
By analysis of D1R carboxyl-terminal deletion mutations, both the guanylyltransferase and ATPase activities have been mapped to the same 60 kDa amino-terminal domain. This fact suggests a close structural linkage between the two activities (12). However, the inability of ATP to inhibit either EϳGMP formation or the transfer of GMP to RNA clearly demonstrates that the two active sites do not overlap. This conclusion is noteworthy given that GTP is a substrate for both reactions. Based on the apparently rate-limiting kinetics for the transfer of GMP to RNA (2,12), one would also expect some degree of inhibition by any nucleotide binding which interferes with this transfer. Clearly ATP does not inhibit this step. Two additional lines of evidence also argue for separate active sites. First, ATP does not effectively cross-link to the 14.2-kDa CNBr fragment containing the lysine 260 required for EϳGMP formation. This result demonstrates that the major ATP binding site is not proximal to this lysine, at least not in the primary sequence. Second, the mutation of this lysine to methionine, while eliminating EϳGMP formation (9,16), had no effect on ATPase activity, indicating that this lysine is not required for phosphohydrolase activity.
Employing a 23-mer triphosphorylated RNA substrate in vitro, D1R 1-545 exhibits a 50-fold greater RNA 5Ј-triphosphatase than guanylyltransferase activity (12); 50 dephosphorylations occur for each nucleotidyl transfer event. This observation implies that the probability of dissociation of diphosphorylated RNA from the RNA 5Ј-triphosphatase active site is 50-fold greater than the transfer of GMP to the RNA 5Ј terminus. Since guanylyl transfer measured in vitro is linear with respect to time at the earliest time points analyzed (data not shown), the diphosphorylated RNA must serve as an acceptor for this transfer without dissociating from the enzyme. The poor coupling of guanylyltransferase to RNA 5Ј triphosphatase in vitro demonstrates, however, that the capping enzyme lacks the ability to effectively retain the nascent diphosphorylated RNA product and transfer it to the guanylyltransferase active site. It is believed that nascent RNA is capped in vivo without the release of the intermediate products. This presumed greater in vivo catalytic efficiency is likely to be due to the existence of a ternary complex containing the capping enzyme, nascent RNA, and the RNA polymerase (26,27). A tethering of the nascent RNA to the RNA polymerase rather than to a site in the mRNA capping enzyme (12) would retain the intermediate products and ensure a proximity of substrate RNA to each capping enzyme active site. In turn, this favorable arrangement would increase the efficiency of catalysis by physically coupling each reaction in the cap formation pathway.