Poliovirus RNA-dependent RNA polymerase (3D(pol)). Divalent cation modulation of primer, template, and nucleotide selection.

We have analyzed the divalent cation specificity of poliovirus RNA-dependent RNA polymerase, 3D(pol). The following preference was observed: Mn(2+) > Co(2+) > Ni(2+) > Fe(2+) > Mg(2+) > Ca(2+) > Cu(2+), and Zn(2+) was incapable of supporting 3D(pol)-catalyzed nucleotide incorporation. In the presence of Mn(2+), 3D(pol) activity was increased by greater than 10-fold relative to that in the presence of Mg(2+). Steady-state kinetic analysis revealed that the increased activity observed in the presence of Mn(2+) was due, primarily, to a reduction in the K(M) value for 3D(pol) binding to primer/template, without any significant effect on the K(M) value for nucleotide. The ability of 3D(pol) to catalyze RNA synthesis de novo was also stimulated approximately 10-fold by using Mn(2+), and the enzyme was now capable of also utilizing a DNA template for primer-independent RNA synthesis. Interestingly, the use of Mn(2+) as divalent cation permitted 3D(pol) activity to be monitored by following extension of 5'-(32)P-end-labeled, heteropolymeric RNA primer/templates. The kinetics of primer extension were biphasic because of the enzyme binding to primer/template in both possible orientations. When bound in the incorrect orientation, 3D(pol) was capable of efficient addition of nucleotides to the blunt-ended duplex; this activity was also apparent in the presence of Mg(2+). In the presence of Mn(2+), 3D(pol) efficiently utilized dNTPs, ddNTPs, and incorrect NTPs. On average, three incorrect nucleotides could be incorporated by 3D(pol). The ability of 3D(pol) to incorporate the correct dNTP, but not the correct ddNTP, was also observed in the presence of Mg(2+). Taken together, these results provide the first glimpse into the nucleotide specificity and fidelity of the poliovirus polymerase and suggest novel alternatives for the design of primer/templates to study the mechanism of 3D(pol)-catalyzed nucleotide incorporation.

Positive-strand RNA viruses represent an existing and emerging threat to the United States public health. For example, as many as 4 million Americans are currently infected by hepatitis C virus. Hepatitis C virus is capable of establishing a persistent infection, which leads to cirrhosis of the liver and, in some cases, liver cancer (1). Unfortunately, highly effective therapies to treat chronic RNA virus infection do not exist. Replication of the genomes of all RNA viruses requires the virus-encoded RNA-dependent RNA polymerase (RdRP) 1 (2).
RdRP activity is unique to virus-infected, human cells; therefore, the RdRP is a suitable target for the development of antiviral agents. Although RdRPs from several viruses have been purified and characterized to some extent, a large gap remains in terms of our understanding of the biochemical mechanism of this class of nucleic acid polymerase relative to those classes of polymerase involved in cellular processes such as replication and transcription (2)(3)(4)(5)(6)(7)(8). A detailed kinetic and thermodynamic description of RdRP-catalyzed nucleotide incorporation should permit this enzyme to be distinguished from cellular polymerases, thus facilitating the development of RdRP-specific inhibitors useful for the treatment of RNA virus infection.
Biological and biochemical studies of poliovirus genome replication have been ongoing for decades (9). These studies have shown that RNA structures at the 3Ј-end of the genome (10), and possibly at the 5Ј-end of the genome (11), specify the site of assembly of the replication complex. The exact composition and order of assembly of this complex remains to be determined. However, both viral and host factors have been implicated in replicase assembly and/or function (12)(13)(14)(15). After complex assembly, poliovirus RNA-dependent RNA polymerase, 3D pol , initiates RNA synthesis by using the protein primer, 3B (VPg). 3D pol has been studied intensively for many years because of its key role in poliovirus genome replication. Therefore, this enzyme is an ideal model system to use in the study of RdRP mechanism and for the elucidation of RdRP structure-function relationships.
In vitro studies employing pure, active 3D pol have identified many of the biochemical properties and enzymatic activities associated with this enzyme. In addition to oligo(dT)-and oligo(rU)-dependent poly(rU) polymerase activity (3,4), 3D pol is capable of uridylylating VPg and utilizing the resulting VPg-pUpU product as a primer for poly(rU) synthesis (16). 3D pol has terminal transferase activity (17) and strand displacement activity (18). Also, 3D pol has the ability to multimerize (19), and Kirkegaard and colleagues (20) have suggested that multimerization may be required for nucleic acid binding, polymerase activity (21), and virus viability (22). Recently, we demonstrated that 3D pol is sufficient for template switching, and this enzyme is capable of catalyzing primer-independent RNA synthesis (23). Finally, a high resolution crystal structure is available for 3D pol (19). The overall topology of 3D pol is quite similar to that of the other classes of nucleic acid polymerase in that the enzyme resembles a right hand with "fingers," "palm," and "thumb" subdomains. The palm subdomain contains four struc-tural motifs (A-D) found in all polymerases, in addition to a fifth motif (E) found only in enzymes, such as reverse transcriptases, which utilize RNA templates.
Clearly, a great deal of information germane to 3D pol function exists. However, detailed kinetic and mechanistic studies of this enzyme have yet to be performed. The absence of this information greatly limits the extent to which structural information can be exploited to establish the structure-function relationships of this class of polymerase. Detailed kinetic and mechanistic investigations of 3D pol have been limited, primarily, by the inability to establish stoichiometric complexes between 3D pol and primer/template that permit polymerase activity to be monitored by following the extension of end-labeled, heteropolymeric RNA primers. One possible explanation for this is that 3D pol has a low affinity for nucleic acid. We have shown that the K M value of 3D pol for short, homopolymeric primer/templates is in the 10 -20 M range (23), and Kirkegaard and colleagues (20) have reported K d values for 3D pol binding to nucleic acid that are in the M range.
In this report, we have extended our systematic, quantitative analysis of 3D pol by evaluating the divalent cation specificity of this enzyme. Taken together, the data described herein provide evidence for functional similarity between the RdRP and DNA polymerases and suggest novel strategies for the design of primer/template substrates to investigate 3D pol mechanism.

Expression and Purification of 3D pol
Expression and purification of 3D pol was performed as described previously (23,24).

Purification of Synthetic Oligonucleotides
DNA and RNA oligonucleotides were purified by denaturing PAGE. Gels consisted of: 19% acrylamide, 1% bisacrylamide, 7 M urea and 1 ϫ TBE (89 mM Tris base, 89 mM boric acid, and 2 mM EDTA). The oligonucleotide ladder was visualized by UV shadowing. A gel slice containing only the full-length oligonucleotide was removed, and the nucleic acid was electroeluted from the gel in 1 ϫ TBE by using an Elutrap apparatus (Schleicher & Schuell). Oligonucleotides were desalted on Sep-Pak columns (Millipore) as specified by the manufacturer. Oligonucleotides were typically suspended in T 10 E 1 (10 mM Tris, 1 mM EDTA, pH 8.0), aliquoted, and stored at Ϫ80°C until use. Concentrations were determined by measuring the absorbance at 260 nm by using calculated extinction coefficients (25).

Purity of [␣-32 P]NTPs
[␣-32 P]NTPs were diluted to 0.1 Ci/l in ddH 2 O, and 1 l was spotted in triplicate onto TLC plates. TLC plates were developed in 0.3 M potassium phosphate, pH 7.0, dried, and exposed to a PhosphorImager screen. Imaging and quantitation were performed by using the ImageQuant software from Molecular Dynamics. The purity was used to correct the specific activity of NTP in reactions to calculate accurate concentrations of product. Purity was checked before or after each experiment and ranged from 50 to 90%.

5Ј-32 P Labeling of Oligonucleotides
DNA and RNA oligonucleotides were end-labeled by using [␥-32 P]ATP and T4 polynucleotide kinase essentially as specified by the manufacturer. Reactions typically contained 11 M [␥-32 P]ATP, 10 M DNA, or RNA oligonucleotide, and 0.4 unit/l T4 polynucleotide kinase. Unincorporated nucleotide was removed by passing the sample over two consecutive 1-ml Sephadex G-25 (Sigma) spun columns.

5Ј-32 P Labeling of DNA Ladder
Labeling of the DNA ladder was performed by using [␥-32 P]ATP and T4 polynucleotide kinase as specified by Life Technologies, Inc.

5Ј-32 P Labeling of GpG
GpG was end-labeled by using [␥-32 P]ATP and T4 polynucleotide kinase essentially as specified by the manufacturer. Reactions typically contained 1 M [␥-32 P]ATP, 10 M GpG, and 0.4 unit/l T4 polynucleotide kinase. Reactions were quenched by heating the reaction at 60°C for 5 min.

Annealing of Heteropolymeric Primer/Templates
1 M end-labeled RNA primer was mixed with 9 M unlabeled RNA primer and 10 M unlabeled RNA template in T 10 E 1 and heated to 90°C for 1 min and slowly cooled to 10°C at a rate of approximately 5°C/min in a Progene thermocycler.

3D pol Assays
Reactions contained 50 mM HEPES, pH 7.5, 10 mM 2-mercaptoethanol, 5 mM MgCl 2 or MnCl 2 , 60 M ZnCl 2 , 500 M NTP, primer/template and 3D pol . Reactions were quenched by the addition of EDTA to a final concentration of 50 mM. Specific concentrations of primer/template and 3D pol , along with any deviations from the above, are indicated below or in the appropriate figure legend.

Divalent Cation Modulation of 3D pol Poly(rU) and
Poly(rG) Polymerase Activity

Steady-state Kinetic Analysis of 3D pol
Kinetic constants, K M and V max , were determined by using the assay described above. The concentration of 3D pol employed in these experiments ranged from 0.01 to 0.5 M depending upon the substrate and cation employed. The V max values reported in Table III have been normalized to 0.01 M 3D pol to facilitate comparison of the various substrates. Concentrations of the varied substrate, nucleic acid or nucleotide, ranged from 0.25 ϫ K M to 4 ϫ K M . The concentration of the substrate that remained constant was 5-10 ϫ K M . Single time points were taken that were in the linear range for product formation. Reaction rates were plotted as a function of substrate concentration, and these data were fit to a hyperbola by nonlinear regression using the program KaleidaGraph (Synergy Software, Reading, PA) to obtain the kinetic constants. In one instance, the determination of the K M value for GTP in the presence of MgCl 2 , the enzyme was not saturated with dG 15 /rC 30 primer/template, thus the K M (app) is reported. However, the true K M was calculated by using Equation 1. , where X ϭ Zn 2ϩ , Cu 2ϩ , Ca 2ϩ , Mg 2ϩ , Fe 2ϩ , Ni 2ϩ , Co 2ϩ , and Mn 2ϩ . Reactions were initiated by addition of 3D pol and incubated at 30°C for 10 min. Reaction volumes were 25 l. In all cases, 5 mM was the optimal concentration for maximal activity. Products were analyzed by DE81 filter binding. Product Analysis DE81 Filter Binding-10 l of the quenched reaction was spotted onto DE81 filter paper discs and dried completely. The discs were washed three times for 10 min in 250 ml of 5% dibasic sodium phosphate and rinsed in absolute ethanol. Bound radioactivity was quantitated by liquid scintillation counting in 5 ml of Ecoscint scintillation fluid (National Diagnostics).

Phosphatase Treatment of [␣-32 P]GTP-and [␥-32 P]GTP-labeled RNA Products
TLC-1 l of the quenched reaction was spotted onto TLC plates. TLC plates were developed in 0.3 M potassium phosphate, pH 7.0, dried, and exposed to a PhosphorImager screen.
Denaturing PAGE-Sample preparation and electrophoresis were as described previously (26). Briefly, 1 l of the quenched reaction was added to 9 l of loading buffer: 90% formamide, 50 mM Tris borate, 0.025% bromphenol blue, 0.025% xylene cyanol and where appropriate a 10-fold excess of unlabeled RNA (trap strand) relative to the endlabeled RNA under investigation was added. Samples were heated to 70°C for 2-5 min prior to loading 5 l on a 1ϫ TBE, 7 M urea polyacrylamide gel of the appropriate percentage. Highly cross-linked gels contained 2% bisacrylamide. Electrophoresis was performed in 1 ϫ TBE at 75 watts. Gels were visualized and quantitated by using a PhosphorImager.

RESULTS
Transition Metals Support 3D pol -catalyzed Nucleotide Incorporation-We determined the effect of Mn 2ϩ on 3D pol -catalyzed nucleotide incorporation with the following substrates: dT 15 / poly(rA), dT 15 /rA 30 , dG 15 /poly(rC), and dG 15 /rC 30 (Table I). The stimulation of 3D pol activity observed by using Mn 2ϩ was substrate-dependent and varied from 5-to 100-fold the activity determined in the presence of Mg 2ϩ (Table I). The 10-fold decrease in activity with dT 15 /poly(rA) was due, most likely, to a decrease in the solubility in the presence of Mn 2ϩ as a white precipitate could be observed after centrifugation of this reaction mixture. The precipitate formed in the presence or absence of enzyme. This phenomenon was not observed with other primer/template substrates.
The optimal concentration for maximal 3D pol activity was 5 mM (Table II). Concentrations of divalent cation greater than 10 mM were inhibitory (Table II). The observed inhibition did not appear to be due to precipitation of nucleic acid and/or enzyme. The K M value for primer/template substrates was reduced by an average of 25-fold in the presence of Mn 2ϩ relative to the corresponding values measured in the presence of Mg 2ϩ (Table III). A 3-fold reduction in V max was observed for dT 15 /rA 30 by using Mn 2ϩ instead of Mg 2ϩ ; however, a 30-fold increase in V max was observed for dG 15 /rC 30 by using Mn 2ϩ instead of Mg 2ϩ . The K M values for UTP and GTP with the corresponding primer/template substrates were similar, 62 and 116 M, respectively, and the V max values were as expected based on the kinetic analysis of primer/template substrates discussed above. The increase in V max observed in the presence of Mn 2ϩ when dG 15 /rC 30 was employed did not result from a change in K M value for GTP (Table III). Thus, an increase in the number of productive 3D pol -dG 15 /rC 30 complexes formed may occur by using Mn 2ϩ instead of Mg 2ϩ . Mn 2ϩ , Co 2ϩ , Ni 2ϩ , and Fe 2ϩ supported higher levels of activity than Mg 2ϩ (Table  IV). Ca 2ϩ and Cu 2ϩ supported lower levels of activity than Mg 2ϩ (Table IV). Zn 2ϩ was incapable of supporting activity (Table IV).
Mn 2ϩ Increases the Efficiency of 3D pol -catalyzed RNA Synthesis de Novo-Recently, we reported that 3D pol initiates RNA synthesis when poly(rC) and GTP are employed as the sole substrates (23). Primer-independent RNA synthesis did not result from a polynucleotide-phosphorylase-like activity as template was required (data not shown). Poly(rC) and GTP appear to be the most efficient substrates for this reaction as neither poly(rA) and UTP nor poly(rU) and ATP could be used to demonstrate convincingly synthesis of RNA (data not shown). In addition, rC 30 , but not dC 30 , was also a template for primer-independent RNA synthesis, albeit at a level 30-fold lower than observed by using poly(rC) ( Table II). As shown in Table II, Mn 2ϩ stimulated primer-independent RNA synthesis by 15-fold relative to reactions performed in the presence of Mg 2ϩ when either poly(rC) or rC 30 was employed as template.
In the presence of Mn 2ϩ , dC 30 was also utilized as a template (Table II) and utilization of the dC 30 template by 3D pol was now only 2-fold less efficient than utilization of the rC 30 template. Products of this reaction resolved by denaturing PAGE and  visualized by phosphorimaging were greater than unit length and similar in appearance to those produced via template switching when dG 15 /rC 30 and GTP are employed as substrates (23).
What is the mechanism of initiation of these long RNA products? One possibility is that the terminal transferase activity of 3D pol adds GMP to the 3Ј-end of rC 30 , thus creating a "snapback" substrate that is efficiently extended by 3D pol . However, by using a 5Ј-end-labeled rC 30 template, we were only able to show the incorporation of a single GMP into rC 30 (data not shown). Moreover, the kinetics of formation of this product were too slow to support the hypothesis that rC 30 G RNA was the substrate used by 3D pol to produce long products (data not shown). A second possibility is that RNA synthesis is initiated de novo. To test this possibility, we performed an experiment employing rC 30 and [␥-32 P]GTP as substrates. If RNA synthesis initiates de novo, then long RNA products should incorporate the label. Product RNA was labeled by using [␥-32 P]GTP, thus long products most likely result from de novo initiation (Fig. 1B). The primary product of this reaction was the dinucleotide, pppGpG. Also, tri-, tetra-, and pentanucleotide products were observed. The dinucleotide product was assigned based upon the comigration of this product with 32 pGpG on polyacrylamide gels (data not shown). To keep a complete inventory of all products formed during the course of the reaction when [␥-32 P]GTP was employed, reaction mixtures were also resolved by TLC; over half of the nucleotide was utilized based on PP i accumulation (Fig. 1C). Phosphatase treatment of [␥-32 P]GTP-labeled RNA showed a time-dependent loss of label by using the DE81 filter paper method ( Fig. 2A) and PAGE (Fig. 2B) without any change in label associated with the control, [␣-32 P]GTP-labeled RNA (Fig. 2, A and B). Greater than 95% of the counts associated with [␥-32 P]GTP-labeled RNA originated from [␥-32 P]GTP, thus confirming that this RNA was initiated de novo.
Quantitative analysis of the kinetics of product formation in reactions employing either [␥-32 P]GTP or [␣-32 P]GTP as substrates showed that both reactions displayed burst kinetics (Fig. 3A). In both cases, the steady-state rates (linear phases) of product formation were identical; however, the burst amplitude of PP i formation measured by TLC was 5-fold greater than that of RNA measured by using the DE81 filter binding method. This difference likely reflects the inability of DE81 filter paper to retain dinucleotide product. Although the burst of PP i formation cannot be used directly to quantitate active sites, the burst can be exploited to compare the "active" fraction of various enzyme preparations.
In contrast, quantitation of the kinetics of [␥-32 P]GTP-labeled RNA formation should report directly on the concentration of active sites if a burst of product formation is observed. When this analysis was performed, an apparent burst of labeled RNA was observed (Fig. 3B). However, based on the concentration of enzyme employed, the burst amplitude was 6 -7-fold greater than the enzyme concentration. Because the enzyme concentration was determined by measuring the protein absorbance at 280 nm under denaturing conditions and using a calculated extinction coefficient of 71,830 M Ϫ1 cm Ϫ1 (24,27,28), it is unlikely that the enzyme concentration was grossly underestimated. The most reasonable explanation for this observation is that multiple rounds of abortive initiation events occur producing dinucleotide product prior to synthesis of completely elongated RNA.
Steady-state kinetic analysis of this reaction in the presence of Mn 2ϩ showed that the catalytic efficiency of this reaction is greater than either of the primer-dependent reactions characterized (Table III). The K M value of 3D pol for GTP in the de novo reaction was virtually identical to that measured for GTP in the primer-dependent reaction (Table III). The ability to saturate the enzyme with reasonably low levels of template RNA coupled with the high catalytic efficiency could be useful for the rapid characterization of the nucleic acid binding properties of 3D pol by evaluating the ability of "competitor" nucleic acids to inhibit RNA synthesis de novo.
Mn 2ϩ Stimulates 3D pol -catalyzed Extension of Heteropolymeric RNA Primer/Templates-The observation that the number of productive 3D pol -dG 15 /rC 30 complexes that formed was increased by using Mn 2ϩ instead of Mg 2ϩ suggested that an increase in the utilization of heteropolymeric RNA primer/ templates might also be observed by using Mn 2ϩ . Two different primer/template substrates were employed (see Fig. 4, A and  D). Both substrates consist of a 15-nucleotide primer and a 21-nucleotide template, which when annealed form a primer/   4B). Once the primer was extended to the end of template, however, additional nucleotides (ϳ20) were added, most likely a result of slippage synthesis. Products consistent with template switching were not observed. With primer/template II, where a single round of nucleotide incorporation should occur by using ATP as the sole nucleotide substrate, the first nucleotide was incorporated and misincorporation was noted (Fig.  4E). However, with each round of misincorporation, subsequent cycles of misincorporation became less efficient as very few primers could be extended to the end of template.
In both cases, the kinetics of primer extension were biphasic (Fig. 4,C and F). The first phase was faster than could be measured by manual quenching of the reaction. When primer/ template I was employed, the amplitude of the first phase represented 65% of this substrate. When primer/template II was employed, the amplitude of the first phase represented 30% of this substrate. Whereas 85% of primer/template I was utilized during the course of the reaction, only 70% of primer/ template II was utilized. When the kinetics of primer extension in the presence of Mn 2ϩ from primer/template II were compared with the kinetics in the presence of Mg 2ϩ , the primary difference observed was that more complexes formed in the presence of Mn 2ϩ than in the presence of Mg 2ϩ , both productive (note difference in y intercept in Fig. 5A) and nonproductive (note difference in end points in Fig. 5A). This conclusion was the same whether the first correct nucleotide (Fig. 5A) or all four nucleotides (Fig. 5B) were provided. However, it should be noted that the use of all four nucleotides supported higher levels of primer extension in the presence of both Mg 2ϩ and Mn 2ϩ than the use of a single nucleotide.
3D pol Adds Nontemplated Nucleotides to Blunt-ended, Heteropolymeric RNA Primer/Templates-We reasoned that the biphasic nature of the kinetics was a reflection of enzyme binding in the "correct" orientation in some cases (fast incorporation) and in the "incorrect" orientation in others (slower incorporation). We performed experiments with primer/templates I and II in which the template strand was end-labeled instead of the primer strand (Fig. 6, A and C). Nontemplated addition of nucleotides was observed with both primer/templates in the presence of either Mn 2ϩ or Mg 2ϩ (Fig. 6, B and D). The reaction was more efficient in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . Consistent with the amplitudes observed when labeled primers were employed, the template strand of primer/template I was utilized by 3D pol with a lower efficiency than the template strand of primer/template II. The ability of 3D pol to add nontemplated nucleotides to the blunt end of an RNA primer/template was somewhat surprising. However, it has been reported previously that the reverse transcriptase from human immunodeficiency virus has a similar activity (30). Terminal transferase activity of 3D pol would yield similar results if single-stranded RNA were present in the reactions described above. We performed an experiment in which either the end-labeled primer (Fig. 7A) or template (Fig.  7C) strand of primer/template II was incubated with 3D pol , ATP, and either Mg 2ϩ or Mn 2ϩ as the divalent cation. In all cases, the kinetics and/or products of the terminal transferase reaction were substantially different from those observed by using a template-labeled primer/template (Fig. 7, B and D). The ability of 3D pol to partition in both possible orientations on heteropolymeric RNA primer/templates must be considered in any quantitative analysis of 3D pol -catalyzed RNA synthesis.
3D pol Is an RdRP and a Reverse Transcriptase-The ability to monitor 3D pol activity by primer extension permitted us to evaluate the specificity and fidelity of 3D pol -catalyzed nucleotide incorporation. In the presence of Mg 2ϩ , both the correct rNMP and dNMP were incorporated to the greatest extent (Fig.  8B). With the correct rNTP, 40% of primers were extended (Fig.  8B, lane 4). In most cases, primers were extended to the end of template and additional nucleotides were added. The addition of extra nucleotides was most likely the result of slippage synthesis. However, it is also plausible that the extra nucleotides were added in a nontemplated fashion as discussed above. With the correct dNTP, 30% of primers were extended (Fig. 8B,  lane 8). Whereas some primers were extended to the end of template, products with only a single dNMP incorporated accumulated to the greatest extent. Products greater than unit length were not observed. In the presence of Mg 2ϩ , the correct ddNMP was not incorporated at all. The n ϩ 1 product observed in lane 12 of Fig. 8B must arise from rNTP contamination of the ddNTP stock. This conclusion is based on the migration of this product through the polyacrylamide gel; the n ϩ 1 product present in lane 12 of Fig. 8B is migrating slower than expected for a ddNMP-incorporated product (cf. Fig. 8B, lane 12, and Fig. 8C, lane 26).
In the presence of Mn 2ϩ , a 2-fold increase in primer utilization was observed when either the correct rNTP (Fig. 8C, lane  18) or the correct dNTP (Fig. 8C, lane 22) was employed. 84% of primers were extended when the correct rNTP was utilized; and 72% of primers were extended when the correct dNTP was utilized. Interestingly, by using Mn 2ϩ as the divalent cation, the correct ddNTP was utilized, and 34% of primers were extended (Fig. 8C, lane 26). The efficiency of correct rNMP and dNMP incorporation was also stimulated by using Mn 2ϩ . In both cases, 90% of extended primers reached the end of template and addition of extra nucleotides was enhanced significantly (cf. Fig. 8B, lanes 4 and 8, and Fig. 8C, lanes 18 and 22).
Mn 2ϩ Decreases the Fidelity of 3D pol -catalyzed Nucleotide Incorporation-In the presence of Mg 2ϩ , the efficiency of utilization of incorrect rNTPs (Fig. 8B, lanes 1-3), dNTPs (Fig. 8B,  lanes 5-7), and ddNTPs (Fig. 8B, lanes 9 -11) was less than 20% of the value measured for incorporation of the correct rNTP. Incorrect rNTPs were utilized better than incorrect dNTPs, and ddNTPs were not utilized at all. Again, as discussed above, the n ϩ 1 products observed in lanes 9 -11 of Fig. 8B must result from rNTP contamination of the ddNTP stocks. In the presence of Mn 2ϩ , incorrect rNTPs were utilized efficiently. 53, 59, or 68% of primers were extended by using ATP (Fig. 8C,  lane 15), CTP (Fig. 8C, lane 16), or GTP (Fig. 8C, lane 17), respectively, as the incorrect rNTP. In most instances, primers could not be extended to the end of template. The most significant accumulation of products was in the n ϩ 1 to n ϩ 3 range. Whereas utilization of dGTP was increased in the presence of Mn 2ϩ (21% of primers extended), utilization of the other two incorrect dNTPs was similar to that observed in the presence of Mg 2ϩ . The use of Mn 2ϩ had very little effect on utilization of incorrect ddNTPs.
Finally, in some instances, an apparent activation of primer cleavage was observed. This increase in primer cleavage was apparent in the presence of Mg 2ϩ (lanes 6, 9, and 10 of Fig. 8B) and was stimulated by using Mn 2ϩ (lanes 15, 19, 20, 21, 23, 24, and 25 of Fig. 8C). Primer cleavage was not due to ribonuclease contamination of our 3D pol preparations as cleavage was not observed in all reactions. Moreover, in most cases, ribonuclease activity does not require a divalent cation, thus an activity difference in the presence of Mg 2ϩ relative to Mn 2ϩ would not be expected (31). Interestingly, primer cleavage occurred primarily after misincorporation of nucleotides. Primer cleavage was not evident in reactions incorporating correct ribonucleotides (lane 18, Fig. 8C), deoxynucleotides (lane 22, Fig. 8C), or dideoxynucleotides (lane 26, Fig. 8C) Additional studies will be necessary to determine the molecular basis for this observation. DISCUSSION We have performed a comprehensive, quantitative evaluation of the divalent cation specificity of poliovirus RNA-dependent RNA polymerase, 3D pol . The primary, universal effect of Mn 2ϩ on 3D pol was a substantial (20 -30-fold) reduction in the K M value of the enzyme for primer/template (Table III). 3D pol activity was stimulated by an additional 20-fold over that expected based solely on the reduction in K M value for primer/ template when dG 15 /rC 30 was analyzed in the presence of Mn 2ϩ (Table III). This additional increase in activity was not due to changes in the K M value of 3D pol for GTP as this value was not affected significantly by using Mn 2ϩ instead of Mg 2ϩ (Table  III). Therefore, we concluded that by using Mn 2ϩ the number of productive 3D pol -dG 15 /rC 30 complexes formed was increased by 20-fold relative to the number formed by using Mg 2ϩ .
The ability of nucleic acid polymerases to utilize transition metals, especially Mn 2ϩ , as the divalent cation cofactor instead of Mg 2ϩ is well established (33)(34)(35)(36)(37). The primary effect of Mn 2ϩ relative to Mg 2ϩ is that nucleotide specificity is relaxed, that is nucleotides with the inappropriate sugar or base can be incorporated more efficiently (35,36). The classic explanation for the observed relaxation in nucleotide specificity in the presence of Mn 2ϩ is that fewer geometrical constraints exist with this divalent cation for coordination of the nucleotide phosphates and active site ligands, which is a prerequisite to phosphoryl transfer (36). In fact, even for 3D pol , it has been shown by Morrow and colleagues (38) that the use of transition metals as divalent cation cofactor for this enzyme can overcome, to some extent, effects of mutations at positions of the enzyme that alter the position or identity of the active site residues that are involved in metal coordination. Also, it has been noted that Mn 2ϩ is capable of relaxing template specificity (39). In this regard, it worth noting that template specificity of 3D pol was also relaxed by using Mn 2ϩ ; a DNA template supported RNA synthesis in the presence of this cofactor (Table II).
In this study, we observed a dramatic reduction in the K M value for primer/template by employing Mn 2ϩ as the divalent cation cofactor instead of Mg 2ϩ . To date, similar observations have not been made for any other nucleic acid polymerase. However, the ability of Mg 2ϩ to increase formation of productive polymerase-nucleic acid complexes has been noted previously by Modrich and colleagues (40). It has been shown that FIG. 4. Primer Extension by 3D pol in MnCl 2 . A, 15/21-mer primer/template I employed in the experiment described in B. B, reactions contained 3D pol (5 M), end-labeled primer/template I (1 M), UTP (500 M), and MnCl 2 (5 mM). Reactions were initiated by addition of 3D pol and incubated at 30°C; reactions were quenched at the indicated times by addition of EDTA to a final concentration of 50 mM. Products were resolved by electrophoresis on a denaturing 20% polyacrylamide gel. C, kinetics of primer extension in the reaction described in B were determined by quantitating product by using the ImageQuant software. D, 15/21-mer primer/template II employed in the experiment described in E. E, reactions contained 3D pol (5 M), end-labeled primer/ template II (1 M), ATP (500 M), and MnCl 2 (5 mM). Reactions were initiated by addition of 3D pol and incubated at 30°C; reactions were quenched at the indicated times by addition of EDTA to a final concentration of 50 mM. Products were resolved by electrophoresis on a denaturing 20% polyacrylamide gel. F, kinetics of primer extension in the reaction described in E were determined by quantitating product by using the ImageQuant software.
transition metals, such as Mn 2ϩ , bind much more tightly to the phosphodiester backbone of nucleic acid than Mg 2ϩ (41). Therefore, it is plausible that stability and/or concentration of primer/template duplex is increased due to the enhanced charge neutralization of the phosphodiester backbone in the presence of Mn 2ϩ . If more primer/template duplex exists at lower concentrations in the presence of Mn 2ϩ than in the presence of Mg 2ϩ , then an apparent reduction in the K M value for primer/ template would be observed as this is the competent form of the substrate.
Alternatively, it is possible that 3D pol has not evolved to bind to a charged template, that is the enzyme is incapable of effectively neutralizing the phosphodiester backbone. The strong binding of Mn 2ϩ to the phosphodiester backbone would overcome this problem thereby increasing the affinity of 3D pol for primer/template, in addition to possibly increasing the number of complexes that form. If this hypothesis is correct, then a mechanism for neutralization of the biological templates might exist. The virus-encoded 3AB protein may fulfill such a role because it has nonspecific RNA binding activity (42) and can increase the use of homo-and heteropolymeric primer/templates (43,44). Similar scenarios have been well established for negative-strand RNA viruses such as Sendai virus (45).
We reported previously that 3D pol is capable of primer-independent RNA synthesis when poly(rC) is employed as template (23). In this study, we demonstrated that primer-independent RNA synthesis resulted from initiation de novo and was also stimulated by using Mn 2ϩ as the divalent cation cofactor. Overall, the reaction sequence employed by 3D pol in catalyzing RNA synthesis de novo is quite similar to that observed for the same type of reaction catalyzed by replicases for RNA viruses such as Q␤ (46) and brome mosaic virus (47). It is currently unclear whether RNA synthesis de novo catalyzed by 3D pol has any biological significance. However, the ability of a polymerase that clearly uses a protein primer in vivo to support RNA synthesis de novo has significant implications on the conclusions that should be drawn when similar observations are made with polymerases from RNA virus systems which lack significant biological characterization. For example, a recent report on the RdRP from bovine viral diarrhea virus showed that this enzyme is capable of initiating RNA synthesis de novo (48). However, in the absence of data characterizing the 5Ј-end of viral RNA, it may be premature to completely rule out the possibility of primed synthesis in the mechanism of initiation of pesti-and hepacivirus genome replication.
The kinetics of primer extension were biphasic with both heteropolymeric primer/template substrates employed (Fig. 4, C and F). We anticipated that the reaction would be monophasic with the kinetics of formation of extended primers being described best by a single exponential. This assumption was based on the fact that the K M values measured for dT 15  and dG 15 /rC 30 in the presence of Mn 2ϩ were in the 1 M range, and a 3D pol concentration of 5 M was employed in this reaction, thus approximately 90% of the primer/template should be bound to enzyme.
One possible explanation for biphasic kinetics given the aforementioned assumptions was that two different 3D polprimer/template complexes formed. Whereas one complex would be competent for primer extension (first, fast phase), the other would be unproductive requiring some type of rearrangement of the initial enzyme-primer/template complex or enzyme dissociation from primer/template prior to formation of a complex that was competent for primer extension (second, slow phase). By employing a template-labeled primer/template, it was apparent that the enzyme was capable of binding to primer/template in both orientations and adding nucleotides to the blunt-end of the duplex (Fig. 6). That the addition was not terminal transferase activity was ruled out by qualitative and quantitative comparison of the single-stranded RNA primer or template (Fig. 7). Therefore, the "lost" fraction was found, and the slow phase, most likely, reflected dissociation of the enzyme from the unproductive conformation to bind to primer/template in the productive conformation.
Partitioning of the enzyme between the productive and unproductive conformations was not equal and differed for the two primer/template substrates employed in this study. Both substrates are identical in length of primer, template, and duplex region and only differ in three readily apparent ways. First, the sequence around the primer/template junction and template overhang are different. Second, there is a subtle (10°C) difference in the calculated T M values for the two primer/templates. Third, by using primer/template I and UTP, the enzyme can extend to the end of template, whereas by using primer/template II and ATP, extension to the end of template is not efficient as it requires misincorporation. This third possibility was ruled out as being a significant factor by showing that partitioning of primer/template II was not affected when reactions were performed in the presence of all four NTPs (Fig.  5). Given the two remaining possibilities, a sequence dependence for binding seems most likely. Additional experiments will be required to clarify this issue.
However, to gain insight into the nucleotide specificity and fidelity of 3D pol , we performed a series of primer-extension experiments in which the utilization of correct and incorrect rNTPs, dNTPs, and ddNTPs was evaluated. Incorporation of dTMP was more efficient than incorporation of any of the incorrect rNMPs (cf. lane 22 and lanes 15, 16, and 17 of Fig.  8C). This result suggests that appropriate base pairing is more important for nucleotide incorporation than the presence of a 2Ј-OH, and the structural conformation of the duplex region of the nascent chain is important for processive synthesis, perhaps translocation. That the 2Ј-OH is recognized to some extent by 3D pol was evident by evaluating misincorporation in the presence of Mn 2ϩ . Whereas all three incorrect rNTPs could be utilized by 3D pol , only one incorrect dNTP was utilized. However, this observation may result from conformational differences between ribose and deoxyribose other than the difference at the C2Ј position. Thus, 3D pol appears to utilize a two-step process for nucleotide selection. In the first step, the ability of the nucleotide to pair with the template is recognized; in the second step, the composition of the sugar is recognized. The use of Mn 2ϩ permits this second step of nucleotide selection to be bypassed more easily. Consecutive cycles of misincorporation become increasingly more difficult as primer extension beyond n ϩ 2 or n ϩ 3 was rare (see lanes 15-17 of Fig. 8C). Again, these data support the notion that the structural conformation and/or integrity of the duplex region of nascent chain is important for processive synthesis.
Finally, we noted that cleavage of the primer occurs in reac- tions incorporating incorrect rNMPs and dNMPs. Primer cleavage occurred in the presence of Mg 2ϩ and Mn 2ϩ but was most striking in the presence of Mn 2ϩ , because of the increased levels of misincorporation observed by using this cofactor. Primer cleavage may result from pyrophosphorolysis. If an appropriately base-paired duplex is a prerequisite to efficient translocation and PP i release requires translocation (49), then it is conceivable that after misincorporation PP i may have sufficient time to attack the misaligned duplex. However, our results would also suggest that PP i is capable of attacking phosphodiester bonds other than the ultimate bond. Of course, this reaction could provide a mechanism for error correction. Studies are in progress to characterize this reaction more completely.