|Home | About | Journals | Submit | Contact Us | Français|
Protein-priming is an important mechanism that many viruses use to initiate genomic DNA or RNA synthesis. Birnaviruses are the only double-stranded (ds) RNA viruses that use protein priming. The viral-encoded VP1 of birnavirus functions as both a polymerase and a protein primer, and is able to undergo self-guanylylation to acquire a covalently linked rGMP. By employing biochemical assays using recombinant proteins, we have shown that VP1 self-guanylylation does not require an RNA template but is dependent on divalent metal ions. VP1 reacts with all four types of rNTPs but strongly prefers rGTP. Unexpectedly, two fatal polymerase mutants D402A and E421Y, each having an essential catalytic residue mutated and unable to catalyze RNA synthesis, remain active in self-guanylylation. The guanylylation site was further mapped to the VP1 N-terminal domain. Our results support a mechanism in which VP1 self-guanylylation is catalyzed by a novel active site different from the polymerase active site.
Many viruses initiate terminal nucleic acid synthesis using protein molecules as primers. These protein primers are known as terminal protein (TP) in DNA viruses and genome-linked protein (VPg) in RNA viruses (Paul et al., 1998; Salas, 1991). Among these protein-priming viruses, many are important causative agents of serious human and animal diseases such as common colds (rhinovirus), hepatitis (hepatitis A and B viruses), poliomyelitis meningitis (poliovirus), acute febrile respiratory disease (adenovirus), and cattle foot-and-mouth diseases (foot-and-mouth disease virus). Therefore, understanding the mechanism of protein-primed viral genome replication will have important application to the development of new antiviral therapies for both human and animal diseases.
Birnaviruses (family Birnaviridae) include pathogens of economic and environmental importance, such as the infectious bursa disease virus (IBDV) of the genus avibirnavirus, infectious pancreatic necrosis virus (IPNV) and blotched snakehead virus (BSNV) of the genus aquabirnavirus, and Drosophila X virus (DXV) of the genus entomobirnavirus (Da Costa et al., 2003; Leong, 2000). Mature IBDV virion contains two dsRNA segments, RNA-A and RNA-B, that are covalently linked to the viral RNA-dependent RNA polymerase (RdRp) VP1 at the 5′ end. Birnaviruses are the only dsRNA viruses that initiate RNA synthesis using protein primers. The protein-priming function in birnaviruses is carried out by the polymerase VP1 itself (Calvert et al., 1991; Xu, Si, and Dobos, 2004), similar to the situation in hepatitis B virus (HBV) in which the reverse transcriptase (RT) also functions as the protein primer (Wang and Seeger, 1992; Zoulim and Seeger, 1994). The first step in birnavirus protein-priming is the covalent attachment of a GMP moiety to the hydroxyl group of a tyrosine or serine residue within the protein-primer molecule. In this initial guanylylation reaction, the GMP moiety is linked to the protein primer through the α-phosphate group via a phosphodiester bond, yielding VP1-G, which can be further guanylylated to produce VP1-GG (Dobos, 1993; Shwed et al., 2002; Spies and Muller, 1990; Xu, Si, and Dobos, 2004). It has been proposed that the -GG moiety in VP1-GG binds to the -CC sequence at the 3′ ends of the viral RNA template during the initiation of both (+) and (−) RNA synthesis (Magyar, Chung, and Dobos, 1998; Xu, Si, and Dobos, 2004). As a consequence, the 5′ ends of both the (+) and (−) strands of the bi-segmented dsRNA genome of birnaviruses are covalently attached to the viral polymerase VP1. The self-guanylylation activities of birnavirus VP1 have been demonstrated in vivo using virus-infected cells as well as in vitro using purified viral particles (Dobos, 1993; Shwed et al., 2002). In a more recent study, it has been shown that recombinant IPNV VP1 is also able to self-guanylylate (Xu, Si, and Dobos, 2004). The birnavirus VP1 guanylylation site residue has been proposed to be a serine for IBDV and IPNV and a tyrosine for DXV (Calvert et al., 1991; Dobos, 1993; Xu, Si, and Dobos, 2004).
We are interested in elucidating the structural basis of viral protein priming in genome replication using birnaviruses as a paradigm. We have determined the crystal structure of a IBDV VP1 core fragment containing residues 19 to 810 (Pan, Vakharia, and Tao, 2007), and have shown that the polymerase active site has a novel topology with the five essential polymerase motifs arranged in the permuted order of C-A-B-D-E as previously predicted (Gorbalenya et al., 2002). In an effort to elucidate the mechanism of birnavirus VP1 protein priming, we have shown that purified recombinant IBDV VP1 is able to self-guanylylate in vitro in the absence of any template. VP1 polymerase activity, however, requires an RNA template for both de novo (primer-independent) and protein primer-dependent RNA synthesis. Both guanylylation and polymerization activities require divalent metal ions such as Mg2+ and Mn2+ as cofactors. Much stronger activities were achieved when Mg2+ was replaced by Mn2+. Nucleotide triphosphate (rNTP) substrate specificity assays show that VP1 preferably reacts with rGTP over other rNTP types. A total of eight VP1 mutants were generated to probe the VP1 protein priming mechanism. To our surprise, two mutants that are completely inactive in RNA synthesis retain the self-guanylylation activity. Our results suggest that birnavirus VP1 self-guanylylation is catalyzed by an active site that does not completely overlap with or may even be different from that for nucleotide polymerization within the same molecule.
The IBDV (strain D78) VP1 coding region was cloned into pFastBacHTa (Invitrogen) using primer pairs 5′-AATCCATGGGCAGTGACATTTTCAACAGTCC and 5′-ACGCTCGAGTTAGCGGCTCTCCTTTT between the Nco I and Xho I restriction sites. The resulting plasmid pFastBacHTa::ibdvvp1 was used as a template to generate plasmids encoding the VP1 mutants D402A, N403A, S166A, T168A, E421D, E421Y,E421A, and Δ670-672 by using primer pairs containing the designated mutations. These plasmids were used to transform DH10Bac competent cells (Invitrogen) and the resulting recombinant baculovirus DNA was used to produce recombinant baculoviruses by transfecting Sf21 insect cells (Invitrogen). Recombinant VP1 contained six histidine residues and a TEV cleavage site at the N-terminus.
IBDV VP1 and the eight mutants described above were each expressed by infecting Sf21 cells at a multiplicity of infection (MOI) of 2. Infected cells were harvested 48 hours post-infection. The cell pellet was resuspended and sonicated in lysis buffer containing 50 mM Tris-HCl pH 7.5, 300 mM NaCl, 5 mM imidazole, 10% glycerol, 17 μg/ml PMSF, 1 μg/ml leupeptin, 1 μg/ml pepstatin, and 5 mM β-mercaptoethanol, and the lysate was clarified by centrifugation at 20,000×g for 20 minutes. Recombinant VP1 was purified by chromatography using a Ni-NTA column, a HiTrap Heparin-Sepharose column, a Superdex-200 column, and a Mono Q column (Amersham) as previously described (Pan, Vakharia, and Tao, 2007).
An 80-nt RNA containing the 3′ terminal sequence of the positive strand of IBDV RNA-A (5′ GUCUCCCGAC ACCACCCGCG CAGGUGUGGA CACCAAUUCG GCCUUACAAC AUCCCAAAUU GGAUCCGUUC GCGGGUCCCC) was obtained from in vitro transcription using a synthetic DNA duplex containing a T7 promoter and the above sequence of interest. RNA synthesis was carried out using the MEGAscript in vitro transcription kit (Ambion). Following the transcription reaction, the DNA template was removed by digestion with RNase-free DNase I and RNA was purified by phenol extraction and ethanol precipitation using a standard procedure. The purified RNA was suspended in nuclease-free distilled water containing 2 units/μl RNasin (Promega) and aliquots were stored at −70 °C. The quantity and the quality of RNA were estimated by agarose gel electrophoresis followed by ethidium bromide staining. The 80-nt RNA was used in all assays described in this study unless otherwise specified.
In vitro guanylylation assay was performed as previously described with minor modifications (Dobos, 1993). Briefly, pure VP1 protein samples (ca. 0.2 μM) were incubated in a 10-μl reaction mixture containing 20 mM Tris-HCl (pH 8.0), 5 mM MgCl2, 2 mM MnCl2, 4 mM dithiothreitol (DTT), 1× proteinase inhibitor cocktail (Roche), and 0.1% (v/v) NP-40. Unless otherwise indicated, the mixture was supplemented with 10 μCi of [α-32P]rGTP (3,000 Ci/mmol, MP Biomedicals).
In vitro polymerization assay was performed using procedures similar to the guanylylation assay, with several additional components including rNTPs mixture (0.01 mM GTP, 0.5 mM each of ATP, CTP, UTP), RNasin (0.4 units/μl), and RNA template (ca. 1 μM). The RNA template was the 80-nt +ssRNA as described above. The samples were incubated at 37 °C for 2 hours. The reactions were stopped by mixing with equal volumes of 2× SDS sample buffer, and boiled for 5 minutes before being subjected to SDS-PAGE. The gel was then dried and subjected to autoradiography and phosphor-imaging quantification.
The competition assays were performed by supplemented competitor nucleotides into the guanylylation assays. In brief, 10 μCi [α-32P]GTP (specific activity 800 Ci/mmol, MP Biomedicals) was premixed with serial concentrations of cold GTP or UTP in 1× guanylylation buffer. 10 μg IBDV VP1 was then added into each sample and mixed, followed by 10-minute incubation in a 37°C sand bath. The samples were then treated with 2× SDS sample buffer, boiled, and subjected to electrophoresis and qautoradiography.
To confirm the presence of nucleotide base in the labeled VP1 products, guanylylation and polymerization assays were performed as described above except that [α-32P]rGTP was replaced with 5 μM [8-3H]rGTP (5.10 Ci/mmol, GE Healthcare) or 100 μM DIG-11-UTP (3.5 mM, Roche). For [8-3H]rGTP labeling, the SDS-PAGE gel was treated with amplifying fluorographic reagent NAMP-100 (GE Healthcare) before drying and autoradiography at −70 °C. For DIG-11-UTP labeling, the reaction products in the gels were electro-transferred to a hybond-N+ nylon membrane (GE Healthcare). The DIG labeled products were visualized by an enzyme immunoassay using anti-digoxigenin antibody and the chemiluminescent substrate CSPD (Roche).
To characterize the molecular nature of the reaction products, [α-32P]rGTP labeled samples were treated with RNase-free proteinase K (Roche) or RNase V1 (Ambion). Thereafter, the treated samples were resolved by SDS-PAGE and detected by autoradiography.
To examine whether the guanylylated VP1 can be used to prime RNA synthesis, polymerase assays were performed using cold rNTPs and labeled, guanylylated VP1. First, VP1 guanylylation was performed as described above in 1× guanylylation buffer (20 mM Tris-HCl pH 8.0, 5 mM MgCl2 , 2 mM MnCl2, 4 mM DTT, 1× proteinase inhibitor cocktail, and 0.1% (v/v) NP-40 ) supplemented with [α-32P]rGTP. Thereafter, [α-32P]rGTP was removed by passing guanylylated VP1 through a NICK column (GE health) that was pre-equilibrated with 10 column volumes of 1× guanylylation buffer. Purified VP1 was further incubated for 2 hours in the 1× polymerization buffer (20 mM Tris-HCl pH 8.0, 5 mM MgCl2 , 2 mM MnCl2, 4 mM DTT, 1× proteinase inhibitor cocktail, RNasin (0.4 units/μl), and 0.1% (v/v) NP-40), 0.5 mM of each ATP, CTP, UTP and GTP, and RNA template (ca. 2 μM). The reactions samples were subjected to 12% SDS-PAGE as described above.
N-terminal His-tagged VP1 was guanylylated with [α-32P]rGTP and partially digested with trypsin at 1:500 mass ratio overnight at room temperature. The trypsin-treated sample was then incubated with Ni-NTA resin in the guanylylation reaction buffer for 4 hrs, and the resin was subsequently washed with the same buffer containing 1% Triton X-100, 100mM NaCl and 20mM imidazole, and eluted with the same buffer containing 100mM NaCl and 250 mM imidazole. The samples were analyzed on a 12.5% SDS-PAGE gel and visualized by autoradiography.
IBDV VP1 proteins used in this study were obtained by over-expression in insect cells using the baculovirus vector (Pan, Vakharia, and Tao, 2007). A 6×His tag was engineered at the N-terminus to facilitate purification unless otherwise specified. The histidine-tagged IBDV VP1 and all eight mutants (D402A, N403A, S166A, T168A, E421D, E421Y, E421A, and Δ670-672) were purified sequentially using nickel affinity, Heparin affinity, gel filtration, and anion exchange chromatography. All mutant proteins behaved similarly like the wild-type protein on all chromatographic columns. VP1 proteins were eluted from the Superdex-200 gel filtration column as a single peak at the ~100 kD position, consistent with VP1 being a monomer. The final purified protein was at least 95% pure, as overloaded SDS-PAGE gels did not show any obvious contaminating bands (Pan, Vakharia, and Tao, 2007). The A280/A260 ratio of such samples was always higher than 2, suggesting very little, if any, nucleic acids in the sample. Typically, 2 mg of purified VP1 proteins were obtained from each liter of insect cell culture. As a control, non-tagged IBDV VP1 had also been obtained from insect cells. The purification of the non-tagged proteins differed from that of the fusion proteins in that a (NH4)2SO4 precipitation step was used to replace the Ni-NTA affinity column. The non-tagged protein behaved identically to the N-terminal His-tagged protein in both guanylylation and polymerase assays (data not shown). Therefore, unless otherwise noted, our further self-guanylylation assays were performed using the His-tagged VP1.
Recombinant IBDV VP1 exhibits self-guanylylation activity in our in vitro assays. To remove non-covalently bound nucleotides, all protein samples were boiled for 5 minutes before being subjected to SDS-PAGE. As shown in Figure 1, VP1 became radio-labeled in the presence of [α-32P]rGTP. On the other hand, bovine serum albumin (BSA), which was used as a negative control, did not show any radiation above the background level (data not shown). To address whether RNA template was required for the reaction, VP1 was pre-treated with RNase A to remove any potential contaminating RNA. The result shows that VP1 was able to self-guanylylate equally well with or without RNase A treatment (Fig. 1E, lane 1 vs. 2). This result demonstrates that VP1 guanylylation, or at least a basal level of such an activity, does not require additional viral or host factors. Therefore, like potyvirus and calicivirus VPg uridylylation, but unlike picornavirus uridylylation and duck hepatitis B virus (DHBV) deoxyguanylylation (Anindya, Chittori, and Savithri, 2005; Machin, Martin Alonso, and Parra, 2001; Puustinen and Makinen, 2004; Wang and Hu, 2002), birnavirus VP1 self-guanylylation is independent of RNA templates.
When the self-guanylylation reaction of VP1 was carried out at different temperatures, the highest activity was achieved at 37 °C (Fig. 1A). Therefore, all reactions were routinely performed at 37 °C in this study. To test the time-dependent behavior of VP1 self-guanylylation, purified VP1 was incubated with [α-32P]rGTP for up to 24 hours (Fig. 1B). Significant amounts of guanylylated VP1 products were obtained after 2 hours. A prolonged, 24-hour incubation produced only twice as much guanylylated product as that from a 2-hour incubation (Fig. 1B, lane 7 vs. 9). For consistency, all of our guanylylation assays were performed for 2 hours unless otherwise indicated.
The concentration of monovalent salts appeared to have profound effects on VP1 self-guanylylation (Fig. 1C). In order to keep VP1 soluble, 30 mM NaCl was kept in all reaction buffers. When additional NaCl or KCl was included, VP1 guanylylation activity decreased dramatically. For example, with an additional 20mM NaCl, the activity was reduced to less than 50% of that under the standard condition (Fig. 1C, lane 1 vs. 8). The addition of KCl was slightly less disruptive compared to NaCl (Fig. 1C).
As divalent metal ions are essential for the activity of many polymerases, the effects of various metal ions on the VP1 guanylylation activity were examined. In these assays, magnesium was always used as the standard metal ion for comparison purposes. Under our assay conditions, Cu2+ and Cd2+ could not support any detectable guanylylation activity. While Co2+, Ca2+, and Zn2+ were able to support only marginal guanylylation activity (data not shown), Mn2+ dramatically stimulated the guanylylation activity of VP1. To further explore the effects of Mn2+ on VP1 guanylylation, experiments were performed using reaction buffers with various concentrations of Mn2+ or Mg2+ (Fig. 1D). Maximal guanylation activity was obtained at 2 mM Mn2+ and declined when the concentration was above 20 mM (Fig. 1D, lanes 1-8). For Mg2+, the activity also started to decrease with more than 20 mM Mg2+, but little difference was observed for concentrations ranging from 0.01 mM to 10 mM (Fig. 1D, lanes 9-16), presumably due to the increase of the total salt concentration in the buffer. The amount of guanylylated VP1 products in the presence of 2 mM Mn2+ was about 15 times more than that using the same amount of Mg2+. The enhancing effects of Mn2+ on protein priming has also been documented for several other protein-priming viruses (Anindya, Chittori, and Savithri, 2005; Lin, Wan, and Hu, 2008; Machin, Martin Alonso, and Parra, 2001; Paul et al., 2003; Puustinen and Makinen, 2004).
To further characterize the dependence of VP1 self-guanylylation on divalent ions, experiments were performed in 2 mM Mn2+ with and without 4 mM EDTA. In the presence of EDTA, IBDV VP1 could not be labeled with [α-32P]rGTP (Fig. 1E, lane 3 vs.4). It was also noticed that very low concentration of Mg2+ or Mn2+ (0.01 mM) was able to support the VP1 guanylylation activity (Fig.1D, lanes 1 and 9). Interestingly, when neither EDTA nor Mg2+/Mn2+ was included in the buffer, a slight self-guanylylation activity was often observed (data not shown), presumably due to the presence of certain residual divalent metal ions in the reaction solution. The unexpected weak VP1 guanylylation activity due to the trace amount of metal ions possibly explains an earlier report that VP1 guanylylation was Mg2+/Mn2+ independent (Dobos, 1993). Our findings demonstrate that divalent metal ion is an essential cofactor for VP1 self-guanylylation. Indeed, divalent metal ions have been found to be essential for protein-priming in many viruses such as poliovirus (uridylylation), DHBV (deoxyguanylylation), bacteriophage PRD1 (deoxyguanylylation), and adenovirus (deoxycytidylylation) (Caldentey et al., 1992; Lichy, Horwitz, and Hurwitz, 1981; Lin, Wan, and Hu, 2008; Paul et al., 2003).
To determine the substrate specificity of VP1, nucleotidylation assays were performed using different ribonucleotide triphosphates (rNTPs). Our results showed that any of the four rNTPs was able to react with VP1 nucleotidylation, with the preference order of G > U > C > A (Fig. 2A). At all the concentrations we have tested (from 0.004 to 0.4 μM), VP1 was consistently labeled most efficiently with rGTP (Fig. 2A). The same tendency was also observed when the nucleotide concentration was further increased to 1.0 mM (data not shown). Because VP1 was the only viral component included in the in vitro assays and there was no RNA template present, this slight preference for rGTP indicates that the substrate binding site located within VP1 itself possesses some, if not all the motifs for specific recognition of the nucleotide substrate. It is interesting to note that the nucleotidylation of potyvirus VPg also lacks strong substrate specificity (Puustinen and Makinen, 2004).
To further examine the substrate specificity of VP1 nucleotidylation, competition assays were performed by titrating cold rNTPs into [α-32P]rGTP. As indicated in Fig. 2B, the VP1 guanylylation reaction was nearly completely competed by 10× more concentrated cold rGTP, while VP1 could still be radio-labeled even though cold rUTP was 1000× more concentrated than [α-32P]rGTP.
To probe the catalytic mechanism of VP1 self-guanylylation, we have tested the effects of two molecules, phosphonoformic acid (PFA), a pyrophosphate (PPi) analogue, and rifampicin, an RNA polymerase inhibitor that binds tightly in the RNA exit channel of bacterial RNA polymerase (Campbell et al., 2001). PFA completely inhibited VP1 self-guanylylation at >0.6 mM concentration, indicating that PPi is likely a product of the self-guanylylation reaction as we had expected. By contrast, rifampicin had no effect even at a 1mM concentration. A previous report shows that rifampicin also failed to inhibit T7 RNA polymerase activity (Chamberlin and Ring, 1973). Likewise, IBDV VP1, an RNA-dependent RNA polymerases, possesses a different structural fold compared to that of bacterial RNA polymerase and probably does not bind rifampicin.
Using in vitro assays, we have previously shown that IBDV VP1 was able to catalyze RNA synthesis using ssRNA of either IBDV RNA-A or RNA-B segment as a template (Pan, Vakharia, and Tao, 2007). To facilitate the characterization of reaction products, an 80-nt ssRNA, which was derived from the 3′ end of IBDV RNA-A plus strand, was used as a template in our polymerase assays. Three product species were observed on SDS-PAGE at the ~25 kD, ~90 kD, and ~110 kD position, respectively (Fig. 4A, lane 11). The ~90 kD band was located at the same position as radiolabeled band from the VP1 self-guanylylation reaction (Fig.4A, lane 3), and therefore was presumably the self-guanylylation product VP1-G/GG. The ~25 kD band migrated to a position close to the 100 bp DNA band (Fig. 4A, lane 1) or the 32P-labled 80-nt ssRNA band (Fig. 4A, lane 2). Like the ~25 kD band, the ~110 kD band was detected only in the presence of RNA template (Fig.4A, lane 3 vs. 4-11). The intensities of the three bands are closely related to the ratio of [α-32P] rGTP/cold rGTP. When [α-32P] rGTP was kept 0.2μM concentration, increasing concentration of un-labeled rGTP (from 0 to 500 μM) resulted in weaker intensities for all three product bands (Fig. 4A lanes 4-8). Similarly, without adding un-labeled rGTP, increasing concentration of [α-32P] rGTP correlated with higher intensities for the three bands (Fig. 4A, lanes 8-11), indicating that the detected signals are [α-32P] rGTP specific.
To characterize the molecular nature of these products, the reaction samples were subjected to proteinase K and RNase V1 treatment (Fig. 4B). The results showed that the ~90 kD band was susceptible to proteinase K but resistant to RNase V1, consistent with it being the guanylylation product VP1-G/GG. In contrast, the ~25 kD band was easily digested by RNase V1 but resistant to proteinase K, demonstrating that it was a dsRNA. The ~110 kD band is sensitive to both protease K and RNase V1, suggesting that it is a protein-dsRNA conjugate and likely to be the protein-primed RNA synthesis product. Taken together, VP1 exhibits thee different types of activities, namely the self-guanylylation activity, the primer-dependent RNA synthesis activity, and the primer-independent RNA synthesis activities, resulting in the formation of VP1-G/GG (~90 kD), VP1-dsRNA (~110 kD), and the dsRNA (~25 kD) products, respectively.
To further verify that the radioactivity associated with the three products were indeed from rGMP (or α-phosphate) and not from γ-phosphate, VP1 polymerization assay was carried out in the presence of [8-3H]rGTP, which has a tritium atom at the 8-position of the purine ring. Similarly, three major products with similar molecular weights were detected (Fig. 4C, lane 2), indicating that they all contained the 3H-guanine.
Since our previous polymerization assays were performed using the plus-strand RNA template, we then asked either VP1 was able to synthesize RNA using the minus-strand RNA template. An 80 nt RNA fragment derived from 3′ end of minus-stranded RNA-A was used for polymerization assay, and similar band pattern was obtained (Fig. 4C, lane 3).
In the previous experiment, we have showed that VP1 guanylylation was inhibited by high concentrations of monovalent cations (Na+ or K+). We then asked whether the polymerization activity was also sensitive to monovalent cations. Our results show that RNA synthesis was also inhibited in the presence of Na+ or K+ at high concentrations (Fig. 4D). In addition, we have noticed that the ~25 kD band (the product of de novo RNA synthesis activity) is about ~100 times stronger than the ~90 kD and the ~110 kD bands (the products of the protein primer-dependent RNA synthesis activity). Details of these three activities and their dependence on reaction conditions are currently under investigation.
A chasing experiment was also performed to demonstrate that the ~110 kD product is indeed a result of protein priming (Fig. 6C). When radio-labeled, guanylylated VP1 was incubated with cold rNTP and RNA template, we observed that a significant portion (~38%) of the ~90 kD radio-labeled VP1 became shifted to the ~110 kD position on a SDS-PAGE gel (Fig. 6C, lane 2 vs. 4). A complete shift of the ~90 kD band was not observed, possibly due to a strong competition with the primer-independent RNA synthesis activity as described above.
Protein priming in many viruses is carried out by two separate molecules (e.g. VPg and 3DPol in poliovirus). The situation, however, is different in birnaviruses and also DHBV, in which the polymerase and the protein priming activities are provided by the same polypeptide. Despite that the crystal structure of the IBDV VP1 has been solved, the catalytic relationship between the polymerase active site and the guanylylation site of birnavirus VP1 remains unknown. The crystal structure of IBDV VP1 shows that it has a novel polymerase active site topology with only two aspartic acid residues and one asparagines residue at the active site (Garriga et al., 2007; Pan, Vakharia, and Tao, 2007). To determine the roles of these amino acid residues in VP1 self-guanylylation and elongative RNA synthesis, we have replaced the aspartic acid and the asparagine residues of the -401ADN- tripeptide (the so-called motif C) with alanines by site-directed mutagenesis (Fig. 5). The two mutants D402A and N403A folded properly based on their biochemical behaviors on chromatography columns as mentioned above.
When tested by in vitro assays, the D402A mutant completely lost its RNA polymerase activity, as the ~25 kD band and the ~110 kD band, the product of the de novo RNA synthesis (25 kD band) and the primer-dependent RNA synthesis, respectively, were not observed (Fig. 6). Nonetheless, the ~90 kD band was detected, suggesting that D402A might still be capable of self-guanylylation. The polymerase activity of N403A, on the other hand, decreased to ~60% of that of the wild type. This result indicates that D402, but not N403, is essential for the VP1 polymerase activity. Subsequently, D402A and N403A were used for guanylylation assay in vitro. Interestingly, both D402A and N403A remained active in self-guanylylation with a level of activity similar to that of the wild-type VP1 (Fig. 6A, lane 5 vs. 6-7), suggesting that an intact polymerase active site was not required for VP1 self-guanylylation. Unlike the wild type VP1, guanylylated product of the D402A mutant could not be chased into dsRNA (Fig. 6C).
To further confirm the specificity of the detected guanylylation signal, protein priming assays were performed using DIG-11-UTP. DIG-11-UTP has a digoxigenin group covalently attached to the 5-position of the pyrimidine ring. As a UTP analog, DIG-11-UTP was able to react with IBDV VP1 due to the lack of stringent substrate specificity. As expected, DIG-labeled VP1 was observed for D402A as well as the wild-type protein (Fig. 6B). No signal was observed for VP1 priming with UTP (without the DIG label), indicating that the detected signal was DIG specific. Similar results were obtained when the priming assay was performed using [8-3H] rGTP (data not shown).
The crystal structure of IBDV VP1 shows that the second acidic amino acid of the polymerase motif A is a glutamic acid E421 (Fig. 5). In all other RdRps, the equivalent position is invariantly an aspartic acid, which helps to select rNTP over dNTP by interacting with the 2′-OH group of the rNTP substrate. In birnavirus VP1, however, the aspartic acid is replaced by a glutamic acid residue. To further investigate the catalytic activity of VP1, E421 was mutated to generate three mutants E421D, E421Y, and E421A. When the three mutants were used in in vitro assays, E421D and E421A exhibited similar polymerization (Fig. 6A, lanes 9, 10, and 12) and guanylylation activities (Fig. 6A, lanes 13, 14, and 16) as wild-type VP1. E421Y, however, lost its RNA polymerization activity (Fig. 6A, lane 9 vs. 11) but retains its guanylylation activity (Fig. 6A, lane 13 vs. 15), indicating that E421, an essential residue for RNA polymerization, is not required for guanylylation. This result, which is similar to that for the D402A mutant, further supports the notion that VP1 guanylylation is independent of the RNA polymerization site. Since the polymerase active sites of the D402A and E421Y mutants were no longer capable of nucleotidyl transfer, VP1 self-guanylylation must be catalyzed by a separate active site, which may partially overlap with or may be completely different from the polymerase active site. Such a catalytic mechanism would be different from that by DHBV RT, where an intact polymerase active site was essential for the deoxyguanylylation of the terminal protein (Wang and Seeger, 1992; Zoulim and Seeger, 1994).
The guanylylation site of IPNV VP1 has been determined to be S163 using peptide mapping (Xu, Si, and Dobos, 2004). Multiple sequence alignment among the three birnaviruses superimposes S163 of IPNV VP1 to T168 of IBDV VP1 (Figs. (Figs.55 and and6D).6D). However, because phospho-amino acid determination of labeled IBDV VP1 indicated that the guanylylation site was a phospho-serine, and because S166 is the only serine found in this region, S166 was proposed to be the guanylylation site of IBDV VP1 (Xu, Si, and Dobos, 2004). To experimentally confirm the guanylylation site of IBDV VP1, two VP1 mutants (S166A and T168A) were constructed and tested in our guanylylation assays. To our surprise, S166A could still actively self-guanylylate and polymerize nucleotides (Fig. 6A, lanes 4 and 8). Like the S166A mutant, the T168A mutant also retained both self-guanylylation and RNA synthesis activities (data not shown). Our results thus indicate that neither S166 nor T168 is the guanylylation site of IBDV VP1.
Our previous crystal structure of IBDV VP1 shows that the beginning of the C-terminal domain forms a 25-residue plug that inserts deeply into the polymerase active site canyon (Pan, Vakharia, and Tao, 2007). To probe the function of the C-terminal plug in VP1 self-guanylylation, guanylylation assays were performed using Δ670-672, a deletion mutant in which residues -670SEF- located at the tip of the plug near the polymerase active site were removed (Fig. 5). The results showed that the Δ670-672 mutant exhibited activities similar to that of the wild-type VP1 (data not shown), indicating that the C-terminal plug probably does not play an important role in the catalysis of VP1 self-guanylylation.
To further locate the guanylylation site residue in VP1, we performed an affinity pull-down experiment using partially digested VP1 samples. Because VP1 contains an N-terminal 6×His tag, Ni-NTA affinity pull-down enabled us to determine the approximate location of the guanylylation site residue from the N-terminus by identifying the shortest N-terminal fragment of VP1 treated with proteases. Our results showed that Ni-NTA selectively pulled down a ~20 kD fragment that contained the radioactive label [α-32P]rGMP (Supplemental Information, Fig. S1). The N-terminal domain (consisting of residues 1 to 167), along with the N-terminal fusion tag (26 residues in total), has a theoretical molecular weight of 22.2 kD. Therefore, the guanylylation site residue most likely resides in the N-terminal domain of VP1. When the ~20 kD fragment was submitted for HPLC/MS/MS, extensive peptide coverage was found for the polypeptide region from the N-terminus up to residue 175 (data not shown), consistent with our interpretation based on the affinity pull-down results. In the crystal structure of VP1, the N-terminal domain is located at the upper rear end of the molecule when viewed from the dsRNA exit channel (Garriga et al., 2007; Pan, Vakharia, and Tao, 2007) (Fig. 5). This domain is unique to VP1, and is at least 20 Å away from the polymerase active site.
Birnaviruses are the only known dsRNA viruses that use protein primers and serve as excellent models for structural studies of protein-priming mechanism. By using in vitro assays and recombinant proteins, we have systematically characterized the reaction conditions for IBDV VP1 self-guanylylation and RNA synthesis. The molecule nature of the VP1 catalytic products have been determined by enzymatic digestion and several different chemical labeling methods. The catalytic mechanisms of VP1 self-guanylylation and RNA synthesis have also be probed using small molecule inhibitors and site-directed mutagenesis.
Our results indicate that IBDV VP1 is most active at 37 °C in the reaction buffer containing 2 mM Mn2+ and minimal amounts of monovalent ions. The guanylyation activity of IBDV VP1 is significantly stronger in Mn2+ than in Mg2+, as previously reported for several other RdRps (Alaoui-Lsmaili et al., 2000; Arnold, Ghosh, and Cameron, 1999; Gerber, Wimmer, and Paul, 2001; Yang et al., 2003) and also DHBV RT (Lin, Wan, and Hu, 2008). Our results also showed that VP1 guanylylation activity does not require an RNA template, consistent with earlier results using RNase-treated virus samples as well as isolated IPNV VP1 (Dobos, 1993; Xu, Si, and Dobos, 2004). Like birnavirus VP1, potyvirus and calicivirus VPg can also be uridylylated by their cognate viral polymerase in the absence of other viral factors (Anindya, Chittori, and Savithri, 2005; Machin, Martin Alonso, and Parra, 2001; Puustinen and Makinen, 2004). In contrast, protein-priming in picornaviruses and DHBV has been shown to require RNA templates. Stem-loop structures have been predicted in the non-coding regions of the birnaviruse genomes (Boot et al., 1999), but our results showed that these structural elements, or even the entire IBDV genetic RNA sequence, are dispensable for VP1 self-guanylylation in vitro (Pan, Vakharia, and Tao, 2007). The observed differences in template requirements in the initiation of protein-priming in different viruses may reflect the differences in their inherent catalytic mechanisms and/or catalytic active site configurations.
In our in vitro assay, both guanylylation and polymerase activities of IBDV VP1 decreased dramatically when the Na+ or K+ concentration was increased to 100 mM or higher. Since cellular environments often contain ~150mM KCl/NaCl, the question is how IBDV replicates its genome under physiological conditions? Several explanations might exist. First, VP1 is reported to function within an intact viral capsid, in which the chemical environment may be somewhat different from the cytosol. Second, the interaction of VP1 with other viral proteins (e.g. VP2 and VP3) may affect the response of VP1 to monovalent ions. Lastly, it is worthwhile to note that the polymerase activity of the T7 RNA polymerase is also dramatically reduced by 100 mM KCl, although the optimal polymerization activity of E. coli RNA polymerase is found between 100-200mM KCl (Chamberlin and Ring, 1973). Since high concentrations of monovalent ions reduce both guanylylation and polymerase activities of VP1, it is reasonable to speculate that the inhibition occurs at the nucleotide binding step.
IBDV VP1 showed limited specificity for rGTP and was able to prime using other nucleotide trisphospates, albeit to lesser extents. This is different from template-dependent protein-priming viruses, such as poliovirus and DHBV, in which a particular nucleotide is nearly exclusively preferred (Paul et al., 2000). It has also recently been reported that DHBV RT shows less dNTP specificity in the presence of Mn2+ than in the presence of Mg2+, probably due to the stronger, and thus promiscuous chelating activity of Mn2+ (Lin, Wan, and Hu, 2008). More selective substrate specificity, however, was not observed for IBDV VP1 when Mg2+ was used instead of Mn2+ (data not shown). Another possibility is that other viral proteins or host factors may be involved in facilitating rGTP recognition in birnaviruses. Birnavirus VP3 is an abundant internal protein in IBD virion and it has been shown to interact with VP1 both in vivo and in vitro (Garriga et al., 2007; Maraver et al., 2003; Tacken et al., 2002). It has also been reported that the heat shock protein 90 (hsp90) complex is essential for DHBV RT priming (Hu and Seeger, 1996). The viral core protein of rotavirus has also been shown to be important for the viral polymerase activity (Patton et al., 1997). The potential role of VP3 in VP1 guanylylation is currently under investigation.
Both primer-dependent and primer–independent RNA synthesis activities have been observed for IBDV VP1 in vitro, producing VP1-dsRNA and dsRNA molecules, respectively. Is the primer-independent polymerase activity physiologically relevant? It has been shown that both the plus and the minus strand genomic RNAs of birnaviruses contain a covalently linked VP1 molecule (Dobos, 1993). In addition, purified birnavirus samples can support one cycle of plus-strand RNA synthesis in vitro and these products are covalently linked to VP1 (Dobos, 1995). Protein-primed RNA synthesis has also been demonstrated in virus-infected cells (Magyar, Chung, and Dobos, 1998). Although these data demonstrate that protein priming may be the primary or the only mechanism to initiate birnavirus RNA synthesis, we cannot yet rule out the possibility that primer-independent RNA synthesis may occur at certain stages of the virus infection cycle. Meanwhile, regardless of whether primer-independent RNA synthesis is biologically relevant, such activities have been observed under in vitro conditions for several protein priming polymerases including the poliovirus 3Dpol (Richards and Ehrenfeld, 1998). Further experiments are needed to characterize the effects of different solvent conditions and viral factors on primer-dependent and primer-independent activities of IBDV VP1.
In spite of the general expectation that nucleotidylation of protein primers is catalyzed by the polymerase active site, we made the surprising finding that two fatal polymerase mutants D402A and E421Y both retain the self-guanylylation activity. This result has been confirmed by independent guanylylation assays using three differently labeled nucleotide triphosphates, [α-32P]rGTP, [8-3H]rGTP, and DIG-11-UTP. Our protein samples were highly purified, as shown by results from chromatograms and SDS-PAGE, ruling out the possibility that the radioactive band was due to a contaminating host protein. In addition, VP1 self-guanylylation can be inhibited by PFA, a pyrophosphate analogue, suggesting that self-guanylylation involves a pyrophosphate side-product, as is expected for any nucleotidyl transfer reaction. With all of our experimental evidence available to date, we speculate that VP1 may possess another active site specifically for self-guanylylation. Indeed, VP1 guanylylation does not need any RNA template, a distinct difference from the RNA synthesis reaction it catalyzes. Considering that VP1 self-guanylylation requires divalent metal ions as a cofactor, it is possible that the VP1 guanylylation active site also employs acidic residues to anchor divalent metals, which then activate the hydroxyl group from the protein primer for the subsequent nucleophilic attack on the α-phosphate of the incoming GTP, similar to what is seen in the polymerase active site (Joyce and Steitz, 1994).
Our mutagenesis results show that neither S166 nor T168 is the protein priming residue as previously suggested (Xu, Si, and Dobos, 2004). Through protease digestion and affinity pull-down experiments, we have mapped the guanylylation site residue to the VP1 N-terminal domain. According to the VP1 crystal structure, the N-terminal domain, which is unique to VP1, consists of the first 167 residues. This domain is spatially separated from the polymerase active site in its entirety, further supporting our model in which VP1 self-guanylylation is catalyzed by a novel active site. Guanylylation of VP1 by a second VP1 molecule is unlikely because VP1 dimers have not been found, even at high protein concentrations of up to 5 mg/ml in gel filtration chromatography. It is possible that the N-terminal domain undergoes conformational changes after guanylylation, allowing the guanylyl moiety to bind to the polymerase active site of the same VP1 or of another VP1 molecule to prime RNA synthesis.
Among all protein-priming RNA viruses, potyvirus is probably the one that most closely resembles birnavirus in terms of its VPg nucleotidylation behaviors. For instance, the uridylylation of potyvirus VPg is independent of RNA template and exhibits only slight substrate preference for UTP (Anindya, Chittori, and Savithri, 2005; Puustinen and Makinen, 2004). Poliovirus protein-priming, on the other hand, requires specific RNA templates. Despite extensive studies on poliovirus protein-priming, two different catalytic mechanisms for VPg nucleotidylation have been proposed (Ferrer-Orta et al., 2006; Kamtekar et al., 2006; Schein et al., 2006). Based on evidence from mutagenesis and structural modeling, Schein et al. put forward a surface-catalysis model and proposed that the uridylylation reaction in polioviruses may occur on a solvent exposed surface of 3DPol (Schein et al., 2006). It is also possible that poliovirus VPg uridylylation is catalyzed at the polymerase active site. Ferrer-Orta et al. have shown that the uridylated VPg binds to the polymerase active site of 3DPol in a crystal structure (Ferrer-Orta et al., 2006). Non-uridylylated VPg also binds to the polymerase active site, but the uridylylation site residue Tyr3 is located away from the polymerase active site, out of reach for the nucleotide substrate and not suitable for uridylyl transfer (Ferrer-Orta et al., 2006). Thus, in the absence of pre-uridylylation complex structures, how VPg nucleotidylation is catalyzed in different RNA viruses still appears to be an open question. Here, the demonstration that an intact polymerase active site is not required for IBDV VP1 self-guanylylation provides strong evidence that birnaviruses, and maybe some other RNA viruses, might use a novel active site designated for nucleotidylation during protein-priming. We expect that further structural and biochemical characterization of birnavirus VP1 and its functional complexes will provide interesting insights into the catalytic mechanism of protein-primed RNA synthesis.
We thank Vik Vakharia, Tom Guu, and Douglas Mata for helpful discussions and critical reading of the manuscript. This work is supported by the Welch Foundation (C-1565) and the National Institutes of Health (1R01AI067638).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.