Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Virology. Author manuscript; available in PMC 2010 July 20.
Published in final edited form as:
PMCID: PMC2737686

An interdomain RNA binding site on the hepadnaviral polymerase that is essential for reverse transcription


The T3 motif on the duck hepatitis B virus reverse transcriptase (P) is proposed to be a binding site essential for viral replication, but its ligand and roles in DNA synthesis are unknown. Here, we found that T3 is needed for P to bind the viral RNA, the first step in DNA synthesis. A second motif, RT-1, was predicted to assist T3. T3 and RT-1 appear to form a composite RNA binding site because mutating T3 and RT-1 had similar effects on RNA binding, exposure of antibody epitopes on P, and DNA synthesis. The T3 and RT-1 motifs bound RNA non-specifically, yet they were essential for specific interactions between P and the viral RNA. This implies that specificity for the viral RNA is provided by a post-binding step. The T3:RT-1 motifs are conserved with the human hepatitis B virus and may be an attractive target for novel antiviral drug development.

Keywords: Hepatitis B virus, hepadnavirus, reverse transcription, reverse transcriptase, RNA binding


Hepatitis B virus (HBV) is the prototypic member of the family Hepadnaviridae, hepatotropic double stranded (ds) DNA viruses (Seeger, Zoulim et al., 2007). HBV chronically infects over 360 million people and causes more than 600,000 deaths each year world-wide (Shepard, Simard et al., 2006). It is a small DNA virus that replicates by reverse transcription (Summers & Mason, 1982). The virion has a lipid envelope studded with viral glycoproteins that surrounds an icosahedral core particle. Within the core particle are the viral nucleic acids and a virally-encoded reverse transcriptase, called the polymerase (P). Hepadnaviruses have been found in birds and mammals, including ducks, herons, geese, ground squirrels, woodchucks, woolly monkeys and humans (Schoedel, Sprengel et al., 1989;Guo, Mason et al., 2005;Lanford, Chavez et al., 1998). These viruses all have ~3 kb partially dsDNA genomes and replicate primarily in hepatocytes. Duck hepatitis B virus (DHBV) is a common model for HBV.

Reverse transcription occurs in cytoplasmic subviral capsids and is catalyzed by P. Reverse transcription is initiated by binding of P to the viral pregenomic RNA template at a stem loop called ε (Pollack & Ganem, 1994;Hirsch, Lavine et al., 1990;Junker-Niepmann, Bartenschlager et al., 1990;Beck & Nassal, 1997). Binding to ε induces a conformational change in P that is necessary to trigger its enzymatic activity (Tavis & Ganem, 1996;Tavis, Massey et al., 1998). The P:RNA complex is then encapsidated through polymerization of the viral core protein around it to form the capsid. Reverse transcription is templated by a bulge in ε (Wang & Seeger, 1993;Tavis, Perri et al., 1994) and is primed by a tyrosine residue on P itself. The result of this unique protein-priming mechanism is that the minus-strand DNA is covalently attached to P. Binding of P to ε and reverse transcription are dynamic processes involving several conformational changes by P, at least some of which are mediated by essential interactions with host chaperone proteins, including HSP90, HSP70, HSP40, p23 and HOP (Beck & Nassal, 2003;Hu, Toft et al., 2002;Hu & Seeger, 1996;Hu, Toft et al., 1997).

P has 4 domains (Fig. 1A) (Chang, Hirsch et al., 1990;Radziwill, Tucker et al., 1990). The terminal protein domain contains the tyrosine that primes DNA synthesis and covalently links P to the viral DNA (Y96 in DHBV, Y63 in HBV) (Weber, Bronsema et al., 1994;Zoulim & Seeger, 1994;Lanford, Notvall et al., 1997). The spacer domain has no known function other than to link the terminal protein domain to the rest of P. The reverse transcriptase domain contains the DNA polymerase active site for reverse transcription, and the RNAse H domain degrades the RNA template during reverse transcription. The structure of P has not been solved due to an inability to crystallize the protein. The HBV reverse transcriptase and RNAse H domains have been modeled based on structures of the Moloney murine leukemia virus (MMLV) and human immunodeficiency virus (HIV) reverse transcriptases (Das, Xiong et al., 2001;Langley, Walsh et al., 2007;Bartholomeusz, Tehan et al., 2004;Potenza, Salvatore et al., 2007), but there is no model for the terminal protein domain as it has no homology to non-hepadnaviral proteins. Complementation studies with recombinant fragments of HBV P imply that there are multiple contacts between the terminal protein domain and the reverse transcriptase/RNAse H domains (Lanford, Notvall et al., 1997;Lanford, Kim et al., 1999), but the relative arrangement and contact points between the domains are unknown. There are also many contacts between P and its template ε, but the residues involved in these interactions are not known.

Figure 1
Structural organization of P

We previously described a motif in the terminal protein domain of DHBV P that is essential for DNA synthesis (Cao, Badtke et al., 2005;Badtke, Cao et al., 2006). We found that six monoclonal antibodies (mAb) against the terminal protein domain were able to immunoprecipitate enzymatically active P translated in vitro in a partially denaturing buffer (RIPA), but only three were able to immunoprecipitate P in a more physiological buffer (IPP150). Epitope mapping revealed that the epitopes obscured in IPP150 flanked a highly conserved region (aa 176-183), which we named T3. Mutations to T3 exposed the obscured mAb epitopes while simultaneously inhibiting DNA priming by P, and mutant genomes with lesions in T3 of both HBV and DHBV failed to synthesize DNA within cells. Importantly, synthetic peptides containing T3 sequences specifically inhibited priming in a dose-dependent manner. Based on these data, we proposed that T3 is a molecular binding site on P needed for enzymatic activity. This hypothesis was recently extended by Stahl, et al. (Stahl, Beck et al., 2007) who showed that the T3 motif we had identified was exposed by the chaperones and that mutations within T3 could disrupt RNA binding without causing a general disruption of protein folding. These authors proposed that T3 may directly bind RNA following its exposure by the chaperones. We now extend these observations through RNA binding studies, mutagenesis and peptide competition to identify T3 and a novel motif in the reverse transcriptase domain called RT-1 as RNA binding sites essential for reverse transcription.


T3 is needed for P to bind to ε

The T3 motif (aa 176-183, EAGILYKR) is needed for P to prime DNA synthesis (Cao, Badtke et al., 2005), but T3 could contribute to ε binding, enzymatic activation of P, and/or priming itself. To understand the mechanism by which T3 contributes to priming, we first tested the ability of T3 mutants to bind ε. P was translated in vitro, 32P-labeled RNA was added, and the mixture was incubated at 30° for 60 minutes. The complexes were immunoprecipitated using an anti-P polyclonal antibody, and then P and the co-precipitated RNA were resolved by electrophoresis (Fig. 2A). The RNAs employed were ε, ε-dlBulge (a biologically inactive mutant form of ε that binds P poorly) (Pollack & Ganem, 1994) and DRF+ (DHBV nucleotides 2401-2605 as an irrelevant RNA). Following normalization of the data to the amount of P precipitated and the specific activities of the RNAs, P was found to bind ε seven-fold better than ε-dlBulge and DRF+ (Fig. 2B). We also tested several T3 mutants, some of which have been described previously (Cao, Badtke et al., 2005), for their ability to bind ε (Fig. 2A). P(Y170A/L171A) is a mutant with lesions just upstream of T3 that primes DNA synthesis very poorly, and it bound ε ~50% as well as wild-type. P(Y181F), a mutant we had shown previously to be wild-type in its priming ability (Cao, Badtke et al., 2005), bound ε better than wild-type P. However, two mutants with lesions in T3 that were unable to prime DNA synthesis, P(I179D/L180D) and P(K182E/K183E) (Cao, Badtke et al., 2005), were severely reduced in ε binding. Therefore, there was a correlation between the ε binding and priming activities of P derivatives carrying mutations in the T3 domain.

Figure 2
RNA binding by wild-type and T3-mutant P

Binding of P to ε on the pregenomic RNA triggers hepadnaviral encapsidation, and neither P nor the pregenomic RNA enter capsids without formation of this ribonucleoprotein complex (Pollack & Ganem, 1994;Bartenschlager, Junker-Niepmann et al., 1990). Therefore, mutating T3 should block encapsidation of P if T3 is needed for P to bind ε. To test this prediction, overlength wild-type and T3-mutant P(K182E/K183E) DHBV genomic expression vectors were transfected into LMH cells, core particles were isolated, the particles were permeablized by low pH treatment (Radziwill, Zentgraf et al., 1988) and the covalently-attached nucleic acids were degraded with micrococcal nuclease. P was then resolved by SDS-PAGE and detected by western blot with mAb 11. Viral core particles expressed from the wild-type genome contained P, but cores expressed from genomes carrying the T3 mutation that ablated RNA binding by P did not (Fig. 2C). Therefore, mutating T3 ablated encapsidation in cells as was predicted by the failure of this mutant enzyme to bind ε in vitro.

We previously demonstrated that a synthetic peptide containing the T3 motif inhibits priming by P (Cao, Badtke et al., 2005), and we interpreted these data to indicate that the T3 peptide inhibits P by competing for the ligand that binds to the T3 motif. However, T3 could have directly inhibited the DNA polymerase active site of P. To exclude this possibility, we performed a priming assay with in vitro translated P in which T3 or an irrelevant peptide was added either before or after ε. The T3 peptide specifically inhibited P in a dose-dependent manner when added before ε, as we had previously observed (Cao, Badtke et al., 2005), but it inhibited P very poorly when added after ε (Fig. 2D bottom panel). Because the T3 peptide was present at equal concentrations in both cases yet it inhibited the enzyme only when added prior to ε, we conclude that the T3 peptide did not directly target the DNA polymerase active site.

Together, the correlation between the ε binding and priming activities for the T3 mutants in vitro, the failure of P(K182E/K183E) to encapsidated P in cells, and ability of the T3 peptide to inhibit priming without directly inhibiting the DNA polymerase active site indicate that the T3 motif functions to promote binding of P to ε.

T3 does not interact with cellular chaperones

We have hypothesized that T3 is a molecular binding site on P (Cao, Badtke et al., 2005), but binding at T3 could be intramolecular (with another portion of P) or intermolecular (with a different molecule). The most likely candidates for an intermolecular ligand would be the ε-containing pregenomic RNA or the cellular chaperones that bind to P.

To evaluate the possibility that the chaperones may bind to T3, we employed miniRT2 (Fig. 1B), an active truncation of P that can prime DNA synthesis in the absence of the chaperones (Wang, Qian et al., 2003). If the T3 peptide inhibits priming by competing for binding of a chaperone to the T3 motif on P, then it would not inhibit miniRT2 in the absence of the chaperones. Therefore, we purified miniRT2 under denaturing conditions to remove the bacterial chaperones that co-purify with miniRT2 under native conditions (Wang, Qian et al., 2003) and refolded the protein by gradually removing the denaturant. Incubation of miniRT2 with the T3 peptide inhibited DNA priming, but three irrelevant peptides had no effect (Fig. 3 and data not shown). Therefore, an interaction between T3 and the molecular chaperones cannot be needed for DNA priming by miniRT2. This conclusion was confirmed by co-immunoprecipitation studies employing full-length P translated in vitro. Both wild-type P and P(I179D/L180D) (Fig. 2B), which carries mutations in T3 that ablate RNA binding, could be co-immunoprecipitated by antibodies against p23, a member of the HSP90 chaperone complex (data not shown). Together, these data indicate that the ligand for T3 must be another region of P itself and/or ε, as these were the only macromolecules present in the miniRT2 priming reaction.

Figure 3
The T3 peptide can inhibit priming by miniRT2 in the absence of chaperones

Prediction of a putative partner for T3

We first evaluated the possibility that T3 may interact with another region of P by examining the sequence of P for motifs that could be potential ligands for T3. The spacer and RNAse H domains were excluded because these regions are not present in miniRT2. Motifs in the terminal protein domain that we had previously screened for their ability to inhibit priming as synthetic peptides (unpublished data) were considered unlikely to be the ligand for T3. The universally-conserved motifs in the reverse transcriptase domain (Poch, Sauvaget et al., 1989;Delarue, Poch et al., 1990;Xiong & Eickbush, 1990) were also considered to be unlikely to be a ligand for T3 because their molecular functions are well characterized. Finally, we hypothesized that the ligand would be conserved among the avian and mammalian hepadnaviruses because T3 is highly conserved (Fig. 4A). The region that fit these four criteria best was aa 385-415 near the N-terminus of the reverse transcriptase domain (Fig. 4B), which we named RT-1.

Figure 4
T3 and RT-1 are conserved among the hepadnaviruses

If RT-1 binds to another region of P it would have to be on the surface of the reverse transcriptase domain. While no crystal structures exist for P, a reliable model has been developed for the reverse transcriptase domain of HBV P (Das, Xiong et al., 2001). Most of the HBV RT-1 sequences were on the surface of the reverse transcriptase domain model at one end of the DNA binding cleft (Fig. 5). This would allow these sequences to interact with the terminal protein, which contains the Y96 residue that must enter the active site during protein priming. Alternatively, this motif would be equally well-positioned to interact with the viral nucleic acids during reverse transcription.

Figure 5
RT-1 is predicted to be on the surface of the reverse transcriptase domain

Mutations within RT-1 simultaneously inhibit priming, expose an epitope near T3 and reduce binding to ε

Mutating T3 inhibits priming, exposes buried mAb epitopes near T3 in the terminal protein domain and inhibits binding to ε (Cao, Badtke et al., 2005). If RT-1 functions in concert with T3, either directly through intramolecular binding or indirectly through binding of both motifs to ε, then the effects of mutating RT-1 should be similar to the effects of mutating T3. To test this prediction, we created ten sets of mutations within RT-1. The majority of these substitutions were non-conservative to maximize the possibility of disrupting the potential T3:RT-1 interface that would exist if these motifs bound to each other. P derivatives with mutations to either T3 or RT-1 were translated in vitro and their ability to prime DNA synthesis, to expose the occluded mAb 6 epitope and to bind ε was tested (Fig. 6). As we had previously observed, all mutations to the T3 motif except Y181F had the predicted pattern of activities. All ten mutants with lesions in RT-1 were less active than wild-type P in the priming assay and seven retained ≤ 5% activity, indicating that RT-1 is important for DNA priming. Exposure of the mAb 6 epitope was increased in eight of the 10 mutants, whereas seven mutations to P outside of T3 or RT-1 had no effect on exposure of the occluded mAb epitopes (P2A, K153E, P195V, D513H/D514A, T668V/T670V, H693Y, and D715V; data not shown). Finally, seven of the mutants bound ε less efficiently than wild-type P. Overall, seven of the ten RT-1 mutants had the expected pattern of reduced priming, increased epitope exposure and reduced ε binding (indicted by dots in Fig. 6), although the magnitudes of effects were usually smaller than when T3 was mutated.

Figure 6
Effects of mutations to T3 and RT-1 on RNA binding, priming, and exposure of the mAb6 epitope

Six of the seven RT-1 mutants with the predicted pattern of phenotypic effects had lesions near the N- or C-terminal ends of RT-1 (amino acids 383-392 and 404-429), and only one was near the center of RT-1 (N399A/E402A), whereas the three mutants that did not have the predicted pattern had substitutions near the middle of RT-1 (residues 394-404). This central region of RT-1 is predicted to be buried within the reverse transcriptase domain (Fig. 5), and partial proteolysis indicates that residues 401 and 402 are not accessible to digestion in DHBV P (Lin, Wan et al., 2008). Therefore, the complex effects of mutating the central domain appear to have been due to disrupting the fold of the reverse transcriptase domain by altering buried residues. Overall, these data are consistent with sequences of both T3 and RT-1 contributing to RNA binding, either through interaction with each other or with RNA. However, these data do not exclude other possible roles for RT-1 in DNA synthesis.

Synthetic peptides containing T3 and RT-1 sequences inhibit priming

A peptide containing T3 sequences specifically inhibits DNA priming in a dose-dependant manner [Fig. 3 and (Cao, Badtke et al., 2005)]. If RT-1 binds to either T3 or ε, peptides containing RT-1 sequences should also specifically block priming. Therefore, we created 3 peptides containing RT-1 sequences (Table 1). RT1A contains the N-terminal half of RT-1 (aa 383-399), RT1B contains the C-terminal half (aa 400-417) and RT1C contains the entire RT-1 sequence (aa 381-416). We also created three larger T3 peptides in an attempt to improve activity of the T3 peptide because the original T3 peptide was active only at high concentrations. Finally, we created T3-Scramble and T3B-Scramble, negative control T3 and T3B peptides in which the residues were scrambled. To determine the effects of these peptides on priming, increasing amounts of the peptides were added to P following termination of in vitro translation with cycloheximide, and the IC50 value for each peptide was calculated. The scrambled T3 peptide was essentially unable to inhibit priming (IC50=975 μM) and the original T3 peptide had a high IC50 (302 μM) (Table 1). The new T3 peptides all had IC50 values under 50 μM, with T3B and T3C having values near 10 μM. The RT1A and RT1B peptides were poor inhibitors of priming, with IC50 values greater than 500 μM. However, RT1C was an effective inhibitor, with an IC50 of 46 μM. The RT-1 sequences needed to be present on a single molecule because mixing RT1A and RT1B did not inhibit priming any better than RT1A and RT1B individually (data not shown). Therefore, both T3 and RT-1 peptides efficiently inhibited priming, as predicted.

Table 1
Effects of T3 and RT-1 peptides on priming.

T3 and RT-1 peptides bind RNA

The preceding data indicate that T3 and RT-1 function in a similar manner, but they are equally consistent with intramolecular binding between the two motifs producing a conformation in P competent for RNA binding, or with T3 and RT-1 forming a joint RNA binding site. To help resolve these two mechanisms, we asked whether the T3 and RT-1 peptides could bind RNA. The peptides (10 pMol) were bound to a nitrocellulose filter in a slot-blot apparatus and the filter was washed. Radiolabeled ε or its biologically inactive derivative ε-dlBulge were added with or without a 50-fold excess of yeast tRNA, the filter was washed, and retained RNA was detected by autoradiography (Fig. 7A top panel). There was no RNA binding to two irrelevant peptides (UL13 and Pep1), nor to T3-Scramble or T3B-Scramble. However, robust binding to the T3, T3B, T3C, and RT1C peptides was detected. RNA binding was proportional to the amount of RNA loaded onto the filter, and the RT1C peptide bound RNA less well than the T3B peptide (Fig. 7A bottom panel). RNA binding was non-specific because the peptides bound well to both ε and ε-dlBulge and because binding was suppressed by an excess of unlabeled tRNA (Fig. 7B top panel). The T3B and RT1C peptides were also able to bind to an irrelevant RNA derived from the DHBV core gene and to a double-stranded DNA fragment (Fig. 7B bottom panel). Therefore, peptides containing either T3 or RT-1 sequences possess non-specific nucleic acid binding activities. This strongly implies that these motifs bind to RNA in the context of the intact protein.

Figure 7
T3 and RT-1 peptides bind RNA

Priming and RNA binding activities of miniRT2

11We extended these binding studies from peptides to miniRT2 because miniRT2 retains ε-specific priming activity, is able to bind ε without the aid of the cellular chaperones, and can be purified easily from bacteria. Therefore, miniRT2 performs the authentic DNA priming reaction but it is more experimentally tractable than full-length P.

Wild-type miniRT2 and miniRT2 carrying the K182E/R183E mutations in T3 (T3m) or the S410A/S413A mutations in RT-1 (RT1m) that impair RNA binding by the full-length P (Fig. 6) were expressed in E. coli and purified under native conditions by nickel-affinity chromatography. As expected, the miniRT2 proteins co-purified with bacterial chaperones (Fig. 8A). MiniRT2 and equal amounts of the two mutant proteins were incubated with ε and [α32P]dGTP, the reactions were resolved by SDS-PAGE, and priming activity was detected as 32P labeling of miniRT2 due to the covalent linkage of [32P]dGMP to the enzyme (Fig. 8B). The mutations to T3 and RT-1 both ablated priming by miniRT2, as they did in the context of full-length P (Fig. 6).

Figure 8
Priming and RNA binding activities of miniRT2 and miniRT2 derivatives with lesions in T3 and RT-1

We next assessed RNA binding by miniRT2 and its T3m and RT1m mutants employing the filter-binding assay we used previously for the peptides. Wild-type miniRT2 bound both ε and its biologically inactive derivative ε-dlBulge, and this binding was efficiently competed by a 50-fold excess of unlabeled yeast tRNA (Fig. 8C). Therefore, miniRT2 bound RNA non-specifically under these conditions just as the T3 and RT-1 peptides did, although it can also bind ε specifically when conditions favoring specific binding are employed (Hu, Flores et al., 2004). MiniRT2-T3m failed to bind RNA under identical conditions. This was expected because these mutations ablate RNA binding by full-length P. However, the RT1m mutations increased RNA binding by about 2-fold compared to the same amount of wild-type miniRT2. This result was unexpected because the RT1m mutations reduce RNA binding by full-length P to ~5% of wild-type levels. RNA binding by miniRT2 and its T3m and RT1m derivatives was not distorted by the solid-phase assay we employed because identical results were obtained using GST-tagged miniRT2 and its T3 and RT1 derivatives in a GST pull-down RNA co-precipitation assay (data not shown). Therefore, the increased RNA binding by the RT1m mutant has three implications. First, S410 and S413 are not essential for RNA binding. Second, although the mutations to RT-1 did not have the predicted effect, they did modulate RNA binding, and this strengthens our hypothesis that RT-1 is involved in RNA binding by P. Finally, priming activity can be dissociated from ε binding by mutating an RNA binding site on miniRT2.


We previously identified the T3 motif in the terminal protein domain of DHBV P as being essential for DNA synthesis in vitro and in cells, and we demonstrated that ablating the HBV T3 motif eliminates DNA synthesis in cells (Cao, Badtke et al., 2005). We proposed that the T3 motif was a binding site on P and that “proper occupancy” of T3 was essential for DNA synthesis because the effects of mutating T3 could be mimicked by detergent treatment and competition with a soluble peptide. These results were extended by Stahl, et al. who showed that T3 is transiently exposed on the surface of P by cellular chaperones (Stahl, Beck et al., 2007) and that mutations within T3 that ablated RNA binding did not cause large changes in the structure of P. They concluded that T3 most likely directly bound ε, and they presented a model in which cellular chaperones shuttle P between open and closed conformations, with the open conformation exposing T3 so it could bind RNA. Here, we further extend these studies with three novel observations. First, we describe RT-1 as a region of the reverse transcriptase domain that is functionally linked with T3 and that together with T3 promotes binding to ε. Second, we show that T3 and RT-1 peptides directly bind RNA. Finally, we present data implying that the initial RNA binding by P is non-specific.

These data lead us to hypothesize that the T3 and RT-1 motifs form a composite RNA binding site to which ε binds during the first step of reverse transcription. We favor this hypothesis for five reasons. First, T3 and RT-1 peptides directly bind RNA (Fig. 7). Second, mutations to T3 or RT-1 inhibit binding of full-length P to ε (Fig. 6). The participation of T3 and RT-1 in RNA-binding is supported by alanine-scanning mutagenesis across DHBV P (Seeger, Leber et al., 1996); one mutation to T3 and six to RT-1 were analyzed, and the results were fully consistent with our observations. Third, both T3 and RT-1 appear to be at least transiently exposed on the surface of P. Stahl et al. have demonstrated that T3 is transiently exposed during RNA binding (Stahl, Beck et al., 2007), Lin et al. have shown that residues flanking T3 are accessible to partial proteolysis (Lin, Wan et al., 2008), and molecular modeling implies that RT-1 is largely solvent-exposed (Das, Xiong et al., 2001) (Fig. 5). Fourth, three of the four sets of mutations we created in T3 and seven of the 10 sets in RT-1 simultaneously reduced RNA binding, increased exposure of the occluded mAb 6 epitope, and inhibited priming (Fig. 6). Finally, mapping 1studies employing truncated derivatives of HBV and DHBV P revealed that sequences from both the terminal protein domain (including T3) and from the reverse transcriptase domain (including RT-1) are needed for P to bind RNA (Wang, Zoulim et al., 1994;Hu & Boyer, 2006;Pollack & Ganem, 1994;Hu & Anselmo, 2000). The simplest interpretation of these data is that the T3 and RT-1 motifs cooperate to form the initial RNA binding site on P.

T3 and RT-1 may have structural roles in addition to their RNA binding activities because mutating T3 and RT-1 exposes antibody epitopes in the terminal protein domain that are normally obscured [(Cao, Badtke et al., 2005) and Fig. 6]. The simplest scenario would be that the RNA blocks access of the antibodies to their epitopes. However, the epitopes on in vitro translated P remain obscured in the absence of ε (data not shown), so either the non-specific RNAs in the translation mixture compete with the antibodies for binding to P, or T3 and RT-1 help maintain a conformation of P in which the epitopes are obscured. We favor a conformational role for these motifs because the affinity of the non-specific RNA for P is unlikely to be high enough to compete with binding of the monoclonal antibodies. A plausible mechanism would be for T3 and RT-1 to bind to each other in the closed conformation, but experimental evidence for this possibility is lacking.

Our data and those of Stahl et al. (Stahl, Beck et al., 2007) can be combined into a model for the early stages of reverse transcription (Fig. 9). In this model, newly translated P binds to the cellular chaperones and forms the Closed Complex. This complex contains P in an inactive conformation (Tavis & Ganem, 1996;Tavis, Massey et al., 1998), where T3 (Stahl, Beck et al., 2007) and perhaps RT-1 are occluded. Cellular chaperones convert the closed complex into the Open Complex, where T3 (and presumably RT-1) is transiently exposed (Stahl, Beck et al., 2007). If the open complex encounters a non-ε RNA, it forms a Non-productive Complex in which the RNA binds weakly to T3 and RT-1. Dissociation of the RNA causes P to revert to the open complex, which in turn could revert to the closed complex, where the chaperones may restart the cycle. Reversibility of these steps is supported by the shuttling of P between the closed and open complexes (Stahl, Beck et al., 2007), and because the vast excess of non-ε RNAs over ε within cells would effectively prevent P from finding ε if these reactions were not reversible.

Figure 9
Model for the contribution of T3 and RT-1 to the initial events of reverse transcription

If the open complex binds to ε, it would create the Productive Complex, where T3 and RT-1 bind ε. The productive complex may be a transient intermediate in which ε contacts P in much the same manner as other RNAs, or ε could be held in contact with P through novel protein-RNA contacts. The existence of the productive complex as a discrete state is supported by ε derivatives with mutations in the apical loop which bind to P but cannot support priming (Pollack & Ganem, 1994;Tavis & Ganem, 1996;Hu & Boyer, 2006). Following formation of the productive complex, specific interactions between ε and P would cause an induced-fit conformational change in P detectable as partial resistance of P to proteolysis (Tavis & Ganem, 1996;Tavis, Massey et al., 1998). This creates the Priming Complex, the form of the enzyme which can prime DNA synthesis. The P:ε interactions needed for conversion to the priming complex are poorly characterized, but as discussed above, they probably include specific interactions with the apical loop of ε. Alternatively, a cellular protein has been hypothesized to bind to the apical loop of ε (Pollack & Ganem, 1994;Hu & Boyer, 2006), and in this case contacts between the hypothetical protein and P would promote progression to the priming complex.

We propose that the transition from the productive to priming complexes is not readily reversible and that this irreversibility is the primary source of the specificity of P for ε. Conflicting data have been reported on this issue. Wang et al. demonstrated that P can prime DNA synthesis if ε is restored following its removal from in vitro translated P with RNAse A (Wang, Zoulim et al., 1994), but they could not demonstrate that the molecules that primed DNA synthesis had bound ε in the first phase of the experiment. We used partial proteolysis to show that P cannot re-engage ε and return to the priming complex following removal of ε (Tavis & Ganem, 1996), but priming activity was not directly measured. Regardless of the mechanism, P loses its ability to engage exogenous nucleic acids shortly before or after it primes DNA synthesis, a property that has been dubbed “template commitment” (Radziwill, Zentgraf et al., 1988).

Key features of this model are: 1) RNA binding by P involves direct interactions between the T3 and RT-1 motifs and the RNA; 2) initial RNA binding by P at the T3/RT-1 site is non-specific; and 3) the preference of P for ε over non-ε RNAs (Fig. 2B) is primarily due to induced-fit conformational changes to both P and ε which trap ε in the priming complex. The initial non-specific RNA binding step is the most novel part of this model. Its existence is supported by the ability of the T3 and RT-1 peptides and miniRT2 to bind ε and non-ε RNAs equally well. Non-specific RNA binding by P has not been explicitly reported because prior RNA binding studies by us and others have been done in the presence of excess irrelevant RNAs, and consequently were designed to detect specific binding to ε. However, full-length P clearly possesses both non-specific and ε-specific RNA binding activities because the preference of P for ε over non-ε RNAs in Fig. 2 was only ~7-fold.

The model in Fig. 9 implies that the bulge within ε that templates priming must be passed from the T3/RT-1 site to the DNA polymerase active site after ε has bound to P. This could occur either by translocation of ε into the active site or through unfolding of ε to allow the bulge to reach the active site. The unfolding mechanism is plausible because unfolding of the upper stem of ε upon binding to P has been reported (Beck & Nassal, 1998). A role for RT-1 sequences in the transfer to the active site is supported by the S410A/S413A mutation (“RT1m” in Fig. 8). MiniRT2 containing the S410A/S413A lesion bound RNA better than wild-type miniRT2, yet it was unable to prime DNA synthesis (Fig. 8), genetically resolving these events in the context of miniRT2. However, interpreting the role of S410A/S413A in RNA binding and priming is complicated because this lesion reduced RNA binding by full-length P 20-fold (Fig. 6). This discrepancy may be due to our experimental design: RNA binding by full-length P was measured under conditions favoring specific ε binding, and consequently reflects RNA binding by P in the priming complex (Fig. 9). In contrast, RNA binding by miniRT2 was measured under non-specific conditions that primarily reflect RNA binding by enzyme in the productive complex.

Most therapies for HBV employ one of five nucleoside/nucleotide inhibitors that target the DNA polymerase active site on P. Drug resistant mutations frequently arise during therapy, including some that are cross-resistant between drugs (Bartholomeusz & Locarnini, 2006;Ghany & Liang, 2007). Therefore, there is a great need for drugs with novel targets. The T3:RT-1 interaction with RNA is essential for reverse transcription and can be disrupted with soluble competitors, and hence the T3:RT-1 RNA binding site should be a viable drug target. Porphyrin compounds can inhibit binding between P and ε, and these compounds appear to target sequences in both the terminal protein and reverse transcriptase domains, consistent with possible inhibition of the T3:RT-1 site (Lin & Hu, 2008). The T3:RT-1 interface is particularly attractive because resistance mutations to drugs targeting T3:RT-1 would be very unlikely to be cross-resistant to drugs that target the DNA polymerase active site. Therefore, anti-T3:RT-1 drugs would be excellent complements to the nucleoside/nucleotide inhibitors, permitting true combinational therapy against HBV for the first time.

Materials and methods

Viruses and DNA clones

pT7BDPol contains DHBV strain 3 (Sprengel, Kuhn et al., 1985) nucleotides (nt) 170 to 3021 encoding P within pBluescript (Stratagene); the construct contains a 33-nt insertion at DHBV nt 901 encoding the influenza virus hemagglutinin epitope (Kolodziej & Young, 1991) as well as leader sequences from Brome mosaic virus to promote translation in vitro. Mutations (Table 1) were inserted into pT7BDPol. The plasmid pdε contains DHBV nt 2526 to 2845 encoding ε within pBluescript. pdε-dlBulge is pdε with a deletion of nt 2571-2576. pDCore contains DHBV nt 2649-414 in the vector pRSETc (Invitrogen). pDRF-BS contains DHBV nt 2401-2605 in pBluescript.


Amino acid sequences were aligned using Clustal W. Accession numbers for the sequences employed are DQ195079 (Duck hepatitis B virus, DHBV), CAC80820 (Stork hepatitis virus, SHV), AAA45738 (Heron hepatitis virus, HHV), AAA45748 (Ross' goose hepatitis virus, RGHV), AAA46767 (Woodchuck hepatitis virus, WHV), P03161 (Ground squirrel hepatitis virus, GSHV), AAC16908 (Woolly monkey hepatitis virus, WMHV) and AM282986 (Hepatitis B virus, HBV). The coordinates for the model of the HBV reverse transcriptase domain (Das, Xiong et al., 2001) were displayed using Pymol (DeLano Scientific).

In vitro transcription and translation

mRNAs for DHBV P and Core were transcribed with T7 RNA polymerase from pT7BDPol and pDCore respectively. ε and ε–dlBulge RNAs were transcribed with T3 RNA polymerase from pdε and pdε-dlBulge. All RNAs were transcribed using Megascript kits (Ambion) according to the manufacturer's instructions. In some cases 32P-labeled RNAs were transcribed by including 25 μCi of [α32P]UTP (3000 Ci/mmole, GE Healthcare) in the reaction. 35S-labeled DHBV P was translated in vitro employing rabbit reticulocyte lysates (Promega) in 10 or 20 μl total volume containing [35S]methionine (1,000 Ci/mmol; GE Healthcare) at 30°C for 1.5 h according to the manufacturer's instructions.

MiniRT2 purification and refolding

MiniRT2 containing a hexa-histidine tag on the N-terminus was expressed and purified by nickel-affinity chromatography under both denaturing and native conditions. Purification under denaturing conditions was as previously described (Wang, Qian et al., 2003), with minor changes. Briefly, the bacterial pellet was suspended in 6 M guanidine-HCl along with 20 mM imidazole and 0.1% NP40. After cell lysis the protein was bound to Ni-NTA agarose beads (Qiagen) and washed with 8 M urea. While still attached to the beads, the protein was refolded by adding refolding buffer (Wang, Qian et al., 2003) along with decreasing amounts of urea. Elution buffer [50 mM hepes (pH 8.0, 350 mM imidazole, 300 mM NaCl, 0.1% NP40 and 27.5% glycerol] was used to release the protein from the beads, followed by dialyzing overnight in 50 mM hepes (pH 8), 0.1 M NaCl, 20% glycerol, 0.2% NP40, 50 mM dithiothreitol and protease inhibitor cocktail (Sigma). Native purification of miniRT2 and its T3m and RT1m derivatives employed conditions established for hexa-histidine tagged Hepatitis C virus NS5B RNA polymerase (Ferrari, Wright-Minogue et al., 1999).

Priming assay

35S-labeled P was translated in vitro and an aliquot was removed to measure translation efficiency. 10 μCi [α32P]dGTP (3000 Ci/mmole, GE Healthcare), 0.25 μg ε and MgCl2 to 4 mM were added and the mixture was incubated at 30° for 30 minutes. The samples were then resolved by SDS-PAGE and the translation and priming signals were quantified using a phosphorimager. The priming signal was normalized to translation efficiency. DNA priming by miniRT2 was assessed using 200 ng purified miniRT2 in TMNK buffer (20 mM Tris pH 7.5, 2 mM MgCl2, 15 mM NaCl, 20 mM KCl, and 4 mM dithiothreitol) plus 0.5% NP40 as described previously (Wang, Qian et al., 2003). Synthetic peptides (Genscript) were included in some experiments. The sequence of the MBP peptide in Fig. 2 is APRTPGGRR.

RNA binding assays

The P:RNA binding assay described previously (Beck & Nassal, 1997) was used with minor modifications. Wild type or mutant P was translated in vitro in the presence of 250 ng of 32P-labeled RNA. Translation was stopped by addition of cycloheximide (80 μM), and then the P:RNA complex was immunoprecipitated using the anti-DHBV P polyclonal antibodies R2B2 or R2B3 in the presence of IPP150+ (10 mM Tris pH 7.5, 150 mM NaCl, 0.1% NP40 plus 100 μg/mL yeast tRNA) at 4° for two hours. Following binding, the complex was washed with IPP150+ four times. Radiolabeled P and RNA were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and detected by phosphorimager analysis. Where necessary, the 35S signal of P was blocked with a sheet of exposed autoradiography film.

RNA binding by synthetic peptides was measured using a filter-binding assay. Peptides (Genscript) were dissolved in TMNK plus 0.5% NP40 and then 10 pMol (unless otherwise indicated) was applied to a nitrocellulose filter in a slot blot apparatus, and the filter was washed with TMNK plus 0.5% NP40. Radiolabeled RNA or DNA dissolved in TMNK plus 0.5% NP40 was added, the filters were washed with TMNK plus 0.5% NP40, and the retained RNA was detected by phosphorimage analysis. In some experiments a 50-fold excess of non-radioactive yeast tRNA (Sigma) was added as an irrelevant competitor. The sequence of the Pep1 peptide is KETWWETWWTEWSQPKKKRKV and the sequence of the UL13 peptide is APPSPPSHGGRRR. RNA binding by miniRT2 and its derivatives was measured employing the filter-binding assay employing 0.2 ug of protein per reaction.


We thank Edward Arnold and Kalyan Das for the HBV P model coordinates, Jürgen Beck for mapping the epitopes for mAb 6 and 11, Xiaohong Cheng for technical assistance, Duane Grandgenett for helpful comments on the manuscript, and Maureen Donlin for molecular modeling of RT-1 and assistance with the figures.

This work was supported by NIH grants AI38447 to J.E.T. and AI043453 to J.H.


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.


  • Badtke MP, Cao F, Tavis JE. Combining genetic and biochemical approaches to identify functional molecular contact points. Biol Proced Online. 2006;8:77–86. [PMC free article] [PubMed]
  • Bartenschlager R, Junker-Niepmann M, Schaller H. The P Gene Product of Hepatitis B Virus is Required as a Structural Component for Genomic RNA Encapsidation. Journal of Virology. 1990;64:5324–5332. [PMC free article] [PubMed]
  • Bartholomeusz A, Locarnini SA. Antiviral drug resistance: clinical consequences and molecular aspects. Semin Liver Dis. 2006;26:162–170. [PubMed]
  • Bartholomeusz A, Tehan BG, Chalmers DK. Comparisons of the HBV and HIV polymerase, and antiviral resistance mutations. Antivir Ther. 2004;9:149–160. [PubMed]
  • Beck J, Nassal M. Sequence- and Structure-Specific Determinants in the Interaction between the RNA Encapsidation Signal and Reverse Transcriptase of Avian Hepatitis B Viruses. J Virol. 1997;71:4971–4980. [PMC free article] [PubMed]
  • Beck J, Nassal M. Formation of a functional hepatitis B virus replication initiation complex involves a major structural alteration in the RNA template. Mol Cell Biol. 1998;18:6265–6272. [PMC free article] [PubMed]
  • Beck J, Nassal M. Efficient Hsp90-independent in vitro activation by Hsc70 and Hsp40 of duck hepatitis B virus reverse transcriptase, an assumed Hsp90 client protein. J Biol Chem. 2003;278:36128–36138. [PubMed]
  • Cao F, Badtke MP, Metzger LM, Yao E, Adeyemo B, Gong YH, Tavis JE. Identification of an essential molecular contact point on the duck hepatitis B virus reverse transcriptase. J Virol. 2005;79:10164–10170. [PMC free article] [PubMed]
  • Chang LJ, Hirsch RC, Ganem D, Varmus HE. Effects of insertional and point mutations on the functions of the duck hepatitis B virus polymerase. J Virol. 1990;64:5553–5558. [PMC free article] [PubMed]
  • Das K, Xiong X, Yang H, Westland CE, Gibbs CS, Sarafianos SG, Arnold E. Molecular modeling and biochemical characterization reveal the mechanism of hepatitis B virus polymerase resistance to lamivudine (3TC) and emtricitabine (FTC) J Virol. 2001;75:4771–4779. [PMC free article] [PubMed]
  • Delarue M, Poch O, Tordo N, Moras D, Argos P. An attempt to unify the structure of polymerases. Protein Eng. 1990;3:461–467. [PubMed]
  • Ferrari E, Wright-Minogue J, Fang JWS, Baroudy BM, Lau JYN, Hong Z. Characterization of soluble hepatitis C virus RNA dependent RNA polymerase expressed in Escherichia coli. J Virol. 1999;73:1649–1654. [PMC free article] [PubMed]
  • Ghany M, Liang TJ. Drug targets and molecular mechanisms of drug resistance in chronic hepatitis. B Gastroenterology. 2007;132:1574–1585. [PubMed]
  • Guo H, Mason WS, Aldrich CE, Saputelli JR, Miller DS, Jilbert AR, Newbold JE. Identification and characterization of avihepadnaviruses isolated from exotic anseriformes maintained in captivity. J Virol. 2005;79:2729–2742. [PMC free article] [PubMed]
  • Hirsch RC, Lavine JE, Chang LJ, Varmus HE, Ganem D. Polymerase gene products of hepatitis B viruses are required for genomic RNA packaging as well as for reverse transcription. Nature (London) 1990;344:552–555. [PubMed]
  • Hu J, Anselmo D. In vitro reconstitution of a functional duck hepatitis b virus reverse transcriptase: posttranslational activation by HSP90. Journal of Virology. 2000;74:11447–11455. [PMC free article] [PubMed]
  • Hu J, Boyer M. Hepatitis B virus reverse transcriptase and epsilon RNA sequences required for specific interaction in vitro. J Virol. 2006;80:2141–2150. [PMC free article] [PubMed]
  • Hu J, Flores D, Toft D, Wang X, Nguyen D. Requirement of heat shock protein 90 for human hepatitis B virus reverse transcriptase function. J Virol. 2004;78:13122–13131. [PMC free article] [PubMed]
  • Hu J, Seeger C. Hsp90 is required for the activity of a hepatitis b virus reverse transcriptase. Proc Natl Acad Sci USA. 1996;93:1060–1064. [PubMed]
  • Hu J, Toft D, Anselmo D, Wang X. In vitro reconstitution of functional hepadnavirus reverse transcriptase with cellular chaperone proteins. J Virol. 2002;76:269–279. [PMC free article] [PubMed]
  • Hu J, Toft DO, Seeger C. Hepadnavirus assembly and reverse transcription require a multi-component chaperone complex which is incorporated into nucleocapsids. EMBO J. 1997;16:59–68. [PubMed]
  • Junker-Niepmann M, Bartenschlager R, Schaller H. A short cis-acting sequence is required for hepatitis B virus pregenome encapsidation and sufficient for packaging of foreign RNA. EMBO J. 1990;9:3389–3396. [PubMed]
  • Kolodziej PA, Young RA. Epitope Tagging and Protein Surveillance. Methods in Enzymology. 1991;194:508–519. [PubMed]
  • Lanford RE, Chavez D, Brasky KM, Burns RB, III, Rico-Hesse R. Isolation of a hepadnavirus from the woolly monkey, a New World primate. PNAS. 1998;95:5757–5761. [PubMed]
  • Lanford RE, Kim YH, Lee H, Notvall L, Beames B. Mapping of the hepatitis B virus reverse transciptase TP and RT domains by transcomplementation for nucleotide priming and by protein-protein interaction. J Virol. 1999;73:1885–1893. [PMC free article] [PubMed]
  • Lanford RE, Notvall L, Lee H, Beams B. Transcomplementation of Nucleotide Priming and Reverse Transcription between Independently Expressed TP and RT Domains of the Hepatitis B Virus Reverse Transcriptase. J Virol. 1997;71:2996–3004. [PMC free article] [PubMed]
  • Langley DR, Walsh AW, Baldick CJ, Eggers BJ, Rose RE, Levine SM, Kapur AJ, Colonno RJ, Tenney DJ. Inhibition of hepatitis B virus polymerase by entecavir. J Virol. 2007;81:3992–4001. [PMC free article] [PubMed]
  • Lin L, Hu J. Inhibition of hepadnavirus reverse transcriptase-epsilon RNA interaction by porphyrin compounds. J Virol. 2008;82:2305–2312. [PMC free article] [PubMed]
  • Lin L, Wan F, Hu J. Functional and structural dynamics of hepadnavirus reverse transcriptase during protein-primed initiation of reverse transcription: effects of metal ions. J Virol. 2008;82:5703–5714. [PMC free article] [PubMed]
  • Poch O, Sauvaget I, Delarue M, Tordo N. Identification of four conserved motifs among the RNA-dependent polymerase encoding elements. EMBO J. 1989;8:3867–3874. [PubMed]
  • Pollack JR, Ganem D. Site-specific RNA binding by a hepatitis B virus reverse transcriptase initiates two distinct reactions: RNA packaging and DNA synthesis. J Virol. 1994;68:5579–5587. [PMC free article] [PubMed]
  • Potenza N, Salvatore V, Raimondo D, Falanga D, Nobile V, Peterson DL, Russo A. Optimized expression from a synthetic gene of an untagged RNase H domain of human hepatitis B virus polymerase which is enzymatically active. Protein Expr Purif. 2007;55:93–99. [PubMed]
  • Radziwill G, Tucker W, Schaller H. Mutational analysis of the hepatitis B virus P gene product: domain structure and RNase H activity. J Virol. 1990;64:613–620. [PMC free article] [PubMed]
  • Radziwill G, Zentgraf H, Schaller H, Bosch V. The duck hepatitis B virus DNA polymerase is tightly associated with the viral core structure and unable to switch to an exogenous template. Virology. 1988;163:123–132. [PubMed]
  • Schoedel F, Sprengel R, Weimer T, Fernholz D, Schneider R, Will H. Animal hepatitis B viruses. Advances in Viral Oncology. 1989;8:73–102.
  • Seeger C, Leber EH, Wiens LK, Hu J. Mutagenesis of a Hepatitis B Virus Reverse Transcriptase Yields Temperature-Sensitive Virus. Virology. 1996;222:430–439. [PubMed]
  • Seeger C, Zoulim F, Mason WS. Hepadnaviruses. In: Knipe DM, Howley P, Griffin DE, Lamb RA, Martin A, Roizman B, Straus SE, editors. Fields Virology. Lippincott Williams & Wilkins; Philadelphia: 2007. pp. 2977–3029.
  • Shepard CW, Simard EP, Finelli L, Fiore AE, Bell BP. Hepatitis B virus infection: epidemiology and vaccination. Epidemiol Rev. 2006;28:112–125. [PubMed]
  • Sprengel R, Kuhn C, Will H, Schaller H. Comparative sequence analysis of duck and human hepatitis B virus genomes. J Med Virol. 1985;15:323–333. [PubMed]
  • Stahl M, Beck J, Nassal M. Chaperones activate hepadnavirus reverse transcriptase by transiently exposing a C-proximal region in the terminal protein domain that contributes to epsilon RNA binding. J Virol. 2007;81:13354–13364. [PMC free article] [PubMed]
  • Summers J, Mason WS. Replication of the Genome of a Hepatitis B-Like Virus by Reverse Transcription of an RNA Intermediate. Cell. 1982;29:403–415. [PubMed]
  • Tavis JE, Ganem D. Evidence for the Activation of the Hepatitis B Virus Polymerase by Binding of Its RNA Template. J Virol. 1996;70:5741–5750. [PMC free article] [PubMed]
  • Tavis JE, Massey B, Gong Y. The Duck Hepatitis B Virus Polymerase Is Activated by Its RNA Packaging Signal, Epsilon. J Virol. 1998;72:5789–5796. [PMC free article] [PubMed]
  • Tavis JE, Perri S, Ganem D. Hepadnavirus Reverse Transcription Initiates within the Stem-Loop of the RNA Packaging Signal and Employs a Novel Strand Transfer. J Virol. 1994;68:3536–3543. [PMC free article] [PubMed]
  • Wang GH, Seeger C. Novel Mechanism for Reverse Transcription in Hepatitis B Viruses. J Virol. 1993;67:6507–6512. [PMC free article] [PubMed]
  • Wang GH, Zoulim F, Leber EH, Kitson J, Seeger C. Role of RNA in Enzymatic Activity of the Reverse Transcriptase of Hepatitis B Viruses. J Virol. 1994;68:8437–8442. [PMC free article] [PubMed]
  • Wang X, Qian X, Guo HC, Hu J. Heat shock protein 90-independent activation of truncated hepadnavirus reverse transcriptase. J Virol. 2003;77:4471–4480. [PMC free article] [PubMed]
  • Weber M, Bronsema V, Bartos H, Bosserhoff A, Bartenschlager R, Schaller H. Hepadnavirus P Protein Utilizes a Tyrosine Residue in the TP Domain to Prime Reverse Transcription. J Virol. 1994;68:2994–2999. [PMC free article] [PubMed]
  • Xiong Y, Eickbush TH. Origin and evolution of retroelements based upon their reverse transcriptase sequences. EMBO J. 1990;9:3353–3362. [PubMed]
  • Zoulim F, Seeger C. Reverse Transcription in Hepatitis B Viruses is Primed by a Tyrosine Residue of the Polymerase. J Virol. 1994;68:6–13. [PMC free article] [PubMed]