INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Duck hepatitis B virus (DHBV) belongs to the family Hepadnaviridae, a small group of enveloped viruses which cause persistent liver infections and replicate their DNA genomes through reverse transcription of an RNA intermediate (26). Because of the involvement of reverse transcription, the replication of the viral genome is assumed to be error prone and to give rise to a high degree of genetic variation. In humans, HBV can cause acute and chronic liver disease, and genetic variation is believed to be related to the clinical outcome of hepadnavirus infection (for a review, see reference 8). In recent years, it has become more fully appreciated how the presence of a complex mixture of variants in the virus population within a single host, the quasispecies, may enable rapid adaptation to changing environments (5), in particular to selection forces such as antiviral drugs (1), vaccines (4), immunomodulatory substances (20), and other changes in the host's immune response, leading to the emergence of various resistance and escape mutants (25).

Major mechanisms contributing to the complexity of the viral quasispecies are the generation of errors during viral genome replication, environmentally induced mutations, and recombination. Of these, errors in replication are thought to constitute the greatest source for generating variants. The frequency of replication errors in hepadnaviruses is a function of the fidelities of three polymerization reactions which occur during the life cycle in two different compartments. In the nucleus, covalently closed circular DNA (cccDNA)-templated positive-strand RNA synthesis is carried out by cellular RNA polymerase (RNA pregenome synthesis), and in the viral core particles, RNA-templated negative-strand DNA synthesis and DNA-templated positive-strand DNA synthesis are both carried out by reverse transcriptase. The relative contributions of the various enzymes involved in different polymerization steps to the overall mutation rate of the virus are not known.

In spite of the reported widespread occurrence of genetic variation in hepadnavirus infection, little is known about the natural mutation rate of HBV, and most estimates are extrapolations from that of RNA viruses and retroviruses, which have been more intensively studied (6, 9, 19, 23). Experimental approaches used to measure variation in hepadnaviruses have generally focused on estimating the variation that has occurred within infected individuals over a known period of time (2, 3, 7, 10, 22). Two difficulties in interpreting such measurements are the lack of knowledge about selective pressures that may have influenced the frequency of any particular variant in the virus population and the history of the dynamic state of virus replication during the interval considered. Since the frequency of any unselected variant in the population will reflect the number of generations of virus replication that have occurred, and fitness selection depends on the dynamics of the infection (29), it is difficult to relate measurements of mutant frequency to the mutation rate per generation.

In this study, we determined the frequency of mutations at several individual sites in the DHBV genome by assaying the occurrence of spontaneous reversion of a single-base-change cytopathic variant of DHBV (14, 15, 17) in an experimental infection of young ducklings. Because revertants could be coselected in parallel with a known amount of wild-type (WT) virus (16), we were able to measure the frequency of revertants relative to the added WT genomes. We used reasonable assumptions to construct a mathematical model for relating the frequency of revertants to the mutation rate per generation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Plasmids. The DHBV type 16 (DHBV-16) WT genome was a head-to-tail-dimer construct derived from that sequenced by Mandart et al. (18) and cloned into pSP65. The G133E mutant viral genome carried a single-amino-acid codon change from glycine to glutamic acid at residue 133 (G133E) in the large envelope protein (14) produced by site-directed mutagenesis of a guanylate residue at nucleotide (nt) 1198 to adenylate (G1198A). In addition, this mutant genome carried a silent nucleotide substitution of cytidylate at nt position 1178 to adenylate (C1178A), resulting in the destruction of a SmaI site present in the WT genome. The presence or absence of the SmaI site was used to determine the virus variant in the duckling serum. Neither nucleotide substitution altered the DHBV polymerase open reading frame, which overlaps the pre-S region.

Cell culture and transfection. WT and mutant viruses were produced by transfection of the chicken hepatoma cell line LMH (13) with respective DHBV DNA as previously described (27). Supernatants from the transfected LMH cells were collected daily from day 3 to day 10 posttransfection, clarified by low-speed centrifugation, and stored at 4°C before use. The titer of virus of the supernatants was determined by Southern blot hybridization with a ³²P-labeled riboprobe following selective DNA extraction from enveloped virus particles (14). Hybridization was quantitated by phosphorimaging (Molecular Dynamics, Sunnyvale, Calif.). Virus used for injection was concentrated by precipitation with 10% (wt/vol) polyethylene glycol 8000 (Sigma, St. Louis, Mo.) (15).

Animals and infections. One-day-old White Pekin ducklings obtained from Metzer Farms (Redland, Calif.) were used in experiment 1, and ducklings from Privett Hatchery (Portales, N. Mex.) were used in experiment 2. Ducklings testing negative for DHBV DNA by dot hybridization were infected by intravenous injection of a volume of 300 µl of a suspension of DHBV on day 4 posthatching. All birds were fed an unrestricted diet and received humane care throughout the study in accordance with guidelines issued by the National Institutes of Health.

2

6

PCR amplification, determination of virus variants, and cloning of DHBV DNA from PCR products. Amplification was performed on DNA derived from the equivalent of 2.5 µl of serum by mixing serum samples with an equal amount of 10 mM Tris-HCl (pH 7.4)-1 mM EDTA and incubating the mixture at 96°C for 10 min (21). The entire pre-S region was amplified by using sense primer P1 and reverse primer P2. A restriction site for ApaI (New England BioLabs, Beverly, Mass.) was introduced into the 5' end of P1 (underlined below). In addition, this primer was biotinylated. Sequences of the primers were as indicated: P1 5'-CGCGGGGCCCAAGAGCATTTCCTA, nt 584 to 603, sense; P2, 5'-CCGATTAGGCCAGCTAGTAT, nt 1324 to 1305, reverse. A 741-bp fragment was produced from the viral DNA template by using 2.5 U of Taq DNA polymerase (Promega, Madison, Wis.) for 35 cycles of 15 s at 95°C, 15 s at 55°C, and 30 s at 72°C in a GeneAmp PCR System 9700 (Perkin-Elmer Cetus, Norwalk, Conn.).

DNA sequencing of PCR-amplified products and cloned DHBV DNA. For direct sequencing, the biotinylated PCR products were adsorbed to streptavidin-coated beads (Dynabeads M-280; Dynal A.S., Oslo, Norway) and washed according to the manufacturer's instructions with the help of a magnetic particle concentrator (Dynal catalogue no. 120.04; Dynal A.S.). The nonbiotinylated minus-strand products were released from the beads by denaturation in 0.1 N NaOH, and the biotinylated plus-strand products were sequenced on the beads with 5 pmol of P2 and ³⁵S-dATP using the T7 Sequenase version 2.0 dGTP reagent kit (Amersham Life Science, Cleveland, Ohio).

Cloning of PCR-derived DNA into replication-competent genomes. Six candidate revertant mutations identified by sequencing of PCR amplification products of the pre-S region were cloned into the DHBV-16 WT genome. PCR products amplified with P1 and P2 (see above) were purified by phenol extraction, precipitated with ethanol, and digested with the restriction enzymes PflF1 and XhoI or KpnI (all from New England BioLabs) respectively. Subsequently, they were subjected to low-melting-point agarose electrophoresis (GIBCO/BRL) and visualized by ethidium bromide staining. The bands of interest were cut out under low-wave UV illumination and, after heating to 65°C, directly used for ligation into the DHBV-16 WT genome construct which had been digested with the appropriate restriction enzymes to remove the corresponding WT pre-S sequences. Ligation was performed with T4 DNA ligase (GIBCO/BRL) overnight at 14°C. DH5 bacterial cells (GIBCO/BRL) were used for cloning.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

The purpose of this study was to determine the spontaneous mutation rate during an experimental infection of young ducklings with the avian hepadnavirus DHBV. We used the reversion rate of a cytopathic variant of DHBV as a measure of mutation rate. Noncytopathic revertants are easily detected, since we previously showed that they are rapidly selected in vivo (16). In addition, we sought to identify the sites of the various mutations in the DHBV genome that cause reversion of the cytopathic phenotype.

Strategy and rationale of the experiment. Previous studies have shown that infection of ducklings with a variant of DHBV of the viral large envelope (pre-S) domain, namely, G133E, was cytopathic for the birds, leading to severe acute but transient noninflammatory liver injury (17). Recovery from acute liver disease was shown to be accompanied by the appearance of noncytopathic spontaneously arising revertant viruses (16). At the molecular level, hepatocellular injury was preceded by elevated levels of DHBV cccDNA in the hepatocytes, and enhanced cytoplasmic staining for viral capsid proteins was observed within the cells. A single amino acid change of glycine to glutamic acid at position 133 (G133E) in the pre-S domain of DHBV was directly responsible for these phenomena, as the production of new viral cccDNA in hepatocytes is regulated by the viral pre-S protein (27). It is believed that WT pre-S protein directs capsids containing relaxed circular DNA, the precursor of cccDNA, into a pathway for assembly of virus particles, which are secreted from the infected cells, excluding their utilization for cccDNA synthesis in the nuclei, possibly through direct interaction with the capsid protein of the virus itself (14).

Viremia and weight gain after mixed infection. In experiment 1, a total of 24 ducklings (six groups consisting of four birds each) were infected at 4 days of age with inocula of 10⁸ viral genomes of G133E, either alone (group 1) or mixed with WT DHBV at a frequency of 10² to 10⁶ WT genomes per G133E genome (groups 2 to 6, respectively). One group of three ducklings was infected with 10⁸ genomes of WT virus only (group 7). Three birds that served as uninfected controls received no inoculum (group 8). Ducklings were bled and weighed daily from day 3 to day 14 and at 21 and 25 days postinfection. The presence of viral DNA in the serum was monitored by PCR, and the respective viral DNA titers were determined by dot hybridization and phosphorimaging.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Viremias in eight groups of ducklings. Viremias were determined by dot hybridization and phosphorimaging. The minimum viremia detectable by this method was approximately 2 × 106 genomes per ml. The mean log value for all positive samples within a group for each time point was plotted.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Body weights in eight groups of ducklings. All birds in each group were weighed at the indicated times, and the mean body weight at each time point was plotted.

Outgrowth of WT and putative revertant viruses in mixed infections with G133E and WT virus. To see whether recovery from liver disease reflected by the resumption of normal growth was primarily due to outgrowth of WT virus or to spontaneous revertants in mixed-virus-infected birds (groups 2 to 6), we carried out PCR amplifications of the DHBV region which carries the genetic tag and G133E mutation and assayed for the presence (WT virus population) or absence (G133E and/or spontaneous revertant population) of the SmaI site at position 1178. As seen in Fig. 3, the PCR products obtained from the serum of ducklings from groups 5 and 6 lacked any detectable SmaI cleavage. These virus populations therefore consisted predominantly of spontaneous revertants and G133E virus with no detected WT internal standard. In contrast, in serum samples from birds infected with larger amounts of WT internal standard added to the inoculum (groups 2, 3, and 4), SmaI-digestible viral DNA appeared during the course of the experiment in amounts that were directly related to the presence of WT virus in the inoculum. These results suggested that, by 21 days postinfection, coincident with recovery from acute liver injury, the major virus population found in the serum of ducklings shifted from the G133E genotype to the WT genotype (group 2) mixed with spontaneous revertant genotypes (groups 3 to 6). Accordingly, the frequency of all spontaneous revertants would be approximately equal to the proportion of WT reference virus in the inoculum that produced an equal mixture of WT and revertant virus in the ducklings at 21 days. (After melting and reannealing during the final cycles of PCR, only 25% of double-stranded products would contain SmaI sites in both strands and be digested with SmaI.) In this experiment, the inoculum most closely producing such a mixture was that containing WT virus at a frequency of 10⁴.

View larger version (45K): [in this window] [in a new window]   FIG. 3.   PCR assay of the emergence of WT virus in ducklings with mixed infections. Serum samples from individual infected ducks, represented in separate lanes, were amplified by PCR. The products were each mixed with 600 ng of pGEM-7Zf(+) DNA and digested with SmaI to detect the appearance of WT internal reference genomes. The upper band (C) is the control added to each reaction mixture to monitor for complete digestion. The double bands represent the undigested PCR product of G133E-derived genomes (G) and the digested product of WT internal reference genomes (W).

View larger version (79K): [in this window] [in a new window]   FIG. 4.   Direct sequence of PCR products derived from representative birds in groups 2, 4, and 6. Sequencing reactions were performed on biotinylated PCR products as described in the text. The sequence ladder runs from bottom to top in the plus-strand direction. The positions of the SmaI site present in the WT internal reference and codon 133 are indicated. Arrows mark sites of potential reverting mutations. aa, amino acid.

Nature and location of mutations in the pre-S region of spontaneous revertant viruses. In order to resolve the individual spontaneous revertant sequences in G133E-infected ducklings, we cloned and sequenced the C-terminal region of the pre-S domain obtained from PCR products of four mixed-virus-infected birds from group 4 at various time points postinfection. In one bird (no. 440), pre-S sequences were analyzed at multiple times during infection. In a second bird (no. 438), the only viremic serum sample available was from 5 days postinfection, and all subsequent samples were negative by PCR. A total of six libraries of cloned pre-S sequences were prepared, and up to 31 separate sequences were determined for each library. As listed in Table 1, a spectrum of amino acid changes and corresponding codon changes was observed in the libraries, with the libraries from 3 and 5 days postinfection showing little heterogeneity, while libraries prepared from later time points contained various amounts of the internal reference WT virus and low amounts of the cytopathic virus G133E. The remaining sequences in the libraries from 14 and 21 days postinfection contained the G133E-specific genetic tag accompanied by changes at one residue (86 of 93) or occasionally a second residue (14 of 86). In cases where mutations occurred at two sites, one of these mutations was usually a unique example (one exception, E133K + Q135R) and was considered to be possibly introduced by PCR. This interpretation is consistent with the observed frequency of single mutations in the WT internal reference genomes sequenced (1 of 17), which is not significantly different.

Functional tests of candidate pre-S mutants. To test whether the growth rates of the candidate spontaneous revertants allowed them to expand in parallel with the WT internal reference, we determined their ability to compete with the WT virus in a mixed infection of ducklings. Pre-S regions containing the four observed codon changes resulting from the six mutations observed in more than one library were substituted into a WT genome and transfected into LMH cells. To verify the absence of additional mutations, the whole pre-S region between the sites used for cloning was sequenced in each construct. Virus stocks of revertants were prepared, and ducklings were inoculated together with WT virus in a 1:1 ratio (experiment 2). To estimate the ability of the cytopathic virus to compete with WT, G133E was also included among the combinations of mixed infections. The region of viral DNA containing the genetic tag was amplified from the earliest virus-positive serum sample from each bird, and the PCR products were subjected to SmaI digestion and agarose gel electrophoresis. The data obtained are shown in Fig. 5.

View larger version (57K): [in this window] [in a new window]   FIG. 5.   PCR assay for the genotype of virus in ducks with mixed infections. The first viremic serum sample from each group of ducklings infected with WT plus the indicated candidate revertant was amplified by PCR, and the products were mixed with pGEM DNA and digested with SmaI exactly as described for Fig. 3. The positions of the control (C), the G133E-specific (G), and the WT-specific (W) bands are indicated.

Frequency of revertants. For estimating the frequencies of the various spontaneous revertants, the ratios of seven spontaneous revertants which occurred most often in the viral population to the WT internal standard were calculated by dividing the total number of respective revertant clones observed by the number of WT clones observed in the libraries. The frequency of WT clones in this viral population was defined as 10⁴ per G133E genome, since, in group 4, WT virus had been added to the inoculum at a frequency of 10⁴. As seen in Table 2, the relative mutation frequencies for these specific sites in the DHBV genome varied from 1 × 10⁵ to 6 × 10⁵, with the same-site reversion being present at the highest frequency.

Types of mutations observed. The mutations observed in the seven authentic revertants included three transitions and four transversions. In contrast, the remaining 16 sporadic mutations observed consisted of 12 transitions and 4 transversions. Whether the different ratios of transitions to tranversions in the two groups of mutants are significant and reflect two different mechanisms for generating mutations is not known (e.g., errors in replication versus PCR-generated errors). No deletions were detected, and this is not surprising because of the high degree of selection for function in this system.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

In the previous studies cited, we demonstrated that infection of young ducklings with a cytopathic hepadnavirus, G133E, resulted in acute noninflammatory liver injury, followed within 2 to 3 weeks by recovery and restoration of normal liver histology. Recovery was attributed to the emergence of spontaneous noncytopathic revertants, which replaced the cytopathic virus in the liver. Thus, liver injury likely reflected the process of virus replacement, with G133E-infected cells dying and being replaced by revertant-infected cells. Infection with the cytopathic virus could therefore be considered to be transient, involving a limited amount of virus replication before elimination from the liver.

In this study, we determined how frequently revertant viruses were produced during the transient G133E replication that occurred in this system. The frequency of revertant production may be taken as a crude measure of the rate at which hepadnaviruses generate variants by spontaneous random mutation. To measure the frequency of reversion, we utilized a genetically marked WT virus as an internal reference. Since true revertants, by definition, would be indistinguishable from the WT reference in their growth and subsequent selection in the duckling, the ratio of such revertants to the WT reference could be used to calculate the total number of revertants generated.

For example, in an inoculum containing 10⁸ G133E virus produced from transfected plasmid DNA, and a reversion rate at a single site of 10⁵ per G133E genome synthesized, the corresponding revertant population would be 10³ virus particles. After addition to the inoculum of 10⁴ WT virus as the internal reference and injection into ducklings, each infecting viral genome would be converted to cccDNA and begin to produce progeny cccDNA by intracellular amplification and extracellular spread of infection. The increase in cccDNA derived from the input revertant genomes would exactly parallel the increase in cccDNA from the WT reference virus; however, revertant genomes would continue to be generated de novo as a by-product of the replication of G133E cccDNA. These new cohorts of revertants will be converted to cccDNA and replicate, also expanding in parallel with the WT internal reference population. As a result, as all three virus populations expand in vivo, and new cohorts of revertants are generated from the parental G133E, the ratio of revertant virus to WT virus will increase over time. As the proportion of replicating G133E genomes diminishes due to death of infected cells or overgrowth of WT and revertant cccDNA, new cohorts of revertants contributed by G133E replication become less significant, and eventually the ratio of revertants and WT virus stabilizes. The final ratio is a measure of the sum of the revertants present in the inoculum plus those generated in vivo and the total WT reference in the inoculum. Since the ratio of WT reference virus to G133E in the inoculum is known, the amount of each revertant per inoculated G133E genome accumulated over multiple generations required for spread of infection throughout the liver can be inferred. In our experiments, the frequency of revertants determined in this manner was between 1 × 10⁵ and 6 × 10⁵ per G133E genome inoculated.

The relationship between the frequency of revertants that we measured and the mutation rate, i.e., the probability of mutation per nucleotide per generation, is complex because we have measured the frequency of specific mutations over an unknown number of generations of G133E DNA replication. In a mathematical model presented in the Appendix, we have estimated that the final frequency of each specific revertant in our experiment is probably between 4- and 23-fold higher than the reversion rate at that site. For the same-site revertant, E133G, which grows at the same rate as does WT virus, the final revertant frequency was 6 × 10⁵, implying that the rate of the mutation E133G was between 0.26 × 10⁵ and 1.5 × 10⁵ per generation. Accordingly, the error rate for all three possible substitutions at this position would be 0.8 × 10⁵ to 4.5 × 10⁵ errors per genome synthesized, or one error per 7 to 42 viral genomes synthesized. All specific second-site revertants were found at somewhat lower frequencies (Table 2), suggesting that these revertants may have replicated at a slightly reduced rate compared with that of the true WT revertant (E133G) and the reference virus, even though a reduced replication rate was not detected in the competition experiment. In support of this notion, the revertant T140I, which was present at the lowest frequency, appeared to undergo a weak negative selection in mixed infection with WT. In addition, error frequencies may differ at different positions and for specific substitutions (6, 9).

In this limited survey, substitutions at three sites in the vicinity of residue 133, namely, amino acids 129, 135, and 140, were able to revert the G133E cytopathic phenotype, presumably by restoring the normal function of the pre-S protein in virus release and cccDNA regulation. This result does not necessarily mean that the region inclusive of amino acids 129 to 140 is part of a functional site of interaction of the pre-S protein with its ligand, e.g., mature capsids. It is not known how the G133E mutation affects pre-S function in virus production and regulation of cccDNA synthesis. While the mutation may directly alter a functional site on the protein, it is also possible that the formation of a domain of the protein that contains the functional site is affected by the mutation, or that the G133E mutation causes a partial folding defect resulting in reduced amounts of properly folded protein. Each such defect could potentially be suppressed by substitution of amino acids neighboring the original mutation.

Interestingly, the glutamate at position 133 of the pre-S open reading frame, known to be cytopathic in the context of DHBV-16, is naturally present in most DHBVs of Chinese or Australian origin without apparent cytopathicity (28). These viruses all carry a lysine at position 135 of the pre-S protein and not a glutamine, as does DHBV-16. The selection for revertants with Q135R or Q135H as second-site mutations might therefore be anticipated if certain basic amino acid side chains at this position can suppress the phenotype resulting from a glutamate at position 133. Although the subsititution Q135K can theoretically be generated by a single nucleotide substitution (CAGAAG), this change would introduce a stop codon (TCATAA) in the overlapping P open reading frame and would be inconsistent with viability. While P gene mutations may be generated during the reverse transcription step, genomes bearing these mutations were unlikely to be propagated during expansion of the revertant pool.

We do not understand the early emergence of the mutant E133K from the population of cytopathic viruses and its later decline seen for bird 440 (Table 1). We previously showed that this mutant is unable to compete effectively with WT virus in a mixed infection and that infection of ducklings with the E133K virus produces an acute liver injury similar to that seen with G133E (16). It is possible that initially this mutant had some undefined advantage over WT revertants that was confined to early stages of infection of the duckling and that the cytopathic phenotype caused rapid selection against it during later stages. Alternatively, the mutant may be generated by an extraordinarily high mutation frequency at position 1199. We have not adequately tested either of these possibilities.