Ribavirin Treatment Failure-Associated Mutation, Y1320H, in the RNA-Dependent RNA Polymerase of Genotype 3 Hepatitis E Virus (HEV) Enhances Virus Replication in a Rabbit HEV Infection Model

ABSTRACT Chronic hepatitis E virus (HEV) infection has become a significant clinical problem that requires treatment in immunocompromised individuals. In the absence of an HEV-specific antiviral, ribavirin (RBV) has been used off-label, but treatment failure may occur due to mutations in the viral RNA-dependent RNA polymerase (RdRp), including Y1320H, K1383N, and G1634R. Chronic hepatitis E is mostly caused by zoonotic genotype 3 HEV (HEV-3), and HEV variants from rabbits (HEV-3ra) are closely related to human HEV-3. Here, we explored whether HEV-3ra, along with its cognate host, can serve as a model to study RBV treatment failure-associated mutations observed in human HEV-3-infected patients. By utilizing the HEV-3ra infectious clone and indicator replicon, we generated multiple single mutants (Y1320H, K1383N, K1634G, and K1634R) and a double mutant (Y1320H/K1383N) and assessed the role of mutations on replication and antiviral activity of HEV-3ra in cell culture. Furthermore, we also compared the replication of the Y1320H mutant with the wild-type HEV-3ra in experimentally infected rabbits. Our in vitro analyses revealed that the effects of these mutations on rabbit HEV-3ra are altogether highly consistent with those on human HEV-3. Importantly, we found that the Y1320H enhances virus replication during the acute stage of HEV-3ra infection in rabbits, which corroborated our in vitro results showing an enhanced viral replication of Y1320H. Taken together, our data suggest that HEV-3ra and its cognate host is a useful and relevant naturally occurring homologous animal model to study the clinical relevance of antiviral-resistant mutations observed in human HEV-3 chronically-infected patients.

in analogy to HEV-1-associated high mortality during pregnancy, HEV-3ra reportedly induces adverse fetal outcomes as well as severe liver injury in pregnant rabbits (45,48), suggesting that rabbits could be used as a promising animal model to study the potential mechanisms of HEV-1-associated fulminant hepatic failure in pregnant women (49,50).
Taken together, given the close evolutionary relationships between human HEV-3 and rabbit HEV-3ra, rabbits along with HEV-3ra can serve as a model to explore the mechanisms of chronic and extrahepatic HEV-3 infections (38,51). In this study, we aimed to determine the effect of three clinically reported RBV treatment failure-associated HEV-3 RdRp mutations (Y1320H, K1383N, and G1634R) using HEV-3ra and rabbits as a model. Our comprehensive in silico analyses demonstrated the close evolutionary relationships between rabbit HEV-3ra and human HEV-3. We further performed comparative mutational and antiviral analyses of distinct HEV-3ra mutants using reverse genetic and indicator replicon systems in vitro. The mutation (Y1320H) with the most significant enhancement of HEV-3ra replication in vitro was used to experimentally infect rabbits, which exhibited a replicationenhanced phenotype during the acute stage of virus infection in Y1320H-infected rabbits. Our data provide an experimental rationale for using HEV-3ra and its cognate host as a surrogate model to study the potential function of RBV treatment failure-associated HEV-3 RdRp mutations identified in CHE patients.

RESULTS
Rabbit HEV-3ra is evolutionarily closely related to human HEV-3. Thus far, eight major HEV genotypes have been assigned to the species balayani of the genus Paslahepevirus. Each genotype has a distinct host range and infection pattern (4,6). According to the consensus proposal from the ICTV Hepeviridae Study Group, the rabbit HEV strains were assigned a subtype within HEV-3, designated HEV-3ra (6). Phylogenetically, representative members of eight HEV genotypes of the genus Paslahepevirus clustered separately, and the HEV-3ra strains formed a sister clade to the HEV-3 (Fig. 1A). Notably, the HEV-3ra clearly showed an even higher genetic diversity than HEV-3, let alone other HEV genotypes (Fig. 1A), reflecting an extremely longterm virus-host evolutionary history of HEV-3ra in rabbits (52). Given the broad host tropism and frequent cross-species transmission of HEV-3 (5,8), it is therefore speculated that rabbits (order Lagomorpha) accidentally acquired the ancient HEV-3, likely due to the occurrence of intensive animal husbandry farming (53). A total of 52 genomic sequences of HEV-3ra are currently available in the GenBank database (retrieved as of October 2022), which were derived from multiple countries worldwide, with the vast majority from Australia. Although three independent studies from France, Switzerland, and Ireland have reported numerous cases of zoonotic HEV-3ra infection in humans, complete HEV genomes in only five French patients were reported. We found that these five strains dispersed in the rabbit HEV-3ra phylogenetic tree (Fig. 1B), suggesting that other HEV-3ra strains are likely zoonotic as well. The HEV-3ra LR strain (GenBank accession no. LC484431) located within the rabbit HEV-3ra cluster was isolated from Inner Mongolia, China, and its reverse genetic system was recently developed (Fig. 1B) (27).
We further compared the sequence identities between HEV-3ra and four major human-infecting HEV genotypes (HEV-1 to -4). As anticipated, HEV-3ra has the highest genomic sequence homology to human HEV-3 at both the nucleotide (77.5%) and amino acid (88.9%) levels (Fig. 1C). Sequence comparisons of the RdRp region of the HEV genome demonstrated that HEV-3ra shared sequence identity of 80.4% at the nucleotide level and 94.9% at the amino acid level, much higher than that of other genotypes (Fig. 1D). Therefore, our phylogenetic and sequence analyses revealed that the rabbit HEV-3ra is evolutionarily closely related to human HEV-3. The high genomic similarity between rabbit HEV-3ra and human HEV-3 in the RdRp region, as demonstrated in our in silico analyses here, suggests that HEV-3ra would be a valuable surrogate model for studying the effect of RdRp mutations on human HEV-3 replication and pathogenesis. FIG 1 Phylogeny of rabbit HEV-3ra and sequence identities between rabbit HEV-3ra and four major HEV genotypes. (A) A neighbor-joining tree of maximum likelihood distances was generated based on HEV genomic sequences of representative members within the genus Paslahepevirus. Clusters of eight distinct HEV genotypes and the rabbit HEV-3ra were highlighted with different colors. (B) A maximum-likelihood tree was generated based on 52 genomic sequences of rabbit HEV-3ra. The HEV-3 reference Meng strain (GenBank accession no. AF082843) served as an outgroup. Virus designations include GenBank accession number, country origin, and host. HEV-3ra sequences used in this study and those detected in humans are highlighted with red-and orange-filled circles, respectively. Evolutionary analyses were conducted in molecular evolutionary genetics analysis 11 (MEGA 11) based on the multiple sequence alignment in Geneious Prime. General time reversible (GTR) 1 gamma distributed (G) 1 invariable sites (I) nucleotide substitution model with the lowest Bayesian information criterion (BIC) score was selected based on the find best-fit substitution model (ML) in MEGA 11. Bootstrap values (.90%) are presented at specific nodes. Scale bars indicate the estimated number of nucleotide substitutions per site. (C) Comparisons of nucleotide sequences of complete genomes and amino acid sequences of concatenated ORF1 and ORF2 among four major HEV genotypes and HEV-3ra. (D) Comparisons of nucleotide and amino acid sequences of RNA-dependent RNA polymerase (RdRp) among four major HEV genotypes and HEV-3ra. Representative viral genomes of eight HEV genotypes and HEV-3ra were included for analyses. RBV treatment failure-associated human HEV-3 RdRp mutations (Y1320H, K1383N, and G1634R) to the HEV-3ra genome and epidemiological  prevalence of three mutations among 8 HEV genotypes and HEV-3ra. Recently,  multiple HEV-3 RdRp mutations, including Y1320H, K1383N, D1384G, K1398R, V1439I,  Y1587F, and G1634R, have been reportedly associated with RBV treatment failure in CHE patients in several European countries (18,20,54); among them, the Y1320H, K1383N, and G1634R were identified in a patient experiencing RBV monotherapy treatment failure and have therefore been further studied in vitro using the human HEV-3 infectious clone Kernow-C1 p6 (GenBank accession no. JQ679013) and indicator replicon p6Gluc (21), which revealed the roles of enhanced viral replication for Y1320H and G1634R and the increased RBV sensitivity for K1383N. However, the mechanisms of these mutations on RBV sensitivity and viral replication remain poorly understood since in vivo experimental confirmation in a suitable and relevant animal model is still lacking. To investigate the potential impact of RBV treatment failure-associated HEV-3 RdRp mutations on rabbit HEV-3ra, we performed sequence alignment of the HEV-3 reference Meng strain (GenBank accession no. AF082843) and 52 strains of HEV-3ra and successfully mapped the three RBV treatment failure-associated HEV-3 RdRp mutations to HEV-3ra ( Fig. 2A). We found that the amino acid residues at positions 1320 and 1383 of HEV-3ra are Y (Tyr) and K (Lys), respectively, which are identical to those of HEV-3; however, the primary amino acid residues at position 1634 of rabbit HEV-3ra are K (Lys) and R (Arg) but not G (Gly).
Y1320H mutation significantly enhances HEV-3ra (LR) replication in cell culture. Given that the rabbit HEV-3ra is evolutionarily closely related to human HEV-3, to determine the impact of RBV treatment failure-associated HEV-3 RdRp mutations on the replication of rabbit HEV-3ra, we utilized an established reverse genetic system of an HEV-3ra LR strain (GenBank accession no. LC484431) (Fig. 3A), which has been shown to successfully propagate in human hepatoma PLC/PRF/5 cells and induce persistent viral infection in inoculated rabbits (27). Using the HEV-3ra LR infectious clone as the backbone, we first generated four individual LR mutants (LR_Y1320H, LR_K1383N, LR_K1634G, and LR_K1634R), each containing a single amino acid mutation, which corresponds to one of three RBV treatment failureassociated HEV-3 RdRp mutations (21). Because the original amino acid residue in LR at genomic position 1634 is K (Gly), we mutated K1634 to both G1634 and R1634. In each of the four LR mutants, either one or two nucleotides were correctly substituted from wildtype LR infectious clone using site-directed mutagenesis systems ( Fig. 3B; see also Table S2 in the supplemental material). In vitro capped RNA transcripts from wild-type LR (LR_WT) and each of four LR mutants were transfected to Huh7-S10-3 liver cells, and the transfected cells were stained with the anti-HEV ORF2 antibody at 7 days posttransfection (dpt). HEV-3ra-positive foci were observed microscopically in LR_WT, LR_Y1320H, LR_K1634G, and LR_K1634R. Remarkably, there were obviously more HEV-3ra-positive foci in LR_Y1320H than in LR_WT (Fig. 3C). The HEV-3ra-positive cells in the immunofluorescence assays were quantified, and the results showed that there were significantly more HEV-3ra-positive foci in LR_Y1320H and significantly fewer positive foci in LR_K1383N. In contrast, there was no significant difference in HEV-3ra-positive foci in LR_K1634G or LR_K1634R compared with that in LR_WT (Fig. 3D). Similar results were also obtained from the quantification of the amounts of virus in the media of transfected cells by measuring viral RNA loads with an HEV-specific reverse transcription-quantitative PCR (RT-qPCR) (56). We found that the LR_Y1320H replicated and/or assembled more efficiently than LR_WT, while LR_K1383N significantly impaired the virus replication. Although neither LR_K1634G nor LR_K1634R significantly affected virus replication compared to the LR_WT, the LR_K1634G replicated at a reduced level and the LR_K1634R at an enhanced level (Fig. 3E).
Recombinant virus replicons that encode indicator genes provide valuable tools for studying viral replication and sensitivity to small molecule inhibitors (19)(20)(21). To ensure the reproducibility of our in vitro data from the HEV-3ra infectious clone system, we further developed an HEV-3ra indicator replicon system. Briefly, the N-terminal ORF2 sequence of rabbit HEV-3ra LR infectious clone is replaced by a secreted version of the Gaussia luciferase (GLuc) gene (Fig. 4A). Two restriction enzymes, SpeI and EcoNI, and overlapping PCRs were employed to construct the indicator replicon of HEV-3ra LR GLuc (designated LRGluc or LRG; GenBank accession no. OP887158). Restriction digestion analysis of the LRGluc plasmid showed the expected size of DNA fragments (see Fig. S1 in the supplemental material). The daily Gluc activity was measured continuously for 10 days, showing that LRGluc is replication competent in Huh7-S10-3 liver cells, 8 dpt would be sufficient to nearly reach peak luminescence (1.1 Â 10 4 units) (Fig. 4B), and LRGluc is suitable for testing HEV-3ra replication efficiency. Subsequently, we constructed four LRGluc single mutants (LRG_Y1320H, LRG_K1383N, LRG_K1634G, and LRG_K1634R), and the Gluc activity of wild-type LRG (LRG_WT) and LRG mutants was tested at 4 and 8 dpt in the culture supernatant of Huh7-S10-3 liver cells (Fig. 4C). At 8 dpt, compared with that of LRG_WT, the LRG_Y1320H demonstrated significantly enhanced viral replication efficiency with a 1.9-fold increase. On the contrary, the LRG_K1383N significantly decreased viral replication to a level of mock transfection. The LRG_K1634G slightly decreased while LRG_K1634R increased viral replication, although the difference is not statistically significant. Therefore, the results derived from the HEV-3ra LR GLuc indicator replicon system are highly consistent with the HEV-3ra LR infectious clone system. Collectively, the impact of three RBV treatment failure-associated HEV-3 RdRp mutations on in vitro viral replication using the rabbit HEV-3ra LR model is highly comparable with using the human HEV-3 Kernow-C1 p6 (21), a cell culture-adapted strain (57), indicating that HEV-3ra is a suitable surrogate model for studying the RBV treatment failure-associated HEV-3 RdRp mutations. The major advantage of using rabbit HEV-3ra, instead of human HEV-3 Kernow-C1 p6, is that the potential role of any RBV treatment failure-associated HEV-3 RdRp mutations can be further tested in a small homologous animal host, rabbit, for HEV-3ra.
Y1320H mutation rescued the K1383N mutation-associated HEV-3ra replication defects, and Y1320H/K1383N double mutant has a markedly higher sensitivity to RBV treatment. It has been reported that Y1320H, G1634R, and the hypervariable region (HVR) insertion can compensate for the K1383N-associated replication defects Comparative analyses of replication efficiency of LRG wild type and mutants. At 4-and 8-days posttransfection with LRG wild type and mutants, respectively, cell culture media of Huh7-S10-3 cells were harvested, and the Gluc expression activity was measured and compared. Values represent means plus SD (error bars) from independent experiments (n = 4). Statistical differences were determined with one-way ANOVA. **, P , 0.01; ***, P , 0.001; ns, not statistically significant.
on the human HEV-3 Kernow-C1 p6 strain (21). To evaluate whether we can reproduce the human HEV-3 p6 results utilizing the rabbit HEV-3ra LR model system, especially given our observation of the highly significant enhancement of Y1320H on HEV-3ra LR replication ability, we further generated the LRG_Y1320H/K1383N double mutant and compared its replication efficiency with LRG_WT and LRG single mutants. We found that the Y1320H rescued the K1383N-associated HEV-3ra replication defects, even to a similar level of LRG_WT (Fig. 5A), which is consistent with the reported impact of Y1320H on human HEV-3.
Additionally, previous in vitro studies showed that the K1383N mutation significantly altered viral fitness as well as RBV susceptibility and may play a crucial role in RBV treatment failure in clinical cases (20,21). To assess the impact of K1383N on the sensitivity of HEV-3ra to RBV, we added 1 mM or 10 mM RBV to Huh7-S10-3 cells transfected with LRG_WT and LRG single and double mutants, and cells without the addition of RBV served as controls. The results from luminescence-based antiviral assays showed that the administration of 1 mM RBV significantly inhibited viral replication, particularly for the LRG_Y1320H/K1383N, and the administration of 10 mM RBV reduced the luminescence activity of LRG_WT and LRG mutant to mock levels (Fig. 5B). Therefore, like human HEV-3, the K1383N mutation also increased the RBV susceptibility for rabbit HEV-3ra. In line with the previous findings regarding the antiviral sensitivity of Y1320H and G1634R mutations using human HEV-3, we also showed that neither Y1320H nor K1634G/R significantly affected RBV susceptibility of HEV-3ra (21). Overall, these data demonstrated that three RBV treatment failure-associated HEV-3 RdRp mutations have a comparable impact on the in vitro replication and RBV sensitivity of HEV-3ra, further indicating that HEV-3ra and its cognate host are a good and relevant model system for studying the antiviral resistance and pathogenesis of RBV treatment failure-associated HEV-3 RdRp mutations.
Physicochemical and structural analyses of RBV treatment failure-associated HEV-3 RdRp mutations. Based on our experimental results in this study, we showed that each of the three RBV treatment failure-associated HEV-3 RdRp mutations has a distinct impact on the in vitro replication of HEV-3ra. In clinical cases, the K1383N occurred in all four patients with RBV treatment failure, but in vitro studies showed that this mutation significantly increased RBV sensitivity, which contradicts the clinical FIG 5 The Y1320H mutation compensated for the K1383N mutation-associated replication defects, and the Y1320H/ K1383N double mutant has a markedly higher sensitivity to RBV. (A) Comparative analyses of replication efficiency in Huh7-S10-3 cells of HEV-3ra LRG wild type and mutants containing a single RdRp mutation or a double mutation (Y1320H/K1383N). (B) Comparative analyses of RBV sensitivity to HEV-3ra LRG wild type and mutants. Huh7-S10-3 cells were cultured without RBV or with 1 mM RBV or 10 mM RBV. At 7 days posttransfection with LRG wild-type and mutants, respectively, cell culture media of Huh7-S10-3 cells were collected, and the Gluc expression activity was measured and compared. Values represent means plus SDs (error bars) from independent experiments (n = 4). Statistical differences were determined with one-way ANOVA. *, P , 0.05; ***, P , 0.001; ns, not statistically significant.
phenotype (20,21). Of note, we found that the K1383N located in motif I of the HEV RdRp region, which is associated with structural conformations, is therefore conserved across eight HEV genotypes and HEV-3ra ( Fig. 2B and Fig. 6A). It seems reasonable to speculate that a single nucleotide substitution in this region can essentially affect the secondary structure of HEV RdRp, which may explain the significant reduction of viral replication with the K1383N mutation. In contrast, the Y1320H and K1634G/R are located in nonmotif regions of RdRp, which supposedly tolerate amino acid changes. Intriguingly, although we showed that the 1634 position is relatively heterogenic in HEV genomes, the 1320 position is highly conserved across HEV genotypes (Fig. 2B).
We further conducted physicochemical comparisons of the three RBV treatment failure-associated HEV-3 RdRp mutations and found that the positively charged amino acid residues at positions 1320 and 1634 are preferable to hydrophobic amino acids for HEV-3 in vitro replication (Fig. 6B) (19). Specifically, the substitution of hydrophobic amino acid Y (Tyr) to positively charged amino acid H (His) at position 1320 and the substitution of hydrophobic amino acid G (Gly) to positively charged amino acids K (Lys) and R (Arg) at position 1634 increased the replication efficiency of HEV-3 ( Fig. 3D and E and Fig. 4C). Additionally, our structural analyses demonstrated that the K1383 is located at the end of a b-sheet, and the introduction of the N1383 mutation transformed the central three-dimensional structure of the RdRp of HEV-3 (Fig. 6C). On the contrary, the H1320 and G/R1634 do not alter the primary structure of the HEV-3. Therefore, it is hypothesized that the K1383N altered the structure of HEV-3 RdRp and eventually led to RBV treatment failure of HEV-3 in CHE patients. Indeed, the selection of K1383N is reversible when RBV therapy stops in clinical cases (20). Nonetheless, the exact underlying mechanisms of HEV-3 RdRp mutations during RBV monotherapy are still to be determined.
The Y1320H mutant enhances virus replication during the acute stage of infection in HEV-3ra-infected rabbits. It has previously been reported that the Y1320H significantly increased in vitro replication of human HEV-3 p6 as well as p6 chimeric constructs with ORF1 sequence from the CHE patient (21). Similarly, as noted above, in this study we also showed that the Y1320H significantly enhanced in vitro replication of HEV-3ra LR. To determine the impact of Y1320H mutation on viral replication of HEV-3ra LR in vivo using HEV-3ra infection in rabbits as the model, we produced LR_WT and LR_Y1320H infectious virus stocks in transfected Huh7-S10-3 cells and intravenously inoculated female New Zealand White rabbits with LR_WT (n = 5), LR_Y1320H (n = 5), and phosphate-buffered saline (PBS) (n = 5) as negative controls (Fig. 7A).
To monitor the viral infection kinetics in inoculated rabbits, twice-weekly fecal samples and weekly-serum samples from each rabbit in all 3 groups were collected for a total of 77 days and used to quantify the amount of viral RNAs in feces and serum samples by an HEV-specific RT-qPCR (56). We found that fecal virus shedding in the LR_Y1320H-infected rabbits occurred much earlier with higher viral RNA loads during the acute stage of HEV-3ra infection before 24 days postinoculation (dpi), and the fecal viral RNA loads were significantly higher at 14 dpi in LR_Y1320H-infected rabbits than that of animals infected with LR_WT (P = 0.008) (Fig. 7B). Similarly, the serum viral RNA loads in LR_Y1320H-infected rabbits were numerically higher during the acute stage of HEV-3ra infection before 21 dpi, although the difference is not statistically significant (Fig. 7C). Therefore, the Y1320H mutation enhanced virus replication during the acute stage of HEV-3ra infection in rabbits, which is consistent with our in vitro results showing an enhanced viral replication of Y1320H compared to the LR_WT (Fig. 3D and E and Fig. 4C). Interestingly, after 24 dpi (feces) or 21 dpi (blood), the viral RNA loads in both feces and blood of LR_Y1320H-infected rabbits were numerically lower compared to those of LR_WT-infected rabbits until the end of the animal study at 77 dpi ( Fig. 7B and C), although the difference is not statistically significant. It should be noted that the viral RNA loads in feces and blood of both LR_WT-and LR_Y1320H-infected rabbits were generally low and lasted for at least 77 dpi, indicating both LR_WT and LR_Y1320H induced persistent HEV-3ra infection in rabbits, which corroborates with an earlier report that the LR_WT caused persistent infection (27).
(Continued on next page) Y1320H Mutation Enhances HEV Replication mBio Seroconversion to IgG anti-HEV-3ra started at 21 dpi in both LR_WT-and LR_Y1320Hinfected rabbits (Fig. 7D), concomitant with a rapid decrease of serum viral RNA loads in LR_WT-infected rabbits (28 dpi onward) and in LR_Y1320H-infected rabbits (21 dpi onward) (Fig. 7C). Notably, the anti-HEV-3ra IgG antibody titers in LR Y1320H-infected rabbits were numerically higher than those of LR_WT-infected rabbits (Fig. 7D), although the differences were not statistically significant. The anti-HEV IgG antibody titers were maintained in both LR_WT-and LR_Y1320H-infected rabbits until the end of the study at 77 dpi. As expected, all rabbits from the negative control group (PBS) remained seronegative throughout the study (Fig. 7D).
The viruses recovered from rabbits inoculated with LR_Y1320H were sequenced to verify the in vivo stabilities of the introduced mutation in the viral RdRp region. Sequence analyses revealed that the Y1320H mutation was maintained in the viruses recovered from feces, liver, and spleen collected at necropsy at 77 dpi, indicating that the Y1320H mutation is genetically stable in infected rabbits.

DISCUSSION
Zoonotic HEV-3 infections have recently received considerable attention since they are associated with CHE in immunocompromised patients, particularly in solid-organ transplant recipients (10,58), and with a number of neurological sequelae (12,59). Although there is currently no specific therapy for chronic HEV-3 infections, abundant evidence has shown that RBV as monotherapy is effective in the treatment of CHE (11,15). However, owing to the mutagenic effect of RBV on viruses, several amino acid mutations, including Y1320H, K1383N, D1384G, K1398R, V1479I, Y1587F, and G1634R, in the RdRp region of HEV-3 have reportedly occurred during RBV treatment in CHE patients (18,20,54). Of note, the Y1320H/K1383N/G1634R/triple mutant is associated with RBV resistance in a clinical case, and in vitro studies showed that Y1320H and G1634R enhance HEV replication, whereas K1383N decreases HEV replication but increases RBV susceptibility, thereby contradicting the clinical observation (21). More recently, additional amino acid substitutions (P25S, G38S, A64T, G71R, P79S, S95P, V245I, and T324S) in HEV-3 ORF2 protein are reportedly associated with a sustained viral response despite RBV therapy, and the P79S mutant impaired antibody-mediated neutralization of HEV-3, thus potentially acting as an immune decoy (20,22). Nonetheless, due to the lack of a tractable and relevant animal model for chronic HEV-3 infection, the mechanisms of action or clinical relevance of these RBV treatment failure-associated mutations in HEV-3 replication and pathogenesis are largely unknown.
Rabbits are natural reservoirs of variant strains of HEV (25). Since the rabbit HEV variants and human HEV-3 phylogenetically cluster together, the ICTV has assigned rabbit HEV within the same genotype as human HEV-3, termed HEV-3ra (6). In our in silico analyses, we showed that HEV-3ra is evolutionarily closely related to human HEV-3, and HEV-3ra from rabbits has a significantly high level of sequence homology with human HEV-3, especially in the RdRp region (Fig. 1). Therefore, the HEV-3ra along with its cognate host, the rabbit, may potentially serve as a valuable and relevant model to elucidate the underlying mechanisms of RBV treatment failure-associated HEV-3 RdRp mutations identified in CHE patients. Furthermore, it has been shown that the HEV-3ra causes persistent infection and extrahepatic manifestations in rabbits, which could potentially serve as a much-needed smaller animal model to study the mechanism of chronic HEV-3 pathogenesis (27)(28)(29)(30)(31)(35)(36)(37)(38).
By utilizing a rabbit HEV-3ra infectious clone, which is derived from an HEV-3ra LR strain and reportedly induces persistent HEV infection in rabbits (Fig. 1) (27), we determined the functional impact of the three most notable RBV treatment failureassociated HEV-3 RdRp mutations (Y1320H, K1383N, and G1634R) on replication and RBV antiviral sensitivity of the model rabbit HEV-3ra virus. We demonstrated that the Y1320H replicated and/or assembled much more efficiently than the WT in the Huh7-S10-3 liver cells, while the K1383N significantly impaired HEV-3ra replication. Neither K1634G nor K1634R significantly affected HEV-3ra replication compared to the WT. The K1634G replicated at a slightly reduced level and the K1634R at a slightly enhanced level (Fig. 3). Moreover, we constructed an HEV-3ra indicator replicon with the GLuc reporter system and used it to further validate our results through comparative mutational analyses. The results from the GLuc reporter system are consistent with those obtained with the infectious clone system (Fig. 4), suggesting that our results from this study are reproducible in two different systems. Surprisingly, a combinational double mutant containing the Y1320H/K1383N double mutations rescued the K1383N-associated replication defects. Furthermore, we demonstrated that the replication levels of HEV-3ra WT and mutants were significantly decreased, particularly for the Y1320H/K1383N mutant, when the cells were treated with RBV, indicating that the K1383N increased the RBV sensitivity of rabbit HEV-3ra (Fig. 5). Taken together, our data on the effect of three RBV treatment failure-associated HEV-3 RdRp mutations using the rabbit HEV-3ra LR model system is highly comparable with in vitro data obtained with human HEV-3 (21), indicating that rabbit HEV-3ra LR could serve as a relevant and useful model for studying the replication and antiviral sensitivity of RBV treatment failure-associated HEV-3 RdRp mutations. A major advantage of rabbit HEV-3ra over human HEV-3 is that the mutations in HEV-3ra can be readily tested with a homologous animal model since HEV-3ra naturally infects rabbits and causes persistent infection. Nevertheless, it should be noted that the rabbit HEV-3ra LR grows less efficiently in cultured cells than human HEV-3 Kernow-C1/p6 (21,22,50,(60)(61)(62).
There is a discrepancy in amino acid preference at position 1634 between HEV-3 from humans and pigs (G1634) and HEV-3ra from rabbits (K1634); moreover, whether R1634 merely occurs in CHE patients remains to be determined (19). To better understand the potential mechanism of these RdRp mutations in RBV resistance, we performed physicochemical analyses of the above-described three RBV treatment failure-associated HEV-3 RdRp mutations. Our results showed that the positively charged amino acids at positions 1320 and 1634 are preferable to hydrophobic amino acids for HEV-3 replication in vitro, but to what extent the biochemical properties of these distinctive amino acids affect HEV-3 replication remains an open question. Of note, the K1383N occurred in all four CHE patients with RBV treatment failure, and the selection of K1383N is reversible after the cessation of RBV treatment (20,21). On the contrary, a retrospective multicenter study reported that the predominant RdRp mutations in HEV-3, e.g., K1383N, do not affect the viral clearance rate (54). We showed that the K1383 is located at motif I of HEV RdRp region, and structural prediction revealed that substitution of amino acid at position 1383 would alter the central three-dimensional structure of HEV RdRp (Fig. 6), which may explain the significant reduction of HEV-3 replication with the K1383N mutation. Indeed, the K1383 is highly conserved across eight different HEV genotypes, but the N1383 has only been observed in minor HEV-3 genomes of CHE patients (Fig. 2). Nonetheless, the exact mechanism of the K1383N mutation occurrence during chronic HEV-3 infection and RBV treatment remains elusive.
Since the Y1320H mutation had the most significant enhancement on the replication of HEV-3ra in our in vitro study, we proceeded to take advantage of the unique homologous rabbit model system of HEV-3ra to investigate whether the Y1320H mutation has similar impacts on virus replication in vivo. We found that fecal virus shedding in the HEV-3ra LR_Y1320H-infected rabbits occurred much earlier with higher viral RNA loads during the acute stage of virus infection before 24 dpi, and the fecal viral RNA loads were significantly higher at 14 dpi in LR_Y1320H-infected rabbits than those in rabbits infected with LR_WT (Fig. 7). Similarly, the serum viral RNA loads in LR_Y1320H-infected rabbits were numerically higher during the acute stage of virus infection before 21 dpi. Therefore, the Y1320H mutation appears to enhance virus replication during the acute stage of HEV-3ra infection in rabbits, which corroborated with our in vitro results showing an enhanced viral replication of LR_Y1320H compared to that of the LR_WT. However, after 21 dpi (blood) or 24 dpi (feces) until the end of the study, the viral RNA loads in both feces and blood of LR_Y1320Hinfected rabbits were numerically lower compared to those of the LR_WT-infected rabbits, although these differences were not statistically significant. Seroconversion to IgG anti-HEV-3ra occurred at 21 dpi with numerically higher titers in LR_Y1320H-infected rabbits than in LR_WT-infected rabbits. The appearance of anti-HEV IgG antibodies was concomitant with a rapid decrease of serum viral RNA loads in LR_WT-infected rabbits (starting 28 dpi) and in LR_Y1320H-infected rabbits (starting 21 dpi). The heightened humoral immune response in the LR_Y1320H-infected rabbits would explain the decreased viral loads in infected animals after seroconversion. However, it is likely that the neutralizing antibodies in the infected rabbits failed to completely neutralize and clear the virus in rabbits, perhaps due to the presence of the quasi-enveloped form of viruses (9,63), thus leading to the observed persistent infection. It should be noted that the Y1320 is evolutionarily conserved in nature among eight different HEV genotypes including rabbit HEV-3ra. Similar to the N1383, the H1320 merely occurred in a few HEV-3 genomes in CHE patients (Fig. 2), which may explain the less efficient in vivo replication of the LR_Y1320H than LR_WT during the late stage of HEV-3ra infection in rabbits (Fig. 7). Whether H1320 exists in natural strains of HEV-3 remains unknown.
Our in vivo results showed persistent HEV-3ra infection in rabbits, which is consistent with previous studies (27)(28)(29)(30)(31) and, therefore, may offer an opportunity in the future to study prolonged HEV-3 infection and its progression to chronicity. Despite this, whether persistent HEV-3 infection in rabbits could be recognized as a chronic infection, how to clearly define the measurable parameters of chronic HEV-3ra infection in rabbits and how the measurements correlate to chronic human HEV-3 infection remain to be determined in the future. Moreover, given the fact that HEV-3ra can persist in rabbits, it would be interesting, in future studies, to determine if RBV resistance mutations can develop in this unique rabbit HEV-3ra infection animal model when treated with RBV.
In conclusion, in this study, we demonstrated that rabbit HEV-3ra is evolutionarily closely related to human HEV-3. Thus, along with its cognate host (rabbit), HEV-3ra is a relevant and useful model to investigate the impact of RBV treatment failure-associated RdRp mutations of human HEV-3 on in vitro replication of HEV-3ra, as demonstrated by highly comparable in vitro results derived from human HEV-3. Importantly, we further revealed that the RBV treatment failure-associated Y1320H mutation significantly increased HEV-3ra replication efficiency in vitro and also enhanced HEV-3ra replication during the acute stage of virus infection at 14 dpi in rabbits. The appearance of anti-HEV-3ra IgG antibodies in infected rabbits coincided with a decreased virus replication, indicating that humoral immune response partially neutralized but did not clear virus infection in rabbits since the infected animals developed persistent HEV-3ra infections. These results collectively reveal that the HEV-3ra and its infection in rabbits are a relevant and useful model for studying the functional impact of RBV treatment failureassociated HEV-3 RdRp mutations identified in clinical cases. Phylogenetic and sequence analyses. Evolutionary analyses of rabbit HEV-3ra were conducted in Molecular Evolutionary Genetics Analysis (MEGA) (64) software for macOS version 11.0.11 with 1,000 bootstrap reiterations. A neighbor-joining tree of maximum likelihood distances was generated for the genus Paslahepevirus. The general time reversible (GTR) 1 gamma distributed (G) 1 invariable sites (I) nucleotide substitution model with the lowest Bayesian information criterion (BIC) score was selected for rabbit HEV-3ra based on the find best-substitution model (ML). Representative viral genomes of the eight HEV genotypes were included according to the proposed reference sequences for subtypes of HEV (genus Paslahepevirus) (6). In total, 52 rabbit HEV-3ra genomes were available and downloaded in the GenBank database (retrieved as of October 2022). Complete genomes are aligned using the multiple alignment using fast Fourier transform (MAFFT) (65) algorithm in Geneious Prime software version 2022.2.2.

MATERIALS AND METHODS
A total of 1,018 complete viral genomes of eight HEV genotypes and the rabbit HEV-3ra within the genus Paslahepevirus (82 genomes for HEV-1, 2 genomes for HEV-2, 644 genomes for HEV-3, 52 genomes for HEV-3ra, 225 genomes for HEV-4, 2 genomes for HEV-5, 2 genomes for HEV-6, 3 genomes for HEV-7, and 6 genomes for HEV-8) analyzed in this study were downloaded in the GenBank database (retrieved as of October 2022). Genomic sequences from each genotype were aligned using the MAFFT algorithm in Geneious Prime software version 2022.2.2. Nucleotide and amino acid numbering are according to the HEV prototype strain from Burma (GenBank accession no. M73218) (6). Locations of functional domains and motifs within HEV ORF1 are according to the Burma strain (66).
Rabbit HEV-3ra infectious clone and indicator replicon. The rabbit HEV-3ra infectious clone pUC57-T7RHEV-LR (designated LR), which was derived from the HEV-3ra LR strain (GenBank accession no. LC484431), was kindly provided by Tian-Cheng Li (National Institute of Infectious Disease, Tokyo, Japan). The LR strain has been shown to induce persistent HEV-3ra infection in rabbits (27). To generate Gaussia luciferase replicon pUC57-T7RHEV-LR-Gluc (designated LRG), primers LR_Spel_FW and LR_ORF1Gluc_RV were used to amplify fragment A1 and primers LR_ORF2Gluc_FW and LR_EcoNI_RV were used to amplify fragment A2 with pUC57-T7RHEV-LR as the template, and primers Gluc_FW and Gluc_RV were used to amplify fragment Gluc with Kernow-C1 p6Gluc as the template (50). Three fragments were amplified with overlapping PCR with primers LR_Spel_FW and LR_EcoNI_RV and subsequently assembled into the pUC57-T7RHEV-LR vector (SpeI and EcoNI digested) (see Fig. S1 in the supplemental material). All primers used to construct the pUC57-T7RHEV-LR-Gluc are listed in Table S2 in the supplemental material. inoculum, we infected the HepG2 liver cells with LR_WT and LR_Y1320H and determined the 50% tissue culture infectious dose (TCID 50 ) using immunofluorescence assays (IFA). Since HEV is a nonlytic virus and no cytopathic effect can be visualized in infected cells, we therefore determine the TCID 50 by microscopically counting HEV-3ra-positive foci. The TCID 50 determination for the HEV inoculum has been described in detail from our previous study (59). A total of 15 specific-pathogen-free (SPF) 6-month-old female New Zealand White rabbits, which tested negative for anti-HEV IgG and IgM antibodies and for rabbit HEV RNA, were divided into three groups of 5 animals each. One group of rabbits was intravenously inoculated via the ear vein with LR_WT (;3 Â 10 3 TCID 50 infectious doses per rabbit), another group was similarly inoculated with LR_Y1320H (;3 Â 10 3 TCID 50 infectious doses per rabbit), and the remaining group of animals was similarly inoculated with PBS as negative control. This animal study is approved by the Virginia Tech Institutional Animal Care and Use Committee (IACUC).
Fecal samples were collected prior to inoculation and twice weekly, and serum samples were obtained weekly from each of the inoculated rabbits. Fecal materials were diluted with PBS to prepare a 10% (wt/vol) fecal suspension. The fecal suspension was then clarified by centrifugation at 10,000 Â g for 15 min and filtered through a 0.45-mm membrane. Viral RNAs from rabbit fecal and serum samples were extracted using TRI Reagent and Quick-RNA viral kit (Zyno Research, Irvine, CA, USA), respectively. The fecal and serum samples were used for the quantification of HEV RNA loads by RT-qPCR, and the serum samples were also used for the detection of anti-HEV IgG antibodies by enzyme-linked immunosorbent assay (ELISA). All rabbits were necropsied at 11 weeks (77 days) postinoculation.
Statistical analysis. All statistical tests were performed in GraphPad Prism for macOS software version 9.4.1. Comparisons among three or more experimental groups in cell culture were performed with one-way analysis of variance (ANOVA) with Tukey's multiple comparison test. To evaluate the difference between HEV-3ra WT and Y1320H in inoculated rabbits at each time point, a two-sided, unpaired, multiple Student's t test without adjustments was performed. Differences were considered statistically significant when the P value was less than 0.05.
Data availability. The nucleotide sequence of the rabbit HEV RHEV-LR Gluc (LRG) indicator replicon generated in this study has been deposited into GenBank under accession no. OP887158.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, TIF file, 1.1 MB.