Research ArticleHEPATITIS

Preclinical assessment of antiviral combination therapy in a genetically humanized mouse model for hepatitis delta virus infection

See allHide authors and affiliations

Science Translational Medicine  27 Jun 2018:
Vol. 10, Issue 447, eaap9328
DOI: 10.1126/scitranslmed.aap9328

Diving into hepatitis delta virus

Although it is a major burden of hepatitis-related disease, there are no specific treatments for hepatitis delta virus (HDV), which depends on hepatitis B virus to replicate. Winer et al. developed a transgenic mouse model to study HDV and test therapeutic interventions. They observed immune responses and effects on the liver, as well as tested drug candidates. This model could help open up the pipeline for future HDV treatments.


Chronic delta hepatitis, caused by hepatitis delta virus (HDV), is the most severe form of viral hepatitis, affecting at least 20 million hepatitis B virus (HBV)–infected patients worldwide. HDV/HBV co- or superinfections are major drivers for hepatocarcinogenesis. Antiviral treatments exist only for HBV and can only suppress but not cure infection. Development of more effective therapies has been impeded by the scarcity of suitable small-animal models. We created a transgenic (tg) mouse model for HDV expressing the functional receptor for HBV and HDV, the human sodium taurocholate cotransporting peptide NTCP. Both HBV and HDV entered hepatocytes in these mice in a glycoprotein-dependent manner, but one or more postentry blocks prevented HBV replication. In contrast, HDV persistently infected hNTCP tg mice coexpressing the HBV envelope, consistent with HDV dependency on the HBV surface antigen (HBsAg) for packaging and spread. In immunocompromised mice lacking functional B, T, and natural killer cells, viremia lasted at least 80 days but resolved within 14 days in immunocompetent animals, demonstrating that lymphocytes are critical for controlling HDV infection. Although acute HDV infection did not cause overt liver damage in this model, cell-intrinsic and cellular innate immune responses were induced. We further demonstrated that single and dual treatment with myrcludex B and lonafarnib efficiently suppressed viremia but failed to cure HDV infection at the doses tested. This small-animal model with inheritable susceptibility to HDV opens opportunities for studying viral pathogenesis and immune responses and for testing novel HDV therapeutics.


Chronic hepatitis delta (CHD), caused by hepatitis delta virus (HDV), was first described as a distinct form of blood-borne hepatitis in 1977 (1). HDV is a 1679-nucleotide, enveloped, negative-sense RNA satellite virus and sole member of the delta virus genus [reviewed in (2)]. The small HDV circular genome is only ca. 1.7 kb in length and is stabilized by extensive intramolecular base pairing. HDV encodes a single open reading frame encoding the delta antigen (HDAg), which exists as two isoforms: the small and large HDAg (3). Because HDV requires the HBV surface antigens (HBsAgs) for packaging of viral particles and, thus, is dependent on the presence of HBV, it is considered a subviral satellite. Consequently, the early steps of HDV entry into hepatocytes follow the same mechanism as HBV. Once within the hepatocyte, a nuclear localization signal on HDAg-L triggers the translocation of the HDV nucleocapsid to the nucleus where the viral genome is replicated. The small size of the HDV genome results in the reliance of the virus on host enzymes, including cellular RNA polymerases, to successfully replicate [reviewed in (4)]. The incoming RNA serves as the template for transcripts longer than the size of the virus’ genome. Such multimeric, linear RNAs contain at least two copies of the antigenomic ribozyme, releasing unit-length linear RNA after self-cleavage. Antigenomic RNA is circularized and forms the template for new genomic RNAs following similar intermediate steps.

Coinfection with HBV/HDV or superinfection of chronic HBV patients with HDV usually progresses to delta persistence, frequently resulting in fibrosis, cirrhosis, and hepatocellular carcinoma. Although HDV and HBV infection can be prevented by prophylactic vaccination, effective and curative HDV treatments do not exist.

Studies of the mechanisms of viral persistence, pathogenesis, and development of effective therapies for CHD have been hampered by the scarcity of experimentally tractable animal models. HDV has a narrow and poorly understood host tropism limited to infections in humans and a few primate species that are also susceptible to HBV (5). The only available small-animal models susceptible to HBV/HDV co- or superinfections are human liver chimeric mice (6), which are highly immunocompromised animals growing a partially human liver. Woodchucks as a small-animal model have also played an important role in understanding HDV viral infection and persistence in vivo. HDV pseudotyped with the envelope proteins from woodchuck hepatitis virus (WHV), a hepadnavirus related to HBV, can infect cultured woodchuck hepatocytes and lead to HDV persistence in woodchucks chronically infected with WHV (710). Although the woodchuck model has contributed substantially to our understanding of HDV persistence and genome stability, there is a dearth of available reagents to study in-depth the host immune response to persistent HDV infection in this model. An inbred mouse model—such as the one presented in our current study—holds the potential to overcome many of these caveats. The challenge is to systematically identify and overcome any restrictions to HBV and HDV growth in murine cells. Differences between the protein sequences of the sodium taurocholate cotransporting peptide NTCP (SLC10A1), the receptor for HBV and HDV (11, 12), in humans and nonpermissive species, such as rodents, pigs, and certain primate species, explain the block at the level of viral entry (13, 14) but not necessarily at other later stages in the viral life cycles. Expression of human NTCP (hNTCP) is sufficient to mediate HDV uptake and infection in mouse hepatocytes in vitro (15), but these cells remain resistant to HBV (15, 16). These observations were corroborated in mice transgenically expressing hNTCP or a humanized allele of NTCP that supports HDV. However, susceptibility was age-dependent and required inoculation with very high doses of HDV (17, 18).

Here, we created a mouse model in which hNTCP is expressed under the control of the human regulatory elements. Immunodeficient mice coexpressing both hNTCP and 1.3× HBV transgenes support persistent HDV infection for more than 80 days. In contrast, immunocompetent hNTCP/1.3× HBV transgenic (tg) mice became acutely viremic but eventually cleared the infection within 30 days. We demonstrate that although combination treatment with myrcludex B (MyrB) and lonafarnib (LNF)—two drug candidates for CHD—can efficiently suppress viremia, each drug alone or together failed to cure HDV infection at the doses tested. Our model is amenable to genetic manipulations, robust, and high in throughput and thus lends itself for studying chronic hepatitis in vivo and systematically testing novel therapies targeting HDV.


Transgenic expression of hNTCP facilitates HBV uptake into hepatocytes in vivo.

To further probe the HBV and HDV life cycles in an experimentally tractable animal model, we generated mice transgenically expressing a bacterial artificial chromosome (BAC) containing part of human chromosome 14, which includes the SLC10A1 gene encoding NTCP and upstream human regulatory elements (henceforth hNTCP/BAC; fig. S1). Quantitative reverse transcription polymerase chain reaction (RT-qPCR) analysis demonstrated that hNTCP is expressed in the liver of mice but not in other tissues such as the kidney, brain, lung, large or small intestines (Fig. 1A and fig. S2, A to D), thus mirroring the native tissue expression pattern of NTCP.

Fig. 1 hNTCP/BAC-NRG mice facilitate uptake of HBVcc and HDVcc.

(A) hNTCP expression as compared to housekeeping gene HPRT1 in hNTCP-BAC C57BL/6 (n = 4), hNTCP/BAC-NRG (n = 12) with C57BL/6 (n = 2), and NRG (n = 2) mice using relative RT-qPCR and comparing ΔΔCt. Mice were challenged with infectious 5-ethynyl-2′-deoxycytidine (EdC)–labeled HBVcc. (B) Quantification of HBVcc-EdC entry into murine hepatocytes (percent cells positive for HBV DNA in an entire slide section) of hNTCP/BAC-NRG (n = 3) versus NRG (n = 3) wild-type (WT) animals. (C) Representative images of HBVcc-EdC entry in hNTCP/BAC-NRG (top) and NRG WT (bottom) hepatocytes. HBV DNA (red) and nuclei (blue) were observed (scale bars, 200 μm). (D) HBsAg quantification over 2 weeks for hNTCP/BAC-NRG (blue, n = 5) and NRG WT (red, n = 5) mice challenged with HBVcc. Quantitation of HDV RNA in serum (E), HDV genomic RNA in liver (F), HDV antigenomic in liver (G), in hNTCP/BAC-NRG (n = 4) versus WT NRG (n = 4) mice with or without expression of HBV envelope proteins. The red dashed lines separate the hNTCP-expressing (left) from the nonexpressing mice (right) in each of the panels. hNTCP/BAC/1.3× HBV HDD NRG mice were challenged with patient-derived HDV (HDVpat) virions (n = 6), HDVcc, or were noninfected. (H) Longitudinal HBsAg data, (I) longitudinal HDV RNA in the serum of animals, and (J) quantification of genomic HDV RNA in the liver. All data are represented as ±SEM. Statistical significance was as follows: ***P ≤ 0.001, ****P ≤ 0.0001, using an ordinary one-way analysis of variance (ANOVA) with a Bonferroni’s multiple comparisons test. AU, arbitrary units.

Mouse cells expressing hNTCP can support HDV infection but appear to be resistant to HBV infection (15, 16). The apparent block in HBV infection is presumably not due to a dominant negative restriction factor, as heterokaryons of HBV-susceptible human and HBV-resistant mouse hepatoma cells remain permissive to infection (19). HDV and HBV share the same viral envelope, but it remains unclear whether HBV can productively enter mouse hepatocytes expressing hNTCP in vivo given their otherwise distinct natures. To directly test whether tg mouse hepatocytes can support HBV uptake, hNTCP/BAC tg mice on the immunodeficient nonobese diabetic recombinase activating gene 1 knockout (Rag1−/−) interleukin 2 receptor gamma chain null (IL-2RγNULL) background (hNTCP/BAC-NRG), which thus lack functional B, T, and natural killer (NK) cells, were injected with EdC-labeled, cell culture–produced HBV (HBVcc-EdC) (Fig. 1, B and C). EdC labeling provides very sensitive means to visualize HBV DNA even while still contained within the entering viral nucleocapsids. Thus, quantification of viral uptake is uncoupled from the completion of postentry steps. Of note, EdC labeling did not compromise viral fitness as evidenced by the fact that HBsAg concentrations in cell culture supernatants, the frequency of HBV core antigen–positive (HBcAg+) cells, and the amount of pregenomic RNA (pgRNA) (fig. S3, A to C) were equivalent irrespective of whether EdC-labeled or nonlabeled HBVcc was used to infect hNTCP-expressing HepG2 cells (3B10) (20). Quantitative image analysis of liver tissue sections harvested at 18 hours after challenge of hNTCP/BAC-NRG mice revealed that HBV entered up to 25% of hepatocytes (Fig. 1B). HBV entry was dependent on the presence of hNTCP expression, as the EdC signal remained at background in non-tg NRG control animals. Consistent with previously published in vitro data (15), there was no further evidence for productive HBV infection as neither experimental group showed an increase in serum levels of HBsAg (Fig. 1D) or HBV DNA and pgRNA in liver tissue (fig. S4, A and B) at any time up to 14 days after infection with HBVcc.

Previous studies in other hNTCP-expressing mouse lines injected with very large viral inocula showed that small frequencies of hepatocytes stained positive for HDAg, but infection ceased within 10 to 14 days (17, 18). Because HDV assembly and spread are dependent on the presence of HBsAg, we hydrodynamically delivered (HDD) a 1.3× over length HBV genome to the livers of adult hNTCP/BAC-NRG mice. Consistent with previous reports (21), delivery of the 1.3× HBV genome led to sustained secretion of HBsAg (figs. S5, A to D, and S6). HDV RNA levels were 50- to 100-fold higher in the sera of hNTCP/BAC/1.3× HBV-NRG mice than in all control groups expressing one or none of the transgenes (Fig. 1E). Consistent with these observations, HDV genomic RNA (Fig. 1F) was about 10-fold higher in doubly tg than in hNTCP tg mice and about 500-fold higher than in non-tg mice. Likewise, antigenomic RNA was only detectable in the doubly tg mice (Fig. 1G), providing further evidence for persistent infection (22). Antigenomic RNA most likely forms in hNTCP tg mice but was too low to detect with the assays available, reinforcing the importance of the HBV envelope proteins in this model for HDV persistence.

During HDV’s replicative cycle, the adenosine in the amber stop codon in the antigenome of the HDAg is edited to an inosine by ADAR1 (23), causing a change to a tryptophan and expression of the long form of the HDAg (24). This transition from small to large HDAg results in a switch from replication to viral packaging and egress. To ascertain that this editing process takes place in this mouse model, we sequenced HDV RNA isolated from serum and livers of hNTCP/BAC/1.3× HBV-NRG mice. Previous studies showed that a mixture of primarily nonedited and, to a lesser extent, edited HDV RNA is detectable in both tissue compartments (24, 25). Consistent with these clinical data, we detected the expected mutation in the HDV genome from viral RNA isolated from the liver (fig. S7), indicating that both isoforms of the HDAg are present, allowing for replication, virion packaging, and egress.

There are eight genotypes of HDV, which vary in their genome sequence by 30 to 50% (26). It is also known that mutations can arise in the HDV genome over the course of a chronic infection both in the woodchuck model and in humans (25, 27). To demonstrate that the observed HDV viremia was not limited to cell culture–derived HDV (HDVcc), we monitored infection with HDVpat. hNTCP/BAC/1.3× HBV-NRG mice were challenged with either heparin column–purified HDVpat [1 × 108 genomic equivalents (GE) per animal, n = 6], HDVcc (1 × 109 GE per animal, n = 5), or phosphate-buffered saline–injected and HDV viremia–monitored over time. HBsAg levels, regardless of cohort, were stable over the 14 days of the experiment (Fig. 1H). Mice challenged with HDVpat or HDVcc became viremic with no significant difference between the two cohorts (Fig. 1I). After 14 days after HDV infection, mice were euthanized. A substantial amount of HDV RNA was isolated from the livers of both cohorts relative to the noninfected controls, demonstrating that both HDVpat- and HDVcc-challenged mice were productively infected with HDV (Fig. 1J).

HDV is taken up through an HBV glycoprotein–mediated mechanism in hNTCP/BAC tg mice

To demonstrate that HDV enters through an HBV glycoprotein–dependent process, we pretreated hNTCP/BAC/1.3× HBV-NRG mice with an HBV preS1-derived peptide shown to effectively block HBV and HDV uptake in preclinical models and clinical trials (Fig. 2A) (28, 29). HDV RNA levels were significantly decreased in both the serum (P < 0.0437) (Fig. 2B) and liver tissue (P < 0.0477) (Fig. 2C) of mice injected with the blocking peptide, but not with a mutant control peptide. Collectively, our data demonstrate that NTCP is a species tropism-determining factor for HBV and HDV uptake in mouse hepatocytes similar to a variety of primate species (13), but additional postentry steps restrict HBV, but not HDV, infection in mice.

Fig. 2 HDV infects hNTCP/BAC/1.3× HBV HDD NRG mice through native entry mechanisms.

(A) Schematic of time course for peptide inhibition assay. Normalized quantification of HDV RNA in hNTCP/BAC/1.3× HBV HDD NRG mice treated with mock (black, n = 5), preS1–fluorescein isothiocyanate peptide (blue, n = 5), or a noninhibitory control (Ctrl) peptide (red, n = 5) in serum (B) or liver (C). All data are represented as ±SEM. Statistical significance was as follows: *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001, using an ordinary one-way ANOVA with a Bonferroni’s multiple comparisons test.

hNTCP/BAC-NRG mice coexpressing a 1.3× HBV genome support persistent HDV infection.

We next aimed to determine whether HDV could achieve long-term persistence in our model. Thus, we infected hNTCP/BAC-NRG mice with HDV along with either a 1.3× HBV genome by HDD or injection with a recombinant adenovirus (rAdV) expressing only the large, middle, and small HBV envelope proteins (rAdV-HBVenv; Fig. 3A). HBsAg was detectable in the sera of both experimental cohorts (Fig. 3B) but was higher in the 1.3× HBV groups. In the case of the vector approach, expression was further validated through bioluminescence imaging of a luciferase reporter coexpressed in the rAdV-HBVenv (Fig. 3, C and D) and by HBsAg Western blot, which detected all three forms: small (S, 35% total), medium (M, 18%), and large (L, 47%) both glycosylated and nonglycosylated (fig. S8). Of note, this ratio of L-HBsAg to S-HBsAg provided by the AdV-HBV Env vector does not fully reflect the expression ratios that occur during native HBV infection. This could potentially explain the lower levels of HBsAg present in the hNTCP/BAC/AdV-HBVenv-NRG mice. Regardless of how the HBV envelope proteins were delivered, serum HDV RNA copies rose 100- to 500-fold over background and remained largely stable for 84 days when the experiment was terminated.

Fig. 3 Characterization of persistent HDV infection in hNTCP/BAC-NRG mice.

hNTCP/BAC/1.3× HBV HDD NRG (blue, n = 8) or hNTCP/BAC/AdV-HBV Env (red, n = 5) mice challenged with HDV. HDV RNA in serum (A); HBsAg quantification (B); firefly luciferase quantification in mice over time (C). (D) In vivo imaging system image of hNTCP/BAC/AdV-HBV Env mouse (left) compared to NRG WT mice (right). (E) HDV genomic RNA (red) and DAPI (4′,6-diamidino-2-phenylindole) (blue) visualization by PLAYR technique in HDV-challenged hNTCP/BAC/AdV-HBV Env (left) versus NRG WT (right). Scale bars, 200 nM. (F) Quantification of HDV genomic RNA in hNTCP/BAC/AdV-HBV Env versus NRG WT mice (percent cells HDV RNA positive/slide imaged). (G) HDV RNA quantification by RT-qPCR in the liver of hNTCP/BAC/1.3× HBV (blue), hNTCP/BAC/AdV-HBV Env (red), and NRG WT mice (black). (H) Highlighter plot of Rbz domain in hNTCP/BAC/1.3× HBV HDD mice at days 3 and 56 after infection (red, transversion mutations; light blue, transition mutations). (I) Highlighter plot of HDVAg sequence for hNTCP/BAC/1.3× HBV-NRG animals (orange, nonsynonymous mutations). All data are represented as ±SEM. Statistical significance was as follows: *P < 0.05, using an ordinary one-way ANOVA with a Bonferroni’s multiple comparisons test. Adeno, adenosine. bp, base pair.

To quantify the frequency of HDV-infected cells, we used a proximity ligation for viral RNA (PLAYR) method (see Materials and Methods for details) (30). Clusters of HDV genomic RNA–bearing cells were readily visible in hNTCP/BAC-NRG but not in NRG control mice infected with HDV (Fig. 3E). The somewhat low frequency (2 to 6%) (Fig. 3F) is likely attributable to the fact that only a fraction of murine hepatocytes express the HBV envelope. Prolonged HDV persistence was further corroborated by RT-qPCR, which showed an about 100-fold (hNTCP/BAC + AdV-HBVenv) to 1000-fold (hNTCP/BAC + 1.3× HBV HDD) higher HDV RNA copy number (Fig. 3G), and by HDAg Western blot analysis (fig. S9) of liver lysates from HDV-infected tg versus non-tg mice.

Next, we sought to determine whether HDV acquires adaptive mutations to augment replication in the murine cellular environment. We managed to derive full-length sequences from four of the eight persistently infected mice. In comparison to the inoculum, several mutations emerged throughout the genome (figs. S10 to S12, A and B, and S13), including the HDV ribozyme (Rbz; Fig. 3H) and the HDAg (Fig. 3I). Although residues near the HDV Rbz cleavage site remained unaltered, an A/G change at position 46 within the Rbz domain near the P4 stem loop structure could possibly affect the RNA secondary structure of the HDV antigenome. However, none of the mutations were consistently detected in all genomes, suggesting they do not contribute to a putative gain of function favoring HDV RNA replication in mice.

Single and combination therapies with MyrB and LNF efficiently suppress HDV viremia in vivo but do not cure HDV infection

It was previously demonstrated that acylated peptides derived from the large HBsAg (now developed as MyrB) can block HBV and HDV virus entry in cell lines (31), in primary hepatocytes (20), and in human liver chimeric mouse models (28). Recent clinical data showed that administration of MyrB to HBV/HDV coinfected patients resulted in statistical significant decreases in HDV RNA loads (29, 32), consistent with the dependence of HDV on HBsAg. As an alternative approach for HDV treatment, inhibitors interfering with the prenylation of HDAg, and thereby its interaction with HBsAg and the subsequent release of infectious particles (33), have been tested in vitro and in vivo. Encouragingly, treatment of HDV patients with one such inhibitor, LNF, resulted in a mean 1.54-log IU/ml decline in HDV RNA from baseline (34). Here, we aimed to test single and combination treatments with MyrB and LNF as therapeutics in our mouse model. We delivered a 1.3× HBV genome by HDD to hNTCP/BAC-NRG mice and infected them 4 days later with HDV (Fig. 4A). Previously, viral load reductions in the serum effects have been observed in HDV/HBV coinfected patients treated with 200 mg of LNF (34) and 2 to 20 mg of MyrB (29, 32), and thus doses were scaled proportionally to the animals’ body weights. We injected mice intraperitoneally with MyrB (67.6 μg/kg) and/or administered LNF (2.7 mg/kg) orally. Although HBsAg concentrations, as expected, were not affected by either or both treatments (Fig. 4B), HDV RNA levels in the serum dropped significantly (LNF, P = 0.0001; dual, P < 0.0001; MyrB, P < 0.0001) within 1 week of treatment and remained around the limit of quantification at 14 days after treatment (Fig. 4C). At 14 days after treatment, mice from each cohort were euthanized to assess HDV RNA copy numbers in the liver. Notably, HDV RNA levels in the liver were significantly reduced by ~3 log (MyrB, P = 0.0418) and ~4.5 log (dual, P = 0.0415) as compared to the carrier control group, whereas the ~1.5 log drop (LNF) was not significant (P = 0.0564) (Fig. 4D). To determine whether HDV was actually cleared in any of the treatment arms, we stopped treatment and continued monitoring HDV viremia. Upon cessation of treatment, HDV viremia rebounded in LNF, MyrB, and dually treated mice, reaching pretreatment viremia levels by 5 weeks (Fig. 4C). This rebound was corroborated by high HDV RNA levels in the livers of mice in each cohort as determined by RT-qPCR (Fig. 4E). Notably, HDV RNA copies remained largely unaffected in the serum and liver of mice treated with the carrier control. Collectively, these data establish proof of concept for the utility of our model for antiviral drug testing.

Fig. 4 Single and combination therapies with MyrB and LNF efficiently suppress HDV viremia in vivo.

(A) Schematic of drug treatment experimental time course. (B) Longitudinal HBsAg ELISA data. (C) Longitudinal analysis of HDV RNA in serum of mock carrier control, LNF, MyrB, and dually treated groups. (D) HDV RNA in liver at the treatment endpoint (18 days after HDV infection, 14 days after drug treatment) (n = 6 for each treatment condition). (E) HDV RNA in liver of HDV-challenged hNTCP/BAC-NRG mice that had drug treatment stopped (n = 4 for each drug condition). All data are represented as ±SEM. Statistical significance was as follows: *P < 0.05, using an ordinary one-way ANOVA with a Bonferroni’s multiple comparisons test.

HDV infects immunocompetent mice expressing hNTCP and a 1.3× HBV transgene

The pathology observed in HDV patients is attributable at least in part to the intrahepatic inflammatory milieu. To create a model that would possibly enable the analysis of delta immunopathology, we analyzed HDV infection in immunocompetent hNTCP/BAC mice on the C57BL/6 background also stably expressing a 1.3× HBV transgene (Fig. 5A) (35), thereby providing the HBsAg required for HDV propagation (Fig. 5B). Consistent with our data in immunocompromised mice, HDV infection was dependent on expression of both the hNTCP/BAC and 1.3× HBV tgs (Fig. 5, C and D). In the sera of doubly tg mice, HDV RNA copies were 10- to 50-fold higher than all control cohorts until day 14 after infection when viral loads decreased to levels comparable to those of the non-tg control group (Fig. 5C). In the liver tissue of both dually tg and hNTCP tg mice, HDV RNA remained high until day 14 after infection, after which it decreased. In doubly tg mice, HDV RNA levels remained above background until day 30 after infection (Fig. 5D). These data suggest that HDV infection is controlled, albeit not cleared, by the murine immune system during the acute phase of infection. The immune system may exert pressure on HDV, yielding viral variants with escape mutations in the HDAg—HDV’s only protein antigen. However, sequence comparison of HDV genomes from hNTCP/BAC/1.3× HBV mice on the NRG and C57BL/6 cohorts did not provide evidence for this (Fig. 5E and fig. S14).

Fig. 5 Characterization of HDV acute infection in immunocompetent hNTCP-BAC C57BL/6 mice.

(A) Schematic of hNTCP/BAC C57BL/6 and 1.3× HBV tg mice crossed to generate hNTCP/BAC/1.3× HBV tg C57BL/6 mice. (B) HBsAg quantification of hNTCP/BAC/1.3× HBV tg versus HBV tg mice over a month’s time. NEG indicates the threshold for the assay above which HBsAg levels are considered positive. Normalized HDV RNA in serum (C) and liver (D) of hNTCP/BAC/1.3× HBV tg, hNTCP/BAC tg, and HBV tg animals. (E) Highlighter plot analysis of HDVAg sequences in immunocompetent hNTCP/BAC/1.3× HBV tg C57BL/6 mice (day 14) versus hNTCP/BAC/1.3× HBV-NRG mice (day 56) (red, site of mutation). For each time point, hNTCP only (n = 4), hNTCP/BAC/1.3× HBV tg (n = 4), and HBV tg (n = 4) mice were euthanized. All data are represented as ±SEM.

HDV induces innate and adaptive immune responses in immunocompetent hNTCP/BAC/1.3× doubly tg mice

We next aimed to determine whether the murine immune system becomes activated during HDV infection and results in liver damage in our model. Alanine aminotransferase (ALT) levels were in the normal range in all mice regardless of genetic background (Fig. 6A). This is consistent with our histological analysis, where only small changes, but no major histopathology, were observed (Fig. 6B). In accordance with previously published data in human liver chimeric mice showing that HDV coinfection induces markedly greater innate immune responses in comparison to HBV monoinfection (36), we also observed changes in the expression of multiple interferon-stimulated genes (ISGs), including PKR, OAS-L, IP-10, and Mx1 in hNTCP/1.3× HBV double as compared to 1.3× HBV single tg mice (Fig. 6, C and D). These observations are also consistent with the mitochondrial antiviral signaling protein (MAVS)–dependent induction of interferon observed in mice in which the HDV and HBV genomes were expressed by an adeno-associated virus vector system (37). Specifically in the hNTCP/1.3× HBV double tg group that supported viremia, we also observed early postinfection increased frequencies of NK (Fig. 6E), NK T cells (Fig. 6F), and MAIT cells (Fig. 6G). These immune cell subsets have been implicated in clearance of cells infected with hepatotropic pathogens (38), and, altogether, our data suggest that the innate and cellular immune response may antagonize establishment of persistent HDV infection in fully immunocompetent mice.

Fig. 6 Characterization of histopathology in HDV-challenged hNTCP/BAC/1.3× HBV tg animals.

(A) ALT concentrations for hNTCP/BAC/1.3× HBV tg, hNTCP-BAC tg, HBV tg, and WT animals. (B) Histopathological analysis of the liver from hNTCP/BAC/1.3× HBV tg and HBV tg animals. Scale bars, 400 μm. (C) Normalized fold change of ISGs MX1, IP-10, OASL2, and PKR (RNA-activated protein kinase R) in the livers of hNTCP/BAC/1.3× HBV tg animals compared to HBV tg. (D) Cytokine analysis in sera of HDV-challenged hNTCP/BAC/1.3× HBV and HBV tg C57BL/6 animals at days 0, 3, and 14 after infection. Cellular immune response in the spleens of HDV-challenged hNTCP/BAC/1.3× HBV tg, hNTCP tg, HBV tg, and WT mice: NK cells (E), NK T cells (F), and mucosal-associated invariant T cells (MAIT) (G). Frequencies of CD45+ lymphocytes are indicated. Each data point is the average of four different animals. All data are represented as ±SEM. Statistical significance was as follows: **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, using an ordinary one-way ANOVA with a Bonferroni’s multiple comparisons test. GM-CSF, granulocyte-macrophage colony-stimulating factor.


This study represents a step forward in developing a robust animal model for HDV infection and immunity. Here, we demonstrate that the entire HDV life cycle can be recapitulated in an inbred mouse model with inheritable susceptibility to HDV. Both HBV and HDV can enter hNTCP-expressing mouse hepatocytes in vivo after an HBsAg-dependent and, thus, native entry mechanism. Although a postentry block prevents the completion of the HBV life cycle, HDV replication, particle assembly, and egress are supported in the presence of the HBsAg. Both genomic and antigenomic viral RNA is detectable in the livers of HDV-infected mice. Sequence analysis revealed that both edited and nonedited forms of the HDAg are present in the livers of persistently infected animals, demonstrating that virions are assembled through the native mechanism. These RNA editing data are consistent with prior observations in HDV monoinfected hNTCP tg mice (17, 18). In the serum, we were only able to detect nonedited genomic RNA molecules, which is conceivably due to the combination of the overall lower viral titers in our mice as compared to patients, and the limitations in the sensitivity of the sequencing method. It should also be noted that the frequency at which the editing event occurs ranges and has been estimated at 15 to 30% in cell culture experiments (24). Thus, edited RNA represents only a subfraction of the total HDV RNA, which may further impair detection in this model.

Previously developed human liver chimeric mice can also support HDV and HBV coinfections (6). However, such xenotransplantation models are low-throughput, expensive to generate, and hampered by intra- and interexperimental variability as well as donor-to-donor variability. Another recent model relies on vector-mediated overexpression of the HDV and HBV genomes, which arguably does not adequately mimic the inflammatory situation induced during infection (37). The inbred mouse model presented here holds the potential to overcome many of these caveats, as it is amenable to genetic manipulations and can be used for preclinical assessment of the efficacy of entry inhibitors and other putative therapeutics. We demonstrate that single and combined administration of MyrB and LNF, two therapeutics with distinct mechanisms of action, result in HDV RNA suppression in both the serum and liver of these mice. Notably, our data suggest that coadministration of MyrB and LNF may have a synergistic effect, as evidenced by an about 1000-fold decrease in HDV RNA in the liver. However, even the combination of MyrB and LNF does not seem adequate to completely abrogate HDV infection. The serum data for the monotherapies are largely in line with recently reported data from clinical trials (29, 32, 34) and thereby show the utility of our model for assessing the preclinical efficacy of novel therapeutics in a convenient animal model. In clinical trials, LNF and MyrB were administered in fixed doses at twice 200 mg or once 2 to 20 mg of LNF (34) and MyrB (29, 32), respectively. It is important to note that the doses used in our study do not perfectly match those of published clinical studies, corresponding to half the optimal monotherapy dose of LNF, and twice the dose of MyrB used in clinical trials. Thus, a direct comparison of their relative efficacy or respective contributions to the apparent observed synergy in this model may not be not possible. Irrespectively, our model now creates the possibility of assessing combination therapies to eliminate HDV in longer-term follow-up studies and to evaluate their effects on viremia not only in serum but also in liver tissue.

We provide evidence that HDV induces both innate and adaptive immune responses, which collectively contribute to the control of HDV viremia in vivo. These results are distinct from the clinical course of HDV infections in patients in which the virus usually progresses to chronicity. Thus, the murine immune system may be effective at clearing HDV. Mild liver pathology was virally induced and presumably immune-mediated, and even this overall moderate phenotype was remarkable considering the animals are tolerized to the HBV transgene. Immune-mediated contribution to liver damage can thus be largely attributed to the HDV infection. It may remain challenging to establish a bona fide HBV/HDV co- or superinfection given the yet to be resolved blocks in the HBV life cycle. However, future efforts aimed at modeling a phenotype even more similar to that observed in patients could address whether liver disease is more exacerbated in HDV-infected hNTCP/BAC/1.3× HBV tg mice in which immunologic tolerance to HBV has been broken.


Study design

This study was designed to characterize HDV infection in a genetically humanized tg mouse model that expresses the HBV/HDV receptor, hNTCP, and the HBV envelope proteins. In addition, we assessed immune responses and preclinical antiviral combination therapy against HDV in these mice. All human samples and animal work were approved by the Institutional Animal Care and Use Committee (IACUC) and Institutional Review Board (IRB) of Princeton University (IACUC # 1930-16, deemed nonhuman subjects research by IRB). For in vitro studies, three to six biological replicates were performed; in in vivo experiments, four to eight mice were used per time point for longitudinal assays. Experiments were not performed in a blinded fashion.


hNTCP-BAC tg mice on the C57BL/6 and NRG backgrounds were generated as described in the Supplementary Materials and Methods. F. Chisari (The Scripps Research Institute) provided the 1.3× HBV tg mice were, and NRG and C57BL/6J mice were obtained from The Jackson Laboratory. All in vivo experiments were in accordance with protocols reviewed and approved by the IACUC of Princeton University, respectively.

Generation of HBV and HDV stocks

To generate HBV stocks, HepG2.2.15 cells (39) were grown in media containing tetracycline until they reached a confluency of 100%. At this time, media were changed to DMEM (Dulbecco’s modified Eagle’s medium) F12 media supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin. To produce HDV stocks, the plasmid pSVL(D3) (a gift from J. Taylor; Addgene plasmid # 29335) (40) containing three copies of the HDV genome was transfected into a HepG2 cell line that constitutively expresses the large, medium, and small HBV envelope proteins. Virus-containing supernatants were concentrated, and the virus was run over a HiTrap heparin column (GE Healthcare Life Sciences) to purify infectious virus particles from noninfectious subviral particles. After dialysis, virus was aliquoted into cryovial tubes and cryopreserved at −80°C until use.

Mouse infections

All mouse infections with HBV, HDV, or rAdV were done by intravenous injection into the tail vein with 1 × 108 GE of HBV, 1 × 109 GE of HDV, or a multiplicity of infection of 1 × 1010 for rAdV in a total volume of 200 μl per mouse. Mice were bled submandibularly at the indicated time points (0, 3, 7, 14, 21, and 30 days after infection), depending on the experiment.

Statistical analysis

All statistical analysis was performed with GraphPad Prism 6h software. Data are presented as means ± SEM. Multiple group comparisons were analyzed by one-way analysis of variance (ANOVA) with a Bonferroni’s multiple comparisons test. P values of <0.05 were taken as being statistically significant. Additional information can be found in the Supplemental Materials and Methods section.


Materials and Methods

Fig. S1. Schematic of the modified BAC containing the SLC10A1 gene used for generating the hNTCP tg mice on the NRG and C57BL/6 backgrounds.

Fig. S2. hNTCP expression in liver and other organs of hNTCP/BAC-NRG mice on the C57BL/6 and NRG backgrounds.

Fig. S3. EdC labeling of the HBV genome does not affect viral fitness.

Fig. S4. Quantification of HBV replication intermediates in HBV-infected hNTCP/BAC-NRG mice.

Fig. S5. Delivery of 1.3× HBV plasmid leads to sustained HBsAg secretion over time.

Fig. S6. HBsAg levels remain stable in HDV-infected hNTCP/BAC/1.3x HBV HDD-NRG mice over the experimental time course.

Fig. S7. Evidence for RNA editing in HDV genomes isolated from livers of HDV-infected hNTCP/BAC/1.3× HBV-NRG mice.

Fig. S8. Adenoviral delivery of HBV envelope proteins leads to expression of all three forms of HBsAg large (L), medium (M), and small (S).

Fig. S9. Assessment of HDVAg in HDV-challenged hNTCP/BAC-NRG mice.

Fig. S10. Longitudinal sequence analysis of hypervariable domain in hNTCP/BAC/1.3× HBV HDD NRG mice.

Fig. S11. Longitudinal sequence analysis of Rbz domain in hNTCP/BAC/1.3× HBV HDD NRG mice.

Fig. S12. Predicted effect of the A/G mutation at position 46 of HDV Rbz on RNA secondary structure.

Fig. S13. Longitudinal sequence analysis of HDVAg domain in hNTCP/BAC/1.3× HBV HDD NRG mice.

Fig. S14. Longitudinal sequence analysis of HDVAg domain in immunocompetent hNTCP/BAC/1.3× HBV C57BL/6 mice.

Table S1. Primary data.


Acknowledgments: HepG2.2.15 cells were provided by C. Seeger [Fox Chase Cancer Center (FCCC)] and the 1.3× HBV tg mice by F. Chisari (The Scripps Research Institute). The pSVL(D3) and 1.3× HBV plasmids were gifts from J. Taylor (FCCC) and Y. Shaul (Weizmann Institute), respectively. S. Urban (University of Heidelberg) provided the MyrB compound. We thank T. Muir and F. Wojcik (both Princeton University) for help with the peptide synthesis. We thank C. DeCoste and the Molecular Biology Flow Cytometry Resource Facility, G. Laevsky and the Nikon Center of Excellence for the generous assistance with imaging, and the staff of the Molecular Biology and Microinjection Cores at The Jackson Laboratory for outstanding technical support. We are grateful to J. Gaska and members of the Ploss laboratory for critical discussions and edits of this manuscript. Funding: This study is supported by grants from the NIH (R01 AI079031, R01 AI107301, and R21AI117213 to A.P.), a Research Scholar Award from the American Cancer Society (RSG-15-048-01-MPC to A.P.), a Burroughs Wellcome Fund Award for Investigators in Pathogenesis (to A.P.), and a Graduate fellowship from the Health Grand Challenge from the Global Health Fund of Princeton University (to B.Y.W.). The Princeton Molecular Biology Flow Cytometry Resource Facility is partially supported by the Cancer Institute of New Jersey Cancer Center support grant (P30CA072720). The New York University Experimental Pathology Immunohistochemistry Core Laboratory is supported in part by the Laura and Isaac Perlmutter Cancer Center support grant NIH/NCI P30CA016087 and the NIH S10 grants NIH/ORIP S10OD01058 and S10OD018338. B.Y.W. is a recipient of an F31 NIH/National Research Service Award Ruth L. Kirschstein Predoctoral award from the National Institute of Allergy and Infectious Diseases and a graduate fellowship from the New Jersey Commission on Cancer Research. J.S. and E.S.-D. are both recipients of postdoctoral fellowships from the German Research Foundation. M.V.W. was funded by The Jackson Laboratory. Author contributions: B.Y.W. and A.P. conceived the study, designed and performed experiments, analyzed the data, and wrote the manuscript. E.S.-D., Y.B., J.S., B.E.L., H.J., J.E.T., M.V.W., and R.E.S. performed experiments and analyzed the data. T.H., G.H., B.H., Y.S., S.G., M.-A.P., A.S.F., and L.C. performed experiments. J.C. performed the histopathological analysis. R.G.N. and K.S. provided patient serum and technical support. K.G. and M.L. provided valuable reagents. Competing interests: The authors declare that they have no competing interests.

Stay Connected to Science Translational Medicine

Navigate This Article