Strain-specific responsiveness of hepatitis D virus to interferon-alpha treatment

Background & Aims Pegylated interferon alpha (pegIFNα) is commonly used for the treatment of people infected with HDV. However, its mode of action in HDV-infected cells remains elusive and only a minority of people respond to pegIFNα therapy. Herein, we aimed to assess the responsiveness of three different cloned HDV strains to pegIFNα. We used a previously cloned HDV genotype 1 strain (dubbed HDV-1a) that appeared insensitive to interferon-α in vitro, a new HDV strain (HDV-1p) we isolated from an individual achieving later sustained response to IFNα therapy, and one phylogenetically distant genotype 3 strain (HDV-3). Methods PegIFNα was administered to human liver chimeric mice infected with HBV and the different HDV strains or to HBV/HDV infected human hepatocytes isolated from chimeric mice. Virological parameters and host responses were analysed by qPCR, sequencing, immunoblotting, RNA in situ hybridisation and immunofluorescence staining. Results PegIFNα treatment efficiently reduced HDV RNA viraemia (∼2-log) and intrahepatic HDV markers both in mice infected with HBV/HDV-1p and HBV/HDV-3. In contrast, HDV parameters remained unaffected by pegIFNα treatment both in mice (up to 9 weeks) and in isolated cells infected with HBV/HDV-1a. Notably, HBV viraemia was efficiently lowered (∼2-log) and human interferon-stimulated genes similarly induced in all three HBV/HDV-infected mouse groups receiving pegIFNα. Genome sequencing revealed highly conserved ribozyme and L-hepatitis D antigen post-translational modification sites among all three isolates. Conclusions Our comparative study indicates the ability of pegIFNα to lower HDV loads in stably infected human hepatocytes in vivo and the existence of complex virus-specific determinants of IFNα responsiveness. Impact and implications Understanding factors counteracting HDV infections is paramount to develop curative therapies. We compared the responsiveness of three different cloned HDV strains to pegylated interferon alpha in chronically infected mice. The different responsiveness of these HDV isolates to treatment highlights a previously underestimated heterogeneity among HDV strains.


Introduction
The hepatitis delta virus (HDV) infects around 20 million people worldwide 1 and recent reports suggested that the number of HDV-positive individuals may be even higher. [1][2][3] Liver disease associated with chronic hepatitis D (CHD) causes substantial global morbidity (cirrhosis, hepatocellular carcinoma) and mortality. 4 HDV RNA replication takes place in the nucleus of hepatocytes by hijacking the host RNA polymerase II, which amplifies the genomic HDV RNA through a double rolling-circle amplification process. 5 During HDV replication, two additional viral RNAs accumulate in the hepatocytes: the antigenomic RNA, which is an exact complement of the genomic RNA and the smaller linear mRNA encoding for the hepatitis delta antigen (HDAg). HDAg binds specifically to the HDV RNA and exists in two different forms: the small HDAg (S-HDAg) that is important for virus replication and the large variant (L-HDAg), which can inhibit replication and promotes virus assembly through a prenylation site. 5 The L-HDAg is generated by post-transcriptional RNA editing at the adenosine 1012 (amber/W site), which is mediated by the RNA-specific adenosine deaminase (ADAR). The balance between viral replication and assembly is orchestrated by the presence of S-and L-HDAg and by post-translational modifications of these proteins, such as prenylation, phosphorylation, methylation, and SUMOylation. 6 HDV is a satellite virus that requires expression of hepatitis B virus (HBV) envelope proteins to release infectious HDV particles. 7 Both HBV and HDV infect the human hepatocytes via the sodium taurocholate cotransporting polypeptide (NTCP). 8 Moreover, and in strong contrast to HBV, HDV can also disseminate through cell division. 9,10 Active HDV infection either can occur upon simultaneous co-infection with HBV or as a super-infection in people already infected with HBV. HDV is classified into eight genotypes. Although HDV-1 is the most prevalent genotype worldwide, HDV-3 is frequently associated with the most severe hepatitis. 11,12 Sequence divergence among the genotypes is as high as 40% over the entire RNA genome with the greatest difference observed between HDV genotype 1 (HDV-1) and genotype 3 (HDV-3). 13 Because of its compact genomic organisation and lack of its own polymerase, HDV offers very few therapeutic targets and current HBV therapies based on the use of nucleos(t)ide analogues (NUCs) inhibiting the HBV polymerase cannot directly target HDV. Understanding the factors that are able to block HDV infection and replication is therefore of utmost importance for the development of HDV curative therapies. In 2020, the HBV/ HDV entry inhibitor bulevirtide (Hepcludex/Myrcludex-B) obtained conditional marketing authorisation by the EMA as the first HDV-specific drug, 14 and pegylated interferon lambda and the prenylation inhibitor lonafarnib are currently being tested in clinical trials. 15 Aside such advances, pegylated interferon-alpha (pegIFNa) has been the most commonly used off-label treatment against HDV for decades. However, treatment is associated with substantial side effects and leads to sustained virological response (defined as undetectable serum HDV RNA 6 months after treatment) in only about 25-30% of patients, with high relapse rates after treatment cessation. Nevertheless, various clinical trials are currently evaluating the contribution of IFN treatment to HDV therapy in novel combination regimens. 16 Apart from acting as an immunomodulatory agent, IFNa induces interferon stimulated genes (ISGs) via the janus kinasesignal transducer and activator of transcription (JAK-STAT) signalling pathway also in hepatocytes, thereby triggering a cellular antiviral state. 17 Nonetheless, the mode of action (MoA) of IFNa in stably HDV-infected primary hepatocytes remains elusive.
HDV is sensed by the pattern recognition receptor melanoma differentiation antigen 5 (MDA5) of the hepatocytes 18 leading to ISG enhancement and chemokine production in HBV/HDV infected cells. 19,20 However, such antiviral state does not limit HDV replication, whereas therapeutically applied more stable (pegylated) IFNs could affect a patient-derived HDV inoculum in vivo. 17 Moreover, IFNa was shown to promote silencing of the HBV genome, thus lowering HBV transcript levels 21 and to destabilise HDV RNA during cell division. 22 To date, only a limited amount of HDV strains have been cloned [23][24][25][26][27] Most studies used a peculiar HDV-1 clone of uncertain human origin 28,29 here dubbed HDV-1a (Table 1 and methods), which turned out to be unaffected by IFNa in vitro, 18,30,31 leading to the assumption that interferon only marginally impairs HDV RNA replication in stably infected cells. 22 In this study, we assessed and compared the antiviral and intrinsic host response to pegIFNa treatment in vivo using human liver chimeric mice stably infected with HBV and either with HDV-1a or two different human-derived cloned HDV strains: a novel genotype 1 (HDV-1p) and an HDV-3 (Table 1). 32

Virus generation
The patient-derived HDV-1 isolate (HDV-1p) was isolated from a male individual with CHD from the university clinic of Hamburg, 33 passaged in human liver chimeric uPA/SCID/beige/IL2RG -/-(USG) mice, sequenced, cloned as genome-sense tandem dimer in pcDNA3.1(+), and infectious particles were produced in cell culture (Table 1). The HDV-3 isolate (Peru-1) was obtained from a young man from Peru, who developed severe acute hepatitis, which was cloned by Casey et al. 32 The origin of the first HDV clone available in the research community is less clear: it is a genotype 1 strain, individual(s) sera were passaged through various chimpanzees at NIH (J. Taylor, pers. commun.), inoculated in a woodchuck and then cloned (Table 1). For the production of the HDV-1a and HDV-3 strains, the HDV recombinant plasmid pSVL(D3) (kindly provided by J. Taylor, Philadelphia, PA, USA) 28 and pCMV3-Peru-1.2 (kindly provided by J. Casey, Washington DC, USA) 34 Research article were used. Infectious HDV-1a, HDV-1p, and HDV-3 particles were generated in HuH7 cells as previously described. 35 In brief, cells were transfected with equimolar amounts of HDV-1a, HDV-1p, or HDV-3 recombinant plasmids and the HBV envelope-expressing vectors pcDNA3.1(+) 36 (HDV-1p) or pT7HB2.7 37 (HDV-1a, HDV-3) encoding the surface proteins of HBV genotype D using Fugene HD Transfection Reagent (Promega, Madison, WI, USA) ( Table 1).

Treatment
In vivo, pegIFNa treatment was started when HBV/HDV-1a-, HBV/ HDV-1p-, or HBV/HDV-3-infected mice reached stable HBV and HDV viraemia levels. Mice received pegIFNa (Pegasys; Roche, Basel, Switzerland) twice a week subcutaneously (25 ng/g body weight) 17,21 for 4 or 9 weeks. Mice were sacrificed 24 h after the last pegIFNa injection. Cultured primary human hepatocytes (PHHs) (isolated from an HBV/HDV-infected mouse) received IFNa (1,000 IU/ml; Roferon-A; Roche, Basel, Switzerland) starting 1 day after plating. The culture medium was changed twice a week. More methods (e.g. generation and infection of mice, virological measurements, RNA in situ hybridisation) can be found in the supplementary information.

Results
Production and infectivity of HDV-1p in vitro and in vivo As summarised in Table 1, the HDV-1p strain was obtained from a person with CHD receiving nucleos(t)ide analogues (NUCs) treatment (lamivudine plus adefovir), which resulted in undetectable serum HBV DNA levels (Fig. 1A). HDV viraemia was detectable and immunohistology confirmed an abundance of HDAg-positive hepatocytes ( Fig. 1A and C). Eventually, after discontinuation of NUC treatment, the individual received pegIFNa for 48 weeks, which resulted in a sustained HDV response and temporary increase of HBV viraemia, 33 but led to HBV and HBsAg loss and seroconversion to hepatitis B surface antibody years later (Fig. 1B).
The HDV-positive isolate obtained before IFN treatment was first shown to be infectious in HBV-infected human liver chimeric USG mice 17 and now sequenced and cloned as described in the Participants and methods section ( Table 1). The full genome sequence of the HDV-1p strain is available at NCBI (accession number: OL825606). HDV-1p virus stocks were produced in HuH7 cells and their infectivity was tested first in vitro using HepG2 hNTCP cells. Seven days after HDV-1p inoculation (multiplicity of infection; MOI = 1), HDAg staining ( Fig. 2A) confirmed in vitro infectivity of the HDV-1p particles produced after cloning the virus. The in vitro infectivity of HDV-1p appeared similar to the cloned strains HDV-3 (MOI = 1) and HDV-1a (MOI = 2) ( Fig. 2A).
To assess the infectivity of the cloned HDV-1p strain in vivo, HBV-infected humanised USG mice were super-infected with HDV-1p from the supernatant of HuH7 cells (4 × 10 6 GE/mouse). In these HBV infected mice, where HDV infection is first established in a low number of PHHs, HDV disseminated among PHHs and viraemia increased up to week 6 after super-infection and remained stable until the end of the observation time (9 weeks post super-infection) (Fig. 2B). In line with previous studies 39, 40 we observed a decrease of HBV viraemia (fivefold) during HDV-1p super-infection (data not shown), while HDAg was clearly detected in human hepatocytes (Fig. 2C). Moreover, amounts and distribution of S-and L-HDAg resembled those detected in the patient liver biopsy (Fig. 2D). Likewise, HDV-3 and HDV-1a reached stable HDV titres after around 5 weeks of HDV super-infection and showed similar ratios of S-and L-HDAg ( Fig. 2B and D). The amount of HDV-infected human hepatocytes determined by immunofluorescence tended to be higher in mice infected with HDV-3 (63%) compared with those infected with HDV-1a (32%) or HDV-1p (42%), suggesting that HDV-3 may display superior spreading capacities in vivo than the HDV-1 strains. Overall, these comparative analyses show that human liver chimeric mice are well suited to study patient-derived viral strains.
PegIFNa treatment in mice infected with different HDV strains Human liver chimeric mice stably infected with HBV and one of the three HDV clones received pegIFNa twice a week (Fig. 3A). In LLoD U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α LLoD U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d U n in f P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d U n in f P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d U n in f P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d U n in f P e g I F N α U n t r e a t e d P e g I F N α U n t r e a t e d P e g I F N α  (Fig. 4A). This confirms that both HDV-1p and HDV-3 isolates are highly responsive to IFN treatment in vivo in mice stably infected with both HBV and HDV. In mice infected with the IFN-resistant HDV-1a strain, the amount of HDAg-positive PHHs appeared comparable between untreated controls (32%) and mice that received pegIFNa for 4 (33%) or 9 weeks (31%) (Fig. 4A). RNA in situ hybridisation showed a clear reduction of genomic HDV RNA in treated HDV-1p and HDV-3 infected mice, while the amount of genomic HDV RNA positive PHHs remained similar in untreated and treated HDV-1a infected mouse livers (Fig. S2). Moreover, genomic and antigenomic HDV RNA levels determined by a strain-specific qPCR assay using biotinylated magnetic beads 17 revealed that pegIFNa reduced both HDV RNA forms in HDV-1p-and HDV-3-infected mice, but mice harbouring HDV-1a remained unaffected (Fig. 4B). Accordingly, levels of S-and L-HDAg decreased in HDV-1p and HDV-3 infected mouse livers receiving pegIFNa administration but remained comparable in untreated and treated HDV-1a infected mice (Fig. 4C).
PegIFNa-mediated human ISG induction in HBV/HDV-1a, HBV/ HDV-1p, and HBV/HDV-3-infected mice Since mice infected with HBV/HDV-1p or HBV/HDV-3 responded to IFNa treatment and mice infected with HBV and the HDV-1a strain were resistant to therapy, we investigated whether these differences in IFN responsiveness could be explained by the different ability of these viruses to induce intrinsic innate responses in infected hepatocytes. In line with a previous study, 19 human ISGs (e.g. hISG15, hMxA, hOAS1, hSTAT1), pattern recognition receptors (hMDA5), and chemokines (e.g. hCXCL10) were similarly and strongly upregulated (between 3-and 88-fold) upon HDV infection, regardless of whether the mice were infected with HBV/HDV-1a, HBV/HDV-1p, or HBV/HDV-3 ( Fig. 6A and B, Table 2).
PegIFNa treatment induced a further enhancement (between 2and 29-fold) of human ISGs compared with untreated, infected mice ( Fig. 6A and B, Table 2). Interestingly, despite the different HDV treatment outcomes, expression levels of analysed genes appeared comparable among treated animals, suggesting that virological differences, rather than the different enhancement of innate host responses, might be responsible for the antiviral effect of IFN in vivo. Furthermore, pegIFNa had no further effect on the expression of human cytokines (hIL28AB, human transforming growth factor-b [hTGF-b]) and on the apoptosis marker human caspase 8, but led to a substantial decrease (3-to 4-fold) of hNTCP expression levels (Fig. 6B, Table 2). Of note, hIL28AB and hTGF-b baseline expression levels are close to the detection limit in PHHs, therefore slight expression changes among groups must be interpreted cautiously.
Remarkably, simultaneous visualisation of HDV RNA and human ISGs by RNA ISH revealed that MxA-positive human hepatocytes were still expressing high levels of antigenomic HDV-1a RNA. These results demonstrate that the IFN-resistant HDV-1a isolate does not hamper the IFN-mediated induction of classical human ISGs at the single cell level (Fig. 6C).
Sequencing of the three distinct HDV genome strains from infected mice Genome sequencing of intrahepatic HDV RNA in mice infected with HDV-1a, HDV-1p, or HDV-3 revealed that no mutations emerged after 4 or 9 weeks of pegIFNa treatment in vivo (data not shown). In addition, the occurrence of genomes encoding for the small (ACC) or large HDAg (ATC) (RNA editing at the amber/ W site) remained similar in treated and untreated mice and was comparable between the different HDV isolates (Fig. S3A). The ribozyme site showed 100% identity between the two HDV-1 isolates and 89.6% identity between HDV-1 and HDV-3 (Fig. S3B). The open reading frame for the large HDAg (214 amino acids) showed 89.5% identity between HDV-1p and HDV-1a and 71.2% identity between HDV-1p and HDV-3, which translated into several amino acid changes ( Fig. S3B and C). However, the amber/W site (aa196), the prenylation site (aa212), and other known post-translational modification sites remained fully conserved between the three different isolates (Fig. S3C). In line with Le Gal et al., 41 the two HDV-1 strains showed a proline residue at the nuclear export site at position 205, whereas HDV-3 harbours a glycine, suggesting different virion secretion efficiencies across genotypes. Two unique differences that exclusively occurred in the IFN-resistant HDV-1a-isolate were detected in the coiled coil domain at position aa41 (leucine instead of isoleucine) and aa44 (isoleucine instead of leucine) (Fig. S3C). Although these are conservative mutations, we cannot exclude their impact on SUMOylation rates at position 42.

Discussion
In CHD, PegIFNa is commonly used as an off-label treatment, although its mode of action in HDV-infected hepatocytes is still unclear and responsiveness to IFN treatment remains limited. 16 Understanding HDV diversity and the mechanisms determining different antiviral responsiveness to treatment is paramount for improving therapeutic options. Experimental studies with the aim to unravel IFN MoA in HDV infection were hampered not only by the paucity of infection models available, but also by the limited availability of well-characterised HDV isolates. By chance, the first HDV-1 clone that was generated in 1988 (Kuo et al., 29 herein HDV-1a), appeared IFN-resistant 28,29 both in transfected hepatoma cells and infected PHHs. 18,30,31 HDV was reduced only when IFNa was administered around HDV infection time, 18,30 indicating that IFNa mainly limited new infection events. These in vitro studies led to the general assumption that HDV is resistant to IFN treatment when infection is already established. However, we observed that human liver chimeric mice infected with an HDV-positive patient-derived serum responded to pegIFNa. 17 Interestingly, this HDV isolate (HDV-1p) was obtained from an individual who later achieved sustained HDV response upon pegIFNa treatment ( Fig. 1 and Bockmann et al. 33 ). To assess the IFN responsiveness of the most commonly used HDV-1a clone in vivo, in HBV/HDV infected human hepatocytes, and to compare these data with distinct HDV strains, we cloned this new HDV-1p isolate and also used an additional, genetically distant clone from an HDV-3 strain. 32 In stable HBV/HDV-1p-and HBV/HDV-3-infected humanised mice, pegIFNa treatment clearly reduced HDV markers in serum and liver, including the levels of genomic and antigenomic HDV RNA, as well as S-and L-HDAg. In striking contrast, we did not detect any antiviral effect on HDV parameters when stable HBV/ HDV-1a infected mice received pegIFNa, although all HBV markers, including HBsAg, were clearly reduced. Of note, the IFN dose commonly applied for studies in humanised mice is rather high (adapted to mouse metabolism) and, most importantly, it is given twice per week. Nevertheless, even extending treatment to 9 weeks did not alter HDV RNA levels or the number of HDV-positive human hepatocytes (as determined by immunofluorescence and RNA in situ hybridisation) when the HDV-1a strain was used.
Notably, HBsAg is needed for the release of HDV particles and the IFN-mediated reduction of HBV serological markers, particularly those linked to cccDNA activity, was shown to be associated with virological responses in people with CHD. 42 Accordingly, IFNmediated suppression of HBsAg production should also lower HDV RNA in serum. However, HBV infection and HBsAg levels are generally high in humanised mice and despite the decrease induced by IFN, HBsAg changes were not sufficient to induce substantial decrease of HDV-1a in IFN-treated mice. Our results indicate that also HDV viraemia changes need to be monitored to predict virological responses during IFN treatment.
Consistent with previous in vitro studies, 28,29 IFNa treatment of PHHs obtained from a stable HBV/HDV-1a-infected mouse had no effect on HDV RNA amounts. However, IFNa led to the reduction of HDV RNA levels in PHHs isolated from a stable HBV/ HDV-1p-infected mouse. Because the virus strain was the only variable in these experimental settings (we used the same human hepatocyte donor and HBV inoculum), the sharp differences determined by using different HDV isolates demonstrate for the first time the existence of virus-derived differences. The key IFNmediated antiviral mechanism acting in stably infected PHHs needs further investigation and the new genotype 1 HDV isolate could also serve for studies aiming at investigating HDV biology, as well as the antiviral activity of new compounds used alone and in combination with IFN.
As we did not measure induction of caspase 8 or detected substantial PHH loss in vitro and in vivo during treatment, the death of HDV-positive cells unlikely provides the main mechanism of intrahepatic HDV decline. Although we cannot rule out an increase of cellular stress induced both by HDV infection and interferon treatment, the strong HDV decrease determined in mice infected with HDV-1p or HDV-3 rather suggests that IFN inhibited HDV RNA synthesis and/or promoted its destabilisation.
The intrinsic innate responses of the hepatocytes are thought to be central to counteract HDV infection. 18,20 To  investigate whether these three HDV strains may alter the ability of the human hepatocytes to sense the infection and to respond to IFN administration, we comparatively analysed classical human ISGs in infected and treated humanised mice. 19 Both the two HDV-1 isolates and the HDV-3 strain similarly enhanced human ISGs, including chemokines and pattern recognition receptors, compared with uninfected control mice. PegIFNa treatment increased the levels of human IFN signalling genes even further andinterestinglyin a comparable manner among all three HDV isolates used, irrespective of their responsiveness to treatment. Even at the single-cell level, HDV-1a-RNA-positive PHHs strongly coexpressed hMxA mRNA, suggesting that this IFN-resistant HDV-1a isolate does not hamper the broad IFN-mediated upregulation of intrinsic innate responses of human hepatocytes in vivo. Nevertheless, as we analysed a limited number of genes, we cannot exclude the possibility that other genes may be differentially expressed. Based on these results, it appears that the hepatocyte innate responses cannot be solely responsible for the strikingly different IFN responsiveness of distinct HDV isolates determined in stable HDV-infected cells, again hinting at the existence of virus-specific factors affecting the strength of IFN responsiveness. We did not identify clear mutations in the ribozyme and HDAg-coding region on the HDV genome or at the RNA editing site upon IFN treatment.
Interestingly, all known post-translational modification sites in the HDAg open reading frame, including the prenylation site at position 212 and the nuclear localisation signal (aa-66-75) remained fully conserved among these three isolates. The two only differences that exclusively occurred in the IFN-resistant HDV-1a-isolate were detected in the coiled coil domain (leucine instead of isoleucine and vice versa) of the HDAg close to a SUMOylation site. However, leucine and isoleucine are both aliphatic, branched hydrophobic amino acids and these mutations may not influence SUMOylation rates at position 42. It remains to be investigated whether more complex sequence and conformational changes residing outside of the more conserved coding regions of HDV account for the intrinsic primary resistance to IFN determined with HDV-1a.
Recently, Zhang et al. 43 treated proliferating HDV-1a-infected hepatoma cells with IFNa and observed a strong block of cell division-mediated HDV spread, 43 suggesting that HDV RNA molecules can be targeted by intrinsic innate responses preferentially during cell division, a mechanism permitting NTCP-independent HDV spreading among daughter cells. 10 Experiments in human liver chimeric mice are usually started when human hepatocyte engraftment is completed and cell turnover is low. 9,10 Herein, we treated HBV/HDV-infected mice with IFN several weeks after PHH repopulation was accomplished and rates of PHH proliferation did not differ among mice infected with the distinct HDV isolates (Ki67-staining, data not shown). Although additional studies are required to explore the impact of IFN on distinct HDV strains during hepatocyte proliferation, cell turnover appears unlikely to explain the different IFN responses determined in the experimental setting used here. The ability of IFN to limit not only HDV infection events, but also to lower intrahepatic viral loads even in resting hepatocytes, would provide a rationale for the stronger anti-HDV effects determined in people with CHD receiving pegIFNa and the entry inhibitor bulevirtide in combination (MYR-203 clinical trial, NCT02637999). 44 In conclusion, our study showed that two new patientderived HDV isolates of genotype 1 and 3 respond to IFNa treatment in immune-deficient human liver chimeric mice. We also provide evidence that the commonly used HDV-1a isolate bears intrinsic capacities to resist IFNa treatment in vivo, which were not determined in two other patient-derived isolates. The existence of virus-specific determinants of IFNa responsiveness raises awareness of the need to use different HDV strains to evaluate virological and host-mediated mechanisms of IFNresponsiveness in HDV infection. The availability of new IFNsensitive HDV strains could also contribute to the development of new therapies aiming at HDV cure.

Financial support
The study was supported by the German Research Foundation (DFG) by a grant to MD and ML (SFB 841, A8), and to DG (GL595/9-1 and SFB1021, B08). MD and DG also received funding from the German Center for Infection Research (DZIF-BMBF; TTU-hepatitis 05.820; 05.822; 05.714). The National Reference Center for Hepatitis B Viruses and Hepatitis D Viruses is supported by the German Ministry of Health via the Robert Koch Institute. All funding sources supporting the work are acknowledged and authors have nothing to disclose. genotypes and HDV-1 strains [4]. HDV RNA levels in cell culture supernatant were also measured using the cobas6800 automated system (Roche, Basel, Switzerland) as previously described [5]. HBV DNA levels in cell culture supernatant and mouse serum as well as HBV pregenomic (pg) RNA levels in cells and mouse liver were determined by qPCR using specific primers and probe (Taqman Gene Expression Assay Pa03453406_s1, Applied Biosystems and [6]) under conditions previously described [3]. Known amounts of an HDV-or HBVcontaining plasmid were used as standard for HDV RNA and HBV DNA quantification in serum and cell culture supernatants. Steady-state levels of intracellular viral RNA and DNA amounts were normalized to the median of human specific hGAPDH and hRPL30 (Taqman Gene Expression Assay Hs99999905_m1 and Hs00265497_m1, Applied Biosystems) using the ∆∆ct method.

Strain-specific responsiveness of hepatitis D virus to interferon
Genomic and antigenomic HDV RNA qPCR assay. Genomic and antigenomic HDV RNA were determined using a biotinylated magnetic beads based qPCR assay as described previously [7]. In brief, RNA extracted from 1 μl mouse liver was reverse transcribed with 0.5 μM of a biotinylated HDV specific forward primer (biotin-GCGCCGGCYGGGCAAC) for genomic HDV RNA or a biotinylated HDV specific reverse primer (biotin-TTCCTCTTCGGGTCGGCATG) for antigenomic HDV RNA detection and the ABI Fast 1-Step Virus Master (Applied Biosystems, Carlsbad, USA). Biotinylated cDNA was purified with the MinElute PCR Purification Kit (Qiagen, Hilden, Germany) and isolated with dynabeads specifically interacting with biotin (Dynal Kilobase Binder Kit, Invitrogen, Darmstadt, Germany) following the manufacturer's instructions. For qPCR 1 μl of purified biotinylated cDNA bound to dynabeads, HDV specific primers and probes [4] and the ABI Fast Advanced Master (Applied Biosystems, Carlsbad, USA) were used. The median of two human-specific housekeeping genes (hGAPDH, Hs99999905_m1, and hRPL30, Hs00265497_m1, Applied Biosystems) were used for normalization.
HDAg Western blot. Western blot of mouse or patient liver tissue was performed as previously described [8]. In brief, protein lysates were obtained by extracting tissue with T-Per Tissue Protein Extraction Reagent (Pierce, Rockford, United States) supplemented with protease and phosphatase inhibitors. Protein content was measured by Pierce BCA Protein Assay Kit Sequencing. For HDV genome sequencing, serum RNA from HBV/HDV-1p-or HDV-1Ainfected mice was extracted as described above and cDNA was synthesized with the Transcriptor First Strand cDNA Synthesis Kit (Roche, Basel, Switzerland) using random hexamer primer according to the manufacturer's instructions. To generate the full genome sequence of HDV-1p overlapping PCR fragments were generated using 5 HDV-specific primer pairs [9] and a Red-Taq Polymerase (Sigma-Aldrich, St. Louis, USA) under conditions described previously [9]. To analyze the occurrence of mutations in treated HDV-1p-or HDV-1A-infected mice, the HDV-specific primer pairs R1 and R2 [10] were used as described previously [11]. PCR product length was analyzed on a 0.8% agarose gel and DNA fragments were purified with the MinElute PCR Purification Kit (Qiagen) as recommended by the manufacturer. The forward and reverse strand was sequenced with Sanger sequencing (Mix2seq kit) by Eurofins Genomics (Ebersberg, Germany) and data was analyzed using Geneious R6 (BioMatters, Auckland, New Zealand).