Establishment of a novel hepatitis B virus culture system using immortalized human hepatocytes

Recent development of hepatitis B virus (HBV) culture systems has made it possible to analyze the almost all steps of the viral life cycle. However, the reproducibility of interaction between HBV and host cells seemed inaccurate in those systems because of utilization of cancer cell lines with a difference from hepatocytes in the majority of cases. In this study, in order to resolve this point, a novel HBV culture system using non-cancer-derived immortalized human hepatocytes derived cell lines, producing exogenous human sodium taurocholate cotransporting polypeptide, was developed. One of the cell clones, E/NtG8 cells, was permissive to both blood-borne HBV (HBVbb) and culture-derived recombinant HBV when cultured in the three-dimensional condition. Furthermore, the production of infectious HBV particles, which showed the similar physicochemical properties to HBVbb, was observed for about a month after HBVbb infection in this system, suggesting that it may reproduce whole steps of the HBV lifecycle under the condition analogous to human liver cells infected with HBV. This system seemed to contribute not only to find novel interactions between HBV and host cells but also to understand mechanism of HBV pathogenesis.

Scientific Reports | (2020) 10:21718 | https://doi.org/10.1038/s41598-020-78655-x www.nature.com/scientificreports/ some cases, high cost, and the limited availability for detailed mechanistic analysis. Therefore, for studying the physiological interactions between HBV and the host cells and the production of HBV particles, HBV culture systems using non-cancer-derived cells that are applicable for molecular, biological, and pathogenesis studies are required. It has been reported that the immortalized human hepatocytes, HuS-E/2 cells treated with 2% DMSO for twelve days became susceptible to HBV infection albeit with low efficiency 21 . HuS-E/2 cells have also been shown to support life cycle of blood-borne hepatitis C virus (HCVbb) to some extent 22 . Recent findings of NTCP, and the low level expression of NTCP gene in HuS-E/2 cells suggested the possibility that the high level production of NTCP in HuS-E/2 cells might improve the potential of this cell clone to support HBV life cycle. In this study, therefore, it was intended to develop a novel HBV culture system using HuS-E/2 cells, ectopically producing NTCP.

Results
Establishment of immortalized human hepatocytes stably producing NTCP and the verification of the permissiveness to HBV infection. After selection with G418, several G418-resistant clones were isolated as independent colonies from HuS-E/2 cells transfected with RG210241. Next, eight clones among the G418-resistant clones were selected as candidates according to the expression level of NTCP-tGFP mRNA and named as E/NtG1 to E/NtG8 cells. Among these clones, E/NtG3 and E/NtG8 cells were selected as HuS-E/2-derived cell clones that were high producers of NTCP-tGFP proteins (Fig. 1A). Localization of NTCP-tGFP around the region of the plasma membranes in the cells was observed as expected (Fig. 1B). Next, the permissiveness of each cell clone to HBVcc infection was examined. As shown in Fig. 1C, the copy numbers of HBV pgRNA in E/NtG8 cells appeared to be maintained relatively from 7 to 10 dpi, compared with E/NtG3 cells in which the obvious decrease of HBV pgRNA was detected over the same period. The level of HBV pgRNA expression in E/NtG8 cells (approximately 5 × 10 3 copies/μg total RNA), however, was lower than that in the other existing HBV culture systems 4 (see Fig. 2C, approximately 10 4 to 10 5 copies/μg total RNA). In addition, the extracellular HBV DNA production from HBVcc-infected E/NtG8 cells was marginal (Fig. 1D). These results suggested that E/NtG3 cells are not permissive to HBV infection, but E/NtG8 cells possess the potential to support HBV infection to some extent with limited ability. The possible contribution of a cellular restriction factor of HBV, SMC6 23 , to the low efficiency of HBV infection in E/NtG8 cells was examined by SMC6 knockdown experiment next. As shown in Supplemental Fig. 1C, the obvious increase of HBV pgRNA was not observed in those cells after HBVcc infection despite effective knockdown of SMC6 mRNA (Supplemental Fig. S1D). As the suppressive effect of SMC6 on the expression of HBV pgRNA was observed in HepG2-NTCP-C4 cells (Supple-   Fig. S1A,B) as previously reported 23 , the low level expression of pgRNA observed in E/NtG8 cells after HBVcc infection was not due to the effect of SMC6 (Supplemental Fig. S1C).

HBVcc infection in 3D-cultured E/NtG8 cells. Next, the possible improvement of HBV infection in E/
NtG8 cells cultured in 3D condition was examined, because parental HuS-E/2 cells have already shown to support the infection and proliferation of HCVbb under the 3D culture condition 24 . In the present study, Cellbed scaffold was utilized for the 3D culture because it was likely to be a better system for the study of virus infections in vitro because cells can be cultured simply and easily in 3D conditions and the cultured cells are directly exposed to the culture medium containing virus particles. Fifteen days after infection with HBVcc, the higher copy number of HBV pgRNA was detected in the 3D-cultured cells treated with myr-47N9K than in the cells treated with myr-47WT (Fig. 2B), while the parental HuS-E/2 cells cultured similarly did not show significant difference (Fig. 2B). The statistical significance of this data suggested that 3D-cultured E/NtG8 cells are permissive to HBVcc infection, unlike 3D-cultured HuS-E/2 cells. Although the level of HBV pgRNA expression in the cells (approximately 5 × 10 3 copies/μg total RNA) was not much higher than that of the cells in the flat culture condition (Fig. 1C), the noticeable production of extracellular HBV DNA from 3D-cultured E/NtG8 cells was observed at 15 dpi (Fig. 2C). These results suggested that 3D-cultured E/NtG8 cells could support almost all the steps of HBV life cycle.

HBVbb infection in 3D-cultured E/NtG8 cells. Next, the permissiveness of 3D-cultured E/NtG8 cells to
HBVbb was examined in the same way. In the case of infection with HBVbb genotype C, HBV pgRNA (2.5 × 10 5 copies/µg total RNA) in myr-47N9K-treated cells showed a significantly higher copy number than that in myr-47WT-treated cells at 15 dpi without the effect of ETV treatment (Fig. 3B). The formation of HBV cccDNA in HBVbb-infected 3D-cultured E/NtG8 at 15 dpi was also detected (Supplemental Fig. S2). The copy numbers of extracellular HBV DNA from myr-47N9K-treated cells was higher by tenfold at 15 dpi and reached to approximately 1.4 × 10 6 copies/ml. ETV treatment of myr-47N9K-treated cells caused a large decrease in the extracellular HBV DNA, indicating that the detection of HBV DNA was HBV genome replication-dependent (Fig. 3C). The infection of HBVbb genotype A was also examined in the same way. At 15 dpi, the amount of extracellular HBV DNA from myr-47N9K-treated cells was detected at higher copy number than in myr-47WT-treated cells as in the case of HBVbb genotype C (Supplemental Fig. S3), suggesting that 3D-cultured E/NtG8 cells can support the infection and proliferation of HBVbb of either genotype. Furthermore, the HBs-positive cells were detected in the cells inoculated with HBVbb genotype C in the presence of myr-47N9K at 15 dpi by indirect immunofluorescence analysis (Fig. 3D), indicating that HBs was actually produced in the cells.

Production of infectious HBV particles from 3D-cultured E/NtG8 cells infected with HBVbb. As
shown in Fig. 4B, the production of extracellular HBV DNA was constantly detected until 27 dpi. The infectious HBV particle production was examined by the infection experiment using pooled culture media as infection source (Fig. 4A). The naïve 3D-cultured E/NtG8 cells were inoculated with the pooled media at 5 HBV GEq/cell as described above. As shown in Fig. 4C, the copy numbers of extracellular HBV DNA from myr-47N9K-treated cells was apparently higher than that from myr-47WT-treated cells at 15 dpi and reached to approximately 10 6 copies/ml, indicating that the pooled media included the HBV particles which were as infectious as HBVbb. These data suggested that 3D-cultured E/NtG8 cells infected with HBVbb could produce highly infectious HBV particles for about a month. All these data suggested that 3D-cultured E/NtG8 cells could be susceptible to both HBVcc and HBVbb and especially reproduce all the steps of the life cycle of HBVbb including the production of the infectious HBV particles.

Properties of the HBV particles produced from 3D-cultured E/NtG8 cells.
In the case of infection experiment using PXB cells, the cells only had to be infected with HBVbb on the 5 GEq/cell condition to obtain the assessable and quantitative results (Supplemental Fig. S4A 20 ). The infectivity of 3D-cultured E/NtG8-derived HBV, therefore, seemed to be equivalent to the HBVbb as above. On the other hand, when HBVcc was used as an  S4B). As the infectivity deemed less different among HBV genotypes 20 , it seemed possible to suppose that the inocula of HBV from 3D-cultured E/NtG8 cells and HBVbb include the factor enhancing the HBV infection or that of HBVcc includes a suppressor. However, the presence of those affecting the HBV infection was not observed when the inoculum including both HBVbb and HBVcc was utilized in the HBV infection experiment using PXB cells (Supplemental Fig. S5). Next, therefore, the physicochemical properties of those HBV particles were investigated further by CsCl density gradient ultracentrifugation analysis.
Overall distribution patterns of HBV DNA, HBV antigens, small HBs (SHBs) and HBc, and buoyant densities in CsCl densities were similar to those have already reported previously 25 , except the obscure presence of unenveloped core particles. The fractions indicated with c, c' , and c'' showing the maximum contents of HBV DNA and HBc were likely to contain the infectious HBV particles (Dane particles) (Fig. 5, left panels). And the fractions indicated with b, b' , and b'' showing the maximum contents of small HBs antigen were presumed to include subviral particles (Fig. 5, right panels). The fractions indicated with e, e' , and e'' included the minor peak of HBV DNA, probably implying the presence of capsid associated with HBV DNA (Fig. 5). It should be note that those features of 3D-cultured E/NtG8-derived HBV were largely similar to those of HBVbb. Especially the pattern of HBV DNA content showed as a single peak around fractions c and c' . However, that of HBVcc fractions showed the peak with obvious shoulder from fractions c to d (Fig. 5C), although what kinds of particles were included in the fractions corresponding to the shoulder were unknown yet. These results suggested that the physicochemical properties of the HBV particles produced from 3D-cultured E/NtG8 cells bear a resemblance to those of HBVbb, rather than HBVcc. On the other hand, it was observed that the distribution pattern of HBc in each fraction in the right panel of Fig. 5 looked similar between those of panel A and panel C. The difference of the pattens seemed to be caused by higher relative amount of HBc in the fraction e' than those in fractions e and e'' . In addition, overall distribution pattern of sHBs seemed similar between those of panel B and panel C. However, the contribution of those pattern differences to the infectivity of the viruses was unknown at this moment.

Candidates of the genes supporting HBV life cycle in 3D-cultured E/NtG8 cells. Our data
indicated that the 3D culture condition makes E/NtG8 cells permissive to HBV infection and proliferation. As shown in Fig. 6A,B, the morphology of E/NtG8 cells cultured on Cellbed for one week showed hepatic cord-like structures 26 . In order to obtain the information about cellular factors involved in HBV life cycle in the cells, the expression of several HBV related genes in 3D cultured E/NtG8 cells was investigated as reported previously in the case of HCV 27 . As shown in Fig. 6C, the significantly upregulated expression of the hepatocyte nuclear factor (HNF) family genes, HNF-1α and HNF-4α, which are host transcription factors essential for HBV gene expression 28 , was observed in 3D-cultured E/NtG8 cells, although only low expression levels were detected in www.nature.com/scientificreports/ the cell cultures in 2D conditions, suggesting that one of the advantages of 3D-cultured E/NtG8 cells would be the high capacity of HBV gene transcription by the increased expression of hepatocyte-specific genes, including HNF family genes. Next, the expression of genes associated with the HBV particle egression, especially genes for fatty acid biosynthesis enzymes 29 , was investigated. Figure 6C shows that acyl-CoA carboxylase (ACC1) and fatty acid synthase (FAS) genes showed increased expression in 3D-cultured E/NtG8 cells, although the increase in ACC1 gene expression seemed rather limited. Next, the expression of the genes for the factors reported to have anti-HBV potential was also investigated. The expression of the gene for BST-2/tetherin, that was reported to inhibit the release of HBV particles 30 , was significantly lower in E/NtG8 cells cultured in both 2D-and 3D-culture conditions than in HepG2-NTCP-C4 cells (Fig. 6C). The expression of the SMC6 gene in 3D-cultured E/NtG8 cells was also analyzed as described above. The amount of SMC6 mRNA in 2D-cultured E/NtG8 cells was similar to that in HepG2-NTCP-C4 cells treated with control siRNA (Fig. 6D). However, this expression was significantly decreased in 3D-cultured E/NtG8 cells just like that in SMC6 siRNA-transfected HepG2-NTCP-C4 cells (Fig. 6D), in which HBV pgRNA expression was largely enhanced after HBVcc infection (Supplemental Fig. 1A). This data indicated that 3D culture affects SMC6 gene expression in E/NtG8 cells. As observed that the 3D-cultured E/NtG8 cells became more hepatocytelike cells (Fig. 6A-C), the SMC6 gene expression was examined in several other cells or tissue that share similar features with normal human hepatocytes. As shown in Fig. 6D, compared with HepG2-NTCP-C4 cells and E/ NtG8 cells cultured in the 2D culture condition, only low-level expression of SMC6 mRNA was observed in PXB mouse liver and cryopreserved human hepatocytes cultured for 48 h. Furthermore, SMC6 mRNA was less Figure 5. Culture media from 3D-cultured E/NtG8 cells infected with HBVbb (A) and HBVcc producing HepG38-7Tet cells (C) and serum including HBVbb (B) were fractionated by CsCl density gradient ultracentrifugation. For each fraction, the amount of HBV DNA was indicated (black bar) against buoyant density (g/ml, gray dot) (right panels of (A-C)). The fractions with small letters on horizontal axis were selected for the detection of HBV antigens by ELISA. The amounts of small HBs (SHBs, white bars) and HBc antigens (black bars) in the selected fractions are shown (left panels of (A-C)).

Scientific Reports |
(2020) 10:21718 | https://doi.org/10.1038/s41598-020-78655-x www.nature.com/scientificreports/ expressed in cryopreserved human hepatocytes before culturing on the culture plate. These results suggest that SMC6 gene expression is limited in human hepatocytes in the liver tissue, although the functional potential of the protein in the hepatocytes of human liver remains unclear. It was appeared that the expression of genes related with HBV lifecycle was regulated in the E/NtG8 cells in 3D-culture conditions to support HBV proliferation.

Discussion
Numerous HBV culture systems have been reported and used for HBV research. Liver cancer-derived cells such as HepG2, HuH-7, and HepaRG cells have been commonly used in these systems, while primary human hepatocytes and PXB cells have also utilized in some cases 31,32 . Hepatocyte-like cells developed by differentiation from induced pulmonary stem (iPS) cells for the reproduction of HBV infection and proliferation have been previously reported [33][34][35] . Each system mentioned above has both merits and demerits. In the case of systems using cancer-derived cells, especially recently developed cell lines constitutively producing NTCP, almost the entire HBV life cycle can be easily studied. Selected cell clones from those cells recently showed better reproduction of the HBV life cycle 36,37 . Nevertheless, the investigation of the interactions between the host cell and HBV under physiological condition seems to be limited because of the abnormalities in cancer-derived cells. The iPS-derived hepatocyte-like cells may be one of the most ideal systems to study this subject, but the efficacy of HBVcc infection is quite low at this moment. Those cells also seemed to require investigation for their potential to produce infectious HBV particles and if they can be highly permissive to HBVbb infection such as PXB cells and 3D-cultured E/NtG8 cells, although the exact differences between HBVcc and HBVbb are not yet known. Because HuS-E/2 cells, the parental cells of E/NtG8 cells, have already shown to support the infection and proliferation of HCVbb through becoming matured to hepatocyte-like cells in 3D culture conditions 22,24,27 , it www.nature.com/scientificreports/ seemed possible that the 3D culture conditions improve cellular environment of E/NtG8 cells for HBV proliferation as well. Then the expression profiles of several genes, which contribute to HBV lifecycle, in 3D-cultured E/ NtG8 cells were investigated. The expression levels of the genes for hepatocyte-specific transcription factors such as HNF-1α and HNF-4α that are known as regulators for HBV gene expression 38,39 were induced in 3D-cultured E/NtG8 cells (Fig. 6C). It is likely that the relatively high level expression of those factors are required for the adequate transcription of HBV genes from HBV cccDNA as well as host genes supporting HBV infection and proliferation. In this study, the production of infectious HBV particles that have similar characteristic features to HBVbb from 3D-cultured E/NtG8 cells were also observed (Figs. 3, 5). Since the fatty acid biosynthesis pathway contributes to HBV particle production 29 , the elevated expression levels of ACC1 and FAS genes observed in 3D-cultured E/NtG8 cells suggests the advantage in the production of infectious HBV particles. Although the expression of the genes for several factors of the endosomal sorting complex, that are required for the transport pathway in HBV particle egression, such as α-glycosidase, AAA-type ATPase Vps4, charged multivesicular body protein 3 (CHMP3), and CHMP4 [40][41][42] , were also observed in E/NtG8 cells, an obvious alteration of the gene expression was not found (Supplementary Fig. S6). In addition, the gene expression of an antiviral factor, BST-2/ tetherin, which was suggested to inhibit HBV release from the cells 30 , was observed to be low in E/NtG8 cells irrespective of the culture condition. These data of the gene expression levels may partly explain the advantage of 3D-culture conditions for E/NtG8 cells in terms of production of infectious HBV particles. In this study, some crucial differences were noted between 2D-and 3D-cultured E/NtG8 cells. A recent study suggested that the SMC 5/6 complex functions as a host restriction factor of HBV replication by suppressing transcription from the extrachromosomal DNA such as HBV cccDNA 23 . This suppressive function of SMC6 was confirmed in HBVcc-infected HepG2-NTCP-C4 cells by decreased expression of this factor by siRNA, although a significant effect was not observed in HBVcc-infected 2D-cultured E/NtG8 cells, which could be attributed to the limited expression levels of the essential transcription factors. Therefore, it was expected that the suppression of SMC6 expression improved the efficacy of HBV infection in 3D-cultured E/NtG8 cells producing those factors at a level higher than that produced in 2D-cultured cells. Unexpectedly, however, SMC6 gene expression was downregulated in 3D-cultured E/NtG8 cells. Since the mRNA level of SMC6 in 3D-cultured E/NtG8 cells was largely similar to that in SMC6 knockdown HepG2-NTCP-C4 cells, the suppressive effect of SMC6 on HBV transcription seemed to be limited in 3D-cultured cells, suggesting that one of the advantages of 3D culture could be cancellation of this cellular restriction machinery against HBV. Nevertheless, it seemed possible that SMC6 gene is expressed at marginal levels in the human liver tissue because the apparent expression of the gene was not detected in both cryopreserved primary human hepatocytes and in the liver tissues of a chimeric mouse with human hepatocytes. The detection of SMC6 mRNA in primary human hepatocytes at two days after plating on the culture dish suggested that SMC6 gene expression could be induced by 2D culture conditions. Further investigations are required for understanding the role of SMC6 in HBV infection in the human liver.
In contrast to HepG2-NTCP-C4 cells, 3D-cultured E/NtG8 cells showed much higher permissiveness to HBVbb infection than HBVcc. Under the condition that HBVbb inoculum was used only at 5 GEq/cell, the infection of HBVbb was observed in 3D-cultured E/NtG8 cells as well as PXB cells 20 , but not HepG2-NTCP-C4 cells (Supplemental Fig. S7). Since the detectable infection of HBVcc was similarly observed when both HepG2-NTCP-C4 cells and 3D-cultured E/NtG8 cells were infected with HBVcc at 8,000 GEq/cell, the different susceptibility to HBVbb between HepG2-NTCP-C4 cells and 3D-cultured E/NtG8 cells seemed due to the lack of requirement(s) in HepG2-NTCP-C4 cells to support HBVbb infection. Further analysis of these requirements may provide an insight into understanding HBV infection in physiological conditions. In addition, it will be important to examine that the expression of genes in the host cells is affected during HBVbb or HBVcc infection.
In conclusion, we developed a novel in vitro culture system combining 3D-culture scaffold and NTCPexpressing immortalized hepatocytes and demonstrated the relatively efficient support of HBV lifecycle, including production of highly infectious HBV particles. This system may be of benefit for studying the interactions between HBV and host cells at the molecular level in physiological conditions and for understanding mechanism of HBV pathogenesis.
Cell culture. HuS-E/2 cells and the cell clones derived from HuS-E/2 cells were cultured using a specialized culture medium for the maintenance of HuS-E/2 cells in the presence and absence of 500 μg/ml G418 (Nacalai Tesque, Kyoto, Japan), respectively, as previously described 22 . Transfection of RG210241 into HuS-E/2 cells was performed using Effectene Transfection Reagent (Qiagen, Hilden, Germany). For 3D cell culture, E/NtG8 cells were cultured with serum-free medium (SFM)(TOYOBO, Osaka, Japan) on Cellbed (the high-purity silica fiber Scientific Reports | (2020) 10:21718 | https://doi.org/10.1038/s41598-020-78655-x www.nature.com/scientificreports/ scaffold, Japan Vilene, Tokyo, Japan) 44 in a 24-well plate (5 × 10 5 cells per well) for one week before the infection experiment. In the case of 3D culture for HepG2-hNTCP-C4 cells, the cells were similarly cultured on Cellbed except for the cell number (5 × 10 5 cells per well) and three days of culture before infection experiment. The overlaid culture medium was changed every 2 days. The 293FT cells were cultured as previously described 17 .
Immunoblot analysis. Immunoblot analysis was performed as previously described 46 . The HRP-catalyzed chemiluminescence in the Western Lightning reagent (Perkin Elmer, Massachusetts, USA) was detected using ImageQuant LAS-4000 system (GE Healthcare, Chicago, USA) according to the manufacturer's protocol.
Indirect immunofluorescence analysis. Indirect immunofluorescence analysis was performed as previously described 47 . The anti-HBs antibody was used as the primary antibody at 1:200 dilution, and the secondary antibody, Alexa 546-conjugated anti-mouse IgG antibody (Invitrogen, California, USA), was used at 1∶1000 dilution. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Fluorescence imaging was performed with AF6000, a fluorescence microscope (Leica, Wetzlar, Germany), or FV1000, a laser scanning confocal microscope (Olympus, Tokyo, Japan). Total RNA extraction from cultured cells, reverse transcription polymerase chain reaction, and quantitative RT-PCR. Total cellular RNA was isolated from the cells using the RNeasy Plus Mini Kit (Qiagen, Hilden, Germany) after treatment with RNase-free DNase (Qiagen). RT-PCR was performed using One-Step RNA PCR Kit (Takara, Osaka, Japan) with target-specific primer sets (listed in Table 1 with PCR cycles). HBV RNA was quantified using ReverTra Ace qPCR RT Kit (TOYOBO, Osaka, Japan) and Universal SYBR Green Master Mix (Applied Biosystems, CA, USA). All these experiments were performed according to the manufacturers' protocols.

HBV preparation and conditions of infection.
Extracellular HBV DNA extraction and quantitative PCR. Extracellular HBV DNA was isolated from the culture medium or sera with DNeasy Blood and Tissue kit (Qiagen) following MNase treatment with 10 U per reaction at 37 °C for 1 h, as previously described 29 . MNase-resistant extracellular HBV DNA was quantified by quantitative PCR using THUNDERBIRD Probe qPCR Mix (TOYOBO, Osaka, Japan) with a targetspecific TaqMan probe and primer sets as previously described 29,48 . In Supplemental Fig. S5, genotype D-specific HBV DNA was quantified using PowerUp SYBR Green Master Mix (Thermo Fisher Scientific, MA USA) and specific primers (listed in Table 1).
Enzyme-linked immunosorbent assay (ELISA). Quantitation of small HBs (SHBs) and HBc were performed by HBs S Antigen Quantitative ELISA Kit, Rapid-II (Beacle; Kyoto, Japan) and QuickTiter HBV Core Antigen ELISA Kit (Cell Biolabs, CA, USA), respectively, according to the manufacturer's protocols.

Statistical analysis.
All experiments were independently performed at least three times. Statistical analyses were performed using Student's t-test; p < 0.05 was considered statistically significant.

Data availability
All data generated or analyzed during this study are included in this published article and its Supplementary Information files.