Human embryonic stem cell–derived hepatoblasts are an optimal lineage stage for hepatitis C virus infection

CELL CULTURE AND DIFFERENTIATION

The H9 Human ESC line was purchased and maintained following the protocols provided by the WiCell Research Institute (Madison, WI). H9 cells were passaged and grown for 1 day in MTeSR (Stem Cell, Vancouver, Canada), then cultured in Kubota Media with Activin A (100 ng/ml) and Wnt3a (50 ng/ml) for 4 days to induce cells to definitive endoderm (DE) cells. The DE cells were differentiated in Hepatocyte Culture Medium (HM, Lonza, Allendale, NJ), Wnt3a (50 ng/ml), FGF4 (30 ng/ml), retinoic acid (RA, 0.1 μM) for 6 days to achieve lineage restrictions to hHpSCs. The hHBs were obtained by treating the hHpSCs with the HM, FGF4 (30 ng/ml), BMP4 (25 ng/ml) and HGF (25 ng/ml) for another 6 days. Finally differentiation to hepatocytes was achieved with HM with HGF (25 ng/ml), OSM (10 ng/ml) and dexamethasone (Dex, 0.1 μM). Recombinant human EGF (40 ng/ml) and VEGF₁₆₅ (50 ng/ml) were added from the HpSC stage to obtain functional hHePs. Matrigel, Laminin, and collagen IV were from Becton Dickinson (BD, San Jose, CA). Recombinant human Activin A, Wnt3a, FGF4, BMP4, EGF, HGF and OSM were from R&D Systems (Minneapolis, MN). Dex was from Sigma-Aldrich (Saint Louis, MI). Recombinant human VEGF165 was from PeProtech (Rocky Hill, NJ).

IMMUNOFLUORESCENT AND IMMUNOHISTOCHEMISTRY ANALYSES

Immunofluorescent and Immunohistochemistry analyses were carried out as previously described.16, 17 Details and dilutions about antibodies were listed in Supporting Table 1.

FLUORESCENCE-ACTIVATED CELL SORTING (FACS)

FACS analysis was carried out as previously described.17 Details and dilutions about anti-Albumin and anti-AFP antibody were listed in Supporting Table 1.

TRANSMISSION ELECTRON MICROSCOPY (TEM)

The cells were washed twice with PBS and fixed by incubation for 48 hours in 4% paraformaldehyde and 1% glutaraldehyde in 0.1 M PBS. After washing with PBS, the cells were post-fixed by incubation for 1 hour with 1% osmium tetroxide and dehydrated in a graded series of ethanol solutions. Then the cells were embedded in Polybed 812 epoxy resin (Polysciences, Warrington, PA). Ultrathin sections were cut and collected on 200 mesh copper grids, stained with 4% aqueous uranyl acetate for 15 minutes, and then with Reynolds' lead citrate for 7 minutes. Stained sections were examined with a H7650 transmission electron microscope (HITACHI, Tokyo, Japan).

QUANTITATIVE REAL-TIME PCR

Total RNA was extracted from cells or freshly isolated tissue using an RNeasy Micro or Mini Extraction kit (Qiagen, Hilden, Germany). Total RNA (1 µg) was used for reverse transcription with ReverTra Ace qPCR RT master mix (Toyobo, Kita-ku Osaka, Japan). The qRT-PCR analysis was performed on a Bio-Rad iQ5 System using the SYBR Green PCR Master Mix (Toyobo). The relative expression of each gene was normalized against GAPDH. Primer sequences are listed in Supporting Table 2.

FUNCTIONAL ASSAYS

hESCs DIFFERENTIATED TO HEPATOCYTE-LIKE CELLS THROUGH MULTIPLE LINEAGE STAGES AND IN A STEP-WISE PROCESS

We previously set up protocols of a step-wise differentiation procedure with serum-free media to efficiently and induce either hESCs16 or multipotent stem cells17 into several hepatic lineages efficiently and reproducibly (Fig. 1A). Following this differentiation procedure, we observed dynamic morphological changes from hESCs to hHeps (Fig. 1B). The expression of lineage-specific markers was used to define empirically each lineage stage and to evaluate the differentiation efficiency. Under the undifferentiated hESC maintenance conditions, almost all of the cells expressed pluripotency markers OCT4 and NANOG. After 4 days of culture in the defined medium used to lineage-restrict the hESCs to hDECs, they lost expression of the pluripotency markers, while >90% of them expressed DE-specific markers SOX17 and FOXA2 (Fig. 1B). The culture medium was changed into a defined medium to lineage-restrict these hDECs to hHpSCs within 6 days, which resulted in high expression of the stem cell marker epithelial cell adhesion molecule (EpCAM); the liver-specific transcription factor hepatocyte nuclear factor 4α exhibited a high level of expression complemented by the activation of albumin (ALB). The specification markers indicative of hepatic lineage cells were robust after treatment with the induction medium for hHBs for another 6 days. About 70% of cells were doubly positive for ALB and alpha-fetoprotein (AFP) (Fig. 1D). Hepatocyte growth factor, dexamethasone, and oncostatin M were used as the final induction factors for at least 6-20 days to generate functional hHeps. At the final stage of maturation, about 80% of the cells were positive for ALB, negative for AFP, and polygonal in shape, all traits consistent with maturation to hepatocytes. Quantitative RT-PCR (Fig. 1C) also confirmed the transition from pluripotent stem cells to SOX17-positive and FOXA2-positive hDECs, followed by induction to AFP-positive hHBs and finally to ALB-secreting, AFP-negative hHeps. The hHeps were also found to display the known hepatocyte functions including glycogen storage, indocyanine green uptake, and urea secretion (Fig. 1E). Ultrastructurally, the hESC-derived hHeps resembled primary human hepatocytes with accumulation of glycogen rosettes, mitochondria, endoplasmic reticulum, canaliculi, and exocytotic vesicles (Fig. 1F). These results indicate that this protocol is highly efficient for generating differentiated hepatocytes from hESCs.

IN VITRO DIFFERENTIATED hHBs WERE THE MOST SUSCEPTIBLE LINEAGE STAGE TO HCV INFECTION

We divided the entire hepatic lineage restriction process into five stages: hESC, hDE, hHpSC, hHB, and adult hHep. These stages were used to define the potential for HCV infections. We further subdivided the stages into pre-Hep and Hep stages after the cells had gone through the hHB stage and were about to enter the hepatocyte stage. To demonstrate the kinetics of HCV infection during the hepatic lineage differentiation process, we inoculated a genotype 1b HCV produced in cell culture (HCVcc), the most common HCV genotype in China, generated with Vero-E6 cells18 at each stage (Fig. 2A). In agreement with previous work, there was no detectable HCV RNA in hESCs or in hDECs. HCV permissiveness started during hepatic specification at the hHpSC stage (Fig. 2B,C,F). Quantitative RT-PCR on cell lysates at 5 days postinfection (dpi) showed that HCV genomes were most abundant in hHBs. This result was consistent with immunofluorescent staining and western blotting (Fig. 2B,C; Supporting Fig. S1). Intracellular HCV RNA levels generally showed an inverted S shape and reached a peak, followed by a decrease and then a relatively stable phase when lineage-restricted cells acquired mature functions (Fig. 2F).

Compared to the classical HCV culture system Huh7.5.1 cells, hHBs were more suitable for intracellular HCV replication. The intracellular HCV from hHBs achieved almost 4 × 10⁶ copies/μg RNA compared to 3 × 10⁴ copies/μg Huh7.5.1 cell RNA. In addition, the supernatant from HCV-infected hHBs contained much more infectious virions that were able to infect Huh7.5.1 cells at a titer of ∼200 RNA copies/μg, 2-3 times that from mature hepatocytes or from Huh7.5.1 cells (Fig. 2C,D). Treatment of infected cells with IFN or cyclosporine A for 2 days inhibited 88% or 80% of HCV replication correspondingly in the hHB stage, confirming that this is authentic HCV replication. HCV-infected hHpSCs and hHeps were not as sensitive as those at the hHB stage to IFN or cyclosporine A treatment (Fig. 2E).

Together, the dynamic HCV infections during hESC hepatic lineage restriction have been outlined; hESC-derived hHB proved to be the most suitable stage for HCV entry and replication, as well as showing the best anti-HCV drug response, compared with cells at the HpSC and Hep stages. Clinically, the liver immunohistochemical (IHC) data of HCV patients also revealed that HCV nonstructural protein 5 (NS5) was always located in and around the AFP⁺/EpCAM⁺ early hepatic epithelial cells that appeared in the regenerating pseudo-lobules (Supporting Fig. S2), which is consistent with the in vitro results.

DISTINCT EXPRESSION LEVELS OF HCV-RELATED CELLULAR DETERMINANTS WERE FOUND DURING MATURATION OF hESCs TO HEPATOCYTES

Subsequently, we identified the cellular determinants that correlated with HCV infection during hESC differentiation. Expression and localization of the viral receptors are the important determinants for cell susceptibility to infection and viral tropism of particular tissues. The expression of at least 10 reported receptors (e.g., LDLR, SRBI, claudin 1, epidermal growth factor receptor)19, 20 related to HCV entry or infection was checked during the whole differentiation process from hESCs to hHeps by quantitative RT-PCR. The results indicated that epidermal growth factor receptor, OCLN, and claudin 1 might play more important roles in HCV permissiveness (Supporting Fig. S3A).

Among all of these HCV entry factors, cluster of differentiation 81 (CD81) and OCLN are the minimal necessary human-specific HCV entry factors.15 The quantitative RT-PCR results showed that the expression of CD81 was maintained at a stable level, whereas OCLN gradually increased during lineage restriction through hepatic stages (Supporting Fig. S3A). Immunostaining results indicated that CD81 is at the surface throughout the maturational lineage stages. By contrast, OCLN showed a gradual translocation process from cytoplasm at the ES and DE stages, to the apical poles little by little, to the membranes when cells entered the hHpSC stage. They finally settled stably on the lateral cell membranes of hHBs and hHeps. OCLN localization occurred increasingly at the cell apex, paralleling the formation of hepatic epithelial polarization. As a TJ protein, the amount of OCLN on the cell surface was reported to be correlated with cellular polarity formation.21, 22 During the maturation of induced hepatocytes, more and more OCLN protein transferred to plasma membrane and formed intermittent belts at the hHB stage and sealed circles at the mature hHep stage (Fig. 3A). These data suggested that the location of OCLN, rather than its expression, might play an important role for hepatic epithelia in the transition from nonsusceptibility to susceptibility of HCV infection. In our study, epithelial polarization and TJ formation occurred with hESC maturation to hepatocytes and were speculated to be barriers to HCV infection, which was consistent with the published data of reduced susceptibility to infection following polarization.23-25

To further explore the host factors involved in HCV replication, assembly, and secretion, genome-wide gene expression in cells at different stages was compared using RNA sequencing. In general, the gene expression profile of the cells at each stage clustered closer to the HCV-infected cells at the same stage, which indicated that the genome profile of lineage specifications could not be affected significantly by HCV infection (Fig. 3B). Genes involved in the regulation of liver development were up-regulated along with processes of maturation (Fig. 3C). Among the genes positively associated with HCV replication,26 ABCB6, CEBPD, MAPK1, RAB5A, and UBE2J1 were highly expressed in mature hHeps, while CRLF3, ITGA7, and TFAP2A were highly expressed in the progenitor hHB stage (Fig. 3D), which indicated that HCV might exploit stage-specific host signal pathways that support its replication in immature and mature cells, respectively. HCV assembly and secretion have been reported to be associated with the lipoprotein secretion pathway and particularly with pathways mediated by apolipoprotein B (APOB) and APOE.20, 27 The expression of both APOB and APOE increased as demonstrated by quantitative RT-PCR and enzyme-linked immunosorbent assay (Supporting Fig. S3). Most of the 66 genes known to be involved in HCV assembly and secretion26 remained unchanged, except that CX3CL1 and PROX1 increased in mature hepatocytes, while MYO3A increased in immature hepatic cells.

Families of viral restriction factors involved in HCV infection10, 26, 28 were significantly enriched in hHeps (Fig. 3E), including IFN-stimulated effector genes (ISGs, e.g., IFIT1, oligoadenylate synthetase 2), leukocyte recruiting chemokines (e.g., chemokine [C-X-C motif] ligands 3 and 8), inflammatory genes (e.g., interleukin 1β, tumor necrosis factor), and recently reported protein tyrosine phosphatase family genes (e.g., PTPRH). Furthermore, 1,101 genes significantly up-regulated in hHeps versus hHBs were analyzed by the Kyoto Encyclopedia of Genes and Genomes. We found that the tumor necrosis factor, nuclear factor-κB, and Janus kinase–signal transducer and activator of transcription signaling pathways, which mediate host antiviral immune responses, were enriched. Interestingly, hepatitis C pathway genes were also significantly up-regulated in hHeps versus hHBs. This pathway contained five claudin family genes and 14 innate immune genes, including IFN pathway genes (e.g., IFNAR2, IFN regulatory factor 9, Janus kinase 1, signal transducer and activator of transcription 1, oligoadenylate synthetase 2) and inflammatory genes (e.g., tumor necrosis factor, chemokine [C-X-C motif] ligand 8) (Fig. 3F). Together, except for the specific physiological characteristics of the hepatic host cells at different lineage status, it was the comprehensive effect on both acceptance and defense to the virus (proviral and antiviral effects) that resulted in different HCV infection features. Low level of cell polarity, integrated TJ, and innate immunity might comprise the main reasons for hHB being the optimal stage of HCV infection.

VEGF ENHANCED HCV INFECTION BY MINIMIZING hHB POLARIZATION AND OBVIATING FURTHER MATURATION

Because HCV infectivity occurs optimally in hHBs, optimizing conditions for keeping the cells at the hHB stage should facilitate a highly effective HCV infection cell model system. Mature hepatocytes are highly polarized, with TJs separating the basolateral (sinusoidal) and apical (canalicular) domains. VEGF and the VEGF–receptor 2 pathway were reported to inhibit hepatocellular TJ integrity and cell polarity in the hepatocellular carcinoma cell line HepG2, through reorganization of OCLN.24 The VEGF–VEGFR2 pathway has also been reported to play an important role in liver regeneration, which suggests that the VEGF pathway could be significant for hepatic epithelial proliferation and differentiation.29 We hypothesized that VEGF might regulate HCV infection through managing cell polarity status.

RNA was extracted from HCV-infected cells treated with VEGF at 5 dpi or not treated. We found that upon VEGF treatment, the expression of intracellular HCV increased during each stage of hepatic lineage restriction (Fig. 4A). Infectivity assays using Huh7.5.1 cells infected with supernatants collected from cells of each stage showed that VEGF also enhanced the production of infectious HCV particles (Fig. 4B). HCV replication in cells was confirmed with immunofluorescence staining (Fig. 4C). We also used HCV-dependent relocalization assays30 to indicate HCV-infected hHBs and hHeps induced by the differentiation protocol with or without VEGF. hHBs and hHeps were transduced with IPS-NLS-EGFP reporters and then infected with HCV. Infected cells exhibited relocalization of fluorescence reported by protease NS3-4A activity. More translocation could be observed in VEGF-treated cells as early as 2 dpi (Fig. 4D). Transmission electron microscopy showed that VEGF increased the area occupied by the membranous web in HCVcc-infected HBs, without altering the morphology of the double-membrane vesicles, the site of genome replication (Fig. 4E),31 which meant that VEGF also could enhance HCV replication. Further mechanisms are expected in the future on this phenomenon.

The 5(6)-carboxy-2′,7′-dichlorofluorescein diacetate assay and canaliculus marker multidrug resistance protein 2 immunostaining were used to check the status of cell polarity. Less cell polarity was found in VEGF-treated cells at both the HB and hepatocyte stages (Fig. 5A). Immunoprecipitation assays showed there was more phosphorylated OCLN protein at the hepatocyte stage than at the HB stage, though a similar level of OCLN existed at both cell stages (Fig. 5B). In addition, when we added VEGF to the serum-free, stage-specific defined medium following the step-wise differentiation, the expression of fetal-specific AFP was up-regulated, while the hepatocyte functional genes, e.g., ALB, transferrin, AAT, and UGT1A3, were suppressed in both hHB and hHep cells by quantitative RT-PCR (Fig. 5C). Immunofluorescence staining showed that VEGF increased AFP-positive cells in all stages of the hepatic lineage cells, consistent with RNA level, and reduced the percentage of cytokeratin 18–positive cells, which indicated that VEGF could postpone hepatic maturation (Fig. 5D). Thus, VEGF treatment might inhibit cell polarity through phosphorylation of OCLN and, secondarily, minimize hepatic differentiation, maintain the cells at an immature stage, and enhance HCV infectivity.

SUSTAINED EXPRESSION OF HCV PROTEIN IN THE LIVER PARENCHYMA OF ENGRAFTED F/R MICE

Wild-type primary hepatocyte transplantation has been proven to rescue liver failure and death of Fah^–/– mice when they are removed from being fed with 2-nitro-4-trifluoromethylbenzoyl-1,3 cyclohexane dione, which provides an ideal model to characterize in vivo repopulation and functions of hepatic cells as well as construction of a humanized murine model for HCV infection.32 In our study, we used immune-deficient F/R mice33 for transplantation to avoid immune rejection and enhance the chimera efficiency. We injected hESC-derived hHBs infected with HCV or not into the left and right lower lobes directly. Immunosuppressor FK506 was used to raise engrafted cell repopulation.34 The survival rate and hALB in the sera of the recipient mice were monitored to evaluate the engraftment procedures. F/R mice without transplantation were all dead within 5 weeks after withdrawal of 2-nitro-4-trifluoromethylbenzoyl-1,3 cyclohexane dione. In contrast, 12 out of 16 F/R mice transplanted with hHBs were alive for more than 12 weeks, no matter the transplanted hHBs with or without infected HCV (Fig. 6A). The results indicated that hESC-derived hHBs can be functionally integrated into F/R mice and can rescue them from liver failure.

Hematoxylin/eosin staining of liver sections from mice receiving hHBs showed clusters of donor hepatic cells over the entire lobes, demonstrating that engrafted hHBs could repopulate the F/R murine livers effectively, consistent with findings from a previous study35; vacuolar degeneration appeared in liver sections from mice in the negative control group (Fig. 6C). IHC staining of human Fah showed that HCV-infected or uninfected hHBs repopulated 28.36 ± 1.73% and 25.05 ± 1.26% of the liver parenchyma, respectively, in the surviving mice. Though the repopulation rates were similar, most of the engrafted HCV-hHBs were hAFP-positive, indicating a fetal phenotype rather than a mature one over the time in the mouse liver. By contrast, hAFP-positive cells were fewer in recipient mice transplanted with uninfected hHBs (Fig. 6D), which was consistent with the higher hALB in serum of those mice (Fig. 6B). Together, the results suggested that the maturation of HCV-infected hHBs was much slower than that of the uninfected hHBs in vivo. We found a similar phenomenon in the liver tissues of HCV patients (Supporting Fig. S2). In mice, Fah-positive cells were positively stained for the HCV core protein after repopulation, which means that the HCV persisted in the engrafted HCV-hHBs. However, HCV RNA and protein are undetectable in the sera or liver tissue. Together, our data suggest that HCV-infected HBs can repopulate F/R livers and ameliorate impaired liver functions caused by Fah deficiency. HCV was maintained in engrafted cells for at least 12 weeks.

IFNβ-MEDIATED INNATE IMMUNE RESPONSES SUPPRESSED THE INFECTIVITY OF HCV AND SMALL INTERFERING hIFNβ ENHANCED HCV INFECTION IN THE LIVERS OF ENGRAFTED F/R MICE

Although cells at the hHB stage were susceptible to HCV infection, the replication of HCV in hHBs kept decreasing, with HCV RNA copies stabilizing at around 10⁴ after 14 days of infection. hHB cells could be well maintained without shifting to any other stages by using the defined medium (Fig. 7A). The host antiviral immunity was often regarded to be the major reason for the decrease of HCV replication. In addition to the increase of immune response genes and the related pathways shown in Figs. 3E and 7B, expression of IFN-inducible protein 10, interleukin-8, IFNβ, ISG54, and ISG56 increased in the hHBs after infection with HCV, which indicated a gradual activation of inflammation stimulation and IFN responses, respectively (Fig. 7C). Spreading of the HCV infection in primary human fetal liver cells was also reported to be limited by the IFN-stimulated mechanisms.36 To enhance HCV replication in hHBs, we suppressed the expression of IFNβ with small interfering RNA (siRNA) to counteract IFN induction and antiviral signaling. The addition of siRNA suppressed the induction of hIFNβ and downstream genes ISG54 and ISG56, and intracellular HCV RNA was enhanced in hHBs (Fig. 7D; Supporting Fig. S4). As expected, HCV RNA in mouse liver tissue engrafted with HCV-infected hHBs was detectable, which showed enhanced HCV infection in vivo (Fig. 7E,F). Taken together, introduction of sihIFNβ could improve HCV infection in vitro and in vivo.