Single-cell analysis of mature hepatocytes reveals an IRF1-driven restriction of HDV infection

Abstract

Background & Aims

Stem cell-derived hepatocyte-like cells (HLCs) are an in vitro model of hepatocytes reproducing mature hepatic functions. However, heterogeneous or imperfect differentiation may limit their biological relevance. HLCs are susceptible to all primary hepatitis viruses, including HDV. Importantly, HLCs support limited HDV replication, at a significantly lower level than hepatoma cell lines.

Methods

We used single-cell RNA sequencing (scRNAseq) to analyse control and HDV-infected HLCs. We assessed maturation and heterogeneity of the HLCs population. We visualised viral genomic and antigenomic sequences abundance and innate immune response at the single-cell level. Further functional characterisation was performed in HLCs and hepatoma cell lines.

Results

HLCs form a population of hepatocytes exhibiting various levels of maturation, with a minor hybrid population of myofibroblast/hepatocyte cells associated with immature phenotype. Upon HDV inoculation, Interferon-Stimulated Genes expression was induced in infected cells, but not in bystander HLCs. Moreover, Interferon Regulatory Factor 1 (IRF1) was enriched in infected HLCs with undetectable levels of viral genome replication, suggesting it may restrict viral infection. Decreasing IRF1 expression in HLCs by 50% improved susceptibility to HDV infection by 10-fold (p <0.01). Moreover, IRF1 overexpression in hepatoma cell lines restored physiological basal level of IRF1 effector antiviral genes and inhibited HDV infection (∼50% reduction after 7 days, p <0.01). Importantly, in IRF1 expressing cells, cell division-mediated spread was inhibited and infection was significantly decreased after 2 weeks of culture (>1.5 log decrease, p <0.001).

Conclusions

scRNAseq of HLCs identified IRF1 as a potent restriction factor of HDV infection, through an antiviral mechanism blocking HDV infection at an early step of the viral cycle.

Impact and implications

We performed single-cell RNA sequencing of control and HDV-inoculated stem cell-derived hepatocytes, and characterised them in terms of hepatic differentiation, viral abundance and response to infection. We identified IRF1 as a constitutive cellular factor restricting HDV infection in mature hepatocytes, particularly targeting HDV in the cytoplasm. This work improves our understanding of mature cell culture models for HDV, needed to decipher its host–pathogens interactions. Identification of antiviral effector genes opens the way to the development of new host targeted antiviral strategies, particularly for targeting cell division-mediated spread that is not inhibited by currently used entry inhibitors (Hepcludex).

Graphical abstract

Highlights

• •We performed single-cell RNA sequencing of control and HDV-inoculated hepatocyte-like cells.
• •We characterised the cell population in terms of hepatic differentiation and aberrant gene expression.
• •Sorting the inoculated cells based on viral abundance shows that only infected cells mount an innate immune response.
• •Infected hepatocyte-like cells with no detectable viral genome replication show an enriched expression of IRF1.
• •IRF1 overexpression in Huh7-NTCP restricts HDV infection at a cytoplasmic stage of the viral cycle.

Introduction

Hepatocyte-like cells (HLCs) are an in vitro mature model of hepatocytes reproducing key features of primary adult hepatocytes (reviewed by Blaszkiewicz and Duncan¹). HLCs are differentiated in vitro from human pluripotent stem cells (hPSCs), embryonic or induced (hESCs and hiPSCs), using protocols based on sequential treatment with growth factors and small drugs.² Several protocols of differentiation have been published, but levels of hepatic maturation vary from one to the other. Moreover, maintenance or induction of genes not associated with hepatic phenotype may limit their relevance. For example, a recent single-cell RNA sequencing (scRNAseq) study revealed that HLCs display a hybrid phenotype, with induced expression of genes associated with both hepatic and intestine differentiation.³

Despite these limitations, HLCs have been shown to be a relevant in vitro model to study hepatitis viruses infection in vitro,⁴ particularly to assess mature cellular processes, such as innate immunity, that are poorly reproduced in transformed hepatoma cell lines.^5,6 HLCs are permissive to all primary hepatitis viruses: HAV,⁷ HBV,⁸ HCV,^9,10 and HEV.¹¹ Recently, we described that HLCs are susceptible to HDV infection in vitro.¹²

HDV, a satellite virus of HBV, is the smallest RNA virus to infect humans, with a strict tropism for hepatocytes. HDV relies on HBV to form its envelope, process cellular egress, and enter the hepatocyte using the Na⁺-taurocholate co-transporting polypeptide (NTCP) as receptor (reviewed by Urban et al.¹³). HDV infection occurs as a co- or super-infection of patients who were chronically infected with HBV, and the HDV/HBV infection leads to the most severe form of viral hepatitis, the hepatitis delta virus.¹⁴ Interferon (IFN) alpha remains the main antiviral treatment, but rarely allows viral clearance. Recent therapeutic developments efficiently target viral entry (bulevirtide [Hepcludex]), but to date there is no curative treatment for HDV.[15], [16], [17] It emphasises the need to study host–pathogen interactions of hepatotropic viruses in the cellular context of mature hepatocyte models to identify mechanisms able to restrict viral replication. Importantly, in vitro HDV susceptibility seems to correlate negatively with cell maturity.⁵ HLCs are characterised by a low in vitro susceptibility to HDV infection, similar to mature primary cultured human hepatocytes (PHHs)¹² and dHepaRG,¹⁸ whereas immature IFN-incompetent hepatoma cell lines, upon overexpression of NTCP, support higher levels of viral replication.^12,18 This suggests that cellular maturity and intrinsic immunity may restrict viral replication in mature models.⁶ Although HBV infection of mature hepatocytes does not trigger an innate immune response, HDV triggers a strong cellular response in differentiated HepaRG (dHepaRG), PHHs^19,20 and HLCs.¹² This innate immune response relies on triggering of the RIG-I-like receptor (RLR) MDA5 and induction of interferon-stimulated genes (ISGs). Although this IFN-based immune response efficiently inhibits HBV replication, HDV seems mostly unaffected.²¹

In addition to the IFN-based innate immune response, hepatocytes display a constitutive innate immunity driven by interferon regulatory factor 1 (IRF1). Primarily known as a nuclear factor promoting type I IFN expression, IRF1 is an ISG. It is upregulated upon pathogen-associated molecular pattern (PAMP) recognition (reviewed by Feng et al.²²), and participates in potentiating interferon regulatory factor 3 (IRF3)-mediated immune gene activation. However, in the absence of stimulation, hepatocytes constitutively express IRF1, which drives a basal expression of so-called IRF1 effector genes in an IFN-independent manner. Among these effector genes are genes involved in lipid metabolism or protein degradation, but also conventional ISGs.^23,24 This IRF1-driven immunity renders hepatocytes more resistant to infection by several RNA viruses, among them HAV and HCV.^23,25 Although HDV is insensitive to the IFN-based innate immune response in hepatocytes, the antiviral capacity of the IRF1 constitutive immunity on HDV has not been studied.

In this study, we use scRNAseq to investigate the heterogeneity of cellular phenotype and viral abundance of both control and HDV-inoculated HLCs. We assessed at the single-cell level the innate immune transcriptomic profile of HLC after HDV mono-infection. Moreover, we studied the basal innate immune program of HLCs and identified IRF1 as a restriction factor limiting HDV infection in mature hepatocytes.

Materials and methods

Cellular biology

HLCs were differentiated in vitro according to our published protocols.²⁶ For HDV infection, HLCs were inoculated at Multiplicity of Infection (MOI): 0.5, in presence of polyethylene glycol (PEG), as previously described.¹² For dissociation, HLCs were treated at 37 °C with EDTA 1 mM for 3 min, followed by 0.05% trypsin-EDTA for 15 min. After washes, single HLCs were resuspended in PBS 0.1% BSA.

Single-cell RNA sequencing and data analysis

Resuspended HLCs were processed following the instructions of the Chromium Next GEM Single Cell 3′ Kit (v3.1, Dual Index) on a Chromium Controller (10x Genomics, Leiden, the Netherlands). Briefly, we captured intracellular RNAs using poly-dT probes. Captured RNAs were tagged with unique molecular identifiers (UMIs), reverse transcribed and amplified. Libraries were prepared following manufacturers’ instructions. The final cDNA libraries were sequenced using NovaSeq 6000 PE50 Illumina sequencing (1,000 million reads per library, 100,000 reads per cell).

Sequencing data were analysed using the 10 × Cell Ranger pipeline according to the manufacturer’s instructions. Reads were aligned to the human genome reference GRCh38-2020. Importantly, we added the sequence of the positive and negative strand sequence of the HDV genome (GenBank: M55042.1) as extrachromosomal features in a custom reference to identify infected cells and discriminate the HDV mRNA from the single-stranded RNA genome. Results were visualised using the 10 × Loupe Browser software. Violin plots depict scRNAseq gene expression as log²-transformed raw UMI counts (log₂ exp) or raw UMI counts normalised to RNA content (log norm exp).

Detailed Materials and methods can be found in the Supplementary data and Supplementary CTAT table.

Results

Stem cell-derived HLCs can be successfully analysed by scRNAseq

We resuspended control HLCs and performed high-throughput unbiased separation of single cells in emulsion using the Chromium Next GEM Single Cell 3′ kit (Fig. 1A). Quality control post Illumina sequencing shows that 8,507 cells were successfully sequenced, with a mean of 144,199 reads and 2,410 genes per cell (Fig. 1B). Analysis at the single-cell level of selected hepatic markers alpha-fetoprotein (AFP), albumin (ALB) and SLC10A1, encoding the NTCP transporter (Fig. 1C), are consistent with our bulk analysis of non-dissociated HLCs, both in terms of average gene expression, compared with RNAseq (Fig. 1D), and percentage and heterogeneity of positive cells, compared with immunostaining (Fig. 1E). In conclusion, we successfully dissociated HLC populations and ran them through the 10 × scRNAseq pipeline, to analyse their transcriptomic program at the single-cell level.

Fig. 1.

HLCs dissociation and scRNAseq run validation. (A) Successful dissociation of HLCs using EDTA and trypsin (magnification: 20 × , scale bar: 100 μm); and association with capture beads in emulsion. (B) T-distributed stochastic neighbour embedding (t-SNE) distribution and unique molecular identifier (UMI) counts of HLCs run through the Chromium Controller and sequenced using Illumina sequencing. Analysis of expression of selected hepatic markers by (C) scRNAseq (log norm exp), (D) RNAseq (RPKM value, average of three preparations) and (E) immunostaining (magnification: 10 ×, scale bar: 200 μm) confirms successful analysis of HLCs at the single-cell level. AFP, alpha-fetoprotein; ALB, albumin; HLC, hepatocyte-like cell; NTCP, Na⁺-taurocholate co-transporting polypeptide; scRNAseq, single-cell RNA sequencing; SLC10A1, soluble carrier family 10 member 1 (NTCP); t-SNE, t-distributed stochastic neighbour embedding; UMI, unique molecular identifier.

scRNAseq reveals a hybrid HLCs phenotype

All HLCs express hepatic markers such as AFP, ALB, TTR, and SERPINA1 (Alpha 1 Antitrypsin). SLC10A1 is expressed in 51% of the cells (Fig. 2A). HLCs expressed only low levels of cholangiocyte-associated genes (Fig. 2B). To investigate misguided differentiation, we applied a non-directed approach described by Nell et al.,³ combining RNAseq and scRNAseq analysis. First, we compared the RNAseq transcriptomic profiles of our hPSCs, differentiated HLCs, and PHHs using a differentiation pattern plot (DiPa) (Fig. 2C). A DiPa plot visualises gene expression changes during hepatic differentiation, and compares the final expression in HLCs with PHHs. Individual genes are then sorted within 11 differentiation pattern groups (DPG 0–10) based on their expression in HLCs compared with PHHs (Table S1). Protein–protein interaction (PPI) network analysis shows that genes accurately upregulated during in vitro differentiation (DPG 1–3) are significantly enriched for genes associated with liver and hepatic metabolism (Fig. 2D). DPG 5 lists genes that are adversely upregulated in HLCs compared with PHHs (‘misguided differentiation’). In the Nell et al.³ study, DPG 5 was enriched for intestine differentiation markers. In our data set, PPI network analysis shows that intestine-associated genes are not adversely overexpressed in our HLCs compared with PHHs (DPG 1-3) and are expressed at low level throughout the HLC population (Fig. 2E). Our DPG 5 PPI analysis rather identified genes associated with cell adhesion and collagen synthesis (Fig. 2D). In our scRNAseq data set, cells expressing these divergent genes clustered separately (Fig. 2F, Fig. S1A). Genes distinguishing this cluster were associated with fibroblasts and hepatic stellate cells (HSCs) (Fig. S1B). Importantly, these cells also expressed hepatic markers (Fig. S1C): their global transcriptomic profile is reminiscent of HSCs (tissue expression database [BTO: 0002741], false discovery rate [fdr]: 0.0157) and myofibroblasts ([BTO: 0001946] fdr: 0.0033) but first of all of hepatocytes ([BTO: 0000759], fdr: 1.18E-119), consistent with a hybrid phenotype. Importantly, expression of myofibroblast markers decreased during HLCs maturation (Fig. 1D), suggesting that this hybrid phenotype is associated with HLCs immaturity. Altogether, our data show that the analysed cell population consists of HLCs at different levels of maturation, but also contains an immature hepatocyte/myofibroblast hybrid population.

Fig. 2.

HLCs population characterisation. ScRNAseq, t-SNE projection, and gene expression (log normalised to total UMI count) of (A) hepatic and (B) cholangiocytes markers was assessed in our HLC population. (C) Differentiation pattern (DiPa) plot of global transcriptomic programs of HLCs compared with hPSC and PHHs. (D) Protein–protein interaction analysis of differentiation pattern groups (DPG) 1–3, showing genes accurately upregulated during in vitro differentiation, and DPG 5 listing genes adversely upregulated during in vitro differentiation of HLCs. Log normalised expression of (E) intestine-associated and (F) myofibroblasts-associated genes within our HLCs population. AFP, alpha-fetoprotein; ALB, albumin; DPG, differentiation pattern group; HLC, hepatocyte-like cell; hPSCs, human pluripotent stem cells; PHH, primary culture of human hepatocytes scRNAseq, single cell RNA sequencing; SLC10A1, soluble carrier family 10 member 1 (NTCP); t-SNE, t-distributed stochastic neighbour embedding; UMI, unique molecular identifier.

Only HDV-infected HLCs mount a strong innate immune response to infection

To circumvent the low percentage of infected cells¹² and to assess the heterogeneity of cellular responses to infection, we analysed HLCs by scRNAseq 3 days post HDV inoculation, when they display their highest level of intracellular HDV RNA.¹² An estimated 7,608 HLCs were sequenced, with a mean of 159,907 reads and 1,850 genes per cell (Fig. 3A). Overall, the level of hepatic differentiation and global gene expression of inoculated HLCs did not differ from that of control (ctrl) HLCs (Fig. 3B) (Fig. S2A and B). Only a small group of genes were overexpressed in inoculated cells, among them ISGs (Fig. S2B). HDV sequences were detected in ∼2% of the cells (Fig. 3C, Fig. S2C), consistent with previous immunofluorescence assay-based results.¹² Importantly, these captured sequences mostly aligned with the hepatitis delta antigen (HDAg) coding sequence, consistent with capture of the polyadenylated HDAg mRNA (Fig. S3A). We sorted the HLCs based on HDV antigenome abundance, from negative ‘bystander’ cells to low (up to 5 UMI count/cell) and high levels (>5 UMI count/cell) of captured intracellular HDV antigenome, and compared them with the ctrl non-inoculated HLCs (Fig. 3C). The different viral load-based populations did not differ in terms of hepatic differentiation (Fig. 3D). SLC10A1 expression was enriched in the infected population, consistent with the role of NTCP as viral receptor (Fig. 3D). We then investigated the induction of ∼400 ISGs²⁷ in our HDV-positive HLCs, compared with bystander HLCs and control non-inoculated HLCs. Genes of canonical ISGs such as MXA, Interferon-Stimulated Gene 15 (ISG15), and the pattern recognition receptor (PRR) MDA5 (IFIH1) are expressed proportionally to HDV antigenome abundance (Fig. 3E, Fig. S5A and B). Interestingly, bystander HLCs express ISGs at a similar level to non-inoculated HLCs suggesting that there is very limited paracrine effect of IFN secretion by infected HLCs, and that the virus replication itself drives the ISG induction. Moreover, we describe the pattern of ISG induction in infected cells (Fig. 3F, Table S2), characterised by overexpression of canonical ISGs such as IFI6, MX1, OAS1-3, IFITs, RSAD2, IFITM1, BST2, IFI27, among others (Fig. 3G).

Fig. 3.

scRNAseq analysis of HDV-infected HLCs. (A) t-SNE, UMI count distribution and (B) hepatic genes expression (log normalised to total UMI count) of HDV-inoculated HLCs population at the single-cell level. (C) Detection of the HDV polyadenylated mRNA within sequenced HLCs and sorting based on abundance of the HDAg mRNA. (D) Hepatic markers and (E) ISGs expression based on viral RNA abundance, compared with control non-inoculated HLCs. Colour legend of panel C also applies to the violin plots of panels D and E. (F) Differential expression of ∼400 ISGs²⁷ in HDV-inoculated HLCs, based on HDV RNA abundance. (G) Heatmap of ISGs upregulated in HDV-infected HLCs. AFP, alpha-fetoprotein; ALB, albumin;HDAg, hepatitis delta antigen; HLC, hepatocyte-like cell; ISG, Interferon-Stimulated Gene; scRNAseq, single-cell RNA sequencing; SLC10A1, soluble carrier family 10 member 1 (NTCP gene); t-SNE, t-distributed stochastic neighbour embedding; UMI, unique molecular identifier.

HDV genome capture reveals IRF1 enrichment in infected HLCs with no detectable HDV antigenome

Previous studies showed that poly-dT probes can also capture non-polyadenylated viral RNA.²⁸ Similarly, we could identify mapped reads that did not belong to the HDV mRNA (Fig. S4A and B), including HDV genomic RNA sequences in 0.7% of the cells (Fig. 4A, red). We confirmed that, as expected,²⁹ HDV genomic sequences are more abundant in infected HLCs than HDV antigenomic sequence (Fig. S3B). Coverage analysis (Fig. S4A) confirms full-length sequencing of the HDAg mRNA (Fig. S4B and C). Interestingly, coverage outside of the HDAg mRNA correlated with A-content (Fig. S4D and E), suggesting that poly-dT probes capture not only poly-A tails, but also A-rich viral sequences. Overall, levels of captured HDV genomic sequences correlated with levels of captured HDV antigenomic sequences (Fig. 4B).

Fig. 4.

HDV genomic RNA capture and viral strain specific ISG response. (A) t-SNE projection of HDV-inoculated HLCs highlighting cells with detectable HDV mRNA (green) and HDV genome (red). (B) Correlation between abundance of HDV mRNA and HDV genome at the single-cell level. (C) Cell sorting based on the profile of viral form detection and SLC10A1 expression. Colour legend of panel C also applies to the violin plots of panels D and E. (D, E) Expression of selected ISGs in cells based on their viral form capture. HLC, hepatocyte-like cell; ISG, Interferon-Stimulated Gene; IRF1, interferon regulatory factor 1; t-SNE, t-distributed stochastic neighbour embedding; SLC10A1, soluble carrier family 10 member 1.

We sorted the inoculated HLCs based on the detected viral polarity, from non-infected bystander cells (Fig. 4C, grey), to cells with only detectable HDV antigenome/mRNA (Fig. 4C, green), and cells positive for both HDV antigenome and genome (Fig. 4C, brown). We also detected a smaller population of HLCs in which the viral genome was detectable but no poly-A mRNA or HDV antigenome could be captured (Fig. 4C, red). These HLCs subgroups express NTCP at similar level (Fig. 4C). The infected HLCs, in which no antigenomic viral sequences were detected display low conventional ISG induction, consistent with low or no viral genome replication (Fig. 4D). However, a small panel of immune genes is enriched in these cells (Fig. 4E, brown). We assessed expression of these genes in control PHHs, HLCs, and hepatoma cell lines (Fig. S6A). Among them, only IRF1 showed a level of basal expression inversely correlated with HDV susceptibility, suggesting it may participate in restricting infection. IRF1 is expressed in every HLCs subgroup, but the HDV RNA(+)/HDAg mRNA(-) group (red) shows a higher percentage and higher expression compared with the other groups (Fig. S6B). Moreover, the very limited induction of canonical ISGs in the red group suggests that IRF1 expression in this group is not induced by an IFN-based antiviral response. Accordingly, we did not detect expression of IFN in these cells (data not shown).

Constitutive expression of IRF1 and effector genes in HLCs restricts HDV infection

Various cell types constitutively express IRF1 and IRF1 effector genes (reviewed by Feng et al.²² ), among them genes associated with lipid metabolism, protein degradation and, notably, canonical ISGs and antiviral effectors such as MX1, BST2, IFIH1, IFIT2/3, and RNASEL. Our HLCs constitutively express IRF1 (Fig. 5A). Immunostaining confirms that IRF1 is located in the nuclei of most unstimulated HLCs (Fig. 5B), driving the expression of a panel of known IRF1 effector genes (Fig. 5A). Importantly, constitutive expression of IRF1-driven ISGs is absent in Huh7, whereas PHHs express them more similarly to HLCs (Fig. 5A). Of note, hPSCs express some ISGs, particularly IFITMs, to protect themselves against infection.³⁰ Accordingly, during differentiation, expression of pluripotency associated IFITMs decreased (Fig. 5C;Fig. S7, DPG 8–10, green), while expression of IRF1 and IRF1 effector ISGs increases during hepatic differentiation (Fig. 5D; Fig. S7, DPG 3–5, red).

Fig. 5.

Basal expression of IRF1 and IRF1 effector genes in HLCs. (A) RNAseq analysis of expression of IRF1 and selected effector genes in three individual preparations of HLCs, Huh7, and PHHs. (B) Immunostaining for IRF1 and ALB in control HLCs, using z-stack and 3D reconstruction (magnification: 100 × , scale bar: 50 μm). (C) Expression of selected ISGs during in vitro hepatic differentiation (Day 0 to Day 15) and HLCs maintenance (Day 15 to Day 21). (C) Expression of selected IRF1 and selected effector ISGs during in vitro hepatic differentiation (Day 0 to Day 15) and HLCs maintenance (Day 15 to Day 21), compared with PHHs. ALB, albumin; HLC, hepatocyte-like cell; hPSCs, human pluripotent stem cells; IRF1, interferon regulatory factor 1; PHH, primary culture of human hepatocytes.

IRF1 siRNA transfection in HLCs led to a 50% decrease of IRF1 expression (Fig. 6A), and resulted in a decrease in expression of IRF1 effector genes, including ISGs such as MX1 (Fig. 6A, bottom). Upon HDV inoculation, HLCs pre-treated with anti-IRF1 siRNA exhibited a higher susceptibility to viral infection (Fig. 6B). Moreover, this phenotype could be maintained over time (Fig. 6C).

Fig. 6.

IRF1 knockdown increases HLCs susceptibility to HDV. HLCs were treated with IRF1-targeting siRNAs or scramble control siRNA for 24 h and inoculated with HDV (MOI 0.5). (A) IRF1 and IRF1 effector gene MX1 expression after siRNA transfection measured by RT-qPCR in non-infected HLCs after 48 h. (B) IRF1 expression (top) and intracellular HDV RNA (bottom) at d5pi, measured by RT-qPCR. (C) IRF1 expression (top) and intracellular HDV RNA (bottom) assessed over the course of a 7-day infection period, by RT-qPCR. Levels of significance: ∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001; ∗∗∗∗p <0.0001 (ordinary one-way or two-way ANOVA, with Tukey correction for multiple comparisons). GADPH, glyceraldehyde 3-phosphate dehydrogenase; HLC, hepatocyte-like cell; IRF1, interferon regulatory factor 1.

Taken together, these results show that hepatic maturity is associated with an IRF1-driven intrinsic immunity that partially restricts HDV infection in immunocompetent HLCs.

IRF1 overexpression restores antiviral mechanisms in Huh7-NTCP

We previously showed that ectopic expression of IRF1 in Huh7.5 restores expression of a panel of ∼100 IRF1 effector ISGs at a level similar to their basal expression in PHHs²⁵ and in HLCs (Fig. 7A). In this context, we restored IRF1 expression in Huh7-NTCP, an IFN-incompetent model susceptible to HDV infection. IRF1 overexpression in Huh7-NTCP was confirmed by RT-qPCR (Fig. 7B) and Western blot (Fig. 7C). Moreover, immunostaining confirmed IRF1 location in the nucleus of transduced cells, indicating transcriptional activity (Fig. 7D). Consistently, IRF1 overexpression raised expression of effector genes, but not of IFNs (Fig. S8A), to a level similar to their basal level in HLCs (Fig. 7D). Moreover, IRF1 overexpression partially restored sensitivity to PAMPs, leading to higher induction of IFNB and L but only moderate upregulation of ISGs upon HDV infection or poly(I:C) transfection, in Huh7-NTCP-IRF1 compared with Huh7-NTCP (Fig. S8B).

Fig. 7.

IRF1 overexpression in Huh7-NTCP restricts HDV infection. (A) IRF1 induced transcripts in Huh7.5, Huh7.5-IRF1, PHHs, and HLCs, normalised to Huh7-NTCP, mean fold changes (log₂) of three RNAseq experiments. (B) Validation of IRF1 overexpression in Huh7-NTCP by RT-qPCR (left) and Western blot (right). (C) Immunostaining for IRF1 and ALB in Huh7-NTCP and Huh7-NTCP-IRF1 (magnification: 4 ×, scale bar: 100 μm). (D) Relative expression of selected IRF1-driven ISGs in Huh7-NTCP, Huh7-NTCP-IRF1, and HLCs, measured by RT-qPCR, normalised to Huh7-NTCP, n = 3–9. (E) Intracellular HDV RNA assessed by RT-qPCR of Huh7-NTCP and Huh7-NTCP-IRF1, from d1 every 2 days post inoculation with HDV (MOI 0.5). (F) Intracellular HDV RNA (left) and IRF1 expression (right) measured by RT-qPCR in Huh7-NTCP (blue) and Huh7-NTCP-IRF1 (green). Huh7-NTCP-IRF1 cells were treated with either a control scramble or IRF1-targeting siRNA for 24 h, then inoculated with HDV (MOI 0.5), gene expression was assessed on d3pi. (G) Quantification of HDAg-positive cells at d5pi after HDV inoculation of Huh7-NTCP and Huh7-NTCP-IRF1 cells with different MOIs and visualisation by immunostaining using ImageJ, representative images for MOI 1 are depicted on the left (scale bar: 100 μm). Levels of significance: ∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001 (ordinary one-way ANOVA, with Tukey correction; except for panel B: unpaired t test; and panel G: two-way ANOVA with Sidak correction). ALB, albumin; HDAg, hepatitis delta antigen; HLC, hepatocyte-like cell; IRF1, interferon regulatory factor 1; NTCP, Na⁺-taurocholate co-transporting polypeptide; PHH, primary culture of human hepatocytes.

Huh7-NTCP-IRF1 cells were less susceptible to HDV infection in vitro (Fig. 7E). This antiviral effect was reversed by anti-IRF1 siRNAs (Fig. 7F). Additionally, we could observe up to 80% reduction in the percentage of HDAg-positive Huh7-NTCP-IRF1 cells (Fig. 7G). To further elucidate the antiviral activity of IRF1, we studied HDV maintenance over long-term multi-passage culture of Huh7-NTCP, as recently described.³¹ Although Huh7-NTCPs maintain HDV infection and support cell division-mediated spread, Huh7-NTCP-IRF1 inhibited viral replication over time (Fig. 8A). Single-cell clonal expansion of HDV-infected Huh7-NTCP-IRF1 shows no formation of foci of HDAg-positive cells, confirming that IRF1 overexpressing cells did not support cell division-mediated spread (Fig. 8B).

Fig. 8.

Cell division-mediated spread of HDV infection in Huh7-NTCP and Huh7-NTCP-IRF1. (A) Huh7-NTCP and Huh7-NTCP-IRF1 cells were inoculated with HDV (MOI 1) and passaged on d5pi (dilution 1:6) for two passages, each after a further 5 days. At every time point cells were assessed for HDAg-positive cells via immunostaining and quantification with ImageJ (N = 2 experiments with ≥6 images per experiment) (left, middle) or intracellular HDV RNA by RT-qPCR (right, N = 2) (magnification: 4 ×, scale bar: 300 μm). (B) Huh7-NTCP and Huh7-NTCP-IRF1 cells were inoculated with HDV (MOI 1) and passaged on d5pi in a dilution of 1:800 to allow clonal expansion. Cells stained for HDAg 7 days post passage (magnification: 4 × , scale bar: 100 μm). Protocol adapted from Zhang et al.³¹ Levels of significance: ∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001; ∗∗∗∗p <0.0001 (two-way ANOVA, with Tukey correction). HDAg, hepatitis delta antigen; IRF1, interferon regulatory factor 1; NTCP, Na⁺-taurocholate co-transporting polypeptide.

Altogether, we show that ectopic expression of IRF1 in IFN-incompetent Huh7-NTCP cells restores physiological constitutive expression of antiviral ISGs, sufficient to inhibit HDV infection and cell division-mediated spread of HDV.

Discussion

Heterogeneity of HLCs has mostly been investigated using batch assays.[32], [33], [34], [35] Recently, Nell et al.³ published the first characterisation of HLCs by scRNAseq. They revealed that the HLCs population consists of overall one cluster of cells, but that HLCs display a hybrid state with concomitant expression of hepatic- and intestine-associated genes. Here, we applied the same approach: although we found intestine-associated genes expressed in our HLCs, they were not adversely expressed compared with PHHs. However, we found that some of our HLCs exhibit a hybrid state between hepatocytes and myofibroblasts/HSCs (Fig. 2C–E). Fibroblasts being of mesodermal origin, finding them within endoderm-derived hepatocytes could be counterintuitive. Epithelial-to-mesenchymal transition of hepatocytes into fibroblasts has been hypothesised to occur naturally in the liver and to contribute to liver pathogenesis (reviewed by Lovisa³⁶). However, temporal follow-up of gene expression (Fig. S1E) shows that our HLCs do not undergo an epithelial-to-mesenchymal transition in vitro. The fibroblast-like state is rather associated with hepatoblasts and immature HLCs, with myofibroblast gene expression decreasing during hepatic maturation. Interestingly, Li et al.³⁷ described that hESCs undergoing in vitro hepatic differentiation go through a sequential epithelial-to-mesenchymal-to-epithelial transition, which is consistent with our data. The difference of hybrid phenotype between our HLCs and the HLCs from Nell et al.³ could be explained either by the use of two different differentiation protocols, the level of hepatic maturation reached, or by the origin and epigenetic memory of the pluripotent stem cell used (embryonic vs. induced hPSCs). Our two studies highlight the importance to characterise HLCs population at the single-cell level to assess cell-to-cell variability and optimise protocols of differentiation adapted to each individual hPSC population.

Importantly, we performed scRNAseq analysis of HDV-inoculated HLCs. Innate immune response to HDV infection has so far been studied only using batch analysis.^12,19,31,38 Recently, Lucifora et al.²⁰ studied the cellular response of dHepaRG and PHHs by RNAseq,²⁰ describing for the first time the full panel of ISGs induced upon HDV infection of immune competent cells. Here, we took advantage of the efficient capture by poly-dT probes of both cellular mRNAs and HDAg mRNAs to sort the cells based on their viral mRNA abundance. We describe that HDV replication itself, and not a paracrine effect of IFN on bystander cells, is the main driver of ISG induction in HLCs. Bystander HLCs display a level of immune activation closer to the one of control non-inoculated HLCs, whereas in HDV-infected HLCs, viral replication drives the expression of a panel of canonical ISGs reminiscent to the one described by Lucifora et al.²⁰ Previous works, including in HLCs,¹² showed that the triggered innate immune response does not inhibit HDV replication. Consistently, our scRNAseq showed that viral replication correlated positively with ISGs induction. This is different from a previous scRNAseq analysis of West Nile Virus (WNV) infection of IFN-competent fibroblasts in which antiviral ISGs abundance was inversely correlated with viral RNA abundance,³⁹ because of the antiviral activity of the triggered innate immune response on WNV.

As described for the non-polyadenylated Dengue virus,²⁸ the capture beads were able to capture some HDV RNA genome. Despite a lower efficient capture because of the lack of a poly-A tail, we captured HDV genomic sequences in 0.7% of the HLCs. Importantly, HDV genome abundance correlated with HDAg mRNA titre. Cells in which both mRNA and genome could be detected also displayed the strongest innate immune activation, consistent with active viral genome replication leading to viral genome accumulation and driving innate activation. In addition to these infected HLCs, we observed a peculiar population of HLCs with detectable HDV genome but no detectable poly-adenylated HDAg mRNA or HDV antigenome. These infected cells were characterised by a low ISG induction, consistent with low or no viral replication, but displayed enrichment of a few specific innate immune genes, among them IRF1.

IRF1 is mostly known as an ISG modulating the cellular IFN response upon PRR triggering or IFN stimulation.²² However, IRF1 basal expression itself also exerts antiviral activity in hepatocytes independently from PRR or IFN activation, by driving constitutive expression of antiviral effectors.²³ IRF1 constitutive antiviral immunity has been poorly studied, because of its absence in hepatoma-derived cell lines. Similar to hepatocellular carcinoma tissue,⁴⁰ transformed hepatoma cell lines exhibit a repressed IRF1 expression (reviewed by Chen et al.⁴¹). Because IRF1 knockdown significantly increased susceptibility to HDV infection in HLCs, we restored IRF1 basal expression in Huh7-NTCP, which led to significantly decreased HDV replication (Fig. 7). Cellular mechanisms of HDV control are poorly understood. Zhang et al.³¹ showed that HDV replication is inhibited only during cell division of IFN-competent cells, but not in dividing IFN-incompetent Huh7. Here we show that restoring IRF1 constitutive expression in Huh7-NTCP cells restores antiviral response and clearance of HDV in dividing cells (Fig. 8). Importantly, this result points to a stage of the viral cycle sensitive to antiviral effectors located within the cytoplasm, when nucleus structure disassembles during mitosis. However, mature hepatocyte models, such as PHHs and HLCs, are quiescent cells. Cell division-mediated viral spread does not occur in these cells and thus, what step of the viral cycle is targeted in these cells is unclear. Previous work showing a moderate antiviral effect of IFN alpha, but only as a pre-treatment, suggests that an early step of the HDV cycle could be the target of the innate constitutive immunity (reviewed by Zhang et al.²¹). Altogether, our data suggest that IRF1-competent hepatocytes restrict HDV infection at a very early step, probably before the nuclear translocation of the RNPs, through IRF1-driven basal expression of antiviral effectors.

Recent therapeutic developments have been made in the field of HDV. Noticeably, Hepcludex, an entry inhibitor, showed during a recent phase III clinical trial significant beneficial effect as a monotherapy.¹⁷ Importantly, being an entry inhibitor, BLV does not inhibit cell division-mediated spread of HDV infection. Our work narrows down the search for new potent cellular antiviral effectors, particularly targeting this cell division-mediated viral spread. Their identification may provide valuable insights into the development of new antiviral strategies that could be applied in combination with Hepcludex. Moreover, we improve the understanding and development of mature in vitro models that are necessary for identification and validation of future antiviral therapies.

Abbreviations

AFP, alpha-fetoprotein; ALB, albumin; ctrl, control; dHepaRG, differentiated HepaRG; DiPa, differentiation pattern plot; DPG, differentiation pattern group; EMT, epithelial-to-mesenchymal transition; fdr, false discovery rate; GADPH, glyceraldehyde 3-phosphate dehydrogenase; HBsAg, hepatitis B surface antigen; HDAg, hepatitis delta antigen; HDV, hepatitis D Virus; hESCm human embryonic stem cell; hiPSCm human-induced pluripotent stem cell; HLC, hepatocyte-like cell; hPSCs, human pluripotent stem cells; HSC, hepatic stellate cell; IFN, interferon; IRF1, interferon regulatory factor 1; IRF3, interferon regulatory factor 3; ISG, Interferon-Stimulated Gene; ISG15, Interferon-Stimulated Gene 15; MOI: Multiplicity of Infection; NTCP, Na⁺-taurocholate co-transporting polypeptide; PAMP, pathogen-associated molecular pattern; PEG, polyethylene glycol; PHH, primary culture of human hepatocytes; PPI, protein–protein interaction; PRR, pattern recognition receptor; RLR, RIG-I-like receptor; scRNAseq, single-cell RNA sequencing; SLC10A1, soluble carrier family 10 member 1; t-SNE, t-distributed stochastic neighbour embedding; UMI, unique molecular identifier; WNV, West Nile Virus.

Financial support

This work was supported by Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project ID 158989968–SFB900 (TP), and under the Germany’s Excellence Strategy – EXC 2155 ‘RESIST’–Project ID 390874280 (TP).

Authors’ contributions

Performed experiments and analysed data: FL, JG, MB, AC. Provided critical advice on the study concept, and edited the manuscript: TP. Designed the study concept and wrote the manuscript: FL, AC.

Data availability statement

RNAseq data are publicly available on GEO, using accession numbers GSE132606 and GSE132548. Additional data are available upon request.

Conflicts of interest

The authors have no conflicts of interest to declare.

Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the Supplementary data to this article: