Hepatitis B virus evades immune recognition via RNA adenosine deaminase ADAR1-mediated viral RNA editing in hepatocytes

Abstract

HBV is considered as a “stealth” virus that does not invoke interferon (IFN) responses; however, the mechanisms by which HBV bypasses innate immune recognition are poorly understood. In this study, we identified adenosine deaminases acting on RNA 1 (ADAR1), which is a key factor in HBV evasion from IFN responses in hepatocytes. Mechanically, ADAR1 interacted with HBV RNAs and deaminated adenosine (A) to generate inosine (I), which disrupted host immune recognition and thus promoted HBV replication. Loss of ADAR1 or its deficient deaminase activity promoted IFN responses and inhibited HBV replication in hepatocytes, and blocking the IFN signaling pathways released the inhibition of HBV replication caused by ADAR1 deficiency. Notably, the HBV X protein (HBx) transcriptionally promoted ADAR1 expression to increase the threshold required to trigger intrinsic immune activation, which in turn enhanced HBV escape from immune recognition, leading to persistent infection. Supplementation with 8-azaadenosine, an ADAR1 inhibitor, efficiently enhanced liver immune activation to promote HBV clearance in vivo and in vitro. Taken together, our results delineate a molecular mechanism by which HBx promotes ADAR1-derived HBV immune escape and suggest a targeted therapeutic intervention for HBV infection.

Introduction

Hepatitis B virus (HBV) infection remains a global burden of public health and causes acute and chronic hepatitis, cirrhosis and even hepatocellular carcinoma (HCC). Despite the success of HBV vaccination and available treatments such as interferon (IFN) and nucleoside/nucleotide analogs, the chronically infected population remains high at 250 million worldwide with an annual death toll from HBV of 800,000 people [1].

HBV is a hepatotropic, partially double-stranded DNA virus with a unique life cycle that produces both viral DNAs (relaxed circular DNA, covalently closed circular DNA (cccDNA)) and viral RNA intermediates (3.5, 2.4, 2.1, and 0.7 kb), which should be detected by pattern recognition receptors (PRRs) to induce IFN responses [2]. However, HBV has long been considered as a “stealth” virus since little or no IFN production is detected in most HBV-infected patients, including those carrying high quantities of viral particles and antigens [3]. Consistent with these clinical observations, neither IFN nor IFN-stimulated gene (ISG) expression has been observed in HBV-infected chimpanzees, woodchucks, or HCC cell lines [4, 5]. Earlier studies have suggested that HBV has developed strategies to counteract innate immunity by interfering with either PRRs or their downstream signals [6, 7]. A recent report indicated that the immune recognition of HBV pregenomic RNA (pgRNA) by retinoic acid-inducible gene-I (RIG-I) triggered a low level of type III IFN in hepatocytes [8]. To date, the mechanisms of innate immune recognition and HBV evasion are still poorly understood.

Although hepatocytes lack a functional DNA-sensing pathway due to the absence of cyclic GMP-AMP synthase (cGAS)-stimulator of IFN genes protein (STING) and therefore cannot respond to foreign DNA in the cytoplasm, RNA sensors in hepatocytes are abundant [9–11]. Treatment with poly(I:C) or infection with Sendai virus (SeV) results in robust expression of IFN-β and ISGs in hepatocytes [12]. In contrast to HBV, other hepatotropic viruses, such as HEV (RNA virus), promote a sustained IFN response in hepatocytes, and elevated ISG expression has been detected in patients with chronic HEV infection and HEV-infected mice grafted with human hepatocytes [13, 14]. Therefore, elucidating the mechanism of HBV RNA evasion from RNA sensors in hepatocytes is a very interesting research direction.

RNA modifications, such as N6-methyladenosine (m6A), m5C methylation, pseudouridine (Ψ) and RNA editing, are common mechanisms by which dsRNAs are shielded from PRRs [15]. Enzymes related to adenosine deaminases acting on RNA 1 (ADAR1), the RNA-editing enzyme that converts adenosine (A) to inosine (I) in duplex RNA regions, are among the most important strategic factors controlling innate immune responses to endogenous RNAs. ADAR1-mediated recognition of “self” and “nonself” RNAs limits autoimmunity response driven by RNA sensors, such as cytosolic melanoma differentiation-associated protein 5 (MDA5/IFIH1) and RIG-I (DDX58) [15–19]. In addition to endogenous RNAs, ADAR1 accounts for extensive editing of viral dsRNA and regulates innate immunity against several viruses, such as measles virus and hepatitis C virus [20, 21]. However, it remains unclear whether ADAR1 manipulates HBV RNAs editing to facilitate HBV innate immune evasion and/or viral replication.

Here, we report that ADAR1 blocks PRRs recognition of HBV RNAs in hepatocytes, which depends on the editing activity, thereby limiting IFN production and subsequently enhancing HBV replication. An ADAR1 inhibitor was shown to increase the liver immune response to promote HBV clearance. Further, HBV X protein (HBx) transcriptionally promotes ADAR1 expression to enhance HBV persistence. Our findings provide a new insight into HBV escape from the host immune recognition system and suggest a therapeutic target for HBV clearance.

Materials and methods

HBV infection in human hepatocytes

HBV stock was prepared from HepG2.2.15 culture supernatant with PEG8000 at a concentration previously reported [10]. As a control, the HBV stock was heated to 100 °C for 10 min to abolish the infectivity of HBV virions. HBV receptor sodium taurocholate cotransporting polypeptide (NTCP)-expressing HCC cell lines (HepG2^NTCP and Huh7^NTCP cells) and HLCZ-01 and primary human hepatocytes (PHHs) were infected with HBV at 400 Geq/cell in medium containing 4% PEG8000 (H8671, Solarbio, San Diego, CA, USA) and 2% DMSO (D8371, Solarbio). The cells were cultured in DMEM supplemented with 10% fetal bovine serum for the indicated times after washing with PBS three times 6 h post infection.

RNA-binding protein immunoprecipitation (RIP) assay

Huh7 cells transfected with the indicated plasmids were crosslinked with UV for 10 min. Twenty microliters of the supernatant was saved for assay input. After 2 h of incubation with anti-Flag (M185, MBL, Japan)/anti-Myc antibody (60003-2-ig, PeproTech, Rocky Hill, NJ, USA), protein-A/G Dynabeads (B23201, Bimake, Houston, TX, USA) were added and further incubated for 1 h with gentle shaking at 4 °C. After washing three times, the precipitated RNAs were purified and analyzed by RT-qPCR with appropriate primers to detect the enrichment of target RNA.

In vitro transcription and transfection of HBV pgRNA

The HBV1.3 plasmid was linearized with HindIII enzyme (FD0504, Thermo Fisher), and HBV pgRNA was transcribed with SP6 RNA polymerase according to the instructions (EP0131, Thermo Fisher). The RNAs (10 nM) obtained were digested with alkaline phosphatase (Thermo Fisher) to remove phosphate groups, and these RNAs were subsequently folded in RNA structure buffer (10 mM Tris, 0.1 M NaCl, and 10 mM KCl) by heating for 3 min at 95 °C, placed immediately on ice for 5 min, and rewarmed slowly to room temperature. The folded RNAs were transfected into cells with Lipo2000 (11668019, Invitrogen). To verify ADAR1-mediated viral editing, an edited-HBV1.3 plasmid carrying 11 ADAR1-mediated A>G transitions was constructed and transcribed in vitro and transfected as described above.

Animal experiments

Six- to eight-week-old male C57BL/6 mice purchased from Charles River laboratories (Beijing, China) were maintained under specific pathogen-free conditions. An AAV-HBV1.2 plasmid (6 µg/mouse) in a volume of normal saline equivalent to 8% of the mouse body weight was hydrodynamically injected (HDI) into the tail vein of each experimental mouse. The total volume was delivered within 5–8 s. Four days later, these mice received 8-azaadenosine (8-aza) (2 mg/kg B.W., 6868, R&D Systems) through intraperitoneal injection or entecavir (ETV) (0.1 mg/kg B.W., HY-13623, MCE) by oral gavage once daily. All animal experimental procedures were approved by the Guide for the Care and Use of Laboratory Animals.

Statistical analysis

All statistical analyses were performed using Prism software package version 8 (GraphPad software). Unpaired Student’s t test, one-way ANOVA, and two-way ANOVA were performed to determine the statistical significance of differences between groups. Significance levels are indicated by asterisks: *P < 0.05; **P < 0.01; ***P < 0.001.

Results

Hepatocytes efficiently respond to in vitro-transcribed HBV RNA but not natural HBV infection

Although innate immunity is the first line of defense against pathogens, it is unclear whether HBV induces an IFN response in hepatocytes [5, 8]. Hence, we tested IFN responses during HBV infection in human hepatocytes, including HepG2^NTCP and HLCZ-01 cells [22], and PHHs. HBV infection triggered low levels of IFNB1 and IFNL1 expression in all hepatocytes compared with inactivated HBV (Fig. 1a, b and Supplementary Fig. 1a). Considering the relatively low efficiency of HBV infection and the differences between HBV genotypes, we transfected HepG2 cells with HBV1.1 (genotype C) or HBV1.3 (genotype D) plasmids to drive high levels of viral replication. However, these transfected HBV plasmids induced weak expression of IFNB1 and IFNL1 even when pgRNA abundance had reached a plateau, and no obvious differences in IFN expression between HBV genotypes was observed (Supplementary Fig. 1b, c).

Fig. 1.

Hepatocytes efficiently respond to in vitro-transcribed HBV RNA but not natural HBV infection. a HepG2^NTCP cells and b PHH cells were infected with HBV at 400 Geq/cell, and HBV heated at 100 °C was used as a control. The levels of IFNB1, IFNL1, and HBV pgRNA were evaluated by quantitative RT-PCR (RT-qPCR) at the indicated times after HBV infection (n = 3). Transcriptional levels of IFNA1, IFNB1, and IFNL1 in HepG2 cells were detected after stimulation for 8 h with c dsRNA (poly (I:C), 10 μg/mL) (n = 3) or VSV (10 MOI) (n = 4) and d dsDNA (CT-DNA, 10 μg/mL) (n = 3) or 2′3′-cGAMP (10 μg/mL) (n = 4). Phosphorylation and total levels of IRF3 and TBK1 were measured by western blotting at the indicated time points after stimulation. *P < 0.05, **P < 0.01, ***P < 0.001 (unpaired t test). e PHH cells (n = 4) and HepG2 cells (n = 3) were transfected with in vitro-transcribed HBV RNA, and IFNA1, IFNB1, and IFNL1 levels were measured by RT-qPCR. Phosphorylation levels of IRF3 and TBK1 were analyzed by western blotting. *P < 0.05, **P < 0.01, ***P < 0.001 (one-way ANOVA)

DNA and RNA intermediates generated during the HBV life cycle are presumably detected by various PRRs; thus, we compared PRRs expression levels in human hepatocytes with those in the THP-1 differentiated macrophage cell line, which expresses high levels of various PRRs. Compared with differentiated THP-1 cells, the levels of DNA sensors, including cGAS, TMEM173, and IFI16, were negligibly detectable in hepatocytes. However, RNA sensors, such as TLR3, DDX58, and IFIH1, were expressed at comparable levels in all hepatocyte cell lines and PHHs compared to THP-1 cells (Supplementary Fig. 1d). Consistently, HepG2 cells exhibited a more selective response to dsRNA analogs than cytosolic DNA. Stimulation with either the RNA mimic poly(I:C) or RNA virus (vesicular stomatitis virus, VSV) triggered mass production of IFNA1, IFNB1, and IFNL1 in HepG2 cells (Fig. 1c). In contrast, the DNA sensor agonists CT-DNA (deoxyribonucleic acid sodium salt obtained from calf thymus) and 2′3′-cGAMP (cyclic GMP-AMP) failed to induce intense IFN responses in HepG2 cells (Fig. 1d) but strongly activated IFN pathways in THP-1 cells (Supplementary Fig. 1e). In accordance with these findings, TBK1 and IRF3 were obviously phosphorylated by poly(I:C) and VSV but not CT-DNA or 2′3′-cGAMP in HepG2 cells (Fig. 1c, d). These data suggest that in human hepatocytes, DNA-sensing pathways are defective, whereas RNA-sensing pathways are relatively abundant.

Since hepatocytes are equipped with an efficiently functioning RNA-sensing pathway, we asked whether hepatocytes react to transcribed HBV RNAs in vitro. Thus, the 3.5 kb HBV pgRNA transcript was transcribed in vitro and dephosphorylated, and then, its secondary structure was restored (tpgRNA) (Supplementary Fig. 1f). Surprisingly, tpgRNA transfection stimulated intense dose-dependent expression of IFNA1, IFNB1, and IFNL1 in both HepG2 and PHH cells. Furthermore, the phosphorylation of TBK1 and IRF3 was also significantly activated by tpgRNA (Fig. 1e and Supplementary Fig. 1f). These observations strongly suggest that HBV RNAs might be modified to escape PRR recognition in hepatocytes.

ADAR1 interacts with HBV RNA intermediates through dsRNA-binding domains

HBV RNA-interacting proteins have been investigated by pgRNA pull-down and mass spectrometric analyses [23]. Upon reanalysis of the published data, we did not find RNA-modifying enzymes related to m6A, m5C, or Ψ, such as methyltransferase-like 3/14, NOP2/Sun RNA methyltransferase 2, or pseudouridine synthases, among the pgRNA-interacting proteins [15]. However, proteins related to RNA processing and infectious disease were enriched, as shown by analyses using Database for Annotation, Visualization and Integrated Discovery v6.8 and Gene Ontology according to the biological functions (Supplementary Tables 1 and 2). Among the enriched proteins, ADAR1 is an RNA-editing enzyme and a specific negative regulator of the RIG-I/MDA5-MAVS (mitochondrial antiviral signaling protein) antiviral response pathway (Fig. 2a) [17]. To verify the interaction of ADAR1 with HBV RNAs, we performed both RNA pull-down and RIP assays. As shown in Fig. 2b and Supplementary Fig. 2a, ADAR1 interacted with biotinylated tpgRNA, and this interaction was significantly inhibited by unlabeled tpgRNA. In accordance with this finding, the RIP assay demonstrated the enrichment of ADAR1 in HBV RNAs (Fig. 2c). In addition, confocal imaging verified the colocalization of ADAR1 and biotinylated tpgRNA in Huh7 cells (Supplementary Fig. 2b). Another RIP analysis revealed that ADAR1 bound to all four viral RNA transcripts (3.5, 2.4, 2.1, and 0.7 kb), although the enrichment of ADAR1 with the 0.7 kb HBV RNA was much lower than that with the other RNA transcripts (Fig. 2d). All these data demonstrate the direct interaction of ADAR1 with HBV RNA intermediates.

Fig. 2.

Interaction of ADAR1 with HBV RNAs. a HBV pgRNA-interacting proteins were adopted on the basis of published data [23]. The proteins related to RNA processing (n = 36) and infectious disease (n = 68) were enriched by the Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.8 and Gene Ontology (GO) according to biological functions. The Venn diagram reveals that ADAR1 and NONO (non-POU domain-containing octamer binding protein) were included in the two groups. b RNA pull-down assay showing the interaction of ADAR1 with tpgRNA. Huh7 cells were cotransfected with biotinylated tpgRNA and unlabeled tpgRNA. tpgRNA was precipitated with anti-biotin antibody, and coprecipitated ADAR1 was detected by western blotting. c RIP analysis confirmed the interaction of ADAR1 and pgRNA. Huh7 cells were cotransfected with Flag-ADAR1 and HBV1.3 and incubated for 72 h, ADAR1 was precipitated with anti-Flag, and coprecipitated pgRNA was measured via RT-qPCR. The level of enriched pgRNA in the control vector was used as the control (n = 3). ***P < 0.001 (unpaired t test). d Flag-ADAR1 and different HBV RNAs (n = 3) were cotransfected into Huh7 cells and incubated for 48 h. e HBV1.3 and a series of truncation mutants or f Flag-ADAR1 mutants were cotransfected into Huh7 cells and incubated for 48 h. A RIP assay was performed with anti-Flag antibody, and coprecipitated RNAs were quantified by RT-qPCR. The control vector was transfected as the control (n = 3). **P < 0.01, ***P < 0.001 (one-way ANOVA)

ADAR1 binds to ~20 bp of the duplex RNA structure [24]. The 5′- and 3′ ends of HBV pgRNA contain the “epsilon (ε)” sequence, which forms a stem-loop secondary structure [25]. We constructed HBV mutants with the 5′/3′-ε regions deleted and evaluated whether the ε structure is a possible ADAR1-binding site. RIP assays showed that a single deletion of either the 5′- or 3′-ε element led to a slight decrease in ADAR1 enrichment and that dual deletion of 5′-ε and 3′-ε greatly decreased ADAR1 recruitment; however, ADAR1 expression was still significantly enriched in complex with pgRNA in which both 5′-ε and 3′-ε regions were deleted, indicating that there are multiple ADAR1-binding sites in HBV RNAs (Supplementary Fig. 2c).

ADAR1 contains a Z-DNA-binding domain in the N-terminus, a highly conserved deaminase domain (DM) in the C-terminus, and three double-stranded RNA-binding domains (dsRBDs) [26]. To map the critical domains of ADAR1 that bind HBV RNA intermediates, a series of ADAR1 mutants, included truncation mutants, were constructed, and an RIP analysis was performed. Compared to full-length ADAR1, ADAR1 with a deleted Z-DNA binding domain (amino acids 503–1222) and deleted dsRBD1 domain (amino acids 615–1222) maintained high binding affinity for HBV RNAs, and when dsRBD2 (amino acids 727–1222) was further deleted, ADAR1 lost the ability to bind with HBV RNAs (Fig. 2e), indicating that the two dsRBDs are dispensable for the interaction of ADAR1 with HBV RNAs. To determine the support for this hypothesis, we constructed a series of ADAR1 dsRBD mutants as previously reported [27]. As shown in Fig. 2f, mutations in any two dsRBDs (mdsRBD1 + 2, mdsRBD1 + 3, or mdsRBD2 + 3) or in all three dsRBDs (mdsRBD1-3) abolished the binding of ADAR1 to HBV RNAs. However, a single mutation in a dsRBD (mdsRBD1, mdsRBD2, or mdsRBD3) did not interfere with the interaction of ADAR1 with HBV RNAs. In addition, mutation at the 912 amino acid site (E912A), which is critical for deaminase activity [16], did not affect the interaction of ADAR1 with HBV RNAs, indicating that the DM domain of ADAR1 is not necessary for HBV RNA binding (Supplementary Fig. 2d). These data suggested that at least two dsRBDs are required for ADAR1 binding with HBV RNAs.

ADAR1 is critical for HBV evasion from the host immune recognition system

It has been established that ADAR1 regulates virus replication by editing viral RNAs to modulate viral immune recognition by the host [20, 21, 28]. To address whether ADAR1 plays a role in HBV editing, HBV-infected HepG2^NTCP-Tet-shADAR1 cells were treated with Dox to induce ADAR1 knockdown (Supplementary Fig. 3a), and then, RNA sequencing was performed. To verify the editing activity of ADAR1, we first analyzed the sequence of whole host RNAs as well as MAVS RNA, which has been identified as an ADAR1 target [29]. Compared with the number in control cells, fewer mutations, including A>G and U>C transitions, in MAVS RNA (Supplementary Table 3), were detected in Dox-induced HepG2^NTCP-Tet-shADAR1 cells. Mutations including A>G and U>C transitions of host genes in HepG2^NTCP-Tet-shADAR1 cells without Dox treatment confirmed the editing activity of ADAR1 (Supplementary Fig. 3b), and most editing sites were found to be located in noncoding sequences and untranslated regions (Supplementary Fig. 3c). Further analysis of the HBV RNA sequence revealed that there were 11 A to I editing sites in HBV RNAs, and the editing frequency of each A to I mutation was obviously reduced after ADAR1 knockdown in Dox-treated HepG2^NTCP-Tet-shADAR1 cells (Fig. 3a, Supplementary Fig. 3b and Supplementary Table 4), suggesting that HBV RNAs are edited by ADAR1.

Fig. 3.

ADAR1 edits HBV RNAs to abolish PRR sensing and related IFN activation. a, b HBV-infected HepG2^NTCP-Tet-shADAR1 cells were treated with 1 μg/mL Dox for 72 h, and then, RNA sequencing was performed. A–G editing frequency of HBV RNAs in cells in the presence of Dox (Dox+) or the absence of Dox (Dox−) was analyzed by RNA sequencing (a). GSEA showed enrichment of the type I IFN production pathway (b). c Correlation analysis of mRNA levels of ADAR1 and IFNB1 or IFNL1 in paratumor liver tissues from HBV-infected patients. d HepG2^NTCP-Tet-shADAR1 cells were infected with HBV at 400 Geq/cell in the presence or absence of Dox for 72 h, and IFNB1, IFNL1, and CXCL10 expression was evaluated by RT-qPCR (n = 4). Phosphorylation and total levels of IRF3 and TBK1 were measured by western blotting. e HepG2 cells transfected with HBV1.3 were treated with the ADAR1 inhibitor 8-aza, and gene expression was measured by RT-qPCR (n = 4). f HepG2 cells were transfected with in vitro-transcribed wild-type HBV pgRNA (tpgRNA-WT) and HBV pgRNA with 11 A–G site mutations (tpgRNA-edited) for 8 h. the IFNB1, IFNL1, and pgRNA levels were measured by RT-qPCR (n = 4). **P < 0.01, ***P < 0.001 (one-way ANOVA). g HepG2^NTCP-Tet-shADAR1 cells were transfected with wild-type HBV1.3 (HBV1.3-WT) and edited-HBV1.3 plasmids with 11 A–G site mutations (HBV1.3-edited) in the presence of Dox for 72 h, and IFNB1, IFNL1, CXCL10, and pgRNA expression was measured by RT-qPCR (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 (one-way ANOVA). h Huh7 cells were pretreated with siADAR1, followed by cotransfection of HBV1.3 and Flag-RIG-I or MDA5-Myc, and the enrichment of HBV RNA was measured by RT-qPCR (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 (two-way ANOVA). i HepG2^NTCP-Tet-shADAR1 cells were treated with Dox for 72 h, and IFIH1 and DDX58 mRNA levels were measured by RT-qPCR (n = 4). **P < 0.01 (unpaired t test). j HepG2^NTCP-Tet-shADAR1 cells transfected with HBV1.3 were treated with Dox and siMAVS, and the expression of the indicated genes was analyzed by RT-qPCR (n = 4). *P < 0.05, ***P < 0.001 (two-way ANOVA)

We evaluated whether ADAR1-mediated RNA editing regulated IFN responses in HBV-infected HepG2^NTCP-Tet-shADAR1 cells. RNA sequencing revealed that ADAR1 knockdown resulted in the upregulation of 865 transcripts (log₂ FC > 1), 92 of which belong to IFN alpha/beta signaling genes (Supplementary Table 5). The results from a GSEA (gene set enrichment analysis) showed a significant enrichment of differentially expressed gene signatures in the type I IFN production pathway in Dox-treated HepG2^NTCP-Tet-shADAR1 cells (Fig. 3b). The expression of IFNs, including IFNL1/3 and IFNB1, and ISGs, such as IFITM3, ISG15, CXCL10, and MX1, was significantly upregulated in Dox-treated HepG2^NTCP-Tet-shADAR1 cells (Supplementary Fig. 3d). Consistently, ADAR1 mRNA levels displayed a significantly negative correlation with both IFNB1 and IFNL1 mRNA levels in paracancerous liver tissues from patients with HBV-related HCC (Fig. 3c).

To confirm ADAR1-mediated immune regulation during HBV replication, IFN and downstream ISG (CXCL10) expression was detected in HBV-infected HepG2^NTCP-Tet-shADAR1 cells. We first confirmed that Dox treatment did not affect the expression of ADAR1, IFNB1, or CXCL10 in HepG2 cells (Supplementary Fig. 3e). As shown in Fig. 3d and Supplementary Fig. 3f, Dox-induced knockdown of ADAR1 significantly enhanced the expression of IFNB1, IFNL1, and CXCL10 and promoted the activation of IRF3 and TBK1, key regulators of the IFN pathway. The enhancement of these regulators was further promoted by HBV infection, although a single incidence of HBV infection had little effect on IFN expression. In accordance with this finding, treatment with the ADAR1 inhibitor 8-aza enhanced IFN and ISG expression in HepG2 cells, while transfection of the HBV1.3 plasmid further promoted this enhancement (Fig. 3e). These data suggested that ADAR1 regulates the innate immune response in HBV-infected hepatocytes on the basis of its editing activity.

To verify the role of ADAR1-mediated HBV RNAs editing in the regulation of IFN responses and subsequent HBV immune evasion, we constructed an HBV1.3-edited plasmid carrying 11 A>G mutated ADAR1 interaction sites that had been identified by RNA-seq, as shown in Fig. 3a. HepG2 cells were transfected with in vitro-transcribed pgRNA prepared from either HBV1.3-edited or HBV1.3-WT cells as described in Fig. 1e. As expected, the unedited tpgRNA (tpgRNA-WT) stimulated high IFNB1 and IFNL1 expression, whereas the IFN expression levels induced by the edited tpgRNA were obviously lower (Fig. 3f). In accordance with these findings, although the cells transfected with unedited HBV1.3 showed significantly elevated expression of IFNB1, IFNL1, and CXCL10, the HepG2^NTCP-Tet-shADAR1 cells transfected with edited HBV1.3 failed to show an activated IFN response (Fig. 3g). More importantly, consistent with the silenced IFN responses, the edited-HBV1.3-transfected HepG2^NTCP cells produced higher levels of pgRNA than the HBV1.3-WT-transfected HepG2^NTCP cells with ADAR1 deficiency (Fig. 3g). All of these data strongly suggest that HBV RNA can be edited by ADAR1 and that this ADAR1 editing prevents IFN induction during HBV replication.

We then sought to determine whether ADAR1 regulates IFN responses by editing and blocking HBV RNA sensing by host cells. To this end, RIG-I and MDA5, two RNA sensors known to recognize HBV RNAs [30], were cotransfected with HBV1.3 plasmid into Huh7 cells in the absence or presence of ADAR1. Subsequently performed RIP assays revealed that both RIG-I and MDA5 bound with HBV RNAs at low levels in the presence of ADAR1, while genetic knockdown of ADAR1 or 8-aza treatment significantly enhanced the enrichment of RIG-I and MDA5 bound to HBV RNA (Fig. 3h and Supplementary Fig. 3g). In addition, knockdown of ADAR1 significantly increased the transcriptional levels of IFIH1 (MDA5) and DDX58 (RIG-I), which may have further strengthened the recognition of HBV RNAs in hepatocytes (Fig. 3i). Moreover, intervention with MAVS, the key adapter protein of MDA5 and RIG-I, significantly attenuated the augmented production of IFNB1, IFNL1, and CXCL10 in Dox-treated HepG2^NTCP-Tet-shADAR1 cells (Fig. 3j). These results show that ADAR1-mediated RNA editing enables HBV RNAs escape from the host RNA-sensing system and prevents IFN responses.

ADAR1 promotes HBV replication on the basis of its editing activity

Accumulating evidence has demonstrated ADAR1 as a replication regulator during many viral infections [20, 31, 32]. The results obtained in this study demonstrated that ADAR1 mediates HBV evasion from the host innate immune system. To further investigate the role of ADAR1 in HBV replication, ADAR1 overexpression or ADAR1 knockdown was induced in different HBV-infected hepatocytes, including HepG2.2.15 cells and HBV-infected Huh7^NTCP cells and HepG2^NTCP cells. As expected, Dox-induced knockdown of ADAR1 significantly decreased the expression of viral proteins (HBsAg and HBeAg) and pgRNA in HepG2.2.15-Tet-shADAR1 cells. Northern blot analysis further confirmed the reduced pgRNA level in ADAR1-deficient cells, while immunofluorescence assays revealed decreased HBsAg (red) levels in Dox-treated HepG2.2.15-Tet-shADAR1 cells, as indicated by high GFP expression. Loss of ADAR1 greatly decreased the protein levels of HBV polymerase (Pol), HBc, and HBx (Fig. 4a). Similar results were observed in HBV-infected HepG2^NTCP-Tet-shADAR1 cells, and ADAR1 deficiency significantly downregulated HBV-related antigens and pgRNA levels (Fig. 4b). In accordance with these findings, overexpression of ADAR1 in HBV1.3-transfected Huh7 cells strongly promoted the expression of viral proteins and viral RNA transcripts (Supplementary Fig. 4a), while ADAR1 knockdown with siRNA greatly decreased the levels of all detected viral antigens and pgRNA in HBV-infected Huh7^NTCP cells (Supplementary Fig. 4b).

Fig. 4.

ADAR1 promotes HBV replication. a HepG2.2.15-Tet-shADAR1 cells (n = 4) and b HBV-infected HepG2^NTCP-Tet-shADAR1 cells (n = 3) were treated with Dox for 72 h, and ADAR1, viral antigens (HBsAg, HBeAg, HBx, Pol, and HBc), and pgRNA levels were measured. For the immunofluorescence assay, Dox induced the expression of both shADAR1 and ZsGreen in the HepG2^NTCP-Tet-shADAR1 cells, and GFP⁺ cells verified ADAR1 knockdown. The scale bar in a is 50 μm, and the scale bar in b is 20 μm. *P < 0.05, **P < 0.01, ***P < 0.001 (unpaired t test). c Huh7 cells were cotransfected with HBV1.3 and wild-type ADAR1 or a series of dsRBD mutants, and HBsAg, HBeAg, and pgRNA levels were measured (n = 3). d HBV1.3-transfected Huh7 cells were pretreated with siADAR1 and then transfected with WT-ADAR1 or ADAR1-E912A (mutation in the catalytic domain), and HBsAg/HBeAg and pgRNA levels were evaluated (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 (one-way ANOVA). e HBV-infected HepG2^NTCP-Tet-shADAR1 cells were treated with Dox and mixture antibodies against type I IFNs and IFNAR2 for 72 h, and viral antigen (HBsAg, HBeAg) and pgRNA levels were evaluated (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 (two-way ANOVA)

To define the key domain of ADAR1 critical for promoting HBV infection, mutants, including serial truncation mutants, were generated. Consistent with the data shown in Fig. 2 indicating that the critical domain for ADAR1 binding with HBV RNAs, the sequences truncated in two ADAR1 mutants (amino acids 503–1222) and ADAR1 (amino acids 615–1222), displayed comparable effects with full-length ADAR1 in promoting HBV antigen expression and pgRNA transcription in HBV1.3-transfected Huh7 cells. In ADAR1, the deletion of two dsRBDs (amino acids 727–1222) failed to enhance HBV infection (Supplementary Fig. 4c). Consistently, ADAR1 with mutations in two or three dsRBDs (mdsRBD1 + 2/mdsRBD1 + 3/mdsRBD2 + 3/mdsRBD1 − 3) failed to regulate the transcription of HBV, while mutation in a single dsRBD did not obviously affect the ADAR1-mediated enhancement of HBV replication (Fig. 4c). Moreover, full-length ADAR1 significantly reversed the reduction in the levels of HBV antigens and pgRNA in ADAR1-depleted cells, while the E912A mutation, which is in the DM of ADAR1, failed to boost HBV replication in HBV1.3-transfected Huh7 cells (Fig. 4d), suggesting that editing activity is critical for ADAR1-mediated HBV promotion. To verify that ADAR1 regulation of HBV replication depending on IFN responses, HBV-infected HepG2^NTCP-Tet-shADAR1 cells were treated with mixed IFN-neutralizing antibodies or siMAVS. As expected, Dox-induced ADAR1 knockdown significantly decreased HBV replication, and the inhibition of HBsAg, HBeAg and pgRNA was almost completely reversed by either IFN-blocking antibodies or silencing MAVS (Fig. 4e and Supplementary Fig. 4d, e). Collectively, these data illustrate that ADAR1 enhances HBV replication via its editing activity and related immune regulation.

Two isoforms of ADAR1 have been identified: ADAR1-p150 is mainly located in the cytoplasm, whereas ADAR1-p110 is located in the nucleus, which is the dominantly expressed isotype [33]. To determine whether the cellular location of ADAR1 is important to its regulation of HBV replication, plasmids expressing the p110 protein and ADAR1 plasmids carrying two in-frame initiation AUGs and expressing both the p150 and p110 proteins were used in HBV-replicating cell models. As shown in Supplementary Fig. 4f, both ADAR1-p110 and the ADAR1 plasmid expressing the two isoform proteins significantly promoted HBV replication. To evaluate whether ADAR1 edits HBV RNAs in the nucleus or cytoplasm, deletion of the N-terminal fragment ΔM708-K715 was performed to abolish the nuclear localization signal of ADAR1 [34], as confirmed by confocal imaging (Supplementary Fig. 4g). Compared to wild-type ADAR1, the ΔM708-K715 mutant showed no ability to regulate HBV replication (Supplementary Fig. 4g). In addition, the ΔM708-K715 mutant also showed reduced binding to HBV RNAs than WT-ADAR1 (Supplementary Fig. 4h). Taken together, our data demonstrate that ADAR1 enhances HBV replication on the basis of its RNA binding and editing activities, likely in the nucleus.

HBx transcriptionally promotes ADAR1 expression via YY1

Because HBV can reduce the expression of multiple host genes to promote self-replication, we asked whether HBV reversely affected ADAR1 expression. Forty clinical HCC patients (Supplementary Table 6) were classified into two groups according to the levels of HBV DNA and HBeAg: HBV-active (HBeAg-positive and HBV DNA > 1000 copies/mL in serum) and HBV-inactive groups (HBeAg negative or HBV DNA < 1000 copies/mL in serum). Immunohistochemical (IHC) staining showed higher ADAR1 expression in the adjacent nontumor tissues of HBV-active patients (Fig. 5a). Further analysis demonstrated a significant positive correlation between ADAR1 and HBcAg expression, and high levels of ADAR1 were observed in the tissues of patients with serum HBV DNA > 1000 copies/mL (Fig. 5b). Moreover, RT-qPCR revealed a positive correlation between ADAR1 and HBV pgRNA in adjacent nontumor tissues (Supplementary Fig. 5a). Similarly, ADAR1 expression was significantly enhanced in both the HBV-infected HepG2^NTCP cells and HBV1.3-transfected Huh7 cells, as well as in HBV transgenic mice (Fig. 5c and Supplementary Fig. 5b). These data illustrate that ADAR1 was highly expressed in these HBV-infected hepatocytes.

Fig. 5.

HBx transcriptionally upregulates ADAR1 by interacting with YY1. a IHC staining of ADAR1 and HBcAg in consecutive sections of paratumor tissue sections obtained from HCC patients. The scale bar represents 200 μm. b ADAR1 staining scores were compared between patients with different HBV DNA levels. The correlation between ADAR1 and HBcAg staining scores was analyzed, ***P < 0.001 (unpaired t test (above), Spearman rank test (below)). c Detection of the protein abundance and mRNA level of ADAR1 in HBV-infected HepG2^NTCP cells (n = 4) or HBV1.3-transfected Huh7 cells (n = 4). ***P < 0.001 (unpaired t test). ADAR1 promoter reporter plasmid was cotransfected with different doses of HBx plasmid (n = 4) (d) or minicircle HBV (MC-HBV)/HBx-deleted MC-HBV (MC-HBV-ΔHBx) (e) for 72 h (n = 3), and the protein levels of ADAR1 and ADAR1 promoter activity were assessed. *P < 0.05, **P < 0.01, ***P < 0.001 (one-way ANOVA). f HepG2 cells were pretreated with siADAR1 followed by transfection of the HBx plasmid for 72 h, and the mRNA levels of IFNB1, IFNL1, and ADAR1 were measured (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001 (two-way ANOVA). g Huh7 cells were transfected with HBx and a series of truncated ADAR1 promoter reporter plasmids (n = 4). h Huh7 cells were pretreated with siYY1, cotransfected with HBx and ADAR1 promoter reporter plasmids (n = 3), and incubated for 36 h, and then, ADAR1 promoter activity was assessed. ***P < 0.001 (unpaired t test was performed on the data in g and two-way ANOVA was performed on the data in h)

To identify the critical viral proteins of HBV that initiate the augmented expression of ADAR1, Huh7 cells were transfected with vectors expressing the entire HBV genome (HBV1.3) or HA-tagged viral proteins (HBx, HBc, pre-S2, and polymerase). Western blot analysis showed that several HBV-encoded proteins induced elevated expression of ADAR1, among which, HBx was the most potent enhancer of ADAR1 expression, compared to HBV1.3 (Supplementary Fig. 5c). Dual luciferase assays showed that ADAR1 promoter activity was activated by HBV proteins, particularly HBx (Supplementary Fig. 5d). Furthermore, ectopic HBx expression led to upregulated protein and transcriptional levels of ADAR1 in a dose-dependent manner (Fig. 5d). To verify the critical role of HBx in enhancing ADAR1 expression, an engineered minicircle HBV cccDNA (MC-HBV), which contains 1.0 copy of the HBV genome to support successful viral replication and viral protein expression, was constructed [35], and MC-HBV with a mutation at site 1376 (ATG→TTG) (MC-HBV-∆HBx) was constructed to inhibit HBx expression. Both vectors were transfected into Huh7 cells. As expected, MC-HBV greatly increased ADAR1 promoter activity and protein expression, while HBx depletion showed reduced ability to enhance ADAR1 expression in Huh7 cells (Fig. 5e). More importantly, overexpression of HBx abrogated the IFNB1 and IFNL1 production induced by ADAR1 interference (Fig. 5f). Together, these results reveal that HBV, especially the HBx protein, upregulates ADAR1 transcription to suppress the IFN response.

To better define the regulatory mechanism of HBx-enhanced transcriptional activation of ADAR1, various constructs truncated at the 5′-flanking region of the ADAR1 promoter, including −1576/+249 (pGL3‐1825), −779/+249 (pGL3‐1028), −442/+249 (pGL3‐691), and −215/+249 (pGL3‐464), were cloned and transiently transfected with HBx plasmids into Huh7 cells. Luciferase reporter assay showed that deletion of −779 to −442 dampened the HBx-mediated augmentation of ADAR1 promoter activity, indicating that the −779 to −442 region contains the core element critical for HBx-mediated trans-regulation of ADAR1 (Fig. 5g). Further, TRANSFAC (http://biogrid-lasagna.engr.uconn.edu/lasagna_search/a) analysis was performed, and the results predicted a binding site for the transcription factor YY1 in the −779 to −442 region (Supplementary Table 7), and HBx overexpression significantly upregulated YY1 levels (Supplementary Fig. 5f), indicating the involvement of YY1 in HBx-mediated promotion of ADAR1 transcription. To verify the critical role of YY1 in the HBx-mediated enhancement of ADAR1, both a mutant ADAR1 promoter reporter plasmid with a depleted YY1-binding motif and siRNA against YY1 were transfected into cells. As shown in Supplementary Fig. 5e, deletion of the YY1-binding motif in the ADAR1 promoter suppressed the HBx-mediated enhancement of ADAR1 promoter activity. In accordance with these results, knockdown of YY1 dampened the HBx-mediated enhancement of ADAR1 transcription (Fig. 5h and Supplementary Fig. 5f). These data support the hypothesis that YY1 is crucial for HBx-mediated transcriptional enhancement of ADAR1 expression.

The ADAR1 inhibitor 8-aza promotes immune activation and HBV clearance

These findings demonstrated ADAR1 is a facilitator of HBV immune evasion via its deaminase activity. We then explored the potential therapeutic effect of ADAR1 inhibitors against HBV. The ADAR1-specific inhibitor 8-aza [36] was introduced into HepG2.2.15- and HBV1.3-transfected Huh7 cells, and ETV (30 nM) was used as a positive control. Figure 6a and Supplementary Fig. 6a, b showed that 8-aza (0.2 μM) had no effects on cell viability but notably inhibited the expression levels of HBsAg, HBeAg, and pgRNA in both the HepG2.2.15 and Huh7 cell lines, compared with untreated control cell lines. The effect of 8-aza on HBV clearance was also verified in vivo. HBV carrier mice hydrodynamically injected with pAAV-HBV1.2 were intraperitoneally injected with 8-aza (2 mg/kg/day) [37]. As expected, 8-aza administration significantly reduced the serum levels of HBsAg and HBV particle DNA, as well as hepatic pgRNA levels and HBcAg expression, similar to the ETV treatment (Fig. 6b–d).

Fig. 6.

Treatment with 8-aza inflames HBV-infected livers and promotes HBV clearance. a HepG2.2.15 cells were treated with 8-aza (0.2 μM) or ETV (30 nM) for 3 days, and HBsAg/HBeAg and pgRNA levels were analyzed (n = 4). b–e pAAV-HBV1.2 plasmids (6 µg) were injected into C57BL/6 mice (6–8-week-old males) by HDI for 4 days, and then 8-aza (2 mg/kg/day) or ETV (0.1 mg/kg/day) treatment (d.p.t) was administered for the indicated times. b HBsAg and HBV DNA levels in serum, c hepatic pgRNA level, and d hepatic HBcAg level were measured (n = 5). e Mice were sacrificed 15 d.p.t., and the transcriptional levels of Ifnb1, Ifnl3, Ifng, Oas1, Mx1, and Tnfa in the livers were measured by RT-qPCR. f Expression of IFN-γ and TNF-α in splenic CD8⁺ T cells and NK cells was detected by flow cytometry. *P < 0.05, **P < 0.01, ***P < 0.001 (two-way ANOVA)

Considering the ADAR1-mediated IFN response in viral control, we evaluated the hepatic immune microenvironment of the treated mice. Consistent with previous in vitro data, 8-aza treatment increased hepatic levels of IFNs, especially Ifnb1 and Ifnl3, and ISGs, such as Oas1 and Mx1 (Fig. 6e). In addition, 8-aza treatment enhanced the mRNA expression of the proinflammatory cytokines Tnfa and Ifng in the HBV carrier mice (Fig. 6e). All of these data suggested that 8-aza promoted immune activation in the HBV-infected mice. In line with these findings, 8-aza treatment rescued IFN-γ and TNF-α secretion levels in splenic CD8⁺ T cells and NK cells impaired by HBV infection (Fig. 6f). Abolishment of ADAR1 activity also inflamed HBV-infected livers by increasing the proportions of T cells and decreasing the proportions of MDSCs and Treg cells (Supplementary Fig. 6c–e). All these results suggest that 8-aza is a potential drug for HBV clearance via enhanced immune responses.

Discussion

HBV is a “stealth” hepatotropic virus that bypasses the innate immune recognition mechanisms [5, 11], although many PRRs exist in hepatocytes [10]. Several mechanisms contributing to HBV evasion have been identified. First, DNA-sensing pathways are defective in hepatocytes, and HBV was reported to block cytosolic DNA-sensing pathways by interacting with STING [38]. Second, although hepatocytes express relatively abundant dsRNA-sensing PRRs and respond effectively to some RNA viruses [5, 10], HBV has developed multiple strategies to counteract RNA sensor-mediated signaling pathways. HBx promotes the degradation of MAVS or deconjugates RIG-I with TNF receptor-associated factor 3 (TRAF3) [6, 39]. However, recent research has shown that HBV does not interfere with innate immune responses to poly(I:C) or SeV in hepatocytes [11]. Therefore, unexplored host factors probably contribute to HBV evasion from RNA sensors in the immune system. Here, we report ADAR1 as a novel host factor that edits HBV RNAs and inhibits the immune response to HBV. HBx-mediated transactivation of ADAR1 facilitates HBV escape from the innate immune system. Moreover, ADAR1 inhibitor efficiently promotes immune activation and HBV clearance, shedding new light on HBV intervention.

ADAR1 is essential for hepatic homeostasis because it suppresses the activation of RIG-I-mediated IFN pathways to protect the liver from injury and extensive damage [40, 41]. In this study, by comparing the transcription profile of HBV-infected HepG2^NTCP-Tet-shADAR1 cells with or without ADAR1 silencing, we confirmed ADAR1-mediated cellular RNA editing, and importantly, 11 A>G transitions were detected in HBV RNAs. These mutations (A>G transitions) in HBV pgRNA greatly damaged the ability of HBV RNA-mediated IFN activation in hepatocytes. Both RIG-I and MDA5 are reported to recognize HBV RNAs [8, 42], ADAR1 silencing increased the interaction of RLRs with HBV RNAs and subsequently upregulated IFN expression. These data provide direct evidence demonstrating the critical role of ADAR1-mediated viral RNA editing in HBV evasion from PRR recognition, which is consistent with previous reports showing that ADAR1 prevents RLR sensing of endogenous dsRNA on the basis of its editing activity [18, 43]. The ADAR1-initiated promotion of HBV infection was largely dependent on the IFN response in hepatocytes. Blocking IFN pathways significantly rescued the ADAR1-mediated enhancement of HBV replication. In accordance with these findings, 8-aza, a potent inhibitor of ADAR1, inflamed HBV-infected livers to promote HBV clearance both in vitro and in vivo. Notably, Liu et al.’s report showed that ADAR1 inhibits HBV replication [44], a finding opposite to that of our study, while two other recent papers both confirmed that ADAR1 promotes HBV replication [29, 45]. These contradictory findings were probably due to the different HBV models used in each study. An HBV replicon plasmid (pPB) containing 1.04 copies of HBV genome-transfected Huh7 cells was the basis of Liu et al.’s work. In our study, HBV1.3 and HBV1.1 plasmids driving HBV replication models, as well as HBV-infected HepG2^NTCP and Huh7^NTCP cells, were adopted, which are generally accepted models for HBV replication research. In addition, the markers we used to evaluate HBV replication were different from those used by Liu et al.

Viruses have developed multiple strategies to hijack ADAR1 for immune system evasion. Virus-encoded products such as adenovirus VAI RNA and the poxvirus E3L protein have been reported to antagonize ADAR1 deaminase enzymatic activity; however, less is known about virus-mediated regulation of ADAR1 expression [41]. In the present study, we detected high levels of ADAR1 in HBV-infected liver tissues. Viral proteins, particularly HBx, promote ADAR1 transcription to further enhance HBV immune evasion and transcription. HBx has been identified as a critical protein derived from HBV. HBx transactivates a variety of viral and cellular proteins to exert biological effect on the HBV life cycle, specifically by binding TFs such as E2F1, SMAD4, and YY1 [46, 47]. Here, we found that HBx transactivated ADAR1 expression by upregulating the expression of the transcription factor YY1. Knockdown of YY1 suppressed the HBx-mediated enhancement of ADAR1 promoter activity. In accordance with this finding, YY1 has been previously shown to be required for HBV-initiated regulation of the DGCR8 and HLJ1 promoters [48, 49]. It has been suggested that ADAR1-mediated editing of dsRNA structures sets a threshold for intrinsic immune system activation. Therefore, the width of the gap between the amount of cellular duplex RNA and the activation threshold determines how much dsRNA can be tolerated by host cells [20]. According to this hypothesis, HBx-driven ADAR1 expression increases the threshold, which inhibits RNA recognition, promotes immune tolerance, and ultimately allows HBV immune escape from the host immune system. In agreement, our data demonstrated that ectopic HBx expression greatly dampened IFN production in hepatocytes. Our data reveal a novel mechanism of HBx-mediated immune system escape.

ADAR1 is reported to be involved in immune regulation. Loss of ADAR1 in tumors heightens IFN sensitivity to overcome resistance to checkpoint blockade and induces inflammation in tumors by regulating CD8⁺ T-cell function [50]. Consistently, our data showed that abolishment of ADAR1 activity also inflamed HBV-infected livers by increasing the proportion and functions of CD8⁺ T cells. In addition, the proportions of MDSCs and Treg cells, both of which contribute to host immunosuppression during HBV infection [51], were decreased in the livers when ADAR1 activity was inhibited. The decreased proportion of MDSCs and Treg cells may overcome immune tolerance and promote the immune response to HBV, which would further facilitate HBV clearance.

In summary, our present work revealed a potential molecular mechanism underlying HBV evasion from host IFN responses. HBx transcriptionally promotes ADAR1 expression to upregulate the threshold of intrinsic immunity activation and promote HBV evasion from innate immune recognition in hepatocytes (Fig. 7). Loss of ADAR1 or treatment with an ADAR1 inhibitor inflames the liver to promote HBV clearance. Our data explain a molecular mechanism of HBV as a “stealth” virus and suggest a targeted therapeutic intervention for HBV infection.

Fig. 7.

Schematic illustration of the potential mechanism. HBx enhances ADAR1 transcription, which in turn edits HBV RNAs, leading to HBV evasion from innate immunity in hepatocytes

Author contributions

LW and ZCW carried out most of the experiments and analyzed data. YS contributed to the establishment of the protocols for the RIP and RNA pull-down assays. YS, ZHW, YZ, XP, XZ, and CL participated in the in vivo experiments. YKZ was involved with the IHC assay. CG provided help in the IFN pathway analysis. XL, NL, and LG were involved in the study design and manuscript preparation. LW wrote the manuscript with the help of CM. Author CM was in charge of the study design, work organization/supervision, and manuscript review. All authors discussed the results and commented on the manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41423-021-00729-1.