Original research: Targeting ADAR1-mediated RNA Editing Inhibits Hepatic Stellate Cell Activation and Liver Fibrosis by Enhancing HSC-intrinsic Innate Immunity

Abstract

Background:

The pathogenesis of liver fibrosis centers on the activation of hepatic stellate cells (HSCs). A-to-I RNA editing, primarily catalyzed by adenosine deaminase acting on RNA1 (ADAR1), is the most prevalent post-transcriptional modification that increases transcriptome diversity.

Objective:

This study aims to elucidate the role of ADAR1-imposed RNA editome in HSC activation and to determine the therapeutic potential of targeting ADAR1 for liver fibrosis.

Design:

ADAR1 expression was measured in fibrotic human and mouse livers, as well as in primary human and mouse HSCs. Adar1 loss-of-function effect was evaluated in Adar1^f/f/Cre-ER, Adar1^△HSC and Adar1^i△HSC mice, whereas viral infection with Ad-Adar1 was employed in gain-of-function studies. Adar1^△HSCIfih1^−/− and Adar1^△HSCIfnar^△HSC mice were used for mechanistic studies. An ADAR1 inhibitor and HSC-selective RNAi were used for therapeutic evaluations.

Results:

ADAR1 is decreased in human and mouse fibrotic livers and activated HSCs. HSC-specific ablation or pharmacological inhibition of ADAR1 ameliorated HSC activation and liver fibrosis. In contrast, forced expression of ADAR1, but not its editing-deficient mutant, exacerbated HSC activation. Mechanistically, ADAR1 ablation accumulated dsRNA and activated HSC-intrinsic innate immunity in an MDA5-dependent manner. IFN-β was identified as a key anti-fibrotic effector via the activation of JAK1/2 pathway. RNA editome analysis revealed the Col3a1 3’ UTR as a novel ADAR1 editing target, leading to increased collagen production.

Conclusion:

ADAR1-imposed RNA editome suppresses HSC-intrinsic innate immunity and promotes collagen production, leading to aggravated HSC activation and liver fibrosis. Targeting ADAR1 with its pharmacological inhibitor or HSC-selective RNAi shows great promise in treating liver fibrosis.

INTRODUCTION

Liver fibrosis is a wound-healing response to iterative liver injuries, characterized by excessive deposition of extracellular matrix (ECM) proteins¹. Advanced liver fibrosis is a major risk factor for liver failure, portal hypertension, and hepatocellular carcinoma. Hepatic stellate cell (HSC) activation is central to the pathogenesis of liver fibrosis, because activated HSCs are the primary source of fibrogenic cytokines and ECM proteins^2–4. Targeting HSCs represents a promising approach to combat liver fibrosis.

RNA editing is a prevalent post-transcriptional modification that diversifies the transcriptome without increasing the genome size⁵. Adenosine-to-inosine (A-to-I) RNA editing catalyzed by adenosine deaminase acting on RNA (ADAR), primarily ADAR1, accounts for more than 95% of editing events in mammals⁶. ADAR1-mediated RNA editing plays a key role in preventing innate immune response and premature apoptosis by suppressing responses to aberrant dsRNA accumulation^{7, 8}. As a suppressor of dsRNA-inflamed innate immunity, ADAR1 has been mechanistically linked to resistance to cancer immunotherapy^{9, 10}, protection against inflammation¹¹, and multiorgan development¹². In the liver, ADAR1 is essential for embryonic liver development¹³. Hepatocyte-specific ADAR1 ablation induces immune cell infiltration and compromised liver functions¹¹, suggesting a protective role of ADAR1 in hepatocytes. However, the role of ADAR1 and RNA editome in nonparenchymal cells, HSCs in particular, remains largely unknown.

In this study, we uncovered an unexpected role of ADAR1 in facilitating HSC activation and liver fibrosis. HSC-specific ADAR1 ablation attenuated HSC activation and liver fibrosis by activating HSC-intrinsic IFN signaling and reducing collagen production. Targeting ADAR1 by its pharmacological inhibitor or HSC-selective RNAi effectively attenuated liver fibrosis. ADAR1 and RNA editing are promising therapeutic targets for the clinical management of liver fibrosis.

MATERIALS AND METHODS

Mice

Wild-type (WT) C57BL/6J mice (Strain#000664); interferon induced with helicase C domain 1 knockout (Ifih1^−/−) mice (Strain#015812), which lack the Ifih1 gene encoding the melanoma differentiation-associated gene 5 (MDA5); interferon-α/β receptor 1 knockout (Ifnar^-/-) mice (Strain#028288), which lack type I IFN receptor function; and Ifnar ^f/f mice (Strain#028256) were purchased from Jackson Laboratory. Lecithin-retinol acyltransferase (Lrat)-Cre³ and tamoxifen-inducible Lrat-Cre-ERT2 transgenic mice that express Cre recombinase under the control of Lrat promoter and enable Cre-mediated recombination specifically in HSCs. Lrat-Cre-ERT2 transgenic mice, generated by knocking Cre-ERT2 into the Lrat gene locus, were custom produced by Ingenious Targeting Laboratory (Ronkonkoma, NY) following a described strategy¹⁴. Adar1^f/f (Strain#016159) and Cre-ER transgenic mice have been described previously¹¹. Adar1^f/f mice were crossbred with Cre-ER, Lrat-Cre and Lrat-Cre-ERT2 transgenic mice to generate Adar1^f/f/Cre-ER, Adar1^△HSC and Adar1^i△HSC mice, respectively. Adar1^△HSCIfih1^−/− and Adar1^△HSCIfnar^△HSC mice were generated by crossing Adar1^△HSC mice with Ifih1^−/− mice and Ifnar ^f/f mice, respectively.

Patient and public involvement

Patients or the public were not involved in the design, conduct, reporting, or dissemination of this research.

Statistical analysis

Statistical analysis was performed using Prism GraphPad 9.0 software (San Diego, CA). Data are shown as means ± SEM. Statistical significance was assessed by the unpaired two-tailed Student’s t-test between two groups and one-way ANOVA among multiple groups. Statistical significance was set as *p < 0.05 or **p < 0.01.

Mouse models of liver fibrosis; Isolation of primary HSCs and cell culture; Histology and immunostaining; Immunofluorescence; Real-time PCR and RNA editing site-specific qPCR; Adenovirus, plasmids, and cell transfection; Western blot analysis; Matrigel invasion assay; RNA-seq and RNA editome analysis; AEAA-conjugated nanoparticle preparation; ALT and AST

See Supplemental Materials.

RESULTS

ADAR1, highly expressed in HSCs, is decreased in human and mouse fibrotic livers and activated HSCs

We first examined the expression of ADAR family members in primary HSCs and found that ADAR1 was the most highly expressed in both human and mouse (Figure 1A–B) HSCs. Bioinformatic analysis of human liver samples revealed that, compared to healthy individuals, ADAR1 expression was significantly reduced in MASH patients with liver fibrosis, and this reduction was more pronounced in advanced fibrosis with the fibrosis stages confirmed by the expression of fibrogenic marker genes (Figure 1C). Consistent with the role of ADAR1 as an RNA editase, our bioinformatic analysis showed that the A-to-I RNA editing activity was significantly reduced in fibrotic human livers compared with non-fibrotic control livers (Figure S1A–B). Decreased Adar1 expression was also observed in carbon tetrachloride (CCl₄)- and thioacetamide (TAA)-induced mouse liver fibrosis, concomitant with increased expression of fibrogenic marker gene Acta2 (Figure 1D). The suppressed protein expression of ADAR1 in CCl₄-, TAA-, and bile duct ligation (BDL)-induced fibrotic livers was verified by Western blotting (Figure 1E).

Fig. 1. ADAR1, highly expressed in HSCs, is decreased in human and mouse fibrotic livers and activated HSCs.

(A and B) The expression of ADAR family genes in primary human (n=3, GSE152250) and mouse HSCs (n=4, GSE256502).

(C) Hepatic mRNA expression in patients with normal or MASH livers (GSE162694, normal n=31, fibrosis stage 0–2 n=92, fibrosis stage 3–4 n=20).

(D and E) Hepatic mRNA (D) and protein (E) expression in mouse models of liver fibrosis.

(F) ADAR1 expression in primary human hepatocytes (HEPs), quiescent HSCs (qHSCs), and activated HSCs (aHSCs) (GSE68000, HEP n=2 donors, qHSC n=3, aHSC n=3; GSE152250, n=3 per group).

(G) ADAR1 protein expression in primary human HSCs activated in culture.

(H) Representative images showing ADAR1 (red), α-SMA (green) and nuclei (blue) in liver sections of patients with MASLD or MASH with advanced fibrosis. Scale bar, 100 μm. Quantification of fluorescence intensity is shown on the right.

(I-L) Adar1 expression in primary mouse hepatocytes and HSCs isolated from mice fed a 16-week fructose-palmitate-cholesterol MASH diet (I, GSE256502, n=4 per group), in HSCs activated in culture for ten days (J, n=3 per group), in HSCs isolated from mice subjected to sham surgery, BDL, TAA, and CCl₄ models (K, n=3 per group), and in HSCs and LSECs isolated from CCl₄-treated mice (L, GSE120281, n=3 per group).

Data are shown as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

The fibrosis-responsive downregulation of ADAR1 was evident in activated human and mouse HSCs. Compared to human hepatocytes, ADAR1 expression was higher in quiescent human HSCs, but it decreased upon culture-induced¹⁵ or TGFβ1-stimulated¹⁶ HSC activation (Figure 1F). The decreased protein expression of ADAR1 in activated human HSCs was confirmed (Figure 1G). ADAR1 expression was also evaluated in human liver specimens from individuals with non-fibrotic metabolic dysfunction-associated steatotic liver disease (MASLD) or metabolic dysfunction-associated steatohepatitis (MASH) with advanced fibrosis. In non-fibrotic MASLD livers, ADAR1 was highly expressed in HSCs, as confirmed by its co-localization with α-SMA immunofluorescence (Figure 1H). In contrast, HSC expression of ADAR1 was markedly reduced in fibrotic MASH livers (Figure 1H). Adar1 expression was higher in quiescent HSCs than in hepatocytes isolated from the same intact mice (Figure 1I). Adar1 expression in HSCs was decreased upon in vivo activation induced by the feeding of a 16-week fructose-palmitate-cholesterol MASH diet¹⁷ (Figure 1I) or upon in vitro activation in cell culture (Figure 1J). The suppression of Adar1 was also observed in activated HSCs isolated from mice subjected to the BDL, TAA or CCl₄ models of liver fibrosis (Figure 1K). In the CCl₄ model, the downregulation of Adar1 was specifically observed in HSCs, but not in liver sinusoidal endothelial cells (LSECs) (Figure 1L).

HSC-specific ADAR1 ablation ameliorates HSC activation and liver fibrosis

To determine the role of ADAR1 in HSC activation, the Adar1^f/f/Cre-ER mice were generated by crossbreeding Adar1^f/f mice with Cre-ER transgenic mice expressing Cre-ER under the control of the CMV early enhancer (CAG) promoter (Figure 2A). Primary HSCs isolated from Adar1^f/f/Cre-ER mice were treated with 4-hydroxytamoxifen (4-OHT) to induce Adar1 deletion (Figure 2B). The decreased mRNA and protein expression of ADAR1 was verified (Figure 2C–D). Adar1 deletion markedly inhibited culture-induced HSC activation, as shown by decreased mRNA and protein expression of fibrogenic genes (Figure 2C–E). Adar1 ablation also suppressed HSC proliferation and invasion, as shown by Ki67 immunostaining and matrigel invasion assay (Figure 2E–F). Gene set enrichment analysis (GSEA) of our RNA-seq data revealed significant downregulation of fibrogenic pathways in 4-OHT-treated Adar1^f/f/Cre-ER HSCs, including ECM, Epithelial mesenchymal transition, Cell cycle, TGFβ signaling pathway, and Platelet derived growth factor (PDGF) binding (Figure 2G). As expected, A-to-I editing activity, measured by the ratio of edited to unedited WT Azin1 transcripts, was reduced in Adar1-deleted HSCs, concomitant with increased endogenous dsRNA accumulation (Figure 2H–I). The potential inhibitory effect of 4-OHT on HSC activation was ruled out, as 4-OHT treatment had minimal impact on fibrogenic gene expression in WT HSCs (Figure S1C). A similar inhibitory effect of Adar1 depletion on HSC activation was observed in WT HSCs infected with adenovirus expressing Adar1 shRNA (Ad-shAdar1) (Figure S1D–E).

Fig. 2. ADAR1 ablation inhibits HSC activation.

(A) Schematic representation of the generation of Adar1^f/f/Cre-ER mice and tamoxifen-inducible Adar1 deletion.

(B-I) Primary HSCs isolated from Adar1^f/f/Cre-ER mice were treated with vehicle or 4-OHT (100 nM) and activated in culture for four days (n=3). Shown are 4-OHT-induced Adar1 deletion (B), the mRNA (C) and protein (D) expression of ADAR1 and fibrogenic marker genes, immunofluorescence of α-SMA and Ki67 with the quantifications shown on the right, scale bar, 100 μm (E), matrigel invasion (F), GSEA (G), A-to-I RNA editing activity (H), and immunostaining of dsRNA using J2 antibody with the quantification shown on the right (I).

(J) Gene expression in 4-OHT-treated Adar1^f/f/Cre-ER HSCs stimulated by TGFβ1 (5 ng/mL) for 24 h (n=3).

Data are presented as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

TGFβ1 is a potent pro-fibrogenic cytokine. We then determined whether Adar1 deletion also inhibits TGFβ1-stimulated HSC activation. TGFβ1 stimulation induced robust upregulation of fibrogenic genes as expected, but this effect was abolished in 4-OHT-treated Adar1^f/f/Cre-ER HSCs (Figure 2J). The decreased sensitivity of Adar1-deficient HSCs to TGFβ1-stimulated activation was verified by reduced protein expression of α-SMA and Ki67 (Figure S1F–G). ADAR1 has two isoforms transcribed from distinct promoters: the constitutively expressed nuclear p110 and the IFN-inducible cytoplasmic p150^{18, 19}. To determine the isoform-specific role of ADAR1 in fibrogenesis, we designed modified siRNA that specifically targets ADAR1 p150 while preserving p110 expression (Figure S1H). Selective knockdown of p150 significantly inhibited TGFβ1-stimulated activation of the human hepatic stellate LX2 cells (Figure S1I), indicating that inhibition of ADAR1 p150 is sufficient to suppress HSC activation.

To investigate the effect of HSC-specific Adar1 ablation in vivo, we generated HSC-specific Adar1 knockout (Adar1^△HSC) and tamoxifen-inducible HSC-specific Adar1 knockout (Adar1^i△HSC) mice, as outlined in Figure 3A–B. Across 16 breeding events from matings between male Adar1^f/f and female Adar1^△HSC mice, a total of 25 Adar1^f/f and 22 Adar1^△HSC male pups, and 29 Adar1^f/f and 10 Adar1^△HSC female pups were born and developed with grossly normal appearance into adulthood (Figure S2A), indicating that HSC-specific Adar1 deletion circumvents the lethality observed in global Adar1 knockout mice. Moreover, Adar1 ablation did not induce HSC death, as evidenced by undetectable apoptosis in TUNEL staining and negligible basal expression of the apoptosis marker cleaved Caspase-3 and the necroptosis markers pRIPK3 and pMLKL in Western blotting, while the Adar1 ^i△HSC HSCs remained responsive to apoptosis or necroptosis challenges induced by TNFα, SM-164, and/or Z-VAD-FMK (Figure S2B–D). Furthermore, perfusions of Adar1^f/f and Adar1^i△HSC mouse livers yielded comparable numbers of primary HSCs (Figure S2E).

Fig. 3. HSC-specific ADAR1 ablation ameliorates HSC activation and liver fibrosis.

(A-B) Schematic representation of the generation of Adar1^△HSC and Adar1^i△HSC mice.

(C-F) Fibrogenic gene expression and immunostaining of α-SMA and Ki67 with the quantifications shown on the right, scale bar, 100 μm, in Adar1^△HSC HSCs activated in culture for five days (C-D) or stimulated by TGFβ1 (5 ng/mL) for 24 h (E-F) (n=3).

(G-H) Eight-week-old male Adar1^f/f and Adar1^△HSC mice were subjected to the CCl₄ model (n=4 per group). Sirius Red staining and α-SMA immunostaining with the quantifications shown on the right, scale bar, 200 μm (G), and hepatic mRNA expression (H).

(I-J) Eight-week-old male Adar1^f/f (n=3) and Adar1^△HSC (n=4) mice were subjected to the BDL model. Shown are Sirius Red staining and α-SMA immunostaining with the quantifications shown on the right, scale bar, 200 μm (I), and hepatic mRNA expression of Ifih1 and fibrogenic genes (J).

(K) Protein expression of ADAR1 in primary HSCs isolated from Adar1^f/f and Adar1^i△HSC mice assessed by Western blotting.

(L) Eight-week-old male Adar1^f/f and Adar1^i△HSC mice were fed an MCD diet for 8 weeks (n=6 per group). Shown are Sirius Red staining and α-SMA immunostaining of liver sections with the quantifications shown on the right, scale bar, 200 μm.

Data are presented as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

Primary HSCs isolated from Adar1^△HSC mice were less sensitive to culture-induced (Figure 3C–D) and TGFβ1-stimulated (Figure 3E–F) activation, as shown by reduced fibrogenic genes and immunofluorescence of α-SMA and Ki67. In vivo, Adar1^△HSC mice were largely protected from CCl₄-induced liver fibrosis, as shown by reduced Sirius Red staining of collagen and α-SMA immunostaining, as well as decreased expression of fibrogenic genes, along with the induction of dsRNA sensor gene Ifih1 (Figure 3G–H). Adar1^△HSC mice were also protected from BDL-induced liver fibrosis (Figure 3I–J). Consistent with the protection observed in Adar1^△HSC mice, the tamoxifen-inducible and post-developmental Adar1^i△HSC mice were resistant to methionine/choline-deficient (MCD) MASH diet-induced liver fibrosis, with the ADAR1 deficiency in HSCs confirmed (Figure 3K–L).

Forced expression of ADAR1, but not its editing-deficient mutant, promotes HSC activation

We next examined whether forced ADAR1 expression exacerbates HSC activation. Primary mouse or human HSCs infected with adenovirus expressing mouse Adar1 (Ad-Adar1) or the control virus (Ad-Ctrl) were activated in culture for four days. Ad-Adar1 infection accelerated mouse HSC activation (Figure 4A–C). As expected, ADAR1 overexpression increased A-to-I RNA editing and decreased dsRNA accumulation (Figure 4D–E). The pro-fibrogenic effect of adenoviral ADAR1 expression was also observed in primary human HSCs, concomitant with increased A-to-I editing activity (Figure 4F–G).

Fig. 4. Forced expression of ADAR1, but not its editing-deficient mutant, promotes HSC activation.

(A-E) Primary mouse HSCs infected with Ad-Ctrl or Ad-Adar1 were activated in culture for four days (n=3). Shown are the mRNA (A) and protein (B) expression of ADAR1 and fibrogenic genes, immunofluorescence of α-SMA and Ki67 with the quantifications shown on the right, scale bar, 100 μm (C), A-to-I RNA editing activity (D), and dsRNA immunostaining with the quantification shown on the right (E).

(F-G) Fibrogenic gene expression (F) and A-to-I editing activity (G) in primary human HSCs infected with Ad-Ctrl or Ad-Adar1 and activated in culture for four days (n=3).

(H-L) Primary mouse HSCs were infected with Ad-Ctrl, Ad-Adar1, or Ad-Adar1 E861A and activated in culture for four days (n=3). Shown are fibrogenic gene expression (H), protein expression of ADAR1 and α-SMA (I), immunofluorescence of α-SMA and Ki67 with the quantifications shown on the bottom left, scale bar, 100 μm (J), mRNA expression of Adar1 (K), and A-to-I RNA editing activity (L).

Data are shown as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

To determine whether the editing activity of ADAR1 is required for its fibrogenic effect, we infected primary mouse HSCs with Ad-Adar1 or Ad-Adar1 E861A, the editing-deficient glutamate (E)-861 to alanine (A) mutant⁸. Infection with Ad-Adar1, but not E861A, promoted HSC activation (Figure 4H–K). As expected, E861A overexpression had little effect on A-to-I RNA editing (Figure 4L). Taken together, these results demonstrated that the catalytic activity of ADAR1 is required for its fibrogenic effect.

ADAR1 ablation inhibits liver fibrosis by triggering HSC-intrinsic innate immunity in an MDA5- and IFN-β-dependent manner

To elucidate the mechanism by which ADAR1 ablation inhibits HSC activation, we performed Gene Ontology (GO) analysis on RNA-seq data from vehicle- or 4-OHT-treated Adar1^f/f/Cre-ER HSCs. The top downregulated genes in Adar1-deficient HSCs were highly enriched in categories of ECM deposition, whereas the top upregulated gene-related pathways were IFN production and response (Figure 5A and S3A), suggesting that the anti-fibrogenic effect of ADAR1 ablation may have been accounted for by activation of HSC-intrinsic innate immunity, specifically IFN signaling. Elevated production of and response to type I and type II IFNs were shown by GSEA plots (Figure 5B–C and S3B-C). IFN-β, also known as fibroblast IFN, is primarily produced by fibroblasts²⁰. Among IFNs, IFN-β showed the highest induction in both expression and secretion (Figure 5D–E). Consistent with the induction of IFNs, Adar1 ablation robustly upregulated the transcription of interferon-stimulated genes (ISGs), including Isg15, Mx1, and Oas1, as well as ISG-encoded IFN pathway components (Figure 5F–G). In contrast, adenoviral overexpression of ADAR1 in HSCs suppressed IFN signaling (Figure S3D–F).

Fig. 5. ADAR1 ablation inhibits liver fibrosis by activating HSC-intrinsic innate immunity in an MDA5- and IFN-β-dependent manner.

(A) Top GO categories with genes downregulated or upregulated by 4-OHT treatment in Adar1^f/f/Cre-ER HSCs. Pathways were ranked by signed log₁₀FDR.

(B and C) GSEA of RNA-seq data derived from vehicle- or 4-OHT-treated Adar1^f/f/Cre-ER HSCs.

(D-G) Adar1^f/f/Cre-ER HSCs treated with 4-OHT were activated in culture for four days (n=3). Shown are the mRNA expression of IFNs (D), secreted IFNs in culture medium (E), and the mRNA expression of ISG marker genes (F) and IFN pathway components (G).

(H) GSEA of dsRNA-related gene clusters in 4-OHT-treated Adar1^f/f/Cre-ER HSCs.

(I-L) Primary mouse HSCs were treated with poly(I:C)/LyoVec^™ (500 ng/mL) for four days (n=3). Shown are fibrogenic gene expression (I), immunofluorescence of α-SMA and Ki67 with the quantifications shown on the right, scale bar, 100 μm (J), and mRNA (K) and secreted protein levels of IFNs (L).

(M) Gene expression in Ifih1^−/− HSCs treated with poly(I:C)/LyoVec^™ (500 ng/mL) (n=3).

(N and O) Eight-week-old male Adar1^f/f, Adar1^△HSC and Adar1^△HSCIfih1^−/− mice were subjected to the CCl₄ model (n=4 per group). Shown are Sirius Red staining and α-SMA immunostaining with the quantifications shown on the right, scale bar, 200 μm (N) and hepatic gene expression (O).

(P and Q) Gene expression in WT and Ifih1^−/− HSCs treated with 8-Aza (1 μM) for four days (n=3).

(R) Adar1^f/f/Cre-ER HSCs incubated with IFN-β Ab or IgG (15 μg/mL) were treated with 4-OHT and activated in culture for four days. Shown is fibrogenic gene expression (n=3).

(S and T) Gene expression in primary HSCs isolated from WT mice and activated in culture in indicated conditional medium (CM) (n=3).

Data are shown as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

The embryonic lethality observed in mice with whole-body knockout of Adar1 or knock-in of its enzyme-dead mutant was fully rescued by concurrent deletion of the dsRNA sensor MDA5^{8, 21}, suggesting that the major function of ADAR1-catalyzed RNA editing is to prevent the sensing of unedited dsRNA as non-self. Indeed, GSEA analyses revealed significant upregulation of dsRNA response and binding in 4-OHT-treated Adar1^f/f/Cre-ER HSCs (Figure 5H). To determine the effect of dsRNA sensing on HSC activation, primary mouse HSCs were treated with polyinosinic-polycytidylic acid (poly(I:C))/LyoVec^™, a complex of synthetic dsRNA analog poly(I:C) and transfection reagent LyoVec^™, which triggers MDA5-mediated dsRNA sensing while avoiding TLR3 recognition of naked poly(I:C)^{22, 23}. Poly(I:C)/LyoVec^™ treatment inhibited HSC activation (Figure 5I–J), concomitant with increased expression and secretion of IFN-β (Figure 5K–L) and augmented ISG signatures (Figure S3J–H).

Having shown that dsRNA analog can directly inhibit HSC activation and knowing that MDA5 is a primary sensor of dsRNA, we speculated that MDA5 is required for the anti-fibrotic effect of ADAR1 ablation. Indeed, the inhibition of fibrogenic gene expression by poly(I:C)/LyoVec^™ was entirely abrogated in HSCs isolated from Ifih1^−/− mice (Figure 5M). In vivo, Adar1^△HSC mice were protected from CCl₄-induced liver fibrosis, but the protection was abolished when Ifih1 was concurrently ablated (Figure 5N–O). Intriguingly, among the fibrogenic genes, Col3a1 expression was not fully restored to the level of Adar1^f/f mice (Figure 5O). In the pharmacological model, the ADAR1-selective inhibitor 8-Aza²⁴ inhibited fibrogenic gene expression in WT HSCs, and this inhibitory effect was abolished in Ifih1^−/− HSCs, accompanied by induction and loss of induction of Ifnb1, respectively (Figure 5P–Q). Furthermore, Ifih1^−/− mice were more susceptible to CCl₄-induced liver fibrosis (Figure S4A–B). HSCs isolated from Ifih1^−/− mice exhibited enhanced activation in culture (Figure S4C), accompanied by a decreased ISG signature (Figure S4D).

Given that IFN-β was most highly induced and secreted in Adar1-deficient HSCs and poly(I:C)-transfected HSCs (Figure 5D–E and 5K–L), we next investigated whether HSC-intrinsic IFN-β production and secretion were required for the anti-fibrotic effect of ADAR1 ablation. Adar1^f/f/Cre-ER HSCs were treated with anti-IFN-β neutralizing antibody (IFN-β Ab) or control IgG prior to 4-OHT treatment. The suppression of fibrogenic genes by 4-OHT-induced Adar1 deletion was notably reversed by IFN-β Ab (Figure 5R). Moreover, conditional medium (CM) collected from 4-OHT-treated Adar1^f/f/Cre-ER HSCs inhibited the activation of WT HSCs in cell culture, but this effect was abolished by IFN-β Ab (Figure 5S). The CM-responsive induction of ISG signature and the loss of induction by IFN-β Ab were verified by the expression of the ISG Ifih1 (Figure 5T).

The anti-fibrotic effect of ADAR1 ablation is dependent on the IFNAR-JAK1/2 signaling axis

IFN-β binds to the type I IFN receptor (IFNAR), activating the JAK-STAT pathway through tyrosine phosphorylation and initiating ISG transcription^{25, 26}. Consistent with increased IFN-β, Adar1-deficient HSCs exhibited upregulation of Receptor signaling pathway via STAT, JAK-STAT signaling pathway, and Tyrosine phosphorylation of STAT protein (Figure 6A).

Fig. 6. The anti-fibrotic effect of ADAR1 ablation is dependent on the IFNAR-JAK1/2 signaling axis.

(A) GSEA of RNA-seq data derived from 4-OHT-treated Adar1^f/f/Cre-ER HSCs.

(B and C) Eight-week-old male Adar1^f/fIfnar^f/f, Adar1^△HSC, and Adar1^△HSCIfnar^△HSC mice were subjected to the CCl₄ model (n=4 per group). Shown are Sirius Red staining and α-SMA immunostaining with the quantifications shown on the right, scale bar, 200 μm (B), and hepatic gene expression (C).

(D-F) Adar1^f/f/Cre-ER HSCs were treated with 4-OHT and Ruxolitinib (10 μM) for four days (n=3). Shown are fibrogenic gene expression (D), immunofluorescence of α-SMA and Ki67 with the quantifications shown on the right, scale bar, 100 μm (E), and protein expression of α-SMA (F).

(G and H) Eight-week-old male Adar1^f/f and Adar1^△HSC mice subjected to the CCl₄ model were treated with Ruxolitinib (60 mg/kg body weight) by intraperitoneal injection daily (n=4 per group). Shown are Sirius Red staining and α-SMA immunostaining with the quantifications shown on the bottom right, scale bar, 200 μm (G), and hepatic gene expression (H).

Data are shown as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

To determine whether IFN signaling and its downstream JAK pathway are required for the anti-fibrotic effect of ADAR1 ablation, we generated HSC-specific Adar1 and Ifnar double-knockout (Adar1^△HSC Ifnar^△HSC) mice by breeding the Ifnar^f/f allele into Adar1^△HSC mice. Adar1^△HSCIfnar^△HSC mice lacked the protective effects of Adar1 ablation against CCl₄-induced fibrosis (Figure 6B). Fibrogenic gene expression in Adar1^△HSCIfnar^△HSC mice was restored to levels comparable to Adar1^f/fIfnar ^f/f mice, except for Col3a1, along with the loss of induction of the ISG Ifih1, indicative of IFN signaling blockade (Figure 6C). These data demonstrate that ADAR1 ablation alleviates fibrosis by activating IFN signaling in an HSC-autonomous and autocrine manner.

Treatment with the JAK1/2 selective inhibitor Ruxolitinib¹⁰ abolished the suppression of activation in 4-OHT-treated Adar1^f/f/Cre-ER HSCs (Figure 6D–F). In vivo, Ruxolitinib exacerbated CCl₄-induced liver fibrosis in Adar1^f/f mice and attenuated the protective effect in Adar1^△HSC mice, as shown by liver histology and fibrogenic gene expression, but again with the exception of Col3a1 (Figure 6G–H). Ruxolitinib sensitized WT mice to CCl₄-induced fibrosis (Figure S5A–B). The incomplete restoration of Col3a1 expression in Adar1^△HSCIfih1^−/−, Adar1^△HSCIfnar^△HSC, and Ruxolitinib-treated Adar1^△HSC mice (Figure 5O, 6C, and 6H) suggests that ADAR1 regulates Col3a1 through a distinct mechanism independent of dsRNA sensing and IFN signaling.

Suppressed ISG signature upon HSC activation regulates ADAR1 expression

ADAR1 is an ISG containing an IFN-stimulated response element in its promoter²⁷. Upon activation in cell culture, HSC expression of IFNA1, IFNB1, and IFNG was significantly reduced, accompanied by downregulated ISG marker genes and ADAR1 (Figure S6A–B). ISG suppression was also observed in activated HSCs isolated from mice subjected to BDL, TAA, or CCl₄ models of liver fibrosis (Figure S6C). IFN-β treatment of primary HSCs robustly upregulated both mRNA and protein expression of ADAR1 (Figure S6D–E), suggesting that constitutive IFN signaling in quiescent HSCs maintains the ISG signature, whereas decreased IFNs during HSC activation downregulated ISGs, including ADAR1 and dsRNA effectors. This downregulation of downstream effectors counteracts the fibro-protective impact of ADAR1 downregulation during fibrogenesis. In contrast, therapeutic intervention or genetic ablation of ADAR1 activates dsRNA sensing and IFN signaling, leading to JAK-STAT pathway activation and ISG transcription, which in turn enhances HSC innate immunity through upregulated effector ISGs, such as Ifih1 (MDA5). IFN-β treatment robustly activated downstream IFN signaling in HSCs despite their reduced Ifnar expression, as evidenced by robust ISG induction, concomitant with suppressed fibrogenic gene expression (Figure S6F–H). The reduction in Ifnar expression following IFN-β treatment may reflect a compensatory response to IFN-β-induced overactivation of IFN signaling.

Editome analysis identifies Col3a1 as an ADAR1 editing target that facilitates collagen production

To gain molecular insight into ADAR1-catalyzed editing in fibrogenesis, we performed editome analysis using the established pipeline REDItools²⁸ on transcriptome data from Adar1^f/f/Cre-ER HSCs treated with vehicle or 4-OHT, identifying 2527 ADAR1 editing sites (Supplementary Table 1). GSEA revealed that edited RNA species were largely clustered in fibrogenic pathways, as shown by KEGG, GO, and Hallmark gene sets (Figure 7A–C). Editing analysis on transcripts of canonical fibrogenic genes further revealed that Adar1 deletion abolished nearly all A-to-I editing events in many of the genes involved in extracellular matrix (ECM) remodeling, pro-fibrogenic signaling, and growth factors and receptors (Figure S7A), indicating that ADAR1 orchestrates a broad pro-fibrogenic RNA editing network. To identify downstream effector targets of ADAR1 editing, 233 edited genes involved in fibrosis and RNA metabolism were selected for further analysis (Supplementary Table 2). Of these, 10 genes exhibited significant expression changes in Adar1-deficient HSCs (Supplementary Table 3). Figure 7D illustrates our unbiased screening using editome analysis, GSEA, and RNA expression profile. Among these 10 genes, the collagen gene Col3a1 was notable for a 3’ UTR editing site (genome location 45387952, 3’ UTR +45) with an editing frequency of 52% in vehicle-treated Adar1^f/f/Cre-ER HSCs, which decreased to 0% in 4-OHT-treated Adar1^f/f/Cre-ER HSCs, accompanied by approximately 5-fold downregulation of Col3a1 transcripts (Figure 7E).

Fig. 7. Editome analysis identifies Col3a1 as an ADAR1 editing target that facilitates collagen production.

(A-C) RNA editome analyses of vehicle- or 4-OHT-treated Adar1^f/f/Cre-ER HSCs. Shown are GSEA of genes containing A-to-I editing sites using KEGG (A), GO (B) and Hallmark gene sets (C).

(D) Schematic representation of unbiased screening of edited RNA species using RNA editome analysis, GSEA, RNA expression profile to identify transcripts involved in ADAR1-facilitated fibrogenesis.

(E) The levels of Col3a1 RNA transcripts and A-to-I editing site of Col3a1 by the combined RNA-seq and RNA editome analyses.

(F) 293T and LX2 cells transfected with reporter genes containing WT or mutant 3’ UTR of Col3a1 downstream of luciferase in the absence or presence of shAdar1 co-transfection and TGFβ1 (10 ng/mL) stimulation (n=4–5 per group).

(G) The mRNA expression of Col3a1 in 293T and LX2 cells transfected with Col3a1-3’ UTR WT or △45 (n=3).

(H) COL3A1 protein expression in 293T and LX2 cells transfected with Col3a1-3’ UTR WT or △45.

Data are shown as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

The 3’ UTRs commonly regulate mRNA fate through post-transcriptional modification^{29, 30}. Bioinformatic visualization of the RNA secondary structure (http://rna.tbi.univie.ac.at/forna/) of Col3a1 3’ UTR revealed an inverted repeat double-stranded structure at the editing site (3’ UTR WT, Figure S7B). To assess its functional relevance, we constructed an editing-resistant deletion mutant (3’ UTR△45, Figure S7C) maintaining a high structural consistency with the 3’ UTR WT compared with single-base substitutions (Figure S7D). Transactivation of the reporter genes containing the 3’ UTR WT or △45 downstream of luciferase was compared in 293T and LX2 cells. In both cell lines, the activity of 3’ UTR WT reporter was increased by TGFβ1 as expected, while deletion of 3’ UTR editing site or shAdar1 co-transfection reduced both basal and TGFβ1-stimulated reporter activity (Figure 7F). Moreover, the 3’ UTR△45 reporter failed to respond to Adar1 knockdown (Figure 7F). We then constructed Col3a1 expression plasmids containing 3’ UTR WT or △45 downstream. Consistent with the reporter assays, the mRNA and protein expression of COL3A1 was markedly reduced in both 293T and LX2 cells transfected with Col3a1-3’ UTR△45 compared with Col3a1-3’ UTR WT (Figure 7G–H). The decreased expression of COL3A1 with editing-resistant 3’ UTR△45 was consistent with our observation that the expression of Col3a1 remained downregulated in Adar1^△HSCIfih1^−/−, Adar1^△HSCIfnar^△HSC, and Ruxolitinib-treated Adar1^△HSC mice (Figure 5O, 6C, and 6H).

The 3’ UTR of Col3a1 is a regulatory hub where trans-acting factors, including RNA-binding proteins, miRNAs, and long noncoding RNAs (lncRNAs), bind to and regulate Col3a1 expression and collagen synthesis^31–33. To identify trans-acting factors affected by ADAR1-mediated editing, we examined Col3a1 expression following individual knockdown or inhibition of these factors. As expected, the editing-resistant Col3a1-3’ UTR△45 showed markedly reduced expression of Col3a1 compared with Col3a1-3’ UTR WT (Figure S7E). Knockdown of the positive regulators DNM3OS and hnRNPA1 decreased Col3a1 expression in the 3’ UTR WT, but not the △45 construct (Figure S7E). Knockdown of hnRNPK, another positive regulator, reduced Col3a1 expression of both WT and △45 constructs (Figure S7E), suggesting an editing-independent regulation. Conversely, inhibition of the negative regulators miR-29a, miR-29b, and miR-29c increased Col3a1 expression of the 3’ UTR WT, but not the △45 construct (Figure S7E). Together, these results suggest that ADAR1-mediated editing is required for the regulatory effects of Col3a1 trans-acting factors DNM3OS, hnRNPA1, miR-29a, miR-29b, and miR-29c.

As summarized in Figure S8, our findings demonstrate that ADAR1 in HSCs exerts a pro-fibrogenic function through two coordinated mechanisms. On one hand, ADAR1-mediated editing within the 3’ UTR of Col3a1 promotes Col3a1 mRNA accumulation and increases the expression of this collagen isoform. On the other hand, ADAR1 suppresses HSC-intrinsic innate immune activation by inhibiting MDA5-mediated dsRNA sensing, thereby reducing IFN-β production and inhibiting the downstream IFNAR-JAK-STAT signaling (Figure S8). Both effects converge to facilitate HSC activation and liver fibrosis. In contrast, ADAR1 plays a protective role in hepatocytes by restraining NF-κB and IFN signaling-mediated inflammation and liver injury through inhibition of aberrant cytosolic MDA5- or RIG-I-dsRNA sensing^{11, 34, 35}.

Targeting ADAR1 ameliorates liver fibrosis and accelerates fibrosis resolution

To determine the therapeutic potential of ADAR1 inhibition in liver fibrosis, we first examined the efficacy of HSC-selective RNAi in reversing the progression of established fibrosis. A pre-existing liver fibrosis was established by 3-week CCl₄ administrations^{2, 36}. After the 3-week initiation, WT mice were intravenously injected with our previously described HSC-selective aminoethyl anisamide (AEAA)-conjugated liposome-protamine-hyaluronic acid (LPH)-siRNA nanoparticles^{2, 37} twice a week for 3 weeks, while the CCl₄ administrations continued (Figure 8A). The efficiency and HSC selectivity of Adar1 knockdown were confirmed by reduced Adar1 expression in HSCs, but not hepatocytes, two days after the last AEAA-LPH-siRNA injection (Figure 8B). Reduced protein expression of ADAR1 was further confirmed in isolated HSCs (Figure 8C). HSC-specific siAdar1 delivery attenuated the progression of pre-established fibrosis, as shown by reduced collagen and α-SMA staining and downregulation of fibrogenic genes (Figure 8D–F), without affecting serum levels of ALT and AST (Figure S9A). Consistent with the observation in Adar1^△HSCIfih1^−/− (Figure 5N) and Adar1^△HSCIfnar^△HSC mice (Figure 6B), the therapeutic effect of siAdar1 was entirely abrogated in Ifih1^−/− and Ifnar^−/− mice (Figure 8D–E), further supporting the requirement of MDA5 and IFNAR for the anti-fibrotic effect of ADAR1 inhibition. The loss of the anti-fibrotic effect of siAdar1 in Ifnar^−/− mice was confirmed by the restoration of fibrogenic gene expression except for Col3a1 (Figure 8F).

Fig. 8. Targeting ADAR1 ameliorates the development and progression of liver fibrosis.

(A-D) CCl₄-treated WT and Ifih1^−/− mice were subjected to AEAA-LPH-siAdar1 nanoparticle regimen (A). Shown are Adar1 expression in HEPs and HSCs (B) and ADAR1 protein levels in HSCs isolated two days after the last siRNA injection (C), and Sirius Red staining and α-SMA immunostaining with the quantifications shown on the right, scale bar, 200 μm (n=4 per group) (D).

(E and F) CCl₄-treated WT and Ifnar^−/− mice were subjected to AEAA-LPH-siAdar1 nanoparticle regimen (n=4 per group). Shown are Sirius Red staining and α-SMA immunostaining with the quantifications shown on the upper right, scale bar, 200 μm (E), and fibrogenic gene expression (F).

(G-L) Primary mouse (G-I) and human (J-L) HSCs were treated with 8-Aza for four days (n=3).

(M-O) Eight-week-old male WT mice subjected to the CCl₄ model were treated with 8-Aza (10 or 20 mg/kg body weight) by intraperitoneal injection three times a week for 6 weeks (M) (n=4 per group). Shown are Sirius Red staining and α-SMA immunostaining with the quantifications shown on the right, scale bar, 200 μm (N), and gene expression in HSCs isolated at the end of experiments (O).

Data are shown as mean ± SEM. *p < 0.05, **p < 0.01. NS, statistically not significant.

We next explored the therapeutic utility of ADAR1-selective pharmacological inhibitor 8-Aza in HSC activation and liver fibrosis. Primary mouse and human HSCs were treated with vehicle or 8-Aza and activated in culture for four days. Treatment with 8-Aza inhibited the activation of mouse (Figure 8G–I) and human (Figure 8J–L) HSCs in a dose-dependent manner. To evaluate the anti-fibrotic effect of 8-Aza in vivo, WT mice subjected to the CCl₄ model were treated with 8-Aza by intraperitoneal injections three times a week for 6 weeks (Figure 8M). Treatment with 8-Aza largely protected mice from CCl₄-induced liver fibrosis in a dose-dependent manner (Figure 8N), without elevating serum levels of ALT and AST (Figure S9B). The fibro-protective effect of 8-Aza was further confirmed by decreased fibrogenic gene expression in HSCs isolated from 8-Aza-treated mice (Figure 8O).

To evaluate the efficacy of 8-Aza in promoting fibrosis resolution and exclude its potential general toxicity, we extended the 8-Aza treatment to up to 12 weeks. After the completion of 6-week CCl₄ administration, WT mice continued receiving vehicle or 8-Aza for an additional 3 or 6 weeks, as outlined in Figure S9C. Our results showed that 8-Aza treatment improved fibrosis resolution by week 9 and nearly fully restored liver architecture by week 12, as indicated by Sirius Red staining and α-SMA immunostaining (Figure S9D). The body weight did not differ significantly among the four groups (Figure S9E). Histopathological examinations of the heart, kidney, and skeletal muscle revealed no discernible abnormalities at either week 9 or week 12 (Figure S9F). Serum LDH levels were significantly reduced in mice treated with 8-Aza for 9 weeks and continued to decline in both vehicle- and 8-Aza-treated mice by week 12 (Figure S9G). These findings indicate that our regimen of 8-Aza was well tolerated in non-hepatic tissues. Taken together, targeting ADAR1 effectively attenuates liver fibrosis and accelerates fibrosis resolution.

DISCUSSION

Liver fibrosis is a progressive disease that predisposes the liver to cirrhosis and malignancy but lacks FDA-approved pharmacotherapies in part due to an incomplete understanding of the intrinsic regulation of HSC activation. Our study uncovers an unexpected role of ADAR1 in facilitating HSC activation and liver fibrosis. Inhibition of ADAR1 by pharmacological intervention or HSC-targeting RNAi prevented and reversed the progression of liver fibrosis.

The endogenous cellular dsRNAs are primarily derived from inverted repeat Alu retroelement transcripts, endogenous retroviral transcripts, mitochondrial dsRNAs, and structured RNAs, such as lncRNAs and pre-miRNAs^38–41. As an RNA-editing enzyme, ADAR1 suppresses immune response by preventing the sensing of dsRNA as non-self^{8, 42, 43}. Having shown that the editing activity of ADAR1 is indispensable for its pro-fibrogenic function, our mechanistic studies integrating RNA-seq and editome analyses revealed that the anti-fibrotic effect of ADAR1 deficiency resulted from altered RNA editome and enhanced HSC-intrinsic IFN signaling. A key novelty of our study is the discovery that HSC-intrinsic IFN signaling and ISG signature can dictate HSC activation and liver fibrosis. While interferons, IFN-γ in particular, have been implicated in inhibiting HSC activation and alcohol-associated liver fibrosis, the reported effects have been attributed to the IFN-γ-producing NK cells that are cytotoxic to activated HSCs^44–46. In the current study, we showed that HSCs possess an intrinsic ISG signature that declines with the onset of HSC activation, potentially contributing to activation. Moreover, ADAR1 inhibition in HSCs enhances the ISG signature through MDA5-dsRNA sensing and IFN signaling. Interestingly, IFN-β seems to play a central fibro-protective role in HSCs, whereas IFN-γ is more prominently involved in NK cell-mediated killing of activated HSCs⁴⁴.

Our study is mechanistic in nature. Although RNA editing is known to be essential for mRNA regulation, only a small handful of ADAR1 editing targets have been experimentally defined. In our study, the 3’ UTR of Col3a1 was identified as a bona fide editing target and pro-fibrogenic effector of ADAR1-catalyzed RNA editing, leading to its mRNA accumulation and increased COL3A1 protein expression, consistent with the reports that ADAR1-catalyzed RNA editing in 3’ UTR stabilizes target transcripts by altering RNA structures and the accessibility of interacting partners^{47, 48}. ADAR1-mediated editing in the 3’UTR of Col3a1 is required for its regulation by multiple trans-acting factors, including DNM3OS, hnRNPA1, miR-29a, miR-29b, and miR-29c. However, the relative contributions of these factors to Col3a1 regulation remain to be defined.

Our results are clinically relevant. In our preclinical models, inhibition of ADAR1 using HSC-targeting RNAi or the small-molecule inhibitor 8-Aza effectively prevented fibrosis development and reversed established fibrotic predisposition, highlighting ADAR1 as a promising therapeutic target for liver fibrosis. However, we cannot exclude the possibility that the effect of 8-Aza on non-HSC hepatic cells may have also contributed to the phenotypic exhibition. IFNs have long been used to treat chronic viral hepatitis and associated fibrosis. Emerging evidence suggests that IFNs have direct anti-fibrotic effects^{44, 49, 50}. IFN-α has been reported to improve serum fibrotic markers and fibrosis degree regardless of the patients’ response to the antiviral therapy^{50, 51}. IFN-β reduced concanavalin A-induced liver fibrosis⁴⁹. IFN-γ deficient mice were more susceptible to CCl₄- or DDC diet-induced fibrosis⁴⁴. However, the exact mechanisms and cellular context of its anti-fibrotic activity remain unclear. Our results showed that IFN-β, the most induced IFN in ADAR1-deficient HSCs, attenuated HSC activation through the IFNAR-JAK/STAT autocrine mechanism. It is highly likely that the documented anti-fibrotic activity of IFNs is attributed to their direct inhibitory effect on HSC activation.

In summary, we showed that ADAR1-imposed RNA editome plays a fibrogenic role in HSC activation. ADAR1 inhibition attenuates HSC activation and liver fibrosis by triggering HSC-intrinsic innate immunity in an MDA5- and JAK/STAT-dependent manner and reducing collagen production. Translationally, targeting ADAR1 in HSCs holds promise in the clinical management of liver fibrosis.

What is already known on this topic

ADAR1 is the main editor responsible for A-to-I RNA editing. However, the role of ADAR1 and RNA editome in HSC activation and liver fibrosis has not been reported.

What this study adds

ADAR1 was established as a fibrosis sensitizing gene. Inhibition of ADAR1-imposed RNA editome triggers HSC-intrinsic innate immune response and attenuates collagen production, leading to the suppression of HSC activation and liver fibrosis.

How this study might affect research, practice or policy

ADAR1-imposed RNA editome exacerbates liver fibrosis by suppressing HSC-intrinsic innate immunity and promoting collagen production. ADAR1 and RNA editing are promising therapeutic targets for the clinical management of liver fibrosis.

Data and materials availability:

All data associated with this study are present in the paper and the supplementary materials. RNA-seq data reported in this paper are available through GEO accession number GSE283569.