ADAR1 Prevents Liver Injury from Inflammation and Suppresses Interferon Production in Hepatocytes

Abstract

Adenosine deaminase acting on RNA 1 (ADAR1) is an essential protein for embryonic liver development. ADAR1 loss is embryonically lethal because of severe liver damage. Although ADAR1 is required in adult livers to prevent liver cell death, as demonstrated by liver-specific conditional knockout (Alb-ADAR1^KO) mice, the mechanism remains elusive. We systematically analyzed Alb-ADAR1^KO mice for liver damage. Differentiation genes and inflammatory pathways were examined in hepatic tissues from Alb-ADAR1^KO and littermate controls. Inducible ADAR1 KO mice were used to validate regulatory effects of ADAR1 on inflammatory cytokines. We found that Alb-ADAR1^KO mice showed dramatic growth retardation and high mortality because of severe structural and functional damage to the liver, which showed overwhelming inflammation, cell death, fibrosis, fatty change, and compensatory regeneration. Simultaneously, Alb-ADAR1^KO showed altered expression of key differentiation genes and significantly higher levels of hepatic inflammatory cytokines, especially type I interferons, which was also verified by inducible ADAR1 knockdown in primary hepatocyte cultures. We conclude that ADAR1 is an essential molecule for maintaining adult liver homeostasis and, in turn, morphological and functional integrity. It inhibits the production of type I interferons and other inflammatory cytokines. Our findings may provide novel insight in the pathogenesis of liver diseases caused by excessive inflammatory responses, including autoimmune hepatitis.

Inflammatory responses are involved in the pathogeneses of most liver diseases, including viral infections,1, 2, 3 chemical liver damage,4, 5 trauma,6, 7 and autoimmune diseases.8, 9 As an essential innate immune component, inflammation plays a critical role in the defense against microbe invasion; however, excessive inflammation can be destructive and cause severe tissue damage.10, 11, 12, 13 It has been well established that pathogen-associated molecular patterns and damage-associated molecular patterns play crucial roles in the activation of inflammatory pathways through pattern recognition receptors,14, 15, 16, 17 which ultimately lead to inflammatory cytokine and chemokine production through NF-κB and/or interferon (IFN) pathway activation.18, 19, 20 In sterile injuries, such as alcoholic liver disease²¹ and liver ischemia/reperfusion,22, 23, 24, 25, 26 IFN pathways play crucial roles in inflammatory response. Excess cytokine production in the liver could instigate inflammatory cells to migrate into liver tissues, where they activate to cause local or even systemic inflammation.

Adenosine deaminase acting on RNA 1 (ADAR1), a primary RNA editing enzyme, had been demonstrated to play critical roles in embryonic and adult livers. Deletion of the ADAR1 gene resulted in severe liver damage in knockout (KO) mouse models.27, 28 However, the mechanism by which ADAR1 functions to prevent liver damage remains currently unknown.²⁹ ADAR1 has been most studied for its RNA editing activities, through which it converts adenosine to inosine on the double-stranded regions of RNA transcripts.30, 31 If the edited sites fall within a coding region or splicing site of an mRNA precursor, it may change either the amino acid sequence and the function of encoded protein or the splicing patterns of the mRNA transcript by generating or eliminating splicing sites.32, 33, 34, 35 miRNAs were also found to be subject to RNA editing, which can alter their biogenesis rate or cause them to suppress different RNAs.36, 37, 38 However, a particular editing site has not been identified that could change the function of a protein that would explain the liver damage observed in the mice.²⁹ RNA editing-independent activities of ADAR1 were also reported recently.39, 40, 41 ADAR1 has been found to inhibit IFN signaling in embryonic hematopoietic stem cells,⁴² and in adult small intestinal stem and progenitor cells.⁴³ As the most potent cytokines in innate immune response, IFNs have been shown to trigger liver damage in sterile injuries.21, 22, 23, 24, 26, 44

The role of ADAR1 in liver biology, especially how it protects the liver from damage, remains an enigma. By analyzing a hepatocyte-specific ADAR1 KO mouse model, we demonstrate herein that ADAR1 plays essential roles in hepatic homeostasis, which, in turn, is critical for liver's structural and functional integrity. Loss of ADAR1 from hepatocytes activated IFN signaling pathways, leading to inflammation, injury, disruption of hepatic zonation, ductular injury, stellate cell activation, fibrosis, and thus extensive liver damage. These results reveal a new mechanism by which ADAR1 contributes to hepatic homeostasis by inhibiting IFN production by hepatocytes and thus modulating inflammation.

Materials and Methods

Mouse Models

A hepatocyte-specific ADAR1 gene KO mouse model was generated by crossing floxed ADAR1 and albumin Cre transgenic mice, as described previously.²⁸ ADAR1 inducible KO mice were generated by crossing a floxed ADAR1 mouse with a Cre-ER transgenic mouse carrying a transgene expressing a fused protein of Cre recombinase and the estrogen receptor.⁴⁵ The Cre-ER transgene is driven by the CAG promoter (CMV early enhancer, actin promoter, and splice acceptor of β-globin gene). In this model, the ADAR1 gene remains intact and expresses the ADAR1 protein at normal levels until tamoxifen is administrated, which induces ADAR1 gene deletion. All animal experiments were performed observing protocols approved by the Institutional Animal Care and Usage Committee at the University of Pittsburgh, School of Medicine (Pittsburgh, PA).

Genotypes were determined by PCR as described previously.28, 46

Sera from the conditional KO mice for ADAR1 and littermate controls were assessed for alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, and alkaline phosphatase levels with a DRI-CHEM 4000 Chemistry Analyzer (HESKA Co., Loveland, CO).

Histology and IHC

Liver samples were fixed in 10% formalin and embedded in paraffin. Tissue sections (4 μm thick) were deparaffinized in xylene and hydrated through graded alcohol rinses from 100% to 95% distilled water, and eventually washed in 1× phosphate-buffered saline. Slides were next incubated in eosin stain for 30 seconds, followed by two washes in 95% ethanol and two washes in 100% ethanol. Slides were then counterstained in Shandon Harris Hematoxylin (Thermo Fisher Scientific, Grand Island, NY) for 1 minute and dehydrated in alcohol from 75% to 100% and eventually to xylene washes before mounting coverslips with 1,3-diethyl-8-phenylxanthine.

For immunohistochemistry (IHC), deparaffinized sections were microwaved for 6 minutes twice in citrate buffer and allowed to cool for 30 minutes. Next, endogenous peroxidases were quenched with 3% hydrogen peroxide for 10 minutes. Slides were next blocked with Super Block (UltraTek, Logan, UT) for 10 minutes, followed by antigen retrieval using antigen-unmasking solution (Vector Labs, Burlingame, CA), as per the manufacturer's instructions. The slides were next incubated with primary antibodies diluted in phosphate-buffered saline (as indicated) for 1 hour at room temperature. Primary antibodies used were anti-glutamine synthetase (GS; Santa Cruz Biotechnology, Dallas, TX; SC-9067; 1:100), anti–cyclin-D1 (Neomarkers, Fremont, CA; RB-9041; 1:100), anti–α-fetoprotein (AFP; Santa Cruz Bioltechnology; SC-8108), anti-IL-6 (7737; Abcam, Cambridge, UK), and anti-tumor necrosis factor-α (TNFα; 1793; Abcam). Sections were then washed three times in phosphate-buffered saline for 5 minutes each, followed by incubation with horseradish peroxidase–conjugated secondary anti-goat (1:200) or anti-rabbit (1:200) antibodies (Millipore, Billerica, MA), for 30 minutes at room temperature. Next, wash steps were repeated, which was followed by detection of secondary antibody signal with diaminobenzidine (Vector Labs), which was quenched with distilled water before counterstaining with Shandon solution. The slides were regraded through the alcohol from 75% to 100%, leading to xylene washes before mounting the coverslips with 1,3-diethyl-8-phenylxanthine (Fluka Labs, St. Louis, MO). Negative controls were generated as above, but without the primary antibody incubation. Images were taken on an Axioskop 40 (Zeiss, Dublin, CA) inverted bright field microscope.

For terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) IHC, slides were stained using the ApopTag peroxidase kit (Intergen Co, Purchase, NY), as per the manufacturer's instructions.

Hepatocyte Preparation and Treatment

The in situ collagenase (type VI; Sigma-Aldrich) liver perfusion technique47, 48 was modified for mouse hepatocyte preparation, as described previously.49, 50, 51 In brief, 75 to 100 mL of 0.13 mg/mL collagenase H (catalog number 11087789001; Roche, Indianapolis, IN) was used to perfuse the mouse liver at a speed of approximately 8 mL/minute to disassociate the liver cells. The total cells were then subjected to differential centrifugation (50 × g) to remove the nonparenchymal cells in Williams’ medium E (catalog number 12551-032; Invitrogen, Carlsbad, CA), which was followed by 30% Percoll (catalog number 17-544-01; GE Healthcare, Aurora, OH) density centrifugation to remove the damaged or dead cells. The purity of yielded hepatocyte typically exceeded 98% monitored by flow cytometry assay, and viability was typically >95% by Trypan blue exclusion assay.49, 51 Hepatocytes (200,000 cells/mL) were plated on gelatin-coated (type A; Sigma-Aldrich) culture plates in Williams' medium E with 10% calf serum, 15 mol/L HEPES, 10⁻⁶ mol/L insulin, 2 mmol/L l-glutamine, 100 U/mL penicillin, and 100 U/mL streptomycin. After 12 hours’ incubation (37°C, 95% air–5% CO₂), the medium was changed, and tamoxifen (T5678; Sigma) was added to the culture (0-hour time point). The cells were collected at 24, 48, and 72 hours after tamoxifen induction to determine the optimal time point for deletion of gene encoding ADAR1. Because our results showed complete gene deletion by 72 hours, we used this time point to assess the expression level of genes regulated by ADAR1.

Reconstitution of ADAR1 Expression in ADAR1-Inducible KO Hepatocytes

Expression of the mouse wild-type ADAR1 in primary hepatocytes to reconstitute ADAR1 expression in the KO cells was achieved by using the adenovirus in which the full-length ADAR1 was expressed. This ADAR1 virus was prepared through a commercial service from Welgen, Inc. (Worcester, MA). Mouse ADAR1 cDNA (NM_001146296) was cloned into Pme1/XhoI sites of a pENTCMV vector and recombined with the Ad5 backbone for virus preparation. Control viruses were also purchased from Welgen, Inc. [scramble shRNA (catalog number V1040) and AdCMV empty (catalog number V1000)]. After 24 hours of tamoxifen addition, the medium was changed, and control adenovirus or ADAR1 cDNA adenovirus at a multiplicity of infection of 1000 was added and incubated for 48 hours, at which time cells were harvested and tested for cytokine expression.

Gene Expression Analysis

Total RNA was isolated from liver tissue or hepatocytes using a RNeasy Mini Kit (Qiagen, Valencia, CA), according to the manufacturer's instructions. Total RNA (1 μg) was reverse transcribed using an iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA), according to the manufacturer's instructions. The samples were then diluted, and the same amount of cDNA was added to each reaction on each plate. The iTaq universal SYBR Green supermix (Bio-Rad Laboratories) and a different primer were used for gene expression analysis. All samples were run in triplicate, and the experiment was repeated three times. Primers used for albumin, glucose-6-phosphatase, α1-antitrypsin, transthyretin, tyrosine aminotransferase, and AFP were described previously⁵²; IL-6 (QT00098875), TNF-α (QT00104006), inducible nitric oxide synthase (QT00941143), STAT1 (QT00162183), and β-actin (QT01136772) were products from Qiagen (Hilden, Germany). Primer sequences were as follows: IFN-α, 5′-CTACTGGCCAACCTGCTCTC-3′ (forward) and 5′-AGACAGCCTTGCCAGGTCATT-3′ (reverse); and IFN-β, 5′-TGACGGAGAAGATGCAGAAG-3′ (forward) and 5′-ACCCAGTGCTGGAGAAATTG-3′ (reverse). The relative quantity of the gene expression was calculated using the 2^−ΔΔCt method with the β-actin as the endogenous control.

Statistical Analysis

Data are presented as means ± SD. Experimental results were analyzed for their significance by the t-test. Significance was established at the 95% confidence level (P < 0.05).

Results

Mortality and Severe Growth Defect of ADAR1 Liver Conditional KO Mice

Global gene KO of ADAR1 is embryonically lethal, with fetal demise at 11.5 to 12.0 days after coitus, because of massive liver damage, thus precluding an in vivo functional study of this essential molecule in adult liver.27, 28 To overcome this barrier, we generated a conditional KO model through breeding floxed ADAR1 and albumin-Cre transgenic mice (ADAR1^Lox/Lox + Alb-Cre⁺, referred to as Alb-ADAR1^KO), which were viable at birth. However, marked abnormalities began to appear starting at approximately 2 to 3 weeks after birth. The most striking phenotypes evident in Alb-ADAR1^KO were growth retardation and high mortality at early ages (Figure 1, A–C). To verify the liver-specific deletion of ADAR1, we analyzed DNA samples from liver with specific primer combinations (Figure 1D) and confirmed that Alb-Cre transgene efficiently excised floxed ADAR1 gene in the liver (Figure 1E). The relative quantity of ADAR1 alleles was estimated by measuring the band densities on the electrophoresis gels, and the efficiency of ADAR1 gene deletion was calculated accordingly. This analysis showed that the floxed ADAR1 was efficiently deleted in the heterozygous mice (Figure 1F). In the homozygous mice, the deletion seemed less pronounced. Nonetheless, ADAR1 gene deletion in the liver in Alb-ADAR1^KO resulted in notable morbidity and mortality.

Figure 1.

Generation of liver-specific ADAR1 conditional knockout mice. A: Growth defect in ADAR1 liver conditional knockout mice (Alb-ADAR1^KO:ADAR1^Lox/Lox;albumin-Cre). The body size of the Alb-ADAR1^KO mice is markedly smaller from their littermates in both B6 and Sv129 background. B: Significant deficit in body weight in Alb-ADAR1^KO indicates growth retardation. C: The survival rate of Alb-ADAR1^KO mice shows that more than half mice die before 9 weeks after birth. D: Schematic showing deletion of floxed alleles after introduction of albumin-cre. Various primer sets (P, P2, and P3) are designed to identify wild-type, floxed, and floxed deleted alleles for genotyping. bADAR1 gene deletion in liver tissue was determined by semiquantitative PCR analysis. Primers for floxed and deleted ADAR1 alleles are indicated by arrows. The standard controls were made from known floxed and knockout DNA samples at the indicated ratios. E: Liver DNA samples from heterozygous (wild/lox;Alb-Cre⁺) and homozygous (lox/lox;Alb-Cre⁺) mice at the ages of 1, 3, and 7 weeks were analyzed. A representative gel electrophoresis shows Alb-Cre transgene (270 bp), the wild-type allele (300 bp), floxed allele (380 bp), and deleted allele (400 bp), as indicated by the arrows. The standard controls were from known floxed and knockout DNA samples at the indicated ratios. Although the gene deletion is marginal at the 1-week stage, it increases significantly at 3 weeks and reaches to approximately 70% at 7 weeks in heterozygous mice. However, the maximum deletion of floxed ADAR1 gene remains approximately 30% in homozygous mice. F: Quantification of ADAR1 gene deletion by Alb-Cre in vivo is represented as percentage deletion. Heterozygous or W/L (ADAR1 wild/lox);Alb-Cre⁺ group has 70% ADAR1 deletion from 1 to 7 weeks, whereas no deletion is evident in W/L;Alb-Cre⁻ livers. The L/L (ADAR1^Lox/Lox);Alb-Cre⁺ livers show approximately 30% ADAR1 gene deletion in the first week and nearly 50% by the seventh week. Data shown are means + SEM (B). n = 10 to 36 (B). ^∗P < 0.05. UTR, untranslated region.

Intact ADAR1 Gene in Mouse Hepatocyte Is Required for Survival at Postnatal Stages

More than 70% of Alb-ADAR1^KO mice died approximately 7 to 9 weeks after birth. However, their survival was better than ADAR1 global KO mice, which did not survive beyond midgestational stage, revealing yet another important role of ADAR1 during early postnatal hepatic development and homeostasis. Although it is known that Alb-Cre most efficiently deletes floxed genes during postnatal stages,⁵³ we wanted to directly assess this in our study. We therefore examined whether the death of Alb-ADAR1^KO mice was associated with the extent of ADAR1 gene deletion. We quantified the deleted ADAR1 gene relative to the undeleted floxed alleles in the liver tissue at the ages from 1, 3, and 7 weeks after birth. In 1-week-old mice, the ADAR1 gene was not efficiently deleted, especially in the homozygous (ADAR1^Lox/Lox;Cre⁺) animals, although the deletion efficiency increased dramatically in mice as they reached 2 to 3 weeks of age (Figure 1, E and F). Alb-ADAR1^KO mice therefore likely survived through embryonic development, and no obvious abnormality appeared at birth. Although the ADAR1 gene deletion in the homozygous mice was not as efficient as in the heterozygous mice, it increased when the mice became older, from 1 to 7 weeks, consistent with the time course of the fatalities. These data support that intact ADAR1 gene in mouse hepatocytes is required for survival at postnatal stages.

A small proportion of the Alb-ADAR1^KO mice survived to adulthood (Figure 1C), raising the question whether ADAR1 was absolutely required. However, additional analysis of Alb-ADAR1^KO mice revealed that the survival of Alb-ADAR1^KO mice beyond 9 weeks was associated with insufficient ADAR1 gene deletion.

In the heterozygous mice, ADAR1 alleles experienced approximately 70% deletion in liver tissues. The nonparenchymal cell populations (approximately 30% of total liver cells) suggested that floxed ADAR1 gene was near completely recombined by the Alb-Cre in hepatocytes. However, in Alb-ADAR1^KO mice that survived, ADAR1 gene deletion was only approximately 30% to 40% of the total liver DNA samples (Figure 1, E and F). Although escape of certain hepatocytes from Alb-Cre–mediated recombination, a phenomenon described previously, might contribute to the survival of some Alb-ADAR1^KO mice,⁵⁴ the remnant ADAR1 expression could be because of nonhepatocyte cell populations expressing ADAR1 in the KO livers. Indeed, a significant inflammatory cell infiltration was evident in the Alb-ADAR1^KO livers. Nevertheless, the phenotype of growth retardation that occurred in 100% of Alb-ADAR1^KO mice confirmed the importance of ADAR1 in hepatocytes.

We backcrossed the mice to either FVB or Sv129 background. The defects in body growth and liver damage were the same in both backgrounds, further confirming that ADAR1 plays an essential role in liver, independent of the mouse genetic background. However, the severity varied in different mice of both strains. Although most of the mice did not survive beyond 10 weeks after birth, approximately 10% of the Alb-ADAR1^KO mice survived to adulthood; low efficiency of ADAR1 gene deletion was observed in these mice (data not shown). The most severely affected mice consistently died at 3 to 4 weeks of age and exhibited small body size and presented ascites.

Notable Injury and Alterations in Liver Function in the Alb-ADAR1^KO Mice

To determine liver function in the Alb-ADAR1^KO mice, we measured serum liver enzyme levels for injury and metabolic function. Alanine aminotransferase and aspartate aminotransferase levels were dramatically increased in Alb-ADAR1^KO mice compared with the controls at all ages tested, indicating hepatocyte injury even as early as 1 to 3 weeks after birth (Figure 2, A and B). Consistently, liver synthetic functions were significantly impaired, as shown by total serum proteins and triglycerides at the corresponding times (Figure 2, C and D). Hypoglycemia was also observed in 1- to 3-week-old young mice (Figure 2E). Mice from a particular litter were analyzed for liver functional blood marks. Table 1 shows that alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, and alkaline phosphatase levels were significantly increased from 5 to 13 weeks, whereas the protein and lipid levels were dramatically decreased, showing the severe defects of protein synthesis and lipid processing in Alb-ADAR1^KO mouse liver.

Figure 2.

Liver damage in Alb-ADAR1^KO mice is indicated by impaired serum biochemistry. A: Serum alanine aminotransferase (ALT) levels are significantly higher in the Alb-ADAR1^KO mice at various ages. B: Serum aspartate aminotransferase (AST) is significantly higher in the Alb-ADAR1^KO mice at various ages when compared with age-matched controls. C: A notable decrease in total serum proteins in Alb-ADAR1^KO mice indicates a compromise in hepatic synthetic function. D and E: A significant decrease in total triglycerides (D) and serum glucose (E) is observed in Alb-ADAR1^KO mice. All data shown are the means ± SEM. n = 4 to 12. ^∗P < 0.05, ^∗∗P < 0.01, and ^∗∗∗P < 0.001.

Table 1.

Levels of Serum Protein, Lipid, and Liver Enzymes in Control and ADAR1 Knockout Mice

Variable	Age, weeks
5			7			13
W/W	W/Δ	Δ/Δ	W/W	W/Δ	Δ/Δ	W/W	W/Δ	Δ/Δ
Total protein, g/dL	4.7	4.5	3.1	4.8	5.2	2.8	5.4	5.2	5.3
Albumin, g/dL	3.4	3.2	2.1	3.5	3.2	1.9	2.9	3	3.2
HDL, mg/dL	85	91	24	81	102	12	106	94	151
Triglyceride, mg/dL	97	126	66	87	144	62	140	129	69
ALT/SGPT, U/L	66	33	834	56	71	691	99	57	489
AST/SGOT, U/L	118	99	1124	110	27	893	147	122	669
LDH, U/L	735	758	3403	734	612	3465	1732	1052	2043
ALP, U/L	312	264	996	210	339	1493	59	84	409

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate transaminase; HDL, high-density lipoprotein; LDH, lactate dehydrogenase; SGOT, serum glutamic oxaloacetic transaminase; SGPT, serum glutamic-pyruvic transaminase; W/W, wild-type ADAR1 genotype; W/Δ, heterozygous ADAR1 genotype; Δ/Δ, homozygous ADAR1 genotype.

Prominent Disruption of Hepatic Architecture in the Alb-ADAR1^KO Livers

Livers of the Alb-ADAR1^KO mice were significantly smaller and paler than those of their littermates (Figure 3A). Histological evaluation was performed next to determine any changes in hepatic microarchitecture after ADAR1 loss. Notable perturbations in histology were evident in the Alb-ADAR1^KO, including distortion of liver lobule structure and the presence of lobule-wide inflammation (Figure 3B). Enlarged hepatocytes with multiple nuclei, often containing two to three nucleoli, were the distinguishing features of the KO mouse livers (Figure 3C). Large quantities of inflammatory cells penetrated into the tissue, and were also evident in high numbers in the portal triad area (Figure 3C). To verify inflammation, we performed IHC for CD45. Although few CD45-positive cells were present in livers of the controls, a noteworthy increase was clearly evident in the Alb-ADAR1^KO (Figure 3C). Subsets of these cells were CD68 positive, indicating the presence of macrophages as well (Figure 3C).

Figure 3.

Gross and histological defects in Alb-ADAR1^KO mice indicate hepatic injury, repair, and ductular proliferation. A: Representative gross image of a liver from a littermate control and an ADAR1^KO mouse shows the knockout (KO) livers are pale and smaller. B: A pronounced inflammation and abnormal hepatocytes showing swelling, lipid accumulation, and mitotic figures in ADAR1^KO livers. C: Additional characterization of liver from a representative 7-week-old ADAR1^KO mouse reveals increased inflammation by hematoxylin and eosin (H&E) staining, verified by staining for CD45 and CD68, which leads to enhanced cell death as shown by increased number of terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)–positive cells. Higher compensatory cell proliferation was also evident in ADAR1^KO livers. Enhanced inflammation and hepatocyte injury coincides with increased fibrosis as seen by Sirius Red (S RED) staining and presence of increased numbers of α-smooth muscle actin (α-SMA)– positive myofibroblasts. Increased lipid accumulation is also evident in ADAR1^KO hepatocytes as shown by oil red O (ORO) staining. Last, disruption of hepatic architecture as indicated by aberrant and irregular localization of glutamine synthetase (GS), a typical pericentral hepatocyte marker, and atypical ductular reaction is shown by CK19 staining in the ADAR1^KO livers. Original magnifications: ×40 [B and C (H&E, TUNEL, Sirius Red, and ORO images)]; ×200 [C (CD45 and CD68, ADAR1^KO, α-SMA, GS, and CK19 images)]. PCNA, proliferating cell nuclear antigen.

To determine the impact of inflammation on hepatocyte biology in the Alb-ADAR1^KO, we next assessed cell death, proliferation, and fibrosis. Alb-ADAR1^KO showed greater numbers of TUNEL-positive cells along with compensatory enhanced hepatocyte proliferation, as indicated by positive staining of hepatocytes for proliferating cell nuclear antigen (Figure 3C). As a result of ongoing cell death, there was also an increase in fibrosis, as indicated by increased Sirius Red staining (Figure 3C). The fibrosis, however, was mostly stage 1 to 2 and predominantly pericellular without any bridging. The livers of Alb-ADAR1^KO were grossly pale, and by hematoxylin and eosin, we observed evidence of steatosis. Examining Alb-ADAR1^KO livers for presence of lipids by oil red O staining showed extensive positive staining for oil red O (Figure 3C).

To further verify alterations in hepatic architecture because of extensive injury in the Alb-ADAR1^KO, we assessed localization of GS, a pericentral hepatocyte marker, and of cytokeratin 19 (CK19), a maker of biliary epithelial cells. Although control livers showed uniform GS-positive hepatocytes around the central vein, the Alb-ADAR1^KO livers displayed irregular GS staining, with some hepatocytes lacking any GS positivity around the central vein (Figure 3C). Similarly, although CK19 staining was limited to cholangiocytes lining the bile ducts in portal triads in control livers, several hepatocytes, especially in the periportal region, showed CK19 positivity in addition to cholangiocytes (Figure 3C). Thus, Alb-ADAR1^KO livers showed disruption of normal hepatic architecture as a result of inflammation and injury.

Decreased Expression of Genes Associated with Hepatocyte Maturation in the Alb-ADAR1^KO Mice

Next, we examined the expression of genes that encode for synthetic factors in hepatocytes and hence indicate the status of hepatocyte differentiation and maturation. Real-time quantitative PCR was used to measure the mRNA levels of albumin, glucose-6-phosphatase, tyrosine aminotransferase, α1-antitrypsin, and transthyretin. Significantly lower expression of all these markers of hepatocyte maturation was noted in the Alb-ADAR1^KO livers compared with littermate controls as early as 3 weeks after birth (Figure 4A). Concomitantly, AFP, a marker of fetal hepatocytes, was actively expressed in the ADAR1^KO livers at all stages (Figure 4A). Because the cellular composition of Alb-ADAR1^KO livers is perturbed, we wanted to complement our observations of altered gene expression of hepatocyte-specific genes in whole livers by IHC as well. Specifically, we examined staining for AFP, which was negligible in control livers but was notably higher in the hepatocytes in the Alb-ADAR1^KO livers (Figure 4B). Thus, in the absence of ADAR1, hepatocytes were relatively undifferentiated.

Figure 4.

Decreased expression of genes associated with hepatocyte maturation in the liver tissue of Alb-ADAR1^ko mice. A: Decreased expression of albumin gene (ALB mRNA) at different times in liver tissues of Alb-ADAR1^KO mice, as assessed by RT-PCR. Decreased expression of glucose-6-phosphokinase (G6P), α1-antitrypsin (AAT), tyrosine aminotransferase (TAT), and transthyretin (TTR), is observed in Alb-ADAR1^KO livers compared with controls. Expression of α-fetoprotein (AFP), a marker of immature or fetal hepatocytes, is significantly induced in the Alb-ADAR1^KO mice. B: Representative immunohistochemistry for AFP confirms enhanced protein expression in hepatocytes in the Alb-ADAR1^KO mice compared with littermate control. Data shown are the means ± SEM (A). n = 4 to 12 (A). ^∗P < 0.05, ^∗∗P < 0.01. IHC, immunohistochemistry.

Deletion of ADAR1 Activates Inflammatory Signaling by Inducing IFN Production in the Hepatocytes

To delineate the mechanism of the inflammation, we examined cytokine levels to test whether the inflammatory pathways were activated in the Alb-ADAR1^KO liver tissues. By using real-time quantitative PCR, we measured the mRNA levels of several typical inflammatory cytokines in liver tissues of KO mice and their littermates. The results showed that cytokine expression levels, including TNF-α, IL-6, Stat1, and IFN-β, were significantly increased in the KO livers, although inducible nitric oxide synthase and IFN-α levels were decreased (Figure 5A). Furthermore, several targets of type I IFN, such as IFN-induced protein 35 and T-cell–specific GTPase, were concomitantly induced in ADAR1^KO livers at different stages compared with controls (Figure 5A). To verify that the dysregulated gene expression is at least, in part, attributable to the changes in expression in hepatocytes as a consequence of ADAR1 loss, especially because of the presence of inflammatory cells, we next performed IHC for IL-6 and TNF-α. A notable increase in both these proteins is apparent in the hepatocytes in ADAR1^KO livers compared with controls, which showed little to no expression (Figure 5B).

Figure 5.

Inflammatory cytokine expression in Alb-ADAR1^KO mouse liver. A: RT-PCR shows increased expression levels of inflammatory cytokines, including IL-6, tumor necrosis factor (TNF)-α, and Stat1 in Alb-ADAR1^KO mouse liver tissue at the ages of 3, 5, and 7 weeks compared with littermates. Inducible nitric oxide synthase (iNOS) expression, however, is decreased in ADAR1^KO mouse liver. Expression of interferon (IFN)-β increases, whereas IFN-α decreases in the ADAR1^KO mouse liver tissue. A significant increase in the expression of IFN downstream targets IFN-induced protein 35 (IIP35) and T-cell specific GTPase (TSGPase) is observed in the ADAR1^KO liver. B: Representative immunohistochemistry for IL-6 and TNF-α confirms enhanced protein expression in both hepatocytes (arrowheads) and the infiltrating inflammatory cells (arrows) in the Alb-ADAR1^KO mice compared with littermate controls. Data shown are means ± SEM (A). n = 3 (A). ^∗P < 0.05, ^∗∗P < 0.01, and ^∗∗∗P < 0.001.

To further validate if altered expression of various cytokines in the Alb-ADAR1^KO livers is because of their expression in the increased inflammatory cell infiltration or it is being contributed by hepatocytes lacking the ADAR1 gene, we used a tamoxifen-inducible ADAR1 KO mouse, as reported elsewhere.46, 55 Hepatocytes free of nonparenchymal cells were isolated from the inducible ADAR1^KO mice by two-step liver perfusion, and the cells were cultured under standard hepatocyte culture conditions.56, 57 Tamoxifen was added to the culture medium to induce ADAR1 gene deletion (Figure 6A). ADAR1 gene deletion was effectively achieved by tamoxifen induction in the hepatocytes, as verified by the PCR analysis of the ADAR1 gene alleles and diminishing levels of ADAR1 protein (Figure 6, B and C). Cell viability was measured by lactate dehydrogenase release, to rule out any deleterious effects of tamoxifen in the presence or absence of ADAR1, and was comparable between the two groups (Figure 6D). Similarly, hepatocyte viability and morphology did not show any differences between tamoxifen-treated ADAR1 KO and control hepatocytes for the entire duration of treatment (Figure 6E). However, expression of IFN-α and IFN-β was dramatically increased in the hepatocytes after ADAR1 deletion compared with tamoxifen-treated control hepatocytes (Figure 6F). Similarly, an increase in TNF-α and IL-6 was also observed in primary hepatocytes on ADAR1 deletion (Figure 6F). Thus, loss of ADAR1 from hepatocytes directly induces the expression of various proinflammatory cytokines.

Figure 6.

Deletion of ADAR1 in hepatocytes induces expression of type I interferon (IFN) and proinflammatory cytokines in vitro. A: Schematic of primary hepatocyte isolation from control or ADAR1^Lox/Lox;Cre-ER⁺ mice by liver perfusion to test for IFN production as the cells from control or inducible knockout mice were treated with tamoxifen (TM) to promote ADAR1 gene deletion in knockouts. B: A dose-dependent increase in deleted allele (400 bp) at the expense of floxed allele (380 bp) is seen after treatment of primary ADAR1^Lox/Lox;CRE-ER⁺ hepatocytes with TM for 72 hours (range, 10 to 100 nmol/L). C: Decreased ADAR1 protein expression is seen in a representative Western blot using lysates from hepatocytes isolated and cultured with TM from inducible (ADAR1^Lox/Lox without Cre) when compared with TM-treated control hepatocytes. Tamoxifen (100 nmol/L) efficiently induces ADAR1 deletion. D: Hepatocyte viability, as indicated by lactate dehydrogenase (LDH) release, does not show noticeable differences between the knockout and controls. E: Comparable morphological characteristics of ADAR1-null and control hepatocytes after culture in the presence of TM (100 nmol/L) for 72 hours. F: Expression of IFN-α, IFN-β, IL-6, and tumor necrosis factor (TNF)-α is dramatically increased at 72 hours after TM treatment of hepatocytes from inducible knockouts. All data shown are means ± SEM (C and F). n = 3 to 5 (D and F). ^∗P < 0.05, ^∗∗P < 0.01, and ^∗∗∗P < 0.001.

ADAR1 Re-Expression in Hepatocytes Normalizes Cytokine Aberrations

To validate the inhibitory effect of ADAR1 on inflammatory cytokine production in hepatocytes, we re-expressed ADAR1 exogenously in the inducible ADAR1 KO hepatocytes in culture through adenovirus. By using this approach (Figure 7A), we were able to successfully reconstitute ADAR1 expression in KO hepatocytes in the presence or absence of tamoxifen (Figure 7B). Next, we determined expression levels of type I IFN, IFN-α, and IFN-β, as well as IL-6 and TNF-α, in the ADAR1 re-expressing hepatocytes compared with control virus infected hepatocytes. The mRNA levels of IFN-α and IFN-β, as well as IL-6 and TNF-α, were significantly inhibited in the ADAR1 reconstituted hepatocytes compared with the control virus, and no virus treated KO hepatocytes (Figure 7C). These data provide direct evidence that lack of ADAR1 in hepatocytes activates type I IFN and other cytokine expression and hence may be the proximal event initiating inflammation and, in turn, progressive hepatic injury in the ADAR1^KO mice.

Figure 7.

ADAR1 re-expression in null hepatocytes abrogates expression of type I interferon (IFN) and proinflammatory cytokines in vitro. A: Schematic of primary hepatocyte isolation and culture from ADAR1^Lox/Lox;Cre-ER⁺ and ADAR1;Cre-ER⁻ mice, followed by addition of tamoxifen (TM) 72 hours after adherence to induce ADAR1 deletion [ADAR1 indicible KO (iKO) cells]. Twenty-four hours later, control or ADAR1 iKO cells were infected with control or ADAR1 carrying adenovirus for 48 hours to test for cytokine expression. B: Increase in ADAR1 mRNA expression is seen by real-time quantitative PCR in hepatocytes after infection with adenovirus containing recombinant ADAR1 cDNA. C: The mRNA levels of IFN-α and IFN-β, as well as IL-6 and tumor necrosis factor-α, are significantly inhibited in the ADAR1 reconstituted hepatocyte compared with the control virus, and no virus treated knockout hepatocytes. Data are means ± SEM (B and C). n = 4 to 12 (B and C). ^∗∗P < 0.01, ^∗∗∗P < 0.001.

Deletion of ADAR1 Leads to Altered Expression of Certain Genes Associated with Hepatocyte Differentiation in Primary Cultures

In addition to the altered expression of proinflammatory cytokines in hepatocytes lacking ADAR1 both in vitro and in vivo, we also identified altered expression of genes associated with hepatocyte differentiation using RNA from whole livers, which was complemented by IHC. However, to address this observation more conclusively, we assessed the RNA isolated from tamoxifen-treated inducible ADAR1^KO hepatocytes (Figure 6A) for various hepatocyte differentiation genes. Intriguingly, no change was observed in the expression of albumin, tyrosine aminotransferase, α1-antitrypsin, and transthyretin (Figure 8, A and C–E), respectively. However, we reproducibly observed a consistent decrease in the expression of glucose-6-phosphatase (Figure 8B) and a consistent increase in the expression of AFP (Figure 8F) after ADAR1 loss. This demonstrates an important role of ADAR1 in regulation of the expression of some key genes involved in hepatocyte maturation.

Figure 8.

Expression of hepatocyte differentiation-associated genes in primary hepatocyte culture after acute ADAR1 gene deletion. Hepatocytes from ADAR1^Lox/Lox;Cre-ER⁺ knockout mice were treated with tamoxifen (TM) for 72 hours. Total RNA was isolated from the cells at 0 and 72 hours after TM treatment and used for quantification of the gene expression by real-time RT-PCR. Expression of albumin (ALB; A), tyrosine aminotransferase (TAT; C), α1-antitrypsin (AAT; D), and transthyretin (TTR; E) is similar in TM-treated and non-treated hepatocytes; expression of glucose-6-phosphatase (G6P; B) decreases significantly. In contrast, α-fetoprotein (AFP; F) expression is significantly elevated after ADAR1 loss. Data are the means ± SEM (A–F). n = 4 (A–F). ^∗P < 0.05, ^∗∗∗P < 0.001.

Discussion

Although ADAR1 was demonstrated to play essential roles in embryonic liver development,27, 28 to the best of our knowledge, no systematic study has been reported on its function in adult liver. Through analysis of liver-specific conditional and an inducible conditional KO mouse, we have identified an indispensable role of ADAR1 in maintaining adult liver homeostasis. The loss of function of ADAR1 in hepatocytes led to a rather drastic phenotype associated with significant morbidity and mortality. We carefully verified the status of the ADAR1 gene in liver to attribute the phenotype to the specific gene deletion. Besides genotyping from tail samples, we confirmed the gene deletion in liver tissues. The striking changes of both serum biochemistry and gene expression profile of liver marker proteins unambiguously indicated extensive damage to the liver in the absence of ADAR1 in the hepatocytes.

ADAR1's functions in other tissues has also been reported, including in bone marrow,⁵⁸ skin,⁵⁹ small intestine,⁴³ and cancer cells.⁵⁵ Germline deletion of ADAR1 in mice results in embryonic lethality at embryonic 11.5 to 12 days after coitus.²⁸ Intriguingly, no obvious morphological defects are observed in organs other than the liver, including the placenta, heart, kidneys, brain, and other tissues, despite ADAR1 expression not being confined to the liver. Thus, the observation that the global KOs of ADAR1 die in utero because of hepatic insufficiency underlines its importance in early hepatic development. Interestingly, fetal livers are the site for hematopoiesis, and ADAR1 loss also causes defects in hematopoiesis,28, 60 suggesting a direct effect on both hepatocytes and the hematopoietic environment within the developing livers. In the current study, we were able to successfully bypass the early liver developmental requirement of ADAR1 through the use of Alb-Cre, which has been shown to induce progressive deletion of the floxed gene.⁵³ Intriguingly, in the homozygous mice, the deletion seemed less efficient because of remnant expression of ADAR1. The cause for this seemingly inefficient deletion may be multifactorial. First, Alb-Cre only deletes floxed genes from hepatocytes and hence any ADAR1 expression in resident nonparenchymal cells is intact. Considering the massive inflammatory cell infiltration evident in KO liver, ADAR1 expression is likely because of the presence of inflammation, which will not be affected by Alb-Cre. Because ADAR1 is an IFN-induced protein, it is likely that IFN induced because of ADAR1 loss from hepatocytes may be paradoxically stimulating ADAR1 expression in the inflammatory cells that have infiltrated into liver. Second, occasionally, Alb-Cre has been shown to be leaky, which implies that in a small subset of mice, because of an as yet unexplained reason, some hepatocytes may shut off Alb-Cre expression, which allows for maintained expression of the floxed allele in those cells. These hepatocytes may have survival and proliferative advantage over time and may repopulate liver.⁵⁴ Third, because hepatocyte dedifferentiation is evident in Alb-ADAR1^KO livers, albumin expression is itself decreased, perhaps as a response to ongoing hepatocyte injury because of inflammation. This may reduce cre-recombinase activity downstream of albumin promoter and yield incomplete ADAR1 deletion. Nonetheless, ADAR1 gene deletion in Alb-ADAR1^KO mice through Alb-Cre led to progressive hepatic damage and eventually lethality in predominant subsets of mice. Thus, intact ADAR1 in hepatocytes is indispensable for hepatic health and animal survival.

ADAR1 was primarily defined to be an RNA editing enzyme that converts adenosine residues to inosine on RNA transcripts. Earlier studies were focused on identifying the RNA target for ADAR1.29, 30, 61 However, after extensive efforts, no particular RNA substrate was found that could explain the massive liver damage. It was also found that ADAR1 interacts with other proteins, such as nuclear factors⁶² and Dicer,⁴⁰ exerting RNA editing independent activities. However, no studies have verified these claims in the liver or elsewhere. Recently, we found that ADAR1 inhibits the cytosolic RNA sensing pathway and suppresses type I IFN production in response to viral and endogenous RNAs.⁴⁶ Hepatocyte is one of the cell types with most abundant RNAs because of its active metabolism. It is thus conceivable that cellular RNA in hepatocyte stimulates the RNA sensing pathway in the absence of ADAR1 that lead to inflammatory cytokine productions. Herein, we provide the evidence that ADAR1 inhibits hepatic inflammation through proactive suppression of IFN production in hepatocytes. Excessive cytokine release from hepatocytes on ADAR1 deletion both in vivo and in vitro induces an inflammatory reaction, which, in turn, leads to cell damage, which is progressive, most likely because of feed-forward mechanisms that may use factors such as damage-associated molecular patterns to perpetuate injury, cell death, regeneration, and fibrosis.

To our knowledge, this is the first study that demonstrates that ADAR1 in hepatocytes prevents inflammation in the liver. As an IFN-stimulated gene, ADAR1 has long been suspected to play a role in infectious diseases.63, 64, 65 A connection of ADAR1 and sterile inflammation has not been established. In the absence of ADAR1, hepatocytes produced high levels of typical inflammatory factors, such as TNF-α, IL-6, inducible nitric oxide synthase, and IFNs. This observation was also supported by positive staining of hepatocytes for TNF-α and IL-6 in the Alb-ADAR1^KO livers. IFNs were directly induced by deletion of ADAR1 in hepatocytes in cultures. These results were further substantiated by rescue experiments in vitro where re-expression of ADAR1 in KO hepatocytes led to normalization of inflammatory cytokine expression.

IFNs are potent cytokines that stimulate the innate immune system to eradicate invading microbes.66, 67, 68 As an inflammatory factor, IFN also participates in inflammatory reaction and plays a role in sterile liver damage.22, 23, 24, 25, 26 Furthermore, in response to infections such as hepatitis C, hepatocytes are known to express and secrete IFNs, especially types I and III.⁶⁹ Furthermore, ADAR1 is known to suppress IFN signaling in hematopoietic and small intestinal cells.42, 43 Considering that the primary event leading to the overall phenotype is ADAR1 deletion from hepatocytes, it reasonable to believe that ADAR1 actively suppresses IFN and cytokine production from hepatocytes. ADAR1 gene deletion in the hepatocyte first results in a type I inflammatory cytokine production from the affected hepatocyte, which, in turn, stimulates the NPCs, including macrophages, to release additional cytokines that further provoke infiltration of inflammatory cells into the hepatic parenchyma. Tissue damage ensues as hepatic injury and fibrosis perpetuate, also leading to aberrant hepatic architecture, perturbed hepatic zonation, hepatocyte dedifferentiation, and some compensatory regeneration, contributing eventually to morbidity and mortality in Alb-ADAR1^KO mice.

Alb-ADAR1^KO livers also showed decreased expression of genes associated with hepatocyte differentiation. Such genes representing synthetic and metabolic functions of hepatocytes were down-regulated in the Alb-ADAR1^KO livers. Simultaneously, ADAR1^KO livers showed overexpression of fetal markers, such as AFP and CK19, in hepatocytes. Indeed, such altered expression of genes associated with hepatocyte maturation is often observed during hepatic injury, such as after exposure to 3,5-diethoxycarbonyl-1,4-dihydrocollidine diet, which induces hepatic expression and secretion of various proinflammatory cytokines.⁷⁰ Furthermore, because ADAR1 is known to interact with Dicer, it is relevant to also point out that disruption of miRNA processing machinery in hepatocytes, such as through conditional deletion of Dicer⁷¹ or by deletion of miR-122, the most abundant miRNA in hepatocytes, leads to similar damage to the liver because of enhanced inflammation, injury, and steatosis.⁷²

Intriguingly, there were some differences between the expression profiles of inflammatory and hepatocyte differentiation–associated genes in vitro (pure hepatocytes) versus in vivo. In vivo in the Alb-ADAR1^KO livers, we observed a significant increase in IFN-β, but not IFN-α, whereas in vitro both IFN-α and IFN-β were induced on ADAR1 deletion. Similarly, we identified decreased expression of several key genes associated with hepatocyte maturation in Alb-ADAR1^KO livers in vivo. However, when hepatocytes were examined after ADAR1 knockdown in primary cultures, there was a notable decrease in expression of only glucose-6-phosphokinase and a notable increase in expression of AFP, whereas others were not affected. These differences suggest that, although ADAR1 may be regulating the expression of some of these genes directly through a yet undefined mechanism, the effects on other genes that are observed in vivo are either cell autonomous or being influenced by some feedback mechanisms.

In summary, we demonstrate an essential role of ADAR1 in the liver, and explore the mechanisms by which ADAR1 functions in hepatocytes. We provide the first evidence that ADAR1 protects the liver from inflammatory damage by inhibiting IFN pathways. This study reveals a novel and essential mechanism for maintenance of the integrity of liver, especially because it is the organ constantly exposed to portal blood that carries toxins of dietary and microbial origin. ADAR1 may be contributing to regulation of innate immunity, especially in hepatocytes, which are already known to be an important component.⁷³ Thus, ADAR1 may have implications in liver diseases involving excessive inflammation. Furthermore, it was recently reported that ADAR1 is involved in the development of liver carcinoma.⁷⁴ It will be interesting to investigate whether there is a connection between its anti-inflammatory effects and its oncogenic functions.