Bile acid-induced IRF3 phosphorylation mediates cell death, inflammatory responses and fibrosis in cholestasis-induced liver and kidney injury via regulation of ZBP1

Abstract

Cell death and inflammation play critical roles in chronic tissue damage caused by cholestatic liver injury leading to fibrosis and cirrhosis. Liver cirrhosis is often associated with kidney damage that is a severe complication with poor prognosis. Interferon regulatory factor 3 (IRF3) is known to regulate apoptosis and inflammation, but its role in cholestasis remains obscure. In this study, we discovered increased IRF3 phosphorylation (p-IRF3) in the liver of patients with primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC). In the bile duct ligation (BDL) model of obstructive cholestasis in mice, we found that tissue damage was associated with increased p-IRF3 expression in the liver and kidney. IRF3 knockout (Irf3^−/−) mice showed significantly attenuated liver and kidney damage and fibrosis compared to WT mice after BDL. Cell death pathways, including apoptosis, necroptosis and pyroptosis, inflammasome activation and inflammatory responses were significantly attenuated in Irf3^−/− mice. Mechanistically, we show that bile acids induced IRF3 phosphorylation in vitro in hepatocytes. In vivo, activated IRF3 positively correlated with increased expression of its target gene, Z-DNA Binding Protein 1 (ZBP1), in the liver and kidney. Importantly, we also found increased ZBP1 in the liver of patients with PBC and PSC. We discovered that ZBP1 interacted with RIP1, RIP3 and NLRP3, thereby revealing its potential role in regulation of cell death and inflammation pathways. In conclusion, our data indicate that bile acid-induced IRF3 phosphorylation and the IRF3-ZBP1 axis play a central role in the pathogenesis of cholestatic liver and kidney injury.

Graphical Abstract

Introduction

Cholestatic liver disease is one of the major indications for liver transplantation in the US (1, 2). Cholestasis is characterized by impaired bile flow, accumulation of bile acids and toxic substances in the liver and systemic circulation triggering inflammation, progression to fibrosis and ultimately cirrhosis. Bile acids (BAs) are synthesized in the liver from cholesterol and have a critical role in lipid digestion and regulate metabolism (3, 4). In humans, cholestasis is one of the most debilitating forms of liver disease with a broad spectrum of etiologies such as primary biliary cholangitis (PBC), primary sclerosing cholangitis (PSC), obstructive gallstones and drug-induced liver injury. Recent therapies targeting bile acid homeostasis, fibrogenesis and immune-mediated pathways have been successfully explored, especially in PBCs (5). However, the underlying pathogenesis of cholestatic disease remain to be fully understood.

Common bile duct ligation (BDL) is a well-established murine model to study cholestatic liver disease that mimics biliary obstruction and BAs accumulation in the liver and serum as observed in human patients (6). BDL closely resembles the pathogenesis of cholestatic liver disease characterized by changes in hepatocellular injury, inflammation, immune cell infiltration and fibrogenesis (6–8). Moreover, BDL also induces bile cast nephropathy that is a consequence of cholestatic liver induced kidney injury in humans and mice (9). Disruption of BA homeostasis or accumulation of BA leads to cell death and organ damage (3). Using Caspase8-, MLKL- or NLRP3-deficient mice, previous studies have showed that apoptosis, necroptosis and pyroptosis, respectively, contribute to liver dysfunction and progression to fibrosis and cirrhosis under cholestatic conditions (8, 10, 11). Moreover, cell death also induces inflammation and excessive hepatic stellate cell (HSC) activation that promote fibrosis (12).

Interferon regulatory factor 3 (IRF3) is a ubiquitously expressed transcription factor that plays a central role in anti-viral responses. Upon activation, IRF3 is phosphorylated (p-IRF3) and p-IRF3 acts as a transcriptional regulator of type-I interferon (IFN) and antiviral genes (13). It has been shown that noncanonical activation of IRF3 can activate apoptotic pathways. For example, IRF3 can directly interact with mitochondrial Bax through its BH2 domain (14) or, depending on IRF3 ubiquitination, translocate to mitochondria to trigger apoptosis (15). Recent studies demonstrated that IRF3 plays roles in the progression of alcoholic liver disease, obesity-associated nonalcoholic fatty liver disease, nonalcoholic steatohepatitis and carbon tetrachloride induced-fibrosis (14, 16–18). However, its role in cholestasis remains unknown.

In this study, we discovered a critical role of IRF3 in regulating cell death, inflammation and fibrosis in cholestatic liver disease. IRF3 phosphorylation was increased in human and mice cholestatic livers. We identified the involvement of IRF3 in kidney injury, indicating a novel pathological mechanism that resembles human bile cast nephropathy (8). Mechanistically, we showed that activated IRF3 was positively associated with Z-DNA binding protein 1 (ZBP1) in the liver and kidney in vivo and in vitro. Our data show that ZBP1 interacted with RIPK1, RIPK3 and NLRP3, likely leading to PANoptosis (pyroptosis, apoptosis and necroptosis) and inflammation (19). In conclusion, we discovered a new role for the IRF3-ZBP1 axis in regulating the pathogenesis of cholestatic liver and kidney injury.

Materials and Methods:

Animals:

C57BL/6J mice from Jackson Laboratory were used as WT controls. Irf3^−/− mice were provided by J. Sprent from Scripps Research Institute, La Jolla, CA, USA. Mice were housed in a specific pathogen-free mouse facility at the Beth Israel Deaconess Medical Center (BIDMC)and all animal handling was performed in compliance with institutional guidelines. Additional procedures were approved by the BIDMC Institutional Animal Care and Use Committee. BDL was performed as previously described (20). Briefly, 10-14-week-old male C57BL/6J or Irf3^−/− mice were anesthetized and placed on operating pad. Mice were shaved and the skin was disinfected with 70% ethanol. Through an abdominal incision, the common bile duct was identified and ligated. The abdomen and peritoneum were closed. For sham mice, the same surgical procedure was performed without the bile duct ligation. Animals were monitored during recovery and treated with buprenorphine (0.1 mg/kg) to avoid pain-induced stress after surgical intervention. On day14 after surgery, blood, liver and kidney tissues were collected and processed for histology or stored at –80°C for RNA and protein extraction.

Additional information is described in the Supplemental Materials and Methods

Results

IRF3 phosphorylation is increased in the cholestatic liver

IRF3 is a multifunctional protein that exerts most of its biological activities in its phosphorylated form (13, 17, 18). To evaluate the role of IRF3 in the pathogenesis of cholestatic liver injury, we first measured the levels of phosphorylated IRF3 (p-IRF) in livers from the two most common human cholestasis diseases: primary biliary cholestasis (PBC) and primary sclerosing cholestasis (PSC). We observed increased p-IRF3 levels in the livers of PBC and PSC patients compared to healthy controls, whereas the expression of total IRF3 was not affected (Fig.1A). To elucidate whether IRF3 activation plays a mechanistic role in the pathophysiology of cholestasis, we utilized bile duct ligation (BDL), which is a murine model of obstructive cholestasis with progressive fibrosis and cirrhosis. Serum levels of bilirubin and the liver cell injury marker, Alanine transaminase (ALT) were dramatically increased at 14 and 28 days after BDL compared to sham controls (Fig.S1A). We also found increased mRNA levels of fibrosis markers Collagen1a1 (Col1a1), α-smooth muscle actin (α-SMA) and tissue inhibitor of metalloproteinases (Timp1), indicating the onset of liver fibrosis (Fig.S1B). As indicated by Sirius red staining, fibrotic areas in the liver were already present at 14 days and to a greater extent at 28 days after BDL compared to sham controls (Fig.1B). We observed significantly increased expression of p-IRF3 in the livers of BDL animals at 14 and 28 days compared to controls (Fig.1C–D) while the expression of total IRF3 was not affected at the protein (Fig.1D–E) or mRNA (Fig.S1C) levels. Therefore, we hypothesized that IRF3 phosphorylation might be involved in cholestasis-induced liver injury.

Figure 1: Increased IRF3 phosphorylation in the cholestatic liver and in bile acids treated hepatocytes and nonparenchymal cells.

(A) Liver lysates were probed for p-IRF and IRF3 from healthy controls and patients with PBC and PSC (n=4–5). The bottom panels indicate quantification. (B) H&E and Sirius red staining were performed on the liver sections of sham-or BDL-operated mice at the indicated time (n=4–6). Representative sections are shown (scale bar=100μm). (C-D) Liver lysates were probed for p-IRF and IRF3 from sham-or BDL-operated mice at day 14 (C) and day 28 (D) (n=4–8). The right panels indicate quantification. (E-F) Cell lysates were probed for p-IRF and IRF3 from primary human hepatocytes and human nonparenchymal cells (hNPC) (E), HepG2 and BMDM (F) treated with indicated bile acids for 1 hour (n=3–4). Right panels indicate quantification. (G) HepG2 and BMDM treated with GDCA for one hour as indicated concentration. *p<0.05, **p<0.01, ***p<0.001.

Bile acids induce IRF3 phosphorylation in hepatocytes and nonparenchymal cells

In addition to increased levels of circulating BAs, the TLR4 ligand lipopolysaccharide (LPS) contributes to the pathophysiology of cholestasis (21). To decipher whether accumulated BAs are involved in triggering IRF3 activation in cholestasis, we treated primary human hepatocytes and nonparenchymal cells with different BAs and measured p-IRF3. We discovered that p-IRF3 was significantly increased in isolated primary human hepatocytes and human liver non-parenchymal cells after stimulation with the primary bile acid chenodeoxycholic acid (CDCA) or the secondary bile acid deoxycholic acid (DCA) (Fig.1E). This effect was reproducible in HepG2 cells and mouse BMDMs, which showed a significant increase in p-IRF3 after CDCA, DCA, taurodeoxycholicacid (TDCA) and taurochenodeoxycholic acid (TCDCA) (Fig.1F), or glycodeoxycholic acid (GDCA) treatment (Fig.1G). These data indicate that BAs are potent inducers of IRF3 phosphorylation in hepatocytes and in nonparenchymal cells in vitro and likely provide a mechanism for the increased p-IRF3 observed in vivo after BDL.

IRF3 deficiency attenuates BDL-induced liver injury and fibrosis

Next, we tested whether IRF3 deficiency could impact pathological outcomes of experimental cholestasis. We performed BDL on Irf3 knockout (Irf3^−/−) and WT mice. There was no mortality during the 14-day postoperative monitoring period in any of the groups (Fig.S2). Serum bilirubin levels remained highly elevated in Irf3^−/− mice after BDL; however, ALT and aspartate transaminase (AST) levels were significantly attenuated in Irf3^−/− compared to WT mice (Fig.2A–B). Importantly, livers from Irf3^−/− mice exhibited significant decreases in fibrotic areas indicated by Sirius red staining compared to WT after BDL on day14 (Fig.2C). Immunohistochemistry revealed a significant reduction in hepatic stellate cell activation markers desmin and α-SMA in Irf3^−/− livers compared to WT controls (Fig.2D–E). Western blot analysis of livers showed reduced expression of α-SMA and vimentin in Irf3^−/− compared to WT after BDL (Fig. 2F). Transforming growth factor β (TGFβ) is known a one of the most potent fibrogenic cytokines (21). In Irf3^−/− mice, we observed significantly reduced TGFβ1 levels compared to WT mice after BDL (Fig.3A). Altogether, these data indicate that IRF3 deficiency protects against fibrosis in cholestatic liver injury.

Figure 2: Absence of IRF3 prevents BDL-induced liver injury and development of fibrosis.

(A-B) Bilirubin (A), ALT and AST (B) levels were measured in the serum of WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (C-E) Liver section were stained with Sirius red (C) and Desmin (D) and α-SMA (E). Representative sections are shown (scale bar=100μm). The right panel indicate quantification. (F) Liver lysates were probed with α-SMA and Vimentin (n=3–6). Right panels indicate quantification. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 3: Inflammation and inflammasome activation are ameliorated in the liver of Irf3^−/− mice compared to WT mice after BDL.

(A-B) Serum levels of TGFβ (A) and MCP1, TNF, IL-6, IL-18 (B) were measured from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (C-D) Liver sections were stained with CD11b (C) and CD68 (D). Representative sections are shown (scale bar=100μm). The bottom panel indicates quantification. (E) Liver lysates were probed for NLPR3, Casp1 and IL-1β from healthy controls and patients with PBC and PSC (n=4–5). The bottom panels indicate quantification. (F) Liver lysates were probed for NLPR3, Casp1 and cleaved- IL-1β from WT and Irf3^−/− mice at day 14 after BDL (n=3–6). Bottom panels indicate quantification. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Inflammation and NLRP3 inflammasome activation are ameliorated in the livers of Irf3^−/− mice

Inflammation plays a critical role in the pathogenesis of cholestatic liver injury and contribute to fibrosis and cirrhosis. We found that BDL induced a robust increase in the levels of circulating proinflammatory mediators, including MCP1, TNF, IL-6 and IL-18, which were all significantly lower in Irf3^−/− than WT mice (Fig.3B). CD11b and CD68 immunostainings revealed significant infiltration of macrophages in WT livers, whereas Irf3^−/− livers showed decreased strongly CD11b+ and CD68+ cell populations 14 days after BDL (Fig.3C–D). Liver mRNA expression of CD11b/CD68 and M2-macrophages markers CD163/Arginase1 was reduced in Irf3^−/− compared to WT mice after BDL, whereas the expression of CD86 and the M2 macrophage marker CD206 remained the same (Fig.S3A). Surprisingly, LY6G+ cells, which indicate neutrophil recruitment, were increased to the same extent both in WT and Irf3^−/− mice after BDL,as shown by immunofluorescence staining (Fig.S3B) and mRNA expression (Fig.S3C),

The robust increase in IL-18 after BDL and its attenuation in Irf3^−/− mice prompted us to assess NLRP3 inflammasome activation. Interestingly, we found increased expression of the NLRP3-inflammasome complex including NLRP3, Caspase1, cleaved-Caspase1 and IL-1β in the livers of patients with PBC and PSC compared to healthy controls (Fig.3E). Consistent with the human data, NLRP3-inflammasome expression and activation were increased in WT mouse livers after BDL compared to sham controls (Fig.3F). Notably, Irf3^−/− mice exhibited a drastic reduction in NLRP3, cleaved-Caspase1 and cleaved-IL-1β expression in the liver compared to WT (Fig.3F), suggesting attenuated inflammasome activation in the absence of IRF3.

Cell death pathways are ameliorated in the livers of Irf3^−/− mice after BDL

Different types of cell death have been implicated in the progression of acute and chronic liver injury (12). Emerging evidence suggests that pyroptosis, apoptosis and necroptosis can promote the progression of liver damage and fibrosis in cholestasis (8, 10, 11). Thus, we evaluated the effects of IRF3 deficiency on cell death pathways.

Gasdermin D (GSDMD) is the most potent executor of pyroptosis via its cleaved N-terminal fragment (N-GSDMD). N-GSDMD forms pores on the cell membrane to enable the release of IL-1ß cytokine and promote pyroptosis. Strikingly, we observed that BDL resulted in increased levels of cleaved-GSDMD (N-GSDM) and total-GSDMD in the liver, while N-GSDMD was significantly reduced in the livers of Irf3^−/− mice (Fig.4A). Correspondingly, the serum levels of ASC which can be released from GSDMD pores was substantially reduced in the Irf3^−/− mice compared to WT after BDL (Fig.4B). Thus, we found that IRF3 deficiency could attenuate BDL-induced pyroptosis.

Figure 4: Irf3^−/− mice show attenuated apoptosis and necroptosis after BDL.

(A) Liver lysates were probed for GSDMD and N terminal-GSDMD from WT and Irf3^−/− mice at day 14 after BDL (n=3–6). (B) Serum levels of ASC were measured from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (C) Liver tissue homogenates from healthy controls and patients with PBC and PSC were assessed for expression of Casp8 and Casp3 (n=4–5). (D) Liver tissue homogenates from WT and Irf3^−/− mice at day 14 after BDL were assessed for expression of Casp8 and Casp3 (n=3–6). The right panels indicate quantification. (E) Liver tissue sections were stained with TUNEL. Representative sections are shown (scale bar=100μm). The right panel indicates quantification. (F) RIP3, p-MLKL and MLKL expression were assessed in the liver tissue from healthy controls and patients with PBC and PSC (n=4–5). (G) RIP3 and MLKL expression were assessed in the liver tissue from WT and Irf3^−/− mice (n=3–4). The right panels indicate quantification. (H) Liver section were stained with p-MLKL. Representative sections are shown (scale bar=50μm). The right panels indicate quantification. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Next, we assessed apoptotic cell death by analyzing cleaved-caspase3 and cleaved-caspase8 in the livers of patient with PBC and PSC and mouse livers after 14 days of BDL. Cleaved-caspase3 and cleaved-caspase8 were significantly increased in PBC and PSC livers and WT mice after BDL (Fig.4C–D). Irf3^−/− livers showed significantly reduced expression of cleaved-caspase8 and cleaved-caspase3 compared to WT mice after 14 days of BDL (Fig.4D). Notably, BDL-induced apoptosis was indicated by increased TUNEL staining in WT livers that was significantly attenuated in Irf3^−/− mice (Fig.4E).

To evaluate necroptotic cell death pathways, we measured RIPK3 and its substrate, MLKL. We observed increased protein expression of RIPK3, MLKL and p-MLKL in human livers with PBC and PSC and WT mouse livers after 14 days of BDL (Fig.4F–G). Interestingly, BDL-induced RIPK3 protein expression in WT mice was significantly attenuated in Irf3^−/− livers (Fig.4G). Immunofluorescence staining revealed that Irf3^−/− livers displayed reduced p-MLKL expression compared to WT controls after 14 days of BDL (Fig.4H), while MLKL expression was comparable with WT controls (Fig.4G).

These data demonstrated that multiple cell death pathways are involved in cholestatic liver injury and highlight the association of death pathways with IRF3 activation. Overall, Irf3-deficient mice showed decreased cell death during cholestatic liver injury.

IRF3 deficiency alters bile acids composition and metabolism after BDL

Next, we examined whether IRF3 deletion affects BA compositions and metabolism after BDL. Serum BA composition profiling revealed a drastic increase in TCA levels after BDL compared to Sham in both genotypes; however, TCA levels were significantly lower in Irf3^−/− compared to WT 14 days post BDL (Fig.4SA). TDCA, TCDCA and GCA levels were similarly elevated in both genotypes after BDL (Fig.4SA).

We also evaluated BA synthesis- and transport-related gene expression. We observed that mRNA expression of cytochrome P450 7A1 (Cyp7a1) was significantly lower in Irf3^−/− livers compared to WT, whereas G-protein-coupled bile acid receptor (TGR5) mRNA expression was increased in Irf3^−/− livers compared to WT after BDL (Fig.S4B). To further elucidate the role of IRF3 in BA metabolism, WT and Irf3^−/− primary hepatocytes were isolated and treated with DCA or CDCA. CYP7A1 expression was attenuated in Irf3-deficient hepatocytes after BA treatment compared to that in Irf3-deficient hepatocytes (Fig.S4C). Interestingly, CYP7A1 baseline level was higher in Irf3-deficient hepatocytes than in WT. We also observed weak TGR5 expression in WT hepatocytes even a BA treatment. Strikingly, TGR5 expression was significantly increased in Irf3-deficient hepatocytes compared to WT after BA treatment (Fig.S4C).

These data indicate that reduced BA synthesis and increased BA transport in the absence of IRF3 after BDL.

Irf3^−/− mice show reduced kidney injury and fibrosis after BDL

Excessive BAs can trigger tubular epithelial injury after BDL, resulting in kidney damage and fibrosis (8, 9, 22). We observed increased p-IRF3 in the liver after BDL, we next examined whether IRF3 activation was involved in cholestasis-induced kidney pathology. We evaluated kidney tissue injury in the BDL model by H&E and periodic acid–Schiff (PAS) histological staining. There was a loss of cellular architecture in the kidney indicated by tubular atrophy both at day 14 and 28 after BDL (Fig.S5A). We also found notable interstitial fibrosis in the renal tissue with Masson’s trichrome staining after BDL at day14 and day28 (Fig.S5A). Histological changes were associated with higher levels of kidney dysfunction markers blood urea nitrogen (BUN) and kidney injury molecule 1 (KIM-1) in the serum of BDL group compared to sham controls (Fig.S5B). BDL-induced renal injury and fibrosis correlated with significant increases in kidney p-IRF3 levels on day14 after BDL compared to sham controls (Fig.5A). This effect was reproducible in human embryonic kidney (HEK293T) epithelial-like cells, which showed increased p-IRF3 after TCA, TDCA, GDCA or CDCA treatment (Fig.5B).

Figure 5: Irf3^−/− mice display reduced severity of kidney injury and fibrosis.

(A) p-IRF3 and IRF3 expression was measured in the kidney tissue after 14 days of sham and BDL (n=4–6). The bottom panel indicates quantification (n=3). (B) Cell lysates were probed for p-IRF and IRF3 from HEK293T cells treated with indicated bile acids for 1hr. The bottom panel indicates quantification (n=3). (C) Serum levels of Creatinine, BUN and KIM-1 were measured from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (D) Kidney injury score at day 14 after Sham or BDL. (E) α-SMA and Vimentin expression were assessed in the kidney tissue (n=3–6). The bottom panels indicate quantification. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Similar to the protective effects of IRF3 deficiency on the liver, Irf3^−/− mice subjected to BDL exhibited a significant decrease in kidney injury, as indicated by reduced serum levels of creatinine, BUN and KIM-1 (Fig.5C), as well as by histopathological evaluation of the kidney via H&E and PAS stainings (Fig.5D and Fig.S5C). Masson’s trichrome staining showed protection from kidney fibrosis (Fig.S5C) and WB indicated a corresponding reduction of α-SMA and vimentin expression in the kidneys of Irf3^−/− mice compared to WT after 14 days of BDL (Fig.5E).

Collectively, these results indicate a critical role of IRF3 in mediating kidney damage after BDL.

Inflammation, inflammasome activation and cell death are reduced in the kidney in Irf3^−/− mice after BDL

Inflammation and cell death are well-known triggers for sustained kidney injury and fibrosis (9). After BDL, we found significantly increased mRNA expression of proinflammatory mediators MCP1, IL-6 and IL-18 levels in the kidneys of WT mice, and this increase was attenuated in Irf3^−/− mice (Fig.6A). Inflammatory cytokines expression correlated with an increase in CD11b+/CD68+ macrophages in the kidney after BDL; however, this effect was significantly reduced in Irf3^−/− mice assessed by IF analysis (Fig.6B–C). IF results correlated with reduced mRNA expression of CD11b and CD68 in Irf3^−/− kidneys compared to WT (Fig.S6A). mRNA expression of CD68, Arginase1 and CD206 was significantly attenuated in Irf3^−/− kidney compared to WT after BDL on day14 (Fig.S6A). Similar to the findings in the liver, BDL-related increase in LY6G+ neutrophils was the same in both WT and Irf3^−/− mice (Fig.S6B–C). These data indicate reduced kidney inflammation in Irf3^−/− mice.

Figure 6: Lack of IRF3 attenuates inflammation, inflammasome activation, apoptosis and necroptosis in the kidney after BDL.

(A) MCP1, IL-6 and IL-18 gene expression were quantified in the kidney tissue from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (B) Kidney sections were stained with CD11b (B) and CD68 (C) (n=3–4). Representative sections are shown (scale bar=100μm). The bottom panel indicates quantification. (D) Casp1, cleaved-IL-1b and NLRP3 expression were measured in the kidney tissue (n=3–6). The bottom panels indicate quantification. (E) Representative images of TUNEL staining (scale bar=100μm). Bottom panel indicates quantification. (F-G) Kidney lysates were probed with Casp8 and Casp3 (F), and RIP3 and MLKL (G) (n=3–6). The bottom panel indicates quantification. (H) Representative images of p-MLKL staining (scale bar=50μm). The bottom panel indicates quantification. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

We also found significant upregulation of NLRP3 protein levels, and more importantly, NLRP3 inflammasome activation was evidenced by increased cleaved- caspase-1 and IL-1ß in the kidney after BDL (Fig.6D). However, NLRP3 inflammasome activation, cleaved-caspase-1 and cleaved Il-1ß were all significantly reduced in Irf3^−/− mice (Fig.6D).

Next, we assessed the role of IRF3 on different cell death pathways including pyroptosis, apoptosis and necroptosis in the kidney after BDL. Consistent with increased inflammasome activation in the kidney, we found that BDL resulted in upregulation of total GSDMD and N-GSDMD expression after BDL compared to sham controls, while Irf3^−/− kidney showed significantly reduced levels of total-GSDMD and N-GSDMD after BDL(Fig.6E), suggesting that IRF3 deficiency attenuated BDL-induced pyroptosis in the kidney. Compared to WT controls, Irf3^−/− kidney showed a significantly diminished apoptosis, a reduction in TUNEL+ cells and decreased expression of cleaved-caspase8 and cleaved-caspase3 after BDL (Fig.6F–G). We also observed increased levels of RIPK3, MLKL and p-MLKL in the kidneys of WT mice after BDL, whereas necroptosis was attenuated in Irf3^−/− kidney as indicated by reduced expression of RIPK3, MLKL and p-MLKL (Fig.6H–I).

In summary, IRF3 regulates inflammatory and cell death response in the kidney contributing kidney damage.

The increase in endoplasmic reticulum (ER) stress induced by BDL is independent of IRF3

As increased circulating endotoxin levels have been described in the mouse BDL model (23), we wondered whether LPS plays a role in BA-induced of IRF3 phosphorylation in vivo. We observed that endotoxin concentration was increased after BDL compared to sham controls, and this increase was independent of IRF3 (Fig.7A). Previously, our group demonstrated that ER stress promotes p-IRF3 via STING (18) and BAs are known inducers of ER stress (3). Thus, we speculated that STING-mediated ER stress may also continue to sustained IRF3 activation in chronic cholestasis. We found that STING expression in the liver and kidney was drastically increased in both WT and Irf3^−/− after BDL compared to sham controls (Fig.7B). Importantly, STING expression was also significantly increased in the livers of patients with PBC and PSC compared to healthy livers (Fig.7C), indicating the importance of STING in cholestatic disease. Next, we evaluated the ER stress by analyzing PERK expression by western blot and CHOP gene expression, we observed that increased PERK and CHOP expression was independent of IRF3 in the liver as well as kidney 14 days after BDL (Fig.7D–E). These data suggested the upstream role of STING mediated-ER stress in IRF3 activation in cholestatic disease.

Figure 7: Increased ER stress induced by BDL is independent of IRF3.

(A) Serum levels of endotoxin were measured from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (B) STING expression were measured in the liver and kidney tissue (n=3–6). The right panels indicate quantification. (C) STING expression were measured in the livers of healthy controls and human patients with PBC and PSC (n=4–5). The bottom panel indicates quantification. (D) PERK expression were measured in the liver and kidney tissue (n=3–6). The bottom panels indicate quantification. (E) CHOP gene expression was quantified in the kidney tissue from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

IRF3 induces ZBP1 expression to promote cholestatic liver injury

Next, we aimed to explore molecular mechanism by which IRF3 promotes cell death and inflammation in the liver and kidney after BDL. IRF3 is a transcription factor that mediates innate immune responses against viral infection via regulating type 1 interferons and nontranscriptional activation to regulate apoptosis (13, 15). However, we found that mRNA levels of IFNb, Ifit1, Ifit3 and ISG15 were comparable between WT and Irf3^−/− both in the liver and kidney after BDL (Fig.8A). Serum levels of IFNβ showed no difference between WT and Irf3^−/− mice after BDL (Fig.S7A) suggesting that the regulatory role of IRF3 in cholestasis is not mediated by type I IFN. Previous studies have identified that noncanonical IRF3 activation can depend on its ubiquitination or interactions with Bax to activate mitochondria-associated apoptosis (14, 15). Our results showed no differences in IRF3 ubiquitination levels in the livers of sham and BDL mice on day14 (Fig.8B). Consistent with this, expression of cytochrome C, Bax and oxidative stress-related genes expression such as p22Phox, p47Phox, p91Phox was not affected by the absence of IRF3 on day14 of BDL in the liver or kidney (Fig.S7B–C).

Figure 8: IRF3 upregulates ZBP1 expression to promote cholestatic liver injury.

(A) IFNb, Ifit1, Ifit3 and ISG15 gene expression were quantified in the liver and kidney from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8). (B) IRF3 was immunoprecipitated using IRF3 antibody from liver of Sham- or BDL-operated mice, followed by immunoblotting of Ubiquitin (Ub), IgG as control. (C) ZBP1 was identified from publicly available RNA-seq and IRF3 CUT&RUN data (17) analyzed by DAVID. (D) ZBP1 expression was assessed in the liver and kidney from WT and Irf3^−/− mice at day 14 after Sham or BDL (n=3–8), and in the livers of healthy controls and human patients with PBC and PSC (n=4–5). The bottom panels indicate quantification. (E) ZBP1 was immunoprecipitated followed by immunoblotting of ZBP1, RIP3, NLRP3 and RIP1, IgG as control. (F) Primary hepatocytes were isolated from WT and Irf3^−/− mice, ZBP1 and IRF3 expression was analyzed (n=4). (G) Primary hepatocytes were treated with DCA, CDCA and LPS for 6hrs, and ZBP1 and IRF3 expression was analyzed (n=3). The bottom panel indicates quantification. (H) ZBP1 and IRF3 expression was analyzed in BMDM from WT and Irf3^−/− mice without or with LPS treatment for 6hrs (n=3). (I) LDH was measured in the supernatant of primary hepatocytes after DCA treatment for 6hrs. (J) Cell lysates of primary hepatocytes were assessed for expression of ZBP1, IRF3, and Casp3.

These observations prompted us to explore potential new targets of IRF3 in cholestasis. A recent study identified multiple potential targets of p-IRF3 in mouse hepatocytes using integrated transcriptome and cistrome datasets (17). We searched for target genes of p-IRF3 involved in apoptosis and inflammation by DAVID functional annotation and identified ZBP1 (Fig.8C), which promotes cell death and inflammation responses (19, 24, 25). The role of ZBP1 in liver diseases is yet to be evaluated. Thus, we hypothesized that ZBP1 was involved in IRF3-mediated effects in BDL. We found a significant upregulation of ZBP1 in livers and kidneys after BDL on day14 in WT mice (Fig.8D). Moreover, Irf3^−/− mice showed reduced levels of ZBP1 in the liver and kidney after BDL (Fig.8D). Importantly, ZBP1 expression was also significantly increased in the livers of patients with PBC and PSC compared to healthy livers, indicating the importance of ZBP1 in cholestatic disease (Fig.8D).

Coimmunoprecipitation experiments revealed that ZBP1 interacted with RIPK1, RIPK3 and NLRP3 in the livers of WT mice after BDL, revealing its potential role in the regulation of cell death and inflammation responses, whereas reduced interaction was observed in Irf3^−/− mice (Fig.8E). These data indicated that IRF3 positively regulates ZBP1 expression and ZBP1 can interact with cell death mediators in cholestasis.

To further elucidate the effect of IRF3 regulation on ZBP1 in cholestasis, WT and Irf3^−/− primary hepatocytes were isolated and treated with DCA, CDCA or LPS. Consistent with the regulatory role of IRF3 on ZBP1, a reduction in the baseline level of ZBP1 was found in Irf3- deficient hepatocytes compared to WT (Fig.8F). Consistent with increased p-IRF3 after BA and LPS treatment, WT primary hepatocytes showed increased ZBP1 expression, while Irf3^−/− primary hepatocytes did not exhibit increased ZBP1 expression after BA or LPS treatment (Fig.8G). Similar to hepatocytes, we found reduced basal levels of ZBP1 in Irf3^−/− BMDM compared to WT (Fig.8H). Moreover, ZBP1 expression was significantly increased in WT BMDMs after LPS treatment, whereas Irf3^−/− BMDMs expressed significantly lower levels of ZBP1 than WT after LPS treatment (Fig.8H). These data suggested that DCA/CDCA or LPS can induce ZBP1 expression in an IRF3-dependent manner.

Next, we observed that administration of DCA induced LDH production and cl-caspase3 expression in WT primary hepatocytes. However, after BA treatment, LDH and cl-caspase3 levels were attenuated in IRF3-deficient primary hepatocytes compared to WT controls (Fig.8I–J). BMDMs from Irf3^−/− mice also showed reduced cl-capsase3 expression after DCA and CDCA treatment compared to WT BMDMs (Fig.S8A). Notably, we observed reduced IL-18 levels in the supernatant of Irf3^−/− BMDMs compared to WT after BA treatment (Fig.S8B). Irf3-deficient cells expressed significantly lower levels of IL-6 with/without BA treatment (Fig.S8C), and IL-6 levels in the supernatant were very low and remained the same in both genotypes (Fig.S8D). Strikingly, TGFβ levels were increased in the supernatant of WT BMDMs after DCA and CDCA treatment, whereas IRF3 deficiency attenuated TGFβ levels after BA treatment (Fig.S8E).

In summary, the absence of IRF3 is associated with reduced ZBP1expression, leading to reductions in cell death, inflammation and fibrosis in response to BAs in the liver and kidney during cholestasis.

Discussion

In this study, we show that IRF3 is activated in PBC and PSC patients and a murine model of cholestatic liver injury and that the lack of Irf3 protects against cholestasis-associated liver inflammation, cell death and fibrosis in mice after BDL. Our data also indicate that the role of IRF3 in promoting tissue damage is not limited to the liver because cholestasis-induced renal injury and fibrosis were also attenuated in Irf3^−/− mice. We discovered that BA directly induced IRF3 phosphorylation in hepatocytes, BMDMs and liver nonparenchymal cells. Mechanistically, p-IRF3-ZBP-1 axis promoted PANoptosis and inflammation to regulate the pathogenesis of cholestatic liver and kidney injury.

An increase in p-IRF3 has been reported in various inflammation-driven liver diseases, including ALD, NAFLD and drug induced-liver injury (14, 16, 18), which can promote inflammation and apoptosis through transcription-dependent and transcription-independent mechanisms. However, no studies have been reported the regulation of IRF3 in the pathogenesis of cholestasis-induced liver injury and its complication. Our data showed that IRF3 phosphorylation was increased in human livers with PBC and PSC, indicating that IRF3 activation may play a role in cholestatic liver disease. Consistent with this hypothesis, we found increased p-IRF3 in the liver in the BDL murine model of obstructive cholestasis. Furthermore, we discovered increased p-IRF3 in kidney injury after BDL, which resembles to human bile cast nephropathy, a severe complication of cholestasis. In vitro, we discovered that primary and secondary bile acids induced rapid IRF3 phosphorylation in hepatocytes, BMDMs, liver nonparenchymal cells and HEK293T cells, indicating the potential direct effect of BAs on IRF3 phosphorylation. The exact mechanism by which BA induces IRF3 phosphorylation has yet to be determined. In vitro, we found that BA induced p-IRF3 in the cells 1 hour after stimulation, which could be mediated by various mechanisms including bile acid receptors, FXR and TGR5 (3). Interestingly, we found that serum BAs composition after BDL was changed in IRF3-deficient mice, which could be due to reduced CYP7A1 and increased TGR5 expression, possibly leading to less BAs production and increased export and ultimately contributing to an improved phenotype in these mice. Notably, LPS and STING-mediated ER stress, which are known p-IRF3 triggers (14, 18), were increased, and IRF3 was dispensable for the increase in these factors in BDL model; Our experiments show the attenuation of hepatic and renal damage, inflammation and fibrosis in Irf3^−/− mice and highlight the role of IRF3 in cholestatic disease. These indicate that IRF3 acts as a gatekeeper to trigger downstream cascade response.

Inflammation is a notable component of cholestatic liver injury (6, 7). An increased in p-IRF3 has been shown to trigger inflammation by targeting type I IFN in liver disease (26). Interestingly, the expression of ISGs, including IFNβ, Isg15, Ifit1 and Ifit3, was highly increased in WT after BDL compared to sham controls; however, Irf3^−/− showed similar levels with WT after BDL. This data indicates that the potential role of IRF3 in BDL is independent of IFN activation. We speculate that IFN production in Irf3^−/− could be explained by compensation via IRF7 activation in the absence of IRF3. Our data also demonstrated that proinflammatory cytokines, including MCP1, IL-6, TNF and IL-18, were reduced in Irf3^−/− mice compared to WT after BDL. We found that IRF3-deficiency attenuated CD11b+/CD68+ macrophages in the liver and kidney after BDL. Notably, M2-macrophage marker expression was lower in IRF3-deficient livers (Arginase1/CD163) and kidneys (Arginase1/CD206) that correlated with reduced fibrosis after BDL.

There was massive NLRP3 inflammasome activation after BDL that was significantly attenuated in Irf3^−/−. We showed that in the BDL model of cholestasis, NLRP3 regulates not only inflammation responses but it also induces pyroptosis both in the liver and kidney. Our coimmunoprecipitation experiments showed that IRF3 interacted with NLRP3 and IRF3-deficience attenuated NLRP3 activation in vivo. We found evidence of NLRP3 activation in human PBC and PSC, as indicated by increase cleaved-caspase-1 and cleaved-IL-1ß, and NLRP3 activation was also present in BDL mice. Proinflammatory cytokines, IL-1ß and pyroptosis was shown to directly activate HSCs that contribute to liver injury and liver fibrosis (8, 27). A previous study suggested that BAs act as DAMPs to trigger NLRP3 inflammasome, which could be a mechanism involved in the BDL model (28). Enhanced inflammasome activation is associated with cholestatic liver injury, and Nlrp3^−/− BDL mice showed reduced inflammation and fibrosis (8). Taken together, our data highlight the role of NLRP3 activation in cholestatic liver and kidney injury, pyroptosis and inflammation and its regulation by IRF3 activation.

Different types of cell death pathways contribute to liver damage and fibrosis (12). Our prior studies revealed that Irf3-deficiency plays a protective role in ALD and in CCL4-induced liver injury (14, 18). In the present study, we dissected the major cell death pathways in BDL and found evidence of apoptosis (indicated by caspase3 and caspase8 activation), pyropotosis (indicated by NLRP3 and GSDMD cleavage) and necroptosis (evidenced by RIPK3 and MLKL activation). Simultaneous activation of these cell death pathways indicates that PANoptosis occurs in cholestatic liver and kidney injury that all three pathways are significantly attenuated in Irf3^−/− mice. More importantly, we also detected evidence of pyroptosis, apoptosis, necroptosis and indicating PANoptosis in the liver of human patients with PBC and PSC. These data indicated that IRF3 may potentiate PANoptosis, thus triggering liver and kidney injury and fibrosis progression.

HSCs are well-known to play a critical role in fibrosis. During fibrogenesis, HSCs are activated and transdifferentiated into proliferative and fibrogenic myofibroblasts, leading to the accumulation of extracelluar matrix (6, 7, 29). TGFβ is considered a key mediator of HSCs activation in response to liver and kidney injury (21, 29, 30). Our data revealed that IRF3 deficiency strongly inhibited TGFβ release after BDL. Stellate cells can be activated by engulfment of apoptotic bodies derived from hepatocytes (29) or by inflammasome particles released from pyroptotic hepatocytes (27). We speculate that attenuation of PANoptosis in absence of IRF3 activation could prevent these stellate cell activation mechanisms in cholestasis. Collectively, these findings suggest reduced liver and kidney fibrosis in the absence of IRF3 after BDL.

Mechanistically, our prior studies revealed that IRF3 associates with the pro-apoptotic adaptor Bax and induces mitochondria-associated apoptosis, especially in hepatocytes (14, 18). In addition, another study demonstrated that IRF3 regulates macrophage apoptosis in response to ethanol exposure through its ubiquitination in the liver (16). However, the molecular mechanisms between IRF3 activation and excessive BA-induced cell death are not yet understood. ZBP1 was described as an IFN-inducing gene that interacts with IRF3 and promotes its activation to initiate the transcription of IFN (31). However, subsequent studies demonstrated that IRF3 activation and IFN production are independent of ZBP1 in HepG2 cells and mice (32, 33). Our study has now identified ZBP1 as a novel target of IRF3 and demonstrates IRF3-mediated regulation of ZBP1. Recent studies have highlighted the vital role of ZBP1 in regulating cell death and inflammation under various pathological conditions, such as heatstroke, inflammatory diseases and infection diseases (24, 25), which also explains the signaling cross-talk between apoptosis, necroptosis and pyroptosis, which is defined as ZBP1-mediated PANoptosis. Our study demonstrated that ZBP1 can interact with RIPK1, RIPK3 and NLRP3 (19), and reduced interactions were observed in Irf3-deficient livers after BDL. Indeed, we observed that ZBP1 upregulation is in line with increased PANoptosis in the cholestatic liver and kidney, whereas the absence of IRF3 prevented this cell death cascade. In addition, increased ZBP1 activation in BDL correlated with the upregulation of MCP1, IL-6 and IL-18 expression and with CD11b+/CD68+ macrophages presence in the liver and kidney in WT but not in IRF3 KO mice, indicating the role of IRF3-ZBP1 axis in inflammation. These cytokines and recruited macrophages can potentiate the inflammatory response and sustain PANoptosis, thus triggering disease worsening and fibrosis progression. Our study demonstrates that IRF3 promotes kidney injury and fibrosis by regulating ZBP1 expression, which shares a similar cell death and inflammatory profiling that takes place in the cholestatic liver.

In summary, our results establish a role for BA-induced IRF3 phosphorylation and the IRF3-ZBP1 axis in regulating cell death and inflammation responses that promote cholestasis-induced liver and kidney injury and fibrogenesis. These observations raise the rationale for inhibition of IRF3 activation as a potential therapeutic strategy in cholestatic liver disease.

ABBREVIATIONS:

ALT

alanine transaminase

AST

aspartate transaminase

α-SMA

α-smooth muscle actin

BA

bile acid

BDL

bile duct ligation

BUN

blood urea nitrogen

CD

cluster of differentiation

CDCA

chenodeoxycholic acid

Col1a1

Collagen1a1

CYP7A1

cytochrome P450 (CYP) isoform 7A1

DCA

Deoxycholic acid

GDCA

Glycodeoxycholic acid

H&E

hematoxylin and eosin

HSC

hepatic stellate cell

IFN

interferon

IHC

immunohistochemistry

IL-

interleukin-

IRF3

Interferon Regulatory Factor 3

KIM-1

kidney injury molecular 1

LPS

lipopolysaccharide

MCP1

monocyte chemoattractant protein-1

MLKL

mixed lineage kinase domain-like protein

KO

knockout

PANoptosis

pyroptosis, apoptosis, and necroptosis

PAS

periodic acid Schiff

PBC

primary biliary cirrhosis

PSC

primary sclerosing cholangitis

RIP

receptor interacting protein

RIPA

RIG-I-like receptors (RLR)-induced IRF-3-mediated pathway of apoptosis

TCA

taurocholic acid

TDCA

taurodeoxycholic acid

Timp1

tissue inhibitor of metalloproteinases

TLRs

toll- like receptors

TNF

tumor necrosis factor

TGR5

G-protein-coupled bile acid receptor

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick end labeling

WT

wide type

ZBP1

Z-DNA Binding Protein 1