Loss of β-catenin reveals a role for glutathione in regulating oxidative stress during cholestatic liver disease

Abstract

Background:

Cholestasis is an intractable liver disorder that results from impaired bile flow. We have previously shown that the Wnt/β-catenin signaling pathway regulates the progression of cholestatic liver disease through multiple mechanisms, including bile acid metabolism and hepatocyte proliferation. To further explore the impact of these functions during intrahepatic cholestasis, we exposed mice to a xenobiotic that causes selective biliary injury.

Methods:

α-naphthylisothiocyanate (ANIT) was administered to liver-specific knockout (KO) of β-catenin and wild-type mice in the diet. Mice were killed at 6 or 14 days to assess the severity of cholestatic liver disease, measure the expression of target genes, and perform biochemical analyses.

Results:

We found that the presence of β-catenin was protective against ANIT, as KO mice had a significantly lower survival rate than wild-type mice. Although serum markers of liver damage and total bile acid levels were similar between KO and wild-type mice, the KO had minor histological abnormalities, such as sinusoidal dilatation, concentric fibrosis around ducts, and decreased inflammation. Notably, both total glutathione levels and expression of glutathione-S-transferases, which catalyze the conjugation of ANIT to glutathione, were significantly decreased in KO after ANIT. Nuclear factor erythroid-derived 2-like 2, a master regulator of the antioxidant response, was activated in KO after ANIT as well as in a subset of patients with primary sclerosing cholangitis lacking activated β-catenin. Despite the activation of nuclear factor erythroid-derived 2-like 2, KO livers had increased lipid peroxidation and cell death, which likely contributed to mortality.

Conclusions:

Loss of β-catenin leads to increased cellular injury and cell death during cholestasis through failure to neutralize oxidative stress, which may contribute to the pathology of this disease.

INTRODUCTION

One of the many metabolic functions performed by the liver is the production of bile, which is required for digestion and absorption of fat. Disrupting or impairing the flow of bile from the liver into the intestine results in cholestasis or accumulation of bile components in the liver. One cause of cholestasis is damage to cholangiocytes, the cells that line bile ducts. Chronic injury to cholangiocytes causes inflammation, abnormal ductular reaction, and fibrosis. If the insult persists or is perpetuated, this can eventually result in cirrhosis, liver failure, or malignancy. Diseases of cholangiocyte origin are called cholangiopathies, and despite diverse etiologies, they all share common pathogenic mechanisms, such as activation of cholangiocyte proliferation, apoptosis, production of inflammatory and profibrotic mediators, and cholestasis.¹

α-naphthylisothiocyanate (ANIT) is a hepatotoxicant that is conjugated to glutathione (GSH) in hepatocytes and transported to the ducts through the canalicular bile acid transporter multidrug-resistance–associated protein 2 (Mrp2).² However, the GSH-ANIT complex is unstable and dissociates in bile, resulting in free ANIT that is then taken up by hepatocytes again and reconjugated, leading to continuous exposure through recirculation of the drug.³ Free ANIT is cytotoxic to cholangiocytes, disrupting their integrity and causing bile acids to accumulate in the liver.²,⁴ Repetitive rounds of conjugation and secretion lead to depletion of GSH from hepatocytes, leaving them vulnerable to hepatocellular toxicity caused by bile acid accumulation in liver.²,³ ANIT-induced GSH depletion may also predispose cholangiocytes and hepatocytes to reactive oxygen species (ROS) generation and oxidative stress, which is a component of many pathophysiological processes in the liver, including cholestasis.⁵ Although it is a chemically induced experimental model, ANIT nonetheless induces pathophysiological changes associated with human cholestatic liver disease, including biliary hyperplasia, inflammation, and biliary fibrosis.⁶

The Wnt/β-catenin pathway has a pleiotropic role in cholestatic liver injury. On the one hand, it is essential for hepatobiliary repair and maintenance of canalicular structures and cell polarity in a model of toxic bile.⁷ However, inhibiting Wnt/β-catenin signaling suppresses aberrant cholangiocyte activation and bile acid synthesis in a model of bile stasis, thus reducing injury.⁸ To determine the role of β-catenin in a model of selective biliary injury, we subjected wild-type (WT) and β-catenin knockout (KO) mice to ANIT diet. We hypothesized that repair would be the dominant function of β-catenin and that its loss would worsen injury. Instead, KO mice died without apparent pathological sequelae. Further investigation revealed that KO mice had decreased expression of glutathione-S-transferase (GST) genes, which regulates the conjugation of GSH to ANIT, and decreased GSH levels after ANIT. Loss of these detoxification mechanisms led to the activation of an alternative redox pathway, nuclear factor erythroid-derived 2-like 2 (Nrf2), and its downstream target NAD(P)H quinone dehydrogenase 1 (NQO1), in KO. A similar increase in compensatory Nrf2 activation was also seen in a subset of primary sclerosing cholangitis (PSC) explanted livers with inactive or absent β-catenin. However, despite the increased Nrf2 activation, KO mice still succumbed to ANIT due to hepatocyte death, which was a result of decreased NF-κB activation and increased oxidative stress. These findings demonstrate the functional consequences of GST loss in hepatotoxicity and further link oxidative stress to the β-catenin pathway in cholestasis.

METHODS

Animal model

β-catenin floxed transgenic mice with loxP sites flanking Exons 2-6 were bred to liver-specific albumin-cre (Alb-cre) mice (both in C57BL/6 background), generating a conditional KO as described.⁸ Alb-cre-negative mice were used as WT littermate controls. Cholestasis was induced by incorporating ANIT (N4525-10G; Sigma Aldrich, St. Louis, MO) into standard mouse chow at a concentration of 0.05% and administering it to male and female mice that were >8 weeks old. ANIT diet was manufactured by Dyets Inc. (Bethlehem, PA) and sterilized by irradiation. Mice were humanely killed at the indicated times, and blood serum and livers were collected. Livers were sectioned, fixed in 10% formalin, and processed for paraffin embedding or frozen in liquid nitrogen and stored at −80 °C. Samples were collected from the following number of animals in each group: WT on control diet, n=3; KO on control diet, n=3; WT on 6 days of ANIT diet, n=5; KO on 6 days of ANIT diet, n=4; WT on 2 weeks of ANIT diet, n=5; KO on 2 weeks of ANIT diet, n=4; WT on 4 weeks of ANIT diet, n=4; KO on 4 weeks of ANIT diet, n=1. All animal studies were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) at the University of Pittsburgh School of Medicine (Pittsburgh, PA; protocol number 20077675) and the NIH (Bethesda, MD). All mice were in a C57BL6 background (Jackson Laboratories) and maintained in ventilated cages under 12 hours light/dark cycles with access to enrichment, water, and mouse chow ad libitum.

Immunohistochemistry on patient tissues

Paraffin sections from explanted tissue from patients with PSC (n=22) were used for β-catenin immunohistochemistry (IHC) staining, which was performed by the Pittsburgh Liver Research Center’s Clinical Biospecimen Repository and Processing Core. Tissue scoring was performed by an experienced research pathologist (Satdarshan P. Monga). β-catenin staining was categorized as 0 (no membrane staining or membrane only staining) or 1 (cytoplasmic and/or nuclear, which indicates activation), according to a similar classification.⁹ NQO1-stained slides were scored as follows: weak (1), moderate (2), and strong (3).

Statistical analysis

Data are presented as means with SD and individual data points. Data were analyzed, and graphs were generated using Prism GraphPad version 9.3.1 (GraphPad Software, San Diego, CA). p-values were determined using one-way ANOVA followed by Tukey multiple comparison test or two-tailed Student t test. Unless indicated, all treatments were compared to WT at baseline. p<0.05 was considered statistically significant.

For patient samples, contingency tables were created to check the association between β-catenin activation and NQO1 expression. The statistical figures and analysis were performed by R programming by means of publicly available packages.

Additional methods are available in the online supplement (http://links.lww.com/HC9/A961).

RESULTS

ANIT diet causes premature mortality of β-catenin KO mice

WT and β-catenin KO mice were fed a 0.05% ANIT diet with the intention of sacrificing these mice at 4 weeks to examine differences in hepatobiliary injury, necrosis, and ductular proliferation. Surprisingly, KO mice began to die after only 6 days of diet. Kaplan-Meier survival curve shows that 50% of KO mice were dead by day 14, and only 1 mouse survived to the 28-day (4 wk) time point (Figure 1A). Numbers of KO mice were therefore increased in order to obtain a sufficient sample size for analysis. We chose 6 days and 14 days for end points since mice first began to succumb to diet at 6 days and because half of the KO mice survived to day 14. At the time of sacrifice, liver weight to body weight ratios were significantly decreased in KO livers at both time points, which is in line with previous studies using these mice¹⁰ (Figure 1B). Histologically, WT mice had a significantly higher number of bile infarcts at 6 days compared to KO, but these were resolved by day 14 (Figure 1C, E). Interestingly, at the same time point (6 d), KO livers showed sinusoidal dilatation without evidence of occlusion, as determined by quantification of sinusoidal diameter (Figure 1D, E). Serum transaminases, alkaline phosphatase, and total bilirubin were only slightly increased in response to ANIT diet at either time point, with no significant differences between WT and KO in any of the parameters (Figure 1F). Collectively, the results indicate that despite inducing only mild hepatocellular injury, ANIT is lethal in mice that lack β-catenin in hepatocytes and cholangiocytes.

FIGURE 1.

KO mice have decreased survival after ANIT diet but lack overt indicators of morbidity. (A) Kaplan-Meier survival curve shows significantly decreased survival in KO mice after exposure to ANIT diet. (B) There is a significant decrease in LW/BW ratio in KO mice after 6 days and 2 weeks of ANIT compared to WT. (C) WT mice have more bile infarcts (expressed as percent area of the image) after 6 days of ANIT compared to KO; however, by 14 days, these infarcts had resolved. KO had few to no infarcts at all time points. (D) Sinusoidal diameter was significantly increased in KO mice after 6 days of ANIT compared to WT mice. (E) H&E stains of liver sections show sinusoidal dilatation in KO livers 6 days after ANIT (magnification ×200). (F) Serum markers of hepatic and biliary injury were essentially unchanged between WT and KO at both time points after ANIT, although some parameters, such as ALT, AST, and bilirubin, were increased in KO over time. Data in B, C, and E represent mean±SD. *p<0.05, **p<0.01 by two-way ANOVA (multiple comparisons). Scale bars:100 μm. Data in A was analyzed by log-rank Mantel-Cox test and found to be significant with a p-value of 0.0005. Data in B was quantified from n≥8 200x fields from n=3 WT and KO at baseline and from 25 to 30 ×200 fields from at least n=4 WT and KO mice at both ANIT time points. Data in D were quantified from 12 to 15 sinusoids per field from 2 randomly selected liver sections per mouse (n=3 WT and KO mice at baseline and n=4 WT and KO mice at both ANIT time points). Inset images were taken from the central vein region for consistency. Abbreviations: ANIT, α-naphthylisothiocyanate; KO, knockout; LW/BW, liver weight to body weight; WT, wild type.

Fibrosis, inflammation, and ductular response are largely unaffected in KO livers

To further characterize liver injury in WT and KO mice on ANIT diet, we analyzed fibrosis by means of Sirius red staining and quantification. Fibrosis was increased in all time points and both genotypes by ANIT. At 6 days, however, fibrosis was significantly exacerbated in KO livers compared to WT, with areas of onionskin-like fibrotic lesions that did not appear in WT until day 14 (Figure 2A). The livers were next assessed for inflammatory cells and reactive cholangiocytes, both of which exacerbate biliary fibrosis. CD45 IHC indicates a robust inflammatory response in both WT and KO, which is located primarily in the periportal region, although in KO, inflammatory cells also invaded the parenchyma at the 6-day time point. Notably, however, KO had less overall inflammation at 14 days after ANIT compared to WT (Figure 2B). Additionally, there was no difference in ductular mass between WT and KO at either time point, as assessed by A6 IHC (Figure 2C), although KO ducts tended to be dilated and contained enlarged atypical ductular cells after 6 days of ANIT. The overall similarities in phenotype suggest that KO livers maintain adequate structure and function despite early mortality.

FIGURE 2.

Increased mortality in KO is not a result of increased fibrosis, inflammation, or ductular response. (A) Sirius red staining shows the early development of sclerosing cholangitis 6 days after ANIT; quantification also shows more fibrosis in KO than in WT at this time point. However, by 2 weeks of ANIT, both WT and KO have equivalent fibrosis. (B) CD45 IHC demonstrates that while inflammation increases after ANIT, it is equivalent between WT and KO after 6 days of ANIT. CD45-positive cells are significantly decreased in KO compared to controls after 2 weeks of ANIT. (C) A6 IHC shows that ductular reaction is equivalent between WT and KO at both time points after ANIT. Notably, however, KO ducts are dilated at both time points and contain elongated ductular cells. Graphs in A–C represent mean±SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by two-way ANOVA (multiple comparisons). All magnifications are ×200. Scale bars:100 μm. Data were quantified from n≥8 ×200 fields from n=3 WT and KO at baseline and from 25 to 30 ×200 fields from at least n=4 WT and KO mice at both ANIT time points. Abbreviations: ANIT, α-naphthylisothiocyanate; IHC, immunohistochemistry; KO, knockout; WT, wild type.

Total bile acid levels are similar in WT and KO despite downregulation of transporter genes

ANIT-induced damage to bile ducts results in bile acid leakage and accumulation, which in turn increases hepatocyte injury and inflammation.¹¹ To determine if this leakage is exacerbated in KO, we measured bile acid content in liver. There was no change in the total levels of bile acids at any of the time points analyzed, although they trended higher in KO at baseline (Figure 3A). Because ANIT also impairs bile acid transport and formation,¹² we next analyzed the expression of synthesis, detoxification, and transport genes, focusing specifically on the earliest changes induced by ANIT, as represented by the 6-day time point. Quantitative PCR shows that Nr0b2, which encodes small heterodimer partner—a nuclear receptor and downstream target of farnesoid X receptor, the master regulator of bile acid homeostasis—was decreased in KO at baseline and also decreased in WT after 6 days of diet (Figure 3B). Both Cytochrome P450 (Cyp7a1), the rate-limiting enzyme in bile acid synthesis, and Cyp27, which regulates the alternative bile acid synthesis pathway, were suppressed in the absence of β-catenin at baseline, supporting studies showing that these genes are either direct or indirect targets of Wnt signaling (Figure 3B).⁸,¹³ However, both Cyp7a1 and Cyp27 were equivalent in WT and KO after 6 days of ANIT. Likewise, Cyp2b10, a target of constitutive androstane receptor involved in bile acid detoxification, is unchanged. Cyp3a11, another P450 that detoxifies hydrophobic bile acids, is decreased in KO after ANIT, although the change is insignificant when compared with WT at the same time point (Figure 3C). Western blot shows that CYP3A11 protein is also lower after ANIT (Figure 3D). On the whole, however, bile acid synthesis and composition do not appear to be significantly different between WT and KO after ANIT.

FIGURE 3.

KO has decreased bile acid export but equivalent total bile acid levels after ANIT. (A) Total bile acid levels are equivalent in WT and KO at both time points after ANIT. (B) Nr0b2 (Shp), Cyp7a1, and Cyp27 levels are not significantly different between WT and KO after 6 days of ANIT, as assessed by quantitative PCR. (C) The expression of detoxification enzyme Cyp2b10 is increased after 6 days of ANIT in both WT and KO but expressed equivalently in both; however, Cyp3a11, another detoxification enzyme, is significantly decreased in KO after ANIT. (D) WB shows that CYP3A11 protein expression is also decreased in KO mice after ANIT. (E) There is no change in the expression of uptake transporters Slc10a1 (Ntcp) and Slco1b2 (Oatp4) after ANIT in either WT or KO. (F) Apical transporter Abcc11 (Bsep) is unchanged in WT and KO after ANIT; however, Abcc2 (Mrp2) is significantly decreased in KO compared to WT both before and after ANIT treatment. (G) WB shows that MRP2 protein expression in KO is approximately half that of WT after 6 days of ANIT. (H) Basolateral transporter Abcc3 (Mrp3) is decreased in KO both before and after 6 days of ANIT, while Abcc4 (Mrp4) is equivalently expressed in both groups before and after treatment. Data represent mean±SD. *p<0.05, **p<0.01, ***p<0.001 by two-way ANOVA (multiple comparisons). For A–C, E, F, and H, each point on the graphs represents a biological replicate that is the average of duplicate technical replicates. Data in D and G represent individual data points from WB quantification. Abbreviations: ANIT, α-naphthylisothiocyanate; KO, knockout; Mrp, multidrug-resistance–associated protein; NQO1, NAD(P)H quinone dehydrogenase 1; WB, western blot; WT, wild type.

Basolateral uptake transporters Slc10a1, which encodes sodium-taurocholate cotransporting polypeptide, and Slco1b2, which encodes organic anion transporting polypeptide-4 were unaffected by ANIT diet in either WT or KO (Figure 3E). Expression of the apical transporter Abcc11, which encodes bile salt export pump, was insignificantly decreased in KO both before and after ANIT (Figure 3F). Conversely, expression of Abcc2, or Mrp2, a canalicular efflux transporter that moves the ANIT-GSH conjugate into bile, was notably suppressed in KO before and after ANIT. Western blot confirms that MRP2 protein was decreased after ANIT in KO versus WT mice, paralleling Mrp2 mRNA decreases (Figure 3G). Abcc3 (Mrp3), a basolateral efflux transporter that is induced under cholestatic conditions, is decreased in KO both before and after ANIT, while Abcc4 (Mrp4) is unchanged (Figure 3H). Overall, although WT and KO have overall equivalent levels of bile acids in the liver, ANIT induces the retention of bile acids in KO hepatocytes to a greater extent than in hepatocytes from WT mice.

Glutathione-S-transferase expression is decreased in KO livers before and after ANIT

GST protect against oxidative stress through catalyzing GSH conjugation to xenobiotic compounds, leading to their subsequent elimination. In rodents, activating mutations of β-catenin is associated with increased expression of GSTs.¹⁴ Further, KO of β-catenin suppresses the expression of the isoenzymes GSTm1, GSTm2, GSTm3, GSTm6, GSTα3, and GSTω1 in mouse liver, which suggests that β-catenin regulates the expression of GSTs.¹⁵,¹⁶ Therefore, we hypothesized that decreased expression of these enzymes in KO could cause increased injury and hepatocyte death in the presence of a compound like ANIT that requires conjugation to GSH to counteract its toxicity. To determine the expression of GSTs in our model, we measured their mRNA expression in WT and KO livers before and after 6 days of ANIT, which was chosen in order to identify the most proximal events leading to dysregulation of metabolic and cell survival processes. Out of 6 GSTs analyzed, 4 (GSTm1, GSTm2, GSTm6, GSTα3) were suppressed in KO at baseline, and 4 either remained suppressed after ANIT (GSTm1, GSTm2, GSTm6) or were reduced only after ANIT (GSTm3) (Figure 4). Thus, the reduction in GSTs is due to the loss of β-catenin from hepatocytes.

FIGURE 4.

Expression of enzymes that facilitate GSH conjugation to xenobiotics are decreased in KO both before and after ANIT. Quantitative PCR for a panel of GSTs shows that 4 of 6 (GSTm1, GSTm2, GSTm6, GSTα3) are repressed in KO at baseline. GSTm1, GSTm2, GSTm3, and GSTm6 are also suppressed in KO after 6 days of ANIT, compared to WT. Data represent mean±SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by two-way ANOVA (multiple comparisons). Each point on the graphs represents a biological replicate that is the average of duplicate technical replicates. Abbreviations: ANIT, α-naphthylisothiocyanate; GST, glutathione-S-transferase; KO, knockout; WT, wild type.

Loss of β-catenin induces activation of the Nrf2 pathway in mice and patients

Next, we analyzed the biochemical effects of GST loss by measuring total GSH in liver before and after 6 days of ANIT, the time point at which mice begin to succumb to liver injury. Total GSH (GSH and its dimer, GSH disulfide) does not increase in KO livers after ANIT, while in WT total GSH is significantly higher after ANIT compared to baseline (Figure 5A). Expression of glutamine synthetase, a β-catenin target gene that synthesizes glutamine, which can, in turn, make GSH,¹⁷ is also decreased after ANIT (Figure 5B).

FIGURE 5.

Nrf2 is activated in ANIT-injured KO mice. (A) Total glutathione levels increase in WT after ANIT but remain the same in KO, leading to a significant difference between WT and KO. (B) β-catenin protein expression is decreased, and GS expression is absent in KO mice before and after ANIT. The trace amounts of β-catenin likely represent the population of nonparenchymal cells in the liver that are still β-catenin-positive. (C) IHC shows that KO livers have increased cytoplasmic and nuclear localization of Nrf2 in the periportal region after ANIT compared to WT. Magnification ×200. (D) Nrf2 target NQO1 is increased in KO compared to WT after 2 weeks of ANIT. For D, the same samples were run on 2 different WB and were standardized to either GAPDH (left) or total protein (right). Data in A represent mean±SD from duplicate technical replicates, averaged for each biological replicate. *p<0.05, **p<0.01 by two-way ANOVA (multiple comparisons). Scale bars:100 μm. Data in B and D represent individual data points from WB quantification. Abbreviations: ANIT, α-naphthylisothiocyanate; GS, glutamine synthetase; GSH, glutathione; IHC, immunohistochemistry; KO, knockout; Nrf2, nuclear factor erythroid-derived 2-like 2; NQO1, NAD(P)H quinone dehydrogenase 1; WB, western blot; WT, wild type.

Nrf2 is a transcription factor involved in the regulation of GSH synthesis as well as other antioxidant genes such as GSTs.¹⁸ We hypothesized that lack of GSTs and decreased GSH in the β-catenin KO may cause compensatory activation of Nrf2. Indeed, we found an increase in Nrf2 protein expression and nuclear translocation in KO after 6 days of ANIT, particularly in the periportal region (Figure 5C), which is simultaneous with GSH depletion. NQO1, a phase II detoxification enzyme important for protection against xenobiotics and another well-characterized Nrf2 target gene,¹⁹ was used as a surrogate marker for Nrf2 activation. NQO1 protein was also increased in KO, although not until 14 days of ANIT (Figure 5D).

We identified a subset of end-stage PSC explants characterized by decreased β-catenin, both at the membrane surface and in the pericentral region.⁹ To determine if loss of β-catenin was correlated with increased Nrf2 activation in patients as well, we assessed the activation of β-catenin and Nrf2, as measured by abundance of its target gene NQO1, in 22 livers explanted for PSC. Of those 22 cases, 12 had activated β-catenin, defined as cytoplasmic and/or nuclear localization, while the other 10 cases were characterized as transcriptionally inactive; these samples either had very low levels of β-catenin protein or β-catenin that was sequestered at the membrane (Figure 6A). β-catenin-active explants had an even distribution of NQO1 expression ranging from weak (1) to moderate (2) to strong (3). However, 6 out of 10 samples with inactive β-catenin showed strong positivity for hepatocyte NQO1, while 2 others in the same category showed moderate positivity (Figure 6B). Thus, loss of β-catenin activation correlated with increased NQO1 protein expression in end-stage PSC explants.

FIGURE 6.

Loss of β-catenin correlates with increased nuclear factor erythroid-derived 2-like 2 activation in explanted livers from patients with end-stage PSC. (A) Representative images of explanted livers from patients with PSC classified as either β-catenin inactive (absent or membranous only) or active (cytoplasmic and/or nuclear). Magnification ×200. (B) A contingency table classifying the data based on the status of β-catenin (inactive or active) shows that the majority of β-catenin inactive samples have moderate or strong levels of NQO1 staining. p=0.63628 by Fisher exact test. Scale bars:100 μm. Abbreviation: NQO1, NAD(P)H quinone dehydrogenase 1.

ANIT increases oxidative stress and cell death and impairs proliferation in KO livers

NF-κB signaling, which regulates cell survival and inflammation, can be negatively regulated by Nrf2 activation.²⁰ Thus, we next assessed the activation status of NF-κB after 2 weeks of ANIT, the time point at which Nrf2 is transcriptionally active as determined by NQO1 expression (Figure 5D). The p65 subunit of NF-κB was strongly detected in both WT and KO cholangiocytes at this time point (Figure 7A). However, WT showed profound nuclear translocation in both hepatocytes and cholangiocytes, while in KO NF-κB is only found in the cytosol of cholangiocytes. Thus, Nrf2 activation in KO coincides with decreased NF-κB transcriptional activation, which may negatively impact both hepatocyte and cholangiocyte survival.

FIGURE 7.

Nuclear factor erythroid-derived 2-like 2 activation in KO livers coincides with decreased NF-κB activation and increased oxidative stress. (A) Immunofluorescence for p65, a subunit of NF-κB, shows nuclear translocation in cholangiocytes and hepatocytes of WT but not KO after 2 weeks of ANIT. Arrowheads point to ducts with nuclear p65. In KO livers, p65 is only detected in the cytoplasm of cholangiocytes. (B) IHC shows more 4HNE, a product of lipid peroxidation, in KO livers than in WT after 6 days of ANIT. Staining is localized mainly around the vessels (central vein and periportal). For A, magnification ×400; for B, magnification ×200. Scale bars:100 μm. Abbreviations: ANIT, α-naphthylisothiocyanate; 4HNE, 4-hydroxynonenal; IHC, immunohistochemistry; KO, knockout; WT, wild type.

GSH regulates cellular redox status, and its depletion, along with loss of GST expression, can lead to oxidative damage.²¹ 4-hydroxynonenal, which is formed during lipid peroxidation, is increased in KO compared to WT as early as 6 days after ANIT (Figure 7B), indicating that increased oxidative stress in the absence of β-catenin precedes Nrf2 activation. Similarly, levels of malondialdehyde, another marker of lipid peroxidation, were higher in KO livers after ANIT, although the difference was not significant (Figure 8A). Because an insufficient antioxidant response can lead to apoptosis or necrosis, we next examined various mechanisms of cell death. Cleaved caspase-3, a marker of apoptosis, was increased in KO at both 6 days and 14 days after ANIT, as assessed by IHC and quantification of positive hepatocytes (Supplemental Figure S1, http://links.lww.com/HC9/A961, Figure 8B). receptor-interacting kinase 3, which mediates programmed necrosis and can be activated by caspase-mediated cleavage and increased ROS,²² is also increased in KO compared to WT at the 2-week time point (Figure 8C). Finally, terminal deoxynucleotidyl transferase dUTP nick end labeling staining revealed significantly more dead and dying hepatocytes in KO livers, particularly after 14 days of ANIT (Figure 8D).

FIGURE 8.

Loss of β-catenin in liver increases cell death and inhibits proliferation after ANIT. (A) Lipid peroxidation is insignificantly increased in KO compared to WT after 6 days of ANIT treatment. (B) KO have significantly more cleaved caspase 3–positive hepatocytes than WT at both 6 days and 14 days after ANIT. (C) WB shows that RIP3 is equivalent in WT and KO after 6 days of ANIT; however, protein expression decreases in WT but not in KO after 14 days of ANIT. (D) TUNEL staining shows levels of cell death in both WT and KO after 6 days of ANIT. By 2 weeks there is significantly more cell death in KO livers than in WT. (E) PCNA IHC shows that proliferation is significantly blunted in KO livers at both time points after ANIT diet, while WT livers show a robust regenerative response at 6 days that has returned to near-normal levels by 2 weeks of ANIT. Data in A–E represent mean±SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by two-way ANOVA (multiple comparisons). All magnifications are ×200. Scale bars:100 μm. For A, assay was performed in duplicate technical replicates and averaged for each biological replicate. For C, data represents individual data points from WB quantification. For D, data were quantified from n≥5 ×200 fields from n=3 WT and KO at baseline, and from n≥5 ×200 fields from n=4 WT and KO mice at both ANIT time points. For B and E, data were quantified from n≥3 ×200 fields from n=3 WT and KO at baseline, from n≥4 ×200 fields from n=3 WT and KO mice at 6 days ANIT, and from n≥4 ×200 fields from n=4 WT and n=2 KO mice at 14 days ANIT. Abbreviations: ANIT, α-naphthylisothiocyanate; IHC, immunohistochemistry; KO, knockout; MDA, malondialdehyde; NQO1, NAD(P)H quinone dehydrogenase 1; PCNA, proliferating cell nuclear antigen; RIP3, receptor-interacting kinase 3; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling; WB, western blot; WT, wild type.

The ability of hepatocytes to proliferate after being subjected to chemical insult is crucial to repair and maintain liver mass, and β-catenin is an important regulator of this process.²³ Proliferating cell nuclear antigen IHC showed a robust proliferative response in WT mice after ANIT, which was significantly blunted in KO livers at all time points (Figure 8E). Thus, there is an imbalance between proliferation and oxidative stress-mediated cell death when β-catenin is deleted, which is the likely cause of premature mortality in these mice.

DISCUSSION

Due to its multifactorial role in maintaining liver function, modulation of β-catenin in the treatment of complex diseases such as cholestasis is highly contextual. For example, inhibiting β-catenin in order to target a specific function, such as bile acid homeostasis, can result in unintended consequences for other functions, such as hepatocyte proliferation. A better understanding of the various underlying pathological origins of cholestasis is necessary in order to develop more precise therapies to treat these diseases. Since no single animal model can encompass all the possible etiologies of cholestatic liver disease, it is important to study multiple models as complementary approaches to evaluate and validate the contribution of a specific signaling pathway such as β-catenin. Previously, we found that loss of β-catenin decreases bile acid synthesis after bile duct ligation, resulting in less severe injury, while inhibiting β-catenin in the Mdr2 KO mouse exacerbates injury due to loss of hepatocyte polarity and impaired regeneration. In this study, we identified an important function for β-catenin in regulating the oxidative stress response during ANIT-induced cholestasis.

Although the precise mechanism of ANIT-induced lethality in KO is still unknown, hepatic insufficiency is a likely cause, as suggested by elevated bilirubin levels. About 80% of bilirubin is made from senescent red blood cells and subsequent breakdown of hemoglobin in the bone marrow, while the remaining 20% is produced in other tissues, such as the liver.²⁴ Bilirubin is conjugated in hepatocytes to increase solubility and facilitate its excretion into bile. Bilirubin levels are thus used as a measure of hepatic synthetic and excretory functions. It has been demonstrated that bilirubin excretion is impaired following ANIT treatment.²⁵ Additionally, bilirubin is a metabolite of heme, and we have shown that suppression of β-catenin downregulates enzymes involved in the synthesis of heme.²⁶ Thus, the increased serum bilirubin in KO relative to WT could have multifactorial causes, including dysregulated levels of heme/hemoglobin, changes in red blood cell counts, or alterations in hemolysis.

Albumin is a plasma protein that has a role in bilirubin and bile acid transport, immunomodulation, and extracellular antioxidant function. Notably, a study has demonstrated that serum albumin modulates cellular responses to oxidative stress through increasing cellular GSH levels and suppressing NF-κB activation.²⁷ Alb-cre-β-catenin floxed mice²⁸ and rodents fed an ANIT diet²⁹ have decreased serum albumin. Therefore, it is likely that KO mice fed ANIT also have decreased serum albumin, which may contribute to decreased total GSH, liver dysfunction, and lethality.

The accumulation of potentially toxic levels of bile acids in the liver is thought to be a major contributor to the pathogenesis of cholestasis, although they do not appear to be a primary driver of ANIT-induced injury.³⁰ Highly concentrated hydrophobic bile acids can stimulate ROS production in cultured hepatocytes;³¹ however, evidence from in vivo studies suggests that the production of ROS by infiltrating neutrophils is the main contributor to oxidative stress during cholestasis.³² Lipid peroxidation was increased in β-catenin KO after ANIT; however, we did not detect an increase in the population of CD45-positive cells, which includes neutrophils. The increased oxidative stress in the KO was likely due to the direct cytotoxicity of ANIT rather than increased accumulation of inflammatory cells.

Evidence pointing to the injury of hepatocytes by ANIT comes from early studies showing that Mrp2-mediated biliary secretion of GSH-ANIT is a prerequisite for the development of cholestasis in rats. Conversely, rats that lack Mrp2 are protected from ANIT-induced hepatotoxicity.² Likewise, the downregulation of Mrp2 in female mice after high-fat diet feeding was protective against ANIT because it prevented its excretion into the bile, reducing its toxic effect on cholangiocytes.¹¹ In β-catenin KO, we noted a significant downregulation of Mrp2 both before and after ANIT, which may be a protective response against biliary toxicity. Presumably, if Mrp2 expression is decreased, then GSH-ANIT complexes remain in the hepatocyte, as they are labile at pH 7.4 and only dissociate in bile.³³ Thus, prevention of injury may be dependent on adequate levels of GSH. However, in the case of β-catenin KO, the amount of GSH may be insufficient to prevent the hepatocyte toxicity and oxidative stress that would occur in the presence of free ANIT.

GSH detoxifies ANIT and plays an important role in reducing oxidative stress, and its depletion can exacerbate injury.³⁴ In particular, several groups have demonstrated a progressive decrease in GSH occurs as cholestasis progresses.³⁵ Thus, the decreased levels of GSH in KO relative to WT after ANIT is a significant finding and likely contributes to increased oxidative stress and injury. Although GSH conjugation and transport occur in hepatocytes, the effect of GSH loss on the biliary epithelium should not be overlooked. GSH depletion is associated with increased cholangiocyte apoptosis in vitro³⁶ and increased cholangiocyte toxicity in a mouse model of biliary injury.³⁷ Decreased cell survival in cholangiocytes was attributed to the inability to respond to increased oxidative stress. Although deletion of β-catenin from hepatocytes is thought to be the primary driver of the KO phenotype after ANIT since the injury originates in cholangiocytes, loss of β-catenin from this cellular compartment may also play an as-yet-unknown role in exacerbating toxicity. A comprehensive exploration of the role of β-catenin in cholangiocyte biology will be necessary in order to determine whether its loss has any impact on bile duct integrity and hepatic function during biliary injury.

GSTs catalyze the conjugation of GSH to xenobiotic substrates and thus facilitate the secretion of the ANIT-GSH complex into bile.³⁸ Induction of several GSTs was attenuated in Nrf2 KO mice, which renders these mice more sensitive to xenobiotics.³⁹ Low expression of GSTs is also associated with elevated oxidative stress.⁴⁰ Studies have shown that loss of β-catenin from hepatocytes leads to increased oxidative stress during injury.⁴¹ Because GSTs are at least partially regulated by β-catenin signaling in the liver,¹⁶ it follows that GST expression is lost in the β-catenin KO, leading to increased oxidative stress, which is a primary mechanism of injury in ANIT-induced cholestasis. Interestingly, increased Nrf2 activity in KO did not result in increased GSTs or GSH, as they remained suppressed despite evidence of Nrf2 activation (Figures 4, 5A). It is likely that Nrf2 activation is a compensatory mechanism induced by injury that is nonetheless insufficient to rescue the loss of β-catenin.

Nrf2 is activated in mice after ANIT treatment, although hepatobiliary toxicity was similar between WT and Nrf2 KO mice after ANIT, likely due to alterations in bile acid synthesis and transport.⁴² However, administration of an Nrf2 activator conferred protection against ANIT-induced toxicity,⁴², while loss of Nrf2 resulted in more pronounced injury after acetaminophen toxicity.⁴³ We noted an upregulation of Nrf2 in the absence of β-catenin, which was intriguing, as Nfe2l2, which encodes Nrf2, is a direct β-catenin target gene,¹³ and Nrf2 is also activated or upregulated in mutated β-catenin tumors.⁴⁴ However, Nrf2 may be activated by multiple other mechanisms, including exposure to toxicants and oxidative stress, both of which are higher in β-catenin KO than in WT. Additionally, toxic bile acids such as lithocholic acid have been shown to activate Nrf2 and promote adaptive antioxidative responses to bile acid toxicity in cholestasis.⁴⁵ Wnts can also directly activate Nrf2 in a β-catenin-independent manner.⁴⁶ Despite the lack of canonical signaling, Wnts are still present in β-catenin KO; indeed, lack of β-catenin may increase the availability of Wnts to activate noncanonical pathways such as Nrf2 signaling. Additional studies will be necessary to determine the complex interplay between β-catenin and Nrf2 during biliary injury.

Nrf2 can negatively regulate NF-κB signaling and cell survival. Because our NQO1 results indicated that Nrf2 was maximally activated at 2 weeks of ANIT treatment, we evaluated the p65 subunit of NF-κB at this time point. Unexpectedly, we found that p65 was absent from KO hepatocytes. One explanation for this is that the threshold activation of p65 in hepatocytes may be below detectable levels for the antibody. Indeed, a study showed that p65 staining is predominantly localized to cholangiocytes at baseline, where it appears in the cytoplasm.⁴⁷ Hepatocyte p65 is faint or undetectable even after recovery from chronic injury, despite p65 activation in cholangiocytes. Another explanation is that activation of Nrf2 can suppress p65 expression. One study showed that Nrf2 can modulate the NF-κB pathway at the post-translational level.⁴⁸ Nrf2 also prevents IκB-α degradation and thus inhibits NF-κB–mediated transcription.⁴⁹ In the β-catenin KO mice after ANIT, Nrf2 is most active in periportal hepatocytes, which lack p65 in either the cytoplasm or nucleus. While the circumstantial evidence supports the hypothesis that Nrf2 can modulate NF-κB protein expression in our model, further studies would be needed to elucidate the mechanism.

Although Nrf2 signaling ultimately does not prevent ANIT-induced death of β-catenin KO mice, its activation could nonetheless be a potential surrogate marker for increased oxidative stress or decreased expression of oxidative stress regulators like GSH. We have shown that a subset of end-stage PSC explants lack hepatocyte β-catenin.⁹ Here, we showed that PSC explants with inactive β-catenin also have moderate to strong induction of NQO1, a surrogate marker of Nrf2 activity. In this cohort, NQO1 positivity could indicate that loss of β-catenin signaling contributes to disease progression through a resulting increase in oxidative stress, which in turn activates Nrf2 signaling. Indeed, one study showed that patients with end-stage PSC have a dysregulated antioxidant response characterized by reactive aldehydes in the periportal region and suppressed hepatic GST activity.⁵⁰ Thus, our findings not only have physiological relevance to patients but also provide a possible mechanism by which oxidative stress may be enhanced in human cholestatic liver disease. Likewise, although it is currently unclear whether loss of β-catenin alleviates or exacerbates end-stage liver disease in PSC,⁹ our study suggests that restoring redox homeostasis in patients with absent or compromised β-catenin may mitigate injury progression and should thus be considered as a therapeutic intervention in cholangiopathies.