JNKs protect from cholestatic liver disease progression by modulating Apelin signalling

Abstract

Background & aims: Cholestatic liver injury is associated with c-Jun N-terminal kinases (JNK) activation in distinct cell types. Its hepatocyte-specific function during cholestasis, however, has not yet been established. Therefore, in our present study, we investigated the role of JNK1/2 during cholestasis and dissected its hepatocyte-specific function.

Methods: A cohort of patients with primary biliary cholangitis (n = 29) and primary sclerosing cholangitis (n = 37) was examined. Wild-type, hepatocyte-specific knockout mice for Jnk2 (Jnk2^Δhepa) or Jnk1 and Jnk2 (Jnk1^Δhepa/2^Δhepa) were generated. Mice were subjected to bile duct ligation (BDL) or carbon tetrachloride (CCl₄) treatment. Finally, Apelin signalling was blocked using a specific inhibitor. As an interventional approach, Jnk1/2 were silenced in wild-type mice using lipid nanoparticles for small interfering RNA delivery.

Results: JNK activation was increased in liver specimens from patients with chronic cholestasis (primary biliary cholangitis and primary sclerosing cholangitis) and in livers of Mdr2^-/- and BDL-treated animals. In Jnk1^Δhepa/2^Δhepa animals, serum transaminases increased after BDL, and liver histology demonstrated enhanced cell death, compensatory proliferation, hepatic fibrogenesis, and inflammation. Furthermore, microarray analysis revealed that hepatocytic Jnk1/2 ablation induces JNK-target genes involved in oxidative stress and Apelin signalling after BDL. Consequently, blocking Apelin signalling attenuated BDL-induced liver injury and fibrosis in Jnk1^Δhepa/2^Δhepa mice. Finally, we established an interventional small interfering RNA approach of selective Jnk1/2 targeting in hepatocytes in vivo, further demonstrating the essential protective role of Jnk1/2 during cholestasis.

Conclusions: Jnk1 and Jnk2 work together to protect hepatocytes from cholestatic liver disease by controlling Apelin signalling. Dual modification of JNK signalling in hepatocytes is feasible, and enhancing its expression might be an attractive therapeutic approach for cholestatic liver disease.

Impact and implications: The cell-specific function of Jnk genes during cholestasis has not been explicitly explored. In this study, we showed that combined Jnk1/2, but not Jnk2 deficiency, in hepatocytes exacerbates liver damage and fibrosis by enhancing Apelin signalling, which contributes to cholestasis progression. Combined cell-specific Jnk targeting may be a new molecular strategy for treating cholestatic liver disease.

Keywords: Apelin; Cholestasis; Fibrosis; Hepatocytes; c-Jun N-terminal kinases (JNK).

Graphical abstract

Fig. 1

JNK activation in human cholestasis. (A) Liver paraffin sections from healthy patients (n = 5) and patients with cholestasis with PSC (n = 37) and PBC (n = 29) stained with H&E show the typical histological pattern of human PSC and PBC. (B) Representative immunohistochemistry staining for pJNK of the same human liver sections. Scale bar = 100 μm. Black arrows indicate pJNK-positive hepatocytes, whereas red arrows indicate positive NPCs. (C) The percentage of pJNK-positive stained areas was measured using ImageJ® software (National Institutes of Health, Bethesda, MD, USA). (D) Liver extracts from individuals with PSC (n = 3) or PBC (n = 3) in addition to healthy control patients (n = 2) were examined by immunoblot analysis using antibodies against GAPDH, JNK, and pJNK. (E) The density of the pJNK bands was quantified and compared with GAPDH intensity accordingly. The optical density of pJNK/GAPDH ratio in healthy controls was set to 1. (F) Linear regression analysis of pJNK-positive area vs. AP demonstrating that increased pJNK staining correlates with increased AP serum values. (G) Correlation between pJNK-positive area and stages of fibrotic PSC and PBC livers. Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Comparisons of two groups were analysed using an unpaired, two-tailed Student t test. Values are represented as mean ± SEM (n, not significant; ∗p <0.05; ∗∗p <0.01). (H) Correlation between pJNK-positive cholangiocytes and the different fibrosis stages of PSC and PBC livers. Liver sections from patients with cholestasis were triple stained for pJNK (red), nuclei with DAPI (blue), and (I) αSMA (green) or (J) CK19 (green) followed by immunofluorescence microscopy. Arrows indicate double positive cells. αSMA, alpha-smooth muscle actin; AP, alkaline phosphatase; CK19, cytokeratin 19; JNK, c-Jun-N-terminal kinase; NPC, non-parenchymal cell; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; pJNK, phosphorylated JNK.

Fig. 2

Murine cholestasis triggers JNK activation and ablation of JNK1/2 in hepatocytes promotes acute cholestasis. (A, B) Liver extracts prepared from BDL-treated or Mdr2^-/- mice (n = 5) and sham or WT mice (n = 3) were examined by immunoblot analysis using antibodies against GAPDH, JNK, and pJNK. (C) Representative immunohistochemistry staining for pJNK of liver sections from Mdr2^-/- animal and BDL-treated non-transgenic mice are shown at different magnifications (20× in the upper panel and 40× in the lower panel; scale bar = 100 μm). Mdr2^+/+ and sham-operated mice were used as controls. Black arrows indicate pJNK-positive hepatocytes, whereas red arrows indicate positive NPCs. (D) Liver sections from Mdr2^-/- and BDL-treated mice were triple stained for pJNK (red), nuclei with DAPI (blue), and CLEC4F (green) followed by immunofluorescence microscopy. Arrows indicate double positive cells. (E) Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa mice were subjected to 2 days of BDL (n = 6) or sham (n = 3). Serum AST and ALT levels were determined. (F) Macroscopic (upper panel) and microscopic (lower panel) appearance (scale bar = 1 cm) and sections of the liver from the same mice were stained with H&E (lower panel). The scale bar represents 100 μm. Necrotic areas were quantified. Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Comparisons of two groups were analysed using an unpaired, two-tailed Student t test. Values are represented as mean ± SEM (∗p <0.05; ∗∗∗p <0.001). ALT, alanine aminotransferase; AST, aspartate aminotransferase; BDL, bile duct ligation; JNK, c-Jun-N-terminal kinase; Mdr2, multidrug resistance 2; NPC, non-parenchymal cell; pJNK, phosphorylated JNK; WT, wild-type.

Fig. 3

Combined Jnk1/2 loss in hepatocytes exacerbates BDL-induced liver damage, 28 days after BDL. (A) Serum AST and ALT levels were determined in Jnk1^f/f/2^f/f (n = 5) and Jnk1^Δhepa/2^Δhepa mice (n = 6), 28 days after BDL. Sham-operated mice were used as controls (n = 3). (B) Macroscopic appearance of Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers, 28 days after BDL surgery. Sham-operated mice were used as controls (n = 4). Scale bar = 1 cm. Arrows indicate yellow dots (bile infarcts) on the liver surface. (C) Representative H&E staining of paraffin liver sections from these mice are shown at different magnifications (5× in the upper panel and 10× in the lower panel; scale bar = 200 μm). Arrows indicate necrotic areas. Number and size of bile infarcts in the same mice were quantified. (D) Frozen sections from Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers (n = 5) were stained for CK19 followed by immunofluorescence microscopy. Scale bar = 100 μm. Arrows indicate positive cells. Relative CK19-positive stained areas were quantified using ImageJ® software (National Institutes of Health, Bethesda, MD, USA). (E) Representative immunohistochemistry staining for Cl. Casp-3 (upper panel) and Ki-67 (lower panel) of paraffin sections from Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers (n = 5–6), 28 days after BDL. Scale bar = 100 μm. Arrows indicate positive cells. (F) Percentage of positive area per view field for Cl. Casp-3 and Ki-67 were quantified in the same livers using ImageJ® software (National Institutes of Health, Bethesda, MD, USA). Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Values are represented as mean ± SEM (n, not significant; ∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001; ∗∗∗∗p <0.0001). ALT, alanine aminotransferase; AST, aspartate aminotransferase; BDL, bile duct ligation; CK19, cytokeratin 19; Cl. Casp-3, cleaved caspase 3; JNK, c-Jun-N-terminal kinase.

Fig. 4

Hepatic inflammatory response and fibrogenesis are aggravated in Jnk1^Δhepa/2^Δhepa livers after BDL. (A) Representative immunofluorescence staining for CD45, F4/80, or CD11b on liver cryosections was performed in Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers (n = 5), 28 days after BDL. Scale bar = 100 μm. Arrows indicate positive cells. (B) Percentage of positive area per view field for CD45, F4/80, or CD11b were quantified in the same livers using ImageJ® software (National Institutes of Health, Bethesda, MD, USA). (C) Representative Sirius Red staining of paraffin sections and immunofluorescence staining of frozen sections with an antibody against Col1A1 were performed in Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers (n = 5), 28 days after BDL. Scale bar = 100 μm. Arrows indicate positive cells. (D) Quantifications of the Sirius Red- and Col1A1-positive area percentage using ImageJ® software (National Institutes of Health, Bethesda, MD, USA) are shown. (E) Immunofluorescence staining of frozen sections with antibodies against desmin and αSMA were performed in the same liver samples (n = 5). Scale bar = 100 μm. (F) Quantifications of desmin- and αSMA-positive area percentages using Image J® software are shown. Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Values are represented as mean ± SEM (n, not significant; ∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001; ∗∗∗∗p <0.001). αSMA, alpha-smooth muscle actin; BDL, bile duct ligation; Col1A1, collagen 1A1; JNK, c-Jun-N-terminal kinase.

Fig. 5

Jnk1/2-dependent pathways during BDL-induced cholestatic liver injury. (A) Gene array analysis was performed in Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers, 28 days after BDL. Log₂ expression values of the individual mice were divided by the mean of the sham-operated mice. Log ratios were saved in a .txt file and analysed using the Multiple Experiment Viewer. Heat map of the top-regulated, upregulated, and downregulated genes (red, upregulated; green, downregulated, n = 3, -5.0 <fold change >5.0). (B) Ingenuity Pathway Analysis was performed in the same liver samples. The top canonical pathways (left panel), the top upstream regulators (middle panel), and the top up- and down-regulated genes (right panel) are represented. (C, D) Representative immunohistochemistry staining for Muc1 and 4HNE of paraffin sections from Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers (n = 5), 28 days after BDL. Scale bar = 100 μm. The positive area percentage was quantified using Image J® software. Arrows indicate positive cells. (E, F) mRNA expression analysis of acute-phase response genes (Saa1, Saa2, and Saa3), Apelin, and ApelinR was quantified by quantitative reverse-transcription PCR. Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Comparisons of two groups were analysed using an unpaired, two-tailed Student t test. Values are represented as mean ± SEM from five to six mice per group (n, not significant; ∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001). 4HNE, 4-hydroxynonenal; ApelinR, Apelin receptor; BDL, bile duct ligation; JNK, c-Jun-N-terminal kinase; LPS, lipopolysaccharide; Muc1, mucin 1; TNF, tumour necrosis factor.

Fig. 6

JNK in hepatocytes regulates Apelin expression. (A) Representative immunohistochemistry staining for Apelin of paraffin sections from Jnk^f/f (WT), Jnk2^Δhepa, and Jnk1^Δhepa/2^Δhepa livers (n = 5), 28 days after BDL. Scale bar = 100 μm. Arrows indicate positive cells. Sham-operated mice were used as controls. The percentage of positive area per view field for Apelin were quantified in the same livers using ImageJ® software (National Institutes of Health, Bethesda, MD, USA). (B) mRNA expression analysis of Apelin and ApelinR in total liver, hepatocytes, HSCs, and KCs isolated from WT mice (n = 3) was quantified by quantitative reverse-transcription PCR. (C) Schematic view of the experimental design. (D) mRNA expression analysis of Apelin and ApelinR in Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa primary hepatocytes stimulated with or without CA and DCA (n = 6). (E) mRNA expression analysis of Apelin and ApelinR in isolated KCs from WT mice and stimulated with the culture supernatant of primary hepatocytes isolated from Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers and previously stimulated with or without CA and DCA (n = 6). (F) mRNA expression analysis of Apelin and ApelinR in Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa livers from CCl₄-treated mice (n = 5). (G) Representative immunohistochemistry staining for Apelin of paraffin sections from the same mice (n = 5). Scale bar = 100 μm. Arrows indicate positive cells. Oil-treated mice were used as controls. Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Values are represented as mean ± SEM (∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001). ApelinR, Apelin receptor; BDL, bile duct ligation; CA, cholic acid; CCl₄, carbon tetrachloride; DCA, deoxycholic acid; HC, hepatocyte; HSC, hepatic stellate cell; JNK, c-Jun-N-terminal kinase; NT, non-treated; KC, Kupffer cell; WT, wild-type.

Fig. 7

Inhibiting Apelin signalling ameliorates cholestasis and siRNA against Jnk1/2 recapitulates Jnk1^Δhepa/2^Δhepa phenotype after BDL. (A) Representative immunohistochemistry staining for Apelin of human PBC and PSC liver sections (n = 3). Scale bar = 100 μm. Arrows indicate Apelin-positive cells. (B, C) Jnk1^f/f/2^f/f and Jnk1^Δhepa/2^Δhepa mice were treated with ML221 with/without BDL for 7 days (n = 4). SR staining from ML221 + BDL or sham-operated mice and positive area of fibrosis was calculated. Scale bars = 100 μm. (D) Representative H&E staining performed on paraffin liver sections. Scale bars = 100 μm. (E) Serum ALT levels of sham-operated mice, BDL-treated, and ML221 + BDL-treated mice were calculated. (F) Serum AST, ALT, GLDH, and AP levels were determined in siGFP (n = 6) and siJnk1/2 mice (n = 6), 2 days after BDL. (G) Macroscopic and (H) microscopic appearance of livers from the same mice. Arrows indicate yellow dots (bile infarcts) on the liver surface. Necrotic areas were quantified. Statistical evaluation was carried out by multicomparison one-way ANOVA followed by post hoc Bonferroni's test among three or more groups. Comparisons of two groups were analysed using an unpaired, two-tailed Student t test. Values are represented as mean ± SEM (∗p <0.05; ∗∗p <0.01; ∗∗∗p <0.001; ∗∗∗∗p <0.001). ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; BDL, bile duct ligation; JNK, c-Jun-N-terminal kinase; ML221, ApelinR antagonist; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; SR, Sirius Red.