Epidermal growth factor signaling protects from cholestatic liver injury and fibrosis

Abstract We have demonstrated that the signal transducer and activator of transcription 3 (STAT3) protects from cholestatic liver injury. Specific ablation of STAT3 in hepatocytes and cholangiocytes (STAT3∆hc) aggravated liver damage and fibrosis in the Mdr2−/− (multidrug resistance 2) mouse model for cholestatic disease. Upregulation of bile acid biosynthesis genes and downregulation of epidermal growth factor receptor (EGFR) expression were observed in STAT3∆hc Mdr2−/− mice but the functional consequences of these processes in cholestatic liver injury remained unclear. Here, we show normal canalicular architecture and bile flow but increased amounts of bile acids in the bile of STAT3∆hc Mdr2−/− mice. Moreover, STAT3-deficient hepatocytes displayed increased sensitivity to bile acid-induced apoptosis in vitro. Since EGFR signaling has been reported to protect hepatocytes from bile acid-induced apoptosis, we generated mice with hepatocyte/cholangiocyte-specific ablation of EGFR (EGFR∆hc) and crossed them to Mdr2−/− mice. Importantly, deletion of EGFR phenocopied deletion of STAT3 and led to aggravated liver damage, liver fibrosis, and hyperproliferation of K19+ cholangiocytes. Our data demonstrate hepatoprotective functions of the STAT3-EGFR signaling axis in cholestatic liver disease. Key message STAT3 is a negative regulator of bile acid biosynthesis. STAT3 protects from bile acid-induced apoptosis and regulates EGFR expression. EGFR signaling protects from cholestatic liver injury and fibrosis.


Introduction
Chronic cholestatic liver diseases are characterized by retention of bile acids in the liver which results in alterations of hepatobiliary bile acid transport and enzyme activities participating in bile acid biosynthesis. Hydrophobic bile acids are particularly toxic and promote cholangiocyte/hepatocyte damage, liver fibrosis, cirrhosis, and formation of hepatocellular carcinoma (HCC) under cholestatic conditions [1][2][3]. Deposition of collagen and other extracellular matrix components is an orchestrated event in cholestatic liver fibrosis and involves several cell types including Kupffer cells and stellate cells. The latter are activated by inflammatory cytokines and responsible for deposition of collagen together with portal myofibroblasts [2]. Despite profound knowledge about cell types that promote cholestatic liver fibrosis and cirrhosis, little is known about hepatoprotective factors that modulate initial events of cholestatic liver injury. The transcription factor STAT3 is required for liver regeneration and hepatoprotection in various chronic liver diseases [4]. STAT3 is mainly activated by IL-6 (interleukin 6) and IL-22 in hepatocytes. These cytokines bind to gp130 (glycoprotein 130) receptors and promote phosphorylation of STAT3 at tyrosine-705 (pY-STAT3) via Janus kinases (JAKs), STAT3 dimerization, and nuclear translocation [4].
Hepatoprotective functions of STAT3 signaling in cholestatic liver disease have been investigated in mice lacking STAT3, IL-6, gp130, or express pathway specific gp130 mutants. Bile duct ligation, cholic acid feeding, and genetic deletion of the Mdr2 gene (Mdr2 −/− ) resulted in aggravated cholestatic liver injury and fibrosis in hepatocyte-specific STAT3 knock-out mice and IL-6 −/− mice [5,6]. Moreover, mice with hepatocyte-specific deletion of gp130 (gp130 hepa ) displayed aggravated liver fibrosis and collagen deposition after DDC (3,5diethoxycarbonyl-1,4-dihydrocollidine) feeding which is a chemical model for sclerosing cholangitis [7]. The use of specific gp130 knock-in mutant alleles that lack either the region for STAT3 activation (gp130 ΔhepaSTAT ) or carry a Y757F mutation that impedes activation of the Ras/ MAPK pathway (gp130 ΔhepaRas ) [7] demonstrated that the hepatoprotective activity of gp130 signaling was due to gp130-mediated STAT3 activation.
Several cellular and molecular mechanisms might account for the hepatoprotective activity of IL-6/gp130/ STAT3 signaling in cholestatic liver injury [4]. We demonstrated that enzymes for bile acid biosynthesis are upregulated in livers of Mdr2 −/− mice, lacking STAT3 in hepatocytes and cholangiocytes (STAT3 Δhc ) [6], which might result in increased production of toxic bile and aggravated cholestatic liver damage. Moreover, expression of epidermal growth factor receptor (EGFR) was downregulated in livers of STAT3 Δhc Mdr2 −/− mice. EGFR signaling protects hepatocytes from bile acid-induced apoptosis which was demonstrated in vitro with hepatocytes harboring a dominant negative ERBB1 allele. A similar protective effect was observed after pretreatment of hepatocytes with the EGFR antagonist Iressa [8][9][10][11].
Here, we employed Mdr2 −/− mice to functionally test the hepatoprotective function of EGFR signaling in bile acid-induced liver injury and fibrosis. Importantly, conditional inactivation of EGFR in hepatocytes and cholangiocytes (EGFR Δhc ) of Mdr2 −/− mice led to severe jaundice and strongly aggravated liver damage and fibrosis. These data suggest a pivotal hepatoprotective function for EGFR signaling in cholestatic liver disease.

Methods
Mice STAT3 Δhc mice were generated by crossing mice carrying floxed alleles of STAT3 [12] to AlfpCre transgenic mice [13]. Furthermore, mice harboring floxed alleles of EGFR [14] were crossed to AlfpCre transgenic mice. Resulting AlfpCre EGFR flox/flox (EGFR Δhc ) mice were bred with Mdr2 −/− mice [15] to generate EGFR Δhc Mdr2 −/− mice. Blood sera, bile, and liver tissue of a 7 week old male mouse were used for analyses. All mouse experiments were performed in accordance with Austrian and European laws and with the general regulations specified by the Good Science Practice guidelines of the Medical University of Vienna.

qPCR
Total RNA was isolated with TRIzol (Life Technologies, 15596-018) and reverse transcribed with QuantiTect Reverse Transcription Kit (Qiagen, 205313). qPCR was performed using Fast SYBR Green Mastermix (Thermo Fisher Scientific, 4385616) and an Applied Biosystems 7500 Fast Real Time PCR System with primers 5′caccctcaagagcctgagtc-3′ and 5′-gttcgggctgatgtaccagt-3′ for COL1, 5′ -caggtgaacccggcaagaacg-3′ and 5′g g g g a c c a g g g c g a c c a c t -3 ′ f o r C O L 3 , 5 ′t c c t c t t g t t g c t a t c a c t g a t a g c t t -3 ′ a n d 5 ′c g c t g g t a t a a g g t g g t c t c g t t -3 ′ f o r T I M P 1 , 5 ′ccagaagaagagcctgaacca-3′ and 5′-gtccatccagaggcactcatc-3′ for TIMP2, 5′-agaggtcacccgcgtgctaa-3′ and 5′t c c c g a a t g t c t g a c g t a t t g a -3 ′ f o r T G F B 1 , 5 ′tcgtccgctttgatgtctca-3′ and 5′-aaatctcgcctcgagctcttc-3′ for T G F B 2 , 5 ′ -c c g a g g a c t a t g a c c g g g a t a a -3 ′ a n d 5 ′c t t g t t g c c c a g g a a a g t g a a g -3 ′ f o r M M P 2 , 5 ′atcccaccaaagtgagaacg-3′ and 5′-taatttccctccccggttac-3′ for C T G F, 5 ′ -a a t c c c a g g a c c a a c t a t g g c a g c -3 ′ a n d 5 ′g a g g c a a a c t t c t g t t c c a a t g g -3 ′ f o r E G F R , 5 ′tgtttgtgatgggtgtg-3′ and 5′-tacttggcaggtttctc-3′ for GAPDH. The expression levels of transcripts were calculated with the comparative CT (threshold concentration) method. The individual RNA levels were normalized for GAPDH and are depicted as relative expression levels.

Serum measurements
Serum levels of bilirubin, alanine aminotransferase (ALT), and alkaline phosphatase (ALP) were measured using the Reflotron® System (Roche Applied Science).

Bile flow measurement and bile composition
To measure bile flow, mice were anesthetised and kept on a heating plate during the experiment. The common bile duct was ligated using a string. The gall bladder was punctuated and a cannula was inserted and fixed. The bile was collected in a tube for 30 min, and afterwards the liver weight was measured. Bile acids in the bile were analyzed by isotope-dilution gas chromatography-mass spectrometry (GCMS) as described previously [17].

Hydroxyproline
Livers were homogenized and hepatic hydroxyproline levels were measured as described previously [18].

Western blot
Protein lysates were obtained according to standard procedures and analyzed by Western blot with antibodies for P-STAT3 (Cell Signaling, 9145) and β-actin (Sigma, A5316).

Statistics
Significant differences were calculated with GraphPad Prism 5. Comparisons of the two groups were analyzed with unpaired t test or Mann-Whitney test. For more than two groups One-Way Analysis of Variance (ANOVA) and Bonferroni post test or Kruskal Wallis, and Dunns post test were used. Significant differences between experimental groups are stated as: *p < 0.05, **p < 0.01, or ***p < 0.001.

STAT3 inhibits bile acid production
We have recently shown that STAT3 is a negative regulator of bile acid biosynthesis gene expression [6]. Consistent with increased mRNA expression of Cyp7a1 (cytochrome P450 family 7, subfamily A, polypeptide 1) and Cyp27a1 [6] significantly elevated bile acid concentrations were observed in bile collected of STAT3 Δhc (AlfpCre STAT3 flox/flox ) mice (Fig. 1a). GCMS (gas chromatography-mass spectrometry) analysis of bile demonstrated no substantial difference of relative bile acid composition between STAT3 flox/flox and STAT3 Δhc m i c e ( F i g . 1 b ) , e x c e p t f o r i n c r e a s e d U D C A (ursodeoxycholic acid) levels in STAT3 Δhc mice. UDCA is a primary bile acid in mice, although its synthesis is not clearly defined [17]. No difference was observed in bile flow (Fig. 1c) which was consistent with normal morphology of bile canaliculi (Fig. 1d). These data indicate that STAT3 inhibits production of excessive amounts of bile acids. Fig. 1 Increased bile acid concentration in the bile of STAT3 Δhc mice. a The total amount of bile acids was measured in collected bile of STAT3 flox/flox and STAT3 Δhc mice. Bars represent mean data +/− SEM (n ≥ 12 animals per genotype; age = 7 weeks). b GCMS analysis for bile acid composition in STAT3 flox/flox and STAT3 Δhc mice. Note that the relative level of UDCA is elevated in STAT3 Δhc mice (1.806 ± 0.1683 % n = 12 for STAT3 flox/flox and 4.468 ± 0.3927 % n = 13 for STAT3 Δhc mice; p < 0.0001). Bars represent mean data +/− SEM (n ≥ 12 animals per genotype; age = 7 weeks). DCA: deoxycholic acid; UDCA: ursodeoxycholic acid; CDCA chenodeoxycholic acid, α-MCA alpha-muricholic acid, β-MCA beta-muricholic acid, CA cholic acid. c The bile flow was measured by gall bladder intubation in STAT3 flox/flox and STAT3 Δhc mice. Bars represent mean data +/− SEM (n ≥ 12 animals per genotype; age = 7 weeks). d Representative LSM fluorescence micrographs demonstrating normal hepatic microarchitecture in STAT3 Δhc mice (n = 3 animals per genotype). Formalin-fixed and paraffin-embedded (FFPE) liver tissues were stained for nuclei in blue (DAPI), bile canaliculi in green (DPPIV/CD26), periportal hepatocytes in white (GS), and hepatic sinusoids in red (DMs). Scale = 100 μm (low magnification), 30 μm (high magnification)

STAT3 protects hepatocytes from bile acid-induced apoptosis
Primary hepatocytes were isolated and treated with bile acids to investigate if elevated bile acid levels in STAT3 Δhc mice affect survival of STAT3-deficient hepatocytes in a cell-intrinsic manner. Primary STAT3 Δhc hepatocytes were more sensitive to treatment with the bile acid DCA (deoxycholic acid) than STAT3 flox/flox hepatocytes (Fig. 2a), and activation of caspase-3 and caspase-8 was observed (Fig. 2b, c). Moreover, we employed loss of the tumor suppressor protein p19 ARF for hepatocyte immortalization [19] to establish STAT3 flox/flox p19 ARF−/− and STAT3 Δhc p19 ARF−/− hepatocyte cell lines that are reminiscent to primary hepatocytes with respect to morphology and hepatocyte-specific gene expression profiles (data not shown). STAT3 flox/flox p19 ARF−/− and STAT3 Δhc p19 ARF−/− immortalized hepatocytes were treated with DCA, and apoptotic cell death was determined. Similar to primary hepatocytes, cell viability of DCA-treated immortalized STAT3-deficient hepatocytes was reduced (Fig. 2d), and caspase 3 was activated (Fig. 2e). These data demonstrate that STAT3 protects hepatocytes from bile acid-induced death in a cell-intrinsic manner. We have recently shown that expression of hepatoprotective EGFR was reduced in STAT3 Δhc and STAT3 Δhc Mdr2 −/− mice. Therefore, we investigated if EGFR expression is reduced in STAT3-deficient hepatocytes and can be indu ce d by IL-6 i n a STAT 3-d epe nd ent m a nne r. Immortalized STAT3 flox/flox p19 ARF−/− but not STAT3 Δhc p19 ARF−/− hepatocytes displayed strong tyrosine-705 phosphorylation after IL-6 treatment which was maintained for 2 h (Fig. 2f). Expression of STAT3, which is regulated by IL-6/pY-STAT3 signaling in a positive feedback loop [20], was induced in STAT3 flox/flox p19 ARF−/− hepatocytes by IL-6 (Fig. 2g). In contrast, EGFR expression was not induced indicating that it is not regulated by canonical IL-6/pY-STAT3 signaling (Fig. 2h). However, EGFR expression was maintained at a constitutively low level in STAT3 Δhc p19 ARF−/− hepatocytes (Fig. 2h). This suggests that reduced expression of EGFR sensitizes STAT3-deficient hepatocytes to bile acid-induced apoptosis.

STAT3 protects from cholestatic liver injury via regulation of EGFR
We have shown that activated STAT3 and the closely related STAT5 protein protect from cholestasis-induced liver injury by partly overlapping molecular mechanisms that include regulation of EGFR [6,21,22]. Therefore, we employed a genetic approach to evaluate if EGFR is a crucial hepatoprotective factor in cholestatic liver injury and used mice with conditional deletion of EGFR in hepatocytes and cholangiocytes (EGFR Δhc ). Because recent evidence has suggested that hepatocyte-specific AlfpCre mice display Cre effects could lead to in vivo artifacts [23], we generated all possible genotypes (wild-type, AlfpCre, AlfpCre EGFR flox/ f l o x = EGFR Δ h c , AlfpCre Mdr2 − / − and AlfpCre EGFR flox/flox Mdr2 −/− = EGFR Δhc Mdr2 −/− ) of mice and performed biochemical and histopathological analyses of liver injury and fibrosis. Importantly, EGFR Δhc Mdr2 −/− mice displayed aggravated liver fibrosis and hepatic damage when compared with control mice. Bilirubin levels were elevated in the serum of EGFR Δhc Mdr2 −/− mice (Fig. 3a, b) and they showed signs of jaundice (Fig. 3c). The liver to body weight ratio was increased in EGFR Δhc Mdr2 −/− mice when compared with Mdr2 −/− mice (Fig. 3d). This was, however, partially due to the AlfpCre transgene because AlfpCre Mdr2 −/− mice without EGFR deletion also displayed an increased liver to body weight ratio (Fig. 3d). H&E staining of liver biopsies revealed prominent periportal fibrosis and immune cell infiltration in EGFR Δhc Mdr2 −/− mice (Fig. 3e) which was reflected by elevated serum levels of liver damage parameters (Fig. 3f, g) and proliferation of bile ducts (Fig. 3h). However, hepatocyte proliferation (Fig. 3i), apoptosis (Fig. 3j) or numbers of macrophages and T cells in the inflammatory infiltrates (Fig. 3k, l) were not changed in EGFR Δ h c Mdr2 − / − mice. These data

STAT3 protects from cholestatic liver fibrosis via regulation of EGFR
Sirius-red staining for collagen deposition and biochemical measurement of hydroxyproline levels, indicative for collagen deposition, demonstrated aggravated liver fibrosis in EGFR Δhc Mdr2 −/− mice (Fig. 4a-c). AlfpCre Mdr2 −/− mice were comparable to Mdr2 −/− mice demonstrating that aggravated liver fibrosis is due to EGFR deletion but not AlfpCre transgene expression (Fig. 4a-c). qPCR analysis demonstrated increased expression of several key genes implicated in fibrosis in EGFR Δhc Mdr2 −/− mice (Fig. 4d). These data demonstrate that EGFR signaling protects from hepatic fibrosis in Mdr2 −/− mice.

Discussion
Hepatic fibrosis is due to chronic liver injury and partially reversible which puts hepatoprotective factors for antifibrotic therapies into the limelight. Genetically modified mouse models for liver fibrosis [24,25] have unraveled effector molecules such as TGF-ß (transforming growth factor beta) [26], PDGF-B [27] (platelet derived growth factor b), PDGF-C (platelet derived growth factor c) [28], or TIMP-1 (tissue inhibitor of metalloproteinase 1) [29], but hepatoprotective factors are not well characterized. We have recently shown that the cytokine IL-6 and the cytokine-inducible transcription factor STAT3 protect from cholestatic liver injury and fibrosis in the Mdr2 −/− mouse model for cholestatic liver disease [6]. Both, IL-6 −/ − Mdr2 −/− and STAT3 Δhc Mdr2 −/− mice showed aggravated liver damage and deposition of collagen in the periportal areas. Gene expression profiling demonstrated that genes for bile acid biosynthesis enzymes were upregulated whereas, EGFR was downregulated in STAT3 Δhc and STAT3 Δhc Mdr2 −/− mice. The mode how STAT3 represses bile acid biosynthesis genes and the implication of known regulators such as FXR-α (farnesoid X receptor alpha) or HNF-4 (hepatocyte nuclear factor 4) [30,31] has to be determined. We show that blunted repression of bile acid biosynthesis genes at the mRNA levels is reflected by a more than two-fold increase of total bile acid concentrations in the bile of STAT3 Δhc mice. Mdr2 −/− mice lack a phospholipid pump in the canalicular Fig. 4 Severe liver fibrosis in EGFR Δhc Mdr2 −/− mice. a Sirius-red staining on liver sections of indicated genotypes showing increased collagen deposition in periportal areas of EGFR Δhc Mdr2 −/− mice (arrowheads). Scale = 100 μm. b Quantitation of Sirius-red-stained area on liver sections of indicated genotypes using histomorphometry (n ≥ 4). c Collagen deposition was quantified using biochemical determination of hydroxyproline levels in livers of indicated genotypes (n ≥ 5). d qPCR for fibrosis markers in livers of indicated genotypes. Bars represent data +/− SEM of n ≥ 4 livers per genotype. COL1 type I collagen, COL3 type III collagen, TIMP1 tissue inhibitor of matrix metalloproteinase 1, TIMP2 tissue inhibitor of matrix metalloproteinase 2, TGFB1 transforming growth factor beta 1; TGFB2 transforming growth factor beta 2, MMP2 matrix metalloproteinase 2, CTGF connective tissue growth factor