Growth hormone resistance exacerbates cholestasis-induced murine liver fibrosis

Abstract

Growth hormone (GH) resistance has been associated with liver cirrhosis in humans but its contribution to the disease remains controversial. In order to elucidate whether GH resistance plays a causal role in the establishment and development of liver fibrosis, or rather represents a major consequence thereof, we challenged mice lacking the Growth hormone receptor gene (Ghr^-/-, a model for GH resistance) by crossing them with Mdr2 knockout mice (Mdr2^-/-), a mouse model of inflammatory cholestasis and liver fibrosis.

Ghr^-/-;Mdr2^-/- mice showed elevated serum markers associated with liver damage and cholestasis, extensive bile duct proliferation and increased collagen deposition relative to Mdr2 ^-/- mice, thus suggesting a more severe liver fibrosis phenotype. Additionally, Ghr^-/-;Mdr2^-/- mice had a pronounced down-regulation of hepato-protective genes Hnf6, Egfr and Igf-1, and significantly increased levels of ROS and apoptosis in hepatocytes, compared to control mice. Moreover, single knockout mice (Ghr^-/-) fed with a diet containing 1% cholic acid displayed an increase in hepatocyte ROS production, hepatocyte apoptosis and bile infarcts compared to their wildtype littermates, indicating that loss of Ghr renders hepatocytes more susceptible to toxic bile acid accumulation. Surprisingly, and despite their severe fibrotic phenotype, Ghr^-/-;Mdr2^-/- mice displayed a significant decrease in tumour incidence compared to Mdr2^-/- mice, indicating that loss of Ghr signaling may slow the progression from fibrosis/cirrhosis to cancer in the liver.

Conclusion

Our findings suggest that GH resistance dramatically exacerbates liver fibrosis in a mouse model of inflammatory cholestasis, therefore suggesting that GH resistance plays a causal role in the disease and provides a novel target for the development of liver fibrosis treatments.

Introduction

Liver fibrosis is the pathological accumulation of fibrous scar tissue in the liver, eventually leading to a loss of hepatic function.¹ Prolonged insults, such as chronic infections (hepatitis), toxins, alcohol abuse, morbid obesity, nonalcoholic fatty liver disease and cholestasis can lead to the development of this severe phenotype.² The disease is often asymptomatic and can progress to liver cirrhosis and may ultimately lead to hepatocellular carcinoma (HCC). Liver damage can result in apoptosis and necrosis of hepatocytes which subsequently leads to the activation of Kupffer cells. These liver resident macrophages produce, amongst other cytokines, tranforming growth factor beta (TGF-β), monocyte chemoattractant protein-1(MCP-1) and platelet-derieved growth factor (PDGF), which activate and mobilise quiescent hepatic stellate cells (HSCs). Once activated, HSCs differentiate into fibrogenic myofibroblasts and secrete TGF-β, Interleukin-6 (IL-6) and tumor necrosis factor alpha (TNF-α), which then act in an autocrine and paracrine manner promoting liver remodeling by secretion of various extracellular matrix (ECM) components and ECM remodeling enzymes which ultimately leads to liver fibrosis.^1–3

Growth hormone (GH) signaling plays a central role in the physiology and pathophysiology of the liver. Binding of GH to its cognate receptor (Growth hormone receptor, GHR) causes receptor dimerization and Janus kinase 2 (JAK2) activation. Activated JAK2 phosphorylates signal transducer and activator of transcription 5 (STAT5) which, upon dimerization, translocates to the nucleus where it acts as a transcription factor controlling the expression of various genes responsible for vital liver function.⁴ Growth hormone resistance [a condition defined by high levels of circulating GH and low levels of insulin-like growth factor 1 (IGF-1)] is commonly found in patients suffering from liver fibrosis and cirrhosis.⁵ It has been extensively debated, whether the alterations of the GH-IGF-1 axis observed in liver cirrhosis are mere consequences of the malfunctioning liver in advanced stages of the disease or, whether they promote the onset and progression of liver fibrosis.⁵ To address this question, we challenged GHR knockout mice (Ghr^-/-, a mouse model for GH resistance⁶),by crossing them with the Multidrug resistance gene 2 knockout mouse (Mdr2 (Abcb4) ^-/-), a mouse model of inflammatory cholestasis and liver fibrosis.⁷

Materials and Methods

Animals

Mice carrying the Ghr/bp deletion (Ghr^-/-) and Mdr2^-/- mice were described previously^6,7 By respective intercrossing we generated double knockout mice (Ghr^-/-;Mdr2^-/-) and maintained animals on a mixed genetic background (129Sv /C57BL/6). Mdr2^+/- mice were disease-free and used as littermate controls. For experimental procedures we used 8 week old male mice unless otherwise stated. Bile acid overload was induced in mice with a diet containing 1% cholic acid for 8 weeks. Mice were kept at the Decentralized Biomedical Facilities, Medical University of Vienna, under standardized conditions, and all animal experiments were carried out according to an ethical animal license protocol and contract approved by the Medical University of Vienna and Austrian Federal Ministry of Education, Arts and Culture authorities.

Additional Materials and Methods

Animal and histology procedures, quantitative reverse transcription polymerase chain reaction (qRT-PCR), serum biochemistry, Western Blot analysis, hydroxyproline measurements, microarray analysis, immunohistochemistry, primary hepatocyte isolation and cell treatments and statistical analysis are described in the Supporting Materials and Methods.

Results

Ghr^-/-;Mdr2^-/- mice develop hepatic injury and impaired bile acid homeostasis

Macroscopic analysis revealed that livers of Ghr^-/-;Mdr2^-/- mice were smaller than littermate controls (Wt, Mdr2^-/- and Ghr^-/-) and displayed enlarged gall bladders at 8 weeks of age (Fig. 1A). Mdr2^-/- mice and Ghr^-/-;Mdr2^-/- mice showed a significant increase in liver weight/body weight ratio, in contrast Ghr^-/- mice showed a significant decrease compared to Wt littermates (Fig. 1B). As expected, Ghr^-/- and Ghr^-/-;Mdr2^-/- mice show increased levels of circulating GH and low serum IGF-1 levels (Fig. 1C). Furthermore, mRNA expression of Als was down-regulated in Ghr^-/- and Ghr^-/-;Mdr2^-/- mice, while Igf1bp1 mRNA expression was up-regulated (Supporting Fig. 1E), thus further reducing IGF-1 bioavailability.⁵ These results are in line with a severe GH resistance phenotype in these mice. Serum parameter levels indicative of liver injury, such as alkaline phosphatase (ALP), aspartate amino transferase (AST) and alanine aminotransferase (ALT), were strongly increased in Ghr^-/-;Mdr2^-/- mice compared to all control mice (Fig. 1D). Furthermore, circulating levels of bilirubin and bile acids levels were greatly elevated in sera obtained from Ghr^-/-;Mdr2^-/- animals, suggesting a severe cholestatic phenotype (Fig. 1D). Expression analysis of genes implicated in bile acid metabolism revealed a down-regulation of the basolateral bile acid importer organic anion transporter polypeptide 1 (Oatp1) in Ghr^-/- and Ghr^-/-;Mdr2^-/- mice (Supporting Fig. 1D). Interestingly, we did not observe Cyp7A1 repression in Ghr^-/-;Mdr2^-/- mice, despite the high levels of bile acids in these animals (Supporting Fig. 1D). These results propose that Ghr^-/-;Mdr2^-/- mice display severe liver injury, strong cholestasis and deregulation of bile acid homeostasis.

Figure 1.

Ghr deletion combined with loss of Mdr2 increases liver damage. (A) Macroscopic analysis reveals significant liver size reduction in Ghr^-/-;Mdr2^-/- mice compared to controls. (B) Liver to body weight ratio (LW/BW) was determined in all 4 genotypes (n = 5/genotype). (C) Levels of GH and IGF-1 within the serum were measured by ELISA and demonstrated GH resistance in Ghr^-/- and Ghr^-/-;Mdr2^-/- mice (n = 6/genotype). (D) Analysis of serum parameters revealed a significant increase of AST, ALT, ALP, bilirubin and bile acids in Ghr^-/-;Mdr2^-/- mice relative to all controls (n ≥ 5/genotype). Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GH, growth hormone; IGF-1, insulin like growth factor 1

Ghr^-/-;Mdr2^-/- mice develop a severe liver fibrosis phenotype

Analysis of hematoxylin and eosin stainings (HE) revealed disruption of liver architecture in Ghr^-/- ; Mdr2^-/- mice compared to all control animals. Additionally, CAB staining analysis of liver sections showed a higher degree of collagen deposition resembling bridging fibrosis in Ghr^-/-;Mdr2^-/- mice compared to all controls (Fig.2A and Supporting Fig.1A). Surprisingly, we did not observe increases of Col1a1 and Col3a1 mRNA expression in Ghr^-/-;Mdr2^-/- mice compared to Mdr2^-/- animals (Supporting Fig. 1B). Nevertheless, to assess the degree of fibrosis, we analysed whole liver collagen content by measuring hydroxyproline levels which revealed Ghr^-/-;Mdr2^-/- mice had a massive increase in hydroxyproline levels compared to all the controls (Fig. 2B). Furthermore, hepatocytes of Ghr^-/-;Mdr2^-/- mice contained less glycogen (as measured by PAS staining), indicating a reduction of hepatocyte function (Fig. 2A). Liver sections stained for the cholangiocyte marker CK19 also showed a higher degree of positive staining in Ghr^-/-;Mdr2^-/- mice compared to all the controls suggesting a greater degree of cholangiocyte proliferation (Fig. 2A). Overall, these findings indicate that Ghr^-/-;Mdr2 ^-/- mice have strongly aggravated fibrotic phenotype typified by extensive collagen deposition, cholangiocyte proliferation and bile acid homeostasis deregulation.

Figure 2.

Ghr^-/-;Mdr2^-/- mice develop a severe liver fibrosis phenotype. (A) Histological and immunohistochemical analysis of liver sections from mice at 8 weeks of age stained with HE, CAB, PAS and CK19. Ghr^-/-;Mdr2^-/- mice showed disrupted liver architecture, increased collagen deposition and cholangiocyte proliferation as well as severe glycogen reduction in hepatocytes compared to all controls. (B) Hydroxyproline was measured in livers from 8 weeks old mice (n ≥ 6/genotype). Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

HE, hematoxylin-eosin; CAB, chromotrope aniline blue; PAS, periodic acid-Schiff; CK-19, cytokeratin 19.

Ghr^-/-;Mdr2^-/- mice exhibit up-regulation of pro-fibrogenic factors

To investigate the Ghr^-/-;Mdr2^-/- fibrotic phenotype further, we performed immunohistochemistry to analyse alpha smooth muscle actin (α-SMA) levels, a marker of activated hepatic stellate cells (HSC). Livers of Ghr^-/-;Mdr2^-/- mice demonstrated increased α-SMA staining (Fig. 3A). This result was further confirmed by qRT-PCR and immunoblot analysis of whole liver homogenates, both demonstrating marked increases in aSma mRNA expression and α-SMA protein levels in Ghr^-/-;Mdr2^-/- mice compared to controls (Fig. 3B and 3C). Interestingly, α-SMA protein levels were also increased in Ghr^-/- mice compared to Wt mice (as measured by immunohistochemistry, western blot and qRT-PCR (Fig. 3A, 3B and 3C), and the average of the latter two methods (Supporting Fig. 1C). This could indicate a larger reservoir of activated HSC residing in livers lacking Ghr^-/-. Additionally, transcription of characteristic pro-fibrogenic factors, such as Pdgfβ, Pdgfrβ, Tgfβ, Tgfβr1 and Tnfα where highly up-regulated in Ghr^-/-;Mdr2^-/- mice (Fig. 3D). In addition, livers of Ghr^-/-;Mdr2^-/-mice showed increased levels of p-SMAD2/3 and an enrichment of a gene signature set characteristic of the TGFβ pathway, thus indicating activation of the TGFβ signalling pathway in this genotype (Supporting Fig. 2B and 7A). Furthermore, expression of matrix metalloproteinases (MMPs) and their inhibitors (tissue inhibitors of metalloproteinases, TIMPs), namely Mmp2, Mmp3, Mmp14, Timp1 and Timp2, were up-regulated in Ghr^-/-;Mdr2^-/- mice compared to all the experimental groups (Fig. 3E). Collectively, these data indicate that deletion of Ghr in Mdr2^-/- mice severely magnifies liver fibrosis in these mice.

Figure 3.

Ghr^-/-;Mdr2^-/- mice display up-regulation of profibrogenic growth factors, cytokines and extracellular matrix remodeling factors. (A) Immunohistochemical analysis of liver sections taken from mice at 8 weeks of age stained for α-SMA. (B) Representative western blot analysis of α-SMA levels in whole liver homogenates indicates increased activated hepatic stellate cells in livers from Ghr^-/-, Mdr2^-/- and Ghr^-/-;Mdr2^-/- mice. HSC70 served as a loading control. (C) Relative mRNA levels of αSma were quantified by qRT-PCR from mouse livers at 8 weeks of age and normalized to Gapdh (n ≥ 4/genotype). (D) Pdgfβ, Pdgfrβ, Tgfβ, Tgfβr1 and Tnfα genes showed significantly increased mRNA levels in Ghr^-/-;Mdr2^-/- mice compared to all controls. Relative mRNA levels were quantified by qRT-PCR from livers of 8 weeks old mice and normalized to Gapdh (n ≥ 4/genotype). (E) mRNA levels of matrix metalloproteinases (MMPs) and tissue inhibitor of metalloproteinases (TIMPs) were extensively increased in Ghr^-/-;Mdr2^-/- mice, suggesting a continuous remodelling of the extracellular matrix. Relative mRNA levels were quantified by qRT-PCR from livers of 8 weeks old mice and normalized to Gapdh (n ≥ 4/genotype). Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

α-SMA, alpha-smooth muscle actin; HSC70, Heat shock protein 70

Deletion of Ghr sensitizes the hepatocytes to bile acid-induced toxicity

To gain mechanistic insights into the development of liver fibrosis we explored potential differences in proliferation and apoptosis in the livers of experimental mice. Liver sections stained and quantified for Ki67 revealed no significant differences in hepatocyte proliferation between the experimental groups (Fig. 4A and 4B). However, quantification of Apoptag stainings showed a highly significant increase in hepatocyte apoptosis in Ghr^-/-;Mdr2^-/- mice relative to all experimental groups (Fig. 4C and 4D). Furthermore, gene set enrichment analysis (GSEA) revealed that livers from Ghr^-/-;Mdr2^-/- compared to Mdr2^-/- mice show a significant enrichment of genes associated with apoptosis (Supporting Fig.7B). This finding prompted us to analyse the expression of genes suggested to play a hepato-protective (pro-survival) role in hepatocytes such as Egfr and Hnf-6.^8–10 Both genes were shown to be diminished in Ghr^-/-;Mdr2^-/- and Ghr^-/- livers relative to control groups (Fig. 4F). Moreover, total EGFR protein was only detectable in Wt and Mdr2^-/- livers with p-EGFR solely observed in Mdr2^-/- mice (Fig. 4E and Supporting Fig. 3B). In line, p-ERK, a downstream kinase in the EGFR/GHR pathways,¹¹ was reduced in Ghr^-/- and Ghr^-/-;Mdr2^-/- compared to Mdr2^-/- hepatocytes (Fig. 4E and Supporting Fig. 3A). Additionally, p-AKT remained unchanged in hepatocytes among each genotype (Supporting Fig. 3C), but we observed a reduction in S6K phosphorylation in livers of Ghr^-/- and Ghr^-/-;Mdr2^-/- mice compared to control mice (Supporting Fig.2A). It has been previously shown that toxic bile acids may induce hepatocyte apoptosis, liver injury and contribute the development of liver fibrosis.¹². As mentioned previously, Ghr^-/-;Mdr2^-/- mice display elevated bile acids levels compared to controls (Fig. 1D). Consequently, we challenged Wt and Ghr^-/- mice with 1% cholic acid diet to observe if lack of Ghr causes increased liver damage under cholestatic dietary conditions. Serum parameters (ALP, AST and ALT) were significantly elevated in Ghr^-/- mice compared to Wt littermates fed with cholic acid (Fig. 5A). Histological analysis of HE liver sections revealed necrotic areas and bile duct infarcts in Ghr^-/- mice, whereas Wt controls displayed a largely normal liver architecture (Fig. 5B). Furthermore, Ghr^-/- mice showed increased hepatocyte proliferation and apoptosis demonstrated by Ki67 and Apoptag staining and quantification (Fig. 5B and 5C). Additionally, in vitro experiments using isolated primary hepatocytes from Ghr^-/- mice treated with Deoxycholic acid, (DCA), TGFβ or TNFα either alone or in combination, showed a significant reduction in cell viability compared to Wt control cells (Fig. 5D, 5E and 5F). Taken together, these results suggest that Ghr^-/- animals are more susceptible to both bile acid and cytokine induced liver injury due to a reduction in pro-survival factors.

Figure 4.

Ghr^-/-;Mdr2^-/- mice show a down-regulation in hepato-protective factors accompanied by an increase in hepatocyte apoptosis. (A) Immunohistochemical analysis of liver sections from mice at 8 weeks of age stained for Ki67. (B) Quantification of Ki67 positive hepatocytes showed no significant difference between Mdr2^-/- and Ghr^-/-;Mdr2^-/- mice (n = 5/genotype). (C) Immunohistochemical staining of liver sections stained with Apoptag shows an increase in hepatocyte apoptosis in Ghr^-/-;Mdr2^-/- mice. (D) Quantification of hepatocytes undergoing apoptosis seen in (C) (n = 5/genotype). (E) Representative western blot analysis of whole liver homogenates from 8 week old mice revealed activation of p-EGFR only in Mdr2^-/- mice and a down-regulation of EGFR, p-ERK and ERK in Ghr^-/- and Ghr^-/-;Mdr2^-/- livers. HSC70 served as a loading control. (F) Relative mRNA levels of Egfr and Hnf6 were quantified by qRT-PCR from mouse livers at 8 weeks of age and normalized to Gapdh (n ≥ 4/genotype). Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

EGFR, epidermal growth factor receptor; p-ERK, phospho-extracellular-signal regulated kinase; ERK, extracellular-signal regulated kinase; HSC70, heat shock protein 70

Figure 5.

Hepatocytes of Ghr^-/- mice are more susceptible to bile acid induced damage in vivo and in vitro compared to Wt littermate controls. (A) Ghr^-/- and Wt mice were fed with a 1% cholic acid enriched diet. Serum levels of ALS, AST and ALP were significantly elevated in Ghr^-/- mice compared to Wt mice, suggesting increased liver damage in Ghr^-/- mice. (B) Immunohistochemical analysis of liver sections from Ghr^-/- and Wt mice fed with a 1% cholic acid enriched diet stained with HE, ki67 and Apoptag. (C) Quantification of Ki67 and Apoptag positive hepatocytes from mice fed with a diet containing 1% cholic acid showed increased apoptosis and proliferation in Ghr^-/- mice relative to Wt littermates. (n = 5/genotype). (D) Primary hepatocytes from Wt and Ghr^-/- mice were treated with 200µm or 250µm DCA for 4 hours. Ghr^-/- hepatocytes showed a significant decrease in cell viability compared to Wt hepatocytes. (E) Primary hepatocytes were treated with either 20ng/µl TNFα alone for 20 hours (left panel) or in combination with 250µm DCA for 3 hours (right panel). (F) Primary hepatocytes were treated with either 10ng/µl TGFβ alone for 48 hours (left panel) or in combination with 250µm DCA for 3 hours (right panel) followed by MTT cell viability assay. Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HE, hematoxylin-eosin; DCA, deoxycholic acid

GHR protects from bile acid-induced ROS in hepatocytes

It has been previously demonstrated that bile acids can induce reactive oxygen species (ROS), oxidative damage, mitochondrial dysfunction and apoptosis in hepatocytes.^13–15 Therefore, we investigated ROS accumulation in our experimental mice by staining liver sections with an antibody against 4-Hydroxynonenal (4-HNE) (a side product of lipid peroxidation caused by ROS¹⁶). We observed highly increased 4-HNE levels in hepatocytes of Ghr^-/-;Mdr2^-/- mice compared to all other experimental mice (Fig. 6A and 6B). Furthermore, qRT-PCR analysis of mRNA from whole liver revealed that compared to all other controls, Ghr^-/-;Mdr2^-/- mice had an up-regulation of genes which are shown to be increased by ROS ^17–19, including Nrf2, Nqo1, Trp53 and its target genes Noxa and Mdm2 (Supporting Fig. 4A), suggesting increased ROS production within the livers of these mice. We confirmed these results by performing immunohistochemical stainings against heme oxygenase 1 (HO-1, Hmox1), a marker of increased cellular stress andROS levels ¹⁷. HO-1 was strongly up-regulated in hepatocytes of Ghr^-/-;Mdr2^-/- mice (Fig. 6C). Furthermore, qRT-PCR analysis of primary hepatocytes vs. non hepatocytes taken from Ghr^-/-;Mdr2^-/- mice revealed a significantly higher expression of Nrf2 and its target genes Hmox1 as well as Nqo1 (all major regulators activated in response to oxidative stress ²⁰) in the hepatocyte fraction of Ghr^-/-;Mdr2^-/- mice (Fig. 6D). Importantly, 4-HNE staining of liver sections from Wt and Ghr^-/- mice fed with a 1% cholic acid enriched diet showed an increase in 4-HNE production in Ghr^-/- compared to Wt hepatocytes (Fig. 6E and 6F). These findings indicate that hepatocytes which are lacking Ghr are more susceptible to bile acid-induced ROS and ROS induced cell damage.

Figure 6.

Hepatocytes of Ghr^-/-;Mdr2^-/- mice show an increase of ROS production due to toxic bile acid accumulation. (A) Immunohistochemical analysis of liver sections from mice at 8 weeks of age stained with an antibody against 4-HNE. (B) Quantification of 4-HNE positive hepatocytes from (A) showed severe increase of lipid peroxidation in hepatocytes of Ghr^-/-;Mdr2^-/- mice (n = 5/genotype). (C) Immunohistochemical analysis of liver sections from mice at 8 weeks of age stained with an antibody against HO-1. (D) Relative mRNA levels of Nrf2, Nqo1 and Hmox1 were quantified by qRT-PCR in isolated hepatocytes compared to non-hepatocytes from 8 week old mice and normalized to Gapdh (n ≥ 5/genotype). (E) Liver sections of Ghr^-/- and Wt mice fed with a cholestatic diet containing 1% cholic acid stained with an antibody against 4-HNE reveals increased lipid peroxidation in hepatocytes of Ghr^-/- mice. (F) Quantification of 4-HNE staining from (E). Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

ROS, reactive oxygen species; 4-HNE, 4-hydroxynonenal; HO-1, heme oxygenase 1

GHR deletion abrogates liver tumor formation

Our results indicate that Ghr^-/-;Mdr2^-/- have a more severe and advanced liver fibrosis phenotype than Mdr2^-/- mice. As fibrosis and cirrhosis can often progress to liver cancer, we further analysed tumor formation in each experimental genotype at 12 months of age. Surprisingly, Ghr^-/-;Mdr2^-/- mice showed a significant decrease in tumor incidence compared to Mdr2^-/- mice despite their severe degree of fibrosis (Fig. 7A and 7B). Only 2 out of 10 Ghr^-/-;Mdr2^-/- mice developed tumors, whereas 11 out of 12 Mdr2^-/- mice developed tumors. Additionally, tumors from Mdr2^-/- mice were larger and more abundant (total tumor number: Ghr^-/-;Mdr2^-/- n = 4; Mdr2^-/- n = 50) (Fig. 7A and 7B). Histopathological analysis revealed that Mdr2^-/- tumors presented a more solid growth pattern, increased cellular polymorphism but had similar populations of oval cells compared to Ghr^-/-;Mdr2^-/- tumors (Fig. 7F and Supporting Fig. 5A and 5B). This indicates that Ghr^-/-; Mdr2^-/- tumors are less abundant and advanced than Mdr2^-/- tumors. Hepatocyte proliferation was increased in non-tumor areas of Mdr2^-/- compared to Ghr^-/-;Mdr2^-/- mice, but we did not find differences between both genotypes in the tumor areas (Fig. 7C). Importantly, Apoptag staining revealed a significant increase in apoptosis in both non-tumorigenic tissue and tumorigenic tissue in Ghr^-/-;Mdr2^-/- compared to Mdr2^-/- mice (Fig. 7D). Additionally, p-ERK levels were substantially decreased in hepatocytes from non-tumorigenic tissue in Ghr^-/-;Mdr2^-/- compared to Mdr2^-/- mice (Supporting Fig. 6A) and slightly decreased within the Ghr^-/-;Mdr2^-/- tumors (Supporting Fig. 6A). EGFR protein expression was absent in tumorigenic and non-tumorigenic hepatocytes in Ghr^-/-;Mdr2^-/- mice (Supporting Fig. 6B ), whereas p-AKT levels remained unchanged (Supporting Fig. 6C). Moreover, cell cycle inhibitors and tumor suppressors such as Cdkn1a, Cdkn1b and Trp53 were highly up-regulated in Ghr^-/-;Mdr2^-/- mice compared to Mdr2^-/- mice (Fig. 7E and Supporting Fig. 4A). In this line, GSEA revealed that Ghr^-/-;Mdr2^-/- livers display an upregulation of genes normally silenced in various kinds of HCCs (Supporting Fig. 7C). These results suggest that impairment of GH signaling exacerbates the liver fibrosis phenotype but decreases tumor development.

Figure 7.

Deletion of Ghr suppresses liver tumorigenesis. (A) Macroscopic pictures of livers of experimental mice show numerous tumors in Mdr2^-/- mice compared to Ghr^-/-;Mdr2^-/- mice at 12 months of age. (B) Quantification of tumor number/mouse in all four genotypes. Each data point represents one animal (n ≥ 10/genotype). (C) Immunohistochemical analysis of liver sections from mice at 12 months of age stained with ki67 showed increased hepatocyte proliferation in Mdr2^-/- relative to Ghr^-/-;Mdr2^-/- mice at the age of 12 months in non-tumor areas, but no differences in cell proliferation rates were observed in tumorigenic tissues from both genotypes (n = 10/genotype) (T: tumor, NT: non-tumor area). (D) Apoptag staining indicated increased apoptosis in Ghr^-/-;Mdr2^-/- non-tumorigenic and tumorigenic tissue at 12 months compared to Mdr2^-/- mice (n = 10/genotype) (T: tumor, NT: non-tumor area). (E) mRNA levels of cell cycle inhibitors Cdkn1a and Cdkn1b were significantly increased in Ghr^-/-;Mdr2^-/- mice compared to controls in isolated hepatocytes of mice at 8 weeks of age (n ≥ 5/genotype). (F) Tumors from Ghr^-/-;Mdr2^-/- mice are less advanced compared to Mdr2^-/- tumors. Tumor grading reveals that Mdr2^-/- mice develop more tumors with cellular polymorphism and solid growth pattern. Results are shown as mean ± standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

Discussion

Clinical symptoms of liver cirrhosis include disturbed energy balance, dysfunction of secondary metabolism and signs of malnutrition.²¹ Dysregulation of several hormones, which control the secondary metabolism, have been associated with liver cirrhosis as is the case in GH resistance syndrome.²¹ This condition manifests itself with high levels of circulating GH and low levels of insulin-like growth factor 1 (IGF-1) and is commonly found in patients suffering from liver cirrhosis.⁵ It has been shown that mice with low or absent IGF-1 levels develop severe liver fibrosis upon challenge with diverse models of cholestasis.^22–24 Furthermore, supplementation of recombinant IGF-1 was shown to be therapeutically beneficial in fibrotic animal models²⁵ and pilot clinical trials.^26,27 Along this line, GH replacement therapy in an adult man suffering from non-alcoholic steatohepatitis (NASH) and fibrosis associated with GH deficiency ameliorated NASH and liver fibrosis.²⁸ Conversely, patients suffering from hypopituitarism develop non-alcoholic fatty liver disease (NAFLD) and cirrhosis.²⁹ Of note, in humans it has been shown that GH resistance is associated with cholestatic diseases, such as primary biliary cirrhosis.³⁰ Here, we provide clear genetic evidence that Ghr deletion aggravates liver fibrosis when challenged with genetic or pharmacological cholestatic models. Ghr^-/-;Mdr2^-/- mice display severe liver injury, as well as high levels of bilirubin and bile acids (Fig. 1D). Furthermore, Ghr^-/-;Mdr2^-/- mice appear to have an impaired bile acid homeostasis and despite their high levels of circulating bile acids, Ghr^-/-;Mdr2^-/- mice failed to repress NTCP and Cyp7a1, a phenomena that may contribute to the pronounced cholestatic phenotype³¹ observed in these mice (Supporting Fig. 1C).

Histological examination of liver sections not only revealed severe disruption of the liver architecture and portal-portal-bridging fibrosis, but also impaired hepatocyte function, as PAS stainings revealed an almos complete lack of glycogen within hepatocytes of Ghr^-/-;Mdr2^-/- mice. Pro-fibrogenic genes were deregulated in Ghr^-/-;Mdr2^-/- mice. Surprisingly, we found α-SMA positive hepatic stellate cells not only in livers of Ghr^-/-;Mdr2^-/- and Mdr2^-/- mice, but also in Ghr^-/- mice (Fig. 3A, B and C). This suggests that hepatic stellate cells are more susceptible to activation in Ghr^-/- mice, a scenario that may in part explain the exaggerated collagen deposition observed in livers of Ghr^-/-;Mdr2^-/- mice. In this context, over-expression of IGF-1 in HSCs protects from liver fibrosis and has been suggested that low levels of IGF-1 may facilitate the activation of hepatic stellate cells in vivo.^32,33

Of note, hepatocyte proliferation between the four genotypes was unchanged. However, we observed a significant increase in hepatocyte apoptosis in Ghr^-/-;Mdr2^-/- mice (Fig. 4C and 4D) that may contribute to the development of liver fibrosis^34,35 in this animal model. Hepatocyte pro-survival factors, such as EGFR, HNF6, IGF1, p-ERK and p-S6K were down-regulated in livers of Ghr^-/- and Ghr^-/-;Mdr2^-/- mice, findings that are in line with recent demonstrations by us and others that GHR-STAT5 regulates the expression of EGFR, HNF6 and IGF1 in the liver ^9,11,22, underpinning the importance of GHR signaling in hepatocytes as a hepato-protective mechanism. Partial hepatectomy experiments have shown that Ghr^-/- and hepatocyte-specific EGFR knock-outs are impaired in liver regeneration.^8,36 Moreover, others confirmed that inhibition of EGFR and IGF1R in primary rat hepatocytes showed enhanced cell death when treated with bile acids.³⁷ Thus, deletion of Ghr renders hepatocytes more vulnerable to apoptosis-inducing stimuli.

We observed high levels of bile acids in Ghr^-/-;Mdr2^-/- mice. Furthermore, it has been proposed that bile acids initiate ROS formation in rat hepatocytes.¹⁵ Analysis of ROS levels in our mouse model indeed revealed an increase in oxidative stress in hepatocytes of Ghr^-/-;Mdr2^-/- mice. We could confirm up-regulation of genes which are expressed upon ROS production. Furthermore, expression analysis of enzymes involved in ROS defence in isolated hepatocytes versus non hepatocytes confirmed that hepatocytes of Ghr^-/-;Mdr2^-/- mice exhibit elevated oxidative stress (Fig. 6D). Importantly, Ghr^-/- mice fed with a bile acidenriched diet displayed increased liver injury, higher ROS levels and a higher number of apoptotic hepatocytes compared to Wt controls. Moreover, we could show that primary isolated hepatocytes from Ghr^-/- mice are more susceptible to bile acids and/or pro-apoptotic cytokine (TNFα and TGFβ)³⁸ treatment compared to Wt cells (Fig.5D, 5E and 5F). This suggests that hepatocytes lacking GHR signaling may undergo apoptosis in response to a combination of bile acid induced oxidative stress and pro-apoptotic cytokines such as TGFβ and TNFα Ghr^-/-;Mdr2^-/- mice show a reduced incidence of tumor development at 12 months of age accompanied with significant decrease of proliferating hepatocytes compared to Mdr2^-/- mice in non-tumorigenic tissue but exhibited similar hepatocyte proliferation in tumor tissue (Fig 7B and 7C). Conversely, hepatocytes in non-tumorigenic and tumorigenic tissue showed higher apoptosis in Ghr^-/-;Mdr2^-/- compared to Mdr2^-/- livers (Fig. 7C and 7D). Furthermore, we observed decreased EGFR and ERK phosphorylation and increased expression of key tumor suppressor genes (Trp53, Cdkn1a and Cdkn1b) in Ghr^-/-;Mdr2^-/- mice, which could possibly contribute to the reduction in tumorigenesis observed in these mice. Thus, Ghr^-/-;Mdr2^-/- may develop fewer tumors due to a loss of pro-survival signals, increased expression of tumor suppressor genes and increased levels of hepatocyte apoptosis.

In agreement with our data, mice bearing the “little” mutation (Ghrh^lit, a phenocopy of the Ghr ^-/- model) show reduced DEN induced-liver tumorigenesis³⁹ and importantly, a clinical study revealed no cases of tumor development in a population suffering from a Ghr mutation.⁴⁰

Collectively we could confirm an aggravated liver fibrosis phenotype, but decreased tumor development in Ghr^-/-;Mdr2^-/- relative to Mdr2^-/- mice signifying the importance of GH signaling in cholestasis-induced liver fibrosis and tumorigenesis. Our findings suggest that GH resistance plays a causal role in the onset and development of cholestasis-induced liver fibrosis and may provide a novel target for the development of new liver fibrosis treatments.

Abbreviations

GH

Growth hormone

GHR

Growth hormone receptor

MDR2

Multi-drug resistance 2

HNF6

Hepatocyte nuclear factor 6

EGFR

Epidermal growth factor receptor ALP - alkaline phosphatase

IGF1

Insulin like growth factor

ROS

Reactive oxygen species

HCC

Hepatocellular carcinoma

TGF-β

Transforming growth factor beta

PDGF

Platelet-derived growth factor

MCP-1

Monocyte chemoattractant protein-1

HSC

Hepatic stellate cells

IL-6

Interleukin 6

TNF-α

Tumor necrosis factor alpha

ECM

Extracellular matrix

JAK2

Janus kinase 2

STAT5

Signal transducer and activator of transcription 5

qRT- PCR

Quantitative reverse transcription polymerase chain reaction

ALS

Acid labile subunit

mRNA

Messenger RNA

IGFBP1

Insulin like growth factor binding protein 1

ALP

Alkaline phosphatase

AST

Aspartate amino transferase

ALT

Alanine aminotransferase

HE

Hematoxylin-eosin

CAB

Chromotrope aniline blue

PAS

Periodic acid-Schiff

α-SMA

Alpha smooth muscle actin

MMP

Matrix metalloproteinase

TIMP

Tissue inhibitor of metalloproteinases

GSEA

Gene set enrichment analysis

ERK

Extracellular-signal Regulated Kinase

DCA

Deoxycholic acid

HO-1

Heme oxygenase 1