The mice used in our experiments were genotyped, and it was confirmed that FVBN mice were Mdr2+/+, while Mdr2KO mice were Mdr2-/- (Suppl Fig. 1).
Time-course of serum Ghr in Mdr2KO mice as compared to FVBN control mice
Ghr peptide level was measured in serum of male and female Mdr2KO and FVBN mice from 2 weeks to 6 months old (Fig. 1). In FVBN controls, Ghr concentration increased gradually with age until four months, when it reached a plateau. In Mdr2KO mice, there was a significant increase in circulating Ghr in mice from two weeks up to one month, but no further enhancement was detected in later age groups. Interestingly, at each timepoint tested, the level of serum Ghr was lower in Mdr2KO mice as compared to FVBN mice.
Expression of Ghr and MBOAT in the stomach of Mdr2KO and FVBN control mice
We assessed Ghr expression in the stomach of Mdr2KO and FVBN mice, since Ghr is mainly produced in this part of the gastrointestinal system (Fig. 2A–C). Ghr mRNA level was significantly lower in cholestatic mice than in normal controls (Fig. 2A). At the translational level, Ghr peptide was also decreased in Mdr2KO mice as compared to FVBN controls (Fig. 2B,C). Similarly, the enzyme which activates Ghr by acylation, MBOAT, was found to be less abundant in Mdr2KO mice over FVBN controls (Fig. 2D–F). There were no significant gender-related differences in Ghr and MBOAT gastric expression.
Cellular distribution of GHS-R1a and GHr in the liver of Mdr2KO and FVBN control mice
To investigate the expression of GHS-R1a and Ghr in isolated cholangiocytes, hepatocytes and hepatic stellate cells (HSC) from livers of Mdr2KO mice vs FVBN controls, we used laser capture microdissection (LCM). For optimal precision, cholangiocytes, hepatocytes and HSC were labeled for two different markers, i.e. cytokeratin (CK) 19 and CK7 for cholangiocytes, CK8 and albumin (Alb) for hepatocytes, desmin and alpha smooth muscle (αSMA) for HSC (Fig. 3A,B, Suppl Fig. 2). While desmin is expressed in all HSC including quiescent, activated and inactivated cells, αSMA is expressed only in activated HSC. GHS-R1a mRNA was more abundant (ΔCt 3–4) as compared to Ghr (ΔCt 8–10), in all groups of mice. In Mdr2KO mice, GHS-R1a mRNA was significantly lower compared to FVBN controls in cholangiocytes, hepatocytes and HSC (Fig. 3A, Suppl Fig. 2A). Ghr mRNA was detected in all tested cell types at the same level in FVBN and Mdr2KO mice, with no gender-related differences (Suppl Fig. 3B). However, the results indicated that Ghr mRNA was significantly lower in αSMA-expressing HSC (Suppl Fig. 2B) than in desmin-expressing HSC (Fig. 3B), suggesting that HSC activation and change from fibroblast to myofibroblast phenotype is associated with a decrease in Ghr mRNA.
The expression of GHS-R1a and Ghr at protein level in different types of hepatic cells was assessed by confocal microscopy colocalization of GHS-R1a or Ghr with CK19 in cholangiocytes, CK8 in hepatocytes and desmin in HSC (Fig. 3C,D for male mice, and Suppl Fig. 3 for female mice). Ghr peptide (Fig. 3D) as well as GHS-R1a (Fig. 3C) were detected in the liver of FVBN mice, colocalizing with CK19 and CK8, and less so with desmin. The level of Ghr and its receptor in the liver of Mdr2KO mice was reduced as compared to FVBN controls. These results indicate that most of Ghr in the liver comes from the systemic Ghr and acts upon GHS-R1a receptors on cholangiocytes and hepatocytes.
Ghr treatment of Mdr2KO mice attenuates serum biomarkers of liver disease
The effects of exogenous Ghr and its des-acylated form, DG, on the serum liver enzymes, when administered to cholestatic vs control mice, were assessed. While DG had no effect on alanine aminotransferase (ALT) and aspartate aminotransferase (AST), Ghr induced a significant reduction in serum levels of these liver enzymes in Mdr2KO mice (Fig. 4A,B). Serum bilirubin and albumin were slightly changed in Mdr2KO mice compared to FVBN control mice treated with vehicle, and the treatment with DG or Ghr corrected these changes (Suppl. Table 1). The inflammatory chemokine CCL2 (C–C chemokine 2), also known as monocyte chemotactic/chemoattractant protein 1 (MCP1) was found to be more than eightfold-increased in serum of male and female Mdr2KO mice compared to FVBN controls (Fig. 4C). Administration of DG did not change the high plasma level of CCL2, however treatment with Ghr significantly reduced CCL2 in Mdr2KO mice by twofold (Fig. 4C). Serum tumor growth factor beta (TGFβ), the product of the most active profibrogenic gene in hepatic cholestasis, was also tested and found to be increased 3 to 4- fold in both male and female Mdr2KO mice compared to FVBN control counterparts (Fig. 4D). Treating the Mdr2KO mice with DG had no effect on TGFβ level, while Ghr reduced approximately by half the plasma TGFβ concentration (Fig. 4D). In summary, Ghr, but not DG, was effective in reducing the serum levels of ALT, AST, CCL2 and TGFβ which were remarkably increased in Mdr2KO mice as compared to FVBN controls.
Ghr decreases intrahepatic bile duct mass (IBDM) and cholangiocyte proliferation in Mdr2KO mice
Because no gender-related differences were found in serum levels of major biomarkers of liver disfunction in Mdr2KO mice when treated with vehicle, DG or Ghr, we used only male mice for the subsequent experiments.
The effects of DG and Ghr on the expression of CK19 in the liver of cholestatic vs control mice was assessed at mRNA and protein level (Fig. 5A–C). Ghr but not DG, caused a significant reduction in the excessively large IBDM of Mdr2KO mice. Proliferating cell nuclear antigen (PCNA) expression was assessed in cholangiocytes of Mdr2KO mice (Fig. 5D–F). Both the mRNA and protein of PCNA were drastically decreased in Mdr2KO mice after treatment with Ghr, but not with DG.
Hematoxylin and eosin (H&E) staining of liver sections from Mdr2KO mice and FVBN controls was used to assess the status of hepatocytes, as well as the size and frequency of biliary mass (Fig. 5G). It can be observed that there were no effects of DG or Ghr on the liver histology in FVBN controls, as expected (top images in Fig. 5G). However, vehicle treated Mdr2KO mice exhibited an increased number of bile ducts, surrounded by thick layers of small cells including string-like shaped HSC. The treatment of Mdr2KO mice with DG had only a small effect on the enlarged biliary mass, however Ghr was very efficient in reducing the size of IBDM. No significant damage was noted in regard to the shape of hepatocytes in Mdr2KO mice, neither steatosis or other histological changes were noted.
These data suggest that exogenously administered Ghr reduces proliferation of cholangiocytes and decreases the size of HSC layers around bile ducts in cholestatic mice.
Ghr treatment alleviates liver fibrosis in Mdr2KO mice
Genes known to be upregulated in cholestasis-induced hepatic fibrosis, including desmin and αSMA markers of HSC, as well as structural proteins of extracellular matrix (ECM) such as collagen types I and III, and integrins, were tested in Mdr2KO mice and FVBN controls treated with vehicle, DG or Ghr (Fig. 6). All the tested genes were downregulated in Ghr-treated Mdr2KO mice while being insignificantly affected by DG. Thus, the expression of both desmin and αSMA was strongly increased in Mdr2KO mice compared to FVBN controls (Fig. 6A–F). The treatment with DG did not change the excessive amount of HSC markers detected in Mdr2KO mice, while Ghr induced a significant decrease of them. Collagen types I and III were assessed using Sirius-Red staining of liver sections, and were found to be produced in excess in Mdr2KO mice treated with vehicle or DG but were reduced in Mdr2KO mice treated with Ghr (Fig. 6G–I). To confirm these results, we also assayed the liver hydroxyproline concentrations (Suppl Fig. 3), and determined that hydroxyproline was excessively expressed in the livers of Mdr2KO mice, and it was not affected by DG, however it was significantly decreased following Ghr treatment (Suppl Fig. 3). The β6 component of integrins was demonstrated to be more than sevenfold increased at mRNA level in the livers of Mdr2KO mice compared to FVBN controls, and it was strongly diminished in Mdr2KO mice treated with Ghr, while DG had only small alleviating effect (Fig. 6J). Moreover, IHC assessment of integrin αvβ6 indicated that the massive increase in this integrin expression within the thick layers of HSC around enlarged bile ducts in Mdr2KO mice was counteracted by treatment with Ghr, while DG had no effect (Fig. 6K–L).
The expression of additional genes with roles in liver fibrogenesis was tested at mRNA level in Mdr2KO mice treated with vehicle, DG or Ghr, and compared to FVBN controls subjected to similar treatments (Fig. 7). Thus, fibronectin (FN1), integrin component αv, matrix metalloproteinase-2 (MMP2), and tissue inhibitor of MMP-1 (TIMP1) were abnormally increased in Mdr2KO mice treated with vehicle. Ghr but not DG, was effective in reducing mRNA expression of these genes (Fig. 7A–D). Profibrogenic genes such as TGFβ, PDGFα (platelets derived growth factor alpha) and connective tissue growth factor (CTGF) which are major indicators of liver fibrosis, were assayed using qPCR and shown to be significantly increased in Mdr2KO mice compared to FVBN controls (Fig. 7E–G). Interestingly, DG, the less active form of Ghr, had a trend to decrease the expression of these genes, and it had a significant effect on CTGF. Ghr was more effective than DG, drastically lowering the expression of all tested growth factors. Finally, several proinflammatory genes including CCL2, interleukin (IL)-1β, IL-6 were assessed in livers of Mdr2KO mice vs FVBN controls (Fig. 7H–J). CCL2 was the most increased (more than 25-fold), followed by IL-6 (fourfold) and IL-1β (twofold) in Mdr2KO mice compared to FVBN controls. DG did not decrease CCL2 mRNA, but had a significant effect on IL-1β and IL-6, while Ghr reduced the expression of all tested cytokines (Fig. 7H–J).
In summary, major indicators of hepatic fibrosis in Mdr2KO mice including biomarkers of proliferating and activated HSC, ECM structural proteins and modulators, profibrogenic and proinflammatory genes were strongly reduced in mice treated with Ghr. The expressions of fewer genes, namely profibrogenic growth factors and cytokines were also diminished by DG.
Ghr reduces apoptosis and necrosis in Mdr2KO mouse livers
To investigate whether DG and Ghr influence apoptosis and necrosis in the Mdr2KO model of cholestasis-induced liver fibrosis, we measured products of these processes using a fluorescence microscopy procedure. Thus, we used a kit for specific staining of phosphatidylserine (PS), a hallmark of apoptosis, and of nuclei of necrosis-damaged cells, and demonstrated that both DG and Ghr significantly decreased apoptosis and necrosis in the liver of Mdr2KO mice (Fig. 8). By image analysis we determined that Ghr was more effective than DG in diminishing apoptosis and necrosis markers.
Ghr attenuates cholangiocyte proliferation via AMP-activated protein kinase (AMPK) and forkhead box protein O1 (FOXO1) activation pathway
Activation of AMPK in response to Ghr was tested by measuring phospho-AMPK (p-AMPK) in mouse cholangiocytes in vitro (Fig. 9A). The cells were treated with Ghr in the absence or presence of Ca2+ chelator BAPTA or AMPK inhibitor dorso morphine (DM), which are known to block AMPK activation. Phospho-AMPK was quantified at various timepoints up to 2 h using an ELISA kit. Ghrelin induced a significant increase in p-AMPK as early as 15 min after treatment and the effect of Ghr lasted up to several hours. The inhibitors of AMPK phosphorylation and activation, BAPTA and DM, counteracted the effect of Ghr up to two hours.
The AMPK-mediated activation of FOXO1 initiated by Ghr signaling in cholangiocytes was also tested (Fig. 9B). FOXO1 activation in cholangiocytes treated with Ghr only or Ghr plus BAPTA or Ghr plus DM was tested, demonstrating that Ghr induced a two-fold increase in binding of activated FOXO1 to specific DNA response elements. This activation of FOXO1 was not detected in the presence of BAPTA and DM (Fig. 8B), suggesting that FOXO1 activation is dependent on AMPK activation.
It is known that FOXO1 is located inside the nuclei when activated, binding promoters of target genes with roles in cell proliferation control36,37. We assessed the nuclear translocation induced by Ghr in cholangiocytes (Fig. 9C). In cells treated with vehicle, FOXO1 was mostly cytoplasmic, detected around nuclei, while after 15 min–2 h of Ghr treatment, FOXO1 was detected inside nuclei in many cells. The inhibitors of AMPK activation, BAPTA and DM, prevented Ghr-induced accumulation of FOXO1 inside nuclei, and even caused upregulation of FOXO1 in the cytoplasm. Some proliferating cells were detected one hour or longer after treatments with BAPTA or DM in addition to Ghr.
To further test the possibility that Ghr affects the proliferation rate of cholangiocytes via AMPK-FOXO1 signaling pathway, we assessed cell proliferation after treatment with Ghr alone or in addition to BAPTA or DM (Fig. 9D). Ghr suppressed cell proliferation while BAPTA and DM prevented this effect (Fig. 9D). To confirm these results, we also measured the incorporation of IDU into mouse cholangiocytes treated with vehicle, Ghr, Ghr plus BAPTA, and Ghr plus DM (Fig. 9E,F). The results suggested that Ghr reduced cell proliferation, however BAPTA and DM attenuated this effect.
Taken together, these data demonstrate that Ghr inhibits cholangiocyte proliferation via a mechanism involving Ca2+ and AMPK-mediated nuclear translocation of FOXO1.
Silencing of GHS-R1a with siRNA in mouse cholangiocytes reverses the effects of Ghr
In order to assess the role of GHS-R1a in signaling FOXO1 nuclear translocation and suppression of cell proliferation rate, we knocked down GHS-R1a mRNA, and confirmed the reduction of GHS-R1a expression (Fig. 10A). Mouse cholangiocytes transfected with GHS-R1a-specific siRNA, were then treated with vehicle or Ghr for 15 to 60 min, and AMPK phosphorylation was measured (Fig. 10B). Similarly, cells transfected with negative control siRNA were used as controls. P-AMPK induced by Ghr, was reduced in GHS-R-siRNA transfected cells compared to cells transfected with negative control siRNA, suggesting that Ghr’s effect is dependent of GHS-R1a. FOXO1 transactivation measurements indicated that GHS-R1a knockdown impaired Ghr-induced FOXO1 transactivation (Fig. 10C). These data were confirmed by the immunofluorescence colocalization of FOXO1 with nuclei in cholangiocytes transfected with GHS-R1a-siRNA vs negative control siRNA (Fig. 10D). Moreover, the assessment of Ghr-induced reduction of cell proliferation rate was performed by two procedures: one measuring the percent of live cells (Fig. 10E), and the other by measuring the incorporation of IDU into newly synthetized DNA during cell replications (Fig. 10F,G). Both procedures indicated that knockdown of GHS-R1a reduced the ability of Ghr to stimulate cholangiocyte proliferation.