Inhibition of hedgehog signaling ameliorates hepatic inflammation in mice with nonalcoholic fatty liver disease (NAFLD)

Abstract

Hedgehog (Hh)-signaling plays a critical role in liver development, regeneration, injury repair and carcinogenesis. Activation of Hh signaling has been observed in patients with nonalcoholic fatty liver diseases (NAFLD); however, the pathobiological function and regulatory mechanism of hepatic Hh signaling in the pathogenesis of NAFLD remain to be further defined. This study was designed to examine the effect and mechanism of hepatic Hh signaling in high fat diet (HFD)-induced NAFLD by using pharmacological Smoothened inhibitors (GDC-0449 and LED225) and by using liver specific Smo knockout (Smo LKO) mice. Administration of Smo inhibitors to HFD-fed wild type mice significantly reduced the numbers of activated macrophages and decreased the expression of pro-inflammatory cytokines (TNFα, IL-1β, MCP1 and IL-6) as assessed by F4/80 immunohistochemistry and qRT-PCR, respectively. The Smo inhibitors were noted to have variable effects on hepatic fat accumulation. We observed that liver specific deletion of Smo also reduced macrophage activation and inhibited pro-inflammatory cytokine expression, while it did not significantly alter fat accumulation in the liver. Mechanistically, we found that activation of Gli1 by Hh signaling in primary hepatocytes increased the production of osteopontin (OPN) which subsequently enhanced macrophage-mediated pro-inflammatory response via a paracrine signaling. Conclusions: Hepatocyte Hh signaling can promote liver inflammation through OPN-mediated macrophage activation; this mechanism importantly contributes to the progression of NAFLD.

INTRODUCTION

Non-alcoholic fatty liver disease (NAFLD) is the most common chronic liver disorder in Western countries(1). A subset of individuals with NAFLD develop nonalcoholic steatohepatitis(2), characterized by hepatocytes ballooning and/or lobular inflammation, which may lead to liver fibrosis that can progress to cirrhosis and the development of primary liver cancer(1, 3). The risk factors that contribute to the progression from simple steatosis to non-alcoholic steatohepatitis include chronic liver inflammation, lipotoxicity, insulin resistance and oxidative stress(4, 5). Among these factors, persistent hepatic inflammation (e.g., pro-inflammatory cytokines and recruitment of inflammatory cells) is considered a major factor to cause NASH progression which predicts poor clinical outcome in patients(6). Unfortunately, at the present time there is no effective pharmacological treatment which can stop or reverse NASH progression.

Hedgehog (Hh) signaling pathway plays an important role in regulation of hepatic progenitor cell fate and liver development(7, 8). In adults, the Hh pathway also modulates the regeneration and repair responses to liver injury(7–10). The components of canonical Hh signaling consist of Hh ligand family (Sonic, Indian and Desert), transmembrane protein receptor Pathced1 (Ptch1), G-protein coupled transmembrane receptor Smoothened and glioma-associated oncogene (Gli) transcription factors(11). In the absence of Hh ligand, Ptch receptor represses Smo and prevents the activation of Hh signaling. When Hh ligands are present, they bind to Ptch receptor and this allows Smo to activate Gli transcription factors and to regulate the expression of target genes (e.g., Ptch, Gli, Hhip)(12). Hh signaling is generally repressed in normal healthy liver, but it becomes activated in liver diseases including liver fibrosis, NASH, hepatocellular carcinoma and cholangiocarcinoma. Recent studies have shown that the production of Sonic Hh (Shh) ligand is increased in the livers of patients with NAFLD, and the levels of Shh correlate with the severity of NAFLD(2, 8). The Shh ligand produced by ballooned hepatocytes is known to stimulate hepatic stellate cells and fibroblasts via a paracrine mechanism, thereby promoting profibrogenic response in mouse model of NASH(13–17). Although it has been generally considered that hepatocytes are Hh non-responsive cells(7), recent studies reported increased expression of Smo and Gli1 in primary hepatocytes from damaged livers(16, 18). The latter observations are in support of the hypothesis that Hh signaling in hepatocytes could be activated during liver injury and this mechanism may contribute to hepatocyte regeneration and repair from liver injury. However, the potential implication of hepatocyte Hh signaling in NAFLD or any other liver diseases has not yet been established.

Osteopontin (OPN) is an extracellular matrix protein produced by various cell types including immune cells (macrophage and neutrophil) as well as smooth muscle, epithelial and endothelial cells(19). It plays both protective and pathologic roles in several liver diseases(20, 21). In particular, OPN expression is significantly upregulated in alcoholic steatohepatitis (ASH), NASH, liver fibrosis and hepatocellular carcinoma; and OPN neutralization effectively attenuated murine liver fibrosis through inhibition of progenitor cells response(22). It is noticeable that OPN is a Hh target gene and functions as a proximal mediator of Hh-mediated fibrogenesis(23, 24). Accordingly, enhanced expression of OPN has been observed in primary hepatic stellate cells from Ptch+/− mice (with overexpression of Hh signaling)(23).

In order to elucidate the role of hepatocyte Hh signaling in the pathogenesis of NAFLD, we crossed Smo^flox/flox mice with Alb-cre mice to generate mice with liver-specific deletion of Smoothened (Smo LKO). The produced Smo LKO mice and their littermate controls were subjected to high fat diet (HFD) protocol. For comparison, we utilized two Smo inhibitors (GDC-0449 and LED225) to assess their effect in HFD-induced NAFLD in WT mice. We observed that SMO inhibitor treatment or liver specific deletion of Smo both reduced macrophage activation and inhibited pro-inflammatory cytokine expression. Our data indicate that hepatocyte Hh signaling can promote liver inflammation through OPN-mediated macrophage activation and that this mechanism importantly contributes to the progression of NAFLD. Our findings support the concept that Hh inhibitors may have promising therapeutic potential for the treatment of NAFLD.

EXPERIMENTAL PROCEDURE

Animals, diet, and drug treatment

To generate liver (hepatocytes)-specific Smo knockout mice (Smo LKO), Alb-cre transgenic (obtained from Jackson Laboratory) mice were crossed with mice bearing a floxed Smo allele (Smo^flox/flox, obtained from Jackson Laboratory), producing Cre⁺; Smo^flox/+. These mice were subsequently crossed with Smo^flox/flox to generate mice that contain Alb-cre and Smo^flox/flox homozygous. We used Cre⁻; Smo^flox/+ or Cre⁻; Smo^flox/flox as control mice. The genotype of mice was determined by PCR. The Smo LKO mice and matched control mice at 8 weeks of age were fed either normal chow or high fat diet (45 kcal% fat, Research Diet #D12451) for 25 weeks. For Smo inhibitor studies, C57/BL6 mice at 8 weeks of age were fed either normal chow or high fat diet (45 kcal% fat, Research Diet #D12451). After 8 weeks of HFD, the mice were treated with vehicle, GDC-0449 (40 mg/kg, Selleckchem, Houston, TX) or LED225 (40 mg/kg, Selleckchem) every other day by intraperitoneal injection for three weeks.

All animal studies were conducted according to the protocol approved by the Tulane Institutional Animal Care and Use Committee (IACUC).

Histology, lipid staining and immunohistochemistry

Liver tissues were fixed in 10% formalin and embedded in paraffin. 4-μm paraffin-embedded sections were stained with hematoxylin and eosin (H&E) using a standard protocol. For lipid staining, OCT-embedded frozen sections were prepared by the Histology Core at the Tulane University Health Sciences Center and fixed in 4% paraformaldehyde for 15 min. After washing with distilled water, the sections were stained with Oil Red O (Sigma, St. Louis, MO) or BODIPY 493/503 (Life technologies, Grand Island, NY) for 1 hour at room temperature. The sections were washed with distilled water and mounted, and then visualized using a light microscope. For measurement of hepatic triglycerides, lipids were extracted from liver tissues and triglycerides were measured using a colorimetric assay (Infinity, Thermo Scientific). Immunohistochemistry was performed as previously described(25). The primary antibodies used for mouse liver sections were F4/80 (Abcam #ab111101, Cambridge, MA) and Osteopontin (Abcam #ab8448).

Isolation and transfection of mouse primary hepatocytes

Mouse primary hepatocytes were isolated by two-step collagenase perfusion as we previously described(26, 27). Briefly, the liver was initially perfused with a Ca²⁺ free buffer followed by perfusion with a buffer containing collagenase D (2 mg/ml) at 37°C. The liver was removed and gently agitated in serum-free William’s medium E (GIBCO). After filtering of digested liver tissue, this solution was centrifuged and cell pellet was resuspended in William’s medium E containing 10 % FBS, 2 mM L-glutamine and hepatocyte maintenance supplement pack (dexamethasone, penicillin-streptomycine, GlutaMax^™ and insulin) (GIBCO). The cells were cultured on collagen-coated plates and treated with the sonic hedgehog ligand (Shh, R&D systems) or the smoothened agonist (SAG, R&D systems) for 48 h. For overexpression of Gli1 in mouse primary hepatocytes, we used Targetfect hepatocytes reagent and Virofect enhancer (Targetingsystems, El Cajon, CA).

Macrophage culture and treatment

Primary Kupffer cells were isolated from mice (detailed methods described in Supporting Information). For treatment with hepatocyte conditioned medium, the conditioned medium was collected from hepatocyte cultures and filtered through a 0.22 μm sterile filter; the conditioned medium was then diluted 1:1 (vol/vol) with Dulbecco’s Modified Eagle Medium (DMEM) containing 5% FBS before addition to cultured Kupffer cells.

Raw264.7 cells (murine macrophage-like cell line) were obtained from ATCC and cultured in DMEM supplemented with 10% FBS and antibiotics at 37°C. For treatment with hepatocyte conditioned medium, the conditioned medium was diluted 1:4 (vol/vol) with DMEM containing 5% FBS prior to cultured Raw264.7 cells.

Treatment with recombinant osteopontin or anti-osteopontin antibody

Mouse recombinant osteopontin (441-OP-050) and neutralizing anti-osteopontin antibody (AF808) were purchased from R&D systems. Mouse primary Kupffer cells and Raw264.7 cells were cultured as described above and treated with mouse recombinant osteopontin (0.5 μg/ml) or vehicle (PBS) for 24 h. Mouse osteopontin neutralizing antibody (5 μg/ml) or control IgG were pre-incubated with hepatocyte conditioned medium for 2 h and then added to the cells for 24 h.

Quantification of osteopontin by ELISA

After collection of supernatants from cultured hepatocytes, secreted mouse osteopontin levels were measured by Mouse/Rat Osteopontin Quantikine ELISA kit (MOST00, R&D systems) according to manufacturer’s instruction. The optical density (O.D.) of samples was measured at 450 nm and corrected at 540 nm.

Statistical analysis

Results were presented as mean ± standard deviation (SD) or ± standard error of the mean (SEM) with a minimum of three replicates. Statistical significances between two groups were analyzed using two tailed Student’s t-test. A p value of less than 0.05 was considered significant.

RESULTS

Effects of the pharmacologic Smoothened inhibitors, LED225 and GDC-0449, on HFD-induced hepatic steatosis and glucose intolerance in mice

Since Hedgehog (Hh) signaling activation has been observed in both human and mouse with nonalcoholic fatty liver disease (NAFLD)(2, 8, 13), it is conceivable that pharmacological Hh inhibitors may have potential for NAFLD treatment. In this study, we assessed the potential therapeutic effect of two Smo inhibitors, GDC-0449 and LED225, in mice with NAFLD. In our protocol, C57/BL6 mice were fed high fat diet (HFD) for 8 weeks; the animals then received intraperitoneal injection of GDC-0449 (40 mg/kg), LED225 (40 mg/kg), or vehicle (DMSO) every other day for additional 3 weeks (with continued HFD feeding). We observed that GDC-0449 or LED225 treatment significantly inhibited hepatic expression of Ptch1 and Gli1 compared to vehicle-treatment (Fig. 1A). GDC-0449 and LED225 treatment of the HFD-fed mice slightly decreased the body weight and the liver/body weight ratio (Fig. 1B). GDC-0449 or LED225 treatment reduced steatosis to some extent as examined by histology, oil red o staining and hepatic triglyceride levels (LED225 appeared to be more effective in reduction of hepatic steatosis compared to GDC-0449) (Fig. 1C, 1D and 1E). Serum levels of triglycerides and cholesterols were also decreased by both GDC-0449 and LED225 treatment (Fig. 1E). To examine the effect of Smo inhibitors on lipogenesis, we performed qRT-PCR to determine the expression of lipogenic transcription factor (Sterol regulatory element binding protein 1; SREBP1) and its target gene (Stearoyl-CoA desaturase-1; SCD1). As shown in Fig 1F, GDC-0449 and LED225 treatment significantly reduced the expression of both SREBP1 and SCD1 in the liver. Furthermore, GDC-0449 and LED225 treatment slightly improved HFD-induced impairment of glucose tolerance at early time point (30 min). Accordingly, the level of hepatic phospho-Akt (Ser473) (a downstream molecule of insulin signaling) was increased in both GDC-0449 and LED225 treated HFD-fed mice (Supporting Fig. 1A and 1B).

Figure 1. Effects of the pharmacological Smo inhibitors, LED225 and GDC-0449, on HFD-induced hepatic steatosis.

WT mice were fed HFD for 8 weeks and then treated with vehicle (DMSO), GDC-0449 (40 mg/kg) or LED225 (40 mg/kg) for 3 weeks (n = 5–8 per group). (A) qRT-PCR analysis of hedgehog target genes Ptch1 and Gli1 in the liver tissues from different groups of mice. (B) Body weight and liver to body weight ratio of mice at the end of treatment. (C) Representative haematoxylin and eosin (H&E) staining; (D) Representative Oil Red O staining (magnification 20×); (E) Hepatic triglyceride levels (normalized to the protein concentration), serum triglyceride and cholesterol levels of HFD-fed WT, GDC-0449- and LED225 treated mice; (F) Relative mRNA expression of sterol regulatory element binding protein-1 (SREBP-1) and stearoyl-CoA desaturase (SCD) by qRT-PCR in the livers of HFD-fed WT, GDC-0449- and LED225 treated mice. The bar graphs represent mean ± SD (*P<0.05, ** P<0.01, ***P<0.001).

Given that a previous study by Hirsova and colleagues has shown a protective effect of GDC-0449 on HFD-induced liver injury(16), we further examined whether the Smo-inhibitor could inhibit liver cell death in our system. We observed that HFD-fed WT mice exhibited increased numbers of TUNEL-positive cells in the liver compared to NCD-fed mice (approximately 2-fold increase); this effect was prevented by treatment with either GDC-0449 or LED225 (Supporting Fig. 2A). Moreover, the NAFLD activity score of Smo inhibitor-treated mice was lower compared to vehicle-treated mice (Supporting Fig. 2B). However, we observed that GDC-0449 or LED225 treatment did not significantly alter the levels of serum AST and ALT (Supporting Fig. 3). We noticed that the latter aspect is different from the study by Hirsova and colleagues who showed that GDC-0449 treatment decreased serum ALT in mice fed the FFC diet (high fat, fructose and cholesterol) for 12 weeks(16). It is possible that the levels of serum transaminases may be influenced by factors including the contents of the diets or the treatment durations of specific inhibitors.

Liver specific Smoothened knockout does not improve HFD-induced hepatic steatosis and glucose intolerance

Although Hh signaling activation is predominantly observed in hepatic stellate cells and fibroblasts in NAFLD, increased expression of Hh signaling components was also detected in damaged hepatocytes(16, 18). To investigate the role of hepatocyte Hh activation in NAFLD, we generated liver (hepatocyte)-specific Smoothened knockout mice (Smo LKO) using Cre/loxP system (Alb-Cre/Smo floxed). Quantitative RT-PCR (qRT-PCR) and Western blot analysis confirmed down-regulation of Smo expression in the livers of Smo LKO mice (Fig. 2A). We observed that the expression of Smo was significantly down-regulated in hepatocytes isolated from the Smo LKO mice (approximately 10-fold reduction) compared to the hepatocytes from the wild type (WT) mice (Supporting Fig. 4A). In contrast, the expression of Smo was not significantly altered in other liver cells (hepatic stellate cells and cholangiocytes) isolated from the Smo LKO and WT mice (Supporting Fig. 4B–4E). These observations confirmed hepatocyte-specific deletion of Smo in the Alb-Cre/Smo floxed mice.

Figure 2. Liver specific knockout of Smo does not suppress HFD-induced hepatic steatosis.

Wild type mice (WT) and liver specific Smoothened knockout mice (Smo LKO) were fed 45 % HFD for 25 weeks (n = 5 per group). (A) qRT-PCR and western blotting of Smo in the livers of WT and Smo LKO mice. (B) Relative mRNA expression of Gli1, Ptch1 and Smo by qRT-PCR in the livers of NCD or HFD-fed WT and Smo LKO mice. (C) Body weight and liver to body weight ratio of WT and Smo LKO mice after NCD or HFD feeding. (D) Representative H&E staining (magnification 20×) (E) Representative BODIPY (magnification 20×) and Oil Red O staining (magnification 20×) in liver tissues, and hepatic triglyceride levels of NCD or HFD-fed WT and Smo LKO mice. (F) Serum AST and ALT levels in NCD or HFD-fed WT and Smo LKO mice. The bar graphs represent mean ± SD (*P<0.05, ** P<0.01, ***P<0.001).

We then subjected the Smo LKO and WT mice to the high-fat diet (HFD) protocol. When WT (Cre-) mice were fed HFD for 25 weeks, the expression levels of Hh receptor and target genes (Gli1, Ptch1 and Smo) in the liver were significantly increased compared to normal chow diet (NCD) feeding. In contrast, Smo LKO mice fed HFD did not show increased expression of Gli1, Ptch1 and Smo (Fig. 2B). These findings demonstrate that HFD induces Hh activation in the liver and that hepatocytes represent a noticeable population of cells that exhibit Hh activation in the setting of NAFLD.

We next examined whether deletion of Smo could protect mice from HFD-induced fatty liver disease. After 25 weeks of HFD feeding, both WT and Smo LKO mice displayed increased body weight and developed hepatic steatosis as determined by histology, lipid staining and hepatic triglyceride analysis (Fig. 2C, 2D and 2E). However, there were no significant differences in hepatic lipid accumulation, serum levels of liver aminotransferases (AST and ALT) and NAFLD activity score between WT and Smo LKO mice (under either NCD or HFD) (Fig. 2E and 2F, Supporting Fig. 5A). Consistent with the in vivo observations, there was no significant difference in lipid accumulation between WT and Smo KO hepatocytes after fatty acid treatment (Supporting Fig. 6). Since NAFLD is associated with abnormal glucose tolerance(28), we performed intraperitoneal glucose tolerance test for WT and Smo LKO mice after 20 weeks of HFD feeding. There was no significant difference in glucose tolerance and hepatic phosphorylation of Akt (Ser473) between HFD-fed WT and Smo LKO mice (Supporting Fig. 1C and 1D). As a parallel approach, we examined the response of WT and Smo LKO hepatocytes to insulin in vitro. We observed that the hepatocytes from WT and Smo LKO mice showed a comparable response to insulin, as reflected by the phosphorylation of Akt and Irs1 (insulin receptor substrate-1) (Supporting Fig. 1E). Thus, deletion of Smo in hepatocytes does not appear to significantly improve HFD-induced hepatic steatosis or glucose intolerance, although it partially prevented liver cell injury (Supporting Fig. 5B) and reduced liver inflammation (see below).

Inhibition of hepatic hedgehog signaling reduces HFD-induced macrophage activation and pro-inflammatory cytokine expression

Excessive lipid accumulation in liver causes hepatic inflammation characterized by Kupffer cell activation and production of pro-inflammatory cytokines, which contributes to the development of nonalcoholic steatohepatitis(29). Given the importance of hepatic inflammation in NASH progression, we examined the effect of Hh inhibition on liver inflammation. We observed that the number of F4/80 positive cells, which is commonly used as a murine macrophage marker, was lower in GDC-0449 (53% reduction) and LED225 (63% reduction) treated HFD-fed mice compared to vehicle-treated HFD-fed mice, as assessed by immunohistochemical staining (Fig. 3A). Moreover, hepatic mRNA levels of pro-inflammatory cytokines including TNFα, IL-1β, MCP1 and IL-6 were significantly decreased by GDC-0449 or LED225 treatment (Fig. 3B). Similar results were also observed in the livers of HFD-fed Smo LKO mice. The F4/80 positive Kupffer cells were reduced by approximately 56% in the livers of HFD-fed Smo LKO mice compared to the livers of HFD-fed WT mice (Fig. 3C). The expression levels of TNFα, IL-1β and MCP1 were also significantly lower in the livers of HFD-fed Smo LKO mice compared to the livers of HFD-fed WT mice (Fig. 3D). These results show that treatment with Smo inhibitors or deletion of hepatocyte Smo can suppress HFD-induced hepatic inflammation by inhibiting macrophage activation and pro-inflammatory cytokine expression.

Figure 3. Pharmacological inhibition and liver-specific deletion of Smo reduce hepatic macrophage activation and pro-inflammatory cytokine gene expression in mice with HFD-induced NAFLD.

(A) Immunohistochemistry of F4/80 and quantification of F4/80 positive cells (magnification 20×); (B) Relative mRNA expression of pro-inflammatory cytokines (TNFα, IL-1β, MCP-1 and IL-6) by qRT-PCR in liver tissues from WT mice treated with vehicle, GDC-0449 (40 mg/kg) or LED225 (40 mg/kg) after 8 weeks of HFD feeding (n = 5–8 per group). (C) Immunohistochemistry of F4/80 and quantification of F4/80 positive cells (magnification 20×); (D) Relative mRNA expression of TNFα, IL-1β, MCP-1 and IL-6 by qRT-PCR in the liver tissues from WT and Smo LKO mice fed NCD or HFD (n = 5 per group). The bar graphs represent mean ± SD (*P<0.05, ** P<0.01, ***P<0.001).

Hepatic lipotoxicity caused by excessive lipid accumulation is known to induce secretion of pro-inflammatory cytokines (such as TNFα) from hepatocytes(30, 31). Given that treatment with Smo inhibitors or deletion of hepatocyte Smo decreased the expression of pro-inflammatory cytokines in the livers of HFD-fed mice, we further examined whether Smo inhibition or depletion might reduce pro-inflammatory cytokine expression in cultured hepatocytes. While down-regulation of Smo expression was confirmed in Smo KO hepatocytes, treatment of Smo-depleted or wild type hepatocytes with Smo agonist (SAG) in vitro did not alter the levels of TNFα, IL-1β, MCP1 and IL-6 (Supporting Fig. 7A). These results suggest that Smo inhibition or depletion does not alter hepatocyte expression of pro-inflammatory cytokines.

Osteopontin production is decreased in livers of HFD-fed Smo LKO and Smo inhibitor-treated mice

Given that osteopontin (OPN) is a key downstream effector of Hh signaling which is overexpressed in NASH livers(23, 32), we next examined the expression of OPN in the livers of WT mice treated with Smo inhibitors (GDC-0449 or LED225) and in the livers of Smo LKO mice with or without HFD feeding. As expected, OPN mRNA expression was induced in HFD-fed WT mice compared to normal chow diet (NCD)-fed mice; treatment of GDC-0449 or LED225 blocked HFD-induced hepatic OPN expression, as determined by qRT-PCR and immunohistochemical staining, respectively (Fig. 4A and 4B). Decreased OPN mRNA and protein expression was also observed in the livers of HFD-fed Smo LKO mice compared to HFD-fed WT mice (Fig. 4C and 4D). Under immunohistochemical analysis, we observed that HFD feeding induced the expression of OPN in hepatocytes and this effect was attenuated by Smo inhibitor treatment or by Smo deletion (Fig. 4B and 4D). While OPN is expressed in biliary epithelial cells, we observed that HFD feeding did not significantly alter OPN expression in these cells (Supporting Fig. 8).

Figure 4. Inhibition or deletion of Smo prevents HFD-induced increase of osteopontin in the liver.

(A) Relative mRNA expression of osteopontin (OPN) by qRT-PCR; (B) Immunohistochemistry of OPN (magnification 60×) in liver tissues from vehicle, GDC-0449 and LED225-treated mice after 8 weeks of NCD or HFD feeding. (C) Relative mRNA expression of OPN by qRT-PCR; (D) Immunohistochemistry of OPN (magnification 60×) in liver tissues from WT and Smo LKO mice fed NCD or HFD. Liver tissues from HFD-fed WT and Smo LKO mice were used for negative controls (lower panels). The bar graphs represent mean ± SD (** P<0.01, ***P<0.001).

Hh signaling regulates osteopontin through Gli1-dependent mechanism in hepatocytes

Although Hh signaling is known to regulate OPN expression in hepatic stellate cells(23), it remains unknown whether Hh signaling is able to modulate OPN production in hepatocytes. To address this issue, we measured osteopontin secretion in isolated hepatocytes from WT and Smo LKO mice. After 48 h hepatocytes culture, we collected cell culture medium from WT and Smo LKO mice and then measured OPN concentration by ELISA. The levels of secreted OPN in the conditioned medium from Smo KO hepatocytes were significantly lower compared to the CM from WT hepatocytes (Fig. 5A). Since OPN promoter contains a putative Gli1-binding site(33), we assessed whether OPN is regulated through a Gli-dependent mechanism in hepatocytes. Specifically, isolated WT hepatocytes were transfected with Gli1-expressing vector for 48 h; the efficiency of Gli1 overexpression was confirmed by qRT-PCR analysis. We observed that Gli1 overexpression in hepatocytes resulted in significant elevation of OPN mRNA and protein levels (Fig. 5B).

Figure 5. Hh signaling regulates osteopontin through Gli1-dependent mechanism in hepatocytes.

(A) Conditioned medium from WT and Smo KO hepatocytes was collected and analyzed to determine the concentration of osteopontin (OPN) by ELISA. (B) Western blot and qRT-PCR analysis of OPN in hepatocytes transfected with Gli1 or control vector for 48 h. (C, D) Relative OPN mRNA and protein expression in WT and Smo KO hepatocytes treated with vehicle, 0.5 μM SAG or 300 ng/ml Shh ligand, as analyzed by qRT-PCR and Western blotting, respectively. (E) WT hepatocytes were treated with vehicle, 0.5 μM SAG or 0.5 μM SAG plus 5 μM GANT61 for 48 h and the expression of OPN was determined by western blotting. The data are expressed as mean ± SD (***P<0.001).

To further determine the role of Hh activation in OPN expression, we treated WT and Smo KO hepatocytes with Smo agonist (SAG) or Sonic Hh ligand (Shh), in vitro. Quantitative RT-PCR and western blot analysis revealed that OPN expression was induced by treatment of SAG or Shh in WT hepatocytes, but not in Smo KO hepatocytes (Fig. 5C and 5D). We observed that the SAG-induced OPN expression was attenuated by the Gli inhibitor, GANT61 (Fig. 5E). These data suggest that activation of Hh signaling in hepatocytes induces OPN expression at least in part through a Gli-dependent mechanism. The latter assertion is further corroborated by the observation that activation of Smo by SAG enhances Gli1 mRNA expression in hepatocytes (Supporting Fig. 7B).

Paracrine osteopontin (OPN) stimulates hepatic macrophage proinflammatory phenotype

Given that Hh-signaling induces osteopontin (OPN) expression in hepatocytes as shown in the above section, we reasoned that Hh induced OPN secretion from hepatocytes might activate hepatic macrophage pro-inflammatory response. To address this issue, we sought to examine the effect of OPN on hepatic macrophage activation. We isolated primary Kupffer cells from wild type mice and treated the Kupffer cell cultures with recombinant OPN to determine their expression of proinflammatory cytokines (the isolated primary Kupffer cells were characterized by immunofluorescence and qRT-PCR for F4/80 as shown in Fig. 6A). We observed that recombinant OPN significantly increased the expression of proinflammatory cytokines, including TNFα, IL-1β, MCP1 and IL-6, in cultured primary Kupffer cells (Fig. 6B). Consistent with the effect of OPN in Kupffer cells, we observed that recombinant OPN also increased pro-inflammatory cytokine expression in Raw264.7 cells (murine macrophage-like cell line) (Fig. 6C).

Figure 6. Recombinant osteopontin stimulates pro-inflammatory cytokine gene expression in macrophage cells.

(A) Characterization of primary Kupffer cells isolated from WT mice by immunofluorescence staining and qRT-PCR analysis for F4/80. (left panel) Representative image of mouse Kupffer cells visualized by phase-contrast (magnification 40×). (mid panel) Expression of Kupffer cell marker F4/80 in cultured Kupffer cells was visualized by fluorescence microscopy (magnification 40×). The purity of the isolated Kupffer cells was routinely more than 95% (right panel). The qRT-PCR analysis for F4/80 in cultured Kupffer cells (compared to mouse primary hepatocytes). (B) Relative mRNA expression of TNFα, IL-1β, MCP-1 and IL-6 in primary Kupffer cells as determined by qRT-PCR. The Kupffer cells cultured for one day after isolation were treated with vehicle or 0.5 μg/ml recombinant osteopontin (rOPN) for 24 h. (C) Relative mRNA expression of TNFα, IL-1β, MCP-1 and IL-6 in Raw264.7 cells treated with vehicle or 0.5 μg/ml rOPN for 24 h. All values were normalized to 18S rRNA. The bar graphs represent mean ± SD (*P<0.05, **P<0.01, ***P<0.001).

To further determine the impact of Hh-regulated, hepatocyte derived OPN on macrophage-mediated pro-inflammatory response, we assessed the effect of hepatocyte conditioned medium on macrophages. Specifically, WT and Smo KO primary hepatocytes were treated with vehicle or SAG for 48 h; the CM was collected and incubated with Kupffer cells or Raw264.7 cells. We observed that the CM from SAG-treated WT hepatocytes increased the expression of pro-inflammatory cytokines in both Kupffer cells and Raw264.7 cells (Fig. 7A) (more prominent cytokine induction was seen in Raw264.7 cells than in primary Kupffer cells). On the contrary, the CM from SAG-treated Smo LKO hepatocytes exhibited no significant effect (Fig. 7A). We noticed that addition of OPN-neutralizing antibody to SAG-treated WT hepatocyte CM inhibited the expression of pro-inflammatory cytokines in both Kupffer cells and Raw264.7 cells (Fig. 7B). On the other hand, addition of SAG directly to Raw264.7 cells failed to induce pro-inflammatory cytokine expression (Supporting Fig. 9A). While both recombinant OPN and SAG-treated WT hepatocyte CM were able to increase Raw264.7 cell migration, the CM from SAG-treated Smo LKO hepatocytes had no effect (Supporting Fig. 9B and 9C). Taken together, our findings suggest a novel paracrine mechanism in which hepatocyte Hh signaling induces the production of OPN which subsequently activates macrophage pro-inflammatory response.

Figure 7. Effects of hepatocyte conditioned media on hepatic macrophage activation.

(A) The conditioned medium from vehicle or SAG-treated WT and Smo KO hepatocytes was added onto cultured primary Kupffer cells or Raw 264.7 cells. The mRNA levels of TNFα, IL-1β, MCP-1 and IL-6 were determined by qRT-PCR analysis (48 h after incubation with the conditioned medium). (B) Primary Kupffer cells or Raw264.7 cells were cultured with SAG-treated WT hepatocyte CM in the absence or presence of 10 μg/ml of osteopontin neutralizing antibody (anti-OPN) for 48 h; the mRNA levels of TNFα, IL-1β, MCP-1 and IL-6 were determined by qRT-PCR. The bar graphs represent mean ± SD (*P<0.05, **P<0.01, ***P<0.001). (C) Schematic diagram illustrating the mechanisms by which hepatic Hh signaling promotes macrophage-mediated inflammation in HFD-induced NAFLD.

DISCUSSION

This study describes a novel mechanism for hepatic Hh signaling in high fat diet (HFD)-induced NAFLD. We show that Smo inhibitors (GDC-0449 and LED225) significantly reduced the activation of macrophages and their expression of pro-inflammatory cytokines (TNFα, IL-1β, MCP1 and IL-6). Deletion of Smo in hepatocytes was also found to inhibit macrophage activation/cytokine production. Our data indicate that hepatocyte Hh signaling can promote liver inflammation through OPN-mediated macrophage activation and that this mechanism importantly contributes to the progression of NAFLD. Our results disclose a novel paracrine mechanism wherein Hh signaling in hepatocytes increases the production and release of osteopontin (OPN) which subsequently enhances liver macrophage-mediated pro-inflammatory response. Such an Hh-OPN paracrine mechanism appears to be critical in the progression of NAFLD (illustrated in Fig. 7C). Our data support the concept that Hh inhibition may represent a promising therapeutic approach for the treatment of NAFLD.

Both canonical (Gli-dependent) and non-canonical (Gli-independent) Hh signaling mechanisms are known to operate in different types of liver cells including hepatic stellate cells/fibroblasts, liver progenitor cells and cholangiocytes(10, 34). However, hepatocytes have been considered as non Hh-responsive cells in part due to their negative staining for Gli2 in patients with NASH(2). In the current study, by employing complementary in vivo and in vitro approaches, we show that hepatocytes actually can become Hh-responsive due to upregulation of Smo in the setting of HFD-induced NAFLD. In our studies in vivo, we observed that the Hh receptor and target genes (Gli1 and Ptch1) were significantly upregulated in livers of HFD-fed WT mice, while these phenomena were not observed in mice with hepatocyte deletion of Smo. In our studies in vitro, we observed that cultured primary hepatocytes responded to both Shh ligand and Smo agonist. Together, our findings suggest that hepatocyte Hh signaling can become active in the setting of liver injury (such as high-fat diet) and contribute to the pathogenesis of NAFLD.

We observed that the Smo inhibitors (GDC-0449 and LED225) decreased cytokine expression more potently than Albumin-Cre mediated deletion of Smo, while both approaches exhibited comparable inhibition of Kupffer cell accumulation. We reason that this may relate to the actions of Smo inhibitors on other cells types in the liver, given that HFD is known to influence other hepatic cell populations which may contribute to cytokine production in the liver.

A recent study shows that GDC-0449 treatment prevents liver injury and fibrosis, but not steatosis, in a diet-induced mouse model of NAFLD(16). This observation was confirmed in our present study. We found that GDC-0449 treatment did not significantly prevent hepatic lipid accumulation in HFD-fed WT mice. In contrast, HFD-fed WT mice treated with LED225 showed significant reduction of hepatic lipid accumulation compared to vehicle-treated mice. Thus, GDC-0449 and LED225 appear to have different effects on hepatocyte lipid accumulation. The exact mechanism for their different effects remains unclear, although possible explanations may include the difference in drug potency, target selectivity or compound stability. Interestingly, both GDC-0449 and LED225 improved glucose intolerance and enhanced insulin signaling (as reflected by increased Akt phosphorylation) in vivo; however, these effects were not observed in the Smo LKO mice. Furthermore, the hepatocytes from WT and Smo LKO mice showed a comparable response to insulin in vitro. These results suggest that Hh signaling in other cell types (non-hepatocytes) may be implicated in the regulation of glucose tolerance and insulin sensitivity in HFD-fed mice. In this context, it is notable that increased Hh signaling has been linked to impaired pancreatic β-cell function/insulin secretion and that this mechanism may contribute to glucose intolerance(35). It is conceivable that this mechanism may provide a possible explanation for improved insulin sensitivity by the Smo-inhibitors. Further studies are needed to investigate the therapeutic effect and mechanism of pharmacological Hh inhibition in diet-induced insulin resistance.

Hepatic steatosis is closely associated with chronic inflammation(36). For example, the chronic inflammation in consequence of lipid peroxidation and oxidative stress is known to accelerate the progression of hepatic steatosis to NASH(36). Thus, it is of critical importance to further understand the cellular and molecular signaling cascades that regulate pro-inflammatory cytokines, chemokines and activation of immune cells during NAFLD progression. In the current study, we observed that Smo inhibitors are able to protect against HFD-induced steatosis, while liver (hepatocyte)-specific deletion of Smo did not reduce steatosis. We observed that both Smo-inhibitor-treated mice and Smo LKO mice showed reduction of cytokine expression, although Smo inhibitors decreased cytokines more potently than Smo deletion. Thus, while hepatocyte deletion of Smo is able to inhibit cytokine production, the magnitude of cytokine inhibition may not be sufficient to reduce HFD-induced steatosis. Alternatively, given that activated Hh signaling is observed in non-parenchymal cells of NASH liver, it is possible that decreased steatosis by Smo inhibitors may require the contribution of non-parenchymal cells. Our findings in the current study suggest that Hh signaling activation in hepatocytes may contribute to NASH progression through regulating hepatic inflammation.

Among the multiple inflammatory regulators, osteopontin (OPN) produced by immune cells, hepatic stellate cells (HSCs) or hepatocytes is known to function as a pro-inflammatory mediator in liver fibrosis and NASH(37–40). In patients with NAFLD, hepatic OPN and its receptor (CD44) expressions are highly correlated with the severity of the disease(32); similarly, the level of Hh signaling activation is also correlated with the grade of NAFLD(2, 8). Recent studies have shown that OPN expression is regulated by Hh signaling since the promoter region of the OPN gene has a putative Gli-binding site(33). Syn et al. have reported that activation of Hh signaling in HSCs increases OPN expression via Gli2 transcription factor in methionine choline deficient (MCD) diet-induced mouse model of NASH(23). Consistent with these reported findings, our data in the current study also showed increased OPN in the livers of HFD-fed WT mice. We observed that HFD-induced OPN expression in the liver was significantly reduced by treatment of the animals with Smo-inhibitors. These findings further support an important role of Hh signaling for OPN activation in HFD-induced liver injury. Moreover, we observed that HFD-induced OPN expression in the liver was significantly reduced by deletion of Smo from hepatocytes; the latter observation supports the role of hepatocyte Hh signaling for OPN expression in the liver. Accordingly, we observed that the secretion of OPN was lower in Smo KO hepatocytes compared to WT hepatocytes. Conversely, we showed that activation of Hh signaling in hepatocytes by Smo agonist or Gli1 overexpression significantly enhanced OPN expression. All of these findings support the notion that hepatocyte Hh signaling enhances OPN production in HFD-induced NAFLD.

OPN regulates hepatic inflammation through modulation of several signaling pathways in cells including hepatocytes, HSCs and natural killer cells(41). In the current study, we examined a paracrine stimulation of macrophages by OPN secreted from hepatocytes. Treatment of cultured macrophages (primary Kupffer cells and Raw264.7 cells) with recombinant OPN or conditioned medium from SAG-treated WT hepatocytes significantly increased macrophage activation and enhanced their expression of pro-inflammatory cytokines. These findings indicate that OPN is an upstream regulator of macrophage-mediated pro-inflammatory response. The underlying mechanism of how the paracrine OPN regulates macrophage activation and pro-inflammatory cytokine expression in NAFLD remains to be further defined.

In summary, the current study provides novel evidence that hepatic Hh signaling regulates macrophage-mediated inflammation through regulation of OPN and that inhibiting Smo attenuates hepatic inflammation in mice with NAFLD. Our study presents important mechanistic insight into the role of hepatocyte Hh signaling in HFD-induced liver injury. Given the noticeable amelioration of liver inflammation by two different Smo inhibitors (GDC-0449 and LED225), it is encouraging that the Smo inhibitors may represent important therapeutic agents for future treatment of NAFLD.

Abbreviations

NAFLD

nonalcoholic fatty liver diseases

NASH

nonalcoholic steatohepatitis

Hh

hedgehog

HFD

high fat diet

Smo LKO

smoothened liver specific knockout

Gli

glioma-associated oncogene

TNFα

tumor necrosis factor alpha

IL-1β

interleukin 1 beta

MCP1

monocyte chemoattractant protein 1

IL-6

interleukin 6

OPN

osteopontin