Increased HO-1 levels ameliorate fatty liver development through a reduction of heme and recruitment of FGF21.

OBJECTIVE: Obese leptin deficient (ob/ob) mice are a model of adiposity that displays increased levels of fat, glucose, and liver lipids. Our hypothesis is that HO-1 overexpression ameliorates fatty liver development.
METHODS: Obese mice were administered cobalt protoporphyrin (CoPP) and stannic mesoporphyrin (SnMP) for 6 weeks. Heme, HO-1, HO activity, PGC1α, FGF21, glycogen content, and lipogenesis were assessed.
RESULTS: CoPP administration increased hepatic HO-1 protein levels and HO activity, decreased hepatic heme, body weight gain, glucose levels, and resulted in decreased steatosis. Increased levels of HO-1 produced a decrease in lipid droplet size, Fatty acid synthase (FAS) levels involving recruitment of FGF21, PPARα, and Glut 1. These beneficial effects were reversed by inhibition of HO activity.
CONCLUSION: Increased levels of HO-1 and HO activity reduced the levels of obesity by reducing hepatic heme and lipid accumulation. These changes were manifested by decreases in cellular heme, increases in FGF21, glycogen content, and fatty liver. The beneficial effect of HO-1 induction results from an increase in PPARα and FGF21 levels and a decrease in PGC1α, levels they were reversed by SnMP. Low levels of HO-1 and HO activity are responsible for fatty liver.

Introduction

Nonalcoholic fatty liver (NAFLD) occurs in a setting of high fat diets, insulin resistance, obesity and dyslipidemia (1, 2). NAFLD is a spectrum of diseases that range from fatty infiltration of the liver all the way to Nonalcoholic Steatohepatitis, NASH. Individuals with NAFLD have an increased risk of developing metabolic syndrome (3), the most severe form can progress to liver failure. If the liver contains fat levels that are in excess of 5 to 10%, fatty liver (hepatic steatosis) occurs. Low levels of antioxidants, excess heme, cytochrome P450-heme and mitochondrial dysfunction are factors that increase oxidative stress and ROS (reviewed in (4)). Increased HO-1 expression enables the cell to resist heme-mediated cell injury (reviewed in (4)). The well-documented protective role of HO-1 against the development of diabetes and the metabolic syndrome has been ascribed to several mechanisms that include a reduction in the levels of cellular heme-dependent proteins that increase oxidative stress and increased formation of the antioxidant, bilirubin (5). Excess heme-iron has also been implicated in liver damage (reviewed in (4)) and increased lipid accumulation generated in adipose tissues (6, 7). Heme iron-mediated oxidative stress magnifies the adverse effects of obesity by inducing inflammation in liver leading to NAFLD and NASH and systemic antioxidant treatment has yielded positive results in small studies (8). Probucol treatments improved NASH through the lowering of lipid levels and an increase in antioxidants level (9). However, broad systemic antioxidant treatment remains a “shotgun” approach. Targeting of HO-1 to adipose tissue and the vascular system (5, 10) has yielded positive effects on adipose dysfunction and insulin resistance (reviewed in (4)). HO-1 overexpression increased cellular antioxidant capabilities by increasing ferritin, CO and bilirubin (11, 12) leading to an increase in insulin sensitivity (12). HO-1 inducers, such as cobalt-protoporphyrin (CoPP), decrease heme levels (13) and increase phosphorylation of AKT in animal models of experimental diabetes (reviewed in (4)). Thus, increasing HO activity results in the reversal of oxidative stress and a decreases in liver damage (reviewed in (4)).

Several transcription factors regulate liver metabolism and food intake including. peroxisome proliferator activated receptors-gamma coactivator-1(PGC-1α) and fibroblast growth factor 21(FGF21)(2, 14). PGC-1α regulates FGE21, and cellular heme biosynthesis (15). FGF21 regulates hepatic carbohydrate and glycogen content and is PGC-1α expression dependent (16). Thus, cellular heme modulates the levels of PGC-1α, lipid metabolism and adipogenesis (14, 17). Increases in hepatic heme content inhibit PGC-1α leading to suppression of FGF21 (18). Glycogen levels in liver are maintained by PPARα and increased FGF21 (19). FGF21 increases the phosphorylation of AMPK and AKT resulting in an increase in insulin sensitivity and the lowering of blood glucose and fatty acid levels (20, 21).

The objective of this study was to examine the consequences of HO-1 induction and increased HO activity on the hepatic heme-HO system, FGF21, glycogen and hepatic fat content in obese mice. CoPP increased HO-1 expression and HO activity, attenuated the development of fatty liver, decreased lipid droplet size, PGC1α and increased FGF21 levels. Inhibition of HO activity increased lipid droplet size, FAS and decreased FGF21 and PPARα, thus substantiating a significant role of HO-1 (reviewed in (4)) against heme mediated adiposity and fatty liver. Pharmacological agents that increase HO-1 levels and gene targeting of HO-1 offer a promising therapeutic target for NAFLD and suggest the existence of a significant link between the heme-HO system and the extent and severity of heme-dependent fatty liver.

Methods and Procedures

Animal Treatment

All experiments were performed following a New York Medical College IACUC approved protocol in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Procedures for animal treatments have been previously published (12). In brief, male obese mice (B6v-Lep obese/J) were purchased from Harlan (Chicago, IL) at the age of 8 weeks. Age- and sex-matched lean mice (B6.V, lean; Harlan) were used as controls. Mice were fed a normal chow diet and had free access to water. Glucose monitoring was performed using an automated analyzer (Life scan, Milpitas, CA). Cobalt protoporphyrin (CoPP), an inducer of HO-1 expression, was administered intraperitoneally (i.p.) once a week (5 mg/kg) for 6 weeks to obese mice. Stannous mesoporphyrin (SnMP), an inhibitor of HO activity, was administered i.p. three times a week (20 mg/kg) for 6 weeks. Metalloporphyrins were dissolved in 10 mmol/l Tris base, and the pH was adjusted to 7.8 with 0.1 N HCl. A Tris/HCl solution free of metalloporphyrins was used to inject control animals. The animals were divided into four groups: 1) lean, 2) obese 3) obese CoPP and 4) obese CoPP+SnMP. Food intake did not change in mice treated with the various treatments. At the time of sacrifice the body weight of all mice was measured. After a 6-hour fast, mice were anesthetized with sodium pentobarbital (65 mg/kg, i.p.) and blood was obtained from a tail vein for glucose measurement using a glucometer. Blood samples were collected in K3EDTA tubes at sacrifice and the plasma was separated. Samples were flash frozen in liquid nitrogen and maintained at -80°C until needed. Other methodological details are provided in the online Data Supplement of Obesity.

Statistical analyses

Statistical significance between experimental groups was determined by ANOVA with Tukey-Kramer post-hoc analysis. The data are presented as means ± SEM and the null hypothesis was rejected at p<0.05.

Results

Effect of CoPP on Obese and Diabetic Phenotypes

A significant (p<0.05) decrease in the expression of HO-1 protein in livers of obese compared to lean mice was observed (Figure 1A). CoPP, and CoPP plus tin mesoporphyrin (SnMP), resulted in a significant (p<0.05) increase of HO-1 protein levels in obese mice when compared to obese mice administered vehicle. In addition, decreased HO-1 expression in liver of obese mice resulted in a significant increase in cellular heme (p<0.05) compared to the lean control (Figure 1B). Treatment with CoPP increased HO-1 protein and HO activity and reduced heme levels. Administration of SnMP to obese mice treated with CoPP reversed the effect of CoPP and increased hepatic heme content (Figure 1B). CoPP increased HO-1 protein levels in obese mice several fold compared to untreated obese mice. As expected obese mice treated with both CoPP and SnMP resulted in a further increase in HO-1 protein levels. CoPP increased HO activity to 1.35 ± 0.17 nmol bilirubin/mg/hr. vs. 0.392 ± 0.029 nmol bilirubin/mg/hr., in untreated obese mice (n=6, p<0.05). Administration of SnMP to CoPP-treated mice resulted in significantly lower hepatic HO activity, 0.032± 0.005 nmol bilirubin/mg/hr., (n=6, P< .05) when compared to CoPP-treated obese mice.

Figure 1

A) Western blot and densitometry analyses of liver HO-1 and actin; B) measurement of Heme levels in liver; C) body weight; and D) blood glucose in lean, obese control, obese treated with CoPP, and obese treated with CoPP and SnMP. Values represent means ± SEM of five independent treatments. *, P < 0.05 vs. lean or #, P < 0.05 vs. obese control.

Obese mice weight gain was significantly (p<0.05) more compared to lean control (Figure 1C). Treatment with CoPP prevented body weight gain in obese mice resulting in a level that was comparable to age matched lean mice. SnMP, administered to CoPP treated mice, prevented weight loss and resulted in an increase in body weight in obese mice. Compared to lean animals, blood glucose was significantly (p<0.05) increased in obese mice. However, obese mice treated with CoPP showed significantly (p<0.05) lower glucose levels compared to obese control (Figure 1D). SnMP ablated the beneficial effects of CoPP on glucose levels (Figure 1D).

Effect of HO Activity on Lipid Accumulation in Liver

As shown in Figure 2A, obese mice have significantly (p<0.05) more lipid accumulation and larger lipid droplets (Figure 2B) in liver compared to age matched lean animals. Oil red O staining of liver from lean and obese mice showed that CoPP decreased both lipid accumulation and droplet size (Figure 2B). The decrease in lipid accumulation and lipid droplet size in mice treated with CoPP was reversed by co-administration of SnMP, Figure 2A, and 2B. SnMP abrogated the beneficial effects of CoPP on both oil red staining and lipid droplet size. Furthermore, obese mice have significantly (p<0.05) higher FAS expression compared to lean (Figure 2C). CoPP-reduced FAS expression to below that of both obese and lean mice. SnMP administration resulted in significantly (p<0.05) higher FAS expression compared to lean controls (Figure 2C).

Figure 2

A) Oil Red O staining of lipids in liver and quantitative analysis of lean (1), obese control (2), obese treated with CoPP (3), and obese treated with CoPP and SnMP(4), magnifications: 40× (n=3). A representative section for each group is shown; B) lipid droplet size from Oil Red O stained livers; and C) Western blot and densitometry analyses of liver fatty acid synthase (FAS) and actin in lean, obese control, obese treated with CoPP, and obese treated with CoPP and SnMP. Values represent means ± SEM of five independent treatments. *, P < 0.05 vs. lean or #, P < 0.05 vs. obese control.

Effect of CoPP on glycogen content in liver

In comparison to lean animals, glycogen content was significantly (p<0.05) decreased in the liver of obese mice (Figure 3A). CoPP administered to obese mice, resulted in significantly (p<0.05) higher glycogen levels in liver. SnMP significantly (p<0.05) lowered glycogen content compared to both lean and obese control animals (Figure 3A). Glycogen synthase 2 (GS2) mRNA expression was suppressed in obese mice and increased by CoPP (Figure 3C). SnMP significantly (p<0.05) reduced GS2 mRNA expression to levels below both lean and obese animals treated with CoPP (Figure 3C). Obese mice had reduced phosphorylation of AKT in liver when compared to lean animals (Figure 3B). CoPP restored the phosphorylation of AKT to levels comparable to lean control while SnMP reversed the beneficial effects of CoPP on AKT phosphorylation.

Figure 3

A) Periodic acid Schiff staining of glycogen in liver of lean (1), obese control (2), obese treated with CoPP (3), and obese treated with CoPP and SnMP(4), magnifications: 40× (n=3). A representative section for each group is shown; B) Western blot and densitometry analyses of AKT phosphorylation (pAKT) and total AKT (AKT) (n=4); and C) Real-time PCR of glycogen synthase 2 (GS2) expression in lean, obese control, obese treated with CoPP, and obese treated with CoPP and SnMP. Values represent means ± SEM of five independent treatments. *, P < 0.05 vs. lean or #, P < 0.05 vs. obese control.

Effect of CoPP on PPARα expression

PPARα expression is decreased 54% (mean, 0.36 ± 0.04 n=5, p<0.05) in obese mice compared to lean control (Figure 4A). CoPP significantly (p<0.05) increased PPARα expression to levels that were comparable to lean control. In contrast SnMP significantly (p<0.05) decreased the perturbations in PPARα expression to levels lower than those of obese mice. In parallel to PPARα expression, phosphorylation of AMPK was significantly (p<0.05) decreased in obese mice as well as in mice treated with SnMP (Figure 4B). CoPP decreased heme levels, and restored AMPK phosphorylation in obese mice to levels comparable to those of lean control animals. The liver specific gene carnitine palmitoyltransferase 1A (CPT1A) was decreased 39% (0.58 ± 0.08 n=5, p<0.05) in obese mice when compared to age matched lean animals. Treatment of obese mice with CoPP resulted in a 487% increase in CPT1A expression (4.65 ± 0.76 n=5, p<0.05) over that of lean control. However, co-administration of SnMP to CoPP treated mice reversed the increase in the levels of CPT1A by 62.37% (0.35 ± 0.04 n=5, p<0.05) (Figure 4C).

Figure 4

Western blot and densitometry analyses of A) PPARα and actin (n=4); and B) AMPK phosphorylation (pAMPK) and total AMPK (AMPK) (n=4) in lean, obese control, obese treated with CoPP, and obese treated with CoPP and SnMP. C) Real-time PCR of CPT1A expression (n=4-5) in lean, obese control, obese treated with CoPP, and obese treated with CoPP and SnMP. Values represent means ± SEM of five independent treatments. *, P < 0.05 vs. lean or #, P < 0.05 vs. obese control.

Effect of CoPP on FGF21 expression in liver

We analyzed FGF21 expression in the liver of both lean and obese mice. In comparison to lean, obese mice displayed an increase of 340% (3.69 ± 0.70, p<0.05) in FGF21 mRNA expression (Figure 5A). CoPP increased FGF21 expression by 868% in obese (9.39 ± 0.61, p<0.05) when compared to lean animals and 255% greater than obese controls. SnMP prevented the CoPP-mediated increase in FGF21 expression resulting in levels similar to those of the obese control (Figure 5A). We examined the glucose lowering effect of increased HO activity to elucidate if it was the result of an increase in the levels of the epigenetic molecules of FGF21 signaling. Glut1 and PGC1α expression were decreased in obese mice, 53% (0.48 ± 0.06) and 39% (0.66 ± 0.08) respectively when compared to age-matched lean animals. CoPP increased Glut1 expression 570% (mean, 5.85 ± 0.99) and PGC1α 639.21% (mean, 6.94 ± 0.91) in obese mice when compared to lean controls. SnMP decreased Glut1 expression 82% (0.18 ± 0.02, p<0.05) compared to lean animals and 63% to obese controls; PGC1α expression was reduced to 86% (0.15 ± 0.01, p<0.05) when compared to lean controls and 77% of obese controls (Figures 5B and C).

Figure 5

A) Real-time PCR of FGF21; B) Glut1; and C) PGC1α expression in lean, obese control, obese treated with CoPP, and obese treated with CoPP and SnMP. Values represent means ± SEM of five independent treatments. *, P < 0.05 vs. lean or #, P < 0.05 vs. obese control.

Discussion

In this report, we demonstrate that increased HO-1 expression and HO activity lowers hepatic heme content, improves glycogen content and FGF21 expression and attenuates hepatic steatosis. In addition, the upregulation of HO-1 expression decreases both hepatic FAS levels and lipid droplet accumulation. Inhibition of HO-1 expression results in the opposite effect. These beneficial effects of HO-1 overexpression are reflected by a reduction the content of heme. PPARα and FGF21 tightly regulate hepatic lipid accumulation and glycogen content. HO activity was lower in untreated obese mice when compared with age matched controls. CoPP increased HO-1 protein and HO activity to levels significantly greater than those seen in age-matched vehicle treated lean animals. Several key findings in this report offer potential mechanisms by which increased HO-1 expression resulted in decreased levels of steatosis. Firstly, cellular heme content in obese mice is increased. Upregulation of HO-1 decreased heme content and increased FGF21 levels in obese mice. Others have shown that intracellular heme mimics the effect of PGC-1α on FGF21 recruitment, i.e. PGC-1α negatively regulates hepatic levels of FGF21 (2) and the heme modulating REV-ERb (alpha) axis. Heme, a ligand of the nuclear receptors REV-ERBα and B (14, 15), decreases FGF21 levels. It is noteworthy that heme levels increased during adipogenesis (14) while HO-1 protein levels and HO activity decreased (12) and that induction of HO-1 decreased adipogenesis, total fat and body weight gain in both obese mice and mice fed a high fat diet (5).

Another potential mechanism by which HO-1 upregulation decreases fat content may be related to increased levels of CO and bilirubin which contribute to vascular cytoprotection (10, 22). Thus an increase in heme degradation resulted in increased antioxidant levels with a resultant increase in PPARα and FGF21 expression. PPARα knockout (KO) mice have steatosis (21, 23), due to decreased FGF21 expression. FGF21 is an important mediator of glycogen storage in liver, most likely through the FGF21 directed phosphorylation of AKT. An increase in HO activity in obese animals resulted in a reduction in oxidative stress, inflammation and lipid accumulation in bone marrow (12). Increases in HO-1 expression and HO activity decreased the mitochondrial release of ROS via an increase in superoxide dismutase, catalase and the mitochondrial signaling pathway (13). A decrease in both human and rodent HO-1 levels caused severe oxidative stress due a concomitant increase in cellular heme iron levels ((24, 25), (reviewed in (4)). Heme, a potent pro-oxidant (26, 27), is critical for the synthesis of NADPH oxidase and ROS (28, 29). Thus, a significant relationship exists between HO-1 expression and steatosis in obese mice. The present data confirm that glucose deprivation increases HO-1 protein levels which are reversed by the addition of glucose (30, 31). The suppressive effect of high glucose on HO-1 levels despite increased levels of ROS is well documented. Elevated glucose levels decrease HO-1 expression (13, 32), analogous to what is observed in humans and rodents with obesity (33). A decrease in HO-1 levels appears to magnify the heme-mediated activation of NADPH oxidase and O-2 generation in diabetes and in obesity (reviewed in (4)). Thus, HO-1 levels control the levels of bilirubin which have potent anti-inflammatory properties both in vitro and in vivo ((34), (reviewed in (4)). In humans, elevated bilirubin levels have been reported in vascular dysfunction and in patients with Gilbert's syndrome (35). In contrast, a decrease in human HO-1 levels and bilirubin results in elevated heme levels and premature aging (24, 25).

Another potential mechanism by which CoPP-mediated upregulation of HO-1 and HO activity decreases hepatic lipids is through elevated levels of FGF21. A decrease in oxidative stress results in the sensitization of the FGF21 signaling cascade. Thus, FGF21 may be considered a therapeutic target for HO-1 in reducing NAFLD. In contrast, inhibition of HO activity decreased the expression of FGF21 target genes confirming previous reports suggest that obesity is a FGF21 resistant state (19). Thus increased heme levels, as a result of reduced HO-1 expression and HO activity, inhibits FGF21 signaling and reduces the attenuation of lipid accumulation. HO-1 has been implicated as a major regulator of several epigenetic signaling pathways by limiting the amount of heme and/or CO, which strongly binds to heme proteins CYP450, 4A10, 2E,1 prevents NAFLD fibrosis and decreases iNOS or COX-2 (36) resulting in the suppression of inflammatory cytokines (4). In contrast, the epigenetic signaling of FGF21 was increased by increased HO activity. Thus, activation of the HO-1-PPARα-FGF21 axis decreases hepatic lipid accumulation but increases hepatic glycogen content.

The HO-1/PPARα activity relationship is an important mediator of blood glucose levels. Increased HO activity improved insulin and glucose sensitivity in Zucker diabetic fat (ZDF) rats (6). These actions reflect an increased phosphorylation of AMPK and AKT, and suggest the existence of a symbiotic relationship between HO and AMPK/AKT. Induction of the HO-1-PPARα axis reduced blood glucose, through increased phosphorylation of AMPK. AMPK plays an important role in lipid metabolism and adipogenesis (37). Phosphorylation of AMPK is elevated in the presence of increased PPARα activity (38). In agreement with increased PPARα expression, CPT1A expression was reduced in obese mice and greatly enhanced by the induction of HO-1, indicative that increased levels of HO-1 expression and HO activity are paralleled by an increase in PPARα activity. The HO modulated increase in PPARα expression is well documented (39). In summary, administration of CoPP to obese mice increased HO-1 levels and HO activity and decreased cellular heme levels. The reduction in hepatic heme content by administration of CoPP resulted in the recruitment of FGF21, PPARα and the abrogation of lipid accumulation in liver of obese mice. Activation of the FGF21 signaling cascade resulted in enhanced glycogen content and AMPK-AKT phosphorylation, as well as, reduced body weight and blood glucose levels. These results clearly imply that the activation of FGF21 level is dependent on HO-1 induction and the reduction in hepatic heme levels (diagrammed in Figure 6). These studies suggest a new functional role for the HO-1 mediated recruitment of PPARα-FGF21. This potential role, as a signaling paradigm, may prove crucial in the control of obesity and the metabolic syndrome related NAFLD, predicted to become the leading cause for liver transplantation in the next decade. This is both an economic burden and a major impact on quality of life. As the epidemic of obesity continues to loom, this problem will only increase.

Figure 6

Schematic diagram of potential mechanisms underlying HO-1-mediated improvement of hepatic steatosis in NAFLD. Fatty liver is accompanied by decreases in HO-1 protein and HO activity, increased heme content and derangement of cell signaling including the increase in PGC1α expression and the decrease in FGF21 levels. Upregulation of HO-1 protein and HO activity by pharmacological agents leads to an increased in heme degradation and the generation of CO and bilirubin. CO and/or bilirubin enhance antioxidant mechanisms, thereby decreasing lipid droplets along an elevation of PPARα expression. Induction of PPARα leads to an increase in FGF21. Increase in HO activity enhances the phosphorylation of AKT and AMPK, resulting in increased Glut1 expression and the lowering of blood glucose levels. Thus, activation of the HO-1 protein and HO- activity along with an increase in PPARα-FGF21 module results in higher glycogen content in the liver and a decrease of liver steatosis.