Prohibitin-induced, obesity-associated insulin resistance and accompanying low-grade inflammation causes NASH and HCC

Obesity increases the risk for nonalcoholic steatohepatitis (NASH) and hepatocarcinogenesis. However, the underlying mechanisms involved in the disease process remain unclear. Recently, we have developed a transgenic obese mouse model (Mito-Ob) by prohibitin mediated mitochondrial remodeling in adipocytes. The Mito-Ob mice develop obesity in a sex-neutral manner, but obesity-associated adipose inflammation and metabolic dysregulation in a male sex-specific manner. Here we report that with aging, the male Mito-Ob mice spontaneously develop obesity-linked NASH and hepatocellular carcinoma (HCC). In contrast, the female Mito-Ob mice maintained normal glucose and insulin levels and did not develop NASH and HCC. The anti-inflammatory peptide ghrelin was significantly upregulated in the female mice and down regulated in the male mice compared with respective control mice. In addition, a reduction in the markers of mitochondrial content and function was found in the liver of male Mito-Ob mice with NASH/HCC development. We found that ERK1/2 signaling was significantly upregulated whereas STAT3 signaling was significantly down regulated in the tumors from Mito-Ob mice. These data provide a proof-of-concept that the metabolic and inflammatory status of the adipose tissue and their interplay at the systemic and hepatic level play a central role in the pathogenesis of obesity-linked NASH and HCC.


Results
Mito-Ob mice display sex differences in adipose tissue structure and function, and in metabolic dysregulation. Visceral adipose tissue inflammation plays a critical role in obesity associated systemic metabolic dysregulation in both humans and mice 36 . The sexually dimorphic metabolic phenotype of Mito-Ob mice 35 prompted us to investigate the inflammatory status of their adipose tissues. Immunohistochemical analysis using macrophage specific marker anti-CD68 antibody (Abcam) revealed a significant increase in inflammatory macrophages in male Mito-Ob mice compared with wild-type mice (Fig. 1a). This difference was not observed between female Mito-Ob mice and control mice, despite comparable obesity and adipocyte hypertrophy in both the male and female Mito-Ob mice (Fig. 1a). Subsequent, data analysis by two-way ANOVA showed statistically significant interaction between sex and genotype (P < 0.01) (Fig. 1a). Consistent with the histological phenotype of adipose tissue, a sex-dimorphic change in adiponectin, leptin and resistin was also found. The adiponectin level was significantly upregulated in female Mito-Ob mice but not in male Mito-Ob mice in comparison with respective control mice (Fig. 1b). For adiponectin, no significant interaction was found between sex and genotype by two-way ANOVA (P < 0.27). Leptin and resistin levels were upregulated in male Mito-Ob but remained unchanged in female Mito-Ob mice in comparison with their respective control mice (Fig. 1b). Two-way ANOVA showed significant interaction between sex and genotype for leptin but not for resistin (Fig. 1b). Fasting serum insulin level was significantly upregulated in male Mito-Ob mice compared with control mice, whereas female Mito-Ob mice had insulin levels similar to wild type mice (Fig. 1c). Interestingly, Mito-Ob mice showed inverse alteration in the anti-inflammatory peptide ghrelin levels between males and females. Serum ghrelin level was significantly increased in female mice compared with wild type control mice (Fig. 1c). In contrast, male Mito-Ob had a significantly reduced ghrelin level compared with wild type control mice (Fig. 1c). Two-way ANOVA showed significant interaction between sex and genotype for insulin and ghrelin (Fig. 1c). This would imply that ghrelin might have a role in sexually dimorphic metabolic phenotypes in Mito-Ob mice. Taken together, these Histograms showing serum adipokine and hormone levels in Mito-Ob mice at 9 months of age. Data are presented as mean ± SEM (n = 5-7 mice in each group). Two-way ANOVA (a-c) macrophage: P < 0.05 for genotype, P < 0.001 for sex, P < 0.01 for interaction; adiponectin: P = 0.17 for genotype, P < 0.001 for sex, P = 0.27 for interaction; leptin: P < 0.02 for genotype, P < 0.001 for sex, and P < 0.005 for interaction; Resistin: P = 0.72 for genotype, P = 0.72 for sex, and P = 0.30 for interaction; insulin: P < 0.0001for genotype, sex, and interaction; ghrelin: P < 0.05 for genotype, P < 0.0005 for sex and interaction. Asterisks indicate comparison between sex matched Mito-Ob vs Wt. *P < 0.05, **P < 0.01, ***P < 0.001 by Student's t test. Wt -wild type; F -female; M -male.
Scientific RepoRts | 6:23608 | DOI: 10.1038/srep23608 data suggest that Mito-Ob mice exhibit sex differences in visceral adipose tissue structure and function, and opposite changes in serum ghrelin levels that correlate with sexually dimorphic metabolic and inflammatory phenotypes of Mito-Ob mice.
In our previous study, we have shown that Mito-Ob mice overexpress PHB in adipose tissue 35 and in histiocytes/macrophages 37 . To confirm that there is no leaky overexpression of PHB from aP2 promoter in the liver of Mito-Ob mice, we measured PHB protein levels by western immunoblotting. No significant difference in PHB protein level was found in the liver from wild type and transgenic mice suggesting that PHB is not overexpressed in the liver of Mito-Ob mice (Fig. 2a).
NAFLD in male Mito-Ob mice progresses to NASH with aging. The sex specific hyperinsulinemia and ectopic fat accumulation in the liver of male Mito-Ob at 6 months of age, as found in our previous study 35 led us to investigate the hepatic phenotype with aging. Histological examination of the liver at 9 months of age showed signs of NAFLD progression to NASH in male Mito-Ob mice as revealed by the characteristic ballooning of hepatocytes and lymphocyte infiltration (Fig. 2b). Increased inflammatory macrophages and lymphocytic infiltration in the liver was further confirmed by immunohistochemical analysis using macrophage and lymphocyte Scientific RepoRts | 6:23608 | DOI: 10.1038/srep23608 specific marker antibodies (Fig. 2c). Infiltrated cells were predominantly positive for CD20 and CD68, and only few were CD3 positive (Fig. 2c). Taken together, these data suggest that obesity associated NAFLD in male Mito-Ob mice spontaneously progress to NASH with aging.
Male Mito-Ob mice spontaneously develop HCC with aging. Interestingly, in our follow-up study we found that male Mito-Ob mice start to develop pale to whitish nodular tumors (2-4 mm) on their liver surfaces around 10-12 months of age, which progressively grew into multifocal larger tumors (6-8 mm) by 12-14 months of age (Fig. 3a), with a prevalence of ~25%. Histological analysis confirmed tissue structure characteristic of HCC (Fig. 3b). Apoptotic cell death as determined by TUNEL staining was significantly increased in the livers of male Mito-Ob mice compared with control mice (Fig. 3c). A parallel increase in Ki67-positive proliferating cells was Scientific RepoRts | 6:23608 | DOI: 10.1038/srep23608 also observed in the liver of male Mito-Ob mice (Fig. 3c). Tissue necrosis and anisocytosis of hepatic nuclei were also apparent (Fig. 3b). Thus, male Mito-Ob mice showed continuous hepatocyte death and compensatory proliferation, a critical process in hepatocarcinogenesis 1,38 .

Mito-Ob mice with NASH/HCC exhibit mitochondrial dysregulation. Glucose and lipid toxicity as
well as chronic low-grade inflammation are known to cause mitochondrial dysregulation 39 , and emerging evidence suggests that mitochondrial dysregulation in hepatocytes precedes the onset of NASH and HCC 40 . Because the male Mito-Ob mice display all these causative signs, therefore, we sought to know the mitochondrial status in NASH and tumor bearing livers. An immunohistochemical analysis using a mitochondrial marker-specific antibody showed a differential staining pattern in the liver tissue with hepatic lesions, which was significantly diminished in the hepatic lesions compared with the area having normal histological architecture (Fig. 4a), suggesting a relationship between the hepatic lesions and mitochondrial dysregulation. Livers from the wild-type mice showed a homogenous staining pattern (Fig. 4a). The mitochondrial transcription factor A (Tfam) level was significantly decreased in the liver from male Mito-Ob compared with wild type mice as determined by western immunoblotting, which was further reduced in the liver from tumor bearing mice (Fig. 4b). This difference was not observed in the expression level of the nuclear transcription factor-1 (Nrf-2), which is required for nuclear encoded mitochondrial proteins (Fig. 4b). The expression level of PPARγ -coactivator-1α (PGC-1α ), showed a trend similar to Tfam protein level, however, it was not statistically different between different groups (Fig. 4b).
The mitochondrial DNA copy number was significantly reduced in Mito-Ob mice with NASH development compared with the control liver (Fig. 4e). However in tumor samples, the mitochondrial DNA copy number was up regulated in comparison with control and NASH samples (Fig. 4e).
Taken together, these data suggest that the hepatic phenotype in male Mito-Ob mice mimics obesity-associated mitochondrial dysregulation in steatohepatitis and HCC in humans 40 .
Mito-Ob mice with HCC showed increased hepatic oxidative DNA damage. To further explore the link between mitochondrial dysregulation and HCC development in male Mito-Ob mice, we measured oxidative DNA damage using the ROS antibody. Hepatic oxidative DNA damage was significantly increased in the liver of tumor bearing Mito-Ob mice compared with Mito-Ob mice without tumors and wild type mice (Fig. 4c,d). An increase in in oxidative DNA damage was also found in Mito-Ob mice without tumors compared to wild type mice; however, it was significantly less than Mito-Ob mice with tumors (Fig. 4c,d). Collectively, these data point towards a role of oxidative stress in the obesity-linked pathogenesis of NASH and HCC.
In this context, it is important to note that mitochondrial impairment and hepatic cell deaths are common features of NASH 15,40 , whereas progression from NASH to HCC also involves mark increase in compensatory cell proliferation 1, 38 . This may be the reason for a negative correlation that was observed between mtDNA copy number and oxidative stress during NASH and HCC development in Mito-Ob mice (Fig. 4c-e). Moreover, this data suggests that an increase in mtDNA may not necessarily mean fully adapted mitochondria.
Mito-Ob mice with HCC showed reduced liver ghrelin level. The pathogenesis of NASH and HCC often associated with increased hepatic inflammation. Because male Mito-Ob mice showed reduced serum ghrelin level and increased hepatic inflammation, therefore, we determined the expression level of ghrelin in the liver from Mito-Ob mice by western immunoblotting. Similar to serum ghrelin levels, tumor bearing liver from male Mito-Ob mice showed significantly reduced ghrelin level in comparison with liver from Mito-Ob mice without tumor and wild type control mice (Fig. 5a). Whereas liver from female Mito-Ob showed significantly increased ghrelin level compared with their male counterparts (Fig. 5a), suggesting a potential role of anti-inflammatory peptide ghrelin in in the development of NASH and HCC in male Mito-Ob mice. ERK1/2 and STAT3 signaling are inversely altered in the liver tumors from Mito-Ob mice. To get insight into the cell signaling pathways involved in HCC development in Mito-Ob mice, we determined the activation level of PI3K/Akt, MAPK/ERK and STAT3 signaling pathways because they have been implicated in the pathogenesis of HCC 1,26,41 , and have been shown to be modulated by PHB in other cell/tissue types 42 . The phopsho-ERK1/2 (p-ERK1/2) level was significantly increased in the liver from male Mito-Ob mice compared with wild type mice, and further increased in the tumor bearing livers (Fig. 5b). In contrast, the phospho-STAT3 (p-STAT3) level showed significant reduction in Mito-Ob mice, and further reduced in tumor bearing mice (Fig. 5b). No significant change in p-Akt levels was found in the liver from Mito-Ob mice with and without tumor in comparison with the wild type mice (Fig. 5b). Taken together, these data suggest a role for increased oncogenic mediator p-ERK and decreased p-STAT3 in the liver tumor development in male Mito-Ob mice.

Discussion
This study reports a role of obesity related hyperinsulinemia and chronic low-grade inflammation in the development of NASH and HCC, independent of diet and carcinogen respectively. This pathological progression is marked by adipose inflammation, an inverse alteration of serum insulin and ghrelin levels, an increase in hepatic lipid accumulation, macrophage and lymphocyte infiltration, and a reduction in hepatic mitochondrial content and function, with a parallel increase in hepatic DNA damage, cell death and compensatory proliferation (Fig. 6). Of note, both female and male Mito-Ob mice have a comparable degree of obesity however, only the males displayed pathological features of NASH and HCC. This would suggest that obesity alone is not sufficient for the development of NASH and HCC but rather require additional effects of adipose inflammation, hyperinsulinemia and the degree of hepatic inflammation.
In this context, it is important to note that there are a number of obese rodent models available that display obesity-associated metabolic dysregulation and NAFLD. However, the majority of them do not develop obesity-linked HCC spontaneously like male Mito-Ob mice, except TSOD mice 29 . However, TSOD mice also  develop diabetes along with obesity and it remains unclear whether TSOD mice develop HCC due to obesity or diabetes. This raises an important question: how and why do obesity-associated abnormalities lead to HCC development in male Mito-Ob mice? This may in part be due to a significant reduction in ghrelin levels in male Mito-Ob mice because emerging evidences suggest that ghrelin is an important anti-inflammatory peptide [43][44][45][46] . Since obesity-associated chronic low-grade inflammation in major metabolic tissues is considered an important driver for obesity related disorders, it is possible that the anti-inflammatory function of ghrelin has a role in the sexually dimorphic metabolic phenotype in Mito-Ob mice. A reduced ghrelin level in male Mito-Ob mice may be permissive in the development of obesity-associated chronic low-grade inflammation, which in turn promotes insulin resistance, NASH and HCC. By the same token, an increased ghrelin level in female Mito-Ob mice may have a protective role against obesity-associated low-grade inflammation, and subsequently against insulin resistance and NAFLD.
Furthermore, a parallel decrease in liver ghrelin level in Mito-Ob mice during disease progression as observed in this study may be due to a sex-dimorphic effect of PHB on ghrelin production from macrophages, as macrophages are known to produce ghrelin 47 , which in turn contribute to the development of NASH and HCC in male Mito-Ob mice and confer a protection in female mice. We propose that PHB overexpressing adipocytes and macrophages respond differently in males and females, especially in obesity, and may have a role in sex differences in metabolic dysregulation and adipose inflammation, with as a consequence sex-dimorphic NASH and HCC Figure 6. Schematic diagram showing proposed mechanism in obesity-linked NASH and HCC development. We propose that an extensive interplay between the metabolic and inflammatory status at the adipose tissue and systemic levels, as well as at the hepatic tissue level is involved in the progression of obesitylinked hepatic steatosis to NASH and HCC. Pro -proinflammatory; Anti -anti-inflammatory. development in male Mito-Ob mice. It would be interesting to know whether normalization of insulin and/or ghrelin levels would reduce or prevent NASH and HCC occurrence in the male Mito-Ob mice. Of note, insulin sensitizers such as thiazolidinediones (TZDs), glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase 4 inhibitors have been studied as therapeutic approaches for NAFLD in recent years and have been shown to have some beneficial effects 45 . In addition, treatment with anti-inflammatory peptide ghrelin has been reported to reduce inflammation and oxidative stress, NAFLD and NASH 45,46 . Thus, a combination therapy of insulin sensitizers with anti-inflammatory agents, especially ghrelin, may provide a better outcome that warrants further investigation. The male Mito-Ob mouse would be a fitting model for such preclinical studies because they display both pathological features: insulin resistance and significantly reduced serum ghrelin levels.
Our data suggest an important role of the oncogenic determinant p-ERK in combination with p-STAT3 down regulation and mitochondrial dysregulation in the development of HCC in male Mito-Ob mice. Whether these changes are related to each other or work independently in HCC development remains to be determined. Glucotoxicity, lipotoxicity, and chronic low-grade inflammation are known to cause mitochondrial dysregulation 39 . A number of studies have reported mitochondrial dysregulation in the liver in the pathogenesis of fatty liver and HCC both in humans and rodents [15][16][17]40 . In addition, recently p-STAT3 has been shown to play an important role in the maintenance of mitochondrial function 48 . A parallel change in mitochondrial dysregulation and the p-STAT3 level during HCC development in male Mito-Ob mice are consistent with the emerging evidence in the literature. However, it is not known whether they coordinate in the development of HCC, which warrants further investigation. In this context, it is important to note that the liver specific Phb knockout (Phb− /− ) mice have been reported earlier, which also develop HCC by 35-46 weeks of age 49 . HCC development in the Phb− /− mice has been attributed to the tumor suppressor function of PHB 49 . However, because PHB is a critical protein for the structural and the functional integrity of mitochondria, and hepatic mitochondrial dysregulation has been associated with HCC development 40 , it is possible that HCC development in the Phb− /− mice may in part be due to compromised mitochondrial function. Nevertheless, because PHB has a role as transcriptional co-regulator as well as in membrane signaling 50,51 , down regulation of liver PHB expression during HCC development in male Mito-Ob mice, as observed in our data, may also affect its extra-mitochondrial functions along with mitochondrial dysregulation. It is possible that such extra-mitochondrial changes may contribute to HCC development in male Mito-Ob mice.
In summary, our data suggest an extensive interplay between the metabolic and inflammatory status at the adipose tissue and systemic levels, as well as at the hepatic tissue level in the progression of obesity-linked NAFLD to NASH and HCC (Fig. 6). Thus, this study provides a proof-of-concept that obesity-associated metabolic and immune dysregulation indeed have a central role in the pathogenesis of NASH and HCC. In addition, the Mito-Ob mouse revealed a sex-dimorphic role of prohibitin in adipose and immune functions.

Materials and Methods
Reagents. Monoclonal anti-PHB antibody was purchased from Cell Signaling Technology (Danvers, MA).
Transgenic mice. All experiments involving animals were performed as per the study protocols (#12-007/1/2/3) approved by the Animal Care and Use Committee at the University of Manitoba and following the guidelines of the Canadian Council of Animal Care (CCAC). The generation and initial characterization of Mito-Ob mice has been reported in our previous study 35,37 . The Mito-Ob transgenic mice were identified by genotyping the tail DNA by PCR using the above forward primer: 5′-GCAGCCCGGGGGATCCACTA-3′ and reverse primer: 5′-GCACACGCTCATCAAAGTCCTCTCCGATGCTG-3′ . The animals were given normal chow (LabDiet, St. Louis, MO) and water ad libitum. The body weight and food intake of the Mito-Ob and the wild-type mice were recorded weekly as described previously 35 . Histology and immunohistochemistry. Visceral adipose tissue and the liver from 9-12 month old Mito-Ob and their wild type littermates (as applicable) were fixed in buffered formaldehyde (Fisher Scientific, Ottawa, ON) and subsequently dehydrated, embedded in paraffin, and 5 μm sections were stained with hematoxylin-eosin or a protein specific antibody for immunohistochemistry 35,37 . Western immunoblotting. Liver expression level of mitochondrial proteins including PHB and cell signaling proteins were determined by Western immunoblotting. In brief, the total liver tissue lysates from Mito-Ob and wild-type mice containing equal amount of proteins (~20 μg/lane) were separated by SDS-PAGE and subsequently analyzed by Western immunoblotting using a protein-specific primary antibody and a HRP-conjugated respective secondary antibody as described before 35,37 . Finally, immunodetection was performed using an Enhanced Chemiluminiscence kit (GE Healthcare, Mississauga, ON).
Adipokines and hormones measurements. Adipokines and hormones in the mouse serum were measured using mouse Bio-Plex ProTM Assays Diabetes panel and Bio-Plex 200TM multiplex suspension array systems (Bio-Rad, Hercules, CA) as per the manufacturer's protocols 35,37 . Apoptosis, cell proliferation and oxidative DNA damage measurements. Apoptotic cell death using TUNEL assay (Trevigen, Gaithersburg, MD) and cell proliferation using anti-Ki67 antibody (Cell Signaling)