Sex Hormone-Binding Globulin Reduction in Metabolic Disorders May Play a Role in NAFLD Development

Abstract

Low plasma sex hormone-binding globulin (SHBG) levels are a hallmark in chronic metabolic diseases, including nonalcoholic fatty liver disease (NAFLD), which represents a spectrum of disease ranging from hepatocellular steatosis through steatohepatitis to fibrosis and irreversible cirrhosis. The functional link between altered SHBG production and NAFLD development and progression remains unclear. We investigated the effects of overexpressing human SHBG in 2 mouse models of NAFLD: a genetically induced double transgenic mouse and a diet-induced model. Remarkably, SHBG overexpression in both NAFLD models significantly reduced liver fat accumulation by reducing key lipogenic enzymes. These findings were corroborated by modulating SHBG expression and by adding exogenous SHBG in HepG2 cells, suggesting the cell autonomous nature of the mechanism. Mechanistically, exogenous SHBG treatment downregulated key lipogenic enzymes by reducing PPARγ messenger RNA and protein levels through activation of extracellular signal-regulated kinase-1/2 mitogen-activated protein kinase pathway. Taking together, we found that SHBG modulates hepatic lipogenesis. This is of importance because reduction of SHBG plasma levels in obese and type 2 diabetic subjects could be directly associated with NAFLD development through an increase in hepatic lipogenesis. Our results point to SHBG as a therapeutic target for preventing or arresting NAFLD development.

Nonalcoholic fatty liver disease (NAFLD) is 1 cause of fatty liver, occurring when fat is deposited (steatosis) in the liver that is not a result of excessive alcohol use (1). NAFLD represents a spectrum of disease ranging from hepatocellular steatosis through steatohepatitis to fibrosis and irreversible cirrhosis (2). The prevalence of NAFLD in the general population of Western countries is 20% to 30% (3). The increased prevalence of diabetes and obesity in the general population is considered to be the most common cause for NAFLD (4). Several studies have shown that obese subjects, type 2 diabetic (T2D) patients, and individuals with NAFLD have low sex hormone-binding globulin (SHBG) levels (5–7), a protein produced by the liver that acts as a carrier of sex steroids and regulates their bioavailability at the tissue level (8).

Apart from the increase in lipogenesis that drives fat accumulation in the liver (1), there is extensive evidence supporting a central role of tumor necrosis factor alpha (TNF-α) and other proinflammatory cytokines in the development of NAFLD (9, 10). Moreover, TNF-α plasma levels have been found elevated in patients with NAFLD (11, 12). We have previously demonstrated that an increase in hepatic lipogenesis and proinflammatory cytokines, such as TNF-α, downregulates SHBG production by decreasing hepatic HNF-4α levels, a key regulator of SHBG transcription (13–15). Therefore, lipid accumulation and low-grade inflammation in NAFLD could be a common link explaining the low circulating SHBG levels in this disease. The latter raises the intriguing question of whether low SHBG levels could contribute to the progression of NAFLD, rather than simply being a consequence and a surrogate biomarker.

To address the importance of SHBG expression in NAFLD in vivo, we decided to overexpress SHBG in an NAFLD mouse model, by creating a double transgenic mouse (SHBG-C57BL/ksJ-db/db), resulting from crossing the C57BL/ksJ-db/db mouse, a model of NAFLD (16), with the human SHBG transgenic mice, and in a hepatic steatosis diet–induced model using the human SHBG transgenic mice and their wild-type littermates fed with a high-fructose diet (HFrD). The SHBG overexpression in both NAFLD and hepatic steatosis models significantly reduced liver fat accumulation by reducing key lipogenic enzymes. The cell autonomous effects of SHBG on hepatic lipogenesis were demonstrated in HepG2 cells by modulating SHBG expression and by treating with exogenous SHBG. Moreover, we elucidated the molecular mechanism by which SHBG reduced the hepatic lipogenesis, which involved PPARγ and extracellular signal-regulated kinase-1/2 (ERK-1/2) mitogen-activated protein kinase (MAPK) pathway.

Overall, our results suggest that SHBG protects against liver fat accumulation and therefore NAFLD development. More specifically, rather than being merely a surrogate marker, SHBG plasma levels regulate hepatic lipogenesis through PPARγ modulation. This is of importance because reduction of SHBG plasma levels in obese and T2D subjects could be directly associated with NAFLD development through an increase in hepatic lipogenesis. Our results point to SHBG as a therapeutic target for preventing or arresting NAFLD development.

Materials and Methods

Animals

The human SHBG transgenic mice were backcrossed onto C57BL/ksJ-db/db background to obtain mice expressing human SHBG and developing obesity and NAFLD. These mice have been previously described (17). Mice were maintained under standard conditions with food (Global Diet 2018; Harlan Interfauna Iberica, Barcelona, Spain) and water provided ad libitum and a 12-hour light/dark cycle. Experimental procedures were approved by the Institutional Animal Use Subcommittees of Vall Hebron University Hospital Research Institute and the Universitat Autonoma Barcelona (registry no. 45/13, Animal Experimental Ethics Committee).

In vivo experiments

Male and female mice of the 4 genotypes (C57BL/ksJ-db/⁺, C57BL/ksJ-db/db, SHBG-C57BL/ksJ-db/⁺, and SHBG-C57BL/ksJ-db/db; n = 5 each) were euthanized at 6 weeks of age and blood and tissue were collected and weighed for RNA and protein isolation.

Human SHBG transgenic and wild-type male mice (n = 5) were fed for 8 weeks with a semisynthetic diet containing 20% protein, 4% soybean fat, and 76% fructose. At the end of the study, mice were euthanized and blood and tissue were collected for RNA and protein isolation.

Cell culture experiments

Cell culture reagents were from Life Technologies Inc (Invitrogen SA, Barcelona, Spain). HepG2 hepatoblastoma cells (cat. no. HB-8065; ATCC, LGC Standards SLU, Barcelona, Spain) were maintained in Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum and antibiotics. HepG2 cells overexpressing or underexpressing SHBG were achieved by stable transfection using an SHBG expression vector (pCMV-SHBG) or a vector expressing small interfering RNA against SHBG (pLKO.1-SHBG; Sigma-Aldrich SL, Madrid, Spain). An empty vector (pCMV) and a pLKO.1 containing random sequences (pLKO.1-Control) were used as controls, respectively. The pCMV and pCMV-SHBG vectors were kindly provided by Dr. Geoffrey Hammond, Univeristy of British, Vancouver, Canada); pLKO.1-Control was kindly provided by Dr. Josep Villena, Vall d’Hebron Institut de Recerca, Barcelona, Spain). Human primary hepatocytes were purchased from Tebu-Bio (Barcelona, Spain). All transfections were performed using lipofectamine 2000 (Invitrogen SA). Transfected HepG2 cells were selected for 10 days by treating with G418 (750 µg/mL). Purified SHBG from human plasma (Abyntek Biopharma SL, Bizkaia, Spain) was used to treat HepG2 cells.

Histology

For morphological studies, 3 animals of each genotype (C57BL/ksJ-db/⁺, C57BL/ksJ-db/db, SHBG-C57BL/ksJ-db/⁺, and SHBG-C57BL/ksJ-db/db) were used. From the diet study, wild-type and human SHBG transgenic mice (n = 3 each) were also analyzed. Livers were fixed in 4% paraformaldehyde for 24 hours and embedded in paraffin. Serial 5-μm-thick sections were used for histological examination and stained with hematoxylin-eosin (H&E).

SHBG measurements

Human SHBG levels from HepG2 cell culture media and from mouse plasma were measured using enzyme-linked immunosorbent assay (ELISA; Demeditec Diagnostics GmbH, Kiel-Wellsee, Germany).

Liver triglyceride

Mouse liver and HepG2 cell triglyceride (TG) levels were measured using a TG assay kit (cat. no. K622-100; BioVision Inc., Aachen, Germany) following the manufacturer’s instructions.

RNA analysis

Total RNA was extracted from HepG2 cells, human primary hepatocytes, and mouse liver samples using TRIzol reagent (Invitrogen SA). Reverse transcription was performed at 42°C for 50 minutes using 3 μg of total RNA, 200 U of Superscript II, and an oligo-dT primer and reagents provided by Invitrogen. An aliquot of the reverse transcription product was amplified in a 25-μL reaction using SYBRGreen (Invitrogen SA) with appropriate oligonucleotide primer pairs corresponding to human SHBG, ACC, FAS, ACLY, PPARG, and 18S and mouse ACC, FAS, ACLY, PPARG, and 18S. Results were analyzed using the 7000 Sequence Detection System program (Applied Biosystems, Thermo Fisher Scientific, Barcelona, Spain).

For microarrays analysis messenger RNA (mRNA) liver samples from db/db and SHBG-db/db mice were isolated using the RNeasy Mini Kit (cat. no. 74104, QIAGEN, Madrid, Spain) following the manufacturer’s instructions.

Western blot analysis

HepG2 cells, human primary hepatocytes, and mouse liver samples were homogenized in RIPA buffer with Complete™ protease inhibitor cocktail (Roche Diagnostics, Barcelona, Spain). Protein extracts were used for Western blotting with antibodies against FAS (cat. no. 22759, Abcam, Cambridge, UK), ACC (cat. no. 63531, Abcam), ACLY (cat. no. 4332, Cell Signaling Barcelona, Spain), PPARγ (H-100, cat. no. sc-7196, Santacruz Biotechnologies, Heidelberg, Germany), P-PPARγ (cat. no. 4888, Biossusa, Woburn, MA), P-ERK-1/2 (cat. no. 9101, Cell Signaling), and PPIA (SA-296; BIOMOL Int., Madrid, Spain) (Table 1). Specific antibody–antigen complexes were identified using the corresponding horseradish peroxidase–labeled goat antirabbit immunoglobulin-G and chemiluminescent substrates (Millipore, Madrid, Spain) by exposure to x-ray film. Quantification was performed using ImageJ free software.

Microarray hybridization and analysis

The goal of the study was to compare hepatic gene expression patterns between C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db mice. Microarrays were carried out using the Affymetrix microarray platform and the Genechip Mouse Gene 2.0 ST Array (an array with >35,000 coding and noncoding transcripts used to explore mouse biology and disease processes). The arrays were performed at the High Technology Unit of our research institute as described elsewhere (www.affymetrix.com). Data obtained from the microarrays were analyzed by the Statistics and Bioinformatics Unit of our research institute. To select differentially expressed genes, the unit used a method described by Smyth and Altman (18). All statistical analysis was done using the free statistical language R and the libraries developed for microarray data analysis by the Bioconductor Project (www.bioconductor.org). To determine pathways and networks that were significantly regulated, we performed pathway analysis using Ingenuity Pathway Analysis program (http://www.ingeniuty.com).

Statistical analyses

Normal distribution of the variables was evaluated using the Kolmogorov-Smirnov test. Comparison of quantitative variables was performed by either the Student t test or Mann-Whitney test according to the data distribution. All data are presented as means ± SEM. Significance was accepted at the level of P < 0.05. Statistical analyses were performed with the SPSS statistical package (SPSS Inc, Chicago, IL).

Results

SHBG overexpression reduces liver weight and lipid accumulation in SHBG-C57BL/ksJ-db/db mice

Previous studies have reported that C57BL/ksJ-db/db mice, a well-established model of NAFLD, have increased liver weight and hepatic lipid accumulation (16). We confirmed these results in the characterization of our mouse model, because we observed a significantly higher liver weight in obese male and female C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db mice compared with their lean C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/⁺ littermates [Fig. 1(A) and 1(B)]. Importantly, SHBG-C57BL/ksJ-db/db male mice had significantly reduced liver weight compared with C57BL/ksJ-db/db mice [Fig. 1(A)].

Histological analysis of livers from male and female mice revealed normal histology in lean C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/⁺ mice [Fig. 1(C) and 1(D)], whereas obese C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db mice showed hepatic lipid accumulation [Fig. 1(C) and 1(D)]. Male and female SHBG-C57BL/ksJ-db/db mice also showed hepatic lipid accumulation, although this was lower than in C57BL/ksJ-db/db mice [Fig. 1(C) and 1(D)]. Analysis of the liver total TG content showed that male and female lean C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/⁺ mice had lower TG than their obese C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db littermates [Fig. 1(E) and 1(F)]. Remarkably, both male and female SHBG-C57BL/ksJ-db/db mice showed lower liver TG accumulation than C57BL/ksJ-db/db mice [Fig. 1(E) and 1(F)].

Human SHBG transgenic mouse models have been previously described, and it has been shown that they have high plasma SHBG levels (19). We confirmed this in our mouse model and found that SHBG-C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/db mice had 0.96 ± 0.2 µM and 0.37 ± 0.1 µM SHBG, whereas no human SHBG was detected in C57BL/ksJ-db/⁺ and C57BL/ksJ-db/db mice (Table 2). The high SHBG plasma levels of these mice results in an increase in their testosterone plasma levels (19). To rule out potential effects of testosterone on hepatic protection against lipid accumulation, we repeated the experiments with castrated SHBG-C57BL/ksJ-db/db mice. Our results showed that castrated these mice had similar liver weights as noncastrated SHBG-C57BL/ksJ-db/db mice [Fig. 1(G)]. The liver histological analysis revealed that castrated SHBG-C57BL/ksJ-db/db mice were also partially protected against hepatic lipid accumulation by both histological and biochemical analysis [Fig. 1(H) and 1(I)].

SHBG overexpression downregulates key lipogenic enzymes in SHBG-C57BL/ksJ-db/db mice

To elucidate the molecular mechanism(s) by which SHBG overexpression protected against lipid accumulation, we performed microarray analyses to detect the differentially expressed genes between C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db livers. We focused our attention on the molecular pathways related to lipogenesis, lipid export, and fatty acid oxidation (Supplemental Table 1). Lipogenic-related pathways had a high number of genes affected, whereas in lipid export or fatty acid oxidation pathways, the number of genes affected was lower. We therefore analyzed pathways that included lipid synthesis, lipid concentration, lipid homeostasis, fatty acid synthesis, fatty acid concentration, fatty acid metabolism, TG synthesis, and TG concentration (Supplemental Tables 2–9). The results showed an increase in downregulated genes in all lipogenic-related pathways in the SHBG- C57BL/ksJ-db/db mice compared with the C57BL/ksJ-db/db mice (Supplemental Tables 2–9). Fifty-four genes were downregulated in SHBG-C57BL/ksJ-db/db mice compared with the C57BL/ksJ-db/db mice (Supplemental Tables 2–9).

Among the downregulated genes in SHBG-C57BL/ksJ-db/db vs C57BL/ksJ-db/db mice, we found 3 key genes regulating hepatic lipogenesis: ACC, FAS and ACLY (20–22). Therefore, we analyzed the ACC, FAS, and ACLY mRNA and protein levels of lean (C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/⁺) and obese (C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db) male and female mice. Although in some cases the relative amounts of mRNA levels did not exactly correlate with those of protein (e.g., FAS in C57BL/ksJ-SHBG-db/db), our results showed that obese C57BL/ksJ-db/db mice had significantly higher ACC, FAS, and ACLY mRNA levels than the lean C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/+ mice in males [Fig. 2(A)] and females [Fig. 2(B)]. However, SHBG-C57BL/ksJ-db/db mice showed substantially lower ACC, FAS, and ACLY mRNA levels [Fig. 2(A) and 2(B)] than C57BL/ksJ-db/db mice. We also found that ACC, FAS, and ACLY protein levels were significantly higher in obese C57BL/ksJ-db/db mice compared with lean C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/⁺ mice, whereas SHBG-C57BL/ksJ-db/db mice had lower ACC, FAS, and ACLY protein levels in both male and female mice [Fig. 2(C) and 2(D)].

SHBG overexpression protects against HFrD-induced fatty liver disease

Because our findings with the humanized SHBG-C57BL/ksJ-db/db mice point to SHBG as a factor that could protect against hepatic steatosis, we next examined whether this protective effect could be exerted in a diet-induced model of NAFLD. To accomplish this, wild-type and human SHBG transgenic male mice were fed with an HFrD for 8 weeks.

The liver histological examination revealed that wild-type mice developed hepatic steatosis, whereas human SHBG transgenic mice did not [Fig. 3(Ai) and 3(Aii)]. Importantly, livers from human SHBG transgenic mice showed approximately threefold less TG content than wild-type mice [Fig. 3(B)]. Analysis of the hepatic ACC, FAS, and ACLY mRNA and protein levels from these mice showed an important suppression in ACC, FAS, and ACLY mRNA and protein levels in the human SHBG transgenic mice in comparison with their wild-type mice littermates [Fig. 3(C) and 3(D)].

SHBG modulates hepatic lipogenesis in HepG2 cells

We first analyzed the effects of under- or overexpression of SHBG on lipogenesis and TG accumulation in HepG2 cells. The results showed that SHBG overexpression (pCMV-SHBG) led to a substantial increase in SHBG mRNA and protein production in HepG2 cells [Fig. 4(A) and 4(B)]. In contrast, SHBG underexpression (pLKO.1-SHBG) led to a substantial decrease in SHBG mRNA and protein production in HepG2 cells [Fig. 4(D) and 4(E)].

We next analyzed the total TG content of the SHBG-overexpressing and SHBG-underexpressing HepG2 cells. The results showed that the cells overexpressing SHBG had significantly lower TG [Fig. 4(C)], whereas cells underexpressing SHBG had significantly higher TG [Fig. 4(F)] compared with their respective controls (pCMV and pLKO.1).

Moreover, the analysis of the mRNA expression abundance of ACC, FAS, and ACLY—crucial enzymes involved in hepatic lipogenesis—showed that cells overexpressing SHBG had reduced ACC, FAS, and ACLY mRNA levels [Fig. 4(G)], whereas cells underexpressing SHBG had higher ACC, FAS, and ACLY mRNA levels compared with their respective controls (pCMV and pLKO.1) [Fig. 4(H)]. Furthermore, ACC and ACLY, but not FAS, protein levels were reduced in overexpressing SHBG cells and increased in underexpressing SHBG cells compared with their respective controls (pCMV and pLKO.1) (Fig. 5).

Exogenous SHBG inhibits lipogenesis by reducing PPARγ levels via the ERK-1/2 MAPK pathway

Although overexpressing or underexpressing SHBG in HepG2 cells confirmed our in vivo results obtained in both genetically and diet-induced NAFLD mouse models, we could not rule out an important role for circulating SHBG, having in mind that HepG2 cells secrete SHBG to the cell culture media as the liver does in our mouse models.

To test this hypothesis, we first treated HepG2 cells with exogenous SHBG (100 nM) during 2, 4, and 6 hours of treatment and then analyzed ACC, ACLY, and FAS mRNA levels. Cell media were changed just before starting the treatments to eliminate accumulated SHBG. The results showed that ACC mRNA levels were downregulated at 2, 4, and 6 hours after SHBG treatment, whereas FAS mRNA levels were downregulated at 4 and 6 hours and ACLY mRNA levels at 6 hours after treatment with exogenous SHBG [Fig. 6(A)].

We next wanted to elucidate the exact molecular mechanism by which SHBG inhibits all 3 key lipogenic enzymes. It has been established in the literature that there are several key transcription factors regulating hepatic lipogenesis, such as SREBP, LXRα, and PPARγ (23, 24). We therefore analyzed if these transcription factors were differentially expressed in our microarray analysis between C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db livers. The results showed that only PPARγ mRNA levels were downregulated in livers from SHBG-C57BL/ksJ-db/db mice compared with C57BL/ksJ-db/db mice (Supplemental Table 10). Furthermore, we analyzed the PPARγ mRNA and protein levels from C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db livers. The results showed a clear and substantial reduction in both PPARγ mRNA and protein levels in SHBG-C57BL/ksJ-db/db mice compared with C57BL/ksJ-db/db mice [Fig. 6(B) and 6(C)].

We then repeated the experiments with exogenous SHBG and treated HepG2 cells with or without SHBG (100 nM) to analyze the PPARγ mRNA levels. The results showed a substantial downregulation of PPARγ mRNA only 6 hours after SHBG treatment compared with vehicle-treated HepG2 cells [Fig. 6(D)]). These results initially suggested that PPARγ downregulation was not involved in the reduction of key lipogenic enzymes because ACC and FAS mRNA levels were already downregulated 4 hours after SHBG treatment [Fig. 6(A)]. However, it has been recently described that PPARγ could be phosphorylated and sent to degradation via the ERK-1/2 MAPK pathway (25, 26). We wanted to test this hypothesis in our in vitro system and therefore repeated the SHBG treatments to determine ERK-1/2 phosphorylation. The results showed that SHBG treatment induced an increase in ERK-1/2 phosphorylation within the first 30 minutes and 1 hour of treatment [Fig. 6(E)]. Importantly, this SHBG-induced ERK-1/2 phosphorylation was abrogated by cotreatment with U0126 (500 nM), a well-known inhibitor of ERK-1/2 phosphorylation [Fig. 6(F)]. Consistently, the U0126 cotreatment blocked the mRNA reduction of the key lipogenic enzymes and PPARγ caused by the SHBG treatment after the 6-hour treatment [Fig. 6(G)]. Furthermore, SHBG treatment resulted in an increased of ERK-1/2 and PPARγ phosphorylation within 2 hours that was blocked by the U0126 cotreatment [Fig. 6(H)]. Finally, because HepG2 cells may not represent human normal liver function, we decided to repeat the experiments using commercially available human primary hepatocytes. The results showed that human primary hepatocytes treated with exogenous SHBG (100 nM) for 6 hours had reduced ACC and PPARγ mRNA levels compared with vehicle-treated cells [Supplemental Fig. 1A and 1B]. However, in our culture conditions, we were not able to detect FAS or ACLY gene expression. In addition, SHBG treatment resulted in an increased of ERK-1/2 phosphorylation within 30 minutes that was blocked by the U0126 cotreatment [Supplemental Fig. 1C].

Discussion

The most well-known role of SHBG is as a carrier protein that transports sex steroids through the blood and regulates their bioavailability and accessibility to target tissues and cells (8). Circulating SHBG is altered in metabolic disorders (27, 28), with low plasma SHBG levels being considered a risk factor for developing T2D and cardiovascular disease (29–33). NAFLD is 1 of the metabolic diseases associated with low plasma SHBG levels (34). In the total plasmatic SHBG protein in humans, 20% in males and 50% in females is unoccupied (27). Whether this excess of unbound protein has a different biological function(s) is unknown. In the current study, we point out that low plasma SHBG levels in metabolic disorders could play a role in the development of these diseases. Specifically, we have demonstrated that SHBG overexpression protects against hepatic steatosis and NAFLD development in genetically and a diet-induced mouse models by reducing hepatic lipogenesis. Importantly, our results suggest that SHBG downregulation mediated by hepatic lipid accumulation and increased proinflammatory cytokines found in obesity and T2D may accelerate NAFLD development, which in turn will further reduce SHBG production (Fig. 7). We also provide strong evidence that SHBG overexpression abrogates hepatic lipogenesis by reducing PPARγ through activation of the ERK-1/2 MAPK pathway, thus reducing key lipogenic enzymes and the liver TG content.

NAFLD, characterized by an increase in hepatic TG content, is usually found in obese subjects (35). We studied the effect of human SHBG overexpression on hepatic TG synthesis and accumulation using 2 mouse models of developing NAFLD and hepatic steatosis. The obese C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db mice showed increased fatty liver compared with their lean littermates (C57BL/ksJ-db/⁺ and SHBG-C57BL/ksJ-db/⁺), whereas obese SHBG-C57BL/ksJ-db/db mice clearly had less hepatic steatosis. We also showed the beneficial effects of overexpressing SHBG in the development of hepatic steatosis induced by an HFrD. Importantly, human SHBG transgenic mice showed an important reduction in liver fat accumulation compared with wild-type mice after 8 week of HFrD feeding. These findings point to SHBG as a protective factor against diet-induced hepatic steatosis.

The testosterone role in NAFLD has been studied in humans with contradictory results. Although 18 weeks of testosterone therapy alleviated fatty liver in men (36), a later study showed no beneficial effects of testosterone in aging men (37). That males and females showed similar hepatic protection against NAFLD in our humanized SHBG-C57BL/ksJ-db/db mouse model indicates that SHBG was important in protecting against hepatic lipid accumulation in this model, regardless of testosterone levels. However, it has been reported that the presence of SHBG in the blood of the human SHBG transgenic male mice led to serum testosterone levels 10 to 100 times higher than their wild-type littermates (19). This important difference in total testosterone levels has also been found in our SHBG-C57BL/ksJ-db/db mouse model (17). Nevertheless, the histological and biochemical analysis of liver lipid accumulation in castrated SHBG-C57BL/ksJ-db/db mice provides strong evidence that SHBG itself protects against lipid accumulation independently of testosterone serum levels. Consistent with our results, low plasma SHBG levels, but not total testosterone or free testosterone, was associated with NAFLD in T2D patients (38). These findings strongly suggest that SHBG has functions independent of sex steroids and may provide a role for SHBG under physiological and pathological conditions.

To elucidate the molecular mechanisms associated with this SHBG protection against NAFLD, we analyzed the differentially expressed genes between livers of C57BL/ksJ-db/db and SHBG-C57BL/ksJ-db/db mice by microarray analyses. We found a clear downregulation of several pathways involved in hepatic lipogenesis. Other lipid-related pathways, such as lipid export and fatty acid oxidation, were also explored, but the number of genes affected was lower. Among the lipogenic genes, we found ACC, FAS, and ACLY, which are 3 key enzymes of hepatic lipogenesis (19–21). Mice lacking ACC, FAS, or ACLY are protected against weight gain and hepatic steatosis (21, 39, 40). We found a substantial reduction of ACC, FAS, and ACLY mRNA and protein levels in livers of SHBG-C57BL/ksJ-db/db mice compared with C57BL/ksJ-db/db. Importantly, similar results were found in livers of human SHBG transgenic mice compared with wild-type mice fed the HFrD, suggesting that SHBG overexpression reduces liver fat accumulation by reducing lipogenesis. This suggestion was further supported by the SHBG overexpression-induced reduction of hepatic TG content in our 2 mouse models. In this regard, it has been recently shown that liver fat, but not visceral fat or total body fat, is an independent predictor of SHBG levels, with a strong association between a decrease in liver fat because of lifestyle intervention and increased plasma SHBG levels (41). Importantly, our results suggest that higher hepatic SHBG production itself may participate in reducing hepatic lipid accumulation by reducing lipogenesis. To test the latter hypothesis and to translate our results from mouse to human, we used an in vitro approach using HepG2 cells. These cells are the only human liver cell line expressing and secreting SHBG, and we have shown previously that when cultured with 10 mM glucose or fructose they are able to accumulate palmitate through an increase in lipogenesis (13). The results supported our hypothesis in which TG content was altered in concert with changes in ACC, FAS, and ACLY by the modulation of SHBG production. Our results suggested that SHBG could be an autocrine factor regulating hepatic lipogenesis (Fig. 7), although we need to keep in mind that HepG2 cells are a human hepatoblastoma cell line and may not be the best model to study human normal liver function.

To understand how SHBG overexpression was downregulating ACC, FAS, and ACLY, we reviewed our microarray data looking for a key transcription factor altered in our SHBG- C57BL/ksJ-db/db mice. We found that PPARγ mRNA and protein levels were reduced in SHBG-overexpressing mice. These results are in agreement with the fact that PPARγ is expressed at elevated levels in the liver of several mouse models of diabetes or obesity, and that hepatic disruption of PPARγ resulted in a clear downregulation of ACC, FAS, and ACLY mRNA levels (42). Remarkably, these results were also corroborated using exogenous SHBG, because SHBG treatment reduced ACC, FAS, and ACLY mRNA levels 6 hours after treatment by reducing PPARγ protein levels through the ERK-1/2 MAPK pathway in HepG2 cells. The SHBG-induced increase in phospho-ERK-1/2 protein levels resulted in an increase of PPARγ phosphorylation, which has been described to result in PPARγ degradation (24). Moreover, the SHBG-induced downregulation of ACC, FAS, and ACLY mRNA levels was abrogated by U0126 cotreatment, a well-known inhibitor of ERK-1/2 phosphorylation. Importantly, these results were partially supported in experiments using human primary hepatocytes.

Overall, our results point to SHBG as a therapeutic target whereby increased expression may protect against NAFLD development and progression. Even if SHBG does not become a feasible therapeutic target, our findings provide clear evidence to explain why obese and type 2 diabetic subjects, 2 patient groups characterized by low plasma SHBG levels, develop NAFLD. Equally important, our results demonstrate that SHBG is more than a simple biomarker, with data highly indicative of the SHBG importance in regulating hepatic lipogenesis.

Abbreviations:

 
• ELISA

  enzyme-linked immunosorbent assay

 
• ERK

  extracellular signal-regulated kinase

 
• HFrD

  high-fructose diet

 
• MAPK

  mitogen-activated protein kinase

 
• mRNA

  messenger RNA

 
• NAFLD

  nonalcoholic fatty liver disease

 
• SEM

  standard error of the mean

 
• SHBG

  sex hormone-binding globulin

 
• T2D

  type 2 diabetic

 
• TG

  triglyceride

 
• TNF

  tumor necrosis factor

Acknowledgments

We thank Dr. Geoffrey L. Hammond, Head of the Department of Cellular & Physiological Sciences, University of British Columbia, because this project would not have been possible without his collaboration in letting us use the human SHBG transgenic mice. We also thank Lorena Ramos Perez and Cristina Sanchez Mora, Research Institute Hospital Vall d’Hebron, for their technical assistance. We acknowledge Professor Michael Underhill (University of British Columbia), Dr. Josep Jimenez Chillaron (Hospital Sant Joan de Deu), Julian Ceron (Institut d’Investigació Biomèdica de Bellvitge), and Dr. Josep Villena Delgado (Vall d’Hebron Research Institute) for the critical review of our manuscript.

This work was supported by the European Regional Development Funds and Grants CP08/00058, PI09/144, and PI12/01357 from the Instituto de Salud Carlos III (to D.M.S.), and by the CIBER de Diabetes y Enfermedades Metabólicas Asociadas, an initiative of Instituto de Salud Carlos III (to C.S.-L., C.H., R.S., and D.M.S.). D.M.S. is the recipient of a Miguel Servet II contract. D.M.S. is the guarantor and takes full responsibility for the manuscript and its originality.

Author contributions: C.S.-L., A.B.-D., and R.A.D. researched data, contributed to discussions, and reviewed/edited the manuscript. C.H. and S.M.I. contributed to discussions and reviewed/edited the manuscript. R.S. and D.M.S. researched data, contributed to discussions, wrote the manuscript, and reviewed/edited the manuscript.

Disclosure Summary: The authors have nothing to disclose.

References

Author notes