Retinoids ameliorate insulin resistance in a leptin-dependent manner in mice


  • Potential conflict of interest: Nothing to report.


Transgenic mice expressing dominant-negative retinoic acid receptor (RAR) α specifically in the liver exhibit steatohepatitis, which leads to the development of liver tumors. Although the cause of steatohepatitis in these mice is unknown, diminished hepatic expression of insulin-like growth factor-1 suggests that insulin resistance may be involved. In the present study, we examined the effects of retinoids on insulin resistance in mice to gain further insight into the mechanisms responsible for this condition. Dietary administration of all-trans-retinoic acid (ATRA) significantly improved insulin sensitivity in C57BL/6J mice, which served as a model for high-fat, high-fructose diet–induced nonalcoholic fatty liver disease (NAFLD). The same effect was observed in genetically insulin-resistant KK-Ay mice, occurring in concert with activation of leptin-signaling pathway proteins, including signal transducer and activator of transcription 3 (STAT3) and Janus kinase 2. However, such an effect was not observed in leptin-deficient ob/ob mice. ATRA treatment significantly up-regulated leptin receptor (LEPR) expression in the livers of NAFLD mice. In agreement with these observations, in vitro experiments showed that in the presence of leptin, ATRA directly induced LEPR gene expression through RARα, resulting in enhancement of STAT3 and insulin-induced insulin receptor substrate 1 phosphorylation. A selective RARα/β agonist, Am80, also enhanced hepatic LEPR expression and STAT3 phosphorylation and ameliorated insulin resistance in KK-Ay mice. Conclusion: We discovered an unrecognized mechanism of retinoid action for the activation of hepatic leptin signaling, which resulted in enhanced insulin sensitivity in two mouse models of insulin resistance. Our data suggest that retinoids might have potential for treating NAFLD associated with insulin resistance. (HEPATOLOGY 2012)

Insulin resistance is as an important factor for the development of metabolic syndrome, obesity, type 2 diabetes, and nonalcoholic fatty liver disease (NAFLD).1 Hyperinsulinemia and hyperglycemia are frequently observed in patients with this disorder, reflecting impaired insulin sensitivity in muscle, adipose, and liver tissues. This symptomology is closely related to that of NAFLD. Because hyperinsulinemia and hyperglycemia are risk factors for the development of hepatocellular carcinoma, ameliorating insulin resistance is important not only for treating NAFLD, but also for preventing NAFLD-associated hepatocellular carcinoma.2

Although several mechanisms underlying insulin resistance have been proposed, leptin resistance has been established as a key mechanism.3, 4 Hyperleptinemia is also a characteristic feature of obesity, and is believed to be a consequence of leptin resistance in the central nervous system, where signal transducer and activator of transcription 3 (STAT3) and Janus kinase 2 (JAK2) signaling via the long isoform of the leptin receptor (LEPRb) lead to reduced food intake and increased energy expenditure.3, 4 The peripheral roles of leptin via the short LEPR isoforms (LEPRa, LEPRc, LEPRd, LEPRe, and LEPRf) remain to be clarified.5 Of interest is the abundant expression of LEPRa in peripheral tissues including the liver.6 However, studies have demonstrated the efficacy of leptin for treating hepatic steatosis and insulin resistance in patients with severe lipodystrophy7, 8 and its direct effect on hepatic insulin sensitivity mediated by adenosine monophosphate-activated protein kinase α2 and insulin receptor substrate-1 (IRS1).9-11 Moreover, leptin stimulation of the short LEPR isoform in db/db mice (genetically LEPRb-deficient) leads to STAT3 phosphorylation as a consequence of p38 mitogen-activated protein kinase activation, thereby resulting in enhanced muscular lipid oxidation.12 The pathophysiological relevance of STAT3 to hepatic insulin sensitivity has also received much attention. Thus, mice lacking liver-specific STAT3 expression exhibit severe insulin resistance associated with suppressed gluconeogenesis.13 Interleukin-6 inhibits the development of liver steatosis by signaling through the gp130-STAT3 pathway.14 Moreover, insulin acts in the brain to facilitate hepatic interleukin-6 production and thereby induces STAT3 activation, which leads to the suppression of hepatic gluconeogenesis.15, 16

All-trans-retinoic acid (ATRA) plays diverse physiological roles as a ligand for retinoic acid receptors (RARs).17 A synthetic retinoid, Am80, which is a more potent and selective RARα/β agonist than ATRA,18 is a new treatment for acute promyelocytic leukemia, even in patients who relapse after complete ATRA-induced remission.19 Am80 and ATRA prevent preadipocyte differentiation, and ATRA ameliorates insulin resistance by enhancing lipolysis in mature adipocytes, which results from the up-regulation and activation of peroxisome proliferator-activated receptor (PPAR) β via fatty acid binding protein 5.20, 21 We have shown that hepatic retinoid signaling is impaired in NAFLD patients, and ATRA signaling through RARα holds great potential for NAFLD treatment.22-24 Moreover, transgenic mice expressing dominant-negative RARα specifically in the liver developed steatohepatitis leading to the development of hepatocellular carcinoma and liver adenomas.24 Gene expression profile analysis of the liver demonstrated reduced levels of insulin-like growth factor-1 (IGF1), suggesting the possible involvement of insulin resistance.25 However, few studies have examined the effect of retinoids on insulin resistance in the liver. We investigated the effect of retinoids on the insulin and leptin-signaling pathways in the livers of insulin-resistant mice.


ATRA, all-trans-retinoic acid; DMSO, dimethyl sulfoxide; DR, direct repeat; HFHFr, high-fat, high-fructose; IGF, insulin-like growth factor; IGFBP2, insulin-like growth factor binding protein 2; IRS1, insulin receptor substrate-1; JAK2, Janus kinase 2; LEPR, leptin receptor; mRNA, messenger RNA; NAFLD, nonalcoholic fatty liver disease; PPAR, peroxisome proliferator-activated receptor; qPCR, quantitative real-time polymerase chain reaction; RAR, retinoic acid receptor; SOCS3, suppressor of cytokine signaling 3; SREBP1, sterol regulatory element-binding protein 1; STAT3, signal transducer and activator of transcription 3.

Materials and Methods

An expanded Materials and Methods section is provided in the Supporting Information.

Nutrients and Mice.

Normal (CE2) and high-fat, high-fructose (HFHFr; 35% fructose, 30% fat by weight) diets with and without 50 mg/kg ATRA, as well as the normal diet containing 20 mg/kg Am80 were purchased from Oriental Yeast (Tokyo, Japan). ATRA was purchased from Sigma (St. Louis, MO). Five-week-old male C57BL/6J or KK-Ay mice and B6.V-LepOb/LepOb (ob/ob) mice were purchased from CLEA Japan (Tokyo, Japan) and Charles River Laboratories Japan (Yokohama, Japan), respectively. The mice were housed under pathogen-free conditions and in a temperature-controlled room on a 12-hour light/dark cycle. Animals received humane care in accordance with study guidelines established by the Tottori University Subcommittee on Laboratory Animal Care.

Following acclimation for 1 week, KK-Ay and ob/ob mice were fed the normal and ATRA- or Am80-supplemented normal diets for 4 weeks. C57BL/6J mice were fed the HFHFr diet for 16 weeks and were divided randomly into two groups, after which they were then fed either the HFHFr diet or the ATRA-supplemented HFHFr diet for another 4 weeks. C57BL/6J mice fed the normal diet for 20 weeks served as controls.

Cell Culture.

The human hepatoma HepG2 cell line and the simian virus 40 temperature-sensitive large T antigen–immortalized mouse hepatocyte cell line TLR326 (Cell Resource Center for Biomedical Research, Tohoku University) were maintained in Dulbecco's modified Eagle's medium (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 10% fetal bovine serum (MBL, Nagoya, Japan), and L-glutamine. Cultures were grown at 37°C (HepG2) or 33°C (TLR3) in 5% CO2 in a humidified incubator.

Statistical Analysis.

All statistical comparisons were performed using the Student t test. P < 0.05 was considered statistically significant. All data are shown as the mean ± SD from 4-10 mice or independent experiments.


Enhancement of Leptin-Dependent Insulin Sensitivity in Mice Genetically Resistant to Insulin by ATRA.

Normal or ATRA-supplemented diets were given for 4 weeks to genetically insulin-resistant KK-Ay or ob/ob mice.27 Although both groups demonstrated similar daily consumption of both diets, ATRA significantly inhibited body weight gain in KK-Ay and ob/ob mice (Supporting Fig. 1A,B). In KK-Ay mice, the ATRA-supplemented diet significantly mitigated hyperglycemia, hyperinsulinemia, glucose intolerance (intraperitoneal glucose tolerance test), and insulin resistance (homeostatic model assessment of insulin resistance), as well as hyperleptinemia. Surprisingly, this diet did not affect these parameters, but rather, increased hyperglycemia in leptin-deficient ob/ob mice (Fig. 1A-E; Supporting Fig. 1C). Because leptin acts as an antidiabetic adipokine in rodents and humans, the increased level of circulating leptin frequently observed in obese patients is explained because of leptin resistance.3, 4 Thus, the decrease in serum leptin levels in KK-Ay mice indicates that ATRA reverses leptin resistance. Based on these observations, we postulated that ATRA might ameliorate insulin resistance in the liver in a leptin-dependent manner. To confirm this hypothesis, we examined the combined effects in vitro of ATRA and leptin on insulin-induced IRS1 phosphorylation. ATRA significantly enhanced insulin-induced IRS1 tyrosine phosphorylation in the presence of leptin (Fig. 1F, Supporting Fig. 2). These data suggest that leptin was required for improved sensitivity of the liver to insulin in response to ATRA.

Figure 1.

ATRA enhances leptin-dependent insulin sensitivity in genetically insulin-resistant mice. (A,B) Fasting blood glucose (A) and serum insulin (B) levels in KK-Ay or ob/ob mice after 4 weeks on a normal diet (white bars) or ATRA-supplemented diet (black bars). (C) Blood glucose kinetics after intraperitoneal glucose administration to KK-Ay (left panel) and ob/ob (right panel) mice after 4 weeks on a normal diet (white circles) or ATRA-supplemented diet (black circles). (D) Area under the curve (AUC) of kinetic profiles of blood glucose levels shown in (C). (E) Homeostatic model assessment of insulin resistance (HOMA-IR) indices in KK-Ay or ob/ob mice after 4 weeks on a normal diet (white bars) or ATRA-supplemented diet (black bars). *P < 0.05, **P < 0.01, ***P < 0.001 versus control. (F) IRS1 phosphorylation in HepG2 cells treated with ATRA (24 hours) and then leptin (3 hours) followed by stimulation with insulin for 2 minutes.

Effect of ATRA on Liver Histology of NAFLD Mice.

Rodents fed a diet enriched with fat and fructose exhibit pathological features of NAFLD.27 Accordingly, we confirmed that C57BL/6J mice developed liver steatosis associated with insulin resistance after being fed a HFHFr diet for 16 weeks (data not shown). Therefore, we fed the mice HFHFr or ATRA-supplemented HFHFr (ATRA + HFHFr) diets for another 4 weeks to investigate the effects of ATRA on NAFLD-associated insulin resistance (Fig. 2A).

Figure 2.

Effects of 4-week ATRA treatment on body weight and liver damage in mice fed an HFHFr diet. (A) Procedure for establishing a diet-induced mouse model of NAFLD. C57BL/6J mice were fed an HFHFr diet (30% fat, 35% fructose by weight) for 16 weeks. The mice were divided randomly into two groups and were further fed an HFHFr diet (HFHFr group) or 50 mg/kg ATRA-supplemented HFHFr diet (ATRA + HFHFr group) for another 4 weeks. C57BL/6J mice fed the normal diet for 20 weeks served as controls. (B) Overall average daily dietary consumption of control mice during the last 4 weeks of feeding normal diet (white bars) and of NAFLD mice during the last 4 weeks of an HFHFr diet (gray bars) or ATRA + HFHFr diet (black bars). (C) Body weight changes of control (white circles), HFHFr (gray circles), and ATRA + HFHFr (black circles) mice during the experimental period. After 16 weeks, NAFLD mice were divided randomly into HFHFr or ATRA + HFHFr groups. (D,E) Serum alanine aminotransferase (ALT) (D) and aspartate aminotransferase (AST) (E) activities of control mice (white bars) and NAFLD mice fed an HFHFr diet (gray bars) or ATRA + HFHFr diet (black bars) for 4 weeks. *P < 0.05, **P < 0.01, ***P < 0.001 versus control diet (white bars). #P < 0.05, ##P < 0.01, ###P < 0.001 between ATRA + HFHFr diet (black bars) versus HFHFr diet (gray bars).

Although daily consumption did not differ between mice fed the HFHFr and ATRA + HFHFr diets, consumption was significantly lower in these two groups than in the control group (Fig. 2B), indicating that ATRA does not induce an anorexigenic effect and that leptin therefore likely does not affect the central nervous system in NAFLD mice. ATRA significantly decreased body weight, liver-to-body weight ratio, visceral fat tissue weight, serum alanine aminotransferase and aspartate aminotransferase levels in NAFLD mice (Fig. 2C-E, Supporting Fig. 3A,B). In agreement with a significant decrease in hepatic lipid levels in the ATRA + HFHFr group as shown by biochemical assays, quantitative real-time polymerase chain reaction (qPCR) demonstrated that ATRA significantly up-regulated the lipolytic transcription factors PPARα and PPARβ, with concomitant down-regulation of lipogenic transcription factors, PPARγ and sterol regulatory element-binding protein 1 (SREBP1), and fatty acid synthase, as described24 (Fig. 3A, Supporting Fig. 3C-E). Note that the primers for detecting SREBP1 were designed to detect splicing variants SREBP1a and SREBP1c, both of which play central roles in regulating lipid synthesis, although each variant differs in some aspects.28 Moreover, ATRA significantly improved hepatic histology as indicated by decreased numbers and area of lipid droplets, and ballooned hepatocytes (Supporting Fig. 3F, Supporting Table 1). The histological study also revealed remarkably diminished periportal macrovesicular steatosis, suggesting enhanced lipid metabolism in the livers of the ATRA + HFHFr group. This finding was consistent with the expression of lipid metabolism-related genes (Fig 3A-E). However, the efficacy of ATRA for treating patients with nonalcoholic steatohepatitis remains to be determined, as the mice in our present study did not develop severe inflammation and fibrosis (Supporting Table 1).

Figure 3.

Effects of 4-week ATRA treatment on gene expression in mice fed an HFHFr diet. (A-E) Hepatic expression levels of PPARα (A), PPARβ (B), PPARγ (C), SREBP1 (D), and fatty acid synthase (FASN) (E) of control mice (white bars), NAFLD mice fed an HFHFr diet (gray bars), or NAFLD mice fed an ATRA-supplemented HFHFr diet (ATRA + HFHFr; black bars) for 4 weeks. *P < 0.05, **P < 0.01, ***P < 0.001 versus control. #P < 0.05, ##P < 0.01, ###P < 0.001 between ATRA + HFHFr versus HFHFr.

Improvement in Insulin Sensitivity by ATRA in NAFLD Mice.

We then explored the effect of ATRA on insulin resistance in a mouse model of HFHFr diet-induced NAFLD. As observed in KK-Ay mice, ATRA significantly normalized hyperglycemia, hyperinsulinemia, glucose intolerance, insulin resistance (homeostatic model assessment of insulin resistance and intraperitoneal insulin tolerance test), as well as hyperleptinemia (Fig. 4A-D, Supporting Fig. 4). In contrast to leptin, the HFHFr diet did not affect circulating levels of adiponectin, another antidiabetic adipokine, or tumor necrosis factor-α, which is known to impair insulin sensitivity (data not shown). IRS1 expression was significantly increased in association with increased tyrosine phosphorylation in ATRA + HFHFr group, compared with those in the HFHFr group (Fig. 4E, Supporting Fig. 5A,B). AKT phosphorylation in the ATRA + HFHFr group was slightly higher than that in the HFHFr group (control, 1.0 ± 0.15; HFHFr, 1.6 ± 0.29; ATRA + HFHFr, 2.1 ± 0.17), albeit not to a statistically significant degree. These results suggest that ATRA also normalizes insulin sensitivity in the diet-induced NAFLD mice, possibly by reversing leptin resistance.

Figure 4.

ATRA enhances insulin sensitivity and activation of the leptin-signaling pathway in a mouse model of diet-induced NAFLD. (A-E) Fasting blood glucose levels (A), serum insulin levels (B), intraperitoneal glucose tolerance test (area under the curve [AUC]) (C), intraperitoneal insulin tolerance test (time course) (D), and hepatic protein expression of insulin and leptin signaling molecules (E) in control (CTRL) mice (white bars) or NAFLD mice after 4 weeks on an HFHFr diet (gray bars) or ATRA-supplemented HFHFr diet (ATRA + HFHFr; black bars). Actin was used as an internal control. *P < 0.05, **P < 0.01, ***P < 0.001 versus control. #P < 0.05, ###P < 0.001 between ATRA + HFHFr versus HFHFr. (F) STAT3 phosphorylation in the livers of KK-Ay or ob/ob mice after 4 weeks of normal (CTRL) or ATRA-supplemented (ATRA) diets.

Association Between ATRA-Induced Improvement in Insulin Sensitivity and Activation of the Leptin-Signaling Pathway.

We examined retinoid signaling in the livers of NAFLD mice, in which ligand-dependent RAR-mediated transcriptional regulation plays a central role. Although hepatic RARα expression was not changed, expression of the RARα target gene Rarb was significantly down-regulated in HFHFr mice, whereas ATRA reversed this effect (Supporting Fig. 6A-C). This result was consistent with our previous observation that hepatic retinoid signaling is impaired in NAFLD patients22 and prompted us to perform transcriptome analysis using complementary DNA microarrays. By identifying genes with elevated expression in the ATRA + HFHFr group compared with the HFHFr group, four independent probes corresponding to Lepr were ranked as the highest 10 (Table 1). This finding was consistent with the leptin-dependent effect of ATRA on insulin sensitivity. qPCR confirmed significant up-regulation of LEPRa as well as IGF binding protein 2 (IGFBP2), which is expressed in the liver in response to systemic leptin administration29 (Supporting Fig. 6D,E). This finding suggests that the leptin-signaling pathway was activated in the livers of mice fed the ATRA + HFHFr diet.

Table 1. The 10 Most Highly Up-regulated Genes in the Livers of Mice Fed an HFHFr Diet After 4-Week ATRA Treatment
GeneGenBank Accession No.Probe Name*Relative Increase
  • Total RNAs from livers of five mice from each group were pooled and simultaneously analyzed via DNA microarray (Whole Mouse Genome Oligo Microarray 44Kx4 pack, Agilent Technologies). Gene expression levels were normalized to those of the control group.

  • Abbreviation: CTRL, control.

  • *

    Probe names were provided by Agilent Technologies for the Whole Mouse Genome Oligo Microarray 44Kx4 pack.


We next examined the expression of leptin-signaling proteins. Consistent with the qPCR data, Western blotting revealed that the expression of the short LEPR isoform was also significantly higher in the ATRA + HFHFr mice compared with that in the HFHFr group (Fig. 4E, Supporting Fig. 5C). Note that LEPRb protein expression was not detected in liver samples. Although the total JAK2 level was lower in HFHFr and ATRA + HFHFr group mice than in controls, its phosphorylation was significantly elevated in the latter group (Fig. 4E, Supporting Fig. 5D,E). Total and phosphorylated STAT3 expression were significantly up-regulated by ATRA, whereas suppressor of cytokine signaling 3 (SOCS3) was not (Fig. 4E, Supporting Fig. 5F-H). STAT3 activation in hepatocytes by ATRA was also demonstrated by immunohistochemistry, showing intense nuclear staining of STAT3 in hepatocytes throughout the liver lobule in the ATRA + HFHFr group, while STAT3 was distributed diffusely in the cytoplasm and nucleus of pericentral hepatocytes in the control and HFHFr groups (Supporting Fig. 7). No changes were observed in the levels of other leptin signaling molecules, adenosine monophosphate-activated protein kinase α, or extracellular signal-regulated kinases 1/29, 30 (data not shown). Interestingly, although a significant increase in LEPRa messenger RNA (mRNA) expression was observed in both KK-Ay and ob/ob mice, STAT3 phosphorylation and IGFBP2 mRNA up-regulation were also observed in the livers of KK-Ay mice fed the ATRA-supplemented normal diet but not in ob/ob mice refractory to ATRA-induced insulin sensitization (Fig. 4F, Supporting Fig. 1D,E). Microarray data for other STAT3-activating cytokines and growth factors and their receptors showed approximately 2.0- and 4.8-fold increases in average in leukemia inhibitory factor receptor (Lifr) and epidermal cell growth factor receptor (Egfr), respectively, in the ATRA + HFHFr group compared with the HFHFr group (Supporting Table 2). However, because ligands of these receptors exist in ob/ob mice, they were unlikely to be significantly involved in hepatic STAT3 activation. These results suggest that ATRA-induced up-regulation of the short LEPR isoform triggered the activation of leptin signaling, consequently leading to the reversal of leptin resistance.

Role of RARα in ATRA-Induced Reversal of Insulin Resistance.

We investigated the involvement of RARα in ATRA action. The Lepra mRNA level in the mouse hepatocyte cell line TLR326 increased as a function of ATRA treatment for up to 12 hours (Fig. 5A). Expression of the short LEPR isoform also increased in a dose-dependent manner at 24 hours (Fig. 5B). As observed in vivo, leptin-induced STAT3 phosphorylation was enhanced by the presence of ATRA (Fig. 5C), suggesting that ATRA-induced LEPRa expression was important for the activation of the hepatic leptin-signaling pathway.

Figure 5.

Leptin-signaling pathway activation in a hepatocyte cell line treated with ATRA. (A) Kinetics of ATRA-induced Lepra mRNA expression. The hepatocyte cell line TLR3 was treated with dimethyl sulfoxide (DMSO) (white circles) or 5 μM ATRA (black circles) for the indicated times. Expression levels were determined via qPCR. (B) Expression of the short LEPR isoform in TLR3 cells treated with 0, 2.5, 5, and 10 μM ATRA for 24 hours. (C) STAT3 activation in TLR3 cells pretreated with DMSO (control) or 5 μM ATRA for 48 hours, followed by stimulation with 100 ng/mL recombinant mouse leptin for 5 minutes. (D) Mouse Lepr promoter-driven luciferase constructs. Putative direct repeats (DRs) for the binding of RAR in the Lepr promoter region were predicted by the NHR-scan program and shown above. The numbers represent base pairs and correspond to the position relative to the transcription start site (+1). The luciferase gene in the pGL3 series is driven by the mouse Lepr promoter from nucleotide positions −2983 (pGL3 −3.0K containing DR1-1, DR1-2, DR3, and DR4), −2140 (pGL3 −2.1K containing DR1-2, DR3, and DR4), −1852 (pGL3 −1.9K containing DR3 and DR4), or −758 (pGL3 −0.8K containing DR4) to +44. The luciferase gene in the pTAL series is driven by a herpes simplex virus-derived TATA-like promoter conjugated with the DR1-2 sequence (enclosed in a square box, pTAL DR1-2) or an inverted DR1-2 sequence (enclosed in a square box, pTAL invDR1-2). (E) The mouse Lepr promoter activity in response to retinoids. At 24 hours after transfection with pGL3 series constructs, TLR3 cells were treated with DMSO (white bars), 5 μM ATRA (gray bars), or 10 μM Am80 (black bars) for an additional 24 hours. (F) Enhancer activity of DR1-2 derived from the mouse Lepr promoter. At 24 hours after transfection of pTAL DR1-2 or pTAL invDR1-2, TLR3 cells were treated with DMSO (white bars), 5 μM ATRA (gray bars), or 10 μM Am80 (black bars) for an additional 24 hours. *P < 0.05, **P < 0.01, ***P < 0.001 versus control (DMSO).

Nuclear hormone receptors, including RARs, bind to a conserved direct repeat (DR) element when they function as transcription factors. In silico analysis of the mouse Lepr promoter region using the NHR-scan program31 revealed four putative DR elements (DR1-1, DR1-2, DR3, and DR4) (Supporting Fig. 8A). A chromatin immunoprecipitation assay using anti-RARα antibody demonstrated that RARα constitutively occupied DR1-2 and, to a lesser extent, DR4, although there was slightly decreased RARα binding to DR1-2 in the presence of ATRA (Supporting Fig. 8B). This is in contrast to the retinoic acid response element of the cytochrome P450 26a1 promoter,32 where ATRA-induced recruitment of RARα was observed. We also performed luciferase reporter assays using Lepr promoter-driven luciferase constructs (Fig. 5D). The mouse Lepr promoter that included DR1-2 responded to ATRA and to the selective RARα/β agonist Am8018 (Fig. 5E). Moreover, DR1-2 enhanced the basal activity of a TATA-like promoter in the presence of retinoids, whereas the inverted DR1-2 sequence showed no such effect (Fig. 5F). Although the possibility of involvement of RARβ remains to be determined, we conclude that retinoids directly regulate Lepr transcription through RARs.

Am80 was shown to induce differentiation of acute promyelocytic leukemia cells with greater efficiency than ATRA.18, 19 Thus, the potential of clinical application of Am80 in the treatment of insulin resistance was evaluated in KK-Ay mice. In contrast to ATRA, mice fed an Am80-supplemented diet did not exhibit changes in whole body weight, daily food consumption, or hepatic lipid content (Supporting Fig. 9A-E). As expected, the Am80 diet significantly ameliorated insulin and leptin resistance (Fig. 6A,B, Supporting Fig. 9F,G). Moreover, Am80 induced significant increases in LEPRa and IGFBP2 mRNA levels and STAT3 phosphorylation in the liver, suggesting activation of hepatic leptin signaling (Fig. 6C-E). The Am80-induced LEPRa mRNA up-regulation was also shown in TLR3 cells in a dose-dependent manner (Fig. 6F).

Figure 6.

Am80-induced insulin sensitization and hepatic leptin-signaling pathway activation in KK-Ay mice. (A-D) Fasting blood glucose level (A), fasting serum insulin level (B), and hepatic LEPRa (C) and IGFBP2 (D) mRNA levels relative to β-actin in KK-Ay mice after 4 weeks on a normal diet (control; white bars) or Am80-supplemented diet (black bars). *P < 0.05 versus control. (E) STAT3 phosphorylation in the livers of KK-Ay mice after 4 weeks on a normal diet (control [CTRL]) or Am80-supplemented diet (Am80). (F) Am80-induced Lepra mRNA expression in the hepatocyte cell line TLR3. Cells were treated with 0 (control; white bar), 2.5 (light gray bar), 5 (gray bar), or 10 (black bar) μM Am80 for 12 hours. Expression levels were determined via qPCR. **P < 0.01 versus control.


This study implicates Lepr as a target of retinoids, suggesting that the mechanism underlying retinoid-induced hepatic insulin sensitization involves the activation of the leptin signaling pathway by increased LEPR expression in the liver (Fig. 7). This hypothesis is strongly supported by our observations that leptin-deficient ob/ob mice were refractory to ATRA-induced STAT3 activation, IGFBP2 expression, and insulin sensitization even though hepatic LEPRa expression was increased. Moreover, homeostatic model assessment of insulin resistance and in vitro data indicated that retinoid-induced activation of the leptin signaling pathway resulted in hepatic insulin sensitization, although this requires verification by clamp assays or other techniques. In the present study, we demonstrated that retinoids have potential for treating diseases related to insulin resistance, which is a fundamental feature of metabolic syndrome, obesity, type 2 diabetes, and NAFLD.

Figure 7.

Model describing retinoid-induced hepatic insulin sensitization. Retinoids transcriptionally induce LEPR expression, which then triggers the activation of STAT3 and IRS1, leading to increased insulin sensitivity.

Researchers are actively investigating the precise functions of leptin in peripheral tissues, including the liver. Leptin is involved in a number of physiological processes, from energy homeostasis to reproduction, immunity, and bone metabolism.3, 4 Leptin exerts its pleiotropic functions primarily as a result of the ubiquitous expression of its receptor, LEPR.6 Leptin resistance was introduced as a concept to explain the high frequency of hyperleptinemia in most obese patients.3, 4 However, the molecular causes and pathological consequences of leptin resistance are not fully understood. Increased endoplasmic reticulum stress is known to inhibit leptin signaling in the central nervous system, thereby resulting in insulin resistance in mice fed a high-fat diet.33 The increased expression of SOCS3 and SH2 domain-containing protein tyrosine phosphatase-2 also abrogates the leptin signaling pathway and decreases insulin sensitivity.34, 35 Moreover, genetic leptin deficiency as occurs in ob/ob mice and in patients with lipodystrophy results in severe insulin resistance, which can be reversed by leptin replacement therapy.7, 8 These results suggest that insulin resistance may be a consequence of leptin resistance. In contrast, one mechanism postulated to explain leptin resistance is the down-regulation of central LEPR expression.36 Although it is not known whether a similar mechanism plays a role in peripheral leptin resistance, we and other investigators have demonstrated the down-regulation of hepatic LEPR expression in diet-induced obese, hyperleptinemic animals.10 Because leptin itself possesses insulin-sensitizing activity,7, 8 the normalization of leptin resistance provides some benefits in treating insulin resistance. However, a clinical study revealed that leptin administration achieved only modest body weight and fat loss in obese patients with hyperleptinemia, advocating the requirement of a leptin sensitizer for enhancing the efficacy of leptin therapy.4, 33, 37 Our data suggest that retinoids might act as a promising leptin sensitizer by restoring hepatic LEPR expression. Future study should examine the effect of retinoids in db/db mice, which genetically lack only LEPRb but express other LEPR isoforms that function in peripheral tissues as well as the liver. However, the relatively highly phosphorylated STAT3 levels in the HFHFr and ATRA + HFHFr groups despite reduced JAK2 phosphorylation and LEPR levels suggest additional mechanisms underlying leptin resistance. Since no difference in SOCS3 expression was observed in the present study, other negative regulators might be involved in this discordance. Microarray data demonstrated that the expression of SH2 domain-containing protein tyrosine phosphatase-2, the hepatocyte-specific deletion of which leads to enhanced and prolonged STAT3 phosphorylation,35 was decreased in the HFHFr and ATRA + HFHFr groups compared with the control group. qPCR also confirmed 1.7- and 2.9-fold down-regulation of SH2 domain-containing protein tyrosine phosphatase-2 in the HFHFr and ATRA + HFHFr groups, respectively (both P < 0.05, compared to the control). Further investigation is necessary to elucidate additional involvement of negative regulators in hepatic leptin resistance.

STAT3 has recently emerged as an important regulator for hepatic gluconeogenesis given its activity to suppress the expression of PPARγ-coactivator 1α, glucose-6-phosphatase and phosphoenolpyruvate carboxykinase 1.13-15 Mice genetically deficient in hepatic STAT3 activation exhibit severe steatosis, hyperinsulinemia, and glucose intolerance when fed a choline-deficient diet.13, 14 In addition to STAT3, the gene encoding IGFBP2 is a target of the leptin signaling pathway in the liver, and plays an important role in leptin's antidiabetic activity.29 Mice with hepatic RARα deficiency exhibit steatohepatitis associated with reduced expression of IGF1, which reduces blood glucose level by acting as an anabolic and metabolic hormone.24, 25 Studies have proposed that IGFBP2 enhances the stability of IGF138; although the expression of IGF1 was not changed in the present study (data not shown), it is still possible that the enhancing effect of IGFBP2 contributes to ATRA action. Interestingly, IGFBP2 administration has been found to mitigate glucose intolerance and hyperinsulinemia not only in ob/ob mice, but also in leptin-resistant mice.29 Taken together, these present and previous findings suggest that either or both STAT3 and IGFBP2 may play a role in retinoid action, at least in part.

Consumption of high-fructose-containing foods is reported to be a risk factor for the development of NAFLD.39, 40 Excessive acetyl coenzyme A production by fructose metabolism leads to increased body fat storage, resulting in the development of insulin resistance.40, 41 In the present study, we induced NAFLD in mice by feeding them an HFHFr diet. ATRA treatment improved their hepatic steatosis and insulin sensitivity and ameliorated liver damage. Expression analysis of genes involved in hepatic lipid metabolism suggested that ATRA enhanced lipolysis and suppressed lipogenesis, possibly contributing to the improvement in hepatic histology. These results are consistent with published observations in transgenic mice expressing liver-specific dominant-negative RARα.24 The slight discrepancy observed in the extent of lipid accumulation as assessed by biochemical or histological assays of mice fed the HFHFr diet implies that other lipids such as diacylglycerols and glycerophospholipids may be involved in the formation of lipid droplets in murine steatosis models.42 Further research will be required to investigate the effect of retinoids on hepatic lipid metabolism in more detail, using techniques such as lipid profiling.42 These data suggest that retinoids have potential for use in treating NAFLD by improving hepatic lipid metabolism and ameliorating insulin resistance.

In contrast to findings in the liver, ATRA treatment modestly decreased visceral fat mass. Others have reported that Am80 and ATRA prevent preadipocyte differentiation.20, 21 In particular, ATRA enhances lipolysis in mature adipocytes by inducing and activating PPARβ, leading to decreased white adipose tissue mass in diet-induced obese mice.21 Moreover, upon the activation of PPARβ, ATRA requires fatty acid binding protein 5, whose expression is undetectable in the liver.21, 43 However, we were unable to detect a difference in the expression of PPARβ between the visceral adipose tissues from the HFHFr and ATRA + HFHFr groups (data not shown). In contrast, because increased expression of PPARβ was observed in the livers of the ATRA + HFHFr group, hepatic PPARβ may be indirectly involved in ATRA-induced improvement of fatty liver. Considering that the RARα/β selective agonist Am80 affected hyperglycemia and hyperinsulinemia with an efficacy similar to that of ATRA, RAR-independent mechanisms are unlikely to be involved in the retinoid-induced insulin sensitization, at least under the conditions of present study.

In conclusion, to our knowledge, the previously unrecognized actions of retinoids described here provide valuable insight into understanding the physiological function of leptin in the liver. Our results strongly suggest that ligand-dependent RAR activation could benefit insulin-resistant patients.