A Small Molecule, UAB126, Reverses Diet-Induced Obesity and its Associated Metabolic Disorders

Obesity in the U.S. is an epidemic, affecting more than one-third of American adults. Obesity elevates risk for a wide range of health problems, including diabetes and cardiovascular diseases (1). Modulation of nuclear receptors (NRs) has been suggested as one of the most attractive therapeutic strategies against obesity-associated cardiometabolic diseases (2). Targeting retinoid X receptor (RXR) may be especially useful to modulate multiple NRs simultaneously because RXR dimerizes with other NRs, including peroxisome proliferator–activated receptors (PPARs), liver X receptor (LXR), retinoic acid receptor, and farnesoid X receptor (2). Moreover, these NRs regulate carbohydrate and lipid metabolism (2). There are three RXR isotypes—RXRα, -β, and -γ (NR2B1–3)—in mammals that are encoded by distinct genes (3). These isotypes of RXR are differentially expressed in various tissues. RXRα is predominantly expressed in mouse liver, while RXRβ is preferentially expressed in brain (4). RXRγ is expressed in skeletal and cardiac muscle (4). 9-Cis-retinoic acid is a potent ligand for RXR in vitro, but has not been found in vivo (5). Thus, RXR is an orphan NR. Several compounds that selectively bind to RXR (rexinoid) have been developed and have been shown to have glucose-lowering, insulin-sensitizing, and antiobesity effects (6–8). However, most of these rexinoids exhibit serious side effects, including hypertriglyceridemia, hepatomegaly, and alterations of the thyroid hormone axis (6). These side effects may be attributed to the potent and nonselective stimulation of RXR (9). Therefore, optimal modulation of RXR activities may advance the therapeutic potential of rexinoids. Using our understanding of the structural interactions between rexinoids and RXR, we designed and synthesized a novel rexinoid-like small molecule, UAB126, from already known rexinoid UAB30 (10). In this study, we introduce a novel compound, UAB126, which reduces obesity, insulin resistance, and dyslipidemia without changes in thyroid hormone level (6).

UAB126 was synthesized in two steps starting from naphthalene boronic acid using Suzuki reaction conditions. Coupling of boronic acid 1 with p-bromobenzaldehyde in the presence of Tetrakis palladium gave the desired aldehyde 3 in 70% yield after purification by chromatography. Knoevenagel-type condensation of aldehyde 3 with malonic acid provided the UAB126 as a single isomer. Purity of the UAB126 was assessed by high-performance liquid chromatography (HPLC) analysis using the Agilent 1260 Infinity II instrument. HPLC analysis was performed using an Agilent Technologies column (poroshell 120, C18, 2.7 μm; 4.6 × 100 mm). The eluent was 65% acetonitrile/35% water and 0.1% formic acid with a flow rate of 1 mL/min (isocratic). Fluorescence (excitation 230, emission 460) and diode array detectors were used to detect the compound.

Six-week-old male C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and acclimated for 1 week before the experiments. UAB126 was dissolved in 1:9 ethanol/PEG400, and then vehicle or UAB126 (100 mg/kg body weight/day) was administered orally by gavage for 5 days. One hour after the final dose, serum was collected and subjected to HPLC analysis. A total of 100 µL of serum samples was diluted with HCl (0.5N, 100 µL) and 0.5 mL of ethyl acetate. Samples were vortexed and centrifuged (10,000 rpm for 10 min), and the supernatant liquid was separated and evaporated to dryness under a stream of nitrogen. Finally, the residue was dissolved in 0.5 mL of acetonitrile. Blank serum samples were also analyzed following the above conditions. Vehicle-treated serum samples spiked with UAB126 prior to extraction were used as a positive control for detection of UAB126.

All animal procedures were performed in accordance with the rules of and approved by the Animal Use and Care Committee at the University of Alabama at Birmingham (UAB). C57BL/6J mice were purchased from The Jackson Laboratory. All animals were maintained in a temperature-controlled facility with a 12:12-h light/dark cycle. Mice were fed low-fat diet (LFD) (10% calories from fat, D132021801; Research Diets, Inc., New Brunswick, NJ) or high-fat diet (HFD) (60% calories from fat, D13021802; Research Diets, Inc.). Diets were mixed without or with UAB126 (1.2 g/kg of diet). Mice had free access to food and water.

These experiments were conducted by the UAB Small Animal Phenotyping Core facility funded by our Nutrition Obesity Research Center. In vivo body composition (total body fat and lean tissue) of mice was determined using an EchoMRI 3-in-1 quantitative magnetic resonance machine (Echo Medical Systems, Houston, TX). Quantification of results was standardized by conducting a system test using a known fat standard prior to experimental measurements being taken.

Intraperitoneal glucose tolerance test (GTT) and insulin tolerance test (ITT) were performed after 6 h fasting. During GTT, tail-vein blood glucose level was determined at 0, 15, 30, 60, and 120 min with a hand-held glucometer (FreeStyle; Abbott, Abbott Park, IL) after intraperitoneal injection of 2 g/kg body weight glucose. During ITT, glucose level was determined 0, 15, 30, 60, and 90 min after intraperitoneal injection of 0.5 units/kg body weight insulin.

Respiration rate, food intake, energy expenditure, and physical activity were accurately quantified as previously described using the Comprehensive Laboratory Animal Monitoring System (Columbus Instruments Inc., Columbus, OH) (11). Respiratory exchange ratio (RER) was calculated as CO₂ generation/O₂ consumption.

Mice were fasted overnight, and the blood was collected during euthanization. The serum was collected after centrifugation (2,200g for 10 min at 4°C) of clotted blood. Serum insulin level was assessed by an ultrasensitive mouse insulin ELISA (Crystal Chem Inc., Elk Grove Village, IL). Free fatty acids were assessed using the NEFA-HR(2) kit (Wako, Richmond, VA). Triglyceride (TG) levels were assessed using Pointe Scientific Inc. (Pittsburgh, PA) TG liquid reagents. Cholesterol level was assessed using Thermo Fisher Scientific total cholesterol reagents (Pittsburgh, PA). The ELISA Multiplex kit (Millipore Sigma-Aldrich, St. Louis, MO) was used according to the manufacturer’s instructions to measure triiodothyronine (T3) and thyroid-stimulating hormone (TSH) levels in serum. HOMA of insulin resistance (HOMA-IR) and QUICKI were calculated as previously described (12).

Mice were fasted for 6 h, and insulin (1 unit/kg) was injected intraperitoneally 10 min before euthanasia. Skeletal muscle, liver, and epididymal adipose tissue were collected and frozen in liquid nitrogen until used. Tissue homogenates were prepared in the cell lysis buffer and processed according to the manufacturer’s instructions for a TissueLyser II (Qiagen, Germantown, MD). Cell lysate was subjected to an immunoblotting with antibodies as described previously (13) and visualized and quantified using the ChemiDoc imaging system and Image Lab 5.0 software (Bio-Rad Laboratories).

Tissues were excised at the time of sacrifice and then fixed in 10% formaldehyde/PBS solution overnight. Paraffin blocks were prepared by the UAB Pathology Core Research Laboratory. Tissue sections were deparaffinized, hydrated, and then treated at 95°C to retrieve antigens in a plastic slide jar containing 10 mmol/L citrate buffer (pH 6.0) for 20 min. The sections were immersed in 0.3% hydrogen peroxide for 30 min, blocked with ABC blocking serum (catalog number PK-6105; Vector Laboratories, Burlingame, CA) for 1 h, and then incubated overnight with an anti-FGF21 antibody (catalog number AF3057; R&D Systems, Minneapolis, MN). The tissue sections were processed using a Vectastain Elite ABC Kit (Vector Laboratories) according to the manufacturer’s instructions to visualize the staining of FGF21. In some cases, the tissue sections were then stained with hematoxylin and eosin and then mounted with Permount. The slides were photographed and analyzed by light microscopy.

Tissues were collected, snap-frozen in liquid nitrogen, and then stored until subjected to an analysis. Total RNA was prepared by homogenizing tissues in TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. One microgram of total RNA was used for cDNA synthesis by using an Omniscript RT kit (Qiagen). Then, the cDNA was subjected to real-time quantitative PCR (qPCR) analysis by using iQ SYBR Green Supermix (Bio-Rad Laboratories). Each gene expression was normalized to β-actin or cyclophilin.

Livers were isolated from HFD- and HFD plus UAB126–treated mice after a 12-week feeding (n = 2). Total RNA was isolated as described above and then subjected to an RNA-sequencing (RNA-seq) analysis. RNA-seq analysis was performed by the UAB Genomics Core facility. NextSeq500 was used for the analysis. A total of 1,334 genes were selected for the over ±2-fold differential expression. These genes were subjected to an online gene enrichment tool, Enrichr (14,15), and Ingenuity.

Stromal vascular cells (SVCs) were isolated from epididymal white adipose tissues (WATs) as described previously (16). Briefly, epididymal fat pads were harvested from mice and chopped into small pieces in PBS containing 2% BSA, and then collagenase type 2 and DNase I were added to the tissue suspension. After the suspension was incubated for 20 min at 37°C with shaking, the cells were filtered through a 100-μm nylon mesh and centrifuged at 1,000 rpm for 5 min. Five minutes before completion of the digestion, 0.1 mol/L EDTA was added. The red blood cells were lysed, and single cells were resuspended in DMEM containing 10% FBS and incubated for 4 h at 37°C with BD GolgiPlug (1:1,000) to measure intracellular cytokine synthesis. Following the 4-h incubation, cells were resuspended in FACS buffer, treated with Fc block (#553142; BD Biosciences), and then stained with fluorochrome-conjugated antibodies specific for Live/Dead (Thermo Fisher Scientific), indicated surface markers, and rat IgG1 isotype controls. Cells were analyzed with the Attune NxT Flow Cytometer (Thermo Fisher Scientific) with ∼1,000,000 events collected for each sample and analyzed with FlowJo (10.6.2) software (BD Biosciences). The flow cytometry gating scheme is shown in Supplementary Fig. 3.

Brown adipocyte differentiation of SVCs from inguinal adipose tissue was performed as described previously (17). The cells were incubated with induction media containing dexamethasone (2 μg/mL), 3-isobutyl-1-methylzanthine (0.5 mmol/L), insulin (5 μg/mL), and T3 (1 nmol/L) for 2 days and then with maintenance media containing insulin (5 μg/mL) and T3 (1 nmol/L) for an additional 6 days. The maintenance media was changed every 2 days.

Statistical significance was assessed by two-tailed Student t test or one-way ANOVA followed by Newman-Keuls or two-way ANOVA by Tukey or Bonferroni post hoc tests. Differences were considered to be statistically significant if the P value was <0.05. Statistical analyses were performed using Prism Version 8 (GraphPad Software, San Diego, CA).

All primary data and animal models used in the study are available to investigators upon reasonable request.

We designed and synthesized UAB126 from a known rexinoid, UAB30 (10,18) (Fig. 1A and B). The purity of the UAB126 was evaluated by HPLC analysis. We confirmed that UAB126 was >99% pure (Fig. 1C). To determine whether UAB126 has an effect on body weight, 6-week-old male C57BL/6J mice were fed LFD or HFD with or without UAB126 (1.2 mg/g diet). Mice fed HFD with UAB126 gained significantly less body weight compared with the mice fed HFD without UAB126 (Fig. 2A). Attenuation of HFD-induced weight gain was mainly due to reduction in fat mass (45% reduction) (Fig. 2B), but not in lean mass (Fig. 2C). To examine whether UAB126 alters energy metabolism and fuel source utilization, metabolic parameters were evaluated by indirect calorimetry. UAB126-treated mice displayed increased fat oxidation rate (indicated by decreased RER in HFD-fed mice (Fig. 2D). The effect of UAB126 on metabolic rate in HFD-fed mice was manifested by the enhanced oxygen consumption (VO₂) and carbon dioxide production (VCO₂) (Fig. 2E and F), whereas no difference in food intake or physical activity was observed in response to UAB126 supplementation (Fig. 2G and H). While UAB126 did not affect RER in LFD-fed mice (Supplementary Fig. 1A), VO₂ and VCO₂ were still increased in LFD-fed mice (Supplementary Fig. 1B and C). To understand the mechanism by which UAB126 regulates metabolic rate, we examined whether UAB126 regulates thermogenic genes in brown adipose tissue (BAT). UAB126 reduced the lipid accumulation in BAT (Fig. 2I) and increased the gene expression of Ucp1, Cytb, Cpt1α, Cpt1β, Cidea, Elov3, and Prdm16 that are associated with thermogenesis and fatty acid oxidation (Fig. 2J). Furthermore, UAB126 did not affect the circulating thyroid hormone level (Fig. 2K). The results suggest that UAB126 prevents HFD-induced weight gain by increasing fat utilization without altering food intake, physical activity, and T3 level.

The size of adipocytes in epididymal fat pad was decreased when UAB126 was supplemented (Fig. 3A). Consistent with the reduced size of WAT, treatment with UAB126 increased the expression of genes associated with lipolysis (hormone-sensitive lipase), insulin sensitivity (adiponectin), and fatty acid oxidation (PPARα and acyl coenzyme oxidase) in the WAT (Fig. 3B). Increased inflammatory response and macrophages in WAT are the typical feature of obesity (19). As such, adipose tissue crownlike structure and the expression of proinflammatory genes were increased in HFD-fed obese mice compared with those in LFD-fed mice (Fig. 3C). These chronic inflammatory markers were reduced when UAB126 was supplemented in HFD-fed mice (Fig. 3C and D). To confirm the anti-inflammatory effect of UAB126 in adipose tissue, FACS analyses were performed on stromal vascular cells isolated from epididymal adipose tissue (gating strategy is shown in Supplementary Fig. 2, and the representative FACS plots are shown in Supplementary Fig. 3). The populations of myeloid cells, monocytes, and macrophages were increased in HFD-fed mice, whereas they were decreased when UAB126 was supplemented (Fig. 3E–G). Furthermore, the population of proinflammatory myeloid cells, TNF-α–positive CD11b⁺ cells, was increased in HFD-fed mice, but it was decreased when UAB126 was added to the HFD (Fig. 3H). The populations of MHC class II–positive macrophages and TNF-α–positive M1 macrophages were increased in HFD-fed mice, whereas they were decreased when UAB126 was supplemented in HFD-fed mice (Fig. 3I and J). In contrast, UAB126 did not affect the populations of myeloid cells in LFD-fed mice. These suggest that UAB126 stimulates lipolysis and decreases proinflammatory response in WAT.

Next, we examined whether UAB126 prevents HFD-induced glucose intolerance. UAB126 improved glucose tolerance in HFD-fed mice only (Fig. 4A), while UAB126 improved insulin tolerance in both LFD-fed and HFD-fed mice (Fig. 4B). UAB126 lowered fasting glucose levels only in HFD-fed mice (Fig. 4C), while UAB126 lowered fasting insulin levels in both HFD-fed and LFD-fed mice (Fig. 4D). HOMA-IR (Fig. 4E) indicates that insulin sensitivity was improved in the UAB126-treated groups compared with the untreated groups. UAB126 enhanced the phosphorylation of insulin signaling pathway molecules, including insulin receptor substrate-1 (IRS-1), protein kinase B (Akt), glycogen synthase kinase 3β (GSK3β), and p70S6 kinase (p70S6K) in skeletal muscle, liver, and WAT in HFD-fed mice (Fig. 4F–V). While UAB126 also enhanced insulin signaling in skeletal muscle (Fig. 4F–K), its effect was only modest in liver (Fig. 4L–Q) and WAT (Fig. 4R–V) of LFD-fed mice. Interestingly, the protein level of IRS-1 in liver was increased by UAB126 in both LFD- and HFD-fed mice (Fig. 4Q). Taken together, these data suggest that UAB126 enhances insulin sensitivity in metabolic tissues and improves glucose homeostasis.

Nonalcoholic fatty liver disease is prevalent in obesity and type 2 diabetes (20). We examined whether UAB126 has an effect on HFD-induced hepatic steatosis. Analysis of liver histology and TG content revealed that supplementation with UAB126 decreased TG accumulation in the liver tissue (Fig. 5A and B). Moreover, UAB126 lowered the circulating TG, nonesterified fatty acids (NEFA), and cholesterol levels (Fig. 5C–E). The lipid-lowering effects of UAB126 were consistent with the decreased lipogenic gene expression in the liver (Fig. 5F). Thus, our data suggest that UAB126 may prevent the HFD-induced nonalcoholic fatty liver disease by modulation of genes that are involved in lipid metabolism.

To further examine the hepatic gene expression regulated by UAB126, we performed RNA-seq analysis on the livers isolated from mice fed HFD with or without UAB126. Because UAB126 was developed as a potential rexinoid, we investigated whether UAB126 can affect the genes that are regulated by RXR and other NRs that heterodimerize with RXR. The genes that were increased or decreased more than twofold by UAB126 were selected for Ingenuity Pathway and Enrichr (https://amp.pharm.mssm.edu/Enrichr) chromatin immunoprecipitation enrichment analysis (14,15). Chromatin immunoprecipitation enrichment analysis demonstrated that RXR, LXR, and PPARα are the major transcription factors involved in UAB126-mediated hepatic gene regulation. Gene enrichment analysis showed that a total of 342 genes were regulated by RXR, LXR, or PPARα (Fig. 6A). Among the 342 genes, 58 genes were commonly regulated by all three NRs (RXR, LXR. and PPARα). The enriched genes regulated by RXR were subjected to a gene ontology test, which indicated that the RXR target genes regulated by UAB126 in the liver were involved in lipid and glucose homeostasis, hepatic steatosis, and hepatokines implicated in adipocyte browning (14,15) (Table 1). The genes enriched for mammalian metabolic phenotypes are demonstrated by a heat map (Fig. 6B). Interestingly, fibroblast growth factor 21 (Fgf21) was the most common gene that appeared in the top selected categories. Thus, FGF21 may be an important mediator of metabolic effects of UAB126. Next, we examined whether UAB126 regulates FGF21 expression. Treatment with UAB126 elevated the expression of Fgf21 protein and mRNA in the liver (Fig. 6C and D) and the circulating levels of serum FGF21 (Fig. 6E). The results suggest that UAB126 regulates genes that are involved in glucose and lipid metabolism by modulation of NRs.

To test the therapeutic ability of UAB126, we examined whether UAB126 can reverse obesity, insulin resistance, and hyperlipidemia. Obesity was induced by feeding mice with HFD for 12 weeks prior to administration of UAB126, and then the mice were treated with or without UAB126. Treatment with UAB126 induced weight loss (Fig. 7A and B). UAB126 improved glucose and insulin tolerance (Fig. 7C–F) as well as reduced fasting serum glucose and insulin levels (Fig. 7G and H). An index of insulin sensitivity, HOMA-IR, was also significantly improved by UAB126 (Fig. 7I). Furthermore, UAB126 lowered circulating and hepatic TG levels (Fig. 7J and K) and circulating cholesterol level (Fig. 7L). UAB126 did not alter the serum T3 and TSH levels (Fig. 7M and N). These data suggest that UAB126 provides a promising therapeutic approach in the settings of obesity with insulin resistance and hyperlipidemia.

To further analyze the molecular mechanisms by which UAB126 may confer its beneficial metabolic effects, we tested whether it also regulated gene transcription in vitro. Interestingly, we found that while UAB30, a proven rexinoid (10), induced RXR element–driven luciferase activity (21) and endogenous Ucp-1 gene expression in cell culture (22), UAB126 did not (Supplementary Fig. 4A and B). Together with the pronounced effects observed in vivo, this raised the possibility that UAB126 has to be metabolized in vivo for its activity. Therefore, we analyzed the UAB126-treated serum samples using HPLC. Indeed, HPLC spectra demonstrated that the serum of vehicle-treated mice spiked with UAB126 showed the same profile with a peak at 3.96 min (Supplementary Fig. 4C) as that of the purified compound (Fig. 1). In contrast, the serum from UAB126-treated mice showed a shifted profile with the major peak at 3.24 min (Supplementary Fig. 4D). These results suggest that UAB126 is metabolized in vivo and may act as a prodrug that must be metabolized either by intestinal or hepatic enzymes to give rise to an active compound.

Nuclear receptors play important roles in energy homeostasis, glucose and lipid metabolism, and nutrient absorption (2). Although RXR can function as a homodimer, it heterodimerizes with other NRs and exhibits diverse physiological and pathophysiological roles (2,23). The known ligand for RXR is 9-cis retinoic acid, a metabolite of vitamin A. However, it is controversial whether 9-cis retinoic acid is an endogenous ligand or not because 9-cis retinoic acid is not found in vivo (24–26). Due to the important roles of RXR in metabolism, several synthetic RXR ligands (rexinoids) have been developed and suggested as treatments for diabetes (6,27,28). However, most of the previously developed rexinoids have side effects, including elevation of TGs, hepatomegaly, and an alteration of the thyroid hormone axis (6). In the current study, we report a novel rexinoid-like small molecule (Fig. 1) that is a potential treatment for metabolic disease, including obesity, insulin resistance, hyperglycemia, and hyperlipidemia without causing hypertriglyceridemia or noticeable adverse effects on the thyroid hormone axis.

Previously known rexinoids such as LG1069 (bexarotene), LG100268, AGN194204, and LG101506 do not have an effect on body weight (6), but demonstrate glucose-lowering effects (6,27,29,30). One study demonstrated that administrations of LG100268 by gavage or intracerebral ventricular injection decrease food consumption and body weight (27,31–33). Although we have not tested the effects of UAB126 on the central nervous system, UAB126 does not affect food intake or physical activity (Fig. 2G and H). Instead, UAB126 stimulated fat utilization (Fig. 2D) and increased metabolic rate (Fig. 2E and F). Thus, the mechanism by which UAB126 regulates body weight seems to be distinct compared with that of the known rexinoids. Interestingly, UAB126 stimulated the expression of genes that are involved in thermogenesis in BAT (Fig. 2J) and lipolysis in WAT (Fig. 3B). The increased fat utilization in UAB126-treated animals may be due to FGF21, which is a known factor that increases fat oxidation (34,35). Previous reports have shown that the promoter of Fgf21 is directly regulated by RXR-PPARα or RXR-LXR heterodimers (36,37). Thus, it is likely that as a rexinoid-like molecule, UAB126 regulates Fgf21 expression via RXR. Furthermore, our RNA-seq analysis supports this idea (Fig. 6). However, our mechanistic studies revealed that unlike UAB30, a proven rexinoid (10), UAB126 was not active in vitro and metabolized in vivo (Supplementary Fig. 4). This suggests that UAB126 acts as a prodrug and must be metabolized in vivo to exert its metabolic functions. In fact, this feature may have contributed to the favorable safety profile observed with UAB126 as opposed to other rexinoids. Obviously, future studies will have to define the exact metabolites of UAB126 and their exact functions.

Although liver is the primary contributor to the circulating FGF21 levels (38), other tissues such as adipose tissue and skeletal muscle may also contribute to the serum FGF21 levels (39,40). However, Fgf21 gene expression was not stimulated by UAB126 in other tissues (data not shown). The data suggest that UAB126 regulates Fgf21 gene expression in a tissue-specific manner possibly by differential recruitment of coactivators to the transcription sites. This hypothesis warrants further studies regarding tissue-specific UAB126 activities.

Because RXR plays important roles in insulin sensitivity, cholesterol metabolism, and macrophage differentiation, UAB126 may also modulate macrophage differentiation and affect the microenvironment of WAT (16,41,42). Indeed, UAB126 increases lipolysis and suppresses proinflammatory responses in WAT (Fig. 3).

UAB126 significantly improved glucose tolerance and insulin sensitivity, although the effects on glucose tolerance seemed to be milder. This could be explained by an incomplete suppression of hepatic glucose production as a result of the dramatically decreased insulin levels observed in UAB126-treated mice (Fig. 4D) in response to the increased insulin sensitivity in skeletal muscle (Fig. 4F–K). This is also consistent with the data showing that reduction of fasting insulin (Fig. 4D) by UAB126 is greater than fasting glucose (Fig. 4C). Another possibility is that UAB126 increases fat oxidation (Fig. 2D), so the fuel source utilization favors fat rather than glucose independent of insulin sensitivity.

The biggest drawback from using potent RXR agonists as treatments for metabolic diseases is that they cause central hypothyroidism in humans and rodents (43,44). Potent RXR agonists decrease TSH and T3 levels, and RXR directly suppresses thyrotropes, which is also related to the increased lipid levels (45). By contrast, UAB126 did not alter the levels of T3 (Fig. 7M) and TSH (Fig. 7N). Despite unaltered thyroid hormone levels, UAB126 is able to decrease TG and cholesterol levels. Moreover, RNA-seq analysis does not indicate any thyroid hormone receptor activity by UAB126. Thus, it is unlikely that thyroid hormone mediates the weight-loss effect of UAB126. Thus, FGF21 seems to be the major contributor in reduction of body weight by activation of BAT and normalization of insulin sensitivity in skeletal muscle without affecting thyroid hormone axis.

UAB126 prevents fatty liver and lowers circulating lipid levels. The lipid-lowering effect of UAB126 is one of the prominent features of UAB126, because one of the well-known side effects of other potent RXR agonists is hypertriglyceridemia (32,45). We observed that UAB126 suppresses hepatic lipogenic genes, including Srebp-1c, Pparγ2, and Scd1 (Fig. 5F). In contrast, other rexinoids such as bexarotene increase circulating TG and cause fatty liver by activation of SREBP-1c (46). Because RXR-LXR heterodimers are the known NRs that regulate hepatic lipogenesis and cholesterol metabolism through an SREBP-1–dependent mechanism (47,48), UAB126 may differentially modulate the activity of RXR-LXR compared with bexarotene. It is unclear whether UAB126 is an agonist, antagonist, or inverse agonist for RXR-LXR. Further studies warrant more detailed structure-function relationship between UAB126 and NRs. As such, many RXR agonists have been reported, but their biological effects are variable, perhaps due to the specificity for partnering NRs, varying potency of the RXR activation, and differential recruiting of coregulators. More recently, partial agonists and RXR homodimer antagonists have been developed and have antidiabetic effects without demonstrating hypertriglyceridemia or hypothyroidism (9). Thus, RXR seems to be a promising target if the specificity and potency are optimally modulated.

In the current study, we introduce a novel rexinoid-like molecule that not only prevents but also reverses diet-induced obesity, insulin resistance, steatosis, and hyperlipidemia without altering food intake and physical activity. Thus, UAB126 may be a powerful pharmacological treatment option for obesity-associated cardiometabolic disorders.

Acknowledgments. The authors thank the members of the UAB Comprehensive Diabetes Center for useful discussions.

Funding. This study was supported by the UAB Diabetes Research Center–sponsored pilot and feasibility program supported by National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases (P30-DK-079626), UAB Center for Clinical and Translational Science pilot and feasibility program by the NIH Clinical Center (UL1-TR-003096), UAB Comprehensive Diabetes Center and National Heart, Lung, and Blood Institute (R01-HL-128695 to J.-a.K.), National Institute on Aging (R03-AG-058078 to J.-a.K.), National Institute of Diabetes and Digestive and Kidney Diseases (R01-DK-112934 to K.M.H., R00-DK-95975 and 1R01-DK-120684 to S.B., and DK-099550 to H.M.T.), and National Institute of General Medical Sciences (T32-GM-109780 to S.I.B.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.-a.K. devised and coordinated the project. G.R., T.K., H.-S.K., and S.I.B. performed experiments. M.E.Y. performed indirect calorimetry. D.D.M. and V.R.A. invented and synthesized UAB126. M.E.Y. and J.-a.K. designed experiments. G.R., T.K., M.E.Y., D.D.M., V.R.A., H.M.T., K.M.H., M.E.Y., S.B., T.C., M.-A.B., A.S., S.J.F., and J.-a.K., provided expertise and wrote the manuscript. J.-a.K. is the guarantor of this work and as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Portions of this work were presented at the 76th Scientific Sessions of the American Diabetes Association, New Orleans, LA, 10–14 June 2016.