Adiponectin is an adipocytokine that signals through plasma membrane–bound adiponectin receptors 1 and 2 (AdipoR1 and -2). Plasma adiponectin depletion is associated with type 2 diabetes, obesity, and cardiovascular diseases. Adiponectin therapy, however, is yet unavailable owing to its large size, complex multimerization, and functional differences of the multimers. We report discovery and characterization of 6-C-β-d-glucopyranosyl-(2S,3S)-(+)-5,7,3′,4′-tetrahydroxydihydroflavonol (GTDF) as an orally active adiponectin mimetic. GTDF interacted with both AdipoRs, with a preference for AdipoR1. It induced adiponectin-associated signaling and enhanced glucose uptake and fatty acid oxidation in vitro, which were augmented or abolished by AdipoR1 overexpression or silencing, respectively. GTDF improved metabolic health, characterized by elevated glucose clearance, β-cell survival, reduced steatohepatitis, browning of white adipose tissue, and improved lipid profile in an AdipoR1-expressing but not an AdipoR1-depleted strain of diabetic mice. The discovery of GTDF as an adiponectin mimetic provides a promising therapeutic tool for the treatment of metabolic diseases.
Introduction
The anti-inflammatory adipocytokine adiponectin (1,2) signals through adiponectin receptors 1 and 2 (AdipoR1 and -2) (3). T-cadherin, a cadherin family member that lacks transmembrane and cytoplasmic domains, also binds adiponectin and is proposed to affect its bioavailability (4). Plasma adiponectin depletion is associated with type 2 diabetes, obesity, and cardiovascular diseases (5–7). Adiponectin administration or overexpression ameliorates insulin resistance, metabolic syndrome, and atherosclerosis in animals (3,8–12) and enhances pancreatic β-cell survival (13). This evidence makes AdipoRs important therapeutic targets for metabolic diseases.
Structurally, adiponectin belongs to the complement 1q family (1,14,15). Adiponectin monomer is a 30-kDa protein consisting of an N-terminal collagenous domain and a COOH-terminal globular domain (1). Mammalian plasma adiponectin is present in several multimeric forms, low-molecular-weight dimer or trimers, medium-molecular-weight hexamers, or high-molecular-weight (HMW) dodecamers and 18 mers (10,16–18). The globular domain of adiponectin (gAd) can form trimers and was initially shown to exist as a proteolytic cleavage product in human plasma (10); although subsequent studies failed to detect it in circulation, its ability to modulate AdipoRs is undisputed. All these forms display different levels of physiological activity, and the HMW complex is considered the most clinically relevant form (10,16–18). The HMW full-length adiponectin and gAd preferentially signal through AdipoR2 and AdipoR1, respectively (3). Given the multimerization-related complexities of adiponectin structure and function, it appears that small molecule AdipoR ligands may provide the only viable therapeutic option against diseases associated with defects in adiponectin expression or action.
We have previously identified 6-C-β-d-glucopyranosyl-(2S,3S)-(+)-3′,4′,5,7-tetrahydroxyflavonol (GTDF), a novel natural analog of the dietary flavonoid quercetin, as a potent orally bioavailable osteoanabolic compound that induced proliferation, differentiation, and mineralization of cultured primary osteoblasts at a nanomolar concentration that was 1,000-fold less than the effective concentration of quercetin or queretin-O-glucoside and restored trabecular bones of osteopenic rats on par with parathyroid hormone (19). While studying its mechanism of action, we found that GTDF induced rapid AMP-dependent protein kinase (AMPK), AKT, and p38 phosphorylation and elevated PPARγ coactivator-1α (PGC-1α) expression in osteoblasts. GTDF also deacetylated tumor suppressor P53 via indirect activation of NAD-dependent deacetylase Sirtuin1 (Sirt1) (S.Sa., N.C., unpublished observations, and Khan et al. [20]). Literature search revealed that adiponectin elicits similar cellular signaling (1,3,11). Interestingly, the quercetin group of compounds display functional properties similar to adiponectin, such as AMPK activation, glucose uptake enhancement, induction of fatty acid oxidation–related genes, and amelioration of diabetes and insulin resistance in vivo (21,22). We thus asked if GTDF, quercetin, or other naturally occurring quercetin analogs could be adiponectin mimetics. Here we report detailed characterization of GTDF as an adiponectin mimetic that improves metabolic health in a rodent model of diabetes.
Research Design and Methods
Materials and Kits
All cell culture reagents were from Invitrogen, Life Technologies (Carlsbad, CA). Fine chemicals were from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated. Epoxy agarose beads conjugated with GTDF were constructed at Shantani Biotech (Pune, India). gAd was purchased from Enzo Life Sciences (Farmingdale, NY) and ATGen Global (Gyeonggi-do, South Korea) and compared. gAd from both sources showed identical activity. gAd from ATGen Global was used in this study. GW7647 was from Sigma-Aldrich, and full-length human AdipoR1 and AdipoR2 mammalian expression constructs were from Open Biosystems (Huntsville, AL). 125I-gAd was from Phoenix Pharmaceuticals (Burlingame, CA), and 125I (20 MBq) used for radiolabeling GTDF was from BARC (Mumbai, India). Glycogen assay kit was from Sigma-Aldrich. Serum concentrations of triglycerides (TGs), nonesterified free fatty acids (NEFAs), HDL, LDL, VLDL, total cholesterol, β-hydroxybutyrate, and creatinine were determined using kits purchased from Pointe Scientific (Canton, MI). Circulating glucagon, insulin, C-peptide, leptin, adiponectin, ghrelin, corticosterone, and markers of inflammation (monocyte/macrophage chemoattractant protein-1 [MCP-1], tumor necrosis factor-α [TNF-α], and interleukin 6 [IL-6]) were estimated using ELISA assay kits (R&D Systems, Minneapolis, MN). Serum concentrations of sodium and potassium were measured using a Cobas c system (Roche Diagnostics, Indianapolis, IN). TUNEL assay kit was from Roche Applied Science (Indianapolis, IN). Plasma membrane (PM) extraction kit was from BioVision (Milpitas, CA).
Cell Culture and Induction of Differentiation
HEK-293, CHO, C2C12, and 3T3L-1 cell lines were cultured as previously reported (23). Mouse stromal vascular fraction (SVF) was prepared from epididymal fat pad by collagenase digestion. Human SVFs were prepared from human lipoaspirates (subcutaneous), collected after approval of the institutional ethics committee. Cells were differentiated in the presence of GTDF or vehicle for 10 days (3T3L-1 and mouse SVF) or 21 days (human SVF) using standard procedure and then analyzed by quantitative PCR (QPCR), immunoblotting, or microscopy.
Iodination and Purification of 125I-GTDF
GTDF was radiolabeled as previously described for quercetin with minor modifications. In brief, 10 μL 125I (20 MBq) was added to 100 μg GTDF in 5% acetic acid/methanol, chloramine-T (4 μg in Milli-Q H2O) was added, and the mixture was allowed to react at room temperature (24°C) for 5 min. The reaction was terminated by adding 60 μL sodium metabisulphite (4 mg/mL in Milli-Q H2O). The reaction mixture was dried by passing nitrogen and was dissolved into methanol (100 μL). Reverse-phase thin-layer chromatography (TLC) (RP-18 F254s, 8 cm in length; Merck) was used to purify 125I-GTDF from free iodine and unlabeled compound using methanol-water (40–60%) as mobile phase. After run, the TLC plate was cut into pieces of 0.5 mm each, and the distribution of radioactivity along the plate was measured in a gamma counter. TLC of the blank reaction suggested the location of free 125I in the TLC plate. The RF value of the labeled compound was determined by gamma counting. The area showing maximum activity at distance of 40–60 mm was eluted from the TLC plate and was washed with methanol, centrifuged, decanted, and dried under N2.
Overexpression and Silencing Experiments
Plasmid transfections were performed with Lipofectamine LTX (Life Technologies) reagent according to the manufacturer’s protocols. For C2C12 transfections, C2C12 myoblasts were trypsinized and transfected. The cells were then grown until 60–70% confluence. They were then differentiated and assessed as required. For RNA interference, small interfering RNAs (siRNAs) (siAdipoR1; L-063377-01-0010 and siC; siRNA against luciferase GL3 duplex D-001400–01-50) were purchased from Thermo Scientific (Pittsburgh, PA). Cells were transfected with 0.1 µmol/L of each siRNA using DharmaFECT 1 transfection reagent (Thermo Scientific). Seventy-two hours after transfection, cells were treated and analyzed as required.
QPCR, Western Blotting, Coimmunoprecipitation
These studies were performed as previously described (23). The list of primer sequences for QPCR has been provided in Supplementary Table 5. Anti–PGC-1α (ST1202; Millipore, Billerica, MA) was used at 1:2,000 dilution, and anti-CD36 (18836-1-AP; Proteintech, Chicago, IL) was used at 1:1,000 dilution. Anti–UCP-1 (ab10983), –UCP-3 (ab3477), and –PPARα (ab8934) (Abcam, Cambridge, MA) were used at 1:1,000 dilutions. Phospho-AMPK (Thr172), AMPK, phospho-p38 (Thr 180/Tyr 182), p38, phospho-ACC, acetyl coA carboxylase (ACC), glucose transporter 4 (Glut4), P53, acetyl-P53 (Lys 382), ccaat/enhancer protein β (C/EBPβ), and β-actin antibodies were from Cell Signaling Technology (Beverly, MA) and were used at 1:1,000, except β-actin, which was used at 1:3,000 dilution. Antibodies against AdipoR1 (sc-46748), AdipoR2 (sc-46755), and N-cadherin (sc-1502) were from Santa Cruz Biotechnology (Dallas, TX) and used at 1:1,000 dilutions. For immunohistochemistry, insulin antibody (Cell Signaling Technology) was used at a dilution of 1:200, phycoerythrin-tagged Ki-67 antibody (BD Biosciences, San Diego, CA) was used at a dilution of 1:100, and anti–UCP-1 (ab10983) was used at a dilution of 1:250. Following detection with one antibody, Western blots were stripped and reprobed with other antibodies whenever possible.
Glucose Uptake
Fully differentiated C2C12 myotubes on 24-well plates were treated with vehicle (0.1% DMSO) or 0.01 μmol/L GTDF for 24 h, after which the cells were serum starved for 3 h. The cells were then washed three times in warm (37°C) HEPES buffer solution (HBS; 140 mmol/L sodium chloride, 20 mmol/L HEPES, 5 mmol/L potassium chloride, 2.5 mmol/L magnesium sulfate, 1 mmol/L calcium chloride, pH 7.4) and then were treated with warm HBS or 0.1 µmol/L insulin (in HBS) for 20 min. Subsequently, cells were washed three times in warm HBS and then incubated in 250 μL transport solution (HBS containing 1 µCi 3H-deoxyglucose [PerkinElmer] and 10 µmol/L unlabeled 2-deoxyglucose [Sigma-Aldrich]) per well for 5 min. Then the transport solution was aspirated and the cells were washed three times with ice-cold stop solution (0.9% NaCl and 25 mmol/L dextrose). Subsequently, the cells were lysed in 100 μL 0.5 N NaOH, and 5 μL lysate was used for determination of protein concentration (by bicinchoninic acid assay; Sigma-Aldrich), and the rest of the lysate was used to measure cellular radioactivity in a beta counter (Beckman Coulter, New Delhi, India).
Fatty Acid Oxidation
C2C12 myotubes plated in 12-well plates were treated with vehicle (0.1% DMSO) or 0.01 µmol/L GTDF for 2, 24, or 120 h. After treatment, the cells were washed three times in warm HBS and were then incubated with medium containing 0.75 mmol/L palmitate (conjugated to 2% fatty acid–free BSA/[14C]palmitate at 2 µCi/mL) for 2 h. After this incubation period, 1 mL of the culture medium was removed and transferred to a sealable tube, the cap of which housed a Whatman (GF/B) filter paper disc that had been presoaked with 1 mol/L potassium hydroxide. 14CO2 trapped in the media was then released by acidification of media using 60% (volume for volume) perchloric acid and gently agitating the tubes at 37°C for 2 h. Radioactivity that had become adsorbed onto the filter discs was then quantified by liquid scintillation counting in a beta counter.
Animal Experiments
Animal studies were approved by the Institutional Animal Ethics Committee of Zydus Research Center. This facility is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. Mice were individually housed at 21°C, in 12-h light/dark cycles. All animals had access to normal chow diet and water ad libitum. Twelve-week-old male C57BL/6J (BL6) (25–30 g), db/db, or BKS-db/db (45–50 g) mice were divided into different treatment groups randomly (db/db, n = 8 per group; BKS-db/db, n = 6 per group). Vehicle groups received 1% carboxymethylcellulose, and the other groups received GTDF or pioglitazone (Pio; 10 mg/kg), once a day, by oral gavage for 30 days. Feed intake and body weight were measured every day. Blood was obtained from tail snips, and glucose levels were measured using a glucometer. Glucose tolerance tests were performed on day 29 after overnight fasting. Pyruvate tolerance test (PTT) was performed on day 31, after an overnight fast. On day 32, the animals were fasted for 6 h and killed. Plasma and tissues were collected and stored at −80°C until further analysis. Hepatic glycogen was estimated using a glycogen assay kit. Intraperitoneal glucose tolerance test (IPGTT), intraperitoneal PTT, estimation of circulating biomarkers, histology, and immunohistochemistry were performed using standard procedures.
Extensor Digitorum Longus Muscle Culture and Treatment
Extensor digitorum longus (EDL) muscle strips along with tendons were dissected out from anesthetized mice and were tied to stainless steel clutch pins by the tendons, without stretching. Muscles were preincubated for 30 min at 37°C in oxygenated (95% O2/5% CO2) Krebs-Henseleit solution (pH 7.4) and then again for 30 min in the same medium with or without 2.5 μg/mL gAd or 0.1 µmol/L GTDF. The incubation media was gassed continuously with 95% O2/5% CO2 through hypodermic needle. At the end of this incubation, tendons were removed and the muscles were blotted on gauze pads, snap frozen in liquid N2, and stored at −80°C until further analysis.
Data Analysis and Statistics
Results are expressed as mean ± SEM. All data were analyzed using GraphPad Prism 5.0 (GraphPad, San Diego, CA). Animal data involving more than two groups were analyzed using two-way ANOVA followed by Bonferroni post test or one-way ANOVA followed by Tukey multiple comparison test as appropriate. In vitro or in vivo data involving two experimental groups were analyzed using two-tailed unpaired Student t test or Mann-Whitney U test as appropriate.
Results
GTDF Binds to AdipoRs and Mimics Adiponectin-Associated Signaling Events In Vitro
We assessed GTDF, quercetin, and its natural analogs (chemical structures in Fig. 1) in a peroxisome proliferator–activated receptor-α (PPARα) ligand activation assay, a previously reported functional assay for AdipoRs (3), in HEK293 cells that express endogenous AdipoR1 (3). All the quercetin analogs tested enhanced PPARα activity in the presence of PPARα agonist GW7647 (Fig. 2A). GTDF maximally activated the reporter at 0.01 µmol/L, whereas dihydroquercetin (DHQ), an enantiomer of the aglycone form of GTDF and quercetin-6-C-β-d-glucopyranoside (QCG), did so at 0.1 µmol/L, and quercetin, quercetin-3-O-β-d-glucopyranoside (QOG), or quercetin-3-O-α-L-rhamnopyranoside (QR) caused maximal activation at 1 µmol/L (Fig. 2A). Amplitudes of activation of PPARα ligand activity by GTDF and gAd were comparable (Fig. 2B). GTDF did not activate PPARα in the absence of GW7647, indicating that it was not a PPARα agonist per se (Fig. 2C).
To assess if GTDF and AdipoRs could physically interact, we immobilized GTDF on agarose beads and performed a pull-down assay with purified PM from C2C12 myotubes that expressed both AdipoR1 and -2 (3). GTDF beads but not control beads successfully pulled down AdipoR1 and -2, and free GTDF or QCG competed with this binding (Fig. 2D). Cold GTDF competed with 125I-adiponectin for binding to C2C12 cells in a radioligand binding assay (Fig. 2E). Radioligand saturation assays with 125I-labeled GTDF (Fig. 2F) or 125I-gAd (Supplementary Fig. 2) revealed that both these ligands bound to AdipoR-deficient CHO cells (24) transfected with AdipoR1 or -2 but not empty expression plasmids (Fig. 2G) (GTDF: Kd and BMAX, 4.90 nmol/L and 1,410 fmol/mg of protein for AdipoR1 and 326 nmol/L and 3,950 fmol/mg of protein for AdipoR2; gAd: Kd and BMAX, 0.25 μg/mL and 130 ng/mg for AdipoR1 and 0.526 μg/mL and 200 ng/mg for AdipoR2). Thus GTDF displayed ∼70-fold more affinity toward AdipoR1 over AdipoR2.
Consistent with adiponectin-associated rapid signaling events (1,25), GTDF induced AMPK, ACC, and p38 phosphorylation in C2C12 myotubes (Fig. 2H). Whereas p38 phosphorylation was rapid and sustained, AMPK and ACC phosphorylation peaked at 1 and 10 min, respectively, and returned to the basal level at 60 min (Fig. 2H). A similar pattern of AMPK and ACC phosphorylation by adiponectin has been reported earlier (11).
Apart from adiponectin, glutamate receptors also activate AMPK and p38 (26), and we therefore assessed if GTDF could modulate these receptors. However, GTDF failed to bind to or activate/repress any of the ionotropic or metabotropic glutamate receptors (mGlu), whereas quercetin strongly activated mGlu2 and inhibited mGlu4, -6, and -8 (Supplementary Tables 1 and 2). To also assess if these signaling events were mediated through one or more G protein–coupled receptors (GPCRs), GPCR modulation by GTDF was assessed, and it failed to activate or repress any of the 158 GPCRs tested (Supplementary Tables 3 and 4).
Consistent with its activation of adiponectin-associated signaling, GTDF enhanced mRNA and protein levels of factors associated with fatty acid transport (CD36 and fatty acid binding protein 3 [FABP3]), oxidation (carnityl palmitoyl transferase 1b [CPT1B], long-chain fatty acyl CoA synthetase [FACS], acetyl CoA oxidase [ACOX1], PPARα, and PPARδ), mitochondrial biogenesis (PGC-1α), mitochondrial uncoupling protein 3 [UCP-3], and Glut4, whereas lipoprotein lipase (LPL) and PPARγ were unchanged (Fig. 2I and J). Adiponectin deacetylates and activates PGC-1α via indirect activation of Sirt1 (25), and consistent with its indirect Sirt1 activation (20), GTDF also deacetylated PGC-1α in C2C12 myotubes (Fig. 2K). This PGC-1α induction and activation led to increased mitochondrial DNA copy number (Fig. 2L). gAd enhances basal glucose uptake (3,11) and increases insulin sensitivity (27) in C2C12 myotubes, and in consistence, GTDF also enhanced basal and insulin-stimulated glucose uptake in these cells (Fig. 2M). Further, GTDF also significantly enhanced basal and insulin-dependent glucose uptake in a C2C12 myotube model of palmitate-induced insulin resistance (Fig. 2N). Similar to rapid enhancement of fatty acid oxidation by adiponectin (3,11,28), GTDF significantly enhanced [14C]palmitate oxidation in C2C12 myotubes within 2 h, which increased further over time (Fig. 2O).
Overexpression or Silencing of AdipoR1 Augments or Mitigates GTDF Functions In Vitro
Consistent with the binding studies, AdipoR1 but not AdipoR2 overexpression in C2C12 myotubes significantly enhanced GTDF-stimulated AMPK, ACC, and p38 phosphorylation (Fig. 3A). siRNA for AdipoR1 (siAdipoR1) but not control siRNA against luciferase (siC) abolished GTDF-induced AMPK, ACC, and p38 phosphorylation, without affecting AdipoR2 expression (Fig. 3B). Further, GTDF-stimulated glucose uptake and fatty acid oxidation were abolished by siAdipoR1 (Fig. 3C and D), whereas insulin-stimulated glucose uptake was unaltered (Fig. 3D). Together with Fig. 1 and Supplementary Tables 1–4, these results demonstrate that GTDF action indeed is AdipoR1 specific.
BKS.Cg-Dock7m+/+ Leprdb/db/J (BKS-db/db) Mice Have Severely Depleted Plasma Membrane–Associated AdipoR1 Compared With B6.BKS(D)-Leprdb/db/J (db/db) Mice
A relevant animal model to test AdipoR specificity of GTDF would have been AdipoR1/R2 knockout mice subjected to diet-induced obesity. However, our inability to obtain the knockout mice led us to search for an alternate model. Since chronic high level of plasma insulin causes an AdipoR depletion–mediated adiponectin resistance (30), we systematically investigated AdipoR expression in major adiponectin target organs, across different age-groups in two different strains of leptin receptor–deficient obese and diabetic mice, db/db (in C57BL/6J background) and BKS-db/db (in BLKS/J background), and wild-type healthy C57BL/6J (BL6).
Comparison of 12-week-old male BL6, db/db, and BKS-db/db mice revealed that both diabetic mice had lower AdipoR mRNAs in skeletal muscle, liver, and epididymal white adipose tissue (eWAT) compared with the healthy BL6 mice. However, between the diabetic strains, only modest differences were observed (Fig. 4A).
Examination of total and PM-associated AdipoR protein levels however, revealed striking differences between the two diabetic strains. Whereas BKS-db/db displayed >80% decrease in total and PM-associated and therefore functional AdipoR1 expression in skeletal muscle, BL6 and db/db did not show a significant difference (Fig. 4B). In liver, PM AdipoR1 but not AdipoR2 was strongly depleted (≥80%) in BKS-db/db but not in db/db, whereas compared with BL6, total AdipoR1 and -2 proteins were diminished in both. In WAT, both diabetic strains displayed >90% reduction in total AdipoR1 protein compared with BL6, whereas AdipoR2 protein was not detected (Fig. 4B). AdipoR2 could not be detected in the PM fraction of skeletal muscle (Fig. 4B), and we failed to generate enough PM extract from WAT for immunoblotting. Together, compared with db/db and BL6, BKS-db/db displayed severely depleted total and PM AdipoR1 protein in skeletal muscle and liver at 12 weeks of age. Consistent with this observation, gAd and GTDF failed to induce AMPK phosphorylation in EDL muscles of BKS-db/db, whereas it did so in EDL muscles from both BL6 and db/db mice (Fig. 4C). Thus, based on above evidence, we selected 12-week-old male db/db mice as adiponectin-sensitive and age- and sex-matched BKS-db/db as adiponectin-resistant models for the in vivo pharmacological studies.
GTDF Fails to Improve Diabetic Phenotype in BKS-db/db Mice
Over a 30-day treatment period, GTDF did not alter feed intake or body weight in BKS-db/db (Fig. 5A and B), whereas PPARγ agonist Pio significantly enhanced body weight 6 days onwards without altering feed intake (Fig. 5A and B). GTDF failed to alter nonfasting blood glucose, fasting blood glucose, and glucose clearance, whereas Pio significantly improved these parameters (Fig. 5C and D). GTDF failed to alter plasma glycated hemoglobin (HbA1c) and insulin (Fig. 5E and F). GTDF was ineffective in improving the fasting lipid profile in these mice, whereas Pio modestly but significantly reduced plasma TG and VLDL (Fig. 5G). These results indicate that GTDF is ineffective in BKS-db/db mice that display severely depleted AdipoR1 protein in skeletal muscle and liver. That Pio improved glycemic parameters in these mice further indicates that GTDF did not function through the PPARγ pathway.
GTDF Ameliorates Diabetic Phenotype in AdipoR1-Expressing db/db Mice
In contrast to its inefficacy in BKS-db/db (Fig. 5), GTDF was remarkably effective in AdipoR1-expressing db/db mice. It reduced feed intake in db/db but not BL6 (Fig. 6A), which could be correlated with a robust reduction in serum ghrelin (Fig. 6N). GTDF-treated db/db mice displayed a reducing trend in body weight, which however was not statistically significant (Fig. 6B), and showed significantly reduced adipose tissue weight (Supplementary Fig. 3). GTDF remarkably reduced nonfasting blood glucose in db/db mice on the 7th day of treatment (Fig. 6C). Reduction in feed intake could not be responsible for the GTDF-induced reduction in nonfasting glucose level in db/db, as the feed intake was significantly different only after 16 days of treatment, whereas nonfasting blood glucose was drastically reduced at the 7th day and did not fall any further.
GTDF modestly yet significantly reduced fasting blood glucose in db/db but not in BL6 (Fig. 6D), which was associated with modestly reduced hepatic gluconeogenesis, as revealed by a PTT (Fig. 6E). GTDF treatment prevented increase in HbA1c in db/db, which increased in the vehicle-treated db/db during the treatment period (Fig. 6F). In IPGTT, GTDF-treated db/db mice showed remarkably improved glucose clearance (Fig. 6G). However, both fasting insulin as well as insulin levels during glucose challenge were higher in these mice over vehicle-treated controls, whereas they were unchanged in BL6 mice (Fig. 6G–I). The enhanced insulin level was associated with increase in pancreatic C-peptide (Fig. 6J), indicating that insulin production was higher in GTDF-treated db/db mice. However, hypokalemia, generally associated with higher insulin, was not observed (Fig. 6K).
Pancreatic histomorphometry revealed that in vehicle-treated db/db mice, islet number was diminished in comparison with BL6, and GTDF significantly increased it in the former but not latter (Fig. 6L and M). Furthermore, ∼30% of the β-cell population in the islets of vehicle-treated db/db mice were apoptotic, as determined by TUNEL staining, whereas GTDF markedly attenuated the number of apoptotic β-cells (Fig. 6L and M). The increased islet number in GTDF-treated db/db pancreas was not due to cellular proliferation as the islets from these mice did not stain for the proliferation marker Ki-67 (Fig. 6L), indicating that GTDF might protect β-cells from gluco- and lipotoxicity-induced apoptosis. Serum glucagon level was unchanged (Fig. 6N).
Consistent with anti-inflammatory properties of adiponectin, GTDF treatment in db/db mice significantly lowered serum TNF-α and IL-6 showed a reducing trend, whereas MCP-1 was unchanged (Fig. 6N). Whereas serum adiponectin showed an increasing trend upon GTDF treatment (P = 0.082), leptin level was greatly diminished (Fig. 6N). We also found a robust fall in serum corticosterone (Fig. 6N), indicating that GTDF might be cardioprotective. Analysis of serum lipid profile revealed significant reductions in total cholesterol, TG, VLDL, NEFAs, and ketone bodies (β-OH butyrate), which supports an enhanced fatty acid oxidation rate, whereas it significantly enhanced HDL level (Fig. 6N), reiterating its cardioprotective promise. LDL and creatinine levels were unchanged (Fig. 6N). In light of the lipid profile data, we examined the hepatic histology, and sections from GTDF-treated db/db mice displayed no vacuolation and lipid accumulation, whereas the vehicle-treated db/db exhibited robust lipid accumulation (Fig. 6O). Hepatic glycogen level was unchanged (Fig. 6P). Together, Figs. 5 and 6 indicated that GTDF stalls diabetes progression in AdipoR-expressing db/db mice and improves their overall metabolic health, whereas it is ineffective in AdipoR1-depleted BKS-db/db.
GTDF Induces Browning of WAT, Evidence of Involvements of Direct and Indirect Pathways
GTDF induced and activated PGC-1α in myocytes, and PGC-1α is reported to induce FNDC5, a myokine that drives browning of myf-5–negative white adipocytes (31). GTDF indeed enhanced FNDC5 expression in C2C12 myotubes and increased its secretion in culture medium (Fig. 7A), indicating that it might induce FNDC5 in circulation and thereby promote browning of WAT. Histological analysis of eWAT from GTDF-treated db/db mice revealed robust fat mobilization characterized by reduction in adipocyte size (Fig. 7B), and these eWATs stained strongly for the brown adipose marker UCP-1 (Fig. 7C), indicating that GTDF indeed induced a browning-like phenomenon.
Adipocyte-specific transgenic overexpression of adiponectin is reported to mobilize WAT, enhance UCPs, and decrease TNF-α and leptin level in this tissue (32), indicating that adiponectin may also promote an FNDC5-independent browning of WAT through a direct action on adipocytes. Differentiation of 3T3L-1 cells in the presence of GTDF from day 0 or day 2 of differentiation reduced oil droplet accumulation in these cells and significantly reduced TG content (Fig. 7D and E). Differentiation of 3T3L-1 and mouse (epididymal) or human SVFs in the presence of GTDF enhanced expressions of UCP-1, UCP-2, PGC-1α, brown adipose determination factor PRDM16, brown fat–enriched protein CIDEA, and adiponectin (Fig. 7F–H). PRDM16 and C/EBPβ transcriptional complex has been implicated in conversion of myf-5–positive myoblasts into BAT (33), and GTDF also induced the expression of C/EBPβ in 3T3L-1 and mouse SVF (Fig. 7H), indicating that this complex may play an important role in the browning phenomenon observed here. Consistent with elevated browning markers, GTDF increased mitochondrial DNA content in mouse SVFs (Fig. 7I). Notably, GTDF did not activate β-adrenergic receptors (Supplementary Tables 3 and 4 and Fig. 7J).
Discussion
After comprehensive molecular characterizations, we have identified GTDF as an orally active, small-molecule adiponectin mimic that improves metabolic parameters in a preclinical disease setting and holds therapeutic promise in metabolic diseases caused by adiponectin deficiency.
GTDF interacted with AdipoRs and displayed a 70-fold higher affinity for AdipoR1 over AdipoR2. It induced adiponectin-associated rapid signaling, gene expressions, and functional events in C2C12 myotubes, which were respectively augmented or abolished by AdipoR1 overexpression or silencing.
Our inability to procure AdipoR knockout mice inspired us to explore alternate model systems and led to the identification of 12-week-old male BKS-db/db mice as AdipoR1 deficient, whereas age- and sex-matched db/db mice were found to be AdipoR1 intact. Intriguingly, these differences were not apparent at the mRNA level but were pronounced in the protein level, especially at the level of PM-associated and hence functional form of AdipoR1 protein. These findings indicate that apart from transcriptional regulation by insulin and Foxo1 (30), AdipoR1 expression and function may also be regulated by posttranscriptional or -translational events. Interestingly, a recent report characterized a developmentally regulated alternate splice variant of AdipoR1 in human subjects that was strongly enhanced during skeletal muscle differentiation and was decreased in diabetic patients, and the protein level of AdipoR1 accurately represented these changes but not total mRNA levels that accounted for all AdipoR1 splice variants (34). Our data gains further support from another recent report, which describes that microRNA-221 and the RNA binding protein polypyrimidine tract binding protein regulate AdipoR1 protein expression and are induced in genetic and diet-induced mouse models of obesity (35).
Consistent with AdipoR1 deficiency, BKS-db/db mice were refractory to GTDF, whereas metabolism in db/db mice was remarkably improved. Interestingly, GTDF reduced feed intake in db/db mice from 16 days onwards. The effect of adiponectin on feed intake is controversial, and all three possible outcomes have been reported (36–38). Although the reason for these differences is unclear, strain of mice, mode of administration, and the administered form of adiponectin appear to be responsible. However, it is important to note that we delivered GTDF by oral route, in contrast to systemic delivery or transgene-based overexpression of adiponectin, and thus GTDF may impact satiety-related hormones from the gut. In agreement with this notion, plasma ghrelin level was greatly reduced in GTDF-treated db/db mice. Ghrelin is chiefly produced by P/D1 cells lining the fundus of the stomach, and given the current lack of information regarding expression of AdipoRs in these cells, it is presently unclear if GTDF-mediated downregulation of ghrelin level is achieved through AdipoRs in these cells. However, since GTDF did not alter feed intake in adiponectin-insensitive BKS-db/db, it seems probable that these cells may express one or more AdipoRs, which may act as sensors for diet-derived putative AdipoR ligands.
In addition to alteration in feed intake, GTDF remarkably reduced nonfasting blood glucose that was significant from day 7 onwards and did not fall thereafter. GTDF also stalled the increase in HbA1c that was observed in vehicle-treated db/db mice. Reduction in feed intake could not account for these changes as earlier reports have demonstrated that caloric restriction not only fails to decrease nonfasting blood glucose or HbA1c in db/db mice with manifested diabetes (39–41) but also fails to prevent onset of hyperglycemia in db/db mice pair fed for 5 weeks since weaning (42). That GTDF decreased feed intake in db/db only but not in BL6 and BKS-db/db clearly indicates that toxicity was not involved. It is further supported by our earlier study where GTDF not only normalized feed intake in high-dose dexamethasone–treated Wistar rats but also prevented dexamethsone-induced mortality (20).
Although the effect of GTDF on nonfasting blood glucose and glucose clearance was robust, it caused a significant but modest decrease in fasting blood glucose and hepatic gluconeogenesis. Since GTDF showed a preference for AdipoR1, and AdipoR2 is the principal AdipoR in liver, it appears that GTDF-mediated reduction of blood glucose might be principally achieved by skeletal muscle AdipoR1-mediated glucose disposal, which is supported by the fact that fasting hepatic glycogen was unaltered upon GTDF treatment. However, a robust reduction in ketone bodies in GTDF-treated db/db mice indicates that in these animals, the liver may efficiently oxidize them.
BKS-db/db display more severe hyperglycemia than db/db and are susceptible to terminal diabetes, characterized by pancreatic β-cell degranulation and death, whereas db/db mice are protected from terminal diabetes due to the unique proliferation capacity of their β-cells (43). Thus, in older db/db mice, hyperglycemia is corrected through a higher plasma insulin level, although these mice remain severely dyslipidemic (43). In our experimental setup, consistent with a lower plasma insulin (which was still much higher than healthy BL6) than db/db, the BKS-db/db mice indeed exhibited higher fasting and nonfasting blood glucose. However, β-cell proliferation was not yet apparent in db/db as evidenced by lack of Ki-67 staining, indicating that during the experimental period, the db/db β-cells did not yet go on a proliferative drive. Further, ∼30% of β-cells from db/db mice were apoptotic, and GTDF strongly mitigated this apoptosis without increasing proliferation, indicating that GTDF may protect these cells from gluco- and lipotoxic stresses, and a similar finding has been reported for adiponectin (44).
Despite higher plasma insulin, GTDF-treated db/db mice were not hypokalemic and displayed a remarkably improved lipid profile, evidenced by decreased serum total cholesterol, TG, VLDL, and NEFA and increased HDL, which, together with a marked decline in plasma corticosterone in these mice, indicates that GTDF may have cardioprotective properties. NEFAs are breakdown products of TG that are released from adipocytes after lipolysis and are important in diabetic pathogenesis (45). The marked decline in plasma NEFAs suggests that in GTDF-treated db/db mice, NEFAs were either used with robust efficiency in liver and skeletal muscle or adipose TG were efficiently oxidized in situ, or a combination of both. These postulates could be corroborated by the facts that livers of GTDF-treated db/db were free from vacuolation and oil droplets, which were characteristically present in vehicle-treated db/db mice, and eWAT depot in GTDF-treated db/db mice showed robust mobilization and increased UCP-1 expression. Since both db/db and BKS-db/db displayed depleted AdipoR1 protein in eWAT, it is possible that GTDF caused browning of eWAT through an indirect mechanism involving the previously described PGC-1α–induced myokine, FNDC5/irisin (31). Consistent with PGC-1α induction and activation, GTDF induced FNDC5 expression in C2C12 myotubes and its release in culture medium. In addition, GTDF was also capable of directly inducing brown adipose markers and increasing mitochondrial content in 3T3L-1 and mouse and human SVFs differentiated in its presence, indicating that GTDF may protect against energy imbalance–related metabolic diseases. However, further studies including GTDF’s effect on energy expenditure are needed to fully understand and realize its potential.
While this article was under review, Okada-Iwabu et al. (46) reported identification and characterization of a small-molecule AdipoR agonist, AdipoRon. AdipoRon is 2-(4-benzoylphenoxy)-N-[1-(phenylmethyl)-4-piperidinyl]acetamide and GTDF is a flavone c-glucoside, and these compounds do not share any structural homology. Functionally, AdipoRon showed comparable affinity to both AdipoR1 and -2 and acted through both these receptors to bring about physiological improvements in diabetic mice, whereas GTDF showed a stronger affinity for AdipoR1, and given its modest effects on hepatic gluconeogenesis, appears to act mainly via AdipoR1. Incidentally, another recent report identified DHQ (also known as taxifoliol) as one of nine small-molecule AdipoR agonists from a library of 10,000 compounds by a fluorescent polarization–based screen; however, the in vivo efficacy of DHQ is yet to be elucidated (47). DHQ is an enantiomer of the aglycone form of GTDF and was active in our PPARα ligand activation as well, albeit at a 10-fold higher concentration than GTDF. In contrast to GTDF, DHQ displays a stronger affinity for AdipoR2 (47), and thus studies with DHQ alone or in combination with GTDF will be needed to further explore their therapeutic potential in metabolic diseases.
In conclusion, discovery of GTDF as a small-molecule adiponectin mimetic that remarkably improves metabolic health in diabetic mice provides a promising therapeutic tool for treatment of adiponectin deficiency–associated metabolic diseases. However, given the lack of AdipoR knockout animal models in this study, it remains to be confirmed if all the beneficial metabolic effects of GTDF were indeed routed through AdipoRs alone.
Article Information
Acknowledgments. The authors acknowledge the Sophisticated Analytical Instrument Facility in CSIR-CDRI for help with confocal microscopy. The authors acknowledge Prem N. Yadav (Division of Pharmacology, CSIR-CDRI) for help with designing of the radioligand binding assays and Durga Prasad Mishra (Division of Endocrinology, CSIR-CDRI) for sharing antibodies. CSIR-CDRI communication number for this article is 8700.
Funding. This work was supported by CSIR grant BSC0201 to N.C. and S.Sa. A.K.S. was supported by a fellowship from CSIR. M.P.K. and M.Y. were supported by Indian Council for Medical Research fellowships. J.S.M., N.S., and Z.H. were supported by fellowships from University Grants Commission. K.K. was supported by a Department of Biotechnology fellowship.
Duality of Interest. A.K.S., M.P.K., J.S.M., N.S., M.Y., K.K., D.P.M., R.M., S.Sh., A.K.T., J.R.G., N.C., and S.Sa. have a pending patent pertaining to some of the work in this article. A.A.J., N.A.D., and M.R.J. are employees of Zydus Research Center, the research and development arm of Cadila Healthcare Ltd., Ahmedabad, India. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. A.K.S. and A.A.J. designed and performed experiments. M.P.K., J.S.M., N.S., M.Y., Z.H., K.K., S.K., N.A.D., and D.P.M. performed experiments. R.M. and M.R.J. supervised experiments. S.Sh. provided materials and contributed to discussion. A.K.T. supervised experiments and contributed to discussion. M.M.G. designed, performed, and supervised experiments. J.R.G. designed and supervised experiments and contributed to discussion. N.C. designed and supervised experiments, contributed to discussion, and wrote, edited, and reviewed the manuscript. S.Sa. conceived the study, designed and supervised experiments, contributed to discussion, and wrote, edited, and reviewed the manuscript. All authors analyzed data. S.Sa. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. Part of this work was presented as an abstract at the International Symposium on Molecular Signaling, Santiniketan, India, 18–21 February 2013.