Insulin resistance, the major metabolic abnormality underlying type 2 diabetes, is associated with chronic inflammation and heavy macrophage infiltration in white adipose tissue (WAT). The therapeutic properties of the synthetic adrenal steroid Δ5-androstene-17α-ethynyl-3β,7β,17β-triol (HE3286) were characterized in metabolic disease models. Treatment of diabetic db/db mice with HE3286 suppressed progression to hyperglycemia and markedly improved glucose clearance. Similar effects were also observed in insulin-resistant, diet-induced obese C57BL/6J mice and genetically obese ob/ob mice. This effect appeared to be a consequence of reduced insulin resistance because HE3286 lowered blood insulin levels in db/db and ob/ob mice. Treatment with HE3286 was accompanied by suppressed expression of the prototype macrophage-attracting chemokine monocyte chemoattractant protein-1 in WAT, along with its cognate receptor C-C motif chemokine receptor-2. Exposure of mouse macrophages to HE3286 in vitro caused partial suppression of endotoxin (lipopolysaccharide)-induced nuclear factor κ-B (NF-κB)-sensitive reporter gene expression, NF-κB nuclear translocation, and NF-κB/p65 serine phosphorylation. Proinflammatory kinases, including IκB kinase, c-Jun NH2-terminal kinase, and p38, were also inhibited by HE3286. In ligand competition experiments HE3286 did not bind to classical sex steroid or corticosteroid receptors, including androgen receptor (AR), progesterone receptor, estrogen receptor (ER) α or ERβ, and glucocorticoid receptor (GR). Likewise, in cells expressing nuclear receptor-sensitive reporter genes HE3286 did not substantially stimulate transactivation of AR, ER, GR, or peroxisome proliferator-activated receptor (PPAR) α, PPARδ, and PPARγ. These findings indicate that HE3286 improves glucose homeostasis in diabetic and insulin-resistant mice and suggest that the observed therapeutic effects result from attenuation of proinflammatory pathways, independent of classic sex steroid receptors, corticosteroid receptors, or PPARs.
Insulin resistance is associated with a state of low-grade, chronic inflammation with increased expression of proinflammatory mediators, cytokines, and chemokines in white adipose tissue (WAT) (Hotamisligil et al., 1995; Kern et al., 1995). Consistent with this observation, high blood levels of proinflammatory cytokines (e.g., tumor necrosis factor α, interleukin 6) and C-reactive protein have been reported in diabetic and insulin-resistant subjects (Shoelson et al., 2007). The proinflammatory responses initiated by these proteins occur primarily through the c-Jun NH2-terminal kinase (JNK)/activator protein 1 and the IκB kinase (IKK)/nuclear factor κ-B (NF-κB) signaling pathways (Schenk et al., 2008). Accordingly, the signaling activities of both JNK and IKK are elevated in skeletal muscle from insulin-resistant humans and mice, thereby leading to chronic activation of the activator protein 1 and NF-κB pathways with the consequent expression of proinflammatory genes (Yuan et al., 2001; Hirosumi et al., 2002; Cai et al., 2005). In mice, genetic ablation [knockout (KO) models] or inhibition of JNK1 or IKKβ results in improved insulin sensitivity in vivo and in vitro (Yuan et al., 2001; Hirosumi et al., 2002; Arkan et al., 2005; Cai et al., 2005; Solinas et al., 2007). In humans, treatment with salicylates (Salsalate), which inhibit the IKKβ/NF-κB axis, results in improved glucose and lipid homeostasis (Goldfine et al., 2008). Thus, chronic activation of these proinflammatory pathways is associated with and contributes to the pathogenesis of obesity-related insulin resistance, and their pharmacological modulation appears to produce therapeutic benefit to enhance insulin sensitivity.
Recent studies have shown that WAT from obese animals or humans is heavily infiltrated with macrophages that express most of the detectable proinflammatory molecules, suggesting that they play an important role in insulin resistance and obesity-related inflammation (Weisberg et al., 2003; Xu et al., 2003). Mice in which IKKβ or JNK1 genes have been disrupted specifically in the myeloid lineage (which includes macrophages) exhibit increased insulin sensitivity and are protected from high-fat diet-induced glucose intolerance (Arkan et al., 2005; Solinas et al., 2007). Because the major proinflammatory JNK1/NF-κB or IKKβ/NF-κB axis is disabled within macrophages in these animal models, the resulting insulin-sensitive phenotype suggests an important contribution of the macrophage NF-κB pathway to inflammation-induced insulin resistance (Arkan et al., 2005). Accordingly, insulin sensitizers of the thiazolidinedione (TZD) class that act through peroxisome proliferator-activated receptor γ (PPARγ) are known to attenuate production of inflammatory mediators that promote insulin resistance (Ricote et al., 1998). Moreover, mice deficient in macrophage-specific PPARγ exhibit significant glucose intolerance with systemic insulin resistance and respond only partially to TZD treatment (Hevener et al., 2007; Pascual et al., 2007).
A major potential contributor to macrophage recruitment into WAT during obesity is the soluble mediator monocyte chemoattractant protein-1 (MCP-1), which can be produced by adipocytes, macrophages, and other cells (Charo and Taubman, 2004). Importantly, MCP-1 is an NF-κB target gene that is expressed at high levels in WAT from obese mice and humans (Sartipy and Loskutoff, 2003; Charo and Taubman, 2004; Bruun et al., 2005). Studies with adipocytes in vitro indicate that MCP-1 inhibits insulin-stimulated glucose uptake, inducing an “insulin-resistant” state in cultured cells (Sartipy and Loskutoff, 2003). Moreover, genetic deficiency of the MCP-1 receptor CCR2 (C-C motif chemokine receptor-2) reduces food intake, delays development of obesity, and is accompanied by a decreased number of macrophages in WAT while enhancing systemic glucose homeostasis and insulin sensitivity (Weisberg et al., 2006). Thus, it appears that adipocyte-derived MCP-1 plays a major role in promoting macrophage recruitment and inflammation in the insulin-resistant state.
We have examined a chemically modified version of a naturally occurring 7β-hydroxylated metabolite of Δ5-androstenediol, which in turn is derived from dehydroepiandrosterone (DHEA). We refer to the resulting compound as HE3286, which is metabolically stable (Auci et al., 2009). Although the precise pharmacological role of DHEA is still controversial, its administration is associated with multiple therapeutic effects when administered to rodents in high doses (Coleman et al., 1984). This diversity of biological actions may reflect the well-known extensive metabolic biotransformation of DHEA in rodents, from which discrete steroidal metabolites possessing intrinsic and distinct biological activities are generated. In this light, DHEA may act as a prohormone or metabolic precursor of biologically active hormonal effectors (Leiter et al., 1987; Marwah et al., 2002). Although DHEA has been shown to improve insulin resistance and glucose homeostasis in diverse animal models of metabolic disease, it has not been possible to demonstrate similar activities in human studies (Nair et al., 2006). It is thought that the lack of efficacy in humans may reflect large species-specific differences in metabolic transformation of DHEA and limited oral bioavailability (Auci et al., 2009). The metabolically stabilized, orally bioavailable compound HE3286 has been previously studied in diverse mouse models of inflammation and septic shock and shown to possess anti-inflammatory properties without immune-suppressive activity (Auci et al., 2009).
In this article, we show that HE3286 exhibits antidiabetic effects in mice, lowering glucose, improving glucose intolerance, and delaying progression to hyperglycemia. This therapeutic activity was associated with reduced expression of MCP-1 and its receptor CCR2 in adipose tissue. Because HE3286 also attenuated NF-κB activity and phosphorylation of proinflammatory kinases in lipopolysaccharide (LPS)-stimulated macrophages, the therapeutic effects shown here may result from broad anti-inflammatory properties that enhance insulin sensitivity.
Materials and Methods
Male BKS.Cg-m +/+ Leprdb/J (db/db) mice (5 or 7 weeks old) and male B6.V-Lepob/J (ob/ob) mice (6–7 weeks old) were purchased (The Jackson Laboratory, Bar Harbor, ME) and housed in an environmentally controlled room under a 12-h light/dark cycle with free access to a standard mouse diet and water. After a 7-day acclimation period, blood samples were collected by tail nick for glucose measurements, and a baseline oral glucose tolerance test (OGTT) was performed, after which mice were randomly assigned to treatment groups according to equivalence of body weight and nonfasting and fasting blood glucose levels. Age-matched db/+ or ob/+ heterozygous animals were used for lean controls. For studies with diet-induced obese (DIO) mice, male C57BL/6J mice (5 weeks old) were first fed a high-fat diet (20% kcal protein, 60% kcal fat, 20% kcal carbohydrate; ResearchDiets, New Brunswick, NJ) for 8 weeks, until they reached a target body weight of ≥30 g. Blood glucose level tests and a baseline OGTT were performed, and animals were randomly assigned to groups as described above. All test articles for animal dosing consisted of soluble formulations of HE3286 prepared shortly before each study in a cyclodextrin-based vehicle consisting of 30% (w/v) sulfobutyl-β-cyclodextrin (Captisol; CyDex, Lenexa, KS) dissolved in water at 10 mg/ml and adjusted with NaOH or HCl to a final pH of 6.5 ± 1. Stability of the test articles was verified by high-performance liquid chromatography analysis at least for equivalent periods of time to the duration of the efficacy studies. All animal procedures were performed following protocols designed to adhere to the Guide for Care and Use of Laboratory Animals of the National Institutes of Health and approved by the Institutional Animal Care and Use Committee of Hollis-Eden Pharmaceuticals, Inc.
Oral Glucose Tolerance Tests.
Glucose tolerance was assessed by standard OGTT after an overnight fast. Mice received a bolus of glucose (2 g/kg on days 0 and 14 or 1 g/kg on days 28 or longer) by oral gavage, and blood samples were collected by tail nick 15, 30, 60, and 120 min thereafter. A blood sample for baseline glucose (time 0) was also collected before initiating the OGTT. Blood glucose levels were measured with a glucometer (OneTouch Ultra Meter; LifeScan, Milpitas, CA), but samples that were >600 mg/dl were collected separately with heparin-coated microcapillary tubes and processed by a standard enzymatic method (Sigma-Aldrich, St. Louis, MO).
Serum MCP-1 and Insulin Levels.
Insulin levels were measured in serum by enzyme-linked immunosorbent assay using 96-well microtiter plates coated with mouse-specific anti-insulin monoclonal antibodies [Insulin (Mouse) Ultrasensitive EIA; Alpco Diagnostics, Salem, NH]. Serum MCP-1 levels were determined by enzyme-linked immunosorbent assay with an affinity-purified anti-MCP-1 polyclonal antibody (Quantikine Mouse CCL2/JE/MCP-1 Immunoassay; R&D Systems, Minneapolis, MN). Assays were conducted following the manufacturer's protocol.
RAW264.7 Cells and Transient Transfections.
Murine RAW264.7 macrophages [American Type Culture Collection (Manassas, VA) TIB-71] were maintained in Dulbecco's modified Eagle medium (DMEM; Mediatech, Herndon, VA), with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA) in a humidified incubator at 37°C. For transient transfections, cells were seeded and transfected after 24 h by using the Lipofectamine reagent (Invitrogen) with pNF-κB luciferase plasmid (Stratagene, La Jolla, CA) and pRLTK Renilla luciferase control plasmid (Promega, Madison, WI). After 24 h, cells were exposed to the compounds indicated for 1 h and then stimulated with 100 ng/ml of LPS for 6 h. Cell lysates were prepared, and both firefly and Renilla luciferase activity were determined sequentially by using the Dual-Luciferase reporter assay system (Promega). Luminescence was measured with a GENios Pro plate reader (Tecan, Durham, NC). Results were corrected for well-to-well relative transfection efficiency with respect to Renilla luciferase activity.
Real-Time Quantitative Reverse Transcription-Polymerase Chain Reaction.
Epidydymal adipose tissue was dissected and immediately placed in RNAlater solution (Ambion, Austin, TX) until processed. Total RNA was extracted with a RiboPure RNA purification kit (Ambion), following the manufacturer's instructions. The quality and integrity of RNA was confirmed by OD260/OD280 ratios >1.9 and denaturing agarose gel electrophoresis. First-strand cDNA synthesis was accomplished with 100 ng of RNA and the iScript cDNA Synthesis kit (Bio-Rad, Hercules, CA), and polymerase chain reaction (PCR) amplification of the resulting reverse transcription (RT) products was performed in the presence of iQSYBR Green I Supermix dye (Bio-Rad) and target-specific primer pairs. The mouse acidic ribosomal protein P0 (RPLP0) was used as a reference housekeeping gene in the same PCRs to normalize expression of target genes to a suitable endogenous standard (Dheda et al., 2004). PCR primers used were as follows: MCP-1, 5′-ACTCACCTGCTGCTACTCATTCAC-3′ (forward), 5′-CTTCTTTGGGACACCTGCTGCT-3′ (reverse); tumor necrosis factor α, 5′-CTTGTCTACTCCCAGGTTCTCTT-3′ (forward), 5′-GATAGCAAATCGGCTGACGG-3′ (reverse); CCR2, 5′-GAGCCTGATCCTGCCTCTACTTG-3′ (forward), 5′-CTCTTCTTCTCATTCCTACAGCGA-3′ (reverse); and RPLP0, 5′-CTGAGATTCGGGATATGCTGTTG-3′ (forward), 5′-GTCCTAGACCAGTGTTCTGAGC-3′ (reverse).
Thermocycling conditions included an initial 3-min denaturing step at 95°C followed by 40 successive cycles of denaturation at 95°C for 10 s, annealing at 60°C for 30 s, and extension at 72°C for 20 s. Each PCR amplification was routinely followed by a 15-min melting curve program (95°C for 1 min, 55°C for 1 min and ramping from 55°C to 94°C in 0.5°C increments in 80 cycles of 10 s) and a final cooling step to 4°C. Real-time detection of PCR amplification products was determined by fluorescence with an iCycler iQ Multicolor Detection System (Bio-Rad). Single Tm peak melting curves showed no evidence of primer-dimer formation. Relative quantification of target gene expression was calculated based on real-time PCR efficiency of amplification and the relative difference in threshold crossing points between a sample and a control (Livak and Schmittgen, 2001). Results are expressed as a ratio in comparison to the reference gene (Pfaffl, 2001).
RAW264.7 macrophages were cultured on glass coverslips and treated with 100 nM HE3286 or vehicle [0.1% dimethyl sulfoxide (DMSO)] for 1 h, followed by 100 ng/ml of LPS for 15 min. Cells were fixed with 4% paraformaldehyde in phosphate-buffered saline, washed, and permeabilized in 0.1% Triton X-100 (Sigma-Aldrich). Cells were washed again and blocked with 1% normal goat serum (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) for 1 h, followed by the addition of an anti-NF-κB/p65 monoclonal antibody (IgG1; Santa Cruz Biotechnology, Inc.) and staining with fluorescein isothiocyanate-conjugated goat-anti-mouse IgG1 (Santa Cruz Biotechnology, Inc.). The resulting immunofluorescence was visualized and captured at 40× with a fluorescence microscope.
Murine intraperitoneal macrophages were elicited by thioglycolate and isolated as described (Welch et al., 2003). In brief, cells were seeded in six-well plates (7 × 106/10-cm plate) and maintained in DMEM with 15 mM glucose and 10% FBS for 3 days. Media were changed every 24 h, and cells were then serum-starved in DMEM (15 mM glucose) with 0.5% FBS overnight. For experiments with RAW264.7 mouse macrophages, cells were cultured as described above and serum-starved (0.5% FBS) overnight. In both cases, cells were pretreated with DMSO control (0.01% final) or 100 nM HE3286 for 2 h (intraperitoneal macrophages) or overnight (RAW264.7 macrophages), followed by 100 ng/ml of LPS stimulation for various times. Cell lysates were collected in RIPA buffer [20 mM Tris-HCl (pH 7.4), 0.15 M NaCl, 1 mM EDTA, 1% Triton X-100, and a cocktail of protease (Roche Diagnostics, Indianapolis, IN) and phosphatase (Sigma-Aldrich) inhibitors]. Proteins in cell lysates were resolved by 10% SDS-polyacrylamide gel electrophoresis and then transferred onto polyvinylidene difluoride membranes. Membranes were blocked with 5% nonfat dry milk in TBST [10 mM Tris-HCl (pH 8), 0.15 M NaCl, 0.05% Tween 20] and developed for specific proteins by using the appropriate primary antibodies (all from Cell Signaling Technology Inc., Danvers, MA): phospho-IKKα-Ser180/IKKβ-Ser181, phospho-SAPK/JNK-Thr183/Tyr185, phospho-p38 mitogen-activated protein kinase (MAPK)-Thr180/Tyr182, and phospho-NF-κB-p65/Ser536. For loading controls, blots were stripped and developed with anti-α-tubulin rabbit polyclonal antibody (Cell Signaling Technology Inc.). Immune complexes were detected by enhanced chemiluminescence. Relative band intensities were quantified by densitometry scanning and normalized to untreated controls.
Nuclear Receptor Binding Assays.
Assessment of binding activity for various nuclear receptors was performed by homogeneous competition assays using the PolarScreen fluorescence polarization system (Invitrogen). In brief, serial dilutions of HE3286 (or a reference competitor) were incubated on 384-well plates for 2 h at room temperature in the presence of an appropriate fluorescent ligand (Fluormone) and a nuclear receptor of recombinant origin (AR, kit P3018; GR, kit P2816; ERα, kit P2614; ERβ, kit P2615; PR, kit P2895) in a total volume of 30 μl, essentially following the manufacturer's protocol. Fluorescence polarization in each well was determined with a GENios Pro reader (Tecan), and based on the extent of fluorescence polarization suppression detected IC50 competition values were derived by using GraphPad Prism software (GraphPad Software Inc., San Diego, CA). Most nuclear receptors used in these assays were full-length recombinant proteins of human origin, with the exception of AR (His/glutathione S-transferase-tagged rat AR ligand binding domain) and PR (glutathione S-transferase-tagged human PR ligand binding domain). Reference ligands used for each receptor are indicated in Table 1.
Transactivation activity of sex steroid or corticosteroid receptors was assessed primarily in established stable transfectant human cancer cell lines expressing nuclear receptor-sensitive luciferase reporter genes. For AR and GR, the cell line MDA-kb2 (American Type Culture Collection CRL-2713) harboring a mouse mammary tumor virus/luciferase cassette was used (Wilson et al., 2002). For ERα and ERβ, the cell line T47D-kBluc (American Type Culture Collection 2865) stably transfected with a synthetic plasmid containing three copies of an estrogen response element (ERE) fused upstream of a luciferase gene was used (Wilson et al., 2004). In brief, cells were plated at 20,000 cells/well/100 μl in 96-well clear-bottom white assay microtiter plates (Corning Life Sciences, Lowell, MA) and kept in phenol red-free RPMI 1640 medium supplemented with 4 mM l-glutamine and 10% charcoal-stripped FBS (CHAR-DEX; Invitrogen). Cells were exposed to the various compound dilutions as needed, and after an overnight incubation at 37°C, media were aspirated, cells were lysed, and luciferase activity was then determined. In some cases, transactivation of ERβ and GR was also performed by transient transfection of HEK293 fibroblasts using expression plasmids encoding full-length human GR or ERβ and appropriate luciferase reporter vectors (see below). For transactivation activity of the human PPARs PPARγ, PPARδ, and PPARα, the fluorimetric Gene Blazer β-lactamase assay system (Invitrogen) was used following the manufacturer's instructions. Assays were conducted on 384-well plates using a fluorogenic substrate (CCF4-AM; Invitrogen). Reference compounds (Tocris Bioscience, Ellisville, MO) used for each receptor were: rosiglitazone and GW1929 for PPARγ (EC50 = 1.3 nM); L165,041 for PPARδ (EC50 = 1.1 nM); and GW7647 for PPARα (EC50 = 0.3 nM).
cDNA Expression Constructs.
Full-length cDNA fragments encoding GR and ERβ were cloned by PCR using double-stranded PCR-ready cDNA templates (QUICK-Clone; Clontech, Mountain View, CA) from human adipose tissue (GR) or prostate and ovary (ERβ). Appropriate PCR primers were designed to obtain blunt-end, full-length target amplicons by using AccuPrime Taq polymerase (Invitrogen), thereby making them compatible with topoisomerase-mediated directional Gateway cloning (pENTR-TOPO; Invitrogen). Primers used were: GR, 5′-CACCTGATATTCACTGATGGACTC-3′ (forward), 5′- GGCAGTCACTTTTGATGAAAC-3′ (reverse); and ERβ, 5′- CACCTCTCAAGACATGGATATAAA-3′ (forward), 5′- TCACTGAGACTGTGGGTTCT-3′ (reverse).
Recombinant inserts were subsequently transferred to a pDEST40 expression vector by LR recombination (Gateway-LR Clonase; Invitrogen). The identity of the inserts was verified by full bidirectional sequencing, and their functional competency was assessed by transient cotransfection of HEK293 fibroblasts, using appropriate GRE- or ERE-reporter plasmids (pGRE-SEAP or pERE-TA-SEAP, Clontech; or ERE-/GRE-luciferase constructs generated in pGL3 vectors; Promega, Madison, WI). Cells were exposed to increasing concentrations of reference ligands (dexamethasone for GR and 17β-estradiol for ERβ), and reporter enzyme activity was determined in whole-cell extracts. These titration experiments revealed concentration-dependent increases in reporter activity for each reference ligand and cognate receptor (EC50 = 0.06 nM for 17β-estradiol in ERβ-transfected cells; EC50 = 8 nM for dexamethasone in GR-transfected cells; see Table 1).
All data are expressed as mean ± S.E.M. Statistical significance was assessed by one-way analysis of variance with a Bonferroni post-test and in some cases with a nonparametric Mann-Whitney test. Calculations were performed with GraphPad Prism software. In general, P < 0.05 was considered statistically significant.
Pharmacokinetic Considerations and Dose Rationale.
Although a full description of the pharmacokinetic profile of HE3286 is beyond the scope of this article, there are a few key observations that bear on the rationale for the dosing schedule used in the studies described here. In mice, HE3286 has an absolute oral bioavailability of 16%, and drug exposure is linearly dose-proportional in the range of 20 to 80 mg/kg when administered orally (C. Ahlem, M. Kennedy, T. Page, and J. Frincke, manuscript in preparation). A mean AUC of approximately 2240 ± 960 ng per h/ml is typically achieved after oral administration of a 40 mg/kg HE3286 dose. Because the terminal half-life of HE3286 in mice is relatively short (2.04 ± 1.2 h) and its toxicological evaluation has shown low potential for acute systemic toxicity, the daily dose was limited primarily by volume considerations. In the absence of an established minimal effective dose or optimized dosing schedule, a dose of 80 mg/kg b.i.d. was chosen to provide maximal exposure using our standard research formulation. This dose level is near the maximal practical dose volume (approximately 5 ml/kg for obese mice), and the twice-daily schedule was adopted to compensate for the short half-life of HE3286 to avoid relatively long periods of negligible or no drug exposure (drug holiday), which could negate efficacy if HE3286 was administered only once daily. No evidence of hepatic or renal dose-limiting toxicity has been found in the course of the studies reported here. Furthermore, the doses used are below the minimum nontoxic dose of 200 mg/kg established in 90-day toxicology studies in mice.
Delayed Progression to Hyperglycemia and Enhanced Glucose Tolerance in db/db Mice by HE3286.
Glucose-lowering activity of HE3286 was initially evaluated in two studies using C57BLKs/J-m Leprdb (db/db) mice in which the drug was orally administered (gavage) in doses ranging from 20 to 80 mg/kg b.i.d. In the first study, treatment of 8-week-old diabetic db/db mice with HE3286 (40 or 80 mg/kg b.i.d. for 28 days) significantly reduced progression of hyperglycemia (P < 0.01 at days 10 and 21). In contrast, vehicle-treated animals showed a steady increase in blood glucose levels, reaching 350 to 400 mg/dl (Fig. 1A). In the second study with younger (6 weeks old), “prediabetic” db/db mice, HE3286 (20 or 40 mg/kg b.i.d.) maintained blood glucose levels below 200 mg/dl at all times, comparable with lean db/+ control animals (Fig. 1C; P < 0.001 versus vehicle). As expected, animals treated with rosiglitazone (25 mg/kg b.i.d.) showed normal glucose levels throughout the study (Fig. 1C). In contrast, vehicle-treated mice became hyperglycemic with glucose levels exceeding 450 mg/dl after 32 days. These glucose-lowering effects occurred in the absence of significant changes in body weight because its rate of accretion was virtually the same among all groups in both studies (Fig. 1, B and D). In addition, OGTTs indicated that HE3286, like rosiglitazone, markedly enhanced glucose clearance in db/db mice after 14 days (P < 0.05) or 28 days (P < 0.01) of treatment (Fig. 2, A and B). To determine whether these effects may have resulted from amelioration of insulin resistance, serum insulin levels were also measured (Fig. 2C). Treatment with HE3286 for 4 weeks, like rosiglitazone, caused a significant reduction in insulin levels compared with vehicle (P < 0.01–0.001).
Suppression of Glucose Intolerance in Insulin-Resistant DIO and ob/ob Mice by HE3286.
The previous results suggested that HE3286 might mitigate insulin resistance before the β-cell deficit becomes a major contributor to impaired glucose homeostasis. Therefore, the effect of HE3286 on two insulin-resistant nondiabetic models, DIO C57BL/6J mice and genetically obese B6.V-Lepob/J (ob/ob) mice, was examined. As shown in Fig. 3, A and B, treatment of DIO mice with HE3286 significantly enhanced glucose clearance (P < 0.05–0.001). Moreover, a reduction of basal blood glucose levels was already apparent in the HE3286-treated mice before gavaging the glucose load (see Fig. 3A, time 0). OGTT studies indicated that treatment of ob/ob mice with HE3286 significantly improved (P < 0.05–0.001) glucose intolerance, in a similar fashion as rosiglitazone (Fig. 3, C and D), and also markedly reduced serum insulin levels (see below).
Association of Anti-Inflammatory Activity of HE3286 with Reduced Insulin Resistance.
Given the link between insulin resistance and MCP-1-induced macrophage infiltration and inflammation in adipose tissue (Sartipy and Loskutoff, 2003; Weisberg et al., 2006), it was of interest to assess parallel changes that might occur in the expression of MCP-1 and its receptor CCR2. Expression of mRNA levels of MCP-1 and CCR2 in adipose tissue of diabetic db/db mice receiving vehicle only was markedly increased relative to nondiabetic lean db/+ littermates (Fig. 4A). However, MCP-1 and CCR2 expression was significantly reduced in db/db mice treated with HE3286 or rosiglitazone (P < 0.05) relative to vehicle-treated animals (Fig. 4A). This HE3286-induced inhibition of MCP-1 expression in adipose tissue was accompanied by reduced serum levels of MCP-1 (P < 0.05) in the same animals, but not rosiglitazone (Fig. 4B). The anti-inflammatory activity of HE3286 was also evident in obese insulin-resistant ob/ob mice treated with HE3286, in which serum MCP-1 protein levels decreased (P < 0.01; Fig. 5A). In addition, the marked increase in serum insulin levels observed in obese ob/ob mice relative to lean ob/+ mice was significantly abrogated in animals treated with HE3286 (P < 0.05) but not with vehicle (Fig. 5B), suggesting that insulin resistance was ameliorated. These effects were accompanied by markedly reduced expression of MCP-1 mRNA expression (P < 0.05) in WAT (Fig. 5C). Thus, amelioration of insulin resistance and improved glucose utilization by HE3286 treatment is associated with decreased MCP-1 production in WAT.
Modulation of Macrophage Activation by HE3286.
Because genes encoding MCP-1 and other proinflammatory effectors are under the control of the key transcription factor NF-κB, it was of interest to explore a possible effect of HE3286 on this central regulator of the inflammatory response. LPS stimulation of cultured RAW 264.7 mouse macrophages pretreated with HE3286 displayed markedly reduced p65 nuclear staining, indicating decreased nuclear translocation of NF-κB (Fig. 6A). The effect of HE3286 on NF-κB activation was also demonstrated in time course experiments using freshly isolated mouse peritoneal macrophages, which revealed that the extent of NF-κB/p65 serine phosphorylation was decreased by HE3286 (Fig. 6B). A similar effect was also observed in RAW264.7 macrophages (see below and Fig. 7). To confirm that these effects translate into a functional change in NF-κB-driven gene expression, experiments were conducted with RAW264.7 cells transiently transfected with NF-κB-sensitive promoter/reporter constructs. Prior treatment of RAW264.7 cells with HE3286 decreased subsequent activation of NF-κB-driven reporter gene expression in response to LPS stimulation by 50 to 60% (Fig. 6C). As expected, NF-κB-dependent luciferase expression was also inhibited by dexamethasone or the MAPK inhibitor PD98059. Taken together, these observations indicate that HE3286 limits activation of NF-κB in macrophages in response to LPS and suggest that this anti-inflammatory action may contribute to the observed amelioration of glucose intolerance in vivo.
Suppression of Proinflammatory Signaling Kinase Cascades.
To gain further insight into the mechanism for the observed anti-inflammatory activity of HE3286, the impact of the compound on major proinflammatory kinase cascades engaged by endotoxin stimulation was studied in RAW264.7 macrophages. As expected, stimulation with LPS led to increased phosphorylation of IKK and NF-κB/p65 (see Fig. 6B) and two major proinflammatory MAPK signaling cascades, JNK and p38 (Fig. 7). However, prior exposure to HE3286 resulted in marked suppression in the extent of LPS-induced phosphorylation of these proteins (P < 0.05 for all kinases and NF-κB/p65 at least after 60 min of LPS stimulation; Fig. 7). Although it was not possible to distinguish between IKKα and IKKβ in these experiments given the specificity of the antibodies used (phospho-IKKα/β-Ser 180/181), it is of interest to note that IKKβ has been implicated in down-regulation of insulin receptor substrate-1 signaling through increased phosphorylation on serine residues (Arkan et al., 2005). These findings suggest that HE3286 causes a broad anti-inflammatory effect characterized by impaired LPS-induced upstream activation of IKK and attendant suppression of NF-κB activation, as well as reduced activation of other TLR4-sensitive proinflammatory signaling kinase cascades.
Nuclear Receptor Activity Profile of HE3286.
The possibility that the observed HE3286 effects might occur through nuclear receptors known to influence inflammation and/or glucose homeostasis (i.e., GR, ERs, and PPARs) was addressed by binding and transactivation experiments. As shown in Table 1, HE3286 does not bind to the major sex steroid or corticosteroid receptors, AR, ERα, ERβ, PR, and GR, as indicated by IC50 values of > 10,000 nM in competition binding assays. In addition, transactivation assays revealed no activity of HE3286 in MDA-kb2 cells coexpressing AR and GR (EC50 > 10,000 nM), and essentially no transactivation in transiently transfected HEK293 fibroblasts expressing ERβ (EC50 > 3600 nM). In contrast, evidence for weak activity (EC50 = 268 nM) was found in T47D-kBluc cells coexpressing ERα and ERβ (Table 1), suggesting preferential, but weak, potential transactivation of ERα. However, under the conditions of these assays, the natural ER ligand 17β-estradiol transactivates ERα/ERβ (T47D-kBluc cells) with an EC50 of 0.002 nM. Finally, as shown in Fig. 8, HE3286 did not cause transactivation of PPARα, PPARγ, or PPARδ. Assuming that the transactivation experiments described here are equivalent in other cell types, the combined results suggest that the observed therapeutic effects of HE3286 to improve glucose metabolism in vivo do not result from transactivation of sex steroid receptors, corticosteroid receptors, or PPARs.
In this article, we have studied the novel synthetic compound HE3286, a 17α-ethynyl-substituted analog of the naturally occurring DHEA metabolite Δ5-androstene-3β,7β,17β-triol (Marwah et al., 2002), and show that HE3286 improves glucose metabolism and exhibits antidiabetic effects. Thus, when orally administered to diabetic db/db mice twice a day for 4 weeks, HE3286 delays progression to hyperglycemia in the absence of any significant changes in body weight. Treatment with HE3286 also led to markedly enhanced glucose tolerance in diabetic db/db mice, insulin-resistant DIO mice, and genetically obese ob/ob mice. These effects were accompanied by decreased hyperinsulinemia and reduced expression levels of MCP-1 mRNA in WAT, which is consistent with diminished serum levels of MCP-1 as well. Likewise, reduced mRNA expression of the MCP-1 receptor CCR2 in WAT from db/db mice was observed. These results indicate that, in addition to ameliorating glucose intolerance and insulin resistance, HE3286 possesses anti-inflammatory properties that act to abrogate excessive MCP-1 production in WAT. Preliminary clamp studies in db/db mice suggest that HE3286 enhances insulin-stimulated glucose uptake in skeletal muscle and improves hepatic insulin action by suppressing excessive liver glucose output.
Because the MCP-1 pathway is known to play a role in recruiting macrophages into adipose tissue and developing obesity-related insulin resistance (Xu et al., 2003; Bruun et al., 2005; Weisberg et al., 2006), it is possible that the HE3286-induced improvement in glucose metabolism is, at least in part, a result of anti-inflammatory effects via decreased MCP-1 production. The link between MCP-1 and overall glucose metabolism and insulin action has been uncovered by a number of studies. Thus, transgenic mice overexpressing the MCP-1 gene under the control of the aP2 promoter/enhancer develop insulin resistance and hepatic steatosis, whereas an insulin-sensitive phenotype is observed in homozygous MCP-1 KO mice or animals in which 7ND, a dominant-negative mutant of MCP-1, is acutely expressed after injection of an expression plasmid (Kanda et al., 2006). Furthermore, treatment of DIO-C57BL/6J mice with INCB3344, a selective CCR2 antagonist (Brodmerkel et al., 2005), results in enhanced glucose disposal and insulin sensitivity (Weisberg et al., 2006). Therefore, it seems reasonable to suggest that a drug that attenuates expression and production of MCP-1 can have therapeutic benefit, possibly as a result of diminished macrophage infiltration into WAT with attendant improved glucose homeostasis and increased insulin sensitivity.
The ability of HE3286 to inhibit MCP-1 expression appears to result from broader anti-inflammatory actions as indicated by our results on NF-κB function in macrophages. We found that LPS-induced nuclear translocation of NF-κB was inhibited by HE3286 in cultured RAW264.7 mouse macrophages. HE3286 also caused inhibition of NF-κB-dependent reporter gene expression and a decrease in LPS-induced p65 phosphorylation. These three observations strengthen the conclusion that HE3286 attenuates NF-κB function in macrophages, leading to reduced MCP-1 gene expression. On the other hand, attenuation of NF-κB function per se could translate into therapeutic effects on glucose metabolism. Other reports have shown that inhibition of NF-κB by adenoviral-mediated expression of the IκBα superrepressor SR-IκBα prevents insulin resistance in cultured hepatocytes in vitro (Iimuro et al., 1998) and improves hepatic insulin action and glucose homeostasis in db/db mice in vivo (Tamura et al., 2007). These studies provide strong support to the notion that inhibition of NF-κB activity in diabetes/insulin resistance can improve disordered glucose homeostasis.
Although HE3286 inhibits activation and function of NF-κB in LPS-stimulated macrophages, the basis for this inhibition appears to reside upstream of NF-κB activation. Thus, prior exposure of macrophages to HE3286 resulted in suppression of LPS-induced activation of the IKK/NF-κB axis and two major proinflammatory MAPK pathways (JNK and p38). In macrophages, these kinase signaling cascades are typically activated by pattern-recognition Toll-like receptors, of which TLR4 is the major target of LPS (Medzhitov, 2001). Because it has been shown that TLR4 can transduce the proinflammatory signals of fatty acids (Shi et al., 2006), which are often elevated in insulin-resistant states, it is tempting to speculate that HE3286 might interfere with TLR4 function, leading to suppression of proinflammatory cascades.
Despite the structure of HE3286, which defines it as a C-19 Δ5-androstene species different from glucocorticoids, we tested its ability to bind or transactivate GR and found a lack of activity (see Table 1). Similarly, HE3286 does not exhibit any detectable binding activity toward AR, ERα, ERβ, or PR. The low transactivation activity (EC50 = 268 nM) observed in cells coexpressing ERα and ERβ, but not ERβ alone, suggested a weak potential transactivation of ERα, albeit this activity is virtually negligible when compared with 17β-estradiol (EC50 = 0.002 nM). Furthermore, preliminary results indicate that HE3286 inhibits macrophage NF-κB function in the presence of ERα-selective antagonists and its beneficial effects on glucose homeostasis in vivo are preserved in ERα-deficient mice (unpublished work).
In contrast to TZD-based insulin sensitizers that target PPARγ, HE3286 did not induce PPARγ-mediated transactivation activity compared with rosiglitazone or GW1929 (see Fig. 7). This is an important pathway to explore because it is likely that at least part of the insulin-sensitizing effects of TZDs involve PPARγ-mediated anti-inflammatory responses caused by transrepression of inflammatory mediators (Ricote et al., 1998; Welch et al., 2003). Moreover, deletion of the PPARγ gene from macrophages results in impaired glucose tolerance, insulin resistance, and increased expression of inflammatory effectors, indicating that macrophage PPARγ is required for the full insulin-sensitizing effects of TZDs (Hevener et al., 2007). The fact that HE3286 fails to transactivate PPARγ suggests that its anti-inflammatory and antidiabetic effects occur through a PPARγ-independent pathway.
Although GRs, ERs, and PPARs are important in metabolic regulation and inflammation, there are certain nuclear receptor subtypes that also play an important role in these functions. For example, the liver X receptors LXRα and LXRβ serve as oxysterol sensors that can suppress hepatic gluconeogenesis and attenuate NF-κB function through transrepression mechanisms (Bensinger and Tontonoz, 2008). Likewise, the ER-related receptors ERRβ and ERRγ function through ligand-independent binding of inducible coactivators involved in the liver gluconeogenesis program and mitochondrial function, such as PGC-1α and PGC-1β (Mootha et al., 2004). Similarly, the retinoic acid receptor-related orphan receptor RORα is a key modulator of lipid homeostasis that can increase fatty acid oxidation in skeletal muscle (Lau et al., 2004) and also suppress inflammation and NF-κB function (Delerive et al., 2001). Thus, it is clear that additional work will be necessary to assess the possibility that the beneficial effects of HE3286 might also reflect modulation of nonclassic nuclear receptors.
A major drawback of efficacy studies in rodent models of metabolic disease is the possibility that the pharmacological activity of a new drug may fail to translate in humans. In the particular case of HE3286, we believe that this concern is diminished in light of emerging clinical data in obese, insulin-resistant human subjects, in whom daily treatment with HE3286 for 4 weeks is associated with improved whole-body insulin sensitivity as determined by glucose clamp studies (unpublished work). In these studies, it will be important to continue to explore the underlying hypothesis of the link between inflammation and insulin resistance by determining whether the beneficial effects of HE3286 on glucose homeostasis are associated with anti-inflammatory activity. Furthermore, completion of phase II clinical trials with HE3286 in type 2 diabetic subjects currently in progress should provide additional support for the potential translation of the therapeutic effects described here to long-term glycemic control.
In summary, the current studies demonstrate the antidiabetic and anti-inflammatory properties of HE3286, a synthetic analog of a naturally occurring Δ5-androstene metabolite of DHEA. In rodent models of diabetes and insulin resistance (db/db, ob/ob, and DIO-C57BL/6J mice), HE3286 ameliorated glucose intolerance and suppressed hyperinsulinemia, serum levels, and adipose tissue expression of MCP-1. In addition, in vivo expression of MCP-1 and its receptor CCR2 in adipose tissue was abrogated by HE3286 in db/db mice. These effects do not appear to result from binding or transactivation of nuclear receptors known to influence inflammation and/or glucose homeostasis, including PPARγ, GR, ERα, or ERβ. In LPS-activated macrophages, HE3286 attenuated nuclear translocation of NF-κB, NF-κB-p65 phosphorylation, and NF-κB-dependent reporter gene expression. Likewise, phosphorylation of IKK, JNK, and p38 was also inhibited by HE3286, possibly reflecting upstream disruption of TLR4 signaling. We propose that the antidiabetic profile of HE3286 stems from broad anti-inflammatory activity, which ultimately mitigates the negative impact that obesity-related inflammatory mediators have to impair insulin action and cause insulin resistance.
We thank S. Williams for immunofluorescence images, D. Bell for quantitative real-time RT-PCR analysis, and N. Hein for animal studies.
- Received September 1, 2009.
- Accepted January 8, 2010.
J.M.O. is a consultant for Hollis-Eden Pharmaceuticals, Inc., and J.M.O. and M.L. received a 1-year research grant (2007–2008) from Hollis-Eden Pharmaceuticals.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- monocyte chemoattractant protein-1
- nuclear factor κ-B
- inhibitor of κ-B protein
- IκB kinase
- c-Jun NH2-terminal kinase
- androgen receptor
- glucocorticoid receptor
- estrogen receptor
- progesterone receptor
- peroxisome proliferator-activated receptor
- white adipose tissue
- oral glucose tolerance test
- diet-induced obese
- mitogen-activated protein kinase
- Toll-like receptor-4
- C-C motif chemokine receptor-2
- Dulbecco's modified Eagle medium
- fetal bovine serum
- reverse transcription
- polymerase chain reaction
- dimethyl sulfoxide
- estrogen response element
- glucocorticoid receptor element.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics