Abstract
A radioiodinated ligand, [125I]SB-236636 [(S)-(−)3-[4-[2-[N-(2-benzoxazolyl)-N-methylamino]ethoxy]3-[125I]iodophenyl]2-ethoxy propanoic acid], which is specific for the γ isoform of the peroxisomal proliferator activated receptor (PPARγ), was developed. [125I]SB-236636 binds with high affinity to full-length human recombinant PPARγ1 and to a GST (glutathioneS-transferase) fusion protein containing the ligand binding domain of human PPARγ1(KD = 70 nM). Using this ligand, we characterized binding sites in adipose-derived cells from rat, mouse and humans. In competition experiments, rosiglitazone (BRL-49653), a potent antihyperglycemic agent, binds with high affinity to sites in intact adipocytes (IC50 = 12, 4 and 9 nM for rat, 3T3-L1 and human adipocytes, respectively). Binding affinities (IC50) of other thiazolidinediones for the ligand binding domain of PPARγ1 were comparable with those determined in adipocytes and reflected the rank order of potencies of these agents as stimulants of glucose transport in 3T3-L1 adipocytes and antihyperglycemic agents in vivo: rosiglitazone > pioglitazone > troglitazone. Competition of [125I]SB-236636 binding was stereoselective in that the IC50 value of SB-219994, the (S)-enantiomer of an α-trifluoroethoxy propanoic acid insulin sensitizer, was 770-fold lower than that of SB-219993 [(R)-enantiomer] at recombinant human PPARγ1. The higher binding affinity of SB-219994 also was evident in intact adipocytes and reflected its 100-fold greater potency as an antidiabetic agent. The results strongly suggest that the high-affinity binding site for [125I]SB-236636 in intact adipocytes is PPARγ and that the pharmacology of insulin-sensitizer binding in rodent and human adipocytes is very similar and, moreover, predictive of antihyperglycemic activity in vivo.
Thiazolidinedione insulin sensitizers are a new class of antihyperglycemic agent that after chronic administration to animal models of non–insulin-dependent diabetes, improve glycemic control by enhancing insulin action in target tissues rather than by increasing insulin secretion. Numerous studies using a number of thiazolidinediones indicate that these agents enhance insulin-mediated suppression of hepatic glucose production and promote insulin-stimulated glucose transport into skeletal muscle and adipose tissue (Hulin et al., 1996). Increased glucose disposal, in adipose tissue at least, results from a combination of increased expression of GLUT-4 and increased translocation of GLUT-4 to the adipocyte cell surface in response to insulin (Young et al., 1995). In addition to improving insulin action in diabetic animal models, thiazolidinediones can influence adipocyte functionin vitro. They increase glucose transport in differentiated 3T3-L1 adipocytes in culture, an action mediated by an insulin-independent increase in the expression of glucose transporters (Gibbs et al., 1989), and they can stimulate preadipocyte differentiation and expression of adipose-specific genes (Ibrahimiet al., 1994; Kletzien et al., 1992). The relevance of actions in adipose cells is underscored by the finding that the rank order of potency of thiazolidinediones as antidiabetic agents in vivo is highly correlated with their potencies as adipogenic stimulants in vitro (Lenhard et al., 1996).
Rosiglitazone (BRL-49653), a potent thiazolidinedione insulin sensitizer, has recently been identified as a high-affinity ligand for PPARγ, a nuclear hormone receptor that is abundantly expressed in adipocytes and plays a central role as a regulator of terminal adipocyte differentiation (Lehmann et al., 1995; Tontonezet al., 1994). Two splice variants of PPARγ have been identified. Human and murine PPARγ1 and PPARγ2 bind rosiglitazone with comparable affinities, and both γ subtypes are activated equally by the compound in transactivation assays (Elbrecht et al., 1996). The activation of another PPAR isoform, PPARδ (also known as fatty acid-activated receptor, or NUC-1), by high concentrations of rosiglitazone has been reported by some (Ibrahimi et al., 1994) but not all (Lehmann et al., 1995) researchers, and the relative contributions of activation of PPARδ and PPARγ to the adipogenic response to thiazolidinediones remain controversial.
The implication that one or more PPARs play a key role in the antidiabetic mechanism of thiazolidinediones has arisen entirely from correlations of compound potencies in animal models with potencies/affinities determined in binding and functional studies using recombinant PPAR isoforms (Willson et al., 1996). Direct evidence for the existence of specific binding sites for insulin sensitizers in normal rodent- and human-derived cells is lacking. In this report, to identify specific binding sites in intact rodent and human adipocytes, we describe the use of a high-specific-activity radioiodinated ligand, [125I]SB-236636, that binds with high affinity and specificity to the LBD of recombinant hPPARγ1. The properties of these binding sites, assessed with a variety of PPAR subtype-selective ligands, are comparable in rat and human adipocytes and highly correlated with the pharmacology of binding to recombinant PPARγ1.
Materials and Methods
All cell culture reagents were purchased from GIBCO (Paisley, Scotland, UK). 2-Deoxy-D-[2,6-3H]glucose was purchased from Amersham International (Berkshire, UK). SB-217092 [(±)-3-[4-[2-[N-(2-benzoxazolyl)-N-methylamino]ethoxy]3-iodophenyl]2-ethoxypropanoic acid], [125I]SB-236636 [(S)-(−)-3-[4-[2-[N-(2-benzoxazoyl)-N-methylamino]ethoxy]3-[125]iodophenyl]2-ethoxypropanoic acid; 2500 Ci/mmol], rosiglitazone (BRL-49653) [(±)-5-[[4-[2-[N-methyl-N-(2-pyridyl)amino]ethoxy]phenyl]methyl]2,4-thiazolidinedione], SB-219993 [(R)-(+)-3-[4-[2-[N-(2-benzoxazolyl)-N-methylamino]ethoxy]phenyl]2-(2,2,2-trifluoroethoxy)propanoic acid], SB-219994 [(S)-(−)-3-[4-[2-[N-(2-benzoxazolyl)-N-methylamino]ethoxy]phenyl]2-(2,2,2-trifluoroethoxy) propanoic acid], pioglitazone and troglitazone were synthesized in-house (fig. 1). Collagenase (136–169 U/mg CLS1) was obtained from Lorne Laboratories (Reading, UK), and bovine serum albumin (fraction V) was purchased from Boehringer-Mannheim (Sussex, UK). AEBSF was purchased from Calbiochem (Nottingham, UK). LY 171883, Wy 14643 and ETYA were purchased from BIOMOL Research Laboratories (Plymouth Meeting, PA). 2-Bromopalmitic acid was purchased from Aldrich (Gillingham, Dorset, UK), and 15-deoxy-Δ12,14-prostaglandin J2 was obtained from Cayman Chemical (Ann Arbor, MI). HAWP 02500 (0.45-μm) filters were obtained from Millipore (Watford, UK). All other chemicals were purchased from Sigma (Poole, Dorset, UK). Kits for the estimation of glucose were obtained from Ciba-Corning, Halstead, Essex, UK. The 3T3-L1 fibroblast cell line was obtained from American Type Culture Collection (Rockville, MD).
Animals
Male Sprague-Dawley rats (weight, 200–250 g) used for adipocyte studies were from Charles River (Kent, UK), and female C57Bl/6ob/ob obese mice (age, 9–10 weeks) were from Harlan Olac (Bicester, UK) and Jackson Laboratories (Bar Harbor, ME). Animals were maintained under a 12-hr light/dark cycle. Rats were fed RM1 diet (SDS; Special Diet Services, Witham, Essex, UK) and water ad libitum and maintained at 21 ± 2°C, whereas obese mice were maintained at 26 ± 2°C and fed powdered RM3 diet (SDS) and water ad libitum.
Test Systems
Baculovirus expression of full-length hPPARγ1.
The PPARγ1 insert was excised from a clone containing the coding sequence of hPPARγ1 and ligated into pBacPAK8 [Clontech (Cambridge Bioscience), Cambridge, UK] for sense orientation (pBacPAK/PPARγ1) and into pBacPAK9 (Clontech) for antisense orientation (pBacPAK/PPARγ1-rev). The baculovirus transfer vectors were amplified in Escherichia coli JM 109 (Promega UK, Southampton, UK). The presence of insert was confirmed by PCR using pBacPAK-specific primers (Clontech), and the plasmids were purified using a commercial DNA purification system (Wizard; Promega).
pBacPAK/PPARγ1 or pBacPAK/PPARγ1-rev were cotransfected into Sf9 cells and recombinant protein was produced according to standard procedures (O’Reilly et al., 1994).
E. coli expression of GST-hPPARγ1 LBD-fusion protein.
A cDNA encoding amino acids 174 to 475 of hPPARγ1 was inserted into bacterial expression vector pGEX-5X-1. The chimera was expressed in XL-1 blue E. coli. Extracts were prepared by lysing the cells in 50 mM HEPES ( pH 7.9), 100 mM KCl, 1 mM dithiothreitol and 1% Triton X-100, followed by centrifugation for 30 min at 100,000 × g.
Culture of 3T3-L1 adipocytes.
3T3-L1 cells were grown to confluence and differentiated according to Frost and Lane (1985). Mature adipocytes were used between days 8 and 10 after differentiation. Differentiated cells were treated with compounds, dissolved in DMSO for 48 hr. Fresh compound was added each day, and the concentration of DMSO did not exceed 0.2% (v/v).
Preparation of adipocytes from rat and human adipose tissue.
Rat adipocytes were prepared from epididymal adipose tissue according to Rodbell (1964) with the following modifications. Collagenase was used at a concentration of 2 mg/ml in a HEPES-buffered (30 mM, pH 7.4) Krebs’ solution supplemented with 5.6 mM glucose, 200 nM adenosine and 4% (w/v) BSA.
Human breast adipose tissue was obtained from patients undergoing surgery for carcinoma or cosmetic reduction at St. George’s Hospital Medical School (University of London, Tooting, UK). Care was taken to ensure that adipose tissue samples used in the preparation of adipocytes was freed of blood vessels, lymph or any malignant tissue. All tissue removal and its use had been approved by the St. George’s Hospital Ethical Committee. Adipocytes were prepared from adipose tissue through collagenase digestion (2 mg/ml collagenase; 4 ml medium/g of tissue) according to Rodbell (1964). In these experiments, Krebs-Henseleit medium was used, supplemented with 5.6 mM glucose and gassed with 95% O2/5% CO2. In addition to 4% (w/v) BSA, the medium contained the alpha-2 adrenoceptor agonistp-aminoclonidine (100 nM) and adenosine (200 nM) to inhibit basal lipolysis.
Experimental Procedures
Determination of antihyperglycemic activity in the obese mouse.
Insulin-sensitizer compounds were administered by dietary admixture to glucose-intolerant C57Bl/6 ob/ob obese mice. After 8 days of administration, the antihyperglycemic activity of each compound was determined from the measurement of changes in tolerance to an oral glucose load (Cantello et al., 1994). A 25% reduction in the area under the blood glucose vs. time curve (ED25), compared with controls, was considered to be a half-maximal effective dose of compound (Cantello et al., 1994).
2-Deoxyglucose transport in 3T3-L1 adipocytes.
For determination of glucose transport rates, differentiated 3T3-L1 cells were washed three times in DPBS and incubated in serum-free DMEM for 2 hr. After removal of the DMEM and three washes with DPBS, Krebs-Ringer phosphate buffer was added, and the cells were incubated for 15 min. 2-Deoxyglucose uptake was initiated by the addition of 2-deoxy-d-[2,6-3H]glucose (25 μM; 1 μCi/well). After 10 min, the medium was aspirated, and the cells were washed three times with ice-cold DPBS. The cell monolayers were allowed to air-dry and then dissolved in 1 M NaOH. Aliquots were removed for scintillation counting and protein estimation.
Western blot analysis of PPARγ.
3T3-L1 fibroblasts and 3T3-L1 adipocytes were homogenized in lysis buffer (50 mM Tris·HCl, pH 7.5, 150 mM NaCl, 0.2 μM AEBSF, 1 mM EDTA, 10 μM leupeptin, 1 μM pepstatin and 1 μg/ml aprotinin). Then, 25 μg of protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membranes (Millipore, Watford, UK) (Young et al., 1995) and blocked overnight in blocking buffer (10 mM Tris·HCl, 150 mM NaCl, 5% nonfat dry milk, 0.05% Tween-20). Blots were sequentially incubated with rabbit polyclonal anti-PPARγ serum and a horseradish peroxidase-coupled anti-rabbit secondary antibody for 1 hr, followed by three washes with Tris-buffered saline containing 0.05% Tween-20. Polyclonal antibodies to PPARγ were raised in the rabbit, using the keyhole limpet hemacyanin-conjugated peptide SEKTQLYNRPHEEPSN (amino acids 87–102 of mouse PPARγ1 and amino acids 117–132 of mouse PPARγ2) as antigen. Blots were visualized with the supersignal CL-HRP substrate system according to the manufacturer’s instructions (Pierce Chemicals, Chester, UK).
PPAR isoform mRNA expression.
Total RNA was extracted from 3T3-L1 cells or adipocytes using a single-step extraction solution according to the manufacturer’s instructions (TRIzol; Life Technologies, Renfrewshire, Scotland, UK). For PCR, 1 μg of total RNA was treated with RNase-free DNase I (Life Technologies) to remove contaminating genomic DNA before cDNA synthesis using Superscript II (Life Technologies) and oligo(dT) priming. PCR primers to each PPAR isoform were designed using a primer design package (Oligo v. 5.0; NBI, Plymouth, MN) and are shown in table 1. PCR was performed in a total volume of 50 μl with 0.5 μM concentration of primers, 25 ng of cDNA and 0.25 U of Taqpolymerase (Life Technologies). After an initial denaturing step for 4 min, a two-step cycling protocol was used (68/94°C for 1 min each) for 30 cycles. Adipocyte content of specific PPAR isoforms was also determined by slot-blot analysis in which 15 μg of total RNA was hybridized with digoxygenin-labeled oligonucleotide probes for PPARα and PPARδ (table 1). PPARγ was hybridized with a digoxygenin-labeled cDNA probe synthesized by PCR from human adipose cDNA using the primers shown in table 1. A detailed description of this methodology is given elsewhere (Clapham et al., 1997).
Radioligand Binding Studies
Human PPARγ1LBD-GST fusion protein.
Binding of [125I]SB-236636 to crude lysates of E. coli expressing GST-hPPARγ1 LBD fusion protein was carried out in 96-well plates at 4°C for 16 hr. The assay consisted of 1.4 μg of crude protein and 150 pM [125I]SB-236636 in a total volume of 50 μl of lysis buffer. Competing compounds were dissolved in DMSO (final concentration in the incubation, <0.02%) and present at the concentrations indicated in the legends to figures. Bound ligand was separated from free on mixed cellulose acetate filters (Inoue et al., 1983). Nonspecific binding was assessed with 100 μM rosiglitazone.
Full-length hPPARγ1.
Cell pellets from Sf9 cells transfected with PPARγ1 were solubilized with 30 mM HEPES (pH 7.9) containing 300 mM KCl, 1 mM dithiothreitol, 1 mM AEBSF and 1% Triton X-100. Binding experiments were performed as detailed for GST-hPPARγ1 LBD fusion protein.
3T3-L1 adipocytes.
3T3-L1 cells were grown to confluence and differentiated in 35-mm-diameter six-well plates. Cell monolayers were washed three times in phosphate-buffered saline and incubated for 1 hr at 37°C in DMEM containing 30 pM [125I]SB-236636 and varying concentrations of competing compounds. The cell monolayers were then washed three times with phosphate-buffered saline and dissolved in 1 ml of 1 M NaOH. Aliquots of the cell lysate were counted for cell-associated [125I]SB-236636. Nonspecific binding of [125I]SB-236636 (obtained in the presence of 10 μM rosiglitazone) was 30% of total binding.
Rat adipocytes.
After preparation, adipocytes were rinsed in DMEM/Ham’s F-12 nutrient mix medium containing 15 mM HEPES (pH 7.4) supplemented with 200 nM adenosine. After three washes to remove collagenase and BSA, 0.5 ml of cells was aliquoted (triplicate incubations) into tubes containing [125I]SB-236636 to yield a final concentration of the radioligand of 30 pM. Each adipocyte preparation was diluted to an adipocrit of 10% (v/v) before aliquoting. Binding was carried out at 37°C for 1 hr in a shaking water bath. Cell-associated radioactivity was assessed after separation of cells from the medium by centrifugation of cells (200 μl of incubation medium, in duplicate) through silicone oil (Dow Corning 200/200 cs) for 20 sec at 10,000 × g using a Beckman Instruments (Palo Alto, CA) microfuge (Green, 1983). Cell pellets were then obtained by cutting the microfuge tube, and cell-associated radioactivity was measured using a Wallac (Gaithersburg, MD) Wizard γ Counter. Nonspecific binding was assessed in the presence of 10 μM rosiglitazone and was 50% of total binding.
Human adipocytes.
The methodology was essentially the same as that described for rat adipocyte binding except incubations were carried out in quintuplicate and the DMEM/F-12 medium was supplemented with 25 mM HEPES, pH 7.4, containing 200 nM adenosine and 100 nMp-aminoclonidine to suppress basal lipolysis (Berlan and Lafontan, 1982; Larrouy et al., 1991). Specific binding was 30% of total binding. Nonspecific binding was determined in the presence of 10 μM rosiglitazone.
Results
Evaluation of SB-236636 as a Radioligand
To assess the potential of [125I]SB-236636 as a radioiodinated ligand to detect molecular targets for rosiglitazone in insulin-responsive cells, the insulin-sensitizing properties of the nonradioactive racemic compound, SB-217092, were determined in the obese mouse and differentiated 3T3-L1 cells. SB-217092 was equipotent with rosiglitazone as an antihyperglycemic agent in the obese mouse (table 2). In addition, SB-217092 was a potent stimulant of glucose transport in differentiated 3T3-L1 adipocytes. The EC50 value was 9 nM, which is comparable with that of rosiglitazone (EC50 = 11 nM) (table 2).
Binding of [125I]SB-236636 to Human PPARγ1 LBD and Recombinant Full-Length Human PPARγ1
Saturation binding of [125I]SB-236636 to a crude lysate of E. coli expressing GST-hPPARγ1 LBD fusion protein yielded aKD value of 70 nM (fig.2). Competition of [125I]SB-236636 binding by a number of thiazolidinedione antihyperglycemic agents with different in vivo potencies is shown in table 3. The rank order of binding affinities (IC50values) was rosiglitazone (41 nM) > pioglitazone (4830 nM) > troglitazone (7970 nM). In addition, binding was stereoselective, because SB-219994 [(S)-enantiomer; fig. 1) possessed an IC50 value of 2.1 nM, whereas the corresponding (R)-enantiomer, SB-219993, had an IC50 value of 2770 nM (table 3). The PPARγ1 LBD may indeed have absolute specificity for the (S)-configuration because the batch of SB-219993 used in these studies contained a small percentage (0.26%) of SB-219994.
Specific binding of [125I]SB-236636 was detected in crude extracts of Sf9 cells transfected with hPPARγ1. No specific binding was detectable in extracts of nontransfected cells or cells transfected with PPARδ or the PPARγ1 insert in the reverse (antisense) orientation. Competition experiments showed that rosiglitazone was a potent competitor of binding (IC50 = 62 nM; table3). Again, competition of [125I]SB-236636 binding showed stereoselectivity; the IC50 for SB-219994 [(S)-enantiomer] was 1.8 nM compared with 1210 nM for SB-219993 [(R)-enantiomer]. Thus, ligand binding affinities measured using only the LBD of PPARγ1 (GST-PPARγ1 LBD) are comparable with those determined using the full-length receptor.
Binding Studies in Intact Cells
3T3-L1 cells.
In nonconfluent, nondifferentiated 3T3-L1 fibroblasts, no specific binding of [125I]SB-236636 was detected. However, differentiation of these cells to lipid-filled adipocytes lead to the appearance of specific binding of [125I]SB-236636 (2000 dpm/mg of protein at a free ligand concentration of 30 pM). The appearance of [125I]SB-236636 binding sites during differentiation coincided with the expression of PPARγ. PPARγ1 and γ2 mRNA, determined by PCR and slot-blot analysis, and PPARγ1 and γ2 protein, as assessed by Western blotting, were present only in differentiated 3T3 cells displaying adipocyte-like morphology (table4, fig. 3). Competition of [125I]SB-236636 binding in differentiated 3T3-L1 cells by insulin sensitizers showed the same rank order of potency as that determined at recombinant PPARγ in cell-free extracts. IC50 values were SB-219994 (0.6 nM) > rosiglitazone (4 nM) > SB-219993 (700 nM) > troglitazone (2000 nM).
Rat adipocytes.
Initial experiments showed that specific binding of tracer amounts of [125I]SB-236636 to intact rat white adipocytes reached equilibrium within 40 min at 37°C (table 5). Scatchard analysis of saturation binding of [125I]SB-236636, assessed in three separate adipocyte preparations, yielded aKD value of 2.4 nM and aBmax of 170 fmol/mg of protein (fig.4).
Competition of [125I]SB-236636 binding in rat adipocytes by rosiglitazone and a number of other insulin sensitizers and potential PPAR ligands is shown in table6. Rosiglitazone had a much higher binding affinity (IC50 = 12 nM) than other thiazolidinedione insulin sensitizers, such as pioglitazone (580 nM) and troglitazone (420 nM). This is consistent with the higher affinity of rosiglitazone for the LBD of recombinant hPPARγ and, furthermore, is in agreement with its greater potency in vitro as a stimulant of glucose transport in 3T3-L1 cells (EC50 values for rosiglitazone, pioglitazone and troglitazone were 11, 650 and 1050 nM, respectively) and as an antidiabetic agent in vivo (Cantello et al., 1994) (table 2).
In confirmation of results obtained with recombinant hPPARγ1, competition of [125I]SB-236636 binding in rat adipocytes by insulin sensitizers was stereoselective (table 6). SB-219994 was ≈550-fold more potent than SB-219993. Again, the much higher affinity of the (S)-enantiomer for the specific binding site in the rat adipocyte correlated with the 100-fold greater potency of SB-219994 as an antihyperglycemic agent in the obese mouse (table 2).
Other PPAR ligands were tested for their ability to compete with [125I]SB-236636 binding. The leukotriene D4 antagonist LY 171883, the potent hypolipidemic agent Wy 14643 and ETYA, all cited as PPARα ligands (Bocos et al., 1995; Kliewer et al., 1994) were weak competitors of [125I]SB-236636 binding. The putative PPARδ ligand 2-bromopalmitic acid (Ibrahimi et al., 1994) displaced only 35% of specifically bound [125I]SB-236636 at the maximum concentration tested (10 μM).
Antidiabetic agents that are structurally distinct from thiazolidinediones but claimed to enhance insulin action also were tested in competition binding assays. The biguanide metformin was inactive, although the sulfonylurea insulin secretagogue glibenclamide competed with an IC50 value of 1000 nM. Interestingly, glibenclamide also competed with [125I]SB-236636 binding to GST hPPARγ1-LBD with an IC50of ≈5000 nM (data not shown).
Human adipocytes.
To determine whether human adipocytes possess a population of high-affinity binding sites similar to those present in rat adipocytes, a parallel series of experiments was conducted using adipocytes freshly prepared from human mammary adipose tissue. In human adipocytes, as in rat adipocytes, cell-specific binding of [125I]SB-236636 reached equilibrium within 40 min at 37°C (table 5). Scatchard analysis of a saturation binding study performed with a typical adipocyte preparation yielded aKD value of 0.8 nM with aBmax of 49 fmol/mg of protein (fig.5). Thus, the affinity and number of insulin-sensitizer binding sites in rat and human adipocytes are comparable.
IC50 values for thiazolidinediones and other putative PPAR ligands are shown in table 6. Rosiglitazone had a high-binding affinity (IC50 = 10 nM), which was identical to that determined in rat adipocytes. In addition, rosiglitazone bound with higher affinity than either pioglitazone or troglitazone. As noted with rat adipocytes, the human adipocyte insulin-sensitizer binding site showed stereoselectivity. Thus, the IC50 value for displacement of [125I]SB-236636 binding by SB-219994 was 55-fold lower than that of SB-219993. 2-Bromopalmitic acid, LY 171883, ETYA and Wy 14643 were all weak competitors of [125I]SB-236636 binding.
In contrast, 15-deoxy-Δ12,14-prostaglandin J2, which has been shown to bind to the LBD of mPPARγ, induce transcriptional activity of an mPPARγ-containing chimeric receptor and promote adipocyte differentiation (Formanet al., 1995; Kliewer et al., 1995), competed for [125I]SB-236636 binding with high affinity (IC50 = 45 nM).
Discussion
It is well established that thiazolidinediones improve glycemic control in animal models of non–insulin-dependent diabetes mellitus by increasing insulin sensitivity of the liver, muscle and adipose tissues (Hulin et al., 1996). The potency with which these agents bind to, and activate, PPARγ, a ligand-activated nuclear receptor that is a key regulator of adipogenesis, is highly correlated with antidiabetic activity in vivo and suggests that this receptor might be pivotal in the insulin-sensitizing mechanism (Lehmannet al., 1995). This link has, however, been derived exclusively from studies using recombinant PPARγ in cell-free or transfected cell systems, and evidence for the existence of specific binding sites for insulin sensitizers in intact rodent- or human-derived cells in a more physiological setting is lacking.
To identify and characterize specific binding sites for insulin sensitizers in cells and tissues, a high-specific-activity radioiodinated ligand, [125I]SB-236636, was developed. The choice of ligand was predicated on the following: (1) it must contain iodine to allow radiolabeling to a high specific activity, thereby enabling detection of low abundance receptors. (2) It must be cell penetrant and have a high affinity for the molecular target or targets, as demonstrated by potent insulin-sensitizing activity bothin vivo and in vitro. SB-217092 [derived from the α-ethoxyacid SB-213068, which is a highly potent antihyperglycemic agent and high-affinity PPARγ ligand (Buckleet al., 1996)], was chosen as the iodine-containing insulin sensitizer because it had comparable potency to rosiglitazone as an antidiabetic agent in the obese mouse and as a stimulant of glucose transport in 3T3-L1 adipocytes. The observation that the (S)-enantiomer of α-substituted β-phenylpropanoic acid antihyperglycemic agents is considerably more potent than the (R)-enantiomer prompted us to radioiodinate only the (S)-enantiomer of SB-217092 to produce [125I]SB-236636 (D. Haigh, personal communication).
Before attempting to identify binding sites in whole cells and subcellular fractions, the characteristics of [125I]SB-236636 binding to recombinant PPAR isoforms were determined. Binding of [125I]SB-236636 was PPARγ specific. Although [125I]SB-236636 bound with high affinity to both a fusion protein containing the LBD of hPPARγ1 and to full-length hPPARγ1, no specific binding was detectable in extracts of Sf9 cells transfected with a plasmid containing PPARγ1 in antisense orientation. No specific binding could be detected to either a hPPARα LBD GST fusion protein (data not shown) or full-length mouse PPARδ. This could simply be a consequence of low intrinsic affinity of [125I]SB-236636 for PPARα and PPARδ or may result from incorrect folding of the recombinant proteins after cell lysis. The latter possibility is considered unlikely because we also were unable to detect specific radioligand binding in intact cell types known to express high levels of PPARα or PPARδ (data not shown). Moreover, direct binding of [3H]leukotriene B4 to E. coli-expressed PPARα LBD-GST fusion protein has recently been reported, suggesting that the LBD, of this PPAR isoform at least, folds correctly after cell lysis (Devchand et al., 1996).
The binding affinity of the thiazolidinedione rosiglitazone at GST-hPPARγ1 LBD (IC50 = 41 nM), which was assessed using [125I]SB-236636 as ligand, is comparable to that reported for rosiglitazone at mPPARγ (KD = 43 nM) using [3H]rosiglitazone as ligand (Lehmann et al., 1995); this is predicted from the high homology (98%) of the LBDs of the human and murine receptors. However, the binding affinity of rosiglitazone for full-length hPPARγ1determined in our studies is ≈8-fold higher than the affinity of rosiglitazone for recombinant hPPARγ1 and γ2 reported recently (Elbrecht et al., 1996) using [3H]AD-5075 as a radioligand. The reason for this discrepancy is unclear. In the present study, we also now show that ligand binding to PPARγ1 is stereoselective, with the (S)-enantiomer of the α-trifluoroethoxy propanoic acid insulin sensitizer, SB-219994, having an IC50value of ≈1000-fold lower than that of the (R)-enantiomer, SB-219993. Thus, ligand enantioselectivity of PPARγ is the same as that displayed by PPARα, which is preferentially activated by 8-(S)- but not 8-(R)-HETE (Yu et al., 1995).
In a search for thiazolidinedione binding sites in insulin-sensitive target tissues, [125I]SB-236636 binding studies were performed initially using cytosolic and membrane fractions and nuclear extracts prepared from rat liver, skeletal muscle and adipose tissue. No specific radioligand binding was detected in any subcellular fraction, possibly reflecting low abundance of the receptor or liberation of endogenous ligand on cell lysis, which masks specific binding. However, specific, saturable binding was observed in freshly prepared rat epididymal and human mammary tissue-derived adipocytes. The rank order of potency and enantioselective competition of [125I]SB-236636 binding by insulin sensitizers in intact adipocytes was identical with the pharmacology shown by recombinant PPARγ1 and strongly supports the contention that the specific binding site in adipocytes is PPARγ. In addition, specific binding was undetectable in undifferentiated 3T3-L1 cells, in which PPARγ is not expressed (table 4, fig. 3). The appearance of specific binding sites coincided with terminal differentiation and expression of PPARγ (Tontonez et al., 1994). Further support that the adipocyte binding site for [125I]SB-236636 is PPARγ is provided by the demonstration that putative PPARα-, and PPARδ-, selective agonists, including the nonmetabolizable fatty acid 2-bromopalmitate, the hypolipidemic agent Wy 14643, the leukotriene D4antagonist LY 171883 and the synthetic arachidonic acid analog ETYA, were all poor competitors. All of these agents also are low-potency PPARγ ligands, as assessed from radioligand binding or functional transactivation studies.
Scatchard analysis of saturation binding curves produced binding site densities of 170 and 49 fmol/mg of protein for rat and human adipocytes, respectively, values that are close to those of other nuclear hormone receptors such as thyroid (Inoue et al., 1983) and glucocorticoid (Sheppard and Funder, 1996) receptors. The affinity of the insulin-sensitizer binding site for [125I]SB-236636 in intact adipocytes was ≈30-fold (rat) to ≈140-fold (human) higher than theKD value measured with recombinant PPARγ1 LBD. In addition, thiazolidinedione and acyclic antihyperglycemic agents were more potent competitors (1.5–38-fold) of radioligand binding in intact adipocytes than of binding to either full-length hPPARγ1 or hPPARγ1 LBD in cell-free extracts, although the rank order of binding affinities was maintained in all the test systems. Moreover, in absolute terms, potencies of insulin sensitizers as stimulants of glucose transport in 3T3-L1 adipocytes were in closer agreement with IC50 values determined in intact adipocytes than with those assessed using recombinant PPARγ protein.
Murine adipocytes express two forms of PPARγ, γ1 and γ2, which arise from differential promoter use and alternative splicing (Zhu et al., 1997). Recently, the molecular cloning of hPPARγ1 and γ2 homologs was reported (Elbrecht et al., 1996). The question arises of whether binding sites in intact adipocytes represent PPARγ1, PPARγ2, or both. Although relative expression of the two isoforms in adipocytes at the protein level has not been rigorously quantified, because isoform-specific antibodies are not available, murine and human recombinant γ1 and γ2isoforms appear to have identical ligand-binding properties and show comparable transactivational responses to rosiglitazone and other thiazolidinediones (Elbrecht et al., 1996; Lehmann et al., 1995). It therefore is likely that [125I]SB-236636 binding sites in adipocytes consist of γ1 and γ2components.
The endogenous ligand or ligands for PPARγ have not been unequivocally identified, although recently, Forman et al. (1995) and Kliewer et al. (1995) demonstrated that prostanoids can activate PPARγ in transactivation assay systems. Of a range of arachidonic acid metabolites tested, 15-deoxy-Δ12,14-prostaglandin J2 was the most potent activator of PPARγ and inducer of adipogenesis. The prostanoid also displaced [3H]rosiglitazone binding to PPARγ LBD, although with only low affinity (Ki ≈ 2.5 μM). In contrast, the present results show that 15-deoxy-Δ12,14-prostaglandin J2 was a potent competitor of [125I]SB-236636 binding in human adipocytes (IC50 = 45 nM) and further supports the conclusion, drawn from the competition studies with rosiglitazone and a number of other insulin sensitizers, that the high-affinity binding site for [125I]SB-236636 in human adipocytes is PPARγ. The discrepancy in binding affinities of 15-deoxy-Δ12,14-prostaglandin J2 determined in intact adipocytes and at recombinant PPARγ in cell lysates is entirely consistent with the data obtained for thiazolidinediones and acyclic insulin sensitizers. The apparent increase in ligand affinity and potency in the intact cell assay is perhaps not surprising because the receptor would be in the optimal conformation and also have access to partner proteins, such as retinoid × receptorα, with which PPARγ heterodimerizes. Although 15-deoxy-Δ12,14-prostaglandin J2 can undoubtedly promote adipogenesis in vitro, the physiological relevance is unclear because its synthesis by adipose tissue has not been reported. Furthermore, the role of prostanoids in the regulation of PPARγ function in fully differentiated lipid-filled adipocytes, as opposed to preadipocytes, is unknown.
Examples from two other chemical classes of orally active antidiabetic agent were tested for their ability to bind to PPARγ1 in cell-free extracts and intact adipocytes. The biguanide metformin was inactive, but surprisingly, the sulfonylurea insulin secretagogue glibenclamide had an IC50 value of 1000 nM in rat adipocytes and an IC50 value at recombinant PPARγ1 LBD of 5000 nM. Direct insulin-sensitizing properties of relatively high millimolar concentrations of sulfonylureas in rat adipocytes in culture have been reported, and activation of PPARγ may be associated with this response (Altan et al., 1985; Zuber et al., 1985).
In summary, this is the first report detailing the use of a radioiodinated ligand that binds with high affinity and specificity to both rodent and hPPARγ to identify specific binding sites in intact rat and human adipocytes. We have shown that for a range of thiazolidinedione and acyclic insulin sensitizers and other PPAR ligands, there is a high correlation between binding affinities determined in rat and human adipocytes with those measured in binding assays using recombinant PPARγ. It therefore is very likely that PPARγ is the molecular species to which [125I]SB-236636 binds in intact adipocytes. This ligand will prove useful in identifying and characterizing binding sites in cell types other than adipocytes, in which insulin action is enhanced by antihyperglycemic compounds such as rosiglitazone.
Footnotes
-
Send reprint requests to: Dr. S. A. Smith, SmithKline Beecham Pharmaceuticals, New Frontiers Science Park, Third Avenue, Harlow, Essex, CM19 5AW UK. E-mail: Stephen A Smith{at}sbphrd.com
- Abbreviations:
- PPAR
- peroxisomal proliferator-activated receptor
- hPPAR
- human peroxisomal proliferator-activated receptor
- mPPAR
- murine peroxisomal proliferator-activated receptor
- GLUT-4
- glucose transporter isoform 4
- GST
- glutathione S-transferase
- AEBSF
- 4-(2-aminoethyl)benzenesulfenylfluoride hydrochloride
- ETYA
- 5,8,11,14-eicosatetraynoic acid
- HETE
- 8-hydroxyeicosatetraenoic acid
- PCR
- polymerase chain reaction
- HEPES
- 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid
- DMSO
- dimethylsulfoxide
- BSA
- bovine serum albumin
- DPBS
- Dulbecco’s phosphate-buffered saline
- DMEM
- Dulbecco’s modified Eagle’s medium
- LBD
- ligand binding domain
- Received June 17, 1997.
- Accepted October 17, 1997.
- The American Society for Pharmacology and Experimental Therapeutics