Pharmacogenetic Evidence That Cd36 Is a Key Determinant of the Metabolic Effects of Pioglitazone*

Pioglitazone, like other thiazolidinediones, is an insulin-sensitizing agent that activates the peroxisome proliferator-activated receptor γ and influences the expression of multiple genes involved in carbohydrate and lipid metabolism. However, it is unknown which of these many target genes play primary roles in determining the antidiabetic and hypolipidemic effects of thiazolidinediones. To specifically investigate the role of the Cd36 fatty acid transporter gene in the insulin-sensitizing actions of thiazolidinediones, we studied the metabolic effects of pioglitazone in spontaneously hypertensive rats (SHR) that harbor a deletion mutation in Cd36 in comparison to congenic and transgenic strains of SHR that express wild-typeCd36. In congenic and transgenic SHR with wild-typeCd36, administration of pioglitazone was associated with significantly lower circulating levels of fatty acids, triglycerides, and insulin as well as lower hepatic triglyceride levels and epididymal fat pad weights than in SHR harboring mutant Cd36. Additionally, insulin-stimulated glucose oxidation in isolated soleus muscle was significantly augmented in pioglitazone-fed rats with wild-type Cd36 versus those with mutantCd36. The Cd36 genotype had no effect on pioglitazone-induced changes in blood pressure. These findings provide direct pharmacogenetic evidence that in the SHR model, Cd36is a key determinant of the insulin-sensitizing actions of a thiazolidinedione ligand of peroxisome proliferator-activated receptor γ.

Pioglitazone, like other thiazolidinediones, is an insulin-sensitizing agent that activates the peroxisome proliferator-activated receptor ␥ and influences the expression of multiple genes involved in carbohydrate and lipid metabolism. However, it is unknown which of these many target genes play primary roles in determining the antidiabetic and hypolipidemic effects of thiazolidinediones. To specifically investigate the role of the Cd36 fatty acid transporter gene in the insulin-sensitizing actions of thiazolidinediones, we studied the metabolic effects of pioglitazone in spontaneously hypertensive rats (SHR) that harbor a deletion mutation in Cd36 in comparison to congenic and transgenic strains of SHR that express wild-type Cd36. In congenic and transgenic SHR with wild-type Cd36, administration of pioglitazone was associated with significantly lower circulating levels of fatty acids, triglycerides, and insulin as well as lower hepatic triglyceride levels and epididymal fat pad weights than in SHR harboring mutant Cd36. Additionally, insulin-stimulated glucose oxidation in isolated soleus muscle was significantly augmented in pioglitazone-fed rats with wild-type Cd36 versus those with mutant Cd36. The Cd36 genotype had no effect on pioglitazone-induced changes in blood pressure. These findings provide direct pharmacogenetic evidence that in the SHR model, Cd36 is a key determinant of the insulin-sensitizing actions of a thiazolidinedione ligand of peroxisome proliferator-activated receptor ␥.
Thiazolidinediones such as pioglitazone and rosiglitazone have attracted considerable attention for the treatment of type II diabetes. The insulin-sensitizing effects of these compounds are believed to be related at least in part to their ability to bind to the peroxisome proliferator activated receptor ␥ (PPAR␥), 1 a nuclear hormone receptor that regulates the expression of multiple genes involved in the control of carbohydrate and lipid metabolism (1)(2)(3)(4).
Although many genes and metabolic pathways are likely to be involved in the insulin-sensitizing action of PPAR␥ ligands, it has been proposed that the effect of thiazolidinediones on genes regulating free fatty acid (FFA) transport and metabolism may be one of the key mechanisms responsible for the antidiabetic effects of these drugs (2,3,(5)(6)(7)(8)(9)(10). Chronic elevations of FFA levels are a well known determinant of insulin resistance (11)(12)(13)(14). Accordingly, a number of investigators have suggested that thiazolidinediones may attenuate insulin resistance by increasing the metabolic clearance of FFA and decreasing FFA levels in the circulation.
The fatty acid transporter CD36 is one of a number of molecules that mediate the uptake of FFA by adipocytes and muscle cells (15)(16) and is a well known target of PPAR␥ ligands (17). Mutations in Cd36 have been found to be associated with impaired carbohydrate and lipid metabolism in both humans and laboratory animals (18 -22) Recently, Hevener et al. (6) reported that infusion of FFA induces reductions in systemic glucose disposal rate that are paralleled by decreases in muscle expression of CD36. These investigators also found that oral administration of the thiazolidinedione troglitazone can attenuate the decreases in CD36 expression and glucose disposal rate otherwise induced by infusion of FFA (6). Based on these correlational observations, Hevener and colleagues have proposed that the insulin-sensitizing effects of thiazolidinediones may at least be partly mediated by their effects on Cd36 expression and FFA metabolism (6).
The availability of animal models lacking Cd36 provides an opportunity to use a pharmacogenetic approach to specifically test the proposed role of Cd36 in the insulin-sensitizing actions of thiazolidinediones. Based on the hypothesis of Hevener and colleagues, it might be predicted that the insulin-sensitizing effects of thiazolidinediones would be attenuated in animals lacking Cd36 (6). We now report a positive test of this hypothesis by showing that the insulin-sensitizing actions of pioglitazone are impaired in spontaneously hypertensive rats harboring mutant Cd36 compared with congenic and transgenic strains of spontaneously hypertensive rats expressing wildtype Cd36. Whereas previous studies have demonstrated that PPAR␥ ligands can alter the expression of multiple genes related to carbohydrate and lipid metabolism (3,4), the current findings provide direct pharmacogenetic evidence that Cd36 is a key determinant of the insulin-sensitizing actions of pioglitazone, a PPAR␥ ligand widely used for the treatment of type II diabetes.

EXPERIMENTAL PROCEDURES
Animals-We compared the SHR progenitor strain carrying a spontaneous mutation in Cd36 that abolishes membrane expression of CD36 protein to a congenic strain and transgenic strain of SHR that express wild-type Cd36 (19 -21). The progenitor strain of SHR with defective Cd36 descends from inbred SHR originally obtained from National Institutes of Health and was beyond the F90 generation when derivation of the congenic and transgenic strains was initiated. The defect in Cd36 in the SHR progenitor strain has been previously described in detail (21). The SHR.BN-Il6/Npy (abbreviated as SHR-4) congenic strain was derived by backcrossing an inbred Brown Norway strain (Charles River Laboratories) with the SHR progenitor strain for 8 generations (19,21). The SHR-4 congenic strain is 99% genetically identical to the SHR progenitor strain, which differs only with respect to the region of chromosome 4 that includes Cd36. Details regarding derivation and characterization of the SHR-4 congenic strain have been previously described (19,21). SHR-4 congenic rats of the N8F8 generation were used in the current study. The SHR/Ola-TgN(EF1aCd36)19Ipcv (abbreviated as SHR-TG19) transgenic rats were derived by microinjecting SHR progenitor zygotes with wild-type Cd36 cDNA under control of the elongation factor 1␣ promoter (20). Details regarding derivation of the transgenic strain and the expression pattern of the Cd36 transgene have been previously described (20). In the current studies, we used SHR-TG19 transgenic rats of the F7 generation that were homozygous for the Cd36 transgene as confirmed by progeny testing. For in vivo studies, a minimum of 8 animals were studied in each group and for in vitro studies, a minimum of 5 animals were studied in each group. All experiments were performed in agreement with the Animal Protection Law of the Czech Republic (311/1997) which corresponds fully to the European Community Council recommendations for the use of laboratory animals 86/609/ECC. The experiments were approved by the Ethics Committee of the Institute of Physiology, Czech Academy of Sciences, Prague.
Experimental Protocol-Control groups from each strain (without pioglitazone) were compared with treatment groups from each strain (with pioglitazone). The control groups was fed standard chow from weaning and then given a semi-synthetic diet containing 60% fructose (TD 00202, Harlan Teklad, Madison, Wisconsin) beginning at 8 weeks of age. The treatment groups from each strain were treated identically to the control groups but in addition were given pioglitazone mixed in the high fructose diet (300 mg/kg of diet). After 13 days of fructose feeding, oral glucose tolerance tests were performed using a glucose load of 300 mg/100 g body weight after 7 h of fasting. Blood was drawn from the tail without anesthesia before the glucose load (0 min time point) and at 30, 60, and 120 min thereafter. The area under the glucose curve (AUC) above the baseline at 0 min was calculated using the trapezoidal rule. After 15 days of fructose feeding, serum levels of glucose, insulin, cholesterol, triglycerides, and nonesterified fatty acids (NEFA) were measured in nonfasted animals. Blood samples were clotted at 4°C, centrifuged, and the sera were kept frozen until analysis. At the end of the study, body weights were determined, the rats sacrificed by decapitation, and blood and tissues were collected for experimental measurements. Additional groups of SHR progenitor and SHR-4 congenic rats were used to investigate the effects of Cd36 genotype on pioglitazone-induced changes in arterial blood pressure and heart rate.
Insulin-stimulated Glucose Oxidation, Glycogen Synthesis, and Lipogenesis-Insulin-stimulated glucose oxidation was determined in isolated soleus muscle by measuring the effects of insulin on incorporation of [U-14 C]glucose into CO 2 according to Vrana et al. (23). After decapitation, the soleus muscles were attached to a stainless steel frame in situ or at in vivo length by special clips and separated from other muscles and tendons and immediately incubated for 2 h in Krebs-Ringer bicarbonate buffer, pH 7.4 that contained 5.5 mM unlabeled glucose, 0.5 Ci/ml of [U-14 Cglucose, and 3 mg/ml bovine serum albumin (Armour, Fraction V) with or without 250 microunits/ml insulin. After 2 h incubation, 0.3 ml of 1 M hyamine hydroxide was injected into central compartment of the incubation vessel and 0.5 ml of 1 M H 2 SO 4 into the main compartment to liberate CO 2 . The vessels were incubated for another 30 min, the hyamine hydroxide was then quantitatively transferred into the scintillation vial containing 10 ml of toluene-based scintillation fluid for counting of radioactivity. For measurement of insulin-stimulated incorporation of glucose into glycogen, soleus muscles were incubated for 2 h in 95% O 2 ϩ 5% CO 2 in Krebs-Ringer bicarbonate buffer, pH 7.4, containing 0.1 Ci/ml of [U-14 C]glucose, 5 mmol/L of unlabeled glucose, and 2.5 mg/ml of bovine serum albumin (Armour, Fraction V), with or without 250 units/ml insulin. Glycogen was extracted, and insulin-stimulated incorporation of glucose into glycogen determined as previously described (23). Measurement of insulin-stimulated lipogenesis from glucose was performed as previously described (24). Briefly, the distal segments of epididymal adipose tissue (200 -250 mg) were incubated in vials containing 3 ml of modified Krebs-Ringer bicarbonate buffer with 2.5 mg/ml of bovine serum albumin (Armour, Fraction V.), 0.1 Ci/ml [ 14 C]glucose, and 5.5 mmol/l unlabeled glucose at pH 7.4 with a gas phase of 95% O 2 ϩ 5% CO 2 . After incubation with and without 250 microunits of insulin for 120 min, pieces of adipose tissue were rinsed in saline and homogenized, and the total lipids were extracted for counting to determine the nmol of glucose converted into lipid per gram of adipose tissue as previously described (24).
Tissue Triglyceride Measurements-For determination of triglycerides in liver and soleus muscle, tissues were powdered under liquid N 2 and extracted for 16 h in chloroform:methanol, after which 2% KH 2 PO 4 was added, and the solution was centrifuged. The organic phase was removed and evaporated under N 2 . The resulting pellet was dissolved in isopropyl alcohol, and triglyceride content was determined by enzymatic assay (Lachema, Brno, Czech republic).
Hemodynamic Studies-Arterial blood pressures and heart rates were measured continuously in unanaesthetized, unrestrained male SHR progenitor rats and SHR-4 congenic rats from 80 days of age using radiotelemetry. All rats were allowed to recover for at least 10 days from surgical implantation of radiotelemetry transducers. Pulsatile pressures and heart rats were recorded in 5-s bursts every 5 min throughout the day and night for 10 weeks. 24-hour averages for mean arterial blood pressure and heart rate were calculated for each rat in each day of the study. The results from each rat in the same group were then averaged to obtain the group means. Rats were fed 60% fructose powder chow (TD 00202, Harlan Teklad, Madison, Wisconsin) and tap water ad libitum throughout the study. Pioglitazone was added in the high fructose diet (300 mg/kg of diet) from 94 to 122 days and 136 to 150 days of age.
Biochemical and Molecular Analyses-Blood glucose levels were measured by the glucose oxidase assay (Lachema, Brno, Czech Republic) using tail vein blood drawn into 5% trichloroacetic acid and promptly centrifuged. Serum NEFA levels were determined using an acyl-CoA oxidase-based colorimetric kit ((Roche Diagnostics GmbH, Mannheim, Germany). Serum triglycerides and total cholesterol concentrations were measured by standard enzymatic methods (Lachema, Brno, Czech Republic). Serum insulin concentrations were determined using a rat insulin radioimmunoassay kit (Amersham Biosciences). Cd36 gene expression was determined by Northern blot analysis of epididymal fat tissue as previously described (20,21) and further quantified by real-time PCR using the cyclophilin (peptidylprolyl isomerase A) gene as an internal control and primers specific for wild-type Cd36. The upstream primers were CAAATGCTGGACCCAACACA (cyclophilin A) and TCAAGGTGTGCTCAACAGCC (Cd36); the downstream primers were TGCCATCCAACCACTCAGTC (cyclophilin A) and AG-GATAAAACACACCAACTGT (Cd36). The RNA was reverse transcribed using random primers and the cDNA was amplified by real-time PCR using QuantiTect SYBR Green reagents (Qiagen) on an Opticon continuous fluorescence detector (MJ Research) running 35 cycles of the following protocol: 1 min denaturation at 95°C, 30 s annealing at 51.2°C for Cd36 or 58.9°C for cyclophilin, followed by 30 s extension at 72°C. Post-PCR melting curves confirmed the specificity of singletarget amplification and -fold expression of Cd36 relative to cyclophilin was determined in triplicate using the preferred method of Muller et al. (25).
Statistical Analysis-All data are expressed as means Ϯ S.E. Twoway analysis of variance was used to test for effects of strain, pioglitazone treatment, and strain ϫ treatment interactions on the dependent variables of interest. Differences between control versus treatment groups within a strain were evaluated by Student's t test, and differences among strains within groups were evaluated by one-way analysis of variance with Dunnett's method to adjust for multiple comparisons using the SHR strain as the control. Statistical significance was defined as p Ͻ 0.05.

RESULTS
Before starting the high fructose diet, mean body weights of the three strains were similar with the SHR-4 congenic strain (193 Ϯ 3 g) weighing slightly less (p Ͻ .05) than the SHR progenitor strain (209 Ϯ 3 g) and SHR-TG19 transgenic strain (214 Ϯ 3 g) (Fig. 1A). Within each strain, rats treated with pioglitazone gained the same amount of weight as untreated rats (Fig. 1B), so no excess weight gain was observed during the pioglitazone treatment. The SHR-4 congenic rats gained ϳ10 grams more than the SHR progenitor rats and the SHR-TG10 transgenic rats (both p Ͻ .05), perhaps because of their slightly lower body weights at the beginning of the study (Fig. 1B).
The effects of pioglitazone on serum insulin levels are shown in Fig. 3. Analysis of variance showed a significant effect of strain (p Ͻ .002) and treatment (p Ͻ .001) on serum insulin levels. Pioglitazone treatment was associated with significantly lower serum insulin levels in the SHR congenic and transgenic strains but not in the SHR progenitor strain harboring mutant Cd36. Serum insulin levels in pioglitazonetreated SHR congenic rats, 0.48 Ϯ 0.03 nmol/liter, and in pioglitazone-treated SHR-TG19 transgenic rats, 0.58 Ϯ 0.05 nmol/liter, were significantly lower than in pioglitazonetreated SHR, 0.75 Ϯ 0.05 nmol/liter (both p Ͻ .05). There was no significant effect of pioglitazone on serum glucose levels in the three strains (SHR, 5.7 Ϯ 0.2 mmol/liter with pioglitazone versus 5.3 Ϯ 0.1 mmol/liter without pioglitazone; SHR-TG19 transgenic, 5.7 Ϯ 0.1 mmol/liter with pioglitazone versus 5.6 Ϯ 0.1 mmol/liter without pioglitazone; SHR-4 congenic, 6.2 Ϯ 0.2 mmol/liter with pioglitazone versus 6.1 Ϯ 0.2 mmol/liter without pioglitazone). In the absence of pioglitazone, oral glucose tolerance was not significantly different among the three strains by analysis of variance, although there was a tendency for better glucose tolerance in the SHR-TG19 strain compared with the SHR-4 congenic strain and SHR progenitor strain (AUC ϭ 155 Ϯ 23 mmol⅐liter Ϫ1 ⅐2 h Ϫ1 , 189 Ϯ 18 mmol⅐liter Ϫ1 ⅐2 h Ϫ1 , and 181 Ϯ 2 mmol⅐liter Ϫ1 ⅐2 h Ϫ1 , respectively) (Fig. 4). With pioglitazone, oral glucose tolerance was significantly better in both the SHR-TG19 transgenic strain (AUC ϭ 163 Ϯ 10 mmol⅐liter Ϫ1 ⅐2 h Ϫ1 ) and the SHR-4 congenic strain (AUC ϭ 154 Ϯ 7 mmol⅐liter Ϫ1 ⅐2 h Ϫ1 ), compared with the SHR progenitor strain (AUC ϭ 193 Ϯ 16 mmol⅐liter Ϫ1 ⅐2 h Ϫ1 , both p Ͻ .05) (Fig. 4). Glucose tolerance in the SHR-4 congenic rats fed pioglitazone was also significantly better than in the SHR-4 congenic rats not given pioglitazone (p Ͻ .05) (Fig. 4). Glucose tolerance was similar in SHR with or without pioglitazone and in SHR-TG19 rats with and without pioglitazone. In the SHR-TG19 strain, the lack of effect of pioglitazone on the oral glucose tolerance AUC may have been related to its modest level of Cd36 expression and the more modest effects of pioglitazone on NEFA levels in this strain.
To investigate mechanisms that might mediate the influence of Cd36 on the insulin-sensitizing actions of pioglitazone, we studied pioglitazone effects on insulin-stimulated glucose oxi-  2. A, serum levels of nonesterified fatty acids in rats treated with pioglitazone compared with rats not given pioglitazone. B, serum triglyceride levels in rats treated with pioglitazone compared with rats not given pioglitazone. *, significant difference compared with the other two strains. #, significant difference between rats treated with pioglitazone versus rats not treated with pioglitazone within the same strain. Other symbols as in Fig. 1. dation in soleus muscle and on the incorporation of glucose into diaphragmatic muscle glycogen. In the absence of insulin, pioglitazone treatment had no effect on glucose oxidation in soleus muscles from any of the strains (data not shown). However, in the presence of insulin, pioglitazone treatment induced increases in glucose oxidation in both the SHR transgenic and congenic strains, although statistical significance was achieved only in the TG19 strain (p Ͻ .05) (Fig. 5). No effect on insulinstimulated glucose oxidation was observed in the SHR progenitor strain with mutant Cd36 (Fig. 5). As a consequence, insulin-stimulated glucose oxidation in SHR treated with pioglitazone, 141 Ϯ 21 nmol glucose⅐g tissue Ϫ1 ⅐2h Ϫ1 , was significantly lower (p Ͻ .05) than in the pioglitazone-treated SHR-4 congenic strain, 276 Ϯ 51 glucose⅐g tissue Ϫ1 ⅐2h Ϫ1 , and in the pioglitazone-treated SHR transgenic strain, 316 Ϯ 44 glucose⅐g tissue Ϫ1 ⅐2h Ϫ1 (Fig. 4). No differential effects of pioglitazone treatment were observed on incorporation of glucose into diaphragm muscle glycogen or into adipose tissue lipids between any of the strains (data not shown).
Epididymal fat pad weights, which were similar among the three strains, were significantly increased by pioglitazone treatment (Fig. 6). The magnitude of the increase appeared greatest in the SHR progenitors, where epididymal fat pad weight at the end of the treatment, 1.4 Ϯ 0.4 g/100 g body weight, was significantly greater (p Ͻ .05) than in both the SHR-4 congenic strain, 1.1 Ϯ 0.3 g/100 g body weight, and the SHR-TG19 strain, 1.29 Ϯ 0.3 g/100 g body weight (Fig. 6). In the absence of pioglitazone, hepatic triglyceride levels (Fig. 7) were similar in the SHR progenitors with mutant CD36 and in the SHR-TG-19 strain, which expresses CD36 in the liver, 15.6 Ϯ 1.5 mol/g and 14.4 Ϯ 0.6 mol/g, respectively (20). In both strains, hepatic triglycerides were significantly higher than in the SHR-4 congenic strain, 8.1 Ϯ 0.9 mol/g, p Ͻ .05. Administration of pioglitazone appeared to reduce hepatic triglyceride levels in all three strains, however, this effect was greatest in the transgenic rats (Fig. 7). Thus, in rats fed pioglitazone, hepatic triglyceride levels were significantly reduced in both the SHR-4 congenic and SHR-TG19 transgenic strains as compared with the SHR progenitor strain (6.1 Ϯ 0.3 and 8.1 Ϯ 0.3 mol/g versus 12.2 Ϯ 0.6 mol/g respectively, p Ͻ 05). No strain or treatment effects were detected on triglyceride levels in isolated soleus muscle.
Northern blot analysis of fat tissue showed abundant expression of wild-type Cd36 message in the SHR-4 congenic strain and low expression of the Cd36 transgene in the SHR-TG19 strain (Fig. 8). Endogenous mutant Cd36 transcripts were also present in the SHR-TG19 strain and the SHR progenitor (Fig.  8). Administration of pioglitazone was associated with increased expression of wild-type Cd36 transcript in the SHR-4 congenic strain and of the mutant Cd36 transcripts in the SHR progenitor and SHR-TG19 strains consistent with previous reports of a functional binding site for PPAR␥-RXR het-  erodimers in the Cd36 promoter (17). Real-time PCR studies were used to estimate the effect of pioglitazone on expression of wild-type Cd36 using primers designed to specifically amplify the wild-type gene and not the mutant variant. In the SHR-4 congenic strain, administration of pioglitazone was associated with a significant increase in expression of wild-type Cd36 compared with SHR-4 congenic rats not given pioglitazone (Fig.  9). Real-time PCR confirmed low level expression of the wildtype Cd36 transgene in the SHR-TG19 strain (Fig. 9). However, in contrast to the SHR-4 congenic strain, in which pioglitazone increased the expression of wild-type Cd36, no effect of pioglitazone was detected on expression of the Cd36 transgene in the SHR-TG19 strain (Fig. 9). As expected, using either Northern blot or real-time PCR analysis, expression of wildtype Cd36 could not be detected in the SHR progenitor strain that harbors only mutant Cd36 (Figs. 8 and 9).
Consistent with the results of previous studies (19), mean arterial pressures were significantly lower in the SHR-4 congenic strain than in the SHR progenitor strain and administration of pioglitazone caused significant decreases in mean arterial pressure of ϳ10 mmHg in both the SHR progenitor and the SHR-4 congenic strains (data not shown). The pioglitazoneinduced changes in blood pressure were associated with reciprocal changes in heart rate that were similar between the two strains (data not shown).

DISCUSSION
Multiple gene targets and metabolic pathways are thought to be involved in the ability of PPAR␥ ligands to enhance carbohydrate and lipid metabolism (3,4). Large scale gene expression profiling studies have documented effects on expression of numerous genes in multiple insulin-sensitive issues (3). However, the relative roles of specific target genes in the insulinsensitizing effect of a PPAR␥ ligand remain ill defined because such information is difficult to discern from studies that simply correlate changes in gene expression with changes in metabolic phenotypes. A number of studies have proposed that the insulin-sensitizing effects of PPAR␥ ligands may be mediated in part by promoting uptake and oxidation of fatty acids by adipose tissue and by shunting lipids away from liver and skeletal muscle (2,3,(5)(6)(7)(8)(9)(10). However, other findings suggest that improving fatty acid utilization by muscle tissues may also play an important role. Burant and colleagues have reported that the PPAR␥ ligand troglitazone can still improve glucose tolerance in aP2/DTA lipodystrophic mice that lack subcutaneous or intra-abdominal fat (26). Hevener et al. (6) showed that the thiazolidinedione troglitazone could attenuate the decreases in insulin sensitivity and in skeletal muscle expression of CD36 that are induced by systemic infusion of free fatty acids (6). CD36 is a membrane protein that facilitates a major fraction of fatty acid uptake/utilization by fat and muscle tissues (16,27). The Cd36 gene is one of the many whose expression can be strongly influenced by PPAR␥ ligands (3,4).
In the current study, we directly tested the role of Cd36 in the insulin-sensitizing actions of the thiazolidinedione pioglitazone. We examined the ability of pioglitazone to improve blood lipid levels and insulin sensitivity of glucose oxidation in the SHR strain, which expresses mutant CD36 (19 -21) and exhibits loss of CD36 function in fatty acid uptake (22). Con-   genic and transgenic SHR strains that we derived from the SHR progenitor and which differed by expression of wild-type CD36 and by restoration of optimal fatty acid uptake (19 -22) were used to test the specificity of the effects observed in the SHR. The availability of these strains allows for a direct pharmacogenetic test of the role of Cd36 in the insulin-sensitizing effects of thiazolidinediones.
We used the SHR strain, which carries a functionally defective CD36 and an SHR-4 congenic strain, which is 99% genetically identical to the SHR progenitor strain and in which the expression of wild-type, functional Cd36 is under control of its native promoter (19). A transgenic line, SHR-TG19, which is genetically identical to the SHR progenitor strain except for Cd36 was also used. In the transgenic line, the wild-type, functional Cd36 is under control of the elongation factor 1␣ promoter. It is expressed at modest levels in tissues that ordinarily express relatively large amounts of Cd36 (e.g. fat, heart, skeletal muscle) (20) and exhibits significant expression in liver and kidney (20) where levels are ordinarily low. Phenotypic differences between the SHR progenitor and the SHR transgenic strains with respect to the effect of pioglitazone can be solely attributed to Cd36 expression. A comparison of findings in the transgenic to those in the congenic strain would control for the possibility that some of the phenotypic differences observed may be related to the slightly atypical Cd36 expression pattern in the SHR-TG19. Taken together, the results in the transgenic and congenic strains versus the SHR progenitor strain provide a compelling pharmacogenetic test of the role of Cd36 in the metabolic actions of pioglitazone.
A considerable body of evidence suggests that the insulinsensitizing actions of thiazolidinediones may be secondary to their effects on fatty acid levels and fatty acid metabolism (2, 3, 5-10). In addition, chronic elevations in fatty acid levels are a well known cause of insulin resistance (11)(12)(13)(14). Our findings confirm the important role of changes in fatty acid utilization in the insulin-sensitizing effect of thiazolidinediones. More significantly, they directly implicate Cd36 as a key determinant of these metabolic changes. In the current study, the ability of pioglitazone to reduce circulating levels of fatty acids was significantly impaired in the SHR as compared with the congenic and transgenic lines. These observations indicate that administration of pioglitazone decreases fatty acid levels in a Cd36dependent fashion. The decrease in circulating fatty acids then results in an improvement of insulin sensitivity because failure to achieve this decrease in the SHR significantly diminishes the ability of pioglitazone to induce reductions in circulating insulin and to enhance insulin stimulated glucose oxidation in soleus muscle. Rescue of Cd36 function in the congenic and transgenic strains restores the effects of pioglitazone on both fatty acid levels and insulin sensitivity. Thus, at least under the dietary and genetic conditions of the current study, Cd36 is an important determinant of key metabolic actions of pioglitazone. Despite the current findings implicating Cd36, however, it should be noted that in the SHR harboring mutant Cd36, a tendency for a modest effect of pioglitazone on some metabolic parameters including serum NEFA and hepatic triglyceride levels was also observed. Thus, although the current studies establish a role for Cd36 in the metabolic actions of pioglitazone, they should not be interpreted as excluding a role for other targets in the metabolic effects of thiazolidinediones.
Pioglitazone had significant hypolipidemic and insulin-sensitizing effects in both the SHR transgenic line and SHR congenic line although oral glucose tolerance testing was significantly improved only in the SHR congenic line, perhaps because of the much greater level of expression of wild-type Cd36 in this strain. In addition, real-time PCR experiments showed no response of the Cd36 transgene to pioglitazone, whereas expression of endogenous wild-type Cd36 in the SHR congenic line was substantially up-regulated by pioglitazone. Thus, whereas up-regulation of Cd36 expression may contribute to the metabolic effects of pioglitazone in the SHR congenic line, up-regulation of Cd36 would not appear to account for the metabolic effects of pioglitazone in the SHR transgenic strain. It is possible that the antidiabetic effects of pioglitazone in the SHR transgenic strain and SHR congenic strain are partly mediated by the formation of new fat cells expressing functional Cd36 as opposed to defective Cd36. It is also possible that the increased adipose tissue mass in pioglitazone-treated animals results in increased release of adipocyte mediators that can regulate fatty acid utilization in other tissues, notably muscle, provided those tissues have functional Cd36. Examples of such mediators are leptin, resistin, and adiponectin (28,29). This interpretation could also explain the interaction of pioglitazone treatment and Cd36 on the increase in fat pad weight observed in treated animals. The increase in fat pad weight was significantly greater in the SHR progenitors, which would reflect the ineffectiveness of the adipocyte mediators to enhance fatty acid utilization and clearance by peripheral tissues of these animals because they carry defective Cd36.
The current findings provide direct pharmacogenetic evidence that Cd36 is an important determinant of the metabolic effects of pioglitazone. However, they do not allow conclusions to be made with respect to the relative importance of Cd36 in specific tissues. Adipose tissue is generally thought to play an important role in mediating the metabolic effects of thiazolidinediones, but as shown by Burant and colleagues (26) and Chao et al. (10), it is not required for all of these effects. Cd36 facilitates a major fraction (Ͼ60%) of fatty acid uptake by adipose tissue and muscle (16,27,30), and it is possible that CD36 expression in both tissues plays an important role in mediating the antidiabetic effects of thiazolidinediones. The role of Cd36 in liver may also be worth exploring, given the potential effects of hepatic triglyceride levels on glucose production. Future studies could dissect the relative contribution of specific tissues by comparing the insulin-sensitizing effects of thiazolidinediones in strains where Cd36 expression is rescued in a tissue-specific fashion. This can be accomplished not only by deriving new transgenic strains of SHR that express wild-type Cd36 in fat, muscle, or liver, but also by capitalizing on tissue-transplantation experiments between histocompatible congenic, transgenic, and progenitor strains of SHR.