CD36-Facilitated Fatty Acid Uptake Inhibits Leptin Production and Signaling in Adipose Tissue

CD36-null (CD36^−/−) and wild-type mice on the C57BL6 background (17,18) were housed in a facility with a 12-h light cycle and fed ad libitum chow (5001; Purina, St. Louis, MO) or high-fat (F3282; Bio-Serv, Frenchtown, NJ) diets. Mice (5 weeks) were fed the high-fat diet for 15 weeks to induce obesity with body weight measured weekly.

Studies were performed (n = 6 per group) in metabolic chambers (Metabowl; Jencons Scientific), as described previously (19). Energy intake was obtained by measuring food consumption multiplied by the diet caloric content (2.98 kcal/g). Metabolic rate was determined by indirect calorimetry with an Oxymat/Ultramat 6 analyser (Siemens) that monitors O₂ and CO₂ concentrations using data acquisition hardware (Analogic) and software (Labtech). The respiratory quotient equals Vco₂/Vo_2. To measure activity, mice were housed for 24 h individually at 18°C and a 12-h light cycle. Cages were placed into frames with infrared photocells interfaced with a computer (San Diego Instruments). Body composition was by dual energy X-ray analysis (GE Lunar, Madison, WI).

Fasted (16 h) CD36^−/− and wild-type mice were given via an intragastric tube 500 ml saline or saline with glucose (7.46 mg/g) or olive oil emulsion (3.5 mg/g) (20). Blood and organs were collected 2.5 h later. Plasma lipids and glucose were tested enzymatically (16), and insulin was tested by RIA (Linco, St. Louis, MO).

Epididymal fat was formalin-fixed, ethanol-dehydrated, transitioned into xylene, and paraffin-embedded. Sections (3 mm) were stained with hematoxylin-eosin. Sections were viewed at ×10 magnification with a Nikon Eclipse TE2000-U microscope in epiflourescence using an EXFO-Xcite-TM120 PC Illumination System and a Photometrics Cool SNAPcf monochrome camera (Roper Scientific, Tuscon, AZ). Morphometery was by MetaMorph v6.2r6 (Universal Imaging Corp) with 0.45 μm/pixel. Mean cell size was calculated using Excel (Microsoft, Redmond, WA).

Epididymal fat from six to eight wild-type and CD36^−/− mice was minced, and 500-mg samples were placed in 2 ml serum-free M199 under 95/5 O₂/CO₂ with daily medium replenishment (21). Leptin production was the sum of leptin in medium and tissue. Tissue (500 mg) was homogenized (Teflon pestle, clearance B) in 2 ml cold buffer (10 mmol/l Tris, 1 mmol/l EDTA, and 250 mmol/l sucrose, pH 7.4) with 3.5 μg/ml aprotinin, 2.5 μg/ml leupeptin, and 1 mmol/l phenylmethylsulfonyl fluoride. After centrifugation (14,000 rpm, 15 min, 4°C) infranatants and media were assayed for leptin by ELISA (Linco). Adiponectin was assayed as described previously (22) or by ELISA (Linco).

Fatty acid incorporation into lipids was as previously described (17). Fatty acid oxidation was measured from the CO₂ produced from [¹⁴C]oleic acid (23) using isolated adipocytes washed four times in buffer with 1% albumin.

Tissues were homogenized (50 mmol/l Tris, pH 7.5, 150 mmol/l NaCl, 1% Triton X-100, 1% sodium deoxycholic acid, 1 mmol/l phenylmethylsulfonyl fluoride, 50 mmol/l sodium fluoride, 1 mmol/l sodium orthovanadate, 50 μg/ml aprotinin, and 50 μg/ml leupeptin). After centrifugation (15 min at 4°C), proteins in supernatants separated by SDS-PAGE were transferred to nylon-enhanced nitrocellulose membranes that were blocked in 0.1% Tween/PBS with 5% fat-free milk (1 h) before incubation with antibodies for peroxisome proliferator–activated receptor γ (PPARγ) (Upstate, Lake Placid, NY) (1:1,000, 2 h, room temperature), signal transducer and activator of transcription 3 (STAT3), or p-STAT3 (Cell Signaling, Danvers, MA) (1:1,000, overnight, 4°C). After three washings with 0.1% Tween/PBS, the membranes were re-blocked and incubated with peroxidase-labeled secondary antibody (1 h, 1:10,000). Immunodetection was by ECL (Amersham, Arlington, IL), and band intensity was determined by densitometry scanning (Umax) with analysis by ImageJ (National Institutes of Health, Bethesda, MD).

Isolated RNA (Trizol; Invitrogen) was washed in 75% ethanol, dried, and resuspended in UltraPure Water (Gibco). Amplification was using Superscript III Platinum SYBR Green (Invitrogen) on a SmartCycler (Cephied, Sunnyvale, CA). Cyclophilin, bActin, or 18S was used to normalize expression of tested genes. The threshold crossing of each sample after normalization (ΔCt) was used to calculate levels of each mRNA relative to controls using 2^−ΔCt. Primers used are listed in the online appendix (available at http://dx.doi.org/10.2337/db06–1699).

Data are shown as means ± SE and tested for statistically significant differences using t tests and P < 0.05. Differences in growth curves between groups were compared by ANOVA followed by the Tukey-Kramer test when appropriate.

The CD36^−/− mouse exhibits a major defect in fatty acid uptake by adipocytes (17,18), providing a unique model for examining the physiological role of fatty acid uptake on leptin secretion. If adipocyte fatty acid uptake regulates leptin, one might predict that the CD36^−/− mouse, with high fatty acid levels but impaired fatty acid uptake and metabolism, would lack this regulatory effect. We compared blood leptin levels in CD36-null and wild-type mice. As shown in Fig. 1, plasma leptin was twofold higher in CD36^−/− mice compared with age- and sex-matched wild-type mice (Fig. 1A). Leptin levels in females were 40% higher than in males, but similar effects of genotype on leptin were observed regardless of sex (data not shown), therefore, only male mice were used for the rest of the studies. The hyperleptenemia in CD36^−/− mice was associated with increased leptin mRNA in adipose tissue (Fig. 1B), suggesting that it was due to enhanced transcription of the ob gene. To determine specificity of the effect of CD36 deficiency on leptin, we measured levels of the adiponectin (Fig. 1C and D). Blood levels of adiponectin and its adipose tissue expression were significantly lower in CD36^−/− compared with wild-type mice. Thus CD36 deficiency affected adiponectin in the opposite direction to leptin.

The high leptin levels in CD36^−/− mice did not reflect increased adiposity. Epididymal fat pads of CD36^−/− mice (Fig. 1E) weighed less than those of wild-type mice, and mean adipose cell size was similar for tissues from both genotypes (Fig. 1F).

To determine whether the increased leptin levels in CD36^−/− mice were effective in limiting energy intake and fat deposition, food intake, growth curves, fat versus lean body mass, bone density, metabolic rate, and respiratory quotient were examined. CD36^−/− mice gained less body weight than wild-type mice (Fig. 2A) with differences becoming significant (P < 0.01) for the period of 12–16 weeks. Lean body mass was similar, but total body fat was 38% lower in CD36^−/− mice (Fig. 2B), in line with the lower fat pad weight. Bone mass (Fig. 2B, right) was significantly decreased (12%) in CD36^−/− mice, which may be consistent with reports documenting leptin inhibition of bone formation (24). Food intake (Fig. 2C), determined during the 24-h calorimetry, was 20% lower for CD36^−/− mice compared with age-matched controls. CD36^−/− mice had a similar metabolic rate to that of wild-type mice when the metabolic rate is related to unit body weight or lean body mass (Fig. 2D). Without this adjustment, metabolic rate of CD36^−/− mice was 15% lower than that of wild-type mice (P < 0.05). The respiratory quotient was similar for both genotypes (Fig. 2E).

The lack of change in the respiratory quotient (Fig. 2E) suggested that fuel oxidation was similar for both genotypes. CD36^−/− mice exhibit decreased fatty acid utilization in peripheral tissues (16,17). In contrast, hepatic fatty acid uptake is slightly increased (17,25). We examined circulating ketone bodies in wild-type and CD36^−/− mice as an index of hepatic fatty acid oxidation. Plasma β-hydroxybutyrate levels (mg/dl) were increased (12.3 ± 1.0 in CD36^−/− versus 7.8 ± 1.1 in wild-type, P < 0.05, n = 9), suggesting enhanced hepatic fatty acid oxidation. The reductions in energy intake and fat mass and the increase in blood ketone bodies in CD36^−/− mice were consistent with known actions of leptin to reduce energy intake (3) and to increase hepatic fatty acid oxidation (26).

We also compared the effects of fat feeding on body weight and leptin in wild-type versus CD36^−/− mice. As shown (Fig. 3A), CD36^−/− mice gained significantly less weight than wild-type mice. Fat pad mass at the end of the diet period was 80% larger in wild-type compared with CD36^−/− mice (Fig. 3B). Leptin levels in CD36^−/− mice were twice those of wild-type mice, and ob gene expression in CD36^−/− adipose tissue was three times higher (Fig. 3C and D, respectively). These data indicated that CD36^−/− mice were protected from diet-induced obesity. To determine whether CD36 deletion would protect from obesity consequent to leptin deficiency, we generated a mouse double null for CD36 and leptin by crossing CD36^−/− and ob/ob mice. The CD36^−/−-ob/ob mouse (Fig. 3E) was obese, indicating that blocking fatty acid uptake does not reverse obesity of the ob/ob mouse and supporting the contribution of leptin to the lean phenotype of CD36 deficiency.

Levels of insulin and glucose, which chronically induce leptin (10), were similar in fed wild-type and CD36^−/− mice (data not shown). Fasted CD36^−/− mice were hypoinsulinemic and hypoglycemic (data not shown) compared with wild-type mice, consistent with previous findings (16,17). However, leptin levels of fasted CD36^−/− mice were still double those of fasted wild-type mice (data not shown).

The acute response of leptin secretion to nutrient ingestion was examined next by monitoring leptin after intragastric and equicaloric doses of glucose or olive oil. As expected (Fig. 4), the glucose load was cleared faster in CD36^−/− mice (Fig. 4A), whereas no significant effects were observed on blood fatty acid levels (Fig. 4A, inset), which were significantly higher in the CD36^−/− group. Glucose increased plasma leptin in wild-type mice by 21% at 2 h, (from 3.9 to 4.7 ng/ml, P < 0.05). At 4 h, leptin levels did not differ from basal or pre-glucose levels. The response of leptin to intragastric glucose was strongly enhanced in CD36^−/− mice. Levels rose from 5.7 to 8.5 ng/ml at 2 h and up to 9.5 at 4 h after gavage (by 49 and 67%, respectively, P < 0.01) (Fig. 4B). Gavage with olive oil (Fig. 4C and D) produced little effect on blood leptin in wild-type or CD36^−/− mice. Glucose levels were unaltered in both groups, but blood fatty acid increased in CD36^−/− mice (Fig. 4C), consistent with delayed fatty acid clearance. Together, the above data suggest that insulin and glucose are potent stimulators of leptin production, as previously reported (10). However, the stimulatory effect is antagonized by CD36-facilitated fatty acid uptake in vivo. CD36 deletion allows uninhibited rise in leptin levels in response to glucose and insulin.

CD36 recognizes a variety of ligands in addition to fatty acid (rev. in 27). To further establish the link between CD36-facilitated fatty acid uptake and leptin inhibition, we examined adipose tissue from wild-type and CD36^−/− mice ex vivo in the absence of CD36 ligands other than fatty acid. CD36^−/−, compared with wild-type adipose tissue, exhibited a 60% defect in fatty acid uptake and incorporation into triglycerides (data not shown) as previously reported (17,18). Leptin secretion was twofold higher in tissue from CD36^−/− compared with wild-type mice (Fig. 5A). Addition of either palmitic or oleic acids (0.075 mmol/l, 0.3 mmol/l BSA, unbound fatty acid 1.6 and 1.8 nmol/l for palmitate and oleate, respectively) significantly reduced basal leptin production by wild-type tissue but had no effect on CD36^−/− tissue (Fig. 5B). Addition of insulin (Fig. 5C) stimulated leptin production by about threefold in cultures from wild-type or CD36^−/− mice (compare y axes in Fig. 5B and C). Importantly, the insulin effect was inhibited in wild-type tissue when fatty acids were added (∼60%), whereas no inhibition occurred in CD36^−/− tissue. These data indicate that CD36–fatty acid uptake antagonizes the stimulatory effect of glucose and insulin on leptin production. Fatty acid inhibition is lost in CD36 deficiency where insulin plus glucose can fully activate leptin release. Thus fatty acid uptake by CD36-independent mechanisms, which accounts for ∼40% of uptake, is ineffective in suppressing leptin. The in vivo rise in blood leptin in CD36^−/− mice given glucose (Fig. 4B) would reflect uninhibited glucose/insulin stimulation as a result of impaired fatty acid sensing by adipocytes.

PPARs are activated by fatty acid or their derivatives and are essential to the nutrient-sensing function of adipose tissue, which includes leptin secretion. Thiazolidinediones, which activate PPARγ, inhibit leptin release (13,28,29), and mice heterozygotes for PPARγ deficiency exhibit high leptin (30). We examined whether the high leptin of CD36^−/− mice could reflect a defect in fatty acid activation of adipose tissue PPARγ. We tested levels and activity of adipose tissue PPARγ in mice after intragastric administration of saline or olive oil. Levels of PPARγ protein at 2.5 h after saline gavage were reduced in CD36^−/− adipose tissue (Fig. 6, top). The effect of intragastric oil administration on expression of the PPARγ targets INSIG-1 (31) and aP2 (32) is shown in Fig. 6, bottom. Increases were observed in wild-type tissue but not in CD36^−/− tissue. Expression of ap2 in the experiment shown in Fig. 6 trended lower without reaching statistical significance. However, basal ap2 expression measured in two experiments was significantly decreased by 45% (n = 5, P = 0.03). The data suggest impaired basal levels, basal activity, and fatty acid responsiveness of PPARγ in CD36^−/− tissue. This helps explain the reduced adipose tissue expression and blood levels of adiponectin (Fig. 1) in CD36^−/− mice because adiponectin is induced by PPARγ (33).

The energy balance data suggested leptin sensitivity was intact in CD36^−/− mice despite the hyperleptinemia. To examine leptin sensitivity of adipose tissue, we compared expression of fatty acid–oxidative genes previously shown responsive to leptin (9) in tissues from wild-type and CD36^−/− mice (Fig. 7). Expression of carnitine palmitoyl transferase 1 (CPT1), uncoupling protein 2 (UCP2), and acyl-CoA oxidase (ACO) was significantly increased in CD36^−/− adipose tissue. Consistent with this, oleate oxidation in adipocytes isolated from CD36^−/− compared with wild-type mice was enhanced (0.02 vs. 0.01 nmol · 10,000 cells⁻¹ · h⁻¹, respectively; P < 0.01). We also examined leptin signaling. Leptin stimulates several intracellular signaling pathways. The pathway involving STAT3 is central to metabolic effects of leptin, and its dysfunction results in obesity and leptin resistance (34,35). STAT3 phosphorylation (p-STAT3), a sensitive indicator of leptin activity (4,34,35) was ∼10-fold higher in CD36^−/− compared with wild-type tissue, whereas total STAT3 content was similar (Fig. 8). The higher p-STAT3–to–STAT3 ratio reflected the ambient hyperleptinemia (2.2 vs. 4 ng/ml for wild type and CD36^−/−, respectively) and indicated that leptin sensitivity was intact. The fact that the 10-fold increase in p-STAT3 to STAT3 was out of proportion with the 1.8-fold increase in leptin levels suggested enhanced leptin signaling.

The present study examined the role of fatty acid uptake in leptin regulation using both in vivo and in vitro approaches. The following novel findings were obtained related to nutrient regulation of leptin production: First, uptake of dietary fatty acid via CD36 inhibits leptin production by adipocytes blunting stimulation by glucose and insulin. CD36 deletion allows uninhibited stimulation of leptin by glucose and insulin, which uncouples the normal association between adipose tissue mass and leptin level. Second, the effect of CD36-facilitated fatty acid uptake on leptin production appears linked to low PPARγ levels and activity in CD36^−/− adipose tissue. Because fatty acid uptake still occurs in CD36^−/− by CD36-independent mechanisms, the CD36–fatty acid pathway may modulate PPARγ levels and supply the fatty acid–derived ligand for PPARγ activation. Third, leptin sensitivity is intact and possibly enhanced in CD36 deficiency despite the hyperleptinemia. CD36^−/− adipocytes exhibited upregulated p-STAT3 and fatty acid oxidative enzymes responsive to leptin. These data indicate that CD36-facilitated fatty acid uptake modulates leptin production in vivo and may contribute to leptin resistance of adipose tissue.

Glucose and insulin stimulate leptin production in vitro (10,11). In vivo, intragastric glucose had modest and transient effects on leptin levels in wild-type mice versus strong and sustained upregulation in CD36^−/− mice. This indicates that glucose and insulin are potent stimulators of leptin secretion in vivo but their effect is blunted by fatty acid. As further confirmed in vitro, absence of fatty acid influx via CD36 is required for full stimulation of leptin production by insulin and glucose.

Fatty acid inhibition, like glucose stimulation of leptin production, likely contributes to the nutrient sensing function of adipose tissue. Fatty acid inhibition was specific to CD36-facilitated fatty acid uptake because the CD36-independent component (40%) was ineffective in suppressing leptin or inducing adiponectin. The effect of CD36-mediated fatty acid uptake on leptin may in part be exerted via modulating insulin signaling and/or glucose metabolism. It could be argued that CD36 deficiency enhances insulin sensitivity of adipocytes, which would stimulate leptin production. Although such an effect could contribute to chronic leptin regulation, it would not explain why adiponectin expression is decreased because this usually correlates with insulin resistance. Previously, disruption of the adipocyte insulin-responsive glucose transporter (36) did not alter circulating leptin, but the response of leptin to nutrient intake and in particular to fatty acid was not tested. As a result, additional work is needed to fully understand how the metabolism of glucose and fatty acid might interact in regulating leptin production. Although we believe that the data in this study, when taken together, strongly suggest an important role for the transcription factor PPARγ in mediating inhibitory effects of CD36-facilitated fatty acid uptake on leptin levels, they do not rule out contribution of additional mechanisms such as changes in insulin signaling or glucose metabolism. PPARγ is a major fatty acid sensor that, when activated, inhibits release of leptin (14,29,37) and induces that of adiponectin (33). CD36 deletion resulted in lower PPARγ protein levels with reduced basal expression of PPARγ targets: ap2 by 45%, adiponectin by 60% (n = 5, P = 0.008), and lipoprotein lipase by 50% (n = 4, P < 0.04). In addition, there was blunted PPARγ activation by fat administration in CD36-null mice (Fig. 6). Possibly, activity of the fatty acid–CD36-PPARγ pathway with subsequent suppression of leptin and induction of adiponectin would promote clearance of nutrients from the circulation. Less leptin favors fatty acid storage in adipose tissue, whereas more adiponectin promotes their oxidation in muscle and liver (33).

Leptin sensitivity was intact in CD36^−/− mice despite the hyperleptinemia as shown by the reduced food intake, the decreased bone density, the enhanced hepatic fatty acid oxidation, and the induction of leptin-responsive fatty acid–catabolizing enzymes in adipose tissue. In addition at the level of adipose tissue, the increase in leptin signaling estimated from the p-STAT–to–STAT ratio was disproportional to the increased in vivo leptin levels, also supporting enhanced sensitivity. The data suggest that the fatty acid–CD36 pathway may be relevant to the development of leptin resistance of adipose tissue.

The effects of leptin on peripheral tissues are less well characterized than its central effects, and this is especially true of adipose tissue of which little information is available. Leptin does not appear to regulate glucose uptake by adipocytes (38), but there is evidence that it reduces fatty acid uptake by these cells (39) or that it antagonizes the effect of insulin to stimulate it (38). Interestingly, it has been reported that adipose tissue of ob/ob mice has severalfold higher CD36 expression that is normalized by leptin administration (39). The tissue also exhibits enhanced rates of fatty acid uptake that drop early after leptin injection, preceding reductions in food intake (39).

In contrast to its actions to decrease adipocyte fatty acid uptake and CD36, leptin upregulates both in muscle cells (40). This suggests that regulation of fatty acid uptake may mediate some of the metabolic effects of leptin to inhibit fatty acid storage and increase fatty acid catabolism.

It is well established that central resistance to leptin signaling plays a major role in the etiology of obesity (3,41,42) but resistance at the level of muscle and adipose tissue may also be important (4,9,43). Adipocyte-specific downregulation of the leptin receptor in mice results in obesity in the absence of effects on food intake (43). Our findings suggest that suppressing adipocyte CD36 could improve leptin regulation in obesity where a selective increase in fatty acid uptake by adipocytes in rodents and humans (44) has been reported.

CD36 deficiency is relatively high (1–3%) in some subpopulations such as Asians and Africans (45). Humans deficient in CD36 exhibit abnormalities of fatty acid utilization that are similar to those identified in rodents and it was proposed that the deficiency may be related to the phenotypic expression of the metabolic syndrome (46). Common polymorphisms in the CD36 gene in Caucasians have been linked to lipid abnormalities (47) and to diabetes incidence (48). None of the studies addressed leptin levels or sensitivity. However, there is indirect evidence to suggest that what we describe in the CD36-null mouse may be relevant to the human case. CD36 levels are upregulated in adipose tissue of obese subjects (49). Also a common polymorphism in the upstream promoter of the CD36 gene was associated with a higher incidence of diabetes type 2 that was most pronounced in subjects with a high BMI (48), suggesting unfavorable interaction of the polymorphism with BMI as a modifier of diabetes risk. Finally, the CD36-PPARγ-adiponectin link that we propose may help explain the correlation previously noted between low adiponectin levels and a CD36 promoter variant in a French population (50).

N.A.A. has received National Institute of Health Grants DK-33301 and DK-60022. This work has been supported by Washington University Clinical Nutrition Research Center Grant P30-DK-056341.

We are grateful to Dr. Philip Scherer for initial measurements of plasma adiponectin.