|Home | About | Journals | Submit | Contact Us | Français|
Deficiency of the membrane protein FAT/CD36 causes a marked defect in fatty acid uptake by various tissues and is genetically linked to insulin resistance in rats and humans. Here, we examined insulin responsiveness of CD36–/– mice. When fed a diet high in complex carbohydrates and low (5%) in fat, these animals cleared glucose faster than the wild-type. In vivo, uptake of 2-fluorodeoxyglucose by muscle was increased severalfold, and in vitro, insulin responsiveness of glycogenesis by the soleus was enhanced. Null mice had lower glycogen levels in muscle and liver, lower muscle triglyceride levels, and increased liver triglyceride content—all findings consistent with increased insulin-sensitivity. However, when the chow diet was switched to one high in fructose, CD36–/– mice but not wild-type mice developed marked glucose intolerance, hyperinsulinemia, and decreased muscle glucose uptake. High-fat diets impaired glucose tolerance equally in both groups, although CD36 deficiency helped moderate insulin-responsive muscle glucose oxidation. In conclusion, CD36 deficiency enhances insulin responsiveness on a high-starch, low-fat diet. It predisposes to insulin resistance induced by high fructose and partially protects from that induced by high-fat diets. In humans, CD36 deficiency may be an important factor in the metabolic adaptation to diet and in susceptibility to some forms of diet-induced pathology.
High blood fatty acids (FAs) are a common feature of insulin-resistant states (1), and raising the level of plasma FAs can induce acute insulin resistance (2). Plasma FAs and insulin sensitivity are negatively correlated (2), and an even stronger negative correlation can be documented with intramuscular triglycerides (TGs) (3, 4). Randle et al. (5) originally stated that excessive muscle FA oxidation induced insulin resistance by inhibiting glucose oxidation. The mechanism proposed (6) involved inactivation of pyruvate dehydrogenase, phosphofructokinase, and hexokinase. However, recent measurements of muscle concentrations of glucose and glucose-6-phosphate using noninvasive nuclear magnetic resonance spectroscopy indicate that the negative FA effect on glucose utilization may be exerted primarily at the level of membrane transport (7). In addition, the reduction in intracellular glucose utilization associated with transport inhibition appears to mostly reflect a decrease in glucose conversion to glycogen rather than in glucose oxidation (8).
A valuable model for the study of the effects of alteration in FA utilization on insulin responsiveness is the recently generated CD36-deficient mouse (9). CD36, also known as fatty acid translocase (FAT) (10), is a multispecific, integral membrane glycoprotein (11, 12) that has been identified as a facilitator of FA uptake. Its function in binding and transport of FA was documented in vitro by affinity labeling with FA derivatives and by cell transfection studies (13). Generation of mice deficient in or overexpressing CD36 established the physiological role of the protein. The CD36-deficient mouse (9, 14) exhibits greater than a 60% decrease of FA uptake and utilization by heart, oxidative skeletal muscle, and adipose tissues. In contrast, mice with muscle CD36 overexpression (15) have enhanced FA oxidation in response to contraction, documenting the rate-limiting role of the transport step.
CD36 deficiency has been genetically linked to insulin resistance in the spontaneously hypertensive rat (SHR), a rodent model of human metabolic syndrome X (16, 17). Supplementing the diet with short- and medium-chain FA, which would not require CD36-facilitated transport, improves SHR hyperinsulinemia and myocardial hypertrophy (18).
In contrast to the findings in the SHR, early data in the CD36-null mouse documented fasting hypoglycemia (9), while muscle CD36 overexpression was associated with hyperglycemia and hyperinsulin-emia (15). In humans, although incidence of CD36 deficiency (19) is highest (6–18.5%) in subpopulations with a high prevalence of diabetes type 2, such as African-Americans (20) and Africans (21), preliminary findings yielded divergent conclusions regarding a possible link with insulin resistance (22, 23). We speculated that the variable results may reflect an effect of environmental conditions, and especially dietary ones. In this respect, it is worth noting that most studies of insulin resistance in the CD36-deficient SHR reported using a high-fructose diet (17, 24). This interpretation would also be in line with the evidence that manifestation of insulin resistance in humans reflects both genetic and dietary influences (25). So we examined the CD36-null mouse for insulin responsiveness and susceptibility to diet-induced insulin resistance. The CD36-null mouse is genetically simpler than the SHR and may provide a better model for the CD36-deficient human. The defect in myocardial FA uptake, consequent to CD36 deficiency, measured in vivo, is similar and approximates 70% in both mice (14) and humans (26), while it is smaller (25%) in the SHR (18).
CD36-null mice and wild-type (WT) control littermates of identical genetic background (93.75% C57BL/6 and 6.25% 129SvJ) (9, 14) were housed in a facility equipped with a 12-hour light cycle and were fed ad libitum a standard chow diet (diet no. 5001; Purina Mills Inc., St. Louis, Missouri, USA), a fructose-rich diet (diet no. 00202; Harlan Teklad, Madison, Wisconsin, USA), or a high-fat diet (diet no. 99012501; Research Diets Inc., New Brunswick, New Jersey, USA). The chow diet contained 50% per weight complex carbohydrate, 22% protein, and 6.5% fat mostly as polyunsaturated FAs. The fructose diet (24) consisted of 60% fructose, 20% protein, and 7% fat as soybean oil. The high-fat diet (27) contained 18.2% sucrose, 33% casein, and 32% safflower oil. Mice were fed safflower and fructose diets for 16 and 12 weeks, respectively, in line with the length of time required to induce insulin resistance, as reported in previous studies (24, 27).
Age- and sex-matched littermates were used for the studies at 14–18 weeks when weights ranged from 20 to 24 g for females and 28 to 35 g for males. All studies were in compliance with the guidelines of the institutional animal care committee.
Tail-vein blood (about 200 μl) was collected from fasted (16 hours) mice into heparin or EDTA-containing (FA determination) tubes. Plasma FFA was measured using a kit from Wako Chemicals USA Inc. (Richmond, Virginia, USA), TGs and glucose with kits from Sigma-Aldrich (St. Louis, Missouri, USA), and insulin with an RIA kit from Linco Research Inc. (St. Charles, Missouri, USA).
Mice fasted for 16 hours were injected intraperitoneally with a 25% glucose solution (1.5 g/kg). Blood (5–10 μl) was collected from the tail prior to and at 10, 20, 30, 45, 60, 90, 120, and 180 minutes after injection. Glucose was measured using a Precision Q.I.D. monitoring system (MediSense; Abbott Laboratories, Abbott Park, Illinois, USA). A similar protocol was used for measuring the response of plasma insulin to the glucose load except that blood (100 μl) was collected at 0, 30, and 60 minutes after injection for insulin and glucose determination.
Fasted mice were injected in a lateral tail vein with 100 μl of fluorodeoxyglucose (18F-2-FDG) in saline (about 5 μCi, half-life of 110 minutes). For determining decay of blood FDG–specific activity, blood samples (25 μl) were collected and counted at the indicated times after FDG injection, and glucose concentration was measured. The mice were killed after 2 hours, and the tissues were rapidly removed, rinsed with cold saline, and blotted dry. Tissues along with blood samples and an aliquot of the injected solution were counted in an NaI auto–gamma counter. To adjust for the difference in blood glucose between WT and CD36-null mice, specific activity of blood FDG at 2 minutes after injection was used as the 100% value (28, 29).
Tissue FDG at the end of the experiment reflects total uptake, since phosphorylated FDG is trapped intracellularly and not metabolized further. Glucose uptake by each tissue was calculated by dividing tissue 18F-2-FDG counts by the calculated integral of blood specific activity (SA) according to the equation:
where 18F-2-FDG = counts per g tissue at the end of the experiment (2 hours), and SA = counts/μg.
Livers, hearts, and hind limbs were harvested from anesthetized mice, freeze-clamped in liquid nitrogen, and stored at –80°C for later analysis. Tissue glycogen was measured as glucose after hydrolysis with KOH (30%) and HCL (0.6 N) (30). TG content was determined enzymatically after lipid extraction (31) as previously described (32). Tissue protein was determined according to the method of Markwell et al. (33).
Rates of glycogenesis in isolated muscles were determined as previously described (34). Extensor digitorum longus (EDL) and soleus muscles were removed intact from anesthetized mice and preincubated (30 minutes) in Krebs-Henseleit buffer (pH 7.4) containing glucose (8 mM) in the absence or presence of maximal stimulating concentrations of insulin (20 mU/ml). Incubations were for 60 minutes in the same fresh buffer that now also contained [U14C]-D-glucose (0.5 μCi per vial). At the end of the incubation, muscles were blotted dry, frozen in liquid nitrogen, and stored at –80°C until analyzed. Glycogen was precipitated using standard procedures (34). Rate of glycogen synthesis from D-glucose was determined from the 14C content of the glycogen pool. Glucose oxidation was monitored from the 14CO2 produced. This was determined by transferring a 0.5-ml aliquot of incubation buffer to a sealed glass vial and acidifying the buffer with 1 M H2SO4. Liberated 14CO2 was captured by a suspended center well containing benzethonium hydroxide. Center wells were placed in scintillation vials and counted.
Data are shown as means ± SE. They were calculated using InStat (GraphPad Software, San Diego, California, USA) and analyzed using two-tailed unpaired t test. Results for the glucose tolerance tests were analyzed with two-factor repeated-measures ANOVA before application of the two-tailed unpaired t tests.
Compared with those in WT mice, fasting glucose and insulin levels were significantly lower in CD36-null mice. Data shown (Table (Table1)1) are for both sexes combined. However, the hypoinsulinemia was more pronounced in males than in females (insulin was 54% versus 33% lower, respectively). The index 1/insulin was higher in CD36-null mice, suggesting enhanced insulin sensitivity.
CD36-null mice (males and females) had a significantly enhanced ability to clear an intraperitoneal glucose load (Figure (Figure1a).1a). At 20 minutes after the load, plasma glucose reached a peak concentration, which was about 30% lower in the CD36-null mice. Area under the clearance curve (inset) was 22% lower in null mice as compared with age- and sex-matched WTs (P < 0.001). To determine whether the insulin response to glucose was altered, plasma insulin was measured before and at 30 and 60 minutes after the glucose load. Insulin levels after glucose administration (Figure (Figure1b)1b) and the insulinogenic index (Table (Table1),1), which reflects pancreatic β cell function (35, 36), were similar for WT and CD36-null mice.
To identify the tissues responsible for the increased clearance of blood glucose, we measured glucose uptake in vivo using radioactive 18F-2-FDG. As shown in Figure Figure2a,2a, uptake of FDG was increased in hearts (five times), diaphragms (three times), soleus (four times), gastrocnemius (three times), and hind limb muscle (two times) from CD36-null mice as compared with WT mice. Uptake was unaltered in adipose tissue and decreased in liver (Figure (Figure2b).2b). During the experiment, blood glucose level was constant in both WT and CD36-null mice. However, decay of 18F-2-FDG–specific activity in blood (Figure (Figure2c)2c) was faster in CD36-null mice, indicating that more endogenous glucose was being released into the blood to dilute specific activity of the FDG tracer. Since the mice were fasted overnight, the major glucose source would be the liver, suggesting increased hepatic glucose output.
In a subsequent experiment (data not shown), we tested the response of tissue glucose uptake to exogenous administration of a high dose of insulin (0.5 IU/kg body weight). Insulin produced severalfold increases in 18F-2-FDG uptake in heart, diaphragm, soleus muscle, gastrocnemius, and adipose tissues of WT mice. In CD36-null mice, the insulin-induced increases in FGD uptake were significantly smaller as compared with those in WT, for heart, diaphragm, and soleus. This was in line with the observation that oxidative muscles of CD36-null mice had optimal rates of glucose utilization at the endogenous insulin levels present in the fasted state.
To directly examine insulin sensitivity, in vitro tests were carried out using the isolated soleus, a mostly oxidative muscle; and the EDL, a mostly glycolytic muscle. As shown in Table Table2,2, glycogenesis by the soleus in response to a maximal concentration of insulin was superior for muscles from both fed (+ 316%) and fasted (+ 495%) CD36-null mice as compared with corresponding muscles from fed (+ 111%) and fasted (+ 59%) WT mice. In contrast, there was no significant alteration of the insulin response of the EDL. Glucose oxidation by soleus and EDL muscles from WT and CD36-null mice was similar and responded equally well to insulin (data not shown). Thus, in CD36-null mice, the effect of insulin on glycogenesis in the soleus but not in the EDL was enhanced, while insulin effect on glucose oxidation was not altered.
Glycogen and TG contents were measured in heart and hind limb, which are typical of oxidative and glycolytic muscles, respectively. Liver content was determined since it plays an important role in the homeostasis of both plasma glucose and lipids. As shown in Table Table3,3, CD36-null mice had lower glycogen levels in the liver and in heart and hind limb muscles.
TG levels were decreased in muscle and heart of CD36-null mice (by 49% and 42%, respectively) but were increased twofold in the liver (Table (Table33).
We next examined whether the high glucose-to-FA utilization ratio created by CD36 deficiency increases susceptibility to metabolic pathology from diets with a high glycemic load while protecting from that induced by diets high in fat.
A diet rich in fructose induces a syndrome X–like metabolic phenotype in the SHR but not in the normal control rat (16, 17). In WT and CD36-null mice (Table (Table4),4), fructose feeding did not alter blood glucose but it increased blood insulin. The increase was small in WT mice (37 versus 28), while it was more than fourfold for CD36-null mice (50 versus 12). Blood levels of FAs and TGs (Table (Table5)5) were increased by the fructose diet in both groups but were higher in null as compared with WT mice on both the chow and fructose diets.
Glucose tolerance of CD36-null mice was markedly impaired by the fructose diet, while no effect was observed in WT mice (Figure (Figure3a).3a). Area under the clearance curve (inset) was higher in null than in WT mice fed fructose (P < 0.01). The impairment in glucose tolerance in null mice was significant after 3 weeks on the diet (data not shown), although it was less pronounced than at the time the mice were killed at 12 weeks.
CD36-null mice fed fructose, secreted more insulin in response to the glucose load than did WT mice (Figure (Figure3b).3b). Nulls also exhibited peripheral insulin resistance, since FDG uptake by heart and gastrocnemius muscles was significantly lower than uptake by corresponding muscles from WT mice (Figure (Figure4).4). This contrasted with the situation on the chow diet (Figure (Figure2),2), in which FDG uptake in null muscles was severalfold higher than in WT muscles. A comparison of FDG uptake in CD36-null hearts on chow (Figure (Figure2)2) versus fructose (Figure (Figure4)4) yields values of 240 versus 20 μg/h/g tissue, which represents greater than a 90% drop, while FDG uptake by WT hearts averaged 50 μg/h/g for both diets.
We examined whether CD36 deficiency, which impairs FA utilization by muscle, would protect against peripheral insulin resistance consequent to increased consumption of dietary fat. A diet rich in safflower oil (37, 38) has been shown to be the most effective in impairing glucose tolerance in the C57BL/6J mouse line and was used for these studies. This diet induces peripheral insulin resistance without increasing blood insulin and TG, since it inhibits hepatic lipogenesis and VLDL production (38).
As shown in Table Table4,4, blood levels of glucose and insulin were similar in WT and CD36-null mice fed the high-safflower diet, while TG and FA levels (Table (Table5)5) were higher in the CD36-null mice.
Glucose tolerance was impaired by safflower feeding to an equal extent in WT and CD36-null mice (Figure (Figure5a),5a), and the areas under the clearance curves were similar. However, clearance curves were significantly different for each group when compared with those of the mice on the chow diet (P < 0.05). The CD36-null mice fed safflower secreted significantly more insulin than the WT mice at 30 but not at 60 minutes after the glucose load, as shown in Figure Figure55b.
For the high-fructose diet (Table (Table6),6), TG levels in livers from CD36-null mice were higher than those in WT livers, while levels in heart and skeletal muscle were similar for both groups. When data are compared with those for mice on the chow diet, TG levels were increased by fructose feeding in all tissues. The relative increase was most marked in muscle (11- and 26-fold for WT and null, respectively).
For the high-safflower diet (Table (Table6),6), TG levels in heart and hind limbs were about 25% lower for CD36-null than for WT mice; however, the differences did not reach statistical significance (n = 6 per group). Levels were similar in the liver. When data are compared with those for mice on the chow diet, liver TGs for both mouse groups were lower on the safflower diet, which has been reported to inhibit hepatic lipogenesis (38).
Insulin sensitivity of glucose utilization in muscles of mice fed the safflower diet was examined in the incubated soleus in vitro. As shown in Table Table7,7, glucose incorporation into glycogen exhibited similar responsiveness to a maximal concentration of insulin, when muscles from WT and CD36-null mice are compared. In contrast, glucose oxidation by the soleus was almost unresponsive to insulin in WT muscles, while it was still responsive in CD36-deficient muscles.
The CD36-deficient mouse, which exhibits defective FA uptake (14, 18) and severely reduced FA oxidation by heart and oxidative muscle, provides a valuable model for examining the link between FA uptake and insulin responsiveness (39).
As shown in this study, the presence of high blood FAs is not sufficient to induce peripheral insulin resistance. The CD36-null mouse maintained on a chow diet has high blood FAs and TGs but is more insulin-sensitive than the WT mouse. The data illustrate the concept that muscle insulin responsiveness is linked to the balance between FA uptake and FA oxidation, with resistance reflecting conditions where uptake exceeds oxidative capacity. In the CD36-null mouse, FA uptake is decreased below the muscle’s capacity to both oxidize and esterify FAs. This is evidenced by large drops in unincorporated intracellular FA (14) as well as in muscle TG levels (Table (Table3).3). Our data are consistent with previous findings that deactivation of the PPARα nuclear receptor, which regulates expression of proteins related to muscle FA utilization, is associated with an amelioration of insulin resistance in the apoE-null mouse (40). CD36 expression in muscle is markedly decreased by deactivation of PPARα, which would be expected to impair peripheral FA uptake.
The enhanced insulin sensitivity of the CD36-null mouse contradicts the previously documented linkage of CD36 deficiency to insulin resistance in the SHR (16, 24). However, there is an important difference between the CD36-null mouse and the SHR with respect to muscle FA metabolism. In the SHR, there is evidence to indicate that, despite CD36 deficiency, muscle FA uptake exceeds oxidative capacity. First, the defect in FA uptake by SHR muscle is modest (25% versus 70% in the CD36-null mouse). Second, a fivefold rise in unincorporated cellular FAs was measured in SHR muscle (18), while a 50% decrease is observed in the CD36-null mouse muscle (14). Third, net FA incorporation into TGs is unchanged in SHR muscle (18), which contrasts with a 50% drop in the mouse (14). In line with this, there is a small rise in muscle TG levels in the SHR (41), as opposed to the almost 50% drop in muscle TG in the CD36-null mouse (this study). The biochemical mechanism for these differences still has to be determined. One reason may be that a smaller proportion of exogenous FAs is cycled into TGs before oxidation in rat versus mouse muscle (42). So in the rat, defective uptake would impact FA oxidation more than it would impact FA esterification. In addition, CD36 may facilitate a smaller fraction of FA uptake by rat than by mouse muscle (18).
A significant observation in the present study is that the WT mouse fed high fructose does not exhibit glucose intolerance and decreased muscle glucose utilization despite markedly increased muscle TGs. This supports the interpretation that muscle TGs correlate positively with insulin resistance (43, 44) only when levels reflect excess FA uptake relative to oxidation and that resistance is linked to the accumulation of FA metabolites, possibly FA-acyl-CoA (44).
In the CD36-null mouse, the defect in FA uptake results in strongly enhanced glucose uptake and incorporation into glycogen by muscle. This is a mirror image of the findings in humans, where the major effects of increased FA supply are reductions in muscle glucose uptake and glycogenesis (45, 46). The important role of oxidative muscle in mediating the effects on overall insulin responsiveness is highlighted by the data, since the enhanced insulin sensitivity was primarily limited to oxidative muscles, where the FA defect is most pronounced.
Finally, it is important to note the effect of muscle FA uptake on liver metabolism. Although CD36 expression in the liver is normally low (10), the impact of the deficiency on liver metabolism is significant. Consequent to the defect in peripheral FA utilization, FA flux to the liver is increased. As a result, hepatic FA uptake (14) and FA incorporation into TGs and cholesteryl ester (C. Coburn et al., unpublished observations) are enhanced, increasing hepatic TG content and production of VLDLs (9). Livers of CD36-null mice also had decreased glycogen content, reflecting increased hepatic glucose output and the need to supply more glucose to heart and muscle tissues.
Epidemiological data indicate that the Western diet is associated with a higher incidence of type 2 diabetes (recently reviewed in ref. 47). This diet is characterized by a high content of saturated fat and of simple, rapidly absorbed carbohydrates. There is evidence to support a negative effect of both components on insulin responsiveness of glucose utilization. Diets high in fat induce peripheral insulin resistance, in part by promoting obesity (47, 48). Similarly, diets rich in simple sugars, such as sucrose or fructose, which are rapidly absorbed and which have a high glycemic index, when given in high quantities (high glycemic load) promote the glucose intolerance and dyslipidemia characteristic of the metabolic syndrome X (47, 49–51). It has been well documented that humans differ in their susceptibility to the diabetogenic effects of both dietary fat and carbohydrates (47, 52), resulting in variability of the reported data. The findings in this study document how alterations in one gene, in this case CD36, differentially modulate the metabolic adaptation to various diets and the susceptibility to diet-induced pathology.
The high susceptibility of the CD36-null mouse to the diabetogenic effect of a high-fructose diet is similar to that previously documented in the SHR (24). In control rats (53, 54) and mice (this study), high-fructose diets produce no change in peripheral insulin responsiveness. In contrast, the SHR strain and the CD36-null mouse are very sensitive to the effect of a high-fructose diet and develop hyperinsulinemia and insulin resistance. The SHR fed fructose has been widely studied as a valid model of human syndrome X. In healthy human volunteers, short-term feeding of a high-fructose diet is not associated with clear symptoms of insulin resistance (25, 55, 56). However, if our findings could be extrapolated to humans, they would suggest that CD36-deficient humans may be at a high risk of developing insulin resistance from diets high in simple sugars like fructose and sucrose.
CD36 deficiency would be expected to protect against the diabetogenic effects of a high-fat diet, since it impairs peripheral FA utilization. In this respect, deactivation of PPARα, which would downregulate muscle CD36, protects from insulin resistance induced by a Western diet (57). Despite the impaired glucose tolerance observed in CD36-null mice fed high safflower, there was partial protection, as shown with in vitro soleus incubations, where glucose oxidation remained responsive to insulin in CD36-deficient muscles. In addition, preliminary studies of glucose metabolism under hyperinsulinemic clamp conditions indicate a partially protective effect of CD36 defi-ciency with respect to insulin responsiveness of he-patic glucose output and of muscle glucose transport (J. Kim et al., unpublished observations). More studies of the response of CD36-deficient animals to high-fat feeding will be needed, as they should help dissect the respective roles of the liver and muscle in the etiology of insulin resistance.
In summary, based on our data, CD36-deficient humans would be predicted to exhibit enhanced or impaired insulin resistance depending on the diet consumed. They could be particularly susceptible to metabolic pathology from diets with a high glycemic load, which have been positively correlated to incidence of non–insulin-dependent diabetes (58). In contrast, they may be less susceptible to the diabetogenic effects of diets high in fat. It also remains to be determined whether high CD36 expression levels in muscle would increase susceptibility to insulin resistance from high-fat diets. Since incidence of CD36 deficiency in humans ranges from 0.3% to 18.5% depending on the subpopulation (20, 21), it would be important to examine the link between functional levels of CD36 and insulin responsiveness in humans. However, the influence of environmental factors is likely to make such a study very complex. In addition to diet, lifestyle may be important, since contraction influences muscle FA uptake and sarcolemmal levels of CD36 (59, 60). The difficulty environmental interactions create in studies of population genetics has been illustrated recently by work on the LDL and PPARγ genes (61, 62).
We thank the PET scanning facility at North Shore University Hospital (Manhasset, New York, USA) and Ralph Matacchieri in particular for preparation and donation of the fluorodeoxyglucose. This work was supported by American Heart Association Fellow-ship AHA0020639T (to T. Hajri) and NIH grant RO1-DK33301 (to N.A. Abumrad). The work was also supported by a grant from the Canadian Institutes of Health Research to A. Bonen.