Heart-type fatty acid-binding protein (H-FABP) is required for high rates of skeletal muscle long-chain fatty acid (LCFA) oxidation and esterification. Here we assessed whether H-FABP affects soleus muscle glucose uptake when measured in vitro in the absence of LCFA. Wild-type and H-FABP null mice were fed a standard chow or high-fat diet before muscle isolation. With the chow, the mutation increased insulin-dependent deoxyglucose uptake by 141% (P < 0.01) at 0.02 mU/ml of insulin but did not cause a significant effect at 2 mU/ml of insulin; skeletal muscle triglyceride and long-chain acyl-CoA (LCA-CoA) levels remained normal. With the high-fat diet, the mutation increased insulin-dependent deoxyglucose uptake by 190% (P < 0.01) at 2 mU/ml of insulin, thus partially preventing insulin resistance, and it completely prevented the threefold (P < 0.001) diet-induced increase of muscle triglyceride levels; however, muscle LCA-CoA levels showed little or no reduction. With both diets, the mutation reduced the basal (insulin-independent) soleus muscle deoxyglucose uptake by 28% (P < 0.05). These results establish a close relation between FABP-dependent lipid pools and insulin sensitivity and indicate the existence of a nonacute, antagonistic, and H-FABP-dependent fatty acid regulation of basal and insulin-dependent muscle glucose uptake.
- heart-type fatty acid-binding protein
- insulin resistance
“fatty acid overload” due to increased extracellular fatty acid levels (18, 26, 31) or pharmacologically reduced fatty acid oxidation (6) can impair insulin action on skeletal muscle (see review in Ref. 24), and insulin resistance in obesity has been associated with reduced muscle oxidative capacity (see reviews in Refs. 16 and 17). These situations all favor the accumulation of lipids such as triglycerides and fatty acyl-CoA (5, 16). Conversely, genetic interference with long-chain fatty acid (LCFA) uptake was shown to prevent fat-induced accumulation of muscle triglyceride and fatty acyl-CoA, as well as muscle insulin resistance (19). However, the levels of triglycerides and long-chain acyl-CoA (LCA-CoA) in muscles are not always correlated with insulin sensitivity (1, 4, 12, 14, 30), and the exact pathways of fatty acid-induced insulin resistance are still being elucidated (1, 27, 28, 31). In addition, in some cases (29, 32) it has also been found that basal (insulin-independent) glucose uptake can be increased because of fat diet, and basal skeletal muscle glucose uptake was increased in obese patients (16). The LCFA-induced reduction of insulin-sensitive muscle glucose uptake is nonacute, because it is maintained for a while in the absence of LCFA (e.g., 20, 31); similarly, any stimulatory effects of LCFA on basal muscle glucose uptake would be expected to be nonacute, because LCFA oxidation is well known to acutely reduce glucose oxidation in skeletal muscle (25), an effect that is unlikely to be of pathogenetic significance (17, 28).
Because no pharmacological or genetic treatment can be expected to act at only one site, it is important to compare models of altered fatty acid usage with different primary lesions to identify the common denominators of glucose uptake. Here we examined the role of heart-type fatty acid-binding protein (H-FABP) in the nonacute regulation of basal and insulin-stimulated glucose uptake. H-FABP is an important, if not dominant, LCFA-binding protein in heart and skeletal muscle cytosol and is required for high levels of skeletal muscle LCFA oxidation and esterification, at least under standard diet (3). Thus we were interested to know whether the pool of LCFA associated with H-FABP would be relevant for the nonacute regulation of muscle glucose uptake. Accordingly, we measured the glucose uptake into soleus muscles in vitro after subjecting wild-type and H-FABP null mice (2) to either standard or a high-fat diet.
MATERIALS AND METHODS
Mice lacking the H-FABP gene were originally produced on the 129BALB/c background (2). For the experiments reported here, they were backcrossed onto the C57BL/6 background for at least 7 generations, and F1 offspring of heterozygous parents or F1 offspring of null × null and wild-type × wild-type crosses were used. If not indicated otherwise, males were used. Mice were genotyped with a reliable single-tube PCR assay, as previously described (3). The experiments were approved by the Texas A&M University Laboratory Animal Care Committee.
Diets and starvation.
Mice received a standard chow (no. 8604, Harlan Teklad) or a high-fat diet [no. TD 93075, Harlan Teklad; main digestibles in g/kg: 289 protein, 207 starch, 90 sucrose, 274 shortening (Primex), 16 cellulose] for ∼4 wk, or they were starved overnight (14–16 h) in cages with fresh bedding.
Blood metabolite levels.
Blood was drawn with heparinized (for plasma) or nonheparinized (for serum) glass capillaries from tails of overnight-fasted mice. It was centrifuged immediately to remove blood cells, and the resulting plasma or serum was kept on ice for metabolite analysis within the next 24 h. Glucose levels were measured (no. 315, Sigma Diagnostics) in plasma, and free fatty acid levels were measured (no. 994-75409E) in serum. Before determination of triglyceride (no. 336, Sigma Diagnostics) and insulin (kit no. 90060, Chrystal Chemistry, Chicago, IL) levels, plasma was kept frozen (−70°C).
Glucose tolerance tests.
Mice were starved overnight and injected with glucose (20%, 2 mg/g body wt ip) for determination of tail blood glucose and insulin at the indicated time points.
Tissue metabolite levels.
Muscle samples were shock frozen and stored at −80°C. Muscle glycogen concentration was determined essentially as previously described (22) by digestion in 0.5 ml of 1 N KOH, neutralization with 1 N HCl, treatment of an aliquot (0.1 ml) with 0.5 mg/ml of amyloglucosidase (no. A7420, Sigma), and measurement of the released glucose with Sigma's kit no. 315. Muscle triglyceride levels were measured with a modification (21) of a previously described method (11, 29). Levels of LCA-CoA were determined in shock-frozen tissue samples as previously described (22, 33).
Deoxyglucose uptake by isolated muscle.
Overnight-fasted mice were anesthetized with Avertin. Soleus muscles were individually incubated in 1.5 ml of the appropriate buffer (to be described) in continuously gassed (5% CO2-95% O2) 20-ml plastic scintillation vials in a shaking water bath (29°C). Following the procedure of Etgen et al. (10), three-step incubations (each step in a new vial) were performed in Krebs-Henseleit buffer (KHB)-0.1% bovine serum albumin with the following supplements and durations: step 1: 8 mM glucose-32 mM mannitol, 40 min; step 2: 40 mM mannitol, 10 min; step 3: 1 mM 2-deoxy-[3H]glucose at 0.5 μCi/ml (no. NET328, NEN) and 39 mM [U-14C]mannitol at 0.1 μCi/ml (no. NEC314, NEN), 20 min. Insulin (Humulin; no. 0002-8215-01, Eli Lilly) was present throughout the incubations at concentrations that will be indicated. Muscles were then blotted on filter paper, trimmed, weighed, and dissolved in 0.1 ml of hyamine hydroxide at 60°C overnight. Thirty microliters were counted in a dual-label liquid scintillation spectrophotometer. Glucose transport activity (expressed in μmol·20 min−1·g muscle−1) was calculated by subtracting the extracellular from intracellular 2-deoxy-[3H]glucose with the extracellular marker [U-14C]mannitol.
Fatty acid metabolism in isolated soleus muscle.
The setup was the same as for the glucose uptake experiments, but another incubation medium and time schedule were used. The incubation medium consisted of KHB, 2% bovine serum albumin (fatty acid-free, Sigma), 0.1 mM palmitic acid (Sigma), 1 μCi/ml of either [3H]palmitic acid (no. NET043, NEN) or [3H]oleic acid (no. NET289, NEN), and insulin as indicated. Palmitic acid (125 mM stock in ethanol) was mixed with warmed albumin-containing medium until completely clear. Freshly excised soleus muscles were incubated for 60 min at 29°C and freeze-clamped at the end. Fatty acid oxidation was measured through production of tritiated water (7) in 1 ml of incubation medium with a vial incubated without muscle as blank. Fatty acid esterification into triglycerides was quantified by homogenization of the muscles with a glass homogenizer in 1 ml of ice-cold chloroform-methanol (1:1), followed by lipid extraction and thin-layer chromatography (8). Bands of main lipid classes were visualized with iodine vapor, scraped out, and counted in 5 ml of scintillation liquid (Scintisafe, Fisher Scientific).
Data are shown as means ± SE. They were calculated using the GraphPad Prism Statistical software program and analyzed using one-way ANOVA (Bonferroni's multiple comparison test) and a two-tailed Student's t-test, if not otherwise indicated. Comparisons were made between H-FABP null and wild-type mice (★P < 0.05, ★★P < 0.01, ★★★P < 0.001), between standard diet and high-fat/high-sugar diet (+P < 0.05, ++P < 0.01, +++P < 0.001), and between incubations in presence or absence of insulin (⧫P < 0.05, ⧫⧫P < 0.01, ⧫⧫⧫P < 0.001).
LCFA oxidation in isolated soleus muscles.
In the absence of insulin, oxidation of palmitate was decreased by 53% (Fig. 1A) in mutant muscles from nonfasted, chow-fed mice, and oleate oxidation was reduced by 56% (P < 0.01, n = 5 per genotype). With this diet, insulin reduced the rate of palmitate oxidation in wild-type soleus muscle but did not significantly reduce the already lowered rate in H-FABP null muscle (Fig. 1, A and B). Accordingly, the decrement caused by the mutation was significantly (P < 0.05) larger in the absence of insulin than in its presence. A similar result was found with oleic acid (not shown). With the high-fat diet, the genotype-specific reduction was 67% for palmitic acid in the absence of insulin (Fig. 1A), and similar results were obtained from skeletal muscles of fasted mice (Fig. 1B) and females (not shown). With this diet, the insulin-caused reduction of wild-type muscle fatty acid oxidation became nonsignificant.
LCFA esterification in isolated soleus muscles.
In the absence of insulin, esterification of palmitate into triglycerides was lower by 26% in null compared with wild-type muscles from nonfasted, chow-fed mice (Fig. 1C), whereas esterification of oleate was reduced by 38% (P < 0.05, n = 4 per genotype). With this diet, insulin moderately but significantly (P < 0.05) stimulated palmitate esterification in both genotypes. With high-fat diet, esterification of palmitate was reduced by 33% in null compared with wild-type muscles (Fig. 1C); this genotypic reduction occurred on a lowered level, as the high-fat diet reduced the esterification rates in both wild-type (−27%) and null (−34%) muscles. Similar results were obtained with muscles from fasted (Fig. 1D) and female (not shown) mice. Again, insulin moderately stimulated palmitate esterification in both genotypes (Fig. 1, C and D).
Basal deoxyglucose uptake in isolated soleus muscles.
Without insulin, deoxyglucose uptake was lower by ∼28% (P < 0.05) in isolated soleus muscle from H-FABP null compared with wild-type mice, regardless of whether the mice had been fed chow or high-fat diet (Fig. 2A); similar results were seen in female mice (−29 and −27%, P < 0.05) and with extensor digitorum longus (EDL) muscle (−19%, P < 0.05) (data not shown).
Insulin-stimulated deoxyglucose uptake in isolated soleus muscles.
With standard chow, high levels of insulin (2 mU/ml) increased deoxyglucose uptakes by 4.1 μmol·mg−1·20 min−1 (109%) in wild-type, and by 4.7 μmol·mg−1·20 min−1 (+178.3%) in H-FABP null samples (Fig. 2A), but the difference between these increments was not statistically significant. A similar result was obtained with female mice (data not shown). The slightly (but not significantly) larger increment in H-FABP null muscles prompted us to develop a dose-response curve. This curve was significantly steeper in H-FABP null compared with wild-type soleus muscle (Fig. 2B). As a result, insulin-dependent deoxyglucose uptake at 0.02 mU/ml insulin was significantly higher in null vs. wild-type soleus muscle (+141%) and approached plateau levels already at 0.2 mU insulin/ml, but the plateau levels were similar in the two genotypes. With the high-fat diet, total deoxyglucose uptake at 2 mU/ml insulin was only nonsignificantly (+30%) increased in H-FABP null vs. wild-type muscle (Fig. 2A); however, the genotypic difference of the insulin-dependent increments (Fig. 2B) was much larger (+190%) and clearly significant (P < 0.01). In the presence of the high insulin concentration, the total rate of deoxyglucose uptake by wild-type soleus muscle was significantly reduced (−43%) by high-fat diet, whereas the corresponding reduction with H-FABP null muscle was smaller (−24%) and not significant at P < 0.05 (Fig. 2A).
Lipid levels in vivo.
With standard diet, the mutation doubled (+97%) plasma fatty acid levels but did not affect plasma triglyceride levels (Table 1), similar to the phenotype before the backcross (2). With high-fat diet, circulating fatty acid levels more than doubled (+117%) in wild-type mice but increased less (+36%) in H-FABP null mice due to the higher starting level, although the free fatty acid level remained higher (+24%) than in wild-type mice (Table 1). No genotype-specific differences were observed in plasma triglyceride levels of mice on a high-fat diet, although triglyceride levels were increased in both genotypes (Table 1).
Muscle triglyceride levels remained normal in the H-FABP null mice kept with the standard diet, but the H-FABP null mutation effectively prevented the almost threefold increase of muscle triglycerides caused by high-fat diet (Fig. 3A). Total and individual LCA-CoA levels in gastrocnemius muscle were not affected by the mutation with the standard diet, but with the high-fat diet they were moderately reduced (Table 2). Statistical significance at P < 0.05 was reached for C18:3 (−43%) with the two-tailed t-test, and for C16:0 (−39%), C18:3 (−43%), and total LCA-CoA (−26%) with the one-tailed t-test (Table 2). LCA-CoA were also measured in soleus muscles, but statistics could not be applied because solei had to be pooled (1 muscle per mouse, 7 mice per pool). The results (2 pools per genotype) showed total LCA-CoA levels of 23.3 and 27.6 (male wild type) vs. 21.5 and 21.2 (male H-FABP null) nmol/g, and 21.1 and 20.1 (female wild type) vs. 21.3 and 20.0 (female H-FABP null) nmol/g. Thus, with high-fat diet there was moderate decrease of muscle LCA-CoA levels due to the mutation in males.
Carbohydrate and insulin levels in vivo.
With the standard diet, fasting glucose levels were decreased by ∼25% in male and female H-FABP null vs. wild-type mice (Table 3). With the high-fat diet, glucose levels increased in both wild-type (males, +110%; females, +73%) and H-FABP null (males, +127%; females, +87%) mice, but they remained lower in null vs. wild-type mice (males, −19.5%; females, −18%; Table 3). Skeletal muscle glycogen levels were decreased in H-FABP null skeletal muscles by 25% under standard diet and by 26.5% under high-fat diet (Fig. 3B). Mice maintained on standard diet did not exhibit a genotype-specific difference of glucose tolerance, but with the high-fat diet (which significantly impaired glucose tolerance in both genotypes) we observed a moderate yet significant improvement of glucose tolerance in H-FABP null vs. wild-type mice 15 min after injection, but not at the later time points (Fig. 4A). With the standard diet, fasting insulin levels in H-FABP null mice were significantly decreased (−66%) compared with wild-type levels (Table 3) [in females the decrease (−30%) was not significant], but after glucose injection, insulin levels increased faster in H-FABP null mice than in wild-type mice, resulting in identical levels by the end of the assay (Fig. 4B). With the high-fat diet, fasting insulin plasma levels increased by a comparable factor in wild-type (males, +446%; females, +803%) and H-FABP null (males, +469%; females, +549%) mice, meaning that the insulin levels remained substantially lower than in the H-FABP null mice (Table 3). However, unlike those after standard diet, the insulin levels did not increase further during the glucose tolerance test (Fig. 4B).
The present study provides further support for an important role of H-FABP in muscle LCFA and glucose metabolism.
First, our data extend our previous result that muscle H-FABP is a major determinant of myocellular uptake and metabolic availability of palmitate during a standard diet (3, 21) by adding another fatty acid (oleate) or another diet (high-fat diet) and by comparing fasting with feeding. At least with standard chow, LCFA metabolism was reduced without a decrease of total and individual LCA-CoA levels, similar to reduced LCFA oxidation, but normal LCA-CoA levels in livers of mice lacking liver FABP (9). Furthermore, under nutritional stress (high-fat diet), lack of H-FABP completely prevented the threefold accumulation of triglycerides (while reducing LCA-CoA levels only moderately). This finding mirrors our previous observation (3) that the decrease of triglyceride levels normally observed during in vitro incubation is blunted in the H-FABP null soleus muscle. Thus H-FABP is a limiting factor for muscle LCFA metabolism regardless of physiological situation, but more work is needed to clarify the mechanism leading to altered triglyceride levels in some (contraction, fat diet) but not other (standard chow) conditions.
Second, we showed here that FABP is an important mediator of at least some of the fatty acid effects that reduce muscle insulin sensitivity or responsiveness. Current thinking (24, 27, 28) holds that relatively slow effects of nonoxidative LCFA metabolites are important in opposing insulin action. In agreement with this notion, we found not only that H-FABP null muscles showed increased insulin sensitivity (standard chow) and responsiveness (fat diet) but that these effects were seen in the absence of added LCFA (during both preincubation and labeling), making rapid substrate-level competition between glucose and LCFA oxidation (25) an unlikely explanation. Our results vary this notion by showing that insulin sensitivity of glucose uptake can be increased by reduced LCFA binding/metabolism while normal insulin responsiveness is maintained. It is also of note that LCA-CoA levels were not affected by the mutation under this condition (standard chow). With the high-fat diet, when insulin-dependent deoxyglucose uptake by isolated soleus muscle was significantly stimulated by the H-FABP null mutation even at high (2 mU/ml) levels, we saw only a trend toward decreased cellular LCA-CoA levels. Therefore, we suggest that the metabolic availability of LCA-CoA (indirectly determined by H-FABP) may be more relevant for insulin sensitivity or resistance than cellular LCA-CoA levels. This would fit various recent observations that a higher insulin sensitivity can be associated with moderate (15) or no (1, 4) decrease of myocellular LCA-CoA levels. The H-FABP null mice are a useful model with which to study whether the metabolic availability of LCA-CoA or another aspect of LCFA metabolism regulates insulin-dependent glucose transport.
Third, we observed that in vitro soleus muscle deoxyglucose uptake was reduced by the mutation in the absence of insulin, regardless of the nature of prior diet. Thus, at least in vitro and without added fatty acid, the decreased basal uptake antagonized the increased insulin sensitivity/responsiveness, thereby reducing the net change of skeletal muscle glucose uptake. These results resemble our previous observation that glucose oxidation (measured in the absence of insulin but presence of fatty acid) was decreased in soleus muscle isolated from H-FABP null mice fed a standard diet (3), and in the decreased basal glycogen synthesis rates (measured in absence of fatty acid) in isolated soleus muscle from mice lacking the plasma membrane LCFA transporter CD36, a mutation that at the same time increases the insulin responsiveness of glycogen synthesis (13).
Fourth, we found here that blood LCFA levels were increased, and blood glucose and insulin levels were decreased, in H-FABP null mice. We speculate that hormonal and substrate conditions in vivo dampen the nonacute genotype-specific changes of muscle glucose uptake that we observed in vitro. For example, the reduced plasma insulin levels might limit the effect of the nonacutely increased insulin sensitivity/responsiveness, whereas the presence of LCFA might, through the glucose-fatty acid cycle (25), reduce glucose uptake in wild-type muscle more than in H-FABP null muscle (which oxidizes LCFA less efficiently) and thus offset the nonacutely reduced basal glucose uptake. It may not surprise, then, that glucose tolerance was not (chow) or only slightly (fat diet) improved in H-FABP null vs. wild-type mice. Thus both the acute and the nonacute effects of substrate and hormone concentrations will have to be studied to fully understand the role of H-FABP in muscle glucose uptake in vivo. Such information will also be helpful for understanding the energetic impact of reduced LCFA oxidation in vivo that involves reduced soleus muscle ATP and phosphocreatine levels, but not a reduced mitochondrial enzyme capacity (3).
In conclusion, we have described here two nonacute and opposite effects of H-FABP on muscle glucose uptake. Our results are compatible with the view that lipid pools determined by H-FABP are more important than total cellular lipid levels for regulating basal and insulin-dependent skeletal muscle glucose uptake.
This work was funded by the Department of Pathobiology at Texas A&M University, the American Heart Association Texas Affiliate (no. 0355051Y to B. Binas), and National Institute of Diabetes and Digestive and Kidney Diseases Grant U24-DK-59635 (to G. W. Cline and J. K. Kim).
We thank Drs. John Ivy and Desmond Hunt (University of Texas at Austin) for introducing us to the soleus muscle incubation method, and Dr. Gerald Shulman (Yale University) for criticizing the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 by American Physiological Society