Paradoxical Coupling of Triglyceride Synthesis and Fatty Acid Oxidation in Skeletal Muscle Overexpressing DGAT1

MCK-DGAT1 transgenic mice and wild-type littermates were fed standard rodent diet or high-fat diet (HFD) as previously described (3). The swimming exercise regimen was the same as described (3). Male mice were used in this study. Animal protocols were approved by the Columbia University Institutional Animal Care and Use Committee.

Three-month-old mice were fed HFD or standard rodent diet for 8 weeks before the clamp study. The clamp was conducted in awake mice as previously described (19). All procedures were approved by the Yale University Institutional Animal Care and Use Committee. Briefly, after an overnight fast, [3-³H]-glucose (high-performance liquid chromatography purified; Perkinelmer) was infused at 0.05 μCi/min for 2 h to assess the basal glucose turnover. Hyperinsulinemic-euglycemic clamp was conducted for 120 min with primed/continuous infusion of human insulin (Novo Nordisk; 105 pmol/kg prime, 15 pmol · kg⁻¹ · min⁻¹ infusion), keeping plasma glucose at ∼6.7 mmol/l (basal level). [3-³H]-glucose was infused at 0.1 μCi/min throughout the clamp to estimate insulin-stimulated whole-body glucose fluxes. Calculations of hepatic glucose output (HGO), whole-body glycolysis, and whole-body glycogen synthesis were also as described (20). To assess muscle glucose uptake, separate groups of mice were clamped and 2-deoxy-d-[1-¹⁴C]glucose (2-[¹⁴C]DG; Perkinelmer) was injected as a bolus at the 75th min of the clamp as described (20).

Four-month-old mice pretreated with HFD for 8 weeks were used for these experiments. Mitochondria were isolated from muscle tissues using Mitochondria Isolation Kits (Sigma-Aldrich, St. Louis, MO) according to the manufacturer's instructions. Mitochondrial fatty acid oxidation was performed by measuring the rate of ³H₂O production from the substrate [³H]palmitate present in excess (final concentrations: 0.2 mmol/l palmitate with 10 μCi/ml of 9,10(n)-[³H]palmitate and 1.25% BSA) in a mitochondria incubation medium (containing 20 μg mitochondria protein) as described (21).

Four-month-old mice pretreated with HFD for 8 weeks were used for these experiments. Mitochondrial DNA copy number was measured using quantitative real-time PCR as previously described (22). Mitochondrial DNA for NADH dehydrogenases 1 and 4 (ND1 and ND4) and nuclear DNA for β-actin were coamplified within the same reaction wells (for primer sequences, see supplemental Table 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-1096/DC1). Mitochondrial ATP production rates were measured in isolated mitochondria through time-course monitoring of bioluminescence in response to a luciferin-luciferase ATP-monitoring reagent (Biotherma, Haninge, Sweden) as described (23). Maximal activity of citrate synthase was measured in muscle homogenates by photospectrometry as described (24). β-hydroxyacyl CoA dehydrogenase (HAD) activity was also measured in muscle homogenates by following spectrophotometrically the conversion of acetoacetyl-CoA to β-hydroxybutyryl-CoA by β-HAD as described (25). Total carnitine palmitoyl transferase (CPT) activity was assayed radiochemically in isolated mitochondria by following the conversion of palmitoyl-l-carnitine from palmitoyl-CoA and radiolabeled l-carnitine as described (25,26). CPT-I activity was calculated from the total CPT activity after subtraction of CPT-II activity, which was measured in the presence of 0.2 mmol/l malonyl-CoA that inhibits CPT-I.

Four-month-old mice pretreated with HFD for 8 weeks were used. Muscles were isolated, fixed with 2.5% glutaraldehyde in 0.1M Sorenson's buffer (PH 7.2), and treated with 1% OsO4 also in Sorenson's buffer for 1 h. After dehydration and embedment in Lx-112 (Ladd Research Industries), thin sections (60 nm) were cut using a MT-7000 ultramicrotome, stained with uranyl acetate and lead citrate, and examined under a JEOL JEM-1200 EXII electron microscope. Pictures were captured by an ORCA-HR digital camera (Hamamatsu) and recorded with an AMT Image Capture Engine.

Twenty-week-old weight-matched male MCK-DGAT1 mice and wild-type littermates that had been on a HFD for 8 weeks were singly housed in metabolic chambers. After 2–4 days of adaptation, oxygen consumption, locomotor activity, and respiration exchange ratio were measured continuously over at least 5 days during 12:12 light-dark cycles, using a CLAMS (Columbus Instruments, Columbus, OH) open-circuit indirect calorimetry system as previously described (27).

Four-month-old mice with 8-week HFD pretreatment were used for these experiments. RT-PCR (3,28) and Western blot (3,29) were as described. Gene-specific primer sets are listed in supplemental Table 1. Primary antibodies to pyruvate dehydrogenase kinase (PDK)4, uncoupling protein (UCP)-3, CD36, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) were obtained from Abcam (Cambridge, MA). DGAT activity was measured in vitro using [¹⁴C]-labeled palmitoyl-CoA as previously described (29,30). The method to quantify muscle triglyceride content was as described (3,29).

Statistical differences were first analyzed by ANOVA to determine overall treatment effects (genotype or exercise), followed by post hoc comparisons between concerned study groups (Statistica V6.0, StatSoft, OK). A two-tailed P value of <0.05 was used to indicate statistical significance. For indirect calorimetry measurements, ANOVA repeated measures (Prism, GraphPad Software, San Diego, CA) were used. Post hoc analyses were conducted using Tukey post hoc or Bonferroni comparisons. Data are expressed as means ± SE.

Using glucose tolerance and insulin tolerance tests, we showed previously that MCK-DGAT1 mice were protected against HFD-induced insulin resistance (3). Ex vivo 2-deoxyglucose uptake in skeletal muscles isolated from HFD-fed animals demonstrated that insulin sensitivity was greater in DGAT1-overexpressing muscles than in the wild-type muscles (3). In this study, whole-body and tissue insulin sensitivities were assessed in vivo by hyperinsulinemic-euglycemic clamps in combination with radioisotope-labeled glucose infusions. Three-month-old male MCK-DGAT1 and wild-type mice were fed HFD for 8 weeks (WT-HF and DGAT1-HF, respectively) before the clamp study. Age- and sex-matched wild-type mice on standard rodent diet (WT-NC) were used as baseline control. The major findings are depicted in Fig. 1, and other energetic and clamp parameters are summarized in supplemental Table 2. As expected, WT-HF mice showed marked insulin resistance with the mean whole-body glucose infusion rate at only ∼16 mg · kg⁻¹ · min⁻¹ (Fig. 1,A). DGAT1-HF mice were more insulin sensitive than WT-HF mice, achieving a mean whole-body glucose infusion rate of ∼29 mg · kg⁻¹ · min⁻¹ (Fig. 1,A). However, the baseline whole-body glucose infusion rate in the control WT-NC mice was ∼45 mg · kg⁻¹ · min⁻¹ (Fig. 1,A), indicating that DGAT1 overexpression provided ∼45% protection against HFD-induced insulin resistance. Transgenic mice were not protected from HFD-induced hepatic insulin resistance. Compared with a >50% suppression of HGO in WT-NC mice during the clamping, insulin-mediated suppression of HGO was completely absent in both WT-HF and DGAT1-HF mice (Fig. 1,B). To directly assess muscle insulin sensitivity under the same clamping conditions, a bolus infusion of 2-deoxy-d-[1-¹⁴C]glucose was given. Uptake of 2-deoxy-d-[1-¹⁴C]glucose in skeletal muscle was ∼30% greater in DGAT1-HF mice than in WT-HF mice (Fig. 1 C), confirming that increased lipotoxicity tolerance in skeletal muscle was the physiological basis for the protection of MCK-DGAT1 mice, albeit incomplete, against HFD-induced whole-body insulin resistance.

As expected, muscle DGAT activity was threefold increased in MCK-DGAT1 (Fig. 2,A), which was associated with a ∼70% increase in muscle triglyceride content (Fig. 2,B). To characterize skeletal muscle fatty acid oxidation capacity, we first determined mitochondrial DNA (mtDNA) copy number by assessing the ratio of mtDNA ND1 and ND4 to the nuclear DNA β-actin gene in soleus and anterior tibial muscles. There were 30–45% fewer mitochondria per copy genome DNA in DGAT1-overexpressing tibial and soleus muscles than those of the wild-type controls (Fig. 2,C). However, analyses of mitochondrial ATP production rates (MAPR) per unit muscle protein revealed no reduction in transgenic mice as compared with wild-type mice (Fig. 2,D). Maximal tissue citrate synthase activity was significantly greater in the DGAT1-overexpressing soleus muscle than the wild-type counterpart was, although no significant difference was detected in tibialis anterior (Fig. 2,D). Maximal tissue β-HAD activity per unit muscle protein was significantly greater in the transgenic muscle than the wild-type muscle (Fig. 2,E). Similarly, mitochondrial CPT-I activity per unit mitochondrial protein was significantly greater in the DGAT1-overexpressing muscle than the wild-type muscle (Fig. 2,E). The unchanged or potentially enhanced mitochondrial oxidative capacity in the presence of fewer mitochondria suggested a higher mitochondrial efficiency in DGAT1-overexpressing muscles. Direct examination of soleus muscles by electron microscopy confirmed that fewer, but bigger, mitochondria were present in the DGAT1-overexpressing muscle (Fig. 2,F) than in the wild-type muscle (Fig. 2,G). Additionally, in vitro measurement of maximal fatty acid oxidation rates demonstrated a 20% increase in fatty acid oxidation capacity in the MCK-DGAT1 muscle compared with the wild-type muscle (1,217 vs. 1,019 nmol · min⁻¹ · g⁻¹ mitochondrial protein, P < 0.01; Fig. 2 H).

mRNA levels of marker genes in fat metabolism were examined by real-time PCR in soleus muscle from 4-month-old male wild-type and MCK-DGAT1 transgenic mice after 8-week HFD feeding (Fig. 3 A). DGAT1 mRNA levels were approximately threefold as high in MCK-DGAT1 mice as in wild-type mice (P < 0.05) as expected, and mRNA levels for PDK4 and UCP3 were doubled in MCK-DGAT1 compared with wild-type mice (P < 0.05). There was also a trend of approximate doubling of mRNA levels for CD36 (P = 0.051), CPT1 (P = 0.120), ACC2 (P = 0.072), and GLUT4 (P = 0.117) in MCK-DGAT1 mice. No significant changes were detected in mRNA levels for DGAT2, fatty acids, ACS1, ADRP, sterol regulatory element–binding protein-1c, peroxisome proliferator–activated receptor (PPAR)-α, PPARδ, PPARγ, peroxisome proliferator–activated receptor γ coactivator-β, CRAT, ECHS1, UCP1, UCP2, and PEPCK.

PDK4 is believed to play a critical role in regulating the fuel partitioning between fat and glucose oxidation (31,32). To further understand the changes in PDK4 expression in MCK-DGAT1 muscles, we compared a group of DGAT1-overexpressing mice with a group of wild-type mice after 1-week swimming exercise (3). Using sedentary wild-type mice as a baseline control, both DGAT1 overexpression and exercise led to ∼1.5-fold increases (P < 0.05) in PDK4 mRNA levels in skeletal muscle (Fig. 3,B and C), providing two independent cases in which upregulation of PDK4 mRNA expression was dissociated from glucose intolerance. Western blot of CD36, UCP3, and PDK4 showed 1.4- to 1.8-fold increases of these proteins in DGAT1-overexpressing muscle relative to wild-type muscle (Fig. 3 D).

To determine the physiological significance of the above in vitro findings and to help interpret the physiological consequences of the increased triglyceride synthesis and increased mitochondrial fatty acid oxidation capacity (and higher expression levels of the uncoupling protein UCP3) in DGAT1-overexpressing skeletal muscle, we measured whole-body energy metabolism in vivo. Four-month-old weight- and adiposity-matched (Fig. 4,A) male MCK-DGAT1 and wild-type mice were analyzed in metabolic chambers after 8-week HFD feeding. MCK-DGAT1 mice appeared less physically active than wild-type mice (Fig. 4,B), although the changes seemed to be unrelated to food-seeking behavior, as food intake was the same between MCK-DGAT1 and wild-type mice (Fig. 4,G). In contrast to the changes in physical activity, oxygen consumption rates were significantly higher in the transgenic mice than in wild-type mice (Fig. 4,C, wild-type vs. MCK-DGAT1, P < 0.05, repeated-measures ANOVA). No significant difference was detected in respiration exchange rates (RER) under these conditions (Fig. 4,D). Further experimental maneuver with two consecutive 24-h measurements of Vo₂ during fasting followed by refeeding revealed that the difference in energy expenditure was accentuated during refeeding. As expected, energy expenditure reduced and increased in response to prolonged fast and refeeding, respectively, in both wild-type and MCK-DGAT1 mice (Fig. 4,E). However, the increase during refeeding was markedly greater in the transgenic mice than in wild-type mice (P < 0.05, repeated-measures ANOVA) (Fig. 4,F). As was during regular ad lib feeding, no difference in food intake was observed during the 24-h refeeding period between MCK-DGAT1 and wild-type mice (Fig. 4 G).

Long-term food intake was followed in mice fed a standard rodent diet (total 11 weeks) or a HFD (total 8 weeks), starting at 2 months of age. Periodic fluctuations in food intake were observed in both transgenic and wild-type mice, more so in standard rodent diet–fed mice (Fig. 5,A). However, no difference was found between the two genotypes either on standard rodent diet or on HFD (Fig. 5,A and B). A small but significant body weight difference (5–10%) was found between wild-type and MCK-DGAT1 mice on HFD, which became evident after 5 weeks of HFD feeding and persisted throughout the remaining study period (Fig. 5 C). In the absence of reduced food intake or increased physical activities, this finding indicates that metabolic energy expenditure in DGAT1-HF mice was higher than that of the WT-HF mice. No differences in weight were observed between wild-type and MCK-DGAT1 mice on standard rodent diet (data not shown).

In this study, we first demonstrated in vivo that although skeletal muscle of MCK-DGAT1 mice was more resistant to developing HFD-induced lipotoxicity, their liver was as susceptible for lipotoxic insulin resistance as that of wild-type mice. Under hyperinsulinemic-euglycemic clamp conditions, MCK-DGAT1 mice were ∼45% protected against HFD-induced insulin resistance at the whole-body level compared with wild-type mice. The greater whole-body insulin sensitivity in HFD-fed MCK-DGAT1 mice (compared to HFD-fed wild-type mice) was associated with ∼30% higher uptake of bolus-injected 2-deoxyglucose in skeletal muscle under the same feeding and clamp conditions, whereas insulin suppression of HGO was similarly impaired in these mice, indicating that liver did not contribute to the improved whole-body insulin sensitivity.

We showed that although triglyceride synthesis and storage capacities were increased in DGAT1-overexpressing muscle, mitochondrial fatty acid oxidation efficiency was also improved. The latter was initially recognized by decreased mitochondrial copy number in combination with the uncompromised muscle citrate synthase activity and increased maximal β-HAD and CPT-I activities in DGAT1-overexpressing muscles. This was then confirmed by increased maximal mitochondrial fatty acid oxidation capacity per unit mitochondrial protein in DGAT1-overexpressing muscles. Interestingly, MAPR per unit mitochondrial proteins was similar in the two genotypes, suggesting that the increased fatty acid oxidation efficiency does not lead to increased energy efficiency. Gene expression levels of muscle PDK4 and UCP3, normalized to β-actin expression levels, were ∼100% greater in HFD-fed MCK-DGAT1 mice than in age-, sex-, and diet-matched wild-type mice. Although increased PDK4 has been associated with greater fuel partition into fatty acid oxidation (31,32), higher levels of UCP3 may also signify increased use of fat as fuel substrate and/or conversion of fat to thermal energy (33,34). It is plausible that the lack of increase in MAPR in DGAT1-overexpressing muscle, in the presence of increased fatty acid oxidation rates, is because of increased uncoupling of oxidative phosphorylation leading to increased heat production. Although heat production was not directly measured, whole-body 24-h Vo₂ production rates were greater in DGAT1-HF than in WT-HF mice in the absence of increased physical activity in the former. The increased energy expenditure in MCK-DGAT1 mice was further substantiated by the modest but significant lower body weight in the transgenic mice relative to wild-type mice without any difference in energy intake; long-term follow-up of the growth curve showed that MCK-DGAT1 mice weighed 5–10% less than wild-type mice on an at lib HFD feeding regimen, whereas food consumption was the same in both groups. The reduced body weight may contribute to the improved insulin sensitivity shown in this study. However, the small weight difference appears to result from changed fuel metabolism and is unlikely a primary cause of the observed protection against lipotoxicity. Marked differences in insulin sensitivity were seen between “weight-matched” wild-type and MCK-DGAT1 mice and between isolated DGAT1 muscle and wild-type muscle in ex vivo experiments, where weight was not a factor (3).

Insulin resistance is associated with mitochondrial dysfunction in type 2 diabetes (23,35,36) and in the offspring of parents with type 2 diabetes (37). Reduced mitochondrial function and insulin resistance occur in aging in association with reduced mitochondrial DNA copy number (23). However, the issue regarding mitochondrial deficiency and insulin resistance is unresolved (38). The present study does not support the hypothesis that reduction in mitochondrial DNA copy number per se results in insulin resistance. In our model, despite a reduction in mitochondrial number, mitochondrial ATP production is maintained (and heat production probably increased) because of increased fatty acid oxidation efficiency. These findings suggest that mitochondrial quality is more critical. In this regard, a recent study by Koves et al. demonstrated that stressed mitochondria with greater but incomplete fatty acid oxidation are only detrimental to insulin sensitivity (39). The present study further suggests that greater fatty acid oxidation capacity associated with increased uncoupling activity may also be important in maintaining insulin sensitivity. Of course, maintaining adequate oxidative phosphorylation is necessary for muscle contractility, particularly during exercise.

We have previously shown that skeletal muscle overexpressing DGAT1 behaved similarly to exercise-trained muscle in that it had higher triglyceride content but was more insulin sensitive, a phenomenon referred to as the “athlete paradox” or “exercise paradox.” The decreased myocytic DAG and ceramide levels with increased myocytic triglyceride content are the expected effects of DGAT1 actions (3). By converting fatty acyl-CoA and DAG to triglyceride, DGAT1 is believed to be directly responsible for lowering the levels of the lipotoxic fatty acid derivatives. The key finding of the present study is that DGAT1-overexpressing muscle, like exercised muscle, had increased fatty acid oxidation capacity, which was associated with increased steady-state metabolic rates at the whole body level and a smaller HFD-induced weight gain compared with wild-type mice. Importantly, because these findings are made in the muscle-specific DGAT1 transgenic mice, DGAT1 overexpression must be primarily responsible for these paradoxical changes. Because increased DGAT1 expression is also observed with exercise, we further postulate that DGAT1 may play a key mediatory role in exercise-induced augmentation of both triglyceride synthesis and, indirectly, mitochondrial fatty acid oxidation. Together, the increased anabolic (triglyceride synthesis/storage) and catabolic (fatty acid oxidation/degradation) capacities may both help remove fatty acid intermediates that can be overproduced during fatty acid overload. Reducing intramyocellular DAG levels to below the “threshold of lipotoxicity” is believed necessary to prevent the induction of a cascade of signaling events including activation of PKC isoforms and Jun NH₂-terminal kinase-1 and inhibition of IRS-1, Akt, and GLUT4 in HFD-fed mice (3).

Parallel upregulated triglyceride synthesis and fatty acid oxidation seemed perplexing at the first glance because anabolic and catabolic pathways usually undergo changes in the opposite directions, determined by cellular substrate levels and energy status. However, examination of typical myofibers does show interdependency between the two aspects of fatty acid metabolism: myofibers that have a high triglyceride content are typically the “oxidative” type characterized by a high mitochondrial capacity, whereas those with low triglyceride content (and high glycogen content) are typically “glycolytic.” Importantly, compared with glycolytic myofibers, oxidative myofibers with greater capacities for triglyceride synthesis and fatty acid oxidation appear to be intrinsically more insulin sensitive (4). The present study provided an example of the “paradoxical” coupling of lipid synthesis and oxidation in the DGAT1-overexpressing muscles.

Clearly, these are not the usual reciprocal regulations of the two aspects of the fatty acid metabolism driven by the fatty acid substrate availability. DGAT1 overexpression in skeletal muscle increased the capacities of both triglyceride synthesis and fatty acid oxidation and enabled the myocytes to endure fat overload without developing significant lipotoxicity or insulin resistance. However, instead of increasing mitochondrial densities, DGAT1-overexpressing muscle appeared to have greater mitochondrial fatty acid oxidation efficiency. Whether the improved fatty acid oxidation is coupled with increased fatty acid transport (as suggested by upregulation of CD36 and CPT-I) and/or potentially increased de novo fatty acid synthesis will require further investigation. A “futile cycle” of fatty acid metabolism that consists of increased fatty acid oxidation coupled with increased de novo fatty acid synthesis has been proposed as a thermogenic mechanism in muscle (40).

PDK is a key regulator of fuel partitioning between fatty acid and glucose oxidation, and PDK4 is the major isoform in muscle (41,42). PDK is acutely regulated, in general, by small molecular effectors generated from fatty acid β oxidation. Gene expression of PDK4 also responds to fuel substrate availability and insulin levels (43,,–46). Additionally, PDK4 gene expression appears to be modulated by several more complex biological processes, such as insulin resistance or exercise. Insulin resistance is generally associated with upregulation of PDK4 (43,45), which may be because of elevated fatty acid overload (47,48) or deficient/impaired insulin action (45,46,48). While promoting muscle insulin sensitivity, exercise also “paradoxically” upregulates PDK4 gene expression in skeletal muscle (49,50). Therefore, it appears that a different mechanism for upregulating PDK4 is involved in exercise-trained skeletal muscle and that this mechanism is not substrate driven (49). Exercise upregulates PDK4 and fatty acid oxidation but does not inhibit insulin-stimulated glucose oxidation. In the present study, we showed that DGAT1 overexpression produced a result similar to exercise with regard to PDK4 upregulation, increased fatty acid oxidation, and insulin sensitivity, suggesting that the change in PDK4 expression in this model is also not a secondary response to reduced insulin actions. It is currently unclear, in either the exercise model or the DGAT1-overexpression model, what mediates upregulation of PDK4 gene expression and whether increased mitochondrial fatty acid oxidation is directly caused by upregulated PDK4.

In summary, transgenic expression of DGAT1 in skeletal muscle results in a series of “paradoxical” metabolic changes often seen in the exercise-trained muscle, including increased triglyceride synthesis, increased fatty acid oxidation, and preserved insulin sensitivity during the challenging of fat overload. Although a high triglyceride content and PDK4 expression level may serve as markers of insulin resistance in sedentary and obese individuals, the same parameters may reflect a greater capacity for fat metabolism. In the latter situation, insulin sensitivity is preserved because of reduced levels of intracellular fatty acid derivatives. Because DGAT1 is induced in exercise, and because the MCK-DGAT1 model presents a set of metabolic features similar to the exercise model, we hypothesize that DGAT1 mediates a significant part of exercise-induced metabolic remodeling in skeletal muscle.

This study was supported by National Institutes of Health (NIH) Grants DK-60530, DK-071866, DK-063034, HL-73029, and HL45095. Contributions also came from New York Obesity Research Center Grant DK-026687 and Albert Einstein College of Medicine Diabetes Research and Training Center Grant NIH DK-020541.

No potential conflicts of interest relevant to this article were reported.