Metabolic flexibility to lipid availability during exercise is enhanced in individuals with high insulin sensitivity

Abstract

Metabolic flexibility to lipid (MetFlex-lip) is the capacity to adapt lipid oxidation to lipid availability. Hypothetically, impaired MetFlex-lip in skeletal muscle induces accumulation of lipid metabolites that interfere with insulin signaling. Our aim was to compare MetFlex-lip during exercise in subjects with low (Low_IS) vs. high (High_IS) insulin sensitivity. Twenty healthy men were designated as Low_IS or High_IS on the basis of the median of the homeostatic model assessment of insulin resistance index. Groups had similar age, body mass index, and maximum oxygen uptake (V̇o_2max). Subjects cycled at 50% V̇o_2max until expending 650 kcal. Adaptation in lipid oxidation was calculated as the drop in respiratory quotient (RQ) at the end of exercise vs. the maximum RQ (ΔRQ). Lipid availability was calculated as the increase in circulating nonesterified fatty acids (NEFA) at the end of exercise vs. the minimum NEFA (ΔNEFA). ΔRQ as a function of ΔNEFA was used to determine MetFlex-lip. On average, RQ and circulating NEFA changed similarly in both groups. However, ΔRQ correlated with ΔNEFA in High_IS (r = −0.83, P < 0.01) but not in Low_IS (r = −0.25, P = 0.48) subjects. Thus the slope of the ΔRQ vs. ΔNEFA relationship was steeper in High_IS vs. Low_IS subjects (−0.139 ± 0.03 vs. −0.025 ± 0.03 RQ·mmol⁻¹·l⁻¹, respectively; P < 0.05), with similar intercepts. We conclude that in subjects with High_IS lipid-to-carbohydrate oxidation ratio adapts to the increased circulating NEFA availability during exercise. Such MetFlex-lip appears impaired in subjects with Low_IS. Whether a cause-effect relationship exists between impaired MetFlex-lip and low insulin sensitivity remains to be determined.

INTRODUCTION

Metabolic flexibility (MetFlex) is the capacity to adapt fuel oxidation to fuel availability (12). This capacity is required to efficiently face energy challenges, such as the transition from rest to exercise (13). A mismatch between availability and oxidation of lipids might induce ectopic fat deposition and lipotoxicity (17, 24). Lipotoxicity is characterized by the accumulation of lipid metabolites (ceramides and diacylglycerol) that interfere with insulin signaling, thus impairing insulin action (26, 28). In this context, an enhanced capacity to adapt lipid oxidation to lipid availability, i.e., enhanced metabolic flexibility to lipid (MetFlex-lip), may prevent ectopic fat deposition and therefore protect from insulin resistance.

Curiously, most evidence relating MetFlex to insulin resistance has been obtained under conditions of high glucose supply, such as during euglycemic-hyperinsulinemic clamps (7, 10, 18, 20, 21, 27). In those studies, insulin-resistant states are clearly associated with impaired metabolic flexibility to glucose (MetFlex-glu). However, we claimed that MetFlex-glu was not properly examined because the adaptation in fuel oxidation to glucose did not consider the availability of glucose to the tissue, i.e., glucose uptake. Indeed, tissue glucose availability explained ~50% of the increase in glucose oxidation during euglycemic-hyperinsulinemic clamps (10, 27). Insulin-resistant vs. -sensitive individuals may thus have similar MetFlex-glu when tissue glucose availability is considered (10). Therefore, the relationship between insulin resistance and MetFlex, particularly MetFlex-glu, remains elusive.

In the context of impaired MetFlex as the driving factor for lipotoxicity, assessment of MetFlex-lip appears more relevant to fat deposition in tissues not meant to store fat, such as skeletal muscle, liver, or pancreas. If MetFlex-lip influences the development of insulin resistance, impaired MetFlex-lip should be apparent even at early stages of insulin resistance. A suitable physiological model to study MetFlex-lip is moderate-intensity endurance exercise. Circulating nonesterified fatty acid (NEFA) concentration increases during endurance exercise, and such increase is essential to boost muscle lipid oxidation (23, 30). This relationship between circulating NEFA and lipid oxidation is explained by the fact that circulating NEFA are the main determinant of muscle lipid uptake (5). Indeed, exogenous lipid infusions can boost lipid oxidation even during high-intensity exercise (22). Circulating NEFA concentration could thus be considered a valid marker of muscle lipid availability. Here we propose that the increase in lipid oxidation following the increase in lipid availability during exercise represents a true measure of MetFlex-lip.

MetFlex-lip during exercise has been considered previously as the level of exercise-induced lipid oxidation, without considering lipid availability (13). As such, controversial results have been reported in subjects with contrasting metabolic conditions (2, 6, 14, 16). On one hand, similar exercise-induced lipid oxidation has been observed in lean, obese/insulin-resistant, and type 2 diabetes (T2D) subjects (6) and in obese vs. obese/T2D subjects (2). On the other hand, lipid oxidation during exercise was higher in obese/insulin-resistant vs. lean subjects (14, 16). Hence, if exercise-induced lipid oxidation was a reliable and valid marker of MetFlex-lip, insulin resistance would be associated with a normal or even enhanced MetFlex-lip, which is counterintuitive. To be consistent with the definition of MetFlex, both lipid oxidation and lipid availability must be considered. Thus metabolically flexible subjects would be those who display a strong association between exercise-induced increases in lipid oxidation and in lipid availability. In contrast, impaired MetFlex-lip should be characterized by a dissociation between lipid oxidation and lipid availability.

Our aim was therefore to determine whether MetFlex-lip during exercise differs in healthy subjects with low vs. high insulin sensitivity. We hypothesized that MetFlex-lip would be higher in subjects with high insulin sensitivity compared with those with low insulin sensitivity.

METHODS

Subjects.

We used data obtained in 20 healthy men from a previous study (11). In the present analysis, we divided subjects according to the median of their homeostatic model assessment of insulin resistance (HOMA-IR) index (0.72), classifying them as having high (High_IS; low HOMA-IR) or low (Low_IS; high HOMA-IR) insulin sensitivity. Subjects were all normoglycemic nonsmokers and not engaged in competitive sports. The Institutional Review Board of the Pennington Biomedical Research Center approved the study, which was carried out in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all individual participants included in the study.

Study design.

A detailed description was published previously (11). In brief, body composition, maximum oxygen uptake (V̇o_2max), physical activity level (PAL), and energy requirements were determined at baseline in all subjects. One week later, we provided subjects with meals to be consumed on the 2 days before the endurance exercise challenge, to standardize preexercise diet. We used an isoenergetic low-carbohydrate (CHO; 25%), high-fat (67%) diet, with the intention of promoting exercise-induced lipid oxidation (4). Subjects were admitted to our in-patient unit the evening before the day of the endurance exercise protocol. The next morning, resting gas exchange was determined by indirect calorimetry and blood was drawn to determine circulating concentrations of selected metabolites and hormones. A biopsy was obtained from the vastus lateralis muscle at rest (preexercise) with a 5-mm Bergstrom needle. Then subjects performed the endurance exercise protocol consisting of cycling at 50% of V̇o_2max until expending 650 kcal. Gas exchange was measured at 8%, 20%, 40%, 60%, 80%, and 100% of exercise time. Values at 8% of exercise time were used to adjust workload to the expected 50% V̇o_2max if required. Blood samples were drawn at 20%, 40%, 60%, and 100% of exercise. Another biopsy was obtained from the vastus lateralis muscle immediately after exercise (postexercise).

Gas exchange, V̇o_2max, and PAL.

A Deltatrac II metabolic cart (Datex-Ohmeda, Helsinki, Finland) was used for the resting energy measurements (30 min) while subjects were awake in a supine position at a 22°C room temperature. During exercise a TrueOne 2400 metabolic cart (Parvo Medics, East Sandy, UT) was used, and each time point represented the average of 5-min period measurements. Substrate oxidation was calculated according to Frayn’s equations (9). V̇o_2max was determined by an incremental test on a cycloergometer (Lode Excalibur, Groing, The Netherlands) as described previously (11). Activity monitors (SenseWear Pro3 Armband, Body Media) were worn for 2–6 days to determine total energy expenditure (TEE, in kcal/day). Basal metabolic rate (BMR, in kcal/day) was calculated from resting gas exchange. PAL was computed as the ratio TEE/BMR.

Blood analyses.

Glucose, lactate, and NEFA were determined with Beckman Coulter DXC 600 kits (Brea, CA). Insulin was determined with the Siemens 200 kit (Siemens, Los Angeles, CA). Epinephrine and norepinephrine were determined by HPLC with electrochemical detection (Bio-Rad, Hercules, CA).

Calculation of MetFlex-lip during exercise.

Adaptation in lipid oxidation was estimated as the change in respiratory quotient (RQ = V̇co₂/V̇o₂) induced by exercise, i.e., the difference between RQ at the end of exercise vs. the maximum RQ during exercise (ΔRQ). Lipid availability was represented by the circulating NEFA concentration. Since NEFA concentration changes with exercise, the difference between NEFA at the end of exercise vs. the minimum NEFA (basal or during exercise) was used as an index of lipid availability (ΔNEFA). The equation of the line for the ΔRQ vs. ΔNEFA relationship including all subjects was calculated. Then the ΔNEFA value was included in the equation to determine the predicted ΔRQ. MetFlex-lip was represented by the ΔRQ residual (measured ΔRQ − predicted ΔRQ). Thus subjects with negative residuals (measured ΔRQ lower than predicted ΔRQ) were considered “metabolically flexible,” whereas subjects with positive residuals were considered “metabolically inflexible.”

Muscle palmitate oxidation.

Muscle was homogenized in a modified sucrose-EDTA buffer (in mmol/l: 250 sucrose, 1 EDTA, 1 Tris·HCl, and 2 ATP, pH 7.4). Palmitate oxidation ex vivo was determined by liquid scintillation counting of radioactive ¹⁴CO₂ (complete oxidation) and ¹⁴acid-soluble metabolites (incomplete oxidation) as previously described (11). Results were normalized to muscle wet weight.

Fiber typing and intramyocellular lipid.

Muscle was embedded in optimum cutting temperature compound and tragacanth gum powder and then frozen in cold isopentane. Muscle cross sections were then studied by immunohistochemical staining of slow-twitch muscle myosin (MAB1628, Merck Millipore, Darmstadt, Germany) for fiber type determination and BODIPY Green for intramyocellular lipid (IMCL) determination. The detailed procedures have been published previously (11).

Glycogen content.

Muscle was homogenized in a buffer (pH 6.1) containing 20 mmol/l KH₂PO₄, 10 μmol/l CaCl₂, and 1 mmol/l MgCl₂. Glycogen was estimated by spectrophotometry (620 nm) using an oyster glycogen standard curve as described previously (11). Results were normalized to protein content.

Western blotting.

Muscle was homogenized in RIPA buffer with protease/phosphatase inhibitor cocktails (Sigma, St. Louis, MO). The detailed protocol for Western blotting has been described previously (11). The mitochondrial oxidative phosphorylation (OXPHOS) complexes were detected with the antibody cocktail MS601 (MitoSciences, Eugene, OR, USA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was detected with antibody ab9484 (Abcam, Cambridge, MA) and used as loading control.

Real-time polymerase chain reaction (quantitative PCR).

Total RNA was extracted with the miRNEasy Mini Kit (Qiagen, Germantown, MD). The High Capacity cDNA Kit (Applied Biosystems, Foster City, CA) was then used for cDNA production. Gene expression was detected with TaqMan Gene Expression Assays-on-Demand in the 7900HT Fast Real-Time PCR system (Applied Biosystems). The expression level was determined against a standard curve.

Statistics.

Data are presented as means (SD). Prism 7.0c (GraphPad Software, La Jolla, CA) was used for the analyses. Two-way ANOVA with repeated measures was used to determine the effects of time (exercise time, or preexercise vs. postexercise), group (Low_IS vs. High_IS), or time × group interaction. In the case of significant time × group interaction, Bonferroni post hoc test was used. Unpaired Student’s t-test was used to compare single variables between groups. Variances were compared with the F-test. Correlations were analyzed by Pearson coefficients. With regression and covariance analyses (SAS 9.2, SAS Institute, Cary, NC), slopes (±SE) and intercepts of the ΔRQ vs. ΔNEFA relationship were compared between groups. P < 0.05 was considered statistically significant.

RESULTS

Low_IS and High_IS groups had similar age, body mass, body mass index (BMI), body fat, fat-free mass, V̇o_2max, PAL, fiber type composition (preexercise biopsy), and fasting glycemia (Table 1). By design, fasting insulinemia and HOMA-IR were different between groups (Table 1).

Table 1.

Characteristics of subjects

	Low_IS	High_IS
n	10	10
Age, yr	22.9 (SD 4.5)	25.1 (SD 4.4)
Body mass, kg	77.0 (SD 7.1)	76.4 (SD 6.2)
Body mass index, kg/m2	23.8 (SD 1.9)	23.4 (SD 1.7)
Body fat, %	16.6 (SD 2.7)	16.5 (SD 3.7)
FFM, kg	64.1 (SD 5.2)	63.6 (SD 4.4)
V̇o2max, ml O2·FFM−1·min−1	45.3 (SD 5.7)	49.0 (SD 5.3)
Physical activity level*	1.91 (SD 0.34)	1.81 (SD 0.22)
Type I fiber, %	36.9 (SD 14.8)	34.7 (SD 9.7)
Fasting glycemia, mg/ dl	88.4 (SD 5.1)	87.5 (SD 4.2)
Fasting insulinemia, mU/l	5.0 (SD 1.0)	2.2 (SD 0.32)†
HOMA-IR	1.09 (SD 0.26)	0.47 (SD 0.07)†

Values are means (SD) for n subjects. FFM, fat-free mass; V̇o_2max, maximum oxygen uptake; HOMA-IR, homeostatic model assessment of insulin resistance index.

^*One subject in each group did not have data; thus n = 9 per group.

^†P < 0.001, subjects with high (High_IS) vs. low (Low_IS) insulin sensitivity.

Similar metabolic response to exercise in Low_IS and High_IS subjects.

Exercise lasted 84.1 ± 14.3 and 79.1 ± 9.6 min in Low_IS and High_IS subjects, respectively (P = 0.37). Circulating concentrations of glucose (Fig. 1A), lactate (Fig. 1B), NEFA (Fig. 1C), epinephrine (Fig. 1E), and norepinephrine (Fig. 1F) responded similarly to exercise in Low_IS and High_IS subjects. As expected, insulinemia during exercise was higher in Low_IS vs. High_IS subjects, except at the end of exercise (Fig. 1D).

Fig. 1.

Response to exercise of circulating metabolites and hormones in subjects with low (Low_IS) and high (High_IS) insulin sensitivity: circulating levels of glucose (A), lactate (B), nonesterified fatty acids (NEFA; C), insulin (D), epinephrine (E), and norepinephrine (F). Insets: area under the curve (AUC). **P < 0.01, ***P < 0.001, main effect of time or time × group interaction (Inter); ^##P < 0.01, ^###P < 0.001, Low_IS vs. High_IS in post hoc test; ^$$$P < 0.001, Low_IS vs. High_IS. n = 10 per group.

Exercise increased whole body lipid oxidation rate and slightly decreased whole body CHO oxidation rate, without differences between groups (Fig. 2, A–C). The same result was found when lipid (P = 0.12) and CHO (P = 0.96) oxidation rates were divided by fat-free mass (data not shown). Total oxidation of lipid and CHO were similar in both groups (Fig. 2D). Exercise decreased muscle glycogen content similarly in Low_IS and High_IS subjects (Fig. 2E) but had no effect on IMCL levels (Fig. 2F). IMCL levels tended (P = 0.095) to be higher in Low_IS than High_IS subjects (Fig. 2F).

Fig. 2.

Fuel oxidation and intracellular substrate utilization during exercise in subjects with low (Low_IS) and high (High_IS) insulin sensitivity. A–C: response to exercise of respiratory quotient (RQ; A), lipid oxidation rate (B), and carbohydrate (CHO) oxidation rate (C). D: total lipid and CHO oxidation during exercise. E and F: skeletal muscle glycogen (E) and intramyocellular lipid (IMCL; F) levels before (pre) and after (post) exercise. **P < 0.01, ***P < 0.001, main effect of time or group. AU, arbitrary units. n = 10 per group.

MetFlex-lip in response to elevated NEFA is higher in subjects with High_IS vs. Low_IS.

Considering all subjects, ΔRQ correlated inversely with ΔNEFA (r = −0.48, P < 0.05; Fig. 3A). Thus the higher the exercise-induced increase in circulating NEFA concentration, the higher the increase in lipid oxidation relative to CHO (i.e., drop in RQ). The residuals of the ΔRQ vs. ΔNEFA relationship were similar in Low_IS and High_IS groups [0.0018 (SD 0.031) and −0.0018 (SD 0.025) RQ, respectively, P = 0.77; Fig. 3B]. However, visual inspection of ΔRQ residuals vs. ΔNEFA revealed that the residuals became more negative in High_IS than Low_IS at ΔNEFA concentrations > 0.45 mmol/l (Fig. 3C).

Fig. 3.

Relationship between lipid oxidation and lipid availability in subjects with low (Low_IS) and high (High_IS) insulin sensitivity. A: relationship between change in respiratory quotient (ΔRQ) and change in circulating nonesterified fatty acids (ΔNEFA) during exercise, including all subjects (n = 20). B: ΔRQ residuals from the ΔRQ vs. ΔNEFA relationship shown in A (n = 10 per group). C: ΔRQ residuals from the ΔRQ vs. ΔNEFA relationship shown in A as a function of ΔNEFA. D: relationship between ΔRQ and ΔNEFA during exercise for each group separately (n = 10 per group). E: sum of squares residuals for the ΔRQ vs. ΔNEFA relationships shown in D (n = 10 per group). F: ΔRQ/ΔNEFA. ***P < 0.001 for comparison of variance (n = 10 per group). G: ΔRQ/ΔNEFA vs. homeostatic model assessment of insulin resistance (HOMA-IR) index in subjects with ΔNEFA > 0.45 mmol/l (n = 7).

To further explore this ΔNEFA-dependent effect, we analyzed the ΔRQ vs. ΔNEFA relationship in each group separately. The inverse correlation between ΔRQ and ΔNEFA became stronger in High_IS (r = −0.83, P < 0.01) but disappeared in Low_IS (r = −0.25, P = 0.48) subjects. Consistently, the slope of the ΔRQ vs. ΔNEFA relationship was steeper in High_IS vs. Low_IS subjects (−0.139 ± 0.03 vs. −0.025 ± 0.03 RQ·mmol⁻¹·l⁻¹, respectively), while intercepts were similar (Fig. 3D). It is worth mentioning that ΔNEFA correlated strongly with NEFA concentration at the end of exercise (r = 0.92, P < 0.0001); thus subjects with the highest ΔNEFA had also the highest circulating NEFA concentrations at the end of exercise.

If lipid oxidation and lipid availability are dissociated in Low_IS compared with High_IS subjects, a larger variability in the ΔRQ vs. ΔNEFA relationship is expected in Low_IS. We therefore calculated two indexes of variability. First, we computed the residuals of the ΔRQ vs. ΔNEFA relationship for each group separately and then calculated the sum of squares residuals. The sum of squares residuals was 2.4-fold higher in Low_IS than in High_IS subjects (Fig. 3E). Second, the mean of the ΔRQ-to-ΔNEFA ratio was calculated for each group, and the variance of data was compared. The variance was higher in Low_IS than in High_IS subjects (0.049 vs. 0.004 RQ·mmol⁻¹·l⁻¹, respectively; Fig. 3F). These two indexes demonstrate a larger variability in the ΔRQ vs. ΔNEFA relationship in Low_IS than in High_IS subjects.

As an additional method to establish the association between insulin sensitivity and MetFlex-lip, we correlated the ΔRQ-to-ΔNEFA ratio with HOMA-IR. No association was found when all subjects were considered (r = −0.33, P = 0.15, n = 20). We then analyzed only subjects with ΔNEFA > 0.45 mmol/l, because differences in MetFlex-lip seem mainly manifested above this ΔNEFA level. In those subjects, ΔRQ/ΔNEFA correlated directly with HOMA-IR [r = 0.79, P < 0.05, n = 7 (4 Low_IS + 3 High_IS); Fig. 3G].

Expression of markers of lipid metabolism and oxidative capacity differs in muscle of Low_IS vs. High_IS subjects.

In both groups, complete palmitate oxidation ex vivo increased after exercise, whereas incomplete oxidation did not change (Fig. 4, A and B). The mRNA and protein levels of different markers were studied in skeletal muscle before exercise. Regarding mRNA content of muscle lipases, no differences were detected between groups (Fig. 4C). However, High_IS subjects expressed lipid transport-related genes at a higher extent than Low_IS subjects, specifically fatty acid binding protein-3 (Fabp3) and fatty acid transport protein-1 (Fatp1) (Fig. 4D). High_IS subjects also had higher mRNA levels of adiponectin receptor-1 (Adipor1) than Low_IS subjects, and a similar trend (P = 0.051) was observed for Adipor2 (Fig. 4E). The mRNA level of acyl-CoA dehydrogenase C-4 to C-12 straight chain (Mcad) was lower in High_IS than Low_IS subjects, and a similar trend (P = 0.055) was found for acetyl-CoA carboxylase beta (Acc2) (Fig. 4E). Protein levels of the OXPHOS complexes II and III were higher in High_IS than in Low_IS subjects, and complexes I and IV presented similar trends (P < 0.06) (Fig. 4F).

Fig. 4.

Markers of lipid metabolism and mitochondrial capacity in skeletal muscle of subjects with low (Low_IS) and high (High_IS) insulin sensitivity. A and B: complete (¹⁴CO₂; A) and incomplete [¹⁴acid-soluble metabolites (¹⁴ASM); B] palmitate oxidation ex vivo in skeletal muscle before (pre) and after (post) exercise (n = 9–10 per group). C–E: mRNA levels of genes related to lipid oxidation (n = 5–8; C), lipid transport (n = 3–10; D), and regulation of lipid metabolism (n = 3–10; E) in preexercise skeletal muscle. F: protein level of mitochondrial oxidative phosphorylation (OXPHOS) complexes in preexercise skeletal muscle (n = 10 per group). *P < 0.05, main effect of time; ^$P < 0.05, ^$$P < 0.01, ^$$$P < 0.001, Low_IS vs. High_IS. AU, arbitrary units.

DISCUSSION

The concept of metabolic flexibility (MetFlex) has gained attention as a physiological trait potentially influencing the deposition of ectopic fat and the development of insulin resistance (13, 17). Here we showed that at high NEFA increments MetFlex-lip during exercise was impaired in subjects with low insulin sensitivity. This finding agrees with the idea that impaired MetFlex-lip may promote the development of insulin resistance in skeletal muscle.

MetFlex has been previously defined and calculated by different methods. Originally, Kelley et al. used the increase in RQ during an euglycemic-hyperinsulinemic clamp as an indicator of MetFlex-glu (18). Bergouignan et al. considered as indicative of MetFlex-glu a large RQ variance, along with a small variance in insulinemia, in response to standard meals (1). Recently, San-Millán and Brooks considered MetFlex-lip as the capacity to upregulate lipid oxidation during an exercise of incremental intensity (25). All these approaches were aimed at studying the capacity to modify fuel oxidation in response to metabolic challenges (clamp, meal, exercise), but none of them truly considered fuel availability, a key component in the MetFlex concept. Such an oversight makes uncertain the conclusions regarding the putative influence of MetFlex on metabolic homeostasis. Here we used as a model endurance exercise, a metabolic challenge that increases lipid availability with concomitant increases in lipid oxidation. To determine MetFlex-lip during exercise, we assessed the relationship between the change in RQ (ΔRQ; lipid oxidation) and the change in circulating NEFA (ΔNEFA; lipid availability). Such a method more precisely fits the definition of MetFlex, i.e., the capacity to adapt fuel oxidation to fuel availability.

Although the average responses of lipid oxidation (RQ) and availability (NEFA) during exercise were similar in Low_IS and High_IS subjects, oxidation and availability seemed dissociated from each other in Low_IS subjects. This dissociation manifested as 1) a lack of correlation between the variables and 2) a larger variability in the ΔRQ vs. ΔNEFA relationship. Therefore, the change in NEFA led to less predictable changes in RQ in the Low_IS group. The ΔRQ vs. ΔNEFA relationship also showed a steeper slope in High_IS subjects than in Low_IS subjects. This last observation suggests that at small ΔNEFA during exercise (close to the intercept), both groups switched to lipid oxidation to a similar extent (although with larger variability in Low_IS). However, at larger ΔNEFA, the switch to lipid oxidation was more pronounced in High_IS subjects compared with Low_IS. Subjects with high insulin sensitivity would thus have enhanced lipid oxidation in response to elevated lipid availability, i.e. enhanced MetFlex-lip. This idea is supported by the relationship between ΔRQ/ΔNEFA and HOMA-IR at ΔNEFA > 0.45 mmol/l. However, since few subjects had such elevated lipid availability, future studies with larger sample sizes should boost lipid availability as much as possible to confirm our results. Notably, a mismatch between lipid oxidation and lipid availability as observed in Low_IS subjects has been proposed to induce lipotoxicity and insulin resistance (17, 24). Accordingly, IMCL content tended to be elevated in Low_IS subjects, eventually as a result of impaired MetFlex-lip. Our data therefore point to a role of impaired MetFlex-lip in early stages of insulin resistance development.

The present evidence prompts the idea of assessing lipid oxidation during exercise at controlled NEFA concentrations, manipulated through triglyceride/heparin infusions. This would allow us to assess MetFlex-lip individually and to relate those responses to insulin sensitivity. We speculate that insulin-sensitive subjects would show higher MetFlex-lip than insulin-resistant counterparts only at high NEFA concentrations. Another possibility to compare MetFlex-lip in conditions of high NEFA concentration is by extending the exercise time. Indeed, total exercise time correlated directly with NEFA concentration at the end of exercise in our data (r = 0.56, P < 0.05, n = 20). Additional protocols could also be conducted to determine MetFlex-lip to other metabolic challenges that induce high lipid availability, e.g., prolonged fasting. These experimental designs would constitute a proof of concept for our findings.

The subjects in the present study had a narrow range of HOMA-IR index (0.39–1.65). The advantage of this sample homogeneity is that the groups were matched for almost every parameter. Given the strong association between BMI and insulin sensitivity (8), groups with large differences in insulin sensitivity are generally difficult to match for BMI and fat mass, complicating data interpretation. The disadvantage of the sample homogeneity is that differences in MetFlex-lip might be undetectable, which is aggravated by our small sample size. Nevertheless, we did find that insulin sensitivity level discriminated different patterns of MetFlex-lip during exercise.

MetFlex-lip has been analyzed previously during exposure to high-fat meals (>60% of total kcal). In that approach, subjects with T2D had higher increases in circulating NEFA levels, but lower lipid oxidation, than control subjects (19). Similarly, healthy subjects with a family history of T2D showed normal increases in circulating NEFA but reduced lipid oxidation (15). Therefore, previous studies demonstrate that insulin-resistant states have impaired lipid oxidation for a certain circulating NEFA level, i.e., impaired MetFlex-lip. Notably, high-fat meals increase lipid oxidation in several tissues (liver, skeletal muscle, and others), but endurance exercise augments lipid oxidation mostly in skeletal muscle. Our findings thus suggest that skeletal muscle is directly implicated in the impaired MetFlex-lip characterizing insulin-resistant states. Accordingly, MetFlex-lip of human myotubes in vitro correlates directly with the donor’s insulin sensitivity in vivo (29).

The mechanisms responsible for maintaining a tight relationship between lipid oxidation and lipid availability in subjects with High_IS are unknown. To gain some insight, we measured the expression of different markers in skeletal muscle. Low_IS and High_IS subjects had similar palmitate oxidation ex vivo, an assay that depends on the beta-oxidation (β-Ox) and the tricarboxylic acid (TCA) cycle. These processes would thus not explain the difference in lipid oxidation in vivo between Low_IS and High_IS subjects. In contrast, we did find differences between our groups in some steps downstream of TCA and upstream of β-Ox. Specifically, Low_IS subjects had decreased expression of OXPHOS complexes and of genes that promote lipid oxidation (Adipor1, Adipor2), and had increased expression of genes that inhibit lipid oxidation (Acc2). The expression of genes related to lipid transport (Fabp3, Fatp1) was also reduced in Low_IS subjects, suggesting that lipid uptake and/or intracellular transport might determine MetFlex-lip during exercise. Given the limited sample size in our quantitative PCR studies, further analyses, including measurements of protein levels, are required to confirm our results. Studies comprising assays for NEFA uptake and transport, along with activities of key enzymes involved in lipid metabolism, would be useful. Indeed, cellular transport is a major determinant of MetFlex-glu during euglycemic-hyperinsulinemic clamps, because glucose disposal rate explains 50% of the increase in RQ (10, 27). Here we used circulating NEFA concentration as an index of lipid availability, because circulating NEFA concentration has been shown to be the main determinant of muscle lipid uptake at rest (5). However, circulating NEFA may not be the best index of lipid availability during exercise, because muscle contraction induces translocation of NEFA transporters to the sarcolemma, thus intensifying NEFA uptake (3). A more accurate index of cellular lipid availability during exercise is thus muscle NEFA uptake, and this should be considered in future studies. Whether NEFA uptake and/or intracellular transport explain the difference we observed in MetFlex-lip is unknown.

In conclusion, MetFlex assessment must include measurements of the adaptive fuel oxidative capacity and fuel availability. Using a new approach, we showed that MetFlex to elevated lipid availability during endurance exercise is enhanced in subjects with high insulin sensitivity. Whether impaired MetFlex-lip causes, or results from, low insulin sensitivity remains to be determined.

GRANTS

This work was supported by The Obesity Society (Young Scientist Award 2008 to J. E. Galgani), a Clinical Nutrition Research Unit Grant (P30-DK-072476 to E. Ravussin), the National Institute of Diabetes and Digestive and Kidney Diseases (Grant R01-DK-060412 to E. Ravussin), and Fondecyt (no. 1170117 to J. E. Galgani).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.R. and J.E.G. conceived and designed research; S.B. and J.E.G. performed experiments; R.F.-V., S.B., and J.E.G. analyzed data; R.F.-V. and J.E.G. interpreted results of experiments; R.F.-V. prepared figures; R.F.-V. and J.E.G. drafted manuscript; R.F.-V., S.B., E.R., and J.E.G. edited and revised manuscript; R.F.-V., S.B., E.R., and J.E.G. approved final version of manuscript.