Early Mitochondrial Adaptations in Skeletal Muscle to Diet-Induced Obesity Are Strain Dependent and Determine Oxidative Stress and Energy Expenditure But Not Insulin Sensitivity

Abstract

This study sought to elucidate the relationship between skeletal muscle mitochondrial dysfunction, oxidative stress, and insulin resistance in two mouse models with differential susceptibility to diet-induced obesity. We examined the time course of mitochondrial dysfunction and insulin resistance in obesity-prone C57B and obesity-resistant FVB mouse strains in response to high-fat feeding. After 5 wk, impaired insulin-mediated glucose uptake in skeletal muscle developed in both strains in the absence of any impairment in proximal insulin signaling. Impaired mitochondrial oxidative capacity preceded the development of insulin resistant glucose uptake in C57B mice in concert with increased oxidative stress in skeletal muscle. By contrast, mitochondrial uncoupling in FVB mice, which prevented oxidative stress and increased energy expenditure, did not prevent insulin resistant glucose uptake in skeletal muscle. Preventing oxidative stress in C57B mice treated systemically with an antioxidant normalized skeletal muscle mitochondrial function but failed to normalize glucose tolerance and insulin sensitivity. Furthermore, high fat-fed uncoupling protein 3 knockout mice developed increased oxidative stress that did not worsen glucose tolerance. In the evolution of diet-induced obesity and insulin resistance, initial but divergent strain-dependent mitochondrial adaptations modulate oxidative stress and energy expenditure without influencing the onset of impaired insulin-mediated glucose uptake.

The mechanisms responsible for the association between mitochondrial dysfunction and skeletal muscle insulin resistance are incompletely understood (1). Putative mechanisms include increased oxidative stress, which may impair glucose transport, and diminished mitochondrial fatty acid (FA) oxidation, which may lead to accumulation of toxic lipid metabolites that impair insulin signaling (2, 3). Conversely, increased rates of incomplete β-oxidation and acylcarnitine accumulation have been associated with insulin resistance, which can be ameliorated by reducing FA oxidation using malonyl coenzyme A decarboxylase inhibition (4). Moreover, increased FA oxidation in mice that lack acetyl-coenzyme A carboxylase 2 fails to prevent high-fat diet (HFD)-induced insulin resistance (5). These observations challenge the idea that reduced mitochondrial β-oxidation is responsible for the development of skeletal muscle insulin resistance. A HF, high-sucrose diet suppresses mitochondrial oxidative capacity and biogenesis via increased reactive oxygen species (ROS) production in C57B6 mice (6). Long- and short-term HF feeding increased mitochondrial hydrogen peroxide emission in rats, and preventing ROS generation reversed insulin resistance (7). Furthermore, transgenic overexpression of catalase in mitochondria prevented age-associated insulin resistance in muscle, thereby implicating oxidative stress in its pathogenesis (8).

Although these studies suggest potential mechanisms by which mitochondrial dysfunction occurs and contributes to the pathogenesis or maintenance of insulin resistance, it remains unclear whether these mechanisms are generalizable. In addition, the precise temporal relationship between altered mitochondrial function and onset of skeletal muscle insulin resistance remains uncertain. Furthermore, the extent to which these associations might be confounded by the impact of mitochondrial dysfunction on energy expenditure and weight gain remains to be clarified. For these reasons, we compared mitochondrial function, oxidative stress, adiposity, energy expenditure, and insulin sensitivity in obesity-prone C57BL/6J (C57B) and obesity-resistant FVB/NJ (FVB) mice as a function of duration of HF feeding. These two commonly used inbred mouse strains are genetically quite distant (9), and they have a different metabolic profile. Thus, compared with FVB mice, C57B mice have low circulating triglyceride (TG) levels and increased TG clearance (10, 11). On the other hand, lean FVB mice have higher hepatic insulin resistance and reduced glucose-stimulated insulin secretion (12) and are spontaneously hyperactive (13). Finally, FVB mice have less recruitment of small adipose cells in response to HF feeding when compared with C57B mice (14).

Here, we show that C57B mice exhibited reduced mitochondrial oxidative capacity and increased oxidative stress that preceded impaired insulin-mediated glucose uptake. In contrast, FVB mice developed skeletal muscle mitochondrial uncoupling, which prevented oxidative stress, increased energy expenditure, and potentially limited weight gain but did not prevent insulin resistance in response to HF feeding. We also observed that although reducing oxidative stress in C57B mice enhanced mitochondrial function, it failed to prevent impaired insulin-mediated glucose uptake. Thus, the mitochondrial adaptations to diet-induced obesity and the onset of impaired skeletal muscle glucose uptake early in the course of diet-induced obesity represent parallel and distinct processes.

Research Design and Methods

Animals and diets

The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the United States National Institutes of Health (publication no. 85-23, revised 1996) and was approved by the Institutional Animal Care and Use Committee of the University of Utah. Male C57B or FVB mice (The Jackson Laboratory, Bar Harbor, ME) were fed a HFD or normal chow (NC) for 2, 5, or 10 wk, respectively, starting at 10 wk of age (see Supplemental Tables 1–3, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). The detailed composition and the source of the NC and the HFD is provided in Supplemental Table 3. Male uncoupling protein (UCP)3 knockout (KO) mice, kind gift from Bradford B. Lowell (Harvard Medical School, Beth Israel Deaconess Medical Center, Boston, MA), were backcrossed to C57B background for at least ten generations before they were used. Male UCP3KO and their littermate controls were fed NC or HFD for 10 wk.

Glucose tolerance test (GTT) and insulin tolerance test (ITT), serum metabolites, and hormone levels

GTT were performed after a 6-h fast as described (15). ITT were performed on random-fed animals by injecting insulin (0.75 U/kg body weight), and blood was collected from the tail vein 0, 15, 30, and 60 min after insulin administration. Blood glucose was determined using a glucometer (Glucometer Elite; Bayer, Tarrytown, NY). Insulin concentrations were determined using the sensitive rat insulin RIA kit (Linco Research, Inc., St. Charles, MO). Free FA concentrations were determined using the 1/2-micro fatty acid test kit (Roche Diagnostics, Mannheim, Germany), and TG concentrations were determined using the L-type TG H kit (Wako, Richmond, VA).

Hyperinsulinemic-euglycemic clamp studies

Studies were performed in weight-matched nonsedated mice as described (16). Animals recovered for 48 h after jugular vein catheterization, before undergoing the clamp. Insulin was infused at a constant flow rate with no priming dose (pump 33; Harvard Apparatus, Holliston, MA), and 50% dextrose was infused at a variable rate to maintain a target glucose of 100–150 mg/dl for 60 min. [³H]glucose was infused at a constant rate (0.1 μCi/min). Glucose was measured with a glucometer in 10-min intervals using 3 μl of tail vein blood samples obtained via a tail clip.

In vivo glucose uptake

In vivo glucose uptake in skeletal muscle was estimated using 2-deoxyglucose as described previously (17).

Indirect calorimetry

Mice were studied for three consecutive days in a four-chamber Oxymax system (Comprehensive Lab Animal Monitoring System; Columbus Instruments, Columbus, OH) as previously described (18).

Body composition

Mice were anesthetized (single ip injection of 400 mg of chloral hydrate/kg body weight). Fat mass was determined using Dual Energy X-Ray Absorptiometry (Norland Medical Systems, Fort Atkinson, WI).

Mitochondrial function and ATP synthesis

Mitochondrial function was studied using the saponin-permeabilized fiber technique with succinate (5 mm) and rotenone (10 μm) as previously described (19). Mitochondrial ATP measurements were performed by bioluminescence assay.

Tissue TG and diacylglycerol (DAG) determination

Total TG content of hindlimb muscle was determined using a TG reagent kit (Sigma-Aldrich, St. Louis, MO) using glycerol as a standard. Briefly, frozen hindlimb muscle from C57B and FVB mice fed either NC or HF for 5 wk was homogenized in chloroform/methanol (8:1 vol/vol) and shaken for 3–4 h at room temperature. Lipid phases were separated by adding 1 m H₂SO₄. The organic layer containing TG and phospholipids was then aliquoted and dried. The absorbance was read at 562, and TG content was normalized to dry tissue weight. DAG was extracted and quantified using a diglyceride kinase reaction-based method (20). Quantification was performed using a Storm phosphoimager (GE Healthcare life Sciences, Piscataway, NJ) with purified DAG as reference standards (Avanti Polar Lipids, Alabaster, AL).

Western blot analysis

Total proteins were extracted from frozen hindlimb muscle as described (19) and protein concentration measured using Micro BCA reagents (Pierce, Rockford, IL). For insulin signaling, mice received 3 U of insulin, which was injected via the inferior vena cava or an equivalent volume of saline before the hindlimb muscle was removed, dissected, and frozen. Proteins were resolved by SDS-PAGE and electrotransferred onto a polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA). For some blots, the following primary antibodies and protocols were used: rabbit anti-UCP3 (Affinity Bioreagents, Golden, CO), goat anti-UCP2 (Chemicon International, Temecula, CA), mouse antimanganese superoxide dismutase (MnSOD) (BD Biosciences, San Jose, CA), rabbit anti-4-hydroxynonenal (HNE)-Michael adducts (EMD Chemicals, Inc., Gibbstown, NJ), and mouse antiprotein-glutathione (glutathionylated proteins, PSSG) adducts antibody (ViroGen Corp., Watertown, MA). Mouse anti-α tubulin (Sigma, St. Louis, MO) or Coomassie blue R-250 (Bio-Rad, Hercules, CA) staining were used as loading controls. Protein detection was carried out with the appropriate horseradish peroxidase-conjugated secondary antibody and Enhanced Chemiluminescence or Enhanced Chemiluminescence Plus detection systems (Amersham Biosciences, Piscataway, NJ). For other blots, mouse antiserine-threonine kinase (Akt) 1, phospho-Akt Ser473, phospho-Akt Thr308, rabbit antiphospho-stress-activated protein kinase/Jun amino-terminal kinase (Thr183/Tyr185) (Cell Signaling Technology, Inc., Danvers, MA), rabbit antilipoprotein lipase (LPL), mouse anti-CCAAT/enhancer-binding protein-homologous protein and rabbit anti-X-box binding protein-1 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) were used. Alexa Fluor antirabbit 680 (Invitrogen, Carlsbad, CA) and antimouse 800 (VWR International, West Chester, PA) were used as secondary antibodies and fluorescence quantified using the LI-COR Odyssey imager (Lincoln, NE).

Oxidative stress parameters

ROS levels were measured in hindlimb homogenates as described previously (21). Reduced and oxidized glutathione content in hindlimb homogenates from C57B and FVB mice fed NC or HFD for 5 wk was determined using a spectrophotometric assay kit (EMD Chemicals, Inc.).

Mn tetrakis (4-benzoic acid) porphyrin (TBAP) treatment

C57B mice at 10 wk of age were started on HFD for 5 wk and treated at the same time with the cell-permeable SOD mimetic and peroxynitrite scavenger MnTBAP, 20 mg/kg body weight, or saline three times per week (EMD Chemicals, Inc.) as previously described (22).

Reverse transcription-polymerase chain reaction

Total RNA was extracted from hindlimb with TRIzol reagent and purified with the RNeasy kit (QIAGEN, Valencia, CA). Equal amounts of skeletal muscle RNA from six mice were subjected to real-time PCR using an ABI Prism 7900HT instrument in 384-well plate format with SYBR Green I chemistry and 6-carboxyl-X-rhodamine internal reference. Primer sequences are provided in the Supplemental Table 2.

Statistical analysis

Data are expressed as means ± se. All statistics were performed by one-way ANOVA using JMP software (SAS Institute, Cary, NC). Statistical significance was accepted when the P value was less than 0.05.

Results

FVB mice are resistant to diet-induced obesity

Weight gain was monitored weekly for 10 wk in C57B and FVB mice fed either NC or HFD. Although C57B on HFD gained twice as much weight relative to NC mice, FVB mice on HFD and NC gained weight equivalently (Supplemental Fig. 1A). It is important to note that FVB mice both on NC or HF started with lower body weights, but by 2 wk, there were no differences between the groups (Fig. 1A). HFD altered body composition in C57B mice so that fat mass increased by 70% after 5 wk of HFD (Fig. 1B). Interestingly, genetic background influenced baseline body composition so that fat mass was significantly higher in FVB mice on NC compared with C57B mice on NC (Fig. 1B). However, when FVB mice were fed a HFD, fat mass did not significantly change relative to that of NC controls.

Fig. 1.

Resistance to diet-induced obesity and increased energy expenditure in FVB vs. C57B mice. Body weights (A), fat mass/body weight (B), whole-body VO₂ (C), whole-body CO₂ release (D), and heat production (E) in C57B and FVB mice on NC or HF. In C, D, and E, data were collected over 72 h. In B, C, D, and E, animals were fed NC or HFD for 5 wk. Data are mean ± sem *, P < 0.05; **, P < 0.005 vs. NC-fed mice of the same strain; #, P < 0.05; ##, P < 0.005 vs. C57B mice on HF; $$, P < 0.005 vs. C57B mice on NC.

Increased metabolic rate and energy expenditure in HF-fed FVB mice

Food intake measured over 72 h was not different between C57B and FVB mice fed HFD for 5 wk, but FVB mice were less active at night than C57B mice irrespective of the diet (Supplemental Fig. 1, B and D), indicating that resistance to HF-induced weight gain in FVB could not be accounted for by decreased food intake or increased activity. Oxygen consumption (VO₂), averaged over 72 h, was significantly (P < 0.005) increased in FVB mice both at day and night after 5 wk of HFD (Fig. 1C). In parallel, CO₂ production (VCO₂) was significantly increased at 5 wk of HFD in FVB mice (Fig. 1D). Opposite changes were observed in C57B mice. Thus, although VO₂ was not altered, VCO₂ was significantly (P < 0.005) reduced (Fig. 1D). Thus, HFD increased metabolic rate in FVB mice but decreased it in C57B mice.

Increased VO₂ with unchanged VCO₂ in FVB mice on HFD suggested that VO₂ is uncoupled from VCO₂. To further examine this possibility, we measured heat production, which is derived from these two parameters. Heat generation was significantly increased in FVB mice after 5 wk of HFD (Fig. 1E) but was unchanged in C57B mice on HFD.

In mice, brown adipose tissue (BAT) and skeletal muscle are important contributors to peripheral energy expenditure. The thermogenic activity of BAT is mediated by UCP1. BAT weight was significantly increased in FVB mice on NC compared with C57B mice on NC underscoring genetic differences between strains. However, after 5 wk of HFD, the increase in BAT weight was similar between the two strains (Supplemental data). UCP1 mRNA in BAT was almost identical in C57B and FVB mice fed HFD (Supplemental data). Paradoxically, UCP1 protein was induced by HFD only in C57B mice but not in FVB mice (Supplemental data), thus arguing against increased UCP1-mediated uncoupling in BAT as the basis for changes in energy expenditure in HF-fed FVB mice.

Increased skeletal muscle mitochondrial uncoupling in FVB mice

We next tested the hypothesis that increased whole-body energy expenditure could be the consequence of mitochondrial adaptations in skeletal muscle by examining mitochondrial VO₂, ATP synthesis rates, and ATP/O ratios in saponin-permeabilized soleus muscle fibers. HFD significantly impaired state 3 respiration and ATP synthesis at 2 and 5 wk in C57B mice (Fig. 2, A and C). Because the decline in maximal respiration was proportional to the decline in ATP synthesis, ATP/O ratios were unchanged (Fig. 2D). Interestingly, after 10 wk of HFD, although the respiratory capacity declined in C57B mice on NC, it did not fall further in C57B mice on HF. In contrast, mitochondrial uncoupling was increased in soleus muscle of FVB mice on HFD that persisted throughout the 10-wk study. Although maximal VO₂ did not differ between NC-fed and HF-fed FVB mice, ATP synthesis was significantly (P < 0.05) reduced in the HF-fed group at 2, 5, and 10 wk (Fig. 2, E and G), leading to significantly reduced ATP/O ratios (P < 0.05) (Fig. 2H). Increased respirations in the presence of oligomycin that was present after 2 wk of HFD provide additional evidence for increased mitochondrial uncoupling in HF-fed FVB mice (Fig. 2F).

Fig. 2.

Mitochondrial respiration and ATP synthesis in skeletal muscle of C57B and FVB mice fed NC or HF for 2, 5, or 10 wk. State 3 (A and E), state 4 (B and F), ATP synthesis rates (C and G), and ATP/O ratios (D and H) in saponin-permeabilized soleus fibers from C57B and FVB mice, respectively (n = 5–8 per group and per strain). Data are mean ± sem. a, P < 0.05 vs. NC at any time point; b, P < 0.05 comparing NC 2 or 5 wk with NC at 10 wk; c, P < 0.05 comparing HF 2 or 5 wk with HF at 10 wk; d, P < 0.05 comparing NC at 5 or 10 wk with NC at 2 wk; e, P < 0.05 comparing HF at 5 or 10 wk with HF at 2 wk.

To explore the mechanisms for mitochondrial uncoupling in HF-fed FVB mice, we measured mRNA and protein levels of UCP2 and UCP3. UCP3 gene expression was not influenced by HFD in either strain, whereas UCP2 expression was significantly induced (P < 0.05) in FVB mice after 5 wk of HFD (Table 1). In contrast, we observed a 50–60% induction (P < 0.005) of UCP3 protein in both C57B and FVB mice after 2 wk of HFD (Supplemental Fig. 2A). However, the induction in UCP3 protein persisted only in HFD FVB mice (Supplemental Fig. 2B). We also tested the possibility that BAT interspersed in skeletal muscle could account for increased mitochondrial uncoupling in HFD FVB. As shown in Supplemental Fig. 3E, there was no evidence for UCP1 expression in skeletal muscle of HFD FVB or C57B mice, excluding any contribution of ectopic BAT.

Table 1.

Skeletal muscle expression of mitochondrial biogenesis, FA oxidation, and uncoupling pathway genes

	C57B mice		FVB mice
	NC	HF	NC	HF
PPARα	1.23 ± 0.26	1.25 ± 0.15	0.57 ± 0.17	0.79 ± 0.2
PPARγ	0.96 ± 0.11	0.83 ± 0.06	1.13 ± 0.2	1.13 ± 0.2
PGC1α	1.54 ± 0.34	1.08 ± 0.23	0.56 ± 0.12	0.6 ± 0.07
PGC1β	1.24 ± 0.16	1.25 ± 0.13	0.71 ± 0.21	0.73 ± 0.1
TFAM	1.15 ± 0.09	1.25 ± 0.09	0.71 ± 0.08	0.69 ± 0.04
NRF1	1.04 ± 0.13	1.13 ± 0.1	1.15 ± 0.17	1.05 ± 0.12
NRF2	0.9 ± 0.1	0.98 ± 0.12	0.91 ± 0.15	0.9 ± 0.14
UCP2	1.13 ± 0.13	1.24 ± 0.1	0.84 ± 0.1	1.36 ± 0.12a
UCP3	1.3 ± 0.24	1.38 ± 0.16	0.39 ± 0.08	0.37 ± 0.06

mRNA was extracted from hindlimb muscle of mice fed NC or HFD for 5 wk starting at 10 wk of age. mRNA expression was normalized to cyclophilin mRNA expression, which was not significantly different among the groups. Data are expressed as mean ± sem of six mice per group. PPAR, Peroxisome proliferator activated receptor; PGC, PPAr gamma coactivator; NRF, respiratory nuclear factor.

^aP < 0.05 vs. NC in the same strain.

FVB but not C57B mice are resistant to HFD-induced oxidative stress

To determine whether ROS-mediated activation of UCP (23, 24) contributed to, or was associated with, mitochondrial uncoupling in HF-fed FVB mice, we determined parameters of oxidative stress in both strains. ROS production in hindlimb muscle, measured using the ROS-sensitive probe dichlorofluorescin diacetate (DCFDA), was elevated only in C57B mice fed HFD for 5 wk (Fig. 3A). Furthermore, 5 wk of HFD shifted the intracellular redox environment to a more oxidized state in skeletal muscle of C57B mice as evidenced by lower glutathione/glutathione disulfide ratios (P < 0.005) (Fig. 3B). In addition, lipid and protein peroxidation, as detected by an antibody that recognizes, 4-HNE protein adducts, was significantly elevated (P < 0.05) in HF-fed C57B mice compared with their respective NC animals (Fig. 3, C and D). In contrast, HF feeding did not induce lipid and protein peroxidation in the FVB strain, and if anything, these modifications were reduced by 30% (P = 0.07) relative to chow-fed FVB mice. Importantly, FVB mice on HF had significantly less lipid and protein peroxidation (P < 0.005) relative to C57B on HF (Fig. 3, C and D). These data suggest that activation of mitochondrial uncoupling effectively reduced oxidative stress in FVB mice on a HFD.

Fig. 3.

FVB mice are protected against HFD-induced ROS production in skeletal muscle. DCFDA fluorescence (A), reduced/oxidized glutathione (B), 4-HNE protein adduct blot (C), and quantification of 4-HNE protein normalized to α-tubulin (D) in skeletal muscle of C57B and FVB mice fed NC (n = 4–6 for each strain) or HF (n = 6 for each strain) for 5 wk. Data are mean ± sem. *, P < 0.05; **, P < 0.005 vs. NC-fed mice of the same strain; ##, P < 0.005 vs. C57B mice on HF; $, P < 0.05 vs. C57B mice on NC. GSH/GSSG, Glutathione/Glutathione disulfide; AU, arbitrary units.

To test the hypothesis that increased antioxidant capacity decreased ROS in FVB mice, we MnSOD protein in skeletal muscle of C57B and FVB mice after 2 and 5 wk of HF feeding. MnSOD protein was slightly but significantly increased in FVB mice after 2 wk of HFD. However, by 5 wk of HF, MnSOD protein content increased similarly in both strains (Supplemental Fig. 4, A and B).

It was recently suggested that glutathionylation of UCP2 and UCP3 may increase their activation (25). To explore whether this process is differentially regulated in muscle samples from C57B and FVB, we measured total protein glutathionylation in hindlimb homogenates using an anti-PSSG antibody. As shown in Supplemental Fig. 5, anti-PSSG antibody detected less glutathionylation of proteins in samples extracted from C57B mice fed HFD for 5 wk, whereas more glutathionylated proteins were detected in FVB on HF.

Increased im TG, DAG levels, and LPL protein content in FVB mice on HFD

Another possibility for increased mitochondrial uncoupling is enhanced FA flux through the mitochondrial electron transport chain. Oxidation of FA is less efficient, generates more ROS, and increases mitochondrial uncoupling relative to other substrates (24, 26). To test the possibility that FVB mice increased FA use, we measured skeletal muscle TG and DAG and observed a significant (P < 0.005) increase in TG levels in the muscle of FVB under NC condition both at 2 and 5 wk (Fig. 4, A and B). Interestingly, an increase in muscle TG levels was observed only in FVB mice after HFD (Fig. 4, A and B), suggesting either increased FA flux, reduced oxidation, or enhanced lipid partitioning. In accordance with increase FA flux, we observed an approximately 2.5-fold increase in LPL protein content in the skeletal muscle of FVB mice on HFD (Fig. 4, D and E). In addition, FVB mice had higher levels of muscle DAG that was more pronounced when the mice were fed HFD (Fig. 4C).

Fig. 4.

Increased lipid influx and intracellular lipid metabolites in FVB mice on HFD. Skeletal muscle TG content at 2 wk (A) and 5 wk of HFD (B) and skeletal muscle DAG levels at 5 wk of HFD (C) in C57B and FVB mice (n = 5–6 mice per strain and per diet). Representative Western blotting of LPL expression and Coomassie blue (C.B.) staining (D) and the corresponding densitometry of LPL/C.B. (E) in skeletal muscle homogenates from C57B and FVB fed NC or HFD for 5 wk (n = 4 mice per strain and per diet). Data are mean ± sem **, P < 0.005 vs. NC-fed mice of the same strain; #, P < 0.05; ##, P < 0.005 vs. C57B mice on HF.

HFD induces insulin resistance in both strains

Given associations between insulin resistance and oxidative stress, we hypothesized that FVB mice would be less insulin resistant, because increased skeletal muscle mitochondrial uncoupling reduced oxidative stress. Despite reduced mitochondrial oxidative capacity in C57B and enhanced mitochondrial uncoupling in FVB mice, respectively, after 2 wk of HFD, insulin resistance was absent as illustrated by normal glucose tolerance and normal glucose infusion rates during hyperinsulinemic and euglycemic clamps (Supplemental Fig. 6). Glucose intolerance, which was more pronounced in C57B mice on HF compared with FVB mice, developed after 5 wk of HFD (Fig. 5, A and B). Glucose infusion rates and skeletal muscle glucose uptake were similarly reduced by 5 wk of HF feeding in both strains (Fig. 5, C–F). Insulin-stimulated serine and threonine phosphorylation of Akt in skeletal muscle was not reduced by HFD in either strain, and serine phosphorylation did not differ between strains (Fig. 5, G–I). Insulin-stimulated threonine phosphorylation on Akt was reduced in FVB relative to C57B mice irrespective of the diet. In liver, insulin-stimulated Akt phosphorylation on both sites was significantly (P < 0.05) reduced in HF-fed FVB mice relative to NC-fed FVB mice but was not significantly reduced in C57B livers (Supplemental Fig. 7, A–C).

Fig. 5.

Development of insulin resistance in both C57B and FVB mice after 5 wk of HFD independently of changes in Akt phosphorylation. A and B, GTT; C and D, Glucose infusion rates (GIR); E and F, 2-Deoxyglucose (2-DoG) uptake; G, Representative Western blottings for insulin-stimulated Akt phosphorylation on Ser473 and Thr308; and H and I, Densitometry analysis of phosphorylated Akt/total Akt for the 473 and 308 sites, respectively, of C57B and FVB mice on NC (n = 4–6 for each strain) and HF (n = 6–7 for each strain). Data are mean ± sem *, P < 0.05; **, P < 0.005 vs. NC-fed mice; a, P < 0.005 vs. noninsulin stimulated; b, P < 0.05 vs. C57B mice under the same feeding condition.

Modulating oxidative stress in C57B mice enhanced mitochondrial function without changing insulin resistance

We subjected C57B mice to HFD for 5 wk while they were treated with saline or with the SOD-mimetic MnTBAP. Antioxidant treatment effectively reduced skeletal muscle ROS, as evidenced by a 50% reduction in DCFDA fluorescence (Fig. 6A), and improved mitochondrial ATP synthesis (Fig. 6B). However, neither glucose tolerance nor insulin sensitivity was improved by antioxidant treatment (Fig. 6, C and D). We next examined mice with whole-body deletion of UCP3 (UCP3KO) on the C57B6 background after 10 wk of HFD. HFD-induced oxidative stress in hindlimb muscle of wild-type mice was strikingly exacerbated in UCP3KO mice (Fig. 6E). Despite increased ROS levels, UCP3KO on HFD were equivalently glucose intolerant (Fig. 6F). Moreover, despite increased oxidative stress in UCP3KO mice on HF, insulin-stimulated Akt phosphorylation on Ser473 and Thr308 was similar to that of wild-type mice on HF (Supplemental Fig. 8).

Fig. 6.

Modulation of oxidative stress in C57B mice does not affect diet-induced insulin resistance. A and E, DCFDA fluorescence in hindlimb muscle homogenates; B, ATP synthesis in saponin-permeabilized soleus muscle; C and F, GTT; D, ITT. A–D, Data obtained in C57B mice fed HFD for 5 wk and simultaneously treated with saline or the antioxidant MnTBAP. E and F, Data obtained in wild-type and UCP3KO fed either NC or HFD for 10 wk. Data are mean ± sem. *, P < 0.05; **, P < 0.005 vs. HF-fed saline treated or vs. NC; ##, P < 0.005 vs. wild-type HF. WT, Wild type; AU, arbitrary units.

Discussion

The goal of the present study was to determine, in vivo, the relationship between mitochondrial function, oxidative stress, and skeletal muscle glucose uptake in the early stages of insulin resistance that develops in response to diet-induced obesity. We examined two independent mouse strains with differential susceptibility to diet-induced obesity and determined the impact of modulating oxidative stress and mitochondrial uncoupling on insulin-mediated glucose uptake. Our main findings are summarized in Table 2 and show that reduced insulin-mediated skeletal muscle glucose uptake occurs early in the course of HF feeding in two independent mouse strains and largely precede any impairment in insulin-mediated Akt activation. Importantly, changes in mitochondrial function clearly precede the impairment in skeletal muscle insulin-mediated glucose uptake. However, the mitochondrial adaptations and the development of oxidative stress are strain dependent and do not correlate with skeletal muscle insulin resistance. Specifically, in FVB mice, mitochondrial uncoupling prevents oxidative stress and increases energy expenditure but does not prevent skeletal muscle insulin resistance. In C57B mice, oxidative stress and mitochondrial dysfunction clearly precede the onset of insulin resistance, but concurrent scavenging of ROS does not prevent impaired skeletal muscle glucose uptake. Moreover, exacerbation of oxidative stress by deleting UCP3 does not worsen insulin resistance.

Table 2.

Summary of differences in oxidative stress, mitochondrial VO₂, ATP synthesis, mitochondrial uncoupling, and insulin resistance between C57B and FVB mice

	2 wk		5 wk
	C57B HF	FVB HF	C57 HF	FVB HF
Oxidative stress	ND	ND	↑	↓
VO2	↔	↑	↓	↔
ATP levels	↓	↓	↓	↓
Mitochondrial uncoupling	↔	↑	↔	↑
Insulin sensitivity	↔	↔	↓	↓

ND, Not determined.

Our studies confirm the important role for impaired insulin-mediated glucose uptake as a critical early event in the pathogenesis of insulin resistance that precedes major defects in insulin signal transduction to Akt. In L6 cells exposed to palmitate and in HF-fed C57B mice, impaired insulin mediated glucose uptake preceded any measurable defect in insulin-mediated Akt phosphorylation (27). Similarly, after 2 wk of HF feeding, insulin-stimulated glucose uptake and glucose transporter type 4 translocation were impaired in cardiomyocytes of C57B mice in the absence of changes in insulin signaling (28). Oxidative stress has been proposed to play a role in impaired glucose uptake, because mitochondrial ROS scavenging restored insulin-mediated glucose transporter type 4 translocation in palmitate-treated L6 cells (27). Moreover, acute administration of MnTBAP improved glucose tolerance and skeletal muscle glucose uptake in HF-fed mice, although the impact of long-term treatment was not examined (27). Whole-body overexpression of MnSOD modestly improved glucose tolerance after 12 wk and insulin tolerance after 24 wk of HFD (45% fat). Skeletal muscle glucose uptake and hepatic insulin action were not reported (27), and it is not clear whether MnSOD transgenic mice do not develop insulin resistance after short-term HF feeding.

We evaluated skeletal muscle glucose uptake in vivo in response to diet-induced obesity across a range of skeletal muscle ROS. Although we observed that increased skeletal muscle ROS preceded and correlated with a reduction in glucose uptake and insulin resistance, prevention of oxidative stress by concurrent treatment with the antioxidant MnTBAP did not restore insulin sensitivity. Thus, in contrast to acute studies (27), our data suggest that chronic antioxidant administration might lack efficacy in preventing early molecular defects that impair insulin sensitivity despite efficient ROS scavenging. Importantly, we also examined an independent strain of mice (FVB) in which mitochondrial uncoupling that develops in response to HF feeding prevented oxidative stress but did not prevent impaired insulin-stimulated glucose uptake. Finally, increased oxidative stress in UCP3KO C57B mice, which presumably are unable to reduce mitochondrial superoxide overproduction, did not exacerbate glucose intolerance. Finally, we explored alternative mechanisms that could link oxidative stress and insulin resistance, such as increased endoplasmic reticulum stress or stress kinase signaling, but did not find consistent changes in these pathways that could explain our findings (Supplemental Fig. 9). Taken together, these findings suggest that although impaired glucose uptake in cell culture models might be due in part to ROS-mediated mechanisms, the pathophysiology of impaired skeletal muscle glucose uptake in vivo in response to HF feeding is more complex and could be due to ROS-independent mechanisms.

Anderson et al. (7) reported that short-term HF feeding (60% fat) of Sprague Dawley rats increased mitochondrial ROS by 3 d and blunted insulin-mediated Akt phosphorylation after 6 wk. These defects were reversed by concurrent treatment with a mitochondrially targeted antioxidant. Also, mice with mitochondrially targeted catalase did not develop impaired insulin-mediated glucose uptake after 12 wk on this diet. However, Akt phosphorylation was not assessed under these conditions. There are important differences between our study and that of Anderson et al. (7). We used a lower concentration of dietary fat (45%) and examined effects on insulin-mediated glucose uptake at a point that preceded impaired insulin signaling to Akt. Thus, although it is possible that oxidative stress might play a role in skeletal muscle insulin resistance after very high levels of dietary fat, our findings argue that the adaptations of glucose uptake that precedes impaired Akt signaling in response to lesser degrees of dietary lipid excess might be independent of oxidative stress. Our protocol for in vivo Akt activation used a pharmacological dose of insulin, thus the possibility remains that more subtle degrees of impaired Akt activation could exist. In balance, our findings suggest that initial changes in skeletal muscle glucose uptake in the evolution of insulin resistance are likely independent of mitochondrial dysfunction or ROS overproduction. However, as insulin resistance becomes more severe, ROS-mediated mechanisms might sustain defects in insulin signaling, which can be prevented by suitable antioxidant strategies.

We provide strong evidence that skeletal muscle mitochondria adapt very rapidly to caloric excess, which is consistent with earlier studies in rodents and humans (6, 7, 29). These mitochondrial adaptations are strain dependent and determine the susceptibility of skeletal muscle to oxidative stress. Thus, a failure to effectively scavenge ROS in C57B mice likely contributed to the rapid development of mitochondrial dysfunction, which could be reversed by antioxidant treatment with MnTBAP. In contrast, mitochondrial uncoupling in FVB mice likely contributed to reducing mitochondrial oxidative stress. The failure of C57B muscle mitochondria to uncouple in response to HFD is unlikely due to changes in the actual amount of UCP3. Recent evidence for additional mechanisms of activation/deactivation of UCP2 and UCP3 has emerged (25, 30). A modest increase in glutathionylation using diamide, which reacts with small acidic thiols, such as glutathione, and promotes the formation of protein-glutathione mixed disulfides (PSSG), was associated with a robust augmentation of mitochondrial proton leak and glycolysis flux in aortic smooth muscle cells (31). Thus, the increase in glutathiolation of FVB proteins could contribute in part to the increase in proton leak observed in skeletal muscle of FVB mice on HF. A direct measurement of UCP2 or UCP3 protein glutathionylation will be necessary to explore this mechanism further in future studies.

An important role for mitochondrial uncoupling in mediating the skeletal muscle mitochondrial adaptations to HF feeding is also supported by the exacerbation of oxidative stress in UCP3KO C57B mice. Although our studies clearly suggest that oxidative stress might not contribute to impaired skeletal muscle glucose uptake, early in the course of diet-induced obesity, the short-term nature of our studies do not exclude the possibility that longer term antioxidant strategies, or long-term strategies that increase mitochondrial uncoupling, could be protective as has been suggested in mice with transgenic overexpression of UCP1 in skeletal muscle, in which diet-induced insulin resistance was prevented as a result of reduced obesity and increased AMP-activated protein kinase activation (32, 33).

Mitochondrial uncoupling in FVB mice likely increased energy expenditure that contributed to the short-term resistance of FVB mice to diet-induced obesity. The mechanisms that underlie increased mitochondrial uncoupling and energy expenditure in FVB mice are not fully understood but are likely independent of increased BAT mass and UCP1 expression in BAT, because these parameters were equally induced by HFD in the both strains. In addition, it is unlikely that increased activity and energy dissipation affected energy expenditure in FVB mice, because total ambulatory activity was lower in FVB mice when compared with C57B mice. We cannot rule out, however, that increased energy expenditure in FVB mice on HF could be related to increased sympathetic tone, as has been suggested by other studies (13, 34–36). Increased lipid influx in skeletal muscle of FVB mice could also contribute to increase mitochondrial uncoupling in this mouse strain. Indeed, we observed a higher TG content in hindlimb muscle from FVB mice that was exacerbated by 2 and 5 wk of HF feeding HFD. In addition, LPL protein levels were elevated in FVB mice fed HFD, suggesting increased FA uptake by their muscle.

In conclusion, our study underscores complex and strain-specific differences in the mitochondrial adaptations in skeletal muscle to diet-induced obesity. Although mitochondrial dysfunction and ROS overproduction may develop in skeletal muscle in obesity prone strains, ROS-mediated mitochondrial uncoupling in FVB may increase whole-body energy expenditure in response to HFD and contribute to reduced weight gain. However, it is unlikely that oxidative stress directly precipitates the impairment in insulin-mediated glucose uptake that characterizes the earliest stages of insulin resistance that develops as a consequence of diet-induced obesity, and as such, there might be a limited role for antioxidants in the primary prevention of insulin resistance. Current evidence suggests that the impact of antioxidant strategies to prevent insulin resistance might be modest and require a prolonged duration to see a clinically meaningful effect. Therefore, before antioxidant treatment can be advocated for treating insulin resistance, it will be important to identify subjects who lack mitochondrial adaptations, such as mitochondrial uncoupling, which would already limit oxidative stress, and target therapy to those individuals with the highest burden of oxidative stress and potentially the greatest degree of mitochondrial dysfunction.