TFAM Enhances Fat Oxidation and Attenuates High-Fat Diet–Induced Insulin Resistance in Skeletal Muscle

An estimated 9% of the U.S. population has diabetes, and an additional 86 million people representing one in three adults have prediabetes, which is a risk factor for type 2 diabetes mellitus (T2DM) (1). Skeletal muscle insulin resistance (IR) is a cardinal feature of T2DM and it is often present years prior to T2DM onset (2). Ameliorating IR could prevent or reverse T2DM. Altered mitochondrial function has been shown to cause IR. For example, excessive mitochondrial H₂O₂ (mtH₂O₂) emissions following high-fat diet (HFD) results in IR via attenuation of insulin signaling and glucose uptake (3). In women with IR, abnormal mtH₂O₂ emission when normalized by aerobic exercise training improved insulin sensitivity (4). Moreover, age-related IR occurs with reduced mitochondrial DNA (mtDNA) abundance (5), and maintaining a high abundance of mtDNA and reduced oxidative stress ameliorate age-related IR (6). Overall, it seems that efficient mitochondrial function can improve insulin sensitivity. Lipid accumulation in skeletal muscle (7) or metabolic overload in mitochondria (8) have also been shown to occur with IR, and it has been shown that incomplete β-oxidation contributes to skeletal muscle IR (9). These studies support a notion that enhances mitochondrial oxidative capacity and efficiency, especially the upregulation of mtDNA abundance and reduction of oxidative stress ameliorate IR.

Mitochondrial transcription factor A (TFAM) is a key regulator of mitochondrial DNA replication, repair, and gene transcription (10). An ex vivo study demonstrated that overexpressing TFAM in cardiomyocytes decreased mtH₂O₂ (11). In contrast, adipose tissue–specific deletion of TFAM decreased reactive oxygen species (ROS), increased mitochondrial fuel oxidation, and protected mice against obesity and IR (12). These paradoxical studies suggest that the role of TFAM in regulating mitochondrial fuel metabolism and ROS emission is tissue specific and the effect of TFAM on skeletal muscle is currently unknown.

Hence, we sought to determine whether muscle-specific overexpression of human TFAM (hTFAM) in mice (transgenic [Tg]) can increase mtDNA abundance, β-oxidation, and mitochondrial respiration, decrease mtH₂O₂ generation, and prevent HFD-induced IR. We further investigated various molecular and bioenergetics pathways by which TFAM influences insulin sensitivity and oxidative stress in skeletal muscle. These mechanistic insights provide opportunities for potential novel therapeutic approaches to counteract IR, and to prevent and treat many metabolic disorders, including T2DM.

Study procedures are outlined here, but more details are provided in the Supplementary Data.

hTFAM cDNA and muscle-specific creatinine kinase promoter construct was inserted into a TA cloning vector (pcDNA3.1/NT-GFP-TOPO). Purified vector was directly injected into the pronucleus of single-celled FVB mice embryos, and Tg embryos were implanted in a pseudopregnant mother. Sexually mature F1 Tg progeny were derived and used for all analyses. Both Tg and wild-type (Wt) male animals were assigned to either a chow diet (chow) or an HFD for 12 weeks. We chose only male animals to avoid confounding the effects of fluctuating female hormones on our outcomes.

Body composition was measured, after an overnight fast, using EchoMRI (Echo Medical Systems) and DEXA (GE Healthcare Lunar).

Measurements were performed in both fed and fasted states. VO₂ and VCO₂ were measured using a 16-chamber open-circuit cage calorimetry system (Columbus Instruments). Respiratory exchange ratio and energy expenditure was computed using standard equations. Ambulatory levels were measured using the infrared photocell beam interruption method.

Mice were fasted for 5 h, and peripheral insulin sensitivity was measured using a euglycemic clamp (120–130 mg/dL) maintained for 120 min. Mice were primed with a 16 mU/kg bolus of Humulin R followed by 2.5 mU/kg/min continuous infusion. Blood was collected every 10 min for glucose and insulin measurements (as described above).

Mitochondria were isolated from quadriceps muscle and analyzed by high-resolution respirometry (13). Mitochondria were added to a 2-mL chamber (Oxygraph-2K) followed by sequential additions of glutamate, malate, ADP, succinate, and inhibitors. Mitochondrial membrane integrity was verified with cytochrome C.

Differentiated C₂C₁₂ myoblasts were infected with adenovirus-bearing TFAM (Applied Biological Materials Inc.) in either the presence or absence of 0.2 mmol/L palmitate (PM). A similar protocol was followed to generate empty vector (EV) myotubes. All myotubes were differentiated for 5 days prior to collecting them for various analyses.

Quadriceps muscle mitochondria were isolated and treated with GM, and H₂O₂ emission was measured using Amplex Red oxidation as previously published (14). Myotubes were suspended in respiration medium, and membranes were permeabilized using digitonin. GM were added and H₂O₂ was measured as described above.

Mice gastrocnemius muscle was homogenized and the adduct biomarker 8-oxo-2′deoxyguanosine was measured as previously described (6). Muscle redox state was assessed by measuring oxidized glutathione (GSSG) and reduced glutathione (GSH) using a commercial kit (Cayman Chemicals).

Semiquantitative RT-PCR was used to measure PGC-1α, citrate synthase (CS), NDFAU9, SHDH, UQRCR1, UQRCR2, CYCS, MT-CO1, cytochrome c (Cyt c) oxidase subunit (COX) 1, COX4, APT5A, and UCP3 (uncoupling protein 3) genes.

Protein abundance was measured with antibodies (Abs) against PPARβ, ATP synthase subunit α (ATPsyn), succinate-ubiquinone oxidoreductase (SUO), ND ubiquinone oxidoreductase (ND), ubiquinone-Cyt c oxidoreductase Core subunit 1 (Core1) and Core2, COX I and IV, caspase 3, PGC-1α, UCP3, c-Myc-Tag, pS473-AKT, pT308-AKT, total AKT, total AMPK, phosphorylated AMPK (pAMPK), GLUT4, PPARα, SOD2, catalase, NRF-1, MEF-2A, CaMKKβ, CS, and Cyt c.

Anti-Myc or anti-ubiquitin Ab was mixed with magnetic beads for 1.5 h at room temperature. A total of 250 μg of protein from each group was added in an Ab and beads mixture and then rotated overnight at 4°C. Ab-antigen complexes were captured on magnetic beads and probed using Western blot analysis for Cyt c, ND, Core1, or COX1 Ab.

Acylcarnitines (C0, C2, C3, C4, C5, C8, C12, C14, C16, C18, and C18:1), ceramides (cer14, Cer16, Cer18, Cer24, Cer24:1, sphinganine [SPA], sphingosine [SPH], and SPH-1-phosphate [S1P]), saturated (16:0/16:0) diacylglycerols (DAGs), and unsaturated DAGs (18:0/18:2, 18:0/18:1, 18:1/18:1, 18:2/18:2, 16:0/18:1) were extracted from gastrocnemius muscle and analyzed via liquid chromatography–mass spectrometry, as previously described (15–17).

Tissue chunks were weighed and homogenized, and metabolites were extracted. ¹H spectra were recorded using 600 MHz nuclear magnetic resonance spectroscopy. Metabolites were identified and quantified using Chenomx software and expressed in micromoles per gram of tissue analyzed.

TFAM or EV was inserted into differentiated myotubes. After 5 days, cells were preloaded with TMRM (tetramethylrhodamine methyl ester) and Hoechst 33342. Labeled cells were resuspended in imaging buffer supplemented with either glucose or PM for 2 h. Cells were imaged continuously for 480 s in 10-s intervals. During the imaging study, 2 μmol/L FCCP (carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone) was added at 180 s and 8 μmol/L FCCP was added at 300 s. TMRM intensity was calculated as Δt = (t_x − t₀)/t₀ × 100, where t_x is fluorescence intensity at any time point and t₀ is baseline fluorescence.

After 5 days of cell differentiation, C₂C₁₂ cells were incubated for 24 h with 3 μmol/L (final concentration) of 2,4-dinitrophenol (DNP) (provided by Dr. Geisler, Mitochon Pharmaceuticals, Inc, Blue Bell, PA) or DMSO as a control in differentiation medium.

Data were analyzed using Student t test, one-way ANOVA, two-way ANOVA, or Mann-Whitney rank sum test wherever appropriate. Statistical significance was denoted as *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

We generated Tg FVB mice with skeletal muscle–specific overexpression of hTFAM (Fig. 1A). Muscle TFAM mRNA, protein (Fig. 1B), mtDNA copy number (Fig. 1C), and mitochondrial-encoded mRNAs (Supplementary Fig. 1) were higher in Tg mice compared with Wt littermates.

Mice on a 12-week HFD (60% calorie from fat) consumed more calories than chow mice, but no differences in caloric intake were observed between Wt and Tg mice (Supplementary Fig. 2). However, Tg mice had lower body weight and lean mass than Wt mice irrespective of the diet (Fig. 1D and E). Moreover, Tg mice on an HFD gained less body weight and fat mass than Wt mice on an HFD (Fig. 1E and F). Based on indirect calorimetry, we observed that Tg mice on an HFD had higher metabolic rates during both day and night when compared with Wt mice on an HFD, irrespective of whether they were in fasted or fed state (Fig. 1G). The VO₂ and VCO₂ measurements mirrored the metabolic rate differences observed between Tg and Wt mice (Fig. 1H), although the respiratory exchange ratio was unaltered by TFAM (Supplementary Fig. 3). Curiously, an HFD increased nightly activity of Tg mice more than that of Wt mice (Fig. 1I) and this phenomenon was absent with a chow diet (Supplementary Fig. 4); it is possible that this increased activity contributed to the nightly increase in energy expenditure. These increased activity levels potentially can be attributed to higher mtDNA abundance and functional capacity (18). Hyperinsulinemic-euglycemic clamp showed that an HFD significantly reduced the insulin sensitivity in Wt mice when compared with chow diet (Fig. 1J and K). However, Tg mice on an HFD demonstrated a lesser decline in insulin sensitivity than Wt HFD-fed mice based on glucose infusion rate (GIR) during hyperinsulinemic euglycemic clamp and muscle glucose uptake (Fig. 1J and K). We did not observe any difference in endogenous glucose production during the euglycemic clamp (Supplementary Fig. 5), indicating that the effects of TFAM in insulin sensitivity were restricted to peripheral tissues. An intraperitoneal insulin tolerance test indicated that HFD in Wt mice decreased insulin sensitivity, whereas Tg mice on a chow diet or HFD had the same levels of insulin sensitivity as Wt chow mice (Supplementary Fig. 6). Muscle glucose uptake also decreased in both Wt and Tg mice with an HFD; however, Tg mice on an HFD had higher glucose uptake than Wt mice on an HFD (Fig. 1L), indicating that TFAM attenuated the HFD-induced decline in muscle glucose uptake. Blood glucose levels of Tg HFD-fed mice were lower than Wt HFD-fed mice (Fig. 1M).

Collectively, these results indicate that skeletal muscle overexpression of TFAM increased mtDNA abundance and metabolic activity while attenuating HFD-induced weight gain and loss of insulin sensitivity.

We studied the bioenergetic characteristics of muscle mitochondria to understand the potential link to increased metabolic rate and attenuated fat accumulation. Chow Tg mice had significantly higher state 3 respiration when using palmitoyl-carnitine and malate (PCM) as substrate and lower state 3 respiration when using glutamate and malate (GM) as substrates, when compared with Wt chow mice normalized for protein content (Fig. 2A) as well as normalized to the tissue weight (Supplementary Fig. 7). Mitochondrial respiratory efficiency, measured as respiratory control ratio, was significantly higher in Tg mice compared with Wt mice with PCM and significantly lower with GM substrates (Fig. 2B). Similar phenomena were observed in Tg mice on HFD (Fig. 2C and D). These data revealed that TFAM decreases coupling efficiency when respiration is supported through carbohydrate-based substrates but increases coupling efficiency when respiration is supported by fatty acid (FA) substrates. Overall, these results support that Tg mice exhibit marked shifts in skeletal muscle mitochondrial bioenergetics, whereby when fat oxidative capacity is increased, complex I supported respiratory capacity is decreased.

We observed higher amounts of ADP and AMP and lower amounts of ATP and creatine phosphate (CrP) in skeletal muscle of Tg mice compared with Wt mice regardless of diet (Fig. 2E–K) and a higher AMP/ATP ratio is known to increase the activation of AMPK (19). In contrast to the lower ATP in Tg mice, we observed higher levels of citrate, CS protein, CS activity, and other tricarboxylic acid (TCA) cycle metabolites when compared with Wt mice in both the chow and HFD groups (Fig. 2L–S). The energy demands of high metabolic rate, like during exercise, increase TCA cycle activity and decrease CrP (20), as observed in Tg mice who have high energy demand of muscle. TFAM also increased hexokinase enzyme activity (Fig. 2T) and lactate, indicating that TFAM increased glycolysis (Fig. 2U). Additionally, glycogen synthase phosphorylation was increased by TFAM and HFD, indicating impairment in glycogen storage (Supplementary Fig. 8). Increased malate and fumarate as well as aspartate and glutamate with reduced glutamine suggests increased aspartate-malate shuttle and increased the conversion of glutamine to glutamate for use in the TCA cycle. These data also suggest that TFAM increased the aspartate-malate shuttle to increase NAD⁺ import into the mitochondrial matrix. Carnosine, an inhibitor of electron transport chain (ETC) in animals (21), was also lower in Tg mice compared with Wt mice in both the chow and HFD groups (Fig. 2V). The contrasting findings of lower ATP levels and higher levels of TCA activity in the skeletal muscle of Tg mice when combined with the fact that Tg mice had higher metabolic rate and activity levels suggest that Tg mice consume ATP at a higher level.

HFD is known to increase mtH₂O₂ production (3). As expected, Wt mice on an HFD had the highest mtH₂O₂ emissions followed by Wt mice on a chow diet (Fig. 3A). Of interest, Tg mice had the lowest mtH₂O₂ emissions irrespective of the diet (Fig. 3A). This is further supported by reduced ROS-mediated DNA damage (Fig. 3B), and a lower GSSH/GSH ratio indicates lower oxidative stress (Fig. 3C) regardless of diet. Concomitantly, Tg mice had higher levels of endogenous antioxidant defense enzymes (superoxide dismutase 2, catalase, and Cyt c) when compared with Wt mice regardless of diet (Fig. 3D). These data indicate that muscle TFAM can increase FA oxidation while lowering ROS emissions and can enhance antioxidant defense.

Next, we sought to determine the potential mechanisms by which TFAM influences ROS production. Irrespective of the diet, Tg mice had higher levels of PPARβ (Fig. 3E), which can regulate catalase (22) and SOD2 (23). Tg HFD mice had the highest levels of Cyt c when compared with rest of the groups (Fig. 3D), and Cyt c under nonapoptotic conditions can act as an ROS scavenger (24). We found that TFAM decreased mtH₂O₂, increased SOD2 and catalase (Fig. 3F and G), and blocked the release of cleaved caspase (Fig. 3H) with PM treatment in myotube. PM treatment increased cytosolic Cyt c levels in EV cells when compared with sham treatment, whereas mitochondrial Cyt c levels did not change (Fig. 3I). In contrast, TFAM overexpression increased the mitochondrial Cyt c levels regardless of PM or sham treatment while reducing the cytosolic Cyt c levels to basal level when treated with PM (Fig. 3I).

Since Tg mice had higher levels of activated AMPK (i.e., pAMPK) regardless of diet (Fig. 3E), we hypothesized that TFAM may be using pAMPK to hold Cyt c in the mitochondria to act as a ROS scavenger, and found that activated AMPK binds to Cyt c (Fig. 3J). The net effect of reduced ROS emission and increased antioxidant defense by TFAM also attenuated the ROS-induced loss of GLUT4 observed with PM treatment (Fig. 3G). All of these data support a notion that muscle TFAM controls ROS via SOD2, catalase and Cyt c via PPARβ and pAMPK (Fig. 3K).

We measured β-oxidation and found HFD significantly increased LCAD, a primary β-oxidation enzyme (Fig. 4I) and lower medium-chain (C5–C12) and long-chain (C16–C18) acylcarnitines; however, HFD did not increase short-chain (C0–C4) in Wt. These data indicate that HFD induces an incomplete β-oxidation. We also found that, irrespective of diet, Tg mice had lower very long-, long-, and medium-chain acylcarnitines (Fig. 4D–H), and higher levels of LCAD (Fig. 4I). However, Tg mice had higher levels of short-chain acylcarnitines when compared with Wt mice in both the chow and HFD groups (Fig. 4A and B). These data suggest that TFAM overexpression in skeletal muscle activates a complete β-oxidation. DAGs have been reported to reduce insulin sensitivity in skeletal muscle (25). We found that TFAM influences abundance of DAGs (Supplementary Fig. 9). As reported by Amati et al. (26), DAG levels are higher in exercise-trained muscle corresponding to higher insulin sensitivity, and we also found that muscle DAG levels in TFAM Tg muscle are higher concurrent with higher insulin sensitivity. These data indicate that the cellular DAG content cannot explain skeletal insulin sensitivity induced by TFAM.

HFD has been shown to increase the abundance of intramyocellular DAGs, as well as ceramides (27), and certain classes of ceramides are associated with skeletal muscle IR (28). Further, a decrease in palmitoylcarnitine (Fig. 4F), a precursor of palmitoyl-CoA–based ceramide synthesis, in Tg mice suggested that TFAM may affect muscle ceramide levels. We found that Tg mice had higher levels of SPH, SPA, and S1P, regardless of the diet (Fig. 4J). S1P has been shown to increase the ability of muscle to use FA as a fuel for energy needs (29). Cer14, Cer16, Cer24:1, and Cer24 levels were higher in Tg mice on a chow diet when compared with Wt mice on a chow diet, whereas an HFD did not significantly change their levels between Tg and Wt mice (Fig. 4K, L, N, and O). In contrast, Cer18 is associated with IR (28), was higher in Wt mice on an HFD when compared with chow diet, whereas Tg mice on an HFD had levels similar to that of Wt or Tg mice on a chow diet (Fig. 4M). These data suggest that, even though overexpression of TFAM increases intramyocellular de novo ceramide synthesis, TFAM differentially regulates levels of Cer18, a potent inhibitor of insulin signaling, and TFAM controls β-oxidation to block the accumulation of incompletely oxidized lipids that is the IR factor.

Collectively, these data indicate that HFD induces an incomplete β-oxidation. However, skeletal muscle TFAM enhances both the β-oxidation pathway and the de novo ceramide synthesis pathway to increase FA oxidation while attenuating the accumulation of IR-associated bioactive lipids (Fig. 4O).

High levels of free FA (FFA) were shown to alter ATP synthesis in skeletal muscle (30). However, the precise effect of HFD on ETC enzymes has never been studied. Further, Tg mice on an HFD had increased FA oxidative capacity, decreased ROS, and increased energy metabolism (Figs. 2–4), which would require adaptation of the ETC by TFAM to accommodate the high fuel flux. HFD in Wt mice increased protein expression of ND and CORE2 (Fig. 5A) and decreased ATPsyn when compared with the chow diet (Fig. 5A). However, Tg mice had significantly the increased protein expression of ATPsyn and SUO proteins when compared with Wt mice in both the chow and HFD groups (Fig. 5A). These data suggest that TFAM overexpression can attenuate HFD-induced loss of complex V. Tg mice had lower protein expression of ND, CORE1, UCP3, and COX1 proteins in both diet groups compared with Wt mice (Fig. 5A). Ubiquitination of ND, CORE1, and COX1 proteins also increased in Tg mice (Fig. 5B). These data suggest that the TFAM overexpression downregulates complex I, complex III, and complex IV, which pumps proton into mitochondrial intermembrane space via ubiquitination. mRNA measurements showed that protein changes induced by TFAM in the ETC pathway are mediated in a post-translational manner via protein ubiquitination rather than direct transcriptional changes (Fig. 5C).

The asymmetric reformation of key ETC complexes by TFAM is likely to alter the mitochondrial membrane potential (Δψ_m) (31). Further, TFAM mice showed different coupling efficiency with FA or glucose as fuel substrate (Fig. 2B and D). At resting state, TFAM myotubes have lower TMRM intensity (i.e., a surrogate for Δψ_m) in the mitochondria than EV when using glucose as substrate and higher Δψ_m when using PM as a substrate (Fig. 5D). When a small amount (2 µM) of an ionophore (FCCP), which is an uncoupler disrupting ATP synthesis, was added, mitochondria significantly depolarized in EV cells treated with PM when compared with TFAM cells (Fig. 5D and E). In contrast, when glucose was being used as fuel substrate, TFAM cells lost significant Δψ_m compared with EV cells with 8 µM FCCP, however, Δψ_m was protected by TFAM when 2 µM FCCP was added (Fig. 5D and E). Time-resolved changes in the Δψ_m are shown in Fig. 5F and G. These results indicate that TFAM induces Δψ_m mild uncoupling, perhaps via decreased ND, Core1, and COX1 (Fig. 5A). Simultaneously, TFAM is also protecting Δψ_m, perhaps via ATP synthase reversal. This is supported by previous studies showing that Δψ_m protection (i.e., mild uncoupling) could be mediated via ATP synthase reversal (Fig. 5H) when mitochondrial ETC is altered (32) or glycolysis is increased (33). In addition, a previous study (34) showed that an increased Cyt c level (observed with TFAM overexpression) (Fig. 4D) is associated with mild Δψ_m uncoupling. All of these data suggest that Δψ_m coupling efficiency in TFAM-overexpressed muscle is higher with FA than glucose, showing the preference of FA as the energy source in TFAM muscle.

A previous study (35) suggested that HFD-induced IR cannot be explained on the basis of changes in electron chain complexes. Next, we sought to identify the mechanism on how TFAM preserves insulin sensitivity under HFD conditions. We measured skeletal muscle pAKT in response to insulin in Wt and Tg mice on either chow diet or HFD. HFD attenuated the pAKT in Wt mice in response to insulin when compared with chow diet (Fig. 6A). In contrast, Tg mice preserved the insulin-mediated pAKT under the HFD state (Fig. 6A), which is associated with enhanced energy metabolism (Fig. 2), antioxidant buffering system (Fig. 3), and higher β-oxidation (Fig. 4) by TFAM. Further, TFAM overexpression results in an increase GLUT4 (a key regulator of glucose transport) expression, which is directly regulated by MEF2A (36). Further, NRF-1, upstream of MEF2A, is increased by PPARβ (21), these factors also were upregulated by TFAM overexpression irrespective diet. We also found that pAMPK, which translocates GLUT4 to uptake glucose, is activated by TFAM. We also observed increased expression of PGC-1α protein in Tg mice in both diet groups (Fig. 6B), and this pathway also may increase GLUT4 expression and translocation (37). Thus, our results indicate that TFAM alters pAMPK, PPARβ, or PGC-1α pathways (Fig. 6C) and offer an additional mechanistic explanation of the enhanced insulin sensitivity and glucose uptake observed in Tg mice on an HFD (Fig. 1J–L).

The current study demonstrates that human TFAM overexpression in skeletal muscle has a unique and profound impact on multiple molecular pathways regulating energy metabolism. These effects are distinct from what has been reported in other tissues when TFAM is overexpressed (11,12) or PGC-1α is overexpressed in muscle (38). Most of the molecular changes noted are beyond the direct effects of TFAM as a transcription factor of mitochondrial biogenesis. Importantly, Tg mice increased FA oxidation but warded off HFD-induced decline in glucose disposal, thus counteracting the well-known glucose-FA cycle (39). A notable effect of TFAM overexpression is the enhancement energy expenditure in conjunction with higher mitochondrial FA oxidative capacity, which contributed to reduced fat accumulation and metabolites of incomplete fat oxidation. TFAM also despite enhanced FA oxidation–reduced oxidative stress in mice on an HFD. We used hTFAM in the current study that has >70% sequence homology to that of mouse and DNA-binding domains were conserved. The objective was to understand the biological effects of hTFAM, but the effects related to the species differences cannot be fully excluded.

The most striking effect of TFAM was on energy metabolism in muscle by enhancing FA oxidation and reducing the accumulation of fat and incomplete metabolites of FA oxidation. These changes likely contributed to higher insulin sensitivity, as noted previously (28). We further addressed the potential molecular underpinnings on how TFAM preserved muscle glucose uptake on an HFD. As expected (3), 12-week HFD in Wt mice reduced glucose uptake in muscle (Fig. 1L) in conjunction with reduced Akt phosphorylation. TFAM overexpression increased AMPK activity in conjunction with MEF2A expression that likely contributed to increased GLUT4 expression (36) (Figs. 4E and 6B). Experiments in cell lines show a direct link between AMPK activation and AKT phosphorylation (41). Our results support that the reduction of AMPK activation by HFD (Fig. 3E) led to a decrease in AKT activation (Fig. 6A), which seems to be counteracted by increased AMPK and AKT activation by TFAM (Figs. 3E and 6A). This indicates a novel role for TFAM in skeletal muscle glucose uptake via AMPK. Of interest, activity levels in Tg mice on an HFD especially increased in the night, which has been reported to occur with enhanced mitochondrial capacity (18), and this increased activity also could contribute to insulin sensitivity and activation of AMPK. Some of the TFAM effects are similar to the hyperthyroid state, although unlike in the hyperthyroid state oxidative stress is lower in TFAM mice. Another important effect of TFAM was reduction in ROS emission and oxidative stress, even with HFD and high FA oxidation (Fig. 3A), which also could explain enhanced insulin sensitivity in Tg mice. Chronic overproduction of ROS increases oxidative stress, damages DNA, increases apoptosis, and triggers inflammation, which is a key feature in T2DM and metabolic syndrome (42). Previous studies (40,43) showed that IR induced by HFD can be mitigated by the elimination of ROS. In the current study, TFAM appears to use two different mechanisms to prevent HFD-induced overproduction of ROS: TFAM increased antioxidant enzymes SOD2 and catalase (Fig. 4D) and increased the mitochondrial abundance of Cyt c (Fig. 3I), a potent ROS scavenger. TFAM increased PPARβ (Fig. 3E), which can upregulate SOD2 and catalase (22). Hence, the increased insulin sensitivity observed in TFAM HFD mice may also be related to antioxidant buffers via PPARβ and eliminating ROS.

TFAM is a mitochondrial transcription factor. TFAM overexpression in muscle increased β-oxidation of FAs (Fig. 4), TCA cycle (Fig. 2), and some nuclear-encoded protein expression (Figs. 5 and 6). All of these effects cannot be directly regulated by TFAM with its well-known direct effect on mtDNA because many proteins in the TCA cycle and β-oxidation are encoded by nuclear DNA. Δψ_m depolarization is known to elevate the cytoplasmic Ca²⁺ level (44), and we found that TFAM overexpression induces Δψ_m depolarization with glucose substrate (Fig. 5D–G) and an increase in CaMKKβ expression (Fig. 6B), indicating that TFAM signals to the nucleus by the Ca²⁺/CaMKKβ pathway to induce nuclear-encoded proteins such as PGC-1α and PPARβ (Figs. 3E and 6B). PPARβ activation results in an increase in lipid oxidation and glucose metabolism, as previously reported (45), and improves insulin sensitivity via PGC-1α (46). HFD promotes FA oxidation (Fig. 4) and increased PPARα expression in muscle (Fig. 6), and muscle-specific PPARα overexpression without increasing activity (i.e., muscle contraction) has been reported to increase β-oxidation and decrease glucose uptake and oxidation (47). By enhancing the TCA cycle based on TCA cycle substrates (Fig. 2), TFAM thus seems to have counteracted the imbalance between β-oxidation and TCA cycle following HFD (Fig. 4) and prevented the accumulation of acyl-CoAs, their respective acylcarnitines, and perhaps other as yet unidentified metabolites that could contribute to mitochondrial failure (9) and IR. TFAM overexpression in skeletal muscle results in an increase in PGC-1α (Fig. 6), which also controls coping with FA load by coordinately regulating β-oxidation, TCA cycle, and ETC activity (48). The mitochondria in Tg mice muscle prefer FA to glucose to produce ATP (Figs. 2 and 5), and this is likely related to PPARβ and PGC-1α enhancement by TFAM (Figs. 3E and 6B). PPARβ and PGC-1α increased the use of FA by PDK4 (pyruvate dehydrogenase kinase 4) (49,50), and PPARβ also is reported to mediate glucose uptake and glycolysis (45); thus, TFAM overexpression in muscle results in an increase in β-oxidation as well as glycolysis, as noted by increased lactate levels (Fig. 2). These effects are associated with PPARβ-enhanced mitochondrial coupling efficiency, and Δψ_m in Tg muscle seem to be specific with FA (Fig. 5). Thus, HFD-induced defect in glucose uptake was attenuated by TFAM (Fig. 1K and L), possibly related to TFAM preference for FA than for glucose for muscle fuel (Figs. 2A–D and 5D–G).

Δψ_m is an important factor for maintaining cellular energy homeostasis, and small reductions in Δψ_m induced by mild uncoupling were shown to decrease H₂O₂ emissions (51,52) and produce a natural antioxidant effect (53). Hence, mild mitochondrial uncoupling can be potentially therapeutic in disorders induced by oxidative stress. We found that DNP, an established uncoupler, induced the activation of AKT and AMPK and the expression of GLUT4 (Fig. 6D), thus contributing to insulin sensitivity, but had no effect on PPARβ or any mitochondrial enzymes (Fig. 6E). Generally, mitochondrial uncoupling increases proton leak, resulting in an increase in AMP/ATP ratio, and induces various energy-sensing pathways (54). TFAM overexpression, however, did not increase FA-induced proton leak, but rather decreased the expression of UCP3 (Fig. 5A). TFAM also prevented FA-induced depolarization of mitochondria (Fig. 5D–G). We elucidated two potential mechanisms by which TFAM could be maintaining Δψ_m when challenged with FA. First, TFAM is inducing mild uncoupling by asymmetrically changing mitochondrial enzymes (Fig. 5A). Decreasing complexes I, II, and IV would lower Δψ_m but enhanced complex V (ATP synthase) likely enhanced ATP production. Second, Cyt c can regulate Δψ_m and reduce H₂O₂ emission through AMPK activation (34). In the current study, TFAM decreased H₂O₂ emission and increased Cyt c, which was bound to pAMPK (Fig. 3A and J). Thus, our results indicate that TFAM controls mitochondrial Δψ_m uncoupling via Cyt c/AMPK and reformation of ETC.

In conclusion, the current study demonstrated that muscle-specific overexpression of hTFAM has hitherto unknown beneficial metabolic effects. TFAM overexpression attenuated HFD-induced loss of glucose uptake and insulin sensitivity despite increasing FA oxidation. TFAM also prevented the oxidative stress that occurs on an HFD by enhancing endogenous antioxidant defense. TFAM enhanced the TCA cycle and ATP synthase, and, by pathways involving AMPK, PPARβ, and PGC-1α TFAM, also caused Δψ_m uncoupling as potential mechanisms by which TFAM counters many of the adverse effects of an HFD. These identified pathways are rich in therapeutic targets that can ameliorate IR and oxidative stress and potentially treat obesity.

Acknowledgments. The authors thank Katherine Klaus and Dawn Morse (Mayo Clinic) for skillful technical support.

Funding. The study was supported by the David Murdock Dole Professorship (K.S.N.) and by grants from the National Institutes of Health Center for Scientific Review (DK-007198 to M.L.J., TRDK-007352 to G.C.H., and RO1-DK-41973 to K.S.N.) and the Mayo Clinic Metabolomics Core (supported by grant U24-DK-100469).

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

Author Contributions. J.-H.K. conducted all cell line studies and molecular phenotyping studies. J.-H.K., S.D., and I.V. conducted metabolite measurements. J.-H.K., S.D., and K.S.N. drafted the manuscript. J.-H.K., M.L.J., S.D., N.K.L., I.V., G.C.H., S.A.C., G.N.R., G.I.S., and I.R.L. conducted various aspects of the study. M.L.J. performed mitochondrial phenotyping studies. S.D. conducted statistical analyses. N.K.L., G.C.H., and S.A.C. conducted animal studies. G.I.S. performed hyperinsulinemic clamp and muscle glucose uptake measurements at the Yale Core Laboratory. K.S.N. designed and supervised the study and performed the data analysis and interpretation. All authors contributed to the final version of the manuscript. K.S.N. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Data and Resource Availability. The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.