Reducing the mitochondrial oxidative burden alleviates lipid-induced muscle insulin resistance in humans

Abstract

Preclinical models suggest mitochondria-derived oxidative stress as an underlying cause of insulin resistance. However, it remains unknown whether this pathophysiological mechanism is conserved in humans. Here, we used an invasive in vivo mechanistic approach to interrogate muscle insulin action while selectively manipulating the mitochondrial redox state in humans. To this end, we conducted insulin clamp studies combining intravenous infusion of a lipid overload with intake of a mitochondria-targeted antioxidant (mitoquinone). Under lipid overload, selective modulation of mitochondrial redox state by mitoquinone enhanced insulin-stimulated glucose uptake in skeletal muscle. Mechanistically, mitoquinone did not affect canonical insulin signaling but augmented insulin-stimulated glucose transporter type 4 (GLUT4) translocation while reducing the mitochondrial oxidative burden under lipid oversupply. Complementary ex vivo studies in human muscle fibers exposed to high intracellular lipid levels revealed that mitoquinone improves features of mitochondrial bioenergetics, including diminished mitochondrial H₂O₂ emission. These findings provide translational and mechanistic evidence implicating mitochondrial oxidants in the development of lipid-induced muscle insulin resistance in humans.

Fig. 1.. The mtAO mitoquinone ameliorates lipid-induced muscle insulin resistance in humans in vivo.

(A) Schematic overview of the study design. HI clamp, hyperinsulinemic-isoglycemic clamp. (B) Experimental trial workflow, including the mean glucose infusion rate corrected per body mass (BM) and the arterial glucose concentrations. The dotted line indicates arterial glucose concentration corresponding to isoglycemia. Data presented as means ± SEM. (C) Plasma insulin, free fatty acid (FFA), and catecholamine levels measured before the lipid infusion (Baseline), as well as before (Pre-clamp) and after (End-clamp) the HI clamp. Lipid + mtAO, n = 9 (End-clamp). (D) Individual changes in the quantitative insulin sensitivity check index (QUICKI) in response to 3 hours of lipid infusion. (E) Whole-body insulin sensitivity expressed as the clamp-derived index of insulin sensitivity (SI_clamp) (122). Lipid, n = 10; Lipid + mtAO, n = 9. (F to K) Skeletal muscle insulin sensitivity expressed as the leg glucose uptake, as calculated from the arteriovenous difference in plasma glucose concentration and the femoral arterial blood flow corrected per leg muscle mass (LMM). [(F), (H), and (J)] Time course of leg glucose uptake, a-v glucose difference, and leg blood flow. [(G), (I), and (K)] Leg glucose uptake, a-v glucose difference, and leg blood flow before the HI clamp (Basal) and during the steady-state clamp period (Insulin). Between-treatment differences in insulin-stimulated leg glucose uptake, a-v glucose difference, and leg blood flow, calculated as the change (Δ) from the basal to the steady-state clamp period. (L) Pearson’s correlation between mtAO-induced changes (Δ) in glucose infusion rate (GIR) and leg glucose uptake. Linear mixed models were used to estimate between-treatment differences [(C), (E), (G), (I), and (K)]. Data presented as observed individual values with estimated means ±95% confidence limits, unless otherwise stated. n = 10, unless otherwise stated. Illustrations in (A) were created with BioRender.com.

Fig. 2.. The mtAO mitoquinone does not affect muscle substrate oxidation or lactate release under lipid overload.

(A and B) Leg O₂ consumption (${\stackrel{·}{\mathrm{V}}\mathrm{O}}_{\text{2}}$), CO₂ release (${\stackrel{·}{\mathrm{V}}\mathrm{CO}}_{\text{2}}$), and respiratory quotient (RQ) before the hyperinsulinemic-isoglycemic (HI) clamp (Basal) and during the steady-state clamp period (Insulin). (C and D) Glucose and lipid oxidation across the leg, as estimated from the leg ${\stackrel{·}{\mathrm{V}}\mathrm{O}}_{\text{2}}$ and ${\stackrel{·}{\mathrm{V}}\mathrm{CO}}_{\text{2}}$, before the HI clamp (Basal) and during the steady-state clamp period (Insulin). (E and F) Leg lactate release as calculated from the arteriovenous difference in plasma lactate concentration and the femoral arterial blood flow. (E) Time course of leg lactate release (data are estimated means ±95% confidence limits). (F) Leg lactate release before the HI clamp (Basal) and during the steady-state clamp period (Insulin). Linear mixed models were used to estimate between-treatment differences [(A), (B), (C), (D), and (F)]. Data presented as observed individual values with estimated means ±95% confidence limits, unless otherwise stated. n = 10.

Fig. 3.. The mtAO mitoquinone does not affect canonical insulin signaling but enhances insulin-dependent GLUT4 translocation under lipid overload.

(A) Schematic overview of the major signaling events modulating insulin-stimulated glucose uptake in human skeletal muscle. (B) Proximal insulin signaling, as determined by phosphorylation of Akt2 on Thr308 and Ser473 in whole-muscle homogenates. (C and D) Distal insulin signaling, as determined by phosphorylation of the Akt substrates GSK3β on Ser9 (C) and TBC1D4 on Thr642 (D) in whole-muscle homogenates. (E) GLUT4 translocation, as determined by GLUT4 protein abundance in plasma membrane protein fractions. In-gel stain-free technology was used as a loading control. Lipid, n = 8 (Pre-clamp) and n = 9 (End-clamp); Lipid + mtAO, n = 10. Representative blots (n = 2 biological replicates from each muscle biopsy sample for each patient). (F) Purity of the isolated plasma membrane protein fractions (used to determine GLUT4 translocation), as determined by immunoblot analysis of actin (cytosolic protein marker) and Na⁺/K⁺-ATPase subunit α1 (plasma membrane protein marker) in plasma membrane homogenates (PM) as compared with the corresponding whole-muscle homogenates (WM). PM and WM samples were obtained by pooling a given volume of each individual sample. (G) Pearson’s correlation between mtAO-induced changes in plasma membrane GLUT4 and leg glucose uptake under insulin stimulation. n = 9. Data [(B) to (E)] presented as observed individual values with estimated means ±95% confidence limits. Linear mixed models were used to estimate within- and between-treatment differences at End-clamp [(B) to (E)]. *Different from Pre-clamp (P < 0.05). n = 10, unless otherwise stated. Illustrations in (A) were created with BioRender.com.

Fig. 4.. The mtAO mitoquinone rescues palmitate-induced impairments in GLUT4 trafficking during insulin stimulation.

(A) Workflow to determine insulin-stimulated GLUT4 translocation in L6 myotubes. (B) Insulin-stimulated GLUT4 translocation in L6 myotubes treated with either vehicle BSA (basal, n = 5; insulin, n = 5), 50 nM mitoquinone (basal, n = 5; insulin, n = 5), 250 μM palmitate (basal, n = 5; insulin, n = 5), or 250 μM palmitate + 50 nM mitoquinone (basal, n = 6; insulin, n = 6). Values are normalized to vehicle at basal. (C) Workflow to determine insulin-stimulated GLUT4 translocation in mouse FDB muscle fibers. (D) Insulin-stimulated GLUT4 translocation in mouse muscle fibers treated with either vehicle 0.1% ethanol (basal, n = 10; insulin, n = 11) or 10 μM MitoPQ (basal, n = 17; insulin, n = 13). Fibers were pooled from three mice. Values are normalized to vehicle at basal. Data presented as observed values with means ±95% confidence limits. A one-way ANOVA was used to estimate between-treatment differences. *Different from “Basal” (P < 0.05). Illustrations in (A) and (C) were created with BioRender.com.

Fig. 5.. The mtAO mitoquinone reduces the muscle mitochondrial oxidative burden under lipid overload.

(A) Workflow to quantitatively analyze redox-sensitive proteins and lipid peroxidation in skeletal muscle biopsy samples. (B) Cytosolic oxidative burden as determined by protein abundance of peroxiredoxin 2 (PRDX2) dimers relative to monomers in whole-muscle homogenates. (C) Mitochondrial oxidative burden as determined by protein abundance of peroxiredoxin 3 (PRDX3) dimers relative to monomers in whole-muscle homogenates. (D) Overall peroxiredoxin oxidation as determined by protein abundance of oxidized/hyperoxidized peroxiredoxin (PRDX-SO_2/3) dimers in whole-muscle homogenates. (E) Whole-muscle lipid peroxidation as determined by protein abundance of 4-hydroxynonenal (4-HNE) adducts in whole-muscle homogenates. (F) Representative blots related to the experiments in human skeletal muscle biopsy samples (n = 2 biological replicates from each muscle biopsy sample for each patient). Coomassie blue staining was used as a loading control for PRDX-SO_2/3 and 4-HNE. (G and H) Cytosolic and mitochondrial oxidative burden in human myotubes treated with either vehicle BSA (control; n = 5), 250 μM palmitate (n = 5), or 500 μM palmitate. Relative PRDX2 and PRDX3 dimer abundance (dimer-to-monomer ratio) was normalized to cell plate–specific mean dimerization to adjust for N-ethylmaleimide treatment efficiency. Human data are presented as observed individual values with estimated means ±95% confidence limits. Linear mixed models were used to estimate within- and between-treatment differences (B to E). *Different from Baseline (P < 0.05). n = 10 for all measurements. Human muscle cell data are presented as observed values with means ±95% confidence limits. A one-way ANOVA was used to estimate between-treatment differences (H). Illustrations in (A) and (G) were created with BioRender.com.

Fig. 6.. The mtAO mitoquinone improves muscle mitochondrial bioenergetics under lipotoxic conditions.

(A) Workflow to determine ex vivo mitochondrial bioenergetics in human skeletal muscle, i.e., in permeabilized fiber bundles (PmFBs), using a substrate-inhibitor titration (SUIT) protocol reflecting normal (20 μM P-CoA) versus high (60 μM P-CoA) intramyocellular lipid conditions. PmFBs were preincubated in either decylTPP (control compound) or mitoquinone to evaluate the effects of mtAO on lipid-induced mitochondrial stress. (B) SUIT protocol used to simultaneously measure mitochondrial O₂ consumption (JO₂) and H₂O₂ emission (JH₂O₂) rates. Succinate (Succ); pyruvate (Pyr); malate (Mal); adenosine diphosphate (ADP); glutamate (Glut); cytochrome C (Cyt C); oligomycin (Omy). Data are means ± SEM. (C) Muscle mitochondrial content, as determined by citrate synthase (CS) activity in muscle homogenates obtained from separate portions of the biopsy specimen used for assessments of mitochondrial bioenergetics. Data are presented as individual values with estimated mean ± 95% confidence limits. (D) Maximal mitochondrial oxidative phosphorylation capacity (OXPHOS), oligomycin-induced leak respiration (LEAK_Omy), and OXPHOS efficiency [calculated as 1 − RCR = 1 − LEAK_Omy/OXPHOS (100)]. Data are presented as individual values with estimated means ±95% confidence limits. (E) Sensitivity of mitochondrial JO₂ to ADP. The apparent half-maximal effective concentration (EC₅₀) for ADP was determined using [agonist] versus response (three parameters) analysis in GraphPad Prism. Data are means ± SEM. (F) Maximal and submaximal mitochondrial H₂O₂ emission rates. Data are presented as individual values with estimated means ±95% confidence limits. (G) Sensitivity of mitochondrial JH₂O₂ to ADP. The apparent half-maximal inhibitory concentration (IC₅₀) for ADP was determined using [inhibitor] versus response (three parameters) analysis in GraphPad Prism. Data are means ± SEM. A linear mixed model [(D) and (F)] or one-way ANOVA [(E) and (G)] was used to estimate between-treatment differences. n = 10 for all measurements. Illustrations in (A) were created with BioRender.com.

Fig. 7.. Experimental framework of the study.

Conceptual and experimental framework providing mechanistic evidence that mitochondrial redox state influences muscle insulin action in humans. Illustration created with BioRender.com.