Comprehensive interrogation of human skeletal muscle reveals a dissociation between insulin resistance and mitochondrial capacity

Abstract

Insulin resistance and blunted mitochondrial capacity in skeletal muscle are often synonymous, however, this association remains controversial. The aim of this study was to perform an in-depth multifactorial comparison of skeletal muscle mitochondrial capacity between individuals who were lean and active (Active, n = 9), individuals with obesity (Obese, n = 9), and individuals with obesity, insulin resistance, and type 2 diabetes (T2D, n = 22). Mitochondrial capacity was assessed by ex vivo mitochondrial respiration with fatty-acid and glycolytic-supported protocols adjusted for mitochondrial content (mtDNA and citrate synthase activity). Supercomplex assembly was measured by Blue Native (BN)-PAGE and immunoblot. Tricarboxylic (TCA) cycle intermediates were assessed with targeted metabolomics. Exploratory transcriptomics and DNA methylation analyses were performed to uncover molecular differences affecting mitochondrial function among the three groups. We reveal no discernable differences in skeletal muscle mitochondrial content, mitochondrial capacity, supercomplex assembly, TCA cycle intermediates, and mitochondrial molecular profiles between obese individuals with and without T2D that had comparable levels of confounding factors (body mass index, age, and aerobic capacity). We highlight that lean, active individuals have greater mitochondrial content, mitochondrial capacity, supercomplex assembly, and TCA cycle intermediates. These phenotypical changes are reflected at the level of DNA methylation and gene transcription. The collective observation of comparable muscle mitochondrial capacity in individuals with obesity and T2D (vs. individuals without T2D) underscores a dissociation from skeletal muscle insulin resistance. Clinical trial number: NCT01911104.

NEW & NOTEWORTHY Whether impaired mitochondrial capacity contributes to skeletal muscle insulin resistance is debated. Our multifactorial analysis shows no differences in skeletal muscle mitochondrial content, mitochondrial capacity, and mitochondrial molecular profiles between obese individuals with and without T2D that had comparable levels of confounding factors (BMI, age, aerobic capacity). We highlight that lean, active individuals have enhanced skeletal muscle mitochondrial capacity that is also reflected at the level of DNA methylation and gene transcription.

INTRODUCTION

Skeletal muscle (SkM) insulin resistance is a primary defect in the pathology of type 2 diabetes (T2D) (1, 2). Insulin resistance in SkM has historically been characterized by impaired mitochondrial oxidative capacity (3). Dysregulated oxidation of lipid species is proposed to contribute to lipotoxicity in SkM and is associated with insulin resistance (4, 5). However, the apparent association between SkM insulin resistance and impaired mitochondrial capacity is not always evident. Endurance exercise training in individuals with T2D simultaneously improves insulin sensitivity and SkM mitochondrial capacity quantified in vivo as phosphocreatine (PCr) recovery rate (6, 7) and ex vivo by high-resolution respirometry in permeabilized SkM fibers (8). In contrast, thiazolidinediones improve insulin sensitivity in individuals with T2D without impacting SkM mitochondrial capacity quantified in vivo with PCr recovery rate (9, 10).

At a cross-sectional level, individuals with T2D have comparable in vivo SkM mitochondrial capacity (quantified as ATPmax) to obese controls (11). In addition, at least half of a cohort of individuals with T2D showed comparable ATPmax to sedentary nonobese controls (11), highlighting the heterogeneity in the pathophysiology of T2D. Regardless of T2D status, individuals with obesity have comparably lower succinate dehydrogenase enzyme activities (a spectrophotometric measure of mitochondrial activity in SkM tissue) compared with lean individuals (12). In contrast, oxidative pathway enzyme activities (citrate synthase, cytochrome-c oxidase) are lower in SkM tissue from individuals with T2D compared with lean and obese nondiabetic individuals (13). Furthermore, isolated mitochondria from T2D SkM showed modest reductions in ex vivo respiration normalized to citrate synthase activity compared with healthy individuals with obesity (14). However, another group found ex vivo respiration in permeabilized SkM fibers was not impaired in individuals with T2D compared with lean and obese controls when adjusted for citrate synthase activity (15). Although another report (3) found a hierarchical order of impaired capacity of the respiratory chain quantified by NADH:O₂ oxidoreductase activity in SkM from T2D, obese, and lean individuals. Discrepancies among these results may partially be attributed to the techniques employed for measurements of SkM mitochondrial capacities (enzymatic activities, ex vivo respiration, and in vivo ATP max/PCr recovery rates) and phenotypic differences [aerobic capacity, body mass index (BMI), and age] among the groups.

In the study, we leveraged previously deeply phenotyped samples from a previous study to interrogate differences in mitochondrial function in individuals who were lean and active (Active), individuals with obesity (Obese), and individuals with obesity, insulin resistance, and type 2 diabetes (T2D) (16). Importantly, in this cohort, Obese and T2D had comparable age, BMI, and aerobic capacity. The aim of the study was to perform an in-depth multifactorial assessment of mitochondrial capacity using mitochondrial content quantified by mitochondrial DNA (mtDNA) and citrate synthase activity, ex vivo respiration adjusted for mitochondrial content, supercomplex (SC) assembly, and tricarboxylic (TCA) cycle intermediates. In parallel, we performed exploratory multi-omics (transcriptomics and DNA methylation) of SkM to uncover molecular differences affecting mitochondrial function.

RESEARCH DESIGN AND METHODS

Human Participants

The samples and acquired data used in this study were part of a larger clinical trial comparing variations in endurance exercise response (6) under clinical trial number NCT01911104. We selected data and samples from 22 individuals with T2D (T2D, males = 13), 9 participants with obesity (Obese, males = 2), and 9 active participants (Active, males = 9) for this cross-sectional interrogation. Phenotypic differences of this cohort, including aerobic capacity measured by V̇o_2peak, body composition by dual-energy x-ray absorptiometry (DEXA), in vivo mitochondrial capacity measured by PCr recovery rate, and insulin sensitivity using hyperinsulinemic-euglycemic clamp have previously been published (16). These data are included in participant characteristics Table 1 for reference. The study protocol was approved by the AdventHealth Institutional Review Board and carried out in accordance with the Declaration of Helsinki. All participants provided written informed consent. At the time of enrollment individuals with T2D had to have a hemoglobin A1C (HbA1c) ≤ 8% if receiving glucose-lowering medication and between 6% and 8.5% if treated with diet alone. Participants ceased glucose-lowering treatment 2 wk before participation. All measurements were conducted following an overnight fast.

Table 1.

Clinical characteristics

Clinical Characteristics	T2D (n = 22)	Obese (n = 9)	Active (n = 9)
Age, yr	50 ± 8	46 ± 9#	34 ± 9*
Biological sex, F/M	9/13	7/2	0/9
BMI, kg/m2	35.7 ± 5.2	38.4 ± 8.4#	22.9 ± 2.2*
Body mass, kg	106.0 ± 20.3	100.9 ± 27.2#	77.5 ± 9.9*
Fat mass, kg	44.7 ± 14.1	47.3 ± 16.2#	16.2 ± 6.4*
Fat-free mass, kg	61.2 ± 11.1	53.6 ± 13.9	61.4 ± 6.7
HbA1c, %	7.3 ± 0.9&	5.6 ± 0.4	5.2 ± 0.4*
T2D duration, yr	4.6 ± 4.7
M-value, µmol/kgFFM/min	5.0 ± 2.8&	9.4 ± 5.5	13.0 ± 5.0*
Fasting glucose, mg/dL	182.2 ± 56.3&	94.4 ± 11.1	89.8 ± 4.9*
Fasting insulin, µIU/mL	15.7 ± 7.0	10.3 ± 4.1#	2.6 ± 0.9*
Free fatty acids, mmol/L	0.5 ± 0.2	0.5 ± 0.2	0.3 ± 0.1*
Total cholesterol, mg/dL	152.8 ± 21.1	162.9 ± 35.6	156.4 ± 24.3
LDL, mg/dL	90.2 ± 23.6	96.6 ± 24.5	90.2 ± 19.1
HDL, mg/dL	37.7 ± 8.5	44.5 ± 7.9	49.8 ± 7.7*
Triglycerides, mg/dL	125.8 ± 40.6	108.9 ± 86.9	81.9 ± 29.8
PCr recovery rate, 1/s	0.020 ± 0.007	0.023 ± 0.010#	0.041 ± 0.010*
V̇o2peak, L/min	2.29 ± 0.68	2.24 ± 0.86#	4.17 ± 0.64*
V̇o2peak, mL/kg/min	22.7 ± 5.6	22.1 ± 6.1#	54.3 ± 5.2*
V̇o2peak, mL/kg FFM/min	38.1 ± 5.3	40.5 ± 8.1#	67.8 ± 6.6*

Data are the means ± SD. T2D, type 2 diabetic; Obese, obese non-T2D; Active, lean active; F, female; M, male; BMI, body mass index; FFM, fat-free mass; HbA1c, hemoglobin A1C; M-value, whole body insulin sensitivity; LDL, low density lipoprotein; HDL, high density lipoprotein; PCr, phosphocreatine; V̇o_2peak, maximum rate of oxygen consumption. *Difference between Active and T2D. &difference between T2D and Obese. #difference between Active and Obese. One-way ANOVA, P < 0.05.

Skeletal Muscle Biopsy

SkM biopsies were performed in the vastus lateralis using the Bergström needle technique (17) under fasting conditions. Portions of SkM (150 mg) were snap frozen for Supercomplex assembly quantification, targeted metabolomics, and RNA and DNA extractions, and a fresh portion (10 mg) was used for the assessment of mitochondrial respiration. Certain assays were restricted due to SkM availability.

Citrate Synthase Activity

Citrate synthase activity in SkM tissue was assessed spectrophotometrically. Assay buffer consisting of Buffer Z [https://tinyurl.com/2kuthy7r (no BSA)] supplemented with 5',5'-dithiobis 2-nitrobenzoic acid, and acetyl-CoA was dispensed with permeabilizing agent Alamethicin and SkM lysate and incubated at 37°C for 5 min. The reaction was initiated by oxaloacetate. Absorbance was recorded at 412 nm every 1 min for 20 cycles. Citrate synthase activity was calculated using the Beer–Lambert law with a molar absorption coefficient of thionitrobenzoic acid (13.6 mM/cm).

mtDNA Copy Number

Total DNA was isolated from 10–20 mg of SkM tissue using DNeasy blood and tissue extraction kit (QIAGEN Inc, Valencia, CA). Relative amounts of mitochondrial DNA (mtDNA) and nuclear DNA were quantified by real-time quantitative PCR (18).

High-Resolution Respirometry

Mitochondrial respiration on isolated permeabilized SkM fibers using high-resolution respirometry (Oxygraph-2k, Oroboros Instruments, Austria) with a pyruvate-simulated glycolytic protocol and a fatty acid (FA) protocol has previously been reported in this cohort (16). We leveraged this previously published data to assess whether differences in SkM mitochondrial capacity were due to the content of mitochondria or intrinsic capacity by adjusting each respiratory state for the covariates mtDNA or CS activity in an analysis of covariance (ANCOVA) model. For full details of the respirometry protocol the reader is directed to the study by Carnero et al. (16). During each protocol, respiration of individual complexes was quantified; complex I supported LEAK (L_I or L_FAO), complex I supported oxidative phosphorylation (OXPHOS; P_I or P_I+FAO), complex I and II supported OXPHOS (P_I + II or P_{I + II+FAO}), and maximal electron transfer system capacity (E_I + II or E_{I + II+FAO}).

Supercomplex Immunoblots

Electron transport system (ETS) supercomplex assembly was analyzed in a subset of participants (T2D, n = 17; Obese, n = 4; Active, n = 6) by blue native polyacrylamide gel electrophoresis as previously described (19–21). Previously flash-frozen vastus lateralis tissue was homogenized on ice in a sucrose buffer (250 mM sucrose, 20 mM imidazole/HCl, pH 7.0). Samples were centrifuged at 10,000 g for 10 min and the resulting pellet (containing mitochondria and other membranes and organelles) was resuspended in Blue Native (BN)-PAGE extraction buffer [50 mM imidazole/HCl pH 7.0, 50 mM NaCl, 5 mM 6-aminohexanoic acid, 1 mM EDTA with 1.5% digitonin (experimentally determined, final digitonin to tissue ratio of 1:10 wt/wt)] for 30 min. Samples were cleared by centrifugation at 14,000 g for 30 min. Supernatants were loaded with 5% glycerol and a 1:4 ratio of Coomassie Blue G-250:digitonin onto large 3–13% gradient gels. Gels were run and then transferred and blotted with overnight incubation of the following previously validated primary antibodies (22–24) dilutions: 1:20,000 complex I [NADH dehydrogenase (ubiquinone) 1α subcomplex subunit 9, mitochondrial (NDUFA9)] (459100; Thermo Fisher Scientific, Waltham, MA, USA), 1:40,000 complex II (Fp) succinate dehydrogenase complex, subunit A, flavoprotein variant (459200; Thermo Fisher Scientific), 1:20,000 complex III (ubiquinol‐cytochrome c reductase core protein II) [Ab14745; MitoSciences (Abcam, Cambridge, United Kingdom)]; 1:20,000 complex IV (subunit I) (459600; Thermo Fisher Scientific), and 1:50,000 complex V (ATP synthase subunit a, mitochondrial) (Ab14748; MitoSciences). Anti-mouse IgG HRP secondary antibody (1:5,000; Promega) was added for 1 h before visualization. Membranes were exposed to Immobilon Classico Western HRP substrate (Millipore) and captured on X-ray film. ETS supercomplexes were analyzed based on their banding pattern, as previously confirmed (25). Densitometry analysis was performed using Image J software. Supercomplex (SC) and monomers were normalized to nuclear-encoded Complex II which does not participate in mammalian ETS SC formation (20).

Tricarboxylic Cycle Intermediates

Tricarboxylic (TCA) cycle intermediates (TCAis) were measured by targeted LC-MS/MS (26). Samples were spiked with heavy isotope-labeled internal standards and derivatized with O-benzylhydroxylamine (OBA) and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide before being separated on a Waters Acquity UPLC BEH. Quantitation of organic acids (OA) was performed by single reaction monitoring using a Thermo Scientific Quantiva triple quadrupole mass spectrometer (Thermo Fisher Scientific, San Jose, CA). The raw data were processed using Xcalibur 3.0.

RNA Sequencing

RNA-Seq on SkM tissue was performed in a subset of participants from each group (n = 6 per group, clinical characteristics provided in Supplemental Data). RNA was extracted from 30 to 50 mg of SkM tissue using RNeasy Fibrous Tissue Kit (Qiagen, Valencia, CA). Following polyA selection, library preparation, and quality control, sample libraries underwent mRNA sequencing, with an average depth of 20 million paired-end reads per sample using the Illumina Novaseq 6000. The adaptor was trimmed with Cutadapt and low-quality reads were filtered out before aligning with STAR to the reference genome hg38. Raw counts were calculated with featureCounts. Cufflinks were used to generate normalized read counts per gene and isoform in terms of FPKM values.

Reduced Representation Bisulfite Sequencing

Reduced representation bisulfite sequencing (RRBS) on SkM tissue was performed in a subset of participants from each group (n = 3 per group, clinical characteristics provided in Supplemental Data). Total DNA was isolated from 10 to 20 mg of SkM tissue using DNeasy blood and tissue extraction kit (QIAGEN Inc, Valencia, CA). RRBS library was generated from ≥1.5 μg genomic DNA at Novogene. Libraries were sequenced as paired-end 150 bp on Novaseq 6000 sequencer at an average out per sample ≥10 Gb.

Quantification and Statistical Analyses

To compare the differences between groups for clinical characteristics and measures of mitochondrial capacity/content an ANOVA model with Tukey post hoc adjustment was applied and significance was set at P < 0.05. Respirometry data were analyzed using an ANCOVA with citrate synthase or mtDNA used as a covariate followed by Tukey post hoc comparisons with significance set at P < 0.05. Statistical analyses were performed in JMP.

Bioinformatics

Transcriptomics.

The data were first log₂-transformed and then z-scaled for transformation and standardization, respectively. Random forest classification was performed with two iterations to identify the top 953 genes ranked for classification (27). Principal component analysis (PCA) was performed with factoextra and ggplot2 using the top 953 ranked genes to assess the transcriptional diversity between groups. Differential expressed genes (DEGs) were analyzed with limma R package for between-group differences (28) and adjusted for sex, age, and BMI. Gene-ontology (GO) over-representation analysis on DEGs (P < 0.05) was conducted using clusterProfiler, using genes detected as a background list and an false discovery rate (FDR) of <0.005 (29). Redundancy of GO terms from DEGs was reduced with GOSemSim (30).

DNA methylation.

Initially, Trim Galore, a wrapper tool around Cutadapt and FastQC, was applied to FastQ files for quality control and adapter trimming. Once trimmed, Bismark was performed for read mapping and methylation calling (31). Cytosines in the reference sequenced as thymines are labeled unmethylated, and those that remained as cytosines are labeled methylated. Differentially methylated site (DMS) analysis was performed with R package methylKit (32) and adjusted for sex, age, and BMI. Annotation of the genomic location of differentially methylated sites was achieved with the R package ChIPseeker (33). Correlation analysis between methylation % and gene expression (log₂) was conducted using the rcorr() function from Hmisc R package.

RESULTS

Clinical Characteristics

Full participant characteristics are displayed in Table 1 and have been previously reported (16). For contextualization of results, differences in phenotypic differences are detailed in this section. Active had a lower body mass (kg) in comparison with Obese (P < 0.05) and T2D (P < 0.01), which was due to lower fat mass (kg) (P < 0.001; Table 1). There were no body composition differences between T2D and Obese. By design, T2D had significantly reduced insulin sensitivity (M-value) in comparison to both Obese (P < 0.05) and Active (P < 0.001; Table 1). There were no differences in M-value between Active and Obese (Table 1). Aerobic capacity measured by VO₂peak (L/min) was greater in Active compared to Obese and T2D (P < 0.001) with no differences between Obese and T2D (Table 1). This relationship persists when V̇o_2peak is normalized to body mass (kg) or lean body mass (kg) (Table 1). When the clinical characteristics were adjusted for sex as a covariate all significant differences remained except M-Value was no longer statistically greater in Obese compared with T2D.

Mitochondrial Content

Citrate synthase activity was not significantly different between Obese and T2D but was significantly greater in Active compared with T2D (P < 0.01; Fig. 1A). mtDNA copy number was not significantly different between Obese and T2D but was greater in Active compared with both Obese and T2D (P < 0.001; Fig. 1B). These significant differences remained after the values were adjusted for sex.

Figure 1.

Differences in skeletal muscle mitochondrial content and ex vivo mitochondrial respiration adjusted for mitochondrial content between Active, Obese, and type 2 diabetes (T2D). A: citrate synthase activity in skeletal muscle tissue [Active (F/M = 0/7), Obese (F/M = 3/1), T2D (F/M = 8/9)]. B: mitochondrial DNA (mtDNA) copy number in skeletal muscle tissue [Active (F/M = 0/8), Obese (F/M = 5/2), T2D (F/M = 8/9)]. Data analyzed with ANOVA model with Tukey post hoc adjustment. Mitochondrial respiration quantified by high resolution respirometry in permeabilized skeletal muscle fibers during a glycolytic protocol, consisting of leak (L_I), complex I supported OXPHOS (P_I), complex I+II supported OXPHOS (P_I+II), maximal electron transfer system capacity (E_I+II) adjusted for citrate synthase activity [Active (F/M = 0/7), Obese (F/M = 3/1), T2D (F/M = 8/9)] (C) and mtDNA [Active (F/M = 0/8), Obese (F/M = 5/2), T2D (F/M = 8/6)] (D). Mitochondrial respiration during a fatty-acid oxidative protocol, consisting of leak (L_FAO), complex I supported OXPHOS (P_I+FAO), complex I+II supported OXPHOS (P_I+II+FAO), maximal electron transfer capacity (E_I+II+FAO) adjusted for citrate synthase activity [Active (F/M = 0/6), Obese (F/M = 3/1), T2D (F/M = 8/9)] (E) and mtNDA [Active (F/M = 0/7), Obese (F/M = 5/2), T2D (F/M = 8/6)] (F). An analysis of covariance (ANCOVA) was used to analyze data respiration data with mtDNA or citrate synthase activity used as a covariate followed by Tukey post hoc comparisons. Data are presented as means ± SD overlayed with individual values. ***P < 0.001, **P < 0.01, *P < 0.05. Active, lean active; Obese, obese non-T2D.

Ex Vivo Skeletal Muscle Mitochondrial Capacity

We previously showed that Active had greater O₂ flux compared with Obese and T2D during respiratory states: P_I, P_I + FAO, P_I + II, P_{I + II+FAO}, E_I + II, E_{I + II+FAO} (16). There were no differences between Obese and T2D for any of these respiratory states. Both citrate synthase activity and mtDNA are considered confounding factors to mitochondrial capacity given greater mitochondrial content is associated with greater mitochondrial respiration (3). Regression graphs between respiratory states (E_I + II and E_{I + II+FAO}) and mitochondrial content indices (citrate synthase activity and mtDNA), however, revealed that only mtDNA had a linear relationship with capacity (respiration) (Supplemental Fig. S1). The differences between Active and Obese or T2D ex vivo glycolytic- and fatty-acid-supported mitochondrial capacities remained when respiration values were normalized to citrate synthase activity and mtDNA (Fig. 1, C–F) and there were still no significant differences between Obese and T2D (Fig. 1, C–F). When respiration values were adjusted for citrate synthase activity and sex, significant differences between Active and Obese or T2D were retained and there were still no differences between Obese and T2D. When respiration values were adjusted for mtDNA and sex, significant differences between Active and Obese or T2D were still retained and there were no significant differences between Obese and T2D except with P_I respiratory state (P < 0.05).

Supercomplex Formation

Complex IV (CIV)-containing SCs were significantly greater in Active compared with Obese (P < 0.01) and T2D (P < 0.001; Fig. 2, A and B). Complex IV (CIV)- and Complex III (CIII)-containing SCs were higher in Active compared with T2D (P < 0.01; Fig. 2, A and B). Unadjusted SC are shown in Supplemental Fig. S2. When SC formation was adjusted for sex, significant differences between Active and Obese or T2D were retained and there were still no differences between Obese and T2D.

Figure 2.

Differences in skeletal muscle supercomplex assembly and tricarboxylic (TCA) intermediates between Active, Obese, and type 2 diabetes (T2D). Mitochondrial Supercomplex and monomers measured by BN-PAGE and quantified with antibodies; anti-NADH dehydrogenase (ubiquinone) 1 alpha subcomplex subunit 9, mitochondrial (NDUFA9; complex I), antiflavoprotein (complex II), anti-ubiquinol-cytochrome c reductase core protein II (UQCRC2; complex III), anticomplex IV subunit I (complex IV) and anti-ATP synthase subunit alpha, mitochondrial (ATP5a; complex V). Representative images of Active, T2D, and Obese are shown, white spaces indicate different gel lanes (A). Expression of the mitochondrial supercomplexes and monomers relative to complex II monomer [Active (F/M = 0/6), Obese (F/M = 2/2), T2D (F/M = 8/9)] (B). Quantification of the TCA cycle intermediates using target metabolomics [Active (F/M = 0/9), Obese (F/M = 4/2), T2D (F/M = 9/12)] (C). Data were analyzed with ANOVA model with Tukey post hoc adjustment. Data are presented as means ± SD overlayed with individual values. ***P < 0.001, **P < 0.01, *P < 0.05. Active, lean active; Obese, obese non-T2D.

TCAi

Citrate, succinate, fumarate, and malate were all significantly greater in Active (n = 9) compared with Obese (n = 6; P < 0.001) and T2D (n = 19; P < 0.001; Fig. 2C). α-Ketoglutarate was significantly greater in Active compared with Obese only (P < 0.05; Fig. 2C). There were no differences in any of the TCAis between Obese and T2D (Fig. 2C). After adjustment for sex there were no changes in the observed significant differences. Apart from α-ketoglutarate, these differences remained after the data were adjusted for citrate synthase activity (Supplemental Fig. S3A). Differences also remained after the data were adjusted for mtDNA copy number with the exceptions of α-ketoglutarate and succinate (Supplemental Fig. S3B).

Transcriptomic Analyses

Principal component analyses of the top 953 ranked contributing genes in the SkM showed a transcriptional disparity between Active in comparison with T2D and Obese (Fig. 3A), with considerable overlap between T2D and Obese. This separation in skeletal SkM was reflected by the number of differentially expressed genes (DEGs; P < 0.05) being much higher in T2D versus Active (489 downregulated, 2156 upregulated) and Obese versus Active (457 downregulated, 682 upregulated) than in T2D versus Obese (276 downregulated, 593 upregulated; Supplemental Tables S1–S3). Pathway analysis showed that Active had an upregulation of genes related to respiration/mitochondrial capacity and mitochondrial structure compared with both Obese and T2D (Fig. 3B, Supplemental Table S4). In contrast, T2D had an upregulation of genes related to protein degradation and catabolic processes compared with Active (Fig. 3B, Supplemental Table S4); whereas, Obese had an upregulation of genes related to cell-matrix and junction assembly compared with Active (Fig. 3B, Supplemental Table S4). When comparing T2D against Obese, there were no significant GO terms downregulated in T2D, whereas, protein targeting and localization were upregulated in T2D (Fig. 3B, Supplemental Table S4).

Figure 3.

Transcriptomic differences between Active, Obese, and type 2 diabetes (T2D) skeletal muscle. Principal component analysis of 953 top-ranked contributing genes in skeletal muscle shows transcriptional disparity between Active in comparison with T2D and Obese (A). Selected gene ontology biological processes from over-representation analysis reveal Active have an upregulation of mitochondrial-related processes, whereas differences between Obese and T2D are driven by protein targeting and localization (B). n = 6 from each group. Active, lean active; Obese, obese non-T2D.

MitoCarta Analysis

We used the MitoCarta3.0 list (34) to identify which transcriptional aspects of mitochondrial capacity, structure, and processes differ between Active, Obese, and T2D. Out of a total of 1,136 MitoCarta genes, 169 were downregulated in Obese compared with Active and 143 were downregulated in T2D compared with Active (Supplemental Table S5). Downregulated genes in Obese and T2D were related to oxidative phosphorylation (OXPHOS) subunits and assembly factors, mitochondrial translation, metals and cofactors, lipid and carbohydrate metabolism, and TCA cycle (Fig. 4). A subset of MitoCarta genes (n = 65) were upregulated in T2D compared with Active, which were largely related to autophagy, protein homeostasis and nucleotide metabolism (Fig. 4). Eighty MitoCarta genes were upregulated in T2D compared with Obese, which were related to aspects of various metabolic processes including, lipid metabolism, amino acid metabolism, FA oxidation, nucleotide metabolism, and in some instances OXPHOS subunits and assembly factors (Fig. 4). Obese had very few genes upregulated in comparison to Active (10 genes) and T2D (4 genes; Supplemental Table S5).

Figure 4.

Heatmaps of MitoCarta differentially expressed genes (DEGs) between Active, Obese, and type 2 diabetes (T2D) skeletal muscle. Active had an upregulation of genes related to oxidative phosphorylation (OXPHOS) subunits and assembly factors, mitochondrial translation, metals and cofactors, lipid and carbohydrate metabolism, and tricarboxylic (TCA) cycle in comparison with Obese and T2D. Obese had downregulation of genes related to OXPHOS subunits in addition to FA oxidation in comparison with T2D. Data are presented as standardized average (n = 6 for each group) scores for each group. Active, lean active; Obese, obese non-T2D.

DNA Methylation Analyses

We determined how many MitoCarta DEGs contained differentially methylated guanine residues (CpG) sites (DMS) for each comparison (Fig. 5A, Supplemental Tables S6). Between 25% to 70% of MitoCarta DEGs had DMSs (Fig. 5A), with the highest absolute amount of DMS pertaining to downregulated genes in Obese compared with Active. We next ran correlation analyses to determine which DMS located at MitoCarta DEGs were correlated to their gene expressions for each group comparison (Supplemental Table S7). Although each group comparison had similar percentages (25–35%) of DMSs located at MitoCarta DEGs significantly correlated to gene expressions (Supplemental Fig. S4), only 2% of DMSs located at MitoCarta DEGs upregulated in T2D compared with Obese were correlated to gene expression (Fig. 5B). There were only two DMSs located at downregulated MitoCarta DEGs in T2D compared with Obese and neither correlated to gene expression.

Figure 5.

Differences in DNA methylation between Active, Obese, and type 2 diabetes (T2D) skeletal muscle in relation to differentially expressed MitoCarta genes. The number of mitocarta differentially expressed genes (DEGs) for each comparison splut by whether it contains differentially methylated sites (DMSs) (A). Waterfall plot showing the correlation of DMS to MitoCarta DEGs and their significance (P < 0.05) highlighting for the T2D vs. obese upregulated comparison, there are only two DMSs correlated to MitoCarta DEGs (B). Genomic location of DMSs that are correlated to differentially expressed MitoCarta DEGs across all samples shows the majority of correlated methylation sites occur in or near the promoter region regardless of whether it is a positive or negative correlation (C). Correlation between DMSs and gene expression for key mitochondrial genes ATP5PD and MFN2 shows significant negative correlations (D). A total of 9 participants (n = 3 for each group) were used for the analysis of DNA methylation data. Active, lean active; Obese, obese non-T2D.

Independent of directionality of correlations to gene expressions, the majority of the DMSs (79%) were located in either the promoter region or at the first intron (Fig. 5C). There were a number of DMSs negatively correlated with the expressions of genes encoding OXPHOS subunits and assembly factors that generally showed lower gene expressions and higher CpG methylation levels in Obese and T2D compared with Active (Fig. 5D). Mitochondrial fusion gene mitofusin 2 (MFN2) had two promoter regions that were inversely correlated with gene expressions (one displayed in Fig. 5D) and displayed lower gene expression and higher CpG methylation in Obese and T2D.

DISCUSSION

Insulin resistance and blunted mitochondrial capacity in SkM are often synonymous, however, this association remains controversial with previous research reporting conflicting results. Employing a comprehensive multifactorial analysis of SkM mitochondrial capacity, we demonstrate that obese individuals with and without T2D have comparable mitochondrial capacities underscored by similarities in mitochondrial content, ex vivo respiration adjusted for mitochondrial content, supercomplex assembly, and levels of TCA cycle intermediates. This lack of differences extends to the in vivo level where we previously demonstrated comparable PCr recovery rates between Obese and T2D (16). A critical aspect of our study is that the Obese and T2D cohorts had similar levels of confounding factors such as BMI, age, and aerobic capacity, which are known to impact mitochondrial capacity. Compared with sedentary individuals with obesity with and without T2D, we demonstrate that lean active individuals with enhanced aerobic capacity have numerous aspects of superior SkM mitochondrial capacity quantified by mitochondrial content, ex vivo respiration adjusted for mitochondrial content, supercomplex assembly, and levels of TCA cycle intermediates. By adjusting ex vivo respiration by mitochondrial content, we highlight that differences in overall mitochondrial capacity are in part due to the enhanced intrinsic capacity of the mitochondria. These findings are paralleled by a robust upregulation of genes encoding crucial aspects of mitochondrial capacity (i.e., OXPHOS subunits and assembly factors, mitochondrial fusion, lipid and carbohydrate metabolism, and TCA cycle). Furthermore, genes regulating OXPHOS subunits and mitochondrial fusion displayed reduced DNA methylation in the promoter regions correlating to enhanced gene expression.

In agreement with previous research, we show that individuals with obesity with and without T2D have comparable SkM mitochondrial capacities when measured in vivo or ex vivo (11, 15). This contrasts with previous reports showing reduced SkM mitochondrial capacity in individuals with T2D compared with BMI- and age-matched normo-glycemic controls when measured in vivo (35–37) or ex vivo (14, 35). The discrepancies among these studies can be explained by aerobic capacity not being controlled for (14, 37) or the cohort investigated being overweight – rather than obese – and having higher aerobic capacities compared to our cohorts (35). As expected, lean active individuals had greater SkM mitochondrial capacity which was likely due to them being physically active and having greater aerobic capacity (38). Interestingly, ex vivo respiration remained higher in Active and comparable between Obese and T2D, when the results were adjusted for mitochondrial content measured by two indices. This contrasts previous research (39) and indicates that greater SkM respiration in the Active group is not entirely due to greater mitochondrial content.

Research has demonstrated that the multiprotein complexes of the ETS do not simply exist as monomeric units in the mitochondrial inner membrane but instead can form higher-order SC structures that contain varying proportions of CI, CIII, and CIV. The formations of such SCs permit greater electron flow through the ETS, thus maximizing capacity and efficiency for ATP production (22, 40, 41). The Active group had greater SC formation, which aligns with previous research showing increases in SC assembly following endurance exercise training in previously sedentary individuals (22). Our findings, however, are contrary to a previous report demonstrating reduced SC formations in individuals with T2D compared with BMI-matched nondiabetic group (20). This previous study however used a different SkM group (rectus abdominus), participants with a significantly higher BMI and participants undergoing a hypocaloric diet for 3 wk before bariatric surgery. The greater SC formation in Active supports the idea that mitochondrial capacity can be improved without changes in mitochondrial content and may be due to improved stoichiometry of SC formation.

The TCA cycle is the final common pathway for the oxidation of substrates and the major source for the generation of reducing equivalents for oxidative phosphorylation. In parallel with the previous results, Active had greater TCAis and there were no comparable differences in TCAi between Obese and T2D. Most of the differences were retained after adjustment for mitochondrial content, suggesting individual mitochondria have greater TCAis in Active. Scientific investigations of TCAis in human SkM are few in number. Succinate increases in human SkM of sedentary individuals following endurance training (42), aligning with our Active group. In mouse models, TCAis are reduced in lipid-induced insulin-resistant SkM (4) and SkM of genetically obese mice but to a lesser extent in genetically diabetic mice (43). Together this suggests that TCAis are reduced in sedentary individuals compared to Active individuals.

Previous work has suggested a downregulation of genes related to OXPHOS in SkM of individuals with T2D (44); however, this has not been recapitulated in larger clinical cohorts (45, 46). In agreement, we show that global transcriptomic differences between Obese and T2D SkM are not indicative of mitochondrial-related processes but rather appear to be related to protein targeting. OXPHOS gene expressions have been previously correlated with V̇o_2peak (44); thus, our observed upregulation of genes related to mitochondrial capacity in the Active group compared with Obese and T2D are likely due to their enhanced aerobic capacity – although we cannot discount the impact of obesity in our findings.

When a targeted approach was used to detect differences in mitochondria-specific genes [MitoCarta 3.0 (34)], it illustrated that Active had specific upregulation of numerous genes related to OXPHOS subunits and assembly factors, mitochondrial fusion and fission, mitochondrial translation, lipid, and carbohydrate metabolism in comparison with Obese and T2D. Mitochondrial morphology, which is integral to optimal function, is maintained by tightly regulated fusion and fission processes. Fission divides mitochondria and acts as a quality control for damaged mitochondria to undergo mitophagy (47). Fusion binds mitochondria, resulting in mitochondrial elongation and enhanced cristae cross-sectional area. This promotes greater OXPHOS protein distribution throughout the cristae and allows for greater capacity for energy production (48). Endurance exercise training is known to enhance regulators of mitochondrial fusion, MFN2 and OPA1 (49, 50). The upregulation of the fusion regulators [MFN2 and optic atrophy tpye 1 (OPA1)] in our Active group, concurrent with upregulated OXPHOS subunits and assembly factors indicates mitochondria morphology is optimally maintained and can contribute to enhanced function and oxidative phosphorylation. Interestingly, Obese, rather than T2D, had the greatest number of downregulated MitoCarta genes. Obesity and T2D have been inextricably linked, however, the SkM molecular landscape can vary considerably and is underpinned by the pathology itself. Obesity without T2D is characterized by expanded white adipose tissue that serves to store and buffer excess lipid. The lipid overflow hypothesis suggests when white adipose tissue has limited expansion and lipid storage capabilities, ectopic lipid accumulation occurs in peripheral tissues such as SkM (51). Indeed, individuals with T2D have greater SkM intramuscular lipids compared to individuals with Obesity without T2D (52). Ectopic SkM lipid storage can contribute to the development of insulin resistance and the progression of T2D (53). Interestingly, individuals in the T2D group had an upregulation of MitoCarta genes related to FA oxidation, lipid metabolism, and a subset of OXPHOS subunits which could indicate a compensatory mechanism to deal with the influx of lipids in SkM compared with the Obese group. Despite these transcriptomic differences, there were no parallel and discernable differences in mitochondrial capacity (respiration) in the T2D group. The differential pathobiology of impaired mitochondrial capacity observed in both the Obese and T2D groups (compared with the Active group) warrants further investigation.

DNA methylation is an epigenetic modification of DNA that can impact the regulation of gene transcription in human SkM (54). Only 2% of DMS located at MitoCarta DEGs in the T2D compared with the Obese comparison correlated with gene expression. This is in stark contrast to 25–35% of DMS located at MitoCarta DEGs significantly correlated to gene expression for the remaining group comparisons (i.e., T2D vs. Active, Obese vs. Active, etc.), thus, suggesting that differential DNA methylation is not contributing to the observed differences in transcriptional regulation of mitochondrial capacity between T2D and Obese. We found that independent of the directionality of correlations to gene expressions, the majority of DMSs (79%) were located in either the promoter region or at the first intron which is near the transcription start site (TSS) (55). We observed increased gene expression with increased methylation of DNA in almost half the instances. The repressive role of promoter DNA methylation on gene transcription has long been established (56). It is worth noting that transcription factors have varying degrees of sensitivity to CpG methylation with only 22% exhibiting decreased binding to their motifs with hypermethylation (57). Therefore, it is plausible that the DMS positively correlated to gene expressions do not have reduced transcription factor binding and have increased transcription through other epigenetic modifications. A number of critical mitochondrial-related genes linked to OXPHOS subunits and mitochondrial fusion had a negative correlation between methylation and expression that were upregulated in the Active group. Together, this indicates that major components of mitochondrial capacity that are enhanced in the Active group are regulated at the level of DNA methylation.

The Obese group had normal HbA1c (5.7%) and greater insulin-stimulated glucose uptake (M-value) compared with the T2D group, which was comparable to the Active group. Therefore, our T2D group was insulin resistant compared with our BMI-, aerobic capacity-, and age-similar controls. It is noteworthy that T2D duration was on average 4.6 yr, which along with the elevated HbA1c 7.3 ± 0.9 %, places them at a reduced probability of remission (58, 59) (i.e., they are at a “later stage” of T2D progression). Therefore, SkM insulin resistance and reduced glycemic control do not further exacerbate mitochondrial dysfunction in this group of individuals with T2D. This supports previous work reporting that individuals with low birth weight, who have insulin resistance, have comparable mitochondrial capacity measured with SkM PCr recovery and OXPHOS gene expression to insulin-sensitive normal birth weight controls (60).

By leveraging a comparison to an Active lean group, we were able to highlight molecular, stoichiometry, and functional differences that contribute to superior SkM mitochondrial capacity. We hypothesize that this superior mitochondrial capacity is due to their enhanced aerobic capacity (38). Future research should include a lean sedentary control group to assess if obesity contributes to impaired mitochondrial capacity rather than aerobic capacity.

We aimed to recruit both sexes in all cohorts; however, there was an uneven sex distribution, with the Active group comprised of only males and the majority of individuals in the Obese group being females. Females tend to have superior mitochondrial capacity than males (61). Despite this finding, our Obese group had similar mitochondrial capacity to T2D and the lowest expression of mitochondrial genes. When mitochondrial variables were adjusted for sex using an ANCOVA, there were still no significant differences between the Obese and T2D group for all the variables measured except for respiratory state P_I when also adjusted for mtDNA. We therefore conclude that after adjustment for sex, Obese and T2D still have comparable mitochondrial capacity whereas Active largely retain their superior mitochondrial capacity.

In summary, our highly controlled, in-depth multifactorial analysis dissociates SkM mitochondrial capacity from insulin resistance with robust support from ex vivo respiration, targeted metabolomics, SC assembly, global transcriptomics, and DNA methylation. We further highlight that lean, active individuals have enhanced SkM mitochondrial capacity compared with sedentary obese individuals with and without T2D at the ex vivo, metabolomics and global transcriptomics levels that are linked in part to differential DNA methylation.

DATA AVAILABILITY

Raw and processed RNA-Seq and DNA methylation data sets generated and analyzed during the current study are available in the NCBI GEO repository (GSE196387). All other data generated and 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.

SUPPLEMENTAL DATA

All Supplemental material is available at https://doi.org/10.6084/m9.figshare.22779239.

GRANTS

This work is supported by a grant from the American Diabetes Association (#7–13-JF-53). Dr. Lauren Sparks is the guarantor of this manuscript.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

K.L.W., S.R.S., M.J.W., and L.M.S. conceived and designed research; K.L.W., M.F.P., Y.S., R.X.Y., P.G., N.K., H.C., and D.A.P. performed experiments; K.L.W., M.F.P., Y.S., G.Y., F.G.D., R.B.V., P.G., A.D., N.K., F.Y., H.C., and D.A.P. analyzed data; K.L.W., M.F.P., Y.S., G.Y., R.B.V., A.D., D.A.P., M.-E.H., S.J.G., S.R.S., M.J.W., and L.M.S. interpreted results of experiments; K.L.W., M.F.P., and G.Y. prepared figures; K.L.W. drafted manuscript; K.L.W., M.F.P., Y.S., G.Y., A.D., N.K., M.-E.H., S.J.G., S.R.S., M.J.W., and L.M.S. edited and revised manuscript; K.L.W., M.F.P., Y.S., G.Y., F.G.D., R.X.Y., R.B.V., P.G., A.D., N.K., F.Y., H.C., D.A.P., M.H., S.J.G., S.R.S., M.J.W., and L.M.S. approved final version of manuscript.