figure a


Skeletal muscle accounts for almost half of our body weight and, as such, is responsible for up to 75% of insulin-stimulated glucose disposal after a meal [1, 2]. In fact, blood glucose uptake into muscle and its subsequent conversion to glycogen is a major determinant of insulin sensitivity [3, 4]. Given the importance of skeletal muscle for glucose management, as well as its vital role in insulin action, impairments in skeletal muscle health in people with type 1 diabetes diminishes their ability to mitigate dysglycaemic burdens, promotes the development and progression of complications [5, 6] and, ultimately, expedites the accelerated physical disability that characterises diabetes [7]. However, our understanding of the impact of this chronic disease on the health/quality of human skeletal muscle is extremely limited.

Mitochondria play a fundamental role in energy metabolism and ATP production throughout the body. Based on their role in fuel oxidation, mitochondria can also generate considerable amounts of reactive oxygen species (ROS). It remains controversial whether impaired skeletal muscle mitochondrial respiratory function contributes to insulin resistance and dysglycaemia in obesity and type 2 diabetes [8]. However, a consistent observation has been increased mitochondrial ROS in obese/type 2 diabetes relative to lean control individuals [8]. In addition, using 31P-magnetic resonance spectroscopy (MRS), people with type 1 diabetes have been shown to exhibit slowed post-exercise ATP resynthesis [9,10,11]; the underlying mechanism(s) for this impairment is currently unknown and would require a direct and comprehensive exploration of site-specific mitochondrial respiratory function. Moreover, the effect of type 1 diabetes on skeletal muscle ultrastructure, last reported in 1977 [12], has not been re-examined since the introduction of more aggressive insulin therapies (e.g. basal and fast-acting insulins, insulin analogues and pump therapy), and its effect on muscle mitochondrial ROS production remains to be elucidated.

Here, we hypothesised that otherwise healthy, young adults with type 1 diabetes would display site-specific decrements in skeletal muscle mitochondrial respiration, increased ROS production and altered muscle ultrastructure relative to non-diabetic controls.



Twelve untrained participants with type 1 diabetes were recruited and closely matched with twelve participants without diabetes (control group) for age, sex, BMI and level of physical activity. Demographics (Table 1) and participant number was determined by power calculations from our previous studies in human muscle [13, 14]. All participants with type 1 diabetes used insulin (insulin pump or multiple daily injections) and reported no complications. Prior to giving informed consent, all participants were given oral and written information about the experimental procedures. All procedures were approved by the Research Ethics Board at York University (REB number e2013-032) and conformed to the Declaration of Helsinki.

Table 1 Demographics of participants

Study design

Participants reported to York University in the early morning or early afternoon and were instructed to consume a standardised meal 1.5–2 h prior to their visit. Participants with type 1 diabetes were also instructed to continue their habitual use of insulin. Upon arrival, body mass and height measurements were taken to determine BMI, and a blood sample was obtained using venepuncture, for HbA1c analysis. Skeletal muscle samples were then obtained from the vastus lateralis muscle by a microbiopsy percutaneous needle, as described previously [13], and used for mitochondrial bioenergetic analyses, transmission electron microscopy (TEM), histological analysis and western blotting, as described below.

Mitochondrial bioenergetics

All mitochondrial bioenergetic experiments were performed in vitro using permeabilised muscle fibres prepared from fresh muscle samples that were immediately placed in ice-cold BIOPS containing (in mmol/l): 50 MES, 7.23 K2EGTA, 2.77 CaK2EGTA, 20 imidazole, 0.5 dithiothreitol (DTT), 20 taurine, 5.77 ATP, 15 PCr and 6.56 MgCl2.6H2O (pH 7.1), as previously described in detail [13].

Mitochondrial respiration

Using the Oxygraph-2k (Oroboros Instruments, Innsbruck, Austria), high-resolution measurements of mitochondrial oxygen consumption were conducted in permeabilised myofibres placed in 2 ml of respiration medium MiR05 [15], containing 0.5 mmol/l EGTA, 10 mmol/l KH2PO4, 3 mmol/l MgCl2.6H2O, 60 mmol/l K-lactobionate, 20 mmol/l Hepes, 20 mmol/l taurine, 110 mmol/l sucrose and 1 mg/ml fatty acid-free BSA (pH 7.1), at 37°C. Experiments were conducted at an initial oxygen concentration of 350–375 μmol/l with constant stirring at 750 rev/min. Respiration medium was supplemented with 20 mmol/l creatine (C0780; Sigma, St Louis, MO, USA) to enhance mitochondrial phosphate shuttling [16]. ADP-stimulated respiratory kinetics were determined through ADP (A5285; Sigma) titrations in the presence of 5 mmol/l blebbistatin (CAY13013; Cayman Chemical, Ann Arbor, MI, USA), in order to prevent spontaneous contraction of fibres [13, 17, 18], 5 mmol/l pyruvate (P2256; Sigma) and 2 mmol/l malate (M1000; Sigma). Complex I kinetics were determined through standard pyruvate and glutamate (GLU303; BioShop, Burlington, ON, Canada) titrations in the presence of 5 mmol/l blebbistatin, 5 mmol/l ADP and 2 mmol/l malate, while Complex II kinetics were determined through standard succinate (S2378; Sigma) titrations in the presence of 5 mmol/l blebbistatin, 5 mmol/l ADP and 10 μmol/l rotenone (R8875; Sigma; to prevent superoxide generation at Complex I). During each titration, Cytochrome c (192-10; Lee Biosolutions, Maryland Heights, MO, USA) was added last to determine intactness of the outer mitochondrial membrane. Any samples that exceeded a 10% increase in respiration after Cytochrome c addition were excluded [19].

Mitochondrial H2O2 emission

Mitochondrial H2O2 emission was measured fluorometrically (QuantaMaster 40; HORIBA Scientific, Edison, NJ, USA) at 37°C in buffer Z [18], containing 105 mmol/l K-MES, 30 mmol/l KCl, 10 mmol/l KH2PO4, 5 mmol/l MgCl2.6 H2O, 1 mmol/l EGTA, 5 mg/ml BSA (pH 7.1), supplemented with 20 mmol/l creatine, 10 μmol/l Amplex UltraRed reagent (A36006; Life Technologies, Carlsbad, CA, USA), 0.5 U/ml horseradish peroxidase (P8375; Sigma) and 40 U/ml Cu/Zn superoxide dismutase (SOD1; S9697; Sigma), as described previously [13]. Various sites were assessed by titrating the following substrates [20, 21]: (1) 2.5 μmol/l antimycin (A8674; Sigma) for assessment of Complex III; (2) 10 mmol/l pyruvate and 2 mmol/l malate, to assess Complex I via generation of NADH; (3) 10 mmol/l succinate, to assess Complex I (via reverse flow) via generation of FADH2; (4) 10 mmol/l pyruvate and 0.5 μmol/l rotenone, to assess pyruvate dehydrogenase complex (PDC); rotenone blocks NADH entry into Complex I; and (5) 20 mmol/l glycerol-3-phosphate, for the assessment of mitochondrial glycerol-3-phosphate dehydrogenase (mG3PDH; G7886; Sigma). In addition, fibres used for Complex I (pyruvate + malate), PDC and mG3PDH measurements were treated with 35 μmol/l 1-chloro-2,4-dinitrobenzene (237329; Sigma) during permeabilisation to deplete glutathione [22].

Mitochondrial Ca2+ retention capacity

Mitochondrial Ca2+ retention capacity (CRC) was measured in duplicate, fluorometrically (QuantaMaster 40; HORIBA Scientific), at 37°C in buffer Y, containing 250 mmol/l sucrose, 10 mmol/l Tris-HCl, 20 mmol/l Tris base, 10 mmol/l KH2PO4, 2 mmol/l MgCl2.6H2O and 0.5 mg/ml BSA, with 1 μmol/l Calcium Green-5N (C3737; Life Technologies), as described previously [23]. Briefly, Ca2+ uptake was initiated with 8 nmol pulses of Ca2+ (CaCl2) and then subsequent pulses of 4 nmol Ca2+ were titrated until opening of the mitochondrial permeability transition pore (mPTP) was observed [23].


Fresh muscle was immediately fixed in 2% (vol./vol.) glutaraldehyde (111-30-8; Electron Microscopy Sciences, Hatfield, PA, USA) in 0.1 mol/l sodium cacodylate buffer pH 7.4 (Canemco, Lakefield, QC, Canada) and processed as described previously [24]. To quantify mitochondria, representative micrographs from eight unique fibres (containing a portion of the subsarcolemmal region adjacent to the nucleus, with most of the image containing the intermyofibrillar area) were acquired at ×15,000 magnification. Blinded quantification of mitochondrial size (mean area, μm2), distribution (number per μm2) and density (μm2 × number per μm2 × 100) was achieved using Nikon Imaging Software (NIS)-Elements AR (v 4.6; Nikon, Melville, NY, USA) by manually outlining mitochondria and converting to actual size using a calibration grid [25].

Electron tomography

Samples were oriented and trimmed to allow longitudinal sectioning of fibres, which were visually localised. Ultra-microtomy (Leica UC7; Leica Microsystems, Wetzlar, Germany) was performed with a 35° diamond knife (Diatome, Nidau, Switzerland), and 65 nm thick sections were mounted onto copper slot grids (Plano, Wetzlar, Germany). Electron micrographs for tomography tilt series were acquired on a JEM-2200FS electron microscope (JEOL, Akishima, Tokyo, Japan) (200 kV) equipped with a TemCam-F416 camera (TVIPS, Munich, Germany) and using a 25 electron volt (eV) energy filter slit. Under-focus was adjusted to 4000 nm. Micrographs were acquired two-fold binned for tomography. The resulting pixel size was 2.4 nm. Tilt series was collected over a total angular tilt range from −70° to +70°, at 2° increments. Image series was aligned by patch tracking mode using the IMOD software package (v 4.7; [26]. A single reconstructed volume was computed from each tilt series by radially weighted back projection. A 3D density distribution (tomogram) was obtained. A virtual section with a 24 nm thickness was generated with the slicer tool in IMOD. Major organelles were coloured utilising Adobe Photoshop CC 2015 by a blinded researcher.

Histology and immunofluorescence analysis

Biopsied muscle was immediately embedded in Tissue Tek optimum cutting temperature (O.C.T.) compound (4583; VWR, Radnor, PA, USA), frozen in liquid nitrogen-cooled isopentane and stored at −80°C until analysis. Cryosections (−20°C) that were 8 μm thick were transferred onto static-free microscope slides (CA48311-703; VWR), allowed to air dry and fixed with ice-cold 4% (wt/vol.) paraformaldehyde. Sections were then incubated in blocking solution (5% [vol./vol.] normal goat serum in 1X PBS; S-1000; Vector Laboratories, Burlingame, CA, USA) for 30 min and then co-stained with CD31 (platelet endothelial cell adhesion molecule-1 [PECAM-1]) overnight. An antibody against CD41 was added the following day and incubated for 1 h. The appropriate secondary antibodies were then applied for 1 h: Alexa Fluor 594 and Alexa Fluor 488. Nuclei were counterstained with DAPI. CD31 was used as a marker of capillary content and CD41 to measure platelet aggregation. The antibodies were validated by manufacturers for immunofluorescence analyses (see electronic supplementary material [ESM] Table 1 for antibody details). H&E stains were used for the determination of muscle fibre area in order to calculate capillary:fibre ratio, as previously described [14]. Images were captured with a Nikon 90 Eclipse microscope (Nikon) and analysed using NIS-Elements AR software (v 4.6; Nikon) at original magnification of ×40.

Western blotting

A portion of muscle was quickly snap-frozen and stored at −80°C until analysis. Frozen muscle (~10–30 mg) was homogenised with a tapered teflon pestle in ice-cold buffer containing 40 mmol/l Hepes, 120 mmol/l NaCl, 1 mmol/l EDTA, 10 mmol/l NAHP2O7·10H2O pyrophosphate, 10 mmol/l β-glycerophosphate, 10 mmol/l NaF and 0.3% CHAPS detergent (pH 7.1, adjusted using KOH). Protein concentrations were determined using a BCA assay (Life Technologies) and then loaded equally (50 or 100 μg), using Ponceau S (P7170; Sigma) as a loading control. Proteins were separated by SDS-PAGE, transferred to polyvinylidene fluoride membrane, blocked for 1 h and incubated overnight at 4°C using commercially available antibodies validated by manufacturers for western blotting: human total oxidative phosphorylation cocktail (OXPHOS; to detect individual complexes of the electron transport chain [I, II, III, IV, V]), phosphorylated (Thr172) and total AMP-activated protein kinase (AMPKα), phosphorylated (Ser555) and total Unc-51-like autophagy activating kinase 1 (ULK1), sequestosome (SQSTM1; herein referred to as p62) and vinculin. After overnight incubation, membranes were washed in tris-buffered saline, 0.1% (vol./vol.) Tween 20 (TBS-T) and incubated for 1 h at room temperature with anti-mouse infrared fluorescent secondary antibody (IRDye 800CW-conjugated) for OXPHOS, and horseradish peroxidase conjugated secondary antibody for all other proteins (see ESM Table 1 for antibody details). Immunoreactive proteins were detected by infrared imaging (Odyssey CLx; LI-COR Biosciences, Lincoln, NE, USA) or chemiluminescence (Gel Logic 6000 Pro Imager; Carestream, Rochester, NY, USA) and quantified by densitometry using Image J (v 1.51t;

Statistical analysis

All results are expressed as mean ± SEM except for participant characteristics, which are expressed as mean ± SD. An unpaired two-tailed Student’s t test was performed for all experiments except for CRC and mitochondrial respiration kinetics where an unpaired one-tailed t test and a two-way ANOVA were used, respectively. Significance was established at p ≤ 0.05.


Skeletal muscle mitochondrial bioenergetics

Control of oxidative phosphorylation by ADP

ADP is a primary regulator of mitochondrial bioenergetics and signals the net cellular metabolic demand to mitochondria in times of energetic stress. Therefore, we first determined if mitochondria were sensitive to low, physiologically relevant concentrations of ADP, as well as maximal concentrations (oxidative capacity) in type 1 diabetes. We found that both sensitivity to low ADP levels and maximal capacity for respiration were significantly lower (p < 0.05, main effect) in participants with type 1 diabetes compared with control participants (Fig. 1a).

Fig. 1
figure 1

Skeletal muscle mitochondrial bioenergetics in young adults with and without type 1 diabetes. Rates of mitochondrial oxygen consumption (JO2) were measured in permeabilised muscle fibres in the presence of carbohydrate-derived substrates. (a) Submaximal and maximal (oxidative capacity) ADP-stimulated respiration in the presence of pyruvate and malate. Concentrations of ADP required to reach maximum varied between 6 and 18 mmol/l (both groups, n = 12). (b) Pyruvate-supported respiration (Complex I sensitivity and oxidative capacity) in the presence of 5 mmol/l ADP. Concentration of pyruvate required to reach maximum varied between 6 and 38 mmol/l (control group, n = 11; type 1 diabetes group, n = 9). (c) Glutamate-supported respiration (Complex I sensitivity and oxidative capacity) in the presence of 5 mmol/l ADP. Concentration of glutamate required to reach maximum varied between 4 and 20 mmol/l (control group, n = 12; type 1 diabetes group, n = 8). (d) Succinate-supported respiration (Complex II sensitivity and oxidative capacity) in the presence of 5 mmol/l ADP and rotenone. Concentration of succinate required to reach maximum varied between 22 and 26 mmol/l (both groups, n = 11). (e) Rates of mitochondrial H2O2 emission (mH2O2) derived from Complex III (antimycin A [AA]), Complex I forward electron flow (pyruvate + malate [P + M]) and reverse electron flow (succinate [Succ]), PDC (pyruvate + rotenone [P + ROT]) and mG3PDH (glycerol-3-phosphate [G3P]). White bars, control; black bars, type 1 diabetes; white circles, control; grey squares, type 1 diabetes. *p < 0.05, ***p < 0.001, main effect of type 1 diabetes; p = 0.278; p = 0.117; §p < 0.05. Max, maximum; Wt, weight

Substrate-specific impairments

We then determined if mitochondrial sensitivity to specific substrates was different in the muscle of those with type 1 diabetes vs control individuals. No detectable change in Complex I-supported mitochondrial respiration was observed, as evidenced by the lack of a significant difference in the sensitivity or capacity of the mitochondria to oxidise pyruvate (an index of glucose oxidation) (Fig. 1b). This suggests that mitochondrial oxidation of pyruvate from glucose is not impaired in type 1 diabetes, and neither is Complex I oxidation of NADH from pyruvate dehydrogenase. There were also no detectable changes in glutamate-supported respiration (NADH oxidation) (Fig. 1c), which further suggests there is no deficiency in Complex I activity in type 1 diabetes. In contrast, Complex II-supported respiration by succinate (FADH2 synthesis) was significantly lower (p < 0.001, main effect) at both low (sensitivity) and maximal (oxidative capacity) succinate concentrations in participants with type 1 diabetes, suggesting a reduction in Complex II activity in type 1 diabetes compared with control individuals (Fig. 1d).

Mitochondrial H2O2 emission

Next, we measured mitochondrial H2O2 emission potential. In the participants with type 1 diabetes, Complex III-supported mitochondrial H2O2 emission was significantly elevated (p < 0.05) compared with control participants but was not statistically different at Complex I (reverse [succinate] and forward electron flow [pyruvate + malate]), PDC and mG3PDH (Fig. 1e).

Skeletal muscle mitochondrial content

Considering that a reduction in mitochondrial respiration can be caused by lower mitochondrial content, we next examined the spatial organisation and abundance of mitochondria, as well as the expression of individual electron transport chain proteins. Using TEM, no differences in mean mitochondrial size, number of mitochondria per muscle area or mitochondrial area density in either the subsarcolemmal or intermyofibrillar regions of the muscle were observed (Fig. 2a). Similarly, the levels of individual proteins were not significantly different between type 1 diabetes and control participants (Fig. 2b, c), further supporting no difference in total muscle mitochondrial content between the groups. Likewise, there were no differences in p-AMPKαThr172 (relative to total AMPKα; Fig. 3d), a known regulator of mitochondrial biogenesis [27]. Collectively, this data indicates that the lower respiration per mg of muscle was not due to reductions in mitochondrial content, suggesting that impairments intrinsic to the mitochondria themselves might exist in people with type 1 diabetes. However, no differences were observed in ADP-stimulated respiration, Complex I (pyruvate)- and Complex II (succinate)-supported respiration when normalised to OXPHOS protein content (ESM Fig. 1). Instead, a main effect (p < 0.05) for higher Complex I (glutamate)-supported respiration was observed in those with type 1 diabetes vs control participants when normalised to OXPHOS (ESM Fig. 1c). However, we take caution in this observation given the heterogeneity in both respiration and OXPHOS protein expression between individuals, which may warrant larger sample sizes for such ratiometric comparisons.

Fig. 2
figure 2

Skeletal muscle mitochondrial content. (a) Transmission electron micrographs were acquired and analysed for mitochondrial size (mean area, μm2), distribution (number per μm2) and density (μm2 × number per μm2 × 100) within the subsarcolemmal area (SS) and intermyofibrillar area (IMF) of the muscle. Blue arrow, nucleus; yellow arrow, SS mitochondria; green arrow, IMF mitochondria. Scale bar, 2 μm. (b) Protein levels of the individual complexes of the electron transport chain (I, II, III, IV, V; as measured by the OXPHOS antibody cocktail) as well as mean expression of all complexes combined. (c) Representative immunoblot of Complex I (~18 kDa), Complex II (~29 kDa), Complex III (~48 kDa), Complex IV (~22 kDa) and Complex V (~54 kDa) and Ponceau stain (loading control). White circles, control; grey squares, type 1 diabetes. ATP5a, ATP synthase subunit alpha 5; Con, control; COXII, cytochrome oxidase subunit II; NDUFB8, NADH:ubiquinone oxidoreductase subunit B8; SDHB, succinate dehydrogenase [ubiquinone] iron-sulfur subunit B; T1D, type 1 diabetes; UQCRC2, ubiquinol-cytochrome c reductase core protein 2

Fig. 3
figure 3

Mitochondrial CRC, ultrastructural analysis and measurement of autophagic regulatory proteins. (a) Total mitochondrial Ca2+ (mCa2+) uptake (a.k.a. CRC) was measured in permeabilised muscle fibres by sequential titrations of Ca2+ pulses until opening of the mPTP was observed. (b) Representative electron tomography (3D) images of irregular organisation of mitochondrial cristae (highlighted in red) in type 1 diabetes compared with control individuals. Cyan highlighting, intermyofibrillar autophagic remnants; yellow highlighting, triads/sarcoplasmic reticulum. Scale bar, 200 nm. (c) Representative electron tomography (3D) images of increased presence of autophagic remnants in both subsarcolemmal (highlighted in purple) and intermyofibrillar (highlighted in cyan) regions of the muscle, which were also frequently associated with intramyocellular lipid droplets (highlighted in green) in those with type 1 diabetes vs control individuals. Red highlighting, mitochondrial cristae; magenta outline, subsarcolemma; blue outline, nucleus; yellow highlighting, triads/sarcoplasmic reticulum; orange highlighting, Golgi apparatus. Scale bar, 200 nm. (d) Quantification of p-AMPKαThr172 normalised to total AMPKα with representative blot (p-AMPKαThr172 and total AMPKα, ~62 kDa). (e) Quantification of p-ULK1Ser555 normalised to total ULK1 with representative blot (p-ULK1Ser555 and total ULK1, ~150 kDa). (f) Quantification of p62 normalised to loading control vinculin, with representative blot (p62, ~62 kDa; vinculin, ~124 kDa). White circles, control; grey squares, type 1 diabetes. *p < 0.05. Con, control; T1D, type 1 diabetes

Skeletal muscle mitochondrial Ca2+ tolerance and electron tomography

We next tested Ca2+ handling within the mitochondria. Opening of the mPTP has been linked to swelling and energetic dysfunction [28]. We hypothesised that reductions in Ca2+ handling within the mitochondria (i.e. reduced CRC) would lead to an increased sensitisation of the mPTP and, ultimately, attenuations in mitochondrial respiratory function. As hypothesised, mitochondrial CRC was significantly lower (p = 0.046) in skeletal muscle from participants with type 1 diabetes relative to control participants (Fig. 3a).

We next used electron tomography to assess the skeletal muscle ultrastructure. This powerful technique allows for the 3D study of cellular architecture at the nanoscale level. Consistent with our observed attenuations in mitochondrial respiration, electron tomography revealed irregularities in the organisation of cristae in the participants with type 1 diabetes that were uncommon in control participants (Fig. 3b; mitochondria and cristae highlighted in red). Furthermore, consistent with the reductions in mitochondrial Ca2+ handling previously noted, the muscles from participants with type 1 diabetes also exhibited an increase in size and number of autophagic remnants compared with the control group, and this was visible within both the subsarcolemmal and intermyofibrillar areas (Fig. 3c; highlighted in purple and cyan, respectively). A close association of autophagic remnants to intramyocellular lipid droplets was also frequently observed in individuals with type 1 diabetes (Fig. 3c; lipid droplets highlighted in green). Despite these observations, no differences in the autophagic regulatory protein p-AMPKαThr172 (as noted previously) or its downstream targets p-ULK1Ser555 or p62 were observed between groups (Fig. 3d–f).

Skeletal muscle microvasculature

To ascertain whether changes in mitochondrial bioenergetics were the result of microvasculature loss and concomitant hypoxia, capillary density was quantified by staining frozen muscle sections with an antibody against CD31. No significant difference between groups was observed, regardless of whether the data was collated as per cent density or ratio of total number of capillaries to myofibre area (Fig. 4a). It was also worth considering that, while capillary density may not be diminished, the capillaries present may be obstructed given the reported increases in prothrombotic factors, such as plasminogen activator inhibitor 1 (PAI-1), in type 1 diabetes [29]. We speculated that there may be greater prevalence of thrombi within the capillaries of individuals with type 1 diabetes, consequently resulting in a decreased regional blood flow. We assessed platelet aggregation by co-staining frozen muscle sections for CD31 and CD41. No significant difference between groups was demonstrated in the number of CD41+ areas overlaying CD31+ cells (Fig. 4b) indicating that in this physically active, young adult cohort with type 1 diabetes, impairments in skeletal muscle microvasculature did not mediate the observed alterations in mitochondrial bioenergetics and ultrastructure. Moreover, this finding also suggests that deterioration in skeletal muscle quality may precede the eventual vascular complications that arise in type 1 diabetes.

Fig. 4
figure 4

Skeletal muscle capillary density and analysis of platelet aggregation. (a) Capillary density expressed as the percentage of areas positively stained for platelet endothelial cell adhesion molecule (CD31+) relative to total muscle area analysed and as ratio of total number of CD31+ to myofibre area. (b) Positively stained areas of platelet aggregation (CD41+) within capillaries (CD31+) measured as the ratio of CD41+ to CD31+. White circles, control; grey squares, type 1 diabetes. Con, control; T1D, type 1 diabetes


The findings of this work highlight mitochondrial and autophagic differences within the muscles of young adults with type 1 diabetes with moderate glycaemic control and who exceed recommended activity levels (>150 min of moderate-to-vigorous intensity activity per week) according to the American and Canadian diabetes associations [30, 31]. Specifically, the work described herein demonstrates, for the first time, site-specific alterations within the mitochondria of young adults with type 1 diabetes compared with (matched) control participants without diabetes. These alterations in mitochondrial bioenergetics occur in the absence of a loss of mitochondrial content or changes in skeletal muscle microvasculature (as assessed by capillary density and platelet aggregation). Furthermore, TEM investigations revealed that type 1 diabetes negatively affects skeletal muscle ultrastructure, as evidenced by disorganised mitochondrial cristae and an increased presence of autophagic remnants. These early alterations precede the well-characterised complications (e.g. cardiovascular disease, polyneuropathy) of type 1 diabetes and occur in large, proximal muscle groups (in contrast to distal limb muscles, which are more readily affected by polyneuropathy), suggesting that the early alterations seen in this study are unlikely to be owing to neuropathy but, instead, intrinsic alterations in muscle, and that this skeletal muscle metabolic dysregulation is a primary outcome of this chronic disease that may contribute to dysglycaemia.

Skeletal muscle is not only the primary organ system responsible for physical capacity but is also a critical metabolic organ since it is: (1) the major site of fatty acid catabolism [32]; (2) a principal mediator of whole-body glucose homeostasis [2]; and (3) a major determinant of whole-body insulin sensitivity [2, 3]. Thus, impairments in skeletal muscle quality, particularly in conditions of metabolic disease (e.g. diabetes mellitus), could have profound long-term consequences by increasing the incidence of primary contributors to the development of diabetic complications, such as recurrent dysglycaemia, dyslipidaemia and insulin resistance [33,34,35]. Consistent with the importance of skeletal muscle in overall health, if one considers other chronic diseases (e.g. heart disease and cancer), loss of muscle mass, strength and metabolic function (referred to as cachexia) are commonly observed, and this decline in muscle health is an important determinant in overall survival [36, 37].

While changes in skeletal muscle health of rodent models of type 1 diabetes has been more intensely studied (previously referred to as ‘diabetic myopathy’ [38, 39]), the extent to which this is happening in human skeletal muscle of those with type 1 diabetes has yet to be fully elucidated [40]. In the clinical diabetes literature, skeletal muscle mitochondrial ATP synthesis is one of the most widely investigated variables, often assessed indirectly in vivo by monitoring the resonance of phosphate in ATP using 31P-MRS. Previous studies using 31P-MRS have frequently, but not always [41], reported a ~20–25% reduction in mitochondrial ATP synthesis rates in the skeletal muscle of those with type 1 diabetes [9,10,11]. While 31P-MRS has the obvious advantage of capturing ATP kinetics under intact physiological conditions, a drawback of this approach is the lack of mechanistic understanding underlying the changes in mitochondrial function that may be observed.

Consistent with the work of Cree-Green et al [9] and others [10, 11], we observed a ~20% reduction in mitochondrial oxidative capacity in the muscles of those with type 1 diabetes (Fig. 1a, Max), suggesting that our in vitro (permeabilised muscle fibre method) assessment is consistent with 31P-MRS studies. What was unanswered from the aforementioned 31P-MRS studies, however, was whether the mitochondria are reduced in content and/or impaired, or whether nutrient delivery is the cause of the resulting poor mitochondrial function in those with type 1 diabetes. Here we provide the mechanistic insight to answer these questions.

First, regarding whether the deficits observed in previous 31P-MRS studies were directly related to differences in mitochondrial content or impairments of mitochondrial quality, we now offer invaluable insight using ultrastructural analysis. Using TEM, we observed no difference in the amount, distribution and size of mitochondria within the skeletal muscle of those with type 1 diabetes compared with control individuals. Within the mitochondria, we observed no differences in the protein expression of the individual electron transport chain protein subunits using western blot analysis (total OXPHOS cocktail), further suggesting that mitochondrial content is not reduced in young, physically active adults with type 1 diabetes vs control individuals. In contrast, we did observe structural abnormalities in mitochondria using electron tomography; disorganised cristae and more abundant autophagic structures were present in the muscles of individuals with type 1 diabetes compared with control participants. In line with this, we also observed a significant reduction in mitochondria viability in permeabilised myofibres of those with type 1 diabetes, as assessed by CRC [23].

Second, these findings suggest that nutrient delivery in the muscles of young adults with type 1 diabetes is not likely impaired. Histological/immunofluorescent examination of skeletal muscle biopsies revealed that capillary density within the vastus lateralis muscle was not different between those with type 1 diabetes and those without. Furthermore, as increases in prothrombotic factors have been reported in those with type 1 diabetes [29], we investigated whether the capillaries present were occluded by platelet aggregates (thrombi) and found that they were not. While we do not see visible evidence for impaired blood flow in the present study, we cannot rule out the possibility of a functional impairment resulting in reduced blood flow and delivery of oxygen and nutrients within the muscles of those with type 1 diabetes [42].

Collectively, these novel results provide evidence of direct alterations in skeletal muscle mitochondria in those with type 1 diabetes. The underlying mechanism(s) for these mitochondrial alterations in type 1 diabetes may be revealed by the mitochondrial bioenergetics investigations related to mitochondrial H2O2 emission in this study. To the best of our knowledge, mitochondrial-derived ROS production has never been examined in skeletal muscle of humans with type 1 diabetes. We observed a significant increase in mitochondrial H2O2 emission potential at the level of Complex III, with no significant differences owing to Complex I, intermembrane (mG3PDH) or matrix (PDC) proteins. Work using isolated mitochondria [43, 44], has demonstrated that Complex III significantly contributes to ROS production in response to fatty acid oxidation. While we did not assess this in the current study, future work should be undertaken to discern differences in the propensity for ROS production at Complex III in response to glucose and lipid substrate availability in those with type 1 diabetes. The tissue requirements for the comprehensive assessment of mitochondrial phenotype prohibited additional assessment of major mitochondrial antioxidant enzymes (e.g. glutathione peroxidase, thioredoxin reductase and superoxide dismutase) and markers of oxidative stress in the present study. Nonetheless, it has been suggested that Complex III-generated ROS serves as a signalling molecule, particularly in the adaptive response to hypoxia (for review see [45]). This latter point may be physiologically important when considering the peripheral vascular disease that is known to develop in older adults with type 1 diabetes.


This investigation revealed that young, active people with type 1 diabetes possess mitochondrial and muscle ultrastructural abnormalities that precede reductions in capillary content. While we have demonstrated site-specific alterations in mitochondrial bioenergetics, further investigation is required to identify the precise signalling pathway leading to these changes. Though our observations of increased autophagic remnants in muscles of those with type 1 diabetes were not associated with increases in autophagic markers (p-ULK1Ser555 and p62), it is possible that these signals had already been transmitted and pathways initiated. Alternatively, recent work [46] has suggested that p62 activity is redox sensitive and may promote autophagy in the absence of changes in content. Clearly, future work in this area will help further elucidate this important observation.

It seems likely that the relationship between skeletal muscle abnormalities and dysglycaemia in type 1 diabetes are reciprocal, particularly given previous suggestions that excessive glucose provision to mitochondria elevates ROS production [47, 48]. Another possible mechanism may relate to peripheral insulin administration. Specifically, in people without type 1 diabetes, large rises in blood glucose levels following a glucose load are attenuated by delivering insulin to the liver first via the portal circulation. However, in those with type 1 diabetes, peripheral insulin administration bypasses the canonical ‘liver-first’ model and results in greater elevations in circulating blood glucose [49]. The peripheral insulin administration shifts a greater burden of the glycaemic load to skeletal muscle, forcing it to accept a greater glycaemic load then it otherwise would [1, 2], thereby increasing glycaemic stress akin to that previously described [50] for cell types that could not regulate cellular glucose entry. An important avenue for future research is to investigate this altered relationship between blood glucose levels, peripheral insulin injections and the muscle abnormalities reported in the type 1 diabetes population. Indeed, this relationship might be considered in the context of potential cellular abnormalities in other relevant tissues, such as liver and adipose tissue.

A major clinical concern in this study is that the mitochondrial/metabolic alterations observed were in young adults with type 1 diabetes, who had self-reported moderate-to-vigorous activity levels above the American and Canadian diabetes associations’ recommendations. In addition, these changes occurred in large, proximal muscle groups without a detectable loss of capillary density. Clearly, larger cohort studies are needed to assess the impact of type 1 diabetes in those who are sedentary, as well as those who meet or exceed physical activity recommendations, to determine whether an optimal level of physical activity exists that is sufficient to restore/overcome the metabolic abnormalities that characterise the muscle of those with type 1 diabetes.