Comparison of intramyocellular lipid metabolism in patients with diabetes and male athletes

Despite opposing insulin sensitivity and cardiometabolic risk, both athletes and patients with type 2 diabetes have increased skeletal myocyte fat storage: the so-called “athlete’s paradox”. In a parallel non-randomised, non-blinded trial (NCT03065140), we characterised and compared the skeletal myocyte lipid signature of 29 male endurance athletes and 30 patients with diabetes after undergoing deconditioning or endurance training respectively. The primary outcomes were to assess intramyocellular lipid storage of the vastus lateralis in both cohorts and the secondary outcomes were to examine saturated and unsaturated intramyocellular lipid pool turnover. We show that athletes have higher intramyocellular fat saturation with very high palmitate kinetics, which is attenuated by deconditioning. In contrast, type 2 diabetes patients have higher unsaturated intramyocellular fat and blunted palmitate and linoleate kinetics but after endurance training, all were realigned with those of deconditioned athletes. Improved basal insulin sensitivity was further associated with better serum cholesterol/triglycerides, glycaemic control, physical performance, enhanced post insulin receptor pathway signalling and metabolic sensing. We conclude that insulin-resistant, maladapted intramyocellular lipid storage and turnover in patients with type 2 diabetes show reversibility after endurance training through increased contributions of the saturated intramyocellular fatty acid pools. Clinical Trial Registration: NCT03065140: Muscle Fat Compartments and Turnover as Determinant of Insulin Sensitivity (MISTY)

characteristic of type 2 diabetes 5 .Surprisingly, increased intramyocellular fat storage also occurs in the skeletal muscle of healthy, endurance-trained athletes who, in contrast, are highly insulin sensitive 6 .The intramyocellular fat storage in athletes cannot be viewed as 'ectopic deposition' since athletes have very little subcutaneous adipose tissue, and this phenomenon is known as the "athlete's paradox" 7,8 .A mechanistic explanation for the increased intramyocellular lipid storage in participants who are at opposite ends of insulin sensitivity and cardiometabolic risk remains elusive and a matter of debate.
Important differences in intramyocellular lipids are likely to exist between insulin-sensitive athletes and insulin-resistant patients with type 2 diabetes, and if so, questions arise whether these differences could potentially be a target for manipulation in order to provide health improvements for patients.The abundance of specific intramyocellular lipid isoforms (diacylglycerols and sphingolipids) and their sub-cellular localisation have been proposed as potential candidates to subtend such differences and play a role in promoting insulin resistance 9,10 .However, these species represent only a small fraction of the intramyocellular lipid pool, which is largely composed of triacylglycerols.This suggests that further characteristics common to the majority of lipids must be present to explain the opposing phenotypes resulting from an apparently similar intramyocellular metabolic storage.The relative proportions or utilisation of saturated and unsaturated intramyocellular fat has been proposed as one such characteristic but has never been demonstrated.There is evidence from in vitro data that saturated and unsaturated fatty acids partition toward different metabolic pathways in muscle cells 11 and that upregulation of unsaturated lipo-synthetic pathways in human skeletal myocytes is strongly and reversibly associated with insulin resistance 12 .In keeping with this, a saturated high-fat diet increased content and activity of the mitochondrial fatty acid oxidation master regulator sirtuin in human skeletal muscle, whereas unsaturated diets did not 13 .Contrasting with these observations, saturated fatty acids are traditionally associated with increased cardio-metabolic risk.This creates equipoise regarding the relative contributions of saturated and unsaturated intramyocellular fat to insulin sensitivity.
To address this, we designed a clinical trial of age and sexmatched endurance athletes and patients with type 2 diabetes.The intervention was deconditioning for athletes and endurance exercise training for patients with type 2 diabetes, which influenced participants' basal insulin sensitivity status.At the distinct baseline and postintervention stages, we captured the skeletal muscle spectroscopic detection of saturated versus unsaturated carbon bonds contained within intramyocellular lipid stores, stable isotope [U- 13 C] turnover of skeletal muscle saturated and unsaturated fatty acids, lipidomic, metabolic, molecular and physical performance profiling.We characterised the skeletal muscle intramyocellular lipid signature of athletic health versus diabetes dysmetabolism.The hypotheses of the study were: (1) the relative proportions of saturated/unsaturated intramyocellular lipid storage in patients with type 2 diabetes is different to age/sex matched healthy athletes, and (2) exercise endurance training in type 2 diabetes patients is associated with changes in the saturated/unsaturated intramyocellular storage and turnover.

Participants
From 83 volunteers invited (39 endurance athletes and 44 patients with type 2 diabetes), 29 athletes and 30 patients with type 2 diabetes were enroled after screening (Fig. 1) between September 2016 and January 2019.One athlete and one patient withdrew, and two further patients were withdrawn after developing chest pain during the exercise training programme.Athletes were free of any health conditions or medications, whilst a third of patients were on antihypertensive medication and two thirds were on statin therapy for primary prevention (Supplemental Table 1).At screening, athletes performed more daily exercise, but there were no differences in the dietary composition between groups for total fat, saturated, monounsaturated or poly-unsaturated fatty acids although athletes consumed more sugar.All participants had normal resting electrocardiograms and echocardiography-derived left ventricular ejection fraction, although, as expected, global left ventricular longitudinal strain was reduced in patients with diabetes and left ventricular volumes were increased in athletes, consistent with the "athletic heart".Contrast-enhanced cardiac magnetic resonance ruled out the possibility of any previously unknown (silent) myocardial infarction in patients with type 2 diabetes 14 .All participants achieved a respiratory exchange ratio >1.1 on cardiopulmonary exercise testing, and athletes exercised for longer times, achieved more metabolic equivalents and had higher peak oxygen consumption compared to patients with type 2 diabetes (Supplemental Table 1).

Baseline characteristics of athletes and patients with type 2 diabetes
Athletes were leaner than patients (Table 1).First, we established the intramyocellular lipid storage pattern and Fig. 2A shows the schematic representation of the 1 H-magnetic resonance spectroscopy acquisition of total, as well as saturated and unsaturated intramyocellular lipid peaks described in the methods.The fractional lipid mass (fLM) was lower in athletes' skeletal muscle compared to patients with type 2 diabetes (0.035 ± 0.01 vs 0.056 ± 0.01 arbitrary units, p < 0.0001).In addition, athletes had higher fraction of intramyocellular saturated lipid (87.96 ± 4 vs 81.65 ± 6%, p = 0.0004) and lower fraction of intramyocellular unsaturated lipids (12.04 ± 4 vs 18.35 ± 6%, p = 0.0004) compared to patients with diabetes, Table 2 and Fig. 2B.Next, we assessed the skeletal muscle saturated and unsaturated fatty acid turnover, which is represented by fractional incorporation rate of each, calculated after intravenous administration of [U 13 C]-palmitate (16.0) and [U 13 C]-linoleate (18.2 n-6) coupled with pre-and postinfusion skeletal muscle biopsies.Both palmitate and linoleate fractional incorporation rates were higher in athletes compared to patients with diabetes (0.62 ± 0.4 vs 0.15 ± 0.07%/h and 0.27 ± 0.1 vs 0.12 ± 0.06%/h respectively, a 4.1-fold for palmitate and 2.3-fold for linoleate) despite similar plasma rates of appearance, Table 3 and Fig. 3A.When the collective intra-plus extra-myocellular triacylglycerols, diacylglycerols and ceramide species were quantified directly from skeletal muscle biopsies and grouped according to saturation status, there were no differences between athletes and patients with type 2 diabetes in total, saturated or unsaturated lipid pools, Table 4.Western blot analyses of proteins involved in insulin receptor pathways also demonstrated no differences between athletes and patients with type 2 diabetes including the trans-membrane glucose transporters (GLUT1 and GLUT4), phosphorylated insulin receptor (pIR), insulin receptor substrate (IRS), phosphorylated S6 kinase protein (pS6), phosphorylated extracellular signal-regulated kinase (pERK), phosphorylated or total protein kinase B, phosphorylated or total 5' adenosine monophosphate-activated protein kinase or their phosphorylated to total ratios, all assessed in basal state (Table 4).Athletes had higher serum HDL-cholesterol concentrations but lower serum triglycerides, fasting glucose, insulin and plasma non-esterified (free) fatty acids and were highly insulin sensitive whereas patients with type 2 diabetes were insulin resistant (Table 5).Athletes' cardiopulmonary exercise performance was superior (Table 1) as already demonstrated at the screening stage (Supplemental Table 1).The forced expiratory volume in one second (FEV 1 ), resting and peak oxygen saturations as well as the VE/VCO 2 slope were within normal limits, ruling out any pulmonary pathology or cardiac limitation in either group.
Exercise interventions were group-specific: Compliance with deconditioning (from an average of 92 min to <10 min structured daily exercise for 4 weeks) by the athletes was monitored using accelerometer data whereas patients with type 2 diabetes were directly supervised by the investigators during each daily 60 min of cycling, 5 days/week for 8 weeks.To exclude any dietary change that could have been a confounder during the interventions, a repeat 1-week food diary during exercise interventions was undertaken.This showed no change in nutritional intake in either group, except for a decrease in sugar consumption by athletes (Supplemental Table 2), which paralleled the removal of structured training.

Deconditioning in athletes
Athletes gained an average of 1.2 kg in weight following deconditioning (Table 1).However,1 H-magnetic resonance spectroscopy detected no significant changes in total intramyocellular lipid mass or fractions of saturated/unsaturated intramyocellular lipids (post deconditioning 1 H-magnetic resonance spectroscopy intramyocellular lipid mass of 0.038 ± 0.01 arbitrary units, fraction of saturated and unsaturated storage of intramyocellular lipids of 86.69 ± 4% and 13.30 ± 4%, respectively, Fig. 2B).Post-deconditioning, there was an apparent trend towards a decrease in palmitate fractional incorporation rate (to 0.35 ± 0.3%/h), but not for linoleate (which was measured at 0.21 ± 0.2%/h) (Table 3, Fig. 3A) The rates of plasma appearance for both palmitate and linoleate were unchanged and therefore unaffected by the post-exercise intervention status.Skeletal muscle biopsies identified no changes in total, saturated or unsaturated triacylglycerols, diacylglycerols and ceramides or basal state assessment of insulin receptor pathways (Table 4).Athletes' circulating LDL-cholesterol increased with deconditioning, but their basal insulin sensitivity was maintained (Table 5).As expected, their exercise performance declined, registering less metabolic equivalents and lower peak oxygen consumption, Table 1.

Endurance exercise training in patients with type 2 diabetes
Patients lost an average of 2.6 kg in weight following exercise training, Table 1.Although there were no demonstrable changes in total intramyocellular lipid mass detected with 1 H-magnetic resonance spectroscopy (0.052 ± 0.007 arbitrary units), the fraction of intramyocellular saturated lipid increased (to 88.00 ± 3%), with a reciprocal fall in unsaturated lipids (to 12.00 ± 3%), Fig. 2B.This was accompanied by an increase in both palmitate and linoleate fractional incorporation rate [to 0.35 ± 0.3%/h (2.4-fold) for palmitate and to 0.21 ± 0.2%/h (1.7-fold) for linoleate], Table 3 and Fig. 3A, although the rates of plasma appearance for both palmitate and linoleate were also unchanged and therefore unaffected by the post-exercise intervention status.Skeletal muscle biopsies identified no changes in total, saturated or unsaturated triacylglycerols, diacylglycerols and ceramides (Table 4) but there was an increase in the ratios of basal state phosphorylated/total protein kinase B (p/t AKT), and phosphorylated/total protein levels of 5' adenosine monophosphate-activated protein kinase (p/t AMPK) (Table 4 and Fig. 3B) consistent with enhanced post insulin receptor pathway and metabolic sensing.There were additional health improvements in achieving lower serum cholesterol, triglycerides, fasting glucose concentrations as well as HBA1c, which was associated with an improvement of their basal insulin sensitivity (Table 5).Remarkably, after exercise training patients with type 2 diabetes achieved lower serum cholesterol and LDL-cholesterol compared to athletes, although their fasting glucose and plasma non-esterified (free) fatty acids remained higher and serum HDL-cholesterol lower.
Their exercise performance improved, achieving longer exercise duration, more metabolic equivalents and higher peak oxygen consumption compared with before training (Table 1).
Extramyocellular skeletal muscle lipids and subcutaneous adipose tissue were also investigated with 1 H-magnetic resonance spectroscopy with no within or between group differences before or after either exercise intervention (Supplemental Fig. 1, 2).

Discussion
This parallel mechanistic skeletal muscle phenotyping study identified structural and functional differences in intramyocellular lipids between athletes and type 2 diabetes patients.Compared to patients with type 2 diabetes, athletes have a higher saturation of intramyocellular lipid storage as well as a 4.1-fold and 2.3-fold higher turnover of saturated and unsaturated fatty acids respectively.Deconditioning of athletes resulted in no alterations of their skeletal muscle phenotype except a trend in lowering their very high saturated (palmitate) lipid turnover which was a characteristic of their peak exercise performance.Conversely, 8-week endurance exercise training for patients with type 2 diabetes resulted in numerous important re-adaptations.There was an increase in intramyocellular lipid saturation and a larger increase in saturated compared to unsaturated skeletal muscle lipid turnover.These intramyocellular modifications aligned the diabetes skeletal muscle phenotype to that of deconditioned athletes.There were concomitant improvements in basal insulin sensitivity, serum cholesterol and triglycerides, glycaemic control, and physical performance.After endurance exercise training, patients with type 2 diabetes also demonstrated upregulation in basal state AKT (S473) and AMPKα (T172) phosphorylation, suggesting that both insulin receptor and metabolic pathways were sensitised basally within skeletal muscle cells.Our findings show that maladaptive intramyocellular skeletal muscle changes seen in the early stages of type 2 diabetes can be successfully reversed to a great extent.These data support the concept that skeletal muscle insulin resistance could be the primary defect in type 2 diabetes, as some have proposed 15,16 .Although further work would be required to elucidate the causality between muscle insulin resistance and intramyocellular lipid accumulation, these findings suggest that saturation of the ectopic intramyocellular lipid accumulation could be a new target for health improvement in type 2 diabetes.The concept that saturated fat can be beneficial, especially when stored, challenges the traditional view that associates all saturated fat with an increased risk of cardiovascular disease and worse clinical outcomes.
Structured physical exercise is a cheap but effective lifestyle intervention with powerful prognostic outcomes proven in general populations, and patients with cardiac conditions or diabetes 17,18 .Due to known sex differences in intramyocellular lipid stores and utilisation rates (both higher in women 19 ) in this initial exploration we studied only male participants to maintain a reasonable sample size for such an intensive study protocol.To age-match the groups we recruited recreational competitive athletes according to the European Society of Cardiology definition 20 , involved in endurance training, as elite athletes do not exist in this age group.
Our clinical trial reveals that, contrary to its known bad press, saturated fat is in fact essential for high level performance of skeletal muscle and its reduced intramyocellular availability or utilisation is tracking with the insulin resistant status and metabolic dysfunction of people with type 2 diabetes.The skeletal muscle lipidomic analysis detected no baseline or exercise-induced differences in total levels of saturated versus unsaturated tri/di-acylglycerols or ceramides between athletes and patients with type 2 diabetes.This apparent difference between the spectroscopic and lipidomic evaluations is rooted in two important technical aspects: firstly, spectroscopy assesses proportions of saturated/unsaturated carbon bonds present only in intramyocellular lipid stores, whereas lipidomics reports whole fatty acids per se, which are present in the intra-as well as extramyocellular compartments, including the cell membranes.Saturated fatty acids generally represent a much smaller component of the skeletal muscle lipid composition compared to the unsaturated ones, as for any given carbon length there is only one saturated fatty acid and numerous unsaturated ones.However, our MRI assessment of saturated/unsaturated carbon bonds shows that the saturated fraction of the intramyocellular lipid storage is much higher in athletes compared to patients with type 2 diabetes.
Given the dogma that saturated fat is harmful mainly through its contribution to atherosclerosis-related pathology, it is rather surprising that intramyocellular lipid saturation is reduced in patients with type 2 diabetes.Although the total intramyocellular pool did not reduce in size after patients' endurance training, its composition did alter, with increased saturation after exercise training, at least for the  duration of exercise taken during our study.One possible explanation for these findings is that compared to saturated fatty acid metabolism, β-oxidation of unsaturated fatty acids requires two additional energy-consuming catalytic reactions, one by an isomerase and one by a reductase enzyme 21,22 .Therefore, overall, β-oxidation of an unsaturated fatty acid with the same chain length as a saturated fatty acid will yield less energy 23 .Whilst it is understandable that with training, the skeletal muscle of healthy athletes will tend to store its fuel in a more efficient way (i.e.saturated fat), as it yields more energy, it is less clear why insulin resistant status would be associated with unsaturated intramyocellular storage.This finding cannot be attributed to differences in diets or changes in nutritional intake since participants were instructed to avoid dietary changes.
Even though the 1-week food diaries almost certainly under-reported their true intake, which is a well documented phenomenon 24 , importantly, there was no significant change between or within groups (except the athletes' sugar drinks requirements which were discontinued during deconditioning).Therefore, we conclude that skeletal muscle lipid saturation could represent a new biomarker of metabolic health in type 2 diabetes.Previous investigations have shown that specific lipid species (tri-, diacyl-glycerols, ceramides, sphingolipids) can influence insulin sensitivity status of healthy participants, participants with obesity and participants with type 2 diabetes 9,10 .Even though both ceramides and diacylglycerols have been identified as lipotoxic mediators of insulin resistance 25 , it is increasingly more apparent that specific lipid species, or isoforms, or even the sub-cellular localisation of their saturated or unsaturated forms which confers more importance in skeletal muscle insulin sensitivity than their total quantity 10,26-28 .These species are contained in the lipid bi-layers of sarcolemma, sarcoplasmic reticulum, mitochondrial membrane, as well as stored as ectopic intramyocellular fat and present in extramyocellular lipids which cannot be separated in a biopsy 29 .The lack of any significant change in muscle lipidomic analyses of total intramyocellular triacylglycerols, diacylglycerols or ceramides between athletes and patients with type 2 diabetes at baseline or after exercise intervention add onto a body of literature that remains at variance 30,2,31 .Notwithstanding conflicting reports of such total changes, further work will need to identify species, isoforms 26,28 or subcellular localisations 10 for example their proximity to the mitochondria, which can be modulated as key players responsible for cardio-metabolic health improvements.
We investigated the turnover of the two most abundant saturated (palmitate) and unsaturated (linoleate) fatty acids, the latter a first-inman exploratory application of fat metabolism.As expected from previous studies 6 , trained athletes had higher turnover of both palmitate and linoleate compared to patients with type 2 diabetes, being heavily reliant on palmitate utilisation for high physical performance, whereas in sedentary patients with type 2 diabetes, the palmitate turnover was drastically reduced with rates comparable to those of linoleate.Remarkably, improvements in basal insulin sensitivity through chronic aerobic exercise in patients with type 2 diabetes were also accompanied by a re-arrangement of their intramyocellular lipid turnover which re-aligned their phenotype with that of the deconditioned athletes.The structural and metabolic plasticity demonstrated in this study holds promise for further remedial interventions in diabetes but cannot of course be extrapolated to more advanced forms of disease or those on multiple hypo-glycaemic agents.It is remarkable that exercise training alone in patients with type 2 diabetes  significantly increases basal phosphorylation of AMPKα (T172) and Akt (S473) (in keeping with their improvements in systemic basal insulin sensitivity post-exercise training), even though trans-membrane glucose transporters GLUT1/4, IRS, pS6 or pERK remained unchanged after training.This re-emphasises the role of subcellular localisation and saturation of specific lipid fractions, such as diacylglycerols and ceramides 10,32 .The skeletal muscle AMPK activation (increased phosphorylation) in basal state suggests that the energy-sensing mechanism is upregulated in exercise-trained patients with type 2 diabetes, in keeping with their increased fatty acid fluxes.These findings complement previous observations that intramyocellular lipid droplets' morphology and their sub-cellular distribution in type 2 diabetes patients also showed a phenotype shift towards an athlete-like appearance after endurance training 33 .
Although we demonstrated significant beneficial changes in abundance and turnover of saturated intramyocellular lipids in response to exercise training in type 2 diabetes patients, physical exercise is not a practical solution for every individual.Further work should seek to explore which exercise-mimetic pharmacotherapies are capable of recapitulating these findings in type 2 diabetes patients.In particular, solutions targeting insulin resistance through specific intramyocellular lipids or indirectly via upstream nutrient-sensing transcription factors 34 targeting genes involved in specific lipid biosynthesis (such as long-chain fatty acid elongase 6 35 or Stearoyl-CoA desaturase 36 respectively) could be explored.
Limitations: We did not use a skeletal muscle-specific insulin resistance test or a hyperinsulinaemic-euglycaemic clamp for several pragmatic reasons, of added clinical risk, study visit duration, and complexity.For the same reasons of subject safety (avoiding a too prolonged fasting period) as well as pragmatically completing study visits within 8 h, we elected to measure insulin sensitivity by QUICKI and HOMA-IR indices rather than a more accurate oral glucose tolerance test.Our lipidomic analysis was a semi-quantitative approach aimed at detecting relative differences only.Whilst no medication was discontinued in the type 2 diabetes patients for the study duration, it is important to recognise that both metformin and statins use can result in a reduction or increase, respectively, in insulin resistance status.An influence on the results from nutritional intake cannot be definitively excluded.

Study approval
The study was approved by the North of Scotland Ethics Research Committee.All participants provided written informed consent prior to screening procedures and participation in the study (Clinical Trial Registration: NCT03065140: Muscle Fat Compartments and Turnover as Determinant of Insulin Sensitivity (MISTY).

Interventions
The purpose of this parallel non-randomised, non-blinded trial is to understand how the fat within the muscle can be changed to improve blood sugar control, ultimately to reduce the risk of developing heart disease, diabetes and stroke.During the study, 29 male patients with diabetes were investigated at baseline and at the end of the study after following a supervised endurance training program for a period of 8 weeks.Thirty male endurance athletes were investigated during highly trained status at baseline and at the end of the study after following a period of deconditioning for 4 weeks.The primary outcomes of this study were the assessment and comparison of non-invasive 1 H Magnetic Resonance Imaging of the vastus lateralis muscle to determine the saturated and unsaturated intramyocellular lipid storage before and after deconditioning (in athletes) or exercise training (in patients with diabetes).The secondary outcomes were the assessment and comparison of saturated and unsaturated lipid pool turnover examined with stable isotopes before and after deconditioning (in athletes) or exercise training (in patients with diabetes).

Participants
Age-matched male endurance-trained recreational competitive athletes and people with type 2 diabetes were enroled and provided informed consent.Recruitment commenced on 7 September 2016 and concluded on 11 January 2019.For a study of this size, only males were considered for enrolment due to known significant sex differences in intramyocellular lipid storage and utilisation between men and women 19 ).Athletes were recruited from cycling/running/triathlon clubs and had a ≥5-year history of active training in moderate-vigorous intensity ≥420 min/week (total physical activity level ≥570 min/week).Patients with type 2 diabetes were recruited from local primary care centres having been diagnosed by World Health Organisation criteria of fasting glucose and 2-h oral glucose tolerance test or HbA1c ≥ 48 mmol/mol, were treated by diet ± metformin and were sedentary.Exclusion criteria for patients were: known coronary artery disease, untreated endocrine conditions, estimated glomerular filtration rate <60 mL/min/1.73m 2 , blood pressure ≥ 160/100 mmHg, medication influencing glucose or fatty acid metabolism, such as peroxisome proliferator-activated receptors αor γ-agonists (fibrates or thiazolidinediones respectively), niacin, angiotensin-converting enzyme inhibitor therapy, omega-3 fatty acids or any clinical condition that, in the judgement of the investigators, may have interfered with exercise intervention, fatty acid metabolism or compromise the safety of the subject.
At screening, physical activity levels were assessed using the international physical activity questionnaire (IPAQ) and objectively, with a 7-day accelerometer (wGT3X-BT ActiGraph (Actigraph, USA)) to establish habitual baseline activity.All participants completed a one-week food diary and underwent ECG, echocardiography, cardiopulmonary exercise testing, and patients had gadolinium-enhanced cardiac magnetic resonance to exclude any previous myocardial infarction 14 .Participants were studied at baseline and after an exercise intervention (Fig. 1).Both study visits took place in the morning, after a 12-h fast and all study procedures were performed at the University of Aberdeen.
Each study visit comprised of 1 H-magnetic resonance spectroscopy, skeletal muscle biopsy, venous blood sampling and cardiopulmonary exercise testing.Eleven athletes and 12 patients with type 2  diabetes underwent a 4-h stable isotope infusion followed by repeat skeletal muscle biopsy.To eliminate any possibility of the isotope substudy results being influenced by the cardiopulmonary exercise testing, this was performed as the final procedure at the end of each study visit in the sub-study participants.
Non-invasive quantification of intramyocellular lipids by 1 H-magnetic resonance spectroscopy ( 1 H-MRS) This was performed on the right vastus lateralis using a 3 T scanner with a 16-channel receive-only torso coil (Philips Achieva MX; Phillips Healthcare, Best, NL).Participants were positioned supine, with the right leg as parallel to the magnetic field as possible 37 , to separation of intra-and extra-myocellular lipids.A skeletal muscle voxel was positioned to avoid extramyocellular lipid in adipose tissue or fascia and acquired with 1 H-MRS point-resolved spectroscopy (PRESS; Fig. 2A): short TE (TE/TR = 26/1500 ms, number of signals averaged = 128, voxel size 15x15x15 mm) 38,39 with variable pulse power and optimised relaxation delay water suppression 40 .Six saturation bands were interleaved with the water suppression pulses 41 and used for inner volume suppression to minimise the effect of chemical shift displacement 42,43 .B 0 shimming was performed using second-order pencil beam 44,45 .Commercially available spectral fitting software (LCModel Version 6.3, Oakville, ON, Canada) was used to analyse the data.Spectral peak areas related to different lipid components normalised to internal tissue water levels were used to calculate total intramyocellular lipid mass as well as the fractional lipid mass (fLM) [lipid/(lipid+water)] and fractions of saturated (fSL) (saturated/total) and unsaturated (fUL) (unsaturated/total) carbon bonds within intramyocellular lipids.The lipid mass may be estimated from the weighted sum of the lipid signals with weighting factors 13 accounting for the mass for a =CH-group (Lip52 and Lip53), 14 for -CH 2 -(Lip13 and Lip21) and 15 for the -CH 3 group (Lip09) .Calculations were based on the following formula: TOTAL LIPID = 1/2 * (Lip13 + Lip21) * 14 + (Lip52+Lip53) * 13 + 1/3 x (Lip09) * + 1/3 * (Lip09) * 16.The fraction of intramyocellular unsaturated lipids (fUL) was defined as the average percent of protons from double/triple bonds of all mono-and polyunsaturated fatty acids present in the muscle.We assumed one -CH 3 group per side chain-therefore, the intensity of the Lip09 resonance line scales with the number of fatty acid side chains.The 2.1 ppm peak (Lip21) scales with the number of unsaturated bonds, so: fUL = (3/2 * Lip21/Lip09), where 3/2 is the correction for number of protons, (2 for -CH 2 group and 3 for -CH 3 group of Lip09).The fraction of intramyocellular saturated lipid (fSL) (saturated/total intramyocellular lipids) was then calculated as: fSL = 1 -fUL.fLM was expressed in arbitrary units (A.U.) and fSL and fUL as percentages.
Skeletal muscle biopsies lipidomic, molecular biology, and isotope enrichment analysis Percutaneous biopsies were obtained through an aseptic minimally invasive technique 46 , using the Magnum ® Biopsy System (MG1522, Bard) with a 12-gauge core disposable biopsy needle (Magnum®, MN1210) 15 cm above the patella and from the same depth to avoid variability in muscle fibre composition 47 .The biopsy was immediately divided on ice into 3 fragments (5 mg each), snap frozen and placed in dry storage at -80 °C.

Lipidomic analysis of skeletal muscle
Skeletal muscle lipids were extracted according to Folch et al. 48.Briefly, tissue samples were homogenised with chloroform/methanol (2/1 Vol/Vol, ratio solvent/tissue mass 20 mL/1 g) containing 0.05% buthylydroxytolluene and while kept under nitrogen, agitated for 60 min at room temperature.The homogenate was then washed with 0.2 volume of 0.88% KCl solution.After vortexing for 10 s, the mixture was centrifuged (600 g) and the upper phase removed by siphoning.
The solvent phase was filtered and evaporated under nitrogen.Lipids extracted were then reconstituted in 300 μL methanol and kept under nitrogen atmosphere at -80C until analysis.17:0 ceramide 12:0/12:0 DAG (Avanti Polar Lipids, Alabaster, AL, USA) and 17:0/17:0/17:0 TAG (Sigma) were added as internal standards.Lipids were analysed by liquid chromatography-mass spectrometry (LC-MS) using a Thermo Exactive Orbitrap mass spectrometer coupled to a Thermo Accela 1250 UHPLC system with a water isopropanol/acetonitrile gradient.All samples were analysed in positive ion mode over the mass to charge ratio (m/z) range 250-2000.The data sets were processed with Progenesis QI software (Non-linear Dynamics, Newcastle, UK).Ion signals corresponding to the accurate m/z values for individual ceramide, diacylglycerol and triacylglycerol molecular species were extracted with the mass error set to 5 ppm.Quantification was achieved by relating the raw abundance of each lipid species to the raw abundance of internal standards.

Venous blood sampling
Venous samples were collected in EDTA and serum tubes.Plasma and serum samples were isolated by centrifugation (800 g, 10 min, 4 °C) and aliquots were stored at -80 °C.Serum glucose, cholesterol, triglycerides, high-and low-density lipoproteins and plasma free nonesterified fatty acid concentrations were analysed on a KONELAB30 automated analyser (Thermo Fisher Scientific, Waltham, MA, USA).Serum insulin concentration was measured using ELISA Kits (Mercodia, Uppsala, Sweden).Glycosylated haemoglobin (HbA1c) in patients with type 2 diabetes was measured in Aberdeen Royal Infirmary clinical biochemistry laboratory.Indirect indices of insulin resistance (Homeostasis Model Assessment 2, HOMA2 -IR 49 ) and sensitivity (Quantitative Insulin Sensitivity Check Index, QUICKI 50 ) were calculated.

Cardio-pulmonary exercise testing (CPET)
Maximal exercise capacity and oxygen consumption were assessed by a cycle ergometer (Ergoselect E100 K, Cosmed, Italy) and an integrated metabolic system (Quark PFT, Cosmed, Italy) on an incremental ramp protocol until volitional exhaustion.

Fatty acid turnover-mass spectrometry procedures
Plasma lipids were extracted from 400 μL aliquots 51 in the presence of 0.05% butylated hydroxytoluene.Non-esterified fatty acids were isolated by thin-layer chromatography on silica gel 60 plates (Merck Millipore) using elution system isohexane-diethyl ether-acetic acid (90:30:1 by vol) and FAME were prepared by acidic transesterification.Samples were reconstituted in 125 µL hexane; m/z ions of 270 and 286 were monitored for palmitic acid, and 294 and 312 for linoleic acid using electron impact gas chromatography mass spectrometry 52 with a 1-µL injection volume.Skeletal muscle biopsies (5 mg) were homogenised manually, and total lipids were extracted according to et al. 48in the presence of 0.05% butylated hydroxytoluene.FAME were prepared by transesterification with acidified methanol (1% H 2 SO 4 in methanol) adapted from Burge et al. 53 .Briefly, extracted lipids were mixed with toluene (1.0 ml) and incubated under nitrogen for 45 min at 80 °C with 1% in methanol (2 mL).After cooling, FAME were extracted by adding hexane (2 mL) and saturated NaCl solution (2 mL).After vortexing for 1 min, organic and aqueous phases were separated by centrifugation at 1125 g for 10 min at 8 °C and the hexane layer collected and stored under nitrogen at −80 °C before analysis by gas chromatography combustion isotope ratio mass spectrometry (GCCIRMS) as CO 2 with ions of 44, 45, and 46 monitored (isoprime recision fitted with an isoprime GC5 interface and an Agilent 7890B GC; Elementar, Stockport, UK).Both plasma and muscle enrichment data were expressed as tracer:tracee ratios (TTR).

Fatty acid turnover -Palmitic and linoleic fatty acid fractional incorporation rates
Dynamic isotope dilution method was used to study plasma palmitate and linoleate turnover 54 .The GCMS m + 16 and m + 18 isotopomers were used to determine the rate of appearance (Ra) of palmitate and linoleate in plasma as described previously 55 : Ra (mmol/h) = (infusion rate mmol/h * f 1 )/(mean plasma TTR), where f 1 is the isotopic purity of either m + 16 (0.816) in the palmitate infusate or m + 18 (0.779) in the linoleate infusate.The mean plasma TTR was based on samples taken from 2-4 h of infusion.The fractional incorporation rate (FIR) for skeletal muscle palmitate and linoleate was calculated using the following formula 56 : FIR (%/h) = (total 13 C at t 4total 13 C at t 0 ) * 1/(AUC plasma TTR * f 2 )*100, where t 4 = time of post-infusion biopsy (4 h) and t 0 is time of pre-infusion biopsy (background enrichment).The area under the curve (AUC) for plasma enrichments of each fatty acid was calculated by trapezoid analysis based on the ten time points for the m + 16 and m + 18 isotopomer enrichments, respectively.The factor f 2 was to adjust the m + 16 (* 1.202) and m + 18 (* 1.253) values to total fatty acid 13 C supply which was calculated using selective ion monitoring of the methyl esters with summation of the m/z ion currents.

Exercise interventions and compliance monitoring
Athletes were asked to stop all structured exercise training (limit any structured physical activity ≤ 10 min/day for four consecutive weeks).Compliance with deconditioning was monitored with a 7-day accelerometer.Patients with type 2 diabetes were enroled in an 8-week supervised endurance exercise training programme of 5 sessions/week and each session consisted of 60 min cycling (Ergoselect E100 K,Cosmed, Italy) at 65%-85% of peak oxygen consumption.Cardiopulmonary exercise testing was used fortnightly during the training period to re-test peak oxygen consumption and increase intensity of the exercise accordingly.Throughout the study period, participants were asked to not modify their nutritional habits and completed a oneweek food diary during the exercise intervention.

Statistical analysis
Statistical analyses were conducted using SPSS v29.Due to the lack of any spectroscopic data on the fractions of saturated carbon bonds in patients with type 2 diabetes, the sample size was estimated using our own data available from non-trained, young, healthy controls whose saturated intramyocellular lipids were measured at 89.3 ± 4.3%.We Table 6 | Primary and secondary antibodies used for Western blot analysis aimed to detect a 10% difference between type 2 diabetes and athletes as well as a 10% change after exercise training in type 2 diabetes patients, with a standard deviation in athletes of 5; for type 2 diabetes patients we had no prior data so predicted a higher standard deviation of 14 (different duration of disease, sedentarism, higher BMI, medications).For a study power of 0.8 and significance level of 0.05, 25 participants per group were required for detecting betweengroup differences and 22 participants were required in the type 2 diabetes for within group changes.For the stable isotope turnover analyses, we used the palmitate synthesis rate reported by Perreault et al. 57 of 0.23 ± 0.04/h in pre-diabetes patients.We prespecified a change of 20% as the expected effect of exercise training in type 2 diabetes patients, therefore a population mean difference =0.04, with a SD for the difference 0.04 (for studying within group changes in male participants only), study power 0.8, significance 0.05.This required a sample size of n = 10 in the diabetes group, and thus we matched the athletes' group, without prespecifying any athlete-diabetes differences.We anticipated a higher drop-out rate than the one reported 58 .
Both the primary end-point (intramyocellular saturation abundance) and secondary end-points (palmitate and linoleate turnover) analysis were performed on a per protocol set of patients with valid measurements available at baseline and post interventions.Response (ie, outcome) to exercise interventions is presented as the change (variable, change) from baseline to post-intervention calculated for each participant by subtracting the baseline variable measurement from the post-intervention measurement.Two-tailed-tests assessed the significance of between athletes and type 2 diabetes patients differences in their pre-to-post intervention changes.Baseline and postinterventions comparisons between groups were performed using appropriate two tailed t tests, after testing for normality.Tabulated data is presented as mean (95% confidence intervals) of within or between group changes/differences.All other variables reported were analysed in a similar manner.Any other numeric data are presented as mean and standard deviation.
Multiple comparisons were not adjusted for, as many of the differences we report in this exploratory study have clear significance that would remain after multiple adjustment, but should be considered where there is nominal statistical significance.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
The source data and lipidomic source data generated in this study have been deposited in the Figshare Database (https://doi.org/10.6084/m9.figshare.24219934).Participant level data for the study is also available in the Figshare Database.The Lipidomics Minimal Reporting Checklist is included in the Supplemental Material.The trial protocol can be made available upon request by contacting the Corresponding Author.Requests for data should be made and these requests will be fulfilled by the corresponding author (dana.dawson@abdn.ac.uk), providing the data will be used within the scope of the originally provided informed consent.The corresponding author aims to respond to data requests within three months.The data will be made available for a year following publication of the manuscript.

Fig. 1 |
Fig. 1 | Study consort and flow diagram, showing selection of the two study populations after screening, baseline study investigations (V1), exercise interventions and post-intervention study investigations (V2).HbA1C-

Fig. 2 |
Fig. 2 | 1 H-Magnetic resonance spectroscopy of total, saturated and unsaturated intramyocellular lipid bonds.2A: Top: Representative localisation of the spectroscopy voxel for 1 H-magnetic resonance of the right vastus lateralis.Middle: Colour-coded schematic showing protons connected to carbon nuclei in single (saturated) covalent bonds in blue and red and protons connected to double/triple (unsaturated) carbon bonds or adjacent to double/triple bonds, which are shown in green and purple, respectively.Bottom: The same colour-coding shows how the different proton species process at different frequencies in the magnetic field, generating different spectral peaks as shown by the different chemical shifts (parts per million, ppm) of saturated (0.9 ppm, 1.3 ppm) and unsaturated (2.1 ppm, 5.2 to 5.3 ppm) intramyocellular lipid peaks.Main extramyocellular lipid (EMCL) peak is separated at 1.5 ppm (yellow).2B: Intramyocellular total lipids and fractions of saturated and unsaturated carbon bonds within intramyocellular lipid (n = 25 athletes and n = 22 patients with type 2 diabetes), Arb U = arbitrary units.Data is shown as individual data points with means and error bars for standard deviation.Baseline and post-interventions comparisons between groups were performed using t-tests.Top: Intra-myocellular fractional lipid mass in athletes and type 2 diabetes patients before and after exercise intervention Middle: Intra-myocellular fraction of saturated lipids in athletes and type 2 diabetes patients before and after exercise intervention.Bottom: Intra-myocellular fraction of unsaturated lipids in athletes and type 2 diabetes patients before and after exercise intervention.

-
Athletic healthy volunteer baseline -Athletic healthy volunteer after deconditioning -Type 2 diabetes patient baseline -Type 2 diabetes patient after exercise training

Fig. 3 |
Fig. 3 | Skeletal muscle saturated and unsaturated lipid turnover and insulin receptor pathway/metabolic sensing.Data are shown as individual data points with means and error bars for standard deviation.Arb U arbitrary units.3A: Top: Fractional incorporation rate for [U −13 C] Palmitate (16:0) before and after exercise intervention in athletes and patients with type 2 diabetes (n = 11 athletes and n = 12 patients with type 2 diabetes) Baseline and post-interventions comparisons between groups were performed using t tests.Bottom: Fractional incorporation rate for [U −13 C] Linoleate (18.2 n-6) before and after exercise intervention in athletes and patients with type 2 diabetes (n = 11 athletes and n = 12 patients with type 2 diabetes).3B: Top: Ratio of phosphorylated protein kinase B (AKT) to total of AKT (p/t AKT) protein levels before and after exercise intervention in athletes and patients with type 2 diabetes (n = 8 in each group).Baseline and post-interventions comparisons between groups were performed using t tests.Bottom: Ratio of phosphorylated 5' adenosine monophosphate-activated protein kinase (AMPK) to total of AMPK (p/t AMPK) protein levels before and after exercise intervention in athletes and patients with type 2 diabetes (n = 8 in each group).Below are representative western blot examples from n = 4 in each group.
Baseline differences between athletes and type 2 diabetes patients p-value Athletes change between baseline and deconditioning p-value Patients with type 2 diabetes change between baseline and exercise training p-value Difference in changes between athletes and type 2 diabetes patients are shown as mean and 95% confidence intervals of inter-or intra-group differences/changes parametric hypothesis tests.Athletes n = 28; type 2 diabetes patients n = 27.LDL low-density lipoprotein, HDL high-density lipoprotein, NEFA non-esterified (free) fatty acids, QUICKI QUantitative Insulin-sensitivity ChecK Index, HOMA2-IR The homoeostasis model assessment 2 insulin resistance.https://doi.org/10.1038/s41467-024-47843-yNature Communications | (2024) 15:3690

Table 1 |
Weight and cardio-pulmonary exercise test performance in athletes and patients with type 2 diabetes mellitus before and after exercise interventions

Table 2 |
Total, saturated and unsaturated intramyocellular lipid bonds in athletes and Type 2 diabetes patients before and after exercise interventions Data are shown as mean and 95% confidence intervals of inter-or intra-group differences/changes parametric hypothesis tests.Athletes n = 25; type 2 diabetes patients n = 25.Arb U arbitrary units.

Table 5 |
Metabolic profile of athletes and patients with type 2 diabetes mellitus in basal conditions before and after exercise interventions