Introduction

The role of mitochondrial metabolism in the pathophysiology of insulin resistance is currently much debated, particularly with regard to diabetic individuals. On the one hand some scientists have proposed that compromised mitochondrial oxidative function, particularly in skeletal muscle, is responsible for lipid accumulation and development of insulin resistance [1, 2]. There is thus no doubt that insulin-resistant, obese individuals affected by type 2 diabetes have approximately 30% less mitochondria in their skeletal muscles than age-matched healthy controls [3]. Some studies have reported that the mitochondrial alterations found in muscle of type 2 diabetic patients reflect a functional impairment of mitochondria, since these alterations are present even after correction by mitochondrial mass [4, 5]. However, in other studies, differences in electron transport chain or in oxygen consumption have not been detected after correction by surrogates of mitochondrial mass [1, 6]. On the other hand, other scientists have serious reservations about a causative role of mitochondrial dysfunction in this condition. Thus, Asian Indians displaying higher mtDNA content and increased oxidative enzyme activity are more insulin resistant than age-, sex- and BMI-matched North American counterparts [7]. Furthermore, no deficit in mitochondrial function was found in skeletal muscles from obese patients in early or advanced stages of type 2 diabetes [8].

Weight reduction and physical exercise are the best approaches to improve insulin sensitivity [9]. However, compliance to lifestyle changes has been proven to be poor in the long term [10]. In the last decades, bariatric surgery has emerged as a potential therapy for diabetes [11]. Bilio-pancreatic diversion (BPD) is a bariatric surgical technique characterised by massive weight loss mainly due to lipid malabsorption [12, 13]. BPD causes a net improvement in insulin sensitivity, long before normalisation of body weight [14]. In addition, BPD regulates substrate oxidation, modulating the expression of genes involved in lipid synthesis [15, 16] and oxidation [15, 17] in both muscle and adipose tissue.

BPD operation caused a net improvement of insulin sensitivity in type 2 diabetic and in non-diabetic patients [14]. We have previously described increased MFN2 mRNA expression in skeletal muscle from morbidly obese patients with normal glucose tolerance (NGT) after BPD [18]. In the present study, we aimed to determine whether the expression of genes involved in mitochondrial biogenesis/function was induced in response to BPD. To this end we selected: (1) nuclear genes that regulate mitochondrial biogenesis such as PGC-1α (also known as PPARGC1A), PGC-1β (also known as PPARGC1B) and PPARδ (also known as PPARD); (2) genes that regulate mitochondrial metabolism and fusion such as MFN2 [19, 20]; (3) genes that regulate peroxisome proliferator-activated receptor γ coactivator 1 (PGC-1α) such as SIRT1 [21, 22]; and (4) genes that encode for constitutive proteins such as porin or citrate synthase.

In addition, and being interested in the mechanisms leading to the reversal of diabetes after BPD, we investigated the effect of comparable weight loss caused by BPD on the expression of those genes mentioned above, and whether the response differed between morbid obese NGT or type 2 diabetic patients. All patients were characterised before and after weight loss with respect to insulin sensitivity by the euglycaemic–hyperinsulinaemic clamp; their glucose and lipid oxidation was assessed by the respiratory chamber.

Methods

Patients

We studied 21 morbidly obese patients undergoing BPD. Before surgery, 11 patients had NGT and ten had overt type 2 diabetes. Diabetes was newly diagnosed in all diabetic patients at the time of the study, and they were not taking any oral hypoglycaemic agent or insulin. The average HbA1c was 7.93 ± 1.61%. The clinical study was performed twice, once within 2 weeks before the bariatric procedure, and the second time 24 months later in the 2 weeks before plastic reconstruction of the abdominal wall. During surgery, abdominal muscle tissue biopsies were obtained from rectus abdominis muscles. The anthropometric, clinical and metabolic characteristics of the study patients are shown in Tables 1 and 2. The study protocol was approved by the Institutional Review Board and Ethics Committee of the Catholic University–Policlinico Gemelli (Rome, Italy).

Table 1 Anthropometric and metabolic variables measured in NGT and type 2 diabetic patients before and after weight loss induced by BPD
Full size table
Table 2 Insulin sensitivity and dynamic metabolic variables of the patients
Full size table

BPD

This malabsorptive surgical procedure [23] consists of an approximately 60% distal gastric resection with stapled closure of the duodenal stump. The residual volume of the stomach is about 300 ml. The small bowel is transected at 2.5 m from the ileo-caecal valve and its distal end is anastomosed to the remaining stomach. The proximal end of the ileum, comprising the remaining small bowel (involved in carrying bilio-pancreatic juice but excluded from food transit), is anastomosed in an end-to-side fashion to the bowel, 50 cm proximal to the ileo-caecal valve. Consequently, the total length of absorbing bowel is reduced to 250 cm, the final 50 cm of which, the so-called common channel, represents the site where ingested food and bilio-pancreatic juices mix.

Body composition

During the two study stages (before and after BPD), body weight was measured to the nearest 0.1 kg by a beam scale and height to the nearest 0.5 cm using a stadiometer (Holatin, Crosswell, UK). Waist circumference was measured just above the uppermost lateral border of the right ileum using the NHANES protocol [24]. Hip circumference was measured at the maximum extension of the buttocks as viewed from the side. Body composition was estimated by isotopic dilution [25, 26]. Total body water was determined using 0.19 Bq of 3H-labelled water in 5 ml of saline solution administered as an intravenous bolus injection. Blood samples were drawn before and 3 h after the injection. Radioactivity was determined in duplicate on 0.5 ml of plasma using a beta-scintillation counter (1600TR; Canberra-Packard, Meriden, CT, USA). Corrections were made (5%) for non-aqueous hydrogen exchange; water density at body temperature was assumed to be 0.99371 kg/l. Total body water (kg) was computed as 3H2O dilution space (litres) × 0.95 × 0.99371. The within-patient CV for this method is 1.5%. Free-fat mass (FFM) (kg) was obtained by dividing total body water by 0.732 [25].

Euglycaemic–hyperinsulinaemic clamp procedure

Peripheral insulin sensitivity was evaluated by a 2 h euglycaemic–hyperinsulinaemic clamp. The fasting plasma glucose concentration was maintained throughout the insulin infusion by means of a variable glucose infusion and blood glucose determinations every 5 min. Whole-body glucose uptake (M value) in μmol (kgFFM)−1 min−1 was determined during a primed constant infusion of insulin (at the rate of 6 pmol kg−1 min−1). Value of whole-body glucose uptake was normalised for FFM. The quantitative insulin sensitivity check index (QUICKI) was calculated as reported by Katz et al. [27].

Analytical procedures

See Electronic supplementary material (ESM).

Substrate oxidation studies

The patients spent a day (starting at 08.00 hours) in the respiratory chamber of the Metabolism Unit of the Catholic University School of Medicine in Rome, after an overnight fast. The characteristics of the device have been described previously [28]. All patients were assigned a diet with an energy content reflecting their habitually consumed diet. The food given and returned was weighed to the nearest gram on precision scales (KS-01; Rowenta, Berlin, Germany). Resting energy expenditure (REE) and respiratory quotient (RQ) were calculated from oxygen consumption, carbon dioxide production and nitrogen urinary excretion according to standard formulas [28]. Results are presented as kJ/24 h and after normalisation to FFM (kJ [24 h]−1 [kg FFM]−1). Indirect calorimetry [29] was used to determine O2 consumption and CO2 production during the 30 min preceding the clamp and in the last 80 min of the clamp (in steady-state condition) to estimate net rates of carbohydrate and lipid oxidation (SensorMedics 2900; SensorMedics, Yorba Linda, CA, USA). Non-oxidative glucose disposal rate was calculated as total glucose disposal rate minus glucose oxidation rate.

RNA extraction and real-time PCR analysis

See ESM.

Statistical analysis

Data analyses were performed with SPSS statistical software (SPSS, Chicago, IL, USA). Data are reported as mean ± SD, unless otherwise specified. Paired and unpaired t tests were performed to compare data from the same patients before and after BPD and data of NGT and type 2 diabetic patients after BPD.

To assess differences between the two groups, analysis of covariance (ANCOVA) was performed. The dependent variable was the delta value (calculated as the value before the surgery minus the value after BPD), the fixed factor was the time (before BPD) and the covariate used in the analysis was the baseline value for the variable under investigation. Predictors of MFN2 and PGC-1β mRNA level changes were tested using the Pearson correlation. Distribution statistics for the residuals were calculated to check whether assumptions of normality were met (i.e. skewness and kurtosis <2.0). Multiple linear regression analysis was then used to fit models to predict the determinants of fasting glucose and fasting lipid oxidation. Variables were allowed to enter the models if significant at the <0.05 probability level.

Results

Anthropometric and plasma variables in normal glucose-tolerant or type 2 diabetic morbidly obese patients before and after BPD

Table 1 shows anthropometric and body composition data, measured before and 2 years after BPD, in obese NGT and obese type 2 diabetic patients. After BPD, BMI significantly decreased in both groups, with concomitant reductions in fat mass (68.4 and 65.9%, respectively) and FFM (22 and 21.1%, respectively). As to metabolic variables, BPD significantly decreased plasma levels of glucose, insulin, triacylglycerol and cholesterol, while HDL-cholesterol levels increased after surgery in both groups (Table 1). As shown in Table 1, ANCOVA analysis showed no statistically significant differences between groups.

Morbidly obese NGT and type 2 diabetic patients show improved insulin sensitivity but a different substrate oxidation profile in response to BPD

BPD resulted in the normalisation of insulin sensitivity. As reported in Table 2, whole-body glucose disposal (M value) significantly increased both in NGT (from 11.38 ± 1.72 to 42.07 ± 5.72 μmol [kg FFM]−1 min−1) and type 2 diabetic patients (from 13.43 ± 2.05 to 41.01 ± 5.38 μmol [kg FFM]−1 min−1). No differences in M values were detected in NGT and in diabetic patients, which is in keeping with previous reports [14, 30]. Plasma insulin concentrations during the clamp were similar in all patients studied (Table 2). We also calculated the QUICKI index as a surrogate of hepatic insulin sensitivity [31]. The QUICKI index was significantly increased after BPD, both in type 2 diabetic and in NGT patients (Table 2). Altogether these data suggest that BPD causes a net improvement not only of glucose disposal mediated by peripheral insulin, but also of hepatic insulin sensitivity.

However, NGT and type 2 diabetic patients showed a different capacity to oxidise substrates under fasting conditions and during the clamp. After BPD, both groups increased glucose oxidation during fasting but the increase was much greater in NGT than in diabetic patients (3.1- and 1.6-fold respectively). Conversely, both groups showed decreased lipid oxidation during fasting, the effects being greater in NGT than in diabetic patients (44 and 24% decrease respectively). ANCOVA analysis provided further evidence that the changes induced by BPD in fasting glucose and lipid oxidation were significantly different in NGT and type 2 diabetes patients (p = 0.038 and p = 0.005, respectively) (Table 2). During the clamp, glucose oxidation markedly increased in response to BPD in NGT and type 2 diabetic patients, while lipid oxidation decreased in the former but did not change in the latter after BPD (Table 2). These results again highlight a different capacity of the two groups (NGT and type 2 diabetics) to oxidise lipids during the clamp (p < 0.0001).

The respiratory quotient at fasting and during the clamp in the post-surgery state was markedly lower in diabetic patients than in NGT patients (Table 2), indicating that diabetic patients oxidise more lipids. ∆RQ values were not statistically different in the two groups (Table 2). Neither the change in REE nor that in REE/FFM before and after BPD were different between the two groups (Table 2).

Overall, these data indicate that the effect of BPD in terms of substrate utilisation, whether either fasting or insulin-mediated, is larger in NGT patients.

Morbidly obese type 2 diabetic patients show impaired muscle capacity to induce expression of nuclear genes encoding mitochondrial proteins in response to BPD

To further investigate the molecular basis for the differential metabolic adaptations of NGT and type 2 diabetic patients to BPD, and also because it has been reported that genes involved in mitochondrial function are induced in response to weight loss in morbidly obese patients [18], we studied the expression of genes encoding mitochondrial proteins. To this end, abdominal skeletal muscle biopsies were collected before and after BPD and used for gene expression analysis. In particular, real-time PCR analysis was used to evaluate mRNA levels of a marker of mitochondrial fusion that regulates mitochondrial activity (MFN2), two genes that encode for constitutive proteins (PORIN [also known as VDAC1] and CS) and a gene encoded by mitochondrial DNA (COXIII). As shown in Fig. 1, BPD resulted in a significant increase of MFN2, CS and PORIN mRNA levels in NGT patients (1.63-, 2.45- and 2.99-fold increase, respectively). In contrast, MFN2 expression significantly decreased in type 2 diabetic patients (by 0.48-fold) and no changes were detected in CS or PORIN expression in the type 2 diabetic group after surgery (Fig. 1). COXIII transcripts levels were not modulated by BPD either in the NGT or in the type 2 diabetic group (Fig. 1).

Fig. 1
figure 1

Skeletal muscle gene expression of mitochondrial proteins in NGT (white bars) and type 2 diabetic patients (black bars) in response to BPD. Real-time PCR was performed in abdominal skeletal muscle biopsies from NGT and type 2 diabetic patients before and 2 years after BPD surgery. a MFN2; *p = 0.0311, †p = 0.0003 in NGT; *p = 0.00003 in type 2 diabetic; b CS; *p = 0.02, †p = 0.0083; c PORIN; *p = 0.033, †p = 0.034; and d COXIII. Data are means ± standard deviation. Statistical analyses comparing pre-surgery and post-surgery state were performed by paired t test (*). Statistical analyses comparing post-surgery between NGT and type 2 diabetic patients were performed by unpaired t test (†)

Full size image

These results indicate a fundamental difference between NGT and type 2 diabetic patients in the regulation of nuclear expression of genes encoding mitochondrial proteins. Thus NGT patients experienced induction of MFN2, PORIN or CS, whereas type 2 diabetic patients had no capacity to do so.

Morbidly obese type 2 diabetic patients show impaired muscle capacity to induce gene expression of PGC-1α, PGC-1β and SIRT1 in response to BPD

Next, we studied whether BPD affects the expression of genes encoding transcriptional regulators of mitochondrial function. Hence, mRNA levels of PGC-1β, PGC-1α, ERRα (also known as ESRRA) and PPARδ in abdominal skeletal muscle were measured before and after BPD. As shown in Fig. 2, PGC-1β mRNA levels significantly increased after surgery (2.04-fold) in the NGT group, but not in type 2 diabetic patients. PGC-1α and PPARδ mRNA levels increased (1.53- and 1.75-fold respectively; differences not significant) in NGT patients, but not in diabetic patients in response to BPD and as a result marked differences were detected between NGT and diabetic patients after surgery (Fig. 3). No differences were observed in the expression of ERRα in either group after BPD (Fig. 2).

Fig. 2
figure 2

Skeletal muscle gene expression of transcriptional regulators of mitochondrial function in NGT (white bars) and type 2 diabetic patients (black bars) in response to BPD. Real-time PCR was performed as in Fig. 1. a PGC-1β; *p = 0.046, †p = 0.049; b PGC-1α; †p = 0.048; c PPAR-δ; †p = 0.045; and d ERRα. Data are means ± standard deviation. Statistical analyses comparing pre-surgery and post-surgery state were performed by the paired t test (*). Statistical analyses comparing post-surgery between NGT and type 2 diabetic patients were performed by the unpaired t test (†)

Full size image
Fig. 3
figure 3

SIRT1 gene expression is upregulated in the abdominal skeletal muscle of NGT (white bars) as compared with type 2 diabetic patients (black bars) after BPD. Data are means ± standard deviation. Statistical analyses comparing pre-surgery and post-surgery state were performed by the paired t test (*). Statistical analyses comparing post-surgery between NGT and type 2 diabetic patients were performed by the unpaired t test (†). *p = 0.049, †p = 0.020

Full size image

SIRT1, a NAD+-dependent deacetylase, regulates PGC-1α activity in liver and muscle tissues [21, 22]. Based on the observation that muscle PGC-1α is induced in NGT patients but not in type 2 diabetic patients after BPD, we analysed whether SIRT1 participates in the differential gene expression pattern detected in the two groups. In keeping with previous data, SIRT1 mRNA levels significantly increased after BPD (2.27-fold) in the NGT group, but not in the diabetic group (Fig. 3).

Substrate oxidation differentially correlates with gene expression in NGT and type 2 diabetic patients

To gain more insights into the different capacity of the NGT group to oxidise substrates compared with diabetic patients, correlation analyses between glucose and lipid oxidation during fasting with the levels of the different transcripts (before and after BPD) were performed. Pearson’s correlation results for NGT and diabetic patients are reported in Table 3. PPAR-δ, ERR-α and SIRT1 mRNA levels did not correlate with any of the variables under study (data not shown). Interestingly, MFN2 mRNA levels positively correlated with fasting glucose oxidation and negatively with fasting lipid oxidation in the NGT group. In contrast, MFN2 mRNA levels did not correlate with any of the above-mentioned variables in type 2 diabetic patients. Similarly, the expression of CS, PORIN, PGC-1α or PGC-1β positively correlated with fasting glucose oxidation in NGT patients (Table 3). Analysis of delta values also revealed a positive correlation between the expression of genes (CS or PGC-1α) and fasting glucose oxidation in NGT, but not in diabetic patients (Table 3).

Table 3 Correlation coefficients between different variables in NGT and type 2 diabetic patients for glucose and lipid oxidation
Full size table

A stepwise linear regression analysis was performed in the NGT group, including in the model MFN2, CS, PORIN, PGC-1α and PGC-1β as independent variables, and fasting glucose oxidation as the dependent variable. MFN2 expression turned out to be the main predictor of fasting glucose oxidation (R 2 = 0.496, p = 0.001), as shown in Fig. 4a. Such correlations, however, were lost in the diabetic group, as illustrated by the scatter plot in Fig. 4b. The expression of CS, PORIN, PGC-1α or PGC-1β did not predict glucose oxidation (data not shown).

Fig. 4
figure 4

MFN2 mRNA levels in the abdominal skeletal muscle correlated with fasting glucose oxidation in the NGT group. Simple scatterplots of the correlations between MFN2 mRNA expression and fasting glucose oxidation a in the NGT group \( \left( {y = - 0.3164 + 1.6908x} \right) \) and b in type 2 diabetic patients \( \left( {y = 1.7708 - 0.7494x} \right) \). Lines indicate the best fit linear regression

Full size image

CS expression was a good predictor of fasting lipid oxidation (R 2 = 0.311, p = 0.009), whereas MFN2 or PORIN expression were not determinants of lipid oxidation (data not shown). In all, our data indicate that the modulation of glucose oxidation after BPD in NGT patients seems to be dependent upon changes in MFN2 expression.

Discussion

A major finding of this study is the differential pattern of changes in the expression of genes involved in the regulation of mitochondrial biogenesis/activity in skeletal muscle from morbidly obese type 2 diabetic and NGT patients in response to BPD. Thus, whereas the expression of some mitochondrial genes, including MFN2, PORIN or CS, of mitochondrial regulatory genes such as PGC-1β, PGC-1α, PPAR-δ or SIRT1 increased in the skeletal muscle of NGT patients 2 years after BPD, no changes were detected in type 2 diabetic patients. This pattern of changes was parallel to a differential oxidative profile: whereas glucose oxidation during fasting was higher in the NGT group after BPD, lipid oxidation levels (both in the fasting state and during the euglycaemic–hyperinsulinaemic clamp) were higher in the type 2 diabetic group after BPD. The BPD-induced improvement of insulin sensitivity, as shown by insulin-mediated glucose uptake and oxidation, was comparable in both patient groups. This suggests that improvement of insulin sensitivity is independent of the expression of genes involved in mitochondrial activity, and therefore the mechanisms of increased insulin sensitivity triggered by BPD are of a different nature.

The pattern of changes detected in response to BPD consists of enhanced expression of genes encoding functional and regulatory components of mitochondria such as CS, PORIN or MFN2, as well as of genes encoding regulators of mitochondrial biogenesis such as PGCs, ERRα and activators of mitochondrial function such as PPARδ. We propose that the physiological purpose of the pattern of changes in gene expression detected in skeletal muscle in response to BPD is to trigger mitochondrial biogenesis in skeletal muscle, which favours increased glucose oxidation, under conditions of reduced lipid availability. This needs to be confirmed by direct analysis in muscle, but is consistent with the observations that whole-body glucose oxidation was increased. Patients undergoing BPD have reduced lipid availability provided that BPD causes lipid malabsorption. That is probably the reason why we found lower whole-body lipid oxidation after BPD in NGT patients.

Our study shows some limitations, namely: (1) diet was not controlled in the two groups of patients and patients did not follow any specific dietary regimen or have any dietary restrictions; (2) the patients had a sedentary lifestyle either before or after BPD, but this was not controlled; and (3) only gene expression was monitored, so there is a need to determine levels of proteins involved in mitochondrial biogenesis/function in skeletal muscle.

Mitofusin-2 (MFN2) is a mitochondrial fusion protein that regulates mitochondrial metabolism [19, 20]. MFN2 loss of function reduces glucose oxidation and mitochondrial membrane potential in muscle and non-muscle cells [19, 20]. Similarly, MFN2 gain of function enhances glucose oxidation and mitochondrial membrane potential [20]. The mechanisms are only partially known and seem to be dependent on the expression of subunits of the oxidative phosphorylation system [20]. In this study we have demonstrated that MFN2 is upregulated in skeletal muscle from morbidly obese NGT patients in response to BPD, which correlates positively with glucose oxidation. This is in keeping with previous findings observed in women subjected to BPD [18] or in patients in whom weight loss was induced by Roux-en-Y gastric bypass [32]. Our data additionally indicate that MFN2 expression is a good predictor of fasting glucose oxidation in NGT patients, much better than the expression of genes encoding for regulators of mitochondrial biogenesis such as PGC-1α or PGC-1β, and also better than the expression of genes encoding for constitutive proteins such as porin or citrate synthase (and therefore markers of mitochondrial mass). MFN2 induction was not detected in skeletal muscle from type 2 diabetic patients in response to BPD (in fact, a reduction was observed) and these patients also showed very small changes in substrate oxidation. In all, our results further support the view that MFN2 expression is a positive regulator of glucose oxidation.

The nuclear transcription factor PPARδ and the nuclear co-activators PGC-1α or PGC-1β regulate mitochondrial function and/or mitochondrial biogenesis in muscle [33]. Similarly, both regulate MFN2 transcription through activation of ERR-α [34]. Based on these observations, we studied whether expression of PGC-1α, PGC-1β or PPAR-δ may explain the regulation of MFN2, PORIN or CS in skeletal muscle from morbidly obese patients after BPD. PGC-1α, PGC-1β and PPAR-δ, as well as SIRT1, a modulator of PGC-1α [21, 22], were upregulated in NGT patients after BPD. In contrast, the expression of PGC-1α, PGC-1β, PPAR-δ and SIRT1, as well as that of those genes encoding mitochondrial proteins, was unaltered in morbidly obese type 2 diabetic patients in response to BPD, despite massive weight loss and normalisation of insulin sensitivity.

Our data indicate that BPD triggers a complex set of events (perhaps due to malabsorption, reduced food intake, partial loss-of-function of the intestinal tract induced by BPD) that induces genes critical to mitochondrial function such as PGC-1α, PGC-1β, PPAR-δ; as a result, Mfn2 expression and glucose oxidation are concomitantly enhanced. Increased PGC-1α expression has also been reported in muscle from obese patients after weight loss induced by Roux-en-Y gastric bypass [32]. In contrast, PGC-1α, PGC-1β, PPAR-δ and SIRT1 were not induced in morbidly obese type 2 diabetic patients in response to BPD, which explains the lack of stimulation of genes encoding for mitochondrial proteins (citrate synthase, porin or MFN2) and the blunted changes of glucose and lipid oxidation particularly under fasting conditions.

In different insulin-resistant states, including type 2 diabetes, a defective mitochondrial function has been reported in skeletal muscle [1, 2, 35–37]. Muscle mitochondria from patients with type 2 diabetes show reduced size, reduced activity of the electron transport chain and reduced ADP-stimulated and maximal mitochondrial respiratory capacity [1, 36, 37]. In parallel, type 2 diabetes is associated with reduced expression of genes of oxidative metabolism as well as with repression of MFN2 [38–40]. In skeletal muscle of non-diabetic individuals with a family history of type 2 diabetes, decreased expression of nuclear genes encoding proteins of oxidative phosphorylation has been reported [39, 40], along with reduced in vivo oxidative phosphorylation [35, 41]. However, the role of mitochondria in the pathogenesis of insulin resistance and diabetes is still unclear. In vastus lateralis muscles from obese insulin-resistant patients after weight loss, no increase in oxidative enzyme activities was reported, in spite of an improvement of insulin sensitivity [42]. In addition, Toledo and collaborators [37] have reported that weight loss, without physical exercise, in obese patients ameliorates insulin sensitivity with no effects on muscle mitochondrial capacity.

Our findings suggest that weight loss consequent to BPD exerts a beneficial effect on insulin sensitivity via mechanisms that are independent of skeletal muscle expression of genes involved in mitochondrial biogenesis/function. Furthermore, the observation that gene expression is not altered with weight loss in type 2 diabetic patients while it is induced in patients with NGT suggests the presence of a heritable component.