Branched-Chain Amino Acid Metabolism is Regulated by ERRα and is Further Impaired by Glucose Loading in Type 2 Diabetes

Increased levels of branched-chain amino acids (BCAAs) are associated with type 2 diabetes (T2D) pathogenesis. However, most metabolomic studies in T2D are limited to an analysis of plasma metabolites under fasting conditions, rather than the dynamic shift in response to a glucose challenge. Moreover, metabolomic profiles of peripheral tissues involved in glucose homeostasis are scarce and the transcriptomic regulation of genes involved in BCAA catabolism in T2D is partially unknown. Using metabolomic and gene expression approaches, we found that impairments in BCAA catabolism in T2D patients under fasting conditions are exacerbated after a glucose load, concomitant with downregulated expression of BCAA-related genes in skeletal muscle. We identified a key regulatory role for Estrogen-Related Receptor α (ERRα) in PGC-1α-mediated upregulation of BCAA genes and leucine oxidation. Thus, metabolic inflexibility in T2D impacts BCAA homeostasis and the transcriptional regulation of BCAA genes via a PGC-1α/ERRα-dependent mechanism.


INTRODUCTION
Type 2 diabetes (T2D) is a chronic metabolic disease characterized by chronic hyperglycemia and insulin resistance (Leahy, 2005). These metabolic derangements severely affect pathways controlling the appropriate sensing and handling of nutrients, thereby leading to a positive energy balance and metabolic inflexibility, which further compromises whole body glucose homeostasis (Galgani et al., 2008). Overnutrition and T2D-related disturbances also affect non-glucose metabolites such as branched-chain amino acids (BCAAs; leucine, isoleucine and valine) (Felig et al., 1969), essential amino acids whose utilization and metabolism are exquisitely regulated in healthy individuals. While BCAA and related metabolites play a role in protein synthesis, they also modulate liver gluconeogenesis and lipogenesis rates, cell growth and nutrient signaling, and can enter the tricarboxylic acid cycle to produce energy (Nie et al., 2018). Under pathological conditions, altered levels of BCAAs, especially in a context of overnutrition, can disrupt insulin sensitivity and secretion (Patti et al., 1998).
Obese, insulin resistant and T2D individuals exhibit higher serum concentrations of BCAAs than their healthy counterparts (Huffman et al., 2009). Metabolomic profiling of blood metabolites has revealed a signature of altered BCAA catabolism in obese individuals, with a strong association with insulin resistance . In T2D, BCAA-related metabolites are predictive of disease pathogenesis (Wurtz et al., 2013), as well as prognostic markers for intervention outcomes (Menni et al., 2013;Shah et al., 2012). Moreover, Mendelian randomization analysis suggested a causal link between genetic variants associated to impaired BCAA catabolism and higher risk of T2D (Lotta et al., 2016). Therefore, there is growing evidence that high levels of BCAAs and related intermediate metabolites are not only T2D biomarkers, but also pathophysiological factors. However, the vast majority of these studies were conducted in individuals after an overnight fast, under conditions in which both protein degradation in skeletal muscle and a concomitant release of amino acids to support gluconeogenesis (Fukagawa et al., 1985) could mask alterations in BCAA homeostasis. Moreover, they do not provide insight into the dynamic shift in metabolism that occurs in response to nutritional challenges. An analysis of metabolomic signatures in fasted glucose tolerant and T2D individuals before and after a glucose challenge may provide insight into dynamic changes in BCAAs and related metabolites and the consequences of insulin resistance.
Skeletal muscle is the largest contributor to systemic BCAA oxidation (Neinast et al., 2019) and therefore impairments in glucose and BCAA metabolism in myocytes has an impact on whole body metabolic homeostasis. While BCAA-related gene expression and oxidation rates are reduced in vastus lateralis muscle biopsies from insulin resistant subjects (Lerin et al., 2016), studies of BCAA metabolism in human skeletal muscle are scarce. BCAA catabolism occurs in the mitochondrial matrix, implicating alterations in mitochondrial proteins may influence BCAA metabolism. For example, in transgenic mice overexpressing the mitochondrial biogenesis inducer peroxisome proliferator-activated receptor γ (PPAR γ) coactivator 1α (PGC-1α), BCAA levels are reduced in gastrocnemius muscle (Hatazawa et al., 2014). Conversely, administration of the PPARγ agonist thiazolidinedione rosiglitazone improves glycemic control and increases circulating levels BCAAs in individuals with T2D (van Doorn et al., 2007). Nevertheless, mechanisms underlying the role of PGC-1α in the regulation of BCAA catabolism are unclear.
Using metabolomic and transcriptomic approaches, we provide evidence that BCAA catabolism in skeletal muscle is attenuated and further impaired by a metabolic challenge in individuals with T2D. We report that metabolic inflexibility of BCAA catabolism in skeletal muscle is associated with alterations in BCAA gene expression via a PGC-1α/ERRα-dependent mechanism.

Glucose loading further attenuates BCAA catabolism in T2D patients
Glucose ingestion in fasted subjects elicits an insulin-dependent metabolic response that is blunted in individuals with pre-diabetes (Shaham et al., 2008). To assess whether this impaired response also affects the BCAA profile, we measured levels of leucine, isoleucine, valine, and derived metabolites in plasma and vastus lateralis biopsies from NGT and T2D individuals before and after an (OGTT). On average, the three BCAAs were ~10% and ~13% higher in plasma and skeletal muscle from fasted T2D subjects (Figure 2A-C), and non-significant changes were found in the corresponding branched-chain α-keto acids (BCKAs) ( Figure 2D-F). While the levels of BCAA-derived acyl-carnitines were not different between groups ( Figure 2G-K), 3-hydroxisobutyrate was higher in plasma (+37%) and skeletal muscle (+45%) from T2D subjects ( Figure 2L). This valine-derived metabolite has been reported to promote insulin resistance in skeletal muscle cells by increasing fatty acid uptake (Jang et al., 2016).
Glucose ingestion decreased circulating and intramuscular levels of BCAA and BCKA ( Figure 2A-F), albeit to a lesser extent in individuals with T2D. The intramuscular content of BCKA was decreased 37-56% in NGT subjects, whereas levels remained unaltered in T2D.
While the transamination from BCAA to BCKA is reversible, all three BCKA are irreversibly decarboxylated by the branched-chain α-ketoacid dehydrogenase complex (BCKDH), the ratelimiting enzyme in the catabolism of BCAA. These results suggest that skeletal muscle BCKDH activity is impaired in individuals with T2D. Therefore, we determined total abundance and phosphorylated (inactive) BCKDH content in skeletal muscle ( Figure 2M-O).
We found that BCKDH abundance and phosphorylation was not altered between NGT and T2D subjects. Given recent evidence that BCAA catabolic flux cannot be predicted by BCKDH phosphorylation status (Neinast et al., 2019), we cannot exclude the possibility that BCKDH activity is impaired based on this Western blot analysis. Conversely, we observed a trend towards an accumulation of BCAA-derived acyl-carnitines in skeletal muscle ( Figure 2G-K), including isovalerylcarnitine and isobutyrylcarnitine, with levels increased in T2D as compared to NGT subjects after the OGTT. Moreover, glucose loading induced a significant decrease of 3-hydroxyisobutyrate levels in skeletal muscle of NGT but not T2D subjects.
Schematic representation of BCAA degradation pathway and the log2 fold changes of the evaluated metabolites after OGTT relative to fasting conditions in both groups are reported ( Figure 2P-Q). Collectively, these results suggest that a glucose challenge unmasks defects at several steps of BCAA catabolism.
Expression of genes involved in BCAA catabolism is decreased in T2D skeletal muscle.
We next evaluated whether the metabolic alterations in T2D in response to a glucose challenge are associated with changes in skeletal muscle expression of genes involved in BCAA metabolism. Expression of genes encoding for enzymes involved in first steps of BCAA metabolism, namely Branched Chain Amino Acid Transaminase 2 (BCAT2) and three subunits of the BCKDH complex ( Figure 3A) were decreased in skeletal muscle of T2D patients as compared to NGT controls. Furthermore, 69% of the analyzed genes participating in metabolic steps downstream of the oxidative decarboxylation reaction catalyzed by BCKDH showed a similar profile ( Figure 3B), indicating that BCAA gene expression is widely downregulated in T2D skeletal muscle. These differences were also evident after an OGTT, as mRNA levels remained relatively stable after glucose loading. Thus, we next determined potential candidates involved in the regulation of BCAA gene expression that could explain the differences between NGT and T2D.
PPARGC1A is downregulated in skeletal muscle from fed and fasted T2D patients.
Skeletal muscle PGC-1α expression is reduced in T2D patients (Patti et al., 2003), and proposed as a potential upstream regulator of BCAA metabolism (Hatazawa et al., 2014). As a transcriptional coactivator, PGC-1α interacts with several transcription factors and nuclear receptors to enhance gene expression. The orphan nuclear receptor ERRα, encoded by ESRRA, is a canonical functional partner of PGC-1α ( Figure 3C) that regulates metabolic processes in mitochondria including oxidative phosphorylation and metabolism of lipids and ketones (Luo et al., 2003;Svensson et al., 2016). Therefore, we determined the expression of PPARGC1A and ESRRA in skeletal muscle of NGT and T2D subjects. PPARGC1A mRNA was reduced in T2D subjects, whereas ESRRA mRNA showed non-significant differences in skeletal muscle of fasted subjects ( Figure 3D).
PGC-1α mediates BCAA gene expression in primary human skeletal muscle cells.
We next determined whether PGC-1α mediates BCAA gene expression in skeletal muscle.
Primary human skeletal muscle cells (HSMCs) transfected with PPARGC1A siRNA had reduced expression of PPARGC1A ( Figure 4A), which moderately decreased expression of genes of the family of acyl-CoA dehydrogenases (ACAD8, ACADSB, and IVD), and HIBADH (encoding the 3-hydroxyisobutyrate dehydrogenase enzyme) as compared to cells treated with a scramble siRNA ( Figure 4B). Conversely, expression of the three BCKDH subunits, BCKDHB, DBT and DLD remained unaltered. Adenovirus-mediated PPARGC1A overexpression (Ad-PGC1A) in HSMCs ( Figure 4C) increased the expression of 61% of the analyzed genes relative to adenovirus-GFP control cells, including two of the three BCKDH subunits ( Figure 4D). BCKDK, which encodes a kinase that phosphorylates and inactivates BCKDH, was upregulated after PPARGC1A silencing and unaltered by PPARGC1A overexpression, suggesting that not all BCAA genes are equally modulated by PGC-1α.
Paradoxically, BCKDH protein content was significantly lower in Ad-PGC1A overexpressing cells ( Figure 4E), leading to an increase in the pBCKDH/BCKDH ratio ( Figure   4F-G), despite the higher levels of the phosphatase PPM1K ( Figure 4I).
Mice with skeletal muscle-specific modified expression of PPARGC1A exhibit altered levels of BCAA gene transcripts and related metabolites.
Next, we used two different mouse models to assess the impact of muscle-specific altered  Figure 5B) and the BCKDH regulatory enzymes PPM1K and BCKDK ( Figure 5C). Accordingly, a contrasting profile of BCAA gene expression was observed in skeletal musclespecific PGC-1α transgenic mice (mTg) versus the PGC-1α mKO mice, with an upregulation of the BCAA gene profile in skeletal muscle as compared to wild-type (WT) littermates ( Figure  5E-H). These changes in gene expression were not associated with alterations in either body weight or glucose tolerance ( Figure S1D-F). Consistent with our results from human myotubes ( Figure 4), the gene encoding for the enzyme responsible for the transamination of BCAA to BCKA was unchanged in all groups, indicating that BCAT2 transcriptional regulation is PGC-1α-independent.
To ascertain whether PGC-1α-associated alterations in BCAA gene expression have functional implications in BCAA metabolism, we performed LC/MS metabolomics to profile BCAA-related metabolites in plasma and quadriceps muscle from mKO and mTG mice and respective WT littermates. Circulating levels of leucine, isoleucine, and valine in mKO and mTG mice were not statistically different when compared to the respective WT littermates ( Figure 5I-K), whereas the intramuscular content of valine was reduced in mTG mice ( Figure   5L-N). Similar non-significant changes were observed for isoleucine (p=0.09) and isovalerylcarnitine (p=0.12) ( Figure 5M and O). Surprisingly, mKO mice exhibited decreased levels of 3-hydroxyisobutyrate ( Figure 5Q), although the rest of analyzed metabolites were unaltered compared to WT littermates.
Finally, we determined whether overexpression of PPARGC1A would increase BCAA oxidative rates in C2C12 myotubes. Consistent with our results in human myotubes, C2C12 adenovirus-mediated overexpression of PPARGC1A upregulated the expression of all analyzed BCAA genes, with the notable exception of ACADSB and BCKDK ( Figure S1G), which also followed a distinct expression pattern in mTG mice. In line with the trend towards lower intramuscular accumulation of BCAA metabolites in the mTG mice, PGC-1α-mediated upregulation of BCAA genes in C2C12 myotubes increased leucine oxidation ( Figure S1H).
As a transcriptional coactivator, PGC-1α does not directly bind DNA in a sequence-specific manner, but interacts with transcription factors and nuclear receptors to enhance gene transcription (Puigserver and Spiegelman, 2003). After confirming the involvement of PGC-1α in the transcriptional regulation of the BCAA metabolism gene set, we next investigated whether the orphan nuclear receptor ERRα was necessary for PGC-1α-enhanced BCAA gene expression. siRNA-mediated silencing of ESRRA in primary HSMCs ( Figure 6A) decreased the expression of 65% of the analyzed genes ( Figure 6B). Notably, BCAT2 and BCKDK were among the unaltered genes. Next, we determined whether this BCAA gene downregulation affects leucine oxidation in skeletal muscle. Differentiated primary HSMCs were treated with either DMSO or 3,6-dichlorobenzo(b)thiophene-2carboxylic acid (BT2). BT2 is a smallmolecule inhibitor of BCKDK, and consequently activates the BCKDH complex and increases BCAA oxidative flux. Silencing of ESRRA resulted in a significant decrease in leucine oxidation under BCKDH-activated conditions, as measured by CO2 production after incubation with [U-14 C]-leucine ( Figure 6C). A schematic biochemical representation of the leucine oxidation assay is shown ( Figure 6D). We next tested the hypothesis that ERRα is a PGC-1α interacting partner in the transcriptional regulation of the BCAA gene set using human differentiated myotubes treated with either scramble siRNA or ESRRA siRNA and transfected with adenovirus containing the green fluorescent protein gene (Ad-GFP) or PPARGC1A (Ad-PGC-1A). We also tested this hypothesis using PPARGC1A overexpressing cells treated with the inverse ERRα agonist XCT-790 ( Figure S2). Since PPARGC1A and ESRRA mutually regulate their expression, we first confirmed that ESRRA silencing did not abrogate PPARGC1A in cultured cells. Ad-PGC1A cells had high levels of PPARGC1A transcripts regardless the treatment with ESRRA siRNA ( Figure 6H and Figure Figure   6I).

DISCUSSION
A link between increased plasma BCAAs and insulin resistance was established as early as 1969 (Felig et al., 1969), but this association has remained relatively unexplored until the last decade. We performed a metabolomic and transcriptomic analysis of human plasma and skeletal muscle biopsies, as well as experimental studies in transgenic mouse models and primary HSMCs to elucidate the mechanisms underpinning BCAA metabolism in T2D. While we corroborate the association between BCAA metabolites and T2D (Menni et al., 2013;Suhre et al., 2010;Xu et al., 2013;Yu et al., 2016) and the role of PGC-1α in BCAA catabolism (Hatazawa et al., 2016;Hatazawa et al., 2014;Jang et al., 2016), we provide new evidence that an OGTT unmasks impairments in BCAA catabolism in T2D patients. Moreover, we reveal that PGC-1α-mediated regulation of genes important for BCAA catabolism is dependent on ERRα, a canonical PGC-1α-interacting protein.
Circulating plasma BCAA levels are altered in T2D patients, however metabolomic profiles of peripheral tissues involved in glucose homeostasis are scarce and the transcriptomic regulation of genes involved in BCAA catabolism is unknown. Studies focusing on skeletal muscle BCAA content are limited to an analysis of insulin resistant adults (Lerin et al., 2016), rather than T2D patients. We performed untargeted metabolomic analysis on both plasma and vastus lateralis biopsies obtained before and after an OGTT from men with either NGT or T2D. Circulating BCAAs and BCKAs were decreased after an OGTT and this reduction was attenuated in T2D patients, presumably due to insulin resistance, corroborating an earlier study of plasma samples from the Framingham Heart Study Offspring cohort (Ho et al., 2013). In skeletal muscle, insulin inhibits proteolysis (Tessari et al., 1986) and increases the preference for BCAA oxidation (Neinast et al., 2019). Conversely, in the hypothalamus, insulin signaling promotes BCAA catabolism in liver in a glucose-independent manner (Shin et al., 2014), which may explain the reduced BCAA and BCKA metabolism in individuals with T2D as compared to NGT. Generally, plasma metabolites mirrored skeletal muscle BCAA profile, although greater variation was observed in muscle, possibly due to fiber type heterogeneity of the vastus lateralis (Staron et al., 2000). Moreover, the intramuscular accumulation of isovalerylcarnitine, isobutyrylcarnitine, and 3-hydroxyisobutyrate was not reflected in plasma, suggesting defects in processes controlling the export of metabolites from myocytes in T2D. In insulin resistant rodents, BCAA oxidation is shunted from liver and adipose tissue toward skeletal muscle (Neinast et al., 2019). This finding may be clinically relevant, given that the expression of genes regulating BCAA metabolism in skeletal muscle were reduced in T2D patients. In this regard, the accumulation of 3-hydroxyisobutyrate after an OGTT is notable, since this valinederived metabolite promotes insulin resistance through increased fatty acid uptake in skeletal muscle (Jang et al., 2016). This could lead to a vicious cycle in which secretion of 3hydroxyisobutyrate from skeletal muscle may decrease insulin sensitivity and further impair insulin signaling. Although elevated levels of circulating C3 and C5 acylcarnitines have been also detected in obese subjects , whether accumulation of short-chain acylcarnitines mediates in insulin resistance remains to be elucidated (Schooneman et al., 2013). Nevertheless, these results are suggestive of defects in downstream degradation of BCAA-derived metabolites.
Accumulation of BCAA metabolites in the T2D patients was accompanied by an overall decrease in expression of genes involved in BCAA catabolism, suggesting that alterations in the transcriptional regulation of these genes could lead to attenuated BCAA catabolism in skeletal muscle. Thus, considering that BCAA catabolism occurs within the mitochondrial matrix, we assessed proteins involved in mitochondrial homeostasis. In addition to PGC-1α, which is downregulated in T2D skeletal muscle, we identified the orphan nuclear receptor ERRα as a potential transcription regulator of BCAA-related genes. ERRα is encoded by ESRRA and functions as a PGC-1α-interacting partner that coordinates transcriptional programs controlling energy metabolism (Huss et al., 2004;Mootha et al., 2004). In response to the OGTT, ESRRA mRNA was reduced in skeletal muscle from individuals with NGT, but not T2D. Given that ERRα regulates insulin-induced transcriptional responses associated with mitochondrial metabolism (Batista et al., 2019), we next explored whether ERRα and PGC-1α direct the BCAA transcription program.
PGC-1α coordinates metabolic and transcriptomic programs linked to cellular energy homeostasis, such as hepatic gluconeogenesis, lipid metabolism and mitochondrial respiration (Estall et al., 2009;Wu et al., 1999;Yoon et al., 2001). Reduced PGC-1α mRNA and protein are linked to insulin resistance in T2D (Moreno-Santos et al., 2016;Patti et al., 2003). PGC-1α also mediates the expression of genes involved in BCAA catabolism (Hatazawa et al., 2016;Hatazawa et al., 2014;Jang et al., 2016). We found that expression of PGC-1α and several genes involved in BCAA catabolism was reduced in skeletal muscle of T2D patients. Primary HSMCs and mouse skeletal muscle with reduced PPARGC1A expression exhibited a mild reduction in several BCAA genes, whereas its overexpression was associated with a consistent upregulation. Thus, while PGC-1α was not essential for basal BCAA gene transcription, it may play a role in the adaptive response to increased BCAA catabolic demands, such as during exercise (Hagg et al., 1982) or nutritional changes. However, some BCAA catabolic genes were unaltered, suggesting that PGC-1α-independent regulatory transcriptional and posttranslational mechanisms are also involved. Indeed, overexpression of PGC-1α was associated with reduced BCKDH content, which led to higher phosphorylated BCKDH/total BCKDH ratio despite the increase in PPM1K, a phosphatase responsible for the dephosphorylation of BCKDH. Overall, our results highlight the complexity of the enzymatic networks controlling BCAA metabolism and raise a potential caveat of linking transcriptomic and proteomic data with functional outcomes.
We next hypothesized that alterations in the BCAA gene network would impact BCAA metabolism. We found that mice overexpressing PGC-1α in skeletal muscle exhibited an upregulation of BCAA genes and a trend towards reduced intramuscular accumulation of BCAA metabolites, suggestive of increased BCAA catabolic flux. Accordingly, we found that adenovirus-mediated overexpression of PGC-1α in C2C12 myotubes increased leucine oxidation rates. However, intramuscular levels of BCAA metabolites were unaltered in skeletal muscle of PGC-1α mKO mice. Thus, other organs such as adipose tissue may compensate for a putative impairment in BCAA catabolism (Herman et al., 2010). Furthermore, a metabolic challenge may be necessary to reveal functional alterations in BCAA metabolism in PGC-1α mKO mice.
ERRα physically interacts with PGC-1α to elicit the transcription of genes involved in mitochondrial biogenesis and oxidative phosphorylation (Schreiber et al., 2004), lipid metabolism (Luo et al., 2003) and ketone body homeostasis (Svensson et al., 2016). We found that both ESRRA silencing and inverse agonist inhibition of ERRα downregulated BCAA genes and abrogated the PPARGC1A-induced responses, indicating ERRα is necessary for the transcription of BCAA genes mediated by PGC-1α. Whether this is due to a direct regulation by binding of the ERRα/PGC-1α complex to potential ERRα response elements (ERREs) in the promoter regions of the analyzed genes warrants further evaluation. A combination of chromatin immunoprecipitation (ChiP) and genomic DNA arrays in mouse cardiac tissue identified 195 promoters bound by ERRα (Dufour et al., 2007), and among these, only ACADM, which also participates in lipid metabolism, is involved in BCAA catabolism.
However, ERRα/PGC-1α binds near the transcriptional start site of a previously unidentified gene encoding for the rate-limiting ketolytic enzyme OXCT1 in skeletal muscle (Svensson et al., 2016), suggesting that tissue-specific responses and different bioinformatic approaches could unravel genes directly regulated by ERRα/PGC-1α.
In summary, altered expression of PPARGC1As associated with disturbances in BCAA metabolism in human skeletal muscle of T2D individuals. Experimental approaches to reduce PPARGC1A levels partially recapitulates the BCAA gene set profile identified in skeletal muscle from T2D patients ( Figure 6H). Our results have clinical relevance and suggest that people with T2D have metabolic inflexibility in regard to BCAA homeostasis and an associated impaired transcriptomic response. Additionally, our data demonstrate that ERRα is essential for PGC-1α-mediated transcriptional regulation of genes involved in BCAA metabolism, either directly or indirectly, thereby unraveling a novel role for this orphan nuclear receptor.

LIMITATIONS OF STUDY
The T2D patients in our study were slightly older and had significantly higher BMI and waist circumference than the controls, which are confounding variables and may affect other pathophysiological processes. Nevertheless, these clinical features are characteristic of the general T2D population (Public Health England, 2012). Moreover, our analysis was limited to male participants, and therefore we cannot exclude sex-specific differences in the analyzed outcomes. Due to limitations in the LC/MS methodology, quantitation of BCAA metabolites with CoA moieties was not possible. To overcome this, we used specific derived carnitines from these metabolites as a proxy, limiting the understanding of how the different conditions affect BCAA catabolism after BCKA decarboxylation. Moreover, alterations in PGC-1α expression may influence fiber type distribution in mouse models (Handschin et al., 2007;Lin et al., 2002), which could impact the results. Nevertheless, similar results were obtained in transgenic mice, in vitro primary HSMCs and C2C12 myotubes strongly suggesting that alterations in BCAA metabolism was not due to changes in fiber type.

ACKNOWLEDGMENTS
The costs of human metabolomics study was supported by grant from Daiichi Sankyo Co., Ltd.       silencing. Specific significant differences between groups are not shown in cases where statistically significant interaction was found. siRNA, small interfering RNA; Ad-GFP, adenoviral overexpression of green fluorescent protein; BT2, 3,6-dichlorobrenzo(b)thiophene-2carboxylic acid; scr, scrambled siRNA; TCA, tricarboxylic acid cycle. See also Figure S2.  Results are shown as mean ± SEM. Normal Glucose Tolerant (NGT; n=25) and Type 2 Diabetic (T2D; n=25) men. Statistical analysis was performed using Student's t-test.

Lead Contact and Resource Sharing
Further information and requests for resources or reagents should be directed to and will be made available upon reasonable request by the Lead Contact, Juleen R. Zierath, (Juleen.Zierath@ki.se).

Materials Availability
This study did not generate new unique reagents.
Data and Code Availability ****

Subject characteristics
Thirty-two men with NGT and twenty-nine men with T2D aged 44-69 years were recruited to participate in this study. The study was approved by the regional ethical review board in Stockholm. All participants gave oral and written consent to participate in the study. The experimental procedures were conducted according to the Declaration of Helsinki. Participants underwent a clinical health screening consisting of clinical chemistry and anthropometric measurements (Table 1 and 2). After clinical health screening, five individuals initially characterized as NGT and one individual previously diagnosed as T2D were diagnosed as impaired glucose tolerant and excluded from the study (Figure 1).
Subjects with T2D had higher blood glucose and Hb1Ac and HOMA-IR, as well as higher BMI and waist circumference (Table 1 and 2). Moreover, the men with T2D included in the transcriptomic analysis were older than the men with NGT (Table 2).

Primary human skeletal muscle cells culture
Human satellite cells were harvested from vastus lateralis skeletal muscle biopsies of healthy male and female volunteers as described (Al-Khalili et al., 2003). Cells were propagated in growth media (F12/DMEM, 20% FBS, and 1% anti-anti) until ~90% confluence was reached.
Cells were incubated at 37°C in 5% CO2 humidified chambers, and medium was changed every second day during growth and differentiation. C2C12 cells were differentiated for six days before final experiments.

Mouse models
Mice were housed under controlled lighting (12 h light/12 h dark cycle) with free access to food and water. Experiments were performed in accordance with Swiss federal guidelines and were approved by the Kantonales Veterinäramt Basel-Stadt. Skeletal muscle-specific PGC-1α transgenic (mTG) mice are described elsewhere (Lin et al., 2002). The PGC-1α muscle-specific knockout (mKO) mice were generated by breeding PGC-1α loxP/loxP mice (Lin et al., 2004) with transgenic mice expressing the Cre recombinase under the control of the human α-skeletal actin promoter (HSA-Cre) . Male mice aged between 11-14 weeks were used in all experiments.

Method Details
Blood and muscle biopsy sampling and oral glucose tolerance test Participants arrived at the clinic at 07:45 after a 12-h fasting period. A catheter was inserted into an arm vein and fasting blood samples were obtained. After applying local anesthesia (Lidokain hydrochloride 5 mg/mL), a 1 cm incision was made in the skin and fascia of the vastus lateralis portion of the quadriceps muscle using a conchotome (AgnTho's AB, Sweden).
Muscle biopsies were immediately rinsed in ice-cold PBS solution, blotted on a filter paper to dry off excessive liquid and immediately snap-frozen in liquid nitrogen. After ∼15 min, participants were given a standardized solution containing 75 g of glucose. A second blood sample was obtained 30 min after glucose ingestion. Finally, 120 minutes after glucose ingestion another blood sample and a second skeletal muscle biopsy were obtained. All biopsies were kept in liquid nitrogen storage until analysis.
Human plasma and skeletal muscle metabolomics The processing of GC-TOF/MS data and extraction of putative metabolites was conducted as described (Chorell et al., 2016). Briefly, an in-house MATLAB script was used for the