Hypoxia in Combination With Muscle Contraction Improves Insulin Action and Glucose Metabolism in Human Skeletal Muscle via the HIF-1α Pathway

Skeletal muscle is one of the largest organs in the body and plays a major role during whole-body glucose homeostasis. Insulin resistance (IR) of skeletal muscle is one early hallmark in the development of type 2 diabetes (T2D) (1). Therefore, the identification of novel targets for improving skeletal muscle IR represents a fundamental challenge and plays a key role in developing novel therapeutic strategies for T2D. In contrast to the very limited nonexercise-based strategies to combat IR, physical activity is well known to exert multiple beneficial effects on the progression of IR and T2D (2,3). However, our knowledge on the cellular mechanisms mediating these adjuvant actions of physical activity remains incomplete. Exercise is accompanied by a reduction in intramuscular oxygen tension, resulting in enhanced HIF-1α protein expression (4). In response to cellular hypoxia, HIF-1α is activated to regulate the transcription of >100 target genes, e.g., glycolytic enzymes and glucose transporters (5). Furthermore, it could be shown that loss of HIF-1α in murine skeletal muscle cells impairs GLUT4 translocation and glucose uptake, indicating a possible role for HIF-1α in the regulation of skeletal muscle glucose metabolism (6). One particular aspect of skeletal muscle is that HIF-1α protein is highly expressed in this tissue even in normoxic conditions, suggesting that HIF-1α function on muscle homeostasis is not exclusively dependent on the oxygen state of the muscle cell (7). Importantly, endurance exercise under hypoxia is known to be even more efficient to improve glycemic control in individuals with T2D (8), potentially reflecting the synergistic or additive regulation of HIF-1α downstream target genes. Therefore, we hypothesized that the molecular analysis of hypoxic muscle contraction may pave the way to identify novel targets for improving skeletal muscle IR.

Primary human skeletal muscle cells isolated from eight healthy Caucasian donors (four males, 16, 21, 41, and 47 years of age; four females, 16, 25, 33, and 37 years of age) were supplied as proliferating myoblasts and cultured as described previously (9).

Differentiated myotubes were subjected to electrical pulse stimulation (EPS) as recently described (9).

Myotubes were cultured for 24 h with or without EPS in an atmosphere containing 21, 7, or 2% O₂ supplemented with 5% CO₂ and respective concentrations of nitrogen in an Xvivo hypoxia chamber system (Biospherix, Parish, NY).

Glucose uptake and oxidation assays were performed as described previously (9).

We used predesigned primers (Qiagen) in a SYBR Green–based quantitative real-time PCR. Thioredoxin-interacting protein (TXNIP), HIF-1α, anti–phospho Akt^Ser473, anti–phospho AMPKα^Thr172, anti–phospho Akt^Thr308, anti–phospho AS160^Thr642, anti–phospho GSK3α/β^Ser21/9, total Akt, AMPKα, GSK3α/β, and total AS160 antibodies were supplied by Cell Signaling Technology (Frankfurt, Germany). Polyclonal Rab20 antibody was supplied by Abcam (Cambridge, U.K.). Polyclonal antiserum against GLUT4 was described previously (10).

Silencing experiments were performed by using FlexiTube small interfering RNA (siRNA) and HiPerfect (Qiagen) according to the manufacturer’s instructions. Control cells were treated with AllStars Negative Control siRNA (Qiagen).

Immunofluorescence staining was performed as described previously (9).

L-(+)-lactate was detected in the supernatant fraction with a kit (Lactate Assay Kit II; BioVision).

GLUT-myc translocation assay was performed as described previously (11).

The protocol is described in detail by Zhao et al. (12).

In vivo muscle electroporation and ex vivo glucose uptake were adapted from previous reports (13).

This human exercise intervention study is in detail described by Langleite et al. (14).

C57BL/6J mice were trained 5 days per week for 6 weeks during their active dark phase. Each training session consisted of alternating 5-min intervals with moderate to high (15–18 m/min) or low running intensity (10 m/min), respectively. Mice were sacrificed at least 24 h after their last exercise bout. Gastrocnemius muscle was used for the analysis of TXNIP and Rab20 protein or mRNA expression.

Human skeletal muscle cells were treated with 300 µmol/L cobalt chloride (CoCl₂) for 6 h, and RNA sequencing and data analysis were performed by the GATC Biotech GmbH (Konstanz, Germany).

We cultured human myotubes at different oxygen levels (21, 14, 7, and 2%) with or without EPS for 24 h and analyzed basal and insulin-stimulated glucose uptake (Fig. 1D and Supplementary Fig. 1A and B). Only EPS in combination with 7% O₂ significantly increased insulin-stimulated glucose uptake. In accordance, we observed increased insulin signaling (Fig. 1A–C). However, EPS in combination with 7% O₂ enhanced the EPS-induced lactate release, whereas glucose oxidation was not affected (Supplementary Fig. 2A and B). In addition, we observed a significant increase of HIF-1α protein levels at 7% O₂ and in combination with EPS (Fig. 1A). The well-known HIF-1α target GLUT1 was not regulated at these conditions (Supplementary Fig. 3B). However, GLUT4 mRNA expression was significantly upregulated in the combined setting of 7% O₂ and EPS (Supplementary Fig. 3C).

Silencing of HIF-1α (siHIF1A) in human myotubes led to a downregulation of insulin-mediated Akt^Ser473 phosphorylation and substantially reduced the phosphorylation of AS160^Thr642, whereas insulin receptor β^Tyr1150/1151 phosphorylation was not affected (Fig. 1E). In line, in vivo electroporation of siHif1a into murine soleus muscle abrogated insulin-mediated AS160^Thr642 phosphorylation (Supplementary Fig. 4B–D). Insulin- and EPS-stimulated glucose uptake were strongly inhibited by HIF-1α silencing in human myotubes (Fig. 1F). In addition, insulin-stimulated ex vivo glucose uptake was inhibited after in vivo electroporation of siHif1a into murine soleus muscle (Supplementary Fig. 4A).

CoCl₂ substantially augments HIF-1α protein levels (Fig. 1G). In addition, we observed a strong increase in insulin-induced Akt^Ser473 as well as basal and insulin-mediated AS160^Thr642 phosphorylation (Fig. 1G). Both basal and insulin-regulated glucose uptake were increased (Fig. 1H). Pathway analysis from RNA sequencing data indicated that CoCl₂ treatment specifically induced HIF-1α signaling (Supplementary Fig. 5A–C). The top 15 upregulated and 5 downregulated genes are displayed as a heat map (Fig. 1I). RAB20 was one of the most upregulated genes, and real-time PCR and Western blot data confirmed that Rab20 is regulated by CoCl₂ and hypoxia (Fig. 1J–M). Compared with all other detectable Rab genes in human myotubes, Rab20 showed by far the strongest increase after CoCl₂ treatment (Supplementary Fig. 6). In contrast, TXNIP was the most downregulated gene after CoCl₂ treatment (Fig. 1I). Validation by real-time PCR and Western blot confirmed that TXNIP is regulated by hypoxia in human myotubes (Fig. 1N–P).

Rab20 expression was upregulated during human skeletal muscle cell differentiation (Fig. 2A and B). Knockdown of Rab20 inhibited the insulin- and contraction-stimulated glucose uptake (Fig. 2C). In line, in vivo electroporation of siRab20 in murine soleus muscle inhibited insulin-mediated glucose uptake (Fig. 2D). Abrogation of insulin-stimulated glucose uptake was associated with impaired mobilization of GLUT4 to the plasma membrane in human myotubes (Fig. 2K and L). In addition, siRab20 treatment reduced insulin-mediated GLUT4-myc translocation to the cell surface in L6 myoblast overexpressing AS160 and GLUT4-myc (Fig. 2H and I). Importantly, Rab20 was not colocalized with GLUT4 after insulin stimulation (Fig. 2J). Insulin signaling at the level of Akt and AS160 phosphorylation was not altered by Rab20 silencing (Fig. 2E–G).

RAB20 mRNA expression was upregulated after acute and long-term exercise (Fig. 3A and D). Only RAB20 was enhanced after acute exercise compared with all other detectable Rab genes in human skeletal muscle biopsies (Fig. 3G). Also HIF1A mRNA expression was significantly upregulated by acute and chronic exercise (Supplementary Fig. 7A–D), and HIF1A mRNA expression was positively correlated with RAB20 mRNA expression (Fig. 3B). The increase of RAB20 mRNA expression was positively correlated with the increase of HIF1A mRNA expression after acute exercise (Fig. 3C). Furthermore, skeletal muscle RAB20 mRNA expression in the control group was positively correlated with glucose infusion rate (GIR) (Fig. 3E and F). In addition, exercise training increased Rab20 in murine gastrocnemius muscle (Fig. 3H–J).

Acute exercise induced TXNIP in both groups, whereby the control group showed a significantly higher induction (Fig. 4A and B). However, long-term exercise tended to reduce the expression of TXNIP mRNA in skeletal muscle (control; P = 0.055) (Fig. 4C and D). Moreover, TXNIP mRNA expression was negatively correlated with the GIR (Fig. 4E). In addition, endurance training reduced Txnip mRNA expression as well as protein levels in murine muscle (Fig. 4I and J). Importantly, the loss of TXNIP in human myotubes by siRNA enhanced insulin-stimulated Akt as well as As160 phosphorylation (Fig. 4F–H).

The observation that endurance exercise under hypoxic conditions is more efficient to improve glycemic control (8) potentially reflects additive regulation of HIF-1α downstream target genes. We show that HIF-1α protein levels are significantly increased in the combined setting of muscle contraction and hypoxia, indicating that reduced oxygen tension and muscle contraction exert an additive effect on HIF-1α stabilization. Furthermore, we show that HIF-1α improves glucose metabolism and insulin action in human skeletal muscle, which is in line with previous findings showing that HIF-1α regulates insulin-stimulated glucose uptake in murine skeletal muscle cells (6). Collectively, these data identify HIF-1α as an important regulator of insulin-mediated glucose metabolism.

To analyze the underlying mechanism, we used a genome-wide gene expression profiling approach of cells treated with CoCl₂. We show that Rab20 is expressed in skeletal muscle, and the expression is induced by HIF-1α. In accordance, Hackenbeck et al. (15) reported that RAB20 is a direct target of HIF-1α. The fact that RAB20 expression is substantially upregulated by exercise in skeletal muscle underpins the biological significance. Furthermore, skeletal muscle RAB20 mRNA expression was positively correlated with whole-body insulin sensitivity, indicating a potential role of Rab20 for skeletal muscle glucose metabolism. In line, loss of Rab20 inhibited insulin- and contraction-stimulated glucose uptake, which was associated with a reduction of GLUT4 translocation. A change in the proportion of GLUT4 at the cell surface might be caused by alterations in the endocytosis/exocytosis rate. Rab8A and Rab10 are involved in vesicle tethering via interaction with the exocyst, and Rab5 is involved in sorting of GLUT4 from the recycling endosome to the insulin-sensitive compartment (16–18). Rab20 is colocalized with Rab5 on endosomal membranes (19). Furthermore, Rab20 was shown to interact with Rabex-5, a guanine nucleotide exchange factor for Rab5 (20). Thus, we can speculate that high levels of Rab20 during exercise activate Rab5, which in turn increases recycling of endosomes to the insulin-sensitive compartment.

The mechanism underlying HIF-1α–mediated regulation of Akt or AS160 phosphorylation is still unclear. The well-known recognized function of HIF-1α consists of transcriptional regulation, and therefore it is most unlikely that HIF-1α directly affects the phosphorylation process of Akt or downstream targets. Our data indicate that TXNIP is strongly downregulated by HIF-1α. Furthermore, downregulation of TXNIP in human myotubes results in enhanced insulin-stimulated Akt and downstream signaling. In line, whole-body knockout of TXNIP in mice resulted in enhanced Akt signaling and insulin sensitivity (21). Furthermore, we show that exercise training is able to reduce TXNIP expression in muscle and that muscle TXNIP mRNA expression correlates negatively with the GIR. Johnson et al. (22) demonstrated that caloric restriction improves insulin sensitivity, which was also associated with downregulation of TXNIP in the muscle. Therefore, downregulation of TXNIP in muscle could improve insulin signaling as well as glucose metabolism in humans.

In conclusion, this study provides further insights into the molecular regulation of glucose metabolism in skeletal muscle during exercise, implicating a key role for the transcription factor HIF-1α in this process. Therefore, we suggest that the skeletal muscle HIF-1α signaling pathway may represent a novel therapeutic target to improve insulin sensitivity in humans.

Acknowledgments. The authors thank A. Heck and B. Nellemann (Department of Physical Performance, Norwegian School of Sport Sciences) for taking the biopsies and A.R. Enget, A. Kielland, K.J. Kolnes, D.S. Tangen, T.I. Gloppen, T. Dalen, H. Moen, M.A. Dahl, G. Grøthe, E. Johansen, K.A. Krog, and O. Skattebo (Department of Physical Performance, Norwegian School of Sport Sciences and Department of Nutrition, University of Oslo) for being responsible for and helping out on different aspects of the human strength and endurance intervention. The authors thank N. Tennagels and C.W. Jung (Sanofi Deutschland GmbH) for providing the GLUT4-myc L6 myoblasts. The technical assistance of M. Koenen and the secretarial assistance of B. Hurow (Paul-Langerhans-Group for Integrative Physiology, German Diabetes Center [DDZ]) are gratefully acknowledged.

Funding. This work was supported by the Ministry of Science and Research of the State of North Rhine-Westphalia, the Federal Ministry of Health, and the German Academic Exchange Service (project no. 57068553). K.E. is supported by the Deutsche Forschungsgemeinschaft (EC 440/1-1 and EC 440/2-1).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.W.G. and J.E. contributed to the concept; designed and performed the experiments; acquired, analyzed, and interpreted data; and wrote the manuscript. T.B., K.E., C.S., A.Ch., A.M., J.W., A.Cr., and K.I.B. performed the research and contributed to analysis and interpretation of data. J.J., C.A.D., and H.A.-H. performed the research, contributed to analysis and interpretation of data, and reviewed and edited the manuscript. All authors approved the final version of the manuscript. S.W.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.