Acute Systemic Insulin Intolerance Does Not Alter the Response of the Akt/GSK-3 Pathway

Title page

Acute systemic insulin intolerance does not alter the response of the Akt/GSK-3 pathway to environmental hypoxia in human skeletal muscle

Gommaar D'Hulst1, Lykke Sylow2, Peter Hespel1, Louise Deldicque1

1Department of Kinesiology, Exercise Physiology Research Group, FaBeR, KU Leuven,

Tervuursevest 101, 3001 Leuven, Belgium.

2Molecular Physiology Group, Department of Nutrition, Exercise and Sports, August Krogh

Centre, University of Copenhagen, Universitetsparken 13, 2100 Copenhagen, Denmark.

Corresponding author:

Louise Deldicque

Exercise Physiology Research Group

FaBeR-KU Leuven

Tervuursevest 101

3001 Leuven, Belgium

E-mail: [email protected]

Fax: +32 16 329091; Tel: +32 16 329088

1 Abstract

Purpose: To investigate how acute environmental hypoxia regulates blood glucose and downstream intramuscular insulin signaling after a meal in healthy humans.

Methods: Fifteen subjects were exposed for 4h to normoxia (NOR) or to normobaric hypoxia (HYP, FiO2=0.11) in a randomized order 40 min after consumption of a high glycemic meal. A muscle biopsy from m. vastus lateralis and a blood sample were taken before (T0), after

1h (T60) and 4h (T240) in NOR or HYP and blood glucose levels were measured before exposure and every 30min.

Results: In HYP, blood glucose was reduced 100min (110.1±5.4 in NOR vs. 89.5±4.7 mg·dl-1 in HYP) and 130min (98.7±3.8 in NOR vs. 85.6±4.9 mg·dl-1 in HYP) after completion of a meal, which resulted in a 83% lower AUC in HYP compared to NOR (p=0.006). This coincided with 40% lower GLUT4 protein in the cytosolic fraction (p=0.013) and a tendency to increase in the crude membrane fraction (p=0.070) in HYP compared to NOR. At T240, blood glucose concentration was similar between HYP and NOR whereas plasma insulin as well as phosphorylation of muscle Akt and GSK-3 were ~2-fold higher in HYP compared to NOR

(p<0.05). In contrast, Rac1 protein was less abundant in the membrane fraction in HYP compared to NOR (p=0.003), reflecting lower activation.

Conclusion: Acute environmental hypoxia initially reduced blood glucose response to a meal, possibly via an increase in GLUT4 abundance at the sarcolemmal membrane. Later on, whole body insulin intolerance developed independently of defects in conventional insulin signaling in skeletal muscle.

Key words: Rac1, GLUT4, TBC1D4, glucose, muscle biopsy, insulin, hypoxia

2 Abbreviation list:

Akt/PKB Protein Kinase B AMPK 5' adenosine monophosphate-activated protein Kinase ANOVA Analysis Of Variance AS160/TBC1D Akt substrate of 160 kDa 4 BSA Bovine Serum Albumin CaM Calmodulin CaMK Calcium/Calmodulin-dependent protein kinase DTT Dithiothreitol EDTA Ethylene-Diamine-Tetra-Acetic acid eEF2 Eukaryotic Elongation Factor 2 EGTA Ethylene glycol bis(2-aminoethyl ether)tetraacetic acid ELISA Enzyme-linked immunosorbent assay G/I Glucose-to-insulin ratio GAPs Rab-GTPase-activating proteins GLUT4 Glucose Transporter 4 GS Glycogen Synthase GSK Glycogen Synthase Kinase HCl Hydrochloric Acid HIF-1α Hypoxia-Inducible Factor-1alpha HOMA Homeostatic Model Assessment HYP Hypoxic trial LIMK LIM Kinase NAC N-Acetylcysteine NOR Normoxic trial PAK p21 protein-activated kinase PI3K phosphatidylinositol 3-kinase PLN Phospholamban PVDF Polyvinylidene Fluoride Rac1 GTPase Ras-related C3 botulinum toxin 1substrate ROS Reactive Oxygen Species SDS-PAGE Sodium Dodecyl Sulfate - PolyAcrylamide Gel Electrophoresis Ser Serine Thr Threonine TRIS Tris(hydroxymethyl)aminomethane

3 Introduction

Peripheral insulin resistance contributes to the development of type 2 diabetes. Fortunately, factors such as muscle contraction or exercise augment postprandial glucose uptake in skeletal muscles and thereby improve the outcome of the disease (Sigal et al., 2006). Interestingly, lower blood glucose levels and prevalence of type 2 diabetes have been observed in high altitude populations (Castillo et al., 2007) or after long term hypoxic exposure (Marquez et al., 2013), suggesting that environmental hypoxia could modulate glucose uptake and insulin sensitivity as well depending on the severity and duration of the exposure. During acute exposure to hypoxia, whole body insulin sensitivity transiently declines after 30min to 48h partially due to a concomitant increase in catecholamines and cortisol levels (Peltonen et al., 2012; Larsen et al.,

1997; Oltmanns et al., 2004). Thereafter insulin sensitivity is restored and glucose homeostasis is improved, likely because of a transient normalization to pre hypoxic values of these circulating sympathic hormones (Lecoultre et al., 2013; Larsen et al., 1997).

How low oxygen levels affect insulin signaling in various tissues is less clear. Recent reports have shown a direct decrease in downstream insulin signaling by hypoxia in human and murine adipocytes (Regazzetti et al., 2009). In contrast, a stabilization of hypoxia-inducible transcription factors (HIF)-1α or HIF-2α resulted in improved insulin-induced Akt activation via elevated expression of insulin receptor substrate (IRS) in mice liver (Taniguchi et al., 2013) and human

HepG2 cells (Dongiovanni et al., 2008). In C2C12 myogenic cells, deletion of HIF-1α abrogated insulin-stimulated glucose uptake due to reduced Akt substrate of 160 kDa (AS160/TBC1D4) phosphorylation and impaired glucose transporter 4 (GLUT4) translocation to the plasma membrane, whereas constitutively active HIF-1α increased insulin-induced GLUT4 translocation

(Sakagami et al., 2014). Furthermore, chronic hypoxia (4 weeks, 10% O2) increased insulin-

4 induced Akt phosphorylation in incubated mice m.soleus (Gamboa et al., 2011). All together those results indicate that the effects of lowered oxygen availability on Akt signaling are tissue specific and depend on the duration and severity of hypoxia.

Interestingly, hypoxia is able to increase glucose uptake in skeletal muscle both in vitro (Cartee et al., 1991) and in vivo (Roberts et al., 1996) independently of insulin by using very similar, or at least overlapping, signaling pathways as those of contraction-induced glucose uptake. Hypoxia increases calcium efflux from the sarcoplasmic reticulum (Cartee et al., 1991; Brozinick, Jr. et al., 1999), thereby triggering calmodulin and its downstream kinase calcium/calmodulin- dependent protein kinase (CaMK) to facilitate GLUT4 translocation in skeletal muscle. In turn,

CaMK phosphorylates a small protein called phospholamban (PLN) at Thr17, which leads to an activation of the SERCA2a pump to increase calcium influx into the sarcoplasmic reticulum

(Vittone et al., 2008), thereby restoring calcium homeostasis. However, while the molecular mechanisms by which insulin and contraction increase glucose uptake in skeletal muscle are quite well described (Richter and Hargreaves, 2013), those triggered by hypoxia remain to be largely defined.

Finally, the small Rho GTPase Ras-related C3 botulinum toxin substrate 1 (Rac1) is another downstream target of PI3K controlling glucose uptake. In mature skeletal muscle, it acts in parallel, but independently of Akt and is not activated by 5-aminoimidazole-4-carboxamide ribonucleotide, suggesting it is AMP-activated protein kinase (AMPK)-independently regulated

(Sylow et al., 2013a). Rac1 activates p21 protein-activated kinase (PAK) by releasing the latter from its autoinhibitory domain, thereby allowing PAK1 to be phosphorylated at Thr423 and PAK2 at Thr402, although other mechanisms of regulation have been found as well (Radu et al., 2014).

Rac1 has been linked to insulin resistance in obese and diabetic patients (Sylow et al., 2013b), but

5 has never been studied after a normal meal with physiological plasma insulin concentrations or in response to severe hypoxia in humans.

In summary, acute hypoxia seems to induce whole body insulin resistance against the face of enhanced rate of glucose uptake in skeletal muscle. The latter regulation seems therefore independent of insulin but the molecular mechanisms are not defined yet. Based on preliminary results showing a reduced blood glucose peak after a meal due to exposure to environmental hypoxia, we hypothesize that the blunted peak could be explained by enhanced GLUT4 translocation in hypoxic conditions. The aim of the present work was therefore to investigate intracellular mechanisms known to regulate GLUT4 translocation in human skeletal muscle of subjects exposed to environmental hypoxia and to determine how hypoxic conditions alter blood insulin levels and downstream skeletal muscle signaling.

6 Materials and Methods

Subjects and study design

Experimental session 1: Eight healthy young men (age 24.0 ± 0.8 years; BMI 21.1 ± 0.6 kg·m-2) participated in this study, which was approved by the local Ethics Committee (KU Leuven) and was in conformity with The Helsinki Declaration. Subjects were asked to refrain from vigorous physical activity for 2 days, as well as to abstain from alcohol consumption the day before the experiments. Furthermore, they were not exposed for more than 7 days to an altitude higher than

1500 m within a period of 3 months preceding the start of the study. A written consent was obtained from all subjects after explaining all potential risks of the study. A pre-study medical checkup showed no anti-diabetic drug usage and HOMA-IR ranged from 0.4 to 0.9 indicating no signs of pre-diabetes.

All subjects underwent 2 trials (Fig. 1), one in normoxia (NOR) and one in hypoxia (HYP) in a randomized cross-over designed order, with a 4-week interval period in between. After a standardized dinner (58 % carbohydrates, 28 % fat and 14 % protein) and 8h overnight fast, the subjects reported to the laboratory between 6:00 – 8:30 am where they received a standardized high-glycemic meal (65 % carbohydrates, 29 % fat and 6 % protein, glycemic load range: 49 –

67). A period of exactly 20 min was provided to consume the meal and an additional 40-min rest was given. The subjects were then transferred to an air-conditioned (~21°C) hypoxic facility

(Sporting Edge, Leicestershire, UK) where they remained seated for 4 h. The chamber was either maintained at 20.93 % O2 (NOR), or at 11.1 % O2 (HYP) corresponding to ~5000 m of altitude.

Capillary blood glucose was measured with an Glucocard X-meter, (Arkray, Kyoto, Japan), which has been validated by Freckmann et al. (2010). Measurements were taken in the fasted state (T-60), 20 min after meal completion (T-20), 40 min after meal completion (T0,

7 immediately before entering the room), one minute after entering the room (T1) and thereafter every 30 min until the end of the protocol (T240). Another capillary blood sample was taken at

T-60 and T240 for insulin determination. Subjects remained seated during the whole experimental period, while reading a book or watching a movie.

Experimental session 2: Based on the results from experimental session 1 showing a reduced blood glucose peak after a meal due to exposure to environmental hypoxia and knowing the essential role of skeletal muscle in blood glucose homeostasis, we aimed to determine the contribution of skeletal muscle to our preliminary observations. We therefore repeated a similar experiment including muscle biopsies. Samples from experimental session 2 have already been used in a previous work where the protocol has been extensively described (D'Hulst et al., 2013).

In the present study, we used the samples to generate new data except for plasma insulin levels which have been published in our previous work. Briefly, fifteen healthy young men (age 21.3 ±

0.4 years; BMI 21.8 ± 0.5 kg·m-2) participated in this experimental session 2. The protocol was the same as in experimental session 1 except for blood sampling and muscle biopsies. Forty min after meal completion (T0), a first biopsy sample, with the needle pointing proximally (Van

Thienen et al., 2014), was taken from the left m. vastus lateralis under local anesthesia (1-2 ml

Lidocaine) through a 5-mm incision in the skin. Immediately after, a blood sample was taken from an antecubital vein. After those first measurements in normoxia, subjects were transferred to the hypoxic facility, which was either maintained at 20.93 % O2 (NOR) or set at 11.1 % O2

(HYP). Immediately after entering the room (T1), capillary blood glucose was measured and this was repeated every 30 min until the end of the trial. One hour after the first biopsy (T60), a second biopsy was taken through the same incision as the first, yet with the needle pointing distally. Furthermore, a venous blood sample was taken. Finally, at T240 the last biopsy and

8 blood sample were taken. The last biopsy was taken with the needle pointing distally and through a new incision in the skin 3 cm distally to the first incision.

Cellular fractionation

Details of the fractionation protocol are described elsewhere (Gualano et al., 2011). Briefly, ~15 mg of frozen muscle sample was homogenized 3 x 5 s with a polytron mixer in an ice cold buffer

(1:15, w/v) [20 mM Tris–HCl at pH 7.5, 250 mM sucrose, 2 mM EDTA,10 mM EGTA, and a protease inhibitor cocktail (#04693124001, Complete Mini, Roche Applied Science, Vilvoorde,

Belgium)]. The homogenate was centrifuged at 900 g for 10 min (4°C). The obtained supernatant

(s1) was further centrifuged at 100 000g for 30 min (4°C). This resulted in a pellet and a second supernatant (s2). The supernatant (s2) was used as a cytosolic lysate and the pellet was carefully re-dissolved in a 1% Triton X-100 lysis buffer and centrifuged at 100 000g for 30 min (4°C) to obtain the crude membrane fraction (supernatant, s3).

Whole cell lysate

Frozen muscle tissue (~20 mg) was homogenized 3 x 5 s with a Polytron mixer in an ice-cold buffer (1:10, w/v) [50 mM Tris-HCl pH 7.0, 270 mM sucrose, 5 mM EGTA, 1 mM EDTA, 1 mM sodium orthovanadate, 50 mM glycerophosphate, 5 mM sodium pyrophosphate, 50 mM sodium fluoride, 1 mM DTT, 1 % Triton-X 100 and the same protease inhibitor cocktail as described above]. Homogenates were then centrifuged at 10 000g for 10 min at 4°C. The supernatant was collected and immediately stored at -80°C. The protein concentration was measured using the DC protein assay kit (Bio-Rad laboratories, Nazareth, Belgium).

Western blot

9 Proteins (15-30 µg) were separated by SDS-PAGE (7.5 – 12.5 %) and semi-dry transferred to

PVDF membranes (Immobilon Transfer Membrane; Millipore, Hellerup, Denmark).

Subsequently, membranes were blocked with 3 % non-fat milk or 3 % BSA for 1 h and afterwards incubated overnight (4°C) with the following antibodies: p-LIMK1/2 Thr508/Thr505

(#3841), LIMK1 tot (#)3842, p-PAK1/2 Thr423/Thr402 (#2601), PAK1tot (#2602), p-GSK-3α/β

Ser21/9 (#9331), p-TBC1D4 Thr642 (#8881), p-CAMKII Thr286 (12716), CAMKII pan (#4436), eEF2 tot (#2332) (Cell Signaling Technology, Leiden, The Netherlands), Akt Thr308 (#05-802R),

AS160/TBC1D4 tot (#07-741), Rac1 (#05-389) (Millipore, Hellerup, Denmark), p-PLN (#sc-

17024-R) (Santa Cruz, Huissen, The Netherlands), GLUT4 (#PA1-1065) (Thermo Scientific,

Roskilde, Denmark), p-GS3+3a en p-TBC1D4 Ser588 were homemade. Appropriate horseradish peroxidase-conjugated secondary anti-bodies (DAKO) were used for chemiluminescent detection of proteins. Bands were visualized using the BioRad ChemiDocTM MP Imaging system (Bio-

Rad, Copenhagen, Denmark) and enhanced chemiluminescence (ECL+; Amersham Biosciences,

Brondby, Denmark). Then, membranes were stripped and re-probed with the antibody against the respective total protein to ascertain the relative amount of the phosphorylated protein compared to total amount of protein throughout the whole experiment. The results are presented as the ratio phosphorylated/total amount of the proteins, except for p-PLN which was reported to eEF2, which was found not to be modified by any condition during preliminary experiments. For p-

GSK-3α/β and p-GS site 3a+b, we used coomassie blue staining as loading control, because intramuscular glycogen levels can lead to variations in the total form of these proteins (Nielsen and Wojtaszewski, 2004). A value of 1 was arbitrarily assigned for each subject to the respective

T0 conditions (NOR and HYP) to which the values obtained at T60 and T240 were reported.

Analysis of blood samples

10 Experimental session 1: plasma insulin was assayed by ELISA (EIA1825, DRG International,

Inc., USA). Experimental session 2: plasma insulin was assayed by chemiluminescence using the

Siemens DPC kit (Brussels, Belgium) according to the instructions of the manufacturer; the interassay coefficient of variation of this measurement has been reported to be less than 10%

(Gkourogianni et al., 2014). Serum C-Peptide was measured by ELISA (ALPCO Diagnostics,

Salem, USA), all samples were analyzed in duplicate with an averaged coefficient of variation within duplicates of 5.3%.

Indices obtained from blood samples

Using the data from experimental session 2, we calculated glucose-to-insulin ratio (G/I) and C- peptide-to-insulin molar ratio at T240 as well as delta insulin and delta C-peptide (∆T240-T0 and

∆T240-T60) as surrogate markers of insulin resistance and hepatic insulin clearance (Singh and

Saxena, 2010).

Statistical analyses

A repeated measures ANOVA design was used to assess the statistical significance of differences between mean values over time and between conditions. When appropriate, Bonferroni pairwise multiple comparison test was used as post-hoc. A paired student t-test was used to determine significant differences for the area under the curve of blood glucose, glucose-to-insulin ratio, C- peptide-to-insulin molar ratio, delta insulin and delta C-peptide. Area under the glucose curve was calculated using the trapezoid method. Threshold of significance was set at p=0.05. Results are expressed as the means ± SEM. SEM at T0 represents inter-subject variability and SEM at

T60 and T240 represent intra-subject variability.

11 Results

Acute environmental hypoxia decreases the blood glucose response to a high glycemic meal

Experimental session 1. The blood glucose response to the standardized high glycemic meal peaked 20 min after completion (p=0.001), corresponding to 20 min before entering the room, with no difference between NOR and HYP (Fig. 2a). Immediately after entering the room (T1), blood glucose dropped by 12% in NOR and by 21% in HYP but only the decrease in HYP was significant (p=0.008, T1 vs T0 in HYP). Sixty minutes after entering the room (T60), blood glucose increased a second time in NOR compared to fasted levels (p=0.023) although the peak was much less than 20 min after completion of the meal. This increase was not observed in HYP, resulting in a 18% difference in glucose levels between NOR and HYP at T60 (p=0.002). Plasma insulin level did not differ between NOR and HYP at basal but was 25% higher in HYP compared to NOR at T240 (p=0.034, Fig. 2b).

Experimental session 2. As there was no difference between NOR and HYP in the first part of the glucose curve (from T-60 to T0) in the first experimental session, glucose measurements were started when the subjects entered the room (T1) in the second experimental session. Compared to

NOR, blood glucose response in HYP was reduced 30, 60 and 90 min after entering the room

(corresponding to 70, 100 and 130 min after consumption of the meal) (p=0.056, p<0.001 and p=0.015, respectively, Fig. 2c). As a result, the area under the curve was lower in HYP compared to NOR (396.6±605.9 in HYP vs. 2340.9±603.9 mg·dl-1·240min-1 in NOR, p=0.006, Fig. 2d). At

40 min after the meal (T0) when the first blood sample was obtained, plasma insulin had probably already peaked (Ostman et al., 2005). Thereafter plasma insulin decreased over time in both NOR and HYP (p<0.001, Fig. 2e). However, the decrease from T60 to T240 was less in

HYP compared to NOR (-3.2±3.7 in HYP vs. -15.1± 3.4 µU·ml-1 in NOR, p=0.043, Table 1),

12 resulting in a ~2 fold higher insulin level in HYP at the end of the experimental trial (p=0.033,

Fig. 2e). Plasma C-peptide also decreased in both NOR and HYP over time (p<0.001, Fig. 2f) but less in HYP than in NOR (-729.1±101.2 in HYP vs. -983.2±95.1 ng·ml-1 in NOR, p=0.046, Table

1). At T240, plasma C-peptide tended to be higher in HYP compared to NOR (+34%, p=0.092).

The C-peptide-to-insulin molar ratio, considered as an index of hepatic insulin clearance (Osei et al., 1984), was 37% higher in NOR compared to HYP at T240 (p=0.039, Table 1). At the same time point, glucose-to-insulin ratio was 91% (p=0.001) lower in HYP compared to NOR.

Intramuscular response to insulin is unaffected by hypoxia

To test whether acute hypoxia impairs postprandial downstream insulin signaling, we measured phosphorylation of Akt and its downstream targets. Phospho-Akt at Thr308 and phospho-GSK-3α at Ser21 decreased from T0 to T240 in NOR (p<0.05), but not in HYP, resulting in a 2-fold higher phosphorylation in HYP compared to NOR at T240 (p<0.05, Fig. 3a and b) similar to the phosphorylation pattern of p-Akt at Ser479 as published before (D’Hulst et al. 2013). GSK-3β phosphorylation at Ser9 decreased in NOR only (p=0.015, Fig. 3c) and did not differ between conditions at T240. Phosphorylation of GSK-3 decreases its activity resulting in decrease phosphorylation of GS. Following its upstream kinase, GS phosphorylation at site 3a+3b increased only in NOR (p=0.001, Fig. 3d) and tended to be lower in HYP compared with NOR at

T240 (p=0.055).

Environmental hypoxia transiently increases the presence of GLUT4 at the membrane

To further determine the mechanisms contributing to the lowered blood glucose response in

HYP, GLUT4 was quantified in crude membrane and cytosolic fractions. GLUT4 in the cytosolic fraction was ~40% lower at T60 in HYP compared to NOR (p=0.013, Fig. 4a), whereas at the same time point GLUT4 in the membrane fraction tended to display the inverse pattern, namely a

13 30% increase (p=0.070, Fig. 4b). As expected, no change was observed for whole cell GLUT4 over time or between conditions (Fig. 4c). In order to determine possible mechanisms, we measured p-TBC1D4, as it acts distally of Akt and inhibits GLUT4 translocation when dephosphorylated. A general decrease in TBC1D4 phosphorylation at Thr642 was observed over time (p=0.010, Fig. 4d) with no difference between NOR and HYP. Phospho-TBC1D4 at Ser588 followed a similar pattern, but the decrease in phosphorylation was only present in NOR at T240 vs T60 (p<0.001, Fig. 4e). Similarly to Thr642, no difference between NOR and HYP was observed on Ser588 at any time point, despite higher insulin concentrations in HYP at T240.

Calcium and Rac1 signaling

Neither hypoxia nor the meal modified CaMKII phosphorylation (Fig. 5a). Despite this, hypoxia increased phosphorylation of PLN at Thr17 by 25% at T240 (p=0.046, Fig. 5b) and tended to increase it at T60 (p=0.070). Upon activation Rac1 translocates to the sarcolemma and Rac1 is necessary for normal insulin-stimulated regulation of glucose uptake (Satoh, 2014). Therefore we investigated Rac1 localization and downstream signaling. Rac1 was more abundant in the plasma membrane in NOR compared to HYP at T240 (p=0.003) and tended to be more elevated at T60

(p=0.081, Fig. 5d). Furthermore, PAK1, a well-known downstream target of Rac1 (Chiu et al.,

2011), showed a similar phosphorylation pattern. Phospho-PAK1/2 at Thr423/402 increased by 79% from T0 to T240 in NOR only (p=0.005, Fig. 5e). Phospho-LIMK Thr508 was not modified by hypoxia at any time point.

14 Discussion

In the present study, we show that after a high-glycemic meal, acute environmental hypoxia: 1) decreases the blood glucose peak independently of insulin and possibly due to increased GLUT4 at the sarcolemmal membrane; 2) delays the decrease in plasma insulin level after a meal, which results in proportional activation of Akt but not of Rac1 signaling in skeletal muscle.

Plasma insulin and C-peptide levels were higher after 4-h exposure to environmental hypoxia compared to normoxia, as the postprandial decrease of both hormones was blunted in hypoxia. In contrast, blood glucose was identical between conditions at that time point. As skeletal muscle accounts for ~70% of peripheral glucose uptake under high-insulinemic conditions (DeFronzo et al., 1981), we further sought to determine any defects in downstream intramuscular insulin signaling in hypoxic conditions. To our surprise, phosphorylation of Akt, GSK-3α/β and GS as well as S6K1 (D'Hulst et al., 2013), all followed plasma insulin almost completely, indicating that insulin signaling is not impaired in response to acute hypoxia in human skeletal muscle, at least not at the Akt-GSK signaling level. To our knowledge, this is the first study to compare the acute activation of intramuscular insulin signaling in hypoxia to the activation in normoxia in humans. Previous data come from in vitro studies or studies on chronic hypoxia. The main conclusions of those studies is that even though basal downstream insulin signaling is likely impaired with chronic hypoxia in skeletal muscle (Alvarez-Tejado et al., 2001; Favier et al.,

2010), its ability to be stimulated by insulin remains intact, at least to a certain extent and duration.

In addition to modulate insulin sensitivity, hypoxia has been shown to regulate glucose uptake in several tissues (Wojtaszewski et al., 1998; Kelly et al., 2010). We did not assess glucose uptake directly but the lower blood glucose levels and the predominant role of skeletal muscle in glucose

15 disposal (DeFronzo et al., 1981) naturally led to the hypothesis that glucose uptake could indeed be up-regulated in skeletal muscle in our hypoxic conditions. One argument in favor of this is the decrease in cytosolic GLUT4 concomitantly with a tendency to an increase in sarcolemmal

GLUT4 after 1h, suggestive of a higher glucose influx into skeletal muscle. However, our experimental setup cannot rule out a specific role of other insulin dependent tissues. Therefore further research using glucose tracers is needed to determine the specific role of fat, liver and skeletal muscle tissue in postprandial glucose uptake under hypoxic conditions.

Importantly, contrary to previous results in C2C12 cells (Sakagami et al., 2014), changes in

GLUT4 localization were not initiated by HIF-1α as neither its expression nor transcription of downstream targets were increased (D'Hulst et al., 2013). We then seek to determine which HIF-

1α-independent molecular mechanisms could be responsible for the decreased blood glucose levels and increased GLUT4 location at the sarcolemma. One key molecule in the latter process is TBC1D4, which mediates both insulin-dependent and -independent translocation of GLUT4 in skeletal muscle (Treebak et al., 2014), making it an interesting target to analyze in the present study. However, phosphorylation of TBC1D4 at Ser588 and at Thr642 was not modified after 1h either in normoxia or in hypoxia. Together with unchanged phosphorylation of Akt and AMPK

(D'Hulst et al., 2013), two upstream kinases of TBC1D4 (Treebak et al., 2014), our results rule out a major role of TBC1D4 in the regulation of GLUT4 translocation in our conditions.

Confirming previous reports (Wojtaszewski et al., 1998; Hayashi et al., 2000), the decrease in blood glucose by hypoxia was insulin-independent as insulin levels were identical between normoxia and hypoxia after 1h, while glucose levels were significantly lower in hypoxia.

Hypoxia per sé has been postulated to increase glucose transport via AMPK (Mu et al., 2001) and calcium (Wright et al., 2005) -dependent pathways. Although skeletal muscle oxygenation index

16 was decreased by 7.2% (D'Hulst et al., 2013), we did not find any changes in AMPK, or downstream ACC phosphorylation (data not shown) in any condition. AMPK is a very labile protein and absence of changes in AMPK phosphorylation at Thr172 does not mean that this protein is not involved in glucose metabolism in the present study. Indeed, this is in contrast to a recently published study from our laboratory wherein p-AMPK at Thr172 was elevated by 25% at ~5000m

(Masschelein et al., 2014), probably because of a slightly different sampling time, whereby we likely missed variations in AMPK phosphorylation, hypoxic stimulus, nutritional state or reactive oxygen species (ROS) production (Emerling et al., 2009). However, although we did not assess

ROS production, the latter probably does not play a critical role in hypoxia-induced glucose uptake (Sandstrom et al., 2006).

Hypoxia also regulates calcium efflux out of the sarcoplasmic reticulum, resulting in increased

GLUT4 translocation in a CAMK-dependent and insulin-independent ways (Wright et al., 2005).

However, the phosphorylation state of CAMKII was not modified in hypoxia at 1h, suggesting that calcium signaling was not activated at that time in our conditions and therefore did not contribute to increase GLUT4 translocation and to decrease blood glucose levels. Interestingly, p-

PLN was increased after 4-h hypoxia without any alternations in CAMKII phosphorylation. Most likely this can be accounted to the higher and concomitant activation of Akt in hypoxia, which is sufficient to phosphorylate PLN at Thr17 independently of CaMKII (Catalucci et al., 2009) but the physiological significance of the increase in p-PLN at the end of the 4-h exposure remains to be determined.

Recently, the actin cytoskeleton-regulating GTP-ase Rac1 has been put forward as an important regulator of GLUT4 translocation and glucose transport via activation of PAK and its downstream target LIMK which finally results in cytoskeleton remodeling (Chiu et al., 2011).

17 Importantly, activation of Rac1 is regulated by both insulin and contraction in skeletal muscle and is AMPK-independent (Sylow et al., 2013a; Sylow et al., 2014), which made it a particular interesting target in this experimental setup, as there was no hypoxia-induced AMPK activation.

Hypoxia as well has been shown to activate Rac1 in hepatocytes (Mollen et al., 2007), in breast cancer cell lines (Du et al., 2011) and in epithelial cells (Mattagajasingh et al., 2012), but whether hypoxia regulates Rac1 in skeletal muscle was unknown prior to this study. In the present study,

Rac1 abundance in the membrane, a marker for its activation (Bustelo et al., 2007), as well as phosphorylation of PAK1/2 were increased over time in normoxia only. Therefore, contrary to our hypothesis, Rac1 does not mediate the presence of GLUT4 at the sarcolemmal membrane under hypoxia in vivo as the activation patterns of Rac1 and GLUT4 do not fit each other. Those results support the hypothesis that GLUT4 translocation in the present study is not primarily regulated by insulin. Indeed, previous studies identified Rac1 as a crucial element in insulin- dependent GLUT4 translocation in skeletal muscle that could play a major role in the development of insulin resistance (Sylow et al., 2013b; Satoh, 2014). Here, neither insulin nor

Rac1 did play a key role in the higher abundance of GLUT4 at the sarcolemmal membrane but we cannot exclude any contribution of Rac1 to the development of insulin intolerance observed at the end of the exposure to hypoxia. Insulin-induced activation of the Rac1/PAK pathway is known to be down-regulated in skeletal muscle of insulin-resistant mice and humans (Sylow et al., 2013b) but how this pathway is regulated after a meal in healthy people is currently unknown.

We show here that the Rac1/PAK pathway is activated over time after a meal in normoxia and that this activation is blunted in hypoxia. Knowing the close relationship between the Rac1/PAK pathway and insulin resistance, it is possible that the reduced activation of this pathway contributes to insulin resistance during acute hypoxia but its definite involvement can only be determined using KO or in vitro models.

18 In the present study, no glucose tracer or hyperinsulinemic euglycemic clamp was used, which limits the interpretation of the data. Another limitation is the lack of a pre-meal biopsy but it was deliberately chosen to focus on the response to a meal in hypoxic conditions as the effect of a meal itself on blood glucose and plasma insulin has already been documented largely before

(Reaven, 1979; Wolever and Bolognesi, 1996)

Conclusions

In conclusion, our data show that acute environmental hypoxia reduced blood glucose response to a high glycemic meal, possibly via an increase in GLUT4 abundance at the sarcolemmal membrane. Later on, whole body insulin intolerance develops independently of defects in conventional insulin signaling in skeletal muscle as the response of the downstream Akt/GSK-3 pathway paralleled changes in plasma insulin levels during exposure to hypoxia. Further investigation will be required to determine whether alteration in Rac1 signaling in skeletal muscle could contribute to insulin intolerance during acute hypoxia.

19 Acknowledgements

The authors would like to thank Erik Richter for financing the biochemical assays and for his insightful comments while generating the manuscript. We also thank Irene Bech Nielsen, Betina

Bolmgren, Ruud Van Thienen and Monique Ramaekers for their technical support. This study was not externally funded.

Conflicts of interest

The authors do not have any conflicts of interest.

Ethical standards

The authors declare that all experiments comply with the Belgian law.

20 Bibliography

Sigal RJ, Kenny GP, Wasserman DH, Castaneda-Sceppa C, White RD (2006) Physical activity/exercise and type 2 diabetes: a consensus statement from the American Diabetes Association. Diabetes Care 29:1433-1438 Castillo O, Woolcott OO, Gonzales E, Tello V, Tello L, Villarreal C, Mendez N, Damas L, Florentini E (2007) Residents at high altitude show a lower glucose profile than sea-level residents throughout 12-hour blood continuous monitoring. High Alt Med Biol 8:307-311 Marquez JL, Rubinstein S, Fattor JA, Shah O, Hoffman AR, Friedlander AL (2013) Cyclic hypobaric hypoxia improves markers of glucose metabolism in middle-aged men. High Alt Med Biol 14:263-272 Peltonen GL, Scalzo RL, Schweder MM, Larson DG, Luckasen GJ, Irwin D, Hamilton KL, Schroeder T, Bell C (2012) Sympathetic inhibition attenuates hypoxia induced insulin resistance in healthy adult humans. J Physiol 590:2801-2809 Larsen JJ, Hansen JM, Olsen NV, Galbo H, Dela F (1997) The effect of altitude hypoxia on glucose homeostasis in men. J Physiol 504 ( Pt 1):241-249 Oltmanns KM, Gehring H, Rudolf S, Schultes B, Rook S, Schweiger U, Born J, Fehm HL, Peters A (2004) Hypoxia causes glucose intolerance in humans. Am J Respir Crit Care Med 169:1231- 1237 Lecoultre V, Peterson CM, Covington JD, Ebenezer PJ, Frost EA, Schwarz JM, Ravussin E (2013) Ten nights of moderate hypoxia improves insulin sensitivity in obese humans. Diabetes Care 36:e197-e198 Regazzetti C, Peraldi P, Gremeaux T, Najem-Lendom R, Ben-Sahra I, Cormont M, Bost F, Le Marchand-Brustel Y, Tanti JF, Giorgetti-Peraldi S (2009) Hypoxia decreases insulin signaling pathways in adipocytes. Diabetes 58:95-103 Taniguchi CM, Finger EC, Krieg AJ, Wu C, Diep AN, LaGory EL, Wei K, McGinnis LM, Yuan J, Kuo CJ, Giaccia AJ (2013) Cross-talk between hypoxia and insulin signaling through Phd3 regulates hepatic glucose and lipid metabolism and ameliorates diabetes. Nat Med 19:1325-1330 Dongiovanni P, Valenti L, Ludovica FA, Gatti S, Cairo G, Fargion S (2008) Iron depletion by deferoxamine up-regulates glucose uptake and insulin signaling in hepatoma cells and in rat liver. Am J Pathol 172:738-747 Sakagami H, Makino Y, Mizumoto K, Isoe T, Takeda Y, Watanabe J, Fujita Y, Takiyama Y, Abiko A, Haneda M (2014) Loss of HIF-1alpha impairs GLUT4 translocation and glucose uptake by the skeletal muscle cells. Am J Physiol Endocrinol Metab 306:E1065-E1076 Gamboa JL, Garcia-Cazarin ML, Andrade FH (2011) Chronic hypoxia increases insulin- stimulated glucose uptake in mouse soleus muscle. Am J Physiol Regul Integr Comp Physiol 300:R85-R91 Cartee GD, Douen AG, Ramlal T, Klip A, Holloszy JO (1991) Stimulation of glucose transport in skeletal muscle by hypoxia. J Appl Physiol 70:1593-1600 Roberts AC, Reeves JT, Butterfield GE, Mazzeo RS, Sutton JR, Wolfel EE, Brooks GA (1996) Altitude and beta-blockade augment glucose utilization during submaximal exercise. J Appl Physiol (1985 ) 80:605-615 Brozinick JT, Jr., Reynolds TH, Dean D, Cartee G, Cushman SW (1999) 1-[N, O-bis-(5- isoquinolinesulphonyl)-N-methyl-L-tyrosyl]-4- phenylpiperazine (KN-62), an inhibitor of calcium-dependent camodulin protein kinase II, inhibits both insulin- and hypoxia-stimulated glucose transport in skeletal muscle. Biochem J 339 ( Pt 3):533-540 Vittone L, Mundina-Weilenmann C, Mattiazzi A (2008) Phospholamban phosphorylation by CaMKII under pathophysiological conditions. Front Biosci 13:5988-6005 Richter EA, Hargreaves M (2013) Exercise, GLUT4, and Skeletal Muscle Glucose Uptake. Physiol Rev 93:993-1017

21 Sylow L, Jensen TE, Kleinert M, Mouatt JR, Maarbjerg SJ, Jeppesen J, Prats C, Chiu TT, Boguslavsky S, Klip A, Schjerling P, Richter EA (2013a) Rac1 is a novel regulator of contraction- stimulated glucose uptake in skeletal muscle. Diabetes 62:1139-1151 Radu M, Semenova G, Kosoff R, Chernoff J (2014) PAK signalling during the development and progression of cancer. Nat Rev Cancer 14:13-25 Sylow L, Jensen TE, Kleinert M, Hojlund K, Kiens B, Wojtaszewski J, Prats C, Schjerling P, Richter EA (2013b) Rac1 signaling is required for insulin-stimulated glucose uptake and is dysregulated in insulin-resistant murine and human skeletal muscle. Diabetes 62:1865-1875 Freckmann G, Baumstark A, Jendrike N, Zschornack E, Kocher S, Tshiananga J, Heister F, Haug C (2010) System accuracy evaluation of 27 blood glucose monitoring systems according to DIN EN ISO 15197. Diabetes Technol Ther 12:221-231 D'Hulst G, Jamart C, Van TR, Hespel P, Francaux M, Deldicque L (2013) Effect of acute environmental hypoxia on protein metabolism in human skeletal muscle. Acta Physiol (Oxf) 208:251-264 Van Thienen R, D'Hulst G, Deldicque L, Hespel P (2014) Biochemical artifacts in experiments involving repeated biopsies in the same muscle. Physiol Rep 2:e00286 Gualano B, DE SP, V, Roschel H, Artioli GG, Neves M, Jr., De Sa Pinto AL, da Silva ME, Cunha MR, Otaduy MC, Leite CC, Ferreira JC, Pereira RM, Brum PC, Bonfa E, Lancha AH, Jr. (2011) Creatine in type 2 diabetes: a randomized, double-blind, placebo-controlled trial. Med Sci Sports Exerc 43:770-778 Nielsen JN, Wojtaszewski JF (2004) Regulation of glycogen synthase activity and phosphorylation by exercise. Proc Nutr Soc 63:233-237 Gkourogianni A, Kosteria I, Telonis AG, Margeli A, Mantzou E, Konsta M, Loutradis D, Mastorakos G, Papassotiriou I, Klapa MI, Kanaka-Gantenbein C, Chrousos GP (2014) Plasma metabolomic profiling suggests early indications for predisposition to latent insulin resistance in children conceived by ICSI. PLoS One 9:e94001 Singh B, Saxena A (2010) Surrogate markers of insulin resistance: A review. World J Diabetes 1:36-47 Ostman E, Granfeldt Y, Persson L, Bjorck I (2005) Vinegar supplementation lowers glucose and insulin responses and increases satiety after a bread meal in healthy subjects. Eur J Clin Nutr 59:983-988 Osei K, Falko JM, O'Dorisio TM, Adam DR (1984) Decreased serum C-peptide/insulin molar ratios after oral glucose ingestion in hyperthyroid patients. Diabetes Care 7:471-475 Satoh T (2014) Rho GTPases in insulin-stimulated glucose uptake. Small GTPases 5: Chiu TT, Jensen TE, Sylow L, Richter EA, Klip A (2011) Rac1 signalling towards GLUT4/glucose uptake in skeletal muscle. Cell Signal 23:1546-1554 DeFronzo RA, Jacot E, Jequier E, Maeder E, Wahren J, Felber JP (1981) The effect of insulin on the disposal of intravenous glucose. Results from indirect calorimetry and hepatic and femoral venous catheterization. Diabetes 30:1000-1007 Alvarez-Tejado M, Naranjo-Suarez S, Jimenez C, Carrera AC, Landazuri MO, del PL (2001) Hypoxia induces the activation of the phosphatidylinositol 3-kinase/Akt cell survival pathway in PC12 cells: protective role in apoptosis. J Biol Chem 276:22368-22374 Favier FB, Costes F, Defour A, Bonnefoy R, Lefai E, Bauge S, Peinnequin A, Benoit H, Freyssenet D (2010) Downregulation of Akt/mammalian target of rapamycin pathway in skeletal muscle is associated with increased REDD1 expression in response to chronic hypoxia. Am J Physiol Regul Integr Comp Physiol 298:R1659-R1666 Wojtaszewski JF, Laustsen JL, Derave W, Richter EA (1998) Hypoxia and contractions do not utilize the same signaling mechanism in stimulating skeletal muscle glucose transport. Biochim Biophys Acta 1380:396-404

22 Kelly KR, Williamson DL, Fealy CE, Kriz DA, Krishnan RK, Huang H, Ahn J, Loomis JL, Kirwan JP (2010) Acute altitude-induced hypoxia suppresses plasma glucose and leptin in healthy humans. Metabolism 59:200-205 Treebak JT, Pehmoller C, Kristensen JM, Kjobsted R, Birk JB, Schjerling P, Richter EA, Goodyear LJ, Wojtaszewski JF (2014) Acute exercise and physiological insulin induce distinct phosphorylation signatures on TBC1D1 and TBC1D4 proteins in human skeletal muscle. J Physiol 592:351-375 Hayashi T, Hirshman MF, Fujii N, Habinowski SA, Witters LA, Goodyear LJ (2000) Metabolic stress and altered glucose transport: activation of AMP-activated protein kinase as a unifying coupling mechanism. Diabetes 49:527-531 Mu J, Brozinick JT, Jr., Valladares O, Bucan M, Birnbaum MJ (2001) A role for AMP-activated protein kinase in contraction- and hypoxia-regulated glucose transport in skeletal muscle. Mol Cell 7:1085-1094 Wright DC, Geiger PC, Holloszy JO, Han DH (2005) Contraction- and hypoxia-stimulated glucose transport is mediated by a Ca2+-dependent mechanism in slow-twitch rat soleus muscle. Am J Physiol Endocrinol Metab 288:E1062-E1066 Masschelein E, Van Thienen R, D'Hulst G, Hespel P, Thomis M, Deldicque L (2014) Acute environmental hypoxia induces LC3 lipidation in a genotype-dependent manner. FASEB J 28:1022-1034 Emerling BM, Weinberg F, Snyder C, Burgess Z, Mutlu GM, Viollet B, Budinger GR, Chandel NS (2009) Hypoxic activation of AMPK is dependent on mitochondrial ROS but independent of an increase in AMP/ATP ratio. Free Radic Biol Med 46:1386-1391 Sandstrom ME, Zhang SJ, Bruton J, Silva JP, Reid MB, Westerblad H, Katz A (2006) Role of reactive oxygen species in contraction-mediated glucose transport in mouse skeletal muscle. J Physiol 575:251-262 Catalucci D, Latronico MV, Ceci M, Rusconi F, Young HS, Gallo P, Santonastasi M, Bellacosa A, Brown JH, Condorelli G (2009) Akt increases sarcoplasmic reticulum Ca2+ cycling by direct phosphorylation of phospholamban at Thr17. J Biol Chem 284:28180-28187 Sylow L, Kleinert M, Pehmoller C, Prats C, Chiu TT, Klip A, Richter EA, Jensen TE (2014) Akt and Rac1 signaling are jointly required for insulin-stimulated glucose uptake in skeletal muscle and downregulated in insulin resistance. Cell Signal 26:323-331 Mollen KP, McCloskey CA, Tanaka H, Prince JM, Levy RM, Zuckerbraun BS, Billiar TR (2007) Hypoxia activates c-Jun N-terminal kinase via Rac1-dependent reactive oxygen species production in hepatocytes. Shock 28:270-277 Du J, Xu R, Hu Z, Tian Y, Zhu Y, Gu L, Zhou L (2011) PI3K and ERK-induced Rac1 activation mediates hypoxia-induced HIF-1alpha expression in MCF-7 breast cancer cells. PLoS One 6:e25213 Mattagajasingh SN, Yang XP, Irani K, Mattagajasingh I, Becker LC (2012) Activation of Stat3 in endothelial cells following hypoxia-reoxygenation is mediated by Rac1 and protein Kinase C. Biochim Biophys Acta 1823:997-1006 Bustelo XR, Sauzeau V, Berenjeno IM (2007) GTP-binding proteins of the Rho/Rac family: regulation, effectors and functions in vivo. Bioessays 29:356-370 Reaven GM (1979) Effects of differences in amount and kind of dietary carbohydrate on plasma glucose and insulin responses in man. Am J Clin Nutr 32:2568-2578 Wolever TM, Bolognesi C (1996) Prediction of glucose and insulin responses of normal subjects after consuming mixed meals varying in energy, protein, fat, carbohydrate and glycemic index. J Nutr 126:2807-2812

23 Tables

Table 1 Indices for insulin sensitivity and hepatic insulin clearance NOR HYP p-value Glucose-to-insulin ratio 18.5 ± 2.9 9.7 ± 1.89 0.001 C-peptide-to-insulin molar ratio 0.28 ± 0.008 0.20 ± 0.006 0.039 ∆T240-T0 insulin (µU·ml-1) -21.4 ± 3.2 -18.1 ± 3.8 Ns ∆T240-T60 insulin (µU·ml-1) -15.1 ± 3.4 -3.2 ± 3.7 0.043 ∆T240-T0 C-peptide (ng·ml-1) -983.2 ± 95.1 -729.1 ± 101.2 0.046 ∆T240-T60 C-peptide (ng·ml-1) -886.2 ± 141.9 -763.5 ± 214.9 Ns

Glucose-to-insulin ratio and C-peptide-to-insulin molar ratio at T240, delta insulin and C-peptide levels (∆T240-T0 and ∆T240-T60) in normoxia (NOR) and hypoxia (HYP). Values are means ± SEM (n=15).

24 Figures captions

Fig. 1

Experimental protocols of session 1 and session 2.

Fig. 2

Effect of environmental hypoxia and a high glycemic meal on insulin and glucose levels. (a)

Blood glucose during experimental session 1 (see methods section). (b) Plasma insulin during experimental session 1 at T-60 (fasted) and T240 (end). (c) Blood glucose experimental session

2. (d) Area under the curve for blood glucose experimental session 2. (e) Plasma insulin and (f) serum C-peptide at T1, T60 and T240 during experimental session 2. Data shown are expressed as means ± SEM (n=8 experimental session 1, n=15 experimental session 2). SEM at T0 represents inter-subject variability and SEM at T60 and T240 represent intra-subject variability.

$p<0.05 vs T-60; *p<0.05 vs T1; §p<0.05 vs T60; £p<0.05 vs T0; ‡p<0.05 vs NOR.

Fig. 3

Effect of environmental hypoxia and a high glycemic meal on downstream insulin signaling. (a)

Akt phosphorylation at Thr308 (b) GSK-3α phosphorylation at Ser21, (c) GSK-3β phosphorylation at Ser9 and (d) GS phosphorylation at site 3a+3b at basal (T0), after 1h (T60) and after 4h (T240) in normoxia (NOR) or in hypoxia (HYP). (e) Representative blots. Directly compared bands were all from the same subject. Data shown are expressed as means ± SEM (n=15). SEM at T0 represents inter-subject variability and SEM at T60 and T240 represent intra-subject variability.

*p<0.05 vs T1; §p<0.05 vs T60; ‡p<0.05 vs NOR; (‡)p=0.055 vs NOR.

25 Fig. 4

Effect of environmental hypoxia and a high glycemic meal on GLUT4 and downstream signaling.

(a) Cytosolic GLUT4 fraction, (b) membrane GLUT4 fraction, (c) whole cell GLUT4, (d)

TBC1D4 Thr642 phosphorylation and (e) TBC1D4 Ser588 phosphorylation at basal (T0), after 1h

(T60) and after 4h (T240) in normoxia (NOR) or in hypoxia (HYP). (f) Representative blots.

Directly compared bands were all from the same subject. Data shown are expressed as means ±

SEM (n=15 for TBC1D4 phosphorylation, n=11 for membrane and cytosolic fractions). SEM at

T0 represents inter-subject variability and SEM at T60 and T240 represent intra-subject variability. *p<0.05 vs T1; ‡p<0.05 vs. NOR; §p<0.05 vs T60; (‡)p<0.10 vs NOR.

Fig. 5

Effect of environmental hypoxia and a high glycemic meal on downstream calcium and Rac1 signaling. (a) CaMKII Thr286, (b) PLN Thr17 phosphorylation, (d) membrane Rac1 fraction, (e)

PAK1 Thr423 and (f) LIMK1 Thr508 phosphorylation at basal (T0), after 1h (T60) and after 4h

(T240) in normoxia (NOR) or in hypoxia (HYP). (c) and (g) Representative blots. Directly compared bands were all from the same subject. Data shown are expressed as means ± SEM

(n=15). SEM at T0 represents inter-subject variability and SEM at T60 and T240 represent intra- subject variability. *p<0.05 vs. T0; ‡p<0.05 vs. NOR; (‡)p=<0.10 vs. NOR.