TBC1D4-S711 Controls Skeletal Muscle Insulin Sensitization After Exercise and Contraction

Progressive deterioration of insulin-stimulated skeletal muscle glucose uptake is a hallmark in the etiology of type 2 diabetes (1). Remarkably, a single bout of exercise enhances insulin-stimulated glucose uptake in skeletal muscle from healthy and insulin-resistant subjects (2–5) as well as rodent models (6–8). This effect of exercise may last for up to 48 h (6,9) and highlights the huge potential of exercise to improve insulin-stimulated muscle glucose uptake in diseases characterized by muscle insulin resistance. The insulin-sensitizing effect of acute exercise on muscle glucose uptake seems to rely on elevated muscle perfusion (10) in combination with enhanced GLUT4 translocation (11–13), which occurs independently of changes in proximal insulin signaling (4,5,8,14–16) and protein synthesis (14).

Insulin-stimulated muscle glucose uptake is mediated through an Akt-dependent signaling pathway that involves inhibition of the Rab-GTPase–activating protein TBC1D4 (also known as AS160) by multisite phosphorylation (17,18). This releases intracellular GLUT4 storage vesicles to fuse with the muscle surface membrane, which increases glucose transport capacity (19,20). In mouse skeletal muscle, TBC1D4 is phosphorylated at the serine residue 711 (704 in humans) during insulin stimulation as well as by AMP-activated protein kinase (AMPK) in response to exercise and contraction. However, phosphorylation of TBC1D4-S711 does not seem to be necessary to promote glucose uptake in itself (21).

Several studies in humans and rodents have shown that enhanced insulin-stimulated phosphorylation of TBC1D4 occurs in parallel with improvements in muscle insulin sensitivity after a single bout of exercise and contraction (5,8,15,22–25). We have shown that AMPK is necessary for the ability of exercise and contraction to enhance whole-body and muscle insulin sensitivity (26). In addition, we have shown that TBC1D4 is necessary for contraction and AICAR (5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside; AMPK activator) to enhance muscle insulin sensitivity (27), and several human studies have demonstrated enhanced phosphorylation of TBC1D4-S711 concomitant with improved muscle insulin sensitivity after exercise (5,28,29). Together, this inspired us to hypothesize that an AMPK-TBC1D4-S711 signaling axis mediates the beneficial effect of exercise and contraction on muscle insulin sensitivity.

To test our hypothesis, we generated a knock-in mouse model with a point mutation that replaces serine with alanine at residue 711 in the long isoform of Tbc1d4 (TBC1D4-S711A). Our findings reveal that the lack of phosphorylated TBC1D4-S711 does not affect whole-body and muscle insulin sensitivity in resting mice on either chow or high-fat diets (HFDs). In contrast, TBC1D4-S711A mice do not exhibit enhanced whole-body and muscle insulin sensitivity after exercise and contraction. Collectively, these results suggest that phosphorylation of TBC1D4-S711 is necessary for muscle insulin sensitization after exercise and contraction in mice.

All animal experiments were approved by the Danish Animal Experiments Inspectorate (Copenhagen, Denmark) (permit number 2019-15-0201-01659) and complied with the EU convention for the protection of vertebrates used for scientific purposes. TBC1D4-S711A mice were generated using CRISPR-Cas9 technology as previously described (27). In short, Cas9 protein with a single guide RNA (sgRNA) was used together with a repair template encoding the desired mutations to substitute the S711 codon (TCT) with an alanine codon (GCG) in exon 11 that is present only in the long isoform of Tbc1d4. A silent mutation was introduced in the repair template to prevent sgRNA reannealing (Fig. 1A). Heterozygous TBC1D4-S711A male and female mice were used to breed wild-type, heterozygous, and homozygous TBC1D4-S711A mice, all of which were fertile and viable. To ensure correct mutagenesis, the three different genotypes were sequenced around the mutated codons (Fig. 1B) and subsequently genotyped using the following PCR primer sets: WT: FP 5′-GAGTAATTCTTCCTGCAGCCTTCCAA-3′ and RP 5′-GGAGCAGTGAAGGAAGGGGCAGAG-3′; and S711A: FP 5′-GAGTAATTCTTCCTGCAGCCTTCCATC-3′ and RP 5′-GCAGTGAAGGAAGGGGCCG-3′. Heterozygous long isoform–specific TBC1D4 knockout (KO) mice were also generated, genotyped, and used as previously described (27). Mice were bred at the Department of Experimental Medicine, University of Copenhagen, and housed with nesting material in temperature- and humidity-controlled rooms on a 12:12-h light-dark cycle (lights on 06:00 a.m.) with free access to standard rodent chow diet (Altromin #1324) and water. All experiments were performed on female mice unless stated otherwise.

Body weight and food intake were recorded once weekly using a precision scale. Lean mass and fat mass were measured at the end of the diet interventions using an EchoMRI 4in1 Body Composition Analyzer.

Eleven- to thirteen-week-old mice were single-caged and provided free access to an HFD (60% calories from fat; Research Diets #D12492) for 14 weeks.

For fasting and refeeding experiments, fed blood glucose concentrations were measured in single-housed mice at ∼07:00 a.m. Fasting and refeeding blood glucose concentrations were measured in mice fasted 16 h overnight (start ∼03:00 p.m.) as well as 30 and 60 min after being given free access to standard chow diet. For the oral glucose tolerance test (OGTT), single-housed mice were fasted 16 h overnight (start ∼03:00 p.m.) after which d-glucose (2 g/kg body weight) dissolved in saline (0.9%) was administered by oral gavage. For the insulin tolerance test (ITT), single-housed mice were fasted 4 h (start ∼07:00 a.m.), after which insulin (0.75 units/kg, Actrapid; Novo Nordisk) dissolved in saline (0.9%) was administered by a single intraperitoneal injection. Blood glucose levels were determined at t = 0, 15, 30, 45, 60, and 120 min during the OGTT and ITT. At all time points during the fasting refeeding experiments as well as at t = 0, 15, and 30 min during the OGTT, ∼35 μL blood was drawn into heparinized capillary tubes to determine insulin concentrations in plasma, which was prepared from whole blood centrifugation (16,000 g for 5 min at 4°C). HOMA-IR (homeostatic model assessment for insulin resistance) was calculated from the following equation: HOMA-IR = Fasting plasma insulin · Fasting blood glucose/22.5.

Handheld glucose (Contour XT; Bayer) and lactate (Lactate Plus; Nova Biomedical) analyzers were used for all blood glucose and lactate measurements that were sampled from the tail vein. Plasma insulin was determined using an ELISA kit from ALPCO (80-INSMSU-E10).

Muscle glycogen was measured on whole muscle homogenate after acid hydrolysis and determined by a fluorometric method (30).

Procedures for isolation of RNA and reverse transcription have been described previously (31). The mRNA content of the short and long isoform of TBC1D4 was determined by real-time PCR using SYBR Green (SYBR Green PCR Master Mix; Thermo Fischer Scientific, Woolston Warington, U.K.). PCR primer set for the short and long isoform of the mouse Tbc1d4 gene (short: FP 5′-TCCAGCGAACAATGCAGTGA-3′ and RP 5′-GATGCAACCCGGAGGAAAAT-3′; long: FP 5′-CAGCCCGGCGCATGTA-3′ and RP 5′-CAGGAAAGAGGGAGCAGTGAA-3′) was designed using Primer Express software (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA) and obtained from TagCopenhagen (Copenhagen, Denmark). A serial dilution was made separately for subcutaneous white adipose tissue (ScWAT) and skeletal muscle cDNA samples and used to construct tissue-specific standard curves. These were used to convert the cycle threshold values of the unknown samples, taking into account tissue-specific PCR efficiency, while the y-axis intersect value of the muscle standard curve was used for both muscle and ScWAT to obtain comparable mRNA levels. The amount of TBC1D4 mRNA was normalized to the average amount of β-actin and TBP mRNA of each cDNA sample. The β-actin and TBP mRNA were determined using predeveloped assay reagents and Universal Mastermix (Applied Biosystems) and were not affected by the genotype (32). The amount of TBC1D4 mRNA is, in all three tissues, presented as fold-change relative to the level in soleus muscle from wild-type mice, which was set to one.

Fed and single-housed mice were anesthetized by a single intraperitoneal injection of pentobarbital and xylocain (90 mg/kg and 5 mg/kg, respectively) and left to recover for 15 min. Subsequently, a bolus of saline (0.9%) with or without insulin (0.75 units/kg), containing [³H]-2-deoxyglucose (66.7 uCi/mL) (Hartmann Analytic, Braunschweig, Germany), was administered by a single retro-orbital intravenous injection. Blood glucose concentrations were determined at t = 0, 5, 10, and 15 min. Moreover, 25 μL blood was collected at similar time points to determine the specific activity of [³H]-2-deoxyglucose in plasma. Tissues were collected at t = 15 min after euthanization of animals by cervical dislocation to determine phosphorylated [³H]-2-deoxyglucose ([³H]-2-deoxy-glucose-6-P) tissue content as previously described (33). Tissue glucose clearance rates (mL blood ⋅ g tissue⁻¹ ⋅ min⁻¹) were calculated by relating the amount of [³H]-2-deoxy-glucose-6-P accumulated in tissues to the tissue weight, the specific activity in plasma, the time of exposure, and the average blood glucose concentration.

Insulin- and AICAR-stimulated glucose uptake was determined in isolated skeletal muscle using a previously described technique (34). Although AICAR is an unspecific AMPK activator that promotes intracellular accumulation of ZMP (AICAR monophosphate), the effect of AICAR on muscle glucose uptake has repeatedly been shown to be dependent on AMPK (35–37). Fed animals were anesthetized by a single intraperitoneal injection of pentobarbital and xylocain (100 and 5 mg/kg, respectively) after which soleus and extensor digitorum longus (EDL) were removed and suspended in heated (30°C) and continuously gassed (95% O₂ and 5% CO₂) incubation chambers (Multi Myograph System; Danish Myo Technology, Aarhus, Denmark) containing Krebs Ringer buffer supplemented with 0.1% BSA, 8 mmol/L mannitol, and 2 mmol/L pyruvate (KRB). After 10 min preincubation in KRB, muscles were incubated for 30 min in KRB containing AICAR (2 mmol/L) (Toronto Research Chemicals, Toronto, Ontario, Canada) or insulin (100 or 10,000 μU/mL) (Actrapid; Novo Nordisk, Denmark). The uptake of 2-deoxyglucose was measured during the last 10 min of the 30-min stimulation period by adding 1 mmol/L [³H]-2-deoxyglucose (0.056 MBq/mL), 7 mmol/L [¹⁴C]-mannitol (0.0167 MBq/mL) (Hartmann Analytic, Braunschweig, Germany), and 2 mmol/L pyruvate to the incubation medium. After incubation, muscles were washed in ice-cold KRB, dried on filter paper, and frozen in liquid nitrogen. Glucose uptake was assessed by the accumulation of [³H]-2-deoxyglucose in muscle by using [¹⁴C]-mannitol as an extracellular marker. Radioactivity was measured in muscle lysates using liquid scintillation counting (Ultima Gold and Tri-Carb 4910 TR; PerkinElmer) and related to the specific activity of the incubation buffer.

In situ contraction-stimulated muscle glucose uptake was determined as previously described (26,38). In short, an electrode was placed on the common peroneal nerve of one leg of fed mice that had been anesthetized by a single intraperitoneal injection of pentobarbital and xylocain (100 and 5 mg/kg, respectively). The contralateral leg served as a sham-operated resting control. Muscle glucose uptake during 10 min of electrical stimulation was determined by retro-orbital injection of [³H]-2-deoxyglucose as described above, with blood sampling and blood glucose determination at t = 0, 5, and 10 min. For measurements of muscle glucose uptake after contraction, anesthetized and muscle-contracted mice were left to recover on a heating plate (30°C) for 30 min, after which muscle glucose uptake was determined during a 10-min stimulation period as described above.

Measurements of insulin sensitivity in isolated EDL muscle after in situ contraction were performed as previously described (26), except that EDL muscles were allowed to recover in incubation chambers for 2 h. In short, EDL muscles were excised immediately after in situ contraction and suspended in heated (30°C) and continuously gassed (95% O₂ and 5% CO₂) incubation chambers containing KRB supplemented with 5 mmol/L d-glucose and 5 mmol/L mannitol. Following a 2-h recovery period, basal and submaximal (100 μU/mL) insulin-stimulated 2-deoxyglucose uptake were assessed in EDL muscles during the last 10 min of a 30-min stimulation period as described above. These experiments were performed separately in both male and female mice.

Measurements of insulin tolerance after a single bout of exercise were performed as previously described (26). In short, following treadmill running acclimatization on five consecutive days, mice were subjected to a graded maximal running test to assess the maximal running speed of each individual mouse. Blood glucose and lactate concentrations were measured immediately before and after the maximal running test in a subgroup of animals. For ITTs, mice were single-housed and fasted for 2 h (start ∼06:00 a.m.) before performing 30 min of treadmill exercise (55% of individual maximal running speed). Resting control mice were left in their cages. Immediately after exercise, mice returned to their individual cages for 1 h without access to food and were subsequently administered with a submaximal dose of insulin (0.3 units/kg) dissolved in saline (0.9%) by a single intraperitoneal injection. Blood glucose concentrations were determined at t = 0, 20, 40, 60, 90, and 120 min.

Tissues were crushed in liquid nitrogen prior to homogenization, except for soleus and EDL muscle. Tissues were homogenized in ice-cold lysis buffer as previously described (39) and subsequently rotated end over end for 1 h at 4°C to increase protein extraction. Portions of the homogenate were centrifuged at 16,000 g for 20 min at 4°C, after which the supernatant (lysate) was collected. Total protein abundance in tissue lysates and homogenates was determined in triplicates by the bicinchoninic acid method (Thermo Fisher Scientific, Waltham, MA). Equal amounts of muscle protein lysate were heated in Laemmli buffer (96°C, 10 min) before being subjected to SDS-PAGE and Western blotting analyses as previously described (39).

AMPKα2β2γ3-, α2βγ1-, and α1βγ1-associated activity was determined in muscle protein lysate by three consecutive immunoprecipitations using custom-made anti-AMPKγ3, anti-AMPKα2, and anti-AMPKα1 as previously described (40).

Anti-TBC1D4-pT649 (no. 8881), anti-Akt2 (no. 3063), anti-Akt-pS473 (no. 9271), anti-Akt-pT308 (no. 9275), anti-ACC-pS212 (no. 3661), anti-AMPKα-pT172 (no. 2531), anti-p38-MAPK (no. 9212), and anti-GAPDH (no. 2118) were all from CST (Danvers, MA). Anti-TBC1D1-pS231 (no. 07-2268) was from Millipore (Burlington, MA), anti-GLUT4 (PA1-1065) was from Thermo Fischer Scientific (Waltham, MA), and anti-AMPKα2 (sc-19131) and anti-hexokinase-II (sc-130358) were from Santa Cruz Biotechnology (Dallas, TX). Anti-TBC1D4 (ab189890) and anti-TBC1D1 (ab229504) were from Abcam (Cambridge, U.K.), and ACC protein was detected using horseradish peroxidase–conjugated streptavidin (P0397) from Agilent (Santa Clara, CA). Anti-TBC1D4-pS711 was custom made by Capra Science (41).

The statistical analyses were performed using GraphPad Prism software (GraphPad Software, San Diego, CA). Data are presented with individual values, while bar and line graphs represent means ± SEM unless stated otherwise. An unpaired two-tailed Student t test was used for two-group comparisons. Two independent variables were compared using two-way ANOVAs with or without repeated measures, followed by Sidak’s multiple comparisons test when an interaction between variables occurred. When unequal variance was observed between groups, the data were log₁₀ transformed to obtain equal variance. Statistical significance was defined as P < 0.05.

The data and resources generated and/or analyzed during the current study are available from the corresponding authors on reasonable request.

Initial investigations revealed that the TBC1D4-S711A mutation decreased mRNA levels of the long isoform of Tbc1d4 by ∼40–50% in muscle whereas the short isoform of Tbc1d4 remained unaffected in both muscle and ScWAT from homozygous TBC1D4-S711A mice (Fig. 1C and D). This led to a decrease in TBC1D4 protein content in skeletal muscle from homozygous TBC1D4-S711A mice (Fig. 1E), because skeletal muscle predominantly expresses the long isoform of Tbc1d4 (42). In addition, we found that basal and insulin-stimulated phosphorylation of TBC1D4-S711 was reduced in muscle from heterozygous TBC1D4-S711A mice and absent in muscle from homozygous TBC1D4-S711A mice (Fig. 1F and G). These data validate the TBC1D4-S711A mouse model and provide an opportunity to investigate the importance of TBC1D4-S711 for the regulation of muscle insulin sensitivity and whole-body glycemic control.

Analyses of chow-fed homozygous female TBC1D4-S711A mice revealed no phenotype regarding body weight, body composition, and food intake when compared with wild-type littermates (Fig. 2A–C). TBC1D4-S711A mice displayed normal blood glucose and plasma insulin levels in the fed and fasted states (Fig. 2D and E). HOMA-IR levels (Fig. 2F) implied that the TBC1D4-S711A mice exhibited intact whole-body insulin sensitivity. This was supported by a comparable increase in postprandial blood glucose and plasma insulin levels in the two genotypes 30 and 60 min after refeeding (Fig. 2D and E), as well as similar glucose and insulin tolerance (Fig. 2G–I). Together, these data demonstrate that the TBC1D4-S711A mutation does not affect whole-body glycemic control in resting lean female mice.

Mice with a TBC1D4-T649A mutation have reduced rates of insulin-stimulated muscle glucose uptake (19). Therefore, we decided also to investigate the effect of the TBC1D4-S711A mutation on in vivo insulin-stimulated glucose clearance. Retro-orbital intravenous injection of insulin decreased blood glucose levels and increased glucose clearance in skeletal muscle and ScWAT to a similar extent in TBC1D4-S711A and wild-type mice (Fig. 3A–D). These findings were associated with intact proximal insulin signaling in ScWAT and muscle (Akt-T308/S473) but lower phosphorylation of TBC1D4-T649 in muscle from TBC1D4-S711A mice due to decreased TBC1D4 protein levels (Fig. 3E–N). This demonstrates that phosphorylation of TBC1D4-S711 is dispensable for in vivo insulin-stimulated muscle glucose clearance during resting conditions.

To circumvent potential humoral factors that may affect muscle glucose uptake in the in vivo mouse model, we examined the ability of insulin and AICAR to increase glucose uptake in isolated soleus (oxidative) and EDL (glycolytic) muscle. Upon submaximal and maximal insulin stimulation, glucose uptake rates increased to a similar extent in soleus and EDL muscle from TBC1D4-S711A and wild-type mice (Fig. 4A and B). The effect of AICAR on glucose uptake did not differ between genotypes (Fig. 4A and B). We confirmed that insulin increased phosphorylation of Akt-T308/S473 and AICAR increased phosphorylation of TBC1D1-S231 and ACC-S212 similarly in muscle from the two genotypes (Fig. 4C–J). Insulin-stimulated phosphorylation of TBC1D4-T649 was reduced in muscle from TBC1D4-S711A mice due to decreased TBC1D4 protein levels (Fig. 4K–O and Q). Insulin and AICAR increased phosphorylation of TBC1D4-S711 in muscles from wild-type mice (Fig. 4P, R, and S).

To evaluate the potential detrimental effect of the TBC1D4-S711A mutation on whole-body glycemic control in an obesogenic environment, female TBC1D4-S711A mice and wild-type littermates were exposed to a standard HFD for 14 weeks. To our surprise, TBC1D4-S711A mice gained slightly less weight and displayed reduced body fat content when fed an HFD (Fig. 5A and B), although food intake was comparable in the two genotypes (Fig. 5C). Despite this, no differences in blood glucose and plasma insulin levels were observed between genotypes in the fed, overnight fasted, and refed states (Fig. 5D and E), nor did HOMA-IR and glucose tolerance differ (Fig. 5F–H). In agreement, insulin decreased blood glucose levels and stimulated glucose uptake in skeletal muscle and ScWAT to a similar extent in the two genotypes on an HFD (Fig. 5I–L). These findings were associated with comparable insulin-related tissue signaling (Fig. 5M–V). Collectively, our observations demonstrate that the TBC1D4-S711A mutation is not important for regulating whole-body and muscle glucose metabolism in resting obese female mice.

Recent evidence supports a role of AMPK in regulating muscle glucose uptake in the immediate period after exercise and contraction (38). Therefore, we investigated the importance of the AMPK-TBC1D4-S711 signaling axis for the regulation of muscle glucose uptake in response to contraction. Glucose uptake was not impaired during contraction or 30 min after contraction in skeletal muscle from TBC1D4-S711A mice (Fig. 6A and B). In response to contraction, glycogen levels decreased (Fig. 6C) whereas AMPK activity and downstream signaling increased to a similar extent in muscle from the two genotypes (Fig. 6D–G). Phosphorylation of TBC1D4-S711 increased by contraction in wild-type muscle (Fig. 6H and I). These findings suggest that AMPK-mediated phosphorylation of TBC1D4-S711 is not involved in regulating muscle glucose uptake in response to contraction and that the applied contraction protocol elicits a comparable metabolic stress in muscle from the two genotypes.

Because both insulin and exercise-induced AMPK activation converge on TBC1D4-S711 to promote its phosphorylation (21) and insulin-stimulated phosphorylation of TBC1D4-S711 is elevated in prior exercised and insulin-sensitized skeletal muscle (5,28,29), we tested the importance of TBC1D4-S711 for improving muscle insulin sensitivity after contraction. Two hours after contraction, insulin-independent glucose uptake was not significantly different between prior contracted and rested muscles (Fig. 7A and B). However, prior contraction increased submaximal insulin-stimulated glucose uptake in isolated muscle from wild-type mice (Fig. 7A) but failed to do so in muscle from TBC1D4-S711A mice (Fig. 7B). Accordingly, the incremental increase (delta) in insulin-stimulated glucose uptake (glucose uptake of non-insulin-stimulated muscles subtracted from glucose uptake of paired insulin-stimulated muscles) was only elevated after prior contraction in muscle from wild-type mice (Fig. 7C).

Because our findings were derived from experiments in female mice, we wanted to test whether male TBC1D4-S711A mice exhibited a similar phenotype. We observed that the ability of contraction to improve insulin-stimulated glucose uptake was disrupted in muscle from male TBC1D4-S711A mice (Fig. 7D and E). However, there was a tendency for improved muscle insulin sensitivity in male TBC1D4-S711A mice (P = 0.061) (Fig. 7F), suggesting that male TBC1D4-S711A mice retain the ability to increase muscle insulin sensitivity after contraction but less markedly than male wild-type mice.

To investigate whether the reduced abundance of TBC1D4 protein was affecting the observations in muscle from female TBC1D4-S711A mice, we tested the ability of contraction to improve muscle insulin sensitivity in heterozygous muscle-specific TBC1D4 KO female mice in which muscle TBC1D4 protein is reduced by ∼50%. Importantly, prior contraction improved muscle insulin sensitivity in these mice (Fig. 7G and H), suggesting that the inability of contraction to improve muscle insulin sensitivity in female TBC1D4-S711A mice is not due to a reduced abundance of muscle TBC1D4 protein.

GLUT4 and Hexokinase-II protein levels, as well as proximal insulin signaling measured at the level of Akt-T308, were similar in EDL muscles from wild-type and TBC1D4-S711A female mice (Fig. 7I–K). Insulin increased phosphorylation of TBC1D4-T649 in muscle from both genotypes (Fig. 7L), while phosphorylation of TBC1D4-T649 related to TBC1D4 protein was slightly increased by prior contraction in wild-type muscle only (Fig. 7M and N). Importantly, insulin-stimulated phosphorylation of TBC1D4-S711 was elevated in prior contracted and insulin-sensitized EDL muscle from wild-type mice (Fig. 7O and P), which further supports a role of TBC1D4-S711 in regulating muscle insulin sensitivity.

To address the physiological relevance of TBC1D4-S711 for the regulation of muscle insulin sensitivity, we examined in vivo whole-body insulin sensitivity in mice by an intraperitoneal injection of a submaximal insulin dose (0.3 units/kg) after a single bout of treadmill exercise. Initial findings revealed that the maximal running capacity did not differ between genotypes (Fig. 8A) nor did the blood glucose and lactate levels before and after the maximal running test (Fig. 8B and C). Prior exercise enhanced the blood glucose–lowering effect of insulin only in wild-type mice, and thus, improved whole-body insulin tolerance by ∼75% in wild-type mice (Fig. 8D–F). This provides further evidence to support that TBC1D4-S711 is a key phosphorylation site essential for enhancing muscle insulin sensitivity after exercise to improve whole-body glycemic control.

Previously, we have shown that AMPK and TBC1D4 are necessary for improving muscle insulin sensitivity after contractions and that phosphorylation of TBC1D4-S711 is enhanced in prior contracted and insulin-sensitized muscle (26,27). Using mice harboring a single TBC1D4-S711A point mutation, we now show that phosphorylation of TBC1D4-S711 is necessary for improving muscle insulin sensitivity after contractions and whole-body insulin sensitivity after a single bout of exercise.

We found that the mRNA level and protein abundance of TBC1D4 were decreased by ∼50% in skeletal muscle from TBC1D4-S711A mice. We believe that the observed decrease in muscle TBC1D4 protein is a consequence of a decrease in the TBC1D4 mRNA levels. This may be explained by the introduced gene mutations that can interfere with the ability of exonic splicing enhancers to promote efficient gene splicing (43). Gene mutations and alterations in a nucleic acid sequence have also been shown to change the secondary structure of mRNA that may create unstable mRNA and lead to premature mRNA degradation of Tbc1d4 (44). Alternatively, phosphorylation of TBC1D4-S711 may, in itself, regulate the transcription of Tbc1d4. Importantly, the decrease in muscle TBC1D4 protein did not affect muscle and whole-body glucose metabolism in chow- or HFD-fed TBC1D4-S711A mice compared with wild-type littermates. To our knowledge, observations from heterozygous TBC1D4 KO mice have not been reported, but heterozygous carriers of a common Greenlandic Inuit loss-of-function mutation in Tbc1d4 that decreases TBC1D4 protein abundance in muscle by ∼50% are not characterized by impaired glucose tolerance or increased risk of type 2 diabetes in contrast to homozygous carriers (45). Together, this suggests that the ∼50% reduction in TBC1D4 protein does not compromise muscle and whole-body glucose metabolism during insulin stimulation.

We have recently shown that AMPK and its downstream target TBC1D1 are involved in maintaining muscle glucose uptake elevated in the period after contraction and exercise (38). Because parts of TBC1D1 and TBC1D4 are in complex in skeletal muscle (46), we hypothesized that phosphorylation of TBC1D4-S711 by AMPK could also be involved in regulating muscle glucose uptake after contraction. Nevertheless, our findings did not support TBC1D4-S711 having such a role; therefore, we now propose that AMPK promotes insulin-independent muscle glucose uptake in the immediate period after exercise/contraction via TBC1D1 and it enhances muscle insulin sensitivity several hours after exercise/contraction via TBC1D4.

Improvements in muscle insulin sensitivity after a single bout of exercise in humans are associated with enhanced insulin-stimulated phosphorylation of TBC1D4 (5,28,29). Furthermore, improvements in muscle insulin sensitivity after contraction are positively associated with insulin-stimulated phosphorylation of TBC1D4-S711 in muscle from wild-type mice but not in muscle from AMPK-deficient mice (26). We have now provided evidence to support that phosphorylation of TBC1D4-S711 is necessary for improving muscle insulin sensitivity after exercise and contraction in female mice. In male TBC1D4-S711A mice, the insulin-sensitizing effect of contraction is also disrupted but not fully abolished. This is in agreement with recent findings by Zheng et al. (47), who reported that adeno-associated virus delivery of a TBC1D4 3P phosphomutant (S588A, T642A, and S704A) to skeletal muscle of male TBC1D4 KO rats reduced, but did not fully blunt, the insulin-sensitizing effect of exercise. Whether these findings also apply to female rats is currently unknown but may indicate sex-specific differences in TBC1D4-induced augmentation of muscle insulin sensitivity after exercise/contraction.

Lack of TBC1D4 protein in skeletal muscle impairs the ability of exercise and contraction to enhance muscle insulin sensitivity, while restoration of TBC1D4 protein in muscle from TBC1D4 KO animals rescues it (27,47). Although a ∼50% reduction in muscle TBC1D4 protein abundance did not affect basal/resting glucose metabolism in TBC1D4-S711A mice in the current study, such a reduction could potentially have a detrimental effect on the insulin-sensitizing effect of contraction. To illuminate this, we investigated the ability of contraction to improve insulin sensitivity in muscle from heterozygous muscle-specific TBC1D4 KO mice. Notably, prior contraction improved insulin sensitivity in muscle from these mice, demonstrating that the lack of phosphorylated TBC1D4-S711 rather than the reduced TBC1D4 protein abundance is responsible for disrupting improvements in muscle insulin sensitivity after exercise and contraction in TBC1D4-S711A mice.

Exactly how contraction/exercise-induced phosphorylation of TBC1D4-S711 relays improvements in muscle insulin sensitivity remains uncertain, but it is likely related to GLUT4 relocalization to insulin-responsive storage vesicles that augments translocation of GLUT4 to the muscle surface membrane upon subsequent insulin stimulation (11,12). We have recently demonstrated that the exercise-induced phosphorylation of TBC1D4-S711 bound to TBC1D1 is maintained in human skeletal muscle 4 h after a single bout of insulin-sensitizing exercise (46). Because observations from cell-based models support a functional interplay between TBC1D4 and TBC1D1 where phosphorylation of TBC1D4 plays a key role in translating exercise-mimicking stimuli into physical GLUT4-releasing processes that modulate insulin sensitivity for glucose transport (48), it seems reasonable to assume that a disruption of the interplay between TBC1D4 and TBC1D1 in mature skeletal muscle could add further to our understanding of how exercise improves muscle insulin sensitivity.

In conclusion, we show that phosphorylation of TBC1D4-S711 regulates muscle insulin sensitization in response to exercise and contraction. Future studies will have to clarify how phosphorylation of TBC1D4-S711 facilitates enhanced translocation of GLUT4 to the muscle surface membrane upon insulin stimulation and whether TBC1D4 also acts as a major regulator of insulin sensitivity in human skeletal muscle.

Acknowledgments. The authors would like to thank Betina Bolmgren and Irene Bech Nielsen (Department of Nutrition, Exercise and Sports, Faculty of Science, University of Copenhagen, Denmark) as well as Anja Jokipii-Utzon (Institute of Sports Medicine Copenhagen, Bispebjerg Hospital, Denmark) for their skilled technical assistance. The authors would also like to thank undergraduate student Nicoline Meyer Riisberg (Department of Nutrition, Exercise and Sports, Faculty of Science, University of Copenhagen, Denmark) for her assistance during animal experimentation.

Funding. This work was supported by grants given to J.F.P.W. from the Danish Council for Independent Research Medical Sciences (FSS: 610-00498B) and the Novo Nordisk Foundation (NNF0070370). This work was also supported by a postdoctoral research fellow grant to R.K. from the Danish Diabetes Academy, which is funded by the Novo Nordisk Foundation (NNF17SA0031406), as well as a research grant to R.K. from the European Foundation for the Study of Diabetes/Lilly Young Investigator Research Award Programme. J.T.T. was supported by the Novo Nordisk Foundation Center for Basic Metabolic Research. The Center for Basic Metabolic Research is an independent Research Center at the University of Copenhagen that is partially funded by an unrestricted donation from the Novo Nordisk Foundation (NNF18CC0034900).

Duality of Interest. J.F.P.W. has ongoing collaborations with Pfizer Inc. and Novo Nordisk Inc. unrelated to this work. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. R.K. and J.F.P.W. conceptualized and designed the experiment. R.K., N.O.E., D.F.D., N.R.A., J.B.B., A.G., J.T.T., P.S., and H.P. performed the laboratory analyses. R.K. performed the data analyses. R.K., J.M.K., N.O.E., K.K., D.F.D., and J.K.L. performed the animal experiments. R.K. and J.F.P.W. drafted the manuscript. authors interpreted, reviewed, edited, and approved the manuscript. R.K. and J.F.P.W. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.