The p21‐activated kinase 2 (PAK2), but not PAK1, regulates contraction‐stimulated skeletal muscle glucose transport

Abstract Aim Muscle contraction stimulates skeletal muscle glucose transport. Since it occurs independently of insulin, it is an important alternative pathway to increase glucose transport in insulin‐resistant states, but the intracellular signaling mechanisms are not fully understood. Muscle contraction activates group I p21‐activated kinases (PAKs) in mouse and human skeletal muscle. PAK1 and PAK2 are downstream targets of Rac1, which is a key regulator of contraction‐stimulated glucose transport. Thus, PAK1 and PAK2 could be downstream effectors of Rac1 in contraction‐stimulated glucose transport. The current study aimed to test the hypothesis that PAK1 and/or PAK2 regulate contraction‐induced glucose transport. Methods Glucose transport was measured in isolated soleus and extensor digitorum longus (EDL) mouse skeletal muscle incubated either in the presence or absence of a pharmacological inhibitor (IPA‐3) of group I PAKs or originating from whole‐body PAK1 knockout, muscle‐specific PAK2 knockout or double whole‐body PAK1 and muscle‐specific PAK2 knockout mice. Results IPA‐3 attenuated (−22%) the increase in glucose transport in response to electrically stimulated contractions in soleus and EDL muscle. PAK1 was dispensable for contraction‐stimulated glucose transport in both soleus and EDL muscle. Lack of PAK2, either alone (−13%) or in combination with PAK1 (−14%), partly reduced contraction‐stimulated glucose transport compared to control littermates in EDL, but not soleus muscle. Conclusion Contraction‐stimulated glucose transport in isolated glycolytic mouse EDL muscle is partly dependent on PAK2, but not PAK1.

Upon muscle contraction, multiple intracellular signaling pathways are activated that promote GLUT4 translocation and a subsequent increase in muscle glucose uptake. Redundant Ca 2+ -dependent signaling, metabolic stress signaling, and mechanical stress signaling are proposed to regulate distinct steps important for glucose transport in response to muscle contraction (Sylow, Kleinert, Richter, & Jensen, 2017). The group I p21-activated kinase (PAK)-1 and PAK2 are activated in response to electrical pulse stimulation in C2C12 myotubes (Hu et al., 2018;Yue et al., 2019) and muscle contraction/acute exercise in mouse and human skeletal muscle (Sylow et al., 2013). Group I PAKs (PAK1-3) are downstream targets of the Rho family GTPases Cdc42 and Rac1 (Manser, Leung, Salihuddin, Zhao, & Lim, 1994). Rac1 plays a key role in mediating GLUT4 translocation and consequently glucose uptake in response to muscle contraction and acute exercise in skeletal muscle (Sylow et al., 2013(Sylow et al., , 2016Sylow, Møller, et al., 2017). Additionally, the contraction-stimulated increase in PAK1/2 activity is blunted in skeletal muscles from muscle-specific Rac1 knockout (KO) mice (Sylow et al., 2013), suggesting a role for PAK1 and/ or PAK2 in regulating Rac1-mediated effects during muscle contraction. Whereas PAK1 previously has been proposed to regulate insulin-stimulated GLUT4 translocation in skeletal muscle (Tunduguru et al., 2014;Wang, Oh, Clapp, Chernoff, & Thurmond, 2011), we recently identified PAK2, but not PAK1, as a partial requirement for insulin-stimulated glucose transport in mouse extensor digitorum longus (EDL) muscle (Moller et al., 2019). However, the significance of the increased activity of group I PAKs in the regulation of contraction-stimulated muscle glucose transport is unknown. We hypothesized that PAK1 and PAK2 participate in the regulation of glucose transport in response to muscle contraction, due to their well-described role as Rac1 effector proteins. Our results identify PAK2, but not PAK1, as a partial requirement for contraction-stimulated glucose transport in mouse skeletal muscle.

| Animal experiments
All animal experiments complied with the European Convention for the protection of vertebrate animals used for experimental and other scientific purposes (No. 123, Strasbourg, France, 1985; EU Directive 2010/63/EU for animal experiments) and were approved by the Danish Animal Experimental Inspectorate. All mice were maintained on a 12:12-hr light-dark cycle and housed at 22°C (with allowed fluctuation of ±2°C) with nesting material. Female C57BL/6J mice (Taconic, Denmark) were used for the inhibitor incubation study. The mice received a standard rodent chow diet (Altromin no. 1324; Brogaarden) and water ad libitum. The mice were group housed.

| Mouse genotyping by PCR
An ear punch was digested overnight in 100 µl Viagen lysis buffer plus Proteinase K at 55°C followed by 45 min at 85°C. After spin at 1,000 × g for 5 min, the supernatant was diluted 10 times in TE (pH 8.0) with yellow color (50 pg/ml Quinoline Yellow). Five µl of this was used in a 25 µl real-time quantitative PCR reaction containing Quantitect SYBR Green Master Mix (Qiagen), 200 nM of each primer (Table 1), and blue dye (5 pg/ml Xylene Cyanol). The reactions were furthermore spiked (100 times less than the samples) with a heterozygote sample as a positive PCR control. The samples, including no sample controls (TE), were amplified in an MX3005P realtime PCR machine (95°C, 10 min → {95°C, 15 s → 58°C, 30 s → 63°C, 90 s} × 50 → melting curve 55°C → 95°C).
The Ct values were used to access allele presence by comparison to the no DNA controls (spike values) such that the Ct value should be at least 2 Ct below the no sample controls to indicate the presence of the allele. Amplification efficiency in the individual reactions was estimated by the sigmoid method of Liu and Saint (2002) to ensure that the Ct's could be compared within primer sets. The genotype was later verified by immunoblotting on samples from muscle tissue.

| Immunoblotting
Lysate protein concentration was determined using the bicinchoninic acid method using bovine serum albumin (BSA) standards and bicinchoninic acid assay reagents (Pierce). Immunoblotting samples were prepared in 6X sample buffer (340 mM Tris (pH 6.8), 225 mM DTT, 11% (w/v) SDS, 20% (v/v) Glycerol, and 0.05% (w/v) Bromphenol blue). Protein phosphorylation (p) and total protein expression were T A B L E 1 Primers sequences used for mouse genotyping was blocked in Tris-Buffered Saline with added Tween20 (TBST) and 2% (w/v) skim milk or 3% BSA for 15 min at room temperature, followed by incubation overnight at 4°C with a primary antibody (Table 2). Next, the membrane was incubated with horseradish peroxidase-conjugated secondary antibody (Jackson Immuno Research) for 1 hr at room temperature. Total ACC was detected without the use of antibodies. Instead, the membrane was incubated with horseradish peroxidase-conjugated streptavidin (P0397; Dako; 1:3,000, 3% BSA) at 4°C overnight. Bands were visualized using Bio-Rad ChemiDocTM MP Imaging System and enhanced chemiluminescence (ECL+; Amersham Biosciences). Actin protein expression or Coomassie Brilliant Blue staining was utilized as a control to assess total protein loading (Welinder & Ekblad, 2011) and for each sample set, a representative membrane from the immunoblotting is shown. Densitometric analysis was performed using Image LabTM Software, version 4.0 (Bio-Rad; RRID: SCR_014210). For total protein expression, each data point was presented as the average of the protein expression in the left and right muscle from the same mouse.

| Statistical analyses
Data are presented as mean ± SEM. or when applicable, mean ± SEM with individual data points shown. Statistical tests varied according to the dataset being analyzed and the specific tests used are indicated in the figure legends. Datasets were normalized by square root, log10, or inverse transformation if not normally distributed or failed equal variance test. If the null hypothesis was rejected, Tukey's post hoc test was used to evaluate significant main effects of genotype and significant interactions in ANOVAs. p < .05 was considered statistically significant. p < .1 was considered a tendency. Except for mixed-effects model analyses performed in GraphPad Prism, version 8.2.1. (GraphPad Software; RRID: SCR_002798), all statistical analyses were performed using Sigma Plot, version 13 (Systat Software Inc.; RRID: SCR_003210). Due to missing data points, differences between genotypes and the effect of electrically stimulated contractions were assessed with a mixed-effects model analysis in Figure 2h+i.

| Contraction-stimulated glucose transport is partially inhibited by pharmacological inhibition of PAK1/2
To investigate the role of group I PAKs in the regulation of contraction-stimulated glucose transport, we first analyzed 2-deoxyglucose (2DG) transport in isolated soleus and EDL muscle in the presence or absence of a pharmacological group I PAK inhibitor, IPA-3. Electrically stimulated contractions increased 2DG transport in DMSO-treated soleus (2.9-fold) and EDL (3.0-fold) muscles ( Figure 1a+b). IPA-3 partly inhibited contraction-stimulated 2DG transport in both soleus (−22%) and EDL (−22%; Figure 1a+b). The reduction in contraction-stimulated 2DG transport upon IPA-3 treatment was not associated with reduced initial force development in IPA-3-treated muscles ( Figure 1c). While phosphorylated (p)AMPKα T172 was unaffected by IPA-3 in soleus muscle (Figure 1d), contraction-stimulated pAMPKα T172 was reduced (−46%) in IPA-3-treated EDL muscle (Figure 1e). However, AMPK's downstream target pACC1/2 S79/212 was normally phosphorylated in response to electrically induced contractions in both muscles (Figure 1f+g), suggesting that the AMPK-ACC signaling pathway was largely unaffected by IPA-3 treatment. Altogether, these data suggest that pharmacological inhibition of group I PAKs partly reduces contraction-stimulated glucose transport in mouse skeletal muscles.
3.2 | Contraction-stimulated glucose transport partially relies on PAK2, but not PAK1, in mouse EDL muscle IPA-3 is a pharmacological inhibitor of group I PAKs (PAK1-3) of which PAK1 and PAK2 are detectable in skeletal muscle (Arias-Romero & Chernoff, 2008;Joseph et al., 2017;Tunduguru et al., 2014). To identify the relative role of PAK1 and PAK2 in the regulation of contraction-stimulated glucose transport, we investigated contraction-stimulated glucose transport in isolated soleus and EDL muscles from a cohort of whole-body PAK1 KO, muscle-specific PAK2 (m) KO, and double knockout mice with whole-body knockout of PAK1 and muscle-specific knockout of PAK2 (1/m2 dKO) compared to control littermates. PAK1 was not detectable at the protein level in muscles from mice with whole-body knockout of PAK1 (i.e., PAK1 KO and 1/m2 dKO mice) (Figure 2a+b, Figure S1a and b). On the contrary, muscles lacking PAK2 (i.e., PAK2 mKO and 1/m2 dKO mice) displayed only a partial reduction in PAK2 protein expression (Figure 2a+b, Figure S1c and d). As the knockout of PAK2 is muscle specific, other cell types within skeletal muscle tissue contribute to the signal obtained in the PAK2 immunoblots. Moreover, the slightly upregulated PAK2 protein expression in EDL muscle lacking PAK1 (Figure 2b, Figure S1d) could be ascribed to a compensatory upregulation of PAK2 in muscle tissue (PAK1 KO mice) and nonmuscle tissue (PAK1 KO and 1/m2 dKO mice). The whole-body metabolic characteristics of this cohort of mice have previously been described (Moller et al., 2019). In soleus muscle, contraction-stimulated glucose transport was unaffected by the lack of PAK1, PAK2, or both PAKs combined (Figure 2c). In contrast, in EDL lack of PAK2, either alone or in combination with PAK1 KO, partially reduced contraction-stimulated glucose transport compared to PAK1 KO mice (PAK2 mKO: −21%; 1/m2 dKO: −22%) and control littermates (PAK2 mKO: −13%; 1/m2 dKO: −14%; Figure 2d). Lack of PAK2 decreased initial force development in soleus compared to PAK1 KO mice (PAK2 mKO: −30%; 1/m2 dKO: −38%) and control littermates (1/m2 dKO: −27%; Figure 2e). In EDL, lack of PAK1 (+40%) or PAK2 (+38%) alone increased initial force development, while combined knockout of PAK1 and PAK2 decreased initial force development compared to PAK1 KO mice (−31%) and PAK2 mKO mice (−30%; Figure 2f). Due to the technicalities in the measurement of 2DG transport, it was not possible to obtain accurate measures of the total soleus and EDL muscle mass. Thus, we are unable to normalize muscle tension to mass. However, previous investigations have reported atrophy in several distinct muscles, including soleus and EDL, from 1/m2 dKO mice (Joseph et al., 2017 and we also observed reduced muscle mass in tibialis anterior (TA) 1/m2 dKO muscles (−13%; Figure 2g). However, although the reduction in initial force development in 1/m2 dKO muscle potentially could be ascribed to muscle wasting, the decrease in force development over time was similar between all four genotypes in both soleus and EDL muscle (Figure 2h+i). Taken together, similar to insulin-stimulated glucose uptake (Moller et al., 2019), PAK1 is dispensable for contraction-stimulated glucose transport, while contraction-stimulated glucose transport partially relies on PAK2 in glycolytic EDL muscle.

| DISCUSSION
The present study is, to our knowledge, the first to investigate the requirement of PAK1 and PAK2 in contraction-stimulated glucose transport in mouse skeletal muscle. By undertaking a systematic investigation, including pharmacological as well as genetic interventions, we show that contraction-stimulated glucose transport in isolated skeletal muscle partially requires PAK2, but not PAK1, in glycolytic EDL muscle.
In the current study, IPA-3 attenuated the increase in muscle glucose transport in response to the electrically stimulated contractions in both soleus and EDL muscle, whereas genetically targeted knockout revealed an effect of PAK2 in glycolytic EDL only. It is not unusual that pharmacological inhibition and genetically targeted mutations produce different phenotypes (Knight & Shokat, 2007). This likely means that either the effect of the IPA-3 on glucose transport in soleus is unspecific or, alternatively, that the absent effect of genetic ablation of PAK1 and/or PAK2 in soleus is due to compensation by other mechanisms. It is important to stress that any possible compensatory mechanisms cannot be via redundancy with PAK3, as even in 1/m2 dKO muscle, PAK3 cannot be detected at the protein level (Joseph et al., 2017). In contrast, a kinase screen of 214 full-length human kinases revealed that 10 µM IPA-3 significantly inhibited nine kinases (Deacon et al., 2008). Among these kinases, Akt2 and Glycogen Synthase Kinase (GSK)-3α/β were identified. GSK-3 is indirectly involved in glucose uptake via glycogen synthase regulation and glycogen deposition (Embi, Rylatt, & Cohen, 1980), whereas Akt2 is an established regulator of insulin-stimulated glucose uptake in skeletal muscle (Cho et al., 2001;Garofalo et al., 2003;McCurdy & Cartee, 2005). However, despite the reported unspecific inhibition of Akt2 (Deacon et al., 2008), we and others have reported that insulin-stimulated Akt phosphorylation is not significantly affected by IPA-3 in muscle cells (Tunduguru et al., 2014) and mature mouse skeletal muscle (Moller et al., 2019). In contrast, due to its chemical structure, IPA-3 has been proposed likely to alter the redox potential of cells because of the continuous reduction of IPA-3 (Rudolph, Crawford, Hoeflich, & Chernoff, 2013), potentially causing group I PAKindependent effects of IPA-3. Interestingly, a recent analysis of the crosstalk between oxidation and protein phosphorylation in adipocytes suggested that oxidation of key regulatory kinases, including AMPK, influences the fidelity of the kinase (Su et al., 2019), while, in cardiomyocytes, the activity of AMPK has been suggested to be negatively regulated by oxidation (Shao et al., 2014). Thus, changed redox status due to the continuous reduction of IPA-3 could be a possible explanation for the decrease in pAMPKα T172 phosphorylation observed in IPA-3-treated EDL muscle.
The limited role of group I PAKs in contraction-induced glucose transport is in accordance with our recent finding that group I PAKs were largely dispensable for insulin-stimulated glucose transport in isolated mouse skeletal muscle with only a modest reduction in EDL muscles lacking PAK2 (Moller et al., 2019). Thus, group I PAKs are not major essential components in the regulation of muscle glucose transport. Based on recent emerging evidence, the role for group I PAKs in skeletal muscle seems instead to be related to myogenesis and muscle mass regulation (Joseph et al., 2017. Additionally, in embryonic day 18.5 diaphragm, combined genetic ablation of PAK1 and PAK2 was associated with reduced acetylcholine receptor clustering at the neuromuscular junction, (Joseph et al., 2017) suggesting defects in the neuromuscular synapses.
This relatively modest requirement of group I PAKs in contraction-induced muscle glucose transport is in contrast to the marked glucoregulatory role of Rac1 (Sylow et al., 2013(Sylow et al., , 2016Sylow, Møller, et al., 2017), the F I G U R E 1 Contraction-stimulated glucose transport is partially inhibited by pharmacological inhibition of PAK1/2. (a and b) Contractionstimulated (2 s/15 s, 100 Hz) 2-deoxyglucose (2DG) transport in isolated soleus (a) and extensor digitorum longus (EDL; b) muscle ± 40 µM IPA-3 or a corresponding amount of DMSO (0.11%). Isolated muscles were preincubated for 45 min followed by 15 min of electrically stimulated contractions with 2DG transport measured for the final 10 min of stimulation. Data were evaluated with a two-way repeated-measures (RM) ANOVA. (c) Initial force development during electrically stimulated contractions. Data were evaluated with a Student's t test. (d-g) Quantification of phosphorylated (p)AMPKα T172 and pACC1/2 S79/212 in contraction-stimulated soleus (d and f) and EDL (e and g) muscles. Data were evaluated with a two-way RM ANOVA. Some of the data points were excluded due to the quality of the immunoblot, and the number of determinations was n = 5/6 (DMSO/IPA-3) for pACC1/2 S79/212 in soleus muscle. (h-i) Representative blots showing pAMPKα T172 and pACC1/2 S79/212 and actin protein expression as a loading control in soleus (h) and EDL (i) muscles. Main effects are indicated in the panels. Interactions in two-way RM ANOVA were evaluated by Tukey's post hoc test: Contraction versus basal **/*** (p < .01/.001); IPA-3 vs. DMSO #/## (p < .05/.01). Unless stated previously in the figure legend, the number of determinations in each group: Soleus, n = 8/9 (DMSO/IPA-3); EDL, n = 8/9. Data are presented as mean ± SEM with individual data points shown. Paired data points are connected with a straight line. A.U., arbitrary units upstream regulator of group I PAKs. Rac1 is an essential component in the activation of the reactive oxygen-producing NADPH oxidase (NOX)-2 complex (Abo et al., 1991;Bedard & Krause, 2007). Recently, it was reported that NOX2 is required for exercise-stimulated glucose uptake (Henríquez-Olguin et al., 2019). Moreover, it was shown that exercise-induced NOX2 activation was completely abrogated in TA from muscle-specific Rac1 KO mice (Henríquez-Olguin et al., 2019), suggesting that Rac1 mainly regulates muscle glucose uptake through activation of NOX2 in response to exercise. Alternatively, the Ral family GTPase, RalA could signal downstream of Rac1. Overexpression of a constitutively activated Rac1 mutant activated RalA in L6 myotubes (Nozaki, Ueda, Takenaka, Kataoka, & Satoh, 2012). Additionally, GLUT4 translocation induced by a constitutively active Rac1 mutant was attenuated in L6-GLUT4myc myoblasts upon RalA knockdown (Nozaki et al., 2012). The RalA GTPase-activating protein GARNL1 is phosphorylated in response to in situ contraction of mouse skeletal muscle (Chen et al., 2014), but so far no linkage between Rac1 and RalA has been reported in relation to contraction-stimulated glucose transport.

ACKNOWLEDGMENTS
Methodology; Writing -Original Draft; Writing -Review & Editing; Supervision; Project administration; Funding acquisition. EAR is the guarantor of this work and takes responsibility for the integrity of the data and the accuracy of the data analysis.