Cancer causes dysfunctional insulin signaling and glucose transport in a muscle‐type‐specific manner

Metabolic dysfunction and insulin resistance are emerging as hallmarks of cancer and cachexia, and impair cancer prognosis. Yet, the molecular mechanisms underlying impaired metabolic regulation are not fully understood. To elucidate the mechanisms behind cancer‐induced insulin resistance in muscle, we isolated extensor digitorum longus (EDL) and soleus muscles from Lewis Lung Carcinoma tumor‐bearing mice. Three weeks after tumor inoculation, muscles were isolated and stimulated with or without a submaximal dose of insulin (1.5 nM). Glucose transport was measured using 2‐[3H]Deoxy‐Glucose and intramyocellular signaling was investigated using immunoblotting. In soleus muscles from tumor‐bearing mice, insulin‐stimulated glucose transport was abrogated concomitantly with abolished insulin‐induced TBC1D4 and GSK3 phosphorylation. In EDL, glucose transport and TBC1D4 phosphorylation were not impaired in muscles from tumor‐bearing mice, while AMPK signaling was elevated. Anabolic insulin signaling via phosphorylation of the mTORC1 targets, p70S6K thr389, and ribosomal‐S6 ser235, were decreased by cancer in soleus muscle while increased or unaffected in EDL. In contrast, the mTOR substrate, pULK1 ser757, was reduced in both soleus and EDL by cancer. Hence, cancer causes considerable changes in skeletal muscle insulin signaling that is dependent on muscle‐type, which could contribute to metabolic dysregulation in cancer. Thus, the skeletal muscle could be a target for managing metabolic dysfunction in cancer.


| INTRODUCTION
Within the last decades, it has become evident that cancer causes severe systemic alterations of the host. While the unwanted loss of skeletal muscle and fat mass, known as cachexia, 1 is well-described, a lesser described burden of many cancers is the severe metabolic dysregulation. Evidently, several cancers, and in particular cachexiainducing cancers, are associated with poor metabolic regulation, including insulin resistance in both pre-clinical models [2][3][4] and human patients. [5][6][7][8][9] While the underlying mechanisms are still poorly defined, they are crucial to delineate, as dysregulated metabolism is associated with cancer incidence, poor cancer prognosis, and increased recurrence rates. [10][11][12][13][14][15] Skeletal muscle insulin resistance and dysregulated metabolism are detrimental to whole-body glucose homeostasis, as skeletal muscle is responsible for the majority of insulin-stimulated glucose disposal. 16 We recently showed, that cancer causes severe insulin resistance in pre-cachectic tumor-bearing mice 3 on several parameters, including reduced skeletal muscle and white adipose tissue glucose uptake and abrogated insulin-stimulated microvascular perfusion. 3 Yet, the muscle-specific contributions and molecular defects were not identified in that study. In addition, it is unknown whether different muscle types, Type I fiber-or Type II fiber-dominated muscles, are affected by cancer to a similar degree with regards to insulin resistance toward glucose uptake and anabolism.
To elucidate the muscle-intrinsic mechanisms that contribute to skeletal muscle insulin resistance in cancer, we here conducted a detailed investigation of glucose uptake and intramyocellular signaling in response to insulin in isolated oxidative (Type I fiber-dominated) and glycolytic (Type II fiber-dominated) muscles from tumor-bearing mice. It was hypothesized that muscles isolated from tumor-bearing mice would display altered insulin signaling leading to decreased glucose transport and anabolism.

| Animals and ethics
A total of 28 C57Bl/6J (Taconic, Lille Skensved, DK) mice, 12 weeks old, female, were group-housed at ambient temperature (21-23°C) with nesting materials. The mice were held on a 12 h:12 h light-dark cycle with access to a standard rodent chow diet (Altromin no. 1324, Brogaarden, DK) and water ad libitum. All experiments were approved by the Danish Animal Experimental Inspectorate (Licence: 2016-15-0201-01043). The sample size was decided from previous work with the experimental incubation setup. The experimental unit is a given muscle from a single animal.

| Lewis lung carcinoma
Lewis Lung Carcinoma (LLC) cancer was induced as previously described. 3 LLC cells (ATCC® CRL1642™) were cultured in DMEM, high glucose (Gibco, #41966-029, USA) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich, #F0804, USA), 1% penicillin-streptomycin (ThermoFisher Scientific, #15140122, USA) (5% CO 2 , 37°C). Prior to inoculation into mice, LLC cells were trypsinized and washed twice with PBS. LLC cells were suspended in PBS at a final concentration of 2.5 × 10 6 cells/ml. All mice were shaved on the flank two days prior to the inoculation and randomized into two groups with similar average body weight. The mice were subcutaneously injected with PBS with or without 2.5 × 10 5 LLC cells into the right flank. The experiments were carried out 19 and 21 days after cancer cell inoculation. Mice developing ulcerations (human endpoint) were sacrificed by cervical dislocation. The mice were randomly divided into control mice (n = 12) and tumor-bearing mice (n = 16). Mice with a tumor <0.5 g were excluded. Three animals were excluded due to the size of the tumor, leaving n = 13 mice in the tumor-bearing group.

| Ex vivo muscle incubations
On the day of experimentation, fed mice were anesthetized by intraperitoneal injection of pentobarbital/ lidocain (6 mg of pentobarbital sodium and 0.6 mg of lidocain/100 g of body weight) after which soleus and EDL muscles were tied with non-absorbable 4-0 silk suture loops (Look SP116, Surgical Specialities Corporation) at both ends and suspended between adjustable hooks at resting length (1-2 mN tension) in ex vivo incubation chambers (Multi Myograph system, Danish Myo-Technology) at 30°C in continuously 95% O 2 /5% CO 2 -bubbled Krebs-Ringer-Henseleit (KRH) buffer (118.5 mM NaCl, 24.7 mM NaHCO3, 4.74 mM KCl, 1.18 mM MgSO 4 ·7H 2 O, 1.18 mM KH2PO4, 2.5 mM CaCl 2 ·2H 2 O) supplemented with 8 mM mannitol and 2 mM pyruvate (KRH medium). The experimental groups were randomized between chambers. Muscles (left vs. right side of the body) from the tumor-bearing mice were randomized into saline and insulin conditions to minimize the potential effect caused by having the tumor on one side of the body of the mouse. The tumors and spleens were also dissected at this stage, rinsed, and snap-frozen in liquid nitrogen, before the mice were sacrificed by cervical dislocation. After dissection, the muscles were first allowed 15 min of recovery in fresh KRH buffer and then incubated for 10 min in KRH with or without 1.5 nM of insulin (sub-maximal dose). Next, the medium was changed to one containing radioactively labeled 2-[3H] deoxyglucose (2-DG; 0.30 μCi/ml in 1 mM non-radiolabeled 2-DG) and [ 14 C] labeled mannitol (0.28 μCi/ml in 8 mM non-radiolabeled mannitol) and 10 min of tracer labeling was performed. For the insulin-stimulated group, the same insulin concentration was maintained in the tracer medium. Finally, the muscles were harvested, rinsed in ice-cold KRH medium, dabbed dry on paper, and snap-frozen in liquid nitrogen until further analyses.

| Immunoblotting and glucose transport measurements
The frozen soleus and EDL muscles were trimmed free of connective tissue and sutures and weighed. Muscles were homogenized 1 min at 30 Hz using a TissueLyser II bead mill (Qiagen, USA) in 300 μl ice-cold homogenization buffer, pH 7.5 (10% glycerol, 1% NP-40, 20 mM sodium pyrophosphate, 150 mM NaCl, 50 mM HEPES (pH 7.5), 20 mM βglycerophosphate, 10 mM NaF, 2 mM phenylmethylsulfonyl fluoride, 1 mM EDTA (pH 8.0), 1 mM EGTA (pH 8.0), 2 mM Na3VO4, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 3 mM benzamidine). After the homogenization, the samples were rotated end-over-end for 30 min at 4°C, before being subjected to centrifugation (9500 RCF) for 20 min at 4°C. The lysates were then collected. Lysate protein concentrations were measured using the bicinchoninic acid method with bovine serum albumin (BSA) as a standard. A fraction (50 μl) of the lysate was dissolved in 2 ml of βscintillation liquid (Ultima Gold, Perkin Elmer) for measurement of 2-DG transport using [ 14 C] mannitol to estimate extracellular space using βscintillation counting. The 2-DG transport was related to the protein concentration of the lysate. The measurements of 2-DG transport using the βscintillation counter were performed blinded.
The remaining lysate was used for standard immunoblotting of total proteins and phosphorylation levels of relevant proteins. Subsequently, polyvinylidene difluoride (PVDF) membranes (Immobilon Transfer Membrane; Millipore) were blocked in Tris-buffered saline (TBS)-Tween 20 (TBST) containing 2% skim milk or 3% bovine serum albumin (BSA) for 5 min at room temperature. Membranes were incubated with primary antibodies (Table 1) overnight at 4°C, followed by incubation with HRP-conjugated secondary antibody for 45 min at room temperature. Coomassie brilliant blue staining was used as a loading control. 17 The same coomassie brilliant blue staining is presented, when the same four samples are presented for several proteins investigated using the same membrane (see Figures 3 and 4). To ensure quantification within the linear range for each antibody probed, standard curves were made for total proteins, basal and insulin-stimulated conditions. Stripping of the primary antibody and reprobing the PVDF membrane were performed using a ß-mercaptoethanol-based stripping buffer. The membranes were first washed with TBST, then incubated with the stripping buffer for 45-60 min at 50°C, then washed in TBST at room temperature for 10-15 min, before being blocked with TBST containing 2% skim milk or 3% bovine serum albumin (BSA) for 5 min at room temperature. The membranes were checked for a signal using the HRP-conjugated secondary antibody before the new primary antibody was added for incubation. Due to limited material, some sample pairs are not present for pPRAS40 thr246/PRAS40 and pULK ser757. This is indicated in the figure legends. Bands were visualized using the Bio-Rad ChemiDoc MP Imaging System and enhanced chemiluminescence (ECL+; Amersham Biosciences). Bands were quantified using the Bio-Rad Image Lab software 6.0.1.

| Statistics
All statistics were performed using GraphPad Prism, 8.0 (GraphPad Software, La Jolla, CA, USA). Statistical testing was performed using Student's t-test and twoway repeated-measures ANOVA (the two EDL muscles and the two soleus muscles from the same mouse were treated as pairs comparing basal vs. insulin stimulation) as applicable. The main effects and interactions are presented in the figures when significant. For post hoc analyses, Sidak's multiple comparisons test was performed. The Pearson correlation matrixes ( Figure 5) were performed with ©OriginLab (Northampton, Massachusetts, USA). The significance level for all analyses was set at α < .05.

| Data presentation and graphics
All graphs were created using GraphPad Prism, 9.0 (GraphPad Software, La Jolla, CA, USA). All figures were created using Inkscape (Inkscape.org). Illustrations were created using ©BioRender.com.

| Cancer leads to a minor reduction in GLUT4 protein content in soleus muscle
At day 19-21 post tumor inoculation, soleus and EDL muscles were isolated, incubated, and stimulated with or without a submaximal concentration of insulin ( Figure 1A). On the experimental day, the average tumor size was ~2.0 g ( Figure 1B) and body mass tended to be lower (p = .0867) in tumor-bearing mice ( Figure 1C). Spleen weight was increased (+145%, Figure 1D), indicative of pre-cachexia and elevated inflammation in tumor-bearing mice compared to controls.
We first investigated key proteins related to glucose transport and mitochondrial proteins, namely glucose transporter 4 (GLUT4), hexokinase II (HK II), glycogen synthase (GS), pyruvate dehydrogenase (PDH), subunits of the electron transport chain (ETC), citrate synthase (CS), and long-chain fatty acid transport protein 4 (FATP4). Cancer lead to a minor reduction in protein content of GLUT4 (−9%) and complex 4 of the ETC (−13%) in soleus muscle of tumor-bearing mice compared to control mice. No effects of cancer were observed on the   Figure 1I. Collectively, no major changes were observed for key proteins related to glucose handling.

| Cancer selectively causes insulin resistance in oxidative soleus muscle
We next investigated the glucose transport during submaximal (1.5 nM) insulin stimulation. As expected, insulin increased glucose transport in both soleus (+115%, Figure 2A/B) and EDL (+55%, Figure 2C/D) muscles from non-tumor-bearing control mice. Remarkably, this response was abrogated in the oxidative soleus muscle of tumor-bearing mice (Figure 2A/B). This effect was muscletype specific, as insulin increased glucose transport by 70% in the glycolytic EDL muscle from tumor-bearing mice with no effect of cancer ( Figure 2C/D). These data demonstrate that cancer affects muscles differently depending on muscle-type; oxidative or glycolytic. We subsequently investigated insulin signaling pathways ( Figure 2E), in order to determine the molecular underpinnings of the different responses to cancer in muscle.

| Cancer inhibits insulin-stimulated TBC1D4 and GSK3 phosphorylation
Proximal insulin signaling via phosphorylation (p) of Akt threonine(thr)308 ( Figure 3A) and pAkt serine(ser)473 ( Figure 3B) was similarly increased by insulin in control and tumor-bearing mice. The Rab GTPase activating protein TBC1D4, downstream of Akt, is inactivated by phosphorylation, which is necessary for translocation of GLUT4 to the plasma membrane. 18 As TBC1D4 has multiple insulin-sensitive phosphorylation sites, we measured if any alteration in these phospho-sites could explain the lack of effect of insulin on glucose uptake in the soleus muscle. More specifically, we investigated pTBC1D4 ser318, ser588, and thr642 (in mice; ser324, ser595, thr649), which are all phosphorylated during insulin stimulation, 19,20 and are direct targets of Akt, but also other kinases. 21 In soleus muscle of control mice, insulin led to a 35% increased TBC1D4 phosphorylation at ser318 (Figure 3C), 60% increased at ser588 (Figure 3D), and 150% increased at thr642 ( Figure 3E). In contrast, none of these phosphorylations were increased during insulin stimulation in soleus muscle from tumor-bearing mice ( Figure 3C-E). In addition, basal pTBC1D4 at ser588 ( Figure 3D) and thr642 (p = .086, Figure 3E) were increased or trended toward an increase, respectively, in the soleus of tumor-bearing mice compared to control mice. This was in contrast to EDL, where insulin increased the phosphorylation of all the above-mentioned phospho-sites independent of cancer ( Figure 3C-E). The post hoc test demonstrated that the insulin effect on TBC1D4 ser588 and thr642 was mainly driven by the increased phosphorylation of TBC1D4 during insulin stimulation in the EDL muscles from the tumor-bearing mice.
Thus, these data show that in soleus muscle, cancer impairs insulin signal transduction to TBC1D4 on several phosphorylation sites, which could explain the reduced insulin-stimulated glucose uptake observed in soleus muscle of tumor-bearing mice. In contrast, no impairment of TBC1D4 phosphorylation was observed in EDL muscles from tumor-bearing mice that did not display any alterations in glucose uptake compared to control mice.
Thus, these data show that cancer selectively impairs insulin signaling in some (TBC1D4 and GSK3), but not all (PRAS40), Akt downstream targets depending on the muscle. Representative western blots are shown in Figure 3I.

| Cancer promotes AMPK activation in EDL, but not soleus, muscle
AMP-activated protein kinase (AMPK) is a metabolic stress-sensor in muscle 29 that is proposed to be involved in glucose uptake in response to exercise. [30][31][32][33] AMPK also provides input to insulin signaling and is required for the increase in insulin sensitivity after muscle contraction. 34,35 AMPK phosphorylates TBC1D4 on ser588, 21 which was upregulated in EDL muscles from tumorbearing mice ( Figure 3D) and we, therefore, investigated AMPK signaling.
Phosphorylations of AMPK thr172 ( Figure 4A) and pACC ser212 (a direct AMPK substrate) ( Figure 4B) were similar between control and tumor-bearing mice in soleus muscle. In contrast, both AMPK and ACC phosphorylations were upregulated in the EDL muscle of tumor-bearing mice ( Figure 4A,B), consistent with previous reports. 36,37 Total AMPK α2 and ACC1/2 ( Figure 4C) protein contents were not affected by cancer. Representative western blots are shown in Figure 4D. Thus, elevated AMPK activation might be involved in the protection from cancer-induced insulin resistance in the EDL muscle.

| Cancer altered mTORC1 signaling in both soleus and EDL muscle
The mammalian target of rapamycin complex 1 (mTORC1) is a central regulator of cell size and protein synthesis. 38 Cancer can lead to decreased protein synthesis in human muscle, [39][40][41] and reduced/abrogated mTORC1 signaling has been observed in various pre-clinical cancer mouse models. 36,37,[42][43][44][45][46][47] Thus, we next determined the effect of cancer on insulin-stimulated anabolic signaling in muscle.
Insulin-stimulated phosphorylation of mTOR at ser2448, a reported insulin-sensitive site, 48 was not affected by either sub-maximal insulin or cancer in soleus and EDL muscle ( Figure 4E). Despite no increase in mTOR phosphorylation, insulin increased phosphorylation of the downstream target of mTORC1, p70S6K thr389, in soleus (+208%), and trended to in EDL (+134%, p = .066) of control animals ( Figure 4F). This effect of insulin on p-p70S6K thr389 was completely abrogated in soleus muscle of tumor-bearing mice compared to control mice ( Figure 4F). In contrast, p-p70S6K thr389 was augmented in tumorbearing mice during insulin stimulation compared to control mice in EDL muscle (+60%, Figure 4F). p70S6K activity leads to phosphorylation of ribosomal protein S6 (rS6) at ser235/236. 49 In soleus muscle, insulin only increased the phosphorylation in control animals, not tumor-bearing mice (+45%, Figure 4G), as seen for p-p70S6K thr389. In EDL muscle, insulin caused the main effect of increased p-rS6 ser235 with no effect of cancer ( Figure 4G). ULK1 is another downstream target of mTORC1 and phosphorylation of ULK1 at ser757 inhibits autophagy 50 ( Figure 2E). Interestingly, phosphorylation of ULK1 at ser757 was abrogated in both soleus and EDL muscle of tumor-bearing mice, where this site increased in both muscles during insulin stimulation in control mice (soleus: +75%, EDL: +52%) ( Figure 4H). Thus, insulin increases the phosphorylation of ULK in control, but seemingly not in tumor-bearing mice. Total mTOR, p70S6K, rS6, and ULK1 ( Figure 4I) protein contents were unaffected by cancer. Representative western blots are shown in Figure 4J. Taken together, these results suggest that the observed cancer-induced impairment of glucose transport in soleus muscle also manifested as anabolic resistance, indicated by disrupted mTORC1 downstream signaling during insulin stimulation.

| Positive correlations between
insulin-stimulated signaling proteins were selectively lost in response to cancer in soleus but not EDL muscle Because we detected variability in the effect of the tumor on the induction of insulin resistance, we next asked if tumor mass correlated with glucose uptake or intracellular insulin signaling. Tumor mass correlated with neither insulin-stimulated glucose uptake nor with intracellular insulin signaling (phosphorylations) in soleus muscle ( Figure 5, highlighted in the blue rectangle).
In EDL muscle, glucose uptake also did not correlate with the tumor mass. However, several phosphorylation sites correlated positively with tumor mass in EDL muscle, e.g., pTBC1D4 ser318 and p-p70S6K thr389 ( Figure 5, highlighted in the blue rectangle), showing that the effect of cancer is clearly different in the soleus and EDL muscle.
Expectedly, there was a positive correlation in control mice between Akt phosphorylation and phosphorylations of Akt's targets in both soleus and EDL muscle ( Figure 5, highlighted with a red triangle). Cancer markedly diminished these correlations in soleus muscle, but not in EDL, substantiating the distinct effects of cancer in the soleus vs. EDL. Thus, in the insulin-resistant soleus muscle of tumor-bearing mice, there was a clear disconnect between Akt and some, but not all, downstream substrates ( Figure 5).

| DISCUSSION
Here, we present evidence of selective insulin resistance within different muscle types in response to cancer in mice ( Figure 6). A primary finding was that cancer prevented insulin-stimulated glucose transport in oxidative soleus muscle, but not in glycolytic EDL muscle. Secondly, this selective insulin resistance was associated with an inability for insulin to elicit multi-site phosphorylation of the Rab GTPase activating protein TBC1D4 and of GSK3, despite normal stimulation of Akt. Third, we found that insulin-stimulated mTORC1 signaling, specifically p70S6K-S6-and ULK1 phosphorylation, was reduced by cancer. Collectively, these data show that cancer selectively rewires oxidative soleus muscle to cause severe insulin resistance, which could lead to the metabolic dysregulation observed in cancer.
Our discovery that cancer abrogates insulin-stimulated glucose transport in soleus muscle expands on other studies that have reported reduced blood-glucose-lowering effect of insulin in vivo of tumor-bearing rodents 2-4 and patients with cancer. [5][6][7][8][9]51 Such findings are clinically relevant because metabolic disturbances are associated with cancer incidence, poor cancer prognosis, and increased recurrence rates. [10][11][12][13][14][15] Whole-body insulin resistance, measured by hyperinsulinemic-euglycemic clamp has been reported in cancers such as gastrointestinal, 6,8,51 colorectal, 5,8,51 lung, 8,9,51 and pancreatic cancer. 7 Based on our present results, as well as a recent study, 3 whole-body insulin resistance and glucose intolerance in many cancers are likely due to impaired skeletal muscle glucose uptake.
The results of the current investigation would suggest that distorted insulin signaling in muscle leads to insulin resistance specifically in oxidative muscles. In agreement with this observation, proteomic analyses of human 52 and rodent 53,54 skeletal muscle show that proteins involved in oxidative metabolism are highly altered in cancer cachexia. Human skeletal muscles are highly mixed in fiber type composition containing an average of ~53% oxidative type I fibers ranging from 40% to 76% depending on the muscle group. 55 The translatability of our findings into human patients still remains to be documented but should oxidative fibers be more prone to cancer-induced insulin, this would greatly impact whole-body glucose metabolism in patients.
A second important finding was that cancerassociated insulin resistance in soleus was accompanied by dysregulation on multiple phosphorylation sites on TBC1D4 (ser318, ser588, and thr642), of which thr642 previously has been shown to be important for insulinstimulated glucose uptake in skeletal muscle. 18,56 Interestingly, tumor-bearing mice displayed normal phosphorylation of Akt, which phosphorylates TBC1D4 at thr642. Thus, there seems to be a disassociation between Akt and TBC1D4 of this signaling pathway.
Because Akt-mediated PRAS40 phosphorylation was not reduced, not all signal transduction from Akt was impaired by cancer. As also evidenced by the correlations matrix, only certain signaling events related to TBC1D4 phosphorylation in oxidative soleus muscle of tumorbearing mice were impaired. Selective insulin resistance for some, but not all Akt substrates, has previously been described for insulin resistance associated with diabetes F I G U R E 6 Schematic illustration of the findings presented in the current study or obesity. 57,58 Our findings suggest that cancer also elicits selective insulin resistance in a high muscle-type-specific manner. The causes of selective insulin resistance remain to be identified but could be due to altered activity and/ or expression of phosphatases. The muscle-type selectivity could be due to several factors, including a higher TBC1D4 protein expression in soleus muscle compared to EDL muscle in mice. 59 Thus, the expression of TBC1D4 in itself may affect the impact of cancer on mouse muscle. Yet in humans, TBC1D4 is expressed to a greater extent in Type II fibers. 60,61 Thus, the effect of cancer on insulinstimulated TBC1D4 phosphorylation clearly needs further exploration in human skeletal muscle.
Similar to our findings, phosphorylation at several sites on TBC1D4, including ser318, ser588, and ser751, is impaired during insulin stimulation in muscles from patients with T2D, 62 suggesting that reduced TBC1D4 signaling is a common trait in insulin-resistant skeletal muscle. Likewise, T2D has been associated with slightly lower skeletal muscle expression of GLUT4 protein 63,64 (not a consistent finding 62 ), which was also observed in soleus muscle of tumor-bearing mice. Reduced GLUT4 protein content aligns with TBC1D4 dysregulation as lack of TBC1D4 or loss-of-function mutants result in reduced GLUT4 content in mouse 65 and human skeletal muscle. 66 GLUT1, which mediates basal glucose uptake, 67 was not measured in our study, but could possibly also contribute to insulin resistance. 68 The insulin-sensitive EDL muscle displayed normal insulin-induced TBC1D4 phosphorylation, and cancer had no effects on GLUT4 protein content. In fact, TBC1D4 ser588 phosphorylation was elevated in EDL muscles of tumor-bearing mice. AMPK is a kinase for TBC1D4 including at the ser588 site, 21 and we speculate that the elevated AMPK signaling in EDL muscles from tumor-bearing mice may preserve insulin sensitivity. Notably, AMPK is a positive regulator of insulin sensitivity via TBC1D4 after muscle contractions 34,35 and AMPK seems to be required for normal insulin-induced signaling of the TBC1D4 paralogue and Rab GTPase activating protein, TBC1D1, in mouse muscle. 69 Our findings thus identify an intriguing link between AMPK and insulin sensitivity in the context of cancer that should be explored in future studies.
A third major finding was that insulin-stimulated p-p70S6K thr389, p-rS6 ser235/236, and pULK ser757 were inhibited in soleus muscle of tumor-bearing mice, indicative of reduced mTORC1 signaling and anabolic resistance. Signaling of mTORC1 in skeletal muscle was previously reported to be lower in cancer cachexia at baseline, 36,[42][43][44] during muscle contraction, 43,45 and after an intraperitoneal glucose injection. 37 Yet, other studies showed unchanged or increased mTORC1 signaling in cachectic rodent models 70 and humans. 71 Our study shows that altered mTORC1 activity also extends to insulin-stimulated mTORC1 signaling and suggests that cancer-associated insulin resistance expands to the level of anabolism. These findings support the theory that cancer leads to muscle insulin resistance, [72][73][74] which in turn accelerates muscle loss in cancer cachexia. While atrophy of both oxidative and glycolytic muscle fibers was reported in rodent and human cancer cachexia, 75 some evidence suggests that the glycolytic type II fibers may be more prone to muscle mass loss. 76,77 Whether the selective anabolic insulin resistance in the soleus identified in this study contributes to the development of cancer cachexia, as well as the relevance for humans, are exciting topics for future investigations.
Some limitations to this study should be considered when interpreting the data. Our study was only conducted in female mice. In humans, males with cancer have a greater prevalence of muscle mass loss when compared to female patients with cancer. 78,79 Thus, cancer may have a different molecular imprint on the muscle in males compared to females. Noteworthy, cancer-induced metabolic perturbations have been observed in both male 4 and female 3 mice. Another limitation is the generalizability and broad applicability of our findings to other types of cancer. As the effect of cancer on glucose metabolism and the molecular alteration likely is determined by the type of cancer, 80,81 our findings cannot necessarily be generally transferred to all types of cancer.
In conclusion, we show that cancer leads to marked insulin resistance in oxidative mouse soleus muscle evidenced by blocked insulin-stimulated glucose transport and inhibited insulin-induced phosphorylation of TBC1D4 and GSK3 at multiple phosphorylation sites. Furthermore, cancer impaired mTORC1 signaling, measured via p70S6K-rS6 and ULK1 phosphorylation, in soleus muscle, while only ULK1 phosphorylation was impaired in EDL muscle of tumor-bearing mice. Further investigation of the mechanisms underlying this cancerinduced selective insulin resistance might guide future studies and optimize cancer therapy.

DATA AVAILABILITY STATEMENT
For question(s) or access to data, please contact corresponding author Lykke Sylow.