Tissue-Specific Splicing and Dietary Interaction of a Mutant As160 Allele Determine Muscle Metabolic Fitness in Rodents Free

Genetic adaptation to dietary habits plays an important role in evolution and maintains metabolic fitness in human. Changes in nutrient availability may deteriorate such genetic adaptation and lead to metabolic disorders. For example, thrifty genetic variants that evolve to deal with food scarcity in history predispose carriers to metabolic disorders under feast conditions (1). Similarly, changes in dietary composition might also impair genetic adaptation, which results in the development of metabolic disorders. The prevalence of metabolic disorders including type 2 diabetes (T2D) heightens the need to understand the genetic/dietary interaction in the metabolic regulation as well as the pathogenesis of these diseases.

Both genetic and dietary factors can impact on insulin signaling that is a critical regulator of metabolic fitness. Insulin stimulates responses in multiple organs to dispose postprandial glucose, among which skeletal muscle is the largest insulin-sensitive organ in the body, storing 60–80% of postprandial glucose (2). Besides glucose, skeletal muscle also absorbs fatty acids (FAs) as a fuel in response to insulin (3). The dynamics of fuel preference toward glucose or FAs in skeletal muscle is influenced by various factors including physical activities, nutrient availability and dietary composition (4). Whether and how genetic variation alters fuel preference to adapt skeletal muscle to dietary habits remain unclear.

AS160 (also known as TBC1D4) is an Rab GTPase-activating protein (RabGAP) that controls glucose uptake through regulating the glucose transporter GLUT4 (5). Upon insulin stimulation, protein kinase B (PKB) can phosphorylate multiple Ser/Thr sites on AS160 and induce binding of regulatory 14-3-3 proteins to phosphorylated Thr⁶⁴² site on AS160 (6,7). Thr⁶⁴² phosphorylation and/or 14-3-3 binding of AS160 inactivates its GAP activity, which allows translocation of GLUT4 from its intracellular storage sites onto the cell surface. Mutation of Thr⁶⁴² to a nonphosphorylatable alanine inhibits insulin-stimulated GLUT4 translocation and muscle glucose uptake and causes insulin resistance in mice (8). Besides GLUT4 translocation, AS160 also regulates GLUT4 degradation through the lysosome (9). Deletion of AS160 diminishes GLUT4 via promoting its lysosomal degradation, inhibits muscle glucose uptake, and results in postprandial hyperglycemia and hyperinsulinemia in mice (9–11).

Two As160 splicing variants, namely, As160-l (As160 long isoform) and As160-s (As160 short isoform), exist in mammals with distinct tissue distributions (12,13). The As160-l that has 20 exons is mainly expressed in skeletal muscle and heart, whereas As160-s that lacks exons 11 and 12 is the major form in white adipose tissue (WAT). An isoform-specific AS160^R684X mutation (where X represents a stop codon) that selectively affects the As160-l variant has recently been identified in Greenlandic Inuit (allele frequency of 17%) (13) and North American Inuit (allele frequency of 27%) (14). Human patients with the AS160^R684X mutation have a marked reduction of AS160 protein levels in skeletal muscle and an increased risk of a subset of diabetes that features postprandial hyperglycemia and hyperinsulinemia. A single copy of the AS160^R684X mutation causes mild postprandial hyperglycemia, whereas two copies result in marked hyperglycemia and hyperinsulinemia. Whether this mutation confers adaptive advantage in Inuit in history and how it causes diabetes in the present era remain unclear. Inuit have experienced dietary changes in recent decades from their traditional lipid-rich hypoglycemic diets to sugar-rich diets, accompanied by an increased incidence of T2D (15). We hypothesize that the AS160^R684X mutation might be evolutionarily adaptive to lipid-rich hypoglycemic diets but cause diabetes on sugar-rich diets.

Here, we use genetically modified rodent models to investigate the pathophysiological mechanism of the AS160^R684X mutation through its interaction with diets. We also develop a strategy to regulate mRNA splicing of As160 for restoring metabolic homeostasis in AS160^R684X mutant muscle cells.

Recombinant human insulin was bought from Novo Nordisk (Denmark). 2-deoxy-d-[1,2-³H(N)]glucose and d-[1-¹⁴C]-mannitol were from PerkinElmer. All other chemicals were from Sigma-Aldrich or Sangon Biotech (Shanghai, China). The antibodies used are listed in Supplementary Table 1.

A CRISPR/Cas9-based strategy was applied to generate the AS60^R691X knockin mice on a C57Bl/6J background at the transgenic facility of Nanjing University, in which the Arg⁶⁹¹ (the surrounding sequence is PSArRMY, with Arg⁶⁹¹ shown in lowercase boldface type) on AS160 was mutated to a stop codon. An SpeI enzyme restriction site was introduced through synonymous mutagenesis to facilitate genotyping. Genotyping of the AS60^R691X mice was carried out via amplification of the mutated region (605 base pairs [bp]) using two primers (5′-CCCGACCCTTGACGTATCTTCT-3′; and 5′-GAACTCCGAGCATTAGAAGCAGG-3′), followed by restriction digestion with SpeI. The SpeI digestion gave two cleaved products with sizes of 343 bp and 262 bp for the AS60^R691X knockin mice.

The AS60^R693X knockin rats were generated on a Sprague-Dawley background using a CRISPR/Cas9-based strategy at the transgenic facility of Nanjing University, in which the Arg⁶⁹³ (the surrounding sequence is PSArRMC, with Arg⁶⁹³ shown in lowercase boldface type) on AS160 was changed to a stop codon. Six additional nucleotides (TCTAGA, an XbaI enzyme restriction site) were introduced after the mutation site for facilitation of genotyping. two primers The mutated region (533 bp) was amplified with (5′-CCCTTGAAGTCTCTTCTGCCACGT-3′ and 5′-TCCCTCTAAAATTAACCGGTCACG-3′) and digested with XbaI, which produced two cleaved products with sizes of 288 bp and 245 bp for the AS60^R693X knockin rats.

The Ethics Committee at Nanjing University reviewed and approved all animal studies and protocols. Unless otherwise stated, small rodents were raised under a light/dark cycle of 12 h with free access to food and water. Heterozygous breeding pairs were set up to generate homozygous knockins and wild-type (WT) littermates.

Measurement of blood glucose was carried out with a Contour TS glucometer (Bayer). Serum free fatty acid (FFA), triglyceride (TG) and total cholesterol were determined with the Wako LabAssay NEFA (294-63601), LabAssay Triglyceride (29063701), and LabAssay Cholesterol (294-65801) kits (Wako Chemicals USA), respectively. Plasma insulin was measured with an insulin ELISA kit (EZRMI-13K; EMD Millipore). HOMA of insulin resistance (HOMA-IR) was calculated with the formula HOMA-IR = insulin (in mU/L) * glucose (in mmol/L)/ 22.5 as previously described (16).

Glucose tolerance test was performed in small rodents that were deprived of food overnight (16 h). A bolus of glucose (1.5 mg/g) was administered via oral gavage for mice and via intraperitoneal injection for rats. Blood glucose was subsequently determined through tail bleeding with a Contour-TS glucometer.

Mice and rats were deprived of food for 4 h before intraperitoneal injection with insulin (0.75 mU/g). Animals were then tail bled for measurement of blood glucose with a Contour TS glucometer.

Soleus and extensor digitorum longus (EDL) muscles were isolated from mice or rats and used for glucose uptake as previously described (8). Briefly, isolated soleus or EDL muscles were subjected to stimulation with or without insulin for 50 min at 37°C. Afterward, soleus or EDL muscles were transferred into Krebs-Ringer bicarbonate buffer (KRBB) containing 2-deoxy-d-[1,2-³H(N)]glucose and d-1-[¹⁴C]-mannitol and incubated for another 10 min with or without insulin at 30°C. The reaction was terminated via transfer of muscle into ice-cold KRBB containing cytochalasin B. Muscles were blotted dry and weighed. After lysis, ³H and ¹⁴C radioisotopes in muscle lysates were determined with a Tri-Carb 2800TR scintillation counter (PerkinElmer) and normalized with muscle weight for calculation of muscle glucose uptake.

FA uptake and oxidation were determined in soleus muscle isolated from rats as previously described (17) with modifications. Briefly, soleus muscle was treated with or without insulin for 30 min and further incubated in KRBB (with or without) containing ¹⁴C-palmitic acid for 50 min. After incubation, muscles were lysed for measurement of ¹⁴C radioisotopes with the Tri-Carb 2800TR scintillation counter. ¹⁴C radioisotopes in incubation media were evolved by addition of perchloric acid (0.6 mol/L) as gaseous ¹⁴CO₂ that was trapped in benzethonium hydroxide–soaked filters for measurement using scintillation counting. The radioactivity in gaseous ¹⁴CO₂ was used to calculate FA oxidation, and the sum of radioactivity in muscle and gaseous ¹⁴CO₂ was used to calculate FA uptake.

FA uptake in L6 myotubes was measured as previously described (17). Briefly, myotubes were treated with or without insulin for 30 min and then incubated in KRBB (with or without insulin) containing BODIPY 558/568-C12 (D-3835; Thermo Fisher Scientific) for 10 min. Afterward, myotubes were lysed, and BODIPY 558/568-C12 in cell lysates was measured with a microplate reader (Synergy H1; BioTek Instruments, Inc.).

Rat gastrocnemius muscle was dissected out and minced to fine pieces. Minced muscle tissues were incubated with an enzyme solution containing collagenase D (0.75 units/mL), dispase type II (1.0 U/mL), and CaCl₂ (2.5 mmol/L) for 1.5 h at 37°C. The resultant cell suspension was passed through a cell strainer and spun down at 300g for 5 min. Cells were then preplated for 2 h to remove fibroblasts. The resultant satellite cell suspension was plated on collagen-coated petri dishes for expansion. When 70–80% cell confluence was reached, satellite cells were differentiated to myotubes in DMEM medium containing 2% horse serum.

Glucose uptake was performed in rat primary myotubes as previously described (9). Primary myotubes were subjected to serum deprivation for 12 h and then stimulated with or without insulin for 30 min. Cells were incubated in HEPES-buffered saline buffer containing 2-deoxy-d-[1,2-³H(N)]glucose for 10 min. After lysis, radioisotopes in cell lysates were measured using the Tri-Carb 2800TR scintillation counter.

Rat L6 myoblasts were provided by Dr. Amira Klip (University of Toronto, Toronto, Canada) and maintained in DMEM medium containing 10% (v/v) FBS with regular tests for mycoplasma contamination. L6 myoblasts were differentiated into myotubes as previously described (18).

Transfection of rat myotubes with siRNA or morpholinos was carried out using Lipofectamine 3000 reagent (Thermo Fisher Scientific). Two days after transfection, myotubes were lysed as previously described (19). The sequence of siRNA targeting CD36 is 5′-GCACCACATATCTACACAA-3′. The sequence of AS160 morpholino is 5′-ACACTGAGTACAGAAAACACAGCAGA-3′.

A biotinylation-based method was used for measurement of CD36 on the cell surface as previously described (20). After stimulation with or without insulin for 50 min, isolated rat soleus muscles or L6 myotubes were incubated with Sulfo-NHS-SS-Biotin (Thermo Fisher Scientific) in KRBB (with or without insulin) for 30 min. Muscle or cells were lysed, and biotinylated surface CD36 was extracted, using NeutrAvidin Agarose (Thermo Fisher Scientific). After purification, biotinylated CD36 was determined via Western blotting using the CD36 antibody.

Tissues were snap frozen in liquid nitrogen after harvest. Tissue chunks were homogenized with a Polytron tissue processor as previously described (8). Tissue debris was removed through centrifugation, and protein contents of tissue lysates were measured using Bradford reagent (Thermo Fisher Scientific).

After electrophoresis, proteins were transferred onto nitrocellulose membranes and subjected to immunoblotting assay as previously described (19).

Expression levels of target genes were determined via real-time quantitative PCR (QPCR) with an Applied Biosystems StepOnePlus system. The primers for real-time QPCR are summarized in Supplementary Table 2.

Data are given as the mean ± SEM. Comparisons were performed via t test for two groups or via one-way or two-way ANOVA for multiple groups with Prism software (GraphPad, San Diego, CA).

All data needed to evaluate the conclusions in this study are present in the article and/or Supplementary Material. Additional data related to this article may be requested from the authors.

The selective effect of the AS160^R684X mutation on the As160-l variant in skeletal muscle led us to investigate As160 splicing in more detail using rodent models. Exon 11 is 165-nt and exon 12 is 24-nt in human and rodents, which gives four theoretical splicing variants, a full-length isoform (As160-f) containing both exon 11 and exon 12, a long isoform (As160-l) containing only exon 11, a medium isoform (As160-m) only containing exon 12, and a short isoform (As160-s) with neither exon 11 nor exon 12. We designed isoform-specific primers to determine their expression levels in skeletal muscle, WAT, and hearts of mice and rats. We found that both As160-f and As160-m were hardly detectable in the three tissues in both mice and rats (Supplementary Fig. 1A–L). The As160-l isoform dominated in skeletal muscle and heart, whereas As160-s was expressed mainly in WAT in both mice and rats (Supplementary Fig. 1A–L). Similar expression patterns of these four splicing variants were observed in human skeletal muscle and WAT (Supplementary Fig. 1M–P). In the following study, we mainly focused on the As160-l and As160-s.

To investigate the pathophysiological mechanism of human AS160^R684X mutation, we first generated a mouse model in which AS160-Arg⁶⁹¹ (the residue orthologous to human AS160-Arg⁶⁸⁴) was substituted with a stop codon through the knockin mutagenesis. AS160^R691X mice were viable with no overt physical abnormality. Human patients bearing the AS160^R684X mutation exhibit postprandial hyperglycemia and hyperinsulinemia (13). In contrast, AS160^R691X mice displayed normal clearance of blood glucose after oral glucose gavage (Fig. 1A and Supplementary Fig. 2A). Plasma insulin was unchanged in AS160^R691X mice compared with WT littermates in the basal state as well as after glucose administration (Fig. 1B). Refeeding with a chow diet (CD) after an overnight fast resulted in postprandial hyperglycemia and hyperinsulinemia in global AS160 knockout mice (9). However, when AS160^R691X mice were subjected to this refeeding protocol, they again exhibited neither postprandial hyperglycemia nor hyperinsulinemia (Fig. 1C and D and Supplementary Fig. 2B). Expression of key insulin signaling components such as insulin receptor (IR), PKB, and GSK3 was normal in AS160^R691X mice (Supplementary Fig. 2D and E). Phosphorylation of PKB and GSK3 was also comparable between AS160^R691X mice and WT littermates (Supplementary Fig. 2D and E). Together, these data show that the AS160^R691X mutation does not impair glycemic control in mice.

We next sought to find out why AS160^R691X mice had normal glycemic control. Since the mutation only exists in the As160-l isoform, we determined the expression of splicing isoforms, namely, As160-l and As160-s, in different tissues of AS160^R691X mice. The As160-l isoform was dominant in skeletal muscle and heart, while As160-s was the major form in WAT, of WT mice (Supplementary Fig. 1A–C). In skeletal muscle and heart of AS160^R691X mice, the knockin mutation expectedly downregulated expression of the As160-l isoform, most likely through nonsense-mediated mRNA decay (Fig. 1E and F). Interestingly, the AS160^R691X mutation induced expression of the As160-s isoform in both skeletal muscle and heart (Fig. 1E and F). Importantly, this compensatory response left AS160 protein levels unchanged in these two types of muscle tissues in AS160^R691X mice (Fig. 1G and H). In WAT, the AS160^R691X mutation did not alter expression of the As160-s isoform that dominates in this tissue (Supplementary Fig. 2C). Consequently, the amount of AS160 protein remained normal in WAT of AS160^R691X mice (Fig. 1H).

The induction of As160-s isoform in skeletal muscle led us to hypothesize that this alternative splicing may maintain muscle glucose metabolism. Thr⁶⁴² is a key phosphorylation site on AS160 regulating insulin-stimulated muscle glucose uptake (8), which is present in both AS160 isoforms. Insulin stimulated phosphorylation of AS160-Thr⁶⁴² to similar levels in WT and AS160^R691X muscle (Fig. 1G). GLUT4 protein was diminished in skeletal muscle of AS160^R684X human patients and AS160 knockout mice (10,13). In contrast, the switch of AS160 isoforms resulted in normal expression of GLUT4 protein in skeletal muscle of AS160^R691X mice (Fig. 1H). Moreover, no change in GLUT1 protein was observed in skeletal muscle of AS160^R691X mice compared with WT (Fig. 1H). Importantly, insulin-stimulated uptake of glucose into isolated soleus muscle was comparable between AS160^R691X mice and WT littermates (Fig. 1I). Similarly, insulin-stimulated uptake of glucose into isolated EDL muscle also exhibited no difference between AS160^R691X and WT mice (Fig. 1J).

These data suggest that the induction of the As160-s isoform compensates for the loss of the As160-l form due to the AS160^R691X mutation in mouse skeletal muscle, which helps to maintain muscle glucose metabolism and whole-body glucose homeostasis in mice.

Although rats and mice are both rodents, with many similarities in their metabolism, they still have some distinct metabolic features (21). We wondered whether an AS160^R693X mutation (the site orthologous to human AS160-Arg⁶⁸⁴) might affect glycemic control in rats. To this end, we generated a rat model in which AS160-Arg⁶⁹³ was mutated to a stop codon through the knockin substitution. AS160^R693X rats exhibited no gross physical abnormality and gained weight at a rate similar to that of their WT littermates on CD (Supplementary Fig. 3A). These rats had normal fasted blood glucose (Fig. 2A and Supplementary Fig. 3B). When AS160^R693X rats were administered with a bolus of glucose via gavage, they were intolerant of glucose challenge as compared with WT (Fig. 2B). We then refed AS160^R693X rats with the CD after an overnight fast. Interestingly, AS160^R693X rats displayed higher blood glucose than WT littermates after refeeding (Fig. 2C and Supplementary Fig. 3C). Moreover, postprandial plasma insulin levels were higher in AS160^R693X rats than in WT littermates (Fig. 2D). These data show that AS160^R693X rats display postprandial hyperglycemia and hyperinsulinemia, which is similar to human patients bearing the AS160^R684X mutation.

The metabolic phenotype of AS160^R693X rats prompted us to study glucose metabolism in their skeletal muscle, which stores 60–80% postprandial glucose (2). In isolated soleus muscle, basal glucose uptake was comparable between AS160^R693X rats and WT littermates (Fig. 2E). Insulin stimulated uptake of glucose into WT soleus muscle by approximately sevenfold, whereas it only increased glucose absorption into AS160^R691X soleus muscle by approximately threefold (Fig. 2E). Similar effects of the AS160^R693X mutation on glucose uptake were observed in isolated EDL muscle (Fig. 2F). AS160^R693X rats had normal expression of key insulin signaling components such as IR and PKB in their skeletal muscle (Fig. 2G). Insulin stimulated PKB phosphorylation in AS160^R693X skeletal muscle to a level similar to that in WT muscle (Fig. 2G). These data demonstrate that the AS160^R693X mutation impairs insulin-stimulated glucose uptake in rat skeletal muscle, which is not ascribed to any change in proximal insulin signaling.

To further understand these changes in muscle glucose uptake, we investigated impacts of the AS160^R693X mutation on expression of key regulators of muscle glucose metabolism. The AS160^R693X mutation expectedly diminished As160-l mRNA levels in rat skeletal muscle, most likely through nonsense-mediated decay (Fig. 3A). Similar to the human patients (13) and in contrast to mice (Fig. 1E), the AS160^R693X mutation did not induce compensatory expression of the As160-s isoform in skeletal muscle of rats (Fig. 3B). As a consequence, AS160 protein was greatly decreased in various types of skeletal muscle in AS160^R693X rats (Fig. 3F). Phosphorylation of AS160-Thr⁶⁴² was diminished under both basal and insulin-stimulated conditions in soleus and EDL muscles of AS160^R693X rats (Fig. 3G). In contrast to skeletal muscle, the AS160^R693X mutation led to a diminution of As160-l mRNA but induced expression of As160-s mRNA in the heart (Fig. 3C and D). This compensatory splicing resulted in expression of the normal amount of AS160 protein in this organ (Fig. 3F). The AS160^R693X mutation affected expression of neither As160-s mRNA nor AS160 protein in rat WAT (Fig. 3E and F). TBC1D1, an RabGAP similar to AS160, remained normal in AS160^R693X muscle (Supplementary Fig. 4E, G, and I). AS160 regulates GLUT4 protein levels, and its deficiency promotes GLUT4 degradation via the lysosome (9). In agreement, GLUT4 protein was significantly decreased in both soleus and EDL muscles of AS160^R693X rats, whereas Glut4 mRNA levels remained normal in these tissues (Fig. 3H–K and Supplementary Fig. 4A and C). This effect was specific to GLUT4, since GLUT1 protein and its mRNA were unaltered in skeletal muscle of AS160^R693X rats (Fig. 3H–K and Supplementary Fig. 4B and D). In accordance with downregulation of GLUT4, expression of hexokinase was significantly decreased in AS160^R693X muscle (Supplementary Fig. 4F, H, and I).

To gain more insights into the impacts of AS160 mutation on muscle metabolism, we performed deep sequencing of RNA (RNA-Seq) in skeletal muscle of mutant mice and rats. The transcriptome remained largely unaltered in skeletal muscle of AS160^R691X mice compared with their WT littermates (Supplementary Fig. 5A), further suggesting normal muscle metabolism in these mutant mice. In contrast, 1,167 genes displayed differential expression in skeletal muscle of AS160^R693X rats, among which 1,110 genes were upregulated and only 57 genes were downregulated (Supplementary Fig. 5B). Interestingly, pathway analysis revealed that FA catabolism was activated in skeletal muscle of AS160^R693X rats (Fig. 4A and Supplementary Fig. 5C). FA oxidation in skeletal muscle is regulated by PPARδ, whose signaling was upregulated in skeletal muscle of AS160^R693X rats (Fig. 4A and Supplementary Fig. 5C). We validated these RNA-Seq results via QPCR, which confirmed upregulation of key factors for PPARδ signaling and FA oxidation, including Lpl, Cpt1β, Cpt2, and Acad1 (Fig. 4B). PPARδ in skeletal muscle directly regulates expression of PGC1α, which is a key regulator of mitochondrial biogenesis and function (22). In turn, PGC1α upregulates PPARδ in muscle cells, thus resulting in a feed-forward loop (23). Both Pparδ and Pgc1α were increased in skeletal muscle of AS160^R693X rats (Fig. 4B). Moreover, PERM1, a downstream target of PGC1 and muscle-specific regulator of mitochondrial oxidative capacity (24), was upregulated in skeletal muscle of AS160^R693X rats (Fig. 4B). We measured protein levels of key regulators for PPARδ signaling and FA oxidation and found that they were significantly elevated in skeletal muscle of AS160^R693X rats, probably as a result of upregulation of their mRNA levels (Fig. 4C and D). Cellular component analysis showed that mitoch-ondrial components were enriched among the differentially expressed genes in skeletal muscle of AS160^R693X rats (Supplementary Fig. 5D), which is consistent with upregulation of PPARδ and PGC1α. In total, 55 nuclear-encoded mitochondrial genes were significantly increased at the mRNA level in skeletal muscle of AS160^R693X rats, including key factors for mitochondrial transcription, translation, and rRNA methylation, components of mitochondrial respiratory complex I –IV, and mitochondrial transporters (Fig. 4A). We further confirmed the increase of a number of selected mitochondrial gene mRNAs via QPCR (Fig. 4E). Peroxisome is another organelle essential for lipid oxidation. The transcriptomic analysis revealed an upregulation of peroxisomal gene mRNAs in skeletal muscle of AS160^R693X rats (Fig. 4A and Supplementary Fig. 5D). Validation of the transcriptomic analysis was performed via QPCR, confirming the increase of two important regulators for peroxisomal biogenesis, Pex11a and Pex12, as well as key genes for peroxisomal β-oxidation including Acox1 and Ech1 in skeletal muscle of AS160^R693X rats (Fig. 4F).

Long-chain FAs (LCFA) are natural ligands for PPARδ, whose uptake into skeletal muscle is also under control of insulin (3). As expected, LCFA uptake into skeletal muscle of WT rats was increased in response to insulin (Fig. 5A). Interestingly, muscle LCFA uptake was significantly higher in AS160^R693X rats than in WT littermates under both basal and insulin stimulation conditions (Fig. 5A). LCFA uptake into skeletal muscle is mainly mediated by the FA translocase CD36, which also undergoes translocation from intracellular compartments onto cell surface in response to insulin (25). Protein levels of CD36 did not change in skeletal muscle of AS160^R693X rats compared with WT littermates (Fig. 5C and D). In contrast, surface expression of CD36 was significantly higher in AS160^R693X muscle than in WT muscle (Fig. 5C and E), suggesting that the loss of AS160 due to the AS160^R693X mutation promotes CD36 translocation for muscle uptake of LCFA. In agreement with increased expression of lipid oxidation genes, AS160^R693X muscle displayed higher LCFA oxidation rates than WT muscle (Fig. 5B). Moreover, the AS160^R693X mutation accelerated clearance of blood TG and FFA after oral gavage of oil (Fig. 5F and G), suggesting an increase of lipid utilization in the mutant rats.

To further establish the role of the AS160-CD36 axis in regulation of PPARδ-dependent gene expression, we downregulated AS160 in rat L6 myocytes via shRNA (Supplementary Fig. 6A). Knockdown of AS160 in L6 myocytes increased LCFA uptake in response to insulin (Supplementary Fig. 6B). Similarly, downregulation of AS160 did not alter CD36 protein expression but significantly increased CD36 levels on cell surface of L6 myocytes (Supplementary Fig. 6C–E). The mRNA levels of Pparδ, Pgc1α, Cpt1β, Cpt2, and Fabp3 were upregulated in AS160 knockdown (AS160-KD) cells (Fig. 5H). Importantly, suppression of CD36 expression via siRNA blunted the induction of these PPARδ-dependent genes in AS160-KD L6 myocytes (Fig. 5H). Therefore, these results demonstrate that AS160 deficiency upregulates PPARδ-dependent gene expression through promoting CD36-mediated LCFA uptake in skeletal muscle (Fig. 5I).

Together, these data show that the AS160 mutation induces lipid utilization in skeletal muscle of AS160^R693X rats, which probably compensates for the decrease of glucose utilization.

Increased sugar consumption is thought to contribute to the prevalence of T2D in Inuit (15). This led us to hypothesize that AS160^R693X rats might develop more severe metabolic abnormalities when fed with sugar-rich diets. To test this hypothesis, we put the animals on two different diets, CD and a high-sucrose diet (HSD). The AS160^R693X and WT rats gained body weight at similar rates on these two diets (Supplementary Fig. 7A). On both CD and HSD, AS160^R693X rats developed hyperglycemia and hyperinsulinemia when refed after an overnight fast (Fig. 6A–D). Fasting glucose was gradually increased and became significantly higher in AS160^R693X rats after 9 weeks on the HSD (Fig. 6E and F). We calculated the HOMA-IR as a measure to evaluate insulin resistance in AS160^R693X rats on different diets. Importantly, HOMA-IR was significantly increased in AS160^R693X rats when they were fed with the HSD, while it was not altered in WT littermates (Fig. 6G). Expression of IR, PKB, and GSK3 was normal, while phosphorylation of PKB and GSK3 was significantly lower, in skeletal muscle and liver of HSD-fed AS160^R693X rats compared with their WT controls (Fig. 6H–K).

We then examined how sugar consumption impacted on tissue and lipid homeostasis in AS160^R693X rats. Weights of tibialis anterior muscle and liver did not change upon HSD feeding in WT or AS160^R693X rats compared with those on the CD (Supplementary Fig. 7B and C). Interestingly, HSD consumption caused a significant decrease in the weight of epididymal WAT (eWAT) in AS160^R693X rats but not in WT rats (Fig. 7A). Expression of lipogenic genes Fasn and Acc1 remained normal in the two genotypes on the HSD (Fig. 7B). Lipolytic genes Atgl and Mgl were also unaltered in the two genotypes on the HSD (Fig. 7B). Interestingly, another key lipogenic gene, Hsl, was significantly induced in eWAT of HSD-fed AS160^R693X rats but not WT rats at both mRNA and protein levels (Fig. 7B–D). HSL protein was comparable during fasting-refeeding treatment between the two genotypes on the CD, and its phosphorylation was suppressed to similar levels in the two genotypes after refeeding (Supplementary Fig. 7D–F). FABP4 is a key factor for facilitating export of FFA derived from lipolysis out of adipocytes (26). Similar to HSL, FABP4 mRNA and protein were both increased in eWAT of HSD-fed AS160^R693X rats but not WT rats (Fig. 7B–D). Although liver weight did not change, hepatic TG was significantly elevated in AS160^R693X rats when they were fed with the HSD compared with hepatic TG in those fed the CD (Fig. 7E). Expression of lipogenic and lipolytic genes did not differ in the two genotypes (Fig. 7F–H), suggesting that hepatic TG accumulation in HSD-fed AS160^R693X rats might be secondary to elevated lipolysis in eWAT.

Together, these data show that interaction of the AS160^R693X mutation with the sugar-rich diet exacerbates insulin resistance and aggravates metabolic abnormalities in rats.

We then sought to use AS160^R693X rats and their derived cells to develop potential therapeutic strategies. The maintenance of glucose metabolism in skeletal muscle of AS160^R693X mice by the compensatory expression of the As160-s splicing variant suggests that modulation of the splicing switch might have therapeutic potential to treat human patients with the AS160^R684X mutation. As a proof-of-concept experiment, we used a morpholino-mediated exon skipping technology to induce splicing switch in primary myocytes derived from AS160^R693X rats. An antisense morpholino targeting the splicing site of exon 11 was designed, named Exon Switch (ES), and delivered into AS160^R693X myocytes (Fig. 8A). The ES morpholino induced exon skipping and increased expression of As160-s in AS160^R693X myocytes (Fig. 8B). As a consequence of increased As160-s expression, AS160 protein was markedly increased in AS160^R693X myocytes upon treatment with the morpholinos (Fig. 8C). We then examined the effect of ES morpholino on glucose uptake in primary myocytes. As expected, insulin stimulated glucose uptake into the WT myocytes by approximately twofold (Fig. 8D). In contrast, insulin could hardly increase uptake of glucose into AS160^R693X myocytes treated with a control morpholino treatment (Fig. 8D). Importantly, insulin significantly stimulated glucose uptake into AS160^R693X myocytes upon treatment with the ES morpholino (Fig. 8D). These data suggest that exon skipping might provide a strategy to treat muscle insulin resistance in human patients with the AS160^R684X mutation.

Our findings shed light on how the orthologous site of the human diabetogenic AS160^R684X mutation impacts on fuel preference and glycemic control through interaction with diets. Our results are consistent with a model in which the orthologous site of human AS160^R684X mutation switches muscle fuel preference toward FAs and deteriorates glycemic control on the sugar-rich diet in rats (Fig. 8E).

Given the high frequency of the AS160^R684X mutation in Inuit populations (13,14), it is of great value to elucidate how this mutation causes postprandial hyperglycemia and hyperinsulinemia. Whole-body deletion of AS160 causes postprandial hyperglycemia and hyperinsulinemia in mice, which is ascribed to AS160 deficiency–induced lysosomal degradation of GLUT4 and consequent muscle insulin resistance (9). Skeletal muscle–specific knockout of AS160 further demonstrates the importance of muscle AS160 in control of postprandial glucose homeostasis (9). The knockout strategies affect expression of both splicing variants of As160 in skeletal muscle, whereas the AS160^R684X mutation only impairs the As160-l isoform. Our studies using AS160^R693X rats firmly establish the causal relationship between this diabetogenic AS160 mutation and muscle insulin resistance. Based on this rat model and previous studies on human and mice, we propose a model in which this diabetogenic AS160 mutation selectively diminishes As160-l mRNA levels in skeletal muscle through nonsense-mediated decay, which results in a low protein level of AS160 in skeletal muscle. Decreased AS160 protein promotes GLUT4 degradation via the lysosome, which consequently inhibits insulin-stimulated glucose uptake into skeletal muscle and causes postprandial hyperglycemia and hyperinsulinemia.

Individuals bearing genetic variants/mutations may respond differently to various diets. The genetics-diet interaction allows certain populations to get along with their traditional diets but may lead to metabolic disorders after a switch to industrialized Western diets. The Inuit used to consume very little carbohydrate in their traditional diets (15). Based on our results that reveal metabolic reprogramming toward lipid utilization in the mutant rats, we propose that the prevalence of this AS160 mutation in Inuit might be a result of evolutionary selection due to food availability. The Inuit bearing this AS160 mutation might be historically more adapted to their traditional lipid-rich hypoglycemic diets. However, sugar consumption of Inuit has soared in recent decades, accompanied by a dramatic increase in T2D incidence (15). Our findings in using the rat model demonstrate a specific interaction between the diabetogenic AS160 mutation and sugar-rich diets, which may help to explain the pathogenesis of T2D in Inuit patients with this mutation. The sugar-rich diet exacerbates insulin resistance in AS160^R693X rats, and this genetics-diet interaction reprograms whole-body lipid metabolism. One striking phenotype of AS160^R693X rats is their eWAT weights that become lighter on the HSD. It is well established that insulin resistance promotes lipolysis in WAT (27), which both manifest in the HSD-fed AS160^R693X rats. The enhanced lipolysis in eWAT of HSD-fed AS160^R693X rats might primarily meet the energy demand of their skeletal muscle that depends on lipids as a fuel. However, this increase of lipolysis in adipose also results in accumulation of ectopic fat in liver of HSD-fed AS160^R693X rats. Therefore, AS160^R693X rats offer a valuable model for understanding of genetics-diet interactions in the pathogenesis of metabolic diseases. However, caution needs to be taken in interpreting effects of the AS160^R693X mutation in this rat model, since rodents and human may differ in some aspects of their metabolism (28).

Posttranslational modifications of insulin receptor substrates (IRS) play an important role in the development of muscle insulin resistance (29). Besides IRS, distal components of insulin signaling may also contribute to the pathogenesis of muscle insulin resistance. For example, hypophosphorylation of a PKB substrate, RalGAPα1 (also known as GARNL1), causes muscle insulin resistance toward both glucose and FA metabolism (17). Here we show that AS160 deficiency in skeletal muscle causes selective insulin resistance toward glucose metabolism but increases insulin sensitivity toward FA metabolism. Therefore, muscle insulin sensitivity toward glucose and FA metabolism diverges at the signaling node of AS160 that differentially regulates intracellular trafficking of both GLUT4 and CD36. The differential actions of AS160 on GLUT4 and CD36 might require different Rab small G-proteins. Rab8a, one of the downstream Rabs targeted by AS160, regulates insulin-stimulated LCFA uptake through CD36 translocation but not insulin-induced glucose uptake in skeletal muscle (25). Its deficiency causes selective insulin resistance toward FA metabolism (25). Unlike the defined role of Rab8a in regulation of CD36 translocation, the identities of Rabs downstream of AS160 that regulate GLUT4 translocation and lysosomal degradation remain unknown in skeletal muscle, although candidates such as Rab8a and Rab13 have been implicated in regulation of GLUT4 translocation in L6 rat skeletal myocytes (30).

Both mouse and rat diabetic models are commonly used to study type 2 diabetes (31). To our surprise, only AS160^R693X rats display symptoms that are similar to those in human patients with the AS160^R684X mutation (13,14). Unlike AS160^R693X rats, AS160^R691X mice have normal glycemic control, most likely due to alternative splicing of As160 in skeletal muscle. The two As160 splicing variants, As160-l and As160-s, display distinct tissue distribution (13). For instance, As160-s is the major variant in WAT, whereas As160-l dominates in skeletal muscle and heart. Seven insulin-regulated phosphorylation sites, namely, Ser³¹⁸, Ser³⁴¹, Ser⁵⁷⁰, Ser⁵⁸⁸, Thr⁶⁴², Ser⁶⁶⁶, and Ser⁷⁵¹, are present in both AS160-L and AS160-S isoforms (5,7). Switch to the As160-s form seems to functionally compensate for the loss of As160-l in terms of insulin-stimulated glucose uptake in skeletal muscle of AS160^R691X mice. AS160 also plays a key role in mediating glucose uptake in response to muscle contraction (32) and is necessary for increasing muscle insulin sensitivity in response to contraction and AICAR, both of which can activate the AMP-activated protein kinase (AMPK) (33). In addition to PKB, AMPK also phosphorylates AS160 on multiple sites including Ser⁵⁷⁰ and Ser⁵⁸⁸ (7,34). Besides these sites that are common to both isoforms, AMPK can phosphorylate Ser⁷⁰⁴ on the AS160-L isoform though this is absent on AS160-S (35). However, mutation of Ser⁷⁰⁴ to a nonphosphorylatable alanine does not affect muscle glucose uptake in response to AMPK activators, namely, AICAR and contraction (35). Nevertheless, it would be necessary to find out whether switch to the As160-s isoform can compensate for the loss of As160-l form to control muscle glucose uptake in response to stimuli other than insulin in AS160^R691X mice.

Up to 95% of multiexon genes undergo alternative splicing, which enriches the tissue-specific proteome with protein isoforms through different exon combinations (36). Alternative splicing requires trans-acting factors that activate or repress selection of splice sites on pre-mRNA (37). Tissue-specific expression of trans-acting factors such as the SR, hnRNP, RBM, MBNL, CELF/CUGBP, and KH protein families plays a key role in tissue-specific alternative splicing (37–39). Besides these protein factors, RNA components of the spliceosome vary by an order of magnitude across tissues and also contribute to regulation of tissue-specific alternative splicing (40). It is currently unclear why the As160-s isoform is increased in skeletal muscle and heart of AS160^R691X mice as well as in heart of AS160^R693X rats. One possibility is that certain protein or RNA splicing factors might be altered to promote alternative splicing for the As160-s variant in these tissues. Alternatively, the stability of the As160-s variant might be enhanced, leading to its accumulation in these tissues. Nevertheless, our results show that upregulation of the As160-s variant in skeletal muscle might be of therapeutic value in treatment of human patients bearing the AS160^R684X mutation.

Our proof-of-concept experiment shows that morpholino-mediated exon skipping has the potential to increase the As160-s variant in AS160^R693X rat myocytes, which consequently improves insulin sensitivity in these cells. Morpholino-mediated exon skipping has been successfully applied in the clinic to treat Duchenne muscular dystrophy through splicing exon 51 out in the pre-mRNA of mutated dystrophin (41). Exon skipping has also been realized through various gene editing tools, such as Cas13a (42), Cas13d (43), and a CRISPR-guided cytidine deaminase (44). These exon-skipping approaches shall be evaluated in terms of their effects to increase As160-s expression and to restore glycemic control in AS160^R693X rats, which might eventually lead to development of precision medicine to treat human patients bearing the AS160^R684X mutation.

In summary, we demonstrate that the diabetogenic AS160 mutation causes a fuel switch likely to adapt skeletal muscle to nutrient availability and leads to metabolic disorders through interaction with sugar-rich diets. Our findings also have implications for development of precision medicine through morpholino-mediated exon skipping, which might be of therapeutic potential to treat human patients bearing the AS160^R684X mutation.

Acknowledgments. The authors thank Professor Carol MacKintosh (University of Dundee, Dundee, U.K.) for proofreading of the manuscript and members of the resource unit at Nanjing University for technical assistance.

Funding. The authors thank the Ministry of Science and Technology of China (grant 2018YFA0801104 to H.Y.W. and 2018YFA0801102 to S.C.), the National Natural Science Foundation of China (31971067 to H.Y.W., 32025019 to S.C., and 91954109 to Q.C.), and the Science and Technology Foundation of Jiangsu Province of China (BK20190305 [Basic Research Program] to Q.C. and BK20200315 [Basic Research Program] to C.Q.) for financial support.

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

Author Contributions. X.Y., Q.C., Q.O., P.R., W.F., C.Q. and M.L. performed experiments, analyzed data, and reviewed the manuscript. Q.J. and H.L. reviewed and edited the manuscript. T.-J.Z., H.Y.W., and S.C. designed experiments, analyzed data, and wrote the manuscript. All authors approved the final version of the manuscript. S.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.