Modulation of Skeletal Muscle Insulin Signaling With Chronic Caloric Restriction in Cynomolgus Monkeys

The effect of chronic CR to modulate cellular insulin signaling and transcriptional regulation was assessed in skeletal muscle obtained from nonhuman primates subjected to a 4-year period of 30% CR compared with ad libitum (AL) feeding conditions. The metabolic and physiological changes observed in this cohort with CR were previously reported (14). Specifically, 32 feral adult male cynomolgus monkeys (Macaca fascicularis) (mean ± SEM age 8.2 ± 1.2 years) were part of a randomized trial in which the independent effect of CR and its interaction with insulin resistance and atherosclerotic lesion extent and composition were evaluated (Fig. 1).

The animals were acquired directly from the Institute Pertanian (Bogar, Indonesia) and quarantined for 3 months. Animals were housed socially in pairs except when separated at mealtime by sliding a partition to separate them (14). Beginning in the 4th month and throughout the pretrial (months 4–6), all animals were fed a moderately atherogenic diet (0.25 mg cholesterol/calorie) containing 30% of calories from fat. Caloric intake for each animal was assessed by feeding a known allotment and weighing the uneaten food (14). After pretrial evaluations, the animals were assigned to AL or CR diet groups using a stratified randomization. The CR diet was introduced over a 3-month transition period (90% of AL intake during the 1st month, 80% during the 2nd month, and 70% during the 3rd month and thereafter). Additional vitamin and mineral mixture, β-sitosterol, and crystalline cholesterol were added to the CR diet so that the same amount of these components was ingested regardless of the randomized group (14).

Insulin sensitivity was assessed at 6-month intervals using a modified minimal model and at the end of the study using a hyperinsulinemic-euglycemic clamp as described (14). At the baseline of the clamp, a biopsy of the vastus lateralis muscle was obtained. After the insulin infusion, repeat biopsies were obtained at 5, 20, and 40 min of the clamp. Muscle samples (∼200 mg wet wt) were immediately placed into liquid nitrogen and then stored at −80°C. There was no difference between the steady state plasma glucose (5.38 ± 0.10 vs. 5.42 ± 0.10 mmol/l) or plasma insulin levels observed during the clamp (1,472 ± 145 vs. 1,535 ± 155 pmol/l) for either the CR or AL groups, respectively.

Muscle tissue lysates were prepared by dissection and homogenized in buffer A (25 mmol/l HEPES, pH 7.4, 1% Nonidet P-40, 137 mmol/l NaCl, 1 mmol/l phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 1 μg/ml pepstatin, and 5 μg/ml leupeptin), using a PRO 200 homogenizer (PRO Scientific, Oxford, CT). The samples were centrifuged at 14,000g for 20 min at 4°C, and protein content of the supernatant was determined (Bio-Rad protein assay kit; Bio-Rad Laboratories, Hercules, CA). Supernatants (50 μg) were resolved by SDS-PAGE and subjected to immunoblotting using chemiluminescence reagent (PerkinElmer Life Science, Boston, MA) and quantified as described (22). The 19S proteasome base anti-S5A/Rpn10 antibodies were ordered from Calbiochem (Gibbstown, NJ). Antibodies for phospho–insulin receptor substrate (IRS)-1 (Tyr612), phospho–insulin receptor (IR) (Tyr1150/1151), phosphoinositol (PI) 3-kinase protein 85 (p85 of PI 3-kinase), phospho-Akt (Ser473), IRS-1 and IRS-2, 20S proteasome subunit β2i, Akt, serum- and glucocorticoid-inducible kinase 1 (SGK1), signal transducer and activator of transcription 3 (STAT3), and SIRT1 antibodies were obtained from Upstate Biotech (Lake Placid, NY). Anti-19S proteasome lid subunits S9/Rpn6 and S14/Rpn12 antibodies were ordered from BIOMOL International (Plymouth Meeting, PA). GLUT4 monoclonal antibody was obtained from R&D Systems (Minneapolis, MN). Lipoprotein lipase (LPL) antibody was purchased from GeneTex (San Antonio, TX) and β-actin from Affinity Bioreagents (Golden, CO). IR β-subunit was obtained from Santa Cruz Biotechnology (Santa Cruz, CA).

IR tyrosine kinase activity was assessed as described by Le Marchand-Brustel et al. (23) with modification. Briefly, 500 μg of muscle lysate at each time point was added to 50 μl of agarose bound wheat germ agglutinin incubated at room temperature for 2 h. After washing with buffer A, bound receptor was eluted by 3 mol/l N-acetyl-d-glucosamine solution. IR kinase activity was initialed by adding 40 μl of reaction solution consisting of 1 mmol/l DTT, 10 mmol/l MgCl₂, 3 mmol/l MnCl₂, and 5 μmol/l [r-³²P]ATP. Reactions were carried out for 30 min at 22°C and terminated by adding 2 μl of 0.5 mol/l EDTA and 10 μl of 500 μmol/l ATP. Then 2 μg of IR β-subunit antibody and 50 μl of protein A agarose beads were added to the reaction mixture and incubated at room temperature for 30 min. After extensive washing with buffer A, 40 μl of electrophoresis sample buffer was added and heated at 95°C for 4 min. Following SDS-PAGE, gels were dried and bands visualized by autoradiography.

IR β-subunit tyrosine phosphorylation and IR β-subunit abundance were measured by Western blot techniques (24,25). Briefly, 50 μg of lysates prepared as described above was subjected to 8% SDS-PAGE and transferred to nitrocellulose membrane. IR β-subunit phosphorylation was detected with anti–phospho-IR (Tyr1150/1151) antibody. The membrane was striped with strip buffer and reprobed with anti–IR β-subunit and β-actin antibodies, respectively.

To assess IRS-1 phosphorylation, muscle lysates were subjected to SDS-PAGE (24). The bands were detected with anti–phospho-IRS-1 (Tyr612) antibody (26). After measuring phospho-IRS-1 abundance, the membrane was striped with strip buffer and reprobed with anti–total IRS-1 antibody to obtain abundance. IRS-1–associated PI 3-kinase activities of the muscle at each time point were assessed (25,–27). We confirm that wortmannin treatment could completely inhibit IRS-1–associated PI 3-kinase activity of muscle tissues (data not shown). Total Akt and pAkt (Ser473) at each time point were similarly assessed with Western blot techniques.

Protein content for IRS-2, GLUT4, PI 3-kinase (p85), LPL, SGK1, SIRT1, and STAT3 in the lysates was measured also using Western blot analysis. Results of scanning for each gel were normalized by β-actin, and the data are presented as mean ± SEM of fold change in CR versus AL.

Total RNA was extracted from muscle obtained for microarray and real-time quantitative PCR (qPCR) assays. Frozen tissues were placed in a mortar in liquid nitrogen, and the tissue was pulverized into powder using a pestle on dry ice. Total RNA was isolated from the tissue powder using TRIZOL reagent (Invitrogen, Carlsbad, CA). After DNase I (Invitrogen) digestion, RNA was further purified with an RNeasy Mini Kit (QIAGEN, Germantown, MD). RNA concentration and quality were measured by an RNA 6000 Nano LabChip kit (Agilent Technologies, Santa Clara, CA).

The Applied Biosystems human genome survey microarray version 2.0 chip containing 32,878 oligonucleotide probes (60-mer) representing 29,098 individual human genes and more than 1,000 control probes was used for microarray profiling for 10 animals, 5 randomly chosen from each group. It has been demonstrated that human sequence–based DNA arrays can be used effectively to detect differential gene expression in a nonhuman primate (28). Digoxigenin-UTP–labeled cRNA was generated and linearly amplified from 1 μg of total RNA from each sample using an Applied Biosystems Chemiluminescent RT-IVT labeling kit according to the manufacturer's protocol. After cRNA was fragmented by heating at 60°C for 30 min, 10 μg of cRNA fragments were hybridized at 55°C for 16 h. Chemiluminescence detection, image acquisition, and analysis were performed according to the manufacturer's protocol (Applied Biosystems, Foster, CA). Signals were quantified and corrected for background, and final images and feature data were processed by Applied Biosystems 1700 Chemiluminescent Microarray Analyzer software version 1.1.

The primer sequences of candidate genes were designed using Primer Express software version 3.0 (Applied Biosystems). A 1-μg aliquot of total RNA for each sample was reverse transcribed in a 100-μl reaction volume with a commercial High-Capacity cDNA Archive Kit (Applied Biosystems) according to the manufacturer's protocol. The qPCR was conducted in 384-well microtiter plates on the ABI Prism Sequence Detector 7900 (Applied Biosystems) with Bio-Rad iTaqTM SYBR Green Supermix with ROX Kits. For each sample of each gene, PCR amplification was performed in triplicate with β-actin used as an endogenous control. The mRNA content of each candidate gene was determined simultaneously in 10 paired (CR and AL) muscle samples assessed by DNA array analysis. The assay was performed in duplicate.

20S proteasome activity in muscle lysates obtained in the basal state was measured in duplicate using a 20S Proteasome Activity Assay Kit (Chemicon International, Temecula, CA). 20S proteasome chymotrypsin activity was measured by incubating 20 μg of each lysate with fluorophore 7-amino-4-methylcoumarin (AMC)-labeled peptide substrate LLVY-AMC at 37°C for 60 min. The free AMC released by proteasome activity was quantified using a 380/460-nm filter set in a fluorometer (BioTex, Winooski, VT). The AMC standard curve was generated with a series dilution of AMC standard solution. Proteasome activity was confirmed using purified 20S proteasome as the positive control and is reported as μmol/l AMC per mg protein per h. Each sample/substrate combination was measured both in the presence and in the absence of MG132 (10 μmol/l) or epoxomicin (1 μmol/l), a highly specific 20S proteasome inhibitor (Boston Biochem, Cambridge, MA) (29), to account for any nonproteasomal degradation of the substrate.

The effects of CR on the trial evaluations measured at the specified intervals postrandomization were estimated using repeated-measures ANCOVA. Analysis of group differences was adjusted for the prerandomization levels of the outcome measure tested in order to reduce the variance explained by prerandomization predictors. All tests of hypotheses and reported P values were two-sided. Whenever a baseline value was used as a covariate in an ANCOVA model, an interaction term between the group and the covariate was initially included to check the parallelism assumption. If the interaction was not significant at the 0.10 level of significance, and it was always the case, the interaction term was omitted.

For gene expression, within-array normalization was done with the scanning software from Applied Biosystems based on housekeeping genes on each array. Global normalization among arrays was accomplished using quantile normalization (30). ANOVA analysis with Bonferroni adjustment was used for detecting significantly differentiated genes. Differentially expressed genes between the CR and AL groups were determined based on the following criteria: Bonferroni-adjusted P < 0.05 and fold change in the CR over the AL group of ≥1.5. All data analyses were carried out using SAS (Cary, NC). Public databases including David/Ease, GenMAPP, Panther, GOTM, and TreeView version 1.6 were used to assess functional gene cluster analysis (31).

The clinical, phenotypic, and metabolic effects of CR for the cohort of cynomolgus monkeys (Macaca fascicularis) evaluated were previously reported in detail (14). Compared with the AL group, animals randomized to CR were observed to have significantly reduced dietary intake, reduced body weight, increased insulin sensitivity, and reduced intra-abdominal fat with aging (Fig. 2,A–D) (14). The effect of CR was noted during the 1st year of observation and maintained over the 4 years of observation, as demonstrated not only from periodic assessment of insulin sensitivity with the minimal model (Fig. 2,C), but also from the assessment with hyperinsulinemic-euglycemic clamps obtained at study end (Fig. 2 D) (14).

Animals randomized to CR had significantly increased skeletal muscle protein abundance of IRS-1, IRS-2, IR β-subunit, PI 3-kinase (p85), and GLUT4 compared with that in the AL group (Fig. 3). When compared with AL animals, animals randomized to CR had enhanced insulin-stimulated skeletal muscle IR tyrosine kinase activity (data not shown) and increased IR β-subunit phosphorylation (Fig. 4,A) and IRS-1 phosphorylation (Fig. 4,B). IRS-1 protein levels measured at the 5-, 20-, and 40-min time points did not differ from the value assessed as the 0 time point. Although there was no significant difference in basal PI 3-kinase activity between the CR and AL groups, insulin-stimulated PI 3-kinase activities were significantly higher when assessed at all time points post–insulin stimulation in the CR group compared with those in the AL group (Fig. 4,C). Akt phosphorylation post–insulin stimulation was also increased in skeletal muscle with CR when compared with AL (Fig. 4 D).

A total of 241 genes were identified as significantly differentially expressed with CR from 10,163 probes with a satisfactory quality of signals over all the array slides. Among them, 179 genes were observed to be downregulated and 62 genes upregulated. Using cluster analysis, gene expression differed in 11 categories of biological processes with ∼18% of genes involved in either carbohydrate and lipid metabolism or signaltransduction (Fig. 5). Tables 1 and 2 list genes of interest in the muscle that were either significantly upregulated or downregulated with CR for each biological process, respectively.

To confirm the microarray findings, real-time qPCR assays were conducted. Altogether, 27 genes were selected from the microarray analysis to measure transcription levels using RT-qPCR for which 22 were confirmed to have significant changes in the CR compared with the AL group (Table 3). Two genes, i.e., STAT3 and interleukin 6 signal transducer (IL6ST), did not show significant changes between AL and CR conditions as assessed with real-time qPCR assays as opposed to the microarray analysis, and expression of the type 3 iodothyronine deiodinase (DIO3) gene with RT-qPCR did not agree with the findings from the microarray. In addition, SIRT1 transcriptional level in muscle from the CR monkeys was significantly increased compared with that from the AL monkeys, as assessed by RT-qPCR assay.

From the list of genes confirmed by RT-qPCR, LPL, SGK1, SIRT1, and STAT3 were assessed for protein abundance (Fig. 6). The protein abundance of LPL, SGK1, and SIRT1 was significantly increased in the CR compared with the AL group (97, 57, and 35%, respectively). In agreement with the PCR data, protein expression of STAT3 in the CR monkeys was not confirmed by Western blot analysis (Fig. 6). Because increased gene expression of insulin signaling proteins (i.e., IRS-1, IRS-2, and PI 3-kinase) was not noted, yet increased protein content was observed, factors that regulate protein content (i.e., degradation) were sought. The majority of intracellular proteins are degraded by the 26S proteasome (32). Given the inhibitory effect of insulin reported for proteasome-dependent protein degradation (19,20), 20S proteasome activity was assayed in the monkey muscle lysates. 20S proteasome activity in muscle lysates was significantly reduced in the CR group compared with the AL group, although nonproteasomal degradation measured in the presence of two independent proteasome inhibitors, i.e., MG132 or epoxomicin, was unchanged (Fig. 7,A). Decreased 20S proteasome activity with CR was associated with significantly decreased abundance of selected subunits of the 26S proteasome, including a subunit of the 20S catalytic core (β2i) and subunits of the 19S regulatory complex base (S5A, which contains a ubiquitin binding site) and lid (Rpn6 and Rpn12, two of eight non-ATPase subunits) (Fig. 7 B and C), when compared with the AL group.

One of the most consistent physiological changes observed with CR is the favorable effect on glucoregulation, i.e., increased insulin sensitivity, as reported for both rhesus monkeys and rodents (1,–3,15,–17). Our data confirm that CR markedly improved insulin sensitivity in cynomolgus monkeys when evaluated over a 4-year period (14). In addition, we provide novel data regarding the cellular mechanism of action of CR because the enhanced insulin action observed in animals randomized to CR, as opposed to AL, was associated with increased content of proteins of the IR signal transduction pathway.

Enhanced insulin signaling was demonstrated in our study with the finding of increases in IR kinase activities, IR β-subunit phosphorylation, and IRS-1–associated PI 3-kinase activities following insulin stimulation in the muscle tissues in the CR group. The mechanism by which insulin signaling is enhanced with CR has been suggested to be secondary to an effect on transcriptional regulation. For example, it has been reported that liver IR, IGF-1R, and IRS-1 mRNA were greater in older rats subjected to 25 months of 40% CR than in AL-fed rats (33) and that CR also had greater effect on the cardiac gene expression of IR, IRS-1, IGF-1, IGF-1R, and GLUT4 compared with that in age-matched controls (34). In contrast, studies reported for primates subjected to CR do not support the observations in rodents as related to transcription. Kayo et al. (35) evaluated skeletal muscle from rhesus monkeys subjected to CR and reported increases in gene expression of GLUT4, which agree with our data. Moreover, they also reported increased gene expression of PI 3-kinase P110, β-isoform. There were clear differences in our study compared with the other reported studies in that we evaluated a different primate species (i.e., cynomolgus vs. rhesus monkeys), a different dietary content (i.e., atherogenic diet vs. chow), and a later start date for initiation of CR (14,,,–18,35).

Given our observations, the major question would be the mechanism by which insulin signaling protein content in muscle is increased without a significant effect on transcriptional regulation. Ambient protein levels are determined by the coordinated interplay of metabolic processes that involve transcription, mRNA translation, and degradation (36). Transcription and degradation mechanisms have received significant attention, and regulation at the level of mRNA translation as an independent mechanism is suggested when mRNA and protein levels do not correlate (36). It is also reported that the content of IRS-1 and IRS-2 is influenced by many factors, including growth factors and cytokines, and conditions associated with insulin resistance have been reported to exhibit increases in whole-body protein degradation (37,38). Studies using cell lines or isolated primary cells chronically exposed to insulin have revealed that the reduced level of IRS-1 protein is due to enhanced degradation (39). Other studies have demonstrated that a major effect of insulin is regulation of protein degradation mediated by the proteasome (19,20). Based on these reports and from our microarray findings, the next logical step was to evaluate whether the proteins involved in proteasomal degradation are modulated by CR.

Ubiquitin-dependent proteolysis plays an important role in regulating fundamental biological functions, including cell division and cellular differentiation. The proteasome degradation system is composed of two distinct and successive steps: ubiquitin conjugation and proteasome degradation (40). Ubiquitin is first activated by a single ubiquitin-activating enzyme, E1. Following activation, one of several E2 enzymes (ubiquitin-conjugating proteins) transfers ubiquitin from E1 to a member of the ubiquitin-protein ligase family, E3, to which the substrate protein is specifically bound. E3 catalyzes the last step in the conjugation process, covalent attachment of ubiquitin to the substrate (41). Ubiquitin-tagged proteins are then recognized and degraded by the 26S proteasome complex, which consists of the 20S catalytic core complex capped on one or both ends by 19S regulatory complexes (32). In support of the hypothesis that CR may modulate the ubiquitin-proteasome system, transcriptional levels of ubiquitin-specific peptidase 2 and 18 (USP2 and USP18) were reduced. Moreover, skeletal muscle from animals randomized to CR had decreased 20S proteasome activity and greatly reduced protein abundance of selected proteasome subunits. Collectively, our data support the hypothesis that CR enhances the insulin signaling cascade, and one of the contributing mechanisms may be secondary to modulating the ubiquitin-proteasome system in skeletal muscle.

There were other genes of interest that were modulated with CR including those involved in lipid metabolism, e.g., LPL and related enzyme transcriptional levels (Tables 1,TABLE 2–3). In addition, we observed significantly increased SIRT1 protein abundance and gene expression with CR in primates. This result appears to be consistent with data found in rodent studies (11,42), but other investigators suggest that the regulation of SIRT1 activity during CR may be tissue specific (43). Thus, the relationship between SIRT1 and insulin signaling in CR animals needs to be further studied.

Our data demonstrating that insulin signaling proteins are increased in muscle with CR appear to be in contrast to many published studies. For example, Gazdag et al. (18) evaluated CR in rhesus monkeys and reported only that the content of IRS-1 approached significance (P = 0.051). Friedman et al. (44) reported no change in GLUT4 protein in skeletal muscle but observed a significant improvement in insulin sensitivity in obese human subjects after an average weight loss of 36%. Kim et al. (45) reported improvement in insulin sensitivity with enhanced insulin-stimulated receptor and IRS-1 tyrosine phosphorylation without change in protein content in muscle. Previously, we reported that there was no change in GLUT4 levels for heart or diaphragm muscle between AL or CR animals despite increase in insulin sensitivity (22). Thus, it is important to note that an increase in insulin signaling proteins or GLUT4 abundance does not appear to be essential for increased insulin sensitivity with reduction in caloric intake. Taken together, the mechanism by which CR has its effects on insulin action in cynomolgus monkeys may be different from those reported in other species that demonstrate improved insulin action with CR (i.e., humans, rhesus monkeys, rats, and mice) (1,–3,13,,,,–18,42). It is not known at this time whether the differences are related to the species, composition of diet, age at CR initiation, length of CR intervention, or other factors.

In summary, our data demonstrate that CR enhances insulin sensitivity and skeletal muscle content of insulin signaling proteins in cynomolgus monkeys. Because ambient protein levels are determined by transcription, mRNA translation, and degradation, any of these processes could have contributed to the observations. However, the data do support the finding that a contributing cellular mechanism by which CR enhances insulin action in vivo may be secondary to modulation of protein degradation via the ubiquitin-proteasome system. The mechanism for the specificity for the sparing of degradation of insulin signaling proteins, but not most proteins, remains unexplained and will need to be confirmed with more precise mechanistic studies.

This study was supported by National Institutes of Health Grants AG010816 and AG000578 (awarded to W.T.C.). This project used Genomics Core facilities that are supported in part by COBRE (NIH P20-RR021945) and CNRU (NIH 1P30-DK072476) center grants from the National Institutes of Health.

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

We thank Hui Xie for assistance with microarray data analysis and Nicole Mestayer for preparing this article.