Greater Phosphorylation of AMPK and Multiple AMPK Substrates in the Skeletal Muscle of 24-Month-Old Calorie Restricted Compared to Ad-Libitum Fed Male Rats

Abstract

AMP-activated protein kinase (AMPK), a highly conserved, heterotrimeric serine/threonine kinase with critical sensory and regulatory functions, is proposed to induce antiaging actions of caloric restriction (CR). Although earlier studies assessed CR’s effects on AMPK in rodent skeletal muscle, the scope of these studies was narrow with a limited focus on older animals. This study’s purpose was to fill important knowledge gaps related to CR’s influence on AMPK in skeletal muscle of older animals. Therefore, using epitrochlearis muscles from 24-month-old ad-libitum fed (AL) and CR (consuming 65% of AL intake for 8 weeks), male Fischer-344 × Brown Norway F1 rats, we determined: (a) AMPK Thr172 phosphorylation (a key regulatory site) by immunoblot; (b) AMPKα1 and AMPKα2 activity (representing the 2 catalytic α-subunits of AMPK), and AMPKγ3 activity (representing AMPK complexes that include the skeletal muscle-selective regulatory γ3 subunit) using enzymatic assays; (c) phosphorylation of multiple protein substrates that are linked to CR-related effects (acetyl-CoA carboxylase [ACC], that regulates lipid oxidation; Beclin-1 and ULK1 that are autophagy regulatory proteins; Raptor, mTORC1 complex protein that regulates autophagy; TBC1D1 and TBC1D4 that regulate glucose uptake) by immunoblot; and (d) ATP and AMP concentrations (key AMPK regulators) by mass spectrometry. The results revealed significant CR-associated increases in the phosphorylation of AMPK^Thr172 and 4 AMPK substrates (ACC, Beclin-1, TBC1D1, and TBC1D4), without significant diet-related differences in ATP or AMP concentration or AMPKα1-, AMPKα2-, or AMPKγ3-associated activity. The enhanced phosphorylation of multiple AMPK substrates provides novel mechanistic insights linking AMPK to functionally important consequences of CR.

AMP-activated protein kinase (AMPK) is a highly conserved, serine/threonine kinase with a multitude of important sensory and regulatory functions, including the control of energy metabolism, transcriptional regulation, inhibition of cell growth, and activation of autophagy (1). AMPK is a heterotrimeric protein complex comprised of a single catalytic subunit (α1 or α2 isoform) and 2 regulatory subunits (β1 or β2 together with γ1, γ2, or γ3). Increasing AMPK’s enzymatic activity relies on phosphorylation of the Thr172 residue that is localized in the catalytic α subunit (2). Elevated AMP concentration leading to greater AMP binding to the regulatory γ subunit favors enzymatic activation of AMPK via 3 mechanisms: (a) enhanced Thr172 phosphorylation by protein kinases; (b) inhibited Thr172 dephosphorylation by protein phosphatases; and (c) allosteric elevation of enzyme activity (3). Increasing ATP concentration opposes each of these mechanisms.

AMPK has been frequently proposed to participate in the antiaging actions of caloric restriction (CR) (1,4–7). Multiple studies have assessed CR’s effects on AMPK in the skeletal muscle of rats or mice, but the scope of these studies has been narrow, the results have been inconclusive, and only a few have focused on older animals. In most of these previous publications, the analysis was limited to the determination of Thr172 phosphorylation. Although some of these studies did not detect differences between ad libitum (AL) controls and CR groups for Thr172 phosphorylation (8–16), several other studies reported greater pThr172 for skeletal muscles from CR versus AL animals (17–20). Many protein substrates for AMPK have been identified (21–23), but only 1 AMPK substrate, acetyl-CoA carboxylase (ACC), has been compared for skeletal muscle from AL versus CR rodents in 2 earlier studies, each of which found greater ACC phosphorylation in skeletal muscles from CR versus AL animals (14,19).

Whole body knockout of the α1 isoform did not alter glucose tolerance or insulin resistance, but whole body knockout of the α2 isoform was characterized by modest insulin resistance and impaired glucose tolerance (24). The α1 and α2 isoform sequences are 90% identical in the kinase domain (25), and there is significant overlap between the α isoforms with regard to substrate specificity, but their functional differences are likely, at least in part, attributable to differences in subcellular localization (26). Wang et al. reported that the CR-induced improvements in whole body insulin sensitivity and insulin-stimulated glucose uptake by skeletal muscle observed in wildtype mice were eliminated in AMPK-α2-null mice (19). This outcome was observed even though CR had similar effects on the reduction of body mass, fat pad mass, serum nonesterified fatty acids and triglycerides, and serum leptin for wildtype versus AMPK-α2-null mice.

The only previous study that evaluated the influence of CR on AMPK enzymatic activity in skeletal muscle reported no CR-related effects for either α1- or α2-associated AMPK activity in skeletal muscle of young adult mice (27). The AMPKγ3 isoform is selectively expressed in skeletal muscle, with undetectable expression in a large number of other tissues that have been tested (28). This characteristic is attractive for the purpose of creating interventions that specifically activate AMPK in skeletal muscle, a strategy that should reduce the negative side-effects that have been observed with global AMPK activators (eg, myocardial hypertrophy). Earlier research has demonstrated that physiological circumstances, including specific exercise protocols, can selectively activate AMPKγ3 in the absence of detectable increases in AMPKα1 or AMPKα2 activity (29). CR is known to elevate insulin-stimulated glucose uptake in skeletal muscle, and activation of AMPKγ3 has been linked to elevated glucose uptake by skeletal muscle (29,30). To date, the CR influence on AMPKγ3 activity in skeletal muscle has not been evaluated.

In addition to the limited knowledge about CR effects on AMPK in skeletal muscle during young adulthood, even less is known about CR effects late in life. AMPK Thr172 phosphorylation is the only AMPK-related outcome that has been previously reported in skeletal muscle from older rodents aged 21–30 months-old (11,13,15,17). A CR-induced increase in muscle AMPK Thr172 phosphorylation was observed in one of these 4 previous studies (17). Muscle ATP concentration was not different for 26-month-old CR versus AL rats (31). No earlier studies analyzing skeletal muscle from older animals have assessed CR effects on AMPKα1 activity, AMPKα2 activity, AMPKγ3 activity, AMP concentration, or the phosphorylation of any AMPK substrate. The lack of information about phosphorylation of AMPK substrates in the muscle of old rats is especially notable because AMPK’s functional consequences relies on its ability to phosphorylate other proteins, and greater phosphorylation of protein substrates can potentially persist after the reversal of elevated AMPK phosphorylation and/or activity. Therefore, this study aimed to fill some of the important gaps in knowledge related to CR’s influence on AMPK in the skeletal muscle of older animals by evaluating the skeletal muscle of 24-month-old male Fischer-344 × Brown Norway F1 AL and CR rats: (a) AMPK Thr172 phosphorylation; (b) AMPKα1-, AMPKα2-, and AMPKγ3-associated activity; (c) phosphorylation of multiple protein substrates (ACC, Beclin-1, ULK1, Raptor, TBC1D1, and TBC1D4 (also known as Akt substrate of 160 kDa, AS160); and (d) ATP and AMP concentrations.

We studied the epitrochlearis muscle that has a fiber type profile (8% type I, 13% type IIA, 51% type IIB, and 28% type IIX) that is roughly similar to the average fiber type composition of 76 different rat skeletal muscles or muscle portions (32,33). We previously found that CR had effects on glucose uptake and several signaling proteins (including Akt and TBC1D4) in the epitrochlearis that were similar to the effects on several other skeletal muscles, including the tibialis anterior, gastrocnemius, and plantaris (34,35). In addition, we previously found that 8 weeks of CR was sufficient to induce metabolic effects, including enhanced stimulation of insulin signaling proteins (including Akt and TBC1D4) and glucose uptake by the epitrochlearis of 24.5-month-old male Fischer-344 × Brown Norway F1 rats (36). We hypothesized that skeletal muscles from CR versus AL rats would have greater AMPK Thr172 phosphorylation, AMPKγ3 activity, and phosphorylation of AMPK multiple substrates without diet-related differences for AMPKα1 activity, AMPKα2 activity, ATP concentration, or AMP concentration.

Method

Materials

Chemicals were obtained from Sigma–Aldrich (St. Louis, MO) or Fisher Scientific (Hanover Park, IL) unless otherwise noted. The reagents and apparatus for SDS–PAGE and nonfat dry milk (no. 170-6404) were from Bio-Rad (Hercules, CA). Pierce MemCode Reversible Protein Stain Kit (#24585), bicinchoninic acid protein assay (#23225), and Tissue Protein Extraction Reagent (T-PER; #78510) were from Thermo Fisher Scientific (Waltham, MA). Custom-made AMARA peptide was purchased from Schafer-N Aps (Copenhagen, Denmark). p81 Ion exchange cellulose chromatography paper was purchased from Saint Vincent’s Institute of Medical Research (Melbourne, Australia). ³³P γ-labeled ATP (SCF-301) was purchased from Hartmann Analytic (Braunschweig, Germany). The phosphomotifs on the proteins that were evaluated by immunoblotting are highly conserved between rats and humans. The designation of phosphorylated amino acid residues that were detected by antibodies refers to the relevant human protein sequence to correspond with the terminology used by the vendors. Anti-phospho AMPKα Thr¹⁷² (pAMPKα ^Thr172; #50081, which recognizes phosphorylation on both α1 and α2 isoforms), anti-AMPK-α (AMPKα; #5831, which recognizes both α1 and α2 isoforms), anti-acetyl-CoA carboxylase (ACC; #3676), anti-phospho ACC^Ser79 (pACC^Ser79; #3661), anti-phospho ULK1 Ser⁵⁵⁵ (pULK^Ser555; #5869), anti-ULK1 (ULK1; #8054), anti-phospho Raptor Ser⁷⁹² (pRaptor^Ser792; #2083), anti-Raptor (Raptor; #2280), anti-phospho Beclin-1 Ser⁹³ (pBeclin-1^Ser93; #14717), anti-Beclin-1(Beclin-1; #3495), anti-TBC1D1 (TBC1D1; #4629), and anti-rabbit IgG horseradish peroxidase (HRP) conjugate (#7074) were from Cell Signaling Technology (CST; Danvers, MA). CST antibody #3676 recognizes both ACC1 and ACC2 (37), and CST antibody #3661 recognizes the highly conserved AMPK phosphomotif that is found on both ACC1^Ser79 and ACC2^Ser212 (38). Anti-phospho TBC1D4 Ser⁷⁰⁴ (pTBC1D4^Ser704) was custom made by Capra Science (Angelholm, Sweden) and provided by Dr. Jonas Thue Treebak. Anti-AMP-activated protein kinase γ3 (γ3-AMPK) used for immunoblotting was provided by Dr. David Thomson (Brigham Young University, Provo, UT) (39). Anti-Akt substrate of 160 kDa (AS160, also known as TBC1D4; #ABS54), anti-phospho TBC1D1 Ser²³⁷ (pTBC1D1^Ser237; #07-2268), and enhanced chemiluminescence Luminata Forte Western HRP Substrate (#WBLUF0100) were from EMD Millipore (Billerica, MA). The AMPKα1 antibody used for the AMPKα1 assay was custom made by GenScript (Piscataway, NJ). The AMPKα2 and AMPKγ3 antibodies used for AMPKα2 and AMPKγ3 activity assays, respectively, were custom made by Yenzym Antibodies, LLC (Brisbane, CA). Protein G agarose used for AMPK activity assays was purchased from Merck Life Sciences A/S (Copenhagen, Denmark). Supplementary Table 1 provides additional information on the antibodies used (dilutions and validation).

Animal Treatment

Animal care procedures were approved by the University of Michigan Committee on Use and Care of Animals. Male Fischer-344 × Brown Norway F1 generation rats, obtained at 21–22 months of age from the National Institute of Aging Aged Rat Colony, were individually housed in shoebox cages (12:12 h light–dark cycle; lights out at 17:00 h). After 1 week of familiarization in the University of Michigan facility with AL access to food (NIH31 chow), rats were randomly assigned to either AL or CR groups. Before the CR protocol, the body mass of the groups were not significantly different (AL = 539.9 ± 35.7 g; CR = 547.8 ± 35.4 g). For a period of 8 weeks, the CR group received NIH31/NIA fortified chow in a single daily allotment (at 16:00 h) that equaled 65% of the average daily intake of the AL group. The NIH31/NIA fortified chow was supplemented with vitamins so that the CR animals consumed vitamins at levels comparable to AL rats. Each week, the food intake of AL rats was determined based on the difference between the total amount of food provided and the amount of uneaten food. The daily food allotment for the CR rats, which equaled 65% of the mean daily food intake from the AL rats from the previous week, was adjusted on a weekly basis. For CR rats, in the very rare event that a portion of their daily allocation was uneaten, the uneaten food was weighed and discarded. Their daily intake was calculated as the difference between the amount of food provided and the amount of any uneaten food for that animal. Rats were weighed on a weekly basis between 15:00 and 16:00 h.

Tissue Dissection

The terminal experiment was performed when the animals were 24 months old. On the day of tissue dissection, rats were deeply anesthetized (intraperitoneal injection of 50 mg/kg ketamine/5 mg/kg xylazine) between 11:00 and 12:00 h without removing access to food prior to anesthesia. The rationale for not removing access to food before the terminal experiment was to enable muscle collection under the usual feeding conditions for the AL and CR rats. As expected, none of the rats in the CR group had food remaining when they were anesthetized. The skin from each forelimb was dissected away, and epitrochlearis muscles were carefully dissected out using extra-fine curved microdissecting scissors and rapidly freeze-clamped using aluminum tongs cooled in liquid nitrogen, and stored (−80°C) until subsequent processing and analysis.

Muscle Processing and Analysis for Immunoblotting

Frozen muscles were weighed and then homogenized with 1 mL of ice-cold lysis buffer with a glass pestle attached to a motorized homogenizer (Caframo, Georgian Bluffs, ON, Canada). Muscle lysates used for immunoblotting were prepared with a lysis buffer that contained T-PER supplemented with 1 mM EDTA, 1 mM EGTA, 2.5 mM sodium pyrophosphate, 1 mM sodium vanadate, 1 mM β-glycerophosphate, 1 µg/mL leupeptin, and 1 mM phenylmethylsulfonyl fluoride. Homogenates were rotated for 1 hour at 4°C before centrifugation (15 000g for 15 min at 4°C). The supernatants were transferred to microfuge tubes and stored at −80°C until subsequent analyses. Protein concentration was measured with the bicinchoninic acid procedure.

An equal amount of protein from each lysate was combined with 6× Laemmli buffer, boiled (95°C, 5 minutes), subjected to SDS–PAGE, and transferred to polyvinylidene difluoride (PVDF) membranes. MemCode protein stain was used to confirm equal loading. PVDF membranes were blocked using Tris-buffered saline pH 7.5 plus 0.1% Tween 20 (TBST) with either 5% nonfat milk or 5% BSA (60 minutes, room temperature), incubated with appropriate antibody titers, underwent enhanced chemiluminescence, and quantified using densitometry (AlphaView; ProteinSimple, San Jose, CA). Sample values (densitometric units) were determined relative to the mean of all the samples on the blot. These normalized values were divided by the respective MemCode loading control value for each sample (with individual sample MemCode values normalized by dividing the mean MemCode values for all samples on the blot). Phosphorylated proteins values were expressed as a ratio of phosphorylated protein to total protein (determined for each sample using a separate immunoblot probed with a primary antibody for the relevant total protein).

Muscle Processing and Analysis for AMPK Activity

Muscle processing and AMPK activity were determined as described (40). For measurement of AMPK activity, the epitrochlearis muscles were powdered in liquid nitrogen and weighed. The frozen muscle powder was homogenized 1:20 in ice cold MG-buffer (50 mM HEPES [pH 7.5], 10% glycerol, 20 mM sodium pyrophosphate, 150 mM NaCl, 1% NP-40, 20 mM β-glycerophosphate, 10 mM sodium fluoride (NaF), 2 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, 1 mM EGTA, 10 µg/mL aprotinin, 10 µg/mL leupeptin, 2 mM sodium vanadate, 3 mM benzamidine [pH 7.5]) using a Restch TissueLyser II (Qiagen, Germany) shaking 2 × 1 minute at 30 Hz. The homogenates were rotated at 4°C for 1 hour and then centrifuged (13 000 g for 20 minutes at 4°C). The supernatants were transferred to microfuge tubes and stored at −80°C until subsequent analyses. Protein concentration was measured with the bicinchoninic acid procedure.

AMPKγ3 immunoprecipitation (IP) was performed using 150 µg lysate overnight at 4°C with 20 µL Protein G agarose beads (50:50 slurry) and 2 µg of γ3 specific antibody. IP-buffer (20 mM Tris-base [pH 7.4], 50 mM NaCl, 1% [v/v] Triton X-100, 50 mM NaF, 5 mM sodium pyrophosphate, 500 μM phenylmethylsulfonyl fluoride, 2 mM dithiothreitol [DTT], 4 μg/mL leupeptin, 50 μg/mL soybean trypsin inhibitor, 6 mM benzamidine, and 250 mM sucrose) was added to a total volume of 200 µL. After IP the supernatant was transferred to a new tube with Protein G agarose and AMPKα2 antibody for a sequential IP of the remaining α2βγ1 complexes and again incubated overnight with rotation. The agarose beads were kept on ice with the IP of γ3 complexes and were washed once with IP-buffer, then in 6× Assay-buffer (480 mM NaCl and 240 mM HEPES, pH 7.0) and finally twice in 3× Assay-buffer (diluted 1:1 in water). After washing, 30 µL of kinase-reaction-buffer (40 mM HEPES [pH 7.0], 80 mM NaCl, 833 µM DTT, 200 µM AMP, 100 µM AMARA peptide [H-AMARAASAAALARRR-OH], 5 mM magnesium chloride, 200 µM ATP, and 67 kBq/sample ³³P γ-labeled ATP) was added, and after 30 minutes at 30°C, the reaction was stopped by adding 10 µL 1% phosphoric acid. Aliquots of slurry (6 µL) were spotted in duplicate onto P81 Chromatography paper, which was then washed thrice in 1% phosphoric acid. The dried paper was analyzed using a Typhoon FLA 7000IP PhosphorImager (GE Healthcare, Denmark), while the kinase-reaction-buffer was additionally analyzed with liquid scintillation counting (Tri-Carb 4910TR, PerkinElmer, Denmark). Following the AMPKα2 IP, the final sequential IP was performed with the AMPKα1 antibody as described earlier using the AMPKα1 antibody.

Muscle Processing and Analysis for ATP and AMP Concentrations

A portion of frozen epitrochlearis muscle was weighed and transferred into microtubes, and 100 µL of water was added to each sample, and the samples were probe-sonicated. Methanol:Water solution (8:2, 1 000 μL) containing internal standards (including either ¹³C–labeled ATP or ¹³C–labeled AMP) was added to each sample to enable quantitation. Microtubes were vortexed, and then incubated (4°C for 10 minutes). Samples were vortexed a second time, and then centrifuged (14 000 RPM for 10 minutes at 4°C). Sample extracts were dried in an autosampler vial to provide the equivalent of a 20 mg/mL sample. Prior to analysis, samples were reconstituted in 100 µL of the water/methanol solution (8:2) and vortexed. Analysis was performed using an Agilent system consisting of an Infinity Lab II UPLC coupled with a 6545 QT of the mass spectrometer (Agilent Technologies, Santa Clara, CA) using a JetStream ESI source in negative mode.

Statistical Analysis

AL and CR groups were compared using a Student’s t test. Data are presented as mean ± standard deviation (SD). Pearson correlation analysis was used to evaluate linear associations between 2 variables. A p value ≤.05 was considered statistically significant.

Results

Body Mass and Muscle Mass

Rats were randomly assigned to AL (n = 10) or CR (n = 10) groups before initiating the 8 weeks dietary protocol. During the third week of the protocol, one of the rats assigned to the CR group was euthanized after developing a large tumor, and the remaining rats (AL, n = 10; CR, n = 9) completed the dietary protocol. Body mass was measured weekly, with significantly lower values for CR versus AL rats for weeks 2–8 (Supplementary Figure 1). At the end of the dietary protocol, values of AL rats exceeded values of CR rats for body mass (524.6 ± 33.5 vs 406.4 ± 22.4, p < .001), and epitrochlearis muscle mass (134.5 ± 24.7 vs 110.7 ± 18.5 mg, p < .001).

Epitrochlearis Muscle AMPK Activity, Muscle Protein Abundance, and Phosphorylation

No significant diet-related differences were found for AMPKα1 activity, AMPKα2 activity, or AMPKγ3 activity (Figure 1). AMPK activity was determined using a standard procedure, that is, homogenized samples were incubated with a peptide substrate that contains the AMPK phosphorylation motif and AMPK’s key allosteric regulators (ATP and AMP) at standard concentrations that were identical between AL and CR samples. Thus, the results argue against a diet-related change in AMPK’s enzymatic activity under these highly controlled conditions.

Figure 1.

(A) AMPKα1 activity, (B) AMPKα2 activity, and (C) AMPKγ3 activity in rat epitrochlearis muscle. No significant differences were observed. Values (expressed as pmol · min⁻¹ · mg⁻¹) are shown as mean ± SD (n = 9–10/group). AMPK = AMP-activated protein kinase; SD = standard deviation.

No diet-related differences were detected for a total abundance of most of the proteins that were evaluated, including ACC1/2, Raptor, ULK1, TBC1D1, or TBC1D4 (Table 1). However, total protein abundance was modestly, but significantly different between diet groups for AMPKα1/2 (AL > CR, p < .005), Beclin-1 (CR > AL, p < .05), and AMPKγ3 (AL > CR, p < .005; Table 1; Supplementary Figure 2A–E). The CR-related decrease in AMPKγ3 abundance was very similar when determined with the AMPKγ3 antibody that was used to immunoprecipitate and determine AMPKγ3 activity (AL = 1.19 ± 0.19, CR = 0.79 ± 0.14, p < .005).

Table 1.

Total Abundance of Proteins in Epitrochlearis Muscle of AL and CR Rats

Protein	AL	CR	p Value
AMP-activated protein kinase (AMPK), α1/2	1.09 ± 0.07	0.91 ± 0.16	<.005
Acetyl-CoA carboxylase 1/2 (ACC1/2)	1.02 ± 0.10	0.98 ± 0.11	.435
Beclin-1	0.92 ± 0.17	1.09 ± 0.15	<.05
TBC1D1	1.05 ± 0.25	0.89 ± 0.14	.057
TBC1D4	1.08 ± 0.34	0.90 ± 0.36	.283
ULK1	1.04 ± 0.19	0.95 ± 0.15	.098
Raptor	1.00 ± 0.07	1.01 ± 0.09	.792
γ3 AMPK	1.16 ± 0.18	0.81 ± 0.22	<.005

Notes: The table summarizes the total abundance of proteins as determined by immunoblot. Values are expressed in relative units relative to total protein (determined using MemCode total protein stain) as mean ± SD (n = 9–10 per group). p Values are for a Student’s t test. AL = ad libitum; CR = caloric restriction; SD = standard deviation.

Muscles from CR compared to AL rats had significantly greater values for the phosphorylation protein expressed as a ratio with total protein for: (pAMPK^Thr172/AMPKα1/2, pACC1/2^Ser79/Ser212/ACC1/2, pBeclin-1^Ser93/Beclin-1, pTBC1D1^Ser237/TBC1D1, and pTBC1D4^Ser704/TBC1D4), with no significant diet-related differences for pRaptor^Ser792 or pULK1^Ser555 (Figure 2; Supplementary Figure 2A–E). For most of the phosphorylated proteins that were evaluated, the presence or absence of a significant diet-related difference was unaltered whether or not the phosphorylated values were expressed as a ratio using total protein abundance. Significant differences (CR > AL) were identified for phosphorylated proteins that were not expressed as a ratio with total protein for pAMPK^Thr172 (p < .005), pACC1/2^Ser79/Ser212 (p < .001), pBeclin-1^Ser93 (p < .005), and pTBC1D4^Ser704 (p < .05), but not for pTBC1D1^Ser237, pRaptor^Ser792, or pULK1^Ser555 (not shown). Thus, pTBC1D1^Ser237 was the only AMPK phosphorylation site for which the statistical comparison between diet groups was changed when the values of the phospho-protein were expressed as a ratio of phospho-to-total protein. Significant, positive correlations were detected for pAMPK^Thr172/AMPK with pACC1/2^Ser79/Ser212, pTBC1D1^Ser237/TBC1D1, pTBC1D4^Ser704/TBC1D4, pBeclin-1^Ser93/Beclin-1, and pULK1^Ser555/ULK1, but not with pRaptor^Ser792/Raptor (Figure 3).

Figure 2.

Immunoblot results for (A) pAMPK^Thr172/AMPKα1/2, (B) pACC1/2^Ser79/Ser212/ACC1/2, (C) pTBC1D1^Ser237/TBC1D1, (D) pTBC1D4^Ser704/TBC1D4, (E) pBeclin-1^Ser93/Beclin-1, (F) pULK1^Ser555/ULK1, and (G) pRaptor^Ser792/Raptor in rat epitrochlearis muscle. *CR exceeded AL (p < .001); ^†CR exceeded AL (p < .005); ^‡CR exceeded AL (p < .01). Values are shown as mean ± SD (n = 9–10/group). Representative immunoblots are shown. AMPK = AMP-activated protein kinase; SD = standard deviation; AL = ad libitum; CR = caloric restriction.

Figure 3.

Correlations for pAMPK^Thr172 versus phosphorylated AMPK substrates in rat epitrochlearis muscle. (A) pACC1/2^Ser79/Ser212, (B) pTBC1D1^Ser237/TBC1D1, (C) pTBC1D4^Ser704/TBC1D4, (D) pBeclin-1^Ser93/Beclin-1, (E) pULK1^Ser555/ULK1, and (F) pRaptor^Ser792/Raptor. Values for AL muscles are represented by circles (○). Values for CR muscles are represented by triangles (Δ). Pearson’s correlation was used for statistical analysis. AL = ad libitum; CR = caloric restriction.

Epitrochlearis Muscle ATP and AMP Concentrations

Neither ATP nor AMP concentration was significantly different in the muscles from AL rats compared to CR rats (Figure 4). There was also no significant difference in the AMP:ATP ratio in muscles from AL (0.32 ± 0.12) versus CR (0.36 ± 0.19) rats.

Figure 4.

(A) ATP concentration (μmol/g) and (B) AMP concentration (nmol/g) in rat epitrochlearis muscle. No significant differences were observed. Values are shown as mean ± SD (n = 9–10/group). AL = ad libitum; CR = caloric restriction.

Discussion

The results supported most aspects of our hypothesis. As hypothesized, there were significant CR-associated increases in the phosphorylation of AMPK^Thr172 and phosphorylation of four of the 6 AMPK substrates that were evaluated, without alterations in AMPKα1 activity, AMPKα2 activity, or concentrations of AMP and ATP. However, the hypothesis was not completely supported, as no diet-related difference was detected for AMPKγ3 activity.

Muscle pAMPK^Thr172 was significantly greater for CR compared to AL rats, consistent with the results of several previous studies in skeletal muscle from mice and rats (17–20). However, these results differ from the findings of some other studies that did not detect a CR-related increase in pAMPK^Thr172 determined for skeletal muscle from mice or rats (8–16). To what extent can the differing outcomes for pAMPK^Thr172 be attributed to differences in the ages of the animals, duration of CR, or relative magnitude of CR? The studies that reported a CR-related increase in pAMPK^Thr172, including the current study, used rodents ranging from ~4 to 24 months-old (17–20), and the studies that did not detect this CR-related change included rodents from ~2.5 to 30 months (8–16). The duration of CR in the studies that reported a CR-related increase in pAMPK^Thr172 ranged from ~2 to 19 months (17–20), and the studies that did not detect a CR-related change used CR protocols ranging from 20 days to 27 months (8–16). The studies that reported a CR-related increase in pAMPK^Thr172 used CR animals that consumed 60%–70% of the AL intake (17–20), and the studies that did not detect a CR-related change used CR animals that consumed 60%–80% of AL intake (8–16). Although age, duration of CR, and relative extent of CR are important determinants of the consequences of CR, the available information does not provide a simple explanation for the lack of uniform results. Rigorous analysis of the roles of these factors will require a detailed and systematic experiment that includes a range of ages of animals and CR protocols with a range of durations and relative reductions in calorie intake.

Rats that are given AL food access will eat throughout the day, whereas rats that are restricted to 60%–70% of their AL intake typically eat most of their food within approximately 2–3 hours of receiving their daily food allotment (41). In the current study, a longer period of time elapsed between the final food consumption and the skeletal muscle collection for the CR rats versus the AL rats. Therefore, it is possible the diet-related differences in the timing of food intake played a role in the differences between AL and CR rats. Future research in which the timing of food intake is matched between the AL and CR groups will be needed to isolate the influence of differences in amount versus timing of food intake. Notably, Dai et al. (20), who fasted rats (~10 months-old) for 12 hours prior to sampling skeletal muscles, reported greater muscle pAMPK^Thr172 for CR (consuming 60% of AL intake for 6 months) versus AL animals.

Enhanced AMPK activation in skeletal muscle is a hallmark of vigorous exercise (29). Daily activity is modestly greater for CR versus AL rats, with peak physical activity of CR rats occurring slightly before and during their feeding period (41,42). Is it possible that differences in physical activity played a role in the diet-related differences for muscle pAMPK^Thr172? An increase in muscle pAMPK^Thr172 is typically evident immediately after vigorous exercise, and most or all of this effect is often reversed within 3–4 hours postexercise (43). Greater γ3-associated AMPK activity has also been reported immediately postexercise, but this effect has been reported to be sustained at 3–4 hours after completion of vigorous exercise (44). This postexercise pattern differs from the results in the current study (elevated muscle pAMPK^Thr172 without elevated γ3-AMPK activity for CR vs AL rats), providing evidence against a CR-related increase in the physical activity being primarily responsible for elevated muscle pAMPK^Thr172.

What other factors might account for the differing results? Several of the studies that did not report a CR-effect on pAMPK^Thr172 made the analysis in isolated muscles that had undergone 50 minutes of ex vivo incubation (8,9,15), whereas none of the studies that detected a CR-related increase in pAMPK^Thr172 analyzed muscles after ex vivo incubation. Putative differences between AL and CR muscles may be sufficiently labile to be lost after ex vivo incubation. In addition, none of the studies that analyzed the predominantly slow-twitch (type I), antigravity soleus muscle detected a CR-effect on pAMPK^Thr172 (10,15,16). Although differences in experimental design are evident among some of the studies with differing results, there is not a single, obvious explanation to reconcile the lack of a consistently reported CR-effect on muscle pAMPK^Thr172 for all of the studies.

AMPK’s wide range of biological effects is related to its ability to phosphorylate a large number of protein substrates. However, earlier research had only examined CR’s effect on the phosphorylation of a single AMPK substrate, ACC. Therefore, it was an important new finding that muscles from CR versus AL rats had greater values for the phosphorylation of 4 AMPK substrates (pACC1/2^Ser79/Ser212, pBeclin-1^Ser93, pTBC1D1^Ser237, and pTBC1D4^Ser704), with no significant diet-effect on the phosphorylation of the other 2 AMPK substrates that were evaluated (pRaptor^Ser792 and pULK1^Ser555). For each of the phosphorylated substrates with a significant, CR-related increase in phosphorylation, there was a significant, positive correlation with AMPK Thr172 phosphorylation. There was also a modest, but significant positive correlation between ULK1 Ser555 phosphorylation, which was not significantly increased with CR, and AMPK phosphorylation. Although there is a well-known biological relationship between pAMPK^Thr172 and AMPK’s ability to phosphorylate its substrates (3), correlations cannot establish causality. The low to moderate values for the correlations (R² of 0.24–0.53) suggest that factors other than pAMPK^Thr172 were likely important contributors to the extent of AMPK substrate phosphorylation. The observation that CR did not uniformly elevate the phosphorylation of each AMPK substrate is reminiscent of findings reported for other AMPK-stimulating treatments, including exercise, or incubation with AICAR (an AMPK activating compound) (21). The substrate-selectivity in the response to a given stimulus is likely caused by a variety of factors, including the subcellular colocalization of a given substrate with a particular AMPK heterotrimer, altered susceptibility for substrate phosphorylation secondary to covalent binding with regulatory proteins, the influence of various posttranslational modifications on the substrate, and the influence of protein phosphatases. Virtually nothing is known about the effects of CR on any of these factors.

What are the potential functional consequences of the CR-related increase in phosphorylation of these 4 AMPK substrates? Increased pTBC1D1^Ser237 is linked to elevated insulin-independent glucose uptake by skeletal muscle (45). TBC1D1 is a Rab-GTPase protein that, when not phosphorylated on Ser237, favors more GDP-bound Rab proteins. This, in turn, restrains GLUT4 vesicle exocytosis and glucose uptake (46). Results using TBC1D1^Ser231Ala knockin mice (Ser231 in mice is equivalent to Ser237 in humans) revealed that preventing TBC1D1 phosphorylation on this site resulted in lesser elevation of glucose uptake in response to AICAR, a chemical AMPK activator (47). Accordingly, increased pTBC1D1^Ser237/TBC1D1 would be expected to favor AMPK-dependent and insulin-independent glucose uptake. We previously reported that in vivo glucose uptake by rat epitrochlearis muscle was significantly greater for CR compared to AL rats when these measurements were made in the absence of exogenous insulin (48).

We previously found that in vivo insulin-stimulated glucose uptake by the epitrochlearis muscle is greater for CR versus AL rats (34), and TBC1D4 is crucial for regulating insulin-dependent glucose uptake in skeletal muscle. However, phosphorylation of TBC1D4^Ser704 does not appear to be essential for insulin-stimulated glucose uptake (49). There is evidence that TBC1D4 phosphorylation on Ser704 may favor a greater insulin-stimulated increase in pTBC1D4^Thr642 (30). Increased pTBC1D4^Thr642 is important for the full-effect of insulin on glucose uptake, and CR has been shown to elevate insulin-stimulated pTBC1D4^Thr642 concomitant with increased insulin-stimulated glucose uptake (9,50,51). It is uncertain if site-selective TBC1D4 phosphorylation or other AMPK substrates is required for CR-mediated improvement of insulin-stimulated glucose uptake by muscle.

ACC catalyzes the synthesis of malonyl-CoA via the carboxylation of acetyl-CoA, and malonyl-CoA inhibits long chain fatty acid oxidation. Two isoforms of ACC exist, with ACC2 being the predominant isoform expressed by skeletal muscle. Ser79 of ACC1 and Ser212 of ACC2 are equivalent motifs, and phosphorylation on each site inactivates the respective isoform. Therefore, greater Ser212 phosphorylation of ACC2 is expected to favor greater fat oxidation by skeletal muscle. This outcome corresponds to the previously reported results for a lower respiratory exchange ratio (indicative of greater fat oxidation) in CR compared to AL rats measured for CR compared to AL rats during the light portion of the day (52), which is when the muscles were collected in the current study.

Activation of AMPK is linked to the induction of autophagy, and evidence supports the idea that CR may promote autophagy in various tissues, including skeletal muscle (53). In this context it is interesting that Beclin-1 phosphorylation on Ser93, an AMPK motif that is implicated in the induction of autophagy (54), was greater for CR versus AL muscle. However, no CR-related difference was detected for either pRaptor^Ser792 or pULK1^Ser555, 2 other AMPK phosphosites that are also involved in autophagy induction. We previously evaluated the influence of CR on several signaling proteins that are involved in the regulation of autophagy, and found no difference for muscles from CR versus AL rats for pTSC2^Ser939 or p-mTOR^Ser2448 (35). Taken together, these results provide inconclusive evidence with regard to mechanisms that regulate the putative effects of CR on autophagy in skeletal muscle.

The observation of no CR-induced effect on either AMPKα1 or AMPKα2 activity is consistent with the findings of the only previous study that assessed the influence of CR on these outcomes in skeletal muscle of rodents (5-month-old CR mice consuming 65% of AL intake initiated at weaning) (27). The current results were unique because they focused on older animals, and they provided the first evidence that skeletal muscle AMPKγ3 activity is not different between CR and AL groups. The unaltered AMPKγ3 activity in muscles from CR versus AL rats was accompanied by a significantly lower AMPKγ3 subunit protein abundance in CR versus AL rats, which is reminiscent of published results for reduced abundance of this subunit in chronically exercise-trained compared to untrained human skeletal muscle (29). However, with exercise training versus untrained human muscle, AMPKγ3 activity is reduced. It seems likely that the factors regulating AMPK subunit protein abundance and activity will be distinct for CR compared to chronic exercise training.

What might account for CR-related enhancement of pAMPK^Thr172 and phosphorylation of several AMPK substrates concomitant with no detectable CR-related increase in AMPK activity? It should be noted that AMPK activity was determined with an ex vivo assay using muscle lysates incubated with a peptide substrate that contains the AMPK phosphorylation motif along with the most important allosteric regulators of AMPK (ATP and AMP) at standard concentrations that were identical between AL and CR samples (40). In contrast, pAMPK^Thr172 in the current study was influenced by endogenous levels of these molecules. The CR-effect on pAMPK^Thr172 was not attributable to a diet-related difference in either muscle ATP or AMP concentration in the current study. This study was the first to compare AMP concentration in skeletal muscles from AL versus CR animals. The lack of diet-related difference in muscle ATP concentration is consistent with the results of an earlier study that reported gastrocnemius ATP concentration was not different for 26-month-old AL compared to age-matched CR rats (40% below AL intake beginning at the age of 4 months-old) rats (31). Other regulatory mechanisms can also influence AMPK activity. For example, the binding of fructose-1,6-bisphosphate (FBP) to aldolase interferes with Thr172 phosphorylation, and thus AMPK activation, by an AMP-independent mechanism that involves the formation of a complex between AMPK and lysosomal-localized proteins (55). Lower FBP concentration would favor greater AMPK phosphorylation. Interestingly, we previously reported that skeletal muscle concentration of glucose 6-phosphate, a precursor for FBP, is significantly lower for CR versus AL rats (48). However, the influence of CR on muscle FBP concentration has not been reported. FBP and/or subcellular localization of AMPK could potentially differ between muscles of CR versus AL rats, but these physiological regulators would be irrelevant in the ex vivo AMPK activity assays. The spatiotemporal regulation of the phosphorylation of diverse AMPK substrates is influenced by the compartmentalization of multiple molecules, including AMPK heterotrimers, AMPK substrates, AMPK kinases, AMPK phosphatases, and allosteric regulators (1). The potential influence of CR on these complex regulatory factors remains to be determined.

In conclusion, the current study provided the most comprehensive analysis of CR effects on AMPK in skeletal muscle to date. The results revealed that CR can lead to greater phosphorylation of skeletal muscle AMPK and multiple AMPK phosphates in the absence of altered muscle ATP or AMP concentration, and without significant CR-induced effects on AMPKα1, AMPKα2, or AMPKγ3 activity determined in skeletal muscle lysates. Future research will be required to identify the mechanisms responsible for the CR-related increase in skeletal muscle pAMPK^Thr172 and phosphorylation of selected AMPK substrates.

Funding

This work was supported by the National Institutes of Health (R01AG010026 to G.D.C.), the Danish Council for Independent Research (FSS: 8020-002888 to J.F.P.W.), and the Novo Nordisk Foundation (NNF21OC0070370 to J.F.P.W.).

Conflict of Interest

J.F.P.W. has ongoing collaborations with Pfizer Inc. and Novo Nordisk, Inc. that are unrelated to the current study. The other authors declare no conflict.

Author Contributions

Conceptualization: G.D.C.; Project administration: A.Z.; Investigation: A.Z., J.B.B., S.E.K., E.B.A., and D.T.; Writing—original draft: G.D.C.; Writing—critical review and editing: A.Z., J.B.B., S.E.K., E.B.A, D.T., and J.F.P.W.; Data curation: A.Z., J.B.B., and S.E.K.; Formal analysis: A.Z., S.E.K., and J.B.B.; Visualization: S.E.K.; Funding acquisition: G.D.C. and J.F.P.W.; Resources and supervision: G.D.C. and J.F.P.W. All authors reviewed and approved the final version of the manuscript.