Skeletal Muscle Triacylglycerol Hydrolysis Does Not Influence Metabolic Complications of Obesity

SMAKO mice were generated by crossing B6.129-Pnpla2^tm1Eek (ATGL-flox) mice with Myo-Cre mice (24). Ckm-ATGL Tg mice were generated by cloning murine ATGL in front of the Ckm promoter (25). B6.129-Pnpla2^tm1Eek, SMAKO, and Ckm-ATGL mice were generated as described in Figs. 2A and 5A and the Supplementary Data online. Male ATGL^flox/flox Cre/+ mice were mated to female ATGL^flox/flox+/+ mice to generate ATGL^flox/flox Cre/+ (SMAKO) and ATGL^flox/flox +/+ (control) mice. Male Ckm-ATGL Tg mice were mated to female wild-type (WT) mice to generate Tg and WT mice. Global ATGL knockout (ATGL^−/− or GAKO) mice have been previously reported (16,17). Mice were housed under standard conditions (25°C, 14:10-h light/dark cycle) with ad libitum access to chow (Prolab Isopro RMH 3000; 14 kcal% fat) or high-fat diet (HFD; Research Diets D12451i, 45 kcal% fat). Body composition, energy expenditure, and metabolic parameters were performed as described (26). Experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee and conducted in conformity with the Public Health Service Policy for Care and Use of Laboratory Animals.

DAGs (27), ceramides (27), and long-chain FA-CoAs (28) as well as TAG hydrolase activities (26) and radiolabeled palmitate oxidation (29) were determined in muscle homogenates and normalized to tissue wet weight or protein content.

Respirometry was performed in permeabilized myofiber bundles using an Oxygraph-2K (Oroboros) (30). Assays were run at 37°C in oxygen-saturated (∼150–220 μmol O₂) buffer (105 mmol K-MES, 30 mmol KCl, 10 mmol KH₂PO₄, 5 mmol MgCl₂, 5 mg/mL BSA, and 1 mmol EGTA [pH 7.4]) containing 25 μmol blebbistatin using the following titration protocol: 20 μmol palmitoylcarnitine, 2 mmol malate, 4 mmol ADP, 5 mmol glutamate, 10 mmol succinate, 10 μmol cytochrome c, 10 μg/mL oligomycin, and 2 μmol carbonylcyanide-p-trifluoromethoxyphenylhydrazone. Respiration rates were normalized to myofiber dry weight.

Insulin signaling studies were performed (17) using the following primary antibodies: anti-pAkt (pT308) (4056S; Cell Signaling Technology), anti-pAkt (pS473), and anti-Akt (total) (05-736 and 07-416; EMD Millipore). For protein expression under non–insulin-stimulated conditions, the following antibodies were used: MitoProfile Antibody Cocktail (MS604; Mitosciences), anti-HSL (4017s; Cell Signaling Technology), anti-pHSL (S565 and S660) (4137S and 4126S; Cell Signaling Technology), antiperilipint (Plin)2 (20R-AP002; Fitzgerald Industries), anti-Plin5 (03-GP31; American Research Products), anti-ATGL (2439; Cell Signaling Technology), and anti-CGI58 (NB110-41576; Novus Biologicals). For loading controls, the following antibodies were used: anti-Ran GTPase (610340; BD Biosciences) or β-actin (4970; Cell Signaling Technology). Visualization was performed with Immun-Star WesternC Chemiluminescent Kit in a VersaDoc System and quantified using Quantity One 1-D software (Bio-Rad).

Muscles were frozen in isopentane in OCT and cut into 10-μm sections. For muscle morphology, sections were stained with hematoxylin and eosin (H&E). Alternatively, sections were processed for neutral lipid content (using Oil Red O [ORO]) and/or expression of specific proteins (using ATGL or other antibodies noted above) either alone or in combination with fiber type–specific antibodies from the Developmental Studies Hybridoma Bank (A4.840 [1], A4.74 [2A], 6H1 [2X], 10F5 [2B]) followed by secondary antibodies (A21044 and A10035; Invitrogen) (31). At least four random areas were visualized (DM4000B; Leica Microsystems), digitally captured (Retiga-2000R; QImaging), and quantified (Northern Eclipse; Empix Imaging). For transmission electron microscopy (TEM), muscles were fixed in 2.5% glutaraldehyde, postfixed in 1% osmium tetroxide, dehydrated, embedded in epon, and evaluated by TEM (JEM-1011; JEOL).

RNA extraction, reverse transcription, and gene expression analysis were performed using an Eppendorf Realplex System in accordance with Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines (17,26).

Results are expressed as mean ± SEM. Comparisons were made by unpaired two-tailed Student t test or factorial ANOVA followed by determination of simple effects for pairwise comparisons. If indicated, comparisons were made by two-way ANOVA with repeated measures. For energy expenditure data, comparisons were made using generalized estimate equations. P values of <0.05 were considered statistically significant.

Fiber type–specific quantitative immunofluorescence (IF) revealed that endogenous ATGL was predominantly expressed in type 2A and 2X fibers with lower expression in type 1 fibers and minimal expression in type 2B fibers (Fig. 1A and C and Supplementary Fig. 1A and B). Conversely, global deletion of ATGL resulted in fiber type–specific accumulation of IMTGs in types 2A, 2X, and 1 fibers (Fig. 1B and Supplementary Fig. 1C). The distribution of these fibers differs among murine muscles (Supplementary Fig. 1D). Endogenous ATGL protein directly correlated with IMTG content (Fig. 1E) and also increased in parallel with IMTGs following HFD feeding (Fig. 1C–E). Thus, endogenous ATGL is expressed and functional in type 2A > 2X > 1 murine muscle fibers and increases with diet-induced obesity.

To reduce IMTG hydrolysis, SMAKO mice were generated by crossing ATGL-flox mice with Myo-Cre mice (24) (Fig. 2A). SMAKO mice had reduced ATGL mRNA (Fig. 2B) and protein (Fig. 2C) in skeletal but not cardiac muscle or other tissues. Accordingly, TAG hydrolase activity in skeletal muscle of SMAKO mice was dramatically reduced, could not be stimulated with the ATGL coactivator CGI-58, and was completely blunted in the presence of an HSL inhibitor (Fig. 2D). General muscle morphology did not differ between genotypes (Fig. 2E, top). Fiber type–specific quantitative IF of IMTG content, which is more sensitive and specific than whole-muscle biochemical IMTG analysis, revealed a dramatic fiber type–specific (2A > 2X > 1, but not 2B) increase in IMTGs in skeletal muscle of SMAKO mice that was comparable to GAKO mice (Fig. 2E and F). Cardiac muscle, in contrast, was unaffected (Supplementary Fig. 2). Lipidomic analysis of skeletal muscle (Fig. 2G) revealed no significant genotype effect on total DAGs, ceramides, or FA-CoAs. However, there was a trend toward increased total DAGs similar to GAKO mice (17). A detailed analysis of lipid classes in skeletal muscle revealed an effect of genotype on a few subspecies of DAGs (C14:0/C16:0-DAG, C14:0/C18:0-DAG, and C16:0/C18:0-DAG) and FA-CoAs (C14:0 and C16:0) (Supplementary Fig. 3). Finally, serum TAGs (Control-HFD 60 ± 5 vs. SMAKO-HFD 62 ± 5 mg/dL) and nonesterified FAs (Control-HFD 0.59 ± 0.05 vs. SMAKO-HFD 0.65 ± 0.06 mEq/L) did not differ between genotypes. Thus, decreasing IMTG hydrolysis via ATGL increases certain subspecies of DAGs and tends to increase total DAGs, but does not produce significant changes in other non-TAG intramyocellular lipids or systemic lipid homeostasis despite dramatic increases in IMTG content.

Skeletal muscle FA oxidation increased with HFD feeding but was unaffected by genotype (Fig. 3A). Likewise, mitochondrial respiration in isolated muscle fibers failed to identify any genotype effects in basal, substrate-stimulated, or uncoupled conditions (Fig. 3B). Consistent with these results, expression of mitochondrial oxidative phosphorylation proteins was not different between genotypes (Fig. 3C). Furthermore, imaging by TEM (Fig. 3D) and confocal microscopy (Supplementary Fig. 4) revealed no differences between genotypes for mitochondrial morphology or number, despite massive increases in lipid droplet size and number. Gene expression analysis revealed an increase in peroxisome proliferator–activated receptor (PPAR) α expression in chow- but not HFD-fed SMAKO mice (although threshold cycle was >35 using undiluted cDNA), but no effect of genotype on PPARδ, PPARγ coactivator (PGC) 1α, or PPAR target genes (Fig. 3E). ATGL mRNA expression was clearly decreased (>90%) in SMAKO muscle (Fig. 3F). However, expression of total HSL mRNA (Fig. 3F) and protein (Fig. 3G) as well as phosphorylation of HSL at Ser565 (AMPK target site, negative regulator of HSL action) and Ser660 (ERK target site, positive regulator of HSL action) (Fig. 3G) were unchanged. Likewise, no effects of genotype were identified on mRNA expression for genes involved in lipid uptake (FATP1 and CD36) or synthesis (DGAT1 and DGAT2). In contrast, CGI-58 and Plin2/Adrp mRNA (Fig. 3F) as well as CGI-58, Plin2/Adrp, and Plin5/Oxpat protein (Supplementary Fig. 5) were increased in SMAKO mice. However, these mechanisms were clearly inadequate to compensate for loss of ATGL. Thus, reduced TAG hydrolysis in skeletal muscle does not influence lipid oxidation or mitochondrial phenotypes.

As expected, HFD feeding increased adiposity (Fig. 4A and B), altered energy substrate utilization (Fig. 4E), and impaired both glucose tolerance (Fig. 4F) and insulin sensitivity (Fig. 4G–K). In contrast, there were no effects of genotype on body weight (Fig. 4A), fat mass (Fig. 4B), lean mass (Fig. 4C), individual muscle weights (Fig. 4D), nonmuscle tissue weights (data not shown), or total energy intake and expenditure (data not shown). Although alterations in IMTG metabolism could influence metabolic flexibility without affecting overall energy homeostasis, energy utilization (respiratory exchange ratio) during ad lib feeding, fasting, and refeeding transitions was likewise unchanged between genotypes (Fig. 4E). In addition, no differences in serum glucose were identified between genotypes following a physiological (4-h) fast (Fig. 4G, time 0 of insulin tolerance test [ITT]), a prolonged (12-h) fast (Fig. 4F, time 0 of glucose tolerance test [GTT]), an intraperitoneal glucose challenge (Fig. 4F, GTT), or an intraperitoneal insulin challenge (Fig. 4G, ITT). Consistent with these findings, diet but not genotype effects were identified in skeletal muscle–specific insulin-stimulated phosphorylation of key proteins in the insulin signaling cascade including insulin receptor substrate 1 (data not shown) and Akt (Akt pS⁴⁷³ or Akt pT³⁰⁸) (Fig. 4H–K). Thus, decreasing IMTG hydrolysis via ATGL deletion does not influence systemic energy homeostasis in the setting of acute or chronic nutritional challenges, nor does it influence systemic or muscle-specific glucose homeostasis and insulin action.

To increase IMTG hydrolysis, we generated skeletal muscle–specific Ckm-ATGL Tg mice (Fig. 5A). Tg founder lines had increased ATGL mRNA (Fig. 5B) and protein (Fig. 5C) in skeletal but not cardiac muscle or other tissues. Notably, skeletal muscle ATGL mRNA expression in Tg mice was comparable to endogenous ATGL expression in adipose tissue (32). IF analysis confirmed an increase in ATGL protein expression within all skeletal muscle fiber types of Tg mice (Fig. 5D and E). Thus, ATGL transgene expression overlaps with endogenous ATGL expression. Tg muscle TAG hydrolase activity also increased in an ATGL-specific manner since it was increased in the presence of an HSL inhibitor (Fig. 5F). General muscle morphology did not differ between genotypes (Fig. 5G, top). However, IMTGs were dramatically decreased in Tg mice with the greatest reduction in type 2A and 2X fibers (Fig. 5G and H). Lipidomic analysis of skeletal muscle (Fig. 5I) revealed no genotype effects on total DAGs, ceramides, or FA-CoAs. Furthermore, there was an effect of diet but not genotype on multiple subspecies of DAGs, ceramides, and FA-CoAs, with the only exceptions being reduced C16:0 ceramide and sphingosine (Supplementary Fig. 6). Furthermore, serum concentrations of TAGs (WT-HFD 122 ± 12 vs. Tg-HFD 126 ± 12 mg/dL) and nonesterified FAs (WT-HFD 0.62 ± 0.05 vs. Tg-HFD 0.59 ± 0.07 mEq/L) did not differ between genotypes. Thus, increasing IMTG hydrolysis via ATGL overexpression does not produce major changes in non-TAG intramyocellular lipids or systemic lipid homeostasis despite dramatic reductions in IMTG content.

As with SMAKO mice, FA oxidation was increased by HFD feeding but was unaffected by genotype (Fig. 6A). Likewise, mitochondrial respiration in isolated muscle fibers (Fig. 6B) and expression of mitochondrial proteins (Fig. 6C) did not differ between genotypes. Furthermore, TEM revealed no differences between genotypes for mitochondrial morphology or number, despite a clear reduction in lipid droplet size and number (Fig. 6D). Gene expression analysis revealed diet but not genotype effects on PPARα, PPARδ, PGC1α, and their target genes (Fig. 6E). ATGL mRNA expression was clearly increased (>12-fold) in skeletal muscle of Tg mice (Fig. 6F). However, expression of total HSL mRNA (Fig. 6F) and protein (Fig. 6G) as well as phosphorylation of HSL at Ser565 and Ser660 (Fig. 6G) were unchanged in Tg mice. Likewise, expression of other genes involved in lipid uptake, lipid synthesis, and/or modulation of ATGL action were not affected by genotype. Thus, in striking contrast to adipocyte-specific overexpression of ATGL (23), increasing IMTG hydrolysis via ATGL overexpression has minimal to no effect on other lipid metabolic pathways or mitochondrial phenotypes.

Again, the expected effects of HFD feeding were noted. However, there were no effects of genotype on body weight (Fig. 7A), fat mass (Fig. 7B), lean mass (Fig. 7C), individual muscle weights (Fig. 7D), nonmuscle tissue weights (data not shown), total energy intake and energy expenditure (data not shown), or respiratory exchange ratio (RER) during fed-fasted-refed conditions (metabolic flexibility) (Fig. 7E). In addition, no differences in serum glucose were identified between genotypes following a physiological (4-h) fast (Fig. 7G, time 0 of ITT), a prolonged (12-h) fast (Fig. 7F, time 0 of GTT), an intraperitoneal glucose challenge (Fig. 7F, GTT), or an intraperitoneal insulin challenge (Fig. 7G, ITT). Consistent with these findings, diet but not genotype effects were identified in skeletal muscle–specific insulin-stimulated phosphorylation of key proteins in the insulin signaling cascade including insulin receptor substrate 1 (data not shown) and Akt (Akt pS⁴⁷³ or Akt pT³⁰⁸) (Fig. 7H–K). Thus, increasing IMTG hydrolysis via ATGL overexpression does not influence systemic energy homeostasis or either systemic or muscle-specific glucose homeostasis and insulin action.

The goal of the current study was to determine the role of IMTGs and IMTG hydrolysis in skeletal muscle and whole-body metabolism at baseline and in response to diet-induced obesity. Remarkably, we demonstrate that modulation of IMTG content and hydrolysis by altering ATGL action in skeletal muscle does not influence systemic energy, lipid, or glucose homeostasis, nor does it influence muscle-specific insulin action or mitochondrial function. The absence of profound metabolic phenotypes in skeletal muscle–specific ATGL mutant mice is striking considering the dramatic changes in IMTG hydrolysis and accumulation. These data provide compelling evidence that IMTG content, while often highly associated with insulin resistance and metabolic disease, is not causal in these disorders. These data further indicate that modulating IMTG hydrolysis neither promotes nor protects against deleterious metabolic consequences of obesity.

Although the effects of skeletal muscle lipid uptake and oxidation on glucose homeostasis and insulin action have been extensively studied (11), the metabolic consequences of IMTG metabolism itself remain less well-understood (6). With regards to IMTG synthesis, increasing muscle DGAT1 in vivo promotes insulin sensitivity (13,14), whereas increasing DGAT2 promotes insulin resistance (15). While both interventions similarly increase IMTG content, they differentially affect other non-TAG lipid metabolites (ceramides), suggesting that the latter rather than the former influences metabolic phenotypes. Thus, different proteins may have different metabolic effects, despite affecting similar metabolic pathways. With regards to IMTG hydrolysis, acutely modulating ATGL or HSL action in cultured human myotubes alters DAGs and disrupts glucose homeostasis and insulin action (33). However, the current study indicates that chronically modulating ATGL action in murine skeletal muscle in vivo fails to influence metabolic phenotypes. Several potential adaptations (i.e., increased CGI-58, Plin2, and Plin5 in SMAKO mice) may protect against lipid-induced insulin resistance but are clearly not sufficient to fully compensate for altered IMTG hydrolysis. These data indicate that changing IMTG content by altering IMTG hydrolysis is not pathogenic. The validity of these conclusions is supported by the finding that comparable changes in IMTG content by altering IMTG synthesis are sufficient to influence metabolic outcomes (13–15), whereas changes in IMTG hydrolysis are not.

Interestingly, enhancing IMTG hydrolysis in vivo has minimal effect on non-TAG lipid metabolites, and inhibiting IMTG hydrolysis increases certain subspecies of DAGs and tends to increase rather than decrease total DAGs. The lack of DAG accumulation with ATGL overexpression is likely due to sufficient residual capacity to further metabolize DAGs relative to the amount produced by IMTG hydrolysis. Furthermore, recent evidence suggests that the specific DAG species generated by ATGL cannot be directly used for phospholipid synthesis or activation of protein kinase C, mechanisms by which DAGs have been shown to influence glucose homoeostasis and insulin action (34). The reason why DAGs accumulate with ATGL deletion is less clear but also occurs in mice with global deletion of ATGL (17) and knockdown of the ATGL coactivator CGI-58 (35). Studies of the latter suggest that the increase in DAGs may result from compensatory increases in DAG synthesis without concomitant increases in TAG synthesis (35) and further suggests that the lack of a physiological effect likely results from their subcellular localization to lipid droplets rather than membranes (35). Besides subcellular localization, the type of DAG stereoisomer (i.e., sn-1–2 vs. sn-1–3 vs. sni-2–3) and FA composition may also contribute to the dissociation between DAG concentrations and insulin responsiveness (5). Hence, modulating IMTG hydrolysis in vivo may not influence glucose homeostasis and insulin responsiveness because it does not produce physiologically relevant changes in the amount or cellular localization of specific bioactive non-TAG lipid metabolites.

In contrast, altering TAG hydrolysis could also affect metabolic phenotypes by influencing mitochondrial function via activation of PPAR nuclear transcription factors. A role for ATGL-mediated TAG hydrolysis in ligand-dependent (36) and ligand-independent (37,38) PPARα activation has been demonstrated in nonskeletal muscle cells. The physiological relevance of these findings has been corroborated in vivo in adipose tissue (23,39) and cardiac muscle (21). More recently, it has been shown that indirect modulation of ATGL action via its coactivator CGI-58 in human myotubes (40) or via Plin5 in rat muscle (41) influences PPAR target genes and mitochondrial function. Unexpectedly, changes in ATGL-mediated TAG hydrolysis in skeletal muscle in vivo do not significantly impact these parameters. These results are consistent with reports demonstrating IMTG accumulation but unaltered in vivo mitochondrial function in skeletal muscle of GAKO (42) and muscle-specific CGI-58 knockout mice (43). Thus, in striking contrast to cardiac muscle (21), in vivo modulation of TAG hydrolysis in skeletal muscle does not influence PPAR target genes or mitochondrial phenotypes.

The divergent consequences of altering IMTG hydrolysis in cardiac versus skeletal muscle may be due to differences in metabolic flux (10,44). In cardiac muscle (variable energy supply, chronically high energy demand), it has been proposed that FAs are preferentially directed to IMTG storage followed by coordinated ATGL-mediated release as energy substrates and PPAR ligands, thereby coupling hydrolysis to oxidation (21). In skeletal muscle (variable energy supply and demand), the above processes may only become physiologically relevant during functional but not nutritional stresses. Interestingly, enhancing TAG synthesis (via DGAT1) in rodent skeletal muscle (14,45) or TAG hydrolysis (via ATGL) in rodent adipose tissue are sufficient to drive FA oxidation (23) in the absence of functional stress, whereas enhancing TAG hydrolysis (via ATGL) in rodent skeletal muscle is not. While prior studies have suggested that ATGL-mediated TAG hydrolysis is important for working muscle (19,46), these studies were performed using GAKO mice, which have significant cardiac morbidity and defects in TAG catabolism in multiple tissues. Therefore, additional studies are required to determine the relative contribution of extra- versus intramyocellular TAG hydrolysis to skeletal muscle metabolism/function during functional stresses such as endurance or resistance exercise, which are believed to increase flux through the IMTG pool.

The regulated storage and release of both essential and deleterious lipid metabolites from intracellular TAG stores are among the most fundamental processes in metabolism. The complex relationship between intracellular TAGs and metabolic phenotypes is best exemplified in skeletal muscle, where IMTG accumulation is variably associated with insulin resistance (i.e., obesity) or insulin sensitivity (i.e., endurance exercise). The resolution of how IMTG metabolism contributes to these divergent phenotypes holds the key to understanding how these processes may be exploited for therapeutic benefit. Our study provides convincing evidence that IMTGs neither cause nor prevent insulin resistance and that manipulation of IMTG hydrolysis via altered ATGL action does not influence metabolic consequences of HFD-induced obesity. Additional functional studies will further enhance our understanding of TAG hydrolysis in muscle and whole-body physiology. Importantly, with few exceptions (21,47), tissue-specific modulation of ATGL-mediated TAG hydrolysis has largely produced beneficial effects (23,39,48–50). Thus, the lack of adverse metabolic consequences of altering ATGL action in skeletal muscle indicates that modulating TA hydrolysis may still be a viable therapeutic strategy for treating metabolic disorders. Additional studies in other metabolically relevant tissues are still required to more fully understand the health implications of TAG hydrolysis in normal physiology and disease.

This work was supported by the following: National Institutes of Health grants R03-DK-077697 and R01-DK-090166 and a Howard Hughes Medical Institute Physician-Scientist Early Career Award (to E.E.K.); Claude D. Pepper Pilot Grant (to M.T.S.); Erwin Schrödinger Fellowship (to G.S.); and the Endocrine Fellows Foundation's Marilyn Fishman Grant for Diabetes Research (to C.F.Y.). Tissue lipid analysis was supported in part by the Lipidomics Shared Resource, Hollings Cancer Center, Medical University of South Carolina (P30-CA-138313); the Lipidomics Core in the SC Lipidomics and Pathobiology COBRE (P20-RR-017677); the National Center for Research Resources; and the Office of the Director of the National Institutes of Health (C06-RR-018823).

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

M.T.S. and M.K.B. were project leaders for analysis of SMAKO and Ckm-ATGL mice, respectively. L.C. was the technical manager for all experiments and performed imaging analyses. G.S. and C.F.Y. performed experiments. G.D., B.H.G., and P.C. assessed mitochondrial respiration. V.R. and J.P.D. quantified FA-CoAs. R.S. and R.Z. performed TAG hydrolase activities. D.B.S. performed imaging analysis. N.P.G. helped generate and validate the mouse models. P.C.K. and T.P. contributed intellectual expertise and helped prepare the manuscript. E.E.K. generated mouse models, designed, executed, and analyzed experiments, and wrote the manuscript. All authors contributed intellectually to this work and reviewed and edited the manuscript. E.E.K. 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.

Parts of this study were presented in poster form at the 73rd Scientific Sessions of the American Diabetes Association, Chicago, Illinois, 21–25 June 2013.

The authors thank the following for contributions: Jeffrey S. Flier, MD (Harvard Medical School) for mentoring and support; Eric Olson, PhD (University of Texas Southwestern) for providing Myo-Cre mice; C. Ronald Kahn, MD (Joslin Diabetes Center) for providing the Ckm promoter construct; Jacek Bielawski, PhD (Medical University of South Carolina) for expertise related to lipidomics; the University of Pittsburgh Center for Biological Imaging for equipment and expertise related to imaging; and the University of Pittsburgh Division of Endocrinology Metabolic Center for equipment and expertise related to metabolism.