Caloric restriction in leptin deficiency does not correct myocardial steatosis: failure to normalize PPARα/PGC1α and thermogenic glycerolipid/fatty acid cycling

J. Eduardo Rame, Lili A. Barouch, Michael N. Sack, Edward G. Lynn, Mones Abu-Asab, Maria Tsokos, Steven J. Kern, Jennifer J. Barb, Peter J. Munson, Marc K. Halushka, Karen L. Miller, Karen Fox-Talbot, Jianhua Zhang, Joshua M. Hare, Michael A. Solomon, Robert L. Danner


Objective: Evidence supports an antilipotoxic role for leptin in preventing inappropriate peripheral tissue lipid deposition. Obese, leptin-deficient mice develop left ventricular (LV) hypertrophy and myocardial steatosis with increased apoptosis and decreased longevity. Here we investigated the cardiac effects of caloric restriction versus leptin repletion in obese leptin-deficient (ob/ob) mice. Methods: Echocardiography was performed on 7 mo old C57BL/6 wild-type mice (WT) and ob/ob mice fed ad libitum, leptin-repleted (LR-ob/ob), or calorie-restricted (CR-ob/ob) for 4 wk. Ventricular tissue was examined by electron microscopy (EM), triglyceride (TAG) content, oil red O staining, mitochondrial coupling assay, and microarray expression profiling. Results: LR and CR-ob/ob mice showed decreased body and heart weight, and LV wall thickness compared with ad libitum ob/ob mice. LV fractional shortening was decreased in ad libitum ob/ob mice, but restored to WT in LR and CR groups. However, myocardial lipid content by EM and TAG analysis revealed persistent cardiac steatosis in the CR-ob/ob group. Although CR restored mitochondrial coupling to WT levels, PPARα was suppressed and genes associated with oxidative stress and cell death were upregulated in CR-ob/ob animals. In contrast, LR eliminated cardiac steatosis, normalized mitochondrial coupling, and restored PGC1α and PPARα expression, while inducing core genes involved in glycerolipid/free fatty acid (GL/FFA) cycling, a thermogenic pathway that can reduce intracellular lipids. Conclusions: Thus, CR in the absence of leptin fails to normalize cardiac steatosis. GL/FFA cycling may be, at least in part, leptin-dependent and a key pathway that protects the heart from lipid accumulation.

  • leptin
  • cardiac metabolism
  • lipotoxicity
  • glycerolipid/free fatty acid metabolism
  • peroxisome proliferator-activated receptor
  • peroxisome proliferator-activated receptor γ coactivator

obesity is associated with hypertension, atherosclerosis, cardiomyopathy and decreased longevity. Obesity-induced heart disease, including nonischemic cardiomyopathy characterized by intramyocellular lipid excess, has become an important public health problem as the number of overweight children and adults has surged worldwide.

The role of leptin, an adipocyte-derived hormone that regulates appetite and metabolism, remains incompletely understood in obesity-induced cardiovascular disease. Leptin resistance and, more rarely, deficiency have been implicated in cardiovascular disease. Hyperleptinemia and leptin resistance are highly correlated with fat mass in diet-induced obesity and have been linked to adverse cardiovascular outcomes. However, hyperleptinemia appears to be adaptive during caloric excess, protecting the heart and other organs from lipotoxicity by reducing intracellular lipid accumulation. (20, 21). In support of this antilipotoxic hypothesis, obese leptin-deficient (ob/ob) and leptin-resistant (db/db) mice develop left ventricular (LV) hypertrophy, increased intramyocardial triglyceride accumulation, and accelerated cardiomyocyte apoptosis, abnormalities associated with reduced longevity (2, 3). In the ob/ob murine model, leptin administration suppresses caloric intake resulting in weight loss, regression of LV hypertrophy and normalization of myocardial steatosis and apoptosis (2, 3).

Similar to caloric excess, fasting has been linked to steatosis in nonadipose tissues as triglycerides are mobilized from fat depots and metabolism switches from the utilization of glucose to free fatty acids. Steatosis in the heart and liver has been demonstrated in both animal models and humans subjected to fasting and lesser degrees of caloric restriction (17, 19, 38). Fasting upregulates cardiac peroxisome proliferator-activated receptor alpha (PPARα), a key regulator of myocardial energetics and mitochondrial function in normal animals. Mice lacking PPARα (−/−) develop massive lipid accumulation in nonadipose tissues in response to either high fat diets or fasting, indicating the importance of PPAR signaling pathways in protecting organs from excess steatosis (19). Leptin, which induces PPARα (41) and protects the heart from high-fat diets, has not been similarly investigated during caloric deprivation in the setting of clinically relevant obesity.

Here, the cardioprotective effects of leptin were examined during calorie restriction (CR) using electron microscopy, quantitative measures of myocardial lipid content, mitochondrial coupling studies, and global myocardial expression profiling. CR consisted of pair feeding with leptin-repleted animals, a milder form of lipotoxic stress than true fasting. We hypothesized that despite restoration of normal weight, CR in the absence of leptin would fail to reverse the cardiac steatosis of the ob/ob mouse phenotype. (3). In contrast to leptin repletion (LR), which restored the hearts of ob/ob mice to the wild-type (WT) state, CR with equivalent weight loss did not normalize cardiac steatosis and further dysregulated gene expression with induction of some PPARα target genes despite suppression of PPARα. Gene set enrichment analysis (GSEA) identified glycerolipid/free fatty acid (GL/FFA) cycling, a so-called futile metabolic pathway involved in thermogenesis, as a leptin-regulated, antisteatotic network in the heart.



We studied 6-mo-old ob/ob mice with C57BL/6J background, as previously described (2) and age-matched C57BL/6J WT controls. Weight loss was induced in ob/ob mice for 4 wk by either LR or CR and compared with WT and ob/ob controls fed ad libitum. Echocardiography was performed at 4 wk (see Supplemental Methods for details).1 The Institutional Animal Care and Use Committee of The Johns Hopkins University School of Medicine approved all protocols and experimental procedures.

Electron Microscopy, Lipid Quantitation, Mitochondrial Copy Number, and Respiration

Fixed sections of three hearts from each group were examined with electron microscopy. Unfixed ventricular tissue was used to measure myocardial triglycerides (TG) and perform oil red O staining (see Supplementary Methods for details). Mitochondrial copy number was assessed by quantitative PCR of mitochondrial genes ND1 and cytochrome b. Mitochondria were isolated from fresh LV tissue, and respiration was measured by oximetry (see Supplemental Methods for details).

Myocardial RNA Isolation and Oligonucleotide Microarrays

Mice were killed and hearts quickly removed, weighed, minced, and placed in RNALater solution (Qiagen). Total RNA was isolated and cDNA synthesized. Biotin-labeled cRNA was prepared from 1 μg of cDNA, fragmented and hybridized (10 μg) to Affymetrix (Santa Clara, CA) mouse 430_2 oligonucleotide probe arrays for 16 h at 45°C. Signal intensities were measured using Agilent GeneArray Scanner (Affymetrix) (see Supplemental Methods for details).

Gene Expression Validation with Quantitative Real-Time PCR

Total RNA from three samples of each of the four groups was aliquoted, treated with DNase1, and reverse transcribed to cDNA. Quantitative real-time PCR (RT-PCR) was performed on target genes to validate results generated by microarrays (see Supplemental Methods for details).

Statistical Analysis

Phenotypic data.

Data are presented as means ± SE. Statistical significance (P < 0.05) was determined by ANOVA or Student's t-test where appropriate. A Dunnett's test or Student Newman-Keuls test was used for post hoc analysis.

Microarray data.

Signal intensity values (Affymetrix Microarray Suite 5.0) for the oligonucleotide arrays were standardized and transformed using the Symmetric Adaptive Transform ( Principal components analysis was performed on this restricted transformed data matrix to allow for detection of outliers or other relevant patterns. Differentially expressed probe sets were selected based on a 5% false discovery rate (FDR) ≥ 50% fold-change between any group, and ≥ 50% present call in at least one group. Functional themes among differentially regulated genes were explored using Ingenuity Pathway Analysis (Ingenuity Systems, Redwood City, California; To further identify metabolic gene sets enriched with leptin or caloric restriction without imposing arbitrary cutoffs, we utilized GSEA (Broad Institute, MIT; (35) (see Supplemental Methods for details). The entire data set of the above investigation has been submitted to the National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO;, accession number GSE16790.


Characteristics of the Murine Phenotypes

Supplemental Table S1 summarizes anatomic and echocardiographic measurements from the four groups at 7 mo of age. Body weight was increased in ob/ob mice compared with WT mice. Increased wet heart weight was attributable to concentric hypertrophy, as previously reported, (2) with a relative 50% increase in wall thickness in ad libitum ob/ob mice and a decrease in fractional shortening compared with WT. CR and LR both resulted in significant weight loss, reduction in concentric hypertrophy, and restoration of fractional shortening after 4 wk of intervention. At the time of death, no significant difference was identified between the CR and LR ob/ob groups in heart weight, end-diastolic interventricular septal and posterior wall thickness, relative wall thickness, or fractional shortening.

Effects of CR and LR on Cardiac Steatosis

Electron microscopy was performed to identify ultrastructural and mitochondrial changes. As shown by others, (10) ob/ob mice fed ad libitum were found to have increased accumulation of lipid droplets in cardiac myocytes compared with the WT control hearts (Fig. 1). LR restored the heart to a nonsteatotic phenotype, consistent with myocardial tissue TG levels returning to WT (3). However, equivalent weight loss after CR in ob/ob mice did not result in a significant decrease in lipid accumulation (Fig. 1, C and D). Measurement of myocardial triacylglycerol (TAG) confirmed severe cardiac steatosis with leptin deficiency and no significant reduction with CR, while LR normalized intramyocardial lipid to WT levels (Fig. 2). Further electron microscopy analysis revealed the increased lipid deposition within the mitochondrial zones in between the myofibers without evidence of ultrastructural damage (Supplemental Fig. S1). We also observed within the mitochondrial zones the presence of focal lamellar inclusions, suggestive of lipid storage in the CR-ob/ob mice. There were no significant ultrastructural changes observed in mitochondria in any of the ob/ob groups relative to WT, but a qualitative increase in the number of mitochondria was noted in both the ad libitum and CR-ob/ob groups. We also noted enlarged nucleoli and an increase in myocyte width relative to the nucleus in both the ad libitum and CR-ob/ob groups compared with LR and WT, ultrastructural features that are suggestive of myocyte hypertrophy.

Fig. 1.

Electron microscopy (EM) of representative sections of myocardium in wild-type (WT) (A, B), calorie-restricted (CR) ob/ob mice (C, D), ob/ob mice fed ad libitum (E, F), and leptin-repleted (LR) ob/ob mice (G, H). The most significant finding is severe cardiac steatosis in the CR ob/ob mice as well as in the ob/ob ad libitum group. There is significant intramyocardial lipid deposition with focal lamellar inclusions suggestive of lipid storage. The lipid accumulation appears to be restricted to the mitochondrial zone.

Fig. 2.

Measurement of myocardial triacylglycerol (TAG) and oil red O staining revealed increased lipid content in the hearts of ob/ob mice relative to WT. Consistent with EM, both methods identified a significant decrease in myocardial lipid deposition with leptin repletion but not with CR.

Effects of CR and LR on Mitochondrial Function and Biogenesis

We performed cardiac mitochondrial respiration studies to confirm previous observations of mitochondrial dysfunction in the hearts of ob/ob mice, (5) and to investigate whether excess cardiac steatosis in CR-ob/ob mice leads to a further increase in mitochondrial uncoupling. We identified increased mitochondrial uncoupling in the hearts of ad libitum fed ob/ob mice compared with WT based on a decreased ratio of oxygen consumption in the presence and absence of ADP [respiratory control ratio (RCR), Fig. 3]. LR improved coupling of oxygen consumption and ATP production, restoring the RCR to WT levels. Surprisingly, the CR-ob/ob group also restored mitochondrial coupling back to WT levels, despite the presence of severe myocardial steatosis. Mitochondrial copy number, measured by the ratio of mitochondrial genes (ND1 and cytochrome b) to the nuclear gene H19, was found to be significantly increased in the hearts of the ad libitum and CR-ob/ob groups relative to WT (Supplemental Fig. S2). Only LR restored mitochondrial numbers back to WT levels.

Fig. 3.

Summary of mitochondrial respiration study measuring oxygen consumption in the presence (GMA) and absence (GM) of ADP to determine degree of mitochondrial uncoupling (shown on y-axis as GMA/GM ratio of mitochondrial oxygen consumption). A significant increase in mitochondrial uncoupling is observed in the ob/ob ad libitum group relative to WT.

Effects of CR and LR on the ob/ob Mouse Cardiac Transcriptome

Microarray analysis identified 794 differentially regulated probe sets across the four groups (5% FDR, ≥1.5-fold change with any other group and 50% present-absent call). Figure 4 depicts differentially expressed genes comparing the effects of LR (y-axis) versus CR (x-axis) relative to ob/ob mice fed ad libitum (no intervention). From the 794 probe sets that originally met criteria for differential expression between any group-wise comparison, 790 were divided into one of eight areas reflecting all major patterns of gene expression between experimental groups (see Supplemental Methods). Notably, LR consistently corrected ob/ob ad libitum expression profiles toward those of WT animals (Fig. 4, patterns I–III and VI–VIII) and left unaltered genes that were not affected by leptin deficiency (patterns IV and V). Conversely, CR corrected ob/ob ad libitum gene expression for only a small minority of regulated transcripts (Fig. 4; patterns III and VI, representing 139/790 probe sets). For all other probe sets and patterns, CR failed to directionally alter gene expression toward levels of normal WT animals. In two patterns (I and VIII) containing the largest number of probe sets (271 and 139, respectively), LR corrected, while CR further displaced gene expression away from WT controls. For a complete list of genes falling within each expression pattern, please refer to Supplemental Table S2.

Fig. 4.

Major patterns of gene expression. Scatter plot of CR (x-axis) vs. LR (y-axis) effects relative to ob/ob ad libitum animals shown as log fold changes. Vertical and horizontal lines define no effect (<±10% change) regions for CR and LR, respectively. Eight major patterns of gene expression are thereby defined that are further displayed as line plots and labeled I–VIII. The number of probe sets for each pattern is shown. Note that 4/794 probe sets were unaffected by either CR or LR (center square) and are not further depicted.

An analysis of overrepresented functional themes among differentially regulated genes using Ingenuity Pathways Analysis found significant alterations in lipid metabolism, inflammatory response pathways, and cell death-associated genes (P < 0.00001 for all). Within each of these functional categories, CR-ob/ob animals displayed the greatest deviations in gene expression compared with WT (Fig. 5). For lipid metabolism (Fig. 5A, pattern I), CR further suppressed PPARα and two proteins involved in lipid trafficking, caveolin, and phospholipid transfer protein, compared with ob/ob animals fed ad libitum. Pattern VIII (Fig. 5A), capturing genes downregulated by LR toward WT, but further increased by CR, identified several aspects of disordered lipid metabolism, including alterations in transport and processing (fatty acid transporter 1, acyl-CoA synthetase long-chain family member 6), beta-oxidation of fatty acids (acetyl-coenzyme A acyltransferase 2, mitochondrial 2, 4-dienoyl CoA reductase 1), and modification of lipids (CYP1BI, epoxide hydrolase 2). The predominant impact of CR on inflammatory response genes (Fig. 5B) was to exacerbate leptin-induced suppression (P < 0.00001, patterns I and IV), particularly affecting canonical pathways associated with antigen presentation, the complement system, and cytokine signaling (P < 0.0001 for all). Like lipid metabolism and inflammation, cell death pathway genes were overrepresented among differential regulated transcripts (P < 0.006) and displayed all eight of the major expression patterns seen across our four groups (Fig. 5C). However, leptin deficiency primarily downregulated antiapoptotic genes (i.e., hypoxia upregulated 1, heat shock proteins 1 and 90 alpha, E26 avian leukemia oncogenes 1 and 2, disabled homolog 2, and CD53 antigen; Fig. 5C, all in pattern I) and upregulated proapoptotic genes (i.e., DNA damage-inducible transcript 4, plectrin homology domain containing-family F, member 1 and TP53 apoptosis effector; Fig. 5C, patterns VI, VII, and VIII, respectively), a response that may be consistent with lipotoxic stress. Importantly, CR worsened or failed to reverse these abnormalities (patterns I, II, and IV for downregulation of primarily antiapoptotic genes; patterns V, VII, VIII for upregulation of primarily proapoptotic genes). Likewise, oxidative stress-related, free radical scavenging genes, overrepresented among transcripts that were upregulated in the leptin-deficient state, were further aggravated by CR.

Fig. 5.

Biologically relevant functional themes among differential expressed genes. An Ingenuity Pathway Analysis was performed to identify overrepresented biological functions. Selected genes from 3 overrepresented functional categories, lipid metabolism (A), inflammatory response (B), and apoptosis (C) are shown in a heat map format. Red signifies increased and green decreased expression relative to the mean values for each transcript. Average results are shown for genes returning multiple probe sets. Gene symbols and Entrez ID numbers are provided. Please refer to Supplemental Table S2 for full gene titles. As shown, CR in ob/ob mice produced the most extreme changes in gene expression, while LR largely restored the transcriptome to the WT state.

GSEA of Pathways Affected by CR and Leptin

To detect significant coordinated changes (without a fold-change restriction) in the cardiac gene expression of ob/ob mice subjected to CR versus LR, we performed GSEA (Broad Institute, MIT) with a priori defined gene sets of lipid metabolism and cardiac failure (see Table 1 for summary of significantly enriched gene sets and supplemental methods for list of a priori defined gene sets). In the CR compared with ad libitum ob/ob hearts, we identified significantly upregulated pathways of fatty acid biosynthesis, mitochondrial fatty acid beta oxidation and glutathione metabolism.

View this table:
Table 1.

Summary of gene set enrichment analysis

In the hearts of LR compared with ad libitum ob/ob mice, we identified significant upregulation of the GL/FFA cycle, an exergonic process implicated in thermogenesis and energy homeostasis. (26, 27) This pathway of glycerolipid metabolism was also demonstrated to be significantly downregulated in the ad libitum ob/ob group relative to WT. CR failed to restore the glycerolipid metabolism pathway back to WT levels.

We also identified a significant upregulation of a gene set in LR-ob/ob that is downregulated in the ventricular myocardium of end-stage failing human hearts. This is consistent with the echocardiographic finding that LR restored myocardial function as measured by the return of fractional shortening to WT levels.

Effects of Leptin on Cardiac Expression of Peroxisome Proliferator-Activated Receptor γ Coactivator-1α/PPARα and Glycerolipid Cycling

Given the importance of the peroxisome proliferator-activated receptor γ coactivator (PGC)-1α/PPARα regulatory pathway on fatty acid metabolism and lipid homeostasis in models of obesity (39, 40) and fasting, (19, 23) quantitative RT-PCR was performed to verify changes in PPARα and PGC-1α gene expression. These studies confirmed that transcriptional PGC-1α is downregulated in the heart of ob/ob mice (Fig. 6A). With equivalent weight loss, LR but not CR significantly increased cardiac mRNA levels of PGC-1α back to WT. Despite the known increase in intramyocellular fat in the hearts of ob/ob mice (3, 11, 18, 19) cardiac PPARα gene expression is decreased in leptin-deficient states (see Fig. 6B). LR increased, whereas CR further decreased cardiac PPARα mRNA levels.

Fig. 6.

Summary of quantitative real-time PCR measurement of peroxisome proliferator-activated receptor γ coactivator (PGC)-1α (A) and peroxisome proliferator-activated receptor (PPAR)α (B) expression in the hearts of leptin-repleted ob/ob (LR-ob/ob), ob/ob mice fed ad libitum (ob/ob), and calorie-restricted ob/ob mice (CR-ob/ob), relative to WT. Data represent ratio of gene expression relative to GAPDH. Also shown are estimates of SE (m) and significant differences from WT.

Next, we investigated expression changes in key regulators of lipid metabolism and mitochondrial function that are transcriptionally induced by PPARα. In Fig. 7, we demonstrate by RT-PCR reductions in the cardiac expression of pyruvate dehydrogenase kinase 4 (PDK4), uncoupling protein 3 (UCP3), and fasting induced adipose factor (FIAF or angiopoietin-like 4) in ob/ob mice. Cardiac mRNA levels of PDK4, UCP3, ACOT2, SLC27a1 (fatty acid transporter) and Angptl-4 are induced several-fold in the CR-ob/ob group, despite a significant decrease in PPARα mRNA. In the case of malonyl-CoA decarboxylase (Mlycd, Fig. 7), a key regulator of fatty acid oxidation, both ob/ob ad libitum and CR mice show increased expression relative to WT animals, despite robust downregulation of PPARα in both groups. In contrast, cardiac gene expression of acyl-coenzyme A dehydrogenase (Acadvl) and carnitine palmitoyl transferase 1b (cpt1b) is not increased in the leptin-deficient ob/ob ad libitum and CR groups, but induction of gene expression is observed with LR, suggesting maintenance of PPARα-dependence with respect to these two genes involved in fatty acid oxidation. These observations suggest either: 1) the presence of a variable degree of PPARα responsiveness among key regulatory and metabolic genes associated with lipid homeostasis in leptin-deficient hearts challenged by states of overnutrition and CR; or 2) the presence of other factors controlling the activity of PPARα such as unmeasured posttranslational modifications or increased PPARα ligand availability. We identified increased myocardial FFA in the CR group relative to the ob/ob ad libitum group (Supplemental Fig. S3), which may explain some of the significant differences in PPARα target gene expression between these groups (SLC27a, AngptL4, UCP3, ACOT2, PDK4) We also measured by RT-PCR the expression of other nuclear receptors, PPARβ/δ and ERRα, to understand the changes in the PPAR-regulated genes and found no significant differences across the four groups (Supplemental Fig. S4, d and f).

Fig. 7.

Summary of quantitative real-time PCR measurements for 10 PPARα-responsive genes: CD36 (A), SLC27a1 (B), CPT1b (C), Mlycd (D), ACADVL (E), ACAA2 (F), ACOT2 (G), AngptL4 (H), PDK4 (I), UCP3 (J). Data are shown as means ± SE (m) in arbitrary units (y-axis) relative to GAPDH. P value is for ANOVA across the 4 groups.

To validate the findings from GSEA which identified the Glycerolipid/FFA pathway as a potential leptin-regulated mechanism to relieve myocardial steatosis, we performed RT-PCR on additional genes involved in glycerolipid cycling (Fig. 8). Stearoyl CoA desaturase 1 (Scd1, Fig. 8A), which had been shown to be significantly upregulated in the liver of ob/ob mice, (9) is also induced in the myocardium of leptin-deficient animals and suppressed even below WT levels with LR (P = 0.045 for post hoc comparison of LR-ob/ob vs. WT). In contrast, CR failed to normalize Scd1, which catalyzes the conversion of fully saturated to monounsaturated fatty acids and is an important committed step in the synthesis of TAG. In leptin-deficient groups, we identified increased expression (Fig. 8) of gene products involved in the partitioning of FFA toward TAG [long-chain acyl-CoA synthetase (Acsl1)], esterification (Agpat2, Dgat1), and stabilization of lipid droplets storing TAG [Plin2, adipocyte different ion-related protein (ADRP)]. We identified induction of hormone-sensitive lipase (Fig. 8B: Lipe, HSL) with LR, which plays an important role in the hydrolysis of TAG to glycerol and FFA, both TAG → diacylglycerols (DAG) and DAG → monoacylglycerols (MAG).

Fig. 8.

Summary of differential expression with quantitative real-time PCR of genes implicated in glycerolipid cycling: Scd1 (A), Lipe (HSL) (B), Agpat1 (C), Agpat2 (D), Plin2(ADRP) (E), Lpin2 (F), Acsl1 (G), Dgat1 (H). Data are shown as means ± SE (m) in arbitrary units (y-axis) relative to GAPDH. P value is for ANOVA across the 4 groups.


One of the major peripheral roles of leptin may be to regulate intracellular lipid homeostasis in nonadipocytes such as liver, skeletal muscle, and cardiac muscle. Thus, leptin may prevent TG overload while allowing for a sufficient supply of fatty acids for cellular energy substrate utilization during periods of overnutrition. Here we have demonstrated a pivotal role of leptin in preventing cardiac steatosis during periods of weight loss due to CR. The following observations have been made that emphasize the importance of leptin on cardiac liporegulation: 1) Despite equivalent weight loss in CR-ob/ob and LR-ob/ob mice, we demonstrated that the hearts of CR-ob/ob mice had persistent severe cardiac steatosis; 2) Lack of leptin during CR was associated with dysregulation of genes involved in cardiac lipid homeostasis marked by significantly increased expression of some PPAR-target genes despite a decrease in cardiac expression of PPARα and no significant change in PPARβ/δ or ERRα; 3) Core genes of the GL/FFA cycle implicated in lipid detoxification and energy homeostasis, including Scd1 and Dgat1, are differentially regulated in leptin deficiency; and 4) That CR, in the absence of leptin, restored mitochondrial coupling to WT levels despite the oxidative stress and lipoapoptotic risk associated with severe intramyocellular lipid accumulation. In contrast, LR restored mitochondrial coupling in parallel with the resolution of cardiac steatosis in ob/ob mice.

The protective role of leptin in nonadipose tissue, including the myocardium, has been supported by multiple studies (5, 40). Leptin increases myocardial fatty acid oxidation in nonadipose tissue and thereby may prevent lipotoxicity by attenuating TG accumulation. In the heart, this leptin-induced increase in FAO is independent of AMPK/ACC activity or malonyl-CoA levels. (1) However, despite the absence of leptin, ob/ob mice and db/db mice have demonstrated increased levels of fatty acid oxidation in the heart (5, 6, 24). It has also been demonstrated that despite improvement in body weight, diabetes, and serum TG, CR does not decrease the levels of myocardial fatty acid oxidation in ob/ob mice. (33) With enrichment analysis using a priori defined gene sets, we have identified an alternative mechanism of cardiac lipoprotection linked to leptin with significant upregulation of the GL/FFA cycle in the hearts of LR but not CR-ob/ob mice (Table 1). This “futile” cellular process continuously cycling between esterification of fatty acids to a glycerol backbone and hydrolysis has an essential and established role in energy homeostasis and thermogenesis. (26, 27) RT-PCR of core genes implicated in glycerolipid cycling (Fig. 8) identified a pattern of differential gene expression consistent with lipid storage in both ad libitum and CR leptin-deficient groups, while leptin repletion led to the hydrolysis of glycerolipids. Functional studies are necessary to confirm this and investigate whether myocardial utilization and synthesis of TAG through GL cycling is altered in leptin deficiency. Nevertheless, our observation suggests a mechanistic link between leptin signaling and GL/FFA cycling in the heart that could explain the increase in myocardial lipid (Figs. 1 and 2) that persists despite CR as well other findings that have been identified in models of impaired leptin signaling.

The reduced cold tolerance of ob/ob mice and Zucker fatty rats may be related to a compromised thermogenic response due to decreased GL/FFA cycling. A recent study demonstrated that increased cardiac apoptosis in ob/ob mice may be mediated by decreased catalytic activity of phosphoinositide-3 kinase (PI3K) (p110a), which was reversed by LR (37). Given the link between the GL/FFA cycle and cell survival/proliferation, possibly through lysophosphatidic acid-mediated (a GL/FFA intermediate) activation of the PI3K/mTOR/p70S6K pathway, (13) leptin repletion could reduce cardiomyocyte apoptosis and hypertrophy through an increase in Akt and PI3K signaling. Furthermore, there is mounting evidence that altered GL/FFA cycling contributes to insulin resistance (4, 31), which is present in leptin-deficient mice despite CR (32). Our gene set enrichment analysis may thus provide the initial evidence linking leptin deficiency to myocardial insulin resistance mediated by decreased GL/FFA cycling in the heart. Finally, leptin deficiency-induced decreases in GL/FFA cycling may provide an explanation for cardiac and other end-organ lipotoxicity that exists in murine models of obesity (including the ob/ob mouse model), despite the presence of increased fatty acid oxidation in the heart. The protective effects of hyperleptinemia observed in several murine models of obesity and lipotoxicity (1, 8, 20, 21, 30, 34, 41) may thus be explained by increased cycling of GL and FFA with the release of heat.

Myocardial Steatosis in Leptin Deficiency and PPARα Expression

PPARα has been shown to increase the expression of genes invoked in every step of cardiac fatty acid utilization. Cardiac specific overexpression of PPARα results in increased FA uptake, increased FA oxidation and lipotoxic cardiomyopathy similar to the cardiac metabolic profile of the diabetic heart (14, 15). PPARα-null mice have decreased cardiac expression of genes involved in the cellular uptake, mitochondrial transport, and oxidation of fatty acids. This knockout model has provided ample evidence for the central role played by PPARα in cardiac lipid homeostasis, especially with the development of massive cardiac steatosis in PPARα−/− mice subjected to fasting. (19, 23) This and other studies suggesting that PPARα is induced in response to fasting would predict that PPARα should be robustly upregulated in response to CR. In our model, PPARα gene expression was further suppressed with CR again implicating leptin, even in low or falling levels, as necessary for the normal physiological response of PPARα to caloric deprivation and fasting. We cannot exclude the possibility that the decreased expression of PPARα is compensatory in response to the increased rate of fatty acid oxidation that has been identified in leptin deficiency (5, 6, 24, 33). More recent evidence in a model of recurring ischemia/reperfusion injury suggests the possibility that reactive oxygen species (ROS) suppress PPARα as a compensatory phenomenon that prevents further cardiac injury from secondary lipotoxicity (10). Oxidant stress resulting from increased FAO in ob/ob mice might similarly suppress PPARα. Although measures of oxidant stress were not performed in this 6 mo old murine model, serum malondialdehyde (MDA) was increased in 2 mo old ob/ob mice, and CR failed to decrease this marker of oxidant stress (Supplemental Fig. S5). Future studies with double knockout (ob/ob//PPARα −/−) and transgenic (ob/ob//cardiac-specific PPARα overexpression) models should clarify the role of PPARα in mediating or protecting against lipotoxicity in the leptin-deficient heart.

Leptin Deficiency Dissociates PPARα From the Regulation Of Target Genes

Previous models of obesity implicating impaired leptin signaling, including the obese Zucker diabetic fatty (ZDF) rat resulting from a loss of function (fa/fa) mutation in the leptin receptor, have demonstrated a dissociation between PPARα expression and PPARα target gene regulation. In the ZDF rodent heart marked by lipotoxic cardiomyopathy and other models of leptin resistance (db/db), no change in PPARα transcript levels were observed but significant increases were noted in myocardial MCAD, CPT1, and other PPARα-target genes (6, 7, 29). Although our model suggests that decreased PPARα expression in the heart may be a causal determinant of cardiac steatosis by inhibiting fatty acid oxidation, it is now well accepted that cardiac fatty acid oxidation is increased in ob/ob and db/db mice (5, 7). Furthermore, our study cannot exclude the possibility of increased PPARα activation with leptin deficiency, which could be present in the setting of increased PPARα ligand availability. Although we did not measure activation of lipoprotein lipase (Lpl), which has been recently shown to be an important determinant of PPARα ligand activation (12), we did observe an increase in myocardial FFA with CR (Supplemental Fig. S3), which may explain increased expression of some PPARα target genes in this group (Fig. 7).

Mitochondrial Biogenesis and Uncoupling in Leptin Deficiency

The results of the mitochondrial respiration study can also be interpreted in light of our finding that PGC-1α/PPARα gene expression was suppressed in ad libitum and CR ob/ob mice. Ob/ob mice demonstrated increased mitochondrial uncoupling of oxygen consumption and ATP production relative to WT, consistent with the observation by Boudina et al. (5). Notably, this increase in fatty acid-induced mitochondrial uncoupling was not dependent on the expression of UCP3. In fact, we identified a significant decrease in the expression of UCP3 (Fig. 7) and no significant change in UCP2 in the hearts of ob/ob mice fed ad libitum, implicating other potential mechanisms behind the observed increase in mitochondrial uncoupling of obese hearts. We cannot exclude the possibility that increased activation of uncoupling proteins despite decreased expression in ob/ob mice could account for the increased mitochondrial uncoupling that was observed in our study, a mechanism identified in a study of db/db mice. (6) That study demonstrated an increase in ROS and lipid and protein peroxidation in db/db hearts, which is hypothesized to drive the increase in mitochondrial uncoupling by the activation of UCPs. Although we did not directly measure myocardial ROS or peroxidation products, our study did demonstrate in the hearts of both CR and ad libitum ob/ob mice an increase in the expression of genes involved in oxidative stress and glutathione metabolism (Table 1).

Interestingly, despite increased steatosis, CR restored mitochondrial coupling in the ob/ob mice. In CR-ob/ob mice, gene expression of both UCP3 and UCP2 was significantly increased despite suppression of cardiac PPARα. These data also support that mitochondrial uncoupling is not strictly dependent upon UCP3/UCP2 gene expression. (25) CR ob/ob mice had normal mitochondrial coupling despite increased expression of both UCP3 and UCP2, while ob/ob mice fed ad libitum demonstrated increased mitochondrial uncoupling, but decreased UCP3 and unchanged UCP2 expression.

We identified in our model a disruption of the tight link that has been demonstrated between PGC-1α and mitochondrial biogenesis (22). Our study identified an increase in mitochondrial copy number in both ob/ob ad libitum and ob/ob CR groups compared with WT (Supplemental Fig. S2), despite a decrease in PGC-1α expression. Although this concept was not directly explored, this observation might in part be attributed to CR-mediated induction of SIRT1, which would deacetylate PGC-1α and thereby enhance its activity (16, 28). Another explanation may be temporal, as PGC1α has been demonstrated to be initially induced in both ob/ob and db/db mice, leading to mitochondrial biogenesis and then suppressed with time (7). LR alone in ob/ob mice achieved equivalent weight loss without cardiac steatosis and restored mitochondrial copy number and function to WT levels.

In conclusion, we have found that the regulatory effects of leptin in the heart may be mediated by increased PPARα expression and changes in the control of GL/FFA cycling. Our study shows that the disruption in leptin signaling in ob/ob mouse hearts is associated with downregulation of genes encoding for GL metabolism and concomitant dysregulation of the PGC-1α/PPARα pathway. PGC-1α/PPARα gene expression was decreased in ob/ob mice, while PPARα target genes were nonetheless upregulated, consistent with the known increase in fatty acid oxidation rates (5, 24). Further evidence to support this hypothesis is found in the CR ob/ob model, in which suppression of PGC-1α/PPARα gene expression is highly suggestive of a mechanistic link between leptin signaling and PGC-1α/PPARα-mediated control of cardiac lipid homeostasis. Since diet-induced obesity in humans is associated with leptin resistance and myocardial steatosis (36), this study raises the possibility that the PGC-1α/PPARα axis and glycerolipid/FFA cycling are impaired in obese patients and that restoration of these pathways may be cardioprotective by reducing myocardial steatosis.


This work was supported by the National Heart, Lung, and Blood Institute (NHLBI) Cardiovascular Medicine Fellowship (J. E. Rame), the American Heart Association National Scientist Development Grant 0730375N (J. E. Rame), NHLBI Division of Intramural Research (M. N. Sack), National Institutes of Health Clinical Center (R. L. Danner), and NHLBI Grants R01-HL-077785 and K08-HL-076220 (L. A. Barouch).


No conflicts of interest, financial or otherwise, are declared by the author(s).


The authors thank Dr. Roger Unger for advice and review of the manuscript. Carolea Logun provided invaluable experience in expression profiling and Kelly Byrne formatted and edited the manuscript. Maeva Nyandjo is acknowledged for assistance in the lipid protocols.


  • 1 The online version of this article contains supplemental material.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
View Abstract