Relationship of glucose and oleate metabolism to cardiac function in lipin-1 deficient (fld) mice.

Lipin-1 is the major phosphatidate phosphatase (PAP) in the heart and a transcriptional coactivator that regulates fatty acid (FA) oxidation in the liver. As the control of FA metabolism is essential for maintaining cardiac function, we investigated whether lipin-1 deficiency affects cardiac metabolism and performance. Cardiac PAP activity in lipin-1 deficient [fatty liver dystrophy (fld)] mice was decreased by >80% compared with controls. Surprisingly, oleate oxidation and incorporation in triacylglycerol (TG), as well as glucose oxidation, were not significantly different in perfused working fld hearts. Despite this, [³H]oleate accumulation in phosphatidate and phosphatidylinositol was increased in fld hearts, reflecting the decreased PAP activity. Phosphatidate accumulation was linked to increased cardiac mammalian target of rapamycin complex 1 (mTORC1) signaling and endoplasmic reticulum (ER) stress. Transthoracic echocardiography showed decreased cardiac function in fld mice; however, cardiac dysfunction was not observed in ex vivo perfused working fld hearts. This showed that changes in systemic factors due to the global absence of lipin-1 could contribute to the decreased cardiac function in vivo. Collectively, this study shows that fld hearts exhibit unchanged oleate esterification, as well as oleate and glucose oxidation, despite the absence of lipin-1. However, lipin-1 deficiency increases the accumulation of newly synthesized phosphatidate and induces aberrant cell signaling.

We, therefore, hypothesized that lipin-1 defi ciency in the heart would severely inhibit TG synthesis and FAO and, thereby, cardiac function. We determined that there was cardiac dysfunction in the lipin-1-defi cient mice in vivo. Surprisingly, lipin-1 defi ciency did not produce major modifi cations in FA esterifi cation or oxidation in isolated perfused working hearts. There was also no indication of impaired cardiac function in this ex vivo system, indicating that the dysfunction in vivo probably resulted from systemic infl uences on the heart due to global lipin-1 deficiency. However, there were differences in phosphatidate and phosphatidylinositol turnover and in cell signaling.

Animal care and breeding strategy
We established a breeding colony of Balb/cByJ-Lpin1fl d/J mice from the Jackson Laboratory, Bar Harbor, ME. We used two breeding triads, the fi rst consisting of two female fl d mice bred with one male heterozygous mouse, which produced only fl d and heterozygous offspring. The other triad contained two female heterozygous mice and one male heterozygous mouse, resulting in fl d, heterozygous, and wild-type offspring. The wild-type and heterozygous mice were both designated as the control mice. Mice were fed Lab Diet 5058 containing 9% fat by weight. Male mice were used in the experiments. The research was conducted in accordance with the policies of the Canadian Council on Animal Care, as approved by the University of Alberta Animal Policy and Welfare Committee.

Noninvasive cardiac monitoring and tissue collection
Transthoracic echocardiography was performed on mildly anesthetized (1.5% isofl urane and 95% O 2 ) mice using a Vevo 770 Imaging System (VisualSonics, Toronto, ON) ( 36 ). The Tei index was calculated as the sum of the isovolumic relaxation and contraction time divided by the ejection time ( 37 ). One week after echocardiographies, 11-week-old mice were fasted from 0900 to 1300 h (0600-1800 h light/dark cycle) before being euthanized by decapitation, after which cardiac tissues were collected for RT-PCR and Western blot analysis. Serum samples were also collected and assayed with the glucose-c kit (Wako Chemicals, Richmond, VA). Serum TG and unesterifi ed FAs were measured using the TG GPO kit (Pointe Scientifi c, Canton, MI) and NEFA kit (Wako Chemicals), respectively. For serum lipid measurements, food was withheld from 19-to 23-week-old mice from 0900 to 1100 h.

Quantitative real-time PCR
mRNA concentrations were measured by quantitative RT-PCR relative to that of TATA-binding protein ( Tbp ) ( 25 ). Similar results were obtained with hypoxanthine-guanine phosphoribosyltransferase and cyclophilin A as reference genes. Primer sequences are listed in supplementary Table I.

SDS-PAGE and Western blot analysis
SDS-PAGE and Western blots were performed using antibodies as described in supplementary Table II ( 38 ). The antibody against the C terminus of lipin-1 was a gift from Dr. Thurl Harris liver ( 11 ). First, lipin-1 functions as a phosphatidate phosphatase (PAP) through its catalytic DxDxT motif, which generates the diacylglycerol (DG) required for the synthesis of TG, phosphatidylethanolamine (PE), and phosphatidylcholine (PC) ( 14 ). There are three members in the lipin family, of which lipin-1 is the best characterized. All three lipins have PAP activity, but they are expressed in a tissue-specifi c manner ( 14 ). Lipin-1 appears to be the predominant lipin in the heart because of the apparent absence of cardiac PAP activity in lipin-1-defi cient mice ( 15 ). These animals are known as fatty liver dystrophy ( fl d ) mice because they develop transient fatty livers and hypertriglyceridemia, which resolve upon weaning (16)(17)(18). Fld mice are devoid of mature adipose tissue, which is attributed to the role of lipin-1 in inducing peroxisome proliferator activated receptor ␥ (PPAR ␥ ) expression during adipocyte differentiation ( 19 ). The fl d mouse is also insulin resistant and is prone to developing atherosclerosis when fed a high cholesterol/cholate diet ( 16,20 ). In addition, phosphatidate (PA) accumulation leads to demyelination in Schwann cells through aberrant ERK1/2 activation, which subsequently causes the development of peripheral neuropathy in fl d mice ( 21,22 ).
Lipin-1 also acts as a transcriptional coactivator with peroxisome proliferator-activated receptor-␥ coactivator-1 ␣ (PGC-1 ␣ ) and PPAR ␣ to upregulate the expression of proteins involved in FAO in liver ( 23 ). Consequently, fl d mice are defective in the fasting-induced hepatic expression of PPAR ␣ and its downstream targets ( 23 ). We decided to determine whether lipin-1 has a similar function in the heart. Lipin-1 consists of full-length lipin-1B and the lipin-1A splice variant, which lacks a stretch of 20 amino acids ( 24 ). Lipin-1B is the predominant isoform in the heart, and it exhibits both PAP and transcriptional coactivator activities ( 23,24 ).
Lipin-1 expression is dynamically regulated in the liver ( 25,26 ), heart ( 27 ), and adipose tissue ( 28 ). Lpin1 transcription and, thus, PAP activity are increased in the liver in fasting and diabetes due to the synergistic actions of glucagon (or epinephrine) through cAMP production and glucocorticoids ( 11 ). Insulin antagonizes these actions in rat and mouse hepatocytes ( 25 ). Lpin1 gene expression is regulated in neonatal rat cardiomyocytes in a way similar to that in the liver (B. P. C. Kok and D. N. Brindley, unpublished results). Lpin2 and Lpin3 expression are not regulated signifi cantly by glucocorticoids, cAMP, or insulin in hepatocytes ( 25 ). Increased lipin-1 expression is thought to provide a reservoir of PAP activity which, together with the FA-induced translocation of lipin-1 from the cytosol to the endoplasmic reticulum (ER) ( 15,29,30 ), enhances the capacity for FA sequestration into TG in the face of increased FA uptake ( 10,11 ). Furthermore, increased lipin-1 expression in fasting and diabetes could also promote FAO through its role as a transcriptional coactivator. In the fed state, lipin-1 expression is suppressed by insulin action ( 25 ), and this presumably helps to promote glucose utilization with a concurrent decline in FAO. Despite this, cardiac PAP activity has been shown to be decreased in insulin-resistant rat models as well as in Type 2 diabetic cocktail (Sigma-Aldrich), 30 nM microcystin-LR, 0.6 mM PA labeled with [ 3 H]palmitate (approximately 6 × 10 4 dpm per assay), 1 mM EDTA/EGTA, 0.4 mM PC, and 200 µM tetrahydrolipstatin to inhibit the degradation of the DG product by lipase activity ( 39 ). In the second assay, 45 mM Triton X-100 was used to disperse 5 mM PA in micelles and PC was omitted. Each sample was assayed in a total volume of 100 µl consisting of 100 mM Tris/ maleate (pH 6.5) or Tris/HCl (pH 7.4) in addition to 0.6 mM dithiothreitol, 1 or 6 mM MgCl 2 (pH 7.4 and 6.5 respectively), protease inhibitor cocktail, 30 nM microcystin-LR, 1 mM PA labeled with [ 3 H]palmitate (approximately 6 × 10 4 dpm per assay), 9 mM Triton X-100 (from the substrate preparation), and 200 µM tetrahydrolipstatin.
Parallel measurements were performed in the absence of Mg 2+ or in the presence of 8 mM N-ethylmaleimide to determine the contribution from LPP activity ( 14 ). For the assays where Mg 2+ was omitted or for the assays with Mg 2+ concentration curves, all buffers were depleted of bivalent cations with AG 50W-X8 resin Na + -form ( 40 ). LPP activity was determined directly by using the (University of Virginia, Charlottesville, VA). Quantitative densitometric analyses were performed using ImageJ software (National Institutes of Health, Bethesda, MD). Antibodies against lipin-2 were raised in rabbits using the peptide sequence N ′ -PKGELIQERTKGNK-C ′ followed by affi nity purifi cation (Genscript, Piscataway, NJ). The lipin-2 antibody was verifi ed by comparing endogenous lipin-2 in the heart to recombinant lipin-2 protein overexpressed in MCF-7 breast cancer cells ( Fig. 1B , upper panel).

Phosphatidate phosphatase enzymatic assay
PAP assays were performed at pH 6.5 and 7.4 using two methods for preparing the PA ( 14 ). The fi rst method used a dispersion of PA with PC, which was designed to maximize the activity of PAP versus that of lipid phosphate phosphatase (LPP). Each sample was assayed in a total volume of 100 µl consisting of 100 mM Tris/maleate buffer (pH 6.5) or Tris/HCl buffer (pH 7.4) in addition to 0.6 mM dithiothreitol, 1.5 mM MgCl 2 (pH 7.4) and 5 mM MgCl 2 (pH 6.5), 2 mg/ml FA-poor BSA, protease inhibitor (50:10:10:20:5, by vol), followed by a second full-length development with hexane-diethyl ether-acetic acid (60:40:1, by vol) ( 46 ). Individual lipids were collected in scintillation vials containing 10% water and Ecolite scintillation fl uor (MP Biomedicals, Solon, OH), and radioactivity was measured. A two-dimensional TLC system was used to confi rm the levels of radiolabeled PC, phosphatidylserine, phosphatidylinositol (PI), sphingomyelin, and cardiolipin. TLC plates were loaded at one corner and developed with chloroform-methanol-water-NH 4 OH (60:40:4:0.5, by vol) and chloroform-methanol-acetic acid-acetone-water (45:15:10:20:5, by vol) in two different directions. The phospholipids were identifi ed using standards on a separate plate as well as by ninhydrin staining for amine-containing phospholipids and Dragendorff staining for choline-containing phospholipids ( 47 ). Oleate incorporation into lipids was essentially identical whether we used the one-or two-dimensional TLC system.
The two-dimensional TLC system was also used to separate individual phospholipids from the ventricles of fl d and control mice. Different phospholipids were scraped off the plate and quantifi ed using the organic phosphate assay. Glycogen content and glucose incorporation into glycogen were also measured ( 48,49 ). Briefl y, tissues were hydrolyzed at 100°C for 30 min in 40% KOH, followed by precipitation with 100% ethanol. The precipitated glycogen was washed three times with 95% ethanol and then hydrolyzed at 100°C for 3 h in 3 M HCl. After neutralization with 2 M NaOH, glucose equivalents were measured using a Glucose-c kit (Wako Chemicals).

Statistics
Results are expressed as means ± SEM. The two-tailed Student t -test or one-way ANOVA followed by Bonferroni posthoc test was used to test for signifi cance ( P < 0.05).

Characterization of lipins and PAP and LPP activities in fl d hearts
Fld mice were identifi ed by their physical appearance resulting from their lipodystrophy and their lower body weights ( Table 1 ) compared with controls. We found no signifi cant differences in serum glucose or nonesterifi ed fatty acid levels; however, TG levels were lower in the fl d mice compared with controls ( Table 2 ). As expected, we found no signifi cant mRNA expression of lipin-1A or lipin-1B in fl d hearts ( Fig. 1A ). However, mRNA expression of lipin-2 and lipin-3 was increased by 40-50% in fl d hearts ( P < 0.03 and 0.004, respectively), whereas protein levels of lipin-2 were similar between genotypes ( Fig. 1B, C ). Although we could not obtain a successful Western blot for lipin-3, a recent study showed that lipin-3 levels in fl d hearts are similar to those in wild-type hearts ( 50 ).
We then measured PAP activity in fl d and control mice under a variety of assay conditions. Optimum Mg 2+ concentrations were determined for each of the two PAP assay systems at both pH 6.5 and 7.4 ( Fig. 1D ). Mg 2+ concentrations above 2 mM inhibited cardiac PAP activity at pH 7.4 using either PA/PC mixed liposomes or PA/Triton X-100 micelles ( Fig. 1D ). We also verifi ed that addition of different concentrations of Mn 2+ , Co 2+ , and Ca 2+ did not reveal the presence of other PAP activities (results not shown). Earlier work ( 15 ) showed no signifi cant PAP activity in fl d Triton X-100-based procedure (pH 6.5 and 7.4) in the presence of 8 mM NEM, with the exception that [ 3 H]PA was used instead of [ 32 P]PA ( 41 ). DG formed in these assays was extracted in 2 ml chloroformmethanol (19:5, by vol) containing 0.08% olive oil as an acylglycerol carrier ( 40 ). Activated alumina was added to remove the unreacted PA and any liberated [ 3 H]palmitate from the chloroform phase. The chloroform phase was dried, and radioactivity was determined. PAP activity was calculated by subtracting the NEM-insensitive or Mg 2+ -independent LPP activity from the total activity. Each sample was assayed at three different protein concentrations (30-200 g) to ensure a proportional response, and the conversion of PA to DG was restricted to <20%.
PA was measured in lipid extracts as described previously ( 43 ). Briefl y, a PA standard curve (0-4 nmol) and lipids extracted from cardiac tissue were loaded half way up plastic-backed silica TLC plates (VWR International, Radnor, PA) along with PC and PA standards on the outermost lanes. The plates were developed twice in chloroform-methanol-ammonium hydroxide (65:35:7.5, by vol). The migration of PA bands was determined by cutting the outermost lanes and identifying the PA and PC standards. The TLC plates were then cut 1 cm above the migrated PA, thus removing most other phospholipids, and developed in the reverse direction in chloroform-methanol-acetic acid-acetone-water (50:20:12:10:5, by vol), followed by staining for 1 h with 0.3% Coomassie Brilliant Blue R250 in 20% methanol containing 100 mM NaCl. The TLC plates were destained in 20% methanol and scanned at 700 nm using the Odyssey Infrared Imaging scanner (LI-COR Biosciences, Lincoln, NB). PA levels were determined by comparison to the PA standard curve. TG concentrations were measured in extracts of heart using a TG GPO kit (Pointe Scientifi c, Canton, MI).

Perfused working heart studies and measurements of fatty acid and glucose metabolism
Hearts were perfused in working mode ex vivo ( 44 ). Left atrial preload and aortic afterload were set to 11.5 and 50 mm Hg, respectively. For the fi rst 30 min, all hearts were perfused with Krebs-Henseleit buffer containing 118.5 mM NaCl, 25 mM NaHCO 3 , 4.7 mM KCl, 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 2.5 mM CaCl 2 , 0.5 mM EDTA, 5 mM glucose, and 50 µU/ml insulin to deplete TGs as much as possible ( 3 ). Subsequently, hearts were perfused with the same buffer supplemented with 5 mM [U- 14  We conclude that the absence of lipin-1 severely decreases myocardial PAP activity and that lipin-2 and lipin-3 contribute only about 15-20% of normal cardiac PAP activity under the most optimum assay conditions we could fi nd. LPP activity, which is thought not to participate in TG synthesis ( 51 ), was similar in fl d and control hearts ( Fig. 1E ).

Decreased cardiac function in fl d mice in vivo
Having demonstrated that there was a signifi cant reduction in PAP activity in fl d mice, we next determined whether cardiac function was affected in vivo. Noninvasive transthoracic echocardiography was used to evaluate cardiac function in 19-to 23-week-old fl d and control mice. The fl d mice had signifi cantly decreased systolic function as indicated by reduced ejection fraction and fractional shortening ( Table 1 ). This corresponded with decreased end-diastolic left ventricular internal diameter measurements ( Table 1 ). Stroke volume and cardiac output and measurements of pulmonary peak venous fl ows were also drastically decreased in fl d hearts ( Table 1 ). We also determined hearts using Triton X-100-PA micelles as the substrate, and we essentially confi rmed this observation at pH 6.5 or 7.4 ( Fig. 1E ). A Mg 2+ -dependent, NEM-sensitive PAP activity was detected using mixed PA/PC liposomes at both pH 6.5 and 7.4 in the fl d hearts at levels approximately 15-20% of that found in the control hearts ( Fig. 1E ).
This value is similar to the work of Mitra et al. ( 50 ) who used 32 P-labeled PA presented in Triton X-100 and measured the formation of water-soluble 32 P i . However, this method overestimates PAP activity unless precautions are taken to block phospholipase A-type activities, which produce labeled lysophosphatidate and glycerophosphate. These compounds can then be converted to 32 P i by acid or alkaline phosphatases. All of these compounds are extracted into methanol/water rather than chloroform phase and are included in the measurement of "water-soluble" products. We avoided this complication by measuring the formation of [ 3 H]DG from [ 3 H]PA and blocking the degradation of DG with tetrahydrolipstatin. This latter technique provides a valid PAP assay that can be used with tissue homogenates ( 40 ). Analyzed from transthoracic echocardiographies of 19-to 23-week-old fl d (n = 7) and control (n = 8) mice, as well as 10-week-old fl d (n = 8) and control (n = 7) mice. a P < 0.05 compared with 19-to 23-week-old control mice. b P < 0.05 compared with 10-week-old control mice. c P < 0.05 compared with age-matched control mice and both 10-week-old fl d and control mice.  ( 54,55 ). Additionally, oleate incorporation into TG is greater than or equal to that of palmitate ( 54,56 ). When we analyzed functional parameters in the fl d and control hearts, we found that there were no significant differences in contractility and function ex vivo ( Table 3 ), even though hearts from fl d mice were signifi cantly smaller than those of the controls ( Table 1 ). This was surprising because we had found cardiac dysfunction in vivo.
Interestingly, we found no signifi cant differences between control and fl d hearts in the rates of glucose ( Fig.  4A , left and middle panels) and oleate oxidation ( Fig. 5A ) when expressed relative to heart dry weight or cardiac power. Glucose incorporation into glycogen was also similar between genotypes ( Fig. 4A , right panel). Cardiac glycogen accumulated when both fl d and control hearts were perfused with FA-free and glucose-containing buffer for 30 min, as expected ( Fig. 4B ). The perfusate was switched to oleate-and glucose-containing buffer for another 30 min, and we found that cardiac glycogen content was depleted signifi cantly following this period of perfusion in the control, but not the fl d, hearts ( Fig. 4C ).
We also determined cardiac TG content after each perfusion period ( Fig. 5B ). TG levels were depleted in control hearts after perfusion with FA-free and glucose-containing buffer after 30 min, as expected ( Fig. 5B ). However, there was no signifi cant depletion of TG in fl d hearts. When the perfused hearts were switched to oleate-and glucose-containing buffer for another 30 min, there was no signifi cant change in TG levels at the end of the perfusion in either fl d or control hearts. This demonstrated that TG turnover was at a steady state during this 30 min period ( Fig. 5B ). There were no signifi cant differences in the accumulation of [ 3 H]oleate into total glycerolipids or TG in the fl d hearts compared with controls ( Fig. 5C ). However, oleate accumulation in PC, PE, and PI was increased by 1.95-, 1.9-and 3.49-fold, respectively ( Fig. 5C ). Consistent with reduced cardiac PAP activity, we found a 4.36-fold increase in the accumulation of oleate in PA in fl d hearts compared with controls ( Fig. 5C ). Interestingly, the mass of PI and PS were increased 1.21-and 1.35-fold, respectively, in fl d cardiac function in 10-week-old fl d and control mice and found similar results ( Table 1 ). Heart sizes and tibia lengths of fl d mice at 10 weeks of age were not signifi cantly different compared with age-matched control mice ( Table 1 ). However, these parameters were signifi cantly lower in 19-to 23-week-old fl d mice compared with corresponding control mice ( Table 1 ). Interestingly, systolic blood pressure was similar between fl d and control mice ( Table 2 ), even though we might expect differences in vascular capacity and demand because fl d mice are lipodystrophic and smaller in size.

Metabolic gene and protein expression profi les in fl d hearts
To determine whether cardiac dysfunction in fl d mice in vivo was linked to changes in FA or glucose metabolism, we analyzed the expression of genes for key metabolic regulators in control and fl d hearts. We chose to analyze samples from 11-week-old mice to determine whether there were any changes in cardiac metabolism in the absence of differences in heart size. mRNA levels for PGC-1 ␣ and PPAR ␣ were increased in fl d hearts ( Fig. 2A ), which differs from studies in liver where PPAR ␣ induction was blunted in fl d mice ( 23 ). We further examined expression of PPAR ␣ target genes downstream of lipin-1 regulation ( 23 ). Indeed, ACOX-1 and CPT-1B gene expression were upregulated. However, gene expression of CD36 and MCAD were similar between fl d and control hearts ( Fig. 2A ). mRNA expression for enzymes in the TG synthesis and lipolysis pathways were also similar between genotypes, except for small increases in mRNA expression of GPAT1 and DGAT1 ( Fig. 2B ) in fl d hearts. There were no differences in mRNA expression of CD36, GLUT4, lipoprotein lipase, acetyl-CoA carboxylase (ACC), pyruvate dehydrogenase kinase 4, or malonyl-CoA carboxylase, which are known regulators of FA and glucose metabolism ( Fig. 2A, C ).
In addition to profi ling gene expression, we determined the levels and phosphorylation states of several important regulators of cardiac metabolism and function. Expression and phosphorylation of Akt, ACC, and AMP-activated protein kinase (AMPK) were unchanged (supplementary Fig. I). There were also no signifi cant changes in fatty acid transport protein 1 (FATP1), GLUT4, ERK1/2, or sarcoplasmic reticulum Ca 2+ -ATPase (SERCA) (supplementary Fig. I), although mRNA levels of SERCA were slightly increased ( Fig.  2C ). Interestingly, both adipose triglyceride lipase (ATGL) expression and hormone-sensitive lipase (HSL) phosphorylation at serine 660 were decreased ( Fig. 3A , B ), which indicates reduced TG hydrolysis in fl d hearts. We also observed a decrease in long-chain acyl-CoA synthetase 1 (ACSL1) expression and in the phosphorylation of pyruvate dehydrogenase (PDH) ( Fig. 3 ). ACSL1 defi ciency has been shown to affect the acyl-CoA available for FAO ( 52 ), and reduced phosphorylation of PDH leads to its increased activity ( 53 ).

Cardiac function and metabolism in ex vivo perfused working hearts of fl d mice
We next determined the effects of the changes in gene and protein expression profi les on fatty acid and glucose phosphorylation in the fl d hearts is linked to higher levels of mTOR phosphorylation (supplementary Fig. II). Several studies show that activation of the mTORC1 pathway often results in increased protein synthesis, cell enlargement, and eventually, cardiac hypertrophy ( 60, 61 ). However, the hearts of 19-to 23-week-old fl d mice were signifi cantly smaller compared with controls ( Table 1 ). Interestingly, mTORC1 activation has also been recently implicated in the upregulation of ER stress ( 62,63 ). Therefore, we determined the gene expression of 78 kDa glucose-regulated protein (GRP78), which is an ER chaperone induced by ER stress, and CCAAT/enhancer-binding protein homologous protein (CHOP), which is a transcription factor mediating ER stress response, and found that they were increased ( Fig. 6F ). Moreover, there was an increase in the gene expression of spliced X-box binding protein 1 (XBP1), which is highly indicative of ER stress ( Fig. 6F ) ( 64 ). Increased ER stress signaling associated with aberrant mTORC1 activation through phosphatidate signaling could explain why the fl d hearts became smaller at 19 to 23 weeks of age. hearts compared with controls ( Fig. 5D ). The mass of the other major phospholipids, including PA, were not significantly different between groups. The relative composition of different phospholipids in our study was similar to that reported previously ( 57 ).

Examination of the downstream signaling effects of aberrant phosphatidate accumulation
PA accumulation can activate the mTORC1-p70S6 kinase-S6 ribosomal protein signaling cascade ( 58,59 ). As such, we determined whether mTORC1 signaling was increased in the fl d mice due to aberrant PA metabolism as seen in the perfused fl d hearts. There was a very marked 14-fold increase in the phosphorylation of S6 ribosomal protein in the fl d hearts ( Fig. 6C , E , right panel). Activation of S6 ribosomal protein occurs downstream of mTOR complex-1 (mTORC1) and p70S6 kinase. Correspondingly, p70S6 kinase phosphorylation in fl d mice was significantly increased ( Fig. 6B, E , left panel). Although mTOR phosphorylation was not signifi cantly increased ( Fig. 6A, E , left panel), the majority of increased S6 and p70S6 kinase metabolism and function had not been systematically determined. We hypothesized that the dynamic regulation of lipin-1 expression in the heart would be essential for regulating cardiac FA metabolism and function in different physiological and pathological conditions. We expected that complete lipin-1 defi ciency would have dramatic effects on myocardial FA esterifi cation and oxidation, resulting in cardiac dysfunction. Indeed, fl d mice exhibited systolic dysfunction in vivo, as determined by noninvasive echocardiography. To determine the effects of lipin-1 deficiency on cardiac metabolism, we fi rst assessed PAP activity in fl d hearts. We concluded from our assays that fl d hearts have 15-20% residual PAP activity, which is explained by the expression of lipin-2 and lipin-3 in fl d hearts.
We chose to use the isolated perfused working heart model to study the metabolic effects of the absence of lipin-1 in the heart alone. This allowed us to determine DISCUSSION Lipin-1 is a unique protein with dual functions in promoting FA esterifi cation and FAO ( 11 ). During starvation and diabetes, the combined effects of glucocorticoids and cAMP increase Lpin1 gene transcription and thus PAP activity in liver ( 11,25 ). This increase, in addition to the translocation of lipin-1 to membranes stimulated by unsaturated FAs, is thought to provide a reservoir that maintains or increases the capacity for FA utilization and storage ( 11,15,65 ). Furthermore, PAP activity is decreased in the hearts of insulin-resistant JCR:LA corpulent rats ( 32 ), and lipin-1 expression is decreased in the ventricles of Zucker diabetic fatty rats and in Type 2 diabetic patients ( 31 ). A recent study showed that the regulation of lipin-1 expression in the heart was dependent on PGC-1 ␣ as well as ERR ␣ and ERR ␥ ( 50 ). However, the role of lipin-1 in cardiac Densitometric analysis of single Western blots for HSL, ACSL1, and ATGL in addition to (D) PDH in 11-weekold fl d (n = 6-7) and control (n = 6-7) hearts. Total ACSL1, ATGL, PDH, and HSL were normalized to Ran GTPase and then expressed relative to control values. Phospho-HSL and phospho-PDH were normalized to Ran GTPase and expressed relative to total HSL and PDH, which were also normalized to Ran GTPase. * P < 0.05 compared with controls. light period, and thus, we might not have detected abnormalities in the fl uctuation of plasma FA levels occurring during the diurnal cycle of the lipodystrophic fl d mice, which could also negatively affect cardiac function. We expected that fl d hearts would have decreased rates of glucose oxidation because fl d mice are insulin-resistant ( 16,20 ). The fl d mice had similar circulating levels of glucose compared with controls ( Table 2 ), which was shown in a previous study ( 16 ). This latter work also found that fl d mice are hyperinsulinemic. Interestingly, there was no signifi cant decrease in glucose oxidation and no significant difference in glucose incorporation into glycogen under defi ned conditions ex vivo. However, glycogen depletion in the control hearts was signifi cantly greater than in fl d hearts when the perfusate was switched to glucose-and oleate-containing buffer. It is likely that upon the reintroduction of FA in the perfusate, glucose uptake and, hence, glycogen accumulation is decreased more in control hearts than in fl d hearts. Glucose utilization from glycogen stores might also be greater in control hearts compared with fl d hearts. Most importantly, FAO was not different in isolated perfused fl d hearts compared with the controls, suggesting that cardiac lipin-1 expression is not essential for maintaining myocardial FAO. Consistent with unchanged glucose and fatty acid utilization, gene and protein expression of several key proteins involved in regulating glucose and FA metabolism were similar between fl d and control hearts.
Because fasting-induced expression of PPAR ␣ and its target genes in liver was ablated by knocking down lipin-1 expression ( 23 ), we had hypothesized that lipin-1 deficiency would also limit the transcriptional regulation of PPAR ␣ in the heart. However, we found transcriptional upregulation of PPAR ␣ in the fl d hearts in addition to increased gene expression of the PPAR ␣ downstream targets ACOX-1 and CPT-1, whereas mRNA expression of other PPAR ␣ target genes was unchanged. Further work needs to be done to determine how PPAR ␣ is regulated in the absence of lipin-1 in the heart. It is conceivable that lipin-2 and possibly lipin-3 compensate for the lack of lipin-1 as a PPAR ␣ coactivator, as lipin-2 also exhibits transcriptional coactivator activity ( 68 ). Moreover, Grimsey et al. established that lipin-1 and lipin-2 can be reciprocally regulated ( 69 ). However, lipin-1 is required for promoting transcriptional regulation of PPAR ␣ and its target genes in the livers of fl d mice, and there does not appear to be compensation of lipin-1 transcriptional coactivator function by lipin-2 or lipin-3 in the liver ( 23 ), in spite of lipin-2 being highly expressed in hepatocytes ( 14,25 ). Indeed, the changes in mRNA levels of PPAR ␣ , PGC-1 ␣ , and a subset of their targets could be a compensatory response in vivo to the aberrant fuel utilization of fl d mice in the diurnal cycle ( 67 ). Furthermore, there was increased activation of PDH as shown by decreased phosphorylated PDH, which refl ects the inability of the fl d mice to utilize FAs to the same extent as control mice in the fasted state ( 70 ). Overall, our results demonstrate that the absence of lipin-1 in the heart does not overtly affect cardiac FAO and glucose metabolism when assessed ex vivo. It is likely that the the consequences for cardiac metabolism and function under a workload using defi ned conditions in the absence of systemic factors or extraneous signals from the circulatory system. The use of perfused working hearts is also preferable to cultured cardiomyocytes for assessing the role of lipin-1 on FA and glucose metabolism because metabolism is stimulated in a physiologically appropriate manner by the need to perform mechanical work. Surprisingly, cardiac function in the perfused fl d hearts was not signifi cantly different from the control hearts even though the fl d hearts were smaller. This suggests that the cardiac dysfunction we observed in fl d mice in vivo may be related to the systemic changes stemming from global lipin-1 deficiency, which can be described as a combination of the absence of adipose tissue, the corresponding decrease in adipokine secretion, whole-body insulin resistance, aberrant changes in the circadian rhythm of whole-body metabolism ( 16,66,67 ), and higher workload in vivo compared with ex vivo. We also found decreased circulating TG levels in fl d mice ( Table 2 ), which could affect the availability of substrate for cardiac work. We only measured serum FA levels after 2 h of food deprivation in the the specifi c radioactivity in the glycerol 3-phosphate precursor pool is also measured ( 71 ). This specifi c activity depends on the relative activities of glycerol kinase in the fl d and control hearts and on the rate of substrate cycling between glycerol 3-phosphate and dihydroxyacetone phosphate. This cycling involves glycerol-3-phosphate dehydrogenase, which exhibits a signifi cant isotope effect in its use of 3 H, resulting in progressive increases in the specifi c activity of the [2-3 H]glycerol-3-phosphate precursor pool compared with that obtained using [1,3-3 H] or [ 14 C]glycerol ( 71 ). We also measured glycerolipid synthesis in neonatal rat ventricular myocytes by using [ 3 H] oleate. E600 (100 µM), which is a general lipase inhibitor ( 72 ), was added to the incubations to prevent the hydrolysis of TG and provide a more accurate assessment of synthesis. Knocking down lipin-1 by 50% with adenovirus expressing shRNA against Lpin1 did not signifi cantly decrease the incorporation of oleate into TG. Thus, our unpublished results with cultured cardiomyocytes confi rm those from the more physiologically relevant perfused heart system. Our results show that complete depletion of lipin-1 does not signifi cantly decrease the capacity of the heart to synthesize TG.
This result can possibly be explained by the residual PAP activity, attributed to lipin-2 and/or lipin-3, in fl d hearts. Although fl d hearts express relatively high phosphatidate phosphatase activity from their LPPs, these enzymes probably do not participate in glycerolipid synthesis, as their active sites are facing the extracellular space or the changes in cardiac function and metabolic profi ling in vivo are attributable to the effects of global lipin-1 deficiency, as previously mentioned.
The initial concentrations of TG in perfused hearts of fl d and control mice were similar. Considerable lipolysis took place in the control hearts during the 30 min incubation in glucose-containing buffer in the absence of oleate as was seen from the decrease of about 50% in TG content. However, lipolysis was decreased in hearts of fl d mice because there was minimal TG depletion when fl d hearts were perfused with oleate-free buffer. This conclusion is supported by the decreased levels of both ATGL expression and HSL phosphorylation in the 11-week-old fl d hearts. When the perfused hearts were switched onto oleate-and glucose-containing buffer, we demonstrated that TG levels were constant at the start of this perfusion period compared with the end. This indicates that TG turnover was at a steady state during the perfusion with both oleate and glucose ( 13 ), unlike the initial perfusion period with glucose alone. There was no defi ciency in the ability of the fl d hearts to accumulate [   We also detected increased [ 3 H]oleate accumulation in PA in the fl d hearts without observing an increase in total PA mass. The latter analysis of PA mass was performed after completely separating PA from PS and PI by sequential chromatography on silica gel using a basic solvent system followed by an acidic system ( 43 ). Presumably, the labeling experiment in perfused hearts identifi ed the pool of PA that is formed de novo by lipin activity, whereas the total PA pool refl ects the balance of a variety of enzymes, including phospholipase D and diacylglycerol kinase, that contribute to PA turnover. Interestingly, another group has shown that there are increases in total PA levels in fl d hearts as analyzed by LC/mass spectrometry ( 50 ). However, no details of the full molecular species of the PA are lumenal sides of internal membranes ( 51,73 ). By contrast, glycerolipid synthesis occurs on the cytosolic surface of internal organelles. It has been hypothesized that cytosolic PAP activity provides a reservoir of activity that can be recruited to the ER in response to the FA load ( 29,74 ). Presumably, the reservoir of lipin-2 and lipin-3 activity in fl d hearts is suffi cient to provide the capacity for relatively normal rates of TG accumulation. In addition, the accumulation of oleate in the major phospholipids, PC and PE, was not compromised in fl d hearts; in fact, there was increased labeling of PC and PE with [ 3 H]oleate in fl d hearts. This observation might be due to decreased fatty acid remodeling of phospholipids, which occurs rapidly in the heart ( 56,75 ).  6. mTOR-p70S6 kinase signaling in fl d and control hearts. Representative Western blots of (A) mTOR, (B) p70S6 kinase, and (C) S6 ribosomal protein in 11-week-old fl d and control hearts. Densitometric analysis of single Western blots for (D) total and (E) phosphorylated mTOR, p70S6 kinase, and S6 ribosomal protein in 11-week-old fl d (n = 6-7) and control (n = 6-7) hearts. Total mTOR, p70S6 kinase, and S6 ribosomal protein were normalized to Ran GTPase and then expressed relative to control values. Phospho-S6, phospho-p70S6 kinase, and phospho-mTOR were normalized to Ran GTPase and expressed relative to total S6, p70S6 kinase, and mTOR, which were also normalized to Ran GTPase. * P < 0.05 compared with controls. (F) mRNA expression of proteins involved in ER stress response. mRNA expression in 11-week-old fl d (n = 8-11) and control (n = 7-10) hearts was expressed relative to Tbp (TATA-binding protein). Results for fl d mice were then expressed relative to control mice. * P < 0.05 compared with controls. CHOP, CCAAT/ Enhancer-binding protein homologous protein; GRP78, 78 kDa glucose-regulated protein; XBP-1, X-box binding protein 1. signaling axis. This does not compensate for the development of signifi cantly smaller hearts in the fl d mice. Instead, aberrant mTORC1 activation could be associated with the development of ER stress. This work provides novel information contributing to the understanding of lipin-1 in the regulation of glycerolipid synthesis, energy partitioning, and signaling in the heart.
The authors thank Donna Beker and Sandra Kelly from the Echocardiography Core of the Cardiovascular Research Centre (University of Alberta) for performing and analyzing the echocardiographies; Grant Masson for work on the ex vivo perfused working heart system; Brandi Sidlick for blood pressure measurements; and Jay Dewald for excellent technical assistance.
given, and it is unclear whether the authors achieved efficient separation of PA from the other phospholipids, as shown in previous studies ( 76,77 ). This separation would be essential because we have demonstrated an increase in PS and PI mass in the fl d hearts.
Our group demonstrated that decreasing PAP activity diverts PA metabolism to CDP-diacylglycerol and acidic phospholipid production ( 78 ). This could contribute to the increased labeling of PI and the increase in PI mass that we observed in fl d hearts. The accumulation of PA, PS, and PI in the fl d hearts could lead to aberrant cell signaling. For example, PA accumulation in peripheral nerves of fl d mice increases ERK1/2 activation, leading to demyelination ( 22 ). We did not observe signifi cant increases in ERK1/2 phosphorylation in the fl d hearts. However, PA can also activate mTORC1, leading to the downstream activation of p70S6 kinase ( 58,59 ), which phosphorylates and activates S6 ribosomal protein to promote cell growth ( 58,59 ). In fact, activation of mTORC1-p70S6 kinase leads to cardiac hypertrophy ( 79 ). We now show that there is an increase in signaling downstream of mTOR, i.e., increased p70S6 kinase phosphorylation in combination with a 14-fold increase in S6 ribosomal protein activation. However, there was no evidence of cardiac hypertrophy in 11-week-old mice, which is the age at which we fi rst observed cardiac dysfunction. Instead, the 19-to 23-week-old fl d mice have signifi cantly smaller hearts in the absence of defective FAO. Although the mTORC1 signaling pathway is classically linked to cell growth, recent studies have shown that mTORC1 activation can also result in the unfolded protein response and ER stress ( 62,80 ). There was increased GRP78 and CHOP gene expression as well as Xbp1 splicing in the fl d mice, which could be due to mTORC1 activation ( 63 ). The increased ER stress response in fl d mice could explain why the hearts became smaller at 19 to 23 weeks of age. Alternatively, mTORC1 has been implicated in regulating metabolism. Aberrant activation of mTORC1 increases glycolysis and the oxidative arm of the pentose phosphate pathway ( 81 ). Although we did not see differences in glucose oxidation, we found increased PDH activation. Other studies on fl d mice show changes in energy partitioning in vivo, which could aberrantly affect cardiac function ( 66,67 ).
The present work provides a comprehensive assessment of the effects of lipin-1 defi ciency in fl d mice on the work output of the heart in vivo and ex vivo relative to the use of fatty acids and glucose as fuels. Lipin-1 defi ciency led to cardiac dysfunction in fl d mice as measured in vivo, probably because of systemic factors stemming from global lipin-1 defi ciency, such as lipodystrophy, modifi ed hormonal regulation, and fuel availability. When these factors were equalized in the perfused working heart system, there were no signifi cant differences in work output or the use of oleate and glucose for oxidative metabolism. We conclude that TG accumulation in fl d hearts is similar to that in control hearts because of residual PAP activity resulting from lipin-2/3 and reduced TG hydrolysis. Despite this, fl d hearts displayed increased oleate accumulation in PA, which could be linked to the mTORC1-p70S6 kinase