Lipin Deficiency Impairs Diurnal Metabolic Fuel Switching

BALB/cByJ-+/fld mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and bred to produce lipin-deficient fld (fld/fld) and wild-type (+/+ and +/fld) mice for studies. Both wild-type and fld mice ranged from 7 to 9 months of age, with average body weight of 28.3 + 1.7 g for wild-type and 21.6 + 0.8 g for fld mice. Animals were fed standard laboratory chow diet (4.5% fat and 50% carbohydrate by weight) (Lab Diet; Purina 5001, St. Louis, MO) and maintained under 12-h light/dark conditions (6:00 a.m./6:00 p.m.). Ten mice of each type were used for fasting and refeeding experiments of each stable isotope tracer study. Of the 10 mice, 5 (3 male and 2 female) were used for fasting and 5 (3 male and 2 female) for refeeding studies. The high-fat/high-carbohydrate diet contained 35% fat and 33% carbohydrate (by weight) (Diet F3282; Bioserve, Frenchtown, NJ). All animal studies were performed under approved institutional protocols and according to guidelines established in the Guide for the Care and Use of Laboratory Animals.

Measurements of respiratory quotient (RQ) were performed using an Oxymax indirect calorimetry system (Columbus Instruments, Columbus, OH). Mice were individually housed in the chamber for 48 h with lights on from 0600 to 1800 in an ambient temperature of 22–24°C. Food was available ad libitum during the dark cycle (feeding phase 1800–0600) and removed during the light cycle (fasting phase 0600–1800). RQ measurements were made under Oxymax system settings as follows: air flow, 0.6 l/min; sample flow, 0.5l/min; settling time, 6 min; and measuring time, 3 min. Six mice of each genotype were measured.

Metabolic flux analysis was performed using stable-isotope–labeled glucose or lactate as described previously (17,18). Briefly, fasting of animals was initiated at 4:00 p.m. At 7:00 p.m., animals were anesthetized under ∼5% isoflurane, and a preactivated Alza miniosmotic pump (model 2001D; Alza, Palo Alto, CA) was quickly inserted subcutaneously into each mouse. The entire process of anesthesia and minipump implantation was usually <3 min/mouse. The mini pump contained either 50 mg [U-¹³C₃]lactate or 50 mg [U-¹³C₆]glucose, each dissolved in 220 μl water. All stable isotope tracers used in this study were 99.9% enriched and purchased from Isotec (Miamisburg, OH). The quantity of tracer is sufficient to last through a 24-h infusion at a factory-calibrated pump rate of 8 μl/h. Blood and tissue (liver and skeletal muscle) samples from animals in fasting conditions were collected between 10:00 and 11:00 a.m. the next morning. Alternatively, animals undergoing refeeding were given mouse chow at 10:00 a.m., and blood and tissue samples were collected at 3:00 to 4:00 p.m. Under these experimental conditions, isotopic steady states are achieved at 15 h after fasting (12 h after infusion) and at 4–5 h after refeeding (17). Thus, data reported here were obtained under both metabolic and isotopic steady states. Animals were killed by an overdose of isoflurane anesthesia, and tissue samples were rapidly dissected free, snap frozen in liquid nitrogen, and stored at −80°C until further processing.

Blood plasma was isolated by centrifugation at 4°C. Plasma glucose and lactate concentrations were determined by a Cobas Mira analyzer (Roche Molecular Biochemicals) using reagents provided by Raichem (San Diego, CA). Glucose UV Reagent (catalog no. 80017) was used for glucose determinations, and Stat-Pack Rapid Lactate Test (catalog no. 869218) was used for lactate. Glucose residues derived from liver and muscle glycogen were prepared and determined as previously described (19). Gas chromatography and mass spectrometry conditions for measurement of glucose and lactate derivatives were as previously described (17). For glucose and lactate isotopomer determination, the ion clusters monitored were from 327 to 336 m/z.

Hepatic fatty acids were extracted after saponification of liver homogenates in 30% KOH (potassium hydroxide) and 100% ethanol using petroleum ether. Fatty acids were converted to their methylated derivatives using 0.5 N methanolic HCl (20). A glass capillary column BPX70 (SGE, Austin, TX) measuring 30 m × 250 μm (inner diameter) was used to separate fatty-acid derivatives. The gas chromatography conditions were carrier gas (helium) flow rate 1 ml/min, injector temperature 250°C, and oven temperature programmed from 120 to 220°C at 5°C/min. Under these conditions, palmitate derivative gives mass spectra 270 m/z at retention time 9.09 min and stearate 298 m/z at 11.86 min.

Mass isotopomer distribution was determined using methods that correct for the contribution of derivatizing agent and natural ¹³C abundance to mass isotopomer distribution of the compound of interest (19,21). This distribution was used to determine isotope incorporation and dilution by mass isotopomer distribution analysis according to Hellerstein et al. (22). Hepatic glucose and lactate production rates were determined using the following equation: production rate (mg · kg⁻¹ · min⁻¹) = infusion rate × (1/E_Tracer − 1), where infusion rate is the stable-isotope infusion rate, and E_Tracer is the plasma enrichment of stable isotope.

Hepatic glucose production (HGP) from the lactate was determined using the following equations as previously described (17):

In Eq. 3, M represents enrichment of plasma glucose molecules containing 1 to 6 (M1–M6) 13C carbons derived from 13C lactate, and m represents plasma enrichment of lactate molecules containing 1 to 3 (m1–m3) 13C carbons. Apparently, the fractional contribution of lactate to gluconeogenesis, and HGP from lactate, is influenced by multiple factors, such as, for example, dilution of plasma 13C lactate (decrease in m values) after feeding by metabolism of glucose and amino acids in the gut, resulting in increased fractional contribution of lactate to gluconeogenesis. However, plasma 13C glucose could also be diluted (decrease in M values) after feeding by absorption of dietary glucose in the gut, which would decrease fractional contribution of lactate to gluconeogenesis. Thus, multiple tracer dilutions are taken into account in the calculations. De novo fatty-acid synthesis was estimated by measuring incorporation of 13C (derived from infused [U-13C3]lactate) into the palmitate and stearate in the liver. De novo synthesis of palmitate was further quantitatively determined by calculating palmitate synthesis rate, as a fractional rate of lipogenesis, as described previously (23–25). In this study, we used [U-13C3]lactate as isotopic tracer, instead of deuterium oxide, as in the previous study (23–25). In liver, a portion of [U-13C3]lactate is converted to [U-13C2]acetyl-CoA, which is a m2 isotopomer of acetyl-CoA and used for fatty acid synthesis. Incorporation of 1–8 molecules of m2 [U-13C2]acetyl-CoA into palmitate will yield 1 molecule of m2, m4, m6, … , and m16 palmitate isotopomers, respectively. Based on binomial probability of isotopomer distribution, the enrichment of palmitate isotopomers from de novo fatty acid synthesis could be determined using following equation:

where p is the enrichment of endogenous, unlabeled acetyl-CoA, and q is the enrichment of [U-¹³C₂]acetyl-CoA. Thus, the enrichment of palmitate isotopomers is p⁸ for m0 palmitate, 8p⁷q for m2 palmitate, 28p⁶q² for m4 palmitate,… , and q⁸ for m16 palmitate. From the ratio of two consecutive palmitate isotopomers, the value of p and q could be determined. For instance, the ratio of

which gives

Since p + q = 1, P = 1 − q = 1 − (1 − p) = p/[p + (1 − p)] = [p/(1 − p)]/[1 + p/(1 − p)], which combines with Eq. 6 and gives Eq. 7:

Theoretically, when 100% of acetyl-CoA is used for de novo fatty acid synthesis, the theoretical M2 palmitate enrichment is expected to be M2 palmitate = 8p⁷q = 8p⁷ 1 − p, where the value p is obtained from Eq. 4. The de novo synthesis rate of palmitate in the liver is then determined from the ratio of (measured m2 palmitate enrichment)/(theoretic M2 palmitate enrichment) and expressed as a fraction of newly synthesized palmitate molecules (FNS) (23–25).

Total RNA was isolated from liver or skeletal muscle with Trizol (Invitrogen, Carlsbad, CA) and used as template for cDNA synthesis using oligo dT primers. Real-time PCR was performed with the iCycler iQ real-time detection system (BioRad, Hercules, CA) using SYBR Green PCR QuantiTect reagent (Qiagen, Valencia, CA). Each assay was performed in triplicate using 25–50 ng cDNA. In addition, a standard curve of four serial dilution points of control cDNA (ranging from 100 ng to 100 pg), as well as a no-template control, were included in each assay. The relative concentration of genes of interest was determined by plotting the threshold cycle (C_t) versus the log of the serial dilution points and normalizing to expression levels of an endogenous control gene, TBP (TATA box binding protein). Primers used for real-time PCR were designed to span introns and were analyzed by conventional PCR and agarose gel electrophoresis to verify specificity before use in real-time assays. Primer sequences are given in Table 1.

All data are expressed as means ± SEM. Student’s t test or two-way ANOVA were used for statistical analyses, as specified in table/figure legends.

Triglyceride in adipose tissue and glycogen in liver and muscle are the two major energy storage sources in vertebrate species in the fasted state. Although glycogen is utilized during initial fasting periods, glycogen reserves are depleted after prolonged fasting, and fatty acids derived largely from adipose tissue then become the major energy substrate. The reduced capacity for fat storage and lipolysis in the fld mouse raised the possibility that compensatory adaptations in energy substrate use may occur in these animals. This hypothesis is supported by the previous observation that energy expenditure is increased in fld mice and that the average RQ is altered as assessed by indirect calorimeter (14). The RQ reflects the type of metabolic fuels utilized and is calculated as the ratio of carbon dioxide produced to oxygen consumed, where oxidation of exclusively carbohydrate would produce a theoretical RQ of 1.00 and exclusive utilization of fatty acids an RQ of 0.70 (26). To further investigate how lipin deficiency alters energy substrate utilization, we measured RQ throughout the diurnal cycle. We found striking abnormalities in the fld mouse, particularly in the transition between fed and fasted states.

Throughout a 12-h fast, the RQ in wild-type mice drops as expected from ∼0.9 to 0.7, reflecting a shift from reliance largely upon carbohydrate during feeding to a reliance almost exclusively on fatty acids by the end of the fasting period (Fig. 1A). During the same period, the RQ in fld mice also diminishes, but the response is blunted. They maintain significantly higher RQ values than wild-type mice during the last 5 h of the fast, indicating failure to rely exclusively on fatty acid fuels. The most pronounced difference between wild-type and fld mice occurs following the transition from the fasted to fed state. Immediately upon introduction of chow diet (4.5% fat and 50% carbohydrate by weight), for wild-type mice there is a sharp transition from fatty acid utilization to a preference for carbohydrate substrates (RQ ∼0.90) (Fig. 1A). The RQ remains relatively high until the beginning of the fasting period, which induces a gradual decline to an RQ of 0.70 after 8 h of fasting. In contrast, fld mice do not undergo a pronounced shift to carbohydrate utilization upon feeding, reaching a maximum RQ of only 0.83 after 5 h of feeding (Fig. 1A). The fld mice therefore fail to undergo the normal switch to predominantly carbohydrate metabolism at any point during the feeding period. Paradoxically, the RQ in fld mice increases to its highest level (0.85–0.89) as soon as fasting begins (illustrated in both the initial and second fasting periods shown in Fig. 1A). The switch to carbohydrates at the beginning of the fasted period, and substantial utilization of fatty acids during the fed state, are the reverse of what occurs in normal mammals and indicate a fundamental alteration in metabolism in the lipin-deficient fld mouse.

To determine whether the abnormalities observed in fuel utilization of fld mice could be complemented by increasing the supply of fatty acid and carbohydrates in the diet, indirect calorimetry was also performed on animals that had been fed a high-fat/high-carbohydrate (35% fat and 33% carbohydrate by weight) diet for 6 weeks. In wild-type animals, the diet repressed the RQ throughout the feeding period by ∼0.05 units (Fig. 1B), likely reflecting increased utilization of fatty acids from the diet. The RQ profile of wild-type mice during fasting was comparable on the chow and high-fat/high-carbohydrate diets. The fld mouse exhibited only minor changes in response to the high-fat/high-carbohydrate diet compared with the chow diet (Fig. 1C). A comparison of wild-type and fld mice on the high-fat/high-carbohydrate diet emphasizes the greater utilization of carbohydrates by the fld during a period that is normally fueled by fatty acids (Fig. 1D). The abnormality in substrate utilization during the fasting period and transition to feeding in fld mice remains, even when the animals are supplied with diet that contains 35% of fat by weight. (Fig. 1D). The 24-h integrated RQs showed no difference between wild-type and fld mice under regular chow as well as high-fat/high-carbohydrate feeding conditions (data not shown). These results suggest that there is a profound metabolic disturbance in fld mice as a consequence of lipin deficiency.

The reduced capacity for fat storage and elevated RQ observed in fld mice after prolonged fasting raised the possibility that glycogen may have increased importance as an energy source in these animals during fasting. Consistent with this possibility, measurement of tissue glycogen content showed that fld mice had about twice as much glycogen in both liver and skeletal muscle as wild-type mice when refed for 5 h with chow diet after an 18-h overnight fast (Fig. 2). This could not be attributed to differences in food intake, as wild-type and fld mice consumed the same amount of food during the 5-h refeeding period (wild-type 2.01 ± 0.40 vs. fld 2.22 ± 0.39 g). The liver glycogen content in both wild-type and fld mice after 18-h fasting were dramatically reduced to ∼4.5 mg/g tissue weight and showed no significant difference between the wild-type and fld mice (data not shown). Together with the RQ data, these results suggest that fld mice store excess glycogen during the fed state for use after prolonged fasting, when fatty acids released by adipose tissue lipolysis would normally be used but are unavailable in the fld mouse due to lack of adipose tissue.

In addition to increased tissue glycogen content, fld mice exhibited elevated fasting plasma glucose levels compared with wild-type mice (wild-type 106 ± 16 vs. fld 150 ± 13 mg/dl, P < 0.001 by Student’s t test). Fasting insulin levels (wild-type 0.25 ± 0.06 vs. fld 0.10 ± 0.02 ng/ml) and fasting glucagon levels (wild-type 40 ± 2.0 vs. fld 35 ± 4.4 pM) were comparable between the wild-type and fld mice. These data suggested whole-body insulin resistance in the fld mouse, consistent with previous results of glucose tolerance tests in these animals (13). The elevated fasting glucose levels in the fld mouse could result from either increased HGP and/or reduced peripheral glucose uptake. Measurement of fasting total HGP rate using [U-¹³C₆]glucose revealed a 43% reduction in fld mice (Fig. 3), implying a significant decrease in fld peripheral glucose uptake in the 18-h fasting period. These results suggest that whole-body insulin resistance in fld mice is due to decreased peripheral insulin sensitivity and that hepatic insulin sensitivity is probably increased in the 18-h–fasted fld mouse.

Normally, lactate produced by the Cori cycle supplies one-third of the three-carbon substrates to the liver for gluconeogenesis during fasting (27). To determine whether the decreased total HGP was associated with altered Cori cycle activity, we measured lactate production rate using a [U-¹³C₃]lactate infusion. As shown in Fig. 4A, lactate production rate in both wild-type and fld mice increased in response to the 5-h refeeding. However, fld mice had a significantly lower lactate production rate, which was 66% of the wild-types in the fasted state and 74% of the wild-types in the refed state (Fig. 4A). Plasma lactate concentrations reflected lactate production rates, with decreased levels in fld mice (Fig. 4B). The decrease in lactate production in the fld mice contributed to the significantly reduced hepatic gluconeogenesis from lactate in the fld mouse (Fig. 4C). The decrease in both lactate production and HGP from lactate could contribute to the observed decrease in total HGP in the fasted fld mice.

The HGP from lactate showed no difference between the fasted and fed states for the wild-type mice (Fig. 4C). HGP from lactate in the fld mice, however, was 33% lower than wild-type in the fasted state and was further suppressed to a level of 73% lower than wild-type mice in the fed state (Fig. 4C). The decrease in HGP from lactate, particularly in the fed state, was much greater than the decrease in lactate production, despite the relative similarity of plasma lactate levels (Fig. 4). This suggests that fld mice had an overall reduction in flux through the Cori cycle, in which hepatic glucose is supplied to muscle, which subsequently returns lactate to liver as a gluconeogenic substrate. This decrease in hepatic gluconeogenesis from lactate suggested that a higher proportion of lactate flux in the fld mouse, particularly in the 5-h refed state, was directed toward fatty acid synthesis, even though these animals had a profound reduction in lipid storage capacity.

To determine the fate of lactate in the fld mice, we followed the conversion of infused [U-¹³C₃]lactate to hepatic fatty acids after 5 h of refeeding. Through the action of lactate dehydrogenase and pyruvate dehydrogenase (PDH), [U-¹³C₃]lactate can be converted to [U-¹³C₂]acetyl-CoA, an m2 isotopomer that serves as a direct substrate for fatty acid synthesis. Incorporation of one molecule of m2 acetyl-CoA into fatty acid will yield an m2 fatty acid isotopomer, such as m2 palmitate or m2 stearate. As shown in Fig. 5, enrichment of m2 palmitate and m2 stearate extracted from the liver of refed fld mice was ∼29-fold and 5-fold higher than in wild-type mice, indicating that a greater proportion of the infused [U-¹³C₃]lactate was directed to fatty acid synthesis in the fld mouse. To eliminate the possibility that the difference in hepatic m2 fatty acid enrichment between wild-type and fld mice was due to the increased enrichment of plasma [U-¹³C₃]lactate concentration (Table 2), we determined the de novo synthesis rate of palmitate, which is a fractional rate of lipogenesis, in the liver using a modification of methods described by Hellerstein and coworkers (23–25).

The plasma m3 lactate loses one of its ¹³C carbons and is converted to m2 acetyl-CoA after entering the hepatic pyruvate pool. Based on the measured m2 palmitate enrichment in the liver, the hepatic enrichment of theoretical m2 acetyl-CoA and theoretical M2 palmitate can be calculated (research design and methods). As shown in Table 2, the actual analyses of liver extracts showed that 2.6% of newly synthesized palmitate in the fld mice was m2 palmitate, compared with only ∼0.1% of m2 palmitate in wild-type mice. The theoretical value of hepatic M2 palmitate is expected to be 10.4% of total palmitate in the fld mice and 16.2% in the wild-type mice (Table 2, calculations shown in research design and methods). The ratio of measured m2 palmitate to theoretical M2 palmitate (m2/M2 palmitate in Table 2) is the fraction of newly synthesized palmitate (FNS) that is derived from acetyl-CoA, which represents the fractional rate of de novo fatty acid synthesis (23–25). The FNS for wild-type and fld mice was ∼1 and 27%, respectively (Table 2). Considering that the fld mouse exhibits lipodystrophy without ectopic fat accumulation, the rate of de novo fatty acid synthesis in the fld mouse is unusually high. Together with indirect calorimetry, our data indicate that fatty acids in the fld mouse are mobilized to peripheral tissues as energy substrates during the feeding period.

To determine whether the changes revealed by metabolic flux analysis are reflected by altered gene expression in pathways controlling net fatty acid synthesis and substrate utilization, we examined mRNA levels of enzymes that are involved in gluconeogenesis, tricarboxylic acid (TCA) cycle, and fatty-acid/triglyceride synthesis using real-time RT-PCR. These enzymes are phosphoenolpyruvate carboxykinase (PEPCK), pyruvate carboxylase, pyruvate kinase, pyruvate dehydrogenase kinase isoform 4 (PDK4), ATP:citrate lyase, acetyl-CoA carboxylase, and fatty acid synthase.

PEPCK and pyruvate carboxylase are two key enzymes that commit pyruvate/lactate to hepatic gluconeogenesis. The mRNA levels of these two enzymes in the fld mice were significantly lower than in the wild-type mice in both fasted and refed states (Fig. 6A). In response to refeeding, PEPCK gene expression in both wild-type and fld mice were downregulated by a similar degree, indicating that 5-h refeeding had a similar suppressive effect on hepatic gluconeogenesis regardless of genotype. However, liver pyruvate carboxylase mRNA levels in the fld mice remained unchanged in response to refeeding, in contrast to a significant downregulation of pyruvate carboxylase mRNA in the wild-type mice (Fig. 6A). This gene expression profile of PEPCK and pyruvate carboxylase supports our earlier observation of decreased hepatic gluconeogenesis from lactate/pyruvate. Since pyruvate carboxylase plays a important role in replenishment of TCA cycle intermediates (by the conversion of pyruvate to oxaloacetate) to facilitate fatty acid synthesis (28,29), the suppressed PEPCK and unchanged pyruvate carboxylase gene expressions in the refed fld mice would favor the redirection of metabolic flux toward fatty acid synthesis.

Acetyl-CoA, the substrate for hepatic de novo fatty acid synthesis, is produced from pyruvate through the reaction of PDH. The activity of PDH is negatively regulated by PDK4 (30,31). In fld mice, hepatic PDK4 gene expression was 3.6-fold higher than in wild-type mice in the fasted state, and there was no difference in the refed state compared with wild-type mice (Fig. 6A), indicating a more significant downregulation of PDK4 in the fld mice during the fasted to fed transition. Downregulation of PDK4 would release its suppression on PDH, allowing conversion of pyruvate to acetyl-CoA by PDH. In agreement with this, hepatic pyruvate kinase gene expression showed ∼29-fold upregulation in the fld mice in response to refeeding, compared with less than 2-fold upregulation in wild-type mice (Fig. 6A). Thus, PDK4 and pyruvate kinase gene expression data indicate that a greater proportion of metabolic flux through pyruvate in the fld mice is directed to the formation of acetyl-CoA after refeeding. Examination of de novo fatty acid synthesis enzymes showed significantly greater upregulation of ATP:citrate lyase, acetyl-CoA carboxylase, and fatty acid synthase after refeeding in fld compared with wild-type mice (Fig. 6B).

In chow-fed fld mice, the increased fatty acid synthesis together with the higher RQ during fasting, as well as the increased glycogen storage and lower RQ during refeeding, suggested an adaptation to conserve glucose for prolonged fasting. This most likely stems from the impaired fat stores and reduced HGP observed in the fld mouse. Evidence supporting glucose conservation in the fld mouse includes our earlier observation of a twofold increase in both liver and skeletal muscle glycogen stores. In addition, skeletal muscle hexokinase II mRNA levels in fld mice were only 20% of those in wild-type mice after an 18-h fast and were upregulated sixfold after refeeding, in contrast to unchanged levels in wild-type mice (Fig. 6B). Skeletal muscle, by virtue of its mass, is a major determinant of energy metabolism. The hexokinase II gene expression profile indicates that glucose uptake by the skeletal muscle in the fasted fld mouse was decreased, perhaps as a compensatory response to spare glucose for utilization by glucose-dependent organs such as brain. The dramatic upregulation of hexokinase II after refeeding in the fld mouse is consistent with the increased skeletal muscle glycogen storage as a means of glucose conservation.

Fatty liver is a common feature in disorders characterized by “too much” (obesity) or “too little” (lipodystrophy) fat, as a result of excess fatty acids or inadequate adipose tissue storage, respectively. Individuals with these disorders are predisposed to insulin resistance and diabetes, which are thought to be partly associated with ectopic deposition of triglyceride and lipotoxicity (32). A major difference between human lipodystrophies and the lipodystrophic fld mouse is the spontaneous resolution of the fatty liver by 2 weeks of age in the mouse (33). In addition, the fld mouse does not develop severe hyperglycemia, as is seen in human lipodystrophies (3,5). These previous observations suggested that partitioning of energy substrates, namely glucose and free fatty acids, in the adult fld mouse may be altered.

The studies presented here demonstrate that lipin deficiency leads to novel changes in mobilizing and utilizing energy substrates in the fld mice during the diurnal metabolic cycle. Indirect calorimetry indicated that fatty acids serve as an important metabolic fuel for fld mice during the fed state, whereas wild-type animals rely primarily on glucose. Using metabolic flux phenotyping analysis, we determined that the fatty acids utilized by fld mice during the fed state are likely available because of a 27-fold increase in de novo hepatic fatty acid synthesis in fld compared with wild-type mice. Previous studies have shown that levels of plasma triglycerides and free fatty acids are comparable between adult wild-type and fld mice under the ad libtum feeding condition (13), and fatty liver and hyperlipidemia were observed only in neonatal fld mice (13,33). Thus, we propose that the reliance of adult fld mice on fatty acids as an energy source may explain, in part, why ectopic fat deposition, and hepatic steatosis, does not occur in adult fld mice. In contrast to the increased utilization of fatty acids during the fed state, fld mice exhibit increased glucose utilization during the fasting period, especially during the final 5 h of a 12-h fast. The increased glycogen storage appears to be possible because of the reduced reliance on glucose oxidation during the fed state and suggests that glycogen serves as a surrogate energy reserve in fld mice in the face of inadequate adipose tissue reserves.

Adipocyte-derived hormones are established to have important roles in regulating energy metabolism and homeostasis. As with other mouse lipodystrophy models (34,35), it is reasonable to postulate that low levels of adipocyte-secreted hormones may contribute to metabolic alterations in the fld mouse. For example, leptin administration can reverse insulin resistance seen both in human and mouse lipodystrophies (34,36), in part via hypothalamic actions (37,38). Both leptin and adiponectin levels in the fld mouse are substantially reduced (10,12). However, among these lipodystrophic mouse models, only the fld mouse lacks a fatty liver in adulthood. Thus, although impaired adipokine action may contribute, it cannot account for the unique phenotype and metabolic adaptation observed in the fld mouse, suggesting that these are a consequence specifically of lipin deficiency.

Our data reveal that lipin deficiency results in impaired normal diurnal metabolic fuel switching in the fld mice. The key changes in fld compared with wild-type mice were detected at points that direct metabolic flux toward energy storage versus energy expenditure. The integration of metabolic flux studies, calorimetry, and gene expression analysis leads to an integrated view of the altered metabolic profile of the fld mouse, as illustrated in Fig. 7. In the fed state, fld mice cannot store significant amounts of triglyceride in adipose tissue due to impaired adipocyte function. Rather than being stored, fatty acids synthesized in the liver are therefore utilized as energy substrates. As we determined, hepatic fatty acid synthesis is dramatically increased in the fld mouse with refeeding, mainly for peripheral utilization (Fig. 1), sparing glucose molecules for glycogen synthesis. This results in enhanced glucose flux toward glycogen deposition in the liver and muscle. In the fasted state, glycogen stored in the fld liver and muscle becomes a prominent fuel source in the absence of normal adipose tissue stores. The supply of glycerol for gluconeogenesis in fld liver should be reduced, due to decreased lipolysis from limited amounts of adipose tissue. Lactate supply for gluconeogenesis is reduced (Fig. 4) due to increased peripheral glucose oxidation. Glucose uptake by skeletal muscle is also decreased, which could spare glucose for glucose-dependent organs such as brain. Thus, the lipin deficiency results in a reversed diurnal metabolic cycle. This adaptation in fuel selection and energy conservation in the fld mouse may account for the amelioration of fatty liver in the adult animals. It is important to acknowledge that we assessed regulation of metabolic pathway enzymes by examining mRNA expression, which may not be a reliable gauge of resulting enzyme activity due to posttranscriptional effects. However, the stable isotope-based flux phenotyping studies done here provide a “visual” assessment of the fate of labeled molecules as they are metabolized by specific pathways and the rate at which these labeled molecules pass through these pathways.

The changes observed in the metabolism of the fld mouse suggest that lipin may play a role in partitioning nutrients for energy homeostasis. This is consistent with the previous demonstration that lipin is a downstream target of the mammalian target of rapamycin (39), which functions at the interface between nutrient sensing and the regulation of phosphatidylinositol 3-kinase–mediated metabolic responses (40–42). It was shown that in rat adipocytes, lipin becomes phosphorylated in response to mammalian target of rapamycin activation in response to insulin (39), raising the possibility that lipin may have a role as an effector of this response. Recent work has demonstrated that variations in lipin expression levels in adipose tissue of both mice and humans are associated with glucose homeostasis parameters (43). In populations of mice and humans exhibiting a range of lipin mRNA expression levels in adipose tissue, lipin levels were inversely correlated with glucose, insulin, and insulin resistance as measured by the homeostatic model assessment of insulin resistance index. Furthermore, genetic haplotypes in the human lipin gene, LPIN1, were associated with serum insulin levels and BMI. These results demonstrate that lipin levels in humans as well as mice are associated with metabolic homeostasis and suggest that studies performed in mice to further elucidate the mechanisms involved could shed light on these processes in humans as well.

This work was supported by National Institutes of Health Grants DK 58132 (to I.J.K.) and HL28481 (to K.R.), a grant from the American Diabetes Association (to I.J.K.), and a grant supporting gas chromatography/mass spectrometry metabolic tool development from the Center for Biotechnology, SUNY Stony Brook and Ingenious Targeting Laboratories (to I.J.K.).

We thank Ping Xu and Sara Bassilian for excellent technical assistance.