Lipin 2 Is a Liver-enriched Phosphatidate Phosphohydrolase Enzyme That Is Dynamically Regulated by Fasting and Obesity in Mice*

Lipin 1 is a bifunctional intracellular protein that regulates fatty acid metabolism in the nucleus via interactions with DNA-bound transcription factors and at the endoplasmic reticulum as a phosphatidic acid phosphohydrolase enzyme (PAP-1) to catalyze the penultimate step in triglyceride synthesis. However, livers of 8-day-old mice lacking lipin 1 (fld mice) exhibited normal PAP-1 activity and a 20-fold increase in triglyceride levels. We sought to further analyze the hepatic lipid profile of these mice by electrospray ionization mass spectrometry. Surprisingly, hepatic content of phosphatidate, the substrate of PAP-1 enzymes, was markedly diminished in fld mice. Similarly, other phospholipids derived from phosphatidate, phosphatidylglycerol and cardiolipin, were also depleted. Another member of the lipin family (lipin 2) is enriched in liver, and hepatic lipin 2 protein content was markedly increased by lipin 1 deficiency, food deprivation, and obesity, often independent of changes in steady-state mRNA levels. Importantly, RNAi against lipin 2 markedly reduced PAP-1 activity in hepatocytes from both wild type and fld mice and suppressed triglyceride synthesis under conditions of high fatty acid availability. Collectively, these data suggest that lipin 2 plays an important role as a hepatic PAP-1 enzyme.

Spontaneously arising mutations in the gene encoding lipin 1 (Lpin1) cause severe metabolic abnormalities in fatty liver dystrophic (fld) mice (1), which exhibit neonatal growth retardation, fatty liver, and dyslipidemia (2). These phenotypic manifestations abruptly resolve at ϳ14 days of age. However, adult fld mice exhibit lipodystrophy, insulin resistance, and susceptibility to atherosclerotic lesion formation (3,4). Recent evidence regarding the molecular functions of lipin 1 has begun to explain the severe metabolic phenotype of these mice.
Lipin 1 catalyzes the Mg 2ϩ -dependent dephosphorylation of phosphatidic acid (phosphatidic acid (PA) 2 phosphohydrolase (PAP-1) (5)) to form diacylglycerol (DG), the penultimate step in the Kennedy pathway of triglyceride (TG) synthesis (Fig. 1A). As proof of the importance of lipin 1-mediated PAP-1 activity, fld mice almost completely lack this enzymatic activity in adipose tissue, skeletal muscle, and heart (6), and the inability to synthesize TG in adipocytes may explain the defect in adipogenesis in fld mice.
Interestingly, lipin 1 is a soluble protein, and significant lipin 1 localization to the nucleus has also been demonstrated (1). Moreover, recent evidence suggests that lipin 1 functions as a regulator of gene transcription via protein-protein interactions with DNA-bound transcription factors (7,8), including the peroxisome proliferator-activated receptor ␣ (PPAR␣) and its coactivator protein PPAR␥-coactivator 1␣ (PGC-1␣) (7). PAP-1 catalytic activity is not required for peroxisome proliferator-activated receptor ␣ coactivation, and these two molecular functions seem to be completely separable. Thus, lipin 1 is poised to play important roles in controlling the metabolism of fatty acids at multiple regulatory levels.
Given the role of lipin 1 as an enzyme in TG synthesis, the hepatic phenotype of fld mice is somewhat surprising. First, neonatal fld mice exhibit severe hepatic steatosis because of over-accumulation of TG (2). Second, although other tissues of fld mice are severely deficient in PAP-1 activity, the liver of adult fld mice retains significant Mg 2ϩ -dependent PAP activity (6,9) and normal rates of TG synthesis (10). These findings suggest that other PAP-1 enzymes are active in liver. Based on sequence homology in signature N-and C-terminal domains, two additional lipin family proteins (lipin 2 and lipin 3) have been identified (1). Importantly, lipin 2 and 3 exhibit PAP-1 activity, although the relative activity and V max varies (lipin 1 Ͼ lipin 2 Ͼ lipin 3) (9). In the current study we further evaluated the hepatic phenotype of fld mice by using mass spectrometry-based lipidomic profiling of glycerophospholipid species and subsequently characterized the function and regulation of the liver-enriched PAP-1 protein, lipin 2.
Short-term fasting studies were performed with individually housed male mice which were either food deprived for 36 h (beginning at 2000) or food deprived 24 h (beginning at 2000) and then given ad libitum access to standard rodent chow for 12 h. All animal experiments were approved by the Animal Studies Committee of Washington University School of Medicine.
Mammalian Cell Culture-HepG2 cells were maintained in DMEM, 10% fetal calf serum. Primary mouse hepatocytes were isolated from mice as previously described (12). Adenoviral transductions were carried out by incubating hepatocytes with adenoviruses in a 5% fetal bovine serum, DMEM medium overnight at an multiplicity of infection of 8. Endpoints were collected 24 h later in lipin 2 overexpression studies or 72 h later when using RNAi. TG synthesis rates were quantified by using [ 3 H]glycerol as previously described (7,12).
DNA Constructs-Full-length triple HA-tagged mouse lipin 2 was cloned into pCDNA3 expression construct, subcloned into the Ad-track cytomegalovirus vector, and recombined into the Ad-EASY system. The lipin 2 shRNA oligonucleotide was designed to target nucleotides 145-164 of the mouse Lpin2 transcript and was cloned into the Invitrogen pENTR vector (Invitrogen) and then subcloned into the Ad-EASY system. Adenoviral-driven shRNA construct targeted to LacZ was utilized as a control. The pSPORT-lipin 3 expression vector has been previously described (7).
siRNA Oligonucleotides and Transfection-A Silencer siRNA construction kit (Ambion, Austin, TX) was used to synthesize three siRNAs for mouse lipin 2 and a scrambled, nonspecific sequence siRNA (control). The targeted nucleotide cDNA sequence for the three lipin 2 siRNAs are AACTCTA-CAAGGGCATTAACC (siRNA 50), AACCCAAACCAGCT-GCTAAAG (siRNA 1118), and AAAGGCTCTTGAGCCT-GAAGT (siRNA 1215). The siRNAs were transfected onto primary hepatocytes using Lipofectamine 2000 (Invitrogen). Based on efficiency of knockdown (supplemental Fig. 1A), siRNA 1118 was chosen for all metabolic studies. Transfection was carried out after the hepatocytes were cultured for 3 h in 6-well plates, and a dosage of 100 pmol of siRNA/well was used. Experiments were carried out 60 -72 h post-siRNA transfection.
Mass Spectrometric Lipid Analysis-The quantitative analyses of TG and phospholipid species have been described (13,14). In brief, lipids were extracted from mouse liver using a modified Bligh and Dyer technique in the presence of internal standards for electrospray ionization-mass spectrometric (ESI-MS) analysis. Electrospray ionization mass spectrometric analyses were performed utilizing a TSQ Quantum Ultra Plus tri-ple-quadrupole mass spectrometer (Thermo Fisher Scientific, San Jose, CA) equipped with an automated nanospray apparatus (Nanomate HD, Advion Bioscience Ltd., Ithaca, NY).
PAP-1 Activity-Phosphatidic acid phosphatase (PAP-1) activity was determined by using a modification of the method of Carman and Lin (15) to measure phosphate release from Triton X-100/phosphatidic acid mixed micelles. Labeled phosphatidic acid was purified by thin layer chromatography after using [␥-32 P]ATP and Escherichia coli diacylglycerol kinase to phosphorylate 1,2-dioleoyl-sn-glycerol. Samples of cell or tissue extracts (10 l) were added to reaction mixtures (100 l) containing a final concentration of 2 mM Triton X-100, 10 mM ␤-mercaptoethanol, 0.2 mM [ 32 P]phosphatidic acid (10,000 cpm/nmol), 50 mM Tris-maleate (pH 7.0), and either no added Mg 2ϩ or 1.5 mM MgCl 2 . After 20 min at 30°C the reactions were terminated by adding 0.5 ml of 0.1 N HCl in methanol. [ 32 P]Phosphate was extracted by vigorous mixing after adding 1 ml of chloroform and 1 ml of 1 M MgCl 2 . The amount of soluble [ 32 P] phosphate in 0.5 ml of the aqueous phase was measured by scintillation counting. Mg 2ϩ -dependent PAP activity (PAP-1) was determined by subtracting activity obtained in the absence of MgCl 2 from activity measured in the presence of 1.5 mM MgCl 2 .
DG Acyltransferase Activity-Measurements of DG acyltransferase activity was carried out under apparent V max conditions for 10 min at room temperature in a final volume of 200 l as described (16). Briefly, each assay reaction contained 50 g of liver membrane proteins, 5 mM MgCl 2 , 1.25 mg/ml bovine serum albumin, 200 mM sucrose, 100 mM Tris-HCl (pH 7.4), 25 M [ 14 C]oleoyl-CoA (American Radiolabeled Chemicals; final specific radioactivity: 18,000 dpm/nmol), and 200 M sn-1,2dioleoylglycerol (Sigma). The acyl acceptor (sn-1,2-dioleoylglycerol) was introduced into the reaction mixture as a liposome, which was prepared by mixing with phosphatidylcholine in at a molar ratio of 1:5 in a chloroform solution followed by evaporating chloroform under a stream of N 2 and resuspending liposomes in water. Each individual assay reaction was started by adding liver membrane proteins to the reaction mixture and terminated by adding 1.5 ml of chloroform/methanol (1:2.2, v/v) after a 10-min incubation period. Total lipids were extracted from this final mixture and separated on Linear-K Preadsorbent TLC plates (Waterman Inc., Clifton, NJ) with hexane/ethyl ether/acetic acid (80:20:1, v/v/v). The spots corresponding to TG was scraped from the TLC plates and determined for 14 C radioactivity using a scintillation counter. Reaction mixtures in which liver membrane proteins were omitted were used as blanks for background correction.
Northern and Western Blotting Analyses-Northern blotting analyses were performed using radiolabeled cDNA probes. Full-length cDNAs for mouse lipin 2 and lipin 3 were excised from the pCDNA3-HA-lipin 2 or pSPORT-lipin 3 expression vectors by restriction digest, gel-purified, and then labeled with [ 32 P]dCTP using Klenow enzyme. Unincorporated nucleotides were removed by using NucTrap Probe Purification Columns (Stratagene). Hybridizations were performed using Quickhyb (Stratagene) for 2 h at 65°C. Blots were washed and then exposed to autoradiographic film.
Quantitative Real-time Reverse Transcription-PCR-Real-time reverse transcription-PCR was performed using the ABI PRISM 7500 sequence detection system (Applied Biosystems, Foster City, CA) and the SYBR green kit. Arbitrary units of target mRNA were corrected by measuring the levels of 36B4 RNA. The sequence of the oligonucleotides used in quantitative reverse transcription-PCR analyses can be found in supplemental Table 1.
Metabolic Labeling Studies-Primary hepatocytes isolated from adult WT and fld mice were plated onto 12-well plates. Hepatocytes were infected with adenovirus to overexpress lipin 2 and cultured in complete culture media for 20 -24 h. After this initial culture period, the hepatocytes were washed three times with phosphate-buffered saline and incubated in Metand Cys-free DMEM for 1 h. For pulse-chase studies, the medium was replaced with 1 ml of Met-and Cys-free DMEM containing 250 Ci of 35 S-Promix (530 MBq/ml; Amersham Biosciences) for 2 h. For pulse-only studies, labeled methionine was administered for 30 min. After this pulse period, the hepatocytes were chased in 1 ml of DMEM containing 10 mM unlabeled Met and 3 mM Cys (1000ϫ excess) for the specified time periods. After the specified times, cells were lysed in an immunoprecipitation buffer (150 mM NaCl, 5 mM EDTA, 50 mM Tris (pH 7.4), 0.0625 M sucrose, 0.5% Triton X-100, and 0.5% sodium deoxycholate) containing a mixture of protease inhibitors (Roche Applied Science). The 35 S-labeled lipin 2 and albumin proteins in the cell lysates were immunoprecipitated and quantified by SDS-PAGE analyses as described for the analysis of 35 S-labeled apoB proteins (10).
Liver Protein Fractionation by Gradient Centrifugation-Two adult male C57BL/6 mice were fasted overnight (14 h) and then sacrificed for tissue collection. Whole, fresh liver tissue from both mice was finely minced with a razorblade and then pooled. Minced livers were transferred to a Teflon-piston glass homogenizer with 2 ml of fractionation buffer (0.25 M sucrose, 10 mM HEPES, 1 mM EDTA (pH 7.4)) and homogenized with 4 stokes at 240 rpm. The homogenate was centrifuged for 5 min at 1000 ϫ g. The pellet containing the nuclei was resuspended in 1-ml fractionation buffer. The post-nuclear supernatant was layered onto a 10-ml linear 20 -50% (w/w) sucrose gradient. The gradient was centrifuged for 3 h at 150,000 ϫ g in a 14 ϫ 89-mm tube. The floating fraction was recovered with a tube slicer. Twelve fractions were collected, and proteins were separated by SDS-PAGE. An equal volume of each fraction was loaded so that the quantity of protein varied as is noted in Fig. 3C.
Statistical Analyses-Statistical comparisons were made using analysis of variance or t tests. All data are presented as the means Ϯ S.E., with a statistically significant difference defined as a p value Ͻ0.05.

Liver of fld Mice Exhibits Increased TG and Diminished PA
Content-Despite the important role that lipin 1 plays in TG synthesis, livers of neonatal fld mice have a whitish appearance indicative of neutral lipid accumulation and contain very high levels of DG and TG (2,17). Previous work has also demonstrated that hepatic PAP-1 activity is unchanged (9) or moderately decreased (6) in adult fld mice. However, PAP-1 activity in steatotic fld livers has not been assessed. We found that PAP-1 activity actually tended to be increased in post-natal day 8 (P8) fld mouse liver compared with littermate controls (Fig. 1B), confirming that this pathway is operative and that lipin 1 is not required for full PAP-1 activity. Hepatic DG acyltransferase activity was also unchanged in fld liver.
We next sought to analyze levels of hepatic lipid species downstream of lipin 1 in the Kennedy pathway by using ESI-MS. These analyses confirmed that total TG content was increased 20-fold in homozygous fld mouse liver at P8 compared with littermate control (WT) mice (Fig. 1C). The increase in TG content in fld mice was observed across all major TG species (supplemental Table 2). The content of other phospholipids derived from DG and, thus, downstream of PAP-1 (phosphatidylcholine, phosphatidylethanolamine, and phosphatidylserine ( Fig. 1C and data not shown) were unchanged. We next evaluated levels of lipid intermediates upstream of lipin 1 in the TG synthesis pathway. Surprisingly, the hepatic content of the PAP-1 substrate phosphatidic acid (PA) was markedly depleted in lipin 1-deficient livers ( Fig. 2A). Phosphatidylglycerol (PG) levels were also significantly diminished in fld liver (Fig. 2B). This was due primarily to depletion of the most abundant PA and PG species (Fig. 2, A and B). Cardiolipin (CL), which is derived from PG, was also strikingly depleted in fld liver (Fig. 2C). PAP-1 activity is not required for the conversion of PA to PG or CL, and thus, depletion of these phospholipids is likely caused by diminished availability of PA. Conversely, phosphatidylinositol (PI) levels were unchanged between WT (4.18 Ϯ 0.55 nmol/mg of protein) and fld (4.63 Ϯ 0.49 nmol/mg of protein) mice.
Lipin 2 Is a Hepatic-enriched Lipin Family Member-Given the preservation of PAP-1 activity in fld liver, we next sought to characterize the regulation and function of other lipin family proteins. Whereas expression of Lpin2 is quite high, the expression of Lpin3 mRNA was undetectable in liver by northern blotting analyses (Fig. 3A). When probed against mouse tissue lysates, antibodies raised against lipin 2 peptides detected a band in liver (Fig. 3B) that co-migrates with lipin 2 protein overexpressed via an adenovirus (data not shown). Other tissues that were examined were practically devoid of lipin 2 protein.
Lipin 1 is a soluble protein that can also be localized to the nucleus or associate with endoplasmic reticulum membranes (1,6,18). To determine the subcellular localization of lipin 2 in intact mouse liver, hepatic lysates were fractionated by sucrose  gradient centrifugation (19). To reflect the absolute quantity of lipin 2 in the liver, the resulting fractions were loaded by volume so that the total protein loaded in each lane varied (as noted above the representative image in Fig. 3C). Lipin 2 localization was then evaluated by comparison with markers of the subcellular compartments including the lipid droplet (ADRP), soluble (liver fatty acid-binding protein), endoplasmic reticulum membrane (calnexin), and nuclear (acetylated histone H3) fractions. Small amounts of lipin 2 were detected in lipid droplet, endoplasmic reticulum membrane, and nuclear fractions (Fig. 3C). However, the majority of lipin 2 was detected in the soluble cytoplasmic fraction (Fig. 3C). This soluble fraction also contains the highest levels of lipin 2 protein when equal amounts of protein were loaded in each lane (Fig. 3D).
Similar to lipin 1 (6,20), the lipin 2 band often appears as a doublet when well resolved, likely because of phosphorylation (18). The serine 106 residue of lipin 1 is a major site of phosphorylation (6), and the peptide sequence surrounding this site is conserved with lipin 2 (supplemental Fig. 1B). We confirmed that an antibody directed against phosphorylated serine 106 of lipin 1 also detected lipin 2 (supplemental Fig. 1B). Like lipin 1, the majority of phosphoserine106 lipin 2 was detected in the soluble fraction of the liver lysates (Fig. 3D). Indeed, the majority of the phosphoserine 106 lipin 2 was found in the soluble fraction regardless of whether the gel was loaded for equal volume of the fractions or equal amounts of protein per lane (Fig.  3D). This suggests that under these conditions, the majority of lipin 2 protein is found in the cytoplasm but leaves open the possibility that, like lipin 1 (6), lipin 2 might translocate intracellularly in response to post-translational modification.
Lipin 2 Protein Is Increased in fld Mouse Liver via a Posttranscriptional Mechanism-Consistent with previous work (9, 10), Lpin2 mRNA expression was not changed in liver of P8 or P42 fld mice compared with littermate controls (Fig. 4A). However, lipin 2 protein content was increased severalfold in fld mice of either age (Fig. 4A). Moreover, at P42, loss of a single Lpin1 allele resulted in a significant increase in lipin 2 protein without affecting Lpin2 mRNA levels. The increase in lipin 2 protein levels in primary hepatocytes from fld mice is also preserved after several days in culture (data not shown). These data suggest that lipin 2 protein content is increased in fld liver independent of changes in steady-state Lpin2 mRNA levels.
We, therefore, sought to evaluate lipin 2 synthesis rates and turnover in pulse-chase experiments using metabolic labeling with [ 35 S]methionine. As shown in Fig. 4B, the rate of disappearance of 35 S-labeled lipin 2 was equal in hepatocytes isolated from WT and fld mice. However, the rate of lipin 2 synthesis in 30-min metabolic labeling studies (lower autoradiographs in Fig. 4B) was increased in fld hepatocytes, suggesting that the disconnect between mRNA and protein is due to increased rates of translation.
Lipin 2 Is Dynamically Regulated in Liver but Is Not a Target Gene of PGC-1␣-The hepatic expression of Lpin1 is dramatically increased by 24 h of fasting (7). Similarly, both lipin 2 mRNA and protein expression were significantly induced by food deprivation (Fig. 5A), suggesting that lipin 2 levels are also under transcriptional control. Our original interest in lipin 1 stemmed from the finding that the fasting-induced increase in Lpin1 expression required an intact PGC-1␣ pathway in liver (7). However, hepatic Lpin2 expression was induced equally in WT and PGC-1␣-deficient mice (Fig. 5B). Moreover, although Lpin1 expression is robustly activated by PGC-1␣ overexpression in hepatocytes, Lpin2 mRNA and lipin 2 protein were not changed after PGC-1␣ overexpression (Fig. 5C). Thus, the reg- ulatory circuits that govern hepatic Lpin1 and Lpin2 expression are notably divergent.
Lipin 2 Is an Important PAP-1 Enzyme in Liver-To evaluate the effects of lipin 2 on cellular PAP-1 activity, primary mouse hepatocytes were infected with an adenoviral vector to overexpress lipin 2. Lipin 2 adenovirus affected neither PAP-1 activity nor TG synthesis rates in hepatocytes (Fig. 6A). When HepG2 cells, which express little endogenous lipin 2, were infected with lipin 2 adenovirus, PAP-1 activity was robustly increased (Fig.  6B). However, this was also not sufficient to force an increase in  TG synthesis rates, as assessed using [ 3 H]glycerol as a tracer (Fig. 6B). Although we preferred to evaluate the metabolic effects of lipin 2 in bona fide hepatocytes, the high endogenous expression of lipin 2 and the regulated control of PAP-1 activity seem to preclude the interpretation of studies employing a gain-of-function approach in this cell type.
We, therefore, developed an adenovirus expressing lipin 2 shRNA, which markedly reduced lipin 2 protein in primary hepatocytes (Fig. 7A). Interestingly, lipin 2 knockdown led to a compensatory increase in Lpin3 mRNA expression, especially in fld hepatocytes (Fig. 7A). However, lipin 3 protein levels remained undetectable by Western blotting, and lipin 2 shRNA treatment significantly reduced hepatocyte PAP-1 activity in both WT and fld hepatocytes (Fig. 7B). Despite a marked decrease in PAP-1 activity, rates of TG synthesis were not significantly affected by lipin 2 knockdown in either genetic context (Fig. 7B). These findings suggest that even very low PAP-1 activity is sufficient for normal TG synthesis in hepatocytes under basal conditions. Given these surprising findings, we sought to evaluate the effects of lipin 2 knockdown in the context of increased fatty acid availability, which is a potent stimulus to induce TG synthesis. Due to concerns regarding potential increased sensitivity of steatotic hepatocytes to the toxic effects of adenovirus infection, we performed these experiments using siRNA oligonucleotide transfection. Lipin 2 or scrambled control siRNA were transfected into freshly isolated hepatocytes from P42 WT and fld mice. Again, lipin 2 knockdown led to a dramatic decrease in lipin 2 protein (supplemental Fig. 1A) and a marked diminution in hepatocyte PAP-1 activity (data not shown). Nonetheless, basal rates of TG synthesis were unchanged by lipin 2 siRNA even in fld hepatocytes (Fig. 8A). We, therefore, stimulated hepatocytes with oleic acid (0.5 mM) for 30 min before quantifying TG synthesis. As expected, oleate supplementation increased TG synthesis rates (note the difference in scale between the left-and right-hand graphs). In the context of exogenous oleate, lipin 1 deficiency combined with lipin 2 knockdown significantly suppressed TG synthesis rates (Fig.  8A), suggesting that PAP-1 activity had become rate-limiting.
We also probed the effects of lipin 2 knockdown in hepatocytes from P14 fld mice, which have increased cellular lipid levels and elevated rates of TG synthesis even without exogenous oleate (10). Lipin 2 siRNA administration led to a signifi- cant suppression of TG synthesis rates to levels observed in WT hepatocytes at base line and after oleate stimulation (Fig. 8B). Lipin 2 siRNA also tended to reduce TG synthesis rates after oleate treatment in WT hepatocytes of this younger age.

Increased Hepatic Expression of Lipin 2 Plays a Role in Increased Hepatic TG Synthesis Rates in ob/ob Mice-Given
that lipin 2 knockdown impaired TG synthesis rates in response to increased fatty acid availability, we evaluated the role of lipin 2 in controlling TG synthesis in steatotic ob/ob hepatocytes. We found that the expression of Lpin2 mRNA was not induced in db/db and ob/ob mice compared with littermate controls (Fig. 9A). However, Western blotting analyses revealed a striking increase in lipin 2 protein levels in both obese mouse models compared with littermate lean controls (Fig. 9B). This again implicates post-transcriptional control of lipin 2 levels as an important means to influence lipin 2 activity.
Hepatocytes from WT and ob/ob mice were then isolated and treated with lipin 2 siRNA to determine the effects of lipin 2 on hepatic TG synthesis. Interestingly, lipin 2 knockdown suppressed rates of TG synthesis in ob/ob hepatocytes to levels comparable with WT controls (Fig. 9C). Moreover, the oleatestimulated increase in TG synthesis in ob/ob hepatocytes was markedly blunted. These data collectively suggest that the increase in lipin 2 levels observed in ob/ob mice serves to boost the capacity for TG synthesis.

DISCUSSION
Lipin 1 is the first member of a family of Mg 2ϩ -dependent PA phosphohydrolase enzymes that catalyze the penultimate step in TG synthesis (5). Essentially no PAP-1 activity is detected in adipose tissue, muscle, and heart of lipin 1-deficient fld mice (6). However, liver of adult fld mice exhibits substantial hepatic PAP-1 activity (6,9). Herein we found that hepatic PAP-1 activity was not altered in liver of neonatal fld mice, which have profound hepatic TG over-accumulation. It is likely that the high expression of lipin 2 in liver and the lack of expression of other lipins in striated muscle and adipose tissue may explain these tissue-specific differences. Indeed, lipin 2 is enriched in liver, which seems to be the primary site of its expression. Moreover, lipin 2 protein levels were increased in liver of both 8-day-old and adult fld mice compared with littermate controls, suggesting that lipin 2 is up-regulated to compensate for lipin 1 deficiency. Finally, RNAi studies demonstrated that lipin 2 knockdown caused a marked reduction in PAP-1 activity in both WT and fld cells and indicate that lipin 2 is an important PAP-1 enzyme in hepatocytes.
These data sharply contrast recent findings using lipin 2 shRNA in HeLa cells (18). In that study lipin 2 knockdown actually led to increased PAP-1 activity due to a compensatory induction of lipin 1 mRNA and protein. Although there is no obvious explanation for these differences, the relative importance of lipin 1 and 2 as PAP-1 enzymes is clearly cell typespecific. Additionally, our lipin 2 shRNA data also suggest that lipin 3 does not play a significant role as a PAP-1 enzyme in liver. Previous studies suggested that Lpin3 expression is elevated in fld liver and could compensate for lack of lipin 1 (9). In this study Lpin3 mRNA was also significantly induced by lipin 2 knockdown, especially in fld hepatocytes. However, the absolute expression of lipin 3 remains very low, as mRNA and protein were undetectable by Northern blot and Western blotting even after lipin 2 knockdown. When taken with the low specific activity of lipin 3 (9), its role as a PAP-1 enzyme in liver is probably minimal. However, PAP-1 activity was not completely abolished by combined lipin 1 deficiency with lipin 2 knockdown. The remaining activity is likely due to 1) residual lipin 2 protein content, 2) the actions of lipin 3, or 3) the presence of other PAP-1 enzymes that have not yet been identified.
Conversely, lipin 2 overexpression was not sufficient to increase PAP-1 activity in primary hepatocytes. The activity of lipin proteins seems to be highly regulated by compartmentalization and/or post-translational modification (6,18). It is likely ob/ob ob/ob FIGURE 9. Lipin 2 is induced by obesity-related diabetes. A, the graph depicts mean expression of Lpin2 in liver of ob/ob, db/db, or matched littermate lean control mice. B, representative Western blots showing expression of lipin 2 and actin protein in the same mice. C, the graph depicts mean rates of TG synthesis under basal (bovine serum albumin (BSA) with no additional fatty acids) or oleate-stimulated (0.5 mM for 30 min) conditions using hepatocytes isolated from WT or ob/ob mice.
that these factors limit PAP-1 activity in hepatocytes. However, in HepG2 cells lipin 2 overexpression caused a robust increase in PAP-1 activity. It is worth noting that HepG2 cells express little endogenous lipin 2, whereas hepatocytes express an abundance of this protein. Nonetheless, lipin 2 overexpression did not stimulate TG synthesis in HepG2 cells despite a 4-fold increase in PAP-1 activity. Also, marked reductions in hepatocyte PAP-1 activity did not affect rates of TG synthesis under basal conditions, even in fld hepatocytes. This residual activity seems to be sufficient for normal TG synthesis under these conditions. It is generally accepted that the glycerol-3-phosphate acyltransferase step in this pathway is rate-limiting for TG synthesis (21), and our data are consistent with this.
As evidenced by experiments with adult fld mice or adenovirus overexpression, increased lipin 2 expression per se does not drive TG synthesis or accumulation. It is likely that lipin 2 upregulation in neonate fld mice is not the primary driving force for hepatic steatosis. Rather, fatty acid availability strongly impacts triglyceride synthesis rates, and increased supply of lipid to the liver secondary to lipodystrophy and/or high dietary fat content is probably the primary stimulus. Consistent with this, the hyperlipidemia and fatty liver phenotypes of fld mice coincidentally occur and resolve (2). Lipin 2 up-regulation likely plays a permissive role by increasing the capacity for TG synthesis stimulated by high fatty acid availability. When hepatocytes were challenged with high levels of endogenous or exogenous fatty acids, we unveiled the latent phenotype elicited by loss of lipin 2 and demonstrated significant reductions in rates of TG synthesis. Although obesity is at the other end of the phenotypic spectrum from lipodystrophy, obese mouse models also have high circulating lipid levels as well as increased hepatic lipin 2. Our findings suggest that lipin 2 is also involved in the increased rates of TG synthesis in obese hepatocytes as well. Further experiments are warranted to determine whether chronic suppression of lipin 2 in obese liver would show efficacy at reducing hepatic steatosis.
The fatty liver of fld mice is a metabolic signature of this model. Previous work has shown that the hepatic content of DG and TG is significantly increased in these mice (2,17). We explored the hepatic lipid profile of these mice further by using ESI-MS analysis with the hypothesis that the substrate of lipin 1, phosphatidic acid, would accumulate in these livers. Most surprisingly, PA was markedly depleted in liver of fld mice. This contrasts a recent report that PA accumulates in the endoneurium of fld mice (22). Lipin 2 does not seem to be expressed in the endoneural compartment, and it is possible that this somehow explains the observed differences. In addition, ESI-MS analyses demonstrated that phosphatidylglycerol and cardiolipin, which are derived from PA, were also depleted in fld liver. Cardiolipin is a critical component of the mitochondrial membrane (23), and abnormalities in cardiolipin synthesis and remodeling cause mitochondrial dysfunction in several genetic models (24,25). It is tempting to speculate that cardiolipin deficiency contributes to impairments in mitochondrial fatty acid oxidation in hepatocytes from neonatal fld mice (3) and potentially exacerbates hepatic steatosis in this model.
We also demonstrate herein that the regulatory circuits that control the expression of Lpin1 and Lpin2 are significantly divergent. Although both genes are induced by acute food deprivation ( Fig. 5 and Ref. 26), the induction of Lpin1 required the presence of the PGC-1␣ coactivator (7), whereas Lpin2 does not. This fits with data indicating that steady-state Lpin2 mRNA levels are not affected by glucocorticoids (26) 3 or diabetes (Fig. 9), which cause a marked increase in hepatic PGC-1␣ expression (27,28). Our previous work showed that one of the biological consequences of lipin 1 activation by PGC-1␣ during food deprivation was to feed-forward and enhance PGC-1␣ activity as a transcriptional coactivator (7). It remains to be determined whether lipin 2 also acts in the nucleus as a transcriptional regulatory protein and a component of the PGC-1 complex bound to chromatin in the nucleus.
Finally, we also determined that lipin 2 protein content was highly regulated independent of changes in steady-state mRNA levels. In fld mice, lipin 2 protein levels were increased, whereas Lpin2 expression was not. This feature of lipin 2 regulation was not unique to fld mice, as ob/ob and db/db mice also exhibit increased lipin 2 protein independent of changes in Lpin2 mRNA. The results of subsequent metabolic labeling studies indicate that increases in steady-state lipin 2 protein levels are mediated via increased rates of lipin 2 translation rather than enhanced protein stability. Translational control of lipin 2 is likely an important regulatory nodal point, and quantification of protein levels will be crucial for complete interpretation of future results.
Summary-The absence of lipin 1 results in a surprising hepatic phenotype that is likely explained at least in part by a compensatory activation of the liver-enriched lipin family member, lipin 2. The studies herein demonstrate that lipin 2 is an important hepatic PAP-1 enzyme that is under the control of unique regulatory mechanisms. Given recent evidence that genetic variations or mutations in the human gene encoding lipin 2 are linked to the development of diabetes (29) and a rare inflammatory bone disease (30,31), further work will be needed to define the full spectrum of metabolic pathways regulated by lipin 2 and to understand the molecular mechanisms that control lipin 2 activity.