Lipin expression preceding peroxisome proliferator-activated receptor-gamma is critical for adipogenesis in vivo and in vitro.

We recently identified mutations in the lipin gene, Lpin1, as the cause of lipodystrophy in the fatty liver dystrophy (fld) mouse. Here we identify impaired adipocyte differentiation as the basis for lipodystrophy in lipin-deficient mice and demonstrate that lipin is required for normal induction of the adipogenic gene transcription program. We found that the reduced adiposity in chow fed fld mice and resistance to obesity in fld mice fed a high-fat diet is associated with reduced adipogenic gene expression. Using primary mouse embryonic fibroblasts isolated from fld mice, we confirmed that lipin deficiency prevents normal lipid accumulation and induction of key adipogenic genes, including peroxisome proliferator-activated receptor (PPAR)gamma and CCAAT enhancer-binding protein (C/EBP)alpha. However, our previous studies of daily gene expression in differentiating 3T3-L1 preadipocytes indicated that lipin expression is undetectable until about day 3 of differentiation, at a point after PPARgamma and C/EBPalpha gene expression is established. This paradox was resolved by examining gene expression at 10-h intervals during 3T3-L1 cell differentiation, leading to detection of transient lipin expression at 10 h into the differentiation program, prior to the induction of PPARgamma and C/EBPalpha. Consistent with a requirement for lipin expression upstream of PPARgamma, differentiation of lipin-deficient mouse embryonic fibroblasts could be rescued by ectopic expression of PPARgamma. Thus, we conclude that lipin expression is required prior to PPARgamma during adipocyte differentiation.

We recently identified mutations in the lipin gene, Lpin1, as the cause of lipodystrophy in the fatty liver dystrophy (fld) mouse. Here we identify impaired adipocyte differentiation as the basis for lipodystrophy in lipin-deficient mice and demonstrate that lipin is required for normal induction of the adipogenic gene transcription program. We found that the reduced adiposity in chow fed fld mice and resistance to obesity in fld mice fed a high-fat diet is associated with reduced adipogenic gene expression. Using primary mouse embryonic fibroblasts isolated from fld mice, we confirmed that lipin deficiency prevents normal lipid accumulation and induction of key adipogenic genes, including peroxisome proliferator-activated receptor (PPAR)␥ and CCAAT enhancer-binding protein (C/EBP)␣. However, our previous studies of daily gene expression in differentiating 3T3-L1 preadipocytes indicated that lipin expression is undetectable until about day 3 of differentiation, at a point after PPAR␥ and C/EBP␣ gene expression is established. This paradox was resolved by examining gene expression at 10-h intervals during 3T3-L1 cell differentiation, leading to detection of transient lipin expression at 10 h into the differentiation program, prior to the induction of PPAR␥ and C/EBP␣. Consistent with a requirement for lipin expression upstream of PPAR␥, differentiation of lipin-deficient mouse embryonic fibroblasts could be rescued by ectopic expression of PPAR␥. Thus, we conclude that lipin expression is required prior to PPAR␥ during adipocyte differentiation.
The increasing prevalence of obesity and type II diabetes in our society has focused attention on the development and function of adipose tissue. It is well established that adipose tissue performs a variety of metabolic functions, including energy partitioning and production of endocrine hormones. Studies in human and mouse models of obesity have revealed that excess adipose tissue confers increased risk for diabetes, dyslipidemia, and coronary heart disease (1)(2)(3). Studies of lipodystrophy reveal that insufficient adipose tissue can predispose to these same conditions, indicating that normal adipose tissue development and function is critical for metabolic homeostasis (4 -7).
The gene defects and cellular mechanisms underlying con-genital lipodystrophies in humans have become a focus of intensive investigation in recent years (reviewed in Refs. 4, 8 -11). Heterogeneity exists in both the genetic basis and in the potential cellular mechanisms for different forms of lipodystrophy in humans. Thus far, mutations in four genes have been identified in human congenital lipodystrophies. Mutations in LMNA, encoding the nuclear envelope proteins lamin A and C, cause familial partial lipodystrophy (Dunnigan variety) (12,13). A distinct form of partial lipodystrophy has been associated with heterozygous missense mutations in PPARG, encoding the peroxisome proliferator-activated receptor-␥ (PPAR␥) 1 nuclear transcription factor, which plays a key role in adipogenesis (14,15). Mutations in BSCL2, a novel gene of unknown function with prominent expression in the brain, cause Berardinelli-Seip complete lipodystrophy (16). Mutations in AG-PAT2 (acylglycerol-3-phosphate acyltransferase), encoding an enzyme required for triglyceride synthesis, also cause congenital generalized lipodystrophy (17). Although defined mechanisms exist by which PPARG and AGPAT2 mutations might confer adipose tissue deficiency, the role of LMNA and BSCL2 mutations are not well established. It has been proposed that LMNA mutations impair adipocyte differentiation by affecting the activity of sterol regulatory element binding protein-1, a transcription factor for lipogenic gene expression during adipocyte differentiation (18), but further work is required to fully establish such a mechanism. Several mouse models of lipodystrophy and obesity resistance have been generated through gene knockout and transgene technologies (reviewed in Refs. 7,[19][20][21]. Analysis of the physiological basis for fat depletion in these models has revealed that alterations in multiple processes can lead to reduced adipose tissue stores. These alterations include impaired fat cell precursor proliferation (null mutations in the high mobility group protein Ic, Ref. 22 The fatty liver dystrophy (fld) mouse is unique among the lipodystrophic mouse models mentioned above in that it is the result of a spontaneous mutation in an endogenous gene. The fld mouse resembles human patients with generalized lipodystrophy, having dramatically reduced adipose tissue mass throughout the body and acquired insulin resistance and increased susceptibility to atherosclerosis (32). Using a positional cloning approach, we identified the causative mutation in the Lpin1 gene, which encodes a novel protein, lipin (33). Consistent with the lipodystrophic phenotype of fld mice, lipin is prominently expressed in white and brown adipose tissue and is induced during differentiation of 3T3-L1 preadipocytes. Although the molecular function of lipin remains unknown, recent studies of the lipin homolog in Schizosaccharomyces pombe revealed that yeast lipin interacts with three proteins having roles in nuclear envelope structure and nucleocytoplasmic transport (34). This finding is consistent with our previous demonstration that lipin localizes to the nucleus (33). Intriguingly, lipin has also been shown to be phosphorylated in response to insulin, an event that is dependent upon the mammalian target of rapamycin signaling pathway (35).
Here, we investigate the cellular mechanisms underlying lipodystrophy in lipin-deficient fld mice. We found that the lack of adipose tissue in these animals is a direct consequence of impaired adipocyte differentiation due to a critical role for lipin in this process. Specifically, we establish that lipin is required during the initial stages of adipogenesis for the induction of key adipogenic factors, including PPAR␥ and CCAAT enhancer-binding protein (C/EBP)␣, such that lipin deficiency prevents maturation of adipocytes in vitro and causes lipodystrophy in vivo.

EXPERIMENTAL PROCEDURES
Mice and Diets-BALB/cByJ ϩ/fld mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and bred to produce lipin-deficient (fld/fld) and wild-type (ϩ/ϩ and ϩ/fld) mice for studies. Mice were fed a standard laboratory chow diet (Purina 5001) or high-fat diet containing 35% fat and 33% carbohydrate (Diet F3282, Bio-Serve, 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." Adipose Tissue Measurements and Feed Conversion Efficiency-Fat pads (inguinal subcutaneous, gonadal, and retroperitoneal) were dissected and weighed after conclusion of the high-fat diet, or at 5 months of age in chow fed mice. Cell numbers in adipose tissue depots were determined by fluorometric DNA quantitation using Hoechst dye 33258 from pre-weighed inguinal adipose tissue depots (36). Food intake was determined over 10 days of ad libitum feeding. Feed conversion efficiency was calculated as the weight gain per effective food intake (food consumption normalized for fecal lipid output). Fecal lipid content was determined by quantitation of triglycerides and fatty acids in lipid extracts prepared from dried feces (37).
Cell Culture-The 3T3-L1 cell line was maintained in Dulbecco's modified Eagle's medium containing 10% bovine serum. Mouse embryonic fibroblasts (MEFs) were derived from 18 day wild-type (WT) and fld embryos. Early passage cells (passage 3 or earlier) were used for differentiation and retroviral infection of primary cells. Spontaneously immortalized cell lines were developed as described (38) and used after passage 15. Adipocyte differentiation was initiated after 2 days at confluence with the addition of differentiation medium as described (39). For differentiation of MEFs and retrovirus-infected cells, adipocyte differentiation medium was supplemented with the PPAR␥ ligand, rosiglitazone (BRL 49653, a generous gift from Dr. Todd Leff) as described (40). After 2 days, differentiation mixture was removed and culture was continued in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum with insulin and rosiglitazone. At day 6, cells were lysed and homogenized for RNA isolation and triglyceride determinations, or fixed in 4% paraformaldehyde and stained with Oil Red O (41). All experiments were performed in triplicate.
The pBabe-PPAR␥2 retroviral expression vector was obtained from Dr. Bruce Spiegelman. To produce the lipin viral vector, lipin cDNA (GenBank accession no. AF180471) was subcloned into pBabe. Retroviral infection of cells was performed essentially as described (42).
Statistical Analysis-The number of mice or cell culture replicates for each study is indicated in the figure legends. Values are presented as means Ϯ S.E. A two-tailed Student's t test was used to calculate p values.

Lipin Deficiency Prevents Diet-induced Obesity and Impairs
Adipose Tissue Gene Expression-The phenotype of the fld mouse revealed that lipin deficiency prevents adipose tissue accumulation under normal dietary conditions. However, it was unclear whether the effect of lipin deficiency on adipose tissue could be overcome by dietary manipulation to enhance fat accumulation. Therefore, we investigated whether fld mice could be made to gain fat by feeding a high-fat diet. After 15 weeks on the high-fat diet, WT mice increased their body weight by 27% (8.3 Ϯ 1.8 g), whereas fld mice gained virtually no weight (0.4 Ϯ 1.1 g) (Fig. 1a). The fld mice maintained 15-fold lower subcutaneous and retroperitoneal fat pad mass compared with WT animals (Fig. 1, b and c), with no increase over the fat mass of these mice on a chow diet (32). To determine whether reduced adipose tissue mass could result from impaired preadipocyte proliferation, we determined fat pad cellularity in mice fed chow and high-fat diets. Wild-type and fld mice had comparable cellularity on the chow diet, and both exhibited a trend toward slightly increased cellularity on the high-fat diet, indicating that lipin deficiency does not result in reduced fat pad cellularity (Fig. 1d). Nor could reduced fat in fld mice be attributed to reduced food intake or absorption. Wild-type and fld mice had identical food intake (Fig. 1e), but fld mice exhibited a 6-fold reduction in feed conversion efficiency (weight gain per food intake, normalized to food absorption; Fig. 1f).
We investigated whether lipin deficiency impairs adipose tissue accumulation through altered expression of adipogenic genes. On the chow diet, fld mice exhibited ϳ2-fold reductions in expression of the adipogenic transcriptional regulators, PPAR␥ and C/EBP␣, and in the adipocyte fatty acid binding protein (aP2), a marker of adipocyte maturation and PPAR␥ target gene (Fig. 2). The expression of diacylglycerol acyltransferase (DGAT), a rate-limiting enzyme of triglyceride synthesis, was reduced 3-fold in fld tissue, thus likely contributing to reduced triglyceride accumulation. Wild-type mice exhibited modest (PPAR␥, C/EBP␣) to substantial increases (aP2, DGAT) in gene expression levels in response to the high-fat diet. In contrast, fld expression levels were not increased by the highfat diet. These data indicate that lipin deficiency alters basal expression of adipogenic genes in adipose tissue and impairs the response of these genes to changes typically observed with dietary manipulation, reflecting the failure of fld mice to develop increased adipose tissue mass on the high-fat diet.
Lipin Is Required for Adipocyte Differentiation in Vitro-The results described above implicate impaired adipocyte differentiation as the mechanism for diminished adiposity in lipin-deficient mice. To verify that this depends upon the expression of lipin within the adipocyte, we examined the adipocyte differentiation capacity of fld MEFs. Mouse embryonic fibroblasts were induced to differentiate into adipocytes by incubation in adipocyte differentiation mixture supplemented with the . e and f, daily food intake and feed conversion efficiency in mice from panel a, calculated from the weight gain per effective food intake (see "Experimental Procedures") after 6 weeks on the high-fat diet. *, p Ͻ 0.01. PPAR␥ ligand, rosiglitazone (40). After 6 days of treatment, ϳ20% of WT MEFs assumed a mature adipocyte morphology and accumulated large lipid droplets, as detected by Oil Red O staining (Fig. 3a). In contrast, less than 1% of the fld cells accumulated lipid droplets, which were sparse and small in size. Concordant with the reduced lipid staining, intracellular triglyceride content in fld MEFs was only 3.5% of WT levels (Fig. 3b). Real-time RT-PCR quantitation of mRNA for PPAR␥, C/EBP␣, and aP2 showed the expected induction by day 3 of differentiation in WT MEFs, but virtually no induction in fld cells (Fig. 3, c-e). Pref-1, a negative regulator of adipogenesis, which is normally expressed at high levels in preadipocytes and decreases during adipocyte differentiation (43), remained elevated in fld compared with WT MEFs (Fig. 3f).
To confirm that the impaired adipogenesis observed in fld MEFs was a direct result of lipin deficiency, we complemented fld MEFs by infection with a retroviral vector expressing lipin, and then induced adipocyte differentiation. Ectopic expression of lipin, but not vector alone, rescued the adipogenic defects in fld MEFs, leading to the development of cells that morphologically resembled adipocytes (Fig. 4a). Expression of PPAR␥, C/EBP␣, and aP2 was dramatically increased by lipin reconstitution (Fig. 4b), reaching levels similar to those observed in differentiated WT MEFs (compare with Fig. 3, c-e). These data demonstrate that reduced triglyceride accumulation and failure to induce key adipogenic genes in fld cells is a direct consequence of lipin deficiency. Lipin Acts Upstream of PPAR␥ during Adipogenesis-The observation that PPAR␥ gene expression is compromised in lipindeficient adipocytes and adipose tissue suggested that lipin func- tion is required at a time point preceding PPAR␥ induction during adipogenesis. PPAR␥ gene expression is induced between days 1 and 2 of 3T3-L1 preadipocyte differentiation (42). However, our previous studies indicated that lipin expression in differentiating 3T3-L1 cells is virtually undetectable at days 1 and 2, and then increases to levels observed in mature adipocytes at days 3 and 4 (33). We wondered whether lipin expression does in fact occur at an earlier time point that we had not analyzed previously; therefore, we performed a time course to examine gene expression every 10 h for the first 40 h of 3T3-L1 preadipocyte differentiation, and at 6 days in mature adipocytes.
The expression patterns we observed for known adipogenic factors were consistent with what has been reported previously (44 -47). Thus, Pref-1, a negative regulator of adipogenesis was expressed at the highest levels at 0 -10 h, declined as differentiation proceeded, and was undetectable in mature adipocytes (Fig. 5). In contrast, PPAR␥ and C/EBP␣ were expressed at very low levels initially and were not induced until 20 h; aP2 was not expressed at appreciable levels until after 40 h. C/EBP␤ and C/EBP␦, which are known to be transiently induced within the first day of treatment with differentiation mixture (48), showed a spike in expression levels at 10 h, and returned to a lower level at 20 h. Lipin showed a strikingly similar pattern, with a transient induction at 10 h and a return to low levels at 20 -40 h, followed by high levels in mature adipocytes. The early spike in lipin expression would not have been detected in our previous studies because it falls between day 0 and day 1 time points that were analyzed in that study (33). Notably, the induction of lipin expression at 10 h preceded the induction of PPAR␥ expression at 20 h, which is consistent with the possibility that lipin functions upstream of PPAR␥.
To directly test whether lipin acts upstream of PPAR␥ in adipogenesis, we performed a genetic rescue of lipin-deficient cells with ectopic PPAR␥ expression. Infection of an fld MEF cell line with a PPAR␥-expressing retrovirus followed by treatment with differentiation medium led to increased Oil Red O staining and a 9-fold increase in triglyceride accumulation (Fig.  6, a and b). Treatment with the PPAR␥ ligand, rosiglitazone, also produced a substantial 16-fold increase, suggesting that activation of the small amount of PPAR␥ present in fld cells is effective in promoting adipogenesis when provided with sufficient ligand. The combination of viral PPAR␥ and rosiglitazone had the greatest effect, producing triglyceride levels that approached those observed in WT MEFs differentiated in the presence of rosiglitazone (see Fig. 3b). In concert, the expression of genes induced in mature adipocytes, including aP2 and fatty acid synthase, were dramatically increased in response to ectopic PPAR␥ expression or rosiglitazone treatment, with the highest levels occurring when both were present (Fig. 6c). Thus, PPAR␥ expression or activation in fld cells can overcome the block in adipogenesis, suggesting that lipin is required for normal expression or function of this key adipogenic factor. DISCUSSION The findings presented here establish impaired adipocyte differentiation as the basis for lipodystrophy and resistance to diet-induced obesity in fld mice. Our evidence for the role of lipin in adipocyte differentiation derives from a combination of in vivo and in vitro studies. In vivo, lipin deficiency prevents normal adipose tissue development, with dramatically reduced tissue mass and aberrant adipocyte gene expression. The requirement for lipin in adipogenesis also explains why fld mice fail to increase adipose tissue mass on a high-fat diet, even when supplied with excess substrates for lipid storage. Evidence that the differentiation defect is intrinsic to preadipocytes includes the demonstration that MEFs isolated from fld mice proliferate normally in culture but fail to differentiate, mirroring the reduced triglyceride accumulation and aberrant gene expression observed in adipose tissue from fld mice. We further determined that enhancing PPAR␥ expression or activity in fld MEFs overcomes the block in differentiation, indicating a requirement for lipin in attaining optimal PPAR␥ function during early stages of adipogenesis.
The induction of lipin expression at 10 h of 3T3-L1 adipocyte differentiation is consistent with a role for lipin in determining subsequent events in adipogenesis, such as down-regulation of the adipogenic inhibitor, Pref-1, or induction of adipogenesispromoting factors such as PPAR␥ or C/EBP␣. Previous studies (28,42,43) indicated that failure of any of these events is sufficient to prevent normal adipocyte differentiation. The mechanism by which lipin-deficiency attenuates adipogenic gene expression is not known, but recent studies of the fission yeast lipin homolog reveal interactions with three proteins having roles in nuclear envelope structure or nucleocytoplasmic transport (34). This is consistent with the nuclear localization of lipin (33), and suggests that, unlike other known adipogenic factors, lipin may have a novel molecular function.
The expression profile of lipin during the early stages of adipocyte differentiation is nearly identical to that of C/EBP␤ and C/EBP␦, which are also induced at high levels at 10 h after stimulation with differentiation mixture followed by a sharp decline before the onset of PPAR␥ and C/EBP␣ mRNA accumulation (48). The similarity in expression dynamics between lipin and these two C/EBP family members suggests that the three genes may be induced by common factors, or even be co-regulated, during early adipogenesis. The expression of C/EBP␤ and C/EBP␦ genes is known to be induced directly by adipogenic stimuli, with C/EBP␤ responding to dexamethasone through the glucocorticoid signaling pathway, and C/EBP␦ responding to methylisobutylxanthine induction of the cAMP signaling pathway (48). It has also recently been shown (49) that C/EBP␤ and C/EBP␦ are target genes of the sterol regulatory element binding protein-1c transcription factor, raising the possibility that the lipin gene may share some of these same regulatory mechanisms.
It is notable that the adipogenic gene expression defect was more pronounced in isolated fld cells than in adipose tissue from fld mice. A similar situation has been observed in the comparison of gene expression in cells versus adipose tissue from double-knockout mice carrying null alleles for both C/EBP␤ and C/EBP␦. As with lipin-deficient mice, the C/EBP␤-␦ double-knockout mice exhibit profoundly reduced adipose tissue mass in vivo and failure of embryonic fibroblasts to differentiate in vitro, but adipogenic gene expression was much more strongly impaired in vitro (29). It was proposed that this may reflect a compensatory response in vivo that does not occur in vitro. Additional support for an in vivo compensatory response in fld mice is provided by our previous observation (32) that adipose tissue from 2-week-old fld mice exhibits pronounced PPAR␥ expression, perhaps corresponding to a developmental period in which lipin-independent mechanisms are active. Nevertheless, the failure to develop significant adipose tissue or mature adipocytes in fld mice indicates that compensatory mechanisms are not adequate to overcome the requirement for lipin in adipose tissue development.
Our findings that lipin expression is biphasic, with a transient induction during early stages of preadipocyte differentiation, as well as high levels in mature adipocytes (33), suggests that lipin may have dual roles in adipocyte biology. Because lipin-deficient mice never develop mature adipocytes, additional models will be required to evaluate the role of lipin in mature adipose tissue. We have recently developed a transgenic mouse model with enhanced lipin expression driven by regulatory elements expressed specifically in mature adipo- cytes. 2 Preliminary studies with this model indicate that increased lipin expression in mature adipose tissue is associated with increased adiposity, lending support to the proposal that lipin functions in mature as well as differentiating adipocytes.