Berardinelli-Seip congenital lipodystrophy type 2 (BSCL2) is the most severe form of human lipodystrophy, characterized by an almost complete loss of adipose tissue and severe insulin resistance. BSCL2 is caused by loss-of-function mutations in the BSCL2/SEIPIN gene, which is upregulated during adipogenesis and abundantly expressed in the adipose tissue. The physiological function of SEIPIN in mature adipocytes, however, remains to be elucidated. Here, we generated adipose-specific Seipin knockout (ASKO) mice, which exhibit adipocyte hypertrophy with enlarged lipid droplets, reduced lipolysis, adipose tissue inflammation, progressive loss of white and brown adipose tissue, insulin resistance, and hepatic steatosis. Lipidomic and microarray analyses revealed accumulation/imbalance of lipid species, including ceramides, in ASKO adipose tissue as well as increased endoplasmic reticulum stress. Interestingly, the ASKO mice almost completely phenocopy the fat-specific peroxisome proliferator–activated receptor-γ (Pparγ) knockout (FKO-γ) mice. Rosiglitazone treatment significantly improved a number of metabolic parameters of the ASKO mice, including insulin sensitivity. Our results therefore demonstrate a critical role of SEIPIN in maintaining lipid homeostasis and function of adipocytes and reveal an intimate relationship between SEIPIN and PPAR-γ.
Congenital generalized lipodystrophy (CGL), also known as Berardinelli-Seip congenital lipodystrophy (BSCL), is an autosomal recessive disorder characterized by a near total loss of adipose tissue, severe insulin resistance, and fatty liver (1,2). To date, four genes have been linked to CGL/BSCL, including 1-acylglycerol-3-phosphate-O-acyl transferase 2 (AGPAT2)/CGL1, SEIPIN/CGL2, CAVEOLIN/CGL3, and CAVIN/CGL4 (3). The most severe form of human CGL/BSCL is caused by mutations in SEIPIN/BSCL2, which encodes an integral membrane protein of the endoplasmic reticulum (ER) with no recognizable functional domains (3–5). We and others have generated Seipin knockout (KO) mice (6–8), which have severe lipodystrophy and insulin resistance, thereby proving an essential role of Seipin in adipogenesis in vivo. Interestingly, SEIPIN and its orthologs also control the expansion of lipid droplets (LDs) and lipogenesis (9–12). Therefore, SEIPIN can regulate lipid storage at systemic (adipogenesis) and cellular (LD expansion) levels.
Exactly how ER-localized SEIPIN may regulate adipogenesis remains an open question. The differentiation of preadipocytes requires a transcriptional cascade that ultimately leads to the activation of the master regulator of terminal adipogenesis: peroxisome proliferator–activated receptor-γ (Pparγ), which, together with its coactivators, stimulates the expression of a large number of gene products, including those that promote lipogenesis and glucose transport (13–15). White adipose tissue (WAT) and brown adipose tissue (BAT) are completely lost in a mouse model lacking Pparγ, and mutations in PPARγ in the human population are associated with fat loss (14). It has been proposed that the absence of SEIPIN may lead to the accumulation of certain phospholipid species, such as phosphatidic acid (PA), which may serve as strong PPAR-γ antagonists, thereby causing lipodystrophy (3).
Similar to PPARγ, SEIPIN is highly expressed in adipose tissue, but its expression is low in liver and barely detectable in muscle. The expression of Seipin is dramatically increased at later stages of the differentiation of 3T3-L1 cells (16,17). However, little is known about the in vivo function of SEIPIN in mature adipocytes. Here, we used the Cre-loxP system to generate adipose-specific Seipin KO (ASKO) mice, and our results reveal striking phenotypic similarities between the ASKO mice and the fat PPAR-γ–deficient (FKO-γ) mice (18): both show severe adipocyte hypertrophy, progressive lipodystrophy, insulin resistance, and fatty liver. Our results demonstrate that SEIPIN is required not only for the differentiation of preadipocytes but also for the maintenance of lipid homeostasis and long-term survival of mature adipocytes.
Research Design and Methods
All experiments involving mice were approved by the Institutional Animal Care Research Advisory Committee of Peking University Health Science Center. The Principles of Laboratory Animal Care (NIH Publication 85-23, revised 1996) were followed.
Homozygous Seipinfl/fl mice were obtained as described (6). Adipose-specific deletion of Seipin exon 3 was induced by crossing Seipinfl/fl mice to transgenic mice expressing Cre recombinase driven by an aP2 promoter (18). The genotyping was examined by PCR using the following primers: for the Cre transgene: 5′-GCGGTCTGGCAGTAAAAACTATC-3′ and 5′-GTGAAACAGCATTGCTGTCACTT-3′; for the upstream loxP site: 5′-CTTGTCTCAAAGGGGTCT-3′ and 5′-TCAACAGAACAGACGCT-3′. Mice used in most studies were maintained on a mixed genetic background of 129 and C57BL/6. Mice for high-fat diet (HFD) treatment were from a C57BL/6 background after five generations of backcrossing. The HFD (40% kilocalories from fat) was fed to 6-week-old mice for 6 weeks. For rosiglitazone (Rosi) treatment, a chow diet containing Rosi (0.3 mg/g diet; Sigma-Aldrich, St. Louis, MO) was fed to 6-month-old mice for 10 weeks.
Blood was obtained by retro-orbital bleed. Plasma cholesterol, triacylglycerols (TAG), and glucose were determined using enzymatic methods (Sigma-Aldrich kits). Plasma insulin, leptin, and adiponectin were measured by ELISA (Linco Research, St. Charles, MO). Free glycerol content (GPO-Trinder kit, Sigma-Aldrich) and the level of nonesterified fatty acids (NEFA) were measured by a colorimetric assay (Wako Chemical, Osaka, Japan).
Glucose and Insulin Tolerance Tests
Mice were fasted overnight for 16 or 4 h, respectively, followed by intraperitoneal injection of glucose (2 g/kg body weight) or insulin (0.75 mIU/g body weight; Humulin). Blood samples were collected before (time 0) and at 15, 30, 60, and 120 (90 for insulin tolerance test) min after injection for glucose measurement.
Liver was cryostat sectioned at a thickness of 7 μm for Oil Red O staining. Paraffin-embedded WAT and BAT were sectioned at a thickness of 2 μm and stained with hematoxylin and eosin (H&E) or Sirius Red for fibrosis analysis. Adipocyte area was measured using ImageJ software (n = 200 adipocytes per animal, n = 5 animals per group). Immunodetections were performed with Mac2 antibody (Santa Cruz Biotechnology, Dallas, TX) to examine macrophage infiltration. TUNEL assay was done as described (19).
For in vivo lipolysis, mice were fasted for 4 h and given an intraperitoneal injection of the β3-adrenergic–specific agonist CL-316,243 (0.1 mg/kg, Sigma-Aldrich). Blood was collected before and 15 min after injection for determination of NEFA and glycerol levels. For ex vivo lipolysis, epididymal fat was removed, cut into 10-mg fat pads, and stimulated with or without 1 μmol/L isoproterenol (Sigma-Aldrich) as described by Chen et al. (7). The medium was collected for determination of glycerol levels. Intracellular cAMP concentrations were measured by immunoassay (Enzo Life Sciences, Farmingdale, NY).
RNA Isolation and Quantitative Real-Time PCR
Total tissue RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA), and first-strand cDNA was generated with a RT kit (Invitrogen). Quantitative real-time PCR was performed using primers listed in Supplementary Table 1. All samples were quantitated by the comparative CT method for relative quantitation, normalized to Gapdh.
Western Blot Analysis
Mouse tissue was homogenized in radioimmunoprecipitation assay buffer, and the protein content was determined using a bicinchoninic acid protein assay kit (Pierce, Rockford, IL). The following antibodies were used: Seipin (Abnova, Taipei, Taiwan); Akt, phospho-Akt (Ser473), hormone-sensitive lipase (HSL), phospho-HSL (Ser563), adipocyte triglyceride lipase (ATGL), and phospho–protein kinase-A (PKA) substrate (Cell Signaling Technology, Beverly, MA); FSP27 (a gift from Prof. Peng Li); ADRP (Santa Cruz Biotechnology); perilipin A (Abcam, Cambridge, U.K.); and GAPDH (Millipore, Billerica, MA). The protein bands were analyzed using densitometry and ImageJ image analysis software. Arbitrary densitometry units were quantified and are expressed as mean ± SEM.
Analysis of Liver Lipids
Liver (∼100 mg wet weight) was weighed and homogenized in 1 mL PBS. Lipids were extracted as described by Folch et al. (20) and dissolved in 1 mL 3% Triton X-100. The determination of TAGs was done using enzymatic methods, as described earlier.
Lipid Extraction and Lipidomic Analysis
Epididymal (Epi)-WAT (∼100 mg) or BAT (50 mg) was weighed and homogenized in 1 mL PBS. Lipids were extracted by adding methanol/chloroform (1:2), and lipidomic analysis was done as described (21).
Total RNA (100 ng) from fat was labeled and hybridized onto Affymetrix GeneChip Mouse 430 2.0 arrays (n = 4 controls and n = 4 KO, respectively) according to the manufacturer’s instructions. Data were analyzed as described (22).
All data are presented as means ± SEM. Statistical comparison between groups was performed using the Student t test or two-way ANOVA. A value of P < 0.05 was considered statistically significant.
Generation of the ASKO Mice
To examine the role of SEIPIN in mature adipocytes, we used the Cre/loxP system to generate mice in which Seipin is specifically deleted from the adipose tissue. Homozygous Seipinfl/fl mice were obtained as described (6) and crossed with transgenic mice expressing Cre recombinase under the control of the adipose-specific Fabp4/aP2 gene promoter (aP2-CreTg/0) (18). The resulting Seipinfl/+aP2-CreTg/0 progeny were then crossed with Seipinfl/fl mice to generate the ASKO mice. Litter mates lacking the Cre gene (Seipinfl/fl) were used as controls and are referred to as wild type (WT). Because aP2 is expressed only at late stages of adipocyte differentiation, this strategy is expected to delete Seipin after formation of fat depots, allowing normal differentiation of adipocytes. As revealed by real-time PCR, Seipin expression was almost completely lost in adipose tissue (the residual expression is likely from nonadipocytes), but not in liver, kidney, heart, and skeletal muscle (Fig. 1A). Seipin expression was greatly diminished in the Epi-WAT and BAT of 3-, 6- and 10-month-old ASKO mice (Fig. 1B and C). Seipin is not highly expressed in macrophages, and Seipin expression in intraperitoneal macrophages from the ASKO mice appeared unchanged (Fig. 1A). Therefore, the metabolic defects of the ASKO mice (below) most likely result from SEIPIN loss in adipocytes.
Progressive Lipodystrophy in ASKO Mice
ASKO mice fed with a chow diet showed significant and progressive total WAT loss: ∼25% loss at 3 months old, ∼50% at 6 months, and ∼75% at 10 months (Fig. 2A). Notably, the loss of WAT at different fat depots progressed at different rates (Supplementary Fig. 1A). Histological analyses showed that Epi-WAT from WT mice contained normal mature adipocytes, which were characterized by the presence of a unilocular LD. In contrast, adipocytes from 3-month-old ASKO mice showed signs of hypertrophy, and adipocytes from 6-month-old ASKO mice were highly hypertrophic (Fig. 2B and C). The adipocytes in 10-month-old ASKO mice were vastly variable in size, displaying very large unilocular vacuoles (LDs) or very small adipocytes containing brightly eosinophilic cytoplasm. The subcutaneous fat of the ASKO mice showed similar changes (Supplementary Fig. 1B).
We next examined the effect of Seipin deletion on the expression of adipocyte genes (Supplementary Fig. 2A). For Epi-WAT of ASKO mice, the expression of genes involved in lipogenesis, fatty acid uptake, and storage did not change in 3-month-old mice, and only Pparγ, Fabp4, C/ebpα, and Acc were downregulated in 6-month-old mice. In 10-month-old ASKO mice, multiple genes were downregulated dramatically. The expression of adipocytokines (adiponectin and resistin) was also downregulated in Epi-WAT of ASKO mice, but Leptin mRNA did not change significantly (Supplementary Fig. 2B).
BAT mass also decreased progressively, although not as severely as WAT. There was no diminution of the interscapular fat pads in 3-month-old ASKO mice. However, BAT mass in 6- and 10-month-old ASKO mice decreased by ∼40% and ∼50%, respectively (Fig. 2D). Strikingly, brown adipocytes from 3- and 6-month-old ASKO mice displayed giant, white adipocyte–like droplets (Fig. 2E). In 10-month-old ASKO mice, BAT adipocytes were replaced by a coagulum of amorphous eosinophilic material and cytoplasmic debris, implying necrosis (Fig. 2E). Despite these morphological changes, the expression of BAT-specific Ucp1 was unchanged in 3- and 6-month-old animals (Supplementary Fig. 2C). Nevertheless, the ASKO mice were cold-sensitive (Supplementary Fig. 2D).
Metabolic Characterization of the ASKO Mice
Metabolic parameters of ASKO mice were examined in fed and fasting states (Supplementary Table 2). When fed, ASKO mice showed significantly increased plasma TAG and NEFA and decreased adiponectin. Leptin was significantly decreased, whereas insulin increased only at 10 months. Short-term fasting caused little change, but fasting for 16 h led to significant decreases in plasma TAG in old ASKO mice. Plasma glucose increased in all three age groups upon a 16-h fast.
Lipodystrophy often leads to insulin resistance and glucose intolerance in humans and mice. The glucose tolerance test revealed delayed glucose clearance in 6- and 10-month-old ASKO mice and also dramatically increased insulin levels during glucose infusion (Fig. 3A and B). An insulin tolerance test showed that 10-month-old ASKO mice had impaired insulin sensitivity (Fig. 3C). To further assess tissue-specific sites of insulin resistance, we examined the expression of Akt, Glut4, insulin-receptor substrate 1 (Irs1), and Irs2 in WAT and liver. For 10-month-old ASKO mice, the expression of all four genes was markedly decreased in Epi-WAT (Supplementary Fig. 3A), whereas only the expression of Irs2 was decreased in liver (Supplementary Fig. 3B). To determine whether insulin signaling was impaired in WAT and liver of ASKO mice, we detected AKT phosphorylation (S473). As expected, the ratio of phospho-AKT to total AKT was markedly reduced in WAT of 6- and 10-month-old ASKO mice (Fig. 3D), and in the liver (Fig. 3E) of 10-month-old ASKO mice.
Lipodystrophy is often accompanied by fatty liver. Liver weight, gross morphology, and histology from 3-month-old ASKO mice were similar to those of WT mice. With aging, liver weight progressively increased (Fig. 3F). Oil Red O staining of cryosections showed the liver of 6-month-old ASKO mice contained more LDs than WT mice, and the liver of 10-month-old ASKO mice showed significant steatosis (Fig. 3G). Consistent with the histological observations, the amount of liver TAG of 6- and 10-month-old ASKO mice was 20% and 50% higher than that of WT mice (Fig. 3H). The expression of Fas, Scd1, and Pparγ was significantly increased in liver of 6- and 10-month-old ASKO mice (Supplementary Fig. 3C and D), although SREBP-1c mRNA and protein levels appeared unchanged (Supplementary Fig. 3C and E). Notably, no obvious changes in fat metabolism and insulin signaling were detected in ASKO muscle (Supplementary Fig. 3F and G).
ASKO Mice Are Resistant to Diet-Induced Obesity but Susceptible to HFD-Induced Insulin Resistance
To test the effects of HFD on ASKO mice, ASKO mice of mixed background were backcrossed with C57BL/6 for five generations, and the resulting mice were named ASKO-B6. When fed a chow diet, these mice showed similar metabolic properties as the ASKO mice, including lipodystrophy (Fig. 4A). When fed the HFD for 6 weeks, WT-B6 mice gained ∼20% body weight (Fig. 4B) and ∼100% total fat weight (Fig. 4C). In contrast, ASKO-B6 mice gained little fat pad and body weight, except the gonadal fat (Fig. 4B–D). Total plasma cholesterol, glucose, and especially insulin levels were significantly higher in ASKO-B6 mice after fasting for 4 h (Fig. 4E and F). Fatty liver is apparent in the KO but not the WT mice after the HFD (Fig. 4G and H). These results suggest that although ASKO-B6 mice are resistant to diet-induced obesity, they appear to be more susceptible to HFD-induced insulin resistance and fatty liver.
Macrophage Infiltration and Inflammation in Adipose Tissue of ASKO Mice
An early and striking change of the ASKO mice is the enlargement of LDs and adipocyte hypertrophy (Fig. 2B). Adipocyte hypertrophy may promote adipocyte death, macrophage infiltration, and chronic inflammation (22). Indeed, there is more apoptotic cell death in ASKO adipose tissue (Supplementary Fig. 4A). H&E staining of WAT and BAT had suggested infiltration of inflammatory cells in 6- and 10-month-old ASKO mice (Fig. 2B and E). Mac2-stained macrophages were almost absent in the Epi-WAT of WT and 3-month-old ASKO mice but prominent in 6-month-old ASKO mice and abundant in 10-month-old ASKO mice (Fig. 5A). Macrophages occurred individually or surrounding dead adipocytes to form crown-like structures (arrows in Fig. 5A). The number of infiltrating macrophages in the BAT of ASKO mice also increased dramatically with age (Fig. 5B). Consistent with the histological observations, Mac2 expression was increased in Epi-WAT and BAT of 6- and 10-month-old ASKO mice (Fig. 5C and D). The expression F4/80, another marker of macrophages, was also elevated in Epi-WAT and BAT of 10-month-old ASKO mice. A subset of proinflammatory M1 macrophage–associated genes (Mcp1 and Tnfα) was significantly upregulated in the BAT of 6-month-old ASKO mice, and the M1- and the prorepair M2 macrophage–associated genes were both upregulated in WAT and BAT of 10-month-old ASKO mice (Fig. 5C and D). These findings suggest that macrophages (especially M1) were increased in older ASKO mice, reflecting chronic inflammation. Finally, fibrosis was evident in the adipose tissue from 6- and 10-month-old ASKO mice (Supplementary Fig. 4B and C).
Impaired Lipolysis in ASKO Mice
Because impaired lipolysis can contribute to the enlargement of LDs and adipocyte hypertrophy (22), we examined lipolysis in WT and ASKO mice. In vivo lipolysis in ASKO mice was evaluated by measuring plasma NEFA and glycerol before and after administering the β3-adrenergic agonist CL-316,243. Baseline NEFA and glycerol levels were not significantly different between WT and ASKO mice. After 15 min of CL-316,243 treatment, WT mice showed a normal increase in glycerol (∼twofold) and NEFA (∼1.7-fold) levels, indicative of increased lipolysis, whereas little change was observed in 3- and 6-month-old ASKO mice (Fig. 6A). Isoproterenol-stimulated glycerol release was also markedly diminished in fat explants from ASKO mice compared with WT mice, although the basal levels of glycerol were almost the same (Fig. 6B). The level of cAMP was also reduced in ASKO mice after isoproterenol treatment (Fig. 6C). Lipolytic rates are tightly regulated by PKA-mediated phosphorylation of HSL and PERILIPIN1, and the interplay among PERILIPIN1, CGI58, HSL, and ATGL (23,24). Under basal conditions, ASKO mice exhibited reduced ATGL expression and HSL phosphorylation (Fig. 6D). Isoproterenol-induced phosphorylation of HSL was attenuated in fat explants of ASKO mice compared with WT mice (Fig. 6D). We further detected pan–phospho-Ser/Thr PKA substrates and found that isoproterenol-stimulated phosphorylation of PKA substrates was broadly reduced in ASKO mice compared with WT mice (Fig. 6D). Lipolysis in 10-month-old ASKO mice was similarly reduced as in 3-month-old mice (Fig. 6E). Moreover, reduced lipolysis was also detected in ASKO-B6 mice (Supplementary Fig. 4D and E).
Effects of Thiazolidinedione Treatment of ASKO Mice
To investigate the functional relationship between SEIPIN and PPAR-γ, we determined the effects of PPAR-γ agonists, the thiazolidinediones (TZDs) (14). WT and ASKO mice (6 months old) were treated with or without Rosi for 10 weeks. For WT mice, Rosi treatment resulted in an increase in body weight as well as in WAT mass (Fig. 7A). However, ASKO mice did not show such an increase in body weight or in total WAT mass even though adipose mass in the subcutaneous and inguinal areas was significantly increased (Fig. 7B). Many small and newly differentiated adipocytes appeared in subcutaneous WAT after Rosi treatment (arrows in Fig. 7C). Studies have reported differences between subcutaneous and visceral adipose tissues in function as well as in sensitivity to TZDs (25). These results were also consistent with those from lipodystrophy patients with TZDs treatment (26). The expression of PPAR-γ and its target genes increased in most of the examined genes in both genotypes after Rosi treatment (Fig. 7D). BAT also increased in mass (Fig. 7E). As a result of expanded fat storage capacity after Rosi treatment, plasma TAG and NEFA were significantly reduced in WT and ASKO mice (Fig. 7F). Importantly, Rosi improved glucose tolerance and insulin sensitivity in ASKO mice (Fig. 7G and H), resulting in markedly decreased plasma fasting glucose and insulin level (Fig. 7F and I). Plasma adiponectin and leptin were increased in response to Rosi administration in both genotypes (Fig. 7J).
Histologically, the Rosi-treated livers showed decreased lipid deposition in ASKO mice (Supplementary Fig. 5A). Liver mass (Supplementary Fig. 5B) and TAG (Supplementary Fig. 5C) were also decreased in ASKO mice after Rosi administration. The expression of most transcription factors and metabolic enzymes involved in lipid synthesis (Supplementary Fig. 5D), β-oxidation (Supplementary Fig. 5E), and glucose homeostasis (Supplementary Fig. 5F) were upregulated after Rosi treatment in the liver.
SEIPIN and Adipocyte Lipid Homeostasis
A key finding of this work is the progressive loss of mature adipocytes with aging when SEIPIN function is compromised in adipocytes. We performed microarray (Supplementary Table 3) and lipidomic analyses to investigate the molecular basis for the observed fat loss in ASKO mice. From gene and pathway enrichment analyses, we found many significantly enriched pathways belonging to inflammation (Supplementary Table 4), confirming the upregulation of inflammatory markers in the ASKO adipose tissue (Fig. 5). Interestingly, sphingolipid metabolism is the first significantly enriched pathway other than those affecting the immune system. We validated four genes in the sphingolipid metabolism pathway that were significantly enriched from pathway ANOVA analysis (P < 8.22E-03) using quantitative real-time PCR (Fig. 8A). Moreover, from pathway ANOVA analysis (Supplementary Table 5), the following metabolic pathways are prominently implicated: linoleic acid, sphingolipid, glycerolipid, and fatty acid elongation. Importantly, lipidomic analyses revealed significant changes of TAG, phospholipid, sphigomyelin, and ceramide species in the ASKO mice (Fig. 8), consistent with the microarray data. Finally, ER stress was activated in the ASKO mice (Fig. 8I).
Mutations in SEIPIN are associated with the most severe form of human fat loss (i.e., CGL2/BSCL2). We and others have established an essential role of SEIPIN in adipogenesis both in vitro and in vivo (6–8,16,17). However, SEIPIN is highly expressed in mature adipocytes, where its function is completely unknown. Here, we report the generation of the ASKO mice that lack SEIPIN only in mature adipocytes. The ASKO mice exhibit adipocyte hypertrophy, progressive loss of WAT and BAT, insulin resistance, and hepatic steatosis. Interestingly, to our knowledge, the only other adipose-specific KO mice via aP2-Cre that exhibit adipocyte hypertrophy and progressive lipodystrophy are the FKO-γ mice (18). Our results therefore uncover a critical role of SEIPIN in the maintenance of adipose tissue and also reveal an intimate relationship between SEIPIN and PPAR-γ.
SEIPIN, Adipocyte Hypertrophy, and Progressive Lipodystrophy
SEIPIN is abundantly expressed in adipose tissue, and we show here that its absence from mature adipocytes results in adipocyte hypertrophy, progressive fat loss, and associated metabolic disorders. Therefore, SEIPIN is required not only for adipogenesis but also for the normal function and survival of mature adipocytes. The hypertrophy of adipocytes is likely associated with the role of SEIPIN in LD expansion. Although aggregates of small LDs were found in SEIPIN/Fld1-deficient cells (9–11,27), the most striking change was the formation of giant/supersized LDs (8,10,12). Upregulation of PA, a fusogenic lipid, is believed to contribute to the formation of supersized LDs (21,28), and indeed, PA is increased in ASKO adipose tissue (Fig. 8B and E). Increased lipogenesis and reduced lipolysis can also contribute to the formation of supersized LDs. In the ASKO adipocytes, the level of TAG is significantly increased, whereas hormone-stimulated lipolysis is decreased. A recent study showed unrestrained lipolysis in Seipin−/− mouse embryonic fibroblast cells, which accounts for the failure of adipogenesis (7). SEIPIN may differentially regulate lipolysis in preadipocytes and adipocytes. It should be noted that SEIPIN overexpression in adipocytes increased lipolysis (29), consistent with our current finding. Together, these changes (increased PA, TAG, and reduced lipolysis) may underscore the striking enlargement of LDs in adipocytes: ASKO brown adipocytes, which usually contain multiple small LDs, are now full of giant LDs (Fig. 2E). Therefore, we hypothesize that the increased lipid storage in the form of supersized LDs may form the basis of hypertrophic ASKO adipocytes.
The loss of adipose mass in ASKO mice indicates that a large number of fat cells die with aging. Indeed, adipose tissue inflammation becomes evident as ASKO mice age (Fig. 5). The severe hypertrophy of surviving adipocytes, which are known to be susceptible to apoptosis, leads to further fat loss (22). Our results from lipidomic and microarray analyses suggest that SEIPIN may directly regulate fatty acid/sphingolipid/phospholipid metabolism in adipose tissue. Compromised SEIPIN function leads to accumulation of toxic lipid species such as ceramides, which may cause ER stress and eventual cell death (Fig. 8). New adipocytes are continually formed from a preexisting stem cell or preadipocyte pool. However, cell loss eventually outpaces replenishment as ASKO mice age, causing progressive lipodystrophy (18,30).
SEIPIN and PPAR-γ
The similar tissue distribution patterns of SEIPIN and PPAR-γ, as well as the fact that SEIPIN and PPAR-γ are required for adipogenesis and for the maintenance of adipocytes, strongly suggest that they are closely connected. PPAR-γ function appears to be significantly impaired in ASKO adipose tissue. Indeed, although the expression of PPAR-γ is decreased by ∼20% in 6-month-old ASKO adipose tissue, the expression of PPAR-γ target genes is almost identical between ASKO mice and FKO-γ mice (Supplementary Fig. 2).
How might SEIPIN, an ER resident protein, regulate the activity of PPAR-γ, a ligand-activated transcription factor? Previous results and data in this work suggest a fundamental role for SEIPIN/Fld1p in lipid metabolism (3,21). Loss of SEIPIN function can change the quantity and/or distribution of certain lipids, such as PA (21). PAs could inhibit adipocyte differentiation by serving as high-affinity PPAR-γ antagonists (31). In support of this hypothesis, Rosi treatment significantly improved a number of metabolic profiles of the ASKO mice and also rescued the adipogenic defect in Seipin−/− mouse embryonic fibroblasts, as shown recently (Fig. 7) (8). It should also be noted that AGPAT2 and LPIN1 are key mammalian genes linked to severe generalized lipodystrophy, and genetic ablation of either gene also causes accumulation of PA, which could account for the failure in adipogenesis (32–34). Therefore, PA toxicity appears to be a common theme in a few models of mammalian lipodystrophy.
In summary, our findings reveal an essential role of SEIPIN in adipocyte lipid homeostasis and maintenance and, therefore, provide important insights into the physiological function of SEIPIN. Understanding the molecular function of SEIPIN may lead to novel therapeutic strategies against human obesity and insulin resistance.
Acknowledgments. The authors thank Dr. Nigel Turner and members of the Liu and Yang laboratories for helpful discussions.
Funding. This work is supported in part by Major National Basic Research Program of the People’s Republic of China (2011CB503900 and 2012CB517505) to G.L., National Natural Science Foundation of the People’s Republic of China to G.L. (30930037 and 81121061), Y.W. (30971102), and H.Y. (31228014), and a grant from the National Health and Medical Research Council of Australia (1027387) to H.Y. H.Y. is a Future Fellow of the Australian Research Council.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. L.L., Q.J., and X.W. generated the bulk of the results, conceived and designed the experiments, drafted the manuscript, and contributed equally to this work. Y.Z. and R.C.Y.L. contributed to research data. S.M.L. and G.S. performed lipidomics analysis. L.Zho. and Y.W. contributed to discussion. X.C. researched data and provided advice. M.G., L.Zha., and Y.L. contributed to research data. P.L. and G.X. provided advice, expertise, and reagents. G.L., D.Z., and H.Y. designed the experiments, provided advice and reagents, and wrote the manuscript. G.L. and H.Y. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.