Stomatin modulates adipocyte differentiation through ERK pathway and regulates lipid droplet growth and function

Controlling fatty acid uptake, lipid production and storage, and metabolism of lipid droplets (LDs), are closely related to lipid homeostasis, adipocyte hypertrophy and obesity. We report here that stomatin, a major constituent of the lipid raft, participate in adipogenesis and lipogenesis by preferentially recruiting effectors, such as perilipin for LD fusion or transporters for fatty acid uptake. Adipocyte-like cells having increased stomatin expressions exhibit higher levels of fatty acid uptake and LD growth or enlargements. Moreover, transgenic mice fed with a high-fat diet showed increased stomatin expression that facilitated progression of obesity and caused insulin resistance and hepatic impairments. Conversely, inhibitions of stomatin by gene knockdown or pharmacological treatments could block not only LD growth but also adipogenic differentiation through downregulation of PPARγ pathway. Effects of stomatin on PPARγ involved ERK signaling; however, an alternate pathway also exist. Amongst various anti-obesity measures, stomatin serves as another potential therapeutic target. Introduction Stomatin is an ancient, widely expressed, oligomeric, monotopic integral membrane protein that is associated with cholesterol-rich membrane microdomains/lipid rafts. Besides being a major component of the lipid raft, functions of stomatin are largely unclear. The gene was first identified as the causal factor of Overhydrated Hereditary Stomatocytosis (OHSt) disease, and was therefore named “stomatin” 1, . However, stomatin knockout mouse was viable and did not show stomatocytosis . Human stomatin is ubiquitously expressed; high expressions are noted in adipose tissues, bone marrow and placenta (Supplemental Fig. 1). In placenta, stomatin plays an important role in trophoblast differentiation ; and in bone, stomatin promotes osteoclastogenesis . At the cell level, stomatin is associated with the plasma membrane and cytoplasmic vesicles, such as endosomes , lipid droplets (LDs) , and specialized endosomes/granules in hematopoietic cells . We have previously reported that stomatin, with its unique molecular topology, promoted cell-cell fusion by forming molecular assembly that recruited fusogenic protein to the appositional plasma membranes . In addition to the regulations of fusion events, stomatin can also interact with various plasma membrane proteins residing within lipid raft and modulate their activities. For examples, stomatin can regulate the transport activities of Anion Exchanger 1 (AE1) , glucose transporter (GLUT1) 10 and water channel aquaporin-1 (AQP1) , or activity of acid-sensing ion channel (ASIC) family 13, . Whether such regulations are mediated by stomatin’s scaffolding effects that change the biophysical properties of the lipid rafts or by direct protein-protein interactions between stomatin and the effector molecules, are currently unknown. It is also unclear if and how stomatin is involved in signal transduction which is initiated by complex protein–protein interactions between ligands, receptors and signaling molecules that may occur in the vicinity of lipid rafts. It has been speculated that the size and composition of lipid rafts could change in response to intraor extracellular stimuli; such changes might favor specific protein–protein interactions and activate corresponding signaling cascades. In this study, we addressed the roles of stomatin in adipocyte differentiation and functions, focusing on intracellular vesicular fusion among LDs, fatty acid uptake and their associated signal transduction. Obesity, resulting from a positive energy balance between energy harvest and energy expenditure 15, , is a prevalent global healthcare problem that continues to increase in countries around the world . Obesity is characterized by increased adipose tissue mass that has been associated with a strong predisposition towards metabolic diseases such as diabetes, cardiovascular diseases, non-alcoholic fatty liver diseases and some cancers 18, 19, . The expansion of adipose depots, especially the white adipose tissues, are characterized either by the increase in adipocyte size (hypertrophy) or by the formation of new adipocytes from precursor differentiation in the process of adipogenesis (hyperplasia). Genetic and behavioral attributors both contribute to obesity-related energy imbalance 21, . In the presence of excessive energy, mature adipocytes increase in cell size and undergo cellular hypertrophy to store the surplus fat . Hypertrophic adipose morphology is positively associated with insulin resistance, diabetes and cardiovascular disease. The hypertrophic adipocytes are ultimately responsible for dysfunction of lipid homeostasis, along with other pathological consequences . Hypertrophic adipocytes are characterized by excessive growth of LDs; the resulting unilocular LD may occupy more than 90% of the cell volume. The growth of LDs is achieved by incorporation of triacylglyceride (TG) that is synthesized either locally on the surface of LDs , or obtained from the endoplasmic reticulum . Since fatty acid (FA) is the primary energy source assimilated by adipocytes during lipogenesis, measurements of FA uptake are indicative of LD growth 28, . LDs can also grow via fusion of small LDs . Previous studies have shown that several surface proteins of LDs were involved in controlling the LD size in adipocytes. For examples, Perilipin (Plin1), a member of PAT family proteins highly expressed in adipocytes, could regulate the lipolysis functions 31, . The cell death-inducing DNA fragmentation factor alpha (DFFA)-like effector (CIDE) family proteins, Cidea, and Fsp27/Cidec, also functioned as LDs size regulators 33, . Cooperative interactions of Plin1 and Fsp27 enhanced lipid transfer and LD growth . Adipogenesis is the process by which adipose tissues expand through hyperplastic growth and differentiation. The formation of new adipocytes from precursor cells enhance the capacity of energy storage. Adipogenesis is characterized by expressions of C/EBPβ and C/EBPδ, followed by C/EBPα and PPARγ. The master transcription regulator PPARγ controls a variety of genes involved in adipogenesis 36, . Previous studies have shown that activation of MAPK/ERK pathway (also known as the Ras-Raf-MEK-ERK pathway) could down-regulate transcriptional activity of PPARγ 38, 39 and inhibit adipocyte differentiation. Approaches aimed at increasing adipogenesis over adipocyte hypertrophy are now regarded as a means to treat metabolic diseases. Notably, adipocyte expansion through adipogenesis could mitigate the negative metabolic effects of obesity, although the mechanisms and regulators are not fully understood . This study addressed cell biology and signaling pathways of adipocyte-like cells affected by stomatin and provide insight into the control of adipogenesis and hypertrophic adipocytes related to obesity. Results Expressions of Stomatin increased during adipogenic differentiation A cell model for adipogenic differentiation was established . Treating murine 3T3-L1 fibroblasts with MDI cocktail for three days (day0 ~ day3), followed by insulin treatment (day3 ~ day7), rendered the cells to differentiate into adipocytes-like cells, as evidenced by lipid accumulation visualized and quantified by Oil Red O staining (Fig. 1A). Western blotting assays demonstrated that the expression of stomatin, as well as major adipogenic proteins, such as PPARγ, C/EBPα, and Perilipin, progressively increased during adipogenesis (Fig. 1B). After 7-day differentiation, mature adipocytes containing large lipid droplets (LD) were noted under DIC microscopy and stained positive with Oil Red O. Immunofluorescence staining revealed subcellular distributions of stomatin mainly on the vesicular membranes of LDs, colocalized with perilipin proteins (Fig. 1C). High expressions of Stomatin enhanced LD growth To understand the roles of stomatin in adipogenic differentiation, we over-expressed human stomatin gene (hSTOM) in murine 3T3-L1 cells. After chemically induced differentiation, we found that the levels of lipid accumulation and adipogenic proteins were very similar in cells having excessive exogenous stomatin proteins, compared to the control (Fig. 2A-B). However, when measuring the size of individual LDs, we noticed that there were more large-sized LDs in cells expressing hSTOM than the control (Fig. 2C), indicating that stomatin were involved in LD growth, or large LD formation. Such dynamics of LDs could result from either fusion of small LDs into a large one (exemplified by Supplemental Video 1 that captures the sequence of a LDLD fusion event), or from exchanges of lipid content between interacting LDs. Since small LD vesicles were not discernible under light microscopy, we employed fluorescence recovery after photobleaching (FRAP) experiments to investigate LD activities. As shown in Fig. 2E, photobleaching the fluorescent lipid content of a LD resulted in a fast recovery of fluorescence, reaching half of the original intensity in 15 min in cells over-expressing stomatin compared to a very slow recovery, reaching less than 8% of original intensity in 15 min, in the control cells. These results indicate an active role of stomatin in facilitating dynamic interactions between LDs. Highly abundant Stomatin facilitates fatty acid uptake both in vitro and in vivo The process of lipid biogenesis, or lipogenesis, starts with either de novo synthesis, or fatty acid uptake from extracellular environment using various fatty acid transporters . In the experimental assay shown in Fig. 2F, fluorescently labeled palmitic acid analogs, BODIPY-C16, when added to the culture medium, were internalized into the cells and quantified continuously using an ELISA reader. Cells having high abundance of stomatin exhibited more and faster fatty acid uptake compared to the control. Moreover, in vivo fatty acid uptake experiments were performed in STOM transgenic (STOM Tg) mice. BSA-emulsified BODIPY-C16 were injected into tail vein; after 15 min, there were much more fluorescence signals found in the adipose tissues of STOM Tg than the control mice (Fig. 3F). High expressions of Stomatin caused obesity in mice fed with high-fat diet We generated STOM transgenic (STOM Tg) mice by engineering human stomatin gene into the animal. These mice contained high amount of hSTOM proteins in subcutaneous white adipose tissues, distributed mainly on the surfaces of white adipocytes (Fig. 3A-B). STOM Tg and the control wild-type (WT) mice were fed either regular chow diet (CD), or high-fat diet (HFD) beginning at 3 weeks of age; the mice were weighed every week. We noticed that body weight gains were about the same comparing STOM Tg with WT mice fed with CD. When fed with HFD, however, mice having up-regulated STOM gene gained more weights more rapidly than their WT littermates (Fig. 3C-D). After 20 weeks, body weights of HFD-fed STOM Tg were at least 20% higher than WT. Whole-body composition measurements showed that the increase of fat was more significant than increases of lean, free fluid or total water (Fig. 3E). HFD-fed STOM Tg mice exhibited adipocyte hypertrophy, impaired metabolisms and hepatic dysfunctions After 20-week HFD feeding, mice having up-regulated STOM gene showed significant increase in subcutaneous adipose tissue (SAT) and brown adipose tissue (BAT) than WT, whereas mass of visceral adipose tissue (VAT) were about the same (Fig. 4A). Adipocytes from SAT of HFD-fed STOM Tg mice appeared hypertrophic; they were larger in size than those of WT littermates by histogram analyses (Fig. 4B). In addition, Western blotting of 3T3-L1 adipocyte-like cells showed that enzymes for lipolysis, including perilipin and hormone-sensitive lipase (HSL) and its various serinephosphorylated forms, were similar in cells having over-expressed hSTOM and control cells (Supplemental Fig. 2A), indicating that the observed adipocyte hypertrophy caused by stomatin was not due to impaired triglyceride metabolism. The observed obesity was not a result of altered energy expenditure or energy substrate selection, as evidenced by similar respiratory exchange rates comparing HFD-fed STOM Tg mice and their littermate controls (Supplemental Fig. 2B). The calculated heat production during the light-dark cycle showed that HFD-fed STOM Tg mice had only slightly higher heat production which was too little to influence obesity progression (Supplemental Fig. 2C). Biochemical analysis after 20-week HFD feeding showed that total cholesterol (TCHO) of HFD-fed STOM Tg were slightly higher than the control, but triglycerides (TG) levels were the same (Supplemental Fig. 2D). Interestingly, although fasting blood glucose were similar between the groups, plasma insulin and insulin resistance measured by HOMA-IR experiments, and glucose tolerance measured by intraperitoneal glucose tolerance test (IPTGG) were much higher and intolerable in STOM Tg than WT mice (Fig. 4C). These results indicated that HFD-fed, STOM Tg aggravated glucose homeostasis by enhancing insulin resistance. Obesity is usually correlated with ectopic fat accumulation in the liver. Indeed, HFDfed STOM Tg possessed not only larger liver mass (hepatomegaly), but also exhibited phenotypes of macroand microvesicular steatosis. The observed hepatic steatosis appeared to affect liver functions, as evidenced by elevated levels of plasma GPT and GOP in HFD-fed STOM Tg mice (Fig. 4D). Knockdown of STOM inhibited adipogenesis and LD growth Knockdown of STOM expression was done by short hairpin RNA (shRNA) method. Two shRNAs designed to target different sites of murine stomatin were separately packaged into lentiviral particles and introduced into 3T3-L1cells, resulting in shSTOM-1 and shSTOM-2 cells. Both shRNAs could effectively down-regulate stomatin expressions. Knocking-down STOM expression inhibited not only adipogenesis, evidenced by lack of lipid accumulation (Fig. 5A) and inhibitions of genes involved in adipocytic differentiation, such as PPARγ and C/EBPα (Fig. 5B). Knockdown of STOM also inhibited LD growth. As shown by Fig. 5C, histogram analyses of LD sizes revealed more small-LD and fewer large-LD following STOM knockdown. Differential expressions of shSTOM-1 and control cells were analyzed after induction of adipogenic differentiation for seven days by the microarray method. A total of 1478 annotated coding genes were identified from the transcriptomes by a stringent threshold of P value < 0.05 and FDR P value < 0.001. Among them, 379 transcripts were of significant difference between Stom-deficient and control cells (fold change ≥ 3, or ≤ -3); 185 of them were up-regulated and 194 down-regulated by inhibition of stomatin expression (Supplemental Fig. 3A). The global view of these genes was constructed by hierarchical clustering to characterize changes across six samples (Supplemental Fig. 3B). These genes were mapped onto 128 Wikipathways using Transcriptome Analysis Console (TAC). As the top-tanking enriched pathways shown (Fig. 5D), adipogenesis genes were the most profoundly inhibited gene groups, followed by PPAR signaling pathway. In the “adipogenesis genes” pathway, 12 genes involved in this pathway showed significant changes; 9 of them (75%) were downregulated and 3 (25%) were up-regulated (Fig. 5E). To further validate the results, we performed qPCR experiments for Pparg and Cebpa genes and confirmed transcriptomic findings (Fig. 5F). Somatin inhibitor OB-1 affected adipogenic differentiation and inhibited LD growth By inhibiting self-association, OB-1 is an effective inhibitor of stomatin-mediated functions . LD50 for OB-1 in 3T3-L1 cells was first determined (Fig. 6A). Drug treatments using 25 μM OB-1, while maintaining good cell viability, inhibited adipogenic differentiation in a dose-dependent manner as evidenced by decreased lipid accumulation (Fig. 6B). Treating 3T3-L1 cells during adipocytic differentiation with OB-1 also inhibited LD growth; histogram analyses of LD sizes were applied to quantify the increase of small LDs and decrease of large LDs (Fig. 6C). Stomatin modulated adipocyte differentiation through ERK pathway Inhibitions of stomatin by either shRNA (Fig. 5A) or OB-1 (Fig. 6B) could effectively block lipid accumulation by adipocytic 3T3-L1 cells. Using Western blotting assays, we analyzed their effects on other adipogenesis-related genes and signaling pathways (Fig. 7). Both stomatin-deficient shSTOM-1 and shSTOM-2 cells exhibited decreased PPARγ protein. While the Akt pathway was not affected by STOM knockdown, the ERK pathway appeared activated as evidenced by the increased level of phosphorylated ERK (arrow, Fig. 7A). OB-1 treatment also caused increase of phospho-ERK (Fig. 7D). Two early adipogenesis genes, C/EBPβ and C/EBPδ, are upstream of PPARγ regulation. After induction of adipogenic differentiation, C/EBPβ and C/EBPδ, exhibited a transient increase in the first three days, followed by a gradual decrease from day 3 to day 7 (Fig. 7B and 7C, respectively). Knockdown of stomatin had little effect on the protein levels and degrees of C/EBPβ phosphorylation (data not shown); in contrast, inhibition of stomatin was able to maintain C/EBPδ expression at much higher level than the control (Fig. 7B-C). Since activation of ERK pathway has been shown to inhibit PPARγ , we examined whether stomatin positively regulated PPARγ regulation via inhibition of pERK (red arrows, Fig. 8). To this end, we treated shSTOM-1 cells with U0126, a highly selective inhibitor for ERK, under the notion that U0126 might mitigate the stomatin-knockdown effect of pERK activation. However, we found that U0126 treatment could not reverse the lipogenesis-deficit phenotype of hSTOM-1 cells (Fig. 7E). In contrast, treating hSTOM-1 cells with troglitazone (TGZ), a PPARγ agonist, was able to partially rescue lipogenesis-deficit of hSTOM-1 cells; and interestingly, dual treatments of U0126 and TGZ were noted to further recover lipid accumulation (Fig. 7F). These results suggest the presence of a currently unknown mechanism for stomatin to positively regulate PPARγ and activate adipogenesis (black arrows, Fig. 8); this unknown pathway may work synergistically with the ERK pathway. Discussion We propose a working model for the roles of stomatin in adipogenic differentiation and lipogenesis (Fig. 8). Along the differentiation process from undifferentiated progenitor cells to immature adipocytes then to mature adipocyte, expression of stomatin progressively increase. C/EBPβ and C/EBPδ exhibits a transient increase during the early phase, while C/EBPα in the later phase. The various adipogenic genes and signaling processes seem to converge onto the PPARγ pathway, which is considered one of the master regulators of adipocyte differentiation . We report a novel finding that knockdown of stomatin was able to activate ERK signaling as evidenced by the increase of pERK. Since pERK is a well-known negative regulator of PPARγ we wonder if the promoting effect of stomatin on PPARγ and the downstream adipogenic phenotype is mediated through downregulation of pERK (red arrows, Fig. 8, see also ). Our results (Fig. 7E-F), however, did not support this notion. Unlike activation of PPARγ, activation of ERK pathway alone did not rescue the lipogenesis deficit caused by stomatin knockdown. This suggests the presence of another currently unknown pathway (black arrow, Fig. 8) that may work synergistically with ERK pathway to activate PPARγ and by doing so promote adipogenesis and lipogenesis. What signaling events are affected by stomatin which serves a major role as a component of lipid rafts? Through transcriptome and pathway analyses, we identified MAPK signaling as a potential pathway affected by stomatin (Fig. 5D-E and supplemental Fig. 3). In the case of adipocyte differentiation, EGFR is one of the upstream receptor tyrosine kinases (RTKs) for MAPK signaling. RTKs control many fundamental cell behaviors by activating a series of downstream signaling pathways including RAS-RAF-MEK-ERK pathway or AKT-PI3K-mTOR pathway 45, . We found that the RAS-RAF-MEK-ERK pathway is involved in stomatin-mediated adipogenic differentiation (Fig. 7A). Also, EGFR metabolism is highly regulated by lipid-raft mediated internalization, recycling and degradation 48, 49, . Increased expression of stomatin, for example, caused a reduction in EGFR on the plasma membrane (unpublished data), resulting in reduced ERK activity which is necessary for adipocyte differentiation . Besides participating in modulation of signaling during adipogenic differentiation, stomatin also plays a crucial role in LD growth characterized by vesicle enlargement and transformation from multilocular LDs to unilocular LDs typically seen in adipose tissues. LD is a universal cellular organelle that responds to lipid storage. It is believed to be generated from the endoplasmic reticulum (ER) 51 and can enlarge through the incorporation of lipid from the ER , the local synthesis of triacylglycerols (TAGs) 26, , and the fusion of multiple lipid droplets . Stomatin, as a LD-associated protein 7 and a major lipid raft component, could play multiple roles in growth of LD. First, stomatin if present in high abundance could promote formation of potential fusion pores that leads to LD fusion . Second, stomatin could recruit fusion facilitators to LD-LD contact site (LDCS), and by doing so regulate LD-fusion and growth. Third, stomatin could help generate passage tunnels that allow lipid exchange among contacted LDs 35, . Due either to illumination toxicity to the cell or the fact that the majority LD undergoing fusion were of submicron-sized, direct imaging of fusion between micron-sized LD was hard to achieve under light microscopy. We therefore employed FRAP experiments to address the third mechanism. Lipids in the smaller LDs of the contacted pair are transferred to the larger LDs, owing to the internal pressure difference, thus resulting in the fusion and growth of LDs. For the aforementioned second mechanism, we have previously reported that stomatin works as an "enhancer" on the plasma membrane to increase the effectiveness of molecular machinery for membranes fusion . Within a lipid raft, for example, stomatin can restrict existing fusogenic effectors to the sites of membranemembrane contact and enhance their interactions. The same mechanism may apply to intracellular inter-vesicular fusions. Locally enriched stomatin and perilipin can recruit cell death-inducing DFF45-like effector (CIDE) family proteins to LDCS . As CIDE family proteins are crucial regulators of LD-fusion , the “fencing” mechanism by stomatin can promote fusion pore expansion at LDCS, resulting in LD fusion and growth. The ability of adipocyte to uptake fatty acid of superfluous lipids from the extracellular environment is impacted by their physiological functions in energy homeostasis. The amount and rate of fatty acid uptake were noted to increase in the presence of high level stomatin both in vitro and in vivo. Although some fatty acids can cross plasma membrane by passive diffusion , most fatty acid uptake is mediated by membraneassociated transporters; many of them reside and function in the lipid rafts. When bound to long-chain fatty acid (LCFAs), FAT/CD36 may partition into lipid rafts to accelerate the translocation of LCFAs 28, . In lipid rafts, stomatin can function as an anchor or organizer for these cholesterol-rich membrane domains. Stomatin can also modulate the function of effector residing within lipid rafts. For example, stomatin can regulate several ion channels activities 60 and glucose transport GLUT1 functions . In a more general sense, stomatin can capture or trap the lateral diffusion of proteins within the lipid raft or affect the interaction between ligand and receptors, and by doing so, regulate the downstream signaling transduction. Adipose tissues play a central role in regulating energy storage to protect other tissues, such as the muscle and the liver, from the harmful effects of superfluous circulating free fatty acid. So, fatty acid uptake and LD growth are crucial in controlling lipid storage and obesity development. We demonstrated here that high levels of stomatin, combined with chronic energy surplus, may lead to a hypertrophy phenotype of adipocytes (the HFD-fed STOM Tg mice). The resulting limitation in fat storage capacity then cause local and systemic metabolism disorders . In a broader sense, stomatin is involved in regulating various aspects of lipid homeostasis, including adipocyte differentiation, lipid production, lipid storage, lipolysis and lipid secretion. The finding that stomatin regulates fatty acid uptake and LD growth may provide new opportunities for correcting whole-body energy disorders or energy surplus-induced obesity by modulating the molecular events associated with stomatin. Methods Cell Culture and induction of adipogenic differentiation 3T3-Ll murine fibroblasts, purchased from Bioresource Collection and Research Center (no.60159; BCRC, Hsinchu, Taiwan), were maintained in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10 % bovine calf serum and 1% sodium pyruvate. To induce differentiation into adipocyte-like cells, 3T3-Ll cells were first grown to confluence. Two days after reaching confluence, the medium was replaced with DMEM supplemented with 10 % fetal bovine serum (FBS) containing MDI cocktail, including 0.5 mM IBMX, 0.25 μM dexamethasone and 10 μg/ml bovine insulin. After 72 hours, the medium was replaced with DMEM supplemented with 10 % FBS and 10 μg/ ml bovine insulin. The medium was refreshed every other day. Human embryonic kidney (HEK) 293T cells were cultured in DMEM supplemented with 10 % FBS. Lentivirus production and transduction Lenti-Vector pLAS2W.puro, shRNA for Stom vectors (shSTOM-1: clone number TRCN0000112911 and shSTOM-2: clone number TRCN 0000112912) and packaging plasmids pCMV delta R8.91 and PMD.G were obtained from National RNAi Core Facility Platform (Acadamic Sinica, Taiwan). RFP, hSTOM-RFP and hSTOMflag (hSF) were amplified from expression plasmids and cloned into the lenti-Vector. HEK-293T cells were co-transfected with lentivector and packaging plasmids using NTR II Non-liposome transfection reagent II (T-Pro-Biotechnology, Taiwan). The supernatants containing lentivirus were harvested after 24and 48-hr and centrifuged at 1,250 rpm for 5 min to exclude any remaining HEK-293T packaging cells. Subsequently, cells were transduced with the lentivirus-containing supernatants for 24 hrs, followed by 3 μg/mL puromycin selection for 3 days. Quantifications of lipid accumulation Oil Red O stock solution was prepared by dissolving 0.35g Oil Red O (no. O0625, Sigma Aldrich) in 100ml pure isopropanol. Cells were fixed with 3.7% formaldehyde for 1 hr at room temperature, followed by two times wash, then dried entirely using a hairdryer. Post-fixed cells were stained with Oil Red O, diluted in distilled water (6:4) for 1 hr at room temperature, further washed four times with distilled water. The stained dyes in lipid droplets were extracted with pure isopropanol and absorbance at 490 nm (OD 490) was measured. Western blot analysis Cultured cells were washed twice with cold PBS, then lysed using RIPA lysis buffer supplemented with cOmpleteTM (EDTA-free Protease Inhibitor Cocktail) and PhosSTOPTM (both from Sigma Aldrich). The amount of total extracted proteins was quantified using Pierce 660nm protein assay reagent (Thermo Fisher Scientific TM). The protein lysates were separated by SDS-PAGE, then transferred onto PVDF membranes. The membranes were blocked with 5% bovine serum albumin (BSA) in TBS-T (Tris-buffer saline supplement with 0.1% Tween 20) at room temperature for 1 hr, and then incubated with primary antibodies, such as antiSTOM (a166623b, Abcam), anti-perilipin (9349, Cell Signaling), antiPPARγ (2435, Cell Signaling), antiC/EBPα (8178, Cell Signaling), anti-human STOM (M-14, Santa Cruz Biotechnology), anti-βTubulin (T2200, Sigma Aldrich), and anti-αActin antibodies, at 4 °C overnight. After washing three times with TBS-T, the PVDF membranes were incubated with horseradish peroxidase-conjugated secondary antibodies for 1 hr at room temperature, then treated with substrates (SuperSignal West Femto Maximum Sensitivity Substrate, Thermo Fisher Scientific), then visualized by ImageQuant (GE Healthcare Life Science). Immunofluorescence Cells seeded on coverslips were fixed with 3.7% paraformaldehyde (Sigma-Aldrich) at room temperature for 15 min, rinsed twice with PBS for 5 min, permeabilized and blocked using extraction buffer composed of 0.1 % saponin (Sigma-Aldrich) and 1 % BSA in PBS, for 1hr. Cells were then incubated with primary antibodies, including antiSTOM (a166623b, Abcam) and anti-perilipin (GT2781,GeneTex), at 4 overnight. After washing three times with wash buffer (PBS with 0.1 % saponin), cells were incubated with fluorochrome-conjugated secondary antibodies for 1 hr, and observed under a confocal microscope (LSM700, Zeiss). FRAP assay Adipocyte-like 3T3-L1 cells were incubated with 6μM of BODIPYTM-FL-C12 (D3822, InvitrogenTM) at 37°C overnight. After refreshing with complete medium, live cells were viewed under a confocal microscope (Zeiss, LSM700) using a 100x oil immersion objective. Selected regions were first bleached with 15 pulses of 100% laser power (combined 488 with 405 diode lasers), followed by time-lapse recording at 30-sec interval using normal imaging laser power. Fatty acid uptake assay For in vitro analysis, adipocyte-like 3T3-L1 cells were serum-starved in serum-free DMEM containing 1% sodium pyruvate for 1 hr at 37°C/5% CO2, changed to serumfree DMEM containing 1% sodium pyruvate and 10 μg/ml insulin for 30 min, then to 1X HBSS containing 20mM HEPES. Subsequently, 0.2 μM BODIPY-FL-C16 (Thermo) was added. The fluorescence (excitation: 485 nm and emission: 515 nm) was detected by ELISA reader (TECAN) using bottom-read mode and kinetic reading at a 30-sec interval. For in vivo analysis, 400 nM BODIPYTM-FL-C16 was emulsified in 65 mg/ml BSA solution to prepare the probe solution. An amount of 1.2~1.5 ml (6~8% of body weight) probe solution was injected into the animal via tail vein. The mice were sacrificed and subcutaneous adipose tissues were resected for quantifications of fluorescence. Fat pads were homogenized using homogenizer (Roche MagNA Lyser Benchtop Homogenizer) for 45-sec at 3,000 r.p.m., and incubated at 65 °C for 30 min. After centrifugation at 12,000rpm for 10 min, fat layer was collected and subjected to fluorescence signal quantifications. STOM transgenic mice and phenotype analyses Transgenic mice overexpressing human STOM were generated by pronuclear microinjection. STOM was cloned into vector STOM-p1033, and RNA polymerase II large subunit promoter was used to drive expression in C57BL/6 mice as described previously . The integration of the transgene was confirmed by PCR analysis of mouse tail DNA. Three to four weeks male STOM Tg and wild type littermates were fed with either chow diet or high-fat diet (D12492, 60 kcal% fat, Research Diets) for up to 20 weeks under free-feeding conditions. All of the mice were housed on a 12-hr light/dark cycle at 22°C. The blood biochemistry of STOM Tg and WT mice were examined: fasting blood glucose and plasma insulin was determined by human blood glucose meter (Accu-chek Performa, Roche) and Mouse Ultrasensitive Insulin ELISA (80-INSMSU-E01, ALPCO), respectively. GOT and GPT were measured using Automated Clinical Chemistry Analyzer (FUJI DRI-CHEM 4000i). Glucose tolerance tests Mice fasted overnight were injected intraperitoneally with glucose solution (1 g/kg body weight). Blood glucose levels were measured at 0, 15, 30, 60, and 120 min after injection using a glucometer. Histological analyses The liver and adipose tissues were formalin-fixed and paraffin-embedded. Sections of WAT and the liver were stained with hematoxylin and eosin (H&E stain). All tissue images were obtained using the Aperio CS2 Digital Pathology Scanner (Leica) and analyzed via Imagescope software. The size of adipose cells was analyzed using the Fiji Adiposoft software to obtain the area of each and individual adipocyte. Transcriptome expression analyses Total RNA were extracted from cells using TRIzol® reagent and subjected to global transcriptome analysis using Mouse Genome Arrays (MTA-1_0, Affymetrix). The Transcriptome Analysis Console (TAC4.0) was used to process and analyze CEL files. We initially filtered for 11,264 probe sets annotated as locus type “coding”. Subsequently, 1,478 annotated coding genes were identified via exclusively a stringent threshold of p-value< 0.05 and FDR p-value< 0.001. Additionally, we further filtered for 379 genes via exclusively a stringent fold change threshold ≥ 3, or ≤ -3. The results were confirmed using qRT-PCR. The cDNAs were generated using SuperScript III First-Strand Synthesis System following the manufacturer’s protocol. To determine the expression of specific genes, quantitative real-time PCR (qRT-PCR) was performed using diluted cDNA in a total volume of 5ul with SYBR Green (Qiagen). Gene expression was normalized to the internal reference gene Nono, followed by calculation using ∆ ∆CT method.


Introduction
Stomatin is an ancient, widely expressed, oligomeric, monotopic integral membrane protein that is associated with cholesterol-rich membrane microdomains/lipid rafts.
Besides being a major component of the lipid raft, functions of stomatin are largely unclear. The gene was first identified as the causal factor of Overhydrated Hereditary Stomatocytosis (OHSt) disease, and was therefore named "stomatin" 1, 2 . However, stomatin knockout mouse was viable and did not show stomatocytosis 3 . Human stomatin is ubiquitously expressed; high expressions are noted in adipose tissues, bone marrow and placenta (Supplemental Fig. 1). In placenta, stomatin plays an important role in trophoblast differentiation 4 ; and in bone, stomatin promotes osteoclastogenesis 5 . At the cell level, stomatin is associated with the plasma membrane and cytoplasmic vesicles, such as endosomes 6 , lipid droplets (LDs) 7 , and specialized endosomes/granules in hematopoietic cells 8 . We have previously reported that stomatin, with its unique molecular topology, promoted cell-cell fusion by forming molecular assembly that recruited fusogenic protein to the appositional plasma membranes 5 . In addition to the regulations of fusion events, stomatin can also interact with various plasma membrane proteins residing within lipid raft and modulate their activities. For examples, stomatin can regulate the transport activities of Anion Exchanger 1 (AE1) 9 , glucose transporter (GLUT1) 10 and water channel aquaporin-1 (AQP1) 11 , or activity of acid-sensing ion channel (ASIC) family 12,13,14 . Whether such regulations are mediated by stomatin's scaffolding effects that change the biophysical properties of the lipid rafts or by direct protein-protein interactions between stomatin and the effector molecules, are currently unknown. It is also unclear if and how stomatin is involved in signal transduction which is initiated by complex protein-protein interactions between ligands, receptors and signaling molecules that may occur in the vicinity of lipid rafts. It has been speculated that the size and composition of lipid rafts could change in response to intra-or extracellular stimuli; such changes might favor specific protein-protein interactions and activate corresponding signaling cascades.
In this study, we addressed the roles of stomatin in adipocyte differentiation and functions, focusing on intracellular vesicular fusion among LDs, fatty acid uptake and their associated signal transduction.
Obesity, resulting from a positive energy balance between energy harvest and energy expenditure 15,16 , is a prevalent global healthcare problem that continues to increase in countries around the world 17 . Obesity is characterized by increased adipose tissue mass that has been associated with a strong predisposition towards metabolic diseases such as diabetes, cardiovascular diseases, non-alcoholic fatty liver diseases and some cancers 18,19,20 . The expansion of adipose depots, especially the white adipose tissues, are characterized either by the increase in adipocyte size (hypertrophy) or by the formation of new adipocytes from precursor differentiation in the process of adipogenesis (hyperplasia). Genetic and behavioral attributors both contribute to obesity-related energy imbalance 21,22 . In the presence of excessive energy, mature adipocytes increase in cell size and undergo cellular hypertrophy to store the surplus fat 23 .
Hypertrophic adipose morphology is positively associated with insulin resistance, diabetes and cardiovascular disease. The hypertrophic adipocytes are ultimately responsible for dysfunction of lipid homeostasis, along with other pathological consequences 24 . Hypertrophic adipocytes are characterized by excessive growth of LDs; the resulting unilocular LD may occupy more than 90% of the cell volume 25 . The growth of LDs is achieved by incorporation of triacylglyceride (TG) that is synthesized either locally on the surface of LDs 26 , or obtained from the endoplasmic reticulum 27 .
Since fatty acid (FA) is the primary energy source assimilated by adipocytes during lipogenesis, measurements of FA uptake are indicative of LD growth 28,29  Adipogenesis is the process by which adipose tissues expand through hyperplastic growth and differentiation. The formation of new adipocytes from precursor cells enhance the capacity of energy storage. Adipogenesis is characterized by expressions of C/EBPβ and C/EBPδ, followed by C/EBPα and PPARγ. The master transcription regulator PPARγ controls a variety of genes involved in adipogenesis 36,37 . Previous studies have shown that activation of MAPK/ERK pathway (also known as the Ras-Raf-MEK-ERK pathway) could down-regulate transcriptional activity of PPARγ 38, 39 and inhibit adipocyte differentiation.
Approaches aimed at increasing adipogenesis over adipocyte hypertrophy are now regarded as a means to treat metabolic diseases. Notably, adipocyte expansion through adipogenesis could mitigate the negative metabolic effects of obesity, although the mechanisms and regulators are not fully understood 40 . This study addressed cell biology and signaling pathways of adipocyte-like cells affected by stomatin and provide insight into the control of adipogenesis and hypertrophic adipocytes related to obesity.

Expressions of Stomatin increased during adipogenic differentiation
A cell model for adipogenic differentiation was established 41 . Treating murine 3T3-L1 fibroblasts with MDI cocktail for three days (day0 ~ day3), followed by insulin treatment (day3 ~ day7), rendered the cells to differentiate into adipocytes-like cells, as evidenced by lipid accumulation visualized and quantified by Oil Red O staining (Fig.   1A). Western blotting assays demonstrated that the expression of stomatin, as well as major adipogenic proteins, such as PPARγ, C/EBPα, and Perilipin, progressively increased during adipogenesis (Fig. 1B). After 7-day differentiation, mature adipocytes containing large lipid droplets (LD) were noted under DIC microscopy and stained positive with Oil Red O. Immunofluorescence staining revealed subcellular distributions of stomatin mainly on the vesicular membranes of LDs, colocalized with perilipin proteins (Fig. 1C).

High expressions of Stomatin enhanced LD growth
To understand the roles of stomatin in adipogenic differentiation, we over-expressed human stomatin gene (hSTOM) in murine 3T3-L1 cells. After chemically induced differentiation, we found that the levels of lipid accumulation and adipogenic proteins were very similar in cells having excessive exogenous stomatin proteins, compared to the control ( Fig. 2A-B). However, when measuring the size of individual LDs, we noticed that there were more large-sized LDs in cells expressing hSTOM than the control (Fig. 2C), indicating that stomatin were involved in LD growth, or large LD formation. Such dynamics of LDs could result from either fusion of small LDs into a large one (exemplified by Supplemental Video 1 that captures the sequence of a LD-LD fusion event), or from exchanges of lipid content between interacting LDs.
Since small LD vesicles were not discernible under light microscopy, we employed fluorescence recovery after photobleaching (FRAP) experiments to investigate LD activities. As shown in Fig. 2E, photobleaching the fluorescent lipid content of a LD resulted in a fast recovery of fluorescence, reaching half of the original intensity in 15 min in cells over-expressing stomatin compared to a very slow recovery, reaching less than 8% of original intensity in 15 min, in the control cells. These results indicate an active role of stomatin in facilitating dynamic interactions between LDs.

Highly abundant Stomatin facilitates fatty acid uptake both in vitro and in vivo
The process of lipid biogenesis, or lipogenesis, starts with either de novo synthesis, or fatty acid uptake from extracellular environment using various fatty acid transporters 42 . In the experimental assay shown in Fig. 2F, fluorescently labeled palmitic acid analogs, BODIPY-C16, when added to the culture medium, were internalized into the cells and quantified continuously using an ELISA reader. Cells having high abundance of stomatin exhibited more and faster fatty acid uptake compared to the control.
Moreover, in vivo fatty acid uptake experiments were performed in STOM transgenic (STOM Tg) mice. BSA-emulsified BODIPY-C16 were injected into tail vein; after 15 min, there were much more fluorescence signals found in the adipose tissues of STOM Tg than the control mice ( Fig. 3F).

High expressions of Stomatin caused obesity in mice fed with high-fat diet
We generated STOM transgenic (STOM Tg) mice by engineering human stomatin gene into the animal. These mice contained high amount of hSTOM proteins in subcutaneous white adipose tissues, distributed mainly on the surfaces of white adipocytes ( Fig. 3A-B). STOM Tg and the control wild-type (WT) mice were fed either regular chow diet (CD), or high-fat diet (HFD) beginning at 3 weeks of age; the mice were weighed every week. We noticed that body weight gains were about the same comparing STOM Tg with WT mice fed with CD. When fed with HFD, however, mice having up-regulated STOM gene gained more weights more rapidly than their WT littermates ( Fig. 3C-D). After 20 weeks, body weights of HFD-fed STOM Tg were at least 20% higher than WT. Whole-body composition measurements showed that the increase of fat was more significant than increases of lean, free fluid or total water (Fig.   3E).

HFD-fed STOM Tg mice exhibited adipocyte hypertrophy, impaired metabolisms and hepatic dysfunctions
After 20-week HFD feeding, mice having up-regulated STOM gene showed significant increase in subcutaneous adipose tissue (SAT) and brown adipose tissue (BAT) than WT, whereas mass of visceral adipose tissue (VAT) were about the same (Fig. 4A).
Adipocytes from SAT of HFD-fed STOM Tg mice appeared hypertrophic; they were larger in size than those of WT littermates by histogram analyses (Fig. 4B). In addition, Western blotting of 3T3-L1 adipocyte-like cells showed that enzymes for lipolysis, including perilipin and hormone-sensitive lipase (HSL) and its various serinephosphorylated forms, were similar in cells having over-expressed hSTOM and control cells (Supplemental Fig. 2A), indicating that the observed adipocyte hypertrophy caused by stomatin was not due to impaired triglyceride metabolism.
The observed obesity was not a result of altered energy expenditure or energy substrate selection, as evidenced by similar respiratory exchange rates comparing HFD-fed STOM Tg mice and their littermate controls (Supplemental Fig. 2B). The calculated heat production during the light-dark cycle showed that HFD-fed STOM Tg mice had only slightly higher heat production which was too little to influence obesity progression (Supplemental Fig. 2C).
Biochemical analysis after 20-week HFD feeding showed that total cholesterol (TCHO) of HFD-fed STOM Tg were slightly higher than the control, but triglycerides (TG) levels were the same (Supplemental Fig. 2D). Interestingly, although fasting blood glucose were similar between the groups, plasma insulin and insulin resistance measured by HOMA-IR experiments, and glucose tolerance measured by intraperitoneal glucose tolerance test (IPTGG) were much higher and intolerable in STOM Tg than WT mice ( Fig. 4C). These results indicated that HFD-fed, STOM Tg aggravated glucose homeostasis by enhancing insulin resistance.
Obesity is usually correlated with ectopic fat accumulation in the liver. Indeed, HFDfed STOM Tg possessed not only larger liver mass (hepatomegaly), but also exhibited phenotypes of macro-and microvesicular steatosis. The observed hepatic steatosis appeared to affect liver functions, as evidenced by elevated levels of plasma GPT and GOP in HFD-fed STOM Tg mice (Fig. 4D).

Knockdown of STOM inhibited adipogenesis and LD growth
Knockdown of STOM expression was done by short hairpin RNA (shRNA) method.
Two shRNAs designed to target different sites of murine stomatin were separately packaged into lentiviral particles and introduced into 3T3-L1cells, resulting in shSTOM-1 and shSTOM-2 cells. Both shRNAs could effectively down-regulate stomatin expressions. Knocking-down STOM expression inhibited not only adipogenesis, evidenced by lack of lipid accumulation (Fig. 5A) and inhibitions of genes involved in adipocytic differentiation, such as PPARγ and C/EBPα (Fig. 5B).
Knockdown of STOM also inhibited LD growth. As shown by Fig. 5C, histogram analyses of LD sizes revealed more small-LD and fewer large-LD following STOM knockdown.
Differential expressions of shSTOM-1 and control cells were analyzed after induction of adipogenic differentiation for seven days by the microarray method. A total of 1478 annotated coding genes were identified from the transcriptomes by a stringent threshold of P value < 0.05 and FDR P value < 0.001. Among them, 379 transcripts were of significant difference between Stom-deficient and control cells (fold change ≥ 3, or ≤ -3); 185 of them were up-regulated and 194 down-regulated by inhibition of stomatin expression (Supplemental Fig. 3A). The global view of these genes was constructed by hierarchical clustering to characterize changes across six samples (Supplemental Fig. 3B). These genes were mapped onto 128 Wikipathways using Transcriptome Analysis Console (TAC). As the top-tanking enriched pathways shown ( Fig. 5D), adipogenesis genes were the most profoundly inhibited gene groups, followed by PPAR signaling pathway. In the "adipogenesis genes" pathway, 12 genes involved in this pathway showed significant changes; 9 of them (75%) were downregulated and 3 (25%) were up-regulated (Fig. 5E). To further validate the results, we performed qPCR experiments for Pparg and Cebpa genes and confirmed transcriptomic findings (Fig. 5F).

Somatin inhibitor OB-1 affected adipogenic differentiation and inhibited LD growth
By inhibiting self-association, OB-1 is an effective inhibitor of stomatin-mediated functions 43 . LD 50 for OB-1 in 3T3-L1 cells was first determined (Fig. 6A). Drug treatments using 25 μM OB-1, while maintaining good cell viability, inhibited adipogenic differentiation in a dose-dependent manner as evidenced by decreased lipid accumulation (Fig. 6B). Treating 3T3-L1 cells during adipocytic differentiation with OB-1 also inhibited LD growth; histogram analyses of LD sizes were applied to quantify the increase of small LDs and decrease of large LDs (Fig. 6C).

Stomatin modulated adipocyte differentiation through ERK pathway
Inhibitions of stomatin by either shRNA ( Since activation of ERK pathway has been shown to inhibit PPARγ 44 , we examined whether stomatin positively regulated PPARγ regulation via inhibition of pERK (red arrows, Fig. 8). To this end, we treated shSTOM-1 cells with U0126, a highly selective inhibitor for ERK, under the notion that U0126 might mitigate the stomatin-knockdown effect of pERK activation. However, we found that U0126 treatment could not reverse the lipogenesis-deficit phenotype of hSTOM-1 cells (Fig. 7E). In contrast, treating hSTOM-1 cells with troglitazone (TGZ), a PPARγ agonist, was able to partially rescue lipogenesis-deficit of hSTOM-1 cells; and interestingly, dual treatments of U0126 and TGZ were noted to further recover lipid accumulation (Fig. 7F). These results suggest the presence of a currently unknown mechanism for stomatin to positively regulate PPARγ and activate adipogenesis (black arrows, Fig. 8); this unknown pathway may work synergistically with the ERK pathway.

Discussion
We propose a working model for the roles of stomatin in adipogenic differentiation and lipogenesis (Fig. 8). Along the differentiation process from undifferentiated progenitor cells to immature adipocytes then to mature adipocyte, expression of stomatin progressively increase. C/EBPβ and C/EBPδ exhibits a transient increase during the early phase, while C/EBPα in the later phase. The various adipogenic genes and signaling processes seem to converge onto the PPARγ pathway, which is considered one of the master regulators of adipocyte differentiation 37 . We report a novel finding that knockdown of stomatin was able to activate ERK signaling as evidenced by the increase of pERK. Since pERK is a well-known negative regulator of PPARγ we wonder if the promoting effect of stomatin on PPARγ and the downstream adipogenic phenotype is mediated through downregulation of pERK (red arrows, Fig. 8, see also 44 ). Our results (Fig. 7E-F), however, did not support this notion. Unlike activation of PPARγ, activation of ERK pathway alone did not rescue the lipogenesis deficit caused by stomatin knockdown. This suggests the presence of another currently unknown pathway (black arrow, Fig. 8) that may work synergistically with ERK pathway to activate PPARγ and by doing so promote adipogenesis and lipogenesis.
What signaling events are affected by stomatin which serves a major role as a component of lipid rafts? Through transcriptome and pathway analyses, we identified MAPK signaling as a potential pathway affected by stomatin (Fig. 5D-E and supplemental Fig. 3). In the case of adipocyte differentiation, EGFR is one of the upstream receptor tyrosine kinases (RTKs) for MAPK signaling. RTKs control many fundamental cell behaviors by activating a series of downstream signaling pathways including RAS-RAF-MEK-ERK pathway or AKT-PI3K-mTOR pathway 45,46 . We found that the RAS-RAF-MEK-ERK pathway is involved in stomatin-mediated adipogenic differentiation (Fig. 7A). Also, EGFR metabolism is highly regulated by lipid-raft mediated internalization, recycling and degradation 47,48,49,50 . Increased expression of stomatin, for example, caused a reduction in EGFR on the plasma membrane (unpublished data), resulting in reduced ERK activity which is necessary for adipocyte differentiation 45 .
Besides participating in modulation of signaling during adipogenic differentiation, stomatin also plays a crucial role in LD growth characterized by vesicle enlargement and transformation from multilocular LDs to unilocular LDs typically seen in adipose tissues. LD is a universal cellular organelle that responds to lipid storage. It is believed to be generated from the endoplasmic reticulum (ER) 51 and can enlarge through the incorporation of lipid from the ER 52 , the local synthesis of triacylglycerols (TAGs) 26,53 , and the fusion of multiple lipid droplets 54 . Stomatin, as a LD-associated protein 7 and a major lipid raft component, could play multiple roles in growth of LD. First, stomatin if present in high abundance could promote formation of potential fusion pores that leads to LD fusion 4, 5 . Second, stomatin could recruit fusion facilitators to LD-LD contact site (LDCS), and by doing so regulate LD-fusion and growth. Third, stomatin could help generate passage tunnels that allow lipid exchange among contacted LDs 35, 55 . Due either to illumination toxicity to the cell or the fact that the majority LD undergoing fusion were of submicron-sized, direct imaging of fusion between micron-sized LD was hard to achieve under light microscopy. We therefore employed FRAP experiments to address the third mechanism. Lipids in the smaller LDs of the contacted pair are transferred to the larger LDs, owing to the internal pressure difference, thus resulting in the fusion and growth of LDs.
For the aforementioned second mechanism, we have previously reported that stomatin works as an "enhancer" on the plasma membrane to increase the effectiveness of molecular machinery for membranes fusion 4, 5 . Within a lipid raft, for example, stomatin can restrict existing fusogenic effectors to the sites of membranemembrane contact and enhance their interactions. The same mechanism may apply to intracellular inter-vesicular fusions. Locally enriched stomatin and perilipin can recruit cell death-inducing DFF45-like effector (CIDE) family proteins to LDCS 35 . As CIDE family proteins are crucial regulators of LD-fusion 56 , the "fencing" mechanism by stomatin can promote fusion pore expansion at LDCS, resulting in LD fusion and growth.
The ability of adipocyte to uptake fatty acid of superfluous lipids from the extracellular environment is impacted by their physiological functions in energy homeostasis. The amount and rate of fatty acid uptake were noted to increase in the presence of high level stomatin both in vitro and in vivo. Although some fatty acids can cross plasma membrane by passive diffusion 57 , most fatty acid uptake is mediated by membraneassociated transporters; many of them reside and function in the lipid rafts. When bound to long-chain fatty acid (LCFAs), FAT/CD36 may partition into lipid rafts to accelerate the translocation of LCFAs 28, 58 . In lipid rafts, stomatin can function as an anchor or organizer for these cholesterol-rich membrane domains. Stomatin can also modulate the function of effector residing within lipid rafts. For example, stomatin can regulate several ion channels activities 59, 60 and glucose transport GLUT1 functions 10 .
In a more general sense, stomatin can capture or trap the lateral diffusion of proteins within the lipid raft or affect the interaction between ligand and receptors, and by doing so, regulate the downstream signaling transduction.
Adipose tissues play a central role in regulating energy storage to protect other tissues, such as the muscle and the liver, from the harmful effects of superfluous circulating free fatty acid. So, fatty acid uptake and LD growth are crucial in controlling lipid storage and obesity development. We demonstrated here that high levels of stomatin, combined with chronic energy surplus, may lead to a hypertrophy phenotype of adipocytes (the HFD-fed STOM Tg mice). The resulting limitation in fat storage capacity then cause local and systemic metabolism disorders 61 . In a broader sense, stomatin is involved in regulating various aspects of lipid homeostasis, including adipocyte differentiation, lipid production, lipid storage, lipolysis and lipid secretion.
The finding that stomatin regulates fatty acid uptake and LD growth may provide new opportunities for correcting whole-body energy disorders or energy surplus-induced obesity by modulating the molecular events associated with stomatin.
HEK-293T cells were co-transfected with lentivector and packaging plasmids using NTR II Non-liposome transfection reagent II (T-Pro-Biotechnology, Taiwan). The supernatants containing lentivirus were harvested after 24-and 48-hr and centrifuged at 1,250 rpm for 5 min to exclude any remaining HEK-293T packaging cells.
Subsequently, cells were transduced with the lentivirus-containing supernatants for 24 hrs, followed by 3 µg/mL puromycin selection for 3 days.

Quantifications of lipid accumulation
Oil Red O stock solution was prepared by dissolving 0.35g Oil Red O (no. O0625, Sigma Aldrich) in 100ml pure isopropanol. Cells were fixed with 3.7% formaldehyde for 1 hr at room temperature, followed by two times wash, then dried entirely using a hairdryer. Post-fixed cells were stained with Oil Red O, diluted in distilled water (6:4) for 1 hr at room temperature, further washed four times with distilled water. The stained dyes in lipid droplets were extracted with pure isopropanol and absorbance at 490 nm (OD 490) was measured.

Western blot analysis
Cultured cells were washed twice with cold PBS, then lysed using RIPA lysis buffer supplemented with cOmplete™ (EDTA-free Protease Inhibitor Cocktail) and PhosSTOP™ (both from Sigma Aldrich). The amount of total extracted proteins was quantified using Pierce 660nm protein assay reagent (Thermo Fisher Scientific ™).
After washing three times with wash buffer (PBS with 0.1 % saponin), cells were incubated with fluorochrome-conjugated secondary antibodies for 1 hr, and observed under a confocal microscope (LSM700, Zeiss).

FRAP assay
Adipocyte-like 3T3-L1 cells were incubated with 6µM of BODIPY™-FL-C 12 (D3822, Invitrogen™) at 37°C overnight. After refreshing with complete medium, live cells were viewed under a confocal microscope (Zeiss, LSM700) using a 100x oil immersion objective. Selected regions were first bleached with 15 pulses of 100% laser power (combined 488 with 405 diode lasers), followed by time-lapse recording at 30-sec interval using normal imaging laser power.

Fatty acid uptake assay
For in vitro analysis, adipocyte-like 3T3-L1 cells were serum-starved in serum-free DMEM containing 1% sodium pyruvate for 1 hr at 37°C/5% CO2, changed to serumfree DMEM containing 1% sodium pyruvate and 10 µg/ml insulin for 30 min, then to After centrifugation at 12,000rpm for 10 min, fat layer was collected and subjected to fluorescence signal quantifications.

STOM transgenic mice and phenotype analyses
Transgenic mice overexpressing human STOM were generated by pronuclear microinjection. STOM was cloned into vector STOM-p1033, and RNA polymerase II large subunit promoter was used to drive expression in C57BL/6 mice as described

Glucose tolerance tests
Mice fasted overnight were injected intraperitoneally with glucose solution (1 g/kg body weight). Blood glucose levels were measured at 0, 15, 30, 60, and 120 min after injection using a glucometer.

Histological analyses
The liver and adipose tissues were formalin-fixed and paraffin-embedded. Sections of WAT and the liver were stained with hematoxylin and eosin (H&E stain). All tissue images were obtained using the Aperio CS2 Digital Pathology Scanner (Leica) and analyzed via Imagescope software. The size of adipose cells was analyzed using the Fiji Adiposoft software to obtain the area of each and individual adipocyte.

Transcriptome expression analyses
Total RNA were extracted from cells using TRIzol® reagent and subjected to global transcriptome analysis using Mouse Genome Arrays (MTA-1_0, Affymetrix). The Transcriptome Analysis Console (TAC4.0) was used to process and analyze CEL files.
We initially filtered for 11,264 probe sets annotated as locus type "coding".
Subsequently, 1,478 annotated coding genes were identified via exclusively a stringent threshold of p-value< 0.05 and FDR p-value< 0.001. Additionally, we further filtered for 379 genes via exclusively a stringent fold change threshold ≥ 3, or ≤ -3.
The results were confirmed using qRT-PCR. The cDNAs were generated using SuperScript III First-Strand Synthesis System following the manufacturer's protocol.
To determine the expression of specific genes, quantitative real-time PCR (qRT-PCR) was performed using diluted cDNA in a total volume of 5ul with SYBR Green (Qiagen).
Gene expression was normalized to the internal reference gene Nono, followed by calculation using ∆ ∆CT method.

Video 1
The and perilipin (red) on cells after 7-day induction were observed by DIC and immunofluorescence microscopy (IF). Cell nuclei were revealed by DNA stain (blue). Perilipin and mSTOM were found colocalized on vesicular membranes of lipid droplets (LDs) (inset). Bar = 10 μm.   Body weight increments after 20-week feeding were measured. Each dot represents one mouse. While no difference was noticed in animals fed with CD, body weight gains were significantly higher in HFD-fed STOM Tg, compared with HFD-fed WT mice. Mean ± s.d. is shown. n.s = non-significant. **P<0.01 by unpaired t-test. E) The mass of whole body, fat, lean, free fluid, and total water were calculated for HFD-fed STOM Tg and HFD-fed WT mice. F) Fatty acid uptake was measured in vivo. Fluorescently-labeled BODIPY-FL-C16 fatty acid, or BODIPY-BSA, were injected into tail vein of STOM Tg mice or WT mice. After 15 min, aqueous portions of white adipose tissue (WAT) from the animal were extracted and fluorescence signals that represented lipid uptake were quantified. Data shown are fold changes of fluorescence intensity using BODIPY-BSA injected to WT mice as the reference. Each dot represents one mouse. Mean ± s.d. for three independent experiments. *P<0.05, **P<0.01, and ****P<0.0001 by two-way ANOVA. BODIPY-FL-C16: 4,4-Difluoro-5,7-Dimethyl-4-Bora-3a,4a-Diaza-s-Indacene-3-Hexadecanoic Acid; BSA: bovine serum albumin On day 7 of adipogenic differentiation, adipocyte-like shSTOM-1 cells were stained with AdipoRed and observed under phase-contrast (PC) and fluorescence (Flu) microscopy. The size of lipid droplets (LDs) were measured, and their number per 100 adipocytes were calculated. Histogram analyses showed more small LDs (< 40 μm 2 ) and fewer large (> 80 μm 2 ) LDs in shSTOM-1 cells than the control. Mean ± s.d. for three independent experiments. ****P<0.0001 by two-way ANOVA. Bar = 30 μm. D) Transcriptome analyses of adipocyte-like shSTOM-1 and control cells after induction for 7 days. From the data of microarray assays, scatter plots revealed enriched Wiki pathways in shSTOM-1 cells comparing to the control. For a given pathway, ratio of gene number being up-or down-regulated in that pathway were determined, and plotted as function of fold change. Each dot represents one gene. The color of the dots represents the range of P-values related to the indicated pathway. E) Heat map of adipogenesis gene obtained from the enrichment-based cluster analysis of the Wiki pathway. F) Real-time qPCR analyses to validate the changed genes revealed by microarray assays. Data shown are fold changes relative to Nono mRNA level. Mean ± s.d. for three independent experiments. ***P<0.001 by unpaired t-test. shRNA were transducted into 3T3-L1 cells to knockdown stomatin and generating shSTOM-1 and shSTOM-2 cells. After induction of differentiation, Western blotting analyses were done. Although Akt pathway remained unchanged, the ERK pathway was activated as evidenced by increased pERK (arrow). B, C) Expressions of early adipogenic genes, Cebpb and Cebpd, were examined at mRNA level by qPCR. B) Expressions of Cebpb increased in the first three days of adipogenic differentiation, and subsequently decreased. Knockdown of stomatin (exemplified by shSTOM-1 cells) did not significantly influence this pattern. C) Expressions of Cebpd gene also exhibited a transient early increase then declined; however, knockdown of stomatin appeared to significantly increase and maintain Cebpd expression at a relatively higher level than the control. The house-keeping Nono gene was used as the reference for qPCR exoeriments. *P<0.05 by unpaired t-test. D) Treating adipocyte-like cells with stomatin inhibitor OB-1 also resulted in activation of ERK pathway, evidenced by increased pERK (arrow  Figure 8: Stomatin's roles in modulating adipogenic differentiation and lipogenesis. Undifferentiated progenitor cells can be induced to undergo adipogenic differentiation to become immature adipocytes, and further develop into mature adipocytes, characterized by fatty acid uptake, production of lipid, and fusions and enlargements of lipid droplets (LDs). Dynamic expressions of adipogenic genes along this process are depicted. Stomatin progressively increases during adipocyte differentiation and maturation, and participates not only in fatty acid uptake and LD fusions, but also in modulation of adipogenic gene expressions. By inhibiting pERK, stomatin activates PPARγ: resulting in adipocyte maturation and lipogenesis. Stomatin also plays a role in regulating early phase adipogenic genes, such as C/EBPβ and C/EBPδ through currently unknown mechanisms. C/EBP: CCAAT/enhancer binding protein; PPARγ: peroxisome proliferator-activated receptor-γ.

and control cells after induction of adipogenic differentiation. A)
Scatter plots of the normalized signal intensities of 1,478 genes that exhibited differential expressions between shSTOM-1 and control cells.
Log2 intensities for each spot on the microarray are plotted on the x and y axes with signals from root tips stressed for control and shSTOM-1. The diagonal lines represent fold change cutoffs of ±3. The red spots represent up-regulated genes and the green spots indicate down-regulated genes. B) Hierarchical cluster analyses of the 1,478 genes shown in A). Data were collected from three independent cell clones. Relative expression levels of genes are illustrated by the color gradient (Z-score).