Caveolin-1 Expression and Cavin Stability Regulate Caveolae Dynamics in Adipocyte Lipid Store Fluctuation

Caveolae are small flask-shaped invaginations of the plasma membrane (1) that are found with remarkable abundance in endothelial cells, myotubes, and adipocytes. They are considered a subset of the so-called lipid raft domains and segregate a number of membrane-related processes (2). An accepted paradigm is that caveolae formation is primarily driven by the assembly of a cytoplasmic coat consisting of oligomeric caveolins (3), a protein family with 3 highly-related members (caveolin-1 through -3). Invalidation of individual caveolin genes led to the generation of mice models lacking caveolae in all cell types or in a tissue-restricted manner (4). Caveolin-1–null mice, which also lack caveolin-2, suffer from severe vascular dysfunction and pulmonary defects (5,6) and develop lipodystrophy (7), a metabolic phenotype that cannot be reversed by endothelial caveolin reexpression, pointing to an important function of adipocyte caveolae (8). Closely related lipodystrophic diabetes is also present in patients with nonsense mutations for caveolin-1 (9).

Despite early evidence for defective mechanotransduction in blood vessels of caveolin-deficient mice (10), the physiological function of caveolae remained debated until recent discovery of a unifying caveolar function as mechanosensors, revealed by their property to respond to membrane tension by flattening into the plasma membrane (11). Another recent breakthrough came from the identification of novel caveolae adaptors named cavins, such as polymerase I and transcript release factor (PTRF)/cavin-1 required for caveolae formation (12,13). Specifically enriched in caveolae preparations from human adipocytes (14), PTRF/cavin-1 gene invalidation in mice led to absence of caveolae and lipodystrophic phenotype comparable with the one observed for caveolin-1 knockout mice (7,15). Similarly, PTRF/cavin-1 mutations were identified in human patients suffering lipodystrophic syndromes coupled with muscular dystrophy (16–18), states for which caveolin mutations have also been reported (9,19). Three other cavin homologs were identified (14) and have all been linked to caveolae dynamics (20–23). Despite major importance in regulating caveolae (24), their precise role in mechanosensing still remains unknown.

As caveolae are being recognized as specialized structures responding to plasma membrane tension, it makes sense that a high density of caveolae is found in cells that are physiologically submitted to mechanical stress. These include vascular endothelium in which shear stress is applied by blood circulation, as well as skeletal muscle through contraction-induced stretching. Despite remarkable enrichment in caveolae, estimated to cover ~30–50% of their cell surface (25), adipocytes are not generally considered particularly exposed to external mechanical forces, as they mostly participate in lipid metabolism and energy storage. Indeed, adipocyte morphology is dominated by a central unilocular lipid droplet that fills almost the entire fat cell volume, leaving other organelles including the nucleus at cell periphery. Close apposition of adipocyte lipid droplet to plasma membrane is obvious by electron microscopy, with cytoplasmic thickness as low as 300 nm. Another consequence of the storage compartment being the most prominent intracellular organelle with unusually large dimensions (up to 100 µm diameter in obese patients), the lipid droplet size also resolves overall fat cell volume. As a mirror of highly variable energy status, the adipocyte lipid droplet can be subjected to drastic fluctuations, as well as fat cell size. For example, adipose cell diameter, an easily accessible parameter on adipose tissue histological sections, can vary over a twofold range, which translates into an eightfold change in cell volume. In this context, we reasoned that the lipid droplet organelle might be a driving force for mechanical constrains exerted on the adipocyte plasma membrane, to which caveolae might respond. To examine this possibility, we have designed in vitro cell conditions and in vivo experimental settings in mice and humans, in which caveolae dynamics and/or lipid stores could be manipulated. Our data establish a mutual link between caveolae density and adipocyte lipid stores and reveal that lipid droplet to membrane signaling mechanisms can sense lipid store variations via caveolae. These results therefore suggest that lipid droplets might be viewed as mechano-active organelles in adipocytes.

Antibodies against caveolin-1, PTRF, and β-actin were from BD Biosciences; EHD2 from Abcam; SDPR/cavin-2 from R&D Systems. Horseradish peroxidase anti-rabbit, -mouse, and -goat IgGs were from Jackson ImmunoResearch Laboratories.

3T3-L1 cells (J. Pairault, CNRS, Paris, France) maintained in high-glucose DMEM with 10% FCS were induced to differentiate by adding isobutylmethylxanthine (100 μmol/L), dexamethasone (0.25 μmol/L), and insulin (1 µg/µL) for 2 days and then cultured with insulin alone. Differentiated adipocytes were transduced with an adenovirus encoding canine Cav1. Growing 3T3-L1 cells were transduced with retroviral supernatants from Phoenix packaging cells (Indiana University National Gene Vector Biorepository, Indianapolis, IN) expressing Cav1-RFP, PTRF-eGFP, or SRBC-eGFP in pBabe vectors, followed by selection for puromycin resistance (6 µg/mL). Populations of antibiotic-resistant 3T3-L1 fibroblasts were allowed to differentiate into adipocytes as described above.

Animal studies were approved by the Ethics Committee in Animal Experiment Charles Darwin (agreement no. Ce5/2011/015). 3T3-F442A preadipocytes were grown to near confluence and trypsinized, and cells in PBS were injected subcutaneously (3 × 10⁷ cells per site) in 6-week-old BALB/C NU/NU athymic mice (Charles River Laboratories). Four weeks after implantation, mice on a standard diet received drinking water containing 30% sucrose. At 14 weeks, mice were killed by cervical dislocation and fat pads were excised. Small pieces of adipose tissues were either fixed with paraformaldehyde 4% or homogenized in lysis buffer for Western blotting. Adipose tissues were processed for immunohistochemistry as previously described (8). Adipocyte diameters were measured using ImageJ software.

Thirty-five participants (29 male, 6 female; 20 Caucasian, 14 African American, and 1 other race; with mean ± SD for age 26.7 ± 5.3 years and BMI 25.5 ± 2.2 kg/m²) were recruited into the overfeeding study. All were provided written informed consent, all study parameters were approved by the institutional review board, and procedures were followed with regard to the Declaration of Helsinki and registered as a clinical trial. Participants consumed a diet of 44% fat, 15% protein, and 41% carbohydrate, where the total daily intake equated to 140% of their normal caloric intake for 8 weeks. Body composition was assessed by DEXA (QDR 4500A; Hologics, Bedford, MA). Under fasting conditions, subcutaneous adipose tissue was collected by Bergstrom needle biopsy for the measurement of fat cell size according to the methodology of Hirsch and Gallian (26) and determination of caveolin-1 protein content.

2-Deoxyglucose uptake was measured in differentiated 3T3-L1 adipocytes as previously described (27).

Differentiated adipocytes were treated as previously described (28), and glycerol release into the medium was measured using a commercial kit (glycerol, GY105; Randox Laboratories).

3T3-L1 differentiated cells were washed three times with ice-cold PBS and homogenized into 25 mmol/L Mes (pH 6.0), 150 mmol/L NaCl, and 1% (w/v) Triton X-100 containing complete protease inhibitor cocktail (Roche). Detergent-resistant membrane preparations were isolated as previously described (29).

Samples were lysed (in 50 mmol/L Tris, pH 7.4; 0.27 mol/L sucrose; 1 mmol/L Na-orthovanadate, pH 10; 1 mmol/L EDTA; 1 mmol/L EGTA; 10 mmol/L Na-β-glycerophosphate; 50 mmol/L NaF; 5 mmol/L Na-pyrophosphate; 1% [w/v] Triton X-100; 0.1% [v/v] 2β-mercaptoethanol; and protease inhibitors), centrifuged (15,000g, 4°C, 10 min), and stored at −20°C. Protein concentrations were determined by a Bio-Rad kit, and samples were subjected to SDS-PAGE and blotted according to standard procedures. Protein signals were visualized using enhanced chemiluminescence (Perbio Science; Thermo Scientific).

3T3-L1 on coverslips were fixed as previously described (30), stained with 0.1 μg/mL LD540 (C. Thiele, University of Bonn, Bonn, Germany), and visualized by confocal laser fluorescence (Zeiss LSM 710). Fluorescence quantification was performed using ImageJ software.

3T3-L1 adipocytes were fixed by 2.5% glutaraldehyde in 0.1 mol/L cacodylate (pH 7.4) and processed as previously described (31). Ultra-thin sections were observed with a JEOL JEM-1400 microscope equipped with a digital camera.

Statistical analysis used Student t test, variance analysis, or Spearman regressions for correlations. Data were considered statistically significant at P values <0.05.

Indirect evidence for a link between adipocyte lipid droplet expansion and caveolae dynamics is provided by defective lipid storage phenotype of caveolin/caveolae-deficient adipocytes in mice or humans. To examine whether a reciprocal relationship could be demonstrated in gain-of-function experiments, we overexpressed caveolin-1 in fat cells. Because adipocytes already express substantial levels of caveolins, we used a strong cytomegalovirus promoter to drive stable expression of a Caveolin-RFP (Cav1-RFP) fusion in 3T3-L1 adipocytes from retroviral vectors (Fig. 1A). Exogenous Cav1-RFP, as well as endogenous cav1, mainly distributed in detergent-resistant membranes after triton ×100 extraction indicating targeting in lipid rafts domains (Fig. 1B). By probing with caveolin-1 antibody, we estimated that exogenous Cav1-RFP represented approximately half of the endogenous cav1 cell content, indicating significant (1.5–2.0-fold) but moderate overexpression (Fig. 1B). Cav1-RFP distributed to the cell surface of terminally differentiated 3T3-L1 adipocytes consistent with caveolae distribution, even if some punctuate labeling evoking intracellular vesicles could also be observed (Fig. 1C), consistent with lipid raft formation and recycling in Golgi membranes (32). Forced expression of Cav1-RFP had no impact on endogenous expression of caveolin-2 or cavin mRNA (Fig. 1D) but increased PTRF/cavin-1 protein by 1.5-fold (Fig. 1E and Supplementary Fig. 1A). Adipose differentiation program was not affected as judged by unaltered levels of key adipogenic transcription factors, lipogenic mRNAs (Fig. 1F), and protein expression in a differentiation time course (Supplementary Fig. 1B). We next examined whether forced caveolin-1 expression could increase caveolae density at the cell surface, evaluated as the number of caveolae invaginations found in a linear plasma membrane stretch on electron microscopy images (Fig. 1G). We observed comparable caveolae density in 3T3-L1 cells and tissue sections of mice adipose tissue, validating relevance of the 3T3-L1 model. Compared with control cell lines (stably transduced with an empty retroviral vector), caveolae density was increased significantly in Cav1-RFP cell lines and approached values found in endothelial cells evenly identified in adipose tissue sections (Fig. 1H). We also observed slight enrichment of PTRF/cavin-1 in detergent-resistant membrane fractions of Cav1-RFP cell lines (Supplementary Fig. 1C). These data are consistent with previous studies in mice (15) indicating that caveolae structures depend on cavin-to-caveolin ratios. Thus, in the absence of overt effects on fat cell differentiation, moderate caveolin overexpression in 3T3-L1 produced surface caveolae-enriched adipocytes, providing a suitable system to investigate relationships with lipid droplet dynamics. Because caveolin-deficient adipocytes are severely insulin resistant (7), caveolae are viewed as positive regulators in insulin signaling. We therefore examined whether ameliorated insulin response associated with elevated caveolae density in Cav1-RFP adipocytes. Contrary to chronic caveolin-1 deficiency in which loss of adipocyte insulin receptors has been reported (31), we detected no change in insulin receptor, insulin receptor substrate (IRS)-1, or extracellular signal–related kinase (Erk)-1/2 expression, but slightly increased protein kinase B (PKB) mRNA and protein contents were present in Cav1-RFP adipocytes (Fig. 2A and B). In dose-response experiments, Cav1-RFP cell lines exhibited increased basal phosphorylation of IRS-1, PKB, and Erk-1/2 (Fig. 2C–E and Supplementary Fig. 2A), which translated into elevated phosphorylation upon insulin stimulation. Basal and insulin-stimulated glucose transport was also higher in Cav1-RFP compared with control adipocytes (Fig. 2F). Subsequent glycolytic-dependent lactate production was increased both in basal and in insulin-stimulated states in Cav1-RFP (Fig. 2G), suggesting active metabolic activity by cav1 overexpression.

Metabolic changes in Cav1-RFP adipocytes suggest that they might deposit more lipids. Compared with control fat cells, size distribution of Oil Red O–stained lipid droplets from Cav1-RFP adipocytes was right shifted, indicating bigger lipid droplets in Cav1-RFP-caveolae–enriched cell lines (Fig. 3A and B). We also transduced parental 3T3-L1 adipocytes with an adenovirus encoding untagged caveolin-1, which almost doubled cav1 protein expression (Fig. 3C) and right shifted lipid droplet size distribution (Fig. 3D), further indicating that lipid droplet expansion by caveolin-1 overexpression was independent of the presence of the RFP tag. The extent of lipid droplet size enlargement in adenoviral experiments was lower than that seen with stable retrovirus-based overexpression, likely because of limited duration of transgene expression (2 days versus 10 days).

To determine whether lipid droplet–promoting effects of caveolae enrichment was limited to cultured cells or could also be observed in vivo, we compared adipocytes generated in nude mice from control or Cav1-RFP stable transfectants injected subcutaneously according to an established protocol (33). In preliminary experiments, single mice received two separate injections with parental cell line in the dorsal and ventral regions, which generated fat pads with similar morphology and fat cell size in the two locations (Fig. 3E), indicating independent size of newly formed fat cells on the injection site (Fig. 3F). Then, stable transfectants (control empty vector and Cav1-RFP preadipocytes) were injected in eight mice, individually receiving the two different cell lines in the dorsal or ventral region to generate Cav1-RFP or control adipocytes in the same animal. Cav1-RFP expression could be recovered in newly formed fad pads confirming their exogenous origin (Supplementary Fig. 2B). Interindividual variability was found in ultimate fat cell size of newly formed fat pads (Fig. 3G), but within a single mouse we consistently found bigger fat cells in pads originating from Cav1-RFP preadipocytes (Fig. 3G and H), with a significant 1.5-fold increase in mean fat cell area in Cav1-RFP (Fig. 3I). All together, these data indicate that caveolae density of adipocyte membrane can modulate the ability of fat cells to accommodate lipids in vitro and in vivo.

These data strongly suggest that caveolin-1 expression participates in the setting of adipocyte expandability, a key parameter in physiological response to nutrient overload. To address this, we assessed caveolin expression in adipose tissue frozen biopsies obtained from healthy human subjects enrolled in a trial designed to investigate metabolic effects of overfeeding. Thirty-five young subjects (6 females and 29 males) participated in the 8-week overfeeding protocol, and caveolin-1 protein expression showed no significant correlation with fat cell size at baseline (data not shown). During overfeeding, all the subjects gained weight (from 4.4 to 10.7 kg, mean value 7.5 kg) and increased fat mass as well as nonfat mass (Fig. 4A). Fat-cell size measurements at baseline and after overfeeding in each subject revealed two different patterns in the overfeeding response: 15 subjects increased the size of their fat cells (Fig. 4B), whereas it remained unchanged or even decreased in 20 subjects (Fig. 4C). Since total fat mass similarly increased in the two groups during the overfeeding period, this was indicative of different patterns of adipose tissue response dominated either by hyperplastic (increase in adipocyte number) or hypertrophic (increase in adipocyte lipid droplet size) expansion. In subjects responding to overfeeding by adipocyte enlargement, we found that caveolin-1 protein expression at baseline positively correlated with the change in fat cell size (Fig. 4D), whereas no association was observed in hyperplastic responders. Thus, in healthy humans, adipocyte caveolin-1 is positively associated with the ability to expand fat cells in the face of nutrient overload.

We established high expansion capacity in adipocytes with dense caveolae membranes and next investigated reciprocal relation associated with lipid droplet reduction. We used an experimental setting in which acute lipid droplet shrinkage was induced by prolonged stimulation of lipolysis by cAMP/protein kinase A activation. 3T3-L1 fully differentiated adipocytes were treated with lipolytic agents, 8Br-cAMP (a stable cAMP analog) or forskolin (FSK) (an adenylate cyclase activator), combined with methylisobutylxantine (MIX) (a cAMP-degrading phosphodiesterase inhibitor). Prolonged treatment (30 h) did not affected cell density or viability (Supplementary Fig. 3) but resulted in robust lipolytic stimulation (six- to eightfold) as judged by the release of triacylglycerol-derived glycerol (Fig. 5A). In this setting, almost constant lipolytic rates were observed (Fig. 5A), which led to markedly smaller lipid droplets by staining with LD540, a fluorescent neutral lipid probe (Fig. 5B). Striking reshaping of the plasma membrane was evident in adipocytes treated for 30 h with 8Br-cAMP, which almost completely lacked invaginated caveolae, whereas they were normally found in control cells (Fig. 5C). Quantitative analysis of electron microscopy images confirmed drastic reduction of caveolae density by 8Br-cAMP treatment (Fig. 5D). Reduction of caveolae number was also observed in adipose tissue electron microscopy sections of mice induced for adipose tissue mobilization by 48-h fasting (Fig. 5E).

To investigate the mechanism of caveolae disassembly, we focused on the main caveolar coat component, caveolin-1, and observed no sign of decline upon prolonged lipolytic stimulation (Fig. 6A). Instead, total caveolin protein accumulated in 8Br-cAMP–treated cells from 18 h onwards, resulting in a two- to threefold higher content at end point (Fig. 6B). A similar response was observed using FSK and MIX suggesting a cAMP-mediated event (Fig. 6C). This is in agreement with increased caveolin contents already reported in adipose tissue of rodents during fasting (34). Therefore, loss of caveolin protein is unlikely to explain disappearance of caveolae after prolonged lipid mobilization. As caveolae structures are dependent on cholesterol and caveolins are cholesterol-binding proteins (3), we next compared free cholesterol content in 8Br-cAMP– or FSK/MIX-treated adipocytes and found no change, suggesting a process independent of membrane cholesterol reduction (Fig. 6D). However, immunofluorescence revealed marked changes in caveolin-1 distribution, which evolved from a thin continuous labeling at cell periphery in control cells to a broken irregular pattern in 8Br-cAMP–treated adipocytes, consistent with disappearance of surface caveolae structures (Fig. 6E). Therefore, we next examined the possibility that cavin adaptors, which are required for caveolae assembly, might be targeted during lipid droplet shrinkage. In time course experiments, PTRF/cavin-1 and SDPR/cavin-2 markedly changed upon 8Br-cAMP treatment (Fig. 7A). PTRF/cavin-1 showed a biphasic response to 8Br-cAMP with a transient increase within the first 12 h, followed by a rapid decline at later time points (Fig. 7B). SDPR/cavin-2 also markedly declined after 12 h with 8Br-cAMP but did not increase at shorter times (Fig. 7C). We also noticed a similar response of another noncavin protein, EHD2, which also drastically decreased with 8Br-cAMP (Fig. 7D). EHD2 is a dynamin-related ATPase (35) involved in many aspects of membrane dynamics (36). It colocalizes with caveolin-1 (37), interacts with cavin-1 (38), associates with caveolae (14), and links caveolae to actin filaments (39). FSK/MIX fully mimicked the 8Br-cAMP effect in reducing PTRF/cavin-1, SDPR/cavin-2, and EHD2 contents at end points (Fig. 7E). To test for decreased cavin protein stability, we used cavin-GFP reporters stably expressed in 3T3-L1. In this setting, fluorescent cavins are transcribed from a constitutive strong promoter, and changes in cell fluorescence intensity reflect protein stability. In 8Br-cAMP–treated adipocytes, lipid droplet shrinkage (see the information on LD540 staining) coincided with a drop in fluorescence intensity of PTRF-GFP (Fig. 7F) and SRBC-GFP (Fig. 7G), whereas caveolin-1–RFP remained unaltered (Fig. 6H and I). We also observed that PTRF/cavin-1 loss could be prevented by addition of a proteasome inhibitor MG132 (Supplementary Fig. 4A). These data demonstrate posttranslational regulation of PTRF/cavin-1 and SRBC/cavin-2 by 8Br-cAMP and identify cavin protein degradation, likely through proteasome, as a key process for cavin loss and caveolae disassembly upon adipocyte lipid mobilization. We also noticed significant reduction in SDPR/cavin-3 and EHD2 mRNA in 8Br-cAMP–treated cells (Supplementary Fig. 4B), indicating that transcriptional regulation might additionally contribute to the loss of some caveolar proteins.

Our data establish reciprocal relationship linking caveolae dynamics and lipid store fluctuations in adipocytes and define the key steps in cavin/caveolin balance for adaptive maintenance of fat cell caveolae. We demonstrate that caveolar density can be raised by enforced adipocyte caveolin-1 expression, which promotes fat cell expandability through ameliorated insulin response. Furthermore, we found that adipose tissue caveolin-1 expression specifically associated with hypertrophic (and not hyperplastic) fat tissue expansion in healthy human subjects challenged by overfeeding. Conversely, lipid droplet shrinkage by sustained mobilization is shown here to ultimately produce the disassembly of caveolae through cavin-dependent protein degradation. Therefore, our findings highlight bidirectional lipid droplet to membrane cross-talk for the maintenance of fat cell metabolic flexibility and adaptation to highly variable metabolic conditions primarily impacting lipid droplet volume to ultimately resolve fat cell size. They indicate membrane dynamic response to fat storage, in which continuous caveolae formation (through caveolin-1–regulated expression) is required for fat cell membrane expansion, whereas lipid droplet shrinking associates with intensive remodeling (driven by cavin protein targeting) and caveolae disassembly needed to adapt membrane excess. Our observations fit with a prevailing role of caveolae as sensors of cell membrane tension and also suggest a new aspect of lipid droplets as mechano-active organelles driving membrane stretching, at least when of sufficient size as found in adipocytes. In line, a vimentin network tightly enwrapping adipocyte lipid droplets was identified in early morphological studies (40), and recent proteomic analysis of lipid droplets revealed association with cytoskeleton and molecular motors components (29,41–43). All together, these observations contribute to expand the fast-growing list of lipid droplet–related processes related to cell architecture (rev. in 44).

Our study underlines the importance of caveolae density to promote fat cell expandability, a question with physio-pathological relevance in the context of epidemic development of metabolic diseases worldwide. As a hallmark of obesity, lipid droplet expansion defines adipocyte hypertrophy, and a prevailing view is that saturation of adipose tissue storage leads to the spillover and dissemination of fatty acids to contribute to multiorgan metabolic complications. Therefore, identification of cellular effectors participating in the control of adipocyte lipid storage is of interest to target fatty acid lipotoxic effects. Although it was recognized for long that lack of caveolae severely compromises fat cell expansion as seen in lipoatrophic diabetic caveolin-deficient patients (9,17), caveolar protein expression is not generally associated with obesity, suggesting that steady-state caveolae is globally maintained in obese animals or humans. In agreement, a recent study of caveolin-1 and cavin-1 expression in single adipocytes established a close correlation with their cell surface area (45). Here, we bring evidence that adipocyte caveolae, rather, serve in the adaptation to lipid store fluctuations contributing to fat cell flexibility and provide a scheme for adipocyte caveolin/cavin protein interplay in this adaptive response. During fat storage, membrane surface growth is required for adipocyte expansion, and caveolin-1 expression is rate limiting for additional caveolae formation whereas cavin-1 can be recruited from noncaveolar pools. Noteworthy, adipocyte PTRF/cavin-1 is not restricted to caveolae but also associates with nonraft membranes (Supplementary Fig. 1C) as well as lipid droplets (29), which likely provide reservoirs. During extreme fat cell shrinkage, cavins are targeted to degradation, an event reported in other cell types and conditions as instrumental to signal caveolae disassembly (23). Lipid mobilization is shown here to induce a complex biphasic response of cavin-1. Acutely, PTRF/cavin-1 protein content increases, as well as caveolin-1, in agreement with that reported in fasting mice adipose tissue (34,46). Only in later phases, when lipid store shrinks, do cavin proteins and caveolin-1 become differentially regulated and caveolae ultimately disassemble. Whether PTRF/cavin-1 biphasic response is linked to phosphorylation on serine residues, recently demonstrated upon acute cAMP/PKA activation (46), or can be induced after PKG stimulation, another lipolytic pathway in adipocytes (47) remains to be established. Furthermore, we cannot exclude that the loss of EHD2, which is observed within a similar time frame, may also be an upstream molecular event in caveolae disassembly by lipid droplet shrinkage. Indeed, depletion of EHD2 was recently shown to result in more dynamic and short-lived caveolae (38).

Lastly, this study provides evidence that raising caveolae density can ameliorate adipocyte metabolic response and provides indications that this is achieved through raising basal as well as insulin-stimulated effects. This is in line with previous observations with caveolin-1 modulation by miRNAs (48), mirroring a state of metabolic inflexibility in caveolae-deficient mice models by either caveolin-1 or cavin-1 knockout (49,50). In line, one study also reported that mice hepatocytes, which normally express very low caveolin, could be sensitized for insulin action by exogenous caveolin-3 expression (51).

To conclude, our present data establish fat cell caveolae function in a unique cell response to lipid store variation suggesting lipid-induced mechanotension. Even if caveolae response to mechanotension has been described to occur within minutes, kinetics of lipid-induced mechanotension in adipocytes may be somewhat slower but might involve similar end point responses. On the whole, intervention on caveolin/cavin protein balance might be metabolically beneficial to optimize fat storage into adipocytes.

Acknowledgments. Kai Simons (MPI-CBG, Dresden, Germany) provided adenovirus encoding canine Cav1. Plasmids encoding Cav1-RFP, PTRF-eGFP, and SRBC-eGFP were generously provided, respectively, by R.G. Parton (Institute for Molecular Biosciences, University of Queensland, Brisbane, Australia) and R.G. Anderson (Department of Cell Biology, University of Texas Southwestern Medical Center, Dallas, TX).

Funding. Funding from European Community FP7/2007-2013 LipidomicNet grant agreement no. 202272 and from NORC grant no. P30DK072476 (National Institutes of Health–National Institute of Diabetes and Digestive and Kidney Diseases) is acknowledged. S.L.L. is funded by a grant from “Region Pays de la Loire.”

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. N.B. researched data and reviewed and edited the manuscript. C.P., G.M., F.L., X.L.L., and J.D.C. researched data. E.R. contributed to discussion and reviewed and edited the manuscript. S.L.L. and I.D. wrote the manuscript and researched data. S.L.L. and I.D. 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.