Pleiotropic Effects of Cavin-1 Deficiency on Lipid Metabolism*

Background: Cavin-1 expression is critical for the formation of caveolae, exceptionally abundant structure in adipocytes. Results: Adipocytes from cavin-1-null mice are markedly dysfunctional exhibiting decreased insulin-dependent glucose uptake, fatty acid uptake, and lipolysis. Conclusion: Cavin-1/caveolae have major roles in the regulation of adipocyte metabolism. Significance: Fat cell cavin-1/caveolae functions are important for organismal lipid metabolism in mice and men. Mice and humans lacking caveolae due to gene knock-out or inactivating mutations of cavin-1/PTRF have numerous pathologies including markedly aberrant fuel metabolism, lipodystrophy, and muscular dystrophy. We characterized the physiologic/metabolic profile of cavin-1 knock-out mice and determined that they were lean because of reduced white adipose depots. The knock-out mice were resistant to diet-induced obesity and had abnormal lipid metabolism in the major metabolic organs of white and brown fat and liver. Epididymal white fat cells from cavin-1-null mice were small and insensitive to insulin and β-adrenergic agonists resulting in reduced adipocyte lipid storage and impaired lipid tolerance. At the molecular level, the lipolytic defects in white fat were caused by impaired perilipin phosphorylation, and the reduced triglyceride accumulation was caused by decreased fatty acid uptake and incorporation as well as the virtual absence of insulin-stimulated glucose transport. The livers of cavin-1-null mice were mildly steatotic and did not accumulate more lipid after high-fat feeding. The brown adipose tissues of cavin-1-null mice exhibited decreased mitochondria protein expression, which was restored upon high fat feeding. Taken together, these data suggest that dysfunction in fat, muscle, and liver metabolism in cavin-1-null mice causes a pleiotropic phenotype, one apparently identical to that of humans lacking caveolae in all tissues.

The abundance of caveolae in adipocytes suggests that they may play a particular role in their most active functions, namely to take up and store fatty acids (FA) as triglycerides (TG) in the fed state and hydrolyze TG and release FA in response to fasting/starvation and exercise (8). The trafficking of high concentrations of FA in and out of fat cells poses a potential toxicity problem because FA are mild detergents capable of disrupting the adipocyte plasma membrane (22), but caveolae are detergent-resistant membrane domains (23) that can function to buffer the effects of FA. In fact, data from cells expressing caveolins/caveolae support the FA buffering function of caveolins/caveolae that protects against FA cytotoxicity as compared with cells that do not have caveolins, and in addition, caveolins modulate the rate of transmembrane FA flux (24 -26). Data from mice lacking caveolins/caveolae due to gene knock-out experiments (15,27,28) as well as essentially identical data from functionally inactivating mutations in human caveolar proteins (29 -33) document multiple effects of caveolae deficiency on overall fuel metabolism as well as on the regulation of adipocyte lipid levels, thus begging the question of the cellular and molecular basis for these pathologies.
To address aspects of the mechanism(s) underlying the dysregulation of metabolism due to caveolae deficiency, we carried out a series of in vivo and in vitro experiments examining the metabolic properties of several tissues from cavin-1-null mice. We find that the white adipose tissue of the knock-out mice contains small fat cells, is fibrotic and macrophage infiltrated. Isolated white adipocytes from the knock-out respond poorly to ␤-adrenergic and insulin stimuli and show marked inability to take up and store fatty acids. Brown adipose tissue (BAT) mass is relatively normal in the knock-out, but brown adipocytes are smaller and express low levels of uncoupling protein 1 (UCP1). Livers from the knock-out mice are mildly steatotic. Remarkably, the knock-out mice are resistant to dietinduced obesity and show no exacerbation of the above noted metabolic abnormalities upon high fat feeding. Thus, cavin-1 deficiency results in altered metabolic flux among multiple tissues, which suggests a role of cavin-1 in the coordination of peripheral glucose and fatty acid storage and utilization.

EXPERIMENTAL PROCEDURES
Experimental Animals-Cavin-1 Ϫ/Ϫ mice were created as described on a C57BL/6N ϫ sv129 genetic background (15) and were backcrossed for at least 8 generations with the C57 black lineage. The mice used in the present study were homozygous male cavin-1 Ϫ/Ϫ and their wild-type (WT) littermates generated from breeding of cavin-1 ϩ/Ϫ mice. Animals were maintained in a pathogen-free animal facility at 21°C under a 12-h light/12-h dark cycle with access to a chow diet (CD, 2918; Harlan Teklad Global Diet, Madison, WI). For dietary alterations, mice were individually housed and fed with a high-fat diet (HFD, 45% kcal in fat, D12492; Research Diets, New Brunswick, NJ), or chow, starting at the age of 8 weeks and continued for 12 weeks. Body weights were measured weekly and food intakes were measured at intervals. For in vivo evaluation of metabolic parameters, the mice were food-deprived for 4 h prior to experimental procedures except where noted. For tissue harvesting, mice were sacrificed, and tissues were immediately frozen in liquid nitrogen and stored at Ϫ80°C until biochemical analysis, or were fixed for histology and immunohistochemistry. For the preparation of isolated adipocytes, freshly harvested adipose tissue was digested by collagenase in Krebs-Ringer bicarbonate (KRB) buffer (34). All animal studies were performed in accordance with the guidelines and under approval of the Institutional Review Committee for the Animal Care and Use of Boston University.
Cell Culture-MEF cells from cavin-1 Ϫ/Ϫ and WT mice were isolated and cultured as previously described for obtaining Cav1-null MEFs (26). Cavin-1 overexpression in HEK293 cells was achieved as reported in Ref. 16.
Histology and Immunohistochemistry-Tissues were fixed with Z-Fix (Anatech Ltd.), embedded in paraffin, sectioned at a thickness of 6 -8 m, and stained with hematoxylin and eosin (H & E). For collagen detection, sections were stained with Masson's trichrome reagent as specified by the manufacturer (Sigma). Rat anti-mouse Mac-3 antibody (BD Biosciences) was used for Mac3 staining using standard procedures.
Tissue Lipid Content-Lipids were extracted by the Folch procedure (35), dried under N 2 and reconstituted in PBS (1% Triton). Total TG and cholesterol content was measured using commercial enzymatic assays (Pointe Scientific, Canton, MI).
Glucose and Fat Tolerance Tests-Mice were fasted overnight (16 h). For glucose tolerance tests, mice were injected intraperitoneally with glucose (1.5 g/kg body weight). Tail blood glucose was measured at various time points using a glucometer (Bayer). For the fat tolerance test, mice were given an oral bolus (20 l/g body weight) of olive oil. Tail vein blood was sampled before and at various times after olive oil administration for determination of plasma TG levels by enzymatic assay.
Acute ␤-Adrenergic Stimulation in Vivo-Age-matched cavin-1-null mice and wild-type mice were fasted for 4 h and subjected to an intraperitoneal injection of isoproterenol (10 mg/kg) or saline. Blood was collected before and 20 min after injection for determination of free glycerol levels (Sigma) and free fatty acid levels (Biovision). The white adipose tissues were rapidly harvested and snap frozen in liquid N 2 for subsequent analysis.
Adipocyte Isolation and Lipolysis, Glucose Uptake, and FA Uptake-Primary adipocytes were isolated from epididymal fat pads by collagenase as described, in the presence of 200 nM adenosine (Sigma) (34). The isolated cells were filtered through a 250-m nylon mesh and washed three times. Cell diameter was determined from photomicrographs, and the average fat cell weight was calculated (36). Total neutral lipid was measured as described (37). For lipolysis, aliquots (0.5 ml) of cell suspension (7%, v/v) were incubated with 4 g/ml adenosine deaminase (Roche) and 20 nM (Ϫ)-N 6 -(2-phenyl-isopropyl)adenosine (Sigma) for basal lipolytic activity or 1 M isoproterenol for stimulated activity. Glycerol released into the medium was determined as lipolytic activity using a fluorometric assay (38). Triplicate measurements were performed for each mouse and 6 -7 mice of each genotype were tested. For glucose uptake, isolated adipocytes were washed in glucose-free KRH buffer with 4% BSA and 200 nM adenosine, then aliquots (0.4 ml) of cell suspension (10%, v/v) were incubated with or without insulin (1 nM) for 20 min before adding a glucose mixture (0.1 mM 2-deoxyglucose and 0.5 Ci/ml deoxy-D-glucose, 2-[1,2-3 H(N)]-) (PerkinElmer, NET 328) for 5 min. Uptake was terminated by transferring the cell suspension into a microfuge tube containing 100 l of silicone oil and immediately spinning to separate cells from media. Radioactivity associated with the cells was measured with scintillation counting, and zero-time values were subtracted. The data were normalized to cell number and the rate of uptake was expressed as CPM per 10 6 cells per min. For fatty acid uptake, isolated adipocytes were suspended in KRH buffer (0.1% BSA) at a density of 30%. Transport of oleate was started by adding 90 l of a mixed cell suspension to 10 l of isotope solution (containing 0.5 Ci/ml 3 H-labeled oleate (PerkinElmer, NET 289) and oleate complexed to BSA at ratio of 2:1). Uptake was terminated at 15 s and 30 s, respectively by the addition of 5 ml of cold stop solution (200 M Phloretin), and the cells were separated from medium by vacuum filtration using A/E glass filters. Cell-associated radioactivity was obtained by counting the washed filters (39,40).
Cell incorporation of fatty acid was measured using 3 H-labeled oleate (Oleate: BSA 2:1) incubated with aliquots of 10% isolated adipocytes for 60 min. The cell suspension was transferred to a microfuge tube containing silicone oil and immediately spun to separate cells from media. Radioactivity associated with the cells was measured by scintillation counting.
LDH Release in Cultured Adipose Tissue-Epididymal fat pads of cavin-1 Ϫ/Ϫ and WT mice were obtained by dissection, and ϳ50 mg of explants were incubated in KRB buffer at 37°C for 2 h with gentle shaking. The medium was collected at the end of incubation, and LDH activity was measured using a commercial LDH cytotoxicity assay kit (Cayman Chemical Co., Ann Arbor, MI). Zero-time tissue pieces were homogenized in KRB containing 0.1% Triton to assess total intracellular LDH in the tissue.
Quantitative RT-PCR-Total RNA was isolated from indicated tissues or cells with TRIzol reagent (Invitrogen), and the cDNA was synthesized using Reverse Transcription System (Promega). Real-time PCR was performed with the ViiA7 detection system (Applied Biosystems) using Fast SYBR Green Master Mix (Applied Biosystems). Gene expression levels were normalized to 36B4 and presented relative to the wild type. The primer sequences are available upon request.
Statistics-Data are presented as means Ϯ S.E. The significance of differences between groups was evaluated using ANOVA or 2-tailed unpaired Student's t test. A p value Ͻ0.05 was considered significant.

Cavin-1-null Mice Have Reduced Adipose Depot Mass and
Smaller Fat Cells-We reported previously that 8 -12-weekold male cavin-1 Ϫ/Ϫ mice have a slight but nonsignificantly reduced body weight compared with their WT littermates (15), but they have significantly lower levels of overall adiposity. With a larger cohort of mice, we confirmed that cavin-1 Ϫ/Ϫ and WT mice do not show significant differences in body weight (Fig. 1A) at this age, although we do find there is a tendency for cavin-1 Ϫ/Ϫ mice to have lower body weights than WT littermates after 5 months of age (not shown). Epididymal, subcutaneous, and perirenal white fat depot weights were reduced by 60 -70% in the null mice, but brown adipose tissue (BAT) mass was the same (Fig. 1A, right most graph). To determine the basis for the reduced adiposity, we measured cell diameters from each genotype for 200 randomly chosen perigonadal adipocytes. Adipocyte diameter follows a normal distribution in both knock-out and wild-type mice, but fat cells were significantly smaller in cavin-1 Ϫ/Ϫ adipose tissue with the average cell diameters of cavin-1 Ϫ/Ϫ fat cells being 37.6 Ϯ 2.8 as compared with 75.0 Ϯ 4.4 m for WT adipocytes (Fig. 1B). The difference in fat content is even greater when expressed as fat cell weight  MARCH 21, 2014 • VOLUME 289 • NUMBER 12 ( Fig. 1C). H&E staining on fixed epididymal adipose tissue also confirmed the cell size difference in situ, as shown in Fig. 1D.

Cavin-1 and Lipodystrophy
Cavin-1 Deficiency Results in Reduced Isoproterenol-stimulated Lipolysis, Loss of Insulin-stimulated Glucose Uptake and Reduced Fatty Acid Uptake and Incorporation into Lipid-To establish a metabolic explanation for the reduction of adipose tissue mass (15), we examined lipolysis in vivo ( Fig. 2A) and in isolated adipocytes (Fig. 2B). Baseline serum glycerol levels were similar in the WT and cavin-1 Ϫ/Ϫ mice while free FA levels were higher in cavin-1 Ϫ/Ϫ mice as we previously reported, but there was a blunted response to isoproterenol administration in the knock-out (25% lower after 20 min), reflected by decreased levels of both free glycerol and free fatty acid. In isolated adipocytes in vitro, basal lipolysis per adipocyte was slightly lower in the knock-out than in WT, and the response to isoproterenol was markedly blunted so that rate of adrenergically stimulated lipolysis was ϳ85% lower in the knock-out cells. We previously showed cavin-1 Ϫ/Ϫ mice have impaired insulin sensitivity, reduced GLUT4 expression, and lower levels of the insulin receptor signaling pathway components in adipose and muscle tissue (15). As shown in Fig. 2C, rates of basal 2-DOG uptake and the magnitude of the insulin-FIGURE 2. Cavin-1 deficiency results in reduced isoproterenol-stimulated lipolysis, loss of insulin-stimulated glucose uptake, decreased fatty acid uptake and incorporation, and impaired fat tolerance. A, serum free glycerol levels and free fatty acid levels in wild-type and knock-out mice before and 20 min after isoproterenol administration (10 mg/kg) (n ϭ 5-6). B, basal and ␤-adrenergic receptor agonist-stimulated (isoproterenol, 1 M) lipolysis, was measured as glycerol release over a 2-h period for isolated primary adipocytes from WT and Cavin-1-null mice. The lipolytic rate was normalized to adipocyte cell number (n ϭ 6 -7). C, primary adipocytes were treated with 1 nM insulin or not and 2-DOG uptake was determined as described under "Experimental Procedures" (n ϭ 4). D, relative cell-associated radioactivity incorporation of oleate into lipid over a 60-min period in WT and Cavin-1-null adipocytes (n ϭ 3). E, initial uptake kinetics of 33.3 M [ 3 H]oleate into adipocytes, 15 s and 30 s from WT and Cavin-1-null mice (n ϭ 4). F, oral fat tolerance test: mice were fasted for 16 h, followed by an oral bolus of olive oil (20 l/g BW). Aliquots of tail vein blood were sampled at the times indicated after gavage for the determination of plasma TG levels (n ϭ 4 -5). Data are expressed as the mean Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001 for WT versus KO. stimulation uptake, which reflects transport, were markedly lower in the knock-out adipocytes. As glucose uptake and conversion to glyceride-glycerol is needed to esterify fatty acids, the cavin-1 Ϫ/Ϫ fat cells would be expected to have lower triglyceride synthesis. We further examined the rates of free fatty acid uptake into adipocytes and incorporation into TG. We have previously reported decreased FA uptake in Cavin-1 knockdown 3T3-L1 adipocytes (16) and caveolin-1 has been suggested to function in FA uptake by localizing CD36 to caveolae where it can enhance FA uptake (44). Our results showed a markedly blunted initial rate of oleate uptake (Fig. 2E) and incorporation into cells (lipid) (Fig. 2D) in isolated cavin-1-null adipocytes and levels of CD36 protein were also substantially decreased. The impaired glucose and FA uptake explains the much lower triglyceride levels in cavin-1 Ϫ/Ϫ fat cells, and hence we expected reduced clearance of an acute oral fat load. Consistent with this idea, the fat tolerance test showed a significantly greater increase in serum TG after a lipid gavage in the knock-out mice, suggesting decreased fat clearance (Fig. 2F).
Isoproterenol-stimulated Perilipin-1 (Plin1) Phosphorylation Is Blunted in Fat Cells Lacking Cavin-1-We next examined epididymal fat for selected proteins involved in the regulation of lipolysis (45) and found hormone sensitive lipase (HSL), Plin1, comparative gene identification-58 (CGI-58) and adipocyte triglyceride lipase (ATGL) levels were not significantly different between the cavin-1 Ϫ/Ϫ and WT littermates (Fig. 3A). We then measured basal and ␤-adrenergic-activated phosphorylation events in adipose tissue and showed that phosphorylation of HSL was unaffected, but Plin1 phosphorylation was significantly reduced (Fig. 3B). Adrenergic stimulation was repeated for a cohort of 4 mice and HSL Ser-563 and Ser-660 sites were phosphorylated to the same extent in WT and cavin-1 Ϫ/Ϫ mice, but Plin1 phosphorylation was reduced by ϳ70% in cavin-1-null adipose tissue (Fig. 3, C and D). This result suggests that absence of cavin-1 results in selective inhibition of Plin1 phosphorylation, but the upstream PKA signaling is not affected in these cells.
The Adipose Tissue of Cavin-1-null Mice Is Fibrotic, Macrophage-infiltrated, and Leaks LDH-Insulin resistance related to adipocytes may include contributions by macrophage-mediated inflammation, fibrosis caused by collagen deposition, and cell death/necrosis. Accordingly, we stained adipose tissue from WT and cavin-1 Ϫ/Ϫ mice with trichrome for collagen (Fig. 4A), and Mac3 for macrophage infiltration (Fig. 4B) along with mRNA expression of macrophage markers F4/80 and CD68 (Fig. 4C). Lipotoxicity/cell leakiness was determined by measuring the release of LDH from adipose tissue explants (Fig.  4D) and isolated adipocytes (not shown). The data showed increased fibrosis, macrophage infiltration, and LDH release, respectively, for the cavin-1 Ϫ/Ϫ -derived adipose tissue as compared with the WT adipose tissues.
Cavin-1 has been reported to inhibit collagen gene expression by associating with a protein called binding factor of a type-1 collagen promoter (BFCOL1) (46). Accordingly we show in Fig. 4E that the expression of three collagen (Col) genes is  MARCH 21, 2014 • VOLUME 289 • NUMBER 12 up-regulated in cavin-1-null MEFs along with fibronectin1 (Fn1), whereas vimentin (Vim) is not. As might be expected, the opposite occurs in HEK293 cells overexpressing cavin-1 (Fig.  4F), namely two additional Col genes are down-regulated. These Col genes were all seen to be up-regulated in microarrays from cavin-1-null adipocytes (data not shown). Thus, the gene expression data of Fig. 4, E and F are consistent with the histological data of Fig. 4A and with a regulatory role for cavin-1/ PTRF in collagen gene expression.

Cavin-1 and Lipodystrophy
Cavin-1-null Mice Are Resistant to Diet-induced Obesity-To test whether cavin-1 Ϫ/Ϫ mice could increase fat storage, mice were challenged with a high fat diet. As shown in Fig. 5A, WT mice gained more weight on the HF compared with the control chow diet, but the knock-out mice did not. The differences in body weight could not be explained by hypophagia, as daily average food intake (kcal/d) was similar in WT and knockout mice on either diet (Fig. 5C). WT mice were considerably more glucose intolerant than the knock-out mice after high-fat feeding (Fig. 5D). The latter are mildly glucose intolerant whether or not they are on a HFD.
Livers from Cavin-1-null mice Are Steatotic, Which Is Not Worsened by HFD Whereas Skeletal Muscle from the Knockouts Fails to Accumulate Fat on Either Diet-We measured liver (Fig. 6A) and skeletal muscle lipid content (Fig. 6B) in WT and

Cavin-1 and Lipodystrophy
cavin-1 Ϫ/Ϫ mice and observed that the latter had slightly fattier livers than WT on chow, but liver fat content did not increase on a HFD while it significantly increased in WT on a HFD. Skeletal muscle from the knock-out animals did not accumulate fat under either diet, whereas the WT did when subjected to excess calories for 12 weeks. In confirmation of Fig. 6A, the expression level of the liver lipid droplet protein, Plin2 (45) is higher in the knock-out animals fed a chow diet but is essentially the same in both genotypes on a HFD (Fig. 6C). The mRNAs for selected genes involved in the regulation of hepatic lipogenesis (ACC, acetyl-CoA carboxylase; SREBP, sterol response element-binding protein) were also higher in livers of cavin-1 Ϫ/Ϫ mice, but unlike WT mice, they were not increased following a HFD (Fig. 6D).

UCP1 Expression Is Induced in Mice Lacking
Cavin-1 upon a HFD-The null mice do not gain any weight upon exposure to a HFD (Fig. 5A) despite normal food intake, which suggested that they may have enhanced fat oxidation under these circumstances. We determined that brown adipocytes, like white fat cells, were smaller in the null mice and remained smaller after a HFD as compared with WT (Fig. 7A). We then measured the expression of selected mitochondrial proteins in these animals (Fig. 7B) and observed that UCP1 levels were low in cavin-1 Ϫ/Ϫ mice on a chow diet, but were markedly up-regulated after a HFD to the level of the WT mice. We also checked mRNA levels of BAT genes that regulate mitochondrial thermogenesis and biogenesis (Fig. 7, C-E) and found that peroxisome proliferator-activated receptor-␣ (PPAR␣), PPAR␥ coactivator-1A (PGC1-␣), uncoupling protein (UCP1), and CCAAT/enhancer binding protein ␤ (c/EBP␤) were all decreased in cavin-1 Ϫ/Ϫ BAT, but were induced to the WT levels by a high fat challenge (Fig. 7C). A metabolic characteristics of brown adipocytes is their high rate of fatty acid oxidation that performs the beneficial function of expending excess energy, thus we examined the abundance of mRNAs for acyl-CoA dehydrogenases, which are important for fatty acid oxidation in mitochondria, including very long-chain acyl-CoA dehydrogenase (VLCAD), longchain acyl-CoA dehydrogenase (LCAD), and medium-chain acyl-CoA dehydrogenase (MCAD). Those genes were all decreased in BAT of cavin-1 Ϫ/Ϫ mice but were induced to the similar level as WT after high fat feeding.

DISCUSSION
Here we phenotypically characterize a novel model of lipodystrophy caused by cavin-1 deficiency, one that corresponds closely to the phenotype of humans lacking functional cavin-1 as described in patients from at least 4 distinct ethno-geographic regions of the globe (31)(32)(33)47). The absence of cavin-1 protein in mice (15) and humans (31-33, 47) eliminates caveolae and causes a lipodystrophic phenotype entailing reduced  MARCH 21, 2014 • VOLUME 289 • NUMBER 12  overall adiposity, impaired glucose tolerance, and hyperlipidemia. To better understand the phenotype, we described some of the underlying cellular/tissue defects that cause it, the predominant one being reduced adiposity in all major white fat depots due to small adipocytes and their inability to store lipid (Figs. 1-3) as well as other adipocyte pathologies (Fig. 4). Adipocytes lacking cavin-1 are markedly refractive to hormonal stimulation by insulin and isoproterenol, the latter due to diminished Plin1 phosphorylation despite normal adrenergic signaling toward HSL (Fig. 3). These cells also have a severely limited ability to utilize glucose (Fig. 2C), and because glycerol-3-P derived from glucose is needed to form the TG backbone, their capacity to esterify and store fatty acids is limited. We confirmed this in vitro in cavin-1-null adipocytes where fatty acid uptake (Fig. 2E) and incorporation (Fig. 2D) are greatly diminished and in vivo by a fat tolerance test (Fig. 2F). Thus the contribution of fat cells to the cavin-1-null phenotype is characterized by their remarkable incapacity to store lipid, and consequently, the observed hyperlipidemia. Considering our present and previous results, we postulate that adipocyte-related insulin resistance results from the imbalance of lipid uptake and lipolysis such that the former is even more impaired than the latter (8), and impaired lipid uptake is a major contributor to the lipodystrophic, insulin-resistant phenotype.

Cavin-1 and Lipodystrophy
Both cavin-1-null and Cav1-null genotypes are dyslipidemic, but the onset of the phenotype is earlier and more pronounced in the former mice because, unlike the latter, they lack caveolae in all tissues including skeletal and cardiac muscle (8). In that regard, we showed that the machinery for insulin-stimulated glucose transport in muscle is down-regulated in cavin-1 Ϫ/Ϫ mice (15). Although the liver lacks caveolae (17), liver function in Cav1-null animals appears to be phenotypically protective, and these mice show reduced levels of steatosis (48). On the other hand, cavin-1-null mice have mildly fatty livers on a normal diet (Fig. 6, A and C), and the level of liver fat is not exacerbated by high fat feeding. Muscle lipid content is unaffected by lack of caveolae (Fig. 5B) regardless of diet, which on the high fat diet, most likely reflects the reduced FA uptake and storage capacity of cells other than fat cells that lack caveolae (25), a property we have observed in cultured muscle cells where caveolins/caveolae have been knocked down. 5 We subjected cavin-1 Ϫ/Ϫ and WT mice to a HFD and observed that they had similar energy intake, but the former are resistant to weight gain and are equal in mass to WT on a chow diet at the same age of 20 weeks. As noted above, the null mice have slightly fattier livers at this time and slightly increased fat mass, but only in the epididymal fat depot (Fig. 5B). The nulls on a HFD also show no worsening of their mildly impaired glucose tolerance (Fig. 5D), a somewhat surprising result that suggests possible metabolic compensation in the cavin-1-null mice by a tissue other than white fat, muscle, or liver upon exposure to a caloric challenge. We therefore examined the properties of brown fat in WT and knock-out mice on chow and high fat diets. Brown adipose mass was unaffected by lack of cavin-1, but the brown fat cells, like white fat cells, were smaller and expressed lower levels of UCP1 (Fig. 7), a protein that has been suggested as a therapeutic target for the treatment of obesity by virtue of its ability to enhance FA oxidation (49). Interestingly, UCP1 levels are enhanced upon high fat feeding of the cavin-1 Ϫ/Ϫ mice (Fig. 7), which likely contributes significantly to their resistance to a HFD-mediated weight gain.
Additional possible contributors to the pathophysiological phenotype of cavin-1 knock-out mice are adipose tissue fibrosis and macrophage infiltration (50) as well as adipocyte fragility as indicated by higher LDH leakage (Fig. 4). The in vitro results for impaired lipolytic and insulin signaling are, however, necessarily independent of fibrosis and of the acute affects of macrophage infiltration. It is also possible that there are increased levels of autophagy in the cavin-1 knock-out adipocytes as was reported for Cav1-null fat cells (51). Fibrosis and macrophage infiltration have been noted in Cav1-deficient adipocytes (52,53) suggesting the lack of caveolae, per se, and not necessarily their individual protein components underlie adipocyte dysfunction in the mouse in vivo. This conclusion is also supported by developmental nutritional studies that reported similar changes in adipocytes for caveolae-related genes, Cav1, Cav2, cavin-1/PTRF, and cavin-3, as a function of diet (54). We cannot rule out, however, that the acute lack of hormonal responsiveness to lipolytic stimulation in vitro (Fig. 2) may have different contributions from the individual caveolae proteins. Indeed, we show here that transfected cavin-1 has effects in HEK293 cells (Fig. 4F), and these cells do not express the other caveolae proteins (16).
An impairment of the lipolytic response to isoproterenol was more marked in isolated cell incubations than in the in vivo test. Steady state plasma glycerol and FFA levels at a single time point after isoproterenol treatment may not accurately reflect the rate of appearance of glycerol and studies using tracer methods to assess more accurately assess turnover rates are needed. Also, basal lipolysis in the in vitro study was assessed under conditions of high adenosine that clamp cAMP at a very low levels; the situation in vivo likely differs.
Mechanistic information concerning how caveolae components participate in the regulation of lipolysis is relatively scarce and contradictory in part. Cohen et al. proposed that a ternary complex of Cav1-perilipin-PKA was required for hormonalstimulated lipolysis, which would therefore be absent in Cav1null adipocytes and would result in reduced Plin-1 phosphorylation and decreased stimulated lipolysis (55). However, in Cav1-null cultured adipocytes derived from Cav1-null MEFs, ISO-stimulated PKA-dependent phosphorylation of Plin-1 is normal and subsequent lipolytic stimulation in these cells is actually slightly but significantly greater than in these cells "rescued" by the re-introduction (by transfection) of Cav1 (26). Hence it seems unlikely that a Cav1-perilipin-PKA complex is required for regulated adipocyte lipolysis. Cavin-1 is possibly a better candidate for directly participating in lipolysis as it has been shown to be phosphorylated upon adrenergic stimulation (56), albeit at significantly later times than HSL and perilipin when lipolysis is well underway, 6 and experiments are under-way to characterize possible molecular interactions responsible for the cavin-1 dependence of hormonal-regulated lipolysis.
We do however provide some new insight into the mechanism of fibrosis in the cavin-1-null mice, namely that cavin-1 regulates collagen gene expression (Fig. 4, E and F). Indeed, we saw a significant enhancement of collagen gene expression in microarrays of adipose tissue (data not shown) that we verified here by qPCR. Whether or not this is due to direct interaction of cavin-1 with BFCOL1 on the collagen promoter (46) remains to be determined as attempts to coimmunoprecipitate these proteins have as yet been unsuccessful. Moreover, it is likely that the effects of cavin-1/PTRF are also manifest in its regulation of ribosomal RNA (57,58), and this possibility is also under experimental scrutiny.
Genetic lipodystrophies (59) like those due to lack of cavin-1 are relatively rare in the human population but the heterozygous condition will be much more prevalent, and there are data suggesting that caveolin-1 expression is down-regulated in adipocytes from obese humans (60). Thus it will be of interest to study the expression of the caveolae protein components including cavin-1 and cavin-2 in mouse models of diet-induced obesity and in cavin-1 ϩ/Ϫ mice, since the cavin-1-null animals resemble humans closely in their metabolic properties. Such studies are underway and preliminary data indicates that mice heterozygous for cavin-1 have glucose intolerance more severe than their WT counterparts.