Caveolin-1 Expression Is Essential for Proper Nonshivering Thermogenesis in Brown Adipose Tissue

Antibodies and their sources were as follows: guinea pig anti-perilipin, guinea pig anti-adipose differentiation-related protein (ADRP), and rabbit anti-prohibitin (all polyclonal antibodies [pAb]; Research Diagnostics, Flanders, NJ); anti-phospho-ser/thr-protein kinase A (PKA)-specific phosphosubstrate, anti-phospho-AMP-activated protein kinase (AMPK)-α (thr172), and anti–AMPK-α, (all pAb; Cell Signaling, Beverly, MA); polyclonal rabbit mitochondrial dicarboxylate carrier protein (mDIC), which was produced as previously described (24); rabbit anti–UCP-1 (pAb; Chemicon International, Temecula, CA); rabbit anti-heat shock protein 60 (Hsp60; pAb; Stressgen, Victoria, BC, Canada); rabbit anti–CD-36 (pAb; Cayman Chemical, Ann Arbor, MI); rabbit anti-adipocyte-fatty acid binding protein/aP2 (pAb; Alpha Diagnostics, San Antonio, TX); rabbit anti–S3-12, which was produced as previously described (25); and mouse anti-flotillin-1 (mAb; BD Transduction Laboratories, San Jose, CA). Anti-HSL (pAb) was a generous gift from Dr. Constantine Londos (National Institutes of Health, Bethesda, MD).

Caveolin-1 null mice and the corresponding wild-type cohorts, in the C57BL/6 genetic background, were generated and maintained as previously described (26,27). For these experiments, homozygous caveolin-1 null mice were backcrossed into the parental C57BL/6 genetic background eight times. Wild-type and caveolin-1 null littermates used in this study were obtained by interbreeding generation F8 caveolin-1 heterozygous mice. Thus, all mice are on a relatively pure C57BL/6 genetic background. All studies were carried out in mice between the ages of 8 and 10 weeks, a time when it has been shown that body weights of these mice are indistinguishable (26). Throughout the article, “baseline” refers to mice that have been fasted for 4 h to limit the effects of recent food consumption on core body temperature (14,28).

The colonic temperature was measured as an indication of core body temperature. A lubricated digital probe (VWR International, Westchester, PA) was inserted 2 cm into the rectum and held in place until a stabilized measurement was achieved. An average of two measurements was taken from each mouse. Baseline (4 h fasting) temperatures were collected from wild-type (n = 15) and caveolin-1 null (n = 15) mice at 10:00 a.m., when the mice had returned to the euthermic state after daily torpor (29). Mice were either further fasted for 24 h at ambient temperature (wild-type n = 5, caveolin-1 null n = 5), placed at 4°C with free access to food for 24 h (wild-type n = 5, caveolin-1 null n = 5), or further fasted for 24 h at 4°C (wild-type n = 5, caveolin-1 null n = 5), at which time core body temperature was again assessed. The total time of the fasting treatments thus totaled 28 h (4-h baseline fast + 24-h experimental fast).

Baseline serum was collected from wild-type (n = 10) and caveolin-1 null (n = 10) mice by bleeding the tail of each mouse. Mice were either further fasted for 24 h at ambient temperature (wild-type n = 5, caveolin-1 null n = 5) or further fasted for 24 h at 4°C (wild-type n = 5, caveolin-1 null n = 5), at which time serum was again collected from the tail. Nonesterified fatty acid (NEFA; Half-Micro tests; Roche) and triglyceride (GPO-Trinder; Sigma) levels were determined colorimetrically.

Adipocyte triglycerides were extracted according to the methods of Dole and Meinertz as modified by Carpéné (30). Briefly, the interscapular brown fat pads were removed from each mouse (n ≥ 5 mice for each group), and ∼50 mg of tissue was homogenized in 2 ml of homogenization buffer (20 mmol/l Tris, pH 7.3, 1 mmol/l EDTA, 1 mmol/l β-mercapto-ethanol). Then, 1 ml of this homogenate was placed in a glass tube, and 2 ml of extraction buffer (78% vol/vol isopropanol, 20% vol/vol heptane, and 2% vol/vol 1N sulfuric acid) was added. Next, 2 ml of heptane was added and mixed, and the extracts were allowed to stand until the two phases separated. Then, 2 ml of the upper organic phase was transferred to a glass vial and evaporated under N₂. The sample was redissolved in 250 μl of heptane, and triglycerides were measured colorimetrically (Wako Chemicals, Nuess, Germany). Results are expressed as a function of total cellular protein content, which was determined using the bicinchoninic acid reagent (Pierce).

Western blotting of interscapular BAT from wild-type and caveolin-1 null mice (n = 5 mice for each group) was performed essentially as previously described (12,13). Quantification of Western blots was performed by densitometry. A P value of <0.05 was deemed significant.

Mice (n ≥5 for each group) were killed, and the interscapular brown fat pads were removed, processed, and stained with hematoxylin and eosin, as previously described (31).

High-pressure freeze electron microscopy was performed on wild-type (n = 2) and caveolin-1 null (n = 2) mice at baseline, as previously described (13).

Mitochondria were isolated from BAT at baseline (n = 5 mice for each group) with the aid of a commercially available kit (MITO-ISO1; Sigma, St. Louis, MO). The integrity of the inner membrane was assessed according to the manufacturer’s instructions (MITO-ISO1; Sigma). Outer membrane integrity was assessed with the aid of another commercially available kit (CYTOC-OX1; Sigma).

All results were analyzed using a two-tailed t test. A P value of <0.05 was deemed significant.

The ability of small mammals, such as mice, to maintain their body temperature during acute cold exposure or prolonged fasting depends on heat production via two sources: 1) shivering thermogenesis in muscle and 2) nonshivering thermogenesis in BAT. To investigate the role of caveolin-1 in heat production via these two sources, we subjected caveolin-1 null mice to prolonged fasting (24 h) as well as to acute exposure to low temperatures (4°C). Because small, captive, nocturnal mammals regularly conserve energy stores by entering a state of daily torpor during the inactive part of their circadian rhythm (second part of the night through early morning), during which body temperature can approach ∼25°C (29), we performed all temperature measurements after the end of this period. Compared with wild-type mice, caveolin-1 null mice have a marginally, yet significantly, reduced basal core body temperature (37.6 vs. 37.2°C, respectively; P < 0.05) (Fig. 1). Interestingly, when a cohort of these mice was further fasted for 24 h, we found that whereas wild-type mice were capable of maintaining a relatively normal body temperature (37.1°C), caveolin-1 null mice were not (35.6°C, P < 0.05) (Fig. 1).

When we next exposed wild-type and caveolin-1 null mice to the cold (4°C, 24 h) with free access to food, we found that there was no statistical difference between the resultant core body temperature in these mice (35.8 vs. 35.3°C, P = 0.3) (Fig. 1). However, these values were statistically below those of mice at baseline, indicating that acute cold exposure produces hypothermia in these animals.

When we subsequently exposed a cohort of wild-type and caveolin-1 null mice to the cold (4°C, 24 h) with concurrent fasting (fasting/cold treatment), we found that the core body temperature of caveolin-1 null mice fell precipitously (23.4°C), whereas wild-type mice were able to maintain a much higher temperature (34.2°C; P < 0.05) (Fig. 1). Furthermore, after 24 h of this treatment, caveolin-1 null mice were found to be markedly obtunded and shivering intensely, whereas wild-type mice appeared relatively normal. Thus, these results indicate that the mechanisms involved in heat production via both shivering and nonshivering thermogenesis appear intact in caveolin-1 null mice; however, the maintenance of body temperature requires constant access to food, suggesting that the liberation of stored energy is defective in caveolin-1 null mice.

Because we have previously shown that pharmacologically mediated lipolysis from WAT is severely defective in caveolin-1 null mice (13), we next sought to investigate whether an analogous defect may exist after fasting or fasting/cold treatment. At baseline we found that there was no significant difference between serum NEFA levels in wild-type and caveolin-1 null mice (Fig. 2A). However, when mice were subjected to either fasting or fasting/cold treatment, we found that serum NEFA levels rose normally and significantly in wild-type mice (32), but they failed to rise in caveolin-1 null mice. These results, along with those reported previously (13), indicate that lipolysis fails to occur in caveolin-1 null mice under either condition. Because the measurement of serum NEFAs predominantly represents fatty acids released from WAT, we next attempted to more directly assess the lipolytic response to fasting and fasting/cold in BAT, as outlined in subsequent figures.

Studies in small mammals and humans have demonstrated that cold exposure leads to dramatic increases in serum NEFA levels, which is accentuated by fasting (32–35). Additionally, it has been shown that in humans, ingestion of nicotinic acid, which severely limits free fatty acid mobilization from adipocytes, significantly blunts cold-induced heat production via nonshivering thermogenesis (34). This scenario is, at a basic level, similar to that which occurs in caveolin-1 null mice, where attenuated lipolysis restricts the availability of free fatty acids as a source for heat production in BAT, leading to resting and cold-induced hypothermia.

In addition to NEFA levels, we also analyzed serum triglyceride levels in wild-type and caveolin-1 null animals. At both baseline and after fasting/cold treatment, there was no difference between the two cohorts of mice, with total serum triglycerides decreasing approximately twofold after exposure to fasting/cold (Fig. 2B). These results indicate that loss of caveolin-1 does not effect the combustion of circulating triglycerides during cold exposure.

To explore the notion that defective lipolysis results in cold and fasting intolerance in caveolin-1 null mice, we next examined the triglyceride content of the interscapular brown fat pads of wild-type and caveolin-1 null mice at baseline and after combined fasting/cold treatment. Using a protocol designed to isolate triglycerides from tissue, we noticed that at baseline there was no statistically significant difference between the triglyceride content of wild-type and caveolin-1 null BAT (Fig. 3). However, after fasting/cold treatment, the triglyceride content of wild-type and caveolin-1 null BAT pads both decreased significantly; on the other hand, levels in wild-type mice fell nearly 10-fold, whereas levels in caveolin-1 null mice were reduced only ∼3-fold (P < 0.05). These data indicate that the liberation of stored triglycerides is defective in the BAT of caveolin-1 null mice and, most likely, contributes to the observed phenotype.

We next analyzed paraffin-embedded hematoxylin and eosin-stained BAT sections from wild-type and caveolin-1 null mice at baseline and after fasting/cold treatment (Fig. 4). At baseline, the adipocytes of wild-type and caveolin-1 null BAT pads were filled with multilocular lipid droplets that, overall, appeared larger in caveolin-1 null mice. Interestingly, after fasting/cold treatment, the adipocytes of wild-type mice were essentially devoid of lipid droplets, whereas those of caveolin-1 null mice appeared more like BAT at baseline. In addition, it appears as though the smaller lipid droplets in caveolin-1 null cells have been utilized, perhaps indicating that proteins other than perilipin (i.e., S3-12 or adipophilin) surround these smaller droplets and are not subject to the same caveolin-1-mediated regulation as perilipin-coated droplets. Overall, these changes are consistent with those noted above and are further indicative of a major lipolytic failure in caveolin-1 null mice.

We next analyzed the expression levels of several key proteins involved in brown fat lipid metabolism for any alterations from wild-type. Similar to what we have reported previously in WAT (13), we found that although the protein levels of perilipin were equal between the two cohorts of mice (P = 0.08), the phosphorylation of perilipin in response to lipolytic stimulation (i.e., fasting/cold) was ∼4.5-fold less (4.63 ± 0.48, P < 0.05) in caveolin-1 null mice (Fig. 5A). Unlike in WAT (data not shown), it seems that in BAT, both perilipin-A and -B are subjected to major phosphorylation in response to lipolytic stimulation. In caveolin-1 null mice, phosphorylation of both perilipin isoforms is dramatically decreased, consistent with the idea that lipolysis is defective in the BAT of these mice.

Regarding other lipid droplet-associated proteins, such as HSL, S3-12, and adipophilin (ADRP), we noted no statistically significant alterations in caveolin-1 null mice at either baseline or after cold/fasting, except in the case of HSL, which increases approximately twofold in caveolin-1 null mice (2.13 ± 0.22, P < 0.05) but not in wild-type mice, after fasting/cold treatment (Fig. 5B). Several studies have shown that HSL expression in BAT and WAT is altered by dietary manipulations, cold exposure, and chronic β-agonist stimulation; however, a consensus on whether these conditions lead to an increase or decrease in HSL expression is lacking (21,36–39). Furthermore, the combined effects of cold and fasting on HSL expression was not explored in any of these studies. Interestingly, flotillin-1, a caveolar resident protein that has recently been shown to associate with lipid droplets (40), was found to be upregulated by about twofold (1.97 ± 0.16, P < 0.05) in the BAT of caveolin-1 null mice at baseline. After cold/fasting, the levels of this protein increase above baseline wild-type values approximately fourfold in both cohorts of mice (4.30 ± 0.48, P < 0.05) (Fig. 5B). Because the role, if any, of flotillin-1 in lipid droplet metabolism is unknown, it is difficult to speculate on the possible functional significance of these alterations in protein expression.

Additionally, two proteins involved in lipid binding and transport, CD36 and adipocyte-fatty acid binding protein/aP2, were found to be statistically unchanged in caveolin-1 null mice compared with wild-type mice (Fig. 5C). Both of these proteins were upregulated approximately twofold (P < 0.05) in response to fasting/cold treatment in the BAT of wild-type and caveolin-1 null mice.

Using a relatively new technique, high pressure and rapid tissue freezing, we examined the interscapular BAT from wild-type and caveolin-1 null mice by electron microscopy. BAT derived from wild-type mice shows characteristic brown adipocytes, with multilocular lipid droplets (Fig. 6, asterisks) and abundant mitochondria (Fig. 6, arrows). Remarkably, examination of caveolin-1 null BAT reveals dramatic alterations in adipocyte morphology. In these cells, the mitochondria are markedly larger and dilated (Fig. 6, arrows). Furthermore, they are much less electron dense than the mitochondria of wild-type mice, thus appearing much lighter in color.

The results shown here are representative of all of the images gathered in which nearly 95% of the mitochondria appear altered in caveolin-1 null BAT. This dilation is most likely the result of an altered osmotic gradient between the inner mitochondrial membrane and the cytoplasm, causing an influx of fluid. It is possible that caveolin-1 provides a necessary scaffolding structure for protein, water, or ion transport components across the mitochondrial membrane, the disruption of which leads to the accumulation of proteins or ions within the mitochondria. This speculation is not without basis, as ion and water channels have been shown to be associated with plasma membrane caveolae (41,42). In addition, caveolin-1 has been shown to be targeted to a variety of intracellular locations, including mitochondria, in a tissue-specific fashion (43). Although the mitochondria of BAT have not been studied directly, it is possible that caveolin-1 normally localizes to the membrane of this organelle and when absent leads to the phenotype described above.

To investigate whether these morphologic changes were accompanied by alterations in mitochondrial function, we next examined the expression of several mitochondrial proteins by Western blot. Interestingly, expression of mDIC was found to decrease ∼3.5-fold (3.61 ± 0.68, P < 0.05) after fasting/cold treatment in wild-type mice, as shown previously (24); however, in caveolin-1 null mice the expression levels of mDIC remain statistically unchanged (Fig. 7A). Expression of mDIC in white adipocytes has been shown to be affected by cold exposure (downregulation), as well as free fatty acid loading (upregulation). mDIC may play a role in fatty acid biosynthesis/lipogenesis and/or glyceroneogenesis, and its decrease with cold exposure probably reflects a decreased need for these two processes, whereas its increase with fatty acid loading may reflect an increase in lipogenesis (24). Failure of mDIC levels to decrease in caveolin-1 null BAT after fasting/cold treatment may reflect a general uncoupling of the signaling cascades involved in the regulation of lipolysis and lipogenesis, as we have previously characterized in WAT (12,13). In addition to mDIC, we also analyzed the expression levels of other mitochondrial proteins, i.e., prohibitin, which is involved in the processing and stabilization of newly transcribed mitochondrial membrane proteins (44), and Hsp60, a mitochondrial matrix marker protein, neither of which was found to be affected by a loss of caveolin-1.

Prohibitin levels remained statistically unaltered after fasting/cold treatment, whereas Hsp60 levels decreased nearly twofold (1.91 ± 0.21, P < 0.05) (Fig. 7A). Western analysis of UCP-1 in wild-type and caveolin-1 null BAT samples revealed that this protein is upregulated nearly 3.5-fold (3.43 ± 0.53, P < 0.05) after exposure to fasting/cold treatment and that loss of caveolin-1 did not alter either the basal or induced expression levels of this protein (Fig. 7A). Because activation of PKA is known to be involved in the upregulation of UCP-1 mRNA (45), and PKA activity is negatively regulated by an interaction with caveolin-1 (13), it is somewhat surprising that loss of caveolin-1 does not affect the expression of UCP-1. In view of our previous results suggesting an uncoupling of PKA activity and perilipin phosphorylation in caveolin-1 null WAT (13), it is possible that the localized activation of PKA, which would necessarily effect UCP-1 expression, may not be altered by loss of caveolin-1. That is, whereas the catalytic subunits of PKA, which would normally be sequestered and inactivated within caveolae, become hyperactivated in the absence of caveolin-1, those catalytic subunits that are normally associated with the regulatory subunits in other subcellular locations are not effected by the absence of caveolin-1.

In a further assessment of mitochondrial function and β-oxidation, we also analyzed the expression and phosphorylation of AMPK subunits. Interestingly, we found that although the phosphorylation of the AMPK-α catalytic subunit increases approximately fourfold (4.19 ± 0.67, P < 0.05) in wild-type animals after fasting/cold treatment, its activation increases only twofold (2.20 ± 0.19, P < 0.05) in caveolin-1 null mice (Fig. 7B). These alterations occurred without statistically significant changes in the expression of total AMPK-α. Because AMPK is normally activated in response to cellular stress (i.e., glucose deprivation) and functions to essentially turn on fatty acid oxidation and turn off fatty acid synthesis (46), these results suggest that a loss of caveolin-1 leads to decreased β-oxidation in BAT after fasting/cold treatment, which could thus help to explain the observed cold intolerance.

As an assessment of mitochondrial function, we next used two specific techniques that allow one to monitor the integrity of both the inner and outer mitochondrial membranes. Isolated mitochondria were obtained from the BAT of wild-type and caveolin-1 null animals in the basal state. The integrity of the inner membrane, determined via uptake of the fluorescent potential-sensitive dye JC-1, was found to be unchanged in caveolin-1 null mice (Fig. 7D). Similarly, the outer membrane integrity, which can be assayed by measuring cytochrome C oxidase activity in the presence and absence of detergent, was unchanged in caveolin-1 null BAT (Fig. 7C). Thus, although striking morphologic abnormalities are present in the mitochondria of caveolin-1 null animals, no drastic functional alterations in the membrane integrity of these organelles are apparent.

In summary, this study shows that despite the resistance to diet-induced obesity, no evidence exists for the hyperactivation of BAT and consumption of excess caloric intake via this route in caveolin-1 null mice. Instead, BAT appears relatively inactive because of an inhibition of lipolysis in caveolin-1 null mice, in accordance with our findings in WAT (13). In addition, dramatic alterations in BAT mitochondrial structure were observed.

It is interesting to note that the experimental conditions to which these mice were exposed are similar to those mice would be expected to routinely experience in their natural environment. That is, nondomesticated mice do not have constant access to food and are often exposed to cold temperatures, especially during the winter months. On a prolonged basis, it is expected that caveolin-1 null mice would not survive under these conditions and, as such, would not necessarily be able to survive outside of their carefully controlled experimental environment. Thus, caveolin-1 gene expression would likely be required for long-term survival in nature.

We thank Dr. Constantine Londos for donating the HSL antibodies. This work was supported by grants from the National Institutes of Health (NIH) and the Susan G. Komen Breast Cancer Foundation (to M.P.L) and by NIH grant R01-DK55758 and the American Diabetes Association (to P.E.S.). A.W.C. was supported by NIH Medical Scientist Training Grant T32-GM07288. M.P.L is the recipient of a Hirschl/Weil-Caulier career scientist award.