Tumor Necrosis Factor-α Induces Caspase-mediated Cleavage of Peroxisome Proliferator-activated Receptor γ in Adipocytes*

The nuclear receptor peroxisome proliferator-activated receptor γ (PPARγ) is a ligand-dependent transcription factor that acts as a primary regulator of adipogenesis and controls adipocyte metabolism and insulin action. Increased expression of tumor necrosis factor (TNFα) in adipose tissue of obese subjects potently suppresses the expression of PPARγ and attenuates adipocyte functions. Here we show that PPARγ is a substrate of caspase-3 and caspase-6 during TNFα receptor signaling in adipocytes, and the consequent PPARγ cleavage disrupts its nuclear localization. TNFα treatment of 3T3-L1 adipocytes decreases full-length PPARγ while increasing the level of a 45-kDa immunoreactive PPARγ fragment. Specific inhibitors of caspase-3 and caspase-6 attenuate the cleavage of PPARγ protein in response to TNFα in cultured adipocytes. Incubation of nuclear fractions with recombinant caspase-3 and caspase-6 also generates a 45-kDa PPARγ cleavage product. Dispersion of nuclear PPARγ to the cytoplasm in response to TNFα treatment occurs in parallel with detection of activated caspase-3. We suggest that activation of the caspase cascade by TNFα down-regulates PPARγ protein and PPARγ-mediated metabolic processes in adipose cells.

The development of insulin resistance in skeletal muscle of obese humans precedes and contributes to the onset of type 2 diabetes (1)(2)(3). This impaired responsiveness of muscle to insulin may result from high levels of circulating free fatty acids (FFAs) 2 that disrupt insulin signaling pathways in muscle and other tissues (1, 4 -9). Thus, the sequestration and storage of FFAs as triglycerides within adipose cells protects against the deleterious effect of circulating FFAs and, therefore, reduces insulin resistance in skeletal muscle. Adipose tissue also promotes insulin sensitivity in muscle by secreting adipokines including leptin and adiponectin, which promote fatty acid oxidation and decrease intracellular fatty acids (10,11).
A large body of work has identified transcriptional regulators that participate in the control of adipocyte differentiation as well its metabolic and secretory functions (12,13). The nuclear hormone receptor peroxisome proliferator-activated receptor ␥ (PPAR␥) is a master regulator of adipocyte differentiation and plays an important role in glucose and lipid metabolism as well as insulin sensitivity in mature adipocytes (12)(13)(14). The essential role of PPAR␥ in adipocyte gene expression and differentiation has been firmly established by a number of observations, including the high level of PPAR␥ expression in adipose tissue and the onset of its expression coincident with early stages of adipogenesis in culture (15). Coordinated expression and actions of PPAR␥ with other factors, such as C/EBP␣ and C/EBP␤, during fat cell differentiation has been extensively documented (16 -18). Additionally, it has been shown that activation of many adipocyte-specific genes occurs through binding of PPAR␥ to cis-acting promoter elements (19 -21). Ectopic expression of PPAR␥ in fibroblasts induces adipogenesis (22,23), whereas genetic ablation of the Ppar␥ gene disrupts adipogenesis, resulting in lipodystrophy, high levels of blood triglycerides and ectopic lipid deposition (24).
Overall, these observations are consistent with the notion that PPAR␥ is an essential transcriptional regulator of adipogenesis and is required for maintenance of adipocyte functions including FFA storage and lipid metabolism. Furthermore, the discovery that the insulin-sensitizing thiazolidinediones are high affinity ligands for PPAR␥ (25) suggested that the ability of thiazolidinediones to affect whole body glucose homeostasis is mediated at least in part through changes in PPAR␥-dependent adipose gene expression. In fact, amelioration of insulin resistance in obese mice treated with rosiglitazone is coincident with changes in adipocyte gene expression including up-regulation of many genes important for lipid metabolism and storage of triglycerides, as well as glucose disposal (26). Most of these genes are similarly affected by thiazolidinedione treatment of adipocytes in culture (26,27), suggesting that they reflect direct effects of PPAR␥-dependent gene regulation in adipocytes. Thus, modulators of PPAR␥ functions in adipocytes are highly likely to play roles in regulation of adipocyte metabolism and in determining responses to insulin sensitizers, such as thiazolidinediones. Regulation of PPAR␥ expression and function in fat cells also needs to be identified and characterized to fully understand adipocyte biology and its role in whole body glucose and lipid homeostasis.
Development of obesity coincides with substantial infiltration of macrophages into adipose tissue, which is associated with increased expression of inflammatory cytokines including tumor necrosis factor ␣ (TNF␣). Large amounts of TNF␣ are secreted by adipocytes and macrophages within adipose tissue of obese subjects (28). These agents, in turn, up-regulate adipocyte lipolysis and down-regulate triglyceride storage and insulin signaling, contributing to the metabolic complications of obesity such as impaired glucose tolerance and peripheral insulin resistance. Interestingly, TNF␣ is known to be a potent negative regulator of adipogenesis and PPAR␥ function (29,30). A marked down-regulation of PPAR␥ transcription in fat cells treated with TNF␣ has been demonstrated (29,30) and NFB signaling was proposed to mediate the TNF␣ inhibition of PPAR␥ expression in cultured cell systems (31).
Other signaling pathways triggered by TNF␣ may also influence PPAR␥ levels at the post-transcriptional level, as TNF␣-activated protein kinases may also directly target PPAR␥ protein, influencing PPAR␥ activity and/or stability by phosphorylation. For instance, it has been proposed that phosphorylation of PPAR␥ promotes its degradation via a ubiquitin/proteasome-dependent pathway (32). Moreover, PPAR␥ phosphorylation has been claimed to decrease ligand-binding affinity, controlling interaction with its co-regulators (33). Together, these observations suggest that regulation of PPAR␥ activity and stability also appears to be negatively regulated by kinase-mediated phosphorylation (34) and ubiquitination (32) followed by PPAR␥ protein degradation via a ubiquitin/proteasome-dependent pathway (32). Combined, these pathways might be used by TNF␣ receptor signaling to decrease PPAR␥ protein levels in adipocytes from chronically inflamed adipose tissue in obese subjects (1). Additional pathways elicited by TNF␣ signaling could also be operating to down-regulate PPAR␥ function in fat cells (reviewed in Ref. 1). For instance, it has been demonstrated that TNF␣-induced cleavage of insulin-signaling molecules, such as AKT, is mediated by a caspase-dependent pathway in adipocytes and seems to contribute to insulin resistance trigged by TNF␣ (35).
More recently, He and colleagues (36) demonstrated that caspase-1 mediates degradation of PPAR␥ proteins in adipocytes. The authors suggested that PPAR␥ levels could be modulated by TNF␣-stimulated caspase-1 activity in adipose cells (36). In this study, we sought to extend these observations and to determine whether activation of the caspase cascade by TNF␣ plays a role in the reductions of PPAR␥ protein levels induced by this cytokine, similar to its role in the degradation of other signaling molecules in 3T3-L1 adipocytes. Here, we present evidence that TNF␣ triggers a caspase-dependent cleavage of PPAR␥ proteins in cultured adipocytes, which disrupts its nuclear localization. Caspase-3, -6, and -8, but not caspase-1, -2, -5, -7, or -9, are required for TNF␣-elicited PPAR␥ cleavage in these cells. Based on these results, we postulate that a caspase-3/caspase-6-dependent pathway contributes to the mechanisms whereby TNF␣ suppresses PPAR␥ protein levels and its function in adipose cells.
Cell Culture and Treatments-3T3-L1 fibroblasts were grown and differentiated into adipocytes as described (37). The 3T3-L1 adipocytes were cultured in complete Dubecco's modified Eagle's medium (10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin). For TNF␣ stimulation, 7 days post-differentiated 3T3-L1 adipocytes were treated with the indicated concentration of TNF␣ for the appropriate time intervals. Cells were then washed with ice-cold phosphate-buffered saline lysed in 1% SDS-containing lysing buffer on ice.
Preparation of Nuclear and Cytosolic Subcellular Fractions-Nuclear and cytosolic cell extracts were prepared from 3T3-L1 adipocytes by a modification (38) of the procedure of Dignam et al. (39) and described in Ref. 40. Briefly, cell monolayers were rinsed twice with ice-cold phosphate-buffered saline and once in hypotonic lysis buffer containing 20 mM Tris-HCl, pH 7.5, 10 mM NaCl, 3 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, and 5 g/ml aprotinin. Cells were then harvested in hypotonic lysis buffer, incubated on ice for 5 min and homogenized with 16 strokes in a Dounce homogenizer by adding 0.1% Nonidet P-40 detergent. The resulting homogenate was centrifuged at 3500 ϫ g at 4°C for 5 min, and the supernatant was saved as cytosolic extract. The nuclear pellet was once resuspended in 0.5 volume of a hypotonic lysis buffer and centrifuged as before. The nuclear pellet was then resuspended in an extraction buffer containing 17.5 mM Hepes, pH 7.6, 330 mM NaCl, 1.1 M urea, 1.1% Nonidet P-40, 1 mM dithiothreitol, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, and 5 g/ml aprotinin. Nuclei were extracted for 30 min on ice. Finally, the sample was centrifuged at 13,000 ϫ g for 10 min at 4°C. The resulting nuclear extract and the previously obtained cytosolic extract were analyzed for protein content by BCA analysis (Pierce) according to the manufacturer's instructions and stored at 80°C.
Immunoblotting-3T3-L1 adipocytes were incubated without or with the indicated TNF␣ or inhibitor concentrations for the indicated time, and then harvested with lysis buffer containing 1% SDS. Equal amounts of protein from total cell lysates were resolved by SDS-PAGE and electrotransferred to nitrocellulose membranes, which were incubated with the indicated antibodies overnight at 4°C and then with horseradish peroxidase-linked secondary antibodies for 45 min at room temperature. Proteins were then detected with an enhanced chemiluminescence kit.
In Vitro Digestion with Recombinant Caspases-Aliquots of 40 g of nuclear fractions isolated from 3T3-L1 adipocytes were incubated for 12 h at 37°C in 50 l of phosphate-buffered saline in the presence of 10 units of the indicated recombinant caspase. The reaction was stopped by the addition of electro-phoresis sample buffer and samples were subsequently blotted for PPAR␥ protein.
Immunofluorescence-3T3-L1 adipocytes at day 7 were treated or not with TNF␣, washed three times with ice-cold phosphate-buffered saline, and immediately fixed in 4% formaldehyde for 12 min at room temperature. Fixed adipocytes were permeabilized with 0.05% Triton X-100 and 0.05% Tween 20 and then stained with mouse monoclonal anti-PPAR␥ (clone E-8) and rabbit polyclonal anti-cleaved (Asp 175 )-caspase-3 antibodies for 12 h, followed by labeling with Hoechst 33342 for nuclear staining, Alexa 594-conjugated goat anti-mouse and Alexa 488-conjugated goat anti-rabbit secondary antibodies (Molecular Probes) for 30 min. The cells were preserved in Vectashield mounting medium for fluorescence (Vector Laboratories Inc.). The Images were obtained using a Nikon TE2000-E2 microscope with a Yokogawa CSU10b spinning  disk confocal scan head and custom laser launch, NEOS AOTF and relay optics (Solamere Technology Group, Salt Lake City, UT). Multi-wavelength confocal z-series were acquired with a Nikon 40X Plan Apo oil objective (NA ϭ 1.0) and a QImaging Rolera MGi camera using the standard digitizer. The final pixel size was 0.27 m/pixel, the z step was 0.3 m, and the exposure time was 1-s per image. Metamorph Software controlled the microscope hardware and image acquisition.

TNF␣ Decreases PPAR␥ Protein Levels, while Increasing the Formation of a 45-kDa Immunoreactive PPAR␥ Fragment-It
has been previously shown that TNF␣ treatment of cultured adipocytes reduced PPAR␥ protein levels, presumably through suppression of Ppar␥ gene transcription via the NFB signaling pathway (30,31,41). However, it is not fully understood whether the decrease in PPAR␥ protein levels induced by TNF␣ is also mediated by other TNF␣-stimulated pathways in fat cells. Therefore, we conducted experiments to determine whether activation of TNF␣ receptors in 3T3-L1 adipocytes could lead to PPAR␥ protein degradation. Treatment of fully differentiated 3T3-L1 adipocytes (days 7-9) with TNF␣ for 16 h induced a dose-dependent decrease in PPAR␥ protein, with half-maximal and maximal effects occurring at 5 and Ϸ25-50 ng/ml, respectively (Fig. 1B). These concentrations of TNF␣ correspond to previously reported concentrations of TNF␣ that elicited maximal impairment of insulin signaling and inhibit PPAR␥ expression in 3T3-L1 adipocytes (42). Additionally, as depicted in Fig. 1A, treatment of cultured adipocytes with 50 ng/ml TNF␣ for 6 h caused a time-dependent 60% decrease in PPAR␥ protein. Immunoblotting with a mono-  clonal anti-PPAR␥ antibody raised against an epitope located at the C-terminal region of PPAR␥ protein also revealed a lower molecular mass band that migrated at ϳ45 kDa in SDS-PAGE (Fig. 1, A and B). Incubation with TNF␣ increased the level of this 45-kDa PPAR␥ fragment in a dose-and time-dependent manner (Fig. 1, A and B). Thus, whereas TNF␣ treatment of cultured adipocytes reduces total PPAR␥ protein levels, it simultaneously induces the formation of a lower molecular weight fragment of PPAR␥ protein.
Increase in the PPAR␥ Fragment Induced by TNF␣ Does Not Require New Protein Synthesis-A Ppar␥ mRNA, ␥ORF4, which is generated by alternative splicing has been reported recently by Sabatino and colleagues (43). The ␥ORF4 transcript results in the synthesis of a truncated PPAR␥ protein containing the first 273 amino acids of the mature PPAR␥ but lacking the entire ligand-binding domain. Thus, one possibility is that the 45-kDa PPAR␥ immunoreactive band is a translation product of an alternatively spliced Ppar␥ mRNA expressed upon TNF␣ stimulation. To test this possibility 3T3-L1 adipocytes were preincubated with the protein synthesis inhibitor cycloheximide (5 g/ml) and then treated for 6 h with 50 ng/ml TNF␣. Cells were then harvested and cell lysate samples immu-noblotted for PPAR␥ protein. Inhibition of protein synthesis with cycloheximide did not significantly block the ability of TNF␣ to induce the production of the 45-kDa PPAR␥ fragment (data not shown) suggesting that this product is a result of posttranslational truncation of existing PPAR␥ protein.
Reductions in PPAR␥ Protein Levels by TNF␣ Involve the Proteasome Degradation Pathway-We and other groups have shown that PPAR␥ degradation rates are remarkably rapid as measured in the presence of cycloheximide (t1 ⁄ 2 ϭ 2 h) (44,45). Furthermore, treatment of cultured adipocytes with the proteasome inhibitor MG132 increases the protein levels, suggesting that the proteasome degradation pathway regulates PPAR␥ protein levels in 3T3-L1 adipocytes (32). To test whether the 45-kDa fragment seen in cells stimulated with TNF␣ is a product of proteasomal degradation of PPAR␥ in 3T3-L1 adipocytes, we treated cells with TNF␣ plus proteasome inhibitors. As depicted in Fig. 2, a marked enhancement of PPAR␥ protein levels was observed in the presence of the proteasome inhibitors MG132 and lactacystin. Moreover, addition of the proteasome inhibitors partially prevented PPAR␥ protein degradation mediated by TNF␣ treatment of adipocytes for 3 h (Fig. 2). However, the formation of the 45-kDa PPAR␥ fragment was not prevented by these proteasome inhibitors, rather, accumulation of the fragment seemed to be enhanced by inhibition of proteasome function. Thus, these results suggest that suppression of the PPAR␥ protein level through TNF␣ may also involve proteasome-mediated protein degradation of this nuclear receptor. Generation of the 45-kDa fragment does not require activity of the proteasomal pathway, but ultimate degradation of the fragment may involve the proteasome.
TNF␣-induced Cleavage of PPAR␥ Is Caspase-dependent in Intact Cells-TNF␣ signaling leads to activation of caspase cascades in adipocytes and cleavage of several signaling molecules (35). To identify potential caspase cleavage sites in the PPAR␥ protein sequence, we utilized the web-based tool GraBCas (46). Several putative caspase consensus cleavage sites in the PPAR␥ protein sequence were identified as shown in Table 1. These observations prompted us to examine whether caspase activities are involved in PPAR␥ cleavage elicited by TNF␣ treatment of cultured adipocytes. To determine whether PPAR␥ was cleaved by caspases, 3T3-L1 adipocytes were preincubated with the broad-spectrum caspase inhibitor Boc-D-FMK and then stimulated with TNF␣ for up to 6 h (Fig. 3). Lysates from the harvested cells were then immunoblotted with anti-PPAR␥ monoclonal antibody. Consistent with the results depicted in Figs. 1 and 2, treatment with TNF␣ for 6 h reduced full-length PPAR␥ protein levels and induced formation of a 45-kDa fragment. However, the TNF␣-mediated decrease in PPAR␥ protein and appearance of the 45-kDa fragment was abrogated by this broad-spectrum caspase inhibitor (Fig. 3, right panels). This result strongly indicates that TNF␣ receptor signaling in fat cells induces cleavage of PPAR␥ by caspases and/or caspase-like proteases. They also suggest that caspase-mediated PPAR␥ proteolysis may be part of the mechanism whereby TNF␣ suppresses PPAR␥ expression levels in adipocytes.
Caspases-3, -6, and -8 Are Required for TNF␣-induced PPAR␥ Cleavage in Vivo-Because a general caspase inhibitor prevented the TNF␣-induced degradation and cleavage of PPAR␥ protein (Fig. 3), we next examined the involvement of particular caspases in this process. Thus, a similar experiment was performed with inhibitors of specific caspases. Adipocytes were treated with a panel of caspase inhibitors, then exposed to TNF␣ for 6 h and lysates were subjected to immunoblot analy- sis with anti-PPAR␥ antibody. Consistent with the results depicted in Fig. 3, the general caspase inhibitor Boc-D-FMK completely inhibited the decline of PPAR␥ protein and appearance of the 45-kDa PPAR␥ fragment (Fig. 4, upper panel). The caspase-8 inhibitor (Z-IETD-FMK) blocked both TNF␣-induced PPAR␥ degradation and 45-kDa PPAR␥ fragment formation to nearly the same extent as the broadspectrum caspase inhibitor Boc-D-FMK (Fig. 4). Among the remaining compounds tested, inhibitors of caspases-1, -2, -5, and -9 had essentially no effect. Only the caspase-6 (Z-VEID-FMK) and caspase-3 (Z-DEVD-FMK) inhibitors partially blocked the decline of PPAR␥ protein and the cleavage process. These results indicate that caspases-3, -6, and -8 activities are required for TNF␣-induced PPAR␥ cleavage in vivo. They (47) also suggest that TNF␣ activation of the caspase-8 3 caspase-3 3 caspase-6 cascade is necessary for down-regulation of the PPAR␥ protein in cultured adipocytes stimulated with this cytokine.
In Vitro Cleavage of PPAR␥ with Recombinant Caspase-3 and Caspase-6-The results depicted in Fig. 4, suggesting that caspase-3, -6, and -8 are required for PPAR␥ protein degradation induced by TNF␣, prompted us to investigate whether PPAR␥ proteins are directly targeted by the activity of these caspases. Nuclear fractions from 3T3-L1 adipocytes were isolated and incubated with recombinant caspases 1 (data not shown), and caspase-2, -3, -6, -7, -8, and -9, in the presence or absence of Boc-D-FMK. As depicted in Fig. 5, PPAR␥ proteins were cleaved in vitro only by recombinant caspase-3 and -6, as shown by the disappearance of the full-length PPAR␥ band and the generation of a 45-kDa PPAR␥ fragment. As expected, PPAR␥ protein cleavage by recombinant caspase-3 and -6 was abrogated in the presence of the general caspase inhibitor Boc-D-FMK. Interestingly, recombinant caspase-3 and -6 generated 45-kDa PPAR␥ fragments in vitro, similar to the 45-kDa fragment induced by TNF␣ in vivo (Figs. [1][2][3][4]. Treatment with recombinant caspase-8 was unable to cleave PPAR␥ in vitro although inhibition of caspase-8 blocked PPAR␥ protein cleavage in cells (Fig. 4). Taken together, these results shown that in vitro treatment of PPAR␥ with recombi- nant caspase-3 and -6 generated a 45-kDa cleavage product, consistent with that seen in adipocytes stimulated with TNF␣. They also reinforce the notion that PPAR␥ protein is directly targeted by caspase-3 and -6 during TNF␣ signaling in adipose cells in a pathway indirectly involving caspase-8.
Active Caspase-3 and Caspase-6 Are Translocated into Adipocyte Nuclei Upon TNF␣ Stimulation-Given the predominantly nuclear localization of PPAR␥ protein in adipocytes (44,48,49), we next examined whether TNF␣ treatment induces active caspases to be translocated into the nucleus where they could target PPAR␥ protein. Thus cytosolic and nuclear subcellular fractions from 3T3-L1 adipocytes, treated or not with TNF␣, were isolated and then subjected to immunoblot analysis with anti-procaspase and anti-active caspase antibodies to examine the subcellular distribution of active caspases in response to TNF␣. As shown in Fig. 6A, full-length caspase-8 was present in the cytosolic, but not nuclear fraction. Based on previous results, treatment with TNF␣ was expected to induce cleavage and activation of full-length caspase-8. Although the antibody used does not detect the activated cleaved caspase-8, we interpret the decrease in full-length caspase-8 protein levels in the cytosolic fractions (Fig. 6A, upper and lower panel) as indicative of cleavage and activation (50). This result suggests that stimulation of cultured adipocytes with TNF␣ induces activation of caspase-8 in the cytosolic compartment, but not the nucleus. We further examined whether other caspases downstream to caspase-8, such as caspase-3 and -6, predicted to be the activities that directly target PPAR␥, were also activated under these conditions and translocated to the nucleus. Thus, nuclear and cytosolic fractions from adipocytes stimulated or not with TNF␣ were immunobloted for pro-caspase-3 and active caspase-3. As depicted in Fig. 6B (lower panel), procaspase-3 was present mostly in the cytosolic fraction, with minimal detection in the nuclear fraction. Remarkably, treatment of 3T3-L1 adipocytes with TNF␣ for 6 h caused a dramatic increase in active caspase-3 (i.e. caspase-3-p17 subunit) associated with the nuclear subcellular fraction (Fig. 6B, upper  panel). Moreover, incubation of 3T3-L1 adipocytes with the general caspase inhibitor Boc-D-FMK before treatment of cells with TNF␣ blocked recruitment of active caspase-3 into the nucleus (Fig. 6B, upper panel), consistent with activation and translocation of caspase-3 being dependent on activity of an upstream caspase, most likely caspase-8. Similarly, stimulation of cultured adipocytes with TNF␣ also caused a significant recruitment of active caspase-6 into the nucleus (data not shown). Thus, these results are consistent with the hypothesis that recruitment of active caspase-3 and -6 into the nucleus of TNF␣-stimulated adipose cells mediates cleavage of the PPAR␥ protein.
TNF␣-mediated Caspase-3 and Caspase-6 Activation Disrupts Nuclear Localization of PPAR␥ in Cultured Adipocytes-We next investigated whether caspase-dependent cleavage of the N-terminal region of PPAR␥ upon TNF␣ stimulation affects the subcellular localization of PPAR␥. To examine whether caspase-dependent cleavage of PPAR␥ induced by TNF␣ has an impact on its subcellular distribution, we treated 3T3-L1 adipocytes with or without TNF␣ for 6 h, similar to the experiments described in Figs. 1-4. Adipocytes were then fixed and stained with anti-active caspase-3 and anti-PPAR␥ antibodies, followed by immunofluorescence microscopy analysis to visualize the subcellular distribution of endogenous PPAR␥ protein in cells displaying activation of caspase-3. As shown in Fig. 7, TNF␣ induced a remarkable dispersion of PPAR␥ protein from the nuclear region of 3T3-L1 adipocytes exclusively in cells in which active caspase-3 was also detected. In these cells PPAR␥ was detected throughout the cytosolic compartment. Nuclei remained intact under these conditions as evidenced by the persistence of nuclear staining (Fig. 7) as well as continued nuclear localization of the general transcription factor TATAbinding protein (data not shown), illustrating specific relocation of PPAR␥. Overall, this result supports the hypothesis that in TNF␣-stimulated adipose cells, the caspase-mediated cleavage of PPAR␥ protein at its N-terminal end promotes nuclear export and cytosolic degradation of this transcriptional factor. This export to the cytoplasm likely leads to loss of its activation function in the nucleus.

DISCUSSION
The inflammatory state associated with excess adipose tissue in obesity is characterized by the production of cytokines including TNF␣, which may adversely affect the capability of fat tissue to sequester excess lipids. This contributes to impaired insulin responsiveness of other tissues. We have described here a newly recognized mechanism by which TNF␣ may impair adipocyte function by decreasing the levels and activity of the key transcriptional activator PPAR␥. Our data support a model summarized in Fig. 8, in which TNF receptor signaling in adipocytes activates caspase-8 in the cytosol. Caspase-8 in turn promotes the cleavage and activation of caspases-3 and -6. The latter activated caspases are translocated to the adipocyte nucleus where they directly target PPAR␥ resulting in the cytosolic accumulation and ultimate degradation of a 45-kDa PPAR␥ cleavage product (Fig. 8). The size of this fragment as well as the predicted specificity of caspases for the cleavage site (Table 1) are consistent with cleavage at Asp 69 . This is likely to result in loss of PPAR␥ transcriptional activation by several Our results are largely consistent with a study published recently by He and colleagues (36). They report caspase-dependent cleavage of PPAR␥ resulting in the accumulation of a 45-kDa cleavage product in response to TNF␣. In that study, specific inhibition of caspase-1, but not caspase-3, -6, or -8, was able to prevent PPAR␥ cleavage in response to TNF␣. The reasons for this discrepancy between the results published by He et al. (36) and our data are not clear. However, in their report, caspase cleavage and production of the 45-kDa fragment was only observed under conditions of translation inhibition by cycloheximide in addition to TNF␣ treatment. In contrast, we observe PPAR␥ fragment generation irrespective of the presence of cycloheximide. Thus, it is possible that activation of distinct caspase pathways might result from these differing treatments. For instance, it has been established that cycloheximide sensitizes many types of cells to TNF␣-induced apoptosis, mainly due to its ability to block de novo synthesis of the endogenous caspase inhibitors, such as the cellular FLICEinhibitory protein (c-FLIP) and members of the inhibitors of apoptosis protein family. These proteins are up-regulated by TNF␣receptor via the NFB pathway and function through direct interactions to inhibit the activity of several caspases, including caspase-3, caspase-7, caspase-8, and caspase-9 (51,52). Hence, upon inhibition of protein synthesis by cycloheximide treatment, suppression in the levels of these caspase inhibitors could trigger a marked activation of different sets of caspases in adipocytes. In turn, highly active caspase activities due to the absence of their endogenous inhibitors could target the PPAR␥ protein in adipocytes.
Another possible explanation for the differences between the present results and those of He and colleagues (36) is the effect of cycloheximide on PPAR␥ protein turnover. Cycloheximide treatment reveals a remarkably rapid PPAR␥ degradation in adipocytes (44,45). It is possible that under these conditions of disruption in protein synthesis and rapid PPAR␥ protein turnover, this transcriptional factor could be targeted to a different set of caspases.
However, data presented here including inhibitor studies in cells as well as direct in vitro cleavage by recombinant caspases-3 and -6 strongly suggest that the caspase-3, -6, and -8 cascade is involved. In any case we concur with He et al. (36) that caspase-mediated cleavage of PPAR␥ modulates its levels and function in response to TNF␣, and it will be of interest to determine the precise pathway and its relevance in vivo.
TNF␣ has been implicated in suppressing adipose cell metabolism through down-regulation of PPAR␥ function. The present work adds to an emerging picture of this regulation occurring at multiple levels. Transcripitional regulation of PPAR␥ has been long recognized (reviewed in Ref. 53). Rapid turnover of adipocyte PPAR␥ as well as control at the translational level have more recently been described by us and others (44,45,54). Furthermore, regulation of PPAR␥ transcription activation activity through modulation of coactivator or corepressor activities has also been noted (53). Collectively these findings indicate a complex and multifactorial regulation of PPAR␥ in response to inflammatory cytokines like TNF␣. The extent of the contribution of these various pathways to impairment of adipocyte function and resulting dysfunction in lipid metabolism, insulin sensitivity, and glucose homeostasis is yet to be determined. The description here of caspase pathways involved in PPAR␥ down-regulation adds another potential link between obesity-associated inflammation and insulin resistance.
Acknowledgments-We thank Andrea C. Hubbard for excellent technical assistance at the initial stage of this work. We also thank Dr. Joseph V. Virbasius for valuable discussions. We also appreciate the technical help of Dr. Paul Furcinitti, from the Imaging Core Facility of the University of Massachusetts Medical School and the Diabetes and Endocrinology Research Center of the University of Massachusetts Medical School (supported by Grant DK325220). FIGURE 8. Model of caspase-dependent PPAR␥ protein cleavage. Cell-surface TNF␣ receptor signaling induces the activation of caspase-3, -6, and -8. In turn, activated caspases-3 and-6 are translocated to the adipocyte nucleus (green arrows) where they cleave PPAR␥ most likely at the identified consensus site TTVDF 69 (Table 1) to generate a COOH-terminal 45-kDa fragment. Cleaved PPAR␥ protein is translocated to the cytoplasm (dashed red arrow) and degraded via a proteasome-mediated process, thus down-regulating PPAR␥ functions. Moreover, following PPAR␥ protein cleavage, ubiquitinylation might also occur.