FABP4/aP2 Regulates Macrophage Redox Signaling and Inflammasome Activation via Control of UCP2

ABSTRACT

Obesity-linked metabolic disease is mechanistically associated with the accumulation of proinflammatory macrophages in adipose tissue, leading to increased reactive oxygen species (ROS) production and chronic low-grade inflammation. Previous work has demonstrated that deletion of the adipocyte fatty acid-binding protein (FABP4/aP2) uncouples obesity from inflammation via upregulation of the uncoupling protein 2 (UCP2). Here, we demonstrate that ablation of FABP4/aP2 regulates systemic redox capacity and reduces cellular protein sulfhydryl oxidation and, in particular, oxidation of mitochondrial protein cysteine residues. Coincident with the loss of FABP4/aP2 is the upregulation of the antioxidants superoxide dismutase (SOD2), catalase, methionine sulfoxide reductase A, and the 20S proteasome subunits PSMB5 and αβ. Reduced mitochondrial protein oxidation in FABP4/aP2^−/− macrophages attenuates the mitochondrial unfolded-protein response (mtUPR) as measured by expression of heat shock protein 60, Clp protease, and Lon peptidase 1. Consistent with a diminished mtUPR, FABP4/aP2^−/− macrophages exhibit reduced expression of cleaved caspase-1 and NLRP3. Secretion of interleukin 1β (IL-1β), in response to inflammasome activation, is ablated in FABP4/aP2^−/− macrophages, as well as in FABP4/aP2 inhibitor-treated cells, but partially rescued in FABP4/aP2-null macrophages when UCP2 is silenced. Collectively, these data offer a novel pathway whereby FABP4/aP2 regulates macrophage redox signaling and inflammasome activation via control of UCP2 expression.

INTRODUCTION

Obesity is linked to a variety of metabolic diseases, including type II diabetes, dyslipidemia, and cardiovascular disease. Accumulation of visceral adipose tissue (VAT) is critical for disease development, as it contains inflammatory macrophages that secrete cytokines, chemokines, and other signaling molecules, leading to both local and systemic effects (1, 2). Increased abundance of reactive oxygen species (ROS) is one critical consequence of adipose tissue inflammation, and hydrogen peroxide (H₂O₂) serves as a secondary messenger in various immunometabolic signaling processes (3, 4). H₂O₂ leads to reversible cysteine and methionine oxidation that alters protein activity and/or interactions (5, 6). Moreover, chronically elevated oxidative species diminish the cellular pool of chemical antioxidants (glutathione [GSH] and NADPH), leading to a variety of pathophysiologies (7–10). Sustained protein oxidation leads to not only dysregulated signaling cascades but also protein unfolding and potentially aggregation (11, 12). Multiple quality control systems exist in the endoplasmic reticulum (ER) and mitochondria to prevent oxidative damage and protein aggregation (4, 13–15). Despite this, the mitochondria in particular are vulnerable to sustained oxidative stress due to the high capacity for reactive oxygen species synthesis in the electron transport chain (16–18). Mitochondrial uncoupling proteins are one such mechanism to dissipate the proton gradient across the inner membrane and reduce oxidative stress (19).

In macrophages, one major consequence of mitochondrial dysfunction associated with metabolic disease is the activation of the inflammasome and secretion of interleukin 1β (IL-1β) (20). Secretion of IL-1β from macrophages typically involves multiple inputs and at a minimum requires activation of the NF-κB pathway, resulting in expression of the inflammasome complex, including procaspase-1, NACHT, LRR, and PYD domain-containing protein 3 (NLRP3) and pro-IL-1β. Secondary signaling (e.g., ROS and unfolded proteins) leads to the autocleavage of procaspase-1, followed by processing of pro-IL-1β into mature IL-1β and its subsequent secretion (21–23).

A variety of mouse models have been developed to interrogate the molecular relationship between obesity, inflammation, and metabolic disease. Of these, the FABP4/aP2 knockout mouse has been intensely studied as a paradigm for regulatory systems linking lipid metabolism to inflammation. In the FABP4/aP2 system, the loss of this intracellular fatty acid-binding protein from macrophages results in an anti-inflammatory phenotype in both animal- and cell-based models (3, 24–27). At the molecular level, loss of FABP4/aP2 attenuates NF-κB signaling, reduces c-Jun N-terminal kinase (JNK) phosphorylation, diminishes inflammatory cytokine secretion, and polarizes cells from the classically activated M1 phenotype to the alternatively activated M2 (28, 29). Recently, Xu et al. reported that ablation of FABP4/aP2 in macrophages improves mitochondrial function and attenuates NF-κB signaling via monounsaturated fatty acid induction of UCP2 and sirtuin-3 (3, 28). However, the molecular mechanisms that link the increased expression of UCP2 to diminished inflammatory cytokine synthesis are unclear. To that end, in this report, we describe novel findings that demonstrate that the FABP4/aP2-UCP2 axis regulates mitochondrial redox biology, the mitochondrial unfolded-protein response, and activation of the inflammasome.

RESULTS

Loss of FABP4/aP2 in macrophages induces UCP2 and reduces oxidative stress.

Prior work by Xu et al. has shown that loss of FABP4/aP2 results in elevated intracellular abundance of monounsaturated fatty acids (C_16:1 and C_18:1) and upregulates the expression of UCP2. Moreover, treatment of RAW 264.7 cells with exogenous palmitoleic or oleic acid but not palmitate or stearate increased the expression of UCP2, leading to a model in which monounsaturated fatty acids increase UCP2 expression (3). As shown in Fig. 1A, UCP2 protein was significantly increased in FABP4/aP2-null macrophages (AKO_Mϕ) compared to control macrophages from wild-type C57BL/6J mice (WT_Mϕ). Furthermore, macrophages that lack FABP4/aP2 either via genetic means (AKO cells) or by treatment with an FABP4/aP2 inhibitor (HTS01037) (26), exhibit significantly reduced hydrogen peroxide levels (Fig. 1B). AKO macrophages were also protected from the increase in hydrogen peroxide when treated with 100 ng/ml lipopolysaccharide (LPS) for 24 h (Fig. 1C). The reduction in hydrogen peroxide was lost when UCP2 was silenced in either AKO macrophages (AKO-UKD_Mϕ) or RAW 264.7 macrophages (Raw-UKD). In addition, treatment with HTS01037 could not lower hydrogen peroxide levels in Raw-UKD cells, as was observed in control RAW 264.7 macrophages (Raw-enhanced green fluorescent protein [eGFP]) (Fig. 1D and E). These data collectively indicate that UCP2 modulates the effect of reduced hydrogen peroxide in FABP4-null macrophages. This is consistent with previous studies showing that UCP2 can reduce ROS levels independently of proton uncoupling (19, 30).

FIG 1.

Loss of FABP4/aP2 decreases hydrogen peroxide in a UCP2-dependent manner. (A) UCP2 protein in WT and AKO macrophage lines (WT_Mϕ and AKO_Mϕ). (B) Hydrogen peroxide levels measured in WT macrophages with or without 4 h treatment with 30 μM HTS01037 compared to those in AKO macrophages. (C) Hydrogen peroxide levels in WT and AKO macrophages treated with 100 ng/ml LPS for 24 h or not treated. (D) Hydrogen peroxide in WT and AKO control cells (WT-eGFP_Mϕ and AKO-eGFP_Mϕ) and in UCP2-silenced AKO cells (AKO-UKD_Mϕ). (E) Hydrogen peroxide measured in RAW 264.7 control cells (Raw-eGFP) and UCP2-silenced RAW 264.7 cells (Raw-UKD) with or without treatment with 30 μM HTS01037 for 16 h. *, P < 0.05; **, P < 0.005; ***, P < 0.0005. The error bars indicate SEM.

Since the redox environments are different in cultured WT and AKO macrophages, the redox status of cysteine residues in different fractions of the visceral adipose tissue (VAT) of experimental animals was measured using the biotin switch assay. As shown in Fig. 2A, when WT mice were maintained on a high-saturated-fat diet for 12 weeks, the level of mitochondrial cysteine oxidation in the VAT was markedly greater than in mice maintained on a low-fat chow control diet. When such modification was evaluated in WT mice and compared to FABP4/aP2-null animals (maintained on a high-saturated-fat diet), mitochondrial cysteine oxidation levels were significantly blunted in the AKO mice, even though the mice were as obese as the wild-type controls (Fig. 2B). Moreover, the proteins from the whole-cell lysates of the VAT-derived stromal vascular fraction (SVF) enriched for macrophages also exhibited a reduction in cysteine oxidation from AKO mice compared to the control animals (WT_SVF and AKO_SVF) (Fig. 2C). Finally, when the macrophage cell lines were treated with 100 ng/ml LPS for 24 h, there was less cysteine oxidation in the whole-cell lysate of AKO macrophages, albeit to a lesser degree than observed in the tissue samples (Fig. 2D). In sum, these results indicate that the oxidation of mitochondrial cysteine residues is significantly increased due to a high-fat diet in wild-type mice, but this is protected against in mice lacking FABP4/aP2.

FIG 2.

Loss of FABP4/aP2 in macrophages decreases oxidative modification of cysteine residues. (A) Cysteine oxidation in the mitochondrial fraction of visceral adipose tissue from C57BL/6J mice on a chow or high-saturated-fat diet. (B) Cysteine oxidation in the mitochondrial fraction of visceral adipose tissue from WT and AKO C57BL/6J mice maintained on a high-saturated-fat diet. (C) Cysteine oxidation in the whole-cell lysate of the stromal vascular fraction of the visceral adipose tissue from wild-type and FABP4/aP2-null mice (WT_SVF and AKO_SVF). (D) Cysteine oxidation was measured in whole-cell lysates of WT and AKO macrophage lines following a 24-h treatment with 100 ng/ml LPS. The intensity of the entire lane for each sample was normalized to internal control ATP5α or β-actin to determine changes in cysteine oxidation, and the differences in all experiments were significant (a P value of <0.05 or better). The space between the blots in panel C is due to rearrangement of samples from the same membrane and exposure for presentation purposes.

Loss of FABP4/aP2 in the SVF of mice fed a high-fat diet leads to upregulation of antioxidants and 20S proteasome subunits.

Changes in the redox status of cells are frequently associated with correlative changes in the expression of proteins linked to redox biology, particularly antioxidants and proteins involved in quality control systems (9, 31–33). To address this, we evaluated several antioxidants in visceral adipose tissue-derived SVF in WT and AKO mice fed a high-fat diet. While transcript levels of catalase and superoxide dismutase 2 (Sod2) genes were not significantly different between WT and AKO cells (Fig. 3A), there was a significant increase in protein expression for both antioxidants in AKO animals (Fig. 3B and C). Furthermore, the 20S proteasome, which is responsible for recognizing and turning over oxidized proteins in an ATP-independent manner, was also upregulated in both the AKO SVF and the AKO macrophage line (Fig. 3B to D). Finally, in addition to the protection observed in cysteine oxidation, the SVF from AKO mice maintained on a high-saturated-fat diet revealed upregulation in methionine sulfoxide reductase A expression (Fig. 3E). These results demonstrate that loss of FABP4/aP2 reprograms the redox biology of macrophages toward a more reduced environment through two processes: reduced hydrogen peroxide production and upregulation of antioxidants and quality control systems.

FIG 3.

Expression of antioxidants and the 20S proteasome in stromal vascular cells (FABP4/aP2-null macrophages). (A to C) The SVF of visceral adipose tissue from WT and AKO mice fed a high-fat diet was analyzed for manganese Sod2 and catalase mRNA (A), manganese SOD2 protein (B), and catalase and the αβ subunit of the 20S proteasome (C). (D) AKO and wild-type macrophage lines were analyzed for the expression of PSMB5, the catalytic subunit of the 20S proteasome. (E) Methionine sulfoxide reductase (MsrA) levels in the SVF of WT and AKO mice fed a high-fat diet. The space between the blots in panels B to D is due to rearrangement of samples from the same membrane and exposure for presentation purposes. *, P < 0.05; **, P < 0.005. The error bars indicate SEM.

FABP4/aP2 regulates activation of the mitochondrial unfolded-protein response.

One of the consequences of heightened cellular oxidation and concomitant protein modification is activation of the unfolded-protein response (15, 34, 35). Indeed, oxidative modification of mitochondrial proteins often leads to misfolded or aggregated proteins and the increased expression of a cluster of proteases (ClpP and LonP1) designated for proteolysis and clearance of aberrant polypeptides (35). To test this in the FABP4/aP2 model system, mitochondrial unfolded-protein response (mtUPR) markers were evaluated in the adipose tissue-derived stromal vascular fraction isolated from wild-type and FABP4/aP2-null mice fed a high-fat diet. Consistent with the differences in mitochondrial oxidation status shown in Fig. 2, expression of the mtUPR markers in the AKO-derived stromal vascular fraction were almost completely ablated compared to that from wild-type mice (Fig. 4A). Extending this observation from the animal models to cell lines, the AKO macrophage line also had a significantly reduced mtUPR as measured by ClpP and LonP1 (Fig. 4B). This protection was also shown to be dependent on UCP2 expression, as silencing of UCP2 in the AKO macrophage line led to increased expression of LonP1, ClpP, and Hsp60 (Fig. 4C). Similar to the results in the SVF and the AKO cell line, ClpP and LonP1 are also significantly decreased in bone marrow-derived macrophages (BMDMs) isolated from FABP4/aP2-null mice compared to those from wild-type littermate controls (Fig. 4D). These results indicate that loss of FABP4/aP2 and the concomitant change in the redox environment attenuate the oxidation of mitochondrial proteins and the upregulation of the mitochondrial unfolded-protein response through expression of UCP2. Consistent with this, Xu et al. have reported that mitochondrial function, as measured by cellular respiration and responsiveness to LPS challenge, is improved in FABP4/aP2-null macrophages compared to mitochondria from wild-type cells (3).

FIG 4.

Expression of LonP1, Hsp60, and ClpP in stromal vascular cells, BMDMs, and macrophage lines from FABP4/aP2-null and control mice. (A) LonP1, Hsp60, and ClpP in stromal vascular cells from visceral adipose tissue of WT and AKO mice fed a high-fat diet. (B) LonP1 and ClpP in WT and AKO macrophage lines. (C) LonP1, Hsp60, and ClpP in AKO-eGFP (AKO-eGFP_Mϕ) control and AKO-UCP2 knockdown (AKO-UKD_Mϕ) macrophages. (D) Bone marrow-derived macrophages from WT and AKO mice were treated with 100 ng/ml LPS for 24 h, and LonP1 and ClpP protein levels were measured. *, P < 0.05; **, P < 0.005. The error bars indicate SEM.

Maintaining mitochondrial homeostasis through deletion or inhibition of FABP4/aP2 prevents inflammasome activation and IL-1β secretion.

Reducing ROS levels blunts the NF-κB pathway and stabilizes the NF-κB-inhibitory subunit, IκB-α (36, 37). As shown in Fig. 1, hydrogen peroxide is significantly reduced in AKO macrophages, prompting us to measure the protein level of IκB-α. As expected, the SVFs from AKO mice have significantly more IκB-α than those from WT mice (Fig. 5A). NF-κB regulates the expression of the inflammasome members, such as caspase-1, NLRP3, and IL-1β. Consistent with this, caspase-1 protein expression was significantly decreased in AKO macrophage lines and AKO BMDMs upon LPS stimulation compared to the respective WT controls (Fig. 5B to D). This was also observed when message levels of Nlrp3 were measured in macrophage lines and BMDMs. In both cell models, Nlrp3 was significantly decreased when FABP4/aP2 was genetically ablated and when it was pharmacologically inhibited using HTS01037 in BMDMs prior to LPS stimulation (Fig. 5E and F). This finding was corroborated at the protein level with a significant decrease in NLRP3 protein measured in the AKO cell line compared to the WT control (Fig. 5G). NF-κB also regulates the expression of the inflammasome substrate, IL-1β, and upon LPS stimulation, WT macrophages exhibited an increase in IL-1β transcript levels. On the other hand, AKO cells were protected from this increase in IL-1β message, but the protection was lost when UCP2 was silenced (Fig. 6A and B). The reduction in IL-1β was also observed in WT BMDMs pretreated with HTS01037 and AKO BMDMs upon LPS treatment (Fig. 6C).

FIG 5.

Genetic ablation of FABP4/aP2 upregulates IκB-α and reduces caspase-1 and NLRP3 expression. (A) Expression of IκB-α in stromal vascular cells of visceral adipose tissue from WT and AKO mice fed a high-fat diet. (B) Caspase-1 mRNA in WT and AKO macrophages treated with 100 ng/ml LPS for 24 h or not treated. (C and D) Caspase-1 protein expression in WT and AKO macrophage lines and BMDMs following 100 ng/ml LPS for 24 h. (E and F) Nlrp3 mRNA in WT and AKO macrophages treated with 100 ng/ml LPS for 24 h or not treated (E) and in WT and AKO BMDMs treated with 30 μM HTS01037 for 4 h or not treated with HTS01037, followed by treatment with 500 ng/ml LPS for 4 h or no LPS treatment (F). (G) NLRP3 protein expression in WT and AKO macrophages. *, P < 0.05; **, P < 0.005. The error bars indicate SEM.

FIG 6.

Decrease in transcript levels of IL-1β in AKO macrophages is dependent on expression of UCP2. (A) IL-1β mRNA measured in WT and AKO macrophages with or without 100 ng/ml LPS for 24 h. (B) IL-1β mRNA measured in AKO-eGFP control and AKO-UKD macrophages under the same conditions as for panel A. (C) IL-1β mRNA measured in WT and AKO BMDMs treated with 30 μM HTS01037 for 4 h or not treated with HTS01037, followed by 500 ng/ml LPS for 4 h. **, P < 0.005; ***, P < 0.0005. The error bars indicate SEM.

The bioactive form of IL-1β must be generated by inflammasome-dependent cleavage and secretion, which can be triggered by ROS (as well as other sources) through the mitochondrial intrinsic apoptotic pathway (22, 38). To assess activation of the inflammasome in the FABP4/aP2 system, IL-1β secretion was evaluated in response to either pharmacologic or genetic ablation of FABP4/aP2. Secretion of IL-1β was completely absent in FABP4/aP2-null macrophages and wild-type macrophages treated with HTS01037 compared to wild-type controls when treated with LPS and ATP (Fig. 7A). Furthermore, inflammasome activation was shown to be partially activated by the mitochondrial intrinsic apoptotic pathway, as observed when WT macrophages were treated with cyclosporine (CsA) in addition to LPS and ATP stimulation. As shown in Fig. 7B, CsA could blunt the secretion of IL-1β. Conversely, IL-1β secretion could be induced in AKO macrophages by inhibiting the proteasome with MG132 or by silencing UCP2 (Fig. 7C and D), indicating that AKO cells maintain some capacity for inflammasome activation through UCP2 expression and protein unfolding. Both mitochondrion-induced apoptosis and protein unfolding have been shown to activate the inflammasome and may explain the upregulation of the 20S proteasome observed in AKO cells (21, 22, 39, 40). Consistent with the immortalized cell lines, bone marrow-derived macrophages from AKO or WT mice pretreated with HTS01037 exhibited a significant decrease in IL-1β secretion upon stimulation with LPS and ATP (Fig. 7E).

FIG 7.

Secretion of IL-1β is reduced under pharmacologic and genetic ablation of FABP4/aP2. (A) IL-1β secretion with or without 4 h treatment with 30 μM HTS01037, followed by 500 ng/ml LPS for 4 h and by 2 mM ATP for 1 h. (B) IL-1β secretion in WT macrophages in the presence of 2 μM or 5 μM cyclosporine for 1 h, followed by LPS and ATP treatment as described for panel A. (C) AKO macrophages were treated with LPS and ATP alone as for panel A or with the addition of 10 μM MG132 for 2 h after LPS treatment, and IL-1β secretion was measured. (D) IL-1β secretion was measured after LPS treatment with or without ATP as for panel A and compared between AKO-eGFP control and AKO-UKD macrophages. (E) IL-1β secretion was measured in BMDMs from WT and AKO mice treated for 4 h with 500 ng/ml LPS (with or without a 4-h pretreatment with 30 μM HTS01037), followed by 2 mM ATP for 1 h. *, P < 0.05; ***, P < 0.0005. The error bars indicate SEM.

DISCUSSION

Loss of the adipocyte fatty acid-binding protein has been shown to be protective against the low-grade chronic inflammation that leads to metabolic dysfunction under conditions of obesity (24, 27). This is largely due to the phenotypic switch of adipose tissue macrophages from an inflammatory to an anti-inflammatory state with concomitant changes in cytokine and chemokine secretion (25, 26). While the major physiological outcomes have been described, the molecular mechanisms that underlie such regulation have remained enigmatic. Here, we show for the first time the major finding that macrophage redox biology is regulated by the FABP4/aP2-UCP2 axis, resulting in changes in protein oxidation and preventing the induction of the mitochondrial unfolded-protein response and inflammasome activation (Fig. 8).

FIG 8.

Schematic representation of the FABP4/aP2-UCP2 axis and subsequent regulation of the inflammasome. Genetic or pharmacologic inhibition of FABP4/aP2 increases the level of intracellular free fatty acids (e.g., palmitoleic acid). This in turn leads to increased expression and activation of UCP2, which decreases H₂O₂ levels. Lowering H₂O₂ attenuates protein oxidation and activation of the mtUPR. ER-mitochondrion communication activates the inflammasome and subsequent IL-1β secretion.

We have previously shown that loss of FABP4/aP2 leads to upregulation of UCP2 and Sirt3 in macrophages (3, 28). We now observe that AKO macrophages are protected from LPS-induced hydrogen peroxide production and that UCP2 regulates this process. The difference in hydrogen peroxide ultimately protects AKO cells from hyperoxidation of cysteine (and likely methionine) residues following 12 weeks on a high-fat diet. This was observed in the mitochondria and stromal vascular fractions of VAT isolated from AKO mice compared to that from WT mice and further confirmed in immortalized macrophage lines following LPS stimulation. Interestingly, the difference in cysteine oxidation was less dramatic in the cell lines than in the tissue samples, potentially due to different signaling nodes being activated under conditions of obesity as opposed to a single treatment with LPS. Moreover, protein levels of inflammasome complex members and mtUPR markers in Sirt3-silenced macrophages are not different from those in the wild-type cells. This suggests that changes in the mtUPR and inflammasome activation are likely to be independent of Sirt3 but are downstream of UCP2 (results not shown).

Because UCP2 is a mitochondrial protein that reduces oxidative stress and mitochondrial protein oxidation, we focused on downstream effects potentially mediated by the mitochondrial unfolded-protein response. Indeed, LonP1, ClpP, and Hsp60 were markedly reduced in the AKO SVF, BMDMs, and macrophage lines from AKO mice compared to controls. This protection was also shown to be dependent on the expression of UCP2, as the mitochondrial proteases were increased in UCP2-silenced cells compared to eGFP controls in the AKO background. It is also interesting that the most robust changes observed in the different cell lines consistently came from LonP1, which specifically recognizes oxidized proteins and targets them for degradation (35). Compromising homeostasis of the mitochondrion can lead to apoptosis, and one of the primary downstream effectors of the mtUPR in immune cells is activation on the inflammasome (22, 38, 41, 42). This was prevented in cells lacking FABP4/aP2, as observed from the decrease in mRNA and protein levels of the inflammasome members, as well as its activation (IL-1β secretion). The activation of the inflammasome in WT macrophages could be partially rescued with CsA, an inhibitor of the mitochondrial intrinsic apoptosis pathway, and conversely, IL-1β secretion could be induced in AKO cells by inhibiting the proteasome or by silencing UCP2 (22). This indicates that both mitochondrial integrity and protein folding, potentially through the redox status, are maintained through the FABP4/aP2-UCP2 axis. Furthermore, this system was also tested in BMDMs, and while there was a detectable amount of IL-1β from the AKO and WT HTS01037-treated cells, both conditions exhibited significant decreases compared to WT BMDMs following LPS and ATP stimulation (Fig. 7).

ER-mitochondrial communication is a dynamic process driven by dedicated structural domains within mitochondrion-associated membranes ((43–45). Previous work by Xu et al. (3) has shown that ER stress is blunted in FABP4/aP2-ablated macrophages due to increased expression of UCP2. Paralleling this, Bronner et al. demonstrated that ER stress leads to mitochondrial ROS production, which in turn promotes inflammasome-dependent IL-1β secretion (16). These observations suggest that in macrophages, inflammasome activation may be potentiated by either ER stress or the mtUPR and that organelle communication may facilitate signaling. The FABP4/aP2-UCP2 axis is upstream of these events and implicates redox balance as being pivotal in initiating the inflammatory response (Fig. 8).

The major regulator of IL-1β expression is mediated through the NF-κB pathway, and its activation is upregulated through hydrogen peroxide signaling (46, 47). Indeed, hydrogen peroxide signaling is essential for full activation of inflammatory macrophages, and blunting such processes inhibits NF-κB activation (36, 46, 48, 49). Consistent with this, Fig. 5 shows that IκB-α, the major inhibitor of NF-κB, was upregulated in the SVF of AKO mice. The mechanism for this upregulation is unresolved but may be due to enhanced stability from a less oxidizing environment (36, 37). In support of IκB-α upregulation in AKO SVF, NF-κB targets, NLRP3 and caspase-1, were decreased in AKO macrophages. IL-1β message levels were also decreased in AKO macrophages, and in agreement with other studies, were shown to be dependent on UCP2 expression (30). Another transcription factor that may play a role in the altered redox status of FABP4/aP2-null cells is nuclear factor (erythroid-derived 2)-like 2 (Nrf-2). It is the master regulator of antioxidants, such as SOD2 and catalase, as well as the 20S proteasome and UCP2 (40, 50, 51). While the antioxidants and UCP2 are able to decrease hydrogen peroxide, the 20S proteasome is uniquely able to recognize and degrade oxidized proteins to prevent subsequent unfolding (39). Determining the localization and activation of both NF-κB and Nrf-2 will be an essential next step in understanding the role of the FABP4/aP2-UCP2 axis in controlling inflammasome activation in response to obesity.

MATERIALS AND METHODS

Mice.

Male C57BL/6J WT and AKO mice were fed ad libitum a high-saturated-fat (lard) diet (F3282; BioServe, Flemington, NJ) for 12 weeks after weaning. The mice were used between 15 and 16 weeks of age and weighed 40 to 50 g. Such mice fed a high-fat diet developed insulin resistance, as measured by impaired insulin and glucose tolerance, hyperinsulinemia, increased hepatic glucose output, and elevated lipolysis (52). All experimental procedures using animals were reviewed and approved by the University of Minnesota Institutional Animal Care and Use Committee.

Cell culture.

A variety of stable cell lines linked to FABP4/aP2 biology reported here were immortalized from bone marrow-derived macrophages of chow-fed WT and AKO C57BL/6J mice, as previously reported (25). They included stable macrophages from wild-type C57BL/6J mice (WT_Mϕ) and FABP4/aP2-null mice (AKO_Mϕ), wild-type or AKO macrophages expressing eGFP (WT-eGFP_Mϕ and AKO-eGFP_Mϕ), and AKO macrophages in which UCP2 had been silenced (AKO-UKD_Mϕ). All the above-mentioned macrophages were maintained in RPMI 1640 (Invitrogen, Carlsbad, CA) with 5% fetal bovine serum (FBS). RAW 264.7 control and UCP2 knockdown (Raw-eGFP and Raw-UKD) cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (3). BMDMs (2 × 10⁶ cells) were isolated from C57BL/6J mice, plated, and maintained in Iscove's modified Dulbecco's medium (Invitrogen, Carlsbad, CA), 10% FBS, and 10 ng/ml macrophage colony-stimulating factor (M-CSF) for 1 week prior to stimulation (53).

Quantitative RT-PCR.

Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA) from WT and AKO macrophage lines. cDNA synthesis was performed by using iScript (Bio-Rad, Hercules, CA) according to the manufacturer's protocol. Quantitative reverse transcription (qRT)-PCR amplification utilized a Bio-Rad CFX 96 real-time system with SYBR green Supermix. Transcription factor II E (TFIIE) and TATA-binding protein (TBP) were used as internal controls to normalize expression. The primer sequences can be found in Table 1.

TABLE 1.

Sequences of primers used for qRT-PCR

Target	Primer sequence (5′–3′)
	Forward	Reverse
IL-1β	AAATACCTGTGGCCTTGGGC	CTTGGGATCCACACTCTCCAG
Catalase	CCAGCGACCAGATGAAGCAG	CCACTCTCTCAGGAATCCGC
SOD2	GCCCTGGAACCTCACATC	TGACCACCACCATTGAACTT
Nlrp3	GCTCCAACCATTCTCTGACC	AAGTAAGGCCGGAATTCACC
Caspase-1	GGGACCCTCAAGTTTTGCC	GACGTGTACGAGTGGTTGTATT
TBP	ACCCTTCACCAATGACTCCTATG	ATGACTGCAGCAAATCGCTTGG
TFIIE	CAAGGCTTTAGGGGACCAGATAC	CATCCATTGACTCCACAGTGACAC

Stromal vascular fraction isolation.

Isolation of the stromal vascular fraction was performed as described by Xu et al. (3). Visceral fat pads were dissected from wild-type and FABP4/aP2 knockout (WT_SVF and AKO_SVF) mice, minced, and digested in Krebs-Ringer-HEPES (KRH) buffer supplemented with type I collagenase (Worthington, Lakewood, NJ) and bovine serum albumin (BSA) for 1 h at 37°C. Undigested tissue was removed by filtering the mixture through a 100-μm-pore-size nylon cell strainer (Fisher Scientific, Waltham, MA). The SVF was collected by centrifugation at 500 × g for 10 min, followed by two washes with KRH buffer. The newly isolated SVF was then resuspended either in TRIzol reagent for RNA isolation or in cell lysis buffer supplemented with protease inhibitors for protein assays or precipitated with 20% trichloroacetic acid (TCA) for cysteine oxidation detection.

Mitochondrial isolation.

Mitochondrial isolation was carried out as described by Xu et al. (3). Tissue was placed in ice-cold mitochondrial isolation buffer (20 mM Tris, pH 7.4, 220 mM mannitol, 70 mM sucrose, 1 mM EDTA, 0.1 mM EGTA) supplemented with protease inhibitors. The tissue was then lysed with 7 strokes of a Dounce homogenizer, and the homogenates were centrifuged at 900 × g for 10 min to remove nuclei and unbroken cells. To pellet the mitochondria, the supernatant was centrifuged at 10,000 × g for 15 min at 4°C.

Hydrogen peroxide measurements.

An Amplex Red hydrogen peroxide/peroxidase assay kit (Invitrogen, Carlsbad, CA) was used to measure hydrogen peroxide, following the manufacturer's instructions.

Immunoblot analysis.

Cells were lysed with radioimmunoprecipitation assay (RIPA) buffer supplemented with protease inhibitors (Calbiochem, Darmstadt, Germany). Twenty-five to 50 μg of protein from each sample was separated by SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane, as measured by a bicinchoninic acid (BCA) assay. After blocking with Odyssey blocking buffer (Li-Cor Biosciences, Lincoln, NE), the membranes were incubated with primary antibody overnight at 4°C. The membranes were washed and incubated with secondary antibody conjugated to Li-Cor IRDye for 1 h and visualized using Odyssey infrared imaging (Li-Cor Biosciences, Lincoln, NE). The primary antibodies used were anticatalase (Abcam, Cambridge, MA), IRDye 800CW-streptavidin (cysteine oxidation; Li-Cor, Lincoln, NE), anti-PSMB5 (Abcam, Cambridge, MA), anti-20S αβ (Abcam, Cambridge, MA), anti-NLRP3 (Adipogen, San Diego, CA), anti-caspase-1 (Abcam, Cambridge, MA), anti-methionine sulfoxide reductase A (Abcam, Cambridge, MA), anti-β-actin (Sigma-Aldrich, St. Louis, MO), anti-superoxide dismutase 2 (Cell Signaling, Beverly, MA), anti-LonP1 (Abcam, Cambridge, MA), anti-ClpP (Abcam, Cambridge, MA), and anti-HSP60 (Abcam, Cambridge, MA).

Detecting global cysteine oxidation.

Wild-type and AKO SVF, mitochondrial isolates, or macrophage line lysates were directly precipitated with 20% TCA for 2 h at 4°C. The pellets were subsequently washed with 10% TCA, 5% TCA, and 75% ice-cold acetone and then resuspended in labeling buffer (200 mM Tris, pH 8.5, 5 mM EDTA, 0.05% SDS, and 6 M urea). The lysates were labeled with 100 mM iodoacetamide for 15 min at room temperature, followed by addition of 100% TCA (20% final concentration), and maintained at 4°C overnight. The cell pellets were washed again and resuspended in reducing buffer (40 mM sodium arsenite, 150 mM Tris, pH 7.4, 2.5% SDS), along with 200 μM biotin-maleimide, and incubated for 1 h at 50°C. The protein concentration of each sample was measured with a BCA assay, and the proteins were separated by SDS-PAGE. The membrane was placed in blocking buffer overnight at 4°C and incubated for 1 h in IRDye 800CW-streptavidin to detect the cysteine-conjugated biotin residues (cysteine oxidation).

ELISA analysis.

IL-1β was detected in the media of macrophage lines or bone marrow-derived macrophages treated with either 30 μM HTS01037 for 4 h, 2 μM or 5 μM cyclosporine for 1 h, 500 ng/ml LPS for 4 h, 10 μM MG132 for 2 h, or 2 mM ATP for 1 h using an enzyme-linked immunosorbent assay (ELISA) kit (BD Biosciences, San Jose, CA).

Statistical analysis.

All the results are expressed with standard errors of the mean (SEM). For studies using the stable wild-type and AKO cell lines, the results are presented for a sample size of three or more based on at least three repeated independent experiments. For studies using cells derived from the mouse stromal vascular fraction or bone marrow-derived macrophages, results are reported for experiments performed at least 3 times with a sample size of 4 to 6 per experiment. Statistical significance was determined using an unpaired two-tailed Student t test.