Obesity is associated with elevated levels of free fatty acids (FAs) and proinflammatory CD11c+ macrophages. However, whether and how free FAs contribute to CD11c+ macrophage differentiation and proinflammatory functions remain unclear. Here we report that dietary saturated FAs, but not unsaturated FAs, promoted the differentiation and function of CD11c+ macrophages. Specifically, we demonstrated that stearic acid (SA) significantly induced CD11c expression in monocytes through activation of the nuclear retinoid acid receptor. More importantly, cytosolic expression of epidermal FA binding protein (E-FABP) in monocytes/macrophages was shown to be critical to the mediation of the SA-induced effect. Depletion of E-FABP not only inhibited SA-induced CD11c upregulation in macrophages in vitro but also abrogated high-saturated-fat diet–induced skin lesions in obese mouse models in vivo. Altogether, our data demonstrate a novel mechanism by which saturated FAs promote obesity-associated inflammation through inducing E-FABP/retinoid acid receptor–mediated differentiation of CD11c+ macrophages.
In the past several decades, prevalence of obesity has increased dramatically in the United States and worldwide. It is estimated that more than half of American adults will be obese by 2030 (1). Accordingly, obesity-related illnesses, including cardiovascular disease, type 2 diabetes, and various cancers, will become serious threats to human health. It is accepted that chronic inflammation is the most common pathogenic signature underlying obesity and its related disorders. However, the mechanisms of how obesity contributes to chronic inflammation remain largely unknown.
Numerous animal and human studies have shown that macrophages (in particular CD11c+ macrophages) are recruited to adipose tissue during obesity, contributing to obesity-associated chronic inflammation and insulin resistance (2–4). Depletion of CD11c+ cells decreases inflammatory cytokine production and normalizes insulin sensitivity in obese mouse models (5). Our recent studies demonstrated that a high-fat diet (HFD) induces the accumulation of CD11c+ macrophages in the skin, promoting obesity-associated skin inflammation through the upregulation of inflammasome signaling (6). Thus, it is becoming clear that CD11c+ macrophages are a critical pathogenic cell population that mediates obesity-associated chronic inflammation. It is therefore of great interest to determine those factors that enhance the differentiation and proinflammatory function of CD11c+ macrophages during obesity.
Obesity is generally caused by overnutrition. Excess calories are primarily stored in the form of neutral lipids in tissues and in the circulation. Compared with healthy lean subjects, obese individuals have marked elevated levels of plasma free fatty acids (FFAs), which consist of different types of saturated and unsaturated fatty acids (FAs) (7). Palmitic acid (PA; C16:0), stearic acid (SA; 18:0), oleic acid (OA; 18:1), and linoleic acid (LA; 18:2) account for more than 80% of the total plasma FFAs (8). Elevated levels of plasma FFAs are an important contributor to insulin resistance, chronic inflammation, and other obesity-related disorders (9). However, the molecular mechanisms by which individual FAs contribute to obesity-associated inflammation and disorders remain incompletely understood.
FA binding proteins (FABPs) are a family of cytosolic lipid chaperones that solubilize FAs, facilitate their metabolism, and coordinate with various nuclear receptors for gene regulation (10, 11). FABPs exhibit tightly regulated patterns of tissue distribution. For example, adipose tissues predominantly express adipose FABP (A-FABP), whereas skin epithelia highly express epidermal FABP (E-FABP), suggesting unique functions of individual FABPs in specific tissues and cells (12). Our recent research focusing on functions of the FABP family in immune cell regulation demonstrate that A-FABP and E-FABP are also highly expressed in macrophages. Whereas E-FABP abundantly expressed in CD11c-positive macrophages promotes HFD-induced IL-1β–signaling pathways, A-FABP expression in CD11c-negative macrophages facilitates FA metabolism for cytotoxic ceramide production and contributes to saturated FA-induced macrophage cell death (6, 13). Thus, it appears that individual FABPs play a unique role in linking FA-mediated metabolism and proinflammatory functions in different subsets of macrophages.
In the current study, we performed experiments to investigate the role of individual FAs in regulating the differentiation and function of inflammatory CD11c+ macrophages. We demonstrated that saturated FAs (in particular SA), but not unsaturated FAs, greatly promoted CD11c+ macrophage differentiation and proinflammatory cytokine production both in vitro and in vivo. Furthermore, E-FABP was identified to be the major FABP member in mediating SA-induced activation of retinoid acid receptor (RAR) for the differentiation of CD11c+ macrophages. Importantly, deletion of E-FABP significantly reduced the accumulation of CD11c+ macrophages in obese mice and abrogated saturated FA-mediated chronic skin lesions in different HFD-induced obese mouse models.
Materials and Methods
C57BL/6 background E-FABP–deficient mice (E-FABP−/−), A-FABP–deficient mice (A-FABP−/−) and their wild type (WT) littermates were bred and housed in the animal facility of the University of Louisville. All animal manipulations were carried out according to the approval of the Institutional Animal Care and Use Committee of the University of Louisville. Weaned mice were ad libitum fed either with 60% HFD (lard) or 10% low-fat diet (LFD; lard) (Research Diets) for 5–9 mo before they were sacrificed for analysis of macrophage phenotype in various tissues or organs. To dissect the impact of different fat components in inducing skin inflammation, female mice were fed HFD rich in saturated FAs (45% cocoa butter), unsaturated FAs (45% safflower oil), or control LFD (10% soybean oil), respectively, for 9 mo. Skin lesions were measured, and CD11c+ macrophages in the skin tissues were analyzed by confocal or flow cytometric staining.
M-CSF and GM-CSF were purchased from Cell Signaling Technology (Danvers, MA). All FAs were from Nu-Chek Prep. PA (5 mM), SA (5 mM), OA (5 mM), LA (5 mM), eicosapentaenoic acid (5 mM), and α-linolenic acid (2 mM) were prepared with 2 mM of endotoxin-free BSA in PBS (catalog no. BP9705-100; Fisher Scientific), sonicated until dissolved, and filtered through 0.22-mm sterile filter as we previously described (13, 14). The specific reactive oxygen species (ROS) inhibitors 4-amino-2,4-pyrrolidine-dicarboxylic acid and butylated hydroxyanisole, PPARδ agonist GW0742 and antagonist GSK0660, PPARγ agonist rosiglitazone and antagonist GW9662, and retinoid X receptor (RXR) agonist LG100268 and antagonist HX531 were purchased from Sigma-Aldrich. RAR agonist BM753 and antagonist BMS195640 were purchased from Tocris Bioscience. Necrosis inhibitor IM54 and apoptosis inhibitor z-VAD-FMK were purchased from Enzo Life Sciences. TLR4 inhibitor eritoran, STAT3 inhibitor NSC74859, NFҡB inhibitor CAPE, IκB kinase inhibitor BMS-345541, ceramide synthesis inhibitor fumonisin B1 (FB1), and serine palmitoyltransferase (SPT) were purchased from Cayman Chemical.
Cell culture and treatment
For primary macrophage differentiation, bone marrow cells collected from WT, E-FABP−/−, or A-FABP−/− mice were differentiated either by M-CSF (20 ng/ml) or by GM-CSF (20 ng/ml). For some experiments, as in our previous studies (6, 13, 15), immortalized macrophage cell lines established from WT, A-FABP−/−, or E-FABP−/− mice were also used. All cultured cells were grown in RPMI 1640 supplemented with 5% (v/v) FBS (Atlanta Biologicals) and gentamicin (20 μg/ml). Macrophage cell lines (2 ×105 cells/ml per well) or primary bone marrow–derived macrophages (BMMs) (4 ×105 cells/ml per well) were plated in 24-well plates for all the experiments. After FA treatment for 1–3 d, depending on macrophage cell lines or primary BMMs, cells were lifted for flow cytometric analysis and quantitative PCR detection. In the assays including agonists, antagonists, or other inhibitors, cells were pretreated with respective chemicals for 3 h before addition of different concentrations (50, 100, 200 μM) of saturated or unsaturated FAs.
Flow cytometric analyses
For phenotypic analysis of CD11c+ macrophages in obese mice, different tissues, including the peripheral blood, lymph nodes (LNs), spleen, peritoneum, liver, lung and adipose tissues, bone marrow, and skin, were directly collected from lean and obese mice after PBS perfusion. In order to acquire single cells for flow analysis, liver, lung, and adipose tissue were cut into 1-mm3 fragments and digested with collagenase mixture (5% FBS in RMPI 1640, 0.5 mg/ml collagenase A [Roche Diagnostic], 0.2 mg/ml hyaluronidase type V [Sigma-Aldrich], and 0.02 mg/ml DNase I [Sigma-Aldrich]) on a shaker at 100 RPM at 37°C for 30 min (for liver) or 1 h (for other tissues) and then were filtered for use. Skin tissues were digested as in our previous studies (5).
Macrophages from different sources were surface-stained for 30 min at 4°C in 1% FBS–PBS containing indicated Abs (anti-CD11b, clone M1/70; anti-CD11c, clone HL3; anti-F4/80, clone BM8; anti–MHC class II (MHC II), clone M5/114.15.2; anti-CD36, clone HM36; anti-CD54, clone YN1/1.7.4; anti-CD80, clone 16-10A1; anti-CD86, clone GL-1). For C16-Bodipy uptake assay, cells were treated with C16-Bodipy (0, 10, 100 nM) for 30 min, then washed thoroughly with PBS before flow cytometric analysis. All samples were acquired on a BD Biosciences LSRFortessa flow cytometer, and data analyses were conducted using FlowJo software (Tree Star).
For quantitative PCR analysis, RNA was extracted from cells using the PureLink RNA Mini Kit (Ambion by Life Technologies). cDNA synthesis was performed with the QuantiTect Reverse Transcription Kit (Qiagen). Quantitative PCR using SYBR Green PCR Master Mix was performed on an ABI StepOnePlus Real-Time PCR System (Applied Biosystems) to analyze interesting genes, such as FABP family members, common nuclear receptors, FA transport protein 1 (FATP1), CD36, ACSL1, inflammatory cytokines, etc. (see the detailed gene primer sequence in Supplemental Table I). Relative mRNA levels were determined using HPRT1 as a reference gene. Gene expression levels were analyzed by the 2−∆∆Ct method.
Small interfering RNA transfection
Duplex small interfering RNAs (siRNAs) targeting the coding region of RXRα were ordered from Integrated DNA Technologies. To knock down RXR gene expression, 2 × 105 WT macrophage cell lines or 5 × 105 BMMs in 24-well plates were transfected with designated siRNAs (40 nM) using Oligofectamine (Life Technologies) for 4–6 h. Normal RPMI 1640 medium with 5% FBS was added to the transfected cells for 6 h before 100 μM SA treatment.
The fresh tissue of normal or lesion skin from obese WT or FABP−/− mice was embedded in OCT and stored in −80°C. The embedding blocks were sectioned by freezing microtome (CM1900; Leica) at 5-μm thickness. After blocking with anti-mouse CD16/32 (clone: 93, catalog no. 101302, 1:200 in 1% FBS–PBS; BioLegend), anti-mouse CD11c (clone: N418, catalog no. 117310, 1:200 dilution; BioLegend) and DAPI (catalog no. 4083S, 10 mg/ml in H2O, 1:2000 dilution; Cell Signaling) were applied to the sections for 1 h, followed by washing and sealing with antifading mounting medium. The sections were analyzed by Nikon A1 laser scanning confocal microscopy. The CD11c fluorescent intensity in the confocal images was analyzed by ImageJ software.
Western blotting was used to determine the levels of A-FABP and E-FABP in primary BMMs and the levels of RXRα in macrophages transfected with scramble or RXR siRNAs. In brief, 20-μg proteins (quantified by bicinchoninic acid assay) in each sample were loaded for SDS-PAGE, transferred to polyvinylidene difluoride membrane, and blotted with respective Abs (anti-mFABP5, catalog no. AF1476; anti-m/rFABP4 Ab, catalog no. AF1443; anti-RXRα, catalog no. 3085S). β-actin was probed as a loading control. The Image Quant TL system was applied to calculate the relative protein quantification.
Mouse IL-6, TNF-α, IL-1β, and IL-10 levels in cell cultural supernatants or serum from obese and/or lean mice were measured using ELISA MAX standard set from BioLegend.
The Student t test was performed for the comparison of results from different treatments. A p value < 0.05 is considered statistically significant.
Consumption of an HFD systemically enhances CD11c+ macrophages in vivo
We previously have shown that mice fed an HFD exhibited an increased percentage of CD11c+ macrophages in the skin (6), suggesting a possible systemic increase of proinflammatory CD11c+ macrophages during obesity. To address this hypothesis, we put C57BL/6 mice on either an HFD (60% fat) or an LFD (10% fat) for 20 wk to induce obesity and then analyzed the phenotype of CD11c+ macrophages in different tissues and organs. Because myeloid-derived immune cells express the CD11b marker, we divided CD11c+ macrophages (CD11c+F4/80+) into myeloid-derived CD11b+ and nonmyeloid-derived local-residential CD11b− populations (Fig. 1A). Interestingly, CD11b+CD11c+ macrophages were significantly increased in immune organs, including bone marrow, draining LNs, and peripheral blood, in HFD mice compared with LFD mice (Fig. 1B–D). In contrast, the percentage of CD11b−CD11c− macrophages was similar in these immune organs between the HFD and LFD mice, suggesting a specific upregulation of myeloid-derived CD11c+ macrophage differentiation in mice on the HFD. When we further analyzed the phenotype of CD11b+CD11c+ macrophages in nonimmune organs, such as adipose tissue, liver, and lung, we found that consumption of the HFD also elevated the percentage of CD11b+CD11c+ macrophages, but not CD11b−CD11c+ cells, in these organs (Fig. 1E–G). These data indicate that consumption of the HFD systemically promotes the differentiation of myeloid-derived CD11b+CD11c+ macrophages in obese mice.
Saturated FAs, in particular SA, enhance CD11c+ macrophage differentiation
In investigating why consumption of the HFD induces a systemic increase of myeloid-derived CD11c+ macrophage differentiation in vivo, we noticed that the HFD was rich in both saturated FAs (PA, 19.3% and SA, 10.6%) and unsaturated FAs (OA, 32.9% and LA, 24.4%). We speculated that individual FA components in the HFD might promote the differentiation of myeloid-derived CD11c+ macrophages. To this end, we differentiated bone marrow cells into macrophages under different conditions in the presence or absence of each FA in the diet. When bone marrow cells were differentiated with M-CSF, we observed that the percentage of CD11c+ macrophages was elevated in saturated FA–treated groups, particularly in the SA-treated wells (p < 0.01), compared with the BSA-treated control. In contrast, neither OA nor LA treatment increased CD11c expression in macrophages (Fig. 2A). Under the differentiating condition with GM-CSF, we observed the same phenomenon that saturated FAs, but not unsaturated FAs, induced CD11c expression in BMMs (Fig. 2B). Moreover, when bone marrow cells were cultured in vitro without any exogenous CSFs, we also observed enhanced expression of CD11c in CD11b+ macrophages in response to the treatment with PA or SA, although to a lesser extent (Fig. 2C). These results suggest that saturated FAs, in particular SA, are able to drive bone marrow cell differentiation into CD11c+ macrophages under different conditions.
Because of the distinct origins and functions of tissue macrophages (16), it was unclear whether macrophages in peripheral tissues responded similarly to FA treatment. We collected samples from the peripheral blood, spleen, and peritoneum and measured CD11c expression in monocytes and macrophages in response to individual FA treatment. Interestingly, similar to bone marrow cells, blood monocytes responded to the treatment of either PA or SA by enhancing CD11c+ differentiation (Fig. 2D). In contrast, FA treatment, regardless of saturated FAs or unsaturated FAs, exhibited a marginal impact on CD11c expression in macrophages from the spleen and peritoneum (Fig. 2E, 2F). These results suggest that, unlike tissue-resident macrophages, immature monocytes from the bone marrow or the circulation are more susceptible to SA-induced differentiation into CD11c+ macrophages. Altogether, our data indicate that saturated FAs, in particular SA, in the HFD may represent a previously unappreciated factor-enhancing differentiation of CD11c+ macrophages from the bone marrow and peripheral blood.
SA promotes inflammatory functions of CD11c+ macrophages
To further confirm the effect of SA in promoting CD11c expression, we treated M-CSF–induced bone marrow cells with different doses of SA ranging from 0 to 200 μM, and found that SA at 100 μM was enough to significantly promote CD11c protein expression on the cell surface (Fig. 3A, 3B). Moreover, the observation of a significant increase in mRNA levels of CD11c, but not CD11b, by SA treatment confirmed that SA induced a specific de novo expression of CD11c in BMMs (Fig. 3C, 3D). Because elevated CD11c+ macrophages contribute to chronic inflammation and adaptive immune responses in obesity (4, 17), we further examined if SA treatment promoted such functions. First, we assessed whether SA treatment influenced the expression of MHC II and other major costimulatory molecules (such as CD86, CD80, and CD54) that are critical for Ag presentation in CD11c+ macrophages. As shown in Fig. 3E–H, SA treatment enhanced the expression of all measured costimulatory molecules, in particular MHC II (p < 0.01), CD80, and CD86 (p < 0.05), compared with the BSA control. Next, we measured if SA treatment promoted proinflammatory cytokine production by M-CSF–induced BMMs (M-BMMs). Using real-time PCR, we demonstrated that SA treatment promoted the expression of IL-6, TNF-α, IL-1β, IFN-α, and IFN-β but had no impact on IL-10, in macrophages (Fig. 3J–O). We further confirmed that proinflammatory cytokines (such as IL-6, TNF-α, and IL-1β) were elevated in the supernatants of SA-treated macrophages by ELISA (Fig. 3P–S). Taken together, our data suggest that SA promotes not only CD11c+ macrophage differentiation but also its Ag presentation and proinflammatory functions.
SA-induced CD11c+ macrophage differentiation is dependent on RAR
We next sought to determine the molecular mechanisms by which SA induced CD11c expression in macrophages. M-BMMs exhibited a higher expression of F4/80 and a lower expression of CD11c compared with GM-CSF–induced BMMs (Supplemental Fig. 1A), representing bona fide hematopoietic-derived macrophages in vivo (13, 18). We thus focused on primary or immortalized M-BMMs to dissect the molecular mechanisms. Treatment of immortalized macrophages with individual saturated FAs (SA or PA), but not unsaturated FAs (including OA, LA, α-linolenic acid, and eicosapentaenoic acid), significantly induced a dose-dependent upregulation of CD11c expression (Fig. 4A, 4B). To dissect putative mediators or pathways that might be involved in the SA treatment (e.g., the SA/TLRs/NFκb pathway or SA/ceramides/cell stress pathway) (13, 19–21), we respectively inhibited the activation of TLR4 by eritoran, STAT3 by NSC74859, and NF-κB by CAPE or BMS-345541, and observed if such factors played a critical role in SA-induced CD11c expression by macrophages. However, none of these treatments showed any significant impact on SA-induced CD11c expression in macrophages (Supplemental Fig. 1B). We further inhibited ceramide production by myriocin or fumonisin B1, ROS generation by 4-amino-2,4-pyrrolidine-dicarboxylic acid or butylated hydroxyanisole, apoptosis by z-VAD-FMK, or necrosis by IM54 in macrophages; these treatments did not have an obvious effect on SA-induced CD11c expression either (Supplemental Fig. 1C).
Because FAs and their derivatives are able to regulate gene transcription through various nuclear receptors (22), we measured the common nuclear receptor expression profile in bone marrow cells differentiated with M-CSF. PPARβ, RAR, and their heterodimer partner RXR were highly expressed as compared with other nuclear receptors at different time points (Supplemental Fig. 1D). When macrophages were treated with specific agonists or antagonists for PPARβ or PPARγ, none of these treatments demonstrated obvious alterations in CD11c expression (Supplemental Fig. 1E). Interestingly, treatment with BMS753 (a selective RAR agonist) increased, whereas BMS195640 (a selective RAR antagonist) decreased, SA-induced CD11 expression (Fig. 4C, 4D), suggesting that RAR is a major mediator of SA-induced effect in macrophages. As RAR functioned as a heterodimer with RXR to regulate cell differentiation (23) and RXR was able to bind multiple sites on the CD11c promoter region (Supplemental Fig. 1F), macrophages were further treated with RXR-specific agonist LG100268 or antagonist HX531. LG100268 decreased, but HX531 increased, SA-induced CD11c expression in macrophages (Fig. 4E, 4F). Silencing of RXR expression with siRNA also enhanced SA-induced CD11c expression in the immortalized macrophages (Fig. 4G–I). We further used primary M-BMMs to validate these findings. Consistently, RAR antagonist inhibited, but agonist increased, SA-induced CD11c expression (Supplemental Fig. 1G). Inhibition of RXR with antagonist or siRNA enhanced, but activation of RXR with agonist suppressed, SA-mediated effects in primary BMMs (Supplemental Fig. 1H, 1I). Altogether, these results suggest that although RXR binds to a CD11c promoter serving as a gatekeeper to repress CD11c expression in resting macrophages, metabolism of excess SA coactivates RAR and/or RXR heterodimers leading to the transactivation of CD11c expression in macrophages.
E-FABP expression is critical to SA-mediated CD11c upregulation in macrophages
The importance of nuclear receptor RAR in controlling CD11c gene expression suggested that the trafficking of exogenous SA or its derivatives to the nucleus is a key step in SA-induced CD11c+ macrophage differentiation. To this end, we examined the expression of FABPs that mediate FA transport inside macrophages and found that E-FABP and A-FABP were the predominant members expressed in primary BMMs (Fig. 5A, 5B). Considering the fact that both A-FABP and E-FABP can bind SA (24), we next investigated which FABP(s) was (were) critical in facilitating SA-induced CD11c expression using WT, A-FABP−/−, and E-FABP−/− mice. With real-time PCR and Western blotting, we confirmed the phenotype of macrophages generated from WT, A-FABP−/−, and E-FABP−/− mice. As expected, A-FABP−/− and E-FABP−/− macrophages exhibited a compensatory up-expression of E-FABP and A-FABP, respectively (Fig. 5C–E). Importantly, when we measured CD11c expression in BMMs in the presence or absence of SA, we found that deficiency of E-FABP and A-FABP did not impact CD11c expression in macrophages without SA treatment. However, deficiency of E-FABP, but not A-FABP, significantly inhibited SA-induced CD11c expression in macrophages (Fig. 5F, 5G). Further analyses demonstrated that deficiency of both E- and A-FABP impacted neither the expression of membrane FA transport proteins, including CD36, long chain acyl-coA synthetase 1, and FATP1 (Fig. 5H–J), nor the uptake of long-chain FAs by these macrophages (Fig. 5K). On the basis of these results and our previous study showing that A-FABP facilitated SA-induced ceramide production (13), it seemed clear that deficiency of FABPs did not affect FA uptake by macrophages, but individual FABPs coordinated their unique lipid responses inside macrophages. Unlike A-FABP, E-FABP was critical in trafficking SA or its derivatives for coactivation of RAR-mediated CD11c expression in macrophages. Thus, E-FABP is the main cytosolic chaperone facilitating SA-induced differentiation of CD11c+ macrophages.
E-FABP deficiency inhibits saturated FA-induced CD11c+ macrophage differentiation in obese mice
Given our in vitro cellular studies, which showed that E-FABP expression was critical for SA-induced differentiation of CD11c+ macrophages, we reasoned that HFD-induced CD11c+ macrophage differentiation in different tissues would be compromised in the absence of E-FABP in obese mice in vivo. To this end, WT and E-FABP−/− mice were fed an HFD for 20 wk to induce obesity. E-FABP deficiency did not impact body weight increase in E-FABP−/− mice compared with WT mice, but when we measured CD11b+CD11c+ macrophages in the major organs of these seemingly normal mice, we found that obese E-FABP−/− mice exhibited reduced levels of CD11c+ macrophages in all analyzed tissues compared with obese WT mice, including the bone marrow, liver, lung, and skin (Fig. 6A–D). Moreover, serum levels of the proinflammatory cytokines IL-6 and IL-1β (but not TNF-α) were significantly lower in E-FABP−/− mice than in WT mice (Fig. 6E–G). We also analyzed the impact of A-FABP deficiency on CD11c+ macrophage differentiation and proinflammatory cytokine production in obese mice. Interestingly, consistent with the in vitro studies, A-FABP deficiency did not affect CD11c+ macrophage differentiation (Supplemental Fig. 2A–D), or influence cytokine levels (Supplemental Fig. 2E–G). Thus, E-FABP, but not A-FABP, plays a critical role in promoting HFD-induced CD11c+ macrophage differentiation and proinflammatory cytokine production in obese mice.
E-FABP deficiency protects mice from development of high saturated FA diet–induced skin lesions
We previously reported that mice fed an HFD (lard) developed skin lesions (6). As the HFD was rich in both saturated (31.8%) and unsaturated (61.9%) FAs, it was unclear which dietary FA components contributed to skin lesion formation. To this end, we fed mice an HFD rich in either saturated FAs (cocoa butter) or unsaturated FAs (safflower oil) (Supplemental Fig. 3A) and observed mouse body weight and skin lesion development. After 6 mo, both saturated and unsaturated HFDs significantly increased mouse body weight (from 18.32 ± 0.73 g at the beginning to 34.05 ± 2.54 g for the cocoa butter diet and from 19 ± 0.72 to 29.01 ± 1.25 g for the safflower diet). Although average weight of the cocoa butter group appeared higher than that of the safflower oil group, the weight change between the groups was not statistically different (p = 0.167). Of note, both saturated and unsaturated HFDs led to similar weight gain among E-FABP−/− and WT mice (Fig. 7A). After WT mice were fed either the cocoa butter or safflower oil HFD for 9 mo, 6 out of 16 mice (37.5%) in the cocoa butter group versus 1 out of 10 (10%) in the safflower oil group developed skin lesions (p < 0.01), suggesting that an HFD rich in saturated FAs, but not unsaturated FAs, is more proinflammatory. By contrast, none of the E-FABP−/− mice in either group developed skin lesions (n = 15/group) (Fig. 7B). Given that CD11c+ macrophages are the major pathogenic cells in HFD-induced skin lesions (6), we analyzed this population in skin tissue specimens from these mice. In the cocoa butter–fed group, there was a massive infiltration of CD11c+ macrophages in the skin of WT, but not E-FABP−/−, mice (Fig. 7C, 7D). Accordingly, the expression of MHC II, CD80, and CD86 on CD11c+ macrophages was significantly downregulated in E-FABP−/− mice (Fig. 7E–G). Levels of circulating IL-6 and IL-1β were increased in WT but not E-FABP−/− mice (Fig. 7H, 7I). Notably, TNF-α levels were very low and were not increased, suggesting that TNF-α does not drive this inflammatory process (Fig. 7J). As neither WT nor E-FABP−/− mice that consumed the safflower oil developed significant skin inflammation, it was not surprising that we did not observe obvious accumulation of CD11c+ macrophages in the skin of these mice (Supplemental Fig. 3B, 3C). In addition, there were no differences in the expression of costimulatory molecules on CD11c+ macrophages isolated from the skin of safflower oil–fed WT and E-FABP−/− mice (Supplemental Fig. 3D–F), suggesting that unsaturated FAs in the safflower oil diet are not the main factor in the induction of CD11c+ macrophage differentiation and function.
In another independent experiment comparing WT and A-FABP−/− mice on the same cocoa butter or safflower oil diets, we found that A-FABP deficiency did not affect body weight increases of mice on either diet (Supplemental Fig. 3G). In contrast, 40% (6 out of 15) of WT mice on the saturated FA diet, compared with 0% on the unsaturated FA diet, developed a skin lesion. Interestingly, A-FABP deficiency failed to protect mice against cocoa butter–induced skin lesions (7 of 14 mice exhibiting skin lesions in the cocoa butter diet). No mice on the unsaturated FA diet development a skin lesion (Supplemental Fig. 3H). Taken together, our data clearly showed that diets rich in saturated FAs promoted E-FABP–mediated differentiation of CD11c+ macrophages and skin inflammation.
Studies of both mice and humans have documented CD11c as a hallmark for proinflammatory macrophages in obesity (2, 4, 5, 25). However, the etiology of how CD11c+ macrophages are developed and activated during obesity remains unclear. Although new evidence suggests that lipid uptake and metabolism, rather than inflammatory cytokines, may determine the activation of inflammatory macrophages (26, 27), little is known about which species of lipids are critical or how they mechanistically induce proinflammatory macrophage activation. In this article, we reported that dietary FAs, in particular saturated FAs, induced CD11c+ macrophage differentiation in vitro, which contributed to the systematic increase of CD11c+ macrophages in HFD-induced obese mice. Mechanistically, SA metabolized by macrophages was channeled by the cytosolic lipid sensor E-FABP to activate the nuclear receptor RAR, thus inducing CD11c transcription. Importantly, deletion of E-FABP reduced the differentiation of CD11c+ macrophages and led to the complete abrogation of saturated FA diet–induced skin inflammation in E-FABP−/− mice.
It has long been recognized that obesity is associated with elevated levels of circulating FFAs, which contribute to chronic inflammation, insulin resistance, and other metabolic disturbances (7, 28, 29). Monocytes and macrophages function as professional phagocytes that sense the increase of FFAs through surface lipid receptors, such as CD36 and scavenger receptors, and attempt to clear them via enhanced uptake (30). Questions remain as to how macrophages respond to these overloaded lipids. Macrophages can be polarized in vitro into inflammatory and anti-inflammatory (tissue reparative) phenotypes historically designated as M1 and M2, respectively. However, it is now recognized that macrophages are exposed to a wide variety of environmental triggers in vivo; these strictly polarized phenotypes are rare (31). In the context of obesity, excess FAs or abnormal lipid metabolism play a role in the differentiation and activation of monocytes. For example, it has been shown that obesity induces lysosomal-dependent metabolism of lipids by CD11c+ macrophages independent of inflammatory function typical for classical activation (26). Likewise, studies of human adipose tissue macrophages reveal a profile of gene expression distinct from an M1 phenotype; mimicking the influence of obesity-associated metabolic syndrome by treating macrophages with glucose, insulin, and palmitate resulted in a metabolic activation distinct from classical activation (27). Clearly, the influences of various FAs on macrophage differentiation and function are likely to be diverse. In our previous studies, we have demonstrated that high levels of saturated FAs can induce different levels of macrophage cell death, depending on FA concentrations and macrophage origins (13). We were curious as to what happened to these FA-exposed but surviving populations. By gating of the live cells exposed to various FA stimulations, we found that myeloid-derived (CD11b+) monocytes could be driven to differentiate into CD11c+ macrophages, particularly following SA treatment in vitro. The number of CD11b+CD11c+ macrophages is enhanced in various tissues, including lipid-enriched (e.g., visceral fat) and unenriched tissues (e.g., lung), in HFD-induced obese mice in vivo. Thus, our studies suggest a new concept that excess circulating FFAs or abnormal lipid metabolism determine differentiation and activation of monocytes and macrophages during obesity-associated sterile inflammation.
Inside cells, water-insoluble FAs often form complexes with members of the FABP family (32, 33). FABPs exhibit tissue-specific distribution, whereby macrophages mainly express A-FABP and E-FABP (34, 35). Our recent studies further identified that A-FABP and E-FABP are not expressed equally among different subsets of macrophages. Whereas E-FABP is widely expressed in macrophages, with the highest levels in the CD11c+ subset, A-FABP is preferentially expressed in the CD11c-CD36+ subset (13, 36), suggesting unique roles of individual FABPs in macrophages. In dissecting which FABP members are critical to SA-induced CD11c transcription, we have revealed that E-FABP is responsible for facilitating SA-induced transcriptional activity. Given our previous study showing that A-FABP facilitates saturated FA-induced ceramide production, it is becoming clear that individual FABP members exhibit unique functions in determining the phenotype and differentiation of macrophages in response to environmental lipids.
Once bound to a specific ligand, FABPs are activated to deliver the ligand to the nucleus for assistance of transcriptional regulation. For example, L-FABP can deliver fibrate for transactivation of PPARα (37, 38), and A-FABP delivers troglitazone to enhance transcriptional activity of PPARγ (39). Similarly, E-FABP has been shown to transfer retinoic acids and saturated FAs for activation of RAR and PPARβ/δ in mammary tumor cells (40, 41). As BMMs highly expressed PPARβ/δ, RAR, and RXR, we reasoned that they might be involved in the SA-mediated effect in macrophages. When macrophages were treated with specific agonists or antagonists to PPARβ or PPARγ, none of the interventions appeared to be critical in saturated FA-induced CD11c upregulation. However, RAR activation by agonist and inhibition by antagonist significantly enhanced and inhibited SA-induced CD11c expression, respectively. Given the evidence that other SA-associated inflammatory and metabolic pathways, including TLR4, NF-κB, STAT3, ROS, and ceramide production, were not involved in regulation of CD11c expression, it is very likely that the E-FABP/RAR axis mediates SA-induced CD11c activation in macrophages. Of note, in resting macrophages, RXR may function as a checkpoint to repress CD11c expression, but when monocytes and/or macrophages are activated by excess SA (e.g., in the setting of obesity), E-FABP–facilitated SA metabolism will coactivate RAR and/or RXR for CD11c transactivation.
In general, unsaturated FAs are believed to be healthier to consume than saturated FAs, given that consumption of excess saturated FAs increases the risk of cardiovascular disease, type 2 diabetes, and some types of cancer (42, 43). However, the underlying mechanisms by which saturated FAs increase disease risk remain largely unknown. Our studies provide insight into how saturated FAs negatively affect macrophage biology in several ways. When saturated FAs are taken up by macrophages, they are transported by the cytoplasmic lipid chaperones A-FABP and E-FABP. On one hand, excess saturated FAs delivered by A-FABP can be metabolized for de novo production of ceramides, triggering intrinsic macrophage toxicity. On the other hand, E-FABP is able to mediate saturated FA-induced CD11c upregulation in macrophages. Unlike saturated FAs, unsaturated FAs are taken up and stored in the form of lipid droplets in macrophages, averting their metabolism for ceramide production and CD11c transactivation (13). The reasons why unsaturated FAs impose less toxicity to macrophages may be multifaceted: 1) uptake of unsaturated FAs increases macrophage membrane fluidity and neutral-lipid formation, which promotes lipid droplet formation; 2) binding of unsaturated FAs alters A-FABP structure for coordinating other lipid-mediated responses (44); and 3) double bonds in unsaturated FAs pose extra requirements for efficient use and metabolism of unsaturated FAs in macrophages. Therefore, distinct responses induced by saturated FAs and unsaturated FAs in macrophages may determine their different effects in vivo.
The pathogenic role of CD11c+ macrophages in obesity is well established (2, 5). Our previous studies demonstrated that consumption of diets rich in both saturated and unsaturated FAs induced CD11c+ macrophage–mediated skin inflammation (6). To further validate the different effects of saturated versus unsaturated FAs in vivo, we fed mice diets rich in either saturated (cocoa butter) or unsaturated (safflower oil) FAs and measured CD11c+ macrophage differentiation and skin inflammation. Consistent with our in vitro cellular studies, mice that consumed diets rich in saturated FAs exhibited a significantly higher percentage of CD11c+ macrophages and skin lesions as compared with mice fed the unsaturated FA diet. More strikingly, deletion of E-FABP, but not A-FABP, dramatically reduced the development of high saturated FA diet–induced skin lesions. Thus, the current study demonstrates that consumption of an HFD, in particular a high-saturated-fat diet, promotes skin lesions through inducing differentiation of proinflammatory function of CD11c+ macrophages. Recently, we identified specific E-FABP inhibitors through virtual screening and showed the therapeutic efficacy of this inhibitor in the suppression of inflammation in an autoimmune disease model (45). It will be of great interest to test if E-FABP inhibition reduces obesity-associated skin inflammation.
In summary, the current study identified a novel mechanism by which saturated FAs promote the differentiation of CD11c+ macrophages through the E-FABP/RAR axis. This finding not only provides a scientific explanation for harmful effects of consuming excess saturated FAs but also suggests E-FABP as a new target for the treatment of obesity-related chronic inflammation and other diseases.
We thank Dr. Hyeran Jang for assistance with design of the custom rodent diets made by Research Diets, Inc.
This work was supported by start-up funds from the University of Louisville and by Grants R01CA177679 and R01CA180986 from the National Cancer Institute (Bethesda, MD).
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived macrophage
FA binding protein
FA transport protein 1
free fatty acid
- MHC II
MHC class II
retinoid acid receptor
reactive oxygen species
retinoid X receptor
small interfering RNA
The authors have no financial conflicts of interest.