Human monocyte-derived dendritic cells turn into foamy dendritic cells with IL-17A.

Interleukin 17A (IL-17A) is a proinflammatory cytokine involved in the pathogenesis of chronic inflammatory diseases. In the field of immunometabolism, we have studied the impact of IL-17A on the lipid metabolism of human in vitro-generated monocyte-derived dendritic cells (DCs). Microarrays and lipidomic analysis revealed an intense remodeling of lipid metabolism induced by IL-17A in DCs. IL-17A increased 2–12 times the amounts of phospholipids, cholesterol, triglycerides, and cholesteryl esters in DCs. Palmitic (16:0), stearic (18:0), and oleic (18:ln-9c) acid were the main fatty acid chains present in DCs. They were strongly increased in response to IL-17A while their relative proportion remained unchanged. Capture of extracellular lipids was the major mechanism of lipid droplet accumulation, visualized by electron microscopy and Oil Red O staining. Besides this foamy phenotype, IL-17A induced a mixed macrophage-DC phenotype and expression of the nuclear receptor NR1H3/liver X receptor-α, previously identified in the context of atherosclerosis as the master regulator of cholesterol homeostasis in macrophages. These IL-17A-treated DCs were as competent as untreated DCs to stimulate allogeneic naive T-cell proliferation. Following this first characterization of lipid-rich DCs, we propose to call these IL-17A-dependent cells “foamy DCs” and discuss the possible existence of foamy DCs in atherosclerosis, a metabolic and inflammatory disorder involving IL-17A.

of the physiological relevance of IL-17A-dependent foamy DCs in vivo, in the fi eld of atherosclerosis.

RNA extraction and microarrays
Cell lysis and RNA extraction were performed in Trizol (Invitrogen Life Technologies, Cergy Pontoise, France). RNA quality was checked with an Agilent bioanalyzer [RNA integrity number (RIN) >9]. RNA profi ling was performed using a high-density oligonucleotide array covering the whole human genome (Genechip human genome U133 Plus 2.0; Affymetrix, Santa Clara, CA). Sample processing and array hybridization was performed according to the manufacturer's protocols. Expression values and absent/present/marginal calls were calculated using the GCOS v1.4 software (Affymetrix). Absolute expression transcript levels were normalized for each chip by globally scaling all probe sets to a target signal intensity of 500. chemokines, run aggressive enzymatic attacks, and activate lymphocytes of innate and adaptive responses. Remodeling of lipid metabolism has been documented in macrophages in the context of atherosclerosis. When newly recruited monocytes engulf oxidized LDLs, they differentiate into lipid-laden foamy macrophages involved in both infl ammatory responses and tissue remodeling within the arterial intima ( 3,4 ). A subpopulation of lipid-laden foam cells was suggested to be derived from DCs in the ldlr Ϫ / Ϫ mouse model of atherosclerosis ( 5 ); however, very little is known about lipid accumulation in DCs. In recent years, immunogenic DCs with high endogenous lipid content have been characterized at homeostasis in the liver ( 6 ), while tolerogenic lipid-laden DCs have been identifi ed in the malignant microenvironment ( 7 ).
Interleukin (IL) 17A is a proinfl ammatory cytokine produced during innate response by lymphoid or nonlymphoid cells and during adaptive response by T H 17 cells ( 8,9 ). IL-17A is also involved in several chronic infl ammatory disorders including rheumatoid arthritis, multiple sclerosis, Crohn's disease, psoriasis and Langerhans cell histiocytosis [see review ( 10 )], but also in immunometabolic diseases. Obesity promotes expansion of T H 17 cells ( 11 ) while IL-17A inhibits adipocyte development ( 12 ). Atherogenesis correlates to the high number of IL-17Aproducing cells, although the exact role of those cells in this disease remains unclear ( 13 ). Target cells of IL-17A usually express IL-17RA and IL-17RC chains that form the IL-17A receptor. They include epithelial cells, endothelial cells, fi broblasts, and immune cells ( 14 ). IL-17A induces the production of proinfl ammatory mediators, metalloproteases, and antimicrobial peptides ( 15 ). We also demonstrated that IL-17A promotes long-term survival of monocyte-derived DCs and their fusion into multinucleated giant cells ( 16,17 ). In vitro, the short DC life span was extended above 12 days by exposure of DCs to IL-17A, suggesting that IL-17A-treated DCs may contribute to the development of chronic infl ammation resulting from multiple DC-T-cell cross talks, in vivo. For more than 20 years, the biology of DCs has been studied on in vitro-generated monocyte-derived DCs obtained with granulocytemacrophage colony-stimulating factor (GM-CSF) and IL-4. This model, although being restrictive compared with the numerous DC subpopulations existing in vivo, shows functional properties consistent with their counterpart in vivo. The existence of DCs fully differentiated from monocytes with a prominent role in initiating adaptive immunity was demonstrated in mice ( 18 ), while in humans, Segura et al. ( 19 ) identifi ed a population of human DCs present in infl ammatory environments that were most likely derived from monocytes.
For the fi rst time, we report that IL-17A strongly impacts the lipid metabolism of human in vitro-generated monocytederived DCs. IL-17A led to the generation of liver X receptor (LXR)-␣ + foamy DCs, highly competent in fatty acid capture and still able to stimulate allogeneic T-cell proliferation, as demonstrated by phenotypic and functional analysis. Previously published literature incites discussion resulting fatty acid methyl esters were extracted by isooctane and analyzed by GC with a DELSI instrument model DI 200 equipped with a fused silica capillary SP-2380 column (60 × 0.22 mm), with helium as a carrier gas. Cholesterol was extracted by a mixture of ethanol-chloroform (1:2, v/v). The dry residue was derivatized with 100 l N , O -bis-trimethylsilyl-trifl uoroacetamide for 20 min at 60°C. Derivatized cholesterol was then analyzed by GC/MS using positive chemical ionization mode.

Transmission electron microscopy on DCs
Fixation was initiated by adding an equal volume of fi xative solution, previously warmed to 37°C to the cells, either untreated or treated for 10

Oil Red O and Hoechst DNA staining
Cells were fi xed with 4% formaldehyde for 15 min at room temperature and subsequently stained with a solution of 0.4% Oil Red O dissolved in isopropanol (Sigma-Aldrich) for 20 min and gently shaken at room temperature. After three washes in water, DNA was stained with 10 g/ml of Hoechst 33342 (Sigma-Aldrich) for 30 min at room temperature. Pictures were analyzed using a Leica DMiRB microscope equipped with ×40/0.30 NA or ×40/0.55 NA objective lenses (Leica) a Leica DC300F camera and the Leica FW400 software.

Flow cytometry analysis
Cell suspensions were labeled according to standard procedures using antibodies directly coupled to fl uorochrome for a 30 min incubation in 1% BSA (BSA) and 3% human serumphosphate-buffered saline (PBS). After three washes in this buffer, cells were analyzed on a FACSCalibur (Becton Dickinson). For bodipy staining, cells were incubated with 1µg/ml of Bodipy diluted in PBS with 1% BSA for 20 min at room temperature, washed twice, and resuspended in PBS with 1% BSA. Fluorescence was quantifi ed on a LSRII (Becton Dickinson) and analyzed using FlowJo software.

Bodipy-FL-C16 capture
Cells (10 5 ) were resuspended in 100 µl PBS with or without 0.5 µg/ml of Bodipy-FL-C16 (Invitrogen) and incubated at 37°C or at 4°C for 10 min. Cells were washed fi ve times by centrifugation at 450 g for 5 min in ice-cold PBS containing 0.2% BSA. Fluorescence was measured by fl ow cytometry on a LSRII and analyzed using the FlowJo software. The intracellular Bodipy-FL-C16 fl uorescence was estimated from the shift in the mean fl uorescence intensity between 37°C and 4°C.

Alloreactivity measurement
T CD4 + cells were suspended at 10 7 cells/ml in ␣ -MEM medium containing 2% FCS. After 13 min of incubation in the presence

Analysis of microarray data
Data were fi ltered on the detection call so that probe sets with an absent call among all samples were excluded from the analysis. Statistical analyses were performed on 33,253 probe sets with linear models for microarray data (LIMMA) package ( 21 ) in R/Bioconductor. LIMMA uses moderated t -statistics, which provides for greater power at small sample sizes. Probe sets with a Benjamini-Hochberg-corrected P value р 0.01 and a fold change у 2 or р 0.5 in DC-17 versus DC were considered as differentially expressed. The regulated gene ontology (GO) biological pathways were identifi ed using DAVID (Database for Annotations, Visualization and Integrated Discovery) ( 22 ). The data set is available from the Gene Expression Omnibus database (GSE53163).

Reverse transcription quantitative PCR
First-strand cDNA were synthesized from 250 ng of total RNA in the presence of 100 U Superscript II (Life Technologies) and a mixture of random hexamers and oligo(dT) primers (Promega, Charbonnières-les-Bains, France). Reverse transcription quantitative PCR (RT-qPCR) assays were performed using a Rotor-Gene 6000 (QIAGEN, Courtaboeuf, France). For quantifi cation, a standard curve was generated for each target gene and for the housekeeping gene TATA-binding protein ( TBP ), with six different amounts (150 to 30,000 molecules/tube) of purifi ed target cDNA cloned in the pGEM plasmid (Promega). For each gene of interest, the amount of mRNA determined from the appropriate standard curve was divided by the amount of TBP mRNA to obtain a normalized value. Primer sequences are available upon request.

Lipid content analysis
Total lipids were extracted twice from cells with ethanolchloroform (1:2, v/v). Before extraction, 1,2-diheptadecanoyl-snglycero-3-phosphocholine,1,2-diheptadecanoyl-sn -glycero3phosphoethanolamine, stigmasterol, cholesteryl ester 17:0, and tri-17:0 triglyceride (all from Sigma-Aldrich) were added as internal standards. The organic phases were dried under nitrogen, and the different lipids classes were then separated by thin-layer chromatography using the solvent mixture hexane-diethylether-acetic acid (80:20:1, v/v/v) as eluent. Lipids were detected by UV light after spraying with 0.2% dichlorofl uorescein in ethanol and identifi ed by comparison with standards. Silica gel was scraped off. Triacylglycerols, cholesteryl esters, and phospholipids were transmethylated, and the fatty acid methylesters were analyzed by gas chromatography. Briefl y, each fraction was treated separately with toluenemethanol (2:3, v/v) and 14% boron trifl uoride in methanol. Transmethylation was carried out at 100°C for 90 min in screwcapped tubes. The reaction was terminated by cooling the tubes to 0°C and by the addition of 1.5 ml K 2 CO 3 in 10% water. The phospholipids, cholesterol, triglycerides, and cholesteryl esters) compared with DCs ( Fig. 1B-E ). Kinetic study revealed that after 6 days of culture with IL-17A, all lipid species were increased compared with DCs from the three donors. In addition, between day 6 and 12, phospholipids, triglycerides, and cholesteryl esters were further augmented in DC-17s from all donors ( Fig. 1B, D, E ) while the amount of cholesterol was differentially regulated from one donor to another, showing stability, increase, or decrease in donor 1, 2, and 3, respectively ( Fig. 1C ). Overall, at the optimum time point and depending on the donor, phospholipid levels were increased 2-5 times, cholesterol levels 2-4 times, triglyceride levels 5-12 times, and cholesteryl ester levels 3-9 times.

Quantities of palmitic, stearic, and oleic acid are strongly increased in DC-17s while their relative proportion is stable
We then investigated whether IL-17A could infl uence the fatty acid composition of phospholipids, triglycerides, and cholesteryl esters. Similarly to DCs, palmitic (16:0), stearic (18:0), and oleic (18:ln-9c) acid were the main fatty acid chains present in DC-17s from three donors ( Fig. 2A ), each accounting for ‫ف‬ 40%, 25%, and 10% of total fatty acids, respectively ( Fig. 2B ). There was at least a 2-fold increase (ranging from 2-fold in phospholipids of donor 1 to 19-fold in cholesteryl ester from donor 2) in the amount of palmitic acid (16:0) in DC-17s compared with DCs from the three donors, regardless of the class of lipid ( Fig. 2A ). Similarly, an increase in the amount of stearic acid (18:0), ranging from 1.8-fold in phospholipids of donor 1 to 8-fold in cholesteryl ester from donor 2, was observed in DC-17s from the three donors ( Fig. 2A ). Oleic acid (18:ln-9c) was augmented in all three DC-17s, in particular in triglycerides where it was induced 10 to 30 times depending on the donor ( Fig. 2A ). Although the amounts of palmitic, stearic, and oleic acids were increased in response to IL-17A, overall, their relative proportion remained unchanged ( Fig. 2B ). Therefore, this lipidomic analysis demonstrates that IL-17A highly increases the amount of all fatty acids present in DCs with a conserved composition and a variable intensity depending on the donor.

DC-17s become lipid-laden foamy DCs
Based on the increased amount of neutral lipids and phospholipids in DC-17s, we suspected that exposure to IL-17A would lead to the generation of foamy cells. Electron microscopy analysis of monocyte-derived DCs cultured for 10 days with IL-17A revealed the presence of numerous lipid droplets (LDs) in the cytoplasm while very few LDs were visible in untreated DCs ( Fig. 3A ). The LDs found in DC-17s were bigger (mean diameter = 0.4 µm) than those of untreated DCs (mean diameter = 0.1 µm). To investigate the kinetics of LD formation, cells were observed at days 0, 2, 4, 7, and 12 of culture with IL-17A following staining with Oil Red O, a dye specifi c for neutral lipids. In agreement with electron microscopy data, the vast majority of DCs contained few LDs at day 0 ( Fig. 3B, C ). After 2 days of culture with IL-17A, 2.5% of DCs accumulated Oil Red of 10 M of CFSE, the CFSE incorporation was blocked by the addition of a large excess of ␣ -MEM medium containing 2% FCS. T cells were then washed twice by centrifugation at 1,500 rpm for 10 min at 4°C in ␣ -MEM medium containing 2% FCS. Flow cytometry was used to survey that 100% of T cells have been labeled by CFSE. Monocyte-derived DCs were cultured in various numbers (10-10 5 DCs per well) for 5 days, in the presence of a constant number of CFSE + T cells (10 5 cells/well) purifi ed from a different donor (allogeneic), in ␣ -MEM medium containing 10% FCS. Cells were then harvested after 5 days of culture, and expression of CFSE was quantifi ed on an LSRII and analyzed using FlowJo software. The total number of CFSE-diminished daughter T cells per well was quantifi ed by a time-monitored fl ow cytometry analysis during 2 min at high speed (1 µl/s).

Statistical analysis
Statistical analysis of the differences between DCs and DC-17s were performed using LIMMA ( 21 ), with Benjamini-Hochbergcorrected P р 0.01 considered statistically signifi cant.

Gene expression profi le of IL-17A-treated DCs reveals intense remodeling of lipid metabolism
In order to get a comprehensive picture of genes regulated by IL-17A in DCs, we compared the gene expression profi le of in vitro-generated monocyte-derived DCs treated (DC- 17) or not (DC) with IL-17A using whole human genome microarrays. Analysis performed with LIMMA identifi ed 1,184 signifi cantly upregulated probe sets (fold-change у 2 and P р 0.01) and 937 signifi cantly downregulated probe sets (foldchange р 0.5 and P р 0.01) in DC-17s compared with DCs (supplementary Table 1). To identify if any GO classes were enriched in these two sets of genes, analysis were performed using DAVID. We found a total of eight signifi cantly enriched GO "biological process" categories with a P value cutoff of 0.05 in the upregulated gene set ( Fig. 1A ). Strikingly, only one out of the eight highlighted biological processes was linked to the "immune response." The remaining seven signifi cantly enriched pathways were all directly related to lipid metabolism, arguing for a strong modifi cation of lipid metabolism in DC-17s. Regarding the downregulated gene set, none of the GO classes were enriched with a P value cutoff of 0.05 (data not shown). The seven lipid-related GO biological pathways included 117 probe sets corresponding to 70 genes (supplementary Table 2). The top 50% of genes showing the greatest fold change in DC-17s compared with DCs are shown in Table 1 . Those genes encoded proteins involved in various aspects related to lipid metabolic processes and their regulation, lipid transport and localization, as well as sphingolipid metabolism. These results reveal a new and unexpected strong impact of IL-17A on lipid metabolism.

All lipid species are increased in DC-17s
To further explore the biological relevance of transcriptomic analysis, we analyzed the lipid content of DC-17s for 6 and 12 days using thin-layer chromatography and GC or GC/MS. DC-17s from three different donors had marked increased amounts of all the lipids species analyzed (i.e., upregulated in DC-17s versus DCs, but with a weak expression level, and no signifi cant change was observed for TNF-␣ , IL-6, transforming growth factor (TGF)-␤ 1, IL-10, IFN-␥ , and IL-33 expression (supplementary Table 3). Therefore, IL-17A induces the generation of foamy DCs full of LDs without inducing expression of other cytokines involved in the regulation foam cell formation, except a weak level of IL-1 ␤ .

DC-17s uptake lipids from the microenvironment
We then investigated the origin of the lipids accumulated in DC-17s. As shown in Fig. 4A , genes encoding the key metabolic enzymes of lipid synthesis [i.e., ATP citrate lyase ( ACLY ), acetyl-CoA carboxylase ␣ ( ACACA ), fatty acid synthase ( FASN ), and 3-hydroxy-3-methylglutaryl-CoA reductase ( HMGCR )] were signifi cantly downregulated in DC-17s versus DCs. This suggested that lipid synthesis was not the main mechanism responsible for IL-17A-induced lipid accumulation in human DCs. Thus, we postulated that DC-17s have acquired increased ability to uptake lipids from the microenvironment. Using Affymetrix microarrays, we found that the scavenger receptors MSR1 and CD68 were signifi cantly overexpressed in DC-17s compared O-positive LDs. The amount of LDs inside cells, as well as the percentage of Oil Red O-positive cells, gradually increased over time: ‫ف‬ 80% of the cells were full of LDs after 12 days of culture ( Fig. 3C ). High amounts of LDs were found in binucleated cells ( Fig. 3B ) and multinucleated giant cells with more than two nuclei (not shown, <10% of total cells at day 12) resulting from IL-17A-induced cell fusion. Binucleated cells accounted for 10% and 20% of total cells at days 4 and 12, respectively. The increased amount of lipids in DC-17s and percentages of lipid-rich DCs were further confi rmed by fl ow cytometry, using the lipophilic fl uorescent dye Bodipy 493/503 ( Fig. 3D ). In agreement with Oil Red O staining data, IL-17A treatment markedly increased the lipid content after 3 and 6 days of culture. Perilipins are a family of proteins that associate with the surface of LDs. The PLIN 2 gene was upregulated in DC-17s versus DCs, as observed in microarray data ( Table 1 ) and validated by real-time RT-PCR ( Fig. 3E ). PLIN2 was also induced at the protein level after 12 days of culture with IL-17A ( Fig. 3F ). In addition, we analyzed the expression of cytokines known to regulate the formation of foamy macrophages to look for cytokinemediated indirect effect of IL-17A on DCs. IL-1 ␤ was 4-fold higher amounts of intracellular Bodipy-FL-C16 than DCs ( Fig. 4E ). The ability of DCs to uptake the fatty acid Bodipy-FL-C16 was estimated on three different donors from the shift in the mean fl uorescence intensity between 4°C and 37°C, which was markedly enhanced by IL-17A treatment ( Fig. 4F ). Altogether, those data suggest that increased lipid uptake is the main mechanism responsible for lipid accumulation in IL-17A-induced foamy DCs.

IL-17A-induced foamy DCs acquire macrophage markers and keep their immunogenic properties
Then, we performed a phenotypic, genetic, and functional characterization of human monocyte-derived foamy with DCs ( Fig. 4B ). MARCO and CD36 were not signifi cantly upregulated in DC-17s. Among the six fatty acid transport proteins (FATPs) and seven members of the low density lipoprotein receptor (LDLR) family, FATP1 and LDLRrelated protein 1 ( LRP1 ) were signifi cantly overexpressed in DC-17s versus DCs ( Fig. 4C, D ).
To investigate if DC-17s had an increased ability to capture lipids from the microenvironment, we measured cellular uptake of fatty acids using the fl uorescently labeled palmitic acid Bodipy-FL-C16. Untreated DCs or DCs treated for 5 days with IL-17A were incubated with Bodipy-FL-C16 for 10 min at 37°C or 4°C, and intracellular fl uorescence was measured by fl ow cytometry. DC-17s displayed  List of the top 50% of human genes from microarray analyses, showing the greatest mRNA expression fold change in DC-17s compared with untreated DCs among all genes found in the seven lipid-related GO pathways. The GO pathways are as follows: (1) lipid metabolic process, (2) cellular lipid metabolic process, (3) regulation of lipid metabolic process, (4) membrane lipid metabolic process, (5) lipid localization, (6) lipid transport, and (7) sphingolipid metabolic process. When several probe sets were available for a given gene, the probe set with the most signifi cant overexpression in DC-17s versus DCs was selected. The fold change DC-17/DC was calculated from the mean expression in DC-17s (n = 5) and DCs (n = 4).
concomitantly expressed DCs and macrophage markers ( Table 2 ), thus raising the question of the appropriate name for these foamy myeloid cells. DCs have been originally differentiated from macrophages by their ability to activate naïve T-cell proliferation in coculture with allogeneic T cells ( 23 ). In cocultures of DCs and CFSE + T cells from DCs obtained with IL-17A. At day 6 of IL-17A treatment, DC-17s did not express CLEC9A, but they were positive for CD1a, HLA-DR, CD14, CD68, CD206, and CD163 expression ( Fig. 5A ). Compared with expression of these markers on monocytes, monocyte-derived macrophages, and monocyte-derived DCs, we concluded that DC-17s  genes upregulated in DC-17s versus DCs was NR1H3 ( Table 1 ). NR1H3 was 21-fold higher in DC-17s (mean value = 8,599) than that in untreated DCs (mean value = 414; Table 1 ). Affymetrix data were confi rmed by RT-qPCR ( Fig. 5D ) and by Western blot ( Fig. 5E ) on three independent donors for each experiment. LXR-␣ protein expression was induced after 6 days of culture with IL-17A and still maintained at day 12 ( Fig. 5E ). Furthermore, the expression of several NR1H3 target genes such as ABCA1 , a cholesterol transporter, or APO , the structural components of lipoprotein particles, was also increased in DC-17s versus DCs ( Table 1 ). Those data were also validated at the mRNA level ( ABCA1 and APOC1 ; Fig. 5D ) and at the protein level (APOE; Fig. 5E ). Thus, the LXR-␣ genetic program is active in IL-17A-induced foamy DCs, as previously established in foamy macrophages. different donors, DC-17s activated the proliferation of T cells, which underwent up to fi ve cell cycles ( Fig. 5B ). Numeration of CFSE-diminished daughter T cells demonstrated that allogeneic T-cell proliferation obtained with IL-17A-induced foamy DCs was similar as that obtained with untreated DCs ( Fig. 5C ). Therefore, we propose the name of foamy DCs for the IL-17A-treated DCs characterized by a mixed DC/macrophage phenotype and the ability to stimulate allogeneic T cell proliferation.

IL-17A activates the LXR-␣ genetic program in DCs
The nuclear receptor LXR-␣ /NR1H3 is a key regulator of foamy macrophage function. LXR-␣ controls transcriptional programs involved in the regulation of lipid homeostasis in response to rapid changes in cellular lipids and infl ammation ( 24 ). Interestingly, one of the top 50% of

DISCUSSION
Immunometabolism is an emerging fi eld of investigation at the interface between immunological and metabolic processes. Deregulation of intracellular lipid metabolism has been extensively studied in foamy macrophages in the context of atherosclerosis ( 4 ). However, much less is known regarding DCs. Here we show for the fi rst time that in vitrogenerated monocyted-derived DCs respond to the proinfl ammatory cytokine IL-17A by modulating their lipid metabolism thus generating foamy DCs, in vitro.
We report an intense remodeling of lipid metabolism induced by IL-17A in DCs: i ) several genes involved in lipid metabolism were upregulated; ii ) all the analyzed lipid species were quantitatively increased with a qualitatively stable composition of fatty acid chains; and iii ) LDs accumulated in the cytoplasm. Regarding those intracellular metabolic aspects, foamy DCs resemble foamy macrophages characterized in atheroma. In atherosclerosis, lipid overload under the form of LDL is a risk factor because chronic infl ammation oxidizes LDLs that are specifi cally captured by macrophages through the scavenger receptors MSR1 and CD36 ( 4 ), converting those cells into foamy macrophages. In DCs, IL-17A upregulated the scavenger receptors MSR1 and CD68 but also the fatty acid transporter FATP1 , suggesting that the main mechanism supporting lipid accumulation was an increased capture of lipids from the microenvironment. This hypothesis was validated by the enhanced uptake of the fatty acid FL-C16 by DCs in response to IL-17A. Thus, foamy DC formation in response to IL-17A comes from capture of extracellular lipids. IL-17A-induced foamy DCs were generated in longterm cultures and may be indirectly mediated by a cytokine cascade. Several cytokines were previously shown to modulate the formation of foamy macrophages generated in the presence of aggregated, acetylated or oxidized LDL. IFN-␥ , IL-1 ␤ , and TNF-␣ promote macrophage foam cell formation (25)(26)(27)(28), while IL-6, TGF-␤ 1, and IL-33 (29)(30)(31) have been described as anti-foam cell cytokines in humans and IL-10 facilitates both cholesterol uptake and effl ux in macrophages ( 32 ). Only IL-1 ␤ expression was weakly affected by IL-17A treatment, suggesting a role for this models. We demonstrate that IL-17A strongly induces APOE, an apolipoprotein involved in HDL formation allowing the reverse transport of cholesterol to the liver and thereby limiting atherosclerosis. Accordingly, IL-17A may sustain two antagonistic functions in atherogenesis: the proinfl ammatory role of IL-17A would promote plaque formation while the IL-17A-induced APOE expression would counteract plaque formation. In the Apoe Ϫ / Ϫ mice, only the fi rst function can be active and may explain the major proatherogenic role of IL-17A. In the Ldlr Ϫ / Ϫ mouse model, IL-17A would exert both functions and the second function may counteract proinfl ammatory one.
The origin of foamy cells in atherosclerosis should be questioned: do they belong to macrophage or DC lineage? Historically, DCs have been functionally defi ned by their original ability to effi ciently stimulate allogeneic T-cell proliferation ( 23 ). As this property is maintained in IL-17Ainduced foamy cell generated in vitro from monocytederived DCs, we propose to call these cells "foamy DCs." However, we show that IL-17A induces the expression of the macrophage markers CD14, CD68, and CD163 on foamy DCs. In addition, the M2 macrophage marker CD206 is expressed on both DCs and DC-17s. Finally, CLEC9A (also known as DNGR-1), a marker of the BDCA3 + human conventional DC subset, is not expressed by monocytederived DCs, as previously described ( 43 ). In vitro microarray studies showed that in response to oxidized LDL, monocyte-derived foamy macrophages may acquire a DClike gene expression pattern ( 44 ). So, the exact nature of foamy myeloid cells in atherosclerosis remains an intriguing question, which cannot be solved by in vitro experiments. In vivo, foam cell formation and atherosclerotic plaque growth in the artery was fi rst attributed to foamy macrophages defi ned as fat-laden myeloid cells expressing macrophage markers (F4/80 in mice and CD68 in humans) ( 45 ). However, a recent study using the Ldlr Ϫ / Ϫ mouse model have demonstrated that the majority of intimal lipids in nascent lesions were located within foam cells that express CD11c ( 5 ), a marker widely used as a specifi c marker for murine DCs. CD11c is in fact also expressed by several tissue macrophages ( 46 ) as well as monocytes in models of atherosclerosis ( 47 ). CD11c + circulating monocytes can be activated by intracellular lipid accumulation prior to their recruitment to athero-prone regions of the vasculature, confusing the issue of what are CD11c + foam cytokine, which may be further enhanced by infl ammasome activation, in vivo. However, it was recently demonstrated that fatty acid-induced mitochondrial uncoupling abrogated IL-1 ␤ secretion, deviating the cholesterol crystalelicited response toward selective production of IL-1 ␣ ( 33 ). Therefore, IL-17A induction of foamy DCs rather results from direct genetic reprogramming of lipid metabolism than on indirect cascade of cytokines, though we cannot totally exclude indirect effects via other unidentifi ed products secreted by IL-17A-stimulated DCs.
The nuclear receptor LXR-␣ /NR1H3 is a sterol sensor that controls the regulation of lipid homeostasis. LXR-␣ activation is a hallmark of foamy macrophages ( 24 ). DC expression of LXR-␣ was dramatically increased by IL-17A, both at the mRNA and protein levels. LXR-␣ target genes such as ABCA1 and APOC1 , APOC2 , and APOE ( 34 ) were strongly upregulated, indicating that the LXR-␣ transcriptional function is active when DCs are treated by IL-17A. LXR-␣ activation takes place upon ligand binding that leads to a molecular switch replacing a corepressor by a coactivator complex ( 35 ). Previous studies showed that adding exogenous synthetic LXR ligands activated the LXR program in monocyte-derived DCs ( 36,37 ). However, our data provide fi rst evidence that LXR-␣ can be transcriptionally active in the presence of IL-17A without exogenously added synthetic ligands. This implies that natural endogenous ligands, which remain to be determined, are probably generated in response to IL-17A.
Interestingly, the role of IL-17A has been investigated in two mouse models of atherosclerosis, the Apoe Ϫ / Ϫ and the Ldlr Ϫ / Ϫ model, with opposite conclusions. In the Apoe Ϫ / Ϫ model, in vivo administration of an antibody blocking IL-17A decreased atherosclerotic phenotype suggesting that IL-17A is proatherogenic, independently of APOE ( 38 ). In Ldlr Ϫ / Ϫ mice, neutralizing anti-IL-17A antibodies had no effect. Based on prior studies demonstrating that Socs3 negatively regulates IL-17A expression in T cells ( 39 ), Socs3 Ϫ / Ϫ Ldlr Ϫ / Ϫ chimeric mice were generated and had reduced atherogenesis ( 40 ). Anti-IL-17A antibody treatment or IL-17A defi ciency increased plaque formation in those mice, suggesting that IL-17A may be antiatherogenic when the Apoe gene is functional ( 41,42 ). Assuming that IL-17Adependent foamy DCs are physiologically relevant, our in vitro data provide knowledge to solve the apparent discrepancies on the role of IL-17A in atherosclerosis mouse Monocytes (Mo), monocyte-derived macrophages (MP), and monocyte-derived DCs before (DC) and after (DC-17) 6 days of treatment with IL-17A. Expression of the indicated markers was analyzed on Mo, MP, DC, and DC-17 by fl ow cytometry. Representative of n > 3 experiments. -, absence of marker expression compared with isotype control; + and ++, low versus high positive expression, according to the mean fl uorescence intensity. cells ( 47 ). It is not possible to determine whether foamy cells originate from macrophages or DC lineage based on phenotypical analysis. To understand whether foamy DCs exist in vivo, it would be necessary to perform foam cell purifi cation in mouse model of atherosclerosis, treated or not with anti-IL-17A, for ex vivo evaluation of their alloreactive properties. Our in vitro data on the generation of LXR-␣ + APOE + alloreactive foamy DCs in the presence of IL-17A, together with the role of IL-17A in different mouse models of atherosclerosis in vivo, suggest that different subpopulations of foamy macrophages and foamy DCs exist in atherosclerosis. Knowing that T H 17 and other IL-17Apositive cells participate in the atherosclerotic-associated infl ammation in human arteries ( 13 ), IL-17A may participate in the generation of foamy DCs in atherosclerosis. Whether these foamy DCs really exist in vivo as a separate and relevant myeloid entity will have to be addressed in the future.
IL-17A is involved in the pathogenesis of several chronic infl ammatory diseases, not only those associated with metabolic disorders such as atherosclerosis, type 2 diabetes mellitus, and obesity, but also cancer and tuberculosis where foam cells were characterized ( 7,48 ). In addition to its proinfl ammatory functions, IL-17A may participate in the generation of foamy DCs in various chronic infl ammatory contexts, in vivo.
For stimulating discussions on interfacing disciplines, in memoriam of C. Rabourdin-Combe. The authors thank UMS3444/ US8 for the platforms PLATIM imaging and fl ow cytometry, Profi leXpert for array analysis (http://www.profi lexpert.fr), and Hélène Valentin for helpful discussions and critical reading of the manuscript.