Adipose Tissue–Derived Stem Cells From Obese Subjects Contribute to Inflammation and Reduced Insulin Response in Adipocytes Through Differential Regulation of the Th1/Th17 Balance and Monocyte Activation

Obesity is defined as an excessive accumulation of adipose tissue (AT), and its incidence is increasing worldwide, reaching epidemic proportions. The correlation between obesity, type 2 diabetes, and their complications has been clearly established (1). Obesity is associated with low-grade chronic inflammation of AT, due to the infiltration of multiple immune cells, including monocytes, macrophages, natural killer cells, and lymphocytes, resulting in secretion of adipokines and proinflammatory cytokines by both adipocytes and the population of infiltrating immune cells (2,3). Strikingly, in obesity, immune cells are the most represented cell types within AT, with macrophages accounting for up to 50% of the total cells in the AT of obese rodents and humans (4). While monocytes and macrophages have been clearly shown to contribute to the induction of a proinflammatory profile (5), recent research has also implicated the Th17 cells in obesity-dependent inflammation (6,7) and Th17 cell enrichment has been observed in AT of obese subjects and obese type 2 diabetic patients (8,9). Similarly, peripheral Th17 cells were significantly increased in a cohort of obese women (10). Interleukin (IL)-17A and IL-17F are cytokines playing a preponderant role in the propagation of inflammation, due to the ubiquitous expression of its receptors IL-17RA/RC (11). IL-17A also increases the secretion of other proinflammatory cytokines—including IL-1β and IL-6—and chemokines by multiple cell types including macrophages, monocytes, stromal cells, adipocytes, and stem cells. IL-17A may also directly inhibit adipogenesis (12), thus impairing the lipid storage capacities of AT and therefore contributing to the establishment of an insulin-resistant state. In spite of these advances, the mechanisms by which Th17 cells infiltrate and reside within AT remain largely unexplored. Previously, we demonstrated that bone marrow and synovium-derived mesenchymal stem cells (MSCs) amplify IL-17A production by increasing the frequency of Th17 cells, resulting in a vicious circle of proinflammatory cytokine secretion that worsens joint chronic inflammation, such as in rheumatoid arthritis (13). For this study, we postulated that similar mechanisms could occur in human AT, in which adipose stem cells (ASCs) would favor Th17 cell expansion. Indeed, ASCs resemble MSCs in their capacity to 1) differentiate into multiple tissue lineages, 2) secrete multiple cytokines, and 3) display self-renewing capacities (14). To address this question, we cocultured ASCs with blood mononuclear cells (MNCs). We observed that ASCs induced a strong activation of both Th17 cells and monocytes, with inhibition of Th1 cytokines, through pathways involving phosphatidylinositol 3-kinase (PI3K), STAT3, and the inflammasome. Moreover, ASCs differentially modulated STAT3 and STAT5 binding on the IL-17A/F gene locus. ASC-MNC cocultures generated proinflammatory conditioned media (CM) that impaired mRNA differentiation markers of adipocytes and adipocyte responsiveness to insulin. In addition, IL-17A secretion also occurred in stromal vascular fractions (SVFs) issued from obese, but not lean, individuals. These data suggest that ASCs may contribute to the development of chronic low-grade inflammation in obese patients, through polarization of infiltrating T cells toward the Th17 lineage, and facilitation of a proinflammatory environment by monocytes, which may in turn prevent adipogenesis and decrease the insulin response of adipose cells.

Visceral and subcutaneous AT was obtained from residues of bariatric surgery, with the approval of the Person Protection Committee of Hospices Civils de Lyon or from lean patients undergoing surgery with their informed consent. ASCs were isolated from SVF, following a modified described protocol (15), in which collagenase type-Ia (Sigma-Aldrich, Saint-Quentin-Fallavier, France) replaced Liberase. ASCs were selectively expanded in DMEM/Ham F12 cell culture medium (50/50 vol/vol) (Invitrogen) containing 10% FCS, 4 ng/mL human β-FGF (eBioscience), 2 mmol/L l-glutamine, and 100 units/mL penicillin-streptomycin. ASC were differentiated using DMEM/Ham F12 containing 10% FCS, 2 mmol/L l-glutamine, 100 units/mL penicillin-streptomycin, 1.8 μmol/L insulin, 0.5 mmol/L isobutylmethylxantine, 500 nmol/L dexamethasone, 1 μmol/L rosiglitazone, 2 nmol/L triiodothyronine, and 10 μg/mL transferrin (all from Sigma-Aldrich). During differentiation, half of the culture medium was changed every 2–3 days. ASC differentiation lasted 8–15 days. Differentiation was validated by oil red O visualization of lipid droplets (Supplementary Fig. 1A and C). ASCs were identified by immunophenotypic criteria based on expression of CD73, CD90, CD105; absence of expression of CD45 and HLA-DR; and validation of their capacity to also differentiate into osteoblasts when cultured in osteoblast-differentiating medium (StemPro Osteogenesis Differentiation kit; Life technologies). Calcified bone matrix from osteoblasts was visualized by alizarin-red staining (Supplementary Fig. 1D).

MNC were prepared by Ficoll-Hypaque (1.077 g/mL) density gradient centrifugation as previously described (13). Blood samples were obtained through the Lyon Blood-Bank Center (France), following institutionally approved guidelines. ASCs and MNCs were stored in liquid nitrogen prior to use.

Cocultures were initiated by seeding ASCs in 96-well plates. MNCs were coseeded 18–24 h later, for 24–48 h, in the presence or absence of phytohemagglutinin (PHA) (5 μg/mL, Sigma-Aldrich), or anti-CD3/CD28 monoclonal antibodies (5 μg/mL; ImmunoTools, Friesoythe, Germany). Neutralizing anti–IL-6-receptor antibodies (Tocilizumab, Chugai Pharma France, Paris, France); anti–IL-1β monoclonal antibodies (R&D Systems, Minneapolis, MN); caspase inhibitor Z-YVAD-fluoromethylketone (Z-YVAD-fmk; MBL-International, Nanterre, France); PI3K inhibitor Wortmannin (Sigma-Aldrich); STAT3 Inhibitor-VI, S3I-201 (sc-204304; Santa Cruz Biotechnology); or anti–intracellular adhesion molecule (ICAM)-1 monoclonals (Beckman Coulter, Villepinte, France) were added to cocultures as indicated in the figure legends. Caspase-1 activity was measured using a caspase-1 colorimetric assay kit (BioVision, Milpitas, CA) following the manufacturer’s instructions.

MNC and ASC-MNC cocultures were treated for the last 4 h of incubation with 3.6 μmol/L brefeldin-A (GolgiPlug; Becton-Dickinson, Le-Pont-de-Claix, France) and fixed/permeabilized with a Cytofix/Cytoperm kit (Becton-Dickinson) prior to incubation with fluorescein isothiocyanate–conjugated anti–IL-17A (eBioscience), phycoerythrin-conjugated anti–γ-interferon (IFNγ), and APC-conjugated anti-CCR6 (Becton-Dickinson). Thirty minutes of incubation with fluorescently labeled antibodies were followed by several washes in PBS/2% FCS. Cells were analyzed on a LSRII flow cytometer (Becton-Dickinson); image acquisition and data treatment were performed using the Becton-Dickinson FACSDiva Software.

IL-17A, IL-1β, IL-6, IFNγ, and tumor necrosis factor (TNF)α concentrations were evaluated with ELISA using the corresponding antibodies (eBioscience, Paris, France).

RNA was isolated with Tri-Isolation-Reagent (Roche Diagnostics, Meylan, France). cDNA synthesis was performed using the Primescript-RT kit (Takara, Dalian, Japan). Quantitative PCR (qPCR) was performed on a Rotor-Gene Real-Time-PCR System, using ABsolute QPCR SYBRGreen Mix (ABgene, Illkirch, France). Primer sequences are reported in Supplementary Table 1.

Chromatin immunoprecipitation (ChIP) was performed on ∼5 million MNC or MNC-ASC cocultures as previously described (16), with minor modifications. Cells were cross-linked with 1% (w/v) formaldehyde in PBS for 10 min at room temperature and quenched with 0.125 mol/L glycine in PBS. Formaldehyde-fixed cells were harvested, homogenized in 1% SDS cell lysis buffer, and sonicated for 15 min at maximal power on a Diagenode Bioruptor bath sonicator (Liege, Ougrée, Belgium). Prior to immunoprecipitation, DNA shearing was monitored by agarose gel. Chromatin was immunoprecipitated with the indicated antibodies or processed as input sample. Immune complexes were recovered by incubation with protein A–conjugated (for rabbit antibodies) or protein G–conjugated (for mouse antibodies) magnetic beads (Millipore Temecula, Billerica, MA). DNA was recovered from immune complexes and DNA quantified by RT-qPCR. Primer sequences for ChIP amplification on the IL-17A/F genomic locus were designed to match STAT binding regions previously identified in mice (17) (Supplementary Table 1). ChIP experiments were performed three times. ChIP recoveries are expressed as percent of input samples. Immunoprecipitations with negative control antibodies consistently yielded negligible amplifications. The following antibodies were used: anti-STAT3, anti-STAT5 (rabbit monoclonal antibodies; Cell Signaling Technology, Beverly, MA). Polyclonal rabbit anti-mouse immunoglobulins (DakoCytomation, Glostrup, Denmark) were used as negative control.

Proteins were separated by SDS-PAGE and transferred to polyvinylidene fluoride membranes. Primary antibodies were detected with peroxidase-conjugated secondary antibodies and enhanced chemiluminescence. The following antibodies were used: anti–phospho-STAT3-Ser⁷²⁷ (antibody no. 9134), anti-STAT3 (no. 9139), anti–phospho-Akt-Ser⁴⁷³ (no. 9271), and anti–pan-Akt (no. 4691) (all from CST).

Four-week-old male C57BL/6JOlaHsd and ob/ob mice were from Harlan. Mice were housed at 22°C on a 12-h light/dark cycle. Procedures were conducted in accordance with institutional guidelines for the care of laboratory animals. After acclimatization, C57BL/6JOlaHsd mice were divided into two groups: one fed a standard chow diet (Harlan) and the other fed a high-fat, high-sucrose diet (TD99249; Harlan) for 16 weeks. The ob/ob mice were fed a standard chow diet for 8–10 weeks prior to sacrifice. Animals’ body weights and epididymal fat weights are presented in Supplementary Table 2.

For pairwise comparisons, we applied the Student t test statistics. For multiple comparisons, we used one-way ANOVA followed by post hoc multiple comparison tests as indicated in the figure legends. Differences were considered statistically significant when P < 0.05.

In previous reports, we showed that stromal stem cells inhibit the Th1 response (18,19) while simultaneously promoting Th17 cells (13). We therefore analyzed the behavior of MNCs entering into contact with human ASCs derived from visceral fat of obese donors in a coculture model. Graded concentrations of ASCs were cocultured with MNCs. When MNCs were activated with PHA for 48 h, an ASC concentration–dependent increase of IL-17A secretion was observed, reaching significant statistical differences, starting from the lowest ASC:MNC ratio (1:100) compared with PHA-activated MNCs (Fig. 1A). IL-1β secretion was also increased in the presence of ASCs, with the lowest ASC:MNC ratio eliciting a submaximal response. IL-6 secretion was enhanced in the cocultures but independently from T-cell activation. ASC induced a dose-dependent decrease of PHA-mediated TNFα, and IFNγ production showed a dual response, with high levels in the presence of the lowest concentrations of ASC, but with an ASC dose–dependent decrease (Fig. 1A). Because this dual response was presumptive of the presence of two subpopulations of IFNγ-secreting cells, we next analyzed in more detail the phenotype of the ASC-cocultured T cells. As shown in Fig. 1B, a 10-fold increase in Th17 cells expressing CCR6 was observed by flow cytometry in the presence of PHA and ASC at an ASC:MNC ratio of 1:5. CCR6 was expressed by most Th17 cells, as previously reported (20). When IFNγ was analyzed, two different T-cell subsets were observed, including an IFNγ-producing Th1 population, which secreted IFNγ in response to PHA treatment (Fig. 1C, panel II), which was gradually repressed in the presence of ASCs. The second population secreted both IL-17A and IFNγ, with an ASC dose-dependent response curve (Fig. 1C, panels III–V). These results suggest that ASCs from obese individuals negatively modulate IFNγ secretion by Th1 cells, while increasing IL-17A and IFNγ secretion by Th17 cells.

To investigate whether IL-17A secretion also occurs in a more physiologically relevant setting, we evaluated the effects of PHA treatment on SVF from visceral AT derived from obese versus lean donors. We observed increased IL-17A secretion in PHA-activated SVF from obese but not lean AT (Fig. 2A). Similarly, PHA-induced IL-17A secretion was observed in SVF derived from ob/ob mice but not wild-type mice and in AT explants from high-fat high-sucrose fed, but not chow-fed, mice (Supplementary Fig. 2). To define which, from MNC or ASC lineages, plays a role in inducing IL-17A secretion, lean- versus obese-derived ASCs were cocultured with MNCs from lean donors (Fig. 2B). Reciprocally, lean- versus obese-derived MNCs were cocultured with ASCs from lean donors (Fig. 2C). IL-17A secretion was induced by MNCs from lean subjects interacting with ASCs from obese but not lean individuals (Fig. 2B). However, IL-17A secretion was not induced when MNCs from obese donors were cocultured with lean-derived ASCs (Fig. 2C). Thus, MNC polarization toward the Th17 subset is specifically induced by ASCs from obese donors.

Cultured ASCs are known to be negative for HLA class II molecule expression (21) but positive for HLA class I. To investigate the possibility that an alloreactive response against HLA class I molecules could amplify ASC-mediated Th17 promotion, we cocultured three independent MNC preparations with autologous or allogeneic ASCs and measured PHA-mediated IL-17A secretion. IL-17A secretion was independent from the origin of ASCs, indicating that ASCs induce Th17 promotion irrespective of HLA class I expression (Fig. 3A).

Because PHA is a nonspecific mitogen, activating T cells and other mononuclear cells, we next investigated whether specific stimulation of T-cell receptor (TCR) would be sufficient to induce Th17 polarization in the presence of ASCs. Incubation with CD3/CD28 activating antibodies was used to mimic antigen-mediated TCR activation and resulted in IL-17A secretion to an extent similar to that with PHA in the presence of ASCs (Fig. 3B). To evaluate the role of monocytes in this model, we depleted them from MNC preparations prior to coculture assays. We observed that, although not absolutely required, monocytes amplified IL-17A production (Fig. 3C). In contrast, IL-1β production almost completely depended on the presence of monocytes (Fig. 3D). This was not the case for IL-6 secretion, which mostly depended on ASCs (Fig. 3E).

ASCs have been shown to inhibit Th1 cytokine production through secretion of soluble factors but also through cell contact-dependent mechanisms (22). To investigate this issue, we cocultured PHA-activated MNCs with ASCs on a transwell culture system, not allowing for cell-cell contacts between MNC and ASC. In parallel, conditioned media collected from 48-h ASC cultures were added to PHA-activated MNC. ASC-MNC physical interaction played an important role in IL-17A production, since physical separation of ASCs and MNCs strongly inhibited IL-17A production (Fig. 4A). However, soluble factors secreted by ASCs are also involved, albeit to a lesser extent, because even in transwells, a small but significant increase in IL-17A secretion was observed (Fig. 4A). Coincubation with CM collected from ASC cultures induced a significant increase in IL-17A secretion by PHA-activated MNCs (Fig. 4B), albeit at a much lower level than in ASC-MNC cocultures, indicating the need for physical interaction between MNCs and ASCs for a full response. Accordingly, ICAM-1, an adhesion molecule, might mediate the communication between mononuclear and stromal cells, as the presence of inhibitory anti-CD54 (ICAM-1) monoclonal antibodies in coculture assays reduced by 40% IL-17A secretion (Fig. 4C).

To investigate the molecular mechanisms leading to IL-17A induction by ASCs, we then tested whether IL-1β and IL-6, which were increased in ASC-MNC cocultures (Fig. 1A), could contribute to this positive regulation. With this aim, cocultures of PHA-activated MNCs with ASCs were performed in the presence of neutralizing antibodies directed against IL-1β or IL-6 receptor (IL-6R). Results showed that anti–IL-1β neutralizing antibodies significantly, albeit not completely, inhibited IL-17A production, while anti–IL-6R had no significant effect (Fig. 5A and B, respectively). Because IL-1β secretion requires activation of the cysteine protease caspase-1, a component of the inflammasome (23), we then evaluated the effects of the caspase-1 peptide inhibitor Z-YVAD-fmk and found a drastic inhibition of IL-17A production (Fig. 5C), demonstrating the implication of the inflammasome in ASC-mediated Th17 cell promotion. In support of these results, caspase-1 enzymatic activity was found to be significantly increased in PHA-activated ASC-MNC cocultures compared with unstimulated ASC-MNC cocultures or PHA-activated MNCs (Fig. 5D).

PI3Ks are central regulators of Th17 cell differentiation. Inhibition of PI3Kδ and/or PI3Kγ—the two isoforms specifically found in immune cells—has been shown to alleviate inflammatory and autoimmune diseases (24). Thus, we evaluated whether PI3K could contribute to ASC-mediated IL-17A secretion. The PI3K inhibitor Wortmannin prevented IL-17A production and Akt Ser⁴⁷³ phosphorylation in PHA-activated ASC-MNC cocultures (Fig. 6A and B). As IL-17 gene transcription is dependent on STAT3 (25), IL-17A secretion was measured after incubation with the STAT3 inhibitor, STAT3-VI, which induced a dose-dependent inhibition of IL-17A secretion from PHA-activated ASC-MNC cocultures (Fig. 6C). To define whether PI3K and STAT3 signaling events are related, we measured Akt and STAT3 phosphorylation after incubation with either Wortmannin or STAT3-VI. STAT3 inhibition by STAT3-VI did not affect Akt phosphorylation, but Wortmannin decreased STAT3 phosphorylation, indicating that PI3K acts upstream of STAT3 to promote IL-17A secretion (Fig. 6D).

The transcriptional control of IL-17A in T cells is modulated by the reciprocal binding of STAT3 and STAT5 on multiple common binding sites of the IL-17 gene locus (17). According to this model, STAT3 promotes IL-17A transcription, while STAT5 acts as a negative modulator by competing on the same transcription factor binding sites as STAT3. We thus performed ChIP experiments with antibodies directed to STAT3 and STAT5 and analyzed three different IL-17A/F gene loci (p1, p3, and p4) (Fig. 7A) known to be involved in transcription factor binding. Figure 7B shows that STAT3 binding to IL-17A/F p1 and p4 regions increased in PHA-stimulated cells, regardless of ASC presence. However, STAT5 binding to the same IL-17A/F loci did not increase in the presence of ASCs (Fig. 7B), leading thus to an enhancement of the STAT3/STAT5 IL-17A gene binding ratio from 1.06 in PHA-activated MNC to 1.75 in PHA-activated ASC-MNC cocultures (Fig. 7C). Thus, the presence of ASCs appeared to favor the STAT3/STAT5 IL-17A/F binding balance toward STAT3 and to subsequently activate IL-17A gene transcription.

Because IL-17A inhibits adipogenesis (12,26), we analyzed the impact of ASC-MNC coculture CM on ASC differentiation into adipocytes. We differentiated ASCs from subcutaneous AT for 5 or 12 days and cultured them for the last 72 h in differentiation medium containing or not a 50% volume of CM collected from resting or PHA-activated ASC-MNC cocultures or MNC cultures. Differentiation was monitored by gene expression profiling of the adipocyte-specific genes FABP4, PPARγ, adiponectin (AdipoQ), and leptin. Figure 8A shows a strong increase of these genes upon differentiation compared with undifferentiated ASCs. However, differentiating ASCs incubated with CM from PHA-activated MNC, although appearing morphologically similar to control adipocytes (Supplementary Fig. 1), displayed significant downregulation of FABP4, PPARγ, and adiponectin mRNAs at both 8 (Fig. 8A) and 15 days of differentiation (data not shown). In contrast, expression of leptin mRNA did not significantly decrease compared with adipocytes not exposed to CM. Interestingly, only CM collected from PHA-activated ASC-MNC cocultures inhibited the insulin response of differentiating ASCs, as assessed by reduction of phosphorylated Akt on Ser⁴⁷³ (Akt-Ser⁴⁷³) in insulin-treated cells (Fig. 8B and C). To investigate the effect of CM from PHA-activated ASC-MNC cocultures on an insulin-dependent metabolic response, we studied the inhibitory effect of insulin on isoproterenol-induced lipolysis (Fig. 8D). We observed that, whereas isoproterenol-induced lipolysis was inhibited by insulin in differentiated ASCs (Fig. 8D, left), such inhibition was lost in differentiating ASCs treated with CM from PHA-activated ASC-MNC cocultures (Fig. 8D, right), indicating that PHA-activated ASC-MNC CM affect both insulin signaling and insulin-dependent metabolic responses.

Far from being a mere repository of fat mass, the AT is perhaps the most plastic organ of the body, capable of meeting the evolutionarily conserved need of storing energy/nutrients during periods of plenty but also of acting as a thermogenic organ (27). Such plasticity, associated with lifestyle changes in nutrition and physical activity in the population worldwide, has led to the current epidemics of obesity (28). Although differentiated adipocytes constitute the major part of the AT mass, precursor ASCs represent five cells for each differentiated adipocyte (29). Infiltrating immune cells are also present within AT and contribute to the inflammatory state observed in obesity (3). Infiltrating macrophages were initially identified as major immune effectors within hypertrophic AT (4), but a role for T cells in AT inflammation has also been described (2,3,30). However, the extent of inflammatory phenomena arising from the interaction between MNCs—which collectively include monocytes and T cells—and ASCs has not been investigated.

Based on our previous work on MSC-mediated inflammation in rheumatoid arthritis (13), we postulated that interactions between ASCs and MNCs contribute to AT inflammation. Here, using ASC-MNC cocultures, we observed that obese- but not lean-derived ASCs polarize PHA-activated MNCs toward a Th17 phenotype, resulting in increased secretion of IL-17A, with a parallel decrease in TNFα and IFNγ Th1 cytokine secretion. Supporting these results on a more physiological level, we observed that AT-derived SVF from obese, but not lean, donors also secreted TNFα upon activation with PHA but at much lower levels than IL-17A (Fig. 2A).

IL-17A secretion was triggered by TCR activation, as treatment of MNC with anti-CD3/CD28-activating antibodies induced levels of IL-17A secretion similar to those of PHA (Fig. 3B). Furthermore, because IL-17A secretion was promoted by both autologous and allogeneic ASCs, this suggests that ASCs might present an antigen to MNCs, independently of HLA expression. This is in accordance with reports showing that 1) human ASCs are devoid of HLA class II expression (21) and 2) AT antigens are preferentially presented to T cells in obese versus lean experimental models (31), with lipid antigens being potential candidates (32). Supporting that ASCs might present an antigen to T cells, we demonstrated that obese, but not lean-derived, ASCs are able to induce IL-17A secretion by MNCs from either lean or obese subjects, whereas obese-derived MNCs did not secrete IL-17A when interacting with ASCs from lean subjects (Fig. 2B). Moreover, ICAM-1 partly mediated MNC-ASC interaction, as assessed by the inhibitory effect of anti-CD54 antibodies. We observed that proinflammatory IL-1β increased in the presence of ASCs and that the monocyte fraction of MNCs was likely responsible for IL-1β secretion. In contrast, IL-6 secretion depended on ASCs. Taken together, these data suggest that AT inflammation might be governed by the cellular interplay of three cell lineages: ASCs, T cells, and monocytes.

To decipher the molecular events leading to ASC-mediated IL-17A secretion, we added anti–IL-1β and anti–IL-6R neutralizing antibodies to ASC-MNC cocultures. We observed that IL-1β, but not IL-6, blockade decreased ASC-mediated IL-17A secretion. Because caspase-1, an enzyme activated by the inflammasome, controls IL-1β secretion through pro–IL-1β cleavage (33,34), we evaluated the requirement for caspase-1 activation in the induction of ASC-mediated IL-17A secretion by blocking the inflammasome with Z-YVAD-fmk, or IL-1β secretion with neutralizing antibodies, and by measuring caspase-1 catalytic activity, which we found increased in PHA-activated ASC-MNC cocultures (Fig. 5D). Z-YVAD-fmk treatment resulted in a more potent inhibition of IL-17A production than IL-1β–neutralizing antibody, suggesting the involvement of other inflammasome-generated proinflammatory cytokines (35).

To better understand the mechanisms by which ASC favor Th17 polarization, we examined PI3K and STAT3 activation, which both contribute to the signaling pathways leading to Th17 cell differentiation (36,37). Wortmannin inhibited both Akt phosphorylation and IL-17A secretion in PHA-activated ASC-MNC cocultures. In addition, as IL-17A secretion was blocked by STAT3 inhibition, and Wortmannin partly inhibited STAT3 phosphorylation (Fig. 6C and D), we suggest that PI3K acts upstream of STAT3 on a signaling pathway leading to IL-17A secretion. We then demonstrated that the STAT3/STAT5 binding ratio on the IL-17A/F gene locus increased in the presence of ASCs. Because Yang et al. (17) reported that IL-2 inhibits IL-17A secretion by increasing STAT5 binding to the IL-17A/F gene locus over STAT3 binding, we measured IL-2 secretion and observed its inhibition in the presence of ASCs, as expected for a Th1 cytokine (Supplementary Fig. 3). We therefore suggest that ASCs promote IL-17A secretion by decreasing IL-2 mediated STAT5 binding on the IL-17A/F gene. To determine the impact of ASC-MNC interactions on the adipocyte biology, we next investigated the impact of cytokine secretion arising from ASC-MNC cocultures on adipocyte differentiation. We observed impairment of adipogenesis mRNA markers in the presence of CM obtained from PHA-activated MNC or ASC-MNC cocultures. This suggests that Th1 and Th17 cytokines are involved in this inhibitory effect, in agreement with reports showing that both TNFα and IL-17A inhibit adipogenesis (12,38). Interestingly, while adipocyte differentiation markers FABP4, PPARγ, and AdipoQ were inhibited, leptin mRNA expression was not impaired, in agreement with the fact that AT inflammation is associated with increased levels of leptin but reduced levels of adiponectin (39) and the demonstration of a positive effect of IL-17A on leptin secretion (40). Moreover, only CM from PHA-activated ASC-MNC cocultures significantly reduced the insulin response in differentiating adipocytes (Fig. 8B and C), which was supported by absence of insulin-mediated lipolysis inhibition (Fig. 8D). Seventy-two hour incubation of differentiating ASC with ASC-MNC CM did not significantly affect adipocyte morphology and lipid droplet accumulation (Supplementary Fig. 1). However, we cannot rule out that longer lasting treatments with ASC-MNC CM would affect adipocyte morphology. Based on our results, and on previous reports showing an impairment of the insulin response in IL-17A–treated 3T3-preadipocytes or human hepatocytes (9,26), we suggest that immune cells infiltrating AT may send a signal to ASC, which then prevents further adipocyte differentiation through promotion of Th17 cells. Because IL-17A has also been shown to increase in peripheral blood and tissues of obese patients (9,10), and to exacerbate inflammation (41), our results suggest that ASCs could contribute to obesity-mediated inflammation through deviation of the Th1 response toward the Th-17 pathway in AT and subsequent propagation of inflammation in the periphery. Our study reinforces the concept that obese-derived ASCs induce peripheral inflammation through Th17 cell promotion (13,42–44).

In conclusion, we demonstrate that besides their well-known immune-regulatory effect on Th1 cells (45), ASCs derived from obese subjects contribute to AT inflammation by promoting Th17 and monocyte activation. Further deciphering this novel cell interplay might allow defining new therapeutic targets to alleviate the inflammatory status of AT in obesity.

Funding. This work was supported in part by INSERM and research grants from Fondation de l’Avenir and Fondation Groupe Chèque Déjeuner.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.E. conceived the project, directed the project, performed experiments, and wrote the manuscript. M.R., G.V., A.-M.M., E.D., and J.R. provided biopsies and biological material. M.C., S.C., C.D., and N.B. performed experiments. M.L., E.L., and H.V. directed the project. L.P. directed the project, performed experiments, and wrote the manuscript. All authors reviewed and commented on the manuscript and approved the final version of the manuscript. A.E. and L.P. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Annual Congress of the Francophone Society of Diabetes, Paris, France, 11–14 March 2014, and at the 50th Annual Meeting of the European Association for the Study of Diabetes, Vienna, Austria, 15–19 September 2014.