Obesity is associated with numerous inflammatory conditions including atherosclerosis, autoimmune disease and cancer. Although the precise mechanisms are unknown, obesity-associated rises in TNF-α, IL-6 and TGF-β are believed to contribute. Here we demonstrate that obesity selectively promotes an expansion of the Th17 T-cell sublineage, a subset with prominent pro-inflammatory roles. T-cells from diet-induced obese mice expand Th17 cell pools and produce progressively more IL-17 than lean littermates in an IL-6-dependent process. The increased Th17 bias was associated with more pronounced autoimmune disease as confirmed in two disease models, EAE and trinitrobenzene sulfonic acid colitis. In both, diet-induced obese mice developed more severe early disease and histopathology with increased IL-17+ T-cell pools in target tissues. The well-described association of obesity with inflammatory and autoimmune disease is mechanistically linked to a Th17 bias.
Epidemiological evidence has linked obesity to pro-inflammatory conditions such as cancer, asthma and autoimmune diseases 1, 2. The link with cancer is particularly strong, with obesity emerging as a premier risk factor. Obesity predisposes to several (but not all) autoimmune disorders, including inflammatory bowel disease (IBD) and psoriasis 3, 4. The Th17 T-cell sublineage plays core roles in IBD and psoriasis, and recent evidence demonstrates a pathogenic role for the Th17 inducer, IL-23, in solid tumor animal models 5.
Diet-induced obese (DIO) mice and obese humans have elevated serum and tissue levels of IL-6 6, 7. Adipocytes and tissue-derived macrophages both contribute significantly to IL-6 expression, estimated at ∼33 and ∼20%, respectively 8. IL-6 signaling through STAT-3, along with retinoic orphan receptor α- and γt transcription factors, promote Th17 lineage expansion, establishing an inter-organ system communication 9. We further explored the functional sequelae of the potential connection between obesity and Th17 expansion.
We first compared the proportions of CD4+ Th17 cells in non-immunized spleen cells from DIO and regular diet (RD) mice (Fig. 1A, Supporting Information Fig. 1A). There were approximately 3× more IL-17 secreting CD4+ systemic T cells in DIO mice compared with age-matched, naive mice on RD. This expansion was Th17 sublineage selective, for example, DIO neither affects systemic pool sizes of CD4+, IFN-γ+(Th1), nor Foxp3+regulatory T cells (Treg) (Fig. 1B, Supporting Information Fig. 1B and C), and GATA3+ Th2 cell pools were not affected by obesity (Supporting Information Fig. 2A). The enlarged Th17 pools in DIO mice were functional, with enhanced IL-17 production after in vitro stimulation with anti-CD3 plus anti-CD28, while IFN-γ (Fig. 1C) and Th2 cytokine secretion (Supporting Information Fig. 2B) were similar in cultures of DIO and RD splenocytes.
DIO thus generates an immunological phenotype with selective expansion of the Th17 sublineage. As is well established 9, 10, frequencies of Th17 cells were small and we therefore determined whether this Th17 sublineage bias is maintained during active T-cell priming and expansion. We measured Th17 and Th1 cell populations in draining lymph nodes following immunization of DIO and RD mice with a well-characterized H-2b-binding peptide, myelin oligodendrocyte glycoprotein 35–55 (MOG35-55) 11. Six days after immunization, lymphocytes from draining lymph nodes of DIO donors generated a dramatically increased pool size of CD4+ IL-17+ cells, compared with lymph node cells from RD mice (Fig. 1D). In contrast, the CD4+ IFN-γ+ T cell compartment was unaffected by DIO.
Th17 development can proceed through both IL-6-dependent and IL-6-independent pathways, the latter driven by IL-21 and TGF-β 12, 13. We placed IL-6null mice on high fat diet and immunized with MOG35-55 as above (Fig. 1D). Although DIO IL-6null mice gained weight at similar rates as DIO WT mice (Supporting Information Fig. 3), DIO IL-6null mice did not at all develop the Th17 bias of DIO WT mice. These results indicate that obesity-induced Th17 bias utilizes the IL-6-dependent pathway of Th17 development.
DIO predisposes mice to severe, chronic trinitrobenzene sulfonic acid colitis
We analyzed the clinical impact of DIO in two Th17-dependent disease models, trinitrobenzene sulfonic acid (TNBS) colitis and EAE 10. Depending on the mouse facility (e.g. gut flora), C57BL/6 (B6) mice are fairly resistant 14 to TNBS colitis and many animals recover within a week after disease induction. In contrast, DIO rendered B6 mice highly susceptible to TNBS colitis (Fig. 2). TNBS treated, DIO but not RD mice, demonstrated progressive weight loss throughout the observation period (Fig. 2A, B). DIO significantly accelerated the early disease stage, with 50% lethality in the first week (Fig. 2C, p<0.03, Life table), although lethality in RD and DIO mice was ultimately similar. Weight loss continued only in surviving DIO animals, due to more severe colitis, both macroscopically (Fig. 2D) and histologically (Fig. 2E, F) 10 days after disease induction. DIO colons showed obvious thickening, fibrosis and shortening compared with RD mice, with enhanced ulceration, loss of mucosal surface and extensive transmural colon inflammation (Fig. 2D, F). Consistent with increased DIO disease severity, isolated T cells from spleen, draining mesenteric lymph nodes and lamina propria (LP) of TNBS colitis DIO mice all produced significantly more IL-17 than the corresponding T cells of RD mice (Fig. 2G).
DIO exacerbates EAE
Obesity-dependent Th17 expansion was thus associated with progressively enhanced inflammatory and autoimmune tissue lesions. To test the generality of this conclusion, we compared EAE in DIO and RD mice. Lean and obese B6 mice, immunized with MOG35-55 peptide, both showed first clinical symptoms 10 days later (Fig. 3A), DIO mice more rapidly progressed to severe disease, with about half-reaching disease scores of >3 by day 13, which took 20 days in RD mice (Fig. 3B). DIO worsened spinal cord inflammation with more pronounced inflammatory cell penetration in DIO white matter (Fig. 3C and D). T-cell recall responses to MOG35-55 in EAE mice did not differ between DIO and RD mice, suggesting that immunization was equally effective (Fig. 3E). However, as in the TNBS model, T cells from DIO mice, selectively and significantly, had higher IL-17 production than RD mice in spleen, draining lymph nodes as well as the CNS target tissue (Fig. 3F).
As the incidence of obesity increases worldwide, so will its complications and associated disorders. There is an established link between obesity and autoimmune/pro-inflammatory diseases including psoriasis, IBD and asthma, all containing a significant Th17 component 3, 4, 15, 16. There is scant information on a conceivable linkage between obesity and multiple sclerosis, but one study showed reduced disease deterioration in patients on low fat compared with high fat diet 17. Although EAE is not an ideal model for MS, data presented here suggest that this association is worth investigating, in particular, since a role for Th17 effectors in the penetration of the blood–brain barrier as well as neuronal pathology has been reported 18.
In addition to autoimmune diseases, obesity has been associated with risk to develop neoplastic disorders 2. Recently, it was shown that IL-23 and its downstream Th17 effector cells may promote cancer through inhibition of anti-tumor CD8+ T cells and through enhanced production of tumor promoting MMP9 and angio-neogenesis 5, 19.
Th17 development requires critical signals that lead to increased intracellular levels of STAT3. Although several cytokines can promote Th17 expansion 10, we here demonstrate that obesity-associated Th17 expansion is IL-6 dependent. Other cytokines may further amplify this role of IL-6, indeed, serum amyloid A, an acute-phase protein highly elevated in obesity, has been shown to increase dendritic cell IL-23 production in vitro20.
Collectively, our observations demonstrate that DIO predisposes to IL-6-dependent Th17 expansion. DIO appeared not to accelerate initiation, but rather exacerbate early disease progression. The high prevalence of obese patients with spontaneous pro-inflammatory disorders may indicate an additional role of the Th17 axis in disease initiation, not apparent in our experimentally induced disease models. Future experiments will determine whether the increase of Th17 effector cell frequency in DIO mice causes autoimmune exacerbation directly or through involvement of other effector cells/mechanisms that might identify new therapeutic targets. DIO is reversible and it should be interesting to assess, to what extent and at which disease stage weight loss can reduce Th17 expansion and disease exacerbation, information that might have clinical impact. While this present manuscript was in submission, first observations from a small clinical study demonstrated elevated IL-17 serum levels in obese patients 21, suggesting that our observations in rodents are likely relevant to human disease. IL-6-dependent Th17 expansion is a clinically prominent element of pro-inflammatory diseases in obesity.
Materials and methods
All experiments were performed in male C57BL/6J (B6) mice. WT animals were purchased from Jackson Laboratories (Bar Harbor, ME, USA) (http://www.jax.org/) and maintained in our vivarium in a pathogen-free, temperature controlled, 12 h light and dark cycle environment. Animals were fed either a RD or a high fat (60 kcal% fat) diet (Research Diets, New Brunswick, NJ, USA). DIO mice received RD for the first 6 wk of life and then high fat diet for the following 10 wk. In EAE and TNBS colitis experiments, 16 wk-old DIO mice were maintained on high fat diet throughout the experiments. All studies reported here used males under approved protocols and in agreement with animal ethics guidelines.
Induction of EAE and TNBS colitis
EAE was induced in 16 wk-old male B6 mice. Briefly, mice were immunized in each flank with 100 μg of MOG35-55 (Alpha diagnostic, San Antonio, TX, USA) emulsified in CFA (1:1) (Sigma-Aldrich, Oakville, ON, USA). In total 200 ng of pertussis toxin (Sigma-Aldrich) was given i.p. on the day of EAE induction and 48 h later. Animals were followed for at least 25 days and disease was scored using the following scale: 0, asymptomatic; 1, limp tail; 2, abnormal righting reflex and/or hind limb weakness; 3, unilateral hind limb paralysis; 4, bilateral hind limb paralysis and 5, moribund or death.
TNBS colitis was induced in 16 wk-old B6 males. A final concentration of 2.5% TNBS (Sigma-Aldrich, 50% ethanol) was administered per rectum, using a 4F catheter. The catheter tip was inserted 4 cm and 150 μL of TNBS was injected slowly into the colon while pulling out the catheter. The mouse was held in the vertical position (30 s) after TNBS administration. Animals were weighed daily over the next 10 days.
Histological analysis of the colon and CNS
Spinal cords and brains were removed 25 days after EAE induction, and colons were removed 10 days after TNBS administration. Both were fixed (24 h, 10% buffered formalin), and then stained with H&E. Histology was scored by two blinded observers and the average score was used. The following scoring system for spinal cord histology was employed: 0, unremarkable; 1, focal mononuclear infiltration; 2, mononuclear infiltration in<10% of white matter; 3, mononuclear infiltration in 10–20% of white matter and 4, infiltration in >20% white matter.
For colon histology, the following scoring system was used: 0, no evidence of inflammation; 1, lymphocyte infiltration<10% of high power field (hpf) with no structural changes; 2, lymphocytic infiltration 10–25% of hpf and minor structural changes including crypt elongation, mucosal thickening but no ulceration; 3, lymphocyte infiltration in 25–50% of hpf with bowel thickening extending beyond the mucosal surface; and 4, lymphocyte infiltration in >50% of hpf with major structural changes including crypt distortion, transmural bowel wall thickening and ulceration.
Isolation of splenic, LP and CNS mononuclear cells
Spleen cells were isolated as described 22. CNS and LP mononuclear cells were isolated 25 days after EAE or 10 days after colitis induction. Brains and spinal cords were removed and minced in a Stomacher blender. The suspension was then incubated with 1 mg/mL collagenase (Sigma-Aldrich, 1 h/37°C) with gentle resuspension every 5 min. The suspensions were then pelleted and suspended in 4 mL Percoll (Sigma-Aldrich, 30%) and centrifuged at 1500 rpm/15 min over 4 mL 70% Percoll. The mononuclear cells at the 30/70% Percoll interface were collected and washed 2× in serum-free HL-1 medium, supplemented with 2 mM L-glutamine (Lonza, Walkersville, MD, USA).
LP cells: colons were removed and washed in calcium- and magnesium-free HBSS (Gibco, Burlington, ON, USA) and cut into small pieces<0.5 cm and incubated twice (20 min/37°C) in HBSS, 2 mM DTT, 1 mM EDTA (Sigma-Aldrich), with manual shaking every 5 min. The remaining tissue was then digested with 400 U/mL collagenase in complete media (1 h/37°C), with manual shaking every 5 min. LP mononuclear cells were purified by centrifugation through a discontinuous Percoll gradient of 40/70%, washed 2× in serum-free HL-1 medium, supplemented with 2 mM L-glutamine+antibiotics.
T-cell proliferation and cytokine secretion
In all EAE experiments, proliferation or cytokines were measured following 72 h of stimulation with 10 μg/mL MOG35-55. In TNBS colitis experiments, cytokines were measured following 72 h of incubation with plate bound αCD3 (1 μg/mL) and αCD28 (0.25 μg/mL). Splenocytes (4×105/well) were incubated (72 h/37°C) in 96-well plates pre-coated with αCD3 (1 μg/mL) and αCD28 (0.25 μg/mL) or varying concentrations of MOG35-55 peptide. In proliferation experiments, 1 μCi of [H3]Thymidine was added for the last 18 h prior harvesting liquid scintillation counting. Alternatively, supernatants were collected after 72 h of culture and IFN-γ (BD Biosciences, San Jose, CA, USA) and IL-17 (R&D Systems, Minneapolis, MN, USA) were measured by ELISA according to the manufacturer's protocols.
Splenocytes and/or lymph node cells were incubated (15 min/4°C) with 10 μg/mL Fc-blocker (eBioscience) and then stained for 30 min with the following dilutions and conjugated fluorochromes of a given antibody: CD4-PE (1/200), IL-17-APC (1/150), IFN-γ-APC and FOXP3-APC (1/100), (eBioscience). For intracellular IL-17 or IFN-γ staining, all cells were incubated with PMA (50 ng/mL) and ionomycin (750 ng/mL) for 6 h in HL-1 media at 37°C and golgi blocked for the last 3 h (Golgistop, BD Bioscience, San Diego, CA, USA). Flow cytometric data were analyzed using Flowjo software.
Statistical significance between two means was assessed by Mann–Whitney and unpaired t-tests. Welch correction on t-tests was employed for sample sizes<6. Comparisons of curves were drawn using two-way ANOVA or life tables. In total 2×2 tables were analyzed with Fisher's Exact test. Statistical significance was two tailed and set at 5%, all error bars are single SD.
The authors appreciate L. Morikawa for excellent assistance with histopathology, and L. Han for assistance with cytokine studies. This study was funded by CIHR, GP is recipient of a Banting & Best fellowship award.
Conflict of interest: HMD holds shares in Afference Therapeutics Inc., a start-up R&D company (Toronto).