Article Text

PDF

Extended report
Differential impact of obesity on the pathogenesis of RA or preclinical models is contingent on the disease status
  1. Seung-jae Kim1,2,
  2. Zhenlong Chen1,2,
  3. Abdul B Essani1,2,
  4. Hatem A Elshabrawy1,2,
  5. Michael V Volin3,
  6. Giamila Fantuzzi4,
  7. Iain B McInnes5,
  8. Joshua F Baker6,7,
  9. Patricia Finn8,
  10. George Kondos9,
  11. Suncica Volkov2,
  12. William Swedler1,2,
  13. Shiva Arami1,2,
  14. Nadera Sweiss1,
  15. Shiva Shahrara1,2
  1. 1Division of Rheumatology, Jesse Brown VA Medical Center, Chicago, Illinois, USA
  2. 2Division of Rheumatology, Department of Medicine, University of Illinois at Chicago, Chicago, Illinois, USA
  3. 3Department of Microbiology & Immunology, Midwestern University, Downers Grove, Illinois, USA
  4. 4Department of Kinesiology and Nutrition, University of Illinois at Chicago, Chicago, Illinois, USA
  5. 5Institute of Infection, Immunity and Inflammation, University of Glasgow, Glasgow, UK
  6. 6Philadelphia VA Medical Center, Philadelphia, Pennsylvania, USA
  7. 7Division of Rheumatology, Department of Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania, USA
  8. 8Division of Pulmonary, Division of Medicine, University of Illinois at Chicago, Chicago, Illinois, USA
  9. 9Division of Cardiology, Division of Medicine, University of Illinois at Chicago, Chicago, Illinois, USA
  1. Correspondence to Dr Shiva Shahrara, University of Illinois at Chicago Department of Medicine, Division of Rheumatology 840 S. Woods Street, building CSB Room 1113/1114 Chicago, IL 60612, USA; shahrara{at}uic.edu

Abstract

Objective Studies were performed to uncover the significance of obesity in rheumatoid arthritis (RA) and preclinical models.

Methods Preclinical arthritis models were used to examine the impact of obesity on disease onset and remission. Conditioned media from RA adipose tissues were used to investigate the mechanism contributing to joint neutrophil influx and M1 macrophage differentiation observed in early and remission phases of arthritis.

Results We report that mice fed with high fat diet (HFD) have an earlier onset of collagen-induced arthritis (CIA) compared with mice on regular diet. However, the differences in CIA joint swelling between the two diet groups are lost once disease is established. We found that early arthritis triggered by obesity is due to elevated joint MIP2/interleukin-8 levels detected in CIA as well as in the RA and mouse adipose tissues and the effect of this chemokine on neutrophil recruitment. Although active disease progression is similarly affected in both diet groups, arthritis resolution is accelerated in lean mice while joint inflammation is sustained in obese mice. We document that HFD can prolong toll-like receptor (TLR)4-induced arthritis by increasing joint monocyte migration and further remodelling the recruited cells into M1 macrophages. Consistently, we show that adipose condition media can transform RA and wild-type naïve myeloid cells into M1 macrophages; however, this function is impaired by TLR4 blockade or deficiency.

Conclusions We conclude that despite established disease being unaffected by obesity, the early and the resolution phases of RA are impacted by obesity through different mechanisms.

  • Rheumatoid Arthritis
  • Chemokines
  • Inflammation
  • Cytokines
  • Early Rheumatoid Arthritis

Statistics from Altmetric.com

Rheumatoid arthritis (RA) is a chronic autoimmune disorder in which imbalanced leucocyte migration contributes to joint inflammation and bone erosion.1 A number of epidemiological studies have demonstrated an increased incidence of RA among obese individuals.2–4 In addition, obesity has been associated with elevated disease severity and inferior response to treatment in some clinical studies.5–7 However, there are little to no experimental data to explain the mechanisms that might be at play in these epidemiological studies.

Patients with RA have markedly higher prevalence of metabolic syndrome compared with the general population, yet the reasons are undefined.8 Obesity is known to be associated with metabolic syndrome components and elevated body mass index (BMI) is linked to RA onset.9 In RA, adipocytes and their surrounding macrophages in white adipose tissues produce a number of adipokines that modulate systemic inflammation.10 Interestingly, the adipokine, leptin, participates in innate immunity by increasing the T helper (TH)-1 cell polarisation and suppressing the T regulatory cell differentiation.11 Recent studies reveal that joint inflammation is aggravated in collagen-induced arthritis (CIA) mice fed on high fat diet (HFD) compared with regular diet (RD) due to increase in spleen TH-17 cell development.12

Others have demonstrated that both in human and mice number of macrophages recruited into adipose tissue closely correlates with BMI and adipose size and that the expression analysis of myeloid cells validates its critical role in the inflammatory process of obesity.13 Moreover, it was shown that adipose-associated macrophages in non-obese humans exhibit anti-inflammatory phenotype, while myeloid cells recruited into adipose tissue of obese individuals have a pro-inflammatory characteristic.14

Studies were performed to understand the reason why increased BMI contributes to higher risk for RA development15 and to elucidate the underlying mechanism for inferior response to antitumour necrosis factor (TNF) therapy in obese compared with normal weight patients with RA.5–7 ,16 We uncover that neutrophil (interleukin (IL)-8/MIP2, CXCL1, CXCL5) and monocyte (IL-6, IL-1β, CCL2) chemoattractants are elevated in conditioned media obtained from RA and obese mice adipose tissues. We next establish that obesity affects RA inflammatory process in a distinct manner, during the early onset and the late remission phases. Employing RA specimen and preclinical arthritis models, we found that early joint inflammation is potentiated by obesity through IL-8/MIP2-induced neutrophil migration. While in the later stage of the disease, obesity sustains arthritis by remodelling the newly recruited naïve myeloid cells into pro-inflammatory M1 macrophages, in part via toll-like receptor (TLR)4-dependent cascade. In conclusion, this novel study reveals for the first time the mechanism by which obesity initiates RA onset and further identifies the components that may be responsible for worse anti-TNF therapy response in obese patients with RA.

Materials and methods

Study protocol for animal models

DBA/1J mice were fed a RD (19% protein) or 60% HFD (23.1% protein; Research Diet, New Jersey, USA) beginning at the age of 4 weeks. Subsequently, mice aged 8 weeks were immunised with collagen type II (Chondrex) on days 0 and 21.17–19 Mice were sacrificed on days 29, 55 or 106 when they had been on RD or HFD for 8, 12 or 19 weeks.

C57BL6 mice were fed for 10 weeks on RD or 60% HFD prior to being intra-articularly (i.a.) injected with 10 µg of lipopolysaccharide (LPS) (0111:B4) for 72 hours. Ankle circumference was measured by Caliper.19–21

Adipose tissue conditioned media

Conditioned media was generated by incubating adipose tissue obtained from RA (n=14) or osteoarthritis (OA) synovia (n=7) or visceral fat from 10-week fed HFD mouse (n=6) in media and filtered before use. RA was classified according to the 1987 Revised Criteria22 and the OA knee, hand and hip met the classification of 1986, 1990 and 1991.23–25

Cytokine quantification

Human and mouse TNF-α, IL-6, IL-1β, IL-17, interferon (IFN)-γ, CCL2, CCL5, CXCL5, CXCL1, IL-8/MIP2, CCL3 and CCL20 (R&D Systems) were quantified by ELISA in RA and mouse adipose conditioned media as well as in CIA and LPS-induced arthritis ankle homogenates.

Evaluation of adiposity and fatty liver

Lean and fat mass were calculated by the ratio of body composition and weight.26 Formalin-fixed liver samples were stained with H&E and fatty liver was scored on a 0–4 scale.27

Flow cytometry analysis

To quantify CD3, F4/80, dual F4/80 and CD80, B220-positive cells in CIA, splenocytes were harvested on day 55. Thereafter, cells were stained with allophycocyanin (APC)-conjugated anti-CD3, phycoerythrin (PE)-labelled F4/80 or dual PE-labelled F4/80 and APC-conjugated CD80 as well as fluorescein isothiocyanate (FITC)-labelled B220 (eBioscience). The frequency of TH-1 or TH-17 was quantified in CIA splenocytes harvested on day 55 and was stimulated with phorbol myristate acetate (PMA) (5 ng/mL) and ionomycin (500 ng/mL) in the presence of Brefeldin A (3 μg/mL) for 4 hours. Cells were stained with APC-conjugated anti-CD4 antibodies (Abs) (eBioscience). Percentage of TH-1 and TH-17 cells was determined by staining the splenocytes with FITC-conjugated anti-IFN-γ or PE-labelled anti-IL-17 Abs (eBioscience). RA peripheral blood (PB) monocytes19 ,28 or 7 days macrophage colony stimulating factor (M-CSF)-driven mouse bone marrow myeloid cells were either untreated or treated with IFN-γ±LPS, 10% RA or OA adipose conditioned media or 10%–50% mouse adipose conditioned media for 24 hours prior to determining % PE-conjugated CD127 or APC-labelled CD80-positive cells.

Neutrophil isolation and chemotaxis

Polymorphonuclear cells were isolated from blood29 or from mouse bone marrow progenitor cells.30 Human neutrophils were isolated using Histopaque gradient centrifugation followed by 3% dextran separation.29 Mouse neutrophils were isolated from bone marrow cell suspensions that were laid on a three-layer Percoll gradient of 78%, 69% and 52% (Sigma).30 Neutrophil chemotaxis was performed in a Boyden chamber (Neuroprobe) using a 5 μm membrane. Formylmethionine-leucyl-phenylalanine (fMLP) (10 nM) was used as positive control and phosphate-buffered saline (PBS) was used as negative control.31

To demonstrate that RA adipose conditioned media contributes to neutrophil migration different concentrations (5, 10 and 20%) were used in neutrophil chemotaxis. To identify the chemokine responsible for neutrophil migration; neutrophil chemotaxis was tested in response to 10% RA and mouse adipose media that were treated with 10 μg/mL of sham Ab or Abs to IL-8/MIP2, CXCL1 and CXCL5/LIX or IL-6.

Western blot analysis

RA and mouse adipose conditioned media were either untreated or pretreated with IgG as well as Abs to IL-8/MIP2, CXCL1, CXCL5 and/or IL-6 (R&D systems) for 1 hour. Neutrophils from human PB or mouse bone marrow were untreated (PBS) or stimulated with either fMLP (10 nM), 10% RA (2 min) or mouse conditioned adipose media (5 min) after being pretreated with IgG or Abs to IL-8/MIP2, CXCL1, CXCL5 and/or IL-6. Cell lysates were examined by western blot analysis for phospho (p)-p38 (Cell Signaling; 1:1000) or for p38 (1:3000). Ankle homogenates harvested from TLR4-induced arthritis fed on RD or HFD were western blotted for inducible nitric oxide synthase (iNOS) and arginase I expression (1:1000, Santa Cruz Biotechnology) or actin equal loading (1:3000).

Real-time RT-PCR

Total RNA was extracted using TRIzol from RA monocytes19 ,28 or 7-day M-CSF (20 ng/mL)-driven mouse bone marrow myeloid cells that were either untreated (PBS) or treated with IFN-γ±LPS (100 ng/mL each), 10%–50% de-identified RA and/or OA adipose conditioned media or 10%–50% mouse adipose conditioned media for 6 hours. In a different experiment, 7-day M-CSF-driven mouse bone marrow myeloid cells were either untreated or pretreated for 1 hour with 10 μg/mL IgG, anti-IL-6 (eBioscience), anti-MIP2 (R&D Systems), anti-TLR2 (Invivogen) or anti-TLR4 (eBioscience) Abs prior to being stimulated with 10% mouse conditioned adipose media. C57BL6 wild type (WT) and TLR4−/− (Jackson lab) mouse bone marrow cells cultured 7 days in M-CSF (20 ng/mL) were either untreated (PBS) or treated with IFN-γ (100 ng/mL), ultrapure LPS (100 ng/mL, binds only to TLR4) or 10% mouse conditioned adipose media for 6 hours. Further, 7-day M-CSF differentiated WT mouse bone cells were either untreated or pretreated with TLR4 antagonist (1 μg/mL; Invivogen) for 3 hours prior to stimulating the cells with PBS, LPS (10 ng/mL, binds to TLR2 and 4), ultrapure LPS (10 ng/mL) or 10% mouse conditioned media for 6 hours.

Abs and immunohistochemical analysis

Ankles were formalin fixed (10%), paraffin embedded and sectioned. Inflammation, synovial lining and bone erosion were determined using H&E-stained sections on a 0–5 scale.18–21 Mouse ankles were immunoperoxidase-stained with diaminobenzidine as a chromogen. CIA or TLR4-induced arthritis mice fed on RD or HFD were stained with GR1 (1:50; eBioscience), F4/80 (1:100; Serotec), iNOS and arginase I (1:200; Santa Cruz Biotechnology) or isotype IgG Ab. Positive immunostaining was scored on a 0–5 scale by two masked observers.18–21

Statistical analysis

One-way analysis of variance was used for comparisons among multiple groups, followed by Student's post hoc two-tailed t-test. Student's paired and unpaired two-tailed t-tests were used for comparisons between two groups. p Values <0.05 were considered to be significant.

Results

HFD contributes to early CIA onset by provoking neutrophil migration through MIP2/IL-8

To examine the effect of obesity on RA, CIA mice were fed for 8 weeks on HFD or RD. Interestingly, CIA mice fed on HFD had a significantly earlier disease onset on days 28 and 30 (figure 1A). Consistent with the clinical findings, histological analysis of CIA mice sacrificed on day 29, revealed that while there were early signs of joint inflammation and lining thickness detected in HFD group, no clinical manifestations were found in mice fed on RD (figure 1B, C). We document that the early CIA onset observed in the HFD compared with RD is due to a twofold increase in MIP2 protein levels that can markedly potentiate joint neutrophil infiltration (figure 1D–F). In contrast, joint TNF, IL-6, IL-1 and IL-17 levels were comparable in RD and HFD mice on day 29 post-CIA induction (figure 1F). Interestingly, we establish that the weight gain positively correlates with increased joint inflammation in CIA mice on HFD but not in mice on RD (see online supplementary figures S2–S4). These results suggest that in obese CIA mice, early disease is initiated by neutrophils infiltrating into the joint in response to MIP2.

Figure 1

Obesity impacts collagen-induced arthritis (CIA) onset but not active phase by enhancing joint MIP2-induced neutrophil migration. (A) Joint circumference were calculated in CIA mice that were fed on regular diet (RD) (n=13 mice) or 60% high fat diet (HFD) (n=12 mice) for 8 weeks. (B) CIA ankles harvested on day 29 from mice fed on RD and HFD were H&E stained (orig. mag. ×200) and (C) inflammation (inflam), lining thickness (lining) and bone erosion (erosion) were quantified on a 0–5 scale, n=5 mice, 10 back paws. Synovial tissues (STs) obtained from the back paws of day 29 CIA mice fed on RD and HFD were stained with anti-GR1 antibody (orig. mag. ×200) (D) and GR1 staining was quantified on a 0–5 scale, (E) n=8. (F) Joint MIP2, tumour necrosis factor (TNF), interleukin (IL)-6, IL-1 and IL-17 protein levels were assessed by ELISA from the front (carpal joints) and back paws of day 29 CIA mice fed on RD and HFD and normalised by joint weight, n=8. (G) Joint circumference is shown from day 31 to day 55 in CIA mice that were fed on RD or HFD for up to 12 weeks, n=12–13 mice. (H) CIA ankles harvested on day 55 from mice fed on RD and HFD were H&E stained (orig. mag. ×200) and (I) inflammation (inflam), lining thickness (lining) and bone erosion (erosion) were quantified on a 0–5 scale, n=8. (J) % lean and fat mass or H&E staining of liver samples from day 55 CIA mice on RD and HFD were measured by dual-energy X-ray absorptiometry or scored for fatty liver on a 0–4 scale, n=12–13. (K) The frequency of CD3, B220, F480, F480+CD80+ as well as TH-1 and TH-17 cells was assessed in splenocytes harvested on day 55 from CIA mice fed on RD and HFD, n=7. (L) Changes in joint MIP2, TNF, IL-6, IL-1 and IL-17 were determined by ELISA from day 55 CIA mice fed on RD and HFD, n=7. Values are mean±SE. *Indicates p<0.05. IFN, interferon.

HFD does not affect CIA effector phase

To determine the effect of obesity on the effector phase of CIA, mice were fed with HFD or RD for 12 or 19 weeks and sacrificed on days 55 and 106. Interestingly, as disease progresses beyond day 30, joint inflammation differences were lost (figures 1G and 2A) between the obese mice that had greater fat mass, fatty liver and weight (figure 1J, see online supplementary figure S1) compared with the lean mice harvested on days 55 and 106. Consistent with the clinical data, histological findings demonstrate that joint inflammation, lining thickness and bone erosion were comparable in the CIA lean and obese mice harvested on days 55 and 106 (figures 1H, I and 2B, C). We further uncovered that the frequency of the CIA leucocytes (CD3, B220, F480) and T cell population (TH-1, TH-17) as well as levels of joint monocyte (TNF, IL-1, IL-6, IL-17) and neutrophil chemoattractants (MIP2) were unaffected by obesity-induced diet in mice sacrificed on day 55 (figure 1K, L). Interestingly, when studies were extended for 106 days, CIA disease activity did not resolve and there were no differences detected in joint monocyte infiltration, M1 or M2 macrophage differentiation as well as joint TNF, IL-6, IFN-γ, IL-17 and MIP2 protein levels in mice fed with HFD compared with RD (figure 2D–F). We conclude that while CIA disease initiation is exacerbated by HFD, established active disease is unaffected by this process.

Figure 2

Obesity does not impact the late stage of collagen-induced arthritis (CIA). (A) Joint circumference was determined in CIA mice that were fed on regular diet (RD) (n=9) or high fat diet (HFD) (n=8) for 19 weeks. (B) CIA ankles harvested on day 106 from mice fed on RD and HFD were H&E stained (orig. mag. ×200), and (C) inflammation (inflam), lining thickness (lining) and bone erosion (erosion) were quantified on a 0–5 scale, n=8. (D) Synovial tissues harvested on day 106 from CIA mice on RD and HFD were stained with anti-F480, anti-inducible nitric oxide synthase (iNOS) and anti-arginase antibodies (orig. mag. ×200) and (E) positive staining was quantified on a 0–5 scale, n=8. (F) Joint tumour necrosis factor (TNF), interleukin (IL)-6, interferon (IFN)-γ, IL-17 and MIP2 protein levels were quantified from day 106 CIA mice fed on RD and HFD by ELISA and normalised to joint weight, n=8. Values are mean±SE. *Indicates p<0.05.

In RA and mouse adipose tissues, neutrophil infiltration is fostered via IL-8/MIP2

To determine whether findings in CIA are representative of RA pathogenesis, inflammatory factors were quantified in the adipose conditioned media extracted from RA synovial tissue (ST) and obese mice abdominal adipose tissue. We found that IL-8/MIP2 (47 and 14 ng/mL) levels exceeded all other inflammatory factors. In addition, other neutrophil chemoattractants such as CXCL1 (2.6 and 4.7 ng/mL) and CXCL5 (3.4 and 9.6 ng/mL) were also highly expressed in RA and mouse adipose conditioned media (figure 3A). In contrast, CCL2 (21 and 0.5 ng/mL) and IL-6 (7 and 6.9 ng/mL) were the only monokines that were abundantly detected and IL-1β levels (0.05 and 1.2 ng/mL) were low, while TNF (not detectable (ND) and 0.5 ng/mL) and lymphokine IL-17 (ND and 0.4 ng/mL) levels were either undetectable or sparse in RA and mouse adipose conditioned media (figure 3A). Based on the high levels of neutrophil chemokines in the conditioned media, we asked whether RA adipose media could contribute to neutrophil migration. Our results demonstrate that 10% RA adipose media has the greatest neutrophil chemoattractant ability and this is primarily due to IL-8 but not CXCL1, CXCL5 or IL-6 (negative control) function (figure 3B, C). Confirming this notion, we show that similar to the positive control, fMLP, RA adipose media promotes p38 activation and that IL-8 neutralisation can explicitly negate this function (figure 3D). Consistent with the results in RA, blockade of MIP2 (IL-8 analogue) function abrogates mouse adipose media driven neutrophil chemotaxis through a p38-dependent mechanism (60%–66% respectively) (figure 3E, F). In contrast, neutralisation of CXCL1 and CXCL5 was ineffective in this process (figure 3E, F). These results suggest that in obese individuals, onset of RA can be triggered by joint neutrophil recruitment in response to IL-8/MIP2-induced p38 activation.

Figure 3

Elevated MIP2 detected in the conditioned media from rheumatoid arthritis (RA) and mouse adipose tissues is associated with higher neutrophil migration via p38 activation. (A) Conditioned media was generated from RA synovium (n=14) and abdominal fat from 10-week fed high fat diet (HFD) mouse (n=6) and the pro-inflammatory factors were assessed by ELISA. (B) Peripheral blood (PB) neutrophil chemotaxis was performed in response to 5%–20% RA adipose conditioned media, n=3. PB neutrophils was sham depleted or neutralised with anti-interleukin (IL)-6, anti-IL-8, anti-CXCL1 or anti-CXCL5 antibodies (Abs) prior to examining (C) chemotaxis or performing (D) western blot and densitometric analysis of the bands containing cell lysates that are probed for p38 phosphorylation in response to 10% RA adipose media, n=3–5. (E) Neutrophils extracted from mouse bone marrow were either sham depleted or neutralised with anti-MIP2, anti-CXCL1 or anti-CXCL5 Abs prior to performing chemotaxis, n=4 or (F) densitometric analysis of the western blot bands for p38 phoshorylation, in response to 10% mouse adipose media, n=4. In B–F, phosphate-buffered saline (PBS) and fMLP (10 nM) were used as negative and positive controls. Values are mean±SE. *Indicates p<0.05. TNF, tumour necrosis factor.

Obesity sustains joint inflammation through M1 macrophage polarisation

To further investigate the inhibitory effect of obesity on arthritis resolution, we chose to use the TLR4-induced arthritis model. This model was selected due to rapid arthritis resolution observed 72 hours after LPS injection; in contrast to CIA in which ankles remain inflamed 106 days postinitiation (figure 2A).32 ,33 We found that in the first 48 hours, TLR4-driven joint inflammation progresses similarly in the obese and lean mice, thereafter while arthritis resolves in the lean mice, ankle swelling is sustained in the obese mice at 72 and 84 hours post-LPS injection (figure 4A and see online supplementary figure S7). Histological analysis of tissues harvested at 72 hours confirm that mice on HFD have markedly greater joint inflammation and lining thickness with no difference in bone erosion compared with the RD group (figure 4B–D). Although monocyte chemoattractants such as TNF-α (fourfold), CCL20 (threefold) and CCL5 (twofold) were significantly upregulated in the obese arthritic mice, there were no differences detected in IL-1β, CCL3 and CCL2 levels compared with lean mice (figure 4E, F). On the contrary, mice on HFD and RD expressed similar levels of neutrophil chemokines, CXCL1 and CXCL5, at 72 hours post-LPS injection (figure 4F). Corroborating with the higher levels of monocyte chemokines detected in the obese mice following LPS injection, we found that joint monocyte extravasation was potentiated in these HFD arthritic mice compared with RD group (figure 5A, B). To better understand how obesity prolongs arthritis and inhibits disease remission, joint myeloid cell phenotype was assessed in the obese and the lean arthritic mice. We uncovered that the obese arthritic mice, predominantly express iNOS+ M1 macrophages; while iNOS+ cells are reduced and arginase+ M2 macrophages are upregulated in the lean mice (figure 5C–E). Our results conclude that obesity can sustain arthritis by reconstructing the newly recruited joint myeloid cells into pro-inflammatory M1 macrophages.

Figure 4

The obese arthritic mice demonstrate delayed remission compared with their counterpart controls. (A) Ankle circumference was determined in wild-type mice that were fed on regular diet (RD) or 60% high fat diet (HFD) for 10 weeks following intra-articularly (i.a.) injection with 10 µg of lipopolysaccharide (LPS) and mice were sacrificed after 72 hours of LPS injection, n=10. (B) RD and HFD ankles harvested after 72 hours of LPS injection were H&E stained (orig. mag. ×200) and (C) inflammation (inflam), lining thickness (lining) and bone erosion (erosion) were quantified on a 0–5 scale, n=10. (D) % lean mass and fat mass were assessed by dual-energy X-ray absorptiometry in toll-like receptor (TLR)4-mediated arthritis mice (72 hours) fed on RD and HFD, n=10. (E) Protein levels of pro-inflammatory cytokines (tumour necrosis factor (TNF)-α, interleukin (IL)-1β, IL-6 and MIP2) and (F) chemokines (CXCL1, CXCL5, CCL2, CCL3, CCL20 and CCL5) were quantified by ELISA in TLR4 arthritic mice (72 hours) fed on RD and HFD, n=5. Values are mean±SE. *Indicates p<0.05.

Figure 5

Joint myeloid cell recruitment and M1 macrophage polarisation are upregulated in obese toll-like receptor (TLR)4 arthritic mice compared with lean controls. (A) Synovial tissues (STs) from TLR4 arthritic mice (72 hours) on regular diet (RD) and high fat diet (HFD) were immunostained with anti-F4/80 antibody (orig. mag. ×200) and (B) macrophage staining was quantified on a 0–5 scale, n=7. (C) STs from TLR4 arthritic mice (72 hours) on RD and HFD were immunostained with anti-inducible nitric oxide synthase (iNOS) and anti-arginase antibodies (orig. mag. ×200) and (D) positive staining was quantified on a 0–5 scale, n=7. (E) Ankle homogenates harvested following 72 hours of lipopolysaccharide intra-articularly (i.a.) injection from mice fed on RD and HFD were western blotted for iNOS and arginase expression or actin equal loading, n=5. Values are mean±SE. *Indicates p<0.05.

RA and mouse adipose conditioned media promote M1 macrophage differentiation via TLR4-dependent cascade

We next asked whether adipose conditioned media obtained from RA or obese mice could remodel myeloid cells into inflammatory M1 macrophages as observed in the obese arthritic mice. We show that addition of RA or mouse adipose conditioned media to RA monocytes or mouse bone marrow myeloid cells can polarise the naïve cells into M1 macrophages. The M1 differentiation process was determined by TNF-α, IL-6 and/or IL-8 transcription as well as assessing the cell surface CD127 (human) or CD80 (mouse) frequency (figure 6A–D). Interestingly, RA adipose conditioned media had greater capacity in transforming the naïve myeloid cells into M1 macrophages compared with those obtained from patients with OA (figure 6B). Notably, our results indicate that once the concentration of the inflammatory factors in the RA or mouse adipose conditioned media reach a specific level, it no longer has a dose-responsive effect on M1 differentiation as inflammatory mediators may fall in the non-linear saturated range (figures 6A, C, D). Next, to elucidate the mechanism by which adipose media promotes M1 macrophage polarisation, IL-6, MIP2 and TLR2 and TLR4 pathways were blocked in 10% mouse adipose media (most optimal M1 differentiation condition; figure 6C). We found that although levels of IL-6 were elevated in HFD arthritic mice at 72 hours as well as in RA and mouse adipose conditioned media, neutralisation of these cytokine did not impact the M1 polarisation process (figure 6E). We further show that unlike TLR2 immunoneutralisation, blockade or deficiency in TLR4 signalling severely impairs fat-induced M1 macrophage differentiation, suggesting that TLR4 can be activated by factors available in the adipose conditioned media (figure 6E–G). Our results suggest that interaction of TLR4+ myeloid cells in the obese joint with the inflammatory mediators released from the adipose tissues can reconstruct these naïve myeloid cells into M1 pro-inflammatory macrophages.

Figure 6

Rheumatoid arthritis (RA) and mouse adipose conditioned media can transform the RA or mouse naïve myeloid cells into M1 macrophages through toll-like receptor (TLR)4-dependent function. (A) RA peripheral blood (PB) monocytes or (C) 7 days M-CSF-driven wild-type (WT) mouse bone marrow myeloid cells were treated with phosphate-buffered saline (PBS), 10, 25 and 50% RA or mouse adipose conditioned media and/or interferon (IFN)-γ for 6 hours and tumour necrosis factor (TNF)-α, interleukin (IL)-6 and/or IL-8 mRNA expression levels were determined by real-time RT-PCR, n=4. (B) RA PB monocytes or (D) 7 days M-CSF-driven WT bone marrow cells were treated with PBS, IFN-γ, 10% RA or OA adipose conditioned media or 10%–50% mouse adipose conditioned media for 24 hours prior to determining % CD127 (B) or CD80 (D) by flow cytometry, n=4. (E) 7 days M-CSF-driven WT bone marrow cells were either untreated (PBS) or pretreated (1 hour) with IgG, anti-IL-6, anti-MIP2, anti-TLR2 and anti-TLR4 antibodies prior to being stimulated with 10% mouse adipose media for 6 hours, n=3. (F) 7 days M-CSF-driven WT bone marrow cells were either untreated or pretreated with TLR4 antagonist for 3 hours prior to stimulating the cells with PBS, lipopolysaccharide (LPS), ultrapure LPS or 10% mouse adipose media for 6 hours, n=3. (G) 7 days M-CSF-driven WT and TLR4−/− bone marrow cells were treated with PBS, IFN-γ, ultrapure LPS or 10% mouse adipose media for 6 hours, n=3. In E, F and G, TNF-α or IL-6 mRNA expression levels were determined by real-time RT-PCR. Values are mean±SE. *Indicates p<0.05.

Discussion

In this novel study, we show for the first time an early and a late effect of obesity on RA. We uncover that obesity can critically impact disease initiation by fostering neutrophil migration through IL-8/MIP2 cascade. While diet-induced obesity does not affect established active disease in preclinical models, it can markedly delay arthritis remission as a result of TLR4-driven M1 macrophage differentiation. Our results highlight for the first time that obesity can differentially modulate RA pathogenesis contingent on the disease status.

To shed light on the mechanism by which obesity impacts RA, diet-induced obesity was provoked in CIA. In contrast to previous studies which demonstrate that obesity exacerbates CIA due to increase in the number of spleen TH-17 cells,12 our results clearly show no difference in disease activity or inflammatory parameters assessed in CIA effector phase in the obese compared with lean mice. The discrepancy in our results may be due to the differences in the mouse background (C57BL6 vs DBA/1J in our study) or the % diet fat (34.9% vs 60% in our study) between the two studies. However, we found that the obese CIA mice have an earlier disease onset (days 28–30) compared with the lean mice. On the contrary, as disease progresses beyond day 30, the differences in CIA ankle swelling are lost between the two diet groups. Examining RA and mouse adipose conditioned media, we found that IL-8/MIP2 (47 and 13 ng/mL) was the most abundantly expressed neutrophil chemokine compared with CXCL1 and CXCL5. We document that IL-8/MIP2 neutralisation impairs RA and mouse adipose media-induced neutrophil migration via p38 inactivation. In contrast, neutrophil chemotaxis was unaffected by blockade of CXCL1 and CXCL5 function. Interestingly, neutrophil chemokines (IL-8 (41-fold), CXCL1 (4-fold) and CXCL5 (3-fold)) expressed at significantly lower concentration in OA compared with RA adipose conditioned media can still contribute to neutrophil infiltration (see online supplementary figures S5 and S6). We demonstrate that early onset of CIA observed in the HFD compared with RD mice is due to MIP2-dependent neutrophil trafficking, which may explain why obesity is a risk factor for RA initiation. Corroborating with these findings, IL-8 levels closely correlate with the number of neutrophils in RA synovial fluid34 ,35 and others have shown that anti-MIP2 Ab treatment impairs neutrophil infiltration in preclinical models of arthritis, leading to resolution of acute disease.36 IL-8 is produced by RA ST fibroblasts, macrophages or adipocytes in response to pro-inflammatory cytokines such as IL-1β, TNF, IL-17 or lipoproteins, hence justifying its potential as a therapeutic target for RA.37–40 Using multiphoton real-time in vivo microscopy previous studies have described that neutrophils migrate in clusters and surround CIA joint fibroblasts but not myeloid cells at disease onset41 suggesting that fibroblasts are the main source of MIP2/IL-8 secretion in the initial phases of CIA.

Similar to our observation in RA and mouse adipose conditioned media, circulating levels of IL-8 and IL-6 are elevated in obese individuals and these concentrations positively correlate with BMI and/or C reactive protein (CRP).42 ,43 Interestingly, joint MIP2 and IL-6 levels were upregulated during disease progression at 72 hours in obese TLR4 arthritic mice compared with non-obese counterparts. Previous studies demonstrate that adipose tissue express elevated levels of TLR444 and hence i.a. injection of TLR4 ligand can contribute to adipocyte production of IL-6 and MIP2 and with increased myeloid cell migration following 72 hours of LPS administration, both adipocytes and macrophages can potentially elevate levels of these cytokines. Corroborating with this notion, co-culture of macrophages and adipocytes stimulated with low dose of LPS results in 100-fold higher IL-6 production compared with adipocytes alone.45

Studies performed in knockout mice document that expression of myeloid CCR2 and CCR5 and elevation of their ligands, CCL2 and CCL5 in the adipose tissue are indispensable for obesity-mediated monocyte infiltration.46 ,47 Despite the increase in joint monocyte trafficking detected in the obese mice following 72 hours of TLR4-induced arthritis, CCL2 levels were similar in both diet groups while other monocyte chemoattractants such as CCL5, CCL20 and TNF were elevated in the HFD compared with RD mice. Consistently, others show that obesity-associated monocyte recruitment is not restrained by CCL2 deficiency and the increase in cell influx can be promoted through macrophage galactose-specific lectin MAC2 produced from adipose tissues,48 ,49 suggesting that in the absence of CCL2 alternative mechanisms can manoeuvre this process.

It is known that macrophages are plastic cells that can alter their phenotype depending on their milieu. The M1 macrophages are ‘classically activated cells’ that express cell surface CCR7, CD127, CD80/CD86 and can secrete TNF, IL-6, IL-8 and IL-1 upon activation with IFN-γ and/or LPS.50 We found that CIA severity did not resolve and its progression was not affected by diet-induced obesity; in contrast, obese TLR4-induced arthritic mice had delayed remission compared with the lean group. We uncovered that CIA prolonged to 106 days was not impacted by obesity, since there were similar number of joint M1 and M2 macrophages in both diet groups which resulted in comparable levels of M1-associated pro-inflammatory factors. The lack of effect of obesity on active progression of arthritis may explain why there is a disconnection between BMI, RA joint pain and bone destruction.51

We further show that while joint inflammation resolves in lean TLR4 arthritic mice, diet-induced obesity can prolong arthritis by 24 hours through converting the undifferentiated myeloid cells into iNOS+ M1 macrophages that produce higher levels of pro-inflammatory factors. Delay in remission in the obese TLR4 arthritic mice, corroborates with lack of the anti-TNF response in obese compared with normal weight RA,5–7 ,16 which may also be due to sustained joint M1 macrophages.

To examine the underlying mechanism by which adipose conditioned media transforms the undifferentiated myeloid cells into M1 macrophages, role of IL-6, MIP2, TLR2 and TLR4 was explored in this function. Interestingly, we show that only inhibition of TLR4 but not TLR2, IL-6 or MIP2 negates the development of mouse bone marrow myeloid cells into M1 macrophages. Consistently, earlier studies show that while fatty acids like palmitate degrade IκB in WT mouse peritoneal macrophages, nuclear factor-κB signalling is impaired in TLR4−/− cells.44 Notably, activation of TLR4 by free fatty acids is not restricted to macrophages since a recent paper demonstrates that RA ST fibroblasts stimulated by unsaturated and saturated fatty acids produce pro-inflammatory factors via TLR4 signalling.52 However, it is unclear whether fatty acids can directly bind to TLR4 or if they participate in receptor dimerisation procedure.53 ,54 In contrast to these findings, female TLR4−/− mice show increased obesity, yet these mice are partially protected against insulin resistance compared with WT mice.44 Our results support the conclusion that TLR4 plays an important role in obesity-mediated M1 differentiation perhaps in part due to presence of fatty acids. However, our findings also suggest that M1 polarisation driven by HFD is multifactorial and this process may not be explicitly due to TLR4 function. We conclude that obesity is an important risk factor that mainly dysregulates RA onset and remission and may help to explain epidemiological studies suggesting a greater incidence of RA and poor treatment response among obese individuals.

References

View Abstract

Footnotes

  • Handling editor Tore K Kvien

  • S-jK and ZC contributed equally.

  • Contributors All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. SS had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Designed the research: S-jK and SS. Performed the research: ZC, S-jK, ABE, MVV, HAE and SS. Analysed the data: ZC, S-jK, MVV, ABE, HAE, IBM, JFB, WS, SA, SV, NS, GF, GK, PF and SS. Provided essential reagents: WS, SA, SV and NS. Writing the paper: All the authors contributed to writing the paper.

  • Funding This work was supported in part by awards from Department of Veteran's Affairs MERIT Award 1I01BX002286, the National Institutes of Health AR055240 and AR065778, grant from Within Our Reach from the ACR, funding provided by Department of Defense PR093477 and Arthritis Foundation Innovative award. JFB efforts were funded by VA clinical science research and development as well as VA Career Development Award IK2 CX000955.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Approved by Local Ethical Committees.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data have been included in the paper.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.