Aryl hydrocarbon receptor (Ahr)‐dependent Il‐22 expression by type 3 innate lymphoid cells control of acute joint inflammation

Abstract The aryl hydrocarbon receptor (AHR) controls several inflammatory and metabolic pathways involved in various diseases, including the development of arthritis. Here, we investigated the role of AHR activation in IL‐22‐dependent acute arthritis using the K/BxN serum transfer model. We observed an overall reduction of cytokine expression in Ahr‐deficient mice, along with decreased signs of joint inflammation. Conversely, we report worsened arthritis symptoms in Il‐22 deficient mice. Pharmacological stimulation of AHR with the agonist VAG539, as well as injection of recombinant IL‐22, given prior arthritogenic triggering, attenuated inflammation and reduced joint destruction. The protective effect of VAG539 was abrogated in Il‐22 deficient mice. Finally, conditional Ahr depletion of Rorc‐expressing cells was sufficient to attenuate arthritis, thereby uncovering a previously unsuspected role of AHR in type 3 innate lymphoid cells during acute arthritis.


| INTRODUC TI ON
Recent progresses in genome analyses technologies enabled the identification of more than 300 genes involved in autoimmune diseases. 1 However, for a large proportion of patients, causal variants in these genes are not observed, leading to the 'missing heritability' concept. 2 Epigenetic modifications, environmental factors or individual's microbiota are frequently considered as major, non-genetic, additional drivers of autoimmunity. Indeed, the prevalence of autoimmune diseases is increasing worldwide, a feature that has been linked to dramatic changes in human behaviour such as food consumption, but also to worsening air pollution. 3 Environmental pollutants are suspected to trigger several autoimmune diseases, including rheumatoid arthritis (RA). 4 Notably, polycyclic aromatic hydrocarbons (PAH), dioxin (TCDD), phthalates or alkylphenols activate the transcription factor aryl hydrocarbon receptor (AHR), [5][6][7] which has been involved in RA pathophysiology in mice. 8,9 Interestingly, AHR also senses tryptophan metabolites including those of bacterial origin. 10 This suggests that AHR can participate in the surveillance and homeostasis of the microbiota, the composition of which may also impact on RA pathogenesis. 11 Ligand-activated AHR translocates into the nucleus forming a transcriptional complex leading to ligand-dependent transcription of several genes regulating immune responses, such as Rort,  FoxP3. 6,[12][13][14][15] In the intestine, AHR regulates the innate immune response through the modulation of innate lymphoid cells (ILC3) secreting IL-22 following Il-22 gene transcriptional induction. 16 In mice, IL-22 is part of the RORt-dependent Th17 cell differentiation programme. 17 IL-17 and IL-22 are known promoters of allergic diseases, as demonstrated by their role in allergic lung inflammation [18][19][20] which is dependent on AHR activation. 21

| K/BxN serum arthritis induction
Two hundred microliter of serum harvested from K/BxN mice were injected by the intravenous route on day 0 and day 1. Joint inflammation is visible at day 3 and peaks at day 6. 29 Ankle thickness of the fore and hind limbs was measured daily with a caliper and the clinical score (0 = no swelling or erythema, 1 = slight swelling and/or erythema, 2 = low-to-moderate oedema, 3 = pronounced oedema with limited use of the joint, and 4 = excessive oedema with joint rigidity) reflects daily observations of the extent of swelling and reddening of the joints. The clinical indices for all four paws were added as a composite score.

| Microscopy
Mice were euthanized at day 10. Ankle joints were removed and fixed for 72 hours with 10% paraformaldehyde (pH 7.2). The ankle joint was then incubated in 10% EDTA at pH 7.2 for 10 days at room temperature to decalcify the bone. The samples were rinsed in PBS and dehydrated; embedded in paraffin, cut at 3 µm and stained with haematoxylin and eosin (H&E). To eliminate potential bias, the slides were scored by independent observers. The slides were graded using various parameters, such as severity of synovial hyperplasia (pannus formation), cellular exudates, and cartilage depletion/bone erosion, each scored 0 to 5, and extent of synovial infiltrate, scored 0-5, with higher scores indicating greater infiltration. The grades for all parameters were subsequently summed to obtain an arthritis index, with results expressed as the median arthritis score.

| Gene expression analyses
Synovial tissue was collected at 24h, snap-frozen in liquid nitrogen and kept at −80°C. Total RNA was isolated and homogenized with 1 mL of TRI Reagent ® (Sigma) using TRIzol/Chloroform extraction.

| Statistical analyses
Data were analysed using Prism version 5 (Graphpad Software).
The non-parametric Kruskal-Wallis test with Dunn's multiple comparison test or the parametric one-way ANOVA test with multiple Bonferroni's comparison test were used. Mann-Whitney U test was used to compare two groups of values. Values are expressed as mean ± SEM. Statistical significance was defined at a P-value < 0.05.

| Attenuated arthritis in the K/BxN serum transfer model in Ahr-deficient mice
The role of the aryl hydrocarbon receptor (AHR) in autoimmunity 30,31 and in RA in particular 32 has previously been explored in the collagen-induced arthritis model, which highlighted the role of T cells 8 and IL-17 9 in this process. However, the precise mechanisms linking AHR signalling to inflammatory responses remains poorly described. Here, we used the acute and transient model provided by the transfer of arthritogenic serum harvested from K/BxN donor mice into C57/Bl6 controls and Ahr-deficient recipients 24 to better understand the molecular and cellular players in the AHR-mediated control of inflammation. As seen in Figure 1A,B, Ahr-deficient mice display reduced arthritis symptoms, as demonstrated by a diminished clinical score and ankle thickness upon serum transfer, in agreement with other models like collagen-induced arthritis or antigen-induced arthritis. 8,9 Interestingly, the expression of inflammatory cytokines also appeared affected by loss of Ahr. This is the case of IL-6 ( Figure 1C) which does not participate in disease progression in this model, but which quantification represents a reliable marker of inflammation. 25 Furthermore, reduced IL-6 concentration in the synovia is in line with AHR-dependent activation of the NF-B signalling pathways and control of Il-6 gene expression. 33 In addition, the expression and secretion of chemokines such as CXCL1 is also diminished, which is in agreement with reduced expression of myeloperoxydase (MPO), an important marker of neutrophil infiltration ( Figure 1D,E). Of note, the expression of other cytokines like IL-10 and IL-22 was also diminished in Ahr-deleted mice ( Figure 1F,G).
Finally, histological examination of the joints in control (upon PBS injection) mice and following serum transfer in wild type and Ahr KO animals confirmed that mutant mice exhibited less inflammation and bone erosion. Thus, these data suggest that AHR contributes to the development of serum-induced arthritis. As IL-22 secretion was also reduced in Ahr KO mice, we investigated the potential protective function of this cytokine. F I G U R E 1 Attenuated K/BxN serum arthritis in the absence of Ahr. A, Clinical score (expressed in arbitrary units, AU) and B, ankle swelling (in mm). Cytokine expression in the synovium C, IL-6 (in pg/mL), D, CXCL1 (in pg/mL), E, MPO (in ng/mL), F, IL-10 (in pg/mL) and G, IL-22 (in pg/mL). H, Representative histological pictures of the joint in control mice and following K/BxN serum transfer in wild type (C57Bl/6) and Ahr KO animals at day 6. I. Inflammation and bone erosion scores (in arbitrary units, AU) quantified from histological analysis at day 10. Mean values +/− standard deviation are shown (n = 8 to 10 mice/group). Representative data from two independent experiments are shown. Data were analysed following a Mann-Whitney U test. *P <.05**P <.01***P <.001. AUC: area under the curve

| Innate lymphoid cells are increased and contribute to acute joint inflammation
Finally, we aimed at better defining the cellular mechanisms substantiating AHR-and IL-22-dependent protection against acute joint inflammation. For this, we first analysed molecular (MPO, IL-22 secretion) and phenotypic (joint swelling, histological score) markers of inflammation in Rorc deficient mice, which lack a Th17 polarization response. Th17 cells also express IL-22, 37 and their differentiation depends on the RORγt transcription factor encoded by the Rorc gene. 38 As seen in Figure 5A-D, K/BxN-injected Rorc-deficient animals exhibit a markedly reduced inflammation compared to wildtype controls, as demonstrated by lower MPO levels, joint swelling and inflammatory infiltrate and synovial Il-22 levels were slightly elevated. K/BxN serum transfer arthritis is known to be independent of both B and T cells, external antigen and adjuvant for disease development. 39 We confirmed that serum-induced arthritis is independent of B and T cells using RAG-deficient mice ( Figure 5). and dependent on AHR for their ontogeny. 16 To ascertain that the inflammatory response was mediated by Rorc expressing cells, we used Rorc-GFP reporter mice and found a rapid and massive (21%) recruitment of GFP + ILC3 cells (gating strategy is shown in Figure S4

| D ISCUSS I ON
With 1% of the world population being affected, rheumatoid arthritis is the most frequent chronic inflammatory arthritis. 43 In addition to its high prevalence, RA also illustrates the nature of 'complex diseases', associating genetic factors and environmental triggers. 44 So far, variants in more than 100 genes along with all immune cells of the innate and adaptive immune systems have been studied and involved in RA pathogenesis. 45 2 and S1). Intriguingly, similar effects (protection against arthritis) could be noticed in Ahr KO animals and upon AHR activation with the VAG539 agonist (Figures 1 and 4). We observed that gene ablation affects the expression of most pro-inflammatory cytokines and chemokines, as well as anti-inflammatory mediators (IL-10, IL-22). In this case, protection is likely caused by the reduction of a pro-inflammatory environment. In contrast, VAG539 appears to specifically increase IL-22 production and drive anti-inflammatory conditions. It is possible that engagement of this agonist with AHR induces conformational changes of the receptor leading to the activation of pathways particularly controlling Il-22 expression, while Ahr ablation eliminates NF-κB signalling, which is upstream of most immune genes. Of note, in both cases, the impact of the gene deletion could be visible shortly (24h) after serum transfer, indicating that innate cells were more likely to account for the phenotype, rather than T-or B cells which require several days for their differentiation and activation. These data point to a major anti-inflammatory effect of IL-22, which was further confirmed by injections of the recombinant cytokine (rhIL-22, Figure 4). Two major cell types are known to produce this cytokine: Th17 cells and ILC3. As mentioned above, the rapid effects seen in Il-22 KO animals ( Figure 2) and following rhIL-22 administration (Figure 4), as well as the fast expansion of Rorc-GFP + cells upon serum transfer ( Figure 5E) suggest that ILC3 play a major role in this process. This hypothesis is strongly supported by our observations that Rorc KO mice (which lack both Th17 and ILC3), Rag/Il-2R γ chain mutants combined to Rorc deletion (lacking all T cells and ILC3) exhibited a similar phenotypes (with regards to molecular and histological signs of inflammation) compared to mice devoid of ILC3 only, due to a targeted deletion of a floxed Rorc gene in Ahr-expressing cells ( Figure 5). Importantly, wild type (C57Bl/6) and Rag-2 KO (lacking B and T cells) exhibited similar phenotypes.
A cartoon summarizing these events is shown in Figure 6. We propose a hypothetical mechanism in which complement-dependent inflammation triggered by immune complexes present in K/BxN serum is down modulated by IL-22-producing ILC3. However, our data do not rule out a possible involvement of other innate cells, such as neutrophils which are also capable of IL-22 secretion, 54 and accounting for the unaltered IL-22 expression level in the synovial fluid of T cells-and ILC3-depleted animals ( Figure 5B).
Future experiments aiming at depleting ILC3 with an Ahr-driven diphteria toxin transgene might be informative in this regards. We also demonstrate in this work that AHR signalling and subsequent Il-22 expression is induced upon K/BxN serum injection ( Figure 3). However, the molecular details of AHR activation are not identified at this stage and more work will be needed to identify the endogenous ligands responsible for the induction of AHR signalling.

F I G U R E 6 A model integrating ILC3
in serum-dependent acute arthritis. The cartoon illustrates that complementdependent inflammation triggered by antiglucose-6-phosphate isomerase (G6PI) antibodies/G6PI immune complexes activate cells of the synovial membrane. In response, the cells secrete chemokines such as CXCL1 and CXCL16 attracting neutrophils and ILC3s. Once activated, ILC3 secrete IL- 22  Serum transfer arthritis is modified in Indoleamine 2,3-dioxygenase 2 (Ido2) mutant mice, 55 suggesting that still to be identified endogenous tryptophan metabolites might account for this observation.
Altogether, our data indicate that AHR is an interesting hub capable of sensing exogenous (pollutants or microbial derivatives), as well as endogenous molecules, providing an important surveillance system to sample environmental changes that may trigger acute inflammatory responses. Hence, pharmacological targeting (with agonists, as shown in this report, Figure 4) of this receptor might offer novel opportunities to reduce acute inflammation.

ACK N OWLED G EM ENTS
PG thanks Dr Stephan Bluml (Medical University, Vienna) for helpful advice and critical reading of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.