T Cell Receptor-Independent, CD31/IL-17A-Driven Inflammatory Axis Shapes Synovitis in Juvenile Idiopathic Arthritis

Abstract

T cells are considered autoimmune effectors in juvenile idiopathic arthritis (JIA), but the antigenic cause of arthritis remains elusive. Since T cells comprise a significant proportion of joint-infiltrating cells, we examined whether the environment in the joint could be shaped through the inflammatory activation by T cells that is independent of conventional TCR signaling. We focused on the analysis of synovial fluid (SF) collected from children with oligoarticular and rheumatoid factor-negative polyarticular JIA. Cytokine profiling of SF showed dominance of five molecules including IL-17A. Cytometric analysis of the same SF samples showed enrichment of αβT cells that lacked both CD4 and CD8 co-receptors [herein called double negative (DN) T cells] and also lacked the CD28 costimulatory receptor. However, these synovial αβT cells expressed high levels of CD31, an adhesion molecule that is normally employed by granulocytes when they transit to sites of injury. In receptor crosslinking assays, ligation of CD31 alone on synovial CD28^nullCD31⁺ DN αβT cells effectively and sufficiently induced phosphorylation of signaling substrates and increased intracytoplasmic stores of cytokines including IL-17A. CD31 ligation was also sufficient to induce RORγT expression and trans-activation of the IL-17A promoter. In addition to T cells, SF contained fibrocyte-like cells (FLC) expressing IL-17 receptor A (IL-17RA) and CD38, a known ligand for CD31. Stimulation of FLC with IL-17A led to CD38 upregulation, and to production of cytokines and tissue-destructive molecules. Addition of an oxidoreductase analog to the bioassays suppressed the CD31-driven IL-17A production by T cells. It also suppressed the downstream IL-17A-mediated production of effectors by FLC. The levels of suppression of FLC effector activities by the oxidoreductase analog were comparable to those seen with corticosteroid and/or biologic inhibitors to IL-6 and TNFα. Collectively, our data suggest that activation of a CD31-driven, αβTCR-independent, IL-17A-mediated T cell-FLC inflammatory circuit drives and/or perpetuates synovitis. With the notable finding that the oxidoreductase mimic suppresses the effector activities of synovial CD31⁺CD28^null αβT cells and IL-17RA⁺CD38⁺ FLC, this small molecule could be used to probe further the intricacies of this inflammatory circuit. Such bioactivities of this small molecule also provide rationale for new translational avenue(s) to potentially modulate JIA synovitis.

Keywords: CD31; IL-17; TCR-independent; double negative alpha beta T cells; fibrocyte-like cells; juvenile idiopathic arthritis; oxidoreductase; synovial inflammation.

Figure 1

Cytokine profiles in blood and cell-free SF of JIA patients. Data shown (n = 15–39 per group as in Table 1) are box-median-whisker plots of the five most dominant cytokines found in SF compared to plasma. The boxes represent the 25th and 75th percentile of values. The whiskers were the 5th and 95th percentile of values. There was not a global cytokine upregulation as shown by negligible detection of IL-2. The indicated P-values were determined by Kruskal–Wallis ANOVA. Between group comparisons were made using the post hoc Tukey statistic: [***], [**], and [*] P < 0.005, P < 0.01, and P < 0.05, respectively, indicate JIA SF is significantly different from healthy and JIA plasma; [@] indicates polyarticular [poly]/oligoarticular [oligo] SF were not significantly different from poly JIA plasma, but were significantly different (P < 0.05) from oligo JIA and healthy plasma. Abbreviation: NS, not significant.

Figure 2

αβT cell profiles in PBMC and SFMC of JIA patients. (A) Illustrative flow of the electronic gating strategy for cytometric determination of the expression of TCRαβ, CD4, CD8, CD28, and CD31. The gating profile shown was for an SFMC sample, which was very similar with a parallel PBMC gating profile. Following an initial live/dead electronic gate, the height and width of the forward scatter was used to set a single cell gate for lymphocytes that were sequentially examined for CD4, CD8, CD28, and CD31. (B) Box-median-whisker plots shown are the raw frequency and (C) the age-adjusted frequency of CD31⁺CD28^nullDN αβT cells as a proportion of the total parent population of gated αβTCR⁺ cells. The plots were constructed as in Figure 1. The indicated P-values were determined by Kruskal–Wallis analysis of variance. Post hoc group comparisons by Tukey: ***P < 0.005 indicates SFMC of oligoarticular [oligo] and polyarticular [poly] JIA were significantly different from JIA/healthy PBMC; *P < 0.05 indicates JIA PBMC was significantly different from healthy PBMC.

Figure 3

CD31 ligation alone is sufficient to induce intracellular expression of cytokines. (A) Diagram of receptor crosslinking with specific antibody, and the relevant gate of crosslinked (Cy5⁺) cells for subsequent cytometric analyses. (B) The data shown (bar means, SD whiskers; n = 5–11 per group) are intracellular levels of IFNγ, IL-6, IL-17, and TNFα in synovial CD31⁺CD28^null double negative (DN) and CD8⁺ αβT cells within 6 h of stimulation via CD3, TCRαβ, CD31, or IgG control. The data were expressed as stimulation index to normalize intrinsic variability between individual donors and culture batch differences of Jurkat or JRT3 cells. The plots were constructed as in Figure 1. The indicated P-values were determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: [***] and [**] P < 0.005 and P < 0.05, respectively, indicate CD31 crosslinking was significantly different from TCR or CD3 crosslinking and the IgG control [Cont] for synovial DN and CD8⁺ T cells, Jurkat, and JRT3; [@] indicates CD31 crosslinking was not significantly different from TCR and CD3 crosslinking, but was significantly different (P = 0.001) from the IgG control.

Figure 4

CD31 ligation alone is sufficient to elicit phosphorylation of signaling intermediates. Using the same crosslinking bioassay in Figure 3A, the data shown (bar means, SD whiskers; n = 5–9 per group) are phosphorylation levels of ZAP70, cAbl, AKT, and RelA within 15 min of stimulation via CD31, TCR, CD3, or IgG control. The data are normalized stimulation indices as in Figure 3B. The indicated P-values were determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: [***] and [**] P < 0.005 and P < 0.05, respectively, indicate CD31 crosslinking was significantly different from TCR or CD3 crosslinking or the IgG control [Cont]; [@] indicates CD31 crosslinking was not significantly different from TCR or CD3 crosslinking, but was significantly different (P = 0.0001) from the IgG control.

Figure 5

cAbl is a signaling substrate of CD31-driven TCR-independent activation of synovial αβT cells. (A) Micrographs shown are representative confocal images of CD31-driven polarization of cAbl in αβT cells from five independent experiments. The Z-stack (left to right order) were sequential image slices from the top to the bottom of a cell (DAPI and Green staining) and a fluorescent bead (Red) with immobilized anti-CD31. (B) The data shown (bar means, SD whiskers; n = 4–7 per group) are percent inhibition of CD31-driven expression of phospho-cAbl and phospho-RelA, and intracellular expression of TNFα, IFNγ, and IL-17A by 50 nM Imatinib or 34 µM MnT2E in CD31-crosslinked synovial CD31⁺CD28^null DN and CD8⁺ αβT cells CD31 crosslinking was performed as in Figure 3A. Data normalization was done as in Figures 3B and 4. The indicated P-values were determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: *P < 0.05 indicates MnT2E and Imatinib induced greater magnitudes of reduction of CD31-driven IL-17A expression than the inhibitor-induced reductions of cAbl, TNFα, IFNγ, and RelA expression.

Figure 6

CD31-driven expression of RORγT and trans-activation of IL-17A gene promoter. (A) Data shown (bar means, SD whiskers; n = 5–7 per group) are RORγT levels in synovial CD31⁺CD28^null DN and CD8⁺ αβT cells, Jurkat, and JRT3 following CD31, TCR, CD3, or IgG stimulation. Crosslinking assays were performed as in Figure 3A. Data normalization done as in Figures 3B and 4. The indicated P-values were determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: ***P < 0.001 indicates CD31 crosslinking in JRT3 were significantly different from TCR and CD3 crosslinking, and the IgG control; **P < 0.05 indicates CD31 crosslinking on DN T cells were significantly different from TCR or CD3 crosslinking, or IgG control; * indicates crosslinking of CD31, TCR, or CD3 on CD8 T cells and Jurkat were not significantly different but were significantly different (P < 0.001) from the IgG control. (B) Data shown (mean bars, SD whiskers; n = 4–7 per group) are IL-17A gene promoter-driven luciferase reporter activities of Jurkat and JRT3. The data were normalized for transfection efficiency by co-transfection of Renilla luciferase plasmid. Receptor crosslinking was performed as in Figure 3A. The indicated P-value was determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: [**] indicates crosslinking of CD31 and TCR on Jurkat was not significantly different, but either one was significantly different (P < 0.005) than the IgG control; [***] P < 0.001 indicates CD31crosslinking on JRT3 cells were significantly different than TCR crosslinking or the IgG control; [@] P < 0.005 indicates control luciferase activity of phorbol myristyl acetate/Iono-stimulated JRT3 without receptor crosslinking was higher than Jurkat.

Figure 7

Fibrocyte-like cells are non-lymphoid IL-17A-responsive components of SF. (A) The general electronic gating strategy to set a live single cell gate was done as in Figure 2A. The cytograms shown illustrate the gated larger sized cells in SFMC, i.e., boxed higher forward versus side scatter. As depicted, the proportions of these non-lymphoid SFMC varied widely among patients. These cells co-expressed procollagen 1 and proline hydroxylase, known markers of fibroblasts and mesenchymal fibrocytes. These cells, referred to as FLC, also expressed IL-17RA and CD38. (B) Representative micrograph of 10–15 day cultured, plate-adherent FLC showing stellate to ameboid morphology. Their typical cytogram profile showed negative staining for T and B cell markers, but positive staining of proline hydroxylase, procollagen 1, IL-17RA, and CD38. (C) CD38 expression on FLC (bar means, SD whiskers; n = 5 per group) incubated with three doses of recombinant IL-17A for 24 h. The indicated P-value was determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: [@] indicate no significant differences between 200 and 2,000 ng/ml doses of IL-17A; [**] indicates responses to 200 or 2,000 ng/ml was significantly different (P < 0.05) than those seen with 20 ng/ml IL-17A or media control.

Figure 8

IL-17A-mediated production of cytokines and chemokines by FLC, and their sensitivity to MnT2E. Data shown are box-median-whisker plots (n = 5–7 per group), which were constructed as in Figure 1. As depicted, the maximum levels of production of each of the indicated molecules by FLC cultures in 200 ng/ml recombinant IL-17A [recIL-17] were set as 100% response. Production levels of each indicated molecular effector in FLC cultures with recIL-17 combined with 5 µM corticosteroid or IL6i or TNFi, or 34 µM MnT2E were normalized as percent increase or decrease over the maximal response to recIL-17. The indicated P-values were determined by Kruskal–Wallis ANOVA. Post hoc group comparisons by Tukey: ***P < 0.005 indicate corticosteroid- and/or MnT2E-mediated suppression of IL-17A-mediated production of effectors was significantly greater than those elicited by TNFi or IL6i; [**] and [*]P < 0.01 and P < 0.05, respectively, indicate significant suppression of production of effector molecules compared recIL17 only group; [@] no significant differences among inhibitor-treated groups, but significantly different (P < 0.05) from the recIL-17 only group; ns, not significantly different between the inhibitor-treated group(s) and recIL-17 only group.

Figure 9

Sensitivity of IL-17A-mediated production of tissue-destructive proteins by FLC to down modulation by MnT2E. Culture supernatants from the same FLC bioassays in Figure 8 were also examined for MMPs and vascular endothelial growth factor (VEGF). Sample sizes, box-median-whisker plots, and statistical analyses were identical to that of Figure 8.

Figure 10

Model for an inflammatory circuit between CD31⁺CD28^null αβT cells and FLC in JIA synovitis. The present data (Figures 2 and 7) authenticate enrichment of CD31⁺CD28^null DN αβT cells, as well as procollagen [procol]⁺proline hydroxylase [P4H]⁺ FLC in SF of oligoarticular and rheumatoid factor⁻ polyarticular JIA. In previous work, we have shown a similar CD31⁺CD28^null subset in the CD8⁺ compartment (7). Ligation of CD31 with specific antibody on these two αβT cell subsets sufficiently and effectively signals phosphorylation of ZAP70, Akt, cAbl, and RelA (Figure 4). It is also sufficient to induce production of IL-17A, IL-6, IFNγ, and TNFα (Figure 3). These four cytokines are among the most dominant cytokines identified in SF (Figure 1). We have shown previously that similar cytokine production is achieved by CD31 ligation with CD38-Ig (7), indicating CD38 is a true ligand, CD31 ligand on synovial CD31⁺ αβT cells. Relevance of CD31 as a driver IL-17A is indicated by the specific trans-activation of IL-17A promoter and induction RORγT (Figure 6). CD31-driven production of cytokines and phosphorylation of cAbl and RelA are sensitive to down regulation by Imatinib, a known cAbl inhibitor, and by MnT2E, a superoxide mimic, and inhibitor of the DNA-binding activity of NFκB (Figure 5). IL-17RA⁺CD38⁺ FLC is a likely target of IL-17A as indicated by the upregulation of CD38 (Figure 7). Futhermore, IL-17A induces FLC to produce additional inflammatory cytokines/chemokines and tissue-destructive effectors, which are sensitive to down modulation by MnT2E (Figures 8 and 9). Such inhibitory activity of MnT2E on FLC is comparable, and in some cases, better than the biologics IL6i and TNFi. Details of IL-17A and CD38 signaling in FLC [dashed arrows] remain to be examined.