Effects of interactions with synoviocytes on Th17/Th1 cytokine production
Physical interactions between stromal cells, such as synoviocytes or skin fibroblasts, and immune cells are critical to induce cytokine production at the site of inflammation in arthritis or in psoriatic skin [11, 12]. The use of a transwell system, preventing direct cell-cell contact but allowing the exchange of soluble factors, confirmed this role, as the cytokine production significantly decreased without physical interactions [11, 12]. Here, we used a co-culture system between stromal cells, RA synoviocytes or Pso skin fibroblasts, and immune cells to evaluate the production of Th1 and Th17 cytokines, with a specific interest in IL-23. RA and Pso stromal cells were selected as these cells come from two diseases giving different responses to the IL-17/IL-23 inhibition.
Cell interactions with RA synoviocytes induced high IL-6 production independently of cell activation compared to cells alone (p = 0.016, Fig. 1A). Both cell interactions and activation increased IL-1β secretion, about twofold higher in co-cultures compared to PBMCs alone in control and PHA conditions (p = 0.016, Fig. 1A).
We next focused on cytokines of the Th17 pathway, IL-23 and IL-17, and of the Th1 pathway, IL-12 and IFNγ. IL-17 production was almost undetectable in control, PBMCs and co-culture conditions (Fig. 1A). PHA increased IL-17 secretion (p = 0.016, Fig. 1A), but its concentration was much higher in co-cultures compared to PBMCs alone (229.7 ± 57.9 vs. 89.7 ± 21.5 pg/ml; p = 0.016, Fig. 1A).
Regarding IL-23, cell interactions decreased its production by about twofold compared to PBMCs alone (p = 0.016, Fig. 1A). Furthermore, PHA reinforced the IL-23 decrease in co-cultures compared to control (104.8 ± 19.0 vs. 179.5 ± 30.6 pg/ml; p = 0.016; Fig. 1A). This decrease in IL-23 secretion could be explained by consumption of IL-23 that induced IL-17, which explained higher IL-17 production in co-culture.
The results for IL-12 were opposed to those for IL-23, with an increase in cell interactions. As for IL-17, the highest production was obtained in activated co-culture condition (p = 0.016; Fig. 1A).
IFNγ results were like those of IL-17. In control, IFNγ production was low, even if cell interactions increased this secretion (p = 0.016; Fig. 1A). PHA strongly increased IFNγ secretion, but levels were even higher in co-cultures compared to PBMCs alone (1034.3 ± 206.5 vs. 244.6 ± 93.8 pg/ml; p = 0.016, Fig. 1A).
To conclude this part on synoviocytes, cell interactions alone were sufficient to induce IL-6 and IL-1β production, while high IL-17, IL-12, and IFNγ secretion required both cell activation and interactions. In contrast to all other cytokines, IL-23 production decreased by interactions with synoviocytes.
Effects of interactions with skin fibroblasts on Th17/Th1 cytokine production
To compare joint and skin diseases, we looked at the effect of interactions, between PBMCs and skin fibroblasts, on Th17/Th1 cytokine production. Skin fibroblasts came from two biopsies of the same Pso skin area, one from non-lesional (NLSF) skin and one from lesional (LSF) skin.
IL-6, IL-1β, IL-17, IL-12, and IFNγ productions were like those obtained with synoviocytes (Fig. 1B). Interactions with NLSF or LSF induced a high IL-6 production independently of cell activation (p = 0.016, Fig. 1B). Both cell interactions and PHA increased IL-1β secretion (p = 0.016; Fig. 1B). IL-17 concentration in the control condition was almost undetectable (Fig. 1B). PHA increased IL-17 production in PBMCs and in co-cultures (p = 0.016; Fig. 1B), but the highest concentration was observed in activated co-cultures (p = 0.016; Fig. 1B). IL-12 secretion was increased by cell interactions, mainly with PHA (p = 0.016, Fig. 1B). Cell activation and interactions were sufficient to induce IFNγ production, and its highest concentration was obtained in activated co-cultures (p = 0.016; Fig. 1B). It should also be noted that the production of IFNγ was higher with LSF than with NLSF (p = 0.016; Fig. 1B).
The results regarding IL-23 were different than those with synoviocytes, where IL-23 production decreased (Fig. 1A). Such decrease was not observed with skin fibroblasts, either NLSF or LSF, in control or with PHA conditions (Fig. 1B). In order to better understand this difference, we looked at the expression of IL-23 mRNA (Fig. 2). Firstly, as expected, in cells alone, IL-23 mRNA was mainly detected in PBMC (Fig. 2A) and PHA slightly decreased IL-23 mRNA (Fig. 2A) that correlated with the effect on IL-23 production (Fig. 1). In order to observe the effect of cell interactions on IL-23 mRNA, PBMCs and synoviocytes or PBMCs and skin fibroblasts, which had been cultured alone, were pooled, and then RNA was recovered from the combined cell subsets to be compared to the RNA recovered from co-cultures (Fig. 2B). Furthermore, to better compare the different cell types, the expression of cells cultured alone and pooled was normalized to 1 and he expression in the co-culture rationalized accordingly. Cell interactions with synoviocytes significantly increased IL-23 expression in control and PHA (p = 0.03, Fig. 2B). With NLSF, in control, the expression was significantly increased (p = 0.03, Fig. 2B), and also in PHA, but without reaching a significant value (p = 0.06, Fig. 2B). For LSF, it was only a tendency to increase as nothing was significant. The increase of IL-23 mRNA was the highest with synoviocytes (Fig. 2B), while IL-23 production significantly decreased during interactions with synoviocytes but not with skin fibroblasts (Fig. 1). This could involve a higher consumption of IL-23 in co-cultures with synoviocytes compared to skin fibroblasts.
To conclude this part on skin fibroblasts, the results for most cytokines were rather like those with synoviocytes, where interactions induced pro-inflammatory cytokines, mainly resulting from cell activation. On average, LSF were more potent than NLSF. However, a major difference was observed with IL-23. Cell interactions with RA synoviocytes but not with Pso skin fibroblasts decreased IL-23 secretion while IL-23 mRNA was increased with cell interactions. This could reflect a different consumption of IL-23 between co-cultures depending on the stromal cell origin.
Effects of cell interactions with stromal cells from different origins on cell activation
In order to further investigate the role of stromal cells on PBMC activation, we looked at the expression of activation markers (CD69 and CD86) by flow cytometry. CD3 and CD4 markers were used to distinguish the different PBMC subpopulations.
Firstly, we observed that neither PHA nor cell interactions had an effect on the percentage of CD3+ cells (Fig. 3A). Cell activation by PHA significantly decreased the percentage of CD3+CD4+ cells in PBMCs (p = 0.05, Fig. 3A) and trended to diminish this percentage in co-cultures (Fig. 3A). In addition, cell interactions also decreased the percentage of CD3+CD4+ cells, independently of stromal cell origin, compared to PBMCs alone (p = 0.05, Fig. 3A).
Then, we looked at CD69 and CD86 activation markers. In CD3+ cells, PHA increased significantly both activation markers (p = 0.05, Fig. 3B) with no additional effect of cell interactions (Fig. 3B). In CD3− cells, PHA had no clear effect while cell interactions tended to increase CD69 and in a lower extent CD86 (Fig. 3B). Interestingly, if cell interactions had no clear effect on the percentage of CD69+ and CD86+ cells, they seemed to increase the measurement of fluorescence intensity (MFI), mainly for CD86 (Fig. 3C). In CD3+ cells, we have distinguished CD4+ from CD4− cells (Fig. 3D). In CD4+ cells, PHA increased significantly CD69 and CD86 markers (p = 0.05, Fig. 3D), and this was reinforced by cell interactions (Fig. 3D). In CD4− cells, only PHA had an effect, and this reached a significant value in PBMCs (p = 0.05, Fig. 3D). The results were the same between synoviocytes and skin fibroblasts.
To conclude, PHA activated mainly CD3+ cells, thus lymphocyte population and cell interactions reinforced this effect mainly in CD4+ lymphocytes. This increased activation of CD3+CD4+ correlated with high IL-17 production requiring PHA activation and cell interactions (Fig. 1).
IL-23 contribution to Th17/Th1 cytokine production
Based on the differences in cell interactions between synoviocytes and skin fibroblasts on IL-23 production and on clinical results of anti-IL-23 treatment, we next focused on the involvement of IL-23 in cytokine production resulting from these cell interactions, by adding IL-23 or a blocking anti-IL-23 antibody to co-cultures.
For both synoviocytes and skin fibroblasts, treatment of co-cultures with IL-23 or an anti-IL-23 antibody had no effect on IL-6, IL-1β, or IFNγ (Fig. 4A). With exogenous IL-23 addition, the IL-23 concentration in cultures was around 50 ng/ml (data not shown), reflecting the added IL-23. The anti-IL-23 antibody had no impact on IL-23 production, except for one patient with synoviocytes (control, 92.2 pg/ml vs. anti-IL-23, 836.6 pg/ml), explaining the large SEM (Fig. 4A).
The added IL-23 was functional since IL-17 secretion increased compared to control, in co-culture with synoviocytes and LSF (p ≤ 0.03) and to a lower extent with NLSF (Fig. 4A). Conversely, the anti-IL-23 antibody had no effect on co-cultures with synoviocytes or LSF but decreased IL-17 production with NLSF (p = 0.02, Fig. 4A). The effect of anti-IL-23 antibody was specific as a control antibody was used in the first experiments, with no effect on cytokine production (Fig. 4B).
Regarding IL-12, IL-23 had an effect only on NLSF co-cultures by increasing IL-12 secretion (p = 0.03, Fig. 4A) while the anti-IL-23 antibody had no effect (Fig. 4A).
The effects of exogenous IL-23 or anti-IL-23 antibody were possibly less obvious than expected. Interestingly, the results with synoviocytes and LSF were very similar, but different from those with NLSF, regarding IL-17 and IL-12 production.
Effect of IL-23 and cell activation on cytokine receptor gene expression in isolated cells
Following the identification of differences between skin fibroblasts and synoviocytes regarding the IL-23 pathway, one key point to consider was a possible difference in cytokine receptor expression. Response to cytokines is mediated by membrane receptors usually made of two chains: IL-17RA/IL-17RC for IL-17, IL-12β1/IL-12β2 for IL-12, and IL-12β1/IL-23R for IL-23. We measured receptor mRNA expression in PBMCs, synoviocytes, or skin fibroblasts alone and in three conditions, control, PHA activation, and IL-23 treatment.
In isolated cells, the expression values of all subunits, except IL-17RC, were higher in PBMCs than in synoviocytes (Fig. 5), NLSF, and LSF (Fig. 6). In PBMCs alone, PHA decreased IL-17RA (p = 0.03, Figs. 5 and 6) but significantly increased IL-12Rβ2 expression (p = 0.03, Figs. 5 and 6). IL-12Rβ1 and IL-23R expression showed a modest increase, which was not significant. IL-23 treatment significantly increased only IL-17RC expression (p = 0.03, Figs. 5 and 6).
In stromal cells alone, PHA and IL-23 had no major effect (Figs. 5 and 6), except for a significant increase of IL-17RC expression in NLSF (p = 0.03, Fig. 6). In co-cultures, IL-23 had no effect on receptor gene expression while PHA significantly increased IL-12Rβ2 and IL-23R in the three co-cultures with synoviocytes, NLSF and LSF (p = 0.03, Figs. 5 and 6). IL-12Rβ1 was significantly increased in co-cultures with NLSF and LSF (p = 0.03, Fig. 6), but it was just a tendency with synoviocytes without any significance (Fig. 5).
In conclusion, the expression of these cytokine receptor genes was mainly regulated by PHA activation compared to IL-23, in PBMCs and co-cultures. In stromal cells alone, RA synoviocytes, or Pso skin fibroblasts, receptor gene expression did not seem to be regulated by PHA or IL-23. In PBMCs alone, PHA decreased IL-17RA and IL-17RC expression while it increased IL-12Rβ1, IL-12Rβ2, and IL-23R expression. In co-cultures, PHA similarly regulated receptor gene expression with RA synoviocytes and Pso skin fibroblasts, mainly increasing the expression of IL-23 receptor subunits.
Effect of cell interactions on cytokine receptor gene expression
As cytokine receptor gene expression and regulation were similar in synoviocytes and skin fibroblasts alone, we next focused our interest on the effect of cell interactions. To investigate this point more rigorously, PBMCs and stromal cells were cultured alone or in co-cultures, with PHA. At the end of the culture, PBMCs and synoviocytes or PBMCs and skin fibroblasts, which had been cultured alone, were pooled, and then RNA was recovered from the combined cell subsets. This allowed recovering RNA from both cell types, which we could compare to RNA recovered from co-cultures. Using this protocol, we could observe the direct effect of cell interactions on receptor gene expression (Fig. 7).
Interactions with synoviocytes did not influence IL-17 receptor subunit expression, while, in contrast, the latter decreased with skin fibroblasts, both NLSF and LSF (p = 0.03, Fig. 7). This discrepancy between stromal cells was even more pronounced for IL-12Rβ1, IL-12Rβ2, and IL-23R expression. Interactions with synoviocytes increased IL-12Rβ1 and IL-23R expression (p ≤ 0.03, Fig. 7), and those with skin fibroblasts strongly decreased it for NLSF (p = 0.03, Fig. 7) and had no effect on LSF. IL-12Rβ2 expression was strongly increased in synoviocytes (p = 0.03, Fig. 7) while no significant effect was observed with NLSF and LSF. Nevertheless, a slight increase was observed with LSF.
To confirm the difference in IL-23 receptor regulation, we analyzed its expression by flow cytometry. The subunits IL-12Rβ1 and IL-23R were observed in cells alone or in co-cultures. PBMCs and stromal cells were distinguished by CD45, expressed by PBMCs (93.6 ± 6.22% of CD45+ cells in PBMCs alone and 89.1 ± 2.6% in co-cultures, data not shown) but not by stromal cells (less than 1% of positive cells, data not shown). As expected regarding mRNA results, stromal cells did not express the subunits of the IL-23 receptor (< 1% of positive cells, data not shown). In PBMCs, IL-12Rβ1 was constitutively expressed, and its expression was not affected by cell interactions (Fig. 7B). However, the IL-23R subunit was weakly expressed in PBMCs (1.6 ± 0.3% of IL-23R+ cells, Fig. 7B). Its expression was significantly increased by cell interactions with synoviocytes (2.91 ± 1.1% of IL-23R+ cells, p = 0.03, Fig. 7B) but not with NLSF (1.9 ± 0.8% of IL-23R+ cells, Fig. 7B). These results reinforced the differences observed at the mRNA level. Moreover, the expression of the receptors was done within 24 h of culture, and it would be interesting to look at later times to see if this difference persists and increases. With LSF, the results were more difficult to interpret because of a large standard deviation due to various responses, including one patient who increased the percentage of IL-23R+ cells to 22.6%. However, what we could note was that the result was finally between synoviocytes and NLSF, as for mRNA, with a tendency to increase IL-23R expression without reaching significance.
To conclude this part, cell interactions regulated receptor gene expression very differently, and even more in an opposite direction, depending on the stromal cell origin. Briefly, interactions with synoviocytes increased the expression of both IL-23 receptor subunits at mRNA level and IL-23R at the surface expression level while interactions with non-lesional skin fibroblasts decreased their expression at the mRNA level and had no effect at the surface expression level. The situation in LSF was in between that of NSLF and synoviocytes.