CCR6+ memTh subpopulations differ in cytokine production, but not in expression of RORC, in healthy donors and treatment-naïve RA patients
For studying the role and properties of CCR6+ memTh subpopulations, Th17/Th22, Th17.1, double positive (DP) and double negative (DN) cells were sorted from healthy PBMC based on the chemokine receptors CXCR3 and CCR4 (figure 1A). Directly after sorting, Th17/Th22 cells showed a trend towards being the strongest producers of IL-17A, IL-22 and GM-CSF, whereas Th17.1 and DP cells produced the highest levels of IFNγ (figure 1B, upper panel). Furthermore, Th17.1 and DP cells produced significantly more TBX21 than Th17/Th22 cells. Interestingly, RORC was equal in all subpopulations despite the differences in IL-17A expression (figure 1B, upper panel).
Since the phenotype stability of the subpopulations is currently unknown, these sorted cells were cultured for three days under stimulation of anti-CD3 and anti-CD28 and examined for cytokine and transcription factor expression (figure 1B, lower panel). After three days, both Th17/Th22 and DN cells expressed high levels of IL-17A mRNA, although the differences with Th17.1 and DP cells did not reach statistical significance. All subpopulations expressed IL-22, with DP cells producing slightly less than the Th17/Th22 population. After three days of culture, Th17.1 and DP cells expressed significantly more IFNγ mRNA than Th17/Th22 cells similar to the ex vivo expression pattern. The expression pattern of TBX21 after culture matched the expression of IFNγ, with the highest levels in Th17.1 and DP cells and lowest in Th17/Th22 and DN cells (figure 1B, lower panel). The findings on the cytokine expression pattern on the mRNA level were also found on the protein level, both in the percentage of cytokine-expressing cells (figure 1C) and cytokine levels in the culture supernatant (figure 1D). These data suggest that the three-day culture and stimulation of the sorted CCR6+ memTh subpopulations does not significantly alter the profile of the cells based on expression of IL-17A, IFNγ, RORC and T-bet.
Although these data provide information on the healthy properties of the subpopulations, they do not necessarily represent the same phenotype as the subpopulations in inflammatory arthritis where the T cells are abnormally activated. Therefore, the CCR6+ subpopulations were also studied in PBMC from treatment-naïve RA patients. Similar to healthy donors, all four subpopulations were present in the PBMC of these treatment-naïve patients, with the majority being Th17/Th22 cells (figure 2A). Importantly, the subpopulations could also all be distinguished within the synovial fluid mononuclear cells (SFMC) and PBMC of established RA patients (figure 2B). Similar to the findings in healthy cells, all CCR6+ memTh subpopulations expressed equal levels of RORC, whereas IL-17A was predominantly produced by Th17/Th22 cells. Also the expression patterns for IFNγ and TBX21 were similar to the healthy situation, with the highest levels in Th17.1 cells and the lowest in Th17/Th22 cells. These data demonstrate that although all CCR6+ memTh subpopulations express equal levels of RORC, they differ in other parameters such as IL-17A and IFNγ. These properties are similar between healthy donors and treatment-naïve early RA patients.
All CCR6+ subpopulations can activate synovial fibroblasts to create a pro-inflammatory feedforward loop
To determine whether the differences in pro-inflammatory cytokines secreted by the individual CCR6+ memTh subpopulations resulted in different pathogenic potential, the ability of the subpopulations to activate synovial fibroblasts (SF) from patients with rheumatoid arthritis was investigated. CCR6+ memTh subpopulations, but also the classical Th1 cells (CCR6- CXCR3+ CCR4-) as a negative control, were sorted from healthy donors as in figure 1A and subsequently activated and cocultured with SF for three days. When compared to Th1 cells, all CCR6+ subpopulations activated SF as measured by increased IL-6 and IL-8 production (figure 3A). However, there were differences in the extent of activation, with Th17/Th22 cells being more potent inducers of IL-6 and IL-8 than Th17.1 and DP cells. Interestingly, only Th17.1 and DP cells induced MMP1 (figure 3A). The properties of the CCR6+ memTh populations in terms of IL-17A and IFNγ expression did not differ in the coculture with SF compared to the activated CCR6+ memTh populations only cultures in figure 1D, although the overall levels were slightly elevated due to the interaction with SF.
Similar to the CCR6+ memTh subpopulations from healthy individuals, those from treatment-naïve early RA patients also all activated SF upon coculture as demonstrated by the increased levels of IL-6 and IL-8, although this did not reach statistical significance (figure 3B). Interestingly, in contrast to the ‘healthy CCR6+ T cells’, the CCR6+ memTh subpopulations from RA patients did not differ in their capacity to induce IL-6 and IL-8. Furthermore, where in the healthy situation mainly Th17.1 and DP cells induce MMP1, all subpopulations from RA patients demonstrated a trend towards induction of MMP1 in SF. Similar to healthy individuals, the properties of the subsets in terms of IL-17A and IFNγ expression did not drastically change upon coculture with SF, with Th17/Th22 the highest in IL-17A and the lowest in IFNγ, in contrast to Th17.1 (figure 2 and 3B).
Altogether, these data indicate that all CCR6+ memTh subpopulations from both healthy individuals and RA patients can activate SF. However, there are subtle differences between the healthy and arthritic T cells in terms of which pro-inflammatory factors they predominantly activate in RASF.
All CCR6+ subpopulations use IL-17A to induce IL-6 production by SF
Since all CCR6+ subpopulations are more potent activators of SF than Th1 cells despite their differences in cytokine and transcription factor expression, we next studied which factors play a role in this pro-inflammatory feedforward loop. Previously, we have shown that induction of the pro-inflammatory loop by CCR6+ Th memory cells is mediated by IL-17A (10). However, since not all CCR6+ subpopulations express equal amounts of IL-17A, we first investigated the importance of IL-17A for SF activation in all individual CCR6+ subpopulations. Given the similar properties of the T cell subpopulations from healthy donors and RA patients and the limited availability of cells from RA patients, these functional studies were conducted with healthy donor T cells. Upon treatment with anti-IL-17A the induction of IL-6 by all subpopulations decreased, although this did not reach statistical significance for DP cells (figure 4A). IL-8, MMP1 and MMP3 also showed a decreasing trend upon IL-17A neutralization. The SF activation by DP cells was hardly affected by IL-17A neutralization, suggesting that these cells work via a different mechanism.
One of the features of the pro-inflammatory loop between SF and CCR6+ memTh is that increased activation of the SF leads to increased activation of the Th cells. Interestingly, despite the reduced SF activation upon IL-17A blockade, the T cell derived cytokines IL-17A, IL-22, IFNγ and GM-CSF were not affected by anti-IL-17A treatment (figure 4B). Since IL-17A induction by SF depends on PGE2 (17), this lack of effect could be due to non-responsiveness of PGE2 on the anti-IL-17A treatment. However, PGE2 was significantly inhibited by anti-IL17A in Th17 and DP cells, and showed a trend towards reduction in DN and Th17.1 cells (figure 4C). From these data we conclude that IL-17A blockade reduces SF activation by reducing IL-6 in particular, but that these SF are still capable of further activating the Th cells. This suggests that a mechanism that does not include PGE2 or IL-6 is responsible for the Th cell activation in the SF-Th cell cocultures.
Activation of SF by CCR6+ subpopulations is independent of IFNγ and T-bet
Since IL-17A was mainly responsible for the induction of IL-6 by SF, other cytokines may also play a role in the pro-inflammatory loop between CCR6+ subpopulations and SF. Previous studies have shown that IFNγ suppresses production of MMP1 and MMP3 in IL-1β-stimulated SF, suggesting a modulatory role for this cytokine in the early stages of inflammation (18). Therefore, the role of IFNγ in our system was studied. In contrast to IL-17A blockade, neutralizing IFNγ had no significant effect on the production of IL-6, IL-8, MMP1 and MMP3 (figure 5A). Even though less IFNγ was detected in the culture supernatant, there was no effect on the other T-cell derived cytokines (figure 5B).
Next to IFNγ, also TBX21 showed a distinct pattern of expression between the subsets. Therefore, T-bet, the protein encoded by TBX21, was knocked down using lentiviral shRNA. Due to the low cell numbers that can be obtained for the individual subpopulations, we used the entire CCR6+ Th memory pool for these experiments. In the transduced CCR6+ Th memory cells T-bet expression was decreased by 70% (figure 5C). Importantly, this repression did not induce EOMES, a transcription factor known to take over the role of T-bet (figure 5C). When the T-bet-deficient CCR6+ Th memory cells were cultured together with SF, there was no reduction in fibroblast activation as demonstrated by IL-6, IL-8, MMP1 and MMP3 expression, despite a reduction in IFNγ and a slight increase in IL-17A (figure 5D). Altogether, these data indicate that the activation of SF and induction of the pro-inflammatory loop between SF and CCR6+ Th memory cells are independent of IFNγ and T-bet.
Both SF activation and the pro-inflammatory loop between CCR6+ subpopulations and SF can be inhibited by 1,25(OH)2D3
Previously, we have shown that blocking IL-22 does not affect the pro-inflammatory loop between CCR6+ memTh cells and SF (19), and here we found that IFNγ neutralization has no effect, while IL-17A is partially responsible for SF activation. However, the combined blockade of all pro-inflammatory T cell cytokines may have additional benefits. We have previously shown that CCR6+ memTh cells are highly susceptible to modulation by the active vitamin D metabolite 1,25(OH)2D3 (20-22), leading to decreased production of IL-17A, IL-22 and IFNγ (21, 23, 24). Therefore, this may be a valuable strategy to block these cytokines together and evaluate the effects on SF activation.
Since it is unknown whether the individual CCR6+ memTh subpopulations are susceptible to 1,25(OH)2D3-mediated modulation, we first studied its effects in cultures of individual CCR6+ memTh subpopulations. As expected, the percentage of IL-17A-producing cells decreased in both Th17/Th22 cells and DN cells, the cell types that produce the highest levels of IL-17A. IL-22 was decreased by 1,25(OH)2D3 in all subpopulations, and IFNγ in Th17.1 and DP cells (figure 6A-B). Since this suggests that all subpopulations can be affected by 1,25(OH)2D3, the compound was added to cocultures of SF and CCR6+ memTh subpopulations. In these cultures, 1,25(OH)2D3 inhibited SF activation by all individual CCR6+ subpopulations, as demonstrated by a decrease in IL-6, IL-8, PGE2 and MMP1, but not MMP3 (figure 6C). 1,25(OH)2D3 also decreased the production of IL-17A, IL-22 and IFNγ similar to the data in cultures of T cells alone, but not GM-CSF (figure 6D). Therefore, we concluded that 1,25(OH)2D3 significantly inhibits the pro-inflammatory loop of all the individual CCR6+ memTh subpopulations cocultured with SF.