The importance of interleukin-17 (IL-17) is underscored both by its resistance to control by Treg cells and the propensity of Treg cells to produce this highly inflammatory cytokine. This study sought to address whether Th17 cells are inhibited by Treg cells in rheumatoid arthritis (RA) patients responding to anti–tumor necrosis factor (anti-TNF) therapy, and if so defining the underlying mechanisms of suppression.
Inhibition of Th17 cell responses was determined by Treg cell suppression assays. The Treg cell phenotype was analyzed using flow cytometry and enzyme-linked immunosorbent assay. Mechanisms of suppression were tested by cytokine addition or antibody blockade.
Th17 responses were inhibited by Treg cells from RA patients responding to the anti-TNF antibody adalimumab (Tregada), but not by Treg cells from healthy individuals or patients with active RA. Furthermore, Tregada cells secreted less IL-17, even when exposed to proinflammatory monocytes from patients with active RA. Tregada cells suppressed Th17 cells through the inhibition of monocyte-derived IL-6, but this effect was independent of IL-10 and transforming growth factor β, which mediated the suppression of Th1 responses. Adalimumab therapy led to a reduction in retinoic acid receptor–related orphan nuclear receptor C–positive Th17 cells and an increase in FoxP3+ Treg cells lacking expression of the transcription factor Helios. However, this acquisition of IL-17–suppressor function was not observed in RA patients responding to treatment with etanercept, a modified TNF receptor–Fc fusion protein. Indeed, there was no alteration in Treg cell number, function, or phenotype in etanercept-treated patients, and Th17 responses remained unchecked.
Th1 and Th17 responses are controlled through distinct mechanisms by Treg cells from patients responding to anti-TNF antibody therapy. Adalimumab therapy, but not etanercept therapy, induces a potent and stable Treg cell population with the potential to restrain the progression of IL-17–associated inflammation in RA via regulation of monocyte-derived IL-6.
The highly inflammatory cytokine interleukin-17 (IL-17) has attracted considerable attention for being pivotal in the pathogenesis of several autoimmune diseases, including rheumatoid arthritis (RA). The production of IL-17 by T cells defines a distinct T helper cell subset, Th17, governed by the specific transcription factor retinoic acid receptor–related orphan nuclear receptor C (RORC). Mice deficient in IL-17 show reduced severity of inflammatory arthritis (1), while those with increased levels of IL-17 have exacerbated disease (2, 3). In patients with RA, Th17 cells are elevated in the periphery (4) and have been shown to contribute to joint destruction in the synovium (5). IL-17 acts in synergy with tumor necrosis factor (TNF) to induce chemokine and cytokine production from synovial fibroblasts, resulting in cartilage destruction (6). In humans, Th17 cells can be induced by IL-6 and IL-1β (7–9), both of which are abundant in the inflamed joint in patients with active RA (10, 11).
IL-17 is relatively unusual among the proinflammatory cytokines, not only because it can be produced by functional, FoxP3-expressing Treg cells, but also because it has heightened resistance to Treg cell–mediated suppression (8, 12–15). Inflammation can impair the stability of Treg cells and exacerbate this tendency to secrete IL-17, resulting in conversion to a pathogenic phenotype (16–18). It is this broad expression during inflammation, combined with the relative inability of natural Treg cells to suppress its production, that marks IL-17 as an attractive therapeutic target for the control of inflammatory diseases.
Treg cells play a vital role in tolerance to self and have been shown to be defective in a number of autoimmune diseases. Previous studies by our group, and by others, have shown that Treg cells are unable to suppress the production of interferon-γ (IFNγ) in patients with active RA, in contrast to Treg cells from healthy individuals (19–21). However, following treatment of RA patients with a chimeric anti-TNF antibody, infliximab, a novel population of Treg cells expressing low levels of CD62L was identified (21). In this study, we show that treatment of RA patients with the fully humanized anti-TNF blocking antibody adalimumab results in an increased percentage of FoxP3+ cells endowed with the capacity to both suppress and resist conversion to Th17 cells via an IL-6–dependent mechanism. This was not observed in RA patients who did not respond to treatment with adalimumab, but more significantly, it did not occur in RA patients responding to treatment with the TNF soluble receptor etanercept, highlighting potential differences in the mechanisms of action of these agents, which may be relevant to not only their differing therapeutic effects but also the differing incidence of tuberculosis detected in RA patients treated with these 2 agents.
PATIENTS AND METHODS
We recruited 50 patients with active RA, whose diagnosis fulfilled the American College of Rheumatology 1987 revised classification criteria for RA (22), and 15 healthy volunteers for this study. The University College London Hospital Ethics Committee approved the study. Patients classified as having active disease were those awaiting anti-TNF therapy in whom the Disease Activity Score in 28 joints (DAS28) (23) was >5.1. Patients were considered responders if the DAS28 score fell by more than 1.2 points and the C-reactive protein levels were below 5 mg/liter after treatment.
The following antibodies were used: Alexa Fluor 700–conjugated CD4 (RPA-T4), allophycocyanin (APC)–Cy7–conjugated CD25 (M-A251), phycoerythrin (PE)–Cy5–conjugated CD62L (DREG-56), PE–Cy7–conjugated IFNγ (B27), V450-conjugated CD86 (2331), and PE–Cy7–conjugated CD80 (L307.4) (all from BD Biosciences); PE–Cy5–conjugated CD127 (ebioRDR5), Alexa Fluor 647–conjugated FoxP3 (PCH101), PE-conjugated mouse/human RORγt (AFKJS-9), PE-conjugated CD14 (61D3), APC-conjugated CD16 (eBioCD16), and Alexa Fluor 700–conjugated HLA–DR (LN3) (all from eBioscience); and PE-conjugated mouse/human Helios (22F6) and IL-17A (BL168) antibodies (both from BioLegend). T cells were activated using soluble CD3–specific antibodies (HIT-3a) and CD28–specific antibodies (CD28.2) (both from eBioscience). Neutralizing antibodies specific for human IL-10, transforming growth factor β (TGFβ), and IL-6 were obtained from R&D Systems.
Ficoll-Paque (GE Healthcare) was used to isolate peripheral blood mononuclear cells (PBMCs), which were then cultured in RPMI 1640 (Sigma). Specific cell populations were isolated by fluorescence-activated cell sorting (FACS) analysis (BD Aria).
Flow cytometric analysis and soluble cytokine detection.
Cell surface staining and intracellular staining were performed in accordance with the manufacturers' instructions. The eBioscience FoxP3 staining buffer set was used for all intracellular staining. The secretion of IL-17A was determined using enzyme-linked immunosorbent assay (R&D Systems) or cytometric bead array (BD Biosciences), and the secretion of IL-6 and IL-1 was determined by cytometric bead array (BD Biosciences).
Treg cell depletion assay.
Each sample was sorted into a population of whole PBMCs and PBMCs depleted of CD4+CD25+CD127− (Treg) cells. These populations were cultured at 3 × 105 cells per well in 96-well U-bottomed plates (Nunc) with 1 μg/ml soluble anti-CD3/anti-CD28 for 3 days before stimulation for 4 hours with 4 μg/ml phorbol myristate acetate, 1 μg/ml ionomycin, and 2 μl/ml GolgiStop (BD Biosciences). Cells were permeabilized and fixed using the eBioscience FoxP3 staining buffer for 20 minutes, and then incubated for 1 hour with antibodies to IL-17 and IFNγ.
Cells were sorted into populations of Treg cells and CD4+CD25−CD127+ responder T cells. Responder T cells were cultured with Treg cells in a 3:1 ratio, with a total of 1.3 × 105 cells per well. In assays in which monocytes were added, the cells were sorted based on CD14 expression and stained with PKH-26 (Sigma) to identify monocytes for purposes of exclusion in some analyses. Cells were cultured in the wells at a ratio of 1 monocyte:3 responder T cells:1 Treg cell. In all suppression assays, the cells were stimulated with 1 μg/ml soluble anti-CD3/anti-CD28 for 5 days, and the production of IL-17 and IFNγ was analyzed in the same manner as that used for the Treg cell depletion assay. For some experiments, Treg cells were separated from monocytes and responder T cells by a semipermeable membrane, using a transwell chamber.
Treg cell stability assay.
Samples were sorted into populations of Treg cells and CD14+ monocytes. Monocytes were identified by staining with PKH-26. A total of 3 × 104 Treg cells and 105 monocytes per well were cultured for 6 days with 2 μg/ml soluble anti-CD3/anti-CD28.
Statistically significant differences were determined using Student's t-test. P values less than 0.05 were regarded as statistically significant.
Increased percentage of Treg cells in adalimumab responders, but not etanercept responders, with a reciprocal reduction in Th17 cells.
We first sought to determine whether the increase in Treg cell numbers previously observed in patients treated with infliximab (21, 24) may also occur in patients treated with adalimumab, a fully humanized equivalent anti-TNF agent that is now used more commonly for RA therapy. Whereas the percentage of Treg cells in patients with active RA was similar to that in healthy controls, RA patients who responded to adalimumab therapy had an increased percentage of peripheral Treg cells (Tregada) (Figure 1A). In contrast, patients with persistently active disease despite treatment with adalimumab did not have an increased percentage of Treg cells.
To investigate whether this increase in the number of Treg cells is a typical feature of all types of anti- TNF therapies, the percentage of peripheral Treg cells from RA patients responding to etanercept (Tregetan) was determined for comparison. Unexpectedly, there was no difference between the numbers of Treg cells from patients with active RA and the numbers of Tregetan cells (Figure 1A).
Tregada cells were phenotypically distinct from Treg cells found in the other patient groups and in healthy control subjects. Indeed, there was a significantly higher proportion of Helios-negative Tregada cells (Figure 1B), a transcription factor associated with thymically derived Treg cells (25), when compared to that in the other groups. Treg cells from patients not responding to adalimumab did not show this reduced expression of Helios (Figure 1B). In addition, there were more Tregada cells that were negative for both Helios and CD62L, the latter of which has been previously identified as a characteristic marker of anti-TNF–induced Treg cells (21, 24), than there were in the active RA Treg cell and Tregetan cell groups (Figures 1B and C).
In addition, Tregada cells inhibited the production of IFNγ by autologous responder T cells, in contrast to the lack of inhibitory effects of Tregetan cells and active RA Treg cells on autologous responder T cells (Figure 1D), indicating that in the absence of increased levels of peripheral Treg cells, there is no acquisition of cytokine-suppressor function. Furthermore, we found that the suppression of the Th1 response by Tregada cells was dependent on TGFβ and IL-10 (Figure 1D).
We next explored whether this increase in FoxP3+ cell expression in adalimumab-treated patients could influence the expression of RORC, the master transcription factor of Th17 cells, given the unique reciprocal regulation and delicate balance between these opposing transcription factors and the T cell lineages under their control (26, 27). The expression of RORC was significantly elevated in patients with active RA compared to healthy controls, whereas the levels of RORC+ T cells were reduced 2-fold in patients treated with adalimumab, but not in those treated with etanercept (Figure 2A).
The increased ratio of FoxP3+ cells to RORC+ cells in adalimumab-treated patients compared to that in any other group confirmed the shift toward an immunoregulatory T cell phenotype (Figure 2B). Of note, RORC expression within the FoxP3+ cell population from adalimumab-treated patients was diminished compared to that from patients with active RA or patients responding to etanercept. Furthermore, the reduction in RORC expression in adalimumab-treated patients was associated with a reduced percentage of CD4+IL-17+ cells when compared to that in patients with active RA and patients responding to etanercept treatment (Figure 2C).
Reduced secretion of IL-17 by Tregada cells after culture under inflammatory conditions.
In order to test whether this reduced RORC expression was an indication that Tregada cells could resist the induction of IL-17 production under conditions of inflammation (16–18), we cultured active RA Treg cells, Tregada cells, and Tregetan cells in the presence of autologous monocytes. The purity of the isolated Treg cells from all groups, according to FoxP3 expression, was >90% (results not shown). Following 6 days in culture, Tregada cells produced significantly less IL-17 than cultures of monocytes with Tregetan cells or active RA Treg cells (Figure 3A). There was no difference in the FoxP3 expression between any of the groups, although FoxP3 expression is an unreliable marker of Treg cell function after 6 days of in vitro activation (28).
Ex vivo staining of monocytes from patients with active RA showed a greater proportion of CD14+CD16+ proinflammatory monocytes and, within this population, an increased expression of the activation markers CD80, CD86, and HLA–DR, in comparison to monocytes from patients treated with adalimumab (Figure 3B). We therefore tested whether Tregada cells would secrete more IL-17 if exposed to proinflammatory monocytes derived from patients with active RA. However, IL-17 production from Tregada cells remained low, irrespective of the provenance of the cocultured monocytes (Figure 3C). Although the purity of the Treg cells from the different patient groups was greater than 90%, it is possible that a minor proportion of contaminating effector T cells could have influenced these results.
Suppression of Th17 cell responses by Tregada cells via inhibition of monocyte-derived IL-6 production.
The reduced potential for IL-17 production by Tregada cells and the concomitant reduction in RORC expression in FoxP3− T cells in patients treated with adalimumab (Figure 2A) led us to test whether Tregada cells could suppress the production of this cytokine from responder T cells. When Treg cells were depleted from whole PBMCs of adalimumab-treated patients, production of IL-17 in CD4+ cells increased, suggesting that Tregada cells were inhibiting the Th17 response. In contrast, T cell IL-17 production was reduced in Treg cell–depleted PBMCs from patients with active RA or patients responding to etanercept (Figures 4A and B).
A classic coculture suppression assay, in which responder T cells were cultured with monocytes with or without Treg cells, confirmed that only Tregada cells suppressed Th17 cells (Figure 4C). Treg cells isolated from patients not responding to adalimumab lacked the capacity to suppress IL-17. Assaying the levels of IL-17 in the culture supernatants revealed that its production fell only when Tregada cells were added to the cultures of autologous responder T cells and monocytes (Figure 4D). Of interest, there was more IL-17 present in the supernatants of responder T cells and monocytes from adalimumab-treated patients compared to the other patient groups and healthy controls. This was consistent with the data presented in Figure 4B, in which the highest IL-17 production was detected in PBMCs depleted of Treg cells from adalimumab-treated patients.
Although blockade of TGFβ and IL-10 abolished the suppression of Th1 cells by Tregada cells, inhibition of Th17 cells was unaffected (Figure 4E), indicating that Tregada cells control IFNγ and IL-17 production through distinct mechanisms. Contact between Tregada and responder T cells was not required to elicit the suppression of IL-17, as detected by FACS analysis (Figure 4F) or cytometric bead array (Figure 4G), indicating that soluble factors other than TGFβ and IL-10 mediate the suppressive effects of Tregada cells.
We next looked for other pathways that could mediate the suppression of Th17 by Tregada cells. Given that monocytes are present in these Treg cell suppression assays specifically to promote Th17 cell responses (Figure 5A), we focused on the role of IL-6, IL-1β, and IL-23. The production of these cytokines, known to promote expression of Th17 cells (29), was measured in cultures containing monocytes and stimulated responder T cells from each patient group and healthy controls. Substantial quantities of IL-6, when compared to IL-1, were present in the patient groups, particularly in patients with active RA and patients responding to adalimumab (Figure 5B). IL-23 was undetectable in all of the culture supernatants (results not shown).
When Tregada cells were added to these cultures, IL-6 production was significantly suppressed (Figure 5C). In contrast, IL-6 production actually increased following the addition of either active RA Treg cells or Tregetan cells (Figure 5C), paralleling their ability to increase, rather than suppress, IL-17 production (Figure 4C). This suppression was not affected when contact between Tregada cells and their autologous responder T cells was prevented (Figure 5D).
We next investigated whether Tregada cells could directly inhibit monocyte-derived IL-6 in the absence of responder T cells. Production of IL-6 by monocytes isolated from patients with active RA was significantly lower when the monocytes were cultured with Tregada cells than when the same monocytes were incubated with active RA Treg cells or Tregetan cells, as measured by FACS analysis (Figure 5E) and as determined in the culture supernatants (Figure 5F). Antibody blockade of IL-6 in cultures of monocytes and responder T cells reduced the numbers of Th17 cells (Figure 5G) in a manner similar to that in cultures with Tregada cells. Moreover, the addition of IL-6 reversed the capacity of Tregada cells to inhibit IL-17 production (Figure 5H), confirming the critical role of IL-6 in controlling Th17 cells and mediating the inhibitory effects of these Treg cells.
The major finding from this study is that the Th17 response can be suppressed by Treg cells from patients who respond to treatment with adalimumab, via the control of monocyte-derived IL-6 production. Suppression of IL-17 was not apparent when Treg cells from healthy individuals were analyzed, confirming that this cytokine can be resistant to the effects of Treg cells (12). Thus, adalimumab therapy has the potential to regulate this highly inflammatory pathway via the induction of Th17-suppressing Treg cells, which was reflected in the shift away from RORC-expressing cells ex vivo in these patients.
RORC expression was also substantially diminished within the FoxP3+ cell population from adalimumab-treated patients, indicating that there is intrinsic suppression of RORC by FoxP3 in Tregada cells. Previous evidence has demonstrated that FoxP3 suppresses RORC transcription through a direct interaction (26, 30), but high RORC levels can also inhibit FoxP3 expression (27). Environmental cues, such as inflammation, are likely to control these 2 opposing transcription factors and their reciprocal regulation. However, highly inflammatory monocytes from patients with active RA had limited impact on the production of IL-17 by Tregada cells. It is possible that the disease exacerbation that eventually occurs following cessation of anti-TNF therapy is delayed due to the resistance of these Treg cells to conversion to Th17 cells. Indeed, one of the first studies to examine recurrence of disease activity once anti-TNF therapy was stopped found that those patients who were treated with infliximab, the chimeric equivalent of adalimumab, did not experience a disease flare as rapidly as those who received only methotrexate (31).
It is intriguing that the Treg cells induced by adalimumab therapy control the Th1 and Th17 pathways through separate mechanisms. Treg cells are thought to use multiple pathways to control target cells, but our data demonstrate, for the first time, that independent mechanisms of suppression can be utilized to regulate different human effector T cell lineages. Whereas IL-10 and TGFβ from Tregada cells controlled the production of IFNγ, the blockade of these cytokines had no effect on the ability of Tregada cells to suppress IL-17. Rather, an IL-10– and TGFβ-independent modulation of IL-6, which also did not rely on cell contact, appeared to be pivotal.
The pattern of suppression of IL-6 exactly matched that of IL-17, and the neutralization of IL-6 in cultures of responder T cells and monocytes from adalimumab-treated patients imitated the IL-17–suppressing abilities of Tregada cells. Moreover, the addition of IL-6 reversed the ability of Tregada cells to suppress Th17 cells. The activation of monocytes from patients treated with adalimumab was diminished compared to that in patients with active RA, which may reflect the active suppression of monocytes by Tregada cells, leading to a reduction in IL-6 and suppression of Th17 cells.
However, the blockade of TNF was, in itself, not sufficient to induce Th17-suppressing Treg cells, since etanercept does not modulate Treg cell numbers or function. Furthermore, the functional defect in Treg cells from patients with RA that we, and others, have previously described (19–21) is not simply a direct result of disease activity, given the equivalent clinical response to etanercept and adalimumab therapy. Indeed, the persisting Treg cell defect in patients responding to etanercept suggests that the inflammation associated with RA does not account for the failure of Treg cells to suppress autologous responder T cells, and supports our previous findings that responder T cells from patients with active RA are not resistant to Treg cell–mediated suppression (21). These findings support the notion that anti-TNF monoclonal antibodies and etanercept exert their therapeutic effects in RA through distinct mechanisms, and may explain why patients with RA who are unsuccessfully treated with one form of anti-TNF therapy may respond to another anti-TNF agent (32, 33).
Ostensibly, blockade of both the TNF and IL-17 pathways could confer a therapeutic advantage on adalimumab therapy, but could also render the patient more prone to infection, given the importance of both pathways in host defense. Patients treated with either adalimumab or infliximab have a 7–17-fold higher risk of developing tuberculosis (TB) when compared to patients treated with etanercept, most likely due to reactivation of the latent bacterium (34). Although IL-17 may have a limited role in primary infection, it appears to control both granuloma formation and bacterial burden, both of which would be critical to TB reactivation (35, 36). Treg cells can prevent an effective immune response to TB (37, 38), and therefore it is possible that Tregada cells, through the inhibition of Th1 and Th17 responses, would have a profound effect on TB immunity. In contrast, Treg cells from patients treated with etanercept are unable to modulate either Th1 or Th17 cells, consistent with the finding that this agent has little impact on the incidence of TB (34). The challenge will be to refine anti-TNF antibody therapy such that these induced Treg cells can permit the cessation of treatment for these inflammatory diseases, thereby not only inducing tolerance but also protecting patients from developing TB.
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. Professor Ehrenstein 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.
Study conception and design. McGovern, Notley, Ehrenstein.
Acquisition of data. McGovern, Nguyen.
Analysis and interpretation of data. McGovern, Nguyen, Mauri, Isenberg, Ehrenstein.
ROLE OF THE STUDY SPONSOR
Pfizer had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Pfizer.
We thank J. Evans for providing technical assistance with the FACS sorting. We also thank S. Moore, A. Ferenkeh-Koroma, A. Fox, and A. Davis for help with obtaining blood samples.