Runx1 and Runx3 Are Involved in the Generation and Function of Highly Suppressive IL-17-Producing T Regulatory Cells

CD4+Foxp3+ T regulatory cells (Tregs) display phenotypic and functional plasticity that is regulated by cytokines and other immune cells. Previously, we determined that during co-culture with CD4+CD25− T cells and antigen presenting cells, Tregs produced IL-17. Here, we investigated the mechanisms underlying the differentiation of IL-17-producing Treg (Tr17) cells and their molecular and functional properties. We determined that during stimulation via TCR/CD3 and CD28, the combination of IL-1β and IL-2 was necessary and sufficient for the generation of Tr17 cells. Tr17 cells expressed Runx1 transcription factor, which was required for sustained expression of Foxp3 and RORγt and for production of IL-17. Surprisingly, Tr17 cells also expressed Runx3, which regulated transcription of perforin and granzyme B thereby mediating cytotoxic activity. Our studies indicate that Tr17 cells concomitantly express Foxp3, RORγt, Runx1 and Runx3 and are capable of producing IL-17 while mediating potent suppressive and cytotoxic function.


Introduction
Tight regulation of effector T cell responses is required for effective control of infections and avoidance of autoimmunity. Tregs have essential roles in the maintenance of immune homeostasis. During the past few years a series of studies have shown that subsets of Tregs have the propensity to differentiate into Th17 or Tfh cells [1,2,3,4]. By genetic fate mapping using Foxp3-Cre mice to mark cells that expressed Foxp3 at some stage of development, at least one quarter of small intestinal IL-17 producing cells were found to express Foxp3 during their ontogeny [5]. It has been proposed that this might be the outcome of stochastic activation of Foxp3 expression or, alternatively, the coexpression of Foxp3 and IL-17 might mark a distinct differentiation pathway [6]. IL-17 + Foxp3 + T cells not only have been observed under experimental conditions but have been identified in the inflamed intestinal mucosa of patients with Crohn's disease and are associated colon cancer in the context of ulcerative colitis [7,8]. These findings strongly suggest that differentiation of IL-17 + Tregs is a natural process that can occur under physiologic or immunopathologic conditions in vivo.
Foxp3 plays a negative role in IL-17 expression through physical interaction with Runx1 and RORct, thereby inhibiting their transactivation activity [5,9]. Runx1 is highly expressed in CD4 + T lymphocytes [10] as well as Foxp3 + Tregs and interaction between Runx1 and Foxp3 is essential for the function of Foxp3 [11]. Runx1 also has a significant role in Th17 differentiation due to its ability to induce expression of RORct and to associate with and to act together with RORct to induce Il17 transcription [12]. Runx3 also regulates expression of Foxp3 [13,14]. However, in contrast to Runx1 that is expressed on CD4 + T cells, Runx3 is predominantly expressed in CD8 + T cells, and cooperates with Tbox proteins to establish the transcriptional program of effector cytotoxic T lymphocytes [15].
We have previously determined that during co-culture with CD4 + CD25 2 T cells in the presence of antigen presenting cells (APC), a fraction of Tregs differentiated into IL-17-producing, Tr17 cells [16]. In the present study we examined the mechanisms underlying the differentiation of Tr17 cells and their molecular and functional properties.

Tr17 cells retain phenotypic properties of Tregs
We have previously reported that during co-culture of CD4 + CD25 2 cells and Tregs in the presence of APC, IL-1b and IL-2, endogenous products of the co-culture, were essential for the generation of Tr17 cells [16]. To determine whether these cytokines were necessary and sufficient to induce differentiation of Tr17 cells, we used Foxp3.GFP knock-in mice to isolate highly purified Tregs. CD4 + Foxp3 + Tregs were cultured in an APC-free system using anti-CD3/CD28 mAbs in the presence of IL-1b, IL-2 or their combination. A significant amount of IL-17 secretion was observed only when a combination of IL-1b and IL-2 was added (thereafter termed polarizing culture conditions) (Fig. 1A). Intracellular staining confirmed the generation of IL-17 producing Tregs (Fig. 1B). Because IL-17 secretion is associated with expression of RORct [17], we assessed the expression of RORct. Activation of Tregs under polarizing conditions resulted in upregulation of RORct, while expression of Foxp3 was retained (Fig. 1C). Tr17 cells expressed significantly higher levels of RORct compared to Tregs that remained IL-17 2 during polarizing culture ( Fig. 1D and Fig. S1), consistent with previous reports on IL-17 + Foxp3 + T cells identified in different systems [18,19]. Thus although the expression of Foxp3 remained unchanged, the level of RORct expression in Tregs positively correlated with their ability to produce IL-17.
To determine whether Tr17 cells retained phenotypic properties of Tregs we examined well-established markers. Expression of CTLA-4 and GITR was comparable between Tr17 and IL-17 2 Tregs (Fig. 1D). We also examined expression of ICOS that has been suggested to define distinct subsets of Tregs [20]. Interestingly, Tr17 cells displayed increased ICOS levels as compared to IL-17 2 Tregs (Fig. 1D), suggesting that ICOS expression might identify Tregs that are capable of producing IL-17. Tr17 cells also expressed Helios (Fig. S2), a marker previously proposed to identify subsets of Treg cells [21,22].

Expression of Runx1 is required for the differentiation of Tr17 cells
Our results showed that Tr17 cells expressed enhanced levels of RORct while retaining expression of Foxp3. To understand the mechanism underlying this unique property of Tr17 cells, we focused on Runx1, a transcription factor that can regulate the expression of both Rorc and Foxp3. Expression of Runx1 in bulk Tregs was downregulated after culture with anti-CD3/CD28 mAbs (Fig. S3), similarly to previous observations in naïve CD4 + T cells [10]. Expression of Runx1 was also downregulated in bulk Treg polarizing cultures but to a lower extend ( Fig. 2A and Sig. S3). Strikingly, after gating on IL-17 producer Tregs we determined that Tr17 cells did not display Runx1 downregulation Tr17 Cells Produce IL-17 and Mediate Cytolysis PLOS ONE | www.plosone.org but, instead, had increased Runx1 expression compared to untreated Tregs. In contrast, Runx1 was downregulated in IL-17 2 Tregs (Fig. 2B). These results indicate that the increased Runx1 expression in Tr17 cells accounted for the less prominent downregulation of Runx1 expression in bulk Treg polarizing cultures. To determine the role of Runx1 in the generation of Tr17 cells, we attenuated expression of Runx1. Silencing of Runx1 with small interfering RNA (siRNA) during polarizing culture, significantly reduced the numbers of IL-17 + cells (0.423%) compared to transfection with control 'scrambled' siRNA (1.85%) (Fig. 2C), suggesting that Runx1 plays an important role in regulating IL-17 production in Tregs. Moreover, expression of Rorc and Foxp3 mRNA was significantly reduced in cells transfected with Runx1-specific siRNA (Fig. 2D). Thus, Runx1 plays an essential role in the generation of Tr17 cells by regulating expression of both Rorc and Foxp3 transcription factors.
Previous studies in Th17 cells have identified the IL-17 promoter and one conserved noncoding (enhancer) sequence, CNS-5, as the regions required for the optimal transcription of the Il17 gene [12]. Both these regulatory elements contain sequences that bind the transcription factors RORct and Runx1. Because Tr17 cells express Foxp3, which has the capacity to interact with RORct and Runx1, we examined whether this physical interaction might interfere with the ability of these IL-17 regulatory transcription factors to bind on the relevant regulatory elements of the Il17 gene, using chromatin immunoprecipitation (ChIP). Binding of RORct and Runx1 to the Il17 promoter and enhancer was detected in Tregs cultured under polarizing conditions (Fig. 2E, F). However, binding of RORct and Runx1 to these elements in Tr17 was diminished compared to Th17 cells (Fig. S4). To determine whether Foxp3 might indeed form complexes with RORct and/or Runx1 bound on the Il17 promoter and enhancer, we did ChIP using Foxp3-specific antibody. Foxp3 displayed binding to the Il17 promoter and enhancer in Tr17 cells ( Fig. 2E and F) but not in Th17 cells (Fig. S4). Because no Foxp3 binding sites are present on the Il17 promoter and enhancer [12], these findings suggest that Foxp3 was detected on these elements due to complex formation with RORct and/or Runx1 bound on these regions of the Il17 gene.

Tr17 cells display potent suppressive activity
To determine whether Tr17 retained their functional role as Tregs, we performed suppression assay. Upon culture of CD4 + CD25 2 responder T cells (Tresp) with anti-CD3/CD28 mAbs, approximately 37% of Tresp underwent at least one round of cell division (Fig. 3A, first panel). Only 19% of Tresp divided when they were co-cultured with freshly isolated Tregs (Fig. 3A, second panel) and 16% of Tregs divided when they were co-cultured with Treg that were previously stimulated with anti-CD3-plus-anti-CD28 in vitro (Fig. 3A, third panel). Interestingly, further reduced proportion of Tresp (11%) proliferated when co-cultured with Treg that had been previously incubated under polarizing culture conditions (Fig. 3A, forth panel), suggesting that the inhibitory capacity of these Tregs was enhanced. In order to analyze more accurately the suppressive function of Tr17 cells, we used cytokine capture to isolate highly purified populations of IL-17+ (Tr17) and IL-172 Tregs. Suppression assay indicated that Tr17 cells displayed more potent inhibitory function compared to IL-172 Tregs and this effect was dose dependent (Fig. 3B, C). Specifically, at Treg: Tresp ratio of 1:8, co-culture of Tr17 with Tresp led to only 3% of Tresp cells undergoing cell division, whereas twice as many Tresp proliferated when co-cultured with IL-17 2 Tregs. Furthermore, at Treg: Tresp ratio of 1:1, co-culture of Tr17 with Tresp led to almost undetectable response. During this analysis, we noted an increased percentage of dead cells in the co-cultures of Tresp with Tr17 as compared to IL-172 Tregs. To distinguish between anti-proliferative and cytotoxic effects of Tregs, we simultaneously assessed induction of cell death in CFSE-labeled Tresp co-stained with DAPI. There was a dose-dependent increase in Tresp cell death when co-cultured with Tregs and this effect was more prominent upon co-culture with Tr17 cells (Fig. 3B).

Perforin and granzyme B are highly expressed in Tr17 cells
One of the proposed mechanisms for Treg-mediated suppression involves cytolysis of Tresp cells via the granzyme B-perforin pathway [23,24,25]. We investigated whether Tr17 cells expressed granzyme B and perforin and utilized this pathway to induce cytolysis of Tresp. Flow cytometry analysis revealed that Tr17 cells expressed higher levels of perforin and granzyme B compared to IL-17 2 Tregs (Fig. 4A). To identify the degranulation capacity of Tr17 cells we assessed CD107a (LAMP-1), a marker of degranulation in CD8 + and NK cells [26,27]. After polarizing culture, Tregs were cocultured with Tresp isolated from B6 congenic mice (Thy1.1 + ) and expression of CD107a was examined. Mean fluorescence intensity (MFI) of CD107a in Tregs was 4-fold higher than in Tresp cells (Fig. 4B, bottom, left panel). Furthermore, expression of CD107a in Tr17 cells was significantly enhanced compared to IL-17 2 Tregs (Fig. 4B bottom, right panel), indicating that degranulation is a function preferentially exerted by Tr17 cells.
Runx3 plays a critical role in CTL differentiation and regulates expression of Gzmb and Prf1 either as a sole factor or in synergy with other transcription factors [15]. Expression of Runx3 was low in freshly isolated Tregs and was only slightly upregulated in bulk Tregs during polarizing culture (Fig. 4C). However, assessment of Runx3 on Tr17 and IL-17 2 Tregs indicated that Tr17 cells expressed significantly higher levels of Runx3 protein (Fig. 4C). To determine whether Runx3 had a causative role in regulating expression of granzyme B and perforin in Tregs, we used Runx3specific siRNA. Runx3 siRNA reduced Runx3 mRNA expression during polarizing culture by 47% (Fig. 4D). A comparable reduction was also observed in granzyme B and perforin mRNA (Fig. 4D) indicating that Runx3 had an active role in regulating these genes in Tr17 cells.

Discussion
In our present study we determined that similarly to Th17 cells, in Tr17 cells Runx1 and RORct bind on the previously identified binding elements of the Il17 gene but the abundance of these transcription factors on the relevant DNA elements is significantly lower. Tr17 cells express Foxp3, which plays a negative role in IL-17 expression because via its physical interaction with RORct and Runx1 inhibits their transactivation activity [5,12]. Our findings strongly suggest that the reduced abundance of RORct and Runx1 on the IL-17 promoter and enhancer of Tr17 cells might be related to their association with Foxp3, which was also detected in the same DNA elements by CHIP in Tr17 but not in Th17 cells.
Runx1 and Runx3 transcription factors play key roles in T cell development and differentiation [28]. Runx1 is highly expressed in resting CD4 + T cells but signals mediated by TCR during activation rapidly downregulate Runx1 via negative autoregulation of the distal Runx1 promoter [10]. Runx1 plays an essential role in the generation and function of Treg and Th17 cells [11,12]. It was previously shown that Runx1 regulates Il17 transcription in CD4 + T cells [12]. However, in that study the expression pattern of endogenous Runx1 during differentiation of CD4 + T cells to Th17 was not investigated. In our studies we examined the expression pattern of endogenous Runx1 in Tregs during their differentiation to Tr17. We found that in the presence of IL-1b and IL-2, expression of Runx1 not only was retained but was even increased compared to unstimulated Tregs. Furthermore, silencing Runx1 expression by siRNA reduced the number of Tr17 cells. Our results strongly suggest that sustained expression of Runx1 in Tregs is regulated by polarizing cytokines and is required for the differentiation of Tregs to Tr17 cells.
An additional novel and unexpected finding of our studies was that Tr17 cells expressed increased levels of perforin and granzyme B, and displayed enhanced cytotoxic activity compared to IL-17 2 Tregs. Furthermore, Tr17 cells displayed increased expression of Runx3, which is involved in the expression of Prf1 and Gzmb genes, thereby regulating cytolytic function. Unlike Runx1, Runx3 is not expressed in naïve CD4 + T cells but is induced during Th1 differentiation [29]. Runx3 is also expressed in CD8 + T cells and regulates transcription of Prf1 and Gzmb through binding on multiple regulatory regions of both genes, including their transcription start sites [15]. Runx3 has no effect on the differentiation of Th17 cells [12]. In contrast to this finding in Th17 cells, our results showed increased expression of Runx3 in Tr17 cells, which was functionally important. In CD8 + T cells, expression of Runx3 requires the presence of T-bet and is necessary for the induction of Eomes. In that setting, Runx3 cooperates with T-bet to regulate transcription of granzyme B and with Eomes to regulate transcription of perforin [15]. Further studies will determine whether Runx3 is downstream of T-bet and whether it regulates expression of Eomes in Tr17 cells.
The functional efficacy of Tregs in vivo is programmed by environmental cues. Such functional adaptation may improve the ability of Tregs to control different classes of immune responses driven by distinct types of T effector cells. For example, in response to interferon-c Tregs express increased levels of Th1-specific transcription factor T-bet, leading to expression of the chemokine receptor CXCR3, which enables T-bet + Tregs to share similar migratory properties with Th1 cells and to accumulate at the same sites thereby controlling Th1-mediated inflammation [30]. It is intriguing to speculate that Tr17 cells might be differentiated in an inflammatory microenvironment and might be programmed to regulate Th17-mediated inflammatory responses.
In conclusion, our studies indicate that Tr17 cells concomitantly express Foxp3, RORct, Runx1 and Runx3 and are capable of producing IL-17 while mediating potent suppressive function. These findings might have significant implications in our understanding of how environmental cues translate into determinants of Treg fate and might guide the development of therapeutic strategies that use the administration of Tregs for the treatment of autoimmune diseases.

Ethics Statement
All experiments were conducted in accordance with National Institutes of Health guidelines. The studies were reviewed and approved by the Institutional Animal Care and Use Committee (Approved Protocol Number: #122-2010).

Suppression Assay
To assess suppressive activity of IL-17 2 Tregs and Tr17 cells, after polarizing culture IL-17 + cells were enriched by cytokine according to the manufacturer's protocol. Flow cytometry-sorted IL-17 + GFP + and IL-17 2 GFP + cells were used for suppression assay using as responder CD4 + CD25 2 cells from B6.PL congenic mice labeled with CFSE and DAPI. Percent of dividing cells was calculated using a described formula [31].

Degranulation Assay
Tregs were stimulated with PMA plus ionomycin for 4 h and were cultured with CD4 + T cells from B6.PL mice and anti-CD3/ CD28 mAbs in the presence of PE-conjugated antibody against CD107a (BD Pharmingen, clone 1D4B). Cells were collected, stained with antibodies against Thy1.1 and IL-17 and analyzed by flow cytometry.

ChIP assay
Tregs and CD4 + T cells were cultured for 4 days under the indicated conditions and ChIP assay was performed using Pierce Agarose ChIP kit (Thermo Scientific) according to manufacturer's instructions. Antibodies used for IP were specific for Runx1 (Active Motif), RORct (Santa Cruz Biotechnology), Foxp3 (Santa Cruz Biotechnology) or IgG isotype control.

Statistical analysis
To assess statistical significance of quantitative differences of various parameters, t-test was used. Differences were considered statistically significant for p values ,0.05 (* corresponds to p,0.05; ** corresponds to p,0.001). Figure S1 Characterization of Tr17 cells. Purified Foxp3.GFP + CD4 + T cells were cultured with anti-CD3/CD28 (PDF) Figure S4 RORct and Runx1 but not Foxp3 bind to the Il17 promoter and enhancer during Th17 differentiation. Binding of endogenous RORct, Runx1 and Foxp3 to the Il17 promoter and enhancer was assessed by ChIP with antibodies against RORct, Runx1 and Foxp3 in CD4 + T cells cultured for 4 d with aCD3/CD28 in the presence of TGF-b and IL-6 (Th17polarizing conditions) and restimulated for 4 h with PMA and ionomycin. Results were normalized to input and are shown asfold increase in signal relative to isotype control IgG used as ChIP negative control. Bar graphs show mean values and SD of triplicate samples from one out of two experiments. (PDF)

Supporting Information
Author Contributions