Induced Treg Cells Augment the Th17-Mediated Intestinal Inflammatory Response in a CTLA4-Dependent Manner

Th17 cells and Foxp3+ regulatory T cells (Tregs) are thought to promote and suppress inflammatory responses, respectively. However, whether they counteract each other or synergize in regulating immune reactions remains controversial. To determine their interactions, we describe the results of experiments employing mouse models of intestinal inflammation by transferring antigen-specific Th cells (Th1, Th2, and Th17) differentiated in vitro followed by the administration of the cognate antigen via enema. We show that cotransfer of induced Tregs (iTregs) suppressed Th1- and Th2-mediated colon inflammation. In contrast, colon inflammation induced by transfer of Th17 cells, was augmented by the cotransfer of iTregs. Furthermore, oral delivery of antigen potentiated Th17-mediated colon inflammation. Administration of a blocking antibody against cytotoxic T lymphocyte-associated antigen 4 (CTLA4) abrogated the effects of cotransfer of iTregs, while the injection of a soluble recombinant immunoglobulin (Ig) fusion protein, CTLA4-Ig substituted for the cotransfer of iTregs. These results suggest that antigen-specific activation of iTregs in a local environment stimulates the Th17-mediated inflammatory response in a CTLA4-dependent manner.


Introduction
Accumulating evidence indicates that CD4 + helper T cells play a central role in eliciting normal immune responses and in inducing inappropriate reactions leading to allergy and autoimmune diseases [1]. For example, CD4 + regulatory T cells (Tregs) that express the transcription factor FoxP3 represent a distinct cell population with immunnosuppressive function [1][2][3]. In contrast, effector CD4 + helper T cells are classified mainly into Th1, Th2, and Th17 subsets that induce physiological immune responses depending on the infectious pathogens. Unless attenuated after elimination of pathogens, or maintained tolerance to self or innocuous antigens, activation of these effector subsets initiates allergic or inflammatory disorders. The idea that an aberrant Th2-type immune response induces allergy and is regulated by FoxP3 + Tregs is consistent with the results of studies on humans and numerous mouse models [4][5][6]. In contrast, the pathogenic role of Th17 cells on the development of autoimmune and inflammatory disorders remains controversial although the vast majority of recent findings from genome-wide studies of humans and mouse models support the intimate involvement of this subset in promoting the diseases [7][8][9]. This ambiguity may be explained as follows. First, most studies employ mouse models, including spontaneous occurrence of the diseases, which are driven by combinations of various T cell subsets, resembling human disease [10], which impedes the evaluation of the contribution of Th17 cells to pathogenesis. Second, the properties of Th17 cells are diverse and highly plastic in terms of immunological functions, including immune suppression under certain conditions [11][12][13]. Therefore, whether Th17-type immunity is susceptible to immunological tolerance or suppression mediated by FoxP3 + Tregs remains largely unknown. Moreover, evidence indicates that Tregs support the development of Th17 cells or promote Th17-mediated immunological responses [14][15][16][17][18] by secreting TGF-beta [19] or by consumption of IL-2 [17,18]. Irrespective of the outcomes of interactions between Th17 cells and Tregs, the role of antigen specificity must be considered. Therefore, to delineate the outcomes caused by one-to-one interactions between iTregs and each effector T cells from otherwise complex immunological responses, we employed a model in which antigen-specific CD4 + T cells are adoptively transferred in combination followed by antigen delivery. We show here that the differential effects of iTregs depending on the effector subsets, and that CTLA4 is critically involved in both processes, inhibition of Th1/Th2-mediated colon inflammation and stimulation of Th17-mediated colon inflammation.

Results and Discussion
Antigen-specific effector cells induce colon thickening CD4 + T cells were obtained from spleen and mesenteric lymph nodes of DO11.10 transgenic mice with a Rag2-deficient background (DO11.10 + :Rag2 KO). Cells were differentiated in vitro in a mutually exclusive manner to an interferon (IFN)-gamma-producing Th1 subset, an interleukin (IL)-4-producing Th2 subset, and an IL-17A-producing Th17 subset under each polarizing condition described in Materials and Methods (S1A Fig) [20]. Approximately 2 × 10 7 viable cells of each subset were intravenously transferred to wild-type BALB/c mice, and ovalbumin (OVA) was administered via enema once a week for 4 weeks. One day after the last administration of OVA, colon length and weight were measured and the length-to-weight ratio (colon thickness index, CTI), which correlates with histological scores [21], was calculated. Regardless of the Th subset (Th1, Th2, Th17) derived in vitro, all effector subsets increased CTI values, suggesting that each possesses intrinsic activities that induce an inflammatory response in the intestinal tract (Fig 1A and 1B). Furthermore, the induction of colon thickening depended on transfer of cells and delivery of antigen (S2 Fig).
Accumulation of eosinophils defined as CCR3 + CD11b + mononuclear cells in the colon lamina propria (cLP) was more prominent in mice that received Th2 cells (Fig 1C), which is consistent with the increased production of IL-5 (S1B Fig). CCR3 + CD11b + populations specifically expressed the eosinophil marker Siglec-F ( Fig 1D). In contrast, the number of CD11b + Gr-1 + cells, which are thought to be neutrophils involved in Th17-type immunity [22,23], were slightly increased in the cLP of mice that received Th17 cells compared with those in the cLP of mice that received Th1 and Th2 cells (Fig 1C).
Antigen-specific iTregs stimulate Th17-mediated colon thickening Next, we examined the susceptibility of effector subsets to the suppressive function of iTregs that recognized the same epitope. For this purpose, OVA323-339 epitope-specific iTregs that expressed FoxP3 predominantly and increased levels of CD25 and CTLA4 were prepared Adoptive transfer of effector T cells induces colon thickening. Each effector T cell subset (2 × 10 7 viable cells) was intravenously transferred to wild-type BALB/c mice, and OVA protein was administered via enema once a week for 4 weeks. The day after the last OVA challenge, the colonic weight-to-length ratio (mg/mm) was calculated as the colon thickness index (CTI) to evaluate the inflammatory response. (A) Three types of effector cells (Th1/Th2/Th17) induced CTI using this model mouse. (B) Representative histological (S1A Fig). Data acquired using similar models, that employ cotransfer of antigen-specific iTregs, followed by the administration of an antigen, regardless of target organ such as the respiratory or digestive tract [24][25][26], suggest that Th2 cells are highly susceptible to iTregs, however, the susceptibility of Th17 cells to the cotransfer of iTregs has been an arguable issue [27]. We show here that iTregs almost completely suppressed Th1 and Th2 cell-mediated colon thickening, although transfer of iTregs alone had no effect on CTI (Fig 2A), consistent with previous reports [24][25][26]. In contrast, simultaneously prepared iTregs that suppressed Th1 and Th2 cell-mediated colon thickening, stimulated Th17-mediated colon thickening. Accumulating evidence indicates that lineage stability of iTregs in vivo is not as robust as expected [28]. Therefore it might be possible that iTregs co-transferred with Th17 cells, but not with Th1 or Th2 cells may convert to Th17-like population with some effector function. This issue remains to be clarified further.

Oral-administration of OVA stimulates Th17-mediated colon thickening
Oral administration of antigen suppresses not only gut mucosal but also systemic immune responses, particularly Th2-type allergic reactions against challenge with the same antigen (oral tolerance) [29][30][31]. Moreover, the function(s) of FoxP3 + Tregs are prerequisite for establishing oral tolerance to allergic immune reactions such as allergic diarrhea and ear swelling induced by protein antigens [32]. However, it remains unknown whether immune responses derived solely from Th17 cells are affected by oral tolerization. We addressed this issue considering the stimulatory effects of iTregs on the Th17-mediated colon thickening described above. We applied a standard protocol to induce oral tolerance, by providing a continuous supply of OVA in drinking water for 7 days. This procedure resulted in the increased FoxP3 + ratio of transferred naïve DO + T cells not only prepared from cLP but also from spleen (S3A Fig). However, FoxP3 + ratio of endogenous CD4 + T cells was unaltered, presumably because of the scarcity of the OVA-specific CD4 + T cells detectable in this analysis (S3B Fig). Consistent with conventional findings, Th2-and Th1-mediated colon thickening were highly susceptible to immune suppression by prior oral administration of OVA, although the Th17-mediated increase in CTI was not inhibited but accelerated by tolerization ( Fig 2B). Furthermore, histological analysis indicated the contrasting effects of oral tolerance on Th1/Th2-and Th17-mediated pathology.
Colon thickening is induced less efficiently by IL17A-deficient Th17 cells IL-17A, which is expressed specifically by Th17 cells, plays a pivotal role in Th17-type immune responses, particularly in the recruitment of neutrophils to inflamed sites [33]. In fact, there are a couple of reports showing that IL-17A deficiency abrogated the immunopathology driven by Th17 cells [34]. Therefore, we determined the relative contribution of IL-17A to the Th17-mediated immune response in our experimental settings. For this purpose, we employed eosinophil-deficient mice as recipients to evaluate the accumulation of CD11b + Gr-1 + cells in the cLP in addition to the induction of colon thickening. Th17 cells derived and differentiated from IL17A-deficient mice mediated these processes, although to a diminished but significant images of HE-stained mid-colonic sections are shown. Scale bars indicate 500 microm. (C) Mononuclear cells were isolated from the spleen and cLP of recipient mice. CD11b + CCR3 + and CD11b + Gr-1 + cells were gated, and their frequencies (%) were determined using flow cytometric analysis. Representative data of three independent experiments are shown. (D) Mononuclear cells were isolated from the cLP of recipient mice and stained with monoclonal antibodies (mAbs) against CD11b, CCR3, Gr-1, and Siglec-F. The frequencies (%) of Siglec-F + Gr-1 middle cells in the total population of CD11b + CCR3 + cells are shown doi:10.1371/journal.pone.0150244.g001 iTregs Enhance the Th17-Immune Response via CTLA4 PLOS ONE | DOI:10.1371/journal.pone.0150244 March 7, 2016 extent ( Fig 3A). Splenomegaly was present not only in mice engrafted with IL17A-sufficient Th17 cells but also in mice engrafted with IL17A-defficient Th17 cells ( Fig 3A). The expression of IL-17F, which is most closely related to IL-17A, and the lineage-restricted transcription factor RORgammat were comparable to those of IL17A-sufficient Th17 cells in vitro (Figs 2, 3B and 3C). Moreover, migration to the cLP and the probability of survival were unaltered in vivo Co-transfer of iTregs or oral administration of OVA stimulates colon thickening mediated by Th17. (A) Each effector cell (2 × 10 7 cells per mouse: Th1, Th2, Th17) was adoptively transferred or not (none) with or without iTregs (1 × 10 7 cells per mouse) into wild-type BALB/c mice and each mouse was immunized with OVA as described in Fig 1. CTI values are shown as the mean and standard error (SE). Independent experimental sets were designed for histological analysis, and representative images are shown. (B) Before adoptive transfer, BALB/c mice were continuously supplied with OVA (1 mg/mL) in their drinking water for 7 days to induce oral tolerance (indicated as OVA-fed). Each effector cell (2 × 10 7 cells/ mouse: Th1/Th2/Th17) was adoptively transferred, and mice were treated as described in Fig 1. CTI and histological analysis were performed as described above.  Eosinophil-deficient deltadblGATA mice were engrafted and treated with OVA as described above. Spleen weights were measured and analyzed (n = 4). The weight-to-length ratio of the colon was calculated and expressed as CTI. Mononuclear cells of the iTregs Enhance the Th17-Immune Response via CTLA4 ( Fig 3D). Therefore, the induction of colon thickening and accumulation of CD11b + Gr-1 + cells by engrafting Th17 cells may be due to the redundant functional sums of the activities of IL-17A and other Th17-related cytokines. In fact, a couple of cytokines other than IL-17F, which are reported to be involved in Th17-mediated pathology, were more expressed in IL17A-deficient Th17 cells at least in vitro (Fig 3B), suggesting the intrinsic effect of IL-17A on the cytokine profile produced by Th17 cells. Thus the pathology via transferring IL17A-deficient Th17 cells might be driven by these cytokines.
Cotransfer of iTregs inhibits Th2-mediated colon thickening and stimulates Th17-mediated colon thickening using eosinophil-deficient mice as recipients Transfer of effector T cells, particularly Th2 cells, led to the accumulation of eosinophils in the cLP, which is considered an inflammatory response of the intestinal tract (Fig 1). Therefore, we attempted to evaluate the involvement of eosinophils in the induction of CTI using eosinophildeficient mice as recipients. Both Th17 cells and Th2 cells induced colon thickening in eosinophil-deficient recipients (Fig 4A), indicating that eosinophils residing in the cLP were dispensable for the induction of CTI or that other granulocytic populations may compensate in eosinophil-deficient mice. Moreover, the accumulation of CD11b + Gr-1 + cells was increased in eosinophil-deficient mice engrafted with Th17, but not Th2 cells (Fig 4B). These observations indicate that the accumulation of neither eosinophils nor neutrophils directly leads to colon thickening, although subset-specific immune reactions may occur. In contrast, differential effects of iTregs on Th2 and Th17 cell-mediated colon thickening were reconfirmed using eosinophil-deficient mice as recipients ( Fig 4A). Furthermore, CD11b + Gr-1 + cells were much more abundant in mice that received iTreg and Th17 cells together compared with those that received Th17 cells alone (Fig 4B), indicating that iTregs play a dominant role in determining the stimulatory or inhibitory effect on colon thickening, irrespective of the type of granulocytes residing in the cLP.
Several kinds of molecular apparatus have been proposed to account for the immunomodulatory function of FoxP3 + Tregs [2,35]. Of these, we tested the contribution of IL-10 expressed intrinsically in CD4 + T cell to colon thickening. The Il10-deficiency of Th2 cells induced a hyper-Th2 phenotype, that produced massive amounts of Th2-related cytokines in vitro (S4A and S4B Fig)  Anti-CTLA4 antibody abrogates the effects of iTregs and a CTLA4-Ig fusion protein mimics iTreg function Although effector T cells other than Tregs express CTLA4 after stimulation [36], FoxP3 + cellrestricted deletion of Ctla4 leads to a sub-lethal multifocal inflammatory disorder similar to that caused by systemic deletion of Ctla4, albeit with a later age of onset [37]. This finding indicates that FoxP3 + Tregs require CTLA4 to restrain immune responses. We first employed an anti-CTLA4 antibody to evaluate the contribution of CTLA4 to iTreg-mediated modulation of colon thickening. Administration of the anti-CTLA4 antibody simultaneously cell transfer abrogated the effects of the iTregs, i.e., suppression of Th2-mediated colon thickening and enhancement of Th17-mediated colon thickening (Fig 5A). Histological observations were consistent with CTI values. It might be possible that anti-CTLA4 antibody directly targets Th17 cells or Th2 cells since CTLA4 is induced after activation of the effector T cells. However, administration of anti-CTLA4 antibody does not have significant effect on Th2-or Th17-mediated CTI induction in the absence of iTregs (Fig 5B). In addition, anti-CTLA4 antibody shows no effect on viability of engrafted iTregs or status of FoxP3 expression in vivo (S5 Fig). Therefore, it is likely that anti-CTLA4 antibody hampers the function of CTLA4 expressed on iTregs as well as, if any, endogenous Tregs. In this context, we were intrigued that FoxP3 + cell-specific deletion of Ctla4 resulted in an increase of the number of IFN-gamma + or IL-4 + cells, but not that of IL-17 + cells [37], suggesting CTLA4 expressed on FoxP3 + cells plays a less prominent role in regulating the Th17-type response, yet apparent functional role in suppressing Th1-and Th2-type immune responses. Furthermore, deletion of Stat3 from FoxP3 + cells induces hyperactivation of Th17 cells in vivo, while Stat3-deficient Tregs express higher levels of Ctla4 mRNA compared with iTregs Enhance the Th17-Immune Response via CTLA4 iTregs Enhance the Th17-Immune Response via CTLA4 Stat3-sufficient Tregs do [38], suggesting that CTLA4 expressed by FoxP3 + cells may have little effect on suppressing Th17-type immunity or otherwise have a function to promote Th17-type immunity.
In contrast, recent reports show that Th17 cells acquired a follicular helper T cell (T FH cell) phenotype in vivo to induce an antigen-specific IgA production [39] or ectopic lymphoid follicles [40]. Consistent with these findings, transfer of Th17 cells increased OVA-specific IgA production, although Th1 and Th2 cells exhibited enhancement of OVA-specific IgA production compared with the control (S6A Fig). Th17-driven OVA-specific IgA secretion, however, was diminished by cotransfer of iTregs and abrogated again by simultaneous administration of the anti-CTLA4 antibody (S6B Fig). This suggests that the versatile functions of Th17 cells in vivo are distinct in terms of susceptibility to the iTreg-CTLA4 axis.
CTLA4 is regarded as an immunosuppressive molecule and numerous studies elucidating its molecular action [41,42] have advanced the development of therapeutic applications for various immune-related disorders as well as anti-tumor immunotherapy [36,43]. For example, the soluble fusion protein CTLA4-immunoglobulin (CTLA4-Ig) suppresses immune reactions in vitro and in vivo, presumably by binding to costimulatory ligands expressed by antigen presenting cells to block CD28 signaling in effector T cells [44]. In fact, CTLA4-Ig is effectively used to treat autoimmune and inflammatory diseases such as rheumatoid arthritis [45,46]. Therefore, we addressed the effect of CTLA4-Ig on Th2-or Th17-driven intestinal immune responses. Th2 and Th17-mediated colon thickening responded differentially to the administration of CTLA4-Ig, in other words, Th2-mediated colon thickening was inhibited and at the same time and doses, Th17-mediated colon thickening was accelerated (Fig 5C). In addition, the frequency of DO + Th17 cells, but not of DO + Th2 cells among CD4 + T cells was slightly increased in mice administered CTLA4-Ig (S7 Fig). These results indicate that restriction of costimulatory ligand availability and following inhibition of CD28 signaling, both of which are induced by CTLA4-Ig, lead to the opposite outcomes depending on the effector T cell subsets. This finding seems to be compatible with that of study showing that the differentiation of Th17 cells is blocked by an anti-CD28 antibody, indicating the adverse effect of CD28 signaling on the development of Th17 cells, at least in vitro [47]. In fact, we showed that differentiation of DO + Th17 cells as well as iTregs from naïve CD4+T cells was accelerated in the presence of CTLA4-Ig in terms of IL-17A + ratio and FoxP3 + ratio, respectively (Fig 6A and S8 Fig), which is consistent with a report mentioned above [47]. Moreover, we found that even in the re-stimulation phase the addition of CTLA4-Ig maintains or rather augments the frequency of IL-17A-producing cells (Fig 6B). These results might indicate that administration of CTLA4-Ig favor the differentiation of Th17 cells through curbing of CD28 signaling, which is reminiscent of the function of TGF-beta in this process [48]. In this context, we examined the effect of anti-TGF-beta antibody in conjunction with CTLA4-Ig in our in vitro system. As a result, the addition of CTLA4-Ig increased the frequency of IL17A-producing cells even in the presence of anti-TGF-beta antibody in both process, differentiation from naïve to Th17 cells ( Fig 6C) and re-stimulation of Th17 cells (Fig 6D). These observations seem to be consistent with the scenario that both immunomodulatory molecules, CTLA4 and TGF-beta act independently on the cell surface but following events merge to inhibit CD28-signaling cascade within the cells.
In this context, we were intrigued that mice administered an agonistic anti-CD3epsilonantibody, which induces TCR stimulation without CD28 signaling, shows preferential accumulation of IL-17A-producing CD4 + T cells in the small intestine [11]. In the aggregate, Th17-type immune responses may favor the signal input delivered via the T-cell receptor under conditions of costimulation blockade. Therefore, targeting the CTLA4/CD28 signaling pathway should be carefully evaluated in relation to the presence of pathogenic effector subsets and target organs. In this respect, it is worth to note that in phase III trials, CTLA4-Ig had no significant benefit for patients with inflammatory bowel diseases and, on the contrary, exacerbated ulcerative colitis [49]. Moreover, it was reported that patients with rheumatoid arthritis developed ulcerative colitis during the treatment of CTLA4-Ig [50,51].
The relevance of Th17-driven colon inflammatory responses described in the present study must be considered from the perspective of the pathogenesis of inflammatory bowel disease in the future. Taken together, the results of the present study demand reconsideration of Treg/CTLA4-based immunological modulation to suppress or treat autoimmune diseases, particularly in patients with Th17-driven intestinal inflammation.

Materials and Methods Mice
Balb/c mice were purchased from Nihon SLC (Shizuoka, Japan), DO11.10×Rag2-KO mice, Il10-KO mice, and eosinophil-deficient (deltadblGATA) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Il17A-KO mice were provided by Dr. Iwakura. Il10-KO mice and Il17A-KO mice were backcrossed with Balb/c genetic background more than 10 times in our facility to produce Il10-KO or Il17A-KO×DO11.10×Rag2-KO mice. Female mice 8 to 12 weeks older were used and bred in specific pathogen-free facilities at the Tokyo Metropolitan Institute of Medical Science. This study was carried out in strict accordance with the guidelines in the Proper Conduct of Animal Experiments, as defined by the Science Council of Japan and the Animal Care and Use Committees of the Tokyo Metropolitan Institute of Medical Science approved all experimental procedures (Permit Number:15035and 15036).

Evaluation of colon thickening and histology
The colon, starting from just below the cecum to above the anal, was excised. Connective tissues were removed, cut longitudinally, washed to remove fecal material, following which length and weight were measured to calculate the length-to-weight ratio. Independent experimental groups were used to obtain transverse sections of mid-to distal-colons. for histology. Specimens were formalin-fixed, embedded in paraffin, sectioned, and stained with hematoxylin and eosin.
In vitro differentiation and adoptive transfer of OVA-specific T cells Antigen-specific effector T cells were prepared as described previously [20]. Approximately 2 × 10 7 viable effector T cells were transferred intravenously with or without 1 × 10 7 viable iTregs.

OVA Treatment
Two hundred microliters of OVA solution (10 mg/mL dissolved in PBS) was injected intra-rectally with animal feeding needles (1.5 mm od × 52-mm long, FUCHIGAMI, Kyoto, Japan), such that the tip was 4 cm proximal to the anus. This treatment was repeated five times daily at approximately 10 min intervals for each. To establish oral tolerance, mice were fed with drinking water supplemented with 1 mg/mL OVA for 7 days before adoptive transfer of cells.

Quantification of cytokines and OVA-specific IgA
Cells were activated using Dynabeads T-Activator CD3/CD28 (Life Technologies) or by culturing with antigen-presenting cells (irradiated splenocytes derived from Rag2-KO) and OVA (50 mg/mL), and cytokines in the supernatants were measured using ELISA kits for IL-17F, IL-22, IL-21 and GM-CSF (eBioscience) or a multiplex bead array for IL-5, IL-13, IL-4, IL-10, IL-17A, and IL-2 (Millipore). Fecal extracts were obtained by adding weighed pellets to PBS (1 mL/100 mg fecal sample) containing protease inhibitors (P8340; Sigma-Aldrich). The samples were mixed and centrifuged, and the supernatants were collected for assay. OVA-specific IgA titers were determined using an ELISA with OVA as the capture antigen, and immune complexes were detected using horseradish peroxidase-conjugated anti-mouse IgA (Southern Biotech).

Preparation of cells from the cLP and flow cytometric analysis
Cells from the cLP were prepared by cutting the large intestine into 1-cm long pieces, and then stirred for 20 min at 37°C in PBS containing 5 mM EDTA and 5 mM EGTA to dissociate epithelial and intraepithelial cells. After washing with PBS three times, the remaining tissue was treated for 50 min at 37°C with RPMI containing 2 mg/mL collagenase D (Roche) and 1 mg/mL DNase I (Roche). Mononuclear cells were isolated using a discontinuous Percoll gradient (40% and 75%) and subjected to flow cytometry (FACScantoII, BD Biosciences). Following stimulation with PMA (20 ng/mL) and ionomycin (1μM) in the presence of monensin for 4h, intracellular staining of the cells was performed using Foxp3 Fixation/Permeabilization Concentrate and Diluent (eBioscience) according to the manufacturer's instructions.

Statistical analysis
Data were analyzed using an unpaired two-tailed Student's t test. A P-values of <0.05 was considered statistically significant. Rag2-KO mice were stimulated under the condition for iTreg lineage, namely medium supplemented with IL-2, TGF-beta1 and retinoic acid (iTreg recipe), or medium containing IL-2 and TGF-beta1 but not retinoic acid (iTreg recipe without retinoic acid) in the absence (control) or presence of CTLA4-Ig (+CTLA4-Ig, 20microg/mL). After 7 days, cells were subjected to the analysis as described in S1 Fig.  (TIF)