Th17 cells and Th1 cells coordinate to play a critical role in the formation of inflammatory bowel diseases. To examine how Th17 and Th1 cells are regulated at inflammatory sites, we used Th1-dominant CD4+CD45RBhigh T cell-transferred RAG-2−/− and Th1/Th17-mixed IL-10−/− mice. Interestingly, not only did colitic RAG-2−/− mice that were parabiosed with WT mice show significant amelioration of colitis, but amelioration of disease was also observed in those parabiosed with colitic IL-10−/− mice. To assess the interference between Th1 and Th17 colitogenic T cells, we co-transferred colitogenic CD4+ T cells from the lamina propria (LP) of CD4+CD45RBhigh T cell-transferred RAG-2−/− mice and IL-10−/− mice into RAG-2−/− mice. Surprisingly, the co-transferred RAG-2−/− mice showed a vast cellular infiltration of LP CD4+ T cells similar to that seen in RAG-2−/− mice re-transferred with the cells from colitic RAG-2−/− mice alone, but the co-transferred RAG-2−/− mice did not have the wasting symptoms, which are also absent in RAG-2−/− mice transferred with cells from colitic IL-10−/− mice alone. Furthermore, the percentages of Th1 and Th17 cells originating from IL-10−/− mice and those of Th1 cells originating from colitic RAG-2−/− mice were all significantly decreased in the co-transferred mice as compared with the singly-transferred paired RAG-2−/− mice, suggesting that Th1 and Th17 cells are in competition, and that their orchestration results in a merged clinical phenotype of the two types of murine colitis.
IL-17 is a key cytokine in animal models of chronic inflammation and several human chronic inflammatory diseases 1–6. CD4+ T cells that produce IL-17 (IL-17A) and express a specific transcription factor, retinoid-related orphan receptor γt, have been distinguished from other Th1 and Th2 cells, and termed Th17 cells 7, 8. IL-17 has been reported to be important in the initiation and/or progression of several inflammatory diseases through the recruitment of neutrophils or other cells in the immune system and amplifies the inflammation 9–13.
IL-17 and IFN-γ are both important in inflammatory bowel diseases (IBD) pathogenesis 5, 14. IFN-γ-producing Th1 and IL-17-producing Th17 cells are orchestrated to initiate and promote colitis both in human IBD and mouse models 2. For instance, IFN-γ plays a critical role in intestinal damage or wasting disease. This is supported by findings that IFN-γ neutralization ameliorates the colitis in SCID mice reconstituted with CD4+CD45RBhigh T cells 15, and that CD4+CD45RBhigh T cells from IFN-γ−/− mice fail to develop wasting disease in SCID mice 16. However, recent studies suggest the involvement of the IL-23/IL-17 axis at any stage of colitogenesis 17. Although it has been recently reported that peripheral plasticity from Th17 to Th1 cells plays a key role in the development of colitis 18, it remains unclear how Th17 and Th1 cells are reciprocally regulated at the inflammatory site in vivo.
Adoptive transfer of CD4+CD45RBhigh T cells in immunodeficient mice (RAG-2−/− or SCID mice) induces aggressive inflammation in the colon and wasting disease 19, 20. We recently noticed that lamina propria (LP) CD4+ T cells of the model mice show a Th1-dominant phenotype in flow cytometric analysis. On the other hand, LP CD4+ T cells obtained from the colitis model of IL-10−/− mice 21, which develop spontaneous colitis, show increased ratios of both Th1 and Th17 cells and less wasting disease compared with CD4+CD45RBhigh T-cell-transferred model. Thus, we hypothesized that the Th1/Th17 balance controls the pathogenesis, and the competition between colitogenic CD4+ Th1 and Th17 cells regulates the wasting disease and tissue damage. Although it has been reported that IFN-γ inhibits the development of Th17 cells in vitro22, it still remains unknown whether the converse occurs. It also remains to be determined how Th1 and Th17 cells that are fully differentiated to effector/memory CD4+ T cells reciprocally interfere in vivo. To solve these issues, we assessed the Th1/Th17 balance by two in vivo systems, namely parabiosis and adoptive retransfer systems using CD4+ CD45RBhigh T-cell-transferred RAG-2−/− mice and IL-10−/− mice.
RAG-2−/− mice transferred with CD4+CD45RBhigh T-cell and IL-10−/− mice develop different types of colitis
In the present study, we used two murine IBD models 23, RAG-2−/− mice adoptively transferred with splenic CD4+CD45RBhigh T cells from WT mice (hereafter called RAG-2−/− RBhigh mice) and IL-10−/− mice, which spontaneously develop chronic colitis, to investigate reciprocal interaction between different types of colitogenic CD4+ T cells in vivo. In our animal facility in specific pathogen-free (SPF) condition, IL-10−/− mice clinically developed colitis from approximately 18 wk of age, and their clinical and histological scores reached a plateau at approximately 24 wk of age. Thus, we used 12-wk- and 20-wk-old IL-10−/− mice as non-colitic and colitic IL-10−/− mice, respectively. To match these ages, we used approximately 12- and 20-wk-old RAG-2−/− RBhigh mice at 6 wk after transfer as colitic RAG-2−/− RBhigh mice. We also used 12- and 20-wk-old normal WT C57BL/6J mice as a negative control (See experimental designs in Figs. 2A, 4A, and 6A). We confirmed that WT mice (20-wk-old) appeared healthy with no signs of colitis (data not shown) and no apparent thickening of the colonic wall with normal goblet cell formation (Fig. 1A; upper left), whereas RAG-2−/− RBhigh mice (20-wk-old) developed severe colitis 6 wk after transfer, characterized by significant weight loss, diarrhea (data not shown), and thickening of the colonic wall with transmural inflammation with high numbers of mononuclear cells in LP and submucosa, and prominent epithelial hyperplasia with loss of goblet cells (Fig. 1A; upper right). Old IL-10−/− mice (20-wk-old) developed colitis characterized by prominent epithelial hyperplasia with loss of goblet cells, but showed milder infiltration of lymphocytes than colitic RAG-2−/− RBhigh mice (Fig. 1A; lower right). In contrast, young IL-10−/− mice (12-wk-old) histologically showed very mild colitis with slight infiltration of lymphocytes, but no clinical signs (Fig. 1A; lower left).
Intracellular cytokine staining analysis to assess the Th1/Th17 balance in these two models revealed that (i) LP CD4+ T cells in WT mice expressed little IFN-γ and IL-17, (ii) LP CD4+ T cells in colitic RAG-2−/− RBhigh mice preferentially expressed IFN-γ rather than IL-17, and (iii) LP CD4+ T cells in old IL-10−/− mice with colitis were a mixture of IFN-γ- and IL-17-expressing cells, whereas LP CD4+ T cells in young IL-10−/− mice without colitis showed lower percentages of IFN-γ- and IL-17-expressing cells than old IL-10−/− mice with colitis (Fig. 1B). The ratios of IFN-γ+IL-17+ double-positive cells in LP of colitic mice irrespective of RAG-2−/− RBhigh mice and old IL-10−/− mice appeared to be increased as compared with those in non-colitic mice (WT and young IL-10−/− mice) (Fig. 1B). Although small but substantial numbers of IL-22-producing CD4+ T cells could be detected in LP of WT, RAG-2−/− RBhigh mice and old IL-10−/− mice, few IL-22-producing cells co-expressed IL-17A (Fig. 1C), indicating that IL-22-producing cells may be categorized as Th22 cells 24 in our SPF condition. In addition, small numbers of IL-17A+IL-17F+ double-positive CD4+ T cells could be detected in old IL-10−/− mice but not in WT and colitic RAG-2−/− RBhigh mice (Fig. 1C). Therefore, we defined IL-17A as a representative cytokine of Th17 cells for further studies. Collectively, these data indicated that the colitis models of RAG-2−/− RBhigh mice and IL-10−/− mice can be categorized as Th1-dominant and Th1/Th17-mixed models, respectively.
Parabiosis with young non-colitic IL-10−/− mice ameliorates colitis in RAG-2−/− RBhigh mice
To investigate the in vivo control of colitogenic CD4+ T cells in colitic mice, we first used a parabiosis system between colitic RAG-2−/− RBhigh mice and young IL-10−/− mice without colitis (Fig. 2A). To discriminate the origin of colitogenic CD4+ T cells, we used Ly5.1+CD4+CD45RBhigh T cells as donor cells for the RAG-2−/− transfer colitis model. At 6 wk after transfer, but before parabiosis surgery, we confirmed that RAG-2−/− RBhigh mice showed severe wasting disease and colitis (data not shown). In this experiment, mice were divided into four groups as depicted in Fig. 2A. As expected, RAG-2−/− RBhigh mice in control parabionts with WT mice (Group 1) showed a decrease in clinical symptoms over time. At 4 wk after surgery, the clinical score was significantly lower than that of the positive control, Group 4 RAG-2−/− RBhigh mice (Fig. 2B), suggesting that suppressive T cells such as CD4+CD25+Foxp3+ regulatory T (TR) cells 25 and IL-10-producing Tr1 cells 26 in WT mice migrate to the RAG-2−/− RBhigh mice. Also, WT mice in Group 1 parabionts were consistently healthy during the observed period as assessed by the clinical score (Fig. 2B). Since IL-10 is a key molecule for the functions of both CD4+CD25+Foxp3+ TR and Tr1 cells 27, we next questioned whether RAG-2−/− RBhigh mice parabiosed with non-colitic young IL-10−/− mice (Group 2) would remain colitic due to the dysfunctional regulation by these suppressive T cells. Surprisingly, however, RAG-2−/− RBhigh mice in Group 2 also became healthier, and their final clinical score was significantly lower than that of control Group 4 RAG-2−/− RBhigh mice with colitis, but still significantly higher than that of RAG-2−/− RBhigh mice in Group 1 (Fig. 2B). In contrast, IL-10−/− mice in Group 2 parabionts showed mild clinical manifestations especially extraintestinal hunching and wasting disease without colitis, even though their age was still only 16 wk, and thus their clinical score was significantly higher than that of control Group 3 IL-10−/− mice. This suggests that a cytokine storm derived from the colitic RAG-2−/− RBhigh mice at an earlier time point accelerated the extraintestinal manifestation of parabiosed IL-10−/− mice in Group 2. Group 3 IL-10−/− mice at 16 wk were all healthy during the observed period (Fig. 2B).
At 4 wk after parabiosis surgery, the colon in mice of Groups 1–3 was macroscopically almost normal with solid stool in the thin intestine, in sharp contrast to the colon of Group 4 RAG-2−/− RBhigh mice, which was enlarged and had a greatly thickened wall, with a small, but markedly thickened appendix (Fig. 2C). Consistent with this macroscopic observation, histological examination showed prominent epithelial hyperplasia with glandular elongation and massive infiltration of mononuclear cells in LP of the colon of control Group 4 RAG-2−/− RBhigh mice (Fig. 2D). In Groups 1–3, these inflammatory changes were mostly absent and only few mononuclear cells were observed in LP of the colon (Fig. 2D). This difference was also confirmed by histological scoring of multiple colon sections, which revealed a higher histological score in Group 4 mice than the scores in the other groups (Fig. 2E). In addition, the histological score of Group 2 RAG-2−/− RBhigh mice with mild colitis was significantly higher than that of Group 1 RAG-2−/− RBhigh mice (Fig. 2E).
To further assess the expansive activity of LP CD4+ T cells in this setting, we evaluated the numbers and percentages of Ly5.1+ and Ly5.2+ cells by FACS analysis (Fig. 3A). Consistent with the histological assessment (Fig. 2E), the recovered cell number from each mouse of Groups 1–3 mice was significantly lower than that from colitic Group 4 RAG-2−/− RBhigh mice (Fig. 3B), indicating that colitis of Group 2 RAG-2−/− RBhigh mice is suppressed at least in part by an IL-10-independent mechanism, such as clonal competition 28 between CD4+ T cells derived from RAG-2−/− RBhigh cells and WT or IL-10−/− mice. Consistent with this hypothesis, the majority of LP CD4+ T cells both in Group 1 and 2 parabionts were Ly5.2+ cells derived from WT or IL-10−/− mice, which should retain a higher percentage of CD4+ T-cell clones compared with Group 4 RAG-2−/− RBhigh mice (Fig. 3B). In particular, the percentage of Ly5.1+ LP CD4+ T cells was markedly reduced on the donor side of Group 1 and Group 2 parabionts (Fig. 3B and C). These results suggested that clonal competition between LP CD4+ T cells derived from the two types of mice has an important role in this healing mechanism of RAG-2−/− RBhigh mice in parabionts and regulates the expansion of colitogenic LP CD4+ T cells.
Parabiosis with colitic IL-10−/− mice reduces colitis in RAG-2−/− RBhigh mice
On the basis of our findings that LP CD4+ T cells from not only WT mice but also from non-colitic IL-10−/− mice migrated to the LP in colitic RAG-2−/− RBhigh mice, leading to amelioration of the colitis in RAG-2−/− RBhigh mice (Figs. 2 and 3), we next hypothesized that colitogenic LP CD4+ T cells from IL-10−/− mice with established colitis may also interfere with LP CD4+ T cells from colitic RAG-2−/− RBhigh mice, leading to attenuated severity of colitis. Again, we employed a parabiosis system with colitic IL-10−/− mice as illustrated in Fig. 4A. As expected, clinical symptoms in RAG-2−/− RBhigh mice of Group 1 gradually decreased following parabiosis surgery, whereas the control Group 2 RAG-2−/− RBhigh mice showed severe diarrhea, weight loss, and hunched posture (data not shown). At 4 wk after surgery, the clinical score of the Group 1 RAG-2−/− RBhigh mice parabiosed with colitic IL-10−/− mice was significantly lower than that of Group 2 RAG-2−/− RBhigh mice, but parabiosed Group 1 IL-10−/− mice had similar clinical scores to Group 3 IL-10−/− mice (Fig. 4B). Histological examination revealed severe mucosal erosion and massive infiltration of mononuclear cells in colons from control Group 2 RAG-2−/− RBhigh mice (Fig. 4C), whereas Group 1 parabiosed RAG-2−/− RBhigh mice showed milder colitis with prominent epithelial hyperplasia and moderate infiltration of mononuclear cells without erosion (Fig. 4C). Group 1 and Group 3 IL-10−/− mice showed mild colitis (Fig. 4C). This difference was also confirmed by histological scoring of colon sections: Group 1 parabiosed RAG-2−/− RBhigh mice had significantly lower scores than the paired Group 2, RAG-2−/− RBhigh mice, while Group 1 parabiosed IL-10−/− mice tended to have lower scores than Group 2 IL-10−/− mice, although the difference was not significant (Fig. 4D).
We next assessed the numbers and percentages of Ly5.1+ and Ly5.2+ cells by FACS analysis (Fig. 5A). First, the number of infiltrating LP CD4+ T cells recovered from Group 1 parabiosed RAG-2−/− RBhigh mice was significantly decreased as compared with the control Group 2 RAG-2−/− RBhigh mice (Fig. 5B), although a substantial number of donor Ly5.2+ T cells derived from IL-10−/− mice migrated to the parabiosed RAG-2−/− RBhigh mice. In contrast, the absolute number of infiltrating LP CD4+ T cells recovered from Group 1 IL-10−/− mice was comparable to that from Group 2 IL-10−/− mice. In each of the Group 1 parabionts, the ratio of Ly5.1+ to Ly5.2+ cells was approximately 1:1 at 4 wk after parabiosis surgery (Fig. 5C). These data indicate that, in spite of the absence of IL-10, the donor Ly5.2+ T-cell clones from IL-10−/− mice competed with the Ly5.1+ T cells in the recipient RAG-2−/− RBhigh mice, resulting in the amelioration of Group 1 RAG-2−/− RBhigh parabionts. These data are consistent with the results of the previous experiment (Figs. 2 and 3); that is, colitis of RAG-2−/− RBhigh mice in Group 1 is regulated in an IL-10-independent manner. Nevertheless, moderate colitis persists in RAG-2−/− RBhigh mice in Group 1, as evidenced by histological appearance and score in comparison with RAG-2−/− RBhigh mice parabiosed with WT mice, suggesting that colitis of RAG-2−/− RBhigh mice is regulated at least in part in an IL-10-dependent manner (Fig. 4C and D).
Flow cytometric analysis showed that a substantial amount of Ly5.1+ LP CD4+ T cells from RAG-2−/− RBhigh mice remains in LP in Group 1 RAG-2−/− RBhigh mice, and a minor but appreciable amount of Ly5.1+ LP CD4+ T cells is present in LP in colitic IL-10−/− mice (Fig. 5A and B). The proportion of Ly5.1+ LP CD4+ T cells is 66.8±8.88% in colitic RAG-2−/− RBhigh mice in Group 1 and 30.1±4.56% in colitic IL-10−/− mice (Fig. 5C), but the total number of LP CD4+ T cells is lower in RAG-2−/− RBhigh mice in Group 1 than that in Group 2 RAG-2−/− RBhigh mice, supporting the conclusion that colitis is ameliorated in Group 1 RAG-2−/− RBhigh mice after parabiosis surgery (Fig. 5B). Interestingly, Ly5.1+ LP CD4+ T cells in colitic IL-10−/− mice tended to have a lower proportion of IFN-γ-producing cells than those in Group 2 RAG-2−/− RBhigh mice (P=0.08) (Fig. 5D), but no significant difference was found in the ratio of IL-17-producing T cells (Fig. 5E). These results regarding the proportions of cytokine-producing cells indicated that the remaining LP CD4+ T cells in the Group 1 ameliorating parabionts have the character of colitogenic memory T cells, which are able upon in vitro re-stimulation to produce comparable amounts of cytokines to those in colitic control mice. Collectively, all these data show that Ly5.1+ and Ly5.2+ LP CD4+ T cells are mixed in the parabionts, and that an IL-10-independent mechanism underlies this colitis-ameliorating process.
Colitogenic Th1 and Th17 CD4+ T cells mutually compete in the lymphopenic condition
On the basis of our findings that colitic RAG-2−/− mice gradually became healthier after parabiosis surgery with IL-10−/− mice in either a non-colitic or colitic condition (Figs. 2–5), we posited that donor Ly5.2+ T-cell clones from IL-10−/− mice competed with the recipient Ly5.1+ T cells from RAG-2−/− RBhigh mice, resulting in the amelioration of colitis in Group 1 RAG-2−/− RBhigh parabionts, possibly due to a mechanism of clonal competition. However, it is also possible that non-T cells, such as macrophages and DC, could have migrated to the partner mice in parabionts and affected the results. Thus, to specifically assess the competition of different types of colitogenic CD4+ T cells without such possible impact of other cells, we performed an in vivo competition experiment using an adoptive re-transfer system. This lymphopenic adoptive transfer system also offers a useful approach to compare with the lymphosufficient parabiosis system (Figs. 2–5). To this end, the same numbers of Ly5.1+ LP CD4+ T cells from colitic RAG-2−/− RBhigh mice and Ly5.2+ LP CD4+ T cells from colitic IL-10−/− mice were co-injected intraperitoneally into identical RAG-2−/− mice (Group 1) (Fig. 6A). As controls, single injections of LP CD4+ T cells from colitic IL-10−/− mice or LP CD4+ T cells from colitic RAG-2−/− RBhigh mice were given intraperitoneally into RAG-2−/− mice (Groups 2 and 3, respectively) (Fig. 6A). Interestingly, the percentage loss of body weight of Group 2 RAG-2−/− mice transferred with LP CD4+ T cells from colitic RAG-2−/− RBhigh mice was significantly higher than that of Groups 1 and 3 (Fig. 6B). Eight wk after transfer, all mice were sacrificed, and clinical scores indices of weight loss and colitis were assessed.
The macroscopic findings revealed that the colons from Groups 1 and 2, which were transferred with LP CD4+ T cells from colitic RAG-2−/− RBhigh mice, either singly or with cells from IL-10−/− mice, were markedly thickened and shortened as compared with Group 3 mice, which were transferred singly with LP CD4+ T cells from colitic IL-10−/− mice (Fig. 6C). In particular, the wall of the appendix in Groups 1 and 2 was small but severely thickened (Fig. 6C), as is generally seen in colitic RAG-2−/− mice transferred with CD4+CD45RBhigh T cells (data not shown). The total clinical scores revealed that all three groups of mice developed colitis, which tended to be more severe in Groups 1 and 2 (Fig. 6D). Histological examinations revealed that Group 1 and Group 2 mice developed similar, but more severe, colitis accompanied with a marked infiltration of mononuclear cells than Group 3 mice (Fig. 6E). Consistent with this observation, the histological scores of Groups 1 and 2 mice were significantly higher than that of Group 3 mice (Fig. 6F). All data suggest that the re-transfer procedure traces the same type of colitis as the original: Group 2 mice showed severe colitis with body weight loss and thickened the colon wall, which is the character of original RAG-2−/− RBhigh mice injected with CD4+ CD45RBhigh T cells with colitis, whereas Group 3 mice developed mild colitis without weight loss. Thus, co-injected Group 1 mice showed a mixture of two different types of colitis, i.e. appreciable infiltration of mononuclear cells in LP like Group 2 mice, and non-wasting disease like Group 3 mice.
To clarify the interesting mixed characteristics of colitis in the co-injected Group 1 mice, we investigated the character of the LP CD4+ T cells infiltrating the colon by flow cytometric analysis to determine the ratio of T cells originating from Ly5.1+ LP CD4+ T cells from colitic RAG-2−/− RBhigh mice and Ly5.2+ LP CD4+ T cells from colitic IL-10−/− mice (Fig. 7A). The absolute number of LP CD4+ T cells was significantly larger in Group 1 than Group 3, and a similar tendency was found between Group 2 and Group 3 (Fig. 7B). These findings are consistent with the histological data and clinical findings (Fig. 6C and F). As shown in Fig. 7B and C, Ly5.1+ and Ly5.2+ CD4+ T cells in the co-injected Group 1 mice were found in a ratio of approximately 1:1 in accordance with the initially transferred cell numbers (each 3.0×105 cells per mouse), suggesting competition between Ly5.2+ CD4+ T cells from colitic IL-10−/− mice and Ly5.1+ CD4+ T cells from colitic RAG-2−/− RBhigh mice in both spleen and LP of Group 1 mice. It should be noted that the absolute cell numbers of Groups 1 and 2 were comparable, indicating that the expansion of Ly5.1+ T cells in Group1 originated from LP CD4+ T cells of RAG-2−/− RBhigh mice was markedly suppressed (Fig. 7B, white bar). We further assessed the percentages of IFN-γ- and IL-17-expressing cells in each group by intracellular staining of the cytokines using flow cytometry. Interestingly, the percentage of IFN-γ-producing Ly5.1+ LP CD4+ T cells was significantly lower in Group 1 mice than Group 2 mice (Fig. 7D), which are characterized as colitogenic Th1 cells with a high ratio of IFN-γ-producing LP CD4+ T cells (Fig. 1B). We also found a similar characteristic change in Ly5.2+ LP CD4+ T cells from colitic IL-10−/− mice, which are characterized as colitogenic Th1 and Th17 cells (Fig. 1B), since the percentage of IFN-γ-producing Ly5.2+ LP CD4+ T cells and IL-17-producing Ly5.2+ LP CD4+ T cells were significantly lower in Group 1 than Group 3 (Fig. 7D and E). These data collectively demonstrate that in this adoptive re-transfer system, colitogenic Th1 and Th17 CD4+ T cells mutually interfere in lymphopenic condition at the beginning of transfer.
We have demonstrated here that colitogenic CD4+ T cells in inflamed mucosa are controlled by in vivo competition between coliotogenic Th1 and Th17 cells (Fig. 8). Such a competitive mechanism in colitic conditions may affect the form of IBD, in term of the development of wasting disease and local infiltration of colitogenic CD4+ T cells.
Complete and prompt healing of colitis in RAG-2−/− RBhigh mice by parabiosis surgery with WT mice (Figs. 2 and 3) prompted us to investigate the underlying mechanism of healing IBD. Surprisingly, our present results indicated that colitic RAG-2−/− RBhigh mice parabiosed with either non-colitic or colitic IL-10−/− mice, both of which lack IL-10 as a key molecule for the functions of CD4+CD25+Foxp3+ TR and IL-10-producing Tr1 cells 26, also present significant amelioration. This finding indicated that the colitogenesis of CD4+ T cells might be regulated in an IL-10-independent manner. Reports that the number of CD4+ T cells is controlled by the clonal abundance 17 or expansion speed 18 imply that interactions occur between CD4+ T cells that lead to cell depletion or reduction and stabilize the number of one clone of CD4+ T cells. The reduction of colitogenic LP CD4+ T cells means healing of colitis, and reduction must occur after some competition between CD4+ T cells. Our finding that percentages of cytokine-producing T cells decrease significantly when competition exists between two different types of colitogenic LP CD4+ T cells may represent one such case.
Consistent with many previous reports, our re-transfer experiment (Figs. 6 and 7) shows that re-transferred colitogenic LP CD4+ T cells induce the same type of colitis as the original: that is, re-transferred LP CD4+ T cells from colitic RAG-2−/− RBhigh mice develop colitis with wasting and vast infiltration of mononuclear cells into mucosa, while those from IL-10−/− mice develop colitis without wasting and with a higher number of LP CD4+ T cells producing IL-17 than those from colitic RAG-2−/− RBhigh mice. In line with these findings, it has been reported that IFN-γ and IL-17A or IL-17F have a proinflammatory role in colitogenesis to a varying degree and naïve CD4+ T cells derived from knockout mice of each cytokine show a varying feature of colitis after being transferred into lymphopenic mice 29. The co-transfer experiments showed that LP CD4+ T cells from colitic RAG-2−/− RBhigh mice compete equally with those from IL-10−/− mice in the lymphopenic condition like RAG-2−/− mice. Interestingly, these two different types of cells orchestrated and produced another type of colitis in which the colonic wall was greatly infiltrated by mononuclear cells and significant wasting was absent. These characters derive partly from colitic RAG-2−/− RBhigh mice and partly from IL-10−/− mice. In addition to the clinical symptoms and histological findings of colitis, this competition caused significant changes in cytokine production of colitogenic LP CD4+ T cells, especially in the proinflammatory cytokines, IFN-γ and IL-17. LP CD4+ T cells from IL-10−/− mice contained almost the same percentage of CD4+ Foxp3+ T cells (data not shown), and these TR cells may have contributed to the curing colitis in parabiosed colitic RAG-2−/− RBhigh mice. However, these cells lack a key anti-inflammatory molecule, IL-10, and recent reports show that the TR cells in IL-10−/− mice lack the ability to regulate inflammation 19, 20. Thus, our findings about the curing of colitis in the parabiosis models owe less to TR function and more to other mechanisms like clonal competition 28 between the colitogenic CD4+ T-cell clones.
Th1 and Th17 balance may be regulated by various cytokines. IL-6, TGF-β, and IL-23 are reported to induce IL-17 production from murine naïve CD4+ T cells 9, 13, 17. Some recent studies revealed the upregulation of Th17 cells and Th1/Th17 (IFN-γ and IL-17 co-producing cells) in LP CD4+ T cells of human IBD patients 5, 30. Downregulation of anti-inflammatory cytokine, TGF-β, in IBD may be involved in the conversion of Foxp3+ T cells into Th17 cells 30. Th1/Th17 cells are also detected in colitic IL-10−/− mice (Fig. 1B). These specific cells have a distinct mechanism in multiple sclerosis patients to pass through the blood–brain barrier 12, but the function of Th1/Th17 cells is still unclear in IBD. Further study will be needed to address this point.
In the parabiosis between colitic RAG-2−/− RBhigh mice and colitic IL-10−/− mice, we showed that IFN-γ-producing LP CD4+ T cells from colitic RAG-2−/− RBhigh mice decreased after surgery (Figs. 4 and 5). No such decrease was found in the parabiosis between colitic RAG-2−/− RBhigh mice and non-colitic IL-10−/− mice (Figs. 2 and 3). Almost all the LP CD4+ T cells from colitic RAG-2−/− RBhigh mice were eliminated after surgery, and the percentage of IFN-γ-producing LP CD4+ T cells did not decrease. This may be because IFN-γ-producing LP CD4+ T cells are memory T cells that survive long after the mixture and interference between two different types of colitogenic CD4+ T cells. As we previously reported, the longevity of colitogenic memory CD4+ T cells plays a key role in the perpetuity of colitis 21, 22. During the competition between the two types of cells in colitic RAG-2−/− RBhigh mice, effector T cells producing IFN-γ or IL-17 were diminished except for the cells that turned into memory T cells. The parabiosis experiment using IL-10−/− mice with established colitis (Figs. 4 and 5) showed that a certain amount of T cells remain in colitic RAG-2−/− RBhigh mice. These mice show a reduced percentage of Ly5.1+ IFN-γ-producing T cells compared with the non-parabiosised colitic RAG-2−/− RBhigh mice, which suggests that a balance is established between Th1 and Th17 through the competition of the two different types of Th cells. However, some specific innate immune cells have been revealed as key players engaged in the paradigm changes in Th cells, such as CX3CR1+ DC 31, CD103+ DC 32, and regulatory macrophages 33. Co-transfer experiments showed that LP CD4+ T cells interfere with one another in the same platform, RAG-2−/− mice, but the recruitment of these innate immune cells at the inflammatory site may play an independent role in the interference among Th cells. Further studies will be warranted to address these issues in the competition of colitogenic Th1 and Th17 cells.
Colitogenic CD4+ T cells in the current two colitis models, colitis induced by adoptive transfer into lymphopenic mice versus the spontaneous colitis model, IL-10−/− mice, differ significantly from the standpoint of homeostatic regulation of colitogenic memory CD4+ T lymphocytes. The mechanism of the RAG-2−/− colitis model induced by adoptive transfer of CD4+CD45RBhigh T cells seems to fit the enteric bacteria-induced rapid proliferation model in lymphopenic condition 34, whereas IL-10−/− colitis model seems to fit the slow proliferation model in lymphosufficient condition. The adoptive transfer colitis model has been criticized as too artificial to understand human IBD pathophysiology, since the expansive phenomenon may be just a physiological response to the lymphopenic condition, which affords sufficient space and amounts of cytokines for transferred lymphocytes to expand in vivo. However, here we demonstrated that colitogenic CD4+ T cells from colitic RAG-2−/− RBhigh mice competed equally with colitogenic CD4+ T cells from colitic IL-10−/− mice in the parabiosis system using both the colitic mice (Figs. 4 and 5) as well as in the lymphopenic re-transfer system (Figs. 6 and 7). Consistent with our results, King et al. recently proposed that organ-specific autoimmunity is initiated by homeostatic proliferation of CD4+ T cells driven by lymphoid space 35. Moreover, it has been reported that human IBD are often developed after viral infections that induce lymphopenic condition 36. Therefore, the lymphopenic condition itself appears to be very important trigger for the development of IBD colitis, and this adoptive transfer model of colitis is a sufficiently useful IBD model that mimics the developmental process of human IBD.
In conclusion, we have demonstrated the following i.e. that colitogenic Th1 and Th17 CD4+ T cells mutually compete in vivo in mediating the Th1/Th17 balance; two types of colitogenic CD4+ T cells, from a spontaneous colitis model and an adoptive transfer colitis model, are equally competitive in RAG-2−/− mice in a lymphopenic condition; and the competition among the different clones of T cells is an important mechanism to regulate cell expansion to repress colitis, as shown by healing of colitis achieved by parabiosis with IL-10−/− mice despite the dysfunction of TR cells.
Materials and methods
C57BL/6-Ly5.2 mice were purchased from Japan Clea (Tokyo, Japan). C57BL/6-Ly5.1 mice and C57BL/6-Ly5.2-RAG-2 deficient (RAG-2−/−) were obtained from Taconic Laboratory (Hudson, NY) and Central Laboratories for Experimental Animals (Kawasaki, Japan). IL-10−/− mice were purchased from Jackson Laboratories (Maine, USA). Mice were maintained under SPF conditions in the Animal Care Facility of Keio University. Recipient RAG-2−/− mice were used at 6 or 14 wk of age. Non-colitic and colitic IL-10−/− donors were used at 12 and 20 wk of age, respectively. All experiments were approved by the regional animal study committees and were done according to institutional guidelines and Home Office regulations.
In vivo experimental design
Experiments were performed to investigate the in vivo reciprocal interference between different types of colitogenic CD4+ T cells obtained from two animal IBD models, IL-10−/− mice and RAG-2−/− mice transferred with CD4+CD45RBhigh T cells.
Adoptive transfer in combination with parabiosis (Fig. 2A). For adoptive transfer, CD4+ T cells were first isolated from spleen cells of C57BL/6-Ly5.1 mice using the anti-CD4 (L3T4)-MACS system (Miltenyi Biotec, Auburn, CA) according to the manufacturer's instruction. Enriched CD4+ T cells (96–97% pure, as estimated by FACS Calibur (Becton Dickinson, Sunnyvale, CA)) were then labeled with PE-conjugated anti-mouse CD4 (RM4-5; BD PharMingen, San Diego, CA) and FITC-conjugated anti-CD45RB (16A; BD PharMingen). CD4+CD45RBhigh cells were purified using a FACS Aria (Becton Dickinson). This population was >98.0% pure on reanalysis. RAG-2−/− mice (6-wk-old; n=18) were then injected intraperitoneally with 3×105 CD4+CD45RBhigh T cells. At 6 wk after transfer, RAG-2−/− mice transferred with CD4+CD45RBhigh T cells had developed a wasting disease and colitis as previously reported 37. We then carried out parabiosis surgery according to institutional guidelines and Home Office regulations. Briefly, sex-matched mice were anesthetized prior to surgery, and incisions were made in the skin on the opposing flanks of the donor and recipient animals. Surgical sutures were used to bring the body walls of the two mice into direct physical contact. The outer skin was then attached with surgical staples. Four groups were established: Group 1, colitic RAG-2−/− mice (12-wk-old) joined with normal C57BL/6-Ly5.2 mice (12-wk-old) (n=7 pairs); Group 2, colitic RAG-2−/−mice (12-wk-old) joined with non-diseased IL-10−/− mice (12 wk-old) (n=7); Group 3, age-matched IL-10−/− mice (12-wk-old) without parabiosis surgery (n=5); Group 4; age-matched colitic RAG-2−/− mice (12-wk-old) without surgery (n=5).
As Exp. 1, but using colitic IL-10−/− mice (20 wk of age) (Fig. 4A). To match the age of mice, we used 20-wk-old mice at surgery (n=7 per group).
To assess in vivo interference between colitogenic Th1 and Th-17 CD4+ T cells in colitic mice, we performed in vivo competition experiments by adoptive re-transfer using colitic IL-10−/− mice and RAG-2−/− mice previously transferred with CD4+CD45RBhigh T cells (Fig. 6A). The same number (3×105 cells) of CD4+ T cells from colitic IL-10−/− mice (Ly5.2+) and colitic RAG-2−/− (Ly5.1+) mice were co-injected intraperitoneally into new RAG-2−/− mice (n=6), and the recipients were monitored for 6 wk after transfer. As controls, RAG-2−/− mice were transferred with CD4+ T cells from colitic IL-10−/− mice alone or colitic RAG-2−/− (Ly5.1+) mice alone (each, n=6).
All mice were observed for clinical signs such as hunched posture, piloerection, diarrhea, and blood in the stool. At autopsy, mice were assessed for a clinical score 38 that is the sum of three parameters as follows: hunching and wasting, 0 or 1; colon thickening, 0–3 (0, no colon thickening; 1, mild thickening; 2, moderate thickening; 3, extensive thickening); and stool consistency, 0–3 (0, normal beaded stool; 1, soft stool; 2, diarrhea; 3, bloody stool).
Histological examination was performed as described previously 39 and the mean degree of inflammation was calculated using a modification of a previously described scoring system 38.
Cell isolation was performed as described previously 20. For in vitro assay, live cells were counted by trypan-blue staining, and the viability of cells was confirmed to be almost the same (>96% live) among the sample groups.
Flow cytometry was performed as described previously 40. The following mAb were used: anti- CD3 mAb (145-2C11), anti- CD4 mAb (RM4-5), anti-CD45RB mAb (16A), anti-CD45.1 (Ly5.1; A20), anti-CD45.2 (Ly5.2; 104), anti-IFN-γ(XMG1.2), IL-17A (TC11-18H10), IL-17F (eBio18F10), and IL-22 (140301).
The results are expressed as mean±SEM. Groups of data were compared by Mann–Whitney U test. Differences were considered to be statistically significant when p<0.05.
We are grateful to Mina Kitazume for preparing mice, Takayuki Tomita for technical assistance, and Tetsuro Takayama and Tomoharu Yajima for their valuable discussion. This study was supported in part by grants-in-aid for Scientific Research, Scientific Research on Priority Areas, Exploratory Research and Creative Scientific Research from the Japanese Ministry of Education, Culture, Sports, Science and Technology; the Japanese Ministry of Health, Labor and Welfare; the Japan Medical Association; Foundation for Advancement of International Science; Ohyama Health Foundation; Yakult Bio-Science Foundation; Research Fund of Mitsukoshi Health and Welfare Foundation; and Research Fund of Yakult Medical Foundation.
Conflict of interest: The authors declare no financial or commercial conflict of interest.