An Essential Role for Interleukin 10 in the Function of Regulatory T Cells That Inhibit Intestinal Inflammation

Specific pathogen-free BALB/c and C.B-17 SCID mice were obtained from Simonsen Laboratories or Harlan Olac and maintained in the Animal Care Facility of the DNAX Research Institute or in the Biomedical Services Unit (BMSU) at the John Radcliffe Hospital. 129/SvEv WT, 129/SvEv recombination-activating gene (RAG)-2–deficient (RAG-2^−/−) and 129/SvEv IL-10–deficient (IL-10^−/−) mice were bred in isolators and maintained in microisolator cages with filtered air. Mice were used at 8–12 wk of age.

The following mAbs were used for cell purification: 2-43, anti–mouse-CD8 (TIB210; American Type Culture Collection); M1/70, anti–mouse Mac-1 (TIB128; American Type Culture Collection); RA3-6B2, anti–mouse B220 12; FITC-conjugated anti–mouse CD45RB (clone 16A; PharMingen); and Cy-Chrome–conjugated anti–mouse CD4 (clone RM4-5; PharMingen). FITC-conjugated anti–mouse IFN-γ (clone AN-16; gift of Dr. A. O'Garra, DNAX Research Institute) and PE-conjugated anti–mouse TNF-α (clone XT-22; PharMingen) were used for intracellular staining. The following mAbs were used in vivo: 1B1.2 (rat IgG1), a blocking mAb reactive with mouse IL-10 receptor (IL-10R) 13, JES5-2A5 (rat IgG1), an anti–mouse IL-10 mAb 14, and GL113 (rat IgG1), an isotype control mAb reactive with β-galactosidase. Unless otherwise indicated, antibodies were purified from supernatants of hybridomas by ion exchange chromatography and shown to contain <3 EU endotoxin/mg of protein.

CD4⁺ T cell subsets were isolated from spleens as described previously 4. In brief, single cell suspensions were depleted of B220⁺, MAC-1⁺, and CD8⁺ cells by negative selection using sheep anti–rat coated Dynabeads (Dynal Ltd.). The resulting CD4⁺-enriched cells were stained with Cy-Chrome–conjugated anti-CD4 and FITC-conjugated anti-CD45RB mAb and sorted into the CD45RB^high CD4⁺ and CD45RB^low CD4⁺ fractions by two-color sorting on a FACS Vantage™ (Becton Dickinson). All populations were >98% pure on reanalysis.

RAG-2^−/− mice or C.B-17 SCID mice were injected intraperitoneally with sorted CD4⁺ T cell subpopulations in PBS. mAbs were injected intraperitoneally in PBS the day after T cell reconstitution and weekly for the duration of the experiments.

Colons were removed from mice 8–12 wk after T cell reconstitution and fixed in buffered 10% formalin. 6 μm paraffin-embedded sections were cut and stained with hematoxylin and eosin. Tissues were graded semiquantitatively from 0 to 5 in a blinded fashion. A grade of 0 was given when there were no changes observed. Changes typically associated with other grades are as follows: grade 1, minimal scattered mucosal inflammatory cell infiltrates, with or without minimal epithelial hyperplasia; grade 2, mild scattered to diffuse inflammatory cell infiltrates, sometimes extending into the submucosa and associated with erosions, with minimal to mild epithelial hyperplasia and minimal to mild mucin depletion from goblet cells; grade 3, mild to moderate inflammatory cell infiltrates that were sometimes transmural, often associated with ulceration, with moderate epithelial hyperplasia and mucin depletion; grade 4, marked inflammatory cell infiltrates that were often transmural and associated with ulceration, with marked epithelial hyperplasia and mucin depletion; and grade 5, marked transmural inflammation with severe ulceration and loss of intestinal glands.

Lamina propria (LP) lymphocytes were purified as described 5. Colons were cut into 0.5–1-cm pieces and incubated in Ca- and Mg-free PBS containing 10% heat-inactivated FCS (GIBCO BRL) and 5 mM EDTA to release intraepithelial lymphocytes. The remaining tissue was further digested with collagenase/dispase (100 U/ml; Sigma Chemical Co.), and the LP cells were then layered on a Percoll gradient (Amersham Pharmacia Biotech). The lymphocyte-enriched population was recovered after centrifugation (600 g, 20 min) at the 40/75% interface. For cytokine detection, freshly isolated LP lymphocytes were cultured for 12 h in RPMI 1640 (PAA Laboratories GmbH) containing 10% FCS, 2 mM l-glutamine, 0.05 mM 2-ME, and 100 U/ml each of penicillin and streptomycin in 24-well flat-bottomed plates coated with anti-CD3∈ 145-2C11 (10 μg/ml, CRL1975; American Type Culture Collection). Brefeldin A (10 μg/ml; Sigma Chemical Co.) was added for the final 2 h of incubation, and surface and cytoplasmic stainings were performed as described previously 15. Labeled cells were analyzed on a FACSort™ using CELLQuest™ software (Becton Dickinson).

Colitis scores were compared using the Mann-Whitney test, and differences were considered statistically significant with P < 0.05.

To determine whether the CD45RB^low CD4⁺ population from IL-10^−/− mice contained regulatory T cells, this population was compared with CD45RB^low CD4⁺ cells from WT mice for the ability to inhibit colitis induced in RAG-2^−/− mice by transfer of WT CD45RB^high CD4⁺ cells. Expression of CD45RB was similar on CD4⁺ T cells from 10–12-wk-old IL-10^−/− and WT mice (data not shown). WT CD45RB^high CD4⁺ cells (4 × 10⁵) were transferred to syngeneic RAG-2^−/− mice either alone or in combination with 2 × 10⁵ CD45RB^low CD4⁺ cells from WT or IL-10^−/− mice. As controls, the CD45RB^low CD4⁺ populations from IL-10^−/− or WT mice were transferred alone and some mice were not reconstituted. 10–12 wk after T cell reconstitution, mice were killed and the development of colonic inflammation was assessed. As described previously 3,4, transfer of CD45RB^high CD4⁺ cells into RAG-2^−/− mice resulted in the development of severe colitis (Fig. 1 A) in the majority of mice (63.4%; Table, top). This colitis was characterized by an extensive inflammatory cell infiltrate, marked epithelial cell hyperplasia, and depletion of mucin-secreting cells. Cotransfer of the CD45RB^low CD4⁺ population significantly inhibited (P < 0.005) the development of colitis, as the majority of mice in this group (83.8%; Table, top) had minimal changes in the intestine (Fig. 1 B). In contrast, cotransfer of CD45RB^low CD4⁺ cells from IL-10^−/− mice failed to protect mice from colitis, as the majority of mice in this group (64.2%; Table, top) developed intestinal inflammation (Fig. 1 C) identical to that seen in mice restored with CD45RB^high CD4⁺ cells alone. Not only did the CD45RB^low population from IL-10^−/− mice fail to protect from colitis, this population actually induced disease with similar characteristics (Fig. 1 D) and incidence (59.0%; Table, top) to that induced by WT CD45RB^high CD4⁺ cells. As described previously 4, immune-deficient mice either unreconstituted or restored with WT CD45RB^low CD4⁺ cells exhibited no pathological changes in the intestine (data not shown). These data indicate that production of IL-10 by CD45RB^low CD4⁺ cells is essential for cells within this population to mediate their immune-suppressive function.

The finding that CD45RB^low CD4⁺ cells from IL-10^−/− mice lacked T cells capable of regulating inflammatory responses in the intestine appears at odds with previous studies from this laboratory which showed that treatment with a neutralizing anti–IL-10 mAb failed to abrogate protection from colitis transferred by CD45RB^low CD4⁺ cells 8. The simplest explanation for these data is that the anti–IL-10 mAb used (JES5-2A5) in these studies failed to sufficiently neutralize IL-10. Recently, an mAb reactive with the murine IL-10R has been generated 13 and shown to efficiently neutralize the effects of IL-10. To test whether treatment with anti–IL-10R mAb was able to affect the function of the CD45RB^low population, mice restored with a mixture of CD45RB^high and CD45RB^low CD4⁺ cells were treated weekly with anti–IL-10, anti–IL-10R, or isotype control mAb. As can be seen in Table (bottom), treatment with anti–IL-10R abrogated protection from colitis induced by the CD45RB^low CD4⁺ population, as all of the mice in these groups, treated with anti–IL-10R alone or in combination with anti–IL-10, developed colitis. Antibody treatment alone did not induce immune pathology in the absence of T cells, as unreconstituted recipients treated with anti–IL-10R did not develop colitis (data not shown). As reported previously, anti–IL-10 treatment had no effect on the immune-suppressive activities of the CD45RB^low population, as mice in this group, like mice treated with isotype control mAb, failed to develop colitis.

Colitis induced by the CD45RB^high CD4⁺ population was characterized by the accumulation of IFN-γ– and TNF-α–secreting Th1 CD4⁺ T cells in the lesions (Fig. 2 A). A similar expansion of Th1 cells was also present in the colon of mice that developed colitis after transfer of WT CD45RB^high CD4⁺ cells in the presence of WT CD45RB^low CD4⁺ cells plus anti–IL-10R or after transfer of IL-10^−/− CD45RB^low cells alone. Development of disease correlated with a significant increase in the total number of CD4⁺ T cells expressing IFN-γ and TNF-α after polyclonal stimulation (8–20-fold expansion of CD4⁺ IFN-γ⁺ cells, and 6–34-fold expansion of CD4⁺ TNF-α⁺ cells; Fig. 2 A) compared with noncolitic mice restored with a mixture of CD45RB^high and CD45RB^low cells. The increase in total Th1 cells characteristic of colitis was due to the marked increase of CD4⁺ cells in the intestine (Table), as analysis of cytokine production at the single cell level revealed that the few CD4⁺ T cells present in nondiseased colons were capable of producing similar levels of IFN-γ and TNF-α as those isolated from colitic mice (Fig. 2 B). The finding that in cotransfers of CD45RB^high and CD45RB^low CD4⁺ cells, neutralization of the function of IL-10 (by treatment with anti–IL-10R or the transfer of IL-10^−/− CD45RB^low cells) led to the development of colitis with identical immunological and histological features to that induced by transfer of pathogenic CD45RB^high cells indicates that IL-10 is a key mediator of the regulatory activity of the CD45RB^low CD4⁺ T cell population.

One way regulatory T cells could mediate their immune-suppressive properties is to induce the differentiation of naive T cells into cells with a similar function, rather than into pathogenic cells. This phenomenon has been termed infectious tolerance 16,17. Given that production of IL-10 by CD45RB^low CD4⁺ cells was essential for their regulatory function and that culture of T cells in the presence of IL-10 can lead to the generation of regulatory T cells 18, it is possible that immune suppression by these cells requires IL-10 production by the normally pathogenic CD45RB^high CD4⁺ population.

To test this, CD45RB^high CD4⁺ cells from IL-10^−/− mice were transferred alone or in combination with WT CD45RB^low CD4⁺ cells to RAG-2^−/− mice. As described previously 19, mice restored with IL-10^−/− CD45RB^high CD4⁺ cells developed a colitis with similar incidence and severity (68.0%; Table) to that transferred by this population isolated from WT mice. However, protection from colitis was equally effective when WT CD45RB^low CD4⁺ cells were cotransferred with IL-10^−/− CD45RB^high CD4⁺ cells, as the majority of mice (81.2%; Table) had no colitis or minimal changes in the intestine. These results indicate that inhibition of inflammatory responses in the intestine mediated by CD45RB^low CD4⁺ cells is not dependent on the differentiation of the progeny of CD45RB^high CD4⁺ cells into IL-10–secreting cells.

Data presented herein provide direct evidence that IL-10 plays an obligate role in the function of regulatory T cells that control inflammatory responses in the intestine. In contrast to CD45RB^low CD4⁺ T cells from WT mice, which inhibit colitis induced in immune-deficient mice after transfer with CD45RB^high CD4⁺ T cells, the CD45RB^low CD4⁺ population from IL-10^−/− mice failed to mediate this function. In addition, they induced severe colitis when transferred alone to immune-deficient recipients. Previous studies from this laboratory showed that TGF-β was essential for inhibition of colitis by CD45RB^low cells 8. These results, together with the findings reported here, provide the first clear evidence that IL-10 and TGF-β play nonredundant roles in the functioning of regulatory T cells which control inflammatory responses towards intestinal antigens, as the neutralization or absence of one of these cytokines is sufficient to abrogate protection. Furthermore, IL-10 produced by regulatory T cells themselves is crucial for the normal functioning of these cells.

Colitis in the SCID model involves the development of Th1 cells responding primarily to intestinal flora, as transfer of CD45RB^high CD4⁺ T cells to germ-free SCID mice failed to induce disease 20. The fact that the regulatory cells express the phenotype of antigen-experienced cells (CD45RB^low) would suggest that their generation in normal mice is antigen driven; however, whether these antigens are of bacterial or self origin is not known. Recent studies of the immune response elicited by Helicobacter hepaticus infection, a bacterium that colonizes the cecum, showed that normal mice mounted an IL-10–dependent response, whereas IL-10^−/− mice developed a pathogenic Th1 response towards the bacterium 21. These studies support the hypothesis that in immunocompetent hosts, enteric antigens induce IL-10–secreting T cells that are immune suppressive and prevent inflammatory responses towards intestinal antigens. Mucosal T cell unresponsiveness to enteric antigens has similarly been shown in humans to be mediated by antigen-specific CD4⁺ T cells and production of IL-10 and TGF-β 22.

Injection of rIL-10 inhibited the development of colitis in CD45RB^high CD4⁺ T cell–restored SCID mice 5 and in IL-10^−/− mice treated from weaning 10. However, the inhibitory effects of IL-10 treatment were only transitory, as colitis developed after the treatment was stopped, whereas the transfer of CD45RB^low CD4⁺ T cells provided long-lasting protection. This may reflect the capacity of regulatory T cells to provide a constant source of IL-10 upon stimulation by endogenous antigen (bacteria or self). Alternatively, they may produce or induce other regulatory molecules, such as TGF-β, which are not induced by exogenous administration of IL-10.

The finding that CD45RB^low CD4⁺ cells incapable of synthesizing IL-10 failed to inhibit colitis induced by transfer of CD45RB^high cells to SCID mice illustrates that the regulatory T cell population itself is the critical source of IL-10 in this model, despite the fact that both the CD45RB^high CD4⁺ T cells and host cells in the SCID mice were capable of making IL-10. It is not clear whether the lack of regulatory T cells in IL-10^−/− mice is a result of abnormalities in the development or effector function of this population. However, the finding that treatment with anti–IL-10R was able to neutralize the immune-suppressive function of differentiated regulatory cells in cotransfers of CD45RB^high and CD45RB^low cells from normal mice supports the idea that IL-10 is required for the effector function. The reasons for the striking difference in effectiveness between the anti–IL-10 mAb JES5-2A5 and the anti–IL-10R antibody 1B1.2 are not clear, as both are effective neutralizing antibodies in vitro. However, the relative ineffectiveness of JES5-2A5 has been observed in other experimental situations, such as Leishmania-infected mice (Mauze, S., and R.L. Coffman, unpublished observations).

Precisely how IL-10 mediates its immune-regulatory function is not known. Inhibition of colitis mediated by IL-10–secreting CD45RB^low CD4⁺ cells was characterized by substantial reductions in total number of Th1 cells recovered from the intestine. This difference was attributable to the reduced number of CD4⁺ cells present in the intestine (8–24-fold; Table), as analysis of cytokine production at the single cell level revealed a similar percentage of CD4⁺ T cells capable of producing IFN-γ and TNF-α in diseased and nondiseased colons. This suggests that the major activity of IL-10–secreting regulatory T cells is to inhibit the accumulation of pathogenic Th1 cells in the intestine. Whether this is due to reduced expansion, or migration, of these cells is not known. IL-10 has been shown to mediate a range of antiinflammatory activities, including the inhibition of antigen-induced proliferation and cytokine secretion by T cells. Inhibition is thought to be mediated mainly by effects on APCs, particularly downregulation of molecules involved in T cell costimulation 23. It seems likely that IL-10–secreting regulatory T cells act to inhibit Th1 cell activation, and that IL-10 produced locally in the intestine acts on macrophages to prevent their activation and elaboration of proinflammatory molecules and chemokines, thus inhibiting T cell recruitment into the intestine. Consistent with this, mice in which macrophages and neutrophils are unable to respond to IL-10 as a result of a cell type–specific deletion of Stat-3 developed enterocolitis 24, suggesting that IL-10–mediated macrophage and neutrophil deactivation contributes to the immune-suppressive properties of IL-10 in the intestine.

Recently, regulatory cells with activities similar to those within the CD45RB^low CD4⁺ population were cloned in vitro by culturing with IL-10 18. These cells, termed T regulatory 1 (Tr-1) cells, are characterized by their ability to produce IL-10 but not IL-4 and to inhibit T cell activation in vitro and in vivo. Although immune suppression in vitro was shown to be dependent on IL-10 and TGF-β, the mechanism of action of Tr-1 cells in vivo has not been established. It is likely that regulatory T cells contained within the CD45RB^low CD4⁺ subset represent the in vivo counterpart of in vitro–derived Tr-1 cells. This is based on the fact that the function of the former population is dependent on IL-10 but not IL-4 synthesis, a defining feature of Tr-1 cells in vitro. In addition, a subset of CD45RB^low CD4⁺ T cells, identified by expression of CD38, was shown to be immune suppressive in vitro and to produce IL-10 but not IL-4 after stimulation with anti-CD3 and IL-2 25. However, further comparison between the in vivo–derived regulatory cells described here and in vitro–derived Tr-1 cells awaits identification of cell surface markers specific for Tr-1 cells and elucidation of their mechanism of action in vivo.

The finding that IL-10 leads to the differentiation of Tr-1 cells in vitro raised the possibility that part of the mechanism by which IL-10–secreting regulatory T cells contained within the CD45RB^low CD4⁺ population inhibit colitis is to induce the differentiation of the progeny of CD45RB^high CD4⁺ cells into IL-10–secreting cells. In some systems, regulatory T cells have been shown to induce the differentiation of naive cells into cells with similar regulatory function, a phenomenon termed infectious tolerance 16,17. However, differentiation of Tr-1 cells among the progeny of the CD45RB^high population did not appear to be a crucial part of the immune-suppressive activities of the CD45RB^low population, as these cells were able to inhibit colitis induced by IL-10^−/− CD45RB^high CD4⁺ cells which could not differentiate into IL-10–secreting cells. However, this result does not rule out the possibility that IL-10 secretion by the CD45RB^low population leads to the differentiation of the progeny of CD45RB^high cells into regulatory T cells secreting other cytokines, for example TGF-β, and experiments are currently underway to test this hypothesis.

In summary, these studies identify IL-10 as an essential mediator of the function of regulatory T cells contained within the CD45RB^low CD4⁺ T cell population. Previous studies identified TGF-β as a critical component of the immune-regulatory function of these cells, and the finding that IL-10 has an identical effect is the first demonstration that both of these immune-suppressive cytokines play mandatory roles. How the functions of these two cytokines are linked is not known. The finding that IL-10^−/− mice have immune pathology restricted to the intestine 9 whereas TGF-β1^−/− develop multiple organ disease 26 suggests that IL-10 is not required for the production of TGF-β1. However, TGF-β has been shown to induce IL-10 secretion by APCs 27, making it a possibility that TGF-β alters antigen presentation in favor of the generation of IL-10–secreting regulatory T cells. Alternatively, IL-10 and TGF-β may act entirely separately, and further experiments are required to distinguish between these possibilities.

There is now good evidence that regulatory T cells can be induced after oral exposure to antigen and that their function is dependent on TGF-β 28,29. In addition, regulatory T cells that inhibit the development of autoimmune disease have been shown to exist naturally in mice and rats (for a review, see reference 30). However, information regarding the role of IL-10 and TGF-β in the function of these cells is incomplete, as in most cases only one or the other has been examined. Thus, TGF-β–dependent mechanisms were shown to be involved in the regulation of autoimmune nephritis 31 and thyroiditis 32, whereas inhibition of diabetes by NK1.1 T cells was dependent on IL-4 and IL-10 33. It remains to be established whether regulatory T cells that control inflammatory responses in the gut are the same as those shown to regulate organ-specific autoimmune disease. However, more information on the relative roles of IL-10 and TGF-β in the function of these different populations of cells will facilitate their comparison. Differentiation of IL-10–secreting TGF-β–dependent regulatory T cells is one of the host's natural mechanisms for avoiding immune pathology. Further understanding of the antigen specificity, development, and mechanism of action of these cells is crucial for the design of effective immune interventions that seek to capitalize on this potent immune-regulatory mechanism.

We are very grateful to Dr. Don Mason for critical review of the manuscript. We are indebted to N. Rust for excellent technical assistance with cell sorting, P. McShane for help with statistics, and C. Hetherington and staff for care of experimental animals.

F. Powrie and C. Asseman are funded by the Wellcome Trust. DNAX Research Institute is funded by the Schering-Plough Corporation.