Induction of Colonic Regulatory T Cells by Indigenous Clostridium Species

See allHide authors and affiliations

Science  21 Jan 2011:
Vol. 331, Issue 6015, pp. 337-341
DOI: 10.1126/science.1198469


CD4+ T regulatory cells (Tregs), which express the Foxp3 transcription factor, play a critical role in the maintenance of immune homeostasis. Here, we show that in mice, Tregs were most abundant in the colonic mucosa. The spore-forming component of indigenous intestinal microbiota, particularly clusters IV and XIVa of the genus Clostridium, promoted Treg cell accumulation. Colonization of mice by a defined mix of Clostridium strains provided an environment rich in transforming growth factor–β and affected Foxp3+ Treg number and function in the colon. Oral inoculation of Clostridium during the early life of conventionally reared mice resulted in resistance to colitis and systemic immunoglobulin E responses in adult mice, suggesting a new therapeutic approach to autoimmunity and allergy.

The mammalian gastrointestinal tract harbors numerous species of commensal bacteria that constitute the “microbiota.” The microbiota interacts with the host immune system, inducing the accumulation of several different lymphocyte populations at mucosal sites (1, 2). Recent reports have suggested that the induction of each lymphocyte subset may be regulated by a distinct component of the microbiota. For instance, segmented filamentous bacteria (SFB) strongly induce intestinal T helper 17 (TH17) cells, which play a role in host resistance against intestinal pathogens and promote systemic autoimmunity (35).

CD4+ regulatory T cells (Tregs) expressing the transcription factor forkhead box P3 (Foxp3) are present at higher frequencies in the gut lamina propria (LP), particularly in the colon, than in other organs (6) (fig. S1). It has been postulated that the number and function of mucosal Tregs are affected by the presence of intestinal bacteria. Indeed, daily treatment of mice with probiotic strains of bifidobacteria and lactobacilli modifies the inflammatory status of mice, presumably by inducing Tregs (79). Furthermore, colonization of mice with human commensal Bacteroides fragilis facilitates Treg differentiation and interleukin-10 (IL-10) production (10). Given the importance of the community structure of “indigenous” microbial flora in the maintenance of intestinal homeostasis, and that its alteration (dysbiosis) correlates with inflammatory diseases (11), it is important to further investigate whether, and how, indigenous microflora affect the number and function of mucosal Tregs.

We first examined the appearance of Tregs during mouse ontogeny. The frequency of Foxp3+ Tregs in colonic and small intestinal (SI) LP increased after weaning, whereas in inguinal lymph nodes (iLNs) it remained stable from the second week after birth (Fig. 1A). This temporal accumulation of intestinal Tregs suggested an influence of the intestinal microbiota. Therefore, we next examined germ-free (GF) mice. The percentage and absolute number of Foxp3+CD4+ T cells in SI, iLNs, Peyer’s patches, and mesenteric LNs were unchanged, or increased, in GF mice and antibiotic-treated specific pathogen-free (SPF) mice compared with untreated SPF mice (Fig. 1B and fig. S2). These findings are consistent with previous observations that GF mice have increased or unchanged numbers of Tregs in SI (12, 13). In contrast, a significant decrease in the number of Foxp3+ Tregs was observed in the colonic LP of GF mice or antibiotic-treated mice compared with SPF mice (Fig. 1B and fig. S2). This decrease may be attributed to the absence of specific signaling events induced by intestinal microbes rather than to a defect in the development of gut-associated lymphoid tissues (fig. S3). Indeed, when GF mice were colonized with fecal suspensions from SPF mice (“conventionalization”), a marked increase in the frequency of Tregs was observed in colonic LP (Fig. 1C). Therefore, we conclude that interactions between indigenous microflora and the host play a critical role in the accumulation of colonic LP, but not SI LP, Foxp3+ Tregs.

Fig. 1

Indigenous intestinal bacteria-dependent accumulation of colonic Tregs. (A) The percentage of Foxp3+ cells within the CD4+ cell population isolated from iLNs or LP of colon or SI of SPF BALB/c mice at the indicated age was analyzed by flow cytometry. (B) Lymphocytes from SI, colon, and iLN of 8-week-old BALB/c, IQI and C57BL/6 (B6) GF, and SPF mice were analyzed for CD4 and Foxp3 expression. (C) GF IQI mice were conventionalized (Conv) by oral administration of the fecal suspension from B6 SPF mice purchased from Jackson Laboratory. Colonic LP lymphocytes were isolated 3 weeks later and analyzed for Foxp3 expression. (D) Four-week-old SPF B6 mice were treated with polymyxin B (poly B) or vancomycin (Vanco) for 4 weeks and analyzed for the percentage of Foxp3+ cells within the CD4+ cell population. (E) GF mice were gavaged with chloroform-treated feces from SPF mice (+chloro) and analyzed for the percentage of Foxp3+ cells within the CD4+ cell population. Each circle in (B) to (E) represents an individual mouse, and error bars indicate the SD. Data were obtained from more than two independent experiments with similar results (n ≥ 4 mice per group). *P < 0.01; **P < 0.001, unpaired t test.

To determine whether a specific component of the intestinal flora induces colonic Treg accumulation, we treated SPF mice with antibiotics that preferentially target Gram-positive (vancomycin) or Gram-negative (polymyxin B) bacteria (fig. S4). Compared with the controls, only mice treated with vancomycin had significantly lower frequencies of Tregs in the colon (Fig. 1D), suggesting a dominant role for Gram-positive commensal bacteria in Treg accumulation. We next orally inoculated GF mice with 3% chloroform-resistant fecal microorganisms (spore-forming fraction) because this fraction has been shown to regulate intestinal T cell responses (4). Mice inoculated with chloroform-treated feces showed an increased number of Tregs (Fig. 1E) comparable to that in SPF mice or GF mice gavaged with untreated feces (Fig. 1, B and C). Collectively, our results suggest that a specific component of the indigenous microbiota, belonging most likely to the Gram-positive, spore-forming fraction, plays a critical role in the induction of colonic Tregs.

Clostridia are one of the most prominent Gram-positive and spore-forming bacteria indigenous to the murine gastrointestinal tract (14). Moreover, Clostridium clusters IV and XIVa (also known as Clostridium leptum and coccoides groups, respectively) have been implicated in the maintenance of mucosal homeostasis and prevention of inflammatory bowel disease (IBD) (15, 16). We therefore explored the link between clostridia and the accumulation of colonic Tregs. Clostridia became prominent after weaning and persisted in the adult animals, in contrast to Lactobacillus or Enterobacteriaceae, which were more abundant during the neonatal period and declined thereafter (fig. S5). Furthermore, Clostridium clusters IV and XIVa were most abundant in the cecum and proximal colon, which correlated well with the distribution of Tregs (fig. S6).

To directly examine the effect of Clostridium on the induction of colonic Tregs, we generated “gnotobiotic” mice by colonizing GF mice with fecal material obtained from mice colonized with a cocktail of 46 strains of Clostridium (17). Clostridium-colonized gnotobiotic mice exhibited a robust accumulation of Tregs in colonic LP; however, the frequency of SI Tregs was not affected (Fig. 2A and fig. S7, A and B). The 46 strains of Clostridium used were originally isolated from chloroform-treated fecal material from conventionally reared mice (17). These strains primarily belong to clusters IV and XIVa (14) (see also fig. S8). They are normally present in the intestine of several colonies of commercially available SPF mice and in that of mice housed in our own animal facility (fig. S9A). They mainly colonized the cecum and proximal colon (fig. S9B) and formed a thick layer on the mucosal epithelium (Fig. 2B). The induction of colonic Tregs was specific to Clostridium-colonized mice. Colonization with a cocktail of three strains of Lactobacillus, or with SFB, had little effect on the number of colonic Tregs (Fig. 2A and fig. S7, A and B). Furthermore, colonization by B. fragilis, which is reported to induce IL-10 production in Tregs (10), did not significantly affect the frequency of colonic Tregs (fig. S7C). A cocktail of 16 strains of Bacteroides spp. isolated from the murine intestine did induce a significant increase in the number of Tregs in the colon, although the magnitude was dependent on the background of the mice and was less than that seen in Clostridium-colonized mice (Fig. 2A and fig. S7, A and B).

Fig. 2

Clostridia induce Treg accumulation in colonic LP. (A) GF BALB/c or IQI mice were colonized with segmented filamentous bacteria (SFB), 16 strains of Bacteroides (Bactero.), 3 strains of Lactobacillus (Lacto.), or 46 strains of Clostridium (Clost.) for 3 weeks. The percentage of Foxp3+ cells within the CD4+ cell population in the colon and SI of individual mice was analyzed by flow cytometry (n ≥ 5 mice per group). (B) Electron micrograph showing the proximal colon of Clost.-colonized B6 mice. (C) Lymphocytes from the colonic LP of indicated mice were analyzed for the expression of CD4, Foxp3, and Helios by flow cytometry. The percentage of Helios cells within Foxp3+CD4+ lymphocytes is shown (n = 4 mice per group). (D and E) Whole colons (D) or IECs (E) from GF, Lacto.-, or Clost.-colonized mice were cultured for 24 hours. The concentration of TGF-β1 in the supernatant was determined by enzyme-linked immunosorbent assay (ELISA). (F) Splenic CD4+ T cells were cultured with antibodies against CD3 and CD28 and 50% conditioned medium from cultured IECs isolated from GF or Clost.-colonized mice in the presence or absence of anti–TGF-β. After 5 days, T cells were collected and assayed for Foxp3 expression by real-time reverse transcription–polymerase chain reaction. Each circle represents a mouse or a sample, and error bars indicate the SD (n ≥ 3 per group). *P < 0.02; **P < 0.001, unpaired t test. Data are representative of at least two independent experiments with similar results.

A substantial fraction of Foxp3+ Tregs observed in SPF or Clostridium-colonized mice was negative for Helios, a transcription factor reported to be expressed in thymus-derived natural Tregs (18), suggesting that many of these Tregs could be “induced Tregs” (iTregs) (Fig. 2C and fig. S10A). A high number of Tregs was maintained for at least 4 months after colonization (fig. S10B). Furthermore, Clostridium-mediated Treg induction is vertically and horizontally transmissible (fig. S10C). Clostridium did not affect TH1 cells but moderately induced TH17 cells in the colon (fig. S10D). The 46 strains of Clostridium have been reported to affect the accumulation of CD8+IELs (intraepithelial lymphocytes) in the colon (19). Therefore, Clostridium may modulate various aspects of the immune system.

Considering that transforming growth factor–β (TGF-β) is a critical regulator of Treg development, we examined whether Clostridium colonization provides a TGF-β–rich environment within the colon. To this end, we cultured whole colons or colonic intestinal epithelial cells (IECs) from GF mice, or mice colonized by either Clostridium or Lactobacillus, and found a significant increase in the production of active-form TGF-β in the colon and IECs from Clostridium-colonized mice (Fig. 2, D and E). Addition of the culture supernatant of IECs from Clostridium-colonized mice to splenic CD4+ T cells markedly enhanced the differentiation of Foxp3-expressing cells, and this differentiation was significantly inhibited by an antibody against TGF-β (Fig. 2F). Consistent with the production of TGF-β, transcripts for the genes encoding matrix metalloproteinase 2 (MMP2), MMP9, and MMP13, which have been reported to be involved in the activation of latent TGF-β (20), were expressed at higher levels by IECs from Clostridium-colonized mice than those from GF or Lactobacillus-colonized mice (fig. S11). IECs from Clostridium-colonized mice also expressed a high level of indoleamine 2,3-dioxygenase (IDO) (fig. S11), which has been implicated in the induction of Tregs (21). Collectively, our findings suggest that clostridia activate IECs to produce TGF-β and other Treg-inducing molecules within the colon. We also examined the contribution of bacterial pattern-recognition receptors (PRRs) to the induction of Tregs by Clostridium. SPF mice deficient for Myd88 (a signaling adaptor molecule for Toll-like receptors), Rip2 (an adaptor molecule for NOD receptors), or Card9 (a key transducer of Dectin-1 signaling) had a normal number of mucosal Tregs, compared to each of their littermate controls (fig. S12A). Furthermore, the colonization of GF Myd88-deficient mice with Clostridium induced robust accumulation of colonic LP Tregs (fig. S12B). These results indicate that clostridia induce Tregs independently of these PRR signaling pathways.

Intestinal Foxp3+ Tregs exert their immunosuppressive activity, at least in part, through IL-10 production (22, 23). To further probe the nature of Tregs induced by Clostridium, we newly generated IL-10 reporter mice, in which a cassette containing an internal ribosomal entry site (IRES) and Venus, a brighter version of yellow fluorescent protein, was inserted immediately before the polyadenylation signal of the Il10 gene (referred to as Il10Venus mice; fig. S13). In SPF Il10Venus mice, about 60% of Foxp3+ Tregs in colonic LP were Venus+, whereas only about 10% of Tregs in spleen or other organs were Venus+, consistent with previous studies (24, 25) (Fig. 3, A and B, and fig. S14). Under GF conditions or after antibiotic treatment, the frequency and number of Venus+ cells within the CD4+Foxp3+ Treg population were significantly lower than those seen under SPF conditions (Fig. 3, A to C, and fig. S15). In contrast, no significant change in the frequency of any regulatory cell components within the CD4+ T cell population in SI LP was observed in the absence, or reduction, of commensal bacteria (fig. S15). These results indicate that the indigenous microflora provide a signal for the accumulation of IL-10+ Tregs in colonic LP, whereas different mechanisms are operating in the induction of Foxp3- and/or IL-10+–regulatory cells in SI.

Fig. 3

Clostridia induce IL-10–expressing Tregs in the colon. (A to D) Splenic and colonic LP lymphocytes from SPF, GF, or Clostridium-colonized Il10Venus mice were analyzed for expression of CD4, Foxp3, Venus, and CTLA4 by flow cytometry. Representative dot-plots and histograms are shown in (A) and (D), and the percentage and absolute number of Venus+ cells within the CD4+Foxp3+ cell population in individual mice are shown in (B) and (C). Each symbol represents a mouse, and error bars indicate the SD. Data are representative of at least two independent experiments with similar results (n ≥ 5 mice per group). **P < 0.001 versus GF, unpaired t test. (E) Purified CD4+Venus+ cells from the colonic LP of SPF or Clostridium-colonized Il10Venus mice were cultured with splenic CD4+CD25 T cells (Teff) in the presence of irradiated splenic CD11c+ cells and anti-CD3 for 72 hours at the indicated ratios. Proliferation was measured by [3H]thymidine uptake. Data represent the mean ± SD of triplicate cultures.

To determine the effect of Clostridium on IL-10 expression in colonic Tregs, we examined gnotobiotic Il10Venus mice colonized with 46 strains of Clostridium. The frequency and number of Venus+ cells among Foxp3+ cells in Il10Venus mice colonized with Clostridium were similar to those in SPF mice (Fig. 3, A to C), indicating that clostridia are sufficient to induce IL-10–expressing Foxp3+ Tregs. The induction of IL-10+ Tregs was, again, specifically observed in mice colonized with Clostridium but not other bacteria (Fig. 3, B and C). Venus+ Foxp3+ cells in the colonic LP of Clostridium-colonized mice expressed high levels of cytotoxic T lymphocyte antigen 4 (CTLA4) (Fig. 3D) and exhibited an in vitro suppressive activity similar to that of Venus+ cells from SPF mice (Fig. 3E). These results indicate that clostridia sufficiently induce the accumulation of functionally competent IL-10+ CTLA4high Tregs. Many of the Venus+ Foxp3+ cells in the colonic LP of SPF and Clostridium-colonized mice were found to be negative for Helios, and this cell fraction was particularly reduced in GF mice (fig. S16). These findings suggest that clostridia induce a shift in the composition of Tregs in the colonic LP, particularly through accumulation of the IL-10+ CTLA4high Helios subset of Tregs (presumably iTregs). Furthermore, 3 weeks after Clostridium inoculation, we observed a substantial increase in the number of Venus+CD4+ cells in the liver, lung, and spleen, where otherwise a very small number of Venus+ cells can be detected (fig. S17). This finding suggests that colonization of Clostridium also affects the extra-intestinal immune status.

Early exposure to the environment is known to be a key determinant of adult gut microbial ecology. To affect the Clostridium load, we orally inoculated 2-week-old neonatal SPF mice with feces from Clostridium-associated mice, kept them under SPF conditions, and examined their microbial composition and Treg number in adulthood. Despite similar amounts of total bacteria in the feces of the Clostridium-treated and -untreated groups, there was a significant increase in the amounts of Clostridium clusters IV and XIVa in the treated mice (fig. S18A), accompanied by a significantly higher number of colonic Foxp3+ Tregs (Fig. 4A). We then examined the effect of abundance of Clostridium, and the consequent increase in the number of Tregs, on local and systemic immune responses. Mice were subjected to dextran sodium sulfate (DSS)–mediated colitis, a model of colitis resembling human IBD. The symptoms of colitis, such as weight loss and rectal bleeding, were significantly suppressed in Clostridium-abundant mice compared with control mice (Fig. 4B). Colon shortening, edema, and hemorrhage were less severe in the colon of Clostridium-abundant mice than in that of control mice (Fig. 4C). Consistently, Clostridium-abundant mice exhibited milder histological disease characteristics, such as mucosal erosion, cellular infiltration, and crypt loss, than control mice (Fig. 4D). The Clostridium-abundant mice were also subjected to oxazolone-induced colitis, an experimental colitis mediated by TH2-type cells (26). Clostridium-abundant mice had attenuated weight loss and fewer areas of mucosal erosion, edema, cellular infiltration, and hemorrhage (fig. S18, B and C). Finally, we investigated the effect of Clostridium on systemic immunoglobulin E (IgE) production. Ovalbumin (OVA)–specific IgE levels in sera were measured after intraperitoneal injection of alum-absorbed OVA. Clostridium-abundant mice showed significantly lower IgE levels than control mice (Fig. 4E). Moreover, splenocytes from Clostridium-abundant mice immunized with OVA plus alum showed lower IL-4 and higher IL-10 productions after restimulation with OVA in vitro than those from control mice (Fig. 4F). Therefore, the increased the proportion of Clostridium in the gut microbiota affected mucosal inflammation and systemic antibody responses. Although additional mechanisms and cell types could be involved, our findings suggest that Clostridium-mediated induction of Tregs in the colon may be responsible for these effects.

Fig. 4

Effect of Clostridium abundance on regulation of colitis and IgE response. (A) Two-week-old SPF mice were orally inoculated with Clostridium (SPF+Clost.) or untreated (SPF). After 6 weeks, the percentage of Foxp3+ cells within CD4+ cells in colonic LP was analyzed. The experiment was repeated more than three times with similar results. Each circle represents a mouse, and error bars represent the SD (n = 7 mice per group). (B to D) SPF and SPF+Clost. mice were treated with 2% DSS and monitored and scored for body weight loss, stool consistency, and bleeding for 6 days (n = 7 mice per group). (B) On day 6, the colons were collected (C) and dissected for histological analysis by hematoxylin and eosin staining (D). (E and F) SPF and SPF+Clost. mice were immunized with OVA + alum twice at a 2-week interval. Sera were collected and examined for OVA-specific IgE levels by ELISA (E). Splenocytes were collected from mice in each group and examined for IL-4 and IL-10 production upon restimulation with OVA in vitro (F). Error bars represent the SD (n = 5 mice per group). *P < 0.02; **P < 0.001, unpaired t test.

Our findings show that Tregs are abundant in intestinal LP, and their accumulation in the SI and colon is differentially regulated. The induction of colonic Tregs is dependent on commensal microorganisms with specialized properties. Among the indigenous commensal bacteria, Clostridium spp. belonging to clusters IV and XIVa are outstanding inducers of Tregs in the colon. Although alternative mechanisms may also be involved, our findings are consistent with a model in which the presence of Clostridium induces the release of active TGF-β and other Treg-inducing factors from IECs, which presumably cooperate with dendritic cells to induce a general accumulation of Tregs in the colon and at the same time affect the proportions of individual Treg subsets through the preferential induction of IL-10+ CTLA4high iTregs. Several recent studies have focused on the microbial composition in the intestine during health and disease. Notably, Clostridium clusters IV and XIVa constitute a smaller proportion of the fecal community in patients with IBD than in healthy controls (15). Furthermore, some patients with IBD have a specific reduction in Faecalibacterium prausnitzii, a bacterium belonging to Clostridium cluster IV (16). These reports are consistent with our findings and raise the possibility that indigenous Clostridium-dependent induction of Tregs may be required for maintaining immune homeostasis in mice and humans. The factors derived from Clostridium that are required for the induction of mucosal Tregs are currently unknown. Because gnotobiotic mice colonized with three strains of Clostridium showed an intermediate pattern of Treg induction between GF mice and mice inoculated with all 46 strains (fig. S19), we speculate that a diverse set of metabolites that are most efficiently produced by the 46 strains of Clostridium as a whole may be required for the optimal induction of Tregs. Identifying these metabolites and the molecular mechanisms underlying the Clostridium-host crosstalk will provide invaluable information toward understanding how the gut microbiota regulates immune homeostasis and may suggest potential therapeutic options for treating human IBD and allergies.

Supporting Online Material

Materials and Methods

Figs. S1 to S19


References and Notes

  1. We thank Y. Ueda, J. Nishimura, and H. Saiga for technical assistance, R. Eisenman for critically reading the manuscript, and A. Miyawaki for Venus. We thank the staff at the Sankyo Laboratories for gnotobiotic handling of the mice. The work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology, the Mochida Memorial Foundation for Medical and Pharmaceutical Research, the Kato Memorial Bioscience Foundation, the Mishima Kaiun Memorial Foundation, Kanae Foundation for the Promotion of Medical Science, and Inoue Foundation for Science. The authors have applied for a patent for use of bacteria belonging to the genus Clostridium or a physiologically active substance derived from these bacteria to induce proliferation or accumulation of regulatory T cells and suppress immune functions (patent application no. JP 2010-129134). Material transfer agreements are required for the use of the IL-10 Venus mice, SFB, Clostridium spp. 46 strains, Bacteroides spp. 16 strains, and Lactobacillus spp. 3 strains.

Stay Connected to Science

Navigate This Article