Letter | Published:

c-MAF-dependent regulatory T cells mediate immunological tolerance to a gut pathobiont

Nature volume 554, pages 373377 (15 February 2018) | Download Citation

This article has been updated


Both microbial and host genetic factors contribute to the pathogenesis of autoimmune diseases1,2,3,4. There is accumulating evidence that microbial species that potentiate chronic inflammation, as in inflammatory bowel disease, often also colonize healthy individuals. These microorganisms, including the Helicobacter species, can induce pathogenic T cells and are collectively referred to as pathobionts4,5,6. However, how such T cells are constrained in healthy individuals is not yet understood. Here we report that host tolerance to a potentially pathogenic bacterium, Helicobacter hepaticus, is mediated by the induction of RORγt+FOXP3+ regulatory T (iTreg) cells that selectively restrain pro-inflammatory T helper 17 (TH17) cells and whose function is dependent on the transcription factor c-MAF. Whereas colonization of wild-type mice by H. hepaticus promoted differentiation of RORγt-expressing microorganism-specific iTreg cells in the large intestine, in disease-susceptible IL-10-deficient mice, there was instead expansion of colitogenic TH17 cells. Inactivation of c-MAF in the Treg cell compartment impaired differentiation and function, including IL-10 production, of bacteria-specific iTreg cells, and resulted in the accumulation of H. hepaticus-specific inflammatory TH17 cells and spontaneous colitis. By contrast, RORγt inactivation in Treg cells had only a minor effect on the bacteria-specific Treg and TH17 cell balance, and did not result in inflammation. Our results suggest that pathobiont-dependent inflammatory bowel disease is driven by microbiota-reactive T cells that have escaped this c-MAF-dependent mechanism of iTreg–TH17 homeostasis.


We chose H. hepaticus as a model to investigate host–pathobiont interactions. Blockade of IL-10RA induced inflammation of the large intestine in H. hepaticus-colonized Il23rGFP reporter mice5,6, increasing the proportion of green fluorescent protein-positive (GFP+) cells (predominantly TH17 cells) from approximately 10% to 50% of large intestine CD4+ T cells (Extended Data Fig. 1a). We therefore sought to determine why H. hepaticus-induced T cells do not cause disease in wild-type mice at the steady state. To address this question, we first identified the T cell receptor (TCR) sequences and cognate epitopes of H. hepaticus-induced TH17 cells that expand during inflammation, and subsequently traced the fate of these cells at the steady state.

We cloned individual TCR sequences from colitogenic IL-23R–GFP+ T cells (Extended Data Fig. 1b) and found that nine out of twelve clonotypic TCRs were H. hepaticus-specific (Extended Data Fig. 1c). We subsequently identified7,8 a H. hepaticus-unique protein, HH_1713, that contains two immunodominant epitopes. The E1 peptide epitope, presented by I-Ab, was recognized by H. hepaticus-specific TCR HH5-1, whereas E2 was recognized by TCR HH5-5, HH6-1 and HH7-2 (Extended Data Fig. 1c). We next developed two complementary approaches to track H. hepaticus-specific T cells in vivo9,10, HH7-2 and HH5-1 TCR transgenic mice (HH7-2tg and HH5-1tg) and a major histocompatibility complex (MHC) class II tetramer loaded with E2 peptide (HH-E2 tetramer) (Extended Data Fig. 1d–g).

To track what happens to H. hepaticus-specific T cells in healthy mice, we simultaneously transferred naive T cells from HH7-2tg and 7B8tg (segmented filamentous bacteria (SFB)-specific TCRtg control)8 mice into wild-type mice that were stably colonized with H. hepaticus and SFB (Fig. 1a). Two weeks after adoptive transfer, HH7-2tg donor cells were enriched in the large intestinal lamina propria (LILP) and caecal patch, whereas 7B8tg cells predominated in the small intestinal lamina propria (SILP) and Peyer’s patches (Extended Data Fig. 2a, b), consistent with colonization of H. hepaticus in the large intestine and SFB in the small intestine. As previously reported, 7B8 cells developed into TH17 cells that were largely positive for RORγt and negative for FOXP38 (Fig. 1b, c, Extended Data Fig. 2c, d). By contrast, HH7-2tg cells in the LILP were mostly iTreg cells that express both RORγt and FOXP3 (approximately 60% of total donor-derived HH7-2tg cells)11,12, rather than TH17 cells (less than 10% of total HH7-2tg cells) (Fig. 1b, c, Extended Data Fig. 2c, d). Notably, two other colonic Treg cell markers, GATA3 and ST2, were not expressed on HH7-2tg cells13 (Extended Data Fig. 2e). 7B8tg and HH7-2tg T cells that expressed neither RORγt nor FOXP3 were mostly T follicular helper (TFH) cells and were enriched in the Peyer’s patches and caecal patch (Fig. 1b, c, Extended Data Fig. 2c, d). Breeding HH7-2tg mice onto the Rag1−/− background excluded the possibility that HH7-2tg iTreg cells detected after adoptive transfer were contaminated by thymus-derived natural Treg (nTreg) cells or were influenced by the presence of dual TCRs (Extended Data Fig. 3a–c). Adoptively transferred HH5-1tg and HH-E2-tetramer-positive cells had differentiation profiles similar to HH7-2tg cells (Fig. 1d, e, Extended Data Figs 2f and 3d, e). These results indicate that the host responds to H. hepaticus by generating immunotolerant iTreg cells rather than pro-inflammatory TH17 cells.

Figure 1: H. hepaticus induces RORγt+ Treg and TFH cell responses under steady state.
Figure 1

a, Experimental scheme for co-transfer of congenic isotype-labelled HH7-2tg and 7B8tg cells into wild-type (WT) mice colonized with H. hepaticus and SFB. b, c, RORγt, FOXP3, BCL6 and CXCR5 expression (b) and frequencies of Treg (FOXP3+), TH17 (FOXP3RORγt+) and TFH (BCL6+CXCR5+) (c) among donor-derived T cells in indicated tissues. Data are from 1 of 3 experiments, with n = 15 in the 3 experiments. CP, caecal patch; PP, Peyer’s patches. APC (allophycocyanin), Cy5.5, eF780, FITC, PE (phycoerythrin) and PerCP denote fluorochrome labels. d, e, Wild-type mice (n = 6) were colonized with H. hepaticus for 3–4 weeks and analysed for RORγt, FOXP3 and BCL6 expression in total CD4+ (red) and HH-E2 tetramer+ (blue) T cells from the LILP (d) and frequencies of Treg (FOXP3+), TH17 (FOXP3RORγt+) and TFH (BCL6+) cells among HH-E2 tetramer+ T cells in the LILP and caecal patch (e). Data summarize two independent experiments.

To examine whether the iTreg cell-dominant differentiation of H. hepaticus-specific T cells is altered during intestinal inflammation, we co-transferred naive HH7-2tg and control 7B8tg T cells into colonized Il10−/− recipients. Notably, only a small proportion of the transferred HH7-2tg T cells expressed FOXP3 in the LILP. Instead, most of them differentiated into pro-inflammatory TH17 cells with TH1-like features, characterized by the expression of both RORγt and T-bet and high levels of IL-17A and IFNγ upon re-stimulation14 (Fig. 2a–f, Extended Data Fig. 4a, c, d). These results were recapitulated with HH5-1tg T cell adoptive transfer and endogenous HH-E2 tetramer+ T cells (Extended Data Fig. 4e–g). By comparison, disruption of IL-10-mediated immune tolerance did not result in deviation of SFB-specific TH17 cells to the inflammatory TH17–TH1 cell phenotype (Fig. 2c, d, Extended Data Fig. 4a–d). Furthermore, we observed similar deviated T cell responses to H. hepaticus in models of T cell transfer colitis and Citrobacter rodentium-induced colonic inflammation, but not in dextran sulfate sodium (DSS) colitis, an innate immunity-dependent model (Extended Data Fig. 5a–h). Commensal microorganism-specific T cells can thus acquire pro-inflammatory phenotypes during enteric infection15, although the high frequency of such infections suggests the existence of a mechanism to re-establish gut tolerance. Our observations of H. hepaticus-specific iTreg–TH17 skewing during colitis are consistent with a contemporaneous study using two different Helicobacter species16. These findings indicate that dysregulated T cell tolerance to pathobionts may be a general hallmark of inflammatory bowel disease.

Figure 2: H. hepaticus predominantly induces inflammatory TH17 cells in IL-10 deficiency-dependent colitis.
Figure 2

ad, LILP HH7-2tg and SILP 7B8tg donor-derived cells in Il10+/− (n = 8) and Il10−/− (n = 10) mice were analysed for FOXP3 and RORγt expression (a), frequencies of Treg (FOXP3+) and TH17 (FOXP3RORγt+) cells (b), RORγt and T-bet co-expression (c), and frequencies of T-bet expression among TH17 (FOXP3RORγt+) cells (d). Data are from four independent experiments. e, f, IL-17A and IFNγ expression (e) and frequencies of IL-10+, IL-17A+ and IFNγ+ cells (f) among LILP HH7-2tg donor-derived cells in Il10+/− (n = 5) and Il10−/− (n = 6) mice after re-stimulation. Data summarize two independent experiments. All statistics were calculated by unpaired two-sided Welch’s t-test. Error bars denote mean ± s.d. P values are indicated on the figure.

We next wished to determine whether RORγt+ Treg cells are critical for immune tolerance to gut pathobionts. The transcription factor c-MAF attracted our attention as it was highly enriched in these cells11,17 (Extended Data Fig. 6a) and known to promote an anti-inflammatory program, for example, IL-10 expression in other T helper cell subsets18,19. We therefore deleted Maf with Foxp3cre to test its function in Treg cells. In H. hepaticus-colonized Maffl/fl;Foxp3cre (Maf∆Treg) mice, despite incomplete depletion of c-MAF protein (Extended Data Fig. 6b), there was a marked decrease in the proportion of RORγt+ but not RORγt Treg cells among CD4+ T cells in the large intestine, and a concomitant increase in TH17 cell frequency (Fig. 3a). Maf∆Treg mice also had expanded numbers of total CD4+ T cells in the large intestine, reflected by a pronounced accumulation of TH17 cells, but notably not RORγt+ Treg cells (Extended Data Fig. 6c). By contrast, after H. hepaticus colonization, TH17 cell expansion was less striking in Rorcfl/fl;Foxp3cre (Rorc∆Treg) mice (Fig. 3a, Extended Data Fig. 6c), and neither a decrease of RORγt+ Treg cells nor an expansion of TH17 cells was observed in Gata3fl/fl;Foxp3cre (Gata3∆Treg) mice (Extended Data Fig. 6d, e). The altered frequency of RORγt+ Treg and TH17 cell subsets led us to test whether the fate of H. hepaticus-specific T cells would be affected in the Maf∆Treg and Rorc∆Treg mice. Notably, HH-E2 tetramer+ cells were predominantly TH17 in Maf∆Treg mice, but mostly RORγt+ Treg in control mice (Fig. 3b, Extended Data Fig. 6f). By contrast, although Rorc∆Treg mice also had an increased proportion of H. hepaticus-specific TH17 cells, most tetramer+ cells were Treg cells (Fig. 3b, Extended Data Fig. 6f). Collectively, these results suggest that pathobiont-specific RORγt+ iTreg cells are required for the suppression of inflammatory TH17 cell accumulation. Although RORγt expression contributes to gut iTreg cell function, c-MAF has a more substantial role in the differentiation and/or function of these cells. H. hepaticus-specific TFH cell differentiation in the caecal patch did not seem to be affected in Maf∆Treg mice (Extended Data Fig. 6g). Notably, as in IL-10-deficient mice, SFB-specific TH17 cells neither expanded nor adopted a TH1-like phenotype in Maf∆Treg mice (Extended Data Fig. 6h, i). A potential explanation is that SFB- and H. hepaticus-specific TH17 responses are instructed by different innate immune pathways20,21.

Figure 3: c-MAF is required for the differentiation and function of induced Treg cells in the gut.
Figure 3

a, b, Transcription factor staining in total CD4+ (a) and HH-E2 tetramer+ (b) T cells from the LILP of indicated mice. Left, RORγt and FOXP3 expression. Right, frequencies of indicated Treg (FOXP3+) and TH17 (FOXP3RORγt+) cell subsets. Mice were colonized with H. hepaticus for 5–6 weeks before analysis. Data summarize 3 independent experiments for Rorc∆Treg (n = 7) and littermate controls (ctrl; n = 7 for a, n = 6 for b), and 4 independent experiments for Maf∆Treg (n = 10) and littermate controls (n = 8). ce, Co-transfer of Maf∆Treg and control HH7-2tg T cells into wild-type H. hepaticus-colonized mice. c, Left, donor cell composition in the LILP of recipient mice. Right, RORγt and FOXP3 expression in indicated donor-derived cells. d, Ratios of Maf∆Treg (knockout, KO) versus control (WT) HH7-2tg donor-derived cells in the LILP. Dashed line represents ratio of co-transferred cells before transfer. e, Frequencies of Treg (FOXP3+) and TH17 (FOXP3RORγt+) cells among donor-derived cells. Data are a summary of 10 mice from 2 independent experiments. a, b, e, Statistics were calculated by unpaired two-sided Welch’s t-test. Error bars denote mean ± s.d. P values are as follows: a, i = 1.21 × 10−6, ii = 8.82 × 10−7, iii = 0.016, iv = 6.38 × 10−4, v = 8.06 × 10−4 and vi= 9.89 × 10−7. b, i = 0.056, ii = 7.48 × 10−4, iii = 7.64 × 10−7 and iv = 6.01 × 10−6. e, i = 6 × 10−14. NS = 0.86. f, MA plot depicting RNA-seq comparison of donor naive T cell-derived Maf∆Treg versus control FOXP3–YFP+ iTreg cells (mean of two biologically independent experiments). Blue dots indicate 190 upregulated and 75 downregulated genes in the c-MAF-dependent signature. Highlighted blue dots represent downregulated genes related to Treg cell function, and highlighted red dots indicate genes that are also dependent on RORγt11. Differentially expressed genes were calculated in DESeq2 using the Wald test with Benjamini–Hochberg correction to determine the false discovery rate (FDR) (FDR < 0.1 and log2 fold change > 1.5).

To investigate how c-MAF regulates the gut RORγt+ iTreg–TH17 cell axis, we co-transferred equal numbers of naive Maf+/+;Foxp3cre (control) and Maffl/fl;Foxp3cre HH7-2tg cells into H. hepaticus-colonized wild-type mice (Extended Data Fig. 7a, b). Two weeks after adoptive transfer, the Maffl/fl;Foxp3cre HH7-2tg cells were markedly underrepresented compared to control cells in the LILP, mesenteric lymph nodes (mLNs) and spleen, and were unable to form iTreg cells (Fig. 3c–e and Extended Data Fig. 7c, d). Importantly, at homeostasis, mutant donor-derived cells did not give rise to a high frequency of TH17 cells (Fig. 3e). Transcriptomics analysis revealed that the c-MAF-deficient iTreg cells were functionally impaired, as indicated by defective expression of Il10 and other Treg cell signature genes, as well as of RORγt-dependent genes11 (Fig. 3f, Extended Data Fig. 7e–h). Taken together, these findings show that c-MAF is a crucial cell-intrinsic factor for both the generation and function of microorganism-specific iTreg cells. Notably, the vast majority of accumulated TH17 cells in Maf∆Treg mice expressed c-MAF, indicating that the bulk of these cells did not arise from Treg cells in which c-MAF was deleted (Extended Data Fig. 7i). Thus, suppression of TH17 cell expansion is mediated by these iTreg cells in trans.

RORγt expression in iTreg cells has been implicated in the maintenance of gut immune homeostasis under different challenges11,12. However, spontaneous gut inflammation in Rorc∆Treg mice has not been described. We noticed that Maf∆Treg, but not Rorc∆Treg or control littermates, were prone to rectal prolapse (Fig. 4a). Maf∆Treg mice (8–12 weeks old) colonized with H. hepaticus for 5–6 weeks had enlarged large intestine-draining mLNs, and increased cellularity of mLNs and the large intestine (Fig. 4b, c). Histopathological analysis of the large intestine of these mice revealed mixed acute and chronic inflammation (Fig. 4d). Without H. hepaticus colonization, aged (6–12 months old) Maf∆Treg mice also exhibited mild spontaneous colitis (Fig. 4e). Notably, none of the above changes was observed in Rorc∆Treg mice (Fig. 4b–d). Thus, c-MAF but not RORγt expression in iTreg cells is crucial for the suppression of spontaneous inflammation. Indeed, the transcriptional profile of H. hepaticus-specific T effector (Teff) cells from Maf∆Treg mice with spontaneous colitis was highly similar to that of pathogenic TH17 cells in IL-10RA blockade-induced colitis, but differed markedly from homeostatic TH17 cells (which are predominantly SFB-specific) (Extended Data Fig. 8a–f).

Figure 4: RORγt+ iTreg cells are required to maintain gut homeostasis.
Figure 4

a, Frequency of rectal prolapse by genotype. b, Left, spleens and mLNs from Maf∆Treg and littermate controls. Right, total cell numbers in mLNs. Data summarize three independent experiments for Rorc∆Treg (n = 6) and littermate controls (n = 7), and four independent experiments for Maf∆Treg (n = 9) and littermate controls (n = 8). c, Number of leukocytes in the LILP. Data summarize three independent experiments for Rorc∆Treg (n = 7) and littermate controls (n = 8), and four independent experiments for Maf∆Treg and littermate controls (n = 9). d, Representative histology of large intestine (LI) sections (left) and colitis scores (right) of mice with indicated genotypes. Rorc∆Treg (n = 8) and littermate controls (n = 9). Maf∆Treg (n = 11) and littermate controls (n = 9). Scale bars, 50 µm. e, Colitis scores in aged H. hepaticus-negative Maf∆Treg (n = 6) and littermate control (n = 7) mice. f, g, Suppression of H. hepaticus-specific TCRtg cell-mediated transfer colitis by in vitro differentiated iTreg cells. Data summarize two independent experiments with indicated sample size (n) in total. Colitis scores for Rag1−/− mice that received the indicated TCRtg naive T cells and iTreg combinations. All statistics were calculated by unpaired two-sided Welch’s t-test. Error bars denote mean ± s.d. P values are indicated on the figure or as follows: f, i = 5.34 × 10−4, ii = 1.24 × 10−2, iii = 3.64 × 10−4, iv = 3.26 × 10−4 and v = 4.54 × 10−3. NS = 0.056. g, i = 0.013, ii = 2.50 × 10−3, iii = 4.59 × 10−4, iv = 0.024, and v = 0.016. NS = 0.12.

Similar to the Maf∆Treg strain, mice with inactivation of STAT3 in the Treg compartment or impaired TGFβ signalling in CD4+ T cells also lacked RORγt+ Treg cells and developed spontaneous colitis12,22,23 (Extended Data Fig. 9a, b). Consistent with these findings, c-MAF expression in Treg cells required a combination of both TGFβ and STAT3 signals in vitro and in vivo, as it does in other CD4+ T cells18,19 (Extended Data Fig. 9a–c). This suggests that c-MAF integrates anti-inflammatory TGFβ receptor signals with microorganism-induced cytokine-dependent STAT3 activation to mediate RORγt+ Treg induction.

Although c-MAF is also expressed, albeit at a lower level, in nTreg cells, c-MAF-deficient and -sufficient nTreg cells showed equivalent activity in inhibiting Teff cell proliferation in vitro, as well as in suppressing pathogenesis in a model of T cell transfer colitis in vivo (Extended Data Fig. 10a, b). We therefore wondered why, despite their increased numbers (Extended Data Fig. 6c), nTreg cells were not sufficient to establish gut homeostasis in Maf∆Treg mice. Adoptive transfer of 1,000 naive HH7-2tg or HH5-1tg cells into H. hepaticus-colonized Rag1−/− mice led to colitis. Taking advantage of this system, we compared the suppressive function of iTreg cells differentiated in vitro from naive HH7-2tg, HH5-1tg and polyclonal T cells. We found that epitope-specific iTreg were better at suppressing colitis, providing a potential explanation for why pathobiont-specific iTreg cells are required in addition to nTreg cells to maintain gut homeostasis24 (Fig. 4f, g).

Our results reveal a mechanism for how a healthy individual can host a ‘two-faced’ commensal pathobiont such as H. hepaticus without developing inflammatory disease. Our findings suggest that Treg cell induction serves as a strategy to establish commensalism, not only by helping the microorganisms to colonize their niche25, but also by protecting the host from inflammation. A similar requirement for iTreg cells has also been reported in the establishment of food tolerance26. Our observations in MafTreg mice are linked to and help explain the expansion of colitogenic TH17 cells in mice with Treg-specific inactivation of STAT323. Like c-MAF, STAT3 is probably required for the differentiation and/or function of microbiota-induced RORγt+ iTreg cells12. Moreover, microorganism-specific iTreg cells, compared with non-specific nTreg cells, can better suppress inflammatory Teff cells by recognizing the same epitopes. This result raises the prospect of harnessing the mechanisms of pathobiont-specific iTreg cell responses to re-establish homeostasis in patients with inflammatory bowel disease, for example, by engineering non-pathogenic Treg cell-inducing microorganisms27 to express pathobiont antigens.



Mice were bred and maintained in the animal facility of the Skirball Institute (New York University School of Medicine) and the National Institute of Allergy and Infectious Diseases (NIAID) in specific pathogen-free conditions. C57BL/6 mice were obtained from Jackson Laboratories or Taconic Farm. Il10−/− (B6.129P2-Il10tm1Cgn/J) mice were purchased from Jackson Laboratories and bred with wild-type C57BL/6 mice, which subsequently generated Il10+/− and Il10−/− littermates by heterozygous breeding. CD4-dnTGFbRII mice22 were purchased from Jackson Laboratories, and bred with wild-type C57BL/6 mice to generate CD4-dnTGFbRII and wild-type littermates. Cd4cre (Tg(Cd4-cre)1Cwi/BfluJ) and Cd45.1 (B6.SJL-Ptprca Pepcb/BoyJ) mice were purchased from Jackson Laboratories. Foxp3creYFP mice were previously described and obtained from Jackson Laboratories28. Il23rgfp and Maffl/fl strains were previously described29,30 and provided by M. Oukka and C. Birchmeier, respectively. Stat3fl/fl;Cd4cre mice were provided by D. E. Levy. Gata3fl/fl;Foxp3creYFP mice were bred at the NIAID. Littermates with matched sex (both males and females) were used. Except the aged mice (6–12 months old) analysed in the experiments of Fig. 4e, mice in all the experiments were 6–12 weeks old at the starting point of treatments. Animal sample size estimates were determined using power analysis (power = 90% and α = 0.05) based on the mean and standard deviation from our previous studies and/or pilot studies using 4–5 mice per group. All animal procedures were performed in accordance with protocols approved by the Institutional Animal Care and Usage Committee of New York University School of Medicine or the NIAID as applicable.

Antibodies, intracellular staining and flow cytometry

The following monoclonal antibodies were purchased from eBiosciences, BD Pharmingen or BioLegend: CD3 (145-2C11), CD4 (RM4-5), CD25 (PC61), CD44 (IM7), CD45.1 (A20), CD45.2 (104), CD62L (MEL-14), CXCR5 (L138D7), NPR-1 (3E12), ST2 (RMST2-2), TCRβ (H57-597), TCR Vβ6 (RR4-7), TCR Vβ8.1/8.2 (MR5-2), TCR Vβ14 (14-2), BCL6 (K112-91), c-MAF (T54-853), FOXP3 (FJK-16s), GATA3 (TWAJ), Helios (22F6), RORγt (B2D or Q31-378), T-bet (eBio4B10), IL-10 (JES5-16E3), IL-17A (eBio17B7) and IFNγ (XM61.2). 4′,6-diamidino-2-phenylindole (DAPI) or Live/dead fixable blue (ThermoFisher) was used to exclude dead cells.

For transcription factor staining, cells were stained for surface markers, followed by fixation and permeabilization before nuclear factor staining according to the manufacturer’s protocol (FOXP3 staining buffer set from eBioscience). For cytokine analysis, cells were incubated for 5 h in RPMI with 10% FBS, phorbol 12-myristate 13-acetate (PMA) (50 ng ml−1; Sigma), ionomycin (500 ng ml−1; Sigma) and GolgiStop (BD). Cells were stained for surface markers before fixation and permeabilization, and then subjected to intracellular cytokine staining according to the manufacturer’s protocol (Cytofix/Cytoperm buffer set from BD Biosciences).

Flow cytometric analysis was performed on an LSR II (BD Biosciences) or an Aria II (BD Biosciences) and analysed using FlowJo software (Tree Star).

Isolation of lymphocytes

Intestinal tissues were sequentially treated with PBS containing 1 mM DTT at room temperature for 10 min, and 5 mM EDTA at 37 °C for 20 min to remove epithelial cells, and then minced and dissociated in RPMI containing collagenase (1 mg ml−1 collagenase II; Roche), DNase I (100 μg ml−1; Sigma), dispase (0.05 U ml−1; Worthington) and 10% FBS with constant stirring at 37 °C for 45 min (small intestine) or 60 min (large intestine). Leukocytes were collected at the interface of a 40%/80% Percoll gradient (GE Healthcare). The Peyer’s patches and caecal patch were treated in a similar fashion except for the first step of removal of epithelial cells. Lymph nodes and spleens were mechanically disrupted.

Single-cell TCR cloning

Il23rGFP/+ mice were maintained in SFB-free conditions to guarantee low TH17 background levels. To induce a robust TH17 cell response, the mice were orally infected with H. hepaticus and injected intraperitoneally with 1 mg anti-IL-10RA (clone 1B1.3A, Bioxcell) every week from the day of infection. After two weeks, large intestine GFP+ CD4+ T cells were sorted on the BD Aria II and deposited at one cell per well into 96-well PCR plates pre-loaded with 5 μl high-capacity cDNA reverse transcription mix (Thermo Fisher) supplemented with 0.1% Triton X-100 (Sigma-Aldrich). Immediately after sorting, whole plates were incubated at 37 °C for 2 h, and then inactivated at 85 °C for 10 min for cDNA preparation. A nested multiplex PCR approach described previously was used to amplify the CDR3α and CDR3β TCR regions separately from the single cell cDNA31. PCR products were cleaned up with ExoSap-IT reagent (USB) and Sanger sequencing was performed by Macrogen. Open reading frame nucleotide sequences of the TCRα and TCRβ families were retrieved from the IMGT database (http://www.imgt.org)32.

Generation of TCR hybridomas

The NFAT-GFP 58αβ hybridoma cell line was provided by K. Murphy33. To reconstitute TCRs, cDNA of TCRα and TCRβ were synthesized as gBlocks fragments by Integrated DNA Technologies (IDT), linked with the self-cleavage sequence of 2A (TCRα-p2A-TCRβ), and shuttled into a modified MigR1 retrovector in which IRES-GFP was replaced with IRES-mCD4 (mouse CD4) as described previously8. Then retroviral vectors were transfected into Phoenix E packaging cells using TransIT-293 (Mirus). Hybridoma cells were transduced with viral supernatants in the presence of polybrene (8 μg ml−1) by spin infection for 90 min at 32 °C. Transduction efficiencies were monitored by checking mCD3 surface expression three days later.

Assay for hybridoma activation

Splenic dendritic cells were used as antigen-presenting cells (APCs). B6 mice were injected intraperitoneally with 5 × 106 FLT3L-expressing B16 melanoma cells to drive APC proliferation as previously described34. Splenocytes were prepared 10 days after injection, and positively enriched for CD11c+ cells using MACS LS columns (Miltenyi). 2 × 104 hybridoma cells were incubated with 105 APCs and antigens in round bottom 96-well plates for two days. GFP induction in the hybridomas was analysed by flow cytometry as an indicator of TCR activation.

Construction and screen of whole-genome shotgun library of H. hepaticus

The shotgun library was prepared with a procedure modified from previous studies7,8. In brief, genomic DNA was purified from cultured H. hepaticus with DNeasy PowerSoil kit (Qiagen). DNA was partially digested with MluCI (NEB), and the fraction between 500 and 2,000 base pairs (bp) was ligated into the EcoRI-linearized pGEX-6P-1 expression vector (GE Healthcare). Ligation products were transformed into ElectroMAX DH10B competent Cells (Invitrogen) by electroporation. To estimate the size of the library, we cultured 1% and 0.1% of transformed bacteria on lysogeny broth (LB) agar plates containing 100 μg ml−1 Ampicillin for 12 h and then quantified the number of colonies. The library is estimated to contain 3 × 104 clones. To ensure the quality of the library, we sequenced the inserts of randomly picked colonies. All the sequences were mapped to the H. hepaticus genome, and their sizes were 700 to 1,200 bp. We aliquoted the bacteria into 96-well deepwell plates (Axygen) (~30 clones per well) and grew with AirPort microporous cover (Qiagen) in 37 °C. The expression of exogenous proteins was induced by 1 mM isopropylthiogalactoside (IPTG, Sigma) for 4 h. Then bacteria were collected in PBS and heat-killed by incubating at 85 °C for 1 h, and stored at −20 °C until use. Two screening rounds were performed to identify the antigen-expressing clones. For the first round, pools of heat-killed bacterial clones were added to a co-culture of splenic APCs and hybridomas. Clones within the positive pools were subsequently screened individually against the hybridoma bait. Finally, the inserts of positive clones were subjected to Sanger sequencing. The sequences were blasted against the genome sequence of H. hepaticus (ATCC51449) and aligned to the annotated open reading frames. Full-length open reading frames containing the retrieved fragments were cloned into pGEX-6P-1 to confirm their activity in the T cell stimulation assay.

Epitope mapping

We cloned overlapping fragments spanning the entire HH_1713 coding region into the pGEX-6P-1 expression vector, and expressed these in Escherichia coli BL21 cells. The heat-killed bacteria were used to stimulate relevant hybridomas. This process was repeated until we mapped the epitope to a region containing 30 amino acids. The potential MHC class II epitopes were predicted with online software RANKPEP35. Overlapping peptides spanning the predicted region were further synthesized (Genescript) and verified by stimulation of the hybridomas.

Generation of TCRtg mice

TCR sequences of HH5-1 and HH7-2 were cloned into the pTα and pTβ vectors provided by D. Mathis36. TCR transgenic mice were generated by the Rodent Genetic Engineering Core at the New York University School of Medicine. Positive pups were genotyped by testing TCR Vβ8.1/8.2 (HH5-1tg) or Vβ6 (HH7-2tg) expression on T cells from peripheral blood.

MHC class II tetramer production and staining

HH-E2 tetramer was produced by the NIH Tetramer Core Facility37. In brief, QESPRIAAAYTIKGA (HH_1713-E2), an immunodominant epitope validated with the hybridoma stimulation assay, was covalently linked to I-Ab via a flexible linker, to produce pMHCII monomers. Soluble monomers were purified, biotinylated, and tetramerized with phycoerythrin- or allophycocyanin-labelled streptavidin. SFB-specific tetramer (3340-A6 tetramer) was described previously8. To stain endogenous T cells, mononuclear cells from SILP, LILP or caecal patch were first resuspended in MACS buffer with FcR block, 2% mouse serum and 2% rat serum. Then tetramer was added (10 nM) and incubated at room temperature for 60 min, and cells were re-suspended by pipetting every 20 min. Cells were washed with MACS buffer and followed by regular surface marker staining at 4 °C.

Adoptive transfer of TCRtg cells

Recipient mice were colonized with H. hepaticus and/or SFB by oral gavage seven days before adoptive transfer (The method for oral infection of SFB has been previously described8). Spleens from donor TCRtg mice were collected and mechanically disassociated. Red blood cells were lysed using ACK lysis buffer (Lonza). For TCRtg mice in wild-type background, naive Tg T cells were sorted as CD4+CD3+CD44loCD62LhiCD25 Vβ6+ (HH7-2tg), Vβ8.1/8.2+ (HH5-1tg) or Vβ14+ (7B8tg) on the Aria II (BD Biosciences). For HH7-2tg mice bred to the Foxp3creYFP background, naive transgenic T cells were sorted as CD4+CD3+CD44loCD62LhiFOXP3CreYFP-Vβ6+. Cells were resuspended in PBS on ice and transferred into congenic isotype-labelled recipient mice by retro-orbital injection. Cells from indicated tissues were analysed two weeks after transfer.

H. hepaticus culture and oral infection

H. hepaticus was provided by J. Fox (MIT). Frozen stock aliquots of H. hepaticus were stored in Brucella broth with 20% glycerol and frozen at −80 °C. The bacteria were grown on blood agar plates (TSA with 5% sheep blood, Thermo Fisher). Inoculated plates were placed into a hypoxia chamber (Billups-Rothenberg), and anaerobic gas mixture consisting of 80% nitrogen, 10% hydrogen, and 10% carbon dioxide (Airgas) was added to create a micro-aerobic atmosphere, in which the oxygen concentration was 3–5%. The micro-aerobic jars containing bacterial plates were left at 37 °C for 5 days before animal inoculation. For oral infection, H. hepaticus was resuspended in Brucella broth by application of a pre-moistened sterile cotton swab applicator tip to the colony surface. The concentration of bacterial inoculation dose was determined by the use of a spectrophotometric optical density (OD) analysis at 600 nm, and adjusted to OD600 nm readings between 1 and 1.5. 0.2 ml bacterial suspension was administered to each mouse by oral gavage. Mice were inoculated every 5 days for a total of two doses.

H. hepaticus-specific TCRtg cell-mediated transfer colitis

Naive T (Tnaive) cells were isolated from the spleens of HH7-2tg mice as CD4+CD3+CD44loCD62LhiCD25 Vβ6+ by FACS. The sorted cells (1 × 103) were administered by retro-orbital injection into H. hepaticus-colonized Rag1−/− mice. After two weeks, cells from the large intestine were isolated and analysed by flow cytometry.

C. rodentium-mediated colon inflammation

C. rodentium strain DBS100 (ATCC51459; American Type Culture Collection) was used for all inoculations. Bacteria were grown at 37 °C in LB broth to OD600 nm reading between 0.4 and 0.6. Mice were inoculated with 200 μl of a bacterial suspension (1 × 109–2 × 109 colony-forming units (CFU)) by way of oral gavage. After 15 days, cells from the large intestine were isolated, stained for HH-E2 tetramer and other markers as indicated and analysed by flow cytometry.

DSS-induced colitis

Mice were colonized with H. hepaticus 5 days before DSS treatment. To induce colitis, mice were given 2% DSS (50,000MW, Affymetrix/USB) in drinking water for 2 cycles, with each exposure for 7 days with 5 days of untreated water in between. Control mice were given drinking water for the same period. Cells from the large intestine were then isolated, stained for HH-E2 tetramer and other marks as indicated and analyzed by flow cytometry. Animal weights were monitored daily during the entire experiment.

T cell culture

Naive CD4+ T cells were purified from spleen and lymph nodes of mice with indicated genotypes. In brief, CD4+ T cells were positively selected from organ cell suspensions by magnetic-activated cell sorting using CD4 beads (MACS, Miltenyi) according to the product protocol, and then isolated as CD4+CD3+CD44loCD62LhiCD25 (polyclonal) or CD4+CD3+CD44loCD62LhiCD25 Vβ6+ (HH7-2tg) or Vβ8.1/8.2+ (HH5-1tg) by FACS. T cells were cultured at 37 °C in RPMI (Hyclone) supplemented with 10% heat-inactivated FBS (Hyclone), 50 U penicillin–streptomycin (Hyclone), 2 mM glutamine (Hyclone), 10 mM HEPES (Hyclone), 1 mM sodium pyruvate (Hyclone) and 50 μM β-mercaptoethanol (Gibco).

To generate iTreg cells for transfer colitis experiments (see below), wild-type, HH7-2tg or HH5-1tg cells were seeded at 1 × 106 cells in 1.5 ml per well in 12-well plates pre-coated with an anti-hamster IgG secondary antibody (MP Biomedicals), and cultured for 72 h. The culture was supplemented with soluble anti-CD3ε (0.25 μg ml−1, Bioxcell, clone 145-2C11) and anti-CD28 (1 μg ml−1, Bioxcell, clone 37.51) for TCR stimulation, and anti-IL-4 (1 μg ml−1, Bioxcell, clone 11B11), anti-IFNγ (1 μg ml−1, Bioxcell, clone XMG1.2), human TGFβ1 (20 ng ml−1, Peprotech), human IL-2 (500 U ml−1, Peprotech) and all-trans retinoic acid (100 nM, sigma) for optimal iTreg cell polarization. Aliquots of cultured cells were analysed for intracellular FOXP3 staining by flow cytometry. After they were confirmed to be more than 98% FOXP3+, the remaining live cells (DAPI negative) were FACS sorted for adoptive transfer.

To test the conditions inducing c-MAF expression, 200 μl naive T cells isolated from Maffl/fl, Maffl/fl;Cd4cre or Stat3fl/fl;Cd4cre mice were seeded at 1 × 105 cells per well in 96-well plates pre-coated with the anti-hamster IgG secondary antibody, and cultured for 48 h. The culture was supplemented with soluble anti-CD3ε (0.25 μg ml−1) and anti-CD28 (1 μg ml−1) for TCR stimulation. Combinations of the following antibodies or cytokines were added as indicated in Extended Data Fig. 9c: anti-TGFβ (1 μg ml−1, Bioxcell, 1D11.16.8), human TGFβ1 (0.3 or 20 ng ml−1, Peprotech), human IL-2 (500 U ml−1, Peprotech), mouse IL-6 (20 ng ml−1, Thermo), mouse IL-10 (100 ng ml−1, Peprotech), mouse IL-27 (25 ng ml−1, Thermo), mouse IL-12 (10 ng ml−1, Peprotech), mouse IL-1β (10 ng ml−1, Peprotech), mouse IL-4 (10 ng ml−1, R&D systems), mouse IFNγ (10 ng ml−1, Peprotech), and all-trans retinoic acid (100 nM, sigma).

Treg cell in vitro suppression assay

Tnaive cells with the phenotype CD4+CD3+CD44loCD62LhiCD25 were isolated from the spleen and lymph nodes of CD45.1 wild-type B6 mice by FACS and labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE). nTreg (CD45.2) cells with the phenotype CD4+CD3+FOXP3creYFP+NRP1+ were isolated from the spleen and lymph nodes of Foxp3creYFP or Maf∆Treg mice by FACS. B cells were isolated from the spleen and lymph nodes of CD45.2 wild-type B6 mice by positive enrichment for B220+ cells using MACS LS columns (Miltenyi). 2.5 × 104 CFSE-labelled Tnaive cells were cultured for 72 h with B cell APCs (5 × 104) and anti-CD3 (1 μg ml−1) in the presence or absence of various numbers of nTreg cells as indicated. The cell division index of responder T cells was assessed by dilution of CFSE using FlowJo software (Tree Star).

Suppression of adoptive transfer colitis with Treg cells

To compare the suppressive function of c-MAF-sufficient and -deficient nTreg cells, CD4+CD3+CD25CD45RBhi Teff cells were isolated by FACS from B6 mouse spleens and CD4+CD3+FOXP3-YFP+NPR1+ nTreg were isolated from spleen of H. hepaticus-colonized Foxp3creYFP or Maf∆Treg mice. Teff cells (5 × 105) were administered by retro-orbital injection into H. hepaticus-colonized Rag1−/− mice alone, or simultaneously with 4 × 105 nTreg cells as previously described38. Animal weights were measured weekly.

To compare the suppressive function of TCRtg and polyclonal Treg cells, Tnaive cells with the phenotype CD4+CD3+CD44loCD62LhiCD25 and Vβ6+ (HH7-2tg) or CD4+CD3+CD44loCD62LhiCD25 and Vβ8.1/8.2+ (HH5-1tg) were isolated from spleens of TCRtg mice. 1,000 naive HH7-2tg or HH5-1tg cells were co-transferred with different numbers (500,000, 150,000 or 50,000 as indicated in Fig. 4f, g) of in vitro polarized (see T cell culture, above) HH7-2tg, HH5-1tg or polyclonal iTreg cells into H. hepaticus-colonized Rag1−/− mice by retro-orbital injection.

Co-housed littermate recipients were randomly assigned to different treatment groups such that each cage contained all treatment conditions. After four to five weeks (for nTreg cell comparisons) or eight weeks (for the transgenic T cell comparisons), large intestines were collected and fixed with 10% neutral buffered formalin (Fisher). Samples were sectioned and stained with haematoxylin and eosin (H&E) by the Histopathology Core at the New York University School of Medicine.

Histology analysis

The H&E slides from each sample were examined in a blinded fashion. Samples of proximal, mid and distal colon were graded semi-quantitatively from 0 to 4 as described previously39. Scores from proximal, mid and distal sites were averaged to obtain inflammation scores for the entire colon.

Cell isolation for RNA-seq experiment

A T cell reconstitution system was designed to purify c-MAF-sufficient or -deficient iTreg cells from compatible microenvironments. In brief, Tnaive cells were isolated from the spleen of CD45.2 Foxp3creYFP or Maf∆Treg mice as CD4+CD3+CD44loCD62LhiFOXP3-YFP- and Treg cells were isolated from the spleens of CD45.1 wild-type B6 mice as CD4+CD3+CD25+ by FACS. Tnaive cells (2 × 105) and Treg cells (8 × 105) were simultaneously administered by retro-orbital injection into H. hepaticus-colonized Rag1−/− mice. Two weeks after transfer, c-MAF-sufficient or -deficient iTreg cells were purified from the large intestine of reconstituted mice as CD4+CD3+CD45.1CD45.2+FOXP3-YFP+ by FACS and collected into FBS. 20% of the sorted cells were stained for RORγt and FOXP3 and the remaining cells were saved in TRIzol (Invitrogen) for RNA extraction.

To isolate H. hepaticus-specific colitogenic Teff cells, HH7-2tg;Maf∆Treg and HH7-2tg;Foxp3cre mice were colonized with H. hepaticus. HH7-2tg;Foxp3cre mice were further injected intraperitoneally with 1 mg anti-IL-10RA antibody (clone 1B1.3A, Bioxcell) weekly from the day of colonization. HH7-2 Teff cells (CD3+CD4+TCRVβ6+FOXP3YFP) were sorted from the LILP two weeks after colonization. Homeostatic IL-23R-GFP+ T cells (CD3+CD4+IL-23R-GFP+) were sorted from both SILP and LILP of Il23rGFP/+ mice stably colonized with SFB.

RNA-seq library preparation

Total RNA was extracted using TRIzol (Invitrogen) followed by DNase I treatment and cleanup with RNeasy MinElute kit (Qiagen). Treg cell RNA-seq libraries were prepared with the SMART-Seq v4 Ultra Low Input RNA Kit (Clontech 634899 and 634888). TH17 cell RNA libraries were prepared using the Nugen Ovation Ultralow Library Systems V2 (7102 and 0344). All sequencing was performed using the Illumina NextSeq. RNA-seq libraries were prepared and sequenced by the Genome Technology Core at New York University School of Medicine.

Data processing of RNA-seq

RNA-seq reads were mapped to the Mus musculus genome Ensembl annotation release 87 with STAR (v2.5.2b)40. Uniquely mapped reads were counted using featureCounts41 with parameters: -p -Q 20. DESeq242 was used to identify differentially expressed genes across conditions with experimental design: ~condition + gender. Read counts were normalized and transformed by functions varianceStabilizingTransformation and rlog in DESeq2 with the following parameter: blind=FALSE. Gender differences were considered as batch effect, and were corrected by ComBat43. Downstream analysis and data visualization were performed in R44.

Statistical analysis

For animal studies, mutant and control groups did not always have similar standard deviations and therefore an unpaired two-sided Welch’s t-test was used. Error bars represent ± s.d. Animal sample size estimates were determined using power analysis (power = 90% and α = 0.05) based on the mean and s.d. from our previous studies and/or pilot studies using 4–5 mice. No samples were excluded from analysis. For RNA-seq analysis, differentially expressed genes were calculated in DESeq2 using the Wald test with Benjamini–Hochberg correction to determine the FDR. Genes were considered differentially expressed with FDR < 0.1 and log2 fold change > 1.5. Enriched disease pathways in pathogenic HH7-2 TH17 were determined using Ingenuity Pathway Analysis (https://www.ingenuity.com). Gene set enrichment analysis (GSEA, http://www.broad.mit.edu/gsea/) on Maf-deficient or -sufficient iTreg cells was performed using a gene set of 33 RORγt-dependent genes in NRP1 colonic Treg cells (Rorc, Ccr6, Idua, Il1rn, C2cd4b, Nxt1, Tmem176b, Cxcr3, Tnfrsf1a, Adamts7, Pik3ip1, Rrad, Crmp1, Irak3, Fam129b, Ppcs, Tbxa2r, Avpi1, Serpinb1a, Alkbh7, Nckipsd, Havcr2, Il23r, Txnip, Igj (also known as Jchain), Trim16, Pigp, Rras, Samd10, Il1r2, F2rl1, Maff and Ly6c1)11.

Data availability

cDNA sequences of H. hepaticus-specific TCRs yielding data that support the findings of this study have been deposited in GenBank with the accession codes KY964547KY964570. RNA-seq data have been deposited in the Sequence Read Archive (SRA) with accession code SRP126932, and in the Gene Expression Omnibus (GEO) with accession code GSE108184. All other data are available from the corresponding author upon reasonable request.

Change history

  • 05 February 2019

    The 'Competing interests' statement of this Letter should state 'D.R.L. consults for and has equity in Vedanta Biosciences'.


Primary accessions

NCBI Reference Sequence

Sequence Read Archive


  1. 1.

    & The gut microbiota shapes intestinal immune responses during health and disease. Nat. Rev. Immunol. 9, 313–323 (2009)

  2. 2.

    , & Interactions between the microbiota and the immune system. Science 336, 1268–1273 (2012)

  3. 3.

    , , & Role of the gut microbiota in immunity and inflammatory disease. Nat. Rev. Immunol. 13, 321–335 (2013)

  4. 4.

    , & Pathobionts of the gastrointestinal microbiota and inflammatory disease. Curr. Opin. Immunol. 23, 473–480 (2011)

  5. 5.

    et al. IL-23 plays a key role in Helicobacter hepaticus-induced T cell-dependent colitis. J. Exp. Med. 203, 2485–2494 (2006)

  6. 6.

    et al. Interleukin-23 drives innate and T cell-mediated intestinal inflammation. J. Exp. Med. 203, 2473–2483 (2006)

  7. 7.

    , & Identification of a CD4+ T cell-stimulating antigen of pathogenic bacteria by expression cloning. J. Exp. Med. 182, 1751–1757 (1995)

  8. 8.

    et al. Focused specificity of intestinal TH17 cells towards commensal bacterial antigens. Nature 510, 152–156 (2014)

  9. 9.

    , , & Visualization of peptide-specific T cell immunity and peripheral tolerance induction in vivo. Immunity 1, 327–339 (1994)

  10. 10.

    et al. Naive CD4+ T cell frequency varies for different epitopes and predicts repertoire diversity and response magnitude. Immunity 27, 203–213 (2007)

  11. 11.

    et al. Individual intestinal symbionts induce a distinct population of RORγ+ regulatory T cells. Science 349, 993–997 (2015)

  12. 12.

    et al. The microbiota regulates type 2 immunity through RORγt+ T cells. Science 349, 989–993 (2015)

  13. 13.

    et al. The alarmin IL-33 promotes regulatory T-cell function in the intestine. Nature 513, 564–568 (2014)

  14. 14.

    et al. Fate mapping of IL-17-producing T cells in inflammatory responses. Nat. Immunol. 12, 255–263 (2011)

  15. 15.

    et al. Acute gastrointestinal infection induces long-lived microbiota-specific T cell responses. Science 337, 1553–1556 (2012)

  16. 16.

    et al. Helicobacter species are potent drivers of colonic T cell responses in homeostasis and inflammation. Sci. Immunol. 2, eaal5068 (2017)

  17. 17.

    et al. Foxp3+ T cells expressing RORγt represent a stable regulatory T-cell effector lineage with enhanced suppressive capacity during intestinal inflammation. Mucosal Immunol. 9, 444–457 (2016)

  18. 18.

    et al. A validated regulatory network for Th17 cell specification. Cell 151, 289–303 (2012)

  19. 19.

    et al. The aryl hydrocarbon receptor interacts with c-Maf to promote the differentiation of type 1 regulatory T cells induced by IL-27. Nat. Immunol. 11, 854–861 (2010)

  20. 20.

    et al. MyD88 signalling in colonic mononuclear phagocytes drives colitis in IL-10-deficient mice. Nat. Commun. 3, 1120 (2012)

  21. 21.

    et al. Specific microbiota direct the differentiation of IL-17-producing T-helper cells in the mucosa of the small intestine. Cell Host Microbe 4, 337–349 (2008)

  22. 22.

    & Abrogation of TGFβ signaling in T cells leads to spontaneous T cell differentiation and autoimmune disease. Immunity 12, 171–181 (2000)

  23. 23.

    et al. CD4+ regulatory T cells control TH17 responses in a Stat3-dependent manner. Science 326, 986–991 (2009)

  24. 24.

    et al. A requisite role for induced regulatory T cells in tolerance based on expanding antigen receptor diversity. Immunity 35, 109–122 (2011)

  25. 25.

    et al. The Toll-like receptor 2 pathway establishes colonization by a commensal of the human microbiota. Science 332, 974–977 (2011)

  26. 26.

    et al. Dietary antigens limit mucosal immunity by inducing regulatory T cells in the small intestine. Science 351, 858–863 (2016)

  27. 27.

    et al. Treg induction by a rationally selected mixture of Clostridia strains from the human microbiota. Nature 500, 232–236 (2013)

  28. 28.

    et al. Regulatory T cell-derived interleukin-10 limits inflammation at environmental interfaces. Immunity 28, 546–558 (2008)

  29. 29.

    et al. The transcription factor c-Maf controls touch receptor development and function. Science 335, 1373–1376 (2012)

  30. 30.

    et al. Cutting edge: IL-23 receptor GFP reporter mice reveal distinct populations of IL-17-producing cells. J. Immunol. 182, 5904–5908 (2009)

  31. 31.

    et al. Paired analysis of TCRα and TCRβ chains at the single-cell level in mice. J. Clin. Invest. 121, 288–295 (2011)

  32. 32.

    et al. IMGT, the international ImMunoGeneTics information system. Nucleic Acids Res. 37, D1006–D1012 (2009)

  33. 33.

    et al. CTLA-4 suppresses the pathogenicity of self antigen-specific T cells by cell-intrinsic and cell-extrinsic mechanisms. Nat. Immunol. 11, 129–135 (2010)

  34. 34.

    et al. Differences in dendritic cells stimulated in vivo by tumors engineered to secrete granulocyte-macrophage colony-stimulating factor or Flt3-ligand. Cancer Res. 60, 3239–3246 (2000)

  35. 35.

    & Prediction of peptide–MHC binding using profiles. Methods Mol. Biol. 409, 185–200 (2007)

  36. 36.

    , , & Cassette vectors directing expression of T cell receptor genes in transgenic mice. J. Immunol. Methods 180, 273–280 (1995)

  37. 37.

    et al. Phenotypic analysis of antigen-specific T lymphocytes. Science 274, 94–96 (1996)

  38. 38.

    et al. T cell transfer model of chronic colitis: concepts, considerations, and tricks of the trade. Am. J. Physiol. Gastrointest. Liver Physiol. 296, G135–G146 (2009)

  39. 39.

    , & Cytotoxic T lymphocyte-associated antigen 4 plays an essential role in the function of CD25+CD4+ regulatory cells that control intestinal inflammation. J. Exp. Med. 192, 295–302 (2000)

  40. 40.

    . et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013)

  41. 41.

    , & featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014)

  42. 42.

    , & Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014)

  43. 43.

    , & Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8, 118–127 (2007)

  44. 44.

    R Development Core Team. R: A Language And Environment For Statistical Computing ; (R Foundation for Statistical Computing, Vienna, Austria, 2016)

Download references


We thank S. Y. Kim and the NYU Rodent Genetic Engineering Laboratory (RGEL) for generating TCR transgenic mice, A. Heguy and colleagues at the NYU School of Medicine’s Genome Technology Center (GTC) for preparation of RNA-seq libraries and RNA sequencing, the NIH Tetramer Core Facility for generating MHC class II tetramers, K. Murphy for providing the 58αβ hybridoma line, D. E. Levy for providing the Stat3fl/fl;Cd4cre mice, J. Fox for providing the H. hepaticus strain, P. Dash and P. G. Thomas for advice on single-cell TCR cloning, and J. A. Hall, J. Muller and J. Lafaille for suggestions on the manuscript. The Experimental Pathology Research Laboratory of NYU Medical Center is supported by National Institutes of Health Shared Instrumentation grants S10OD010584-01A1 and S10OD018338-01. The GTC is partially supported by the Cancer Center Support grant P30CA016087 at the Laura and Isaac Perlmutter Cancer Center. This work was supported by the Irvington Institute fellowship program of the Cancer Research Institute (M.X.); the training program in Immunology and Inflammation 5T32AI100853 (M.P.); the Helen and Martin Kimmel Center for Biology and Medicine (D.R.L.); the Colton Center for Autoimmunity (D.R.L.); and National Institutes of Health grant R01DK103358 (R.B. and D.R.L.). D.R.L. is an Investigator of the Howard Hughes Medical Institute.

Author information

Author notes

    • Mo Xu
    •  & Maria Pokrovskii

    These authors contributed equally to this work.


  1. Molecular Pathogenesis Program, The Kimmel Center for Biology and Medicine of the Skirball Institute, New York University School of Medicine, New York, New York 10016, USA.

    • Mo Xu
    • , Maria Pokrovskii
    • , Christy Au
    • , Carolina Galan
    •  & Dan R. Littman
  2. Department of Pathology and Laboratory Medicine, University of Rochester Medical Center, Rochester, New York 14642, USA.

    • Yi Ding
  3. Courant Institute of Mathematical Sciences, Computer Science Department, New York University, New York, New York 10003, USA.

    • Ren Yi
    •  & Richard Bonneau
  4. The Howard Hughes Medical Institute, New York University School of Medicine, New York, New York 10016, USA.

    • Christy Au
    •  & Dan R. Littman
  5. Mucosal Immunology Section, Laboratory of Parasitic Diseases, National Institute of Allergy and Infectious Diseases, NIH, Bethesda, Maryland 20892, USA.

    • Oliver J. Harrison
    •  & Yasmine Belkaid
  6. NIAID Microbiome Program, NIH, Bethesda, Maryland 20892, USA.

    • Yasmine Belkaid
  7. Center for Genomics and Systems Biology, Department of Biology, New York University, New York, New York 10003, USA.

    • Richard Bonneau
  8. Center for Computational Biology, Flatiron Institute, Simons Foundation, New York, New York 10010, USA.

    • Richard Bonneau


  1. Search for Mo Xu in:

  2. Search for Maria Pokrovskii in:

  3. Search for Yi Ding in:

  4. Search for Ren Yi in:

  5. Search for Christy Au in:

  6. Search for Oliver J. Harrison in:

  7. Search for Carolina Galan in:

  8. Search for Yasmine Belkaid in:

  9. Search for Richard Bonneau in:

  10. Search for Dan R. Littman in:


M.X. and M.P. designed and performed all experiments and analysed the data. Y.D. performed blinded histology scoring on colitis sections. C.A. and C.G. assisted with in vivo and in vitro experiments. R.Y. and M.P. performed RNA-seq analysis. O.J.H. and Y.B. analysed the Gata3Treg mouse phenotype. R.B. supervised RNA-seq analysis. M.X., M.P. and D.R.L. wrote the manuscript with input from the co-authors. D.R.L. supervised the research and contributed to experimental design.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Dan R. Littman.

Reviewer Information Nature thanks C. Ohnmacht and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Supplementary information

About this article

Publication history






Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.