Report

Lactobacillus reuteri induces gut intraepithelial CD4+CD8αα+ T cells

See allHide authors and affiliations

Science  25 Aug 2017:
Vol. 357, Issue 6353, pp. 806-810
DOI: 10.1126/science.aah5825

Tolerogenic T cells need probiotics

CD4+CD8αα+ double-positive intraepithelial lymphocytes (DP IELs) are a recently discovered class of intestinal T cells believed to take part in a variety of immune responses, including oral tolerance. These cells are absent in germ-free mice, but the mechanisms driving their development are unclear. Cervantes-Barragan et al. found that a particular species of probiotic bacteria, Lactobacillus reuteri, induces DP IELs. This does not occur by stimulating the immune system directly. Instead, L. reuteri generates a specific derivative of dietary tryptophan that promotes differentiation of DP IEL precursors. These findings underscore the delicate interplay between benign bacteria, diet, and gut health.

Science, this issue p. 806

Abstract

The small intestine contains CD4+CD8αα+ double-positive intraepithelial lymphocytes (DP IELs), which originate from intestinal CD4+ T cells through down-regulation of the transcription factor Thpok and have regulatory functions. DP IELs are absent in germ-free mice, which suggests that their differentiation depends on microbial factors. We found that DP IEL numbers in mice varied in different vivaria, correlating with the presence of Lactobacillus reuteri. This species induced DP IELs in germ-free mice and conventionally-raised mice lacking these cells. L. reuteri did not shape the DP-IEL-TCR (TCR, T cell receptor) repertoire but generated indole derivatives of tryptophan that activated the aryl-hydrocarbon receptor in CD4+ T cells, allowing Thpok down-regulation and differentiation into DP IELs. Thus, L. reuteri, together with a tryptophan-rich diet, can reprogram intraepithelial CD4+ T cells into immunoregulatory T cells.

The gut microbiota drives maturation and function of the immune system (13). Several bacterial taxa shape the differentiation of naïve T cells: Segmented filamentous bacteria (SFB) bias CD4+ T cells toward a T helper 17 (TH17) cell fate (4), Clostridia clusters IV and XIVa promote CD4+ regulatory T cell (Treg) differentiation (5, 6), and Bacteroides fragilis (7, 8) and Faecalibacterium prausnitzii (9) induce CD4+ T cells that secrete interleukin-10 (IL-10). The small intestinal epithelium contains a distinctive population of CD4+CD8αα+ T cell receptor (TCR) αβ T cells [referred to as double-positive intraepithelial lymphocytes (DP IELs)] (10). These cells have a regulatory function complementary to that of Tregs and promote tolerance to dietary antigens (11). DP IELs originate from lamina propria CD4+ T cells that include but are not limited to Tregs (11). Upon reaching the epithelium, these cells reactivate the CD8+ T cell lineage program through down-regulation of Thpok and up-regulation of Runx3 and T-bet (1214). This process is facilitated in part by transforming growth factor–β (TGFβ), retinoic acid, interferon-γ (IFN-γ), and IL-27 (12, 14). The lack of DP IELs in germ-free (GF) mice (11, 13) indicates that the intestinal microbiota is required for their differentiation, although the bacterial taxa and metabolites required are not known.

We observed that mice born at one of our vivaria [specialized research facility (SRF)] harbored substantial numbers of DP IELs (DP+), whereas DP IELs were negligible or absent (DP) in mice from another vivarium [clinical sciences research building (CSRB)] (Fig. 1A). Moreover, DP IELs in SRF mice expressed low levels of the transcription factor Thpok (Thpoklo), whereas CSRB mice lacked CD4+ Thpoklo IELs (13), consistent with the observed lack of DP IELs (fig. S1). Thus, the microbiota in SRF mice may contain taxa capable of inducing DP IELs. Consistent with this, DP IELs appeared in CSRB mice after gavage with fecal or ileal microbiota from SRF but not CSRB animals (Fig. 1B).

Fig. 1 Specific microbiota components induce double-positive intraepithelial lymphocytes (DP IELs).

(A) Representative plots and frequencies of DP IELs (gated on CD45+CD3+TCRγδCD8βCD4+ IELs) from C57BL/6 mice born in CSRB and SRF facilities. (B) DP IEL frequencies in CSRB mice 4 weeks after oral gavage with ileal or fecal microbiota harvested from the indicated mice. (C) DP IEL frequencies in C57BL/6 mice from Jackson (JAX), Taconic, and Charles River (CR) Laboratories. (D) DP IEL frequencies in F0 and F1 generations of JAX and CR mice bred in CSRB. (E) DP IEL frequencies in JAX or CR mice housed separately (JAX, CR), or cohoused (JAX co, CR co). (F) DP IEL frequencies in JAX mice treated with ileal or fecal microbiota from the indicated mice. (G and H) Frequencies of DP IELs in CR mice that were either untreated (Ctrl) or treated with (G) vancomycin, neomycin, ampicillin, and metronidazole (VNAM) or (H) ampicillin plus vancomycin (Amp/Van), neomycin (Neo), or metronidazole (Met). Symbols represent individual mice. Data are pooled from two or three independent experiments. Statistical analysis was performed using a Mann-Whitney U test between groups or a Kruskal-Wallis test for multiple comparison analysis. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Bars represent means.

Most SRF (DP+) mice are embryo-rederived using female recipient mice from Charles River (CR) Laboratories and thus acquire a CR microbiota. In contrast, most CSRB (DP) mice originate from the Jackson Laboratories (JAX). Thus, taxa present in CR but not JAX mice may induce DP IELs. C57BL/6 mice purchased from CR had a DP+ phenotype, whereas C57BL/6 mice purchased from JAX had a DP phenotype (Fig. 1C). DP+ and DP phenotypes were vertically transmissible (Fig. 1D). Moreover, the DP+ phenotype was laterally transferable from CR to JAX mice by cohousing animals (Fig. 1E) or by colonizing JAX mice with CR ileal or fecal microbiota (Fig. 1F). Treatment of CR mice with vancomycin, neomycin, ampicillin, and metronidazole (VNAM) abrogated DP IELs (Fig. 1G), as did ampicillin and vancomycin alone, which target Gram-positive bacteria. Treatment with neomycin, which targets Gram-negative bacteria, led to a slight increase in DP IELs (Fig. 1H); metronidazole did not affect DP IELs. Thus, DP IELs are induced by neomycin-resistant Gram-positive bacterial taxa.

To define bacterial taxa that induce DP IELs, we sequenced polymerase chain reaction amplicons from 16S ribosomal RNA (rRNA) genes present in the ileum of CR and JAX mice, as well as neomycin-treated and untreated CR animals. Six operational taxonomic units (OTUs) were selectively present in the CR microbiota and were also present or enriched after neomycin treatment (fig. S2A). These OTUs included L. reuteri and five members of the Bacteroidales S24-7 lineage (Fig. 2A, fig. S2A, and table S1). L. reuteri was present in the ileum of CR control mice and CR mice treated with neomycin or metronidazole but absent in CR mice treated with ampicillin plus vancomycin (fig. S2B).

Fig. 2 L. reuteri induces DP IELs.

(A) Relative abundance of L. reuteri operational taxonomic unit ID 411486, as determined by sequencing of the V4 region of 16S rRNA genes present in the ileal microbiota of 8-week-old JAX mice and CR mice (n = 4 mice) and 4-week-neomycin-treated (Neo) or untreated (Ctrl) CR mice (n = 5 mice per treatment group). Statistical analysis was performed using a Mann-Whitney U test. *P < 0.05. (B to D) DP IEL frequencies in JAX mice colonized with L. reuteri WU, L. reuteri strain 100-23, L. johnsonii WU, L. murinus, or a mixture of B. vulgatus, B. uniformis, and B. acidifaciens. Untreated JAX mice were used as controls. (E) DP IEL frequencies in specific-pathogen–free (SPF) and germ-free (GF) C57BL/6 mice. Duo, duodenum; je, jejunum; ile, ileum. (F) Total number of DP IELs in GF mice colonized with JAX ileal microbiota, JAX ileal microbiota combined with L. reuteri WU, L. reuteri WU alone, or CR ileal microbiota. Symbols represent individual mice. Data are pooled from one to three independent experiments. Statistical analysis was performed using a Mann-Whitney U test between groups or a Kruskal-Wallis test for multiple comparison analysis. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. NS, not significant. Bars represent means; error bars indicate SEM.

Two L. reuteri strains (WU and 100-23) induced DP IELs after colonization of JAX mice (Fig. 2B), whereas L. johnsonii, L. murinus (Fig. 2C), and various Bacteroides species (Fig. 2D) did not. As expected, DP IELs were absent throughout the small intestines of GF animals (11, 13) (Fig. 2E). Concurrent inoculation of GF mice with L. reuteri (WU strain) and ileal microbiota from JAX mice induced DP IELs, whereas JAX ileal microbiota alone was ineffective (Fig. 2F). Monocolonization with L. reuteri resulted in a slight increase of DP IELs. As expected, the transfer of CR intestinal microbiota induced DP IELs. Thus, L. reuteri induces DP IELs, but the presence of other microbes enhances this effect, perhaps by expanding the CD4+ T cells in the lamina propria and epithelium.

We next examined how L. reuteri induces DP IELs. If L. reuteri antigens shape the specificity of DP IELs, these cells would express quasi-clonal TCR αβ repertoires, as shown for SFB-specific TH17 cells (15). Analysis of TCR Vβ expression on DP IELs, CD4+ IELs, and mesenteric lymph node (MLN) naïve CD4+ T cells showed that TCR Vβ usage of DP IELs resembled that of CD4+ IELs but was different from that of naïve CD4+ T cells (Fig. 3A). In each mouse analyzed, DP IELs exhibited a preferential Vβ usage (more than twofold compared with MLN), although no enrichment for a specific Vβ pattern was shared by all mice, suggesting mouse-to-mouse variability in DP IEL TCRs (Fig. 3, A and B, and fig. S3A). Because the great diversity of polyclonal T cells complicates analysis at the individual TCR level, we analyzed the TCR α repertoires of DP IELs, CD4+ IELs, CD8+ IELs, and MLN naïve CD4+ T cells from mice that express a fixed transgenic TCR αβ chain (Tcli+ mice) (16, 17). Tcli+ T cells were very similar in phenotype and frequency to wild-type (WT) T cells (fig. S3B). TCR Renyi diversity profiles showed both decreased diversity (lower entropy at each order) and evidence for increased clonal expansion (greater downward slope) in the repertoires of all three IEL subsets compared with the repertoire of naïve CD4+ T cells (fig. S3C). The TCR α repertoires of CD4+ IELs, CD8+ IELs, and DP IELs were distinct from those of naïve CD4+ T cells (Fig. 3C), consistent with the expansion of T cell clones in response to specific antigens. Moreover, the TCR α repertoires of DP IELs and CD4+ IELs were similar but distinct from those of CD8+ IELs (Fig. 3, D and E), corroborating that DP IELs differentiate from CD4+ IELs. Although the TCR α repertoires of naïve T cells were similar in all mice analyzed, the TCR α repertoires of DP IELs and CD4+ IELs were quite distinct in each mouse (Fig. 3F), suggesting that the specificity of CD4+ IELs and DP IELs is shaped by various environmental and/or self-antigens in each mouse, rather than by a single antigen.

Fig. 3 DP IELs have diverse TCR repertoires in different mice.

(A) Frequencies of indicated TCR Vβ in DP IELs, CD4+ IELs, and mesenteric lymph node (MLN) naïve CD4+ T cells from 10 individual DP+ C57BL/6 mice. (B) For each TCR Vβ, the ratio between its frequency in DP IELs and its frequency in MLN naïve CD4+ T cells was calculated for individual mice; values are displayed as a heat map. Each column represents a single mouse. (C and D) Morisita-Horn similarity analysis of the TCR α-chain repertoires of intraepithelial CD4+ IELs, CD8+ IELs, DP IELs, and MLN naïve CD4+ T cells. Each symbol represents a comparison between the indicated two subsets within a mouse. Index values of 0 indicate that the two samples are completely dissimilar, whereas index values of 1 indicate that they are indistinguishable. (E) Venn diagram showing the number of distinct and overlapping complementarity-determining region 3 (CDR3) sequences found in TCR α chains of CD4+ IELs, CD8+ IELs, and DP IELs. (F) Morisita-Horn similarity analysis of the TCR α-chain CDR3 repertoires of CD4+ IELs, CD8+ IELs, and DP IELs. Each symbol represents a comparison of the TCR α sequences of the same T cell subset in different mice. Lines indicate mean values. Populations from four individual animals were analyzed.

Some bacterial taxa shape T cell differentiation via their physical components [e.g., polysaccharide A (7, 18)] or their metabolic products [e.g., short-chain fatty acids (1921)]. L. reuteri releases reuterin (22) and histamine (23) and also has the ability to metabolize tryptophan (l-Trp) to indole derivatives, some of which activate the aryl-hydrocarbon receptor (AhR) in group 3 innate lymphoid cells (ILC3s) (24, 25). Accordingly, culture supernatants of L. reuteri grown in l-Trp–containing medium activated an AhR reporter cell line, whereas supernatants of L. johnsonii or L. murinus did not (Fig. 4A). To determine whether L. reuteri induces DP IELs through the release of AhR ligands, we added L. reuteri culture supernatants to naïve ovalbumin-specific CD4+ T cells (OTII) stimulated with ovalbumin peptide-pulsed dendritic cells (DCs) with or without TGFβ (12). L. reuteri culture supernatant induced CD8αα+CD4+ T cell differentiation of OTII T cells, as did the AhR agonist 2,3,7,8-tetrachlorodibenzodioxin (TCDD) (Fig. 4B and fig. S4A), but only in the presence of TGFβ. The AhR antagonist CH223191 inhibited CD8αα+CD4+ T cell differentiation, confirming that the L. reuteri supernatant acts through AhR. AhR agonists in combination with TGFβ were as effective as TGFβ plus retinoic acid and IFN-γ (12, 14) (Fig. 4B). We also generated a strain of L. reuteri lacking a functional aromatic aminotransferase and, hence, the ability to convert amino acids into indolic AhR ligands (L. reuteri ΔArAT) (24, 26). Culture supernatants from L. reuteri ΔArAT failed to stimulate the AhR reporter cells (Fig. 4C). Moreover, L. reuteri ΔArAT did not induce DP IELs in JAX mice (Fig. 4D), although this strain colonized mice efficiently (fig. S4B).

Fig. 4 L. reuteri induces DP IELs through AhR activation in T cells.

(A) Representative plots and quantification of green fluorescent protein–positive (GFP+) cells in an AhR reporter cell line after stimulation with medium (minimum essential medium +10% bovine calf serum), peptone-tryptone water + l-Trp (PT-T), L. reuteri WU, L. reuteri (L.r.) 100-23, L. johnsonii (L.j.), or L. murinus (L.m.) supernatants (grown in PT-T) or TCDD (2,3,7,8-tetrachlorodibenzodioxin). FSC, forward scatter. (B) CD4+CD8βCD8α+ T cell frequencies after culture of naïve OTII CD4+ T cells with spleen CD11c+ dendritic cells (DCs), OVA329-337 peptide, and the indicated stimuli. T, TGFβ; L.r., L. reuteri WU supernatant; CH, CH223191; R, retinoic acid. (C) Quantification of GFP+ cells in an AhR reporter cell line after stimulation with medium, PT-T, L. reuteri 100-23 supernatant, L. reuteri ΔArAT supernatant, or TCDD. (D) DP IEL frequencies in JAX C57BL/6 mice 4 weeks after colonization with either L. reuteri 100-23 or L. reuteri ΔArAT, or in mice that were not colonized (ctrl). (E) DP IEL frequencies in CR C57BL/6 mice fed for 4 weeks with diets low (0.11%), standard (0.24%), or high (0.48%) in l-Trp. (F) CD4+CD8αα+ T cell frequencies after culture of naïve OTII CD4+ T cells with spleen CD11c+ DCs, OVA329-337 peptide, and the indicated stimuli. IAId, indole-3-aldehyde; ILA, indole-3-lactic acid. (G) DP IEL frequencies in cohoused WT, Ahr+/–, and Ahr–/– littermate mice. (H) CD4+CD8αα+ T cell frequencies after culture of WT or Ahr–/– CD4+ T cells with WT or Ahr–/– DCs; a combination of SEB, SEE, and TSST1 superantigens; and the indicated stimuli. (I) DP IEL frequencies in Rorc-Cre+ × Ahrfl/fl and Rorc-Cre × Ahrfl/fl littermates. (J) Representative flow cytometry plots showing DP IELs and CD4+ IELs (left) and Thpok expression (right) in CD4+ IELs (blue) and DP IELs (green) in Ahr+/– and Ahr–/– littermates. Symbols represent single animals; data have been pooled from two or three independent experiments. Statistical analysis was performed using a Mann-Whitney U test (I), a Kruskal-Wallis test [(D), (E), and (G)], and one-way analysis of variance (ANOVA) with Tukey’s post hoc test [(A), (B), (C), (F), and (H)]. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Bars represent means; error bars indicate SEM.

We next assessed the impact of dietary l-Trp levels in DP IEL development. More DP IELs were observed in CR mice fed a diet high in l-Trp (0.48%) compared with mice fed standard (0.24%) or low (0.11%) l-Trp diets (Fig. 4E). However, a high l-Trp diet failed to induce DP IELs in GF mice monocolonized with either L. reuteri 100-23 or L. reuteri ΔArAT (fig. S4C). Thus, L. reuteri acts together with a complex microbiota and dietary l-Trp to induce DP IELs. Bioactivity-guided fractionation of L. reuteri 100-23 supernatant identified the presence of indole-3-lactic acid (ILA), an indole derivative of l-Trp, which activated the AhR reporter cell line and induced CD8αα+CD4+ T cells (fig. S5). This compound was as effective as indole-3-aldehyde, which activates AhR in ILC3s (24) (Fig. 4F and fig. S4D). Ahr–/– mice lacked DP IELs, whereas these cells were present in Ahr+/– littermates and WT controls, supporting our observation that L. reuteri induces DP IELs through the release of AhR ligands (Fig. 4G). We next asked whether AhR drives DP IEL differentiation in a T cell–intrinsic or –extrinsic fashion. TGFβ with either L. reuteri culture supernatant or TCDD induced CD8αα+CD4+ T cells from naïve WT CD4+ T cells stimulated in vitro with Ahr–/– or WT splenic DCs pulsed with a combination of three superantigens. In contrast, Ahr–/– T cells did not differentiate into CD8αα+CD4+ T cells, regardless of whether DCs lacked AhR, implying that AhR is required in T cells (Fig. 4H). Moreover, stimulation of OTII T cells with Ahr–/– or WT DCs in the presence of TGFβ and either L. reuteri supernatant or TCDD induced CD8αα+CD4+ T cells (fig. S4E). Finally, DP IELs were significantly reduced in Ahrfl/fl × Rorc-Cre mice (which lack AhR in T cells and ILCs), as well as in Ahrfl/fl × Cd4-Cre mice (which lack AhR in T cells) in comparison with littermate controls (Fig. 4I and fig. S4, F and G), demonstrating that DP IEL development requires AhR activation in T cells. To determine when AhR signaling intervenes in the DP IEL developmental pathway, we analyzed the expression of Thpok in IELs from Ahr+/ and Ahr–/– mice. Thpokhi CD8ααCD4+ IELs, Thpoklo CD8αα+CD4+ IELs (DP IELs), and a small intermediate population of Thpoklo CD8ααCD4+ IELs were evident in Ahr+/ mice, whereas a single Thpokhi CD4+ IEL population predominated in Ahr–/– mice (Fig. 4J). Thus, a lack of AhR signaling arrests DP IEL development before Thpok down-regulation occurs.

Our study reveals that L. reuteri provides indole derivatives of dietary l-Trp, such as ILA, which activate AhR and lead to the down-regulation of Thpok and reprogramming of CD4+ IELs into DP IELs. Other indole derivatives may have a similar effect. This Ahr-mediated mechanism is distinct from those by which AhR affects other T cells with regulatory functions (Tr1, Tregs), intraepithelial γδ T cells, and DCs (2732). Yet undefined microbial species and dietary and/or self-antigens are necessary to expand the CD4+ IEL population with a diverse TCR repertoire that converts into DP IELs. The complete reliance of DP IELs on a single species (L. reuteri) and its tryptophan metabolites for their final maturation may provide a basis for the use of L. reuteri as a probiotic and tryptophan-rich food to treat disorders that may be modifiable by DP IELs, such as inflammatory bowel diseases (11).

Supplementary Materials

www.sciencemag.org/content/357/6353/806/suppl/DC1

Materials and Methods

Figs. S1 to S7

Tables S1 to S6

References (3350)

References and Notes

Acknowledgments: We thank M. Karlsson and D. O’Donell for their assistance in the gnotobiotic facility and J. Hoisington-Lopez from the DNA Sequencing Innovation Lab at the Edison Family Center for Genome Sciences and Systems Biology for her sequencing expertise. We thank members of our laboratory, in particular J. Bando, for their suggestions and critical reading of this manuscript. This work was supported by the Rainin foundation, NIH grants U01 AI095542 and DK103039 (M.Co.), NIH grant RO1-DK094995 and the Burroughs Wellcome Fund (C.-S.H.), NIH grant RO1 CA176695 (M.Ce.), NIH grant DK30292 (J.I.G.), and NIH Director’s New Innovator Award ID 1DP2AI124441 (M.S.D.). P.P.A. is the recipient of a Sir Henry Wellcome Postdoctoral fellowship (096100). L.C.-B. was supported by the Swiss National Science Foundation (fellowship PBSKP3-134332) and the Swiss Foundation for Medical-Biological Grants (fellowship PASMP3-145751). The data reported in this manuscript are presented in the main paper and in the supplementary materials.
View Abstract

Navigate This Article