Natural Aryl Hydrocarbon Receptor Ligands Control Organogenesis of Intestinal Lymphoid Follicles

See allHide authors and affiliations

Science  16 Dec 2011:
Vol. 334, Issue 6062, pp. 1561-1565
DOI: 10.1126/science.1214914


Innate lymphoid cells (ILC) expressing the transcription factor RORγt induce the postnatal formation of intestinal lymphoid follicles and regulate intestinal homeostasis. RORγt+ ILC express the aryl hydrocarbon receptor (AhR), a highly conserved, ligand-inducible transcription factor believed to control adaptation of multicellular organisms to environmental challenges. We show that AhR is required for the postnatal expansion of intestinal RORγt+ ILC and the formation of intestinal lymphoid follicles. AhR activity within RORγt+ ILC could be induced by dietary ligands such as those contained in vegetables of the family Brassicaceae. AhR-deficient mice were highly susceptible to infection with Citrobacter rodentium, a mouse model for attaching and effacing infections. Our results establish a molecular link between nutrients and the formation of immune system components required to maintain intestinal homeostasis and resistance to infections.

Adaptation to environmental cues in multicellular organisms requires a complex composition of the transcriptional output of responding cells. At gut mucosal surfaces, constant integration of microbial and dietary signals is required for the maintenance of intestinal homeostasis (13). Disturbances lead to susceptibility to intestinal infections, inflammatory bowel diseases, and inflammation-induced cancer (4). The intestinal lamina propria harbors numerous lymphoid follicles that in mice develop during the first weeks after birth. Their genesis is a multistep process initiated by the formation of cryptopatch (CP) clusters containing RORγt+ ILC that have lymphoid tissue–inducing (LTi) function (57). B cells are attracted to CP, leading to the formation of isolated lymphoid follicles (ILF), a process under the control of cues from the indigenous microbiota (8, 9). Besides their LTi function, RORγt+ ILC constitutively produce interleukin (IL)–22 (1014), required for the protection against intestinal attaching and effacing (A/E) infections that cause disease in mice and humans such as those with enterohemorrhagic Escherichia coli (1517). The AhR controls transcription of xenobiotic-metabolizing enzymes like cytochrome P450 family members (e.g., Cyp1a1 and Cyp1b1). Exogenous AhR ligands are environmental toxins (e.g., dioxin), bacterial metabolites, or naturally occurring ligands such as dietary ligands (e.g., flavonoids and glucosinolates), abundantly found in plants (18, 19). Although the role of AhR in the response to environmental toxins is widely appreciated, its broader role in adapting the organism’s response to natural ligands is ill defined.

Recent reports have indicated that RORγt+ ILC express AhR, but its role for their development or function is unknown (13). RORγt+ ILC expressed AhR at levels comparable to the AhR-responsive hepatoma cell line Hepa1.6 (fig. S1, A and B). Stimulation of RORγt+ ILC with various AhR ligands in vitro resulted in the induction of Cyp1a1 and Cyp1b1 gene expression, demonstrating that the AhR is functional in RORγt+ ILC (fig. S1C). Given the virtually ubiquitous expression of AhR in lymphocytes (fig. S1, A and B), we generated mouse lines with a tissue-specific deletion of the AhR in RORγt+ ILC (AhrΔLTi,T), dendritic cells (DC) (AhrΔDC), or intestinal epithelial cells (IEC) (AhrΔIEC) and compared these to mice lacking AhR in all cells (Ahr–/–) (20). Subsets of RORγt+ ILC have LTi function, and we investigated the role of AhR for the genesis of lymphoid organs. AhR signaling was dispensable for the prenatal development of secondary lymphoid organs (fig. S2). In contrast, CP and consequently ILF were largely absent in Ahr–/– and AhrΔLTi,T mice, whereas their development was normal in AhrΔDC and AhrΔIEC mice (Fig. 1, A and B, and fig. S3). The residual CP development found in some AhrΔLTi,T mice was most likely due to inefficient deletion of the Ahr allele, as demonstrated by the analysis of Cre efficiency using Rorc(γt)-CreTg x Rosa26R-EYFP mice (fig. S4) (14). AhrΔLTi,T mice lack AhR expression by RORγt+ ILC and T cells (fig. S1B) because RORγt is temporarily expressed in thymic CD4+CD8+ T cells and subsets of γδ T cells. The formation of CP/ILF is independent of T cells (fig. S5), however. Thus, the postnatal initiation of CP development in the small intestine likely requires AhR expression by RORγt+ ILC.

Fig. 1

AhR-mediated signals control postnatal development of intestinal lymphoid follicles. (A) Immature ILF in representative sections of the small intestine from 8-week-old mice were stained as indicated and analyzed by fluorescence microscopy. Scale bar, 50 μm. Original magnification, ×20. (B) Quantification of lymphoid clusters in the small intestine of the indicated mouse strains (20). Data are shown as mean ± SD, n ≥ 3. (C) Representative flow cytometry analysis of NKp46 and RORγt expression by CD3CD19 lymphocytes from 8-week-old mice. Numbers in quadrants represent the percentage of cells. (D) Absolute numbers (mean ± SD, n ≥ 5) of intestinal RORγt+ ILC. (E and F) Absolute numbers (mean ± SD, n = 5) of CD4+ (E) and CD4 (F) RORγt+ ILC per gram of intestinal tissue at various times after birth. (G) Absolute numbers (mean ± SD, n ≥ 5) of CD4+RORγt+ ILC in 8-week-old mice. Data are representative of eight [(A) to (D)] or three [(E) to (G)] independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test).

The failure to develop CP may reflect AhR-dependent defects in the development or maintenance of RORγt+ ILC or in their LTi function. The expression of genes known to be involved in lymphorganogenesis (21)—such as lymphotoxin a and b (Lta, Ltb), Tnfsf14 (encoding LIGHT), and Tnfsf11 (encoding RANKL)—was not impaired in AhrΔLTi,T mice (fig. S6). A general developmental defect of RORγt+ ILC is also unlikely given the normal prenatal development of secondary lymphoid organs, normal numbers of RORγt+ ILC in spleen and peripheral lymph nodes (fig. S2), and normal numbers of intestinal RORγt+ ILC in newborn Ahr–/– mice (fig. S7). However, in adult (8-week-old) Ahr–/– and AhrΔLTi,T mice, the intestinal pool of RORγt+ ILC was diminished (Fig. 1, C and D). With the exception of intestinal γδ T cells and RORγt+ mature T cells (22, 23), the role of the AhR for the development or maintenance of lymphocytes was limited to RORγt+ ILC subsets (fig. S8). However, the effect on T cell subsets does not contribute to the observed phenotype because the development of intestinal lymphoid follicles is independent of T cells (fig. S5). AhrΔDC or AhrΔIEC mice did not show any substantial perturbations in the pool of RORγt+ ILC (fig. S9, A and B). Collectively our data demonstrate that AhR-mediated signals in RORγt+ ILC are not required for their development but rather for the postnatal maintenance or expansion of intestinal RORγt+ ILC.

Intestinal RORγt+ ILC with LTi function are already present at birth and can be further divided into two subsets on the basis of CD4 expression (14, 24). At birth, CD4+ and CD4 RORγt+ ILC were equally represented in the intestinal lamina propria, and their numbers were not affected by AhR deficiency (Fig. 1, E and F). The pool of CD4+RORγt+ ILC remained largely stable after birth, and AhR deficiency led to a reduction by a factor of 2 in cell numbers (Fig. 1, E and G). In contrast, CD4RORγt+ ILC increased in numbers until day 24 when CP and ILF formation occurs (Fig. 1F), which was not observed in Ahr–/– or AhrΔLTi,T mice (Fig. 1, C, D, and F). Consequently, the CD4+ subset of RORγt+ ILC was overrepresented in Ahr–/– mice (fig. S10). In comparison with CD4RORγt+ ILC, AhR expression was lower by more than a factor of 5 in CD4+RORγt+ ILC (fig. S1, A and B), providing a rationale for the differential AhR dependency of the two subsets. We investigated the contribution of CD4+RORγt+ ILC to the development of intestinal lymphoid follicles. Both subsets were comparable in the expression of integrins (fig. S11A) and of genes involved in lymphorganogenesis (fig. S11B). In CP/ILF, the two subsets were present proportional to their representation within the pool of RORγt+ ILC (figs. S10 and S11C). Interestingly, postnatal depletion (20) of CD4+RORγt+ ILC did not impair the development of CP/ILF (fig. S12). Thus, AhR signals are required for the postnatal expansion of the intestinal pool of CD4RORγt+ ILC with LTi function and CD4+RORγt+ ILC are likely dispensable for the genesis of CP/ILF.

Because AhrΔLTi,T mice showed a phenotype similar to Ahr–/– mice, AhR-dependent transcriptional programs directly affect lineage-specified RORγt+ ILC rather than RORγt precursors. Therefore, we considered that expansion of CD4RORγt+ ILC may involve proliferation. In 2-week-old mice, roughly one-third of CD4RORγt+ ILC were Ki-67+, a marker of cellular proliferation (Fig. 2A). In Ahr–/– mice, Ki-67 expression by CD4RORγt+ ILC was reduced to the extent of CD4+RORγt+ ILC that do not expand (Figs. 1E and 2A). Thus, expansion of CD4RORγt+ ILC involves AhR-controlled proliferation. As AhR is a ligand-inducible transcription factor, we investigated the expression of candidate genes known to regulate the pool of RORγt+ ILC such as the IL-7 receptor α (IL-7Rα) (25) and the receptor tyrosine kinase Kit (26). IL-7Rα expression by RORγt+ ILC subsets from Ahr-deficient mice was normal (fig. S13, A and B). CD4RORγt+ ILC from Ahr+/+ and Ahr–/– mice expressed comparable levels of Kit shortly after birth, and Kit levels continuously decreased during the first days of life. Kit expression was restored in a process that required AhR (Fig. 2B). This is unlikely to reflect differential representation of the previously identified Kitlow and Kithigh subsets of CD4RORγt+ ILC (24) because Kit levels of CD4+RORγt+ ILC, considered to have a Kithigh phenotype, also decreased after birth (fig. S13C).

Fig. 2

Postnatal expansion of CD4RORγt+ ILC requires AhR-controlled stabilization of Kit. (A) The fraction of Ki-67–expressing RORγt+ ILC (mean ± SD, n ≥ 5) was determined by flow cytometry in 2-week-old mice. (B) Mean fluorescence intensity (±SD, n = 5) of Kit staining of RORγt+ ILC at various times after birth. (C) Highly purified RORγt+ ILC from 3-week-old mice were cultured for 24 hours in the presence of 400 nM of 6-formylindolo[3,2-b] carbazole (FICZ). Kit and CD127 expression were assessed by flow cytometry. Numbers in contour plot indicate the percentage of cells in each quadrant. (D) RORγt+ ILC were treated for 4 hours, either with 600 nM FICZ or with dimethyl sulfoxide (DMSO), and subjected to AhR ChIP analysis. (E) Cells were transfected with the indicated luciferase reporter plasmids and stimulated with 400 nM FICZ. One day later, relative units of luminescence were determined. (F) Quantification of lymphoid clusters in the small intestine of the indicated mouse strains (20). Data are shown as mean ± SD, n ≥ 3. (G) Representative sections of the small intestine of B6 and KitWv/Wv mice were stained as indicated and analyzed by fluorescence microscopy. Scale bar, 100 μm. Original magnification, ×20. Data are representative of three [(A) to (E)] and two [(F) and (G)] independent experiments. *P < 0.05 (Student’s t test).

The mouse Kit promoter contains two canonical xenobiotic response elements (XRE) (fig. S14), and Kit but not IL-7Rα expression was uniformly enhanced after stimulation of RORγt+ ILC with AhR ligands in vitro (Fig. 2C and fig. S15, A and B). After stimulation with AhR ligands, Kit transcription occurred rapidly, suggesting that it is under direct AhR control (fig. S15C). AhR chromatin immunoprecipitation (AhR ChIP) of primary RORγt+ ILC revealed that the AhR directly interacts with both XRE elements within the Kit promoter and that stimulation with AhR ligands substantially enhanced XRE occupancy by the AhR (Fig. 2D). AhR bound to these XRE elements induced transcription of the Kit gene, which was abrogated when the XRE elements were mutated (Fig. 2E). Previous data demonstrated that Kit signaling in RORγt+ ILC led to their expansion (26). Indeed, KitWv/Wv mice, which express a receptor with impaired kinase activity, had reduced numbers of RORγt+ ILC and impaired development of CP/ILF (Fig. 2, F and G). Collectively, our data link AhR-controlled Kit expression to postnatal expansion or maintenance of CD4RORγt+ ILC and the formation of intestinal lymphoid follicles.

What induces AhR-mediated transcription of Kit in RORγt+ ILC? Microbe-derived AhR ligands are unlikely candidates because the number of RORγt+ ILC and CP development is normal in germ-free mice (5). After birth, active extraction of metabolites by the absorptive epithelium, especially in the small intestine, emerges as the central route of nutrient supply. Most conventional mouse diets such as those used in our previous experiments (diet 3) are grain-based diets. These types of diets contain high concentrations (up to several g/kg) of phytochemicals such as polyphenols and glucosinolates, which are AhR ligands (19, 27). To investigate the role of dietary AhR ligands for CP/ILF development, we established mouse colonies fed with defined diets from nonplant sources (diet 1) that were nearly free of relevant phytochemicals. CP/ILF development in these mice was compared to mice fed with the same diet supplemented with indole-3-carbinol (I3C; diet 2), which is a well-characterized AhR ligand and a hydrolytic product of the glucosinolate glucobrassicin contained in plants of the Brassicaceae family (e.g., broccoli and Brussels sprouts) (19, 27). Mice fed with phytochemical-free diets had a phenotype similar to AhR-deficient mice, as demonstrated by the reduced formation of CP and ILF at 4 weeks of age (Fig. 3, A and B), reduced numbers of RORγt+ ILC (Fig. 3, C and D, and fig. S16A), impaired postnatal proliferation of CD4RORγt+ ILC (Fig. 3E), and low-level Kit expression by RORγt+ ILC subsets that did not increase after day 11 (Fig. 3F and fig. S16, B and C). Total numbers of intestinal lymphocytes (LPL) did not significantly change (fig. S16D). Strikingly, animals fed with the same diet supplemented with I3C recovered in all parameters (Fig. 3, A to F, and fig. S16). Similar effects were observed with another lot of diet 1 and a similarly defined diet from another vendor (fig. S17). Importantly, I3C-containing diets could not rescue postnatal expansion of RORγt+ ILC when fed to Ahr–/– mice (Fig. 3G), which demonstrates that I3C-mediated effects require AhR signals. Thus, dietary AhR ligands that are virtually absent from these defined diets activate AhR-mediated transcriptional programs promoting timely postnatal expansion of CD4RORγt+ ILC and the formation of intestinal lymphoid follicles.

Fig. 3

Dietary AhR ligands regulate the size of the RORγt+ ILC pool. (A to G) Colonies of mice fed with the indicated diets were analyzed between 3 and 4 weeks after birth. (A) Representative sections of the small intestine were stained as indicated and analyzed by fluorescence microscopy. Scale bar, 100 μm. Original magnification, ×20. (B) Quantification of lymphoid clusters in the small intestine of the indicated treatment groups. Data are shown as mean ± SD, n ≥ 3. (C and D) Quantification of the absolute numbers of RORγt+ ILC (C) and of the fraction of RORγt+ ILC among CD3CD19 lymphocytes (D) in the small intestine. Data are shown as mean ± SD, n ≥ 3. (E) The fraction of Ki-67–expressing RORγt+ ILC (mean ± SD, n ≥ 5) was determined by flow cytometry. (F) Mean fluorescence intensity (±SD, n ≥ 5) of Kit staining of RORγt+ ILC at the indicated time points. (G) Fraction (mean ± SD, n ≥ 3) of RORγt+ ILC among CD3CD19 lymphocytes in the small intestine. Data are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 (Student’s t test).

We investigated whether deprivation of dietary AhR ligands after formation of intestinal lymphoid follicles would lead to reduced numbers of lymphoid follicles or RORγt+ ILC. AhR ligand deprivation beginning at 6 weeks of age and lasting for 6 weeks did not reduce RORγt+ ILC numbers or the composition of CP/ILF (fig. S18). Notably, addition of a purified AhR ligand to the diet led to a transient increase in the numbers of RORγt+ ILC (fig. S18B). The data indicate that dietary AhR ligands do not constitute a major factor controlling survival of RORγt+ ILC after lymphorganogenesis is concluded, whereas the potential of RORγt+ ILC to respond to dietary AhR ligands with expansion may be maintained.

Besides being LTi, RORγt+ ILC also constitutively produce IL-22, which controls intestinal homeostasis and protects against intestinal infections (3, 10, 12, 13, 17). Mice lacking all RORγt+ ILC with LTi function can still develop lymphoid clusters in the colon that are instructed by inappropriately activated B cells, but these aberrant RORγt+ ILC-less clusters promote colitis (28). Interestingly, Ahr-deficient mice spontaneously developed the same type of aberrant lymphoid follicles characterized by the absence of RORγt+ ILC. These follicles contained activated B cells, as evidenced by high expression of activation-induced deaminase (Aicda) (fig. S19). On a per cell basis, IL-22 expression by RORγt+ ILC in small intestine and colon was consistently reduced in Ahr–/– and AhrΔLTi,T mice but not in AhrΔDC or AhrΔIEC mice (Fig. 4A and fig. S20, A to C). IL-17 production by RORγt+ ILC was not affected by AhR deficiency (fig. S20D). However, due to the substantially reduced numbers of RORγt+ ILC in small intestine and colon of Ahr–/– and AhrΔLTi,T mice (Fig. 1D), the absolute numbers of IL-22+RORγt+ ILC in adult mice were decreased by a factor of 10 to 15 (Fig. 4B and fig. S20E). IL-22–producing RORγt+ ILC are required to maintain epithelial expression of antimicrobial Reg3 genes, which was substantially reduced in AhR-deficient mice (Fig. 4C and fig. S20F). AhR-deficient mice failed to produce IL-22 in response to Citrobacter rodentium infection (Fig. 4, A and C), were unable to mount protective innate immunity to C. rodentium infection (Fig. 4D), developed more severe colitis (Fig. 4E and fig. S21), and had increased bacterial titers in liver, draining lymph nodes, and feces (Fig. 4F). Collectively, our data demonstrate that AhR-dependent postnatal expansion of RORγt+ ILC is required for protection against intestinal A/E infections.

Fig. 4

AhR-deficient mice are highly susceptible to intestinal C. rodentium infection. (A to F) Groups of Ahr+/+ and Ahr–/– mice were infected orally with 1010 colony-forming units (CFU) of C. rodentium or were left uninfected. (A and B) Percentage (A) and absolute numbers (B) of IL-22+ cells among RORγt+ ILC were determined by flow cytometry. Data are shown as mean ± SD, n ≥ 3. (C) Quantitative RT-PCR analysis of Reg3g expression by colonic epithelial cells. (D) Percent survival. (E) Histological analysis of colon section (hematoxylin and eosin stain) at day 8 of infection. Arrow in second panel shows infiltrating inflammatory cells. Arrows in third panel show submucosal inflammation. Scale bar, 100 μm. (F) Number of bacteria recovered from the indicated organs at day 8 following infection. Data in (A to F) are representative of three independent experiments (n = 8). *P < 0.05, **P < 0.01 (Student’s t test).

Defining the transcriptional networks that control intestinal homeostasis is essential to understand the pathogenesis of inflammatory bowel diseases and susceptibility to intestinal infections. We now show that AhR controls the pool size of RORγt+ ILC and, consequently, postnatal formation of CP/ILF. AhR-mediated transcription stabilizes Kit expression, which correlated with the postnatal expansion of the CD4RORγt+ ILC pool (26). Intriguingly, AhR and Kit signals also control the maintenance of intraepithelial γδ T cells (29, 30). Although CP development was widely perceived to be programmed (5, 7, 9), we now show that the timely formation of CP likely is under the control of environmental dietary signals. Perhaps plant-derived AhR ligands serve as molecular signatures, alerting the vertebrate host to nutrients that are constantly infested with potentially harmful pathogens. Intriguingly, the diverse glucosinolates play an important role in plant resistance to mildew fungi and insect herbivores, suggesting an evolutionary conserved role in immune defense pathways (31). Our data and two recent reports highlight the potential of diets to control the formation of immune system components (29) and the composition of the microbiota (32). Plant-derived phytochemicals seem to enhance the pool of RORγt+ ILC, and derivative compounds may be used in the future to therapeutically modulate their function in the context of necrotizing enterocolitis of preterm infants (33) and intestinal inflammatory diseases (4).

Supporting Online Material

Materials and Methods

Figs. S1 to S21

References (3447)

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: We thank G. Häcker for support and for critically reading the manuscript; the members of the Diefenbach laboratory for valuable discussions; H. Pircher, Y. Tanriver, W. Schachterle, and S. Ganal for comments on the manuscript; J. Brandel for help with the figures; and N. Göppert and K. Oberle for technical assistance. We are grateful to T. Haarmann-Stemmann, C. Johner, D. Littman, R. Nechanitzky, G. Niedermann, M. Rosenbaum, F. Santori, and A. Schuhmacher for experimental support and to J. Wersing for cell sorting. The data reported in this paper are tabulated in the main paper and in the supporting online material. The work was supported by the Deutsche Forschungsgemeinschaft (GRK1104, SGBM, and SFB620/A14 to A.D. and E.A.K.), the Bundesministerium für Bildung und Forschung, Centrum für Chronische Immundefizienz (to A.D.), and the Swiss National Science Foundation (grant SNF 310030_130674/1 to D.F.).

Stay Connected to Science

Navigate This Article