Report

Lineage Relationship Analysis of RORγt+ Innate Lymphoid Cells

See allHide authors and affiliations

Science  29 Oct 2010:
Vol. 330, Issue 6004, pp. 665-669
DOI: 10.1126/science.1194597

Innate Innit?

Innate lymphocytes (ILCs) are a recently described population of immune cells that produce cytokines like those associated with T helper cells, but lack the recombined antigen receptors characteristic of T cells. Again, like some T helper cell lineages, a proportion of ILCs express the transcription factor RORγt. These include lymphoid tissue inducer (LTi) cells required for fetal lymphoid tissue organogenesis and a population of natural killer (NK)–like cells that function in gut immune responses. Sawa et al. (p. 665; see the Perspective by Veldhoen and Withers) wondered whether the RORγt-expressing ILCs all develop from the same progenitor population. Indeed, they found a fetal liver progenitor that gave rise to several phenotypically distinct populations. However, the LTi cells were not progenitors for the NK-like cells. It seems the trajectory of different ILC populations is developmentally regulated, and postnatally ILCs are favored that play a role in intestinal defense before the gut is fully colonized by intestinal microbiota.

Abstract

Lymphoid tissue–inducer (LTi) cells initiate the development of lymphoid tissues through the activation of local stromal cells in a process similar to inflammation. LTi cells express the nuclear hormone receptor RORγt, which also directs the expression of the proinflammatory cytokine interleukin-17 in T cells. We show here that LTi cells are part of a larger family of proinflammatory RORγt+ innate lymphoid cells (ILCs) that differentiate from distinct fetal liver RORγt+ precursors. The fate of RORγt+ ILCs is determined by mouse age, and after birth, favors the generation of cells involved in intestinal homeostasis and defense. Contrary to RORγt+ T cells, however, RORγt+ ILCs develop in the absence of microbiota. Our study indicates that RORγt+ ILCs evolve to preempt intestinal colonization by microbial symbionts.

Lymphoid tissue–inducer (LTi) cells are necessary in the fetal development of lymph nodes and Peyer’s patches, a lymphoid tissue of the gut (1), as well as in the formation of intestinal isolated lymphoid follicles after birth (2). They express membrane lymphotoxin (LT) α1β2 that activates local stromal cells through the receptor LTβR. Activated stromal cells in lymphoid tissue anlagen recruit and organize lymphocytes in a process resembling inflammation (3), involving proinflammatory cytokines, chemokines, and adhesion molecules (1). The nuclear hormone receptor RORγt is required for the generation of LTi cells (4). After birth, RORγt induces a proinflammatory program in subsets of αβ and γδ T cells, characterized by the expression of interleukin-17 (IL-17) (57), also expressed by LTi cells (8). Furthermore, RORγt is required for the generation of a subset of natural killer (NK)–like NKp46+ cells present in the intestinal lamina propria and characterized by the production of high amounts of IL-22 (911), a major factor in intestinal homeostasis and defense (12, 13). It is now suggested that LTi cells, NK cells, NK-like cells (14), and recently discovered T helper 2 (TH2)–like innate cells involved in intestinal immunity to helminths (1517) are members of a growing family of innate lymphoid cells (ILCs), which develop in the absence of the recombinated antigen receptors and clonal selection that are characteristic of adaptive immunity.

Recent evidence indicates that LT or LTi-like cells are not only necessary effectors for lymphoid tissue formation but also precursors to IL-22+ NK-like cells (18, 19). TH2-like ILCs also have multipotent progenitor capacity and generate mast cells, basophils, and macrophages when stimulated with the TH2-promoting cytokine IL-25 (16). To assess in vivo the progenitor capacity of LTi cells, we performed inducible cell fate–mapping experiments that allowed tracking of the progeny of fetal RORγt+ CD3ε cells (termed here RORγt+ ILCs). Mice were generated that express the tetracycline transactivator (tTA) under control of the Rorct) locus on a bacterial artificial chromosome (7, 20) (figs. S1 and S2). The tTA induces expression of the Cre recombinase on a second transgene (21), which in turn induces stable expression of enhanced yellow or red fluorescent protein (EYFP or RFP) in RORγt+ cells and all their progeny (22). Induction of genetic fate mapping can be controlled in time by the administration of doxycycline (dox), which blocks the tTA-mediated labeling cascade. Thus, the progeny of fetal RORγt+ ILCs was determined by perinatal administration of dox to such triple transgenic mice. First, and in contrast to CD4+ T cells that derive from RORγt+ immature thymocytes and down-regulate RORγt at the mature stage (23), RORγt+ ILCs stably expressed RORγt (Fig. 1A and fig. S2). Second, whereas fetal and perinatal RORγt+ ILCs consisted mainly of c-kithigh IL-7Rαhigh CD4+ (LTi4) and CD4 (LTi0) cells (Fig. 1B and fig. S3), a phenotype consistent with LTi cells (24), the progeny of such cells progressively increased in complexity until 4 weeks after birth (Fig. 1B). Besides CD4+ and CD4 LTi cells, the adult progeny of fetal RORγt+ ILCs included c-kitlow IL- 7Rαlow cells, of which 42% expressed NKp46 (911) (Fig. 1C). Finally, in the adult, RORγt+ ILCs localized mainly to the intestinal lamina propria (fig. S4).

Fig. 1

Inducible fate mapping of RORγt+ ILCs. (A) Flow cytometry analysis of cells isolated from the small intestinal lamina propria of dox-untreated 4-week-old triple transgenic mice crossed to Rorct)-EgfpTG reporter mice (fig. S2). The red square indicates cells that have lost RORγt expression. Numbers indicate the percentage of cells per gated type of cells. The data shown are representative of three individual experiments. (B and C) The fate of fetal and perinatal RORγt+ ILCs of the small intestinal lamina propria. Triple transgenic mice (fig. S1) were given dox continuously, starting on day 3 after birth until analysis. (B) Each data point represents the mean percentage ± SD from at least five individual mice. (C) Flow cytometry analysis of 4-week-old triple transgenic mice. The data shown are representative of at least five individual mice.

Even though the fate mapping of fetal RORγt+ ILCs indicated that LTi cells generate the different subsets of RORγt+ ILCs (Fig. 1B), it is possible that fetal RORγt+ ILCs include RORγt+ precursors that are distinct from LTi cells. To test whether LTi cells could generate distinct subsets of RORγt+ ILCs such as NKp46+ cells, LTi cells were isolated from fetal and adult intestine and transferred into irradiated newborn hosts or cultured for 6 days on OP9 stromal cells. Transferred LTi cells generated no detectable progeny, and cultured LTi cells generated no c-kitlow IL-7Rαlow cells, including NKp46+ cells (Fig. 2A and fig. S5). Furthermore, the in vivo depletion of CD4+ cells affected only CD4+ LTi cells and no other subsets of RORγt+ ILCs (fig. S6), and the proportion of CD4+ LTi cells labeled through our fate-mapping strategy was fivefold higher than the proportion of labeled NKp46+ cells (fig. S2A). Collectively, these data indicate that LTi cells are not progenitor cells of other subsets of RORγt+ ILCs. In contrast, transfer of fetal liver lineage-negative cells into irradiated C57BL/6 mice generated LTi cells, as previously described (25, 26), and the other subsets of RORγt+ ILCs (Fig. 2B). Examination of fetal liver cells revealed three subsets of RORγt+ cells characterized by distinct expression levels of RORγt and integrin α4β7 (Fig. 2C), the latter marker having been shown to define a fetal liver precursor committed to dendritic cells, NK cells, and LTi cells (25). Although too small to be used in transfer experiments, these populations generated all subsets of RORγt+ ILCs upon culture on OP9 cells: GFP+ α4β7+ cells generated CD4+ and CD4 LTi cells, whereas GFP+ α4β7 cells generated NKp46+ cells (GFP, green fluorescent protein) (Fig. 2C and fig. S7A). The presence of IL-2 favored the generation of NK cells from integrin α4β7+ cells, as reported (25), but not of RORγt+ NKp46+ cells (fig. S7B). These data demonstrate that fetal liver RORγt+ cells, but not LTi cells, generate the distinct subsets of RORγt+ ILCs. It has been reported, however, that human LTi-like cells isolated from fetal lymph nodes and adult tonsils generate RORγt+ NKp46+ cells in vitro (18, 19). Thus, in humans, LTi-like cells may include RORγt+ precursors specific for NKp46+ cells.

Fig. 2

The diversity of RORγt+ ILCs is generated by committed fetal liver RORγt+ precursors. (A) Sorted fetal LTi cells (LTi4 or LTi0) from the intestinal lamina propria of Rorct)-EgfpTG fetuses 16 days after coitus [embryonic day 16 (E16)] as well as adult c-kitlow RORγt+ ILCs were cultured on OP9 stroma cells in the presence of IL-7 and stem cell factor (SCF). Cells were analyzed by flow cytometry after 6 days. The data are representative of four independent experiments. (B) Reconstitution of irradiated C57BL/6 hosts was analyzed by flow cytometry 6 weeks after adoptive transfer with lineage-depleted 5 × 105 fetal liver cells from E16 Rorct)-EgfpTG mice. Data are representative of three individual mice. (C) Upper left: flow cytometric analysis of lineage-negative fetal liver cells from E16 Rorct)-EgfpTG mice. Three GFP+ subsets were sorted and cultured on OP-9 cells with IL-7 and SCF and analyzed by flow cytometry after 6 days. Gut mature E16 LTi4 cells were used as a reference for the level of c-kit expression. Data are representative of three independent experiments.

Do fetal RORγt+ ILCs generate adult RORγt+ ILCs? To address this issue, we determined the turnover rate of RORγt+ ILCs from the intestinal lamina propria and found that all subsets were replaced by newly generated cells with a half-life of 22 to 26 days (fig. S8). Similar kinetics were found for RORγt+ ILCs in the mesenteric lymph node and spleen (fig. S9). Thus, a likely source of RORγt+ ILCs after birth is the bone marrow, and transfer of bone marrow cells into irradiated hosts generated all subsets of RORγt+ ILCs (fig. S10). The RORγt+ precursors in the bone marrow, however, remain to be identified.

The sharp population change occurring in RORγt+ ILCs after birth, characterized by a drop in the proportion of CD4+ LTi cells (Fig. 1B), is particularly intriguing. Because LTi cells do not generate c-kitlow IL-7Rαlow cells (Fig. 2), this population change may be partly explained by a markedly higher proliferation rate of c-kitlow IL-7Ralow cells, including NKp46+ cells, as compared to LTi cells (Fig. 3A). In contrast, the progeny of RORγt+ ILCs from pre-weaned or adult mice showed no such sharp decrease in the proportion of LTi cells. They rather mimicked the proportions within total LTi cells found at these ages (Fig. 3B and fig. S11). These observations indicate that the proportions of the different subsets of RORγt+ ILCs are determined by mouse age rather than by cell-intrinsic properties of LTi cells or c-kitlow IL-7Rαlow cells.

Fig. 3

The fate of RORγt+ ILCs subsets is programmed. (A) The proliferation rates of RORγt+ ILCs. Triple transgenic mice were treated with dox from day 3 after birth and injected with 5-ethynyl-2′-deoxyuridine (EdU) daily from day 7 until analysis on day 14. PBS, phosphate-buffered saline. Cells were isolated from the small intestinal lamina propria and analyzed by flow cytometry. Data are representative of three independent experiments. (B) Triple transgenic mice (fig. S1) were treated with dox from days 14 or 28, and cells were analyzed by flow cytometry at different time points thereafter. Each data point represents the mean percentage ± SD of at least five individual mice. (C) Triple transgenic mice were treated from birth with a cocktail of streptomycin, ampicillin, and colistin, as well as with dox, from day 7 after birth and analyzed at 8 weeks. Abx, antibiotics. Shown is the proportion of the distinct subsets of RORγt+ ILCs within fate-mapped YFP+ cells. The data are the mean of four mice per group ± SD; n.s, statistically not significant; unpaired t test. (D) Intestinal ILCs were analyzed by flow cytometry in 10-week-old germ-free (GF) or specific pathogen–free (SPF) Rorct)-EgfpTG mice. Data are representative of three individual mice.

To determine whether the symbiotic microbiota controls this age-dependent fate of RORγt+ ILCs, the fate of fetal and perinatal RORγt+ ILCs was assessed in mice treated continuously from birth with a cocktail of antibiotics that efficiently eradicates intestinal bacteria (2) and with dox, used for fate mapping. In such mice, the fate of fetal and perinatal RORγt+ ILCs remained unchanged (Fig. 3C). In accordance with these results, the proportions of the different subsets of RORγt+ ILCs were similar in germ-free (or dox-treated) mice as compared to mice that harbor a normal symbiotic microbiota (Fig. 3D and fig. S12); in mice recolonized with segmented filamentous bacteria that induce the generation of RORγt+ Th17 cells (27, 28) (fig. S13); and in mice deficient in molecules involved in the recognition of microbe-associated molecular patterns, such as Nod1, Nod2, or Myd88 (fig. S14). These data demonstrate that intestinal symbionts, which induce the generation of intestinal RORγt+ T cells (29), have no impact on the development of RORγt+ ILCs. Furthermore, RORγt+ ILCs also develop independently of adaptive immunity (fig. S15).

All subsets of RORγt+ ILCs, with the exception of NKp46+ cells, expressed high amounts of both IL-17 and IL-22 (Fig. 4A). However, LTi cells differed from c-kitlow IL-7Rαlow cells in their expression of the chemokine receptor CCR6 (Fig. 4B). It has been reported that CCR6 is expressed by LTi cells clustered in cryptopatches of the lamina propria (2, 30), but not by RORγt+ ILCs outside cryptopatches (30), such as NKp46+ cells (9), which expand after birth (Fig. 1B). CCR6 is required for the bacteria-induced generation of isolated lymphoid follicles (ILFs) from cryptopatches (2), and in the absence of CCR6 and ILFs, intestinal homeostasis is disrupted and the size of the microbiota is increased 10-fold (2). Although RORγt+ ILC subsets are not altered in CCR6-deficient mice (fig. S16), the expression of IL-22 is significantly increased (Fig. 4C), and, as a consequence (31), the production of antibacterial peptides by epithelial cells is augmented (Fig. 4D). These data indicate that CCR6 regulates the function of RORγt+ ILCs. When CCR6-dependent topographical control is lost, the lymphoid tissue-inducing function of RORγt+ ILCs is ablated and their proinflammatory or epithelial defense–promoting function is abnormally expanded.

Fig. 4

Topographical and functional compartmentalization of RORγt+ ILCs by CCR6. (A) Flow cytometric analysis of intracellular expression of IL-17A and IL-22 by RORγt+ ILCs. Cells were isolated from the small intestinal lamina propria of 2-week-old Rorct)-EgfpTG mice. Data are representative of three independent experiments. (B) Flow cytometric analysis of CCR6 expression by LTi cells. Numbers indicate the percentage of CCR6+ cells as compared to cells from age-matched (6 weeks old) CCR6-deficient mice (blue histograms). Data shown are representative of results obtained in three individual mice. (C) Expression of IL-22 by CCR6-deficient (CCR6KO) RORγt+ ILCs isolated from 6-week-old mice. WT, wild type. The data are the mean of four mice per group ± SD; *P < 0.05, unpaired t test. (D) Antibacterial peptide expression by epithelial cells in 6-week-old CCR6-deficient mice. Transcripts for the antibacterial peptides Reg3β and Reg3γ were measured by quantitative real-time polymerase chain reaction. Data are the mean of four individual mice ± SD; *P < 0.05, unpaired t test.

The major population change occurring in RORγt+ ILCs after birth, resulting in LTi cells being numerically surpassed by c-kitlow IL-7Rαlow cells after weaning, suggested that the colonizing intestinal microbiota directed the development of RORγt+ ILCs. We show, however, that this population change is independent of microbiota, indicating that RORγt+ ILCs undergo a programmed development that preempts exposure to the symbiotic microbiota. Both LTi cells and NKp46+ RORγt+ ILCs express IL-22 (811), a critical cytokine for the activation and defense of epithelial cells (13, 31). Whereas LTi cells are clustered in cryptopatches between crypts of the intestinal lamina propria, where they direct the formation of isolated lymphoid follicles (ILFs) (2), IL-22+ NKp46+ cells are found within villi (9), thus closer to epithelial cells. An important role for IL-22+ NKp46+ cells in epithelial defense has been shown in the case of Citrobacter rodentium infection (9) and in resistance to colitis induced by dextran sodium sulfate (13). Our data suggest that this topographical and functional compartmentalization of RORγt+ ILCs depends on the chemokine receptor CCR6, which responds to CCL20 and β-defensins produced by epithelial cells. We have demonstrated that RORγt+ ILCs, required before birth mainly for the development of lymphoid tissues, undergo a programmed population change after birth to cope with the massive microbiota and maintain intestinal homeostasis, both through the CCR6-dependent generation of ILFs and through the activation of epithelial immunity. The programmed fate of RORγt+ ILCs is an example of the coevolution of the mammalian host immune system with its symbiotic microbiota in order to maintain homeostasis of the host/symbiont superorganism.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1194597/DC1

Materials and Methods

Figs. S1 to S16

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank the members of the Lymphoid Tissue Development lab for discussions and critical reading of the manuscript and L. Polomack, S. Dulauroy, G. Chauveau-Le Friec, and the team of the Centre d’Ingénierie Génétique Murine for technical assistance. We also thank R. Varona for CCR6-deficient mice, I. Gomperts-Boneca for Nod1- and Nod2-deficient mice, and H. Bujard for LC-1-Cre mice. This work was supported by the Institut Pasteur, grants from the Mairie de Paris, the Agence Nationale de la Recherche, and an Excellence Grant from the European Commission. M.L. was supported by the Deutsche Forschungsgemeinschaft and the Schlumberger Foundation. H.J.F. is funded by the Deutsche Forschungsgemeinschaft (FE-578/3-1). The authors have no competing financial interests.
View Abstract

Navigate This Article