Cells that belong to the family of innate lymphoid cells (ILCs) not only form a first line of defense against invading microbes, but also play essential roles in tissue remodeling and immune pathology. Rorγt+ ILCs, producing the cytokines IL-22 and IL-17, include lymphoid tissue inducer (LTi) cells which are critical for the formation of lymphoid structures. Recently another ILC subset has been identified, which is dependent on RORα for its development and is dedicated to the production of the Th2 cytokines IL-5 and IL-13. These ILCs have been termed type 2 ILCs. All ILC subets are considered to belong to the same family that also includes natural killer cells because they all rely on the common γ-chain (γc) of the IL-2 receptor for their development and function, share a lymphoid morphology and depend on the transcriptional repressor Id2 for their development. Other transcription factors, including Notch, and the aryl hydrocarbon receptor (AhR) in RORγt+ ILCs and GATA3 in type 2 ILCs, also play roles in the development, survival, and function of these ILC subpopulations. Here we review the current knowledge with regard to the transcription factors involved in the development and functions of ILCs.
Innate lymphoid cells (ILCs), including RORγt+ ILCs and type 2 ILCs, represent a novel group of cells related to NK cells. ILCs lack antigen receptors encoded by rearranged genes, such as the T-cell receptors expressed by T cells (reviewed in [[1, 2]]). Emerging evidence indicates that ILCs not only function as a first line of defense against invading microbes, but also play essential roles in tissue remodeling. For instance, LTi cells are involved in repair of tissue damage inflicted by lymphocytic choriomeningitis virus following clearance of infection [], and type 2 ILCs mediate tissue repair in the lung following infection with influenza virus []. Two transcription factors appear to define two major subpopulations of ILCs: retinoid acid related orphan receptor transcription factor (ROR)α, and RORγt [[1, 5, 6]]. The signature cytokines produced by RORγt-dependent ILCs are IL-17 and IL-22, hence these cells are referred to as ILC17 and ILC22, respectively, whereas RORα-dependent ILCs have the ability to produce the type 2 cytokines IL-5 and IL-13. As such, RORα-dependent ILCs are referred to as type 2 ILCs (ILC2s).
Interestingly, based on their cytokine expression profiles, the ILC2, ILC22, and ILC17 populations can be considered as the innate equivalents of the T helper (Th) family members, being the Th2, Th22 and, Th17 subsets, respectively. NK cells that are cytotoxic and secrete interferon gamma may be the innate version of CD8+ cytotoxic T cells.
Other transcription factors, including Notch, and the aryl hydrocarbon receptor (AhR) in RORγt+ ILCs and GATA3 in type 2 ILCs, play also roles in the development, survival, and function of these ILC subpopulations. Unraveling the transcriptional networks that regulate ILCs is still work in progress, and much remains yet to be learned; however, important discoveries have already been made and here we review current knowledge with regard to the transcription factors involved in the development and functions of ILCs.
Id2 and E47 regulate development of ILCs and NK cells
E proteins are basic helix-loop-helix (bHLH) transcription factors that control the development and function of various immune cell populations including T cells, B cells, NK cells and plasmacytoid (p) DCs (reviewed in []). The E proteins contain an HLH domain involved in dimerization and a basic DNA binding domain. Id proteins are HLH proteins that lack a basic DNA binding domain; they can form dimers with E proteins, but these complexes are unable to bind DNA and, as a consequence, Id proteins inhibit the transcriptional activities of E proteins. There are 4 major E proteins: two of these are E12 and E47, which are splice-variants encoded by the E2A gene (therefore also referred to as E2A proteins); the other family members are HEB and E2–2. Lack of functional E2A proteins prevents the development of B cells and impedes T-cell development, whereas HEB and E2–2 are needed for the development of T cells [[8, 9]] and pDCs [[10, 11]] respectively. E2A proteins, in particular E47, inhibit the development of NK and LTi cells []. Id2 sequesters E47, thereby promoting NK- and LTi-cell development. As a consequence, Id2 deficiency results in inhibition of NK cell [], Rorγt+ ILC [] and type 2 ILC [] development. Blockage of LTi- and NK-cell development in Id2-deficient mice can be overcome by genetic ablation of E47 []. The finding that overexpression of Id2 in hematopoietic precursors results in inhibition of T-cell, B-cell and pDC development [[16-18]] while promoting ILC and NK-cell differentiation, indicates that the ratio of E and Id proteins serve as a switch, regulating the cell fate of T cells, B cells and pDCs on the one hand, and ILCs and NK cells on the other.
The common dependency of NK cells, Rorγt- and RORα-dependent ILCs on Id2 for their development suggests that these cell populations are derived from a common Id2-dependent precursor (Fig. 1), although it cannot presently be excluded that Id2 is not required for the development of ILCs and NK cells at the level of a common precursor but at later stages of development. It is therefore important to determine whether all ILCs and NK cells are derived from one common NK/ILC precursor or develop independently from an upstream, uncommitted, precursor such as the common lymphoid precursor.
Validation of this idea requires identification of this precursor cell. Using Id2-GFP reporter mice, Beltz and colleagues identified an Id2high CD117intermediateCD127high Flt3− population in the bone marrow []. These cells lack any NK markers but differentiate in vitro to NK cells when cultured with IL-7 plus IL-15. It might be possible that those cells also have the capacity to differentiate into Rorγt+ ILCs under the influence of other cytokines. Regardless of whether Id2 controls differentiation of a common NK-cell and ILC precursor or not, the continued expression of Id2 and the consequent downregulation of the activity of the E proteins may be required for the maintenance of the ILC/NK-cell lineages [], mirroring the requirement of continued expression of E2A proteins for B-cell development [].
TOX regulates the development of LTi and NK cells
TOX is an HMG box transcription factor that is expressed in several stages of T-cell development in the thymus. Genetic ablation of Tox results in strong inhibition of the transition from CD4+CD8+ double positive thymocytes to CD4+ single positive T cells, and, as a consequence, there are no CD4+ T cells in Tox−/− mice []. TOX is also expressed in LTi and NK cells, numbers of which are significantly reduced in Tox-deficient mice [[22, 23]]. As a consequence, almost no lymph nodes are present in these animals, with the exception of small numbers of phenotypically abnormal Peyer's patches. These data suggest that TOX is expressed in a precursor of both LTi and NK cells. The observation that enforced expression of Id2 in Tox−/− precursor cells is insufficient to overcome the Tox deficiency [] may suggest that TOX does not function upstream of Id2; however it cannot be excluded that TOX does act upstream of Id2 but that it also controls other essential targets and that this latter function cannot be overcome by introducing Id2 in Tox-deficient cells. Rorγt+ ILCs are not completely absent in adult Tox-deficient animals, but this may be due to the occurrence of inflammatory responses in the gut as a consequence of the absence of CD4+ T cells, including regulatory T cells in these mice []. Inflammation might have stimulated proliferation of the few Rorγt+ ILCs that are still present in Tox−/− mice; however, the precise mechanisms by which TOX regulates the differentiation of NK cells and ILCs are yet unknown [].
Regulation of RORγt-dependent ILCs
The prototype RORγt+ ILCs are the LTi cells, which play essential roles in the formation of secondary lymph nodes during fetal development, both in mice and humans [[25, 26]]. After birth, LTi cells are important for the formation of cryptopatches (CPs), as well as isolated lymphoid follicles (ILFs), which evolve from CPs. Within the ILFs, LTi cells are required for the production of IgA by B cells []. LTi cells are able to produce predominantly IL-17, but also some IL-22 [].
Other RORγt-dependent ILCs, which emerge after birth, have been identified [[28-35]]. These cells express the natural cytotoxicity receptor NKp46 and mostly produce IL-22, and hence they are referred to as the ILC22 subset. This subset plays several roles in the early stages of the immune response against pathogens, as exemplified by the effacing-attaching bacterium Citrobacter rodentium. This bacterium causes colitis and wasting disease, which is transient, and is cleared by T cells []. IL-22 is essential in the early response against C. rodentium as, in the absence of this cytokine, these cytokine-deficient mice succumb to the infection []. In this setting, IL-22 is mainly derived from ILCs, as deletion of the ILC22 subset in the acute phase of infection is fatal for the C. rodentium-infected mice, illustrating the importance of these cells in this type of immune response [[30, 34, 38]]. IL-22 production from ILCs is regulated by IL-23 and IL-1β [[39, 40]], and IL-22 mediates its protective effects by acting on epithelial cells, inducing proliferation and secretion of antimicrobial peptides (reviewed in []). A RORγt-dependent ILC population that produces IL-17, rather than IL-22, and is therefore called the ILC17 subset, is present in inflamed intestines in a model for inflammatory bowel disease []. Deletion of these cells ameliorates colitis suggesting that they mediate pathology in this model.
Thus far, three transcription factors have been identified that are involved in the control of development, survival, and function of Rorγt-dependent ILCs: Rorγt, Notch and AhR.
The RORC gene encodes two isoforms: RORγ (also referred to as RORγ1) and RORγt (called RORγ2). RORγ is a broadly expressed nuclear receptor []. RORγt is shorter than RORγ at the N-terminus, as the most 5’ end exons are replaced by a specific RORγt exon. ROR contains a ligand-binding domain to which different ligands can bind, such as 7 substitute oxysterols ([], and reviewed in []), but the exact nature of the agonist that binds to RORγt in different cell types is unclear. RORγt is expressed in double positive thymocytes, Th17 cells, a subset of FoxP3+ Treg cells, subsets of iNKT cells and TCRγδ cells, LTi cells and the ILC22 subset [[5, 45-47]], and genetic ablation of RORγt affects the development and functions of those cell types. RORγt-deficient mice completely lack LTi cells and, as a consequence, Rorγt−/− mice fail to develop lymph nodes, Peyer's patches and ILFs []. In Rorγt−/− mice, numbers of IL-22-producing ILCs, which express NKp46, are severely reduced as well as is their capacity to produce IL-22, whereas NK-cell numbers are unaffected [[30, 35, 41]]. The fact that RORγt is required for the development of both IL-17- and IL-22-producing Th17 cells [] and ILCs reinforces the idea that RORγt+ ILCs are the innate equivalent of Th17 cells.
AhR and Notch
AhR is a ligand-dependent transcription factor that belongs to the family of bHLH PER-ARNT-SIM transcription factors. AhR acts as a sensor of a variety of chemicals, including environmental toxins such as 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin (TCDD), and phytochemicals such as indol-3-carbinol, produced by cruciferous vegetables including cauliflower, cabbage, and broccoli []. Endogenous ligands have been identified as well, for instance the tryptophan photoproduct 6-formylindolo-(3,2-b)-carbazole (FICZ). In the cytoplasm, AhR is a component of a complex that includes chaperones like hsp90 and from which AhR is dissociated upon its activation by ligand binding. AhR associates with the AhR nuclear transporter (Arnt) prior to translocation to the nucleus to bind to promoters of a variety of genes (reviewed in []).
Only recently was a role for AhR in immunity identified. In mice, AhR controls the differentiation of Th17 cells [], and negatively affects the development of Treg cells []. Inhibition of Th17-cell differentiation by T cell-specific deletion of AhR resulted in the amelioration of collagen-induced arthritis, indicating that over-stimulation of AhR can result in pathology []. Interestingly, AhR controls the production of IL-22 by T cells, as ablation of AhR in mice completely eliminated the capacity of Th17 cells to produce IL-22 [[49, 52]]. Furthermore, AhR is involved in IL-22 production by Th22 cells in humans []. More recently, another activity of AhR emerged when it was found that AhR controls the maintenance of gut epithelium-residing CD8αα+ TCRαβ and TCRγδ cells (collectively denoted as intraepithelial lymphocytes (IELs)). Genetic ablation of AhR resulted in specific loss of IELs []. Interestingly, dietary components, in particular indol-3-carbinol, serve as ligands for AhR. Furthermore, these dietary products have been shown to be important for IEL maintenance, since mice fed with a vegetable-free diet showed reduced numbers of these cells [].
Recent work has established that AhR is not only important for the maintenance of IELs, but also for both LTi cells and the ILC22 subset that reside in the gut. Several groups reported that AhR-deficient mice had clearly reduced numbers of Rorγt+ ILCs, including LTi cells and ILC22 cells, in the gut [[54-56]]. This was also seen in Ahr−/−Rag1−/− mice, lacking T cells, as well as in sublethally irradiated wt mice receiving Ahr−/– stem cells. In addition, Rorγt+ ILC numbers were also reduced upon specific deletion of AhR in Rorγt-expressing cells (including ILCs) []. Together these data indicate that the effects of AhR-deficiency on Rorγt+ ILCs are cell intrinsic. Interestingly, the reduction of Rorγt+ ILC numbers, induced by ablation of AhR, was observed only after birth. During fetal development, and early after birth, the ILC22 numbers in AhR-deficient mice are comparable to those in wild type mice, indicating that AhR is not required for development of these cells []. However, after weaning, the numbers of Rorγt+ ILCs in AhR-deficient mice steadily decrease []. Maintenance of ILC numbers is not a consequence of AhR activation by products of colonizing microbiota, because the difference in ILC22 numbers between wt and AhR-deficient animals is not affected by treatment with a mix of antibiotics []. Also, the observation that germ-free animals do not show reductions in gut residing Rorγt+ ILC numbers [[55, 57]] is consistent with the notion that products from commensals are not required for the maintenance of these cells.
It is controversial whether dietary products are the AhR ligands responsible for the maintenance of gut-residing Rorγt+ ILCs, as observed for IELs []. In one study, it was found that mice fed with a diet free of AhR-binding phytochemicals showed decreased numbers of Rorγt+ ILCs, causing a lack of CPs and ILFs []. Addition of indole-3-carbinol, a dietary product, restored the Rorγt+ ILC numbers []. Another study, however, suggested that endogenous AhR ligands, including the tryptophane catabolite kynurenine, were potent regulators of Rorγt+ ILC maintenance as removal of dietary AhR ligands in that study did not disturb Rorγt+ ILC homeostasis and function []. The differences may be due to different types of controlled diets used by the different groups. Further experiments should aim to resolve these discrepancies.
The mechanisms by which AhR controls Rorγt+ cell numbers are not fully understood. Microarray analysis of Rorγt+ cells from wt and AhR-deficient mice suggested that Notch 1 is a downstream target of AhR []. Consistent with this, administration by gavage of the toxin TCCD (2,3,7,8-tetrachlorodibenzo-p-dioxin) resulted in the upregulation of Notch1 and Notch2 in gut Rorγt+ ILCs. Evidence for a role of Notch in AhR-mediated maintenance of Rorγt+ ILCs was provided by the observation that mice deficient for RBP-Jk, an essential partner of Notch, showed substantially reduced numbers of NKp46-expressing Rorγt+ ILCs and, although less prominently, of CD4+ Rorγt+ ILCs (LTi cells) also []. However, there were differences between the AhR- and RBP-Jk-deficient mice, in that in the latter, cryptopatches and ILFs were largely intact, whereas they were greatly reduced in AhR-deficient mice []. These data suggest that AhR mediates its effect on Rorγt+ ILCs only partly through Notch signaling; however, it should be remembered that Cre-mediated deletion of a floxed allele is rarely complete and thus there may be sufficient residual RBP-Jk present in the RBP-Jk-deficient mice for development of cryptopatches and ILFs. Notch signaling was found to be important for in vitro development of adult [] and fetal CLPs [] into RORγt+ ILCs. Interestingly, the latter study suggested a stage-specific requirement of Notch signaling in the development of RORγt+ ILCs as Notch signaling was required in an early stage of development of these cells but inhibited a subsequent step []. The relevant Notch for this role could be Notch2 [] but this has yet to be confirmed in in vivo experiments.
Rorγt+ cells in Ahr−/− mice express lower levels of the anti-apoptotic protein Bcl-2 and accordingly are more apoptotic []. Bcl-2 might be induced by the major cytokine receptors expressed on Rorγt+ ILCs, namely IL-7Rα and ckit; however, there are conflicting data with regard to the link of AhR and IL-7Rα. In one study, expression of IL-7Rα was decreased by AhR ablation [], whereas another group did not observe any change in IL-7Rα expression on Ahr−/– ILCs []. cKit, which is the receptor for stem cell growth factor, may be a direct downstream target of AhR since expression of this receptor is strongly decreased in Ahr−/− ILCs []. It is possible that the Rorγt+ ILC numbers are regulated by AhR in a cKit dependent manner. This suggestion comes from observations made in KitWv/Wv mice, which express a ckit variant with impaired kinase activity. These mice not only show diminished numbers of Rorγt+ ILCs, but also reduced numbers and sizes of CPs and ILFs. These findings strongly suggest that AhR regulates maintenance of RORγt-dependent ILCs by controlling ckit expression.
As in Th17 cells, AhR also appears to be required for optimal IL-22 production by the ILC22 population. The reduction of Rorγt+ ILC numbers in the gut, and the decreased capacity of these cells to produce IL-22, has functional consequences because AhR-deficient mice succumb to infection with C. rodentium and hydrodynamic injection of an IL-22-expressing plasmid into the tail vein reestablishes protection against C. rodentium []. In this setting, IL-23, produced by activated macrophages and DCs, controls IL-22 production by ILCs. Interestingly, AhR-deficient mice display reduced IL-23 receptor expression and IL-23 responsiveness []. It is likely that AhR directly controls IL-22 expression, as the Il22 locus contains multiple AhR-responsive elements []. Interestingly these elements are clustered with Ror-responsive elements and, in the Il22 locus, both Rorγt and AhR bind directly to their response elements. Whereas AhR recruitment to the well-known AhR target Cyp1a1 is unaffected by Rorγt, AhR binding to the Il22 locus is strongly enhanced by Rorγt [].
Regulation of RORα-dependent ILCs
Innate type 2 cytokine-producing lymphoid cells were first described in 2001 [[59, 60]]. Administration of IL-25 to mice elicited the release of high levels of IL-5 and IL-13 from a population of RAG-independent, γ-common-chain dependent, non-T, non-B innate lymphoid cells in the gut. Later studies identified several cell populations with similar, but not identical, phenotypes in various organs. These cell populations were lineage negative (Lin−) Sca-1+IL-7R+Thy1+T1/ST2+, and served as critical mediators of parasite expulsion in the murine intestine [[15, 61, 62]]. Transcriptional analysis revealed a number of transcription factors, including Id2, Notch1, Notch2, RORα and GATA3 [[6, 15, 61, 63]] that could potentially control the development and function of these cells.
Like NK cells and RORγt-dependent ILCs, development of type 2 ILCs depends on the transcriptional repressor Id2 [[4, 15]], suggesting, as discussed above, that they are derived from a common precursor; however, type 2 ILCs develop independently of RORγt, as Rorγt−/− mice exhibit numbers of type 2 ILCs comparable to those in wt mice []. Recently, it was reported that ILC2s could be generated from a bone marrow Lin−IL7Rα+Flt3+ CLP, differentiating under the influence of Notch1 signaling [[6, 64]] ILC2s failed to differentiate in mice with a spontaneous deletion in the gene for RORα, the so called staggerer (Rorasg/sg) mice []. In line with this observation, staggerer mice either injected with IL-25 or infected with the helminth parasite N. brasiliensis failed to either generate ILC2s or expel the parasites respectively.
GATA3 is highly expressed by ILC2s [[15, 63, 65]], and mice in which GATA3 was deleted only in IL-13-producing cells, of which the majority were ILC2s during N. brasiliensis infection, are phenocopies of IL13-deficient mice []. These mice exhibited reduced worm clearance, suggesting that GATA3 is critical for IL-13 production in ILC2s. Together, these findings emphasize the striking similarity between Th2 cells and ILC2s, with both cell types relying on GATA3 for their function. Collectively, the studies described in this section indicate that the development and function of ILC2s are controlled by several transcription factors including Id2, RORα, Notch1 and GATA3.
Therapeutic targeting of ILC-related transcription factors
ILC-related transcription factors are potential targets for therapy in those diseases in which ILCs play either a prominent detrimental or beneficiary role. Two recent papers describe the potent effects of RORγt antagonism in inhibiting Th17-cell differentiation and reducing the severity of experimental auto-immune encephalomyelitis (EAE)[[67, 68]]. First, Huh et al reported that digoxin, and the two synthetic, non-toxic, derivatives 20,22-dihydrodigoxin-21,23-diol and digoxin-21-salicylidene, inhibit the differentiation of mouse and human Th17 cells[]. Digoxin was shown to specifically inhibit RORγt transcriptional activity. As a consequence, digoxin treatment delayed the onset and severity of Th17-cell driven EAE. Solt et al. demonstrated very similar effects with the synthetic RORγt ligand SR1001, which prevented Th17-cell differentiation and ameliorated EAE []. In a model for inflammatory bowel disease, RORγt-dependent ILCs can mediate pathology []. Together these results suggest that the RORγt antagonist SR1001 may be utilised therapeutically to target pathogenic ILCs.
Interestingly, in addition to RORγt, SR1001 also inhibits the activity of the type 2 ILC-related transcription factor RORα [] This opens up the possibility of using ROR antagonists such as SR1001 in the treatment of type 2 ILC-related immune pathologies, including airway hyperreactivity in allergic asthma, as well as those mediated by RORγt-dependent ILCs. However, the application of ROR agonists and antagonists needs to be carefully assessed in view of the known beneficial roles of ILCs. Future work needs to reveal how RORα/γt antagonism affects ILC functions, and how this can be applied in the clinical settings.
In addition to RORγt and RORα, AhR plays a prominent role in the survival and function of the ILC22 population. The AhR agonist FICZ increases the number of intestinal IL-22-producing ILCs, cells that are crucial for clearing C. rodentium infection []. This role in the gut makes AhR an interesting target for the treatment of inflammatory bowel disease, a disease in which ILC-derived IL-22 plays a protective role [[28, 30]].
Conclusion and future directions
In summary, as discussed in this review, the transcriptional programs that govern the development of the various branches of the ILC family, including RORγt and RORα dependent ILCs, are beginning to be unraveled. Future studies should aim to address the precise requirements of specific transcription factors at different stages of ILC development and to unravel how these transcription factors are regulated, what the effects of antagonism are, and how the potential interactions between the various transcription factors affect ILC development and function. With such knowledge, attention can be turned to specific therapeutics based on regulating these family members.