Deconstructing deployment of the innate immune lymphocyte army for barrier homeostasis and protection

Summary The study of the immune system has shifted from a purely dichotomous separation between the innate and adaptive arms to one that is now highly complex and reshaping our ideas of how steady‐state health is assured. It is now clear that immune cells do not neatly fit into these two streams and immune homeostasis depends on continual dialogue between multiple lineages of the innate (including dendritic cells, innate lymphoid cells, and unconventional lymphocytes) and adaptive (T and B lymphocytes) arms together with a finely tuned synergy between the host and microbes which is essential to ensure immune homeostasis. Innate lymphoid cells are critical players in this new landscape. Here, we discuss recent studies that have elucidated in detail the development of ILCs from their earliest progenitors and examine factors that influence their identification and ability to drive immune homeostasis and long‐term immune protection.


| INTRODUC TI ON
The world of innate immune cells has greatly expanded in recent years. Broadly it includes innate lymphoid cells (ILCs) together with an array of unconventional lymphocytes such as γδ T cells, CD1restricted NKT cells, and mucosal-associated invariant T (MAIT) cells. ILCs are distinct from other newly described innate cells as they lack recombined antigen-specific receptors characteristic of B and T lymphocytes and many of the phenotypic lineage markers that define other immune cell subsets. Indeed, ILCs are enriched when the two genes Rag1 and Rag2 that regulate recombination machinery, selection, and diversity in other lymphocytes, are deleted. 1 Nevertheless, ILCs exhibit a number of features that are reminiscent of T cells implying that they may be the innate counterparts of adaptive lineages. 2 ILCs have generally been regarded to be an almost exclusively tissue-resident population found at the barrier surfaces such as the skin, lungs, and intestinal tract. 3 New evidence now suggests that colonization of tissues, replenishment, and rapid dissemination of ILCs depends at least partly on the capacity of these cells to move around the body in response to pro-inflammatory signals allowing them to fight infection and maintain immune homeostasis. Here, we discuss the specific transcriptional pathways that are essential to regulate the generation and maintenance of ILCs. We focus on how recent findings are reshaping our understanding of the complexity of homeostatic regulation at barrier surfaces forcing us to rebuild the rules by which we understand how the immune system operates.

| INNATE LYMPHOID CELL SUBS E TS
Innate lymphoid cells are a heterogeneous family of immune cells that have shed new light on the architecture of the immune response and our understanding of how immune protection is orchestrated. ILCs express germline-encoded receptors that enables them to respond rapidly to stimuli. In many cases, precisely how these receptors work has been unclear as little is known about the ligands activating the receptors. Recent evidence, however, suggests that NKp46 can recognize the cognate ligand complement factor P, 4 and NKp44 can recognize platelet-derived growth factor (PDGF)-DD produced by tumors, 5 highlighting additional crucial roles in recognizing soluble tissue components, in addition to recognition of pathogen-derived ligands [6][7][8][9] to protect against infections and to mediate tissue repair.
This feature allows them to deliver front line defense against the continual assault on the body from both foreign and commensal organisms as well as antigens derived from food and environmental sources.
Although we have only recently been readily able to dissect the diversity of ILC populations due to their rarity, NK cells, and lymphoid tissue-inducer (LTi) cells were discovered more than 30 years ago. This established their prototypical roles in tumor immunosurveillance (NK cells) 5,10 and in the formation of secondary lymphoid tissues (LTi cells) 11,12 during embryogenesis, respectively. Our understanding of this family has now greatly expanded with the discovery of new previously unrecognized members that have been classified into three main subsets: ILC1, ILC2, and ILC3s. 13 These groupings are largely aligned with effector T cells and are based on their expression of transcription factors and cytokine profiles.
This reveals a heterogeneous population that can be further separated into NK cells (which express CD49b, also known as DX5) and non-NK ILC1s (which express CD49a or VLA-1α). Both NK cells and ILC1 express the transcription factor T-BET (encoded by

Tbx21), but generally only NK cells express EOMES (encoded by
Eomesodermin, also referred as T-box brain protein 2). These factors are associated with IFNγ production and anti-tumoral activities.
NK cells and ILC1 also differ in their lifecycle as NK cells seem to continuously recirculate around the body while non-NK ILC1s appear to reside mostly in tissues such as the liver. In addition, it is likely that the specific tissues inhabited by ILC1 significantly influence their phenotype and function. For example, it has been shown that salivary gland ILC1 are phenotypically distinct from liver ILC1 or from intraepithelial ILC1. 14 ILC2s produce interleukin(IL)-5, IL-9 and IL-13 together with tissue repair factors such as amphiregulin. They are defined by their expression of the surface markers ICOS (Inducible T cell costimulator), KLRG1 (Killer cell lectin-like receptor subfamily G member 1), Sca1, ST2 (IL-33R), CD25 (IL-2Rα), and IL-7R together with the transcription factors GATA3 (GATA binding protein 3) and nuclear receptor RORα (RAR-related orphan receptor α). 15 Some variability in the expression of ST2, KLRG1, and CD25 has been observed depending on the tissue location and stimulus, 16 while in most tissues, ICOS is reliably expressed and indeed required for their survival and cytokine production. ILC2 are mainly involved in responses to allergic stimuli and parasites and are thus found at several sites throughout the body including the lungs, spleen, gut, liver, and skin. 3,17 ILC3 are characterized by the expression of RORγt and production of IL-22 and/or IL-17. They are found enriched in mucosal tissues such as the intestine and are demarcated into three distinct subpopulations by their expression of CD4 and the NK receptor, NKp46 (encoded by Ncr1, natural cytotoxicity triggering receptor 1). LTi cells, which orchestrate the generation of lymphoid tissues during fetal development, express the coreceptor CD4 together with the chemokine receptor CCR6, but lack NKp46 expression. 11,12 Two additional populations of ILC3 are defined by their lack of CD4 expression combined with their expression, or lack of, NKp46. Although the marker CCR6 has been used to divide ILC populations into "helper" and "cytotoxic" ILC populations, 18

| Development of early innate lymphoid progenitors in bone marrow
ILCs are thought to arise from all-lymphoid progenitors (ALPs) which contains the common lymphoid progenitor (CLP) and the IL-7Rα + multipotent ILC progenitors. 18,[22][23][24][25] The major progenitor potential lies within the α 4 β 7 fraction of the CLP. 26 Although all ILCs derive from an IL-7Rα + progenitor, an additional stage, termed the early innate lymphoid progenitor (EILP) has recently been defined and notably is marked by the expression of the transcription factor T cell factor-1 (TCF-1, encoded by the gene Tcf7). 27 Tcf7 + progenitors expressed only low levels of IL-7Rα, Zbtb16 (also known as Plzf), and Id2 (Inhibitor of DNA binding 2). 27 What was distinct about this cell type was that it did not fit with the known linear progression of ILC differentiation that had been previously described.
Distinct from other members of the progenitor network, the EILP did not express IL-7Rα. This was perplexing but such a step in ILC differentiation could occur if EILPs did not arise from the ALP; or alternately, ILC progenitors could transition through a stage that depended on the downregulation and subsequent re-expression of IL-7Rα as normally occurs in developing thymocytes ( Figure 1). 28 Thus, the EILP would represent an intermediate developmental stage in which IL-7Rα is transiently downregulated. Indeed, when the IL7rCre strain was crossed to a ROSA26-YFP reporter strain and the Tcf7 EGFP reporter, the temporal expression of Tcf7 and IL-7R amongst IL-7R + an IL-7R − cells could be ascertained. 29 Indeed, it was then clear that the IL-7R − population carried the imprint of previous IL-7R expression and that the EILP defines a critical step in ILC generation. Importantly, this work defined the link between the very early progenitor stages of the ALP and ILCP (ILC progenitor), and the EILP, and crucially pinpointed the requirement for differential regulation of receptor expression for this transition that may well have been normally overlooked ( Figure 1). 29 IL-7R expression is therefore highly dynamic and tightly regulated by TCF-1 30 resulting in early expression in development, but subsequently downregulated to allow the EILP to give rise to ILCP.

| The thymic pathway
Although ILCs in the adult typically originate from the bone marrow, emerging data points to an additional network that regulates thymic progenitors that are normally destined to establish T cell  35,36 Overexpression of ID3, which can function similarly to ID2, promotes the generation of NK cells from thymocytes, 37 while overexpression of ID1 in transgenic mice enhances the development of ILC2 in multiple organs, most notably in the thymus. 38,39 ID1 itself is not generally found in immune cells but ectopic expression of Id proteins, or the removal of their E protein binding partners, serves to reciprocally enhance their expression and drive ILC development. The thymus is not essential for the formation of ILC2 but they can be generated from thymic progenitors when they are cultured with IL-7 and IL-33 39,40 suggesting that the balance of innate and adaptive immune cell fate outcomes depends on the combination of transcription factors together with external stimuli encountered by cells. This effect has also been shown in vivo. 35,41 Similarly, deletion of transcription factors that F I G U R E 1 (A) Transcriptional regulation of ILC development from the common lymphoid progenitor (CLP) to mature ILC subsets 1, 2, and 3. It is now clear that the CLP transits through a series of intermediates including the early innate lymphoid progenitor (EILP) which in contrast with stages both preceding and following the EILP, downregulate the expression of IL-7R. (B) Differential regulation of transcription factors and surface receptors is both dynamic and essential for diversification of ILC subsets normally define T cell identify, such as BCL11b, can have the capacity to derepress the dominant T cell developmental pathway in the thymus resulting in NK cells development. 42 In addition, however, BCL11b can fine tune the balance between ILC2 and ILC3 in the periphery to act as a sensitive rheostat for ILC subset development in response to stimuli. 43,44 Within the thymus, maintaining the delicate balance of ILC subsets appears to be crucial for the integrity of the thymus and the emergency generation of ILC.

| NK cells
Several NK cell progenitors have been identified to give rise to different peripheral NK cell subsets. NK cells are guided through progressive developmental stages by the expression of specific transcription factors necessary for NK lineage commitment and maturation. The prepro NK cell precursor population is the earliest identified committed NK cell progenitor that shows a highly enriched capacity to generate NK cells. 55 These cells express the IL-2Rβ chain (CD122), Sca-1, IL-7Rα, and ID2 but lacked markers typically expressed on fully differentiated mature NK cells such as NK1.1, NKp46, and CD49b. 55 In the bone marrow, this progenitor further committed into NK progenitor (NKP) 55,56 and subsequently into immature and mature NK cells under the influence of a core transcription program including but not restricted to Id2, 55 Gata3, 57 Nfil3, 58 Klf2, 59 Eomes, 60 Tbx21, 61 Tcf7, 27 Tox, 62 and Ets-1 ( Figure 2). 63 Interestingly, NFIL3 expression is essential for the development of all ILC subsets 64 but appears to be only required early and transiently to implement the ILC program as Nfil3 deletion in ID2 + mature NK cells does not affect NK cell survival or function. 26,50,65 Immature and mature NK cells are present in peripheral tissues and blood and are characterized by the expression of NK1.1 and NKp46. CD11b and CD27 expression subdivides NK cells into immature (Imm, CD11b −/low CD27 + ), mature 1 (M1, CD11b + CD27 + ) and mature 2 (M2, CD11b + CD27 − ) subsets. 66 These mature subsets may express other surface molecules, such as KLRG1, 67 CD62L, 68 and DNAM-1 69 that further characterize their specific phenotype and functions. In parallel with the progressive change in surface molecule expression during maturation, NK cell function is also affected.
Most strikingly, mature NK cells become less proliferative and produce less cytokine but conversely, they gain cytotoxic function as they further mature from M1 into M2 populations. 70

| ILC1
ILC1 originate from the innate lymphoid cell progenitor (ILCP) which also gives rise to other subsets of ILCs such as ILC2 and some ILC3. 71 ILC1 express NK1.1 and NKp46 but are distinct from NK cells. They generally lack CD49b or EOMES expression but depend strongly on T-bet expression 72 in contrast with splenic NK cells which only partly rely on this factor. 18,73,74 In addition, while BLIMP1 is required to fully upregulate T-BET expression in splenic NK cells, genetic deletion does not result in a marked defect in NK cell populations. 75 Nevertheless, an unexpected synergy between BLIMP1 and HOBIT (homolog of BLIMP1 in T cells or ZNF683), a transcription factor normally controls tissue-residency in CD8 + T cells, is essential for their development IL-12 responsiveness, or IL-12 itself, also acts as a potent driver for this subset to acquire features of ILC1 cells. [85][86][87][88] Thus, in inflammatory settings each subset appears to be able to reprogram its capability to become IFNγ-producing cells with potent effector functions.

| ILC2
The development of ILC2 is guided by the core transcriptional regulators RORα, 40

| ILC3
RORγt is the cornerstone transcription factor identified as essential for the development of ILC3 98  These two cornerstone studies opened the way to begin to tease apart the precise machinery that guides the development of ILC3, particularly NCR + ILC3, revealing that a number of transcription factors are essential for the development of these cells. These include the aryl hydrocarbon receptor (AHR) which is regulated in ILC3 by RUNX3 79 and is sensitive to signals derived from the microbiota and dietary components that generate aryl hydrocarbons. 101-104

| The complexity of the NK cell and ILC1 subsets and consequences of disruption of NKp46 signaling pathway
Our ability to ascertain the functions of NK cells has relied heavily on their identification as NK1.1 + cells. NK1.1 + cells have been attributed the important defense mechanism of immunesurveillance protecting from the emergence of cancer. [105][106][107] With the discovery of ILC subsets and tracking of cells using the NKp46 receptor in combination with transcriptional regulators, it has become clear that the classical NK cell compartment contained not one, but two subsets of cells-ILC1 and NK cells. Thus, it could no longer be assumed that all the functions credited to NK cells were in fact due to NK cells alone but instead may reflect the outcome of a mixed population of cells that also contained ILC1. Further complicating the interpretation of the data attributable to individual subsets was the emergence of plasticity between a number of ILC subsets implying that alternate flexible programs that did not neatly fit into the subset classification could be identified. In an unexpected twist, recently three groups identified a point mutation in the Ncr1 gene in the congenic CD45.1 + mice used for many studies of lymphocyte tracking and function analysis ( Figure 3A, Table 1 Table 1).
The mutant mice, referred to as Ly5.1 C14R , showed normal numbers of NK cells and ILC1 but a modest alteration to the immature and mature NK cell subsets. 81 To date, several models affecting Ncr1 expression have been generated. These include the Ncr1 gfp/ gfp , 108 Ncr1 iCre/iCre , 109  Nevertheless, the loss of stable expression of NKp46 on the surface of cells has been shown to be broadly important and crucial for protection from influenza virus 82 and tumor control 81 but paradoxically appears to confer higher resistance in MCMV infection. 82 In some cases, such as influenza, the hemagglutinin and neurominidase viral proteins have been purported as endogenous ligands 6,108,114 while B16F10 melanoma cells are known to be controlled by NK cells. 106,115 However, the resistance to MCMV infection identified in this study was a surprise and was correlated with enhanced IFNγ expression. 82 Increased IFNγ production was attributed to the reduction in expression of the gene Dok1 which has previously been proposed to augment IFNγ. 116 However, enhanced IFNγ expression was not observed in all studies of mice carrying the C14R mutation. 81 Using exome sequencing, more than 300 genes were found to differ between Ly5.1 C14R mice and their Ly5.2 counterparts. 81 It is therefore likely that many associations between NKp46 expression and function will emerge and that the newly developed NCR B6C14R mouse, generated on an C57BL/6 background and carrying only the NCR mutation without disruption of other genes, will be an important tool in future studies. 81 Despite this, the total loss of NKp46 expression, or unstable expression, was associated with the loss of TRAIL in both NK cells and ILC1. 80,81,113 Similar to NKp46, TRAIL was transcribed but was unable to migrate to the cell surface in the absence of NKp46. 113 This might occur if NKp46 and TRAIL comprise a single protein complex in cytoplasmic vesicles although the mechanism of release is not yet known. Collectively, these studies highlight the confounding nature of some earlier studies and the necessity to systematically ascertain the roles of NK cells and ILC1 in models where genes affecting function are not unknowingly disrupted.

| ILC plasticity and the common default pathway
The broad subsets of ILCs have, through the development of elegant and novel tools and vigorous investigation, been relatively wellelucidated. However, many questions remain around the programs that define each subpopulation, as well as the cellular and molecular triggers that allow so called "plasticity," or the capacity to adopt a

| IMMUNE HOMEOS TA S IS AT MUCOSAL SURFACE S: ILC NE T WORK S IN THE G UT
The mucosa is colonized by the bulk of immune cells found in the body. These cells sense information from intestinal contents such as the trillions of microbes that inhabit the gut and food components.
This landscape poses considerable challenges to maintain health. To that end, the immune system is charged with the task of balancing responses to maintain mucosal homeostasis. Fending off invading pathogens is clearly important, but maintaining immune homeostasis at these highly vulnerable surfaces is perhaps the single most important function that prevents succumbing to disease. In both the gut and the lung, the epithelium physically separates microbes from the immune cells but a constant dialogue between these compartments drives the integration of signals that guides homeostasis.
For example, in addition to physical interactions between microbes and immune cells, it has been uncovered that metabolites generated by microbes provide essential signals to immune cells in the hostmicrobiota homeostatic network.  At steady-state, slow proliferation of ILC within the tissues themselves allows the balance of subsets to be maintained. In an acute transient infection this may also be the case and that any temporary depletion would be rapidly replaced through enhanced local proliferation. If inflammation continues this might result in depletion that is not readily overcome by local proliferation and could lead to a state of "exhaustion" akin to that exhibited by T cells. Alternately, inflammation may drive differentiation of bone marrow progenitors and export in the blood to the affected tissues the induction of type 2 immune responses following parasite infection. IL-25 is essential to drive the amplification of ILC2 131 but until recently, the exact cell type that produced this cytokine was unknown. Analyses of intestinal epithelial cells revealed that only a very small proportion of cells, the Tuft cells, produced IL-25. [132][133][134] Establishing this exclusivity was facilitated by generation of an IL-25

| Maintaining ILC at mucosal surfaces
reporter mouse line. This mouse also revealed that Tuft cells were not the source of other important epithelial cytokines such as IL-33 and thymic stromal lymphopoietin (TSLP) that can also activate ILC2. 35 Elucidation of this pathway is exciting and prompts us to ask whether other novel cell types found in the intestine, which as yet relatively poorly characterized, might also contribute to maintaining the ILC network and the elegant cooperation between epithelial and immune cells that drives homeostatic balance between ILC2 and ILC3.

| Tissue residency and circulation
Except for NK cells which are mostly circulating, 135,136 ILCs have generally been thought to be largely restricted to the tissues in which they are found, having established their niche early in ontogeny. 137,138 This view of ILCs is predicated on several pieces of evidence including (a) ILCs are poorly replaced following transplantation, 137,139 (b) mature ILCs do not appear to exchange between mice in which the circulatory system is conjoined in models of parabiosis, 137 and (c) few ILC that express a mature phenotype are found in the bone marrow. 18,71 In mice ILC replacement is extremely poor following exposure to lethal doses of γ-irradiation, 137 and in patients who display mutations in the common γ chain cytokine receptor subunit IL-2Rγ, or the tyrosine kinase JAK3, tissue ILCs fail to be effectively reconstituted. 139 In part, this has been attributed to the extremely slow turnover of ILCs. 140  Recent studies indicate that circulating ILCs display their own unique molecular program. This appears to be distinct from ILC phenotypes that have been previously described or that might be predicted. 146  is necessary for the accumulation of NK cells in the liver. 155 Thus, multiple organs harbor a significant reservoir of NK cells separate from those found in the blood and under certain physiological conditions such as pregnancy, 156 or atopic or contact dermatitis, 157,158 these are massively expanded.
From an evolutionary perspective, the notion that ILCs might only be replenished from local sources would leave the body extremely vulnerable-ILCs would be exposed to depletion by a severe highly acute infection, or more damaging long-term by a chronic infection without a mechanism to quickly deploy progenitors, or differentiated cells, to replace these cells. Local proliferation could provide some protection, but this is likely to be limited and an infec-

| ILC3 are essential to maintain immune homeostasis
ILC3 are highly enriched in the gut mucosal tissues and rapidly respond to the cytokine milieu elicited by the colonization of microbes.
Often, we view the role of these cells through the lens of driving immune protection. It is, however, the ability to maintain immune homeostasis that is one of the most fundamental aspects that ensures our health. This requires the capacity of the barrier tissues to continually adjust to unpredictable conditions at those surfaces and to integrate signals from the bacterial communities, epithelial cells, and immune cells. How then do ILCs, particularly ILC3s, participate in orchestrating this type of barrier defense is not well-understood yet.
LTi cells in the embryo establish the sites at which lymph nodes and mucosal-associated secondary tissues develop. 16 These CD4 + CD3 − cells were first discovered in 1997 while a related population termed LTi-like cells have been identified in the cryptopatches of mice. 160 This population interacts with B cells to promote IgA production 161  has been implicated in the recruitment of Arid1a to the Ahr promoter to activate AHR expression. 171 This expansion drives the production of IL-22 by ILC3 which is essential for fucosylation of gut epithelial cells via the induction of the fucosyltransferase, Fut2. 172 IL-22 production also promotes the production of the epithelial derived antimicrobial peptide RegIIIγ which is essential for the control of enteric infections such as Citrobacter rodentium. 120,169,173 In C. rodentium infection, loss of IL-22 produced by the NKp46 + ILC3 subset does not in itself compromise the capacity to control bacterial colonization as IL-22 production can be maintained through the NKp46 − ILC3 subset. 174 This raised the notion that innate and adaptive immune cells are highly redundant and challenged our understanding of how overlapping cell types might contribute to maintaining gut homeostasis. The loss of CD4 + T cell input, however, resulted in prolonged phosphorylation of STAT3, an activation step that is normally only induced transiently in response to microbial colonization. 175 Thus, the absence of CD4 + T cells uncouples the functionality of NKp46 + ILC3 and this cannot be retrieved by sustained

AUTH O R CO NTR I B UTI O N S
All authors researched data for the article, made substantial contributions to discussions of the content, wrote the article, and reviewed and/or edited the manuscript before submission.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no competing interests.