Effective Barriers: The Role of NKT Cells and Innate Lymphoid Cells in the Gut

Intestinal homeostasis is dependent on the bidirectional dialogue between the host intestinal immune cells and commensal bacteria (1). Typically, there is a symbiotic relationship between the intestinal cells and the microbiota. However, the intestinal microbiota is highly varied between individuals, and dysregulation of commensals can have pathological consequences, such as inflammatory bowel disease (IBD). Many studies have focused on the role of Th17, regulatory T cells (Tregs), and classic T cell subsets in intestinal immunity. NKT cells and innate lymphoid cells (ILCs) have been implicated in the regulation of mucosal immunity, but their roles are less well understood, and thus this review focuses on their roles in intestinal immunity.

NKT cells are a distinct population of T lymphocytes that display features characteristic of NK cells and conventional T cells (2–4). NKT cells are CD1d restricted and can be broadly delineated into two subtypes. Type I invariant NKT (iNKT) cells express an invariant TCR α-chain encoded by Vα14Jα18 in mice and Vα24Jα18 in humans, and a limited number of TCR β-chains. Type I iNKT cells are activated by the lipid α‐galactosylceramide (α-GalCer). In contrast, type II NKT cells, which express diverse αβ TCRs, recognize different lipid Ags (5) and are not activated by α-GalCer (Table I). Type I iNKT cells have been studied extensively and can be further divided into subsets similar to CD4 T helper subsets, namely NKT1 cells are T-bet⁺ and produce high levels of IFN-γ, NKT2 cells are PLZF^high and produce IL-4 and IL-13, and NKT17 cells are retinoic acid-related orphan receptor (ROR)γt⁺ and secrete Th17-type cytokines (6, 7). NKT1 cells represent most type I NKT cells in liver and spleen in C57BL/6 mice and primarily express NK1.1 and IL-12R. NKT2 cells are the most enriched type I NKT cell subset in BALB/c mice, and they are the primary subset present in the lungs and intestines of C57BL/6 mice. NKT17 cells are mainly present in the lung, skin, and lymph nodes. Although iNKT cells express a semi-invariant TCR α-chain, they recognize a diverse array of lipid Ags. Additionally, most studies that focus on α-GalCer–reactive NKT cells are describing iNKT cells; however, α-GalCer–reactive, CD1d-restricted NKT cells that use different TCR α-chains have been identified in mice (8) and humans (9–11).

In contrast to classic T cell subsets, iNKT cells establish long-term residency within the tissues and do not circulate (12–14). In mice, most iNKT cells express CD4⁺ or are CD4⁻CD8⁻ double negative (15), and in humans iNKT cells express CD4⁺, CD8⁺, or are double negative (16–19). In addition, NKT cells are phenotypically similar to effector T cells due to their expression of non-lymphoid tissue homing chemokine receptors such as CCR2, CCR5, and CXCR3. Additional subsets of iNKT cells have been identified following their activation in vivo. Specifically, there is a small population that, similar to follicular helper T cells, secrete IL-21, support germinal center development, and express Bcl-6, referred to as iNKT_FH cells (20–22). In addition, most iNKT cell subsets express PLZF; however, there is a regulatory iNKT cell population, NKT10 cells, that expresses E4BP4 (23, 24).

NKT cells can be activated through TCR-dependent and independent mechanisms, in response to self and foreign Ags (25, 26). In fact, CD1d molecules present Ags derived from allergens, bacteria, and fungi to iNKT cells. In addition, danger signals such as those induced by TLR signaling can lead to the induction of CD1d-mediated NKT cell responses to self-antigens. Moreover, cytokines such as IL-12 and IL-18 can drive NKT cell activation. Following activation, NKT cells rapidly produce Th1-, Th2-, and Th17-type cytokines (27, 28), and they have been shown to play a critical role in the regulation of intestinal immunity (29–31).

iNKT cells are located within the intestines in the absence of microbiota (32–34). There are differences in mucosal NKT cell subpopulations in mice and humans. Whereas iNKT cells are present at high frequencies in the fetal intestine in humans, this population is absent at birth in mice (32, 35, 36). However, the population of murine intestinal iNKT cells increases during the first few weeks of age (35, 36). CD1d is expressed by numerous cells within the intestines, including intestinal epithelial cells, ILCs, B cells, dendritic cells (DCs), and macrophages, and CD1d expression has been shown to be a key regulator of intestinal inflammation.

A recent study demonstrated a critical role for vitamin A in regulating the mucosal iNKT cell population in mice (37). Vitamin A served as a negative regulator by inducing the P2X7 receptor, resulting in the death of tissue-resident NKT cells, which was important for intestinal homeostasis. Moreover, crosstalk between the NKT/CD1d axis and the colonization of commensal bacteria is evident, as Nieuwenhuis et al. (38) found that CD1d-deficient mice had an increased bacterial burden following exposure to Pseudomonas aeruginosa. It was shown that CD1d is required for intestinal colonization of both Gram-positive and Gram-negative commensals. To further establish a role for CD1d-dependent NKT cells, mice were injected with α-GalCer and Paneth cells degranulated in WT mice, but not in CD1d-deficient animals (38). Interestingly, in the absence of CD1d, there was a defect in the morphology and function of granules containing antimicrobial peptides.

Cell membranes are largely composed of phospholipids, and many of these ubiquitously expressed lipids have been reported to be CD1d-restricted Ags (39). Both type I and type II murine NKT cell hybridomas recognize phosphatidylinositol, phosphatidylethanolamine, and phosphatidylglycerol (40). Mammalian phospholipids differ from their microbial counterparts in the fatty acyl chain composition or the position of the fatty acyl molecules at the sn-1 and sn-2 positions of the glycerol backbone, which can affect iNKT cell responses (39).

As mentioned above, the first α-GalCer Ag shown to activate iNKT cells was isolated from a marine sponge, Agelas mauritianus (41). Then, a gut commensal, Bacteroides fragilis, was shown to contain an α-GalCer, namely α-GalCerBf (42), that activates both human and mouse iNKT cells. Following these studies, Bacteroides vulgatus and Prevotella copri were demonstrated to produce α-GalCer (43). In later studies, this same group performed a screen to identify additional Ags by examining the intestinal tracts of mice (44). They observed 1–15 pmol of α-GalCer per milligram of protein in the cecum and the colon, which was absent in germ-free mice. In addition, they found that diet, inflammation, and infection could negatively impact α-GalCer levels (44).

Commensal bacteria have been shown to regulate NKT cell homeostasis. Elegant studies by Olszak et al. (32) examining iNKT cells in germ-free mice showed that these cells were significantly elevated in the colonic lamina propria and lungs, compared with specific pathogen-free mice. The increase in iNKT cells was correlated with higher levels of IL-1β and IL-13 and with increased susceptibility to oxazolone-induced colitis and OVA-induced allergic lung inflammation. Importantly, restoration of NKT cell homeostasis was age/time-dependent, as colonization of neonatal mice with conventional microbiota was protective against mucosal iNKT accumulation and related pathology. However, colonization of adult germ-free mice did not protect against colitis-induced mortality and lung inflammation. These data suggest that microbial regulation of intestinal iNKT cells occurs early and persists. In addition, the homing of iNKT cells to the mucosa was found to be CXCL16-dependent. In contrast, studies by An et al. (36) found that monocolonization of germ-free mice with WT B. fragilis, but not a mutant strain deficient in serine palmitoyltransferase, restored mucosal iNKT cell levels without inducing CXCL16. These data and others indicate that there are multiple levels of regulation, as studies by Wieland Brown et al. (42) demonstrated that microbial-derived CD1d-binding lipids from the same organism vary in their ability to induce NKT cell activation and proliferation. In contrast to iNKT cells, the regulation of type II NKT cells by commensal microbiota is poorly understood.

NKT cell Ags are present in the bacteria found within the intestinal microbiota, and several CD1d-restricted mucosal Ags have been identified. These include glycosphingolipids from Sphingomonas spp., glycolipid Ags from Borrelia burgdorferi, and diacylglycerols from various pathogenic bacteria (45–48). There are differences in mouse and human iNKT cell responses to α-galactosyl diacylglycerolipid Ags derived from B. burgdorferi (46). Mouse iNKT cells are best activated by BbGL-2c, whereas human iNKT cells are more responsive to Ags with a high number of unsaturated fatty acids, such as BbGL-2f (46). Streptococcus pneumoniae expresses an α-linked diacylglycerol that can activate both mouse and human iNKT cells (49). Lipid extracts from Helicobacter pylori have been reported to induce iNKT cell proliferation (50). In addition to bacterial Ags, a CD1d-specific Ag has been identified for Candida albicans (51). Another fungal glycosphingolipid Ag, asperamide B, has been shown to activate both mouse and human iNKT cells in a CD1d-dependent, MyD88-independent manner (52).

NKT cells play an integral role in intestinal inflammation. IBDs, such as Crohn’s disease (CD) and ulcerative colitis (UC), are characterized by aberrant inflammation within the gastrointestinal (GI) tract (53). In an oxazolone colitis model, which is a murine model of UC, NKT cells were found to induce disease through their production of IL-13, as disease could not be induced in NKT cell–deficient animals (54). In contrast, in a dextran sulfate sodium–induced colitis model, IL-9–producing iNKT cells were found to be protective due to their suppression of IFN-γ and IL-17A production. Collectively, these data suggest that iNKT cells are beneficial in Th1 models, but they can promote disease in Th2 models of intestinal inflammation (30).

In humans, type II NKT cells comprise a large population of cells in the UC lamina propria (55). Fuss et al. (55) stained lamina propria mononuclear cells (LPMCs) with lysosulfatide-loaded tetramer and found that LPMCs from UC patients bind sulfatide glycolipid–loaded CD1d tetramer. Moreover, UC LPMCs produced IL-13 following stimulation by sulfatide glycolipid in a CD1d-dependent manner. To investigate the role of type II NKT cells in colitis, Liao et al. (56) generated a transgenic mouse model that expressed high levels of CD1d and an autoreactive, type II NKT cell TCR (CD1dTg/24αβTg mice). Notably, the double-transgenic CD1dTg/24αβTg mice spontaneously develop IBD, in contrast to single-transgenic CD1dTg or 24αβTg mice. These data suggest that high CD1d expression increased the numbers of type II NKT cells, which drives disease progression. Moreover, the authors found that treatment with broad-spectrum antibiotics ameliorated disease in CD1dTg/24αβTg mice (56). Overall, these studies suggest a pivotal role for the intestinal microbiota in promoting NKT cell–induced inflammation, which is in contrast to the suppressive role that commensals play in regulating iNKT cell expansion.

The link between chronic inflammation in IBD patients and increased risk for colorectal cancer has been well established. As described above, NKT cells can drive intestinal inflammation in IBD, and thus their role in the development of intestinal tumors has been investigated (57, 58). Studies by Cardell and colleagues (57, 58) using two different iNKT cell–deficient mouse strains found that iNKT cells promoted the development of spontaneous intestinal polyps in Apc^Min/+ mice, a model for early stage human colorectal cancer. Interestingly, in a human study of colorectal cancer, an increased infiltration of Vα24 NKT cells was found in colorectal carcinoma cases, compared with controls, and NKT cell infiltration in colorectal carcinomas was correlated with better outcomes (59).

ILCs were identified relatively recently as lineage-negative cells that mount rapid responses against pathogens and contribute to the regulation of adaptive immunity (60). Present at relatively low frequencies in peripheral blood, ILCs are primarily tissue resident and enriched at mucosal barrier sites (61–63), where they contribute to tissue homeostasis and repair (64). Initially categorized in three groups based on master regulators and effector cytokines (65), ILCs were later reclassified in five distinct populations based on developmental stages, that is, NK cells, ILC1s, ILC2s, ILC3s, and lymphoid tissue inducers (60) (Table II). ILC1s, ILC2s, and ILC3s are defined helper ILCs because they are regarded as the innate functional equivalent of T helper subsets. ILC1s, similar to Th1 cells, express T-bet, and their responses are elicited by IL-12 in combination with IL-2, IL-18, or IL-15 (66–68). The ensuing secretion of IFN-γ and TNF-α enables ILC1s to respond to intracellular pathogens (67, 69, 70). ILC2s, the functional equivalent of Th2 cells, are essential for the control of intestinal parasites, but they also take part in allergic inflammation, particularly in the lung (71–76). They express the transcription factor GATA3 and, upon activation by IL-25, IL-33, or stromal lymphopoietin (TSLP), they produce IL-4, IL-5, IL-9, IL-13, and amphiregulin (73–78). ILC2s can also produce IL-10 when activated in the presence of specific soluble factors (79–82), and they appear to be the main source of IL-10 among helper ILCs in murine intestine (79). A controversy is still ongoing about the existence of IL-10–producing ILCs identified as a separate intestinal subset, the regulatory ILCs (83). ILC3s, similar to Th17 cells, express RORγt and produce IL-17A and IL-22 in response to IL-23 and IL-1β stimulation (84–88). They are key for responses to extracellular bacteria and fungi, as well as maintenance of tissue homeostasis (87).

In the GI tract, each helper ILC subset has a specific distribution, with differences between species (61, 63, 89, 90). In humans, lamina propria CD127⁺ ILC1s are most abundant in the duodenum and tend to decrease going toward the colon, whereas ILC3s increase in numbers from the upper GI tract toward the colon (61, 63). NKp44⁺ ILC3s, which represent the main producers of IL-22 among the ILC3 subsets (91), are the most abundant helper subset in the human intestine (612), whereas ILC2s are barely detectable (61, 63, 92). GI infection and inflammation are associated with significantly altered distribution of helper ILCs, indicating a role for these populations in both pathogen control and inflammatory diseases. We will focus on the role of helper ILCs in intestinal infection, inflammation, and cancer.

ILCs in the intestine play an essential role in the first line of defense against multiple pathogens by quickly releasing soluble factors and regulating subsequent adaptive responses. ILCs, in particular ILC3s, are also needed for an optimal symbiotic relationship with the microbiota, as discussed in detail elsewhere (93). Lack of ILCs leads to dysbiosis (94) and peripheral dissemination of commensal bacteria, which can be prevented by IL-22 administration (95), as this cytokine induces secretion of antimicrobial peptides (RegIIIb, RegIIIg, and the S100 family) (96). In contrast, several studies showed fewer ILC3s, or defective ILC functions, in germ-free mice (97–99). These defects could be linked to the fact that ILC function is directly regulated by multiple metabolites (100, 101), including short-chain fatty acids (97, 102), whose levels change in relationship to microbiome composition.

The microbiome also regulates ILC function via other cell types, including epithelial cells, macrophages, and DCs. Human ILC1-mediated production of IFN-γ upon exposure to gut commensals or pathogens depends on myeloid DC release of IL-12p70, IL-18, and IL-1β (103). The microbiota can regulate ILC2 activity via mediators produced by intestinal epithelia cells (TSLP) or plasmacytoid DCs (type I IFNs) in a positive or negative way, respectively (104, 105). ILC3-mediated production of GM-CSF is dependent on the ability of colonic murine macrophages to sense microbial signals and produce IL-1β (106). ILC3s are the most important for the maintenance of intestinal barrier integrity, including tissue repair after infectious or inflammatory damage. ILC3-derived IL-22 promotes epithelium survival and proliferation, mucus production, and increased expression of antimicrobial peptides (95, 107–110). This subset is critical against several bacterial and fungal infections.

Citrobacter rodentium infection in mice is a widely used model to mimic human attaching-and-effacing Escherichia coli infections and chronic intestinal inflammation (111). Whereas bacterial clearance requires adaptive immunity (112), ILC3-secreted IL-22 is essential in the initial phase of C. rodentium infection (99, 110), and multiple studies confirm the protective role of ILC3s (85, 95, 99, 101, 113). Several mechanisms have been described to induce or regulate ILC3 activation during C. rodentium infection, including interaction with mononuclear phagocytes (111–118), engagement of ILC3 receptors, such as G protein–coupled receptors and the aryl hydrocarbon receptor (AhR) (97, 113, 119–122), and sensing of metabolites such as retinoic acid, vitamin D, and short-chain fatty acids (102, 123–129). Among the ILC3 receptors, AhR, a ligand-dependent transcription factor sensing environmental cues and endogenous compounds generated by commensal, dietary, or cellular metabolism (130, 131), is recognized as a key regulator of ILC3 function (101, 113, 121, 132). Additionally, it was demonstrated that ablation of Ahr enhances intestinal ILC2 function and antihelminth immunity, whereas activation of AhR suppresses ILC2 function, suggesting that the intestinal ILC2/ILC3 balance is maintained by engaging the AhR pathway to mount adequate responses against distinct pathogens (132). Two other receptors known to regulate ILC3 responses to C. rodentium are the NK molecule NKR-P1B (133, 134) and GPR183 (119, 120), whose genetic ablation results in increased susceptibility to infection (119, 134).

In terms of phagocyte-mediated activation of ILC3, multiple requirements have been identified for myeloid phagocytes to promote IL-22 secretion. Intestinal macrophages, by releasing IL-1β and IL-23, promote ILC3-mediated production of IL-22 and antimicrobial peptides such as lectins of the RegIII family (115, 116, 135). Cross-talk via CXCL16 (on macrophages) and CXCR6 (on ILC3) is needed for an effective immune response to C. rodentium infection. Similarly, β₂ integrin expression on intestinal macrophages was necessary for macrophage-derived IL-1β release and optimal ILC3-mediated production of IL-22. Finally, ILC3–macrophage cross-talk, mediated by ILC3-derived GM-CSF, results in the differentiation of M1 inflammatory macrophages and the establishment of a positive feedback loop that augments ILC3 activation and Th17-type immunity (136). While this feedback loop enhances antibacterial responses, it can lead to chronic inflammation, as GM-CSF inhibits wound healing and tissue repair programs (136). Following C. rodentium infection, mice with intestinal epithelial cell–specific ablation of IKKα displayed impaired ILC3-mediated production of IL-22 and antimicrobial peptides, severe intestinal inflammation, elevated bacterial dissemination, and mortality (137). Treatment with IL-22 or transfer of ILC3s obtained from WT mice restored protection (137).

Salmonella typhimurium is a facultative intracellular bacterium that causes gastroenteritis in humans, but in mice it only establishes infection after antibiotic treatment (138). Multiple studies have demonstrated that IL-22 is needed for protection against S. typhimurium, as it contributes to shaping an adequate microenvironmental niche for commensal bacteria (139, 140). However IFN-γ, produced by both ILC1s and CCR6⁻T-bet⁺RORγt⁺ ILCs (141, 142), is also required for protection, as it promotes the release of mucus-forming glycoproteins that shield the epithelial barrier during infection (143, 144). The antimicrobial function of ILC3 during S. typhimurium infection is regulated by Runt-related transcription factor 3 (Runx3) (142). S. typhimurium infection induced RORγt⁺T-bet⁺ ILC3 accumulation in the mesenteric lymph nodes, where they migrated from the colon, displaying elevated IFN-γ production (145). The inducible expression of T-bet in CCR6⁻RORγt⁺ ILCs, which appear to have both ILC3 and ILC1 features, is consistent with data on human ILCs (67, 146), and it suggests that ILC plasticity enables a switch from a homeostatic to a proinflammatory phenotype in response to infections.

Although ILC1s are the most important ILCs for protection from Toxoplasma gondii (69), Ahr^−/− mice, characterized by a pronounced ILC3 defect, following infection had more marked intestinal damage than did WT mice, possibly due to exaggerated T cells responses (147), indicating that in this model ILC3 may indirectly mitigate intestinal damage. ILC3s are also involved in protection against fungi. They were identified as the major source of IL-17 in response to C. albicans and essential for IL-17–mediated protective immunity in the oral mucosa in a murine model of oropharyngeal candidiasis (148).

ILC1s protect against a broad range of intracellular pathogens, including bacteria, viruses, and protozoa. Clostridium difficile infection can cause severe colitis and diarrhea when the normal microbiota is perturbed after antibiotic treatment (149). ILCs appear to protect against C. difficile because Nfil^−/− mice, which lack all ILC subsets, succumbed within 3 d, while WT and Nfil3^+/− heterozygous mice recovered (150). Consistent with these results, C. difficile infection caused an upregulation of both ILC1 (IFN-γ, TNF-α, NOS2)- and ILC3 (IL-22, IL-17A RegIII)-derived soluble factors in the intestine of Rag1^−/− mice, but not in Rag1^−/−Il2rg^−/− mice (which lack all ILCs) (151). Rag1^−/−Il2rg^−/− mice displayed increased susceptibility to C. difficile, quickly succumbing to the infection; ILC transfer restored protection from infection (151). Both Rag1^−/−Tbx21^−/− and Rag1^−/−ifng^−/− mice displayed elevated C. difficile susceptibility and lethality compared with Rag1^−/− mice, suggesting that protection against this pathogen is mediated mostly by the ILC1 subset (151).

ILC1s play a pathogenic role in a murine model of enterocolitis driven by Campylobacter jejuni. This common human pathogen causes gastroenteritis and may lead to long-term intestinal dysfunction such as postinfectious immune-mediated irritable bowel syndrome (152, 153). The intestine of infected mice displayed an accumulation of IFN-γ⁺T-bet⁺ ILCs, which correspond to ILC1s derived from fate mapping RORγt⁺ progenitors (154). Adoptive transfer of this population promoted pathology, whereas T-bet inactivation in this subset mitigated disease, consistent with the concept that IFN-γ can drive intestinal inflammation (154).

ILC1s are the key ILC subset for protection against the intracellular protozoan T. gondii. Intestinal ILC1s produce most IFN-γ as well as TNF-α and quickly attract monocytes at the site of infection. T-bet^−/− mice produce no IFN-γ in response to T. gondii, fail to recruit inflammatory monocytes, and control parasite replication; however, transfer of WT ILC1s into Rag2^−/−il2rg^−/− mice restores infection control (69). In addition to ILC1s, T. gondii infection also induced accumulation of NK-derived ILC1-like cells that are distinct from both steady-state NK cells and ILC1s in uninfected mice (155).

ILC2s are critical in the early phase of immune responses against parasitic helminths, and a few worms are known to elicit their responses (156–162). Infection with Nippostrongylus brasiliensis is a widely used, well-described experimental system for ILC2 antiparasitic function in the intestine. After helminth infection, ILC2s rapidly release abundant IL-13, driving goblet cell hyperplasia, smooth muscle hypercontractility, mucus production, eosinophil recruitment, and, ultimately, worm expulsion (73, 74, 77, 163, 164). Multiple studies confirmed the essential role of IL-13 in worm clearance, using a combination of gene-ablated mice and adoptive transfer experiments (73–76). While considered mostly tissue resident, ILC2s can engage in trans-organ migration. Inflammatory ILC2s, induced from resting intestinal ILC2s by IL-25 or helminth infection, migrated from the intestine to diverse sites based on sphingosine 1-phosphate (S1P)–mediated chemotaxis (165). The inflammatory ILC2s were crucial for clearing worms and promoting tissue repair in the lung via amphiregulin secretion (165). Similarly, ILC2s migrated to the lung following infection with the intestinal nematode Trichinella spiralis, providing local cross-protection against a secondary N. brasiliensis infection (161). Generation of inflammatory ILC2s was promoted by the enzyme tryptophan hydroxylase 1 (Tph1), which was upregulated in an IL-33–dependent manner upon helminth-induced activation (166). Conditional ablation of Tph1 in lymphocytes resulted in an impairment of inflammatory ILC2 responses, with consequent increased susceptibility to helminths (166). Tph1-deficient ILC2s displayed defective activation and lower levels of ICOS, and in vivo administration of an ICOS blocking Ab after N. brasiliensis infection resulted in decreased recruitment of inflammatory ILC2s to the mesenteric lymph nodes (166).

In addition to direct effects on intestinal parasites, ILC2s also regulate the response to worm infection via interaction with CD4 T cells, with evidence indicating that ILC2 and Th2 cell responses to helminth infections are interdependent. ILC2s are needed for an efficient development of Th2 responses during N. brasiliensis infection, as their ablation resulted in significantly lower levels of CD4 T cells producing IL-5 and IL-13, with consequent delay in worm expulsion (167). A number of cognate receptors pairs are involved in the ILC2–CD4 T cell cross-talk that promotes mutual maintenance, expansion, and cytokine production (167, 168). These pairs include MCH class II-TCR (167), OX40-OX40L (168), and ICOS-ICOSL (169).

ILC2s also play a role in some bacterial and protozoan infections, including C. difficile, H. pylori, and Trichuris muris infections (170–172). During C. difficile infection, IL-33 activates ILC2s, which in turn promote eosinophilia and protection from epithelial damage (170). Finally, both humans and mice infected with H. pylori display elevated ILC2s (171, 172), which appear to contribute to the IgA-mediated containment of H. pylori infection. ILC2-ablated mice exhibit increased gastric inflammation and bleeding compared with WT mice (172).

The flip side of ILC importance for the maintenance of epithelial integrity and tissue homeostasis is their association with inflammatory disorders, including CD and UC, the two most common forms of IBD. ILC function in intestinal inflammation was first investigated in mouse models of IBD (84, 99, 100, 173), which identified a protective role for ILC3-derived IL-22 (99, 100, 173). Two early investigations of CD patients provided clear evidence of the implication of ILCs in human intestinal inflammation, although these studies did not use current definitions of ILC subsets (174, 175). Later investigations confirmed a decrease of ILC3 frequency in the lamina propria of IBD patients, with a concomitant accumulation of ILC1s (67, 176, 177). The decrease in NKp44⁺ ILC3 frequencies, detected in inflamed intestinal tissue but not in non-diseased tissue, was associated with disease severity in both CD and UC (61). A comparison of IBD patients enrolled at diagnosis and individuals with established disease showed that NKp44⁺ ILC3 levels were already low at diagnosis, suggesting that their decrease is a contributing factor to the chronic inflammatory state, rather than its consequence (61). Treatment of IBD patients with monoclonal Abs of known therapeutic efficacy partially restored the healthy equilibrium between intestinal ILC subsets, suggesting that rebalancing intestinal ILCs may contribute to beneficial effects of the biologics (177). Unlike the NKp44⁺ subset, NKp44⁻ ILC3s, which produce IL-17 and IFN-γ, contribute to IBD immunopathology, as suggested by several studies in murine models and IBD patients (61, 84, 176, 178). In UC patients the increase of NKp44⁻ ILCs directly correlated with disease severity (61).

ILC1s are pathogenic in chronic intestinal inflammation. Multiple studies, both in mouse models and human patients, observed elevated ILC1 levels in lamina propria and epithelium of diseased intestine, with ILC1-derived IFN-γ driving the inflammatory process (61, 67, 68, 177, 179) and its neutralization ameliorating disease progression (85). ILC2s, present at low frequencies in healthy human intestine (61, 63), were increased in UC patients at diagnosis and in both CD and UC patients with established disease (61).

The unbalanced distribution of ILC subsets in the intestinal mucosa of IBD patients is partly explained by ILC plasticity, which appears to contribute to IBD pathogenesis. The accumulation of ILC1s in the intestine of IBD patients is likely exacerbated by the transdifferentiation of both ILC2s and ILC3s into ILC1s in the presence of elevated levels of IL-12, described in vitro (66, 67, 146, 180–183). ILC3 to ILC1 transdifferentiation is associated with increased expression of Aiolos, a transcription factor elevated in ILC1s, and requires expression of Ikaros to suppress the AhR pathway (182, 184, 185). Suppression of Aiolos and Ikaros inhibited in vitro transdifferentiation of tonsillar ILC3s into ILC1s, while preserving ILC3 function (182). Transdifferentiation of ILC1s into ILC3s upon exposure to IL-2, IL-23, and IL-1β was also observed, both in vitro and in vivo, and was accelerated in the presence of retinoic acid (146). In support of the notion that ILC transdifferentiation happens in vivo and has physiological relevance, elevated intestinal levels of IL-12 were reported in CD patients (175, 186) in whom an inverse relationship between the NKp44⁺ ILC3 and ILC1 frequencies was also observed (61, 67, 177, 187). Additionally, RNA velocity analysis of human tonsil ILCs revealed the presence of an intermediate ILC3-ILC1 cluster with multiple subsets, while ILC3 and ILC1 represented the ends of the spectrum (184). A transitional ILC3-ILC1 population, coexpressing both ILC3 and ILC1 signature genes, was also observed in the lamina propria of human ileum (184). Finally, the presence of IFN-γ⁺ “ex-ILC2s” in the intestine of CD patients suggests that ILC2-to-ILC1 transitions also occur in vivo (181). Overall, ILC transdifferentiation in the intestinal mucosa of IBD patients appears to contribute to an unbalanced and pathogenic ILC composition, and it may be a factor of clinical significance in the context of cancer.

The presence of ILCs has been detected in multiple types of cancer (188), and evidence for both anti- and pro-tumorigenic roles has been described for all helper ILCs. The conflicting evidence may be explained by their heterogeneity in terms of function, ontogeny, and location combined with differences in cancer microenvironment (189)

ILC1s, secreting abundant IFN-γ and TNF-α, are a recognized driver of chronic intestinal inflammation, and they may thus promote malignant transformation (190). ILC2s and ILC3s converting to ILC1s in the presence of elevated IL-12 are likely to exacerbate the inflammatory conditions. However, IFN-γ and TNF-α also have well-described anti-cancer functions, which include promoting the accumulation of tumor-infiltrating macrophages and DCs, recruiting and activating effector cells (such as NK and CD8 T cells) to promote their cytotoxic activity, inhibiting angiogenesis, and impairing survival of cancer cells and inducing their apoptosis (191, 192). The outcome of the integration of ILC1 pro- and anti-tumorigenic function is likely to depend on the microenvironment at the lesion site.

There is scarce direct evidence of a role of ILC2s in intestinal cancer. More data are available about the cytokines IL-33 and IL-13. IL-33 appears to contribute to carcinogenesis in multiple ways, due to its ability to modulate angiogenesis and stromal remodeling. In particular, IL-33 was involved in the development of intestinal cancer in two murine models, promoting IL-6 secretion and activation of stromal cells (193, 194). Analysis of human tumors suggests a more pronounced role for IL-33 in the early stage of tumorigenesis (194). Other studies, however, have described the anti-tumor effects of IL-33. Low expression levels or ablation of ST2L, one of the IL-33 receptor isoforms, were associated, respectively, with cancer tissue in humans and enhanced tumor growth in a murine model of colon cancer (195). Moreover, treatment with IL-33 reduced tumor growth in an eosinophil-depended manner (196), suggesting that the IL-33 effect was mediated by ILC2-derived IL-5. The ILC2/IL-13 axis is thought to be involved in the establishment of a protumorigenic milieu, as ILC2-derived IL-13 promotes the recruitment and activity of myeloid derived suppressor cells (197, 198); additionally, high expression of IL-13Ra was associated with poor prognosis (179). However, recent evidence supports a protective role of ILC2s in colon cancer in mice and humans, with ILC2 ablation or blockade resulting in increased tumor burden (199, 200). In humans, high expression of an intratumor ILC2 gene signature was associated with increased survival, possibly due to IL-9–mediated activation of CD8 T cells (200).

Most of the evidence about ILC3 cells indicates that they may play a role in the initiation and progression of colorectal cancer. In human cancers, elevated expression of IL-23 and IL-23R in malignant tissue was associated with progression to fatal metastatic disease (201, 202). Mice with IL-23p19 chain or ILC ablation were protected from intestinal tumorigenesis induced by systemic overexpression of IL-23 (201, 202). Multiple studies showed a pro-tumorigenic effect of ILC3-derived IL-22 in murine and human colon cancer: ILC ablation, IL-22 neutralization, and factors attenuating IL-22 secretion (e.g., lack of Card9) were associated with protective effects (203–206). Conversely, IL-22 secretion induced by C. albicans accumulation in the intestine of dectin-3–deficient mice resulted in tumor progression (207). Evidence pointing toward a protective role of ILC3 was obtained in cases of non-small cell lung cancer and melanoma (208, 209), and in a murine model of colon cancer (210). Finally, in a study of human colon cancer, the numbers of ILC3s and NKp44⁺ ILC3s in tumor tissue were lower than those in distal regions and were negatively correlated with cancer stage, in agreement with previous reports (211, 212).

NKT cells and ILCs are important constituents in the maintenance of mucosal homeostasis (213). As highlighted in (Fig. 1, communication between the gut microbiota, NKT cells, and ILCs is crucial for intestinal homeostasis. iNKT cells are regulated by commensals and they can in turn modulate the functions of other immune cells, such as ILC3s, regulatory T cells, and conventional T cell subsets. ILC3s can produce IL-22, thereby contributing to intestinal homeostasis. Conversely, in certain situations, subsets of NKT cells and ILCs can drive intestinal inflammation, most notably through IL-13–dependent mechanisms, which can have pathological consequences. Collectively, these nonconventional lymphocyte subpopulations influence the composition of the intestinal microbiota in a bidirectional manner, which makes them attractive candidates for therapeutic intervention in the context of infection, chronic inflammatory disorders, and colorectal cancer. However, much more work is needed to determine how to effectively modulate the desired subset of NKT cells and ILCs to develop treatment strategies. Thus, future studies utilizing single-cell RNA sequencing and proteomic approaches to investigate host–microbe interactions will be critical to advance our understanding and move the field forward.

There are many groups examining the role of NKT cells and ILCs in intestinal homeostasis, and we apologize to those whose work may have been omitted due to space considerations. The schematic in Fig. 1 was created with BioRender.com.