Beyond Unconventional: What Do We Really Know about Group 2 Innate Lymphoid Cells?

The agents that comprise innate immune responses have limited specificity compared with the adaptive response, but they still distinguish and sense foreign structures through evolutionarily conserved receptors (1). Although phagocytes, polymorphonuclear leukocytes, the complement system, or the release of a wide variety of defensins tend to act immediately upon danger encounter (2), their role in supporting and linking adaptive immunity is often overlooked. They are overlooked so much that it took almost 50 years since the discovery of the lymphocyte (3) to unmask a whole family of unconventional innate effectors that orchestrate immunity, inflammation, and homeostasis across multiple tissues (4). These are the innate lymphoid cells (ILCs), the immune system’s ultimate blend between the innate and the adaptive branch.

What makes every ILC innate is a lack of recombination activating gene-dependent rearranged Ag receptors (5). Recently, however, ILCs have been found to display mutated genes and aberrant proteins from the TCR loci that suggest that they can develop from T cells that fail to rearrange their TCRs (6). Despite this, ILCs have no Ag specificity; instead, they respond to a wide variety of cell-derived factors or alarmins that are secreted both in homeostasis and inflammatory conditions (4). Upon alarmin engagement with specific membrane receptors, ILCs usually respond by secreting cytokines that not only will help unleash the full potential of the immune response, but will also orient it toward its specific needs (7). Curiously, that same cytokine response is what discriminates the three main types of ILCs. Based on differential expression of cell surface markers and distinct secretion profiles, ILCs can be type 1, 2, or 3.

In contrast, what makes these cells lymphoid is their developmental origin, as they are all derived from a common lymphoid progenitor committed to multiple cell lineages, including B- and T-lymphocytes (4, 8). Eventually, ILCs reach their final effector stages guided by transcriptional regulators that are retained, even when terminally differentiated, and their expression can also be used to distinguish between groups 1, 2, and 3.

Group 1 ILCs play important roles in inflammation, and after extensive studies, this subset has been split into two groups, one comprising a highly cytotoxic innate cell, the NK, and another one that has been established as the ILC1, which promotes inflammation but has less cytotoxic potential (9). Group 3 ILCs are an example of ILC heterogeneity within a subset. They are defined by the expression of the transcription factor RORγt, like their T_H17 counterparts, and can also be split into two different cell types, the lymphoid tissue inducer and the ILC3 (10). In the mammalian fetus, the former helps develop the lymph nodes and Peyer Patches and the latter participates in maintaining intestinal homeostasis and defends against extracellular pathogens (5). The focus of this review will be the remaining type, group 2 ILCs (ILC2s), which are perhaps the least understood ILC subset, given the increasing evidence that recently situates them as a highly heterogeneous population that plays more roles in immunity than previously thought (Fig. 1).

The emerging plasticity in what used to be considered the most homogeneous group of ILCs forces us to address nonplastic ILC2s as conventional versus the unconventional subsets that we present in this review. ILC2s are experts at mediating the expulsion of helminthic parasites and are implicated in the development of asthma and allergic inflammation. These are known as type 2 immune responses. ILC2 developmental stages and functions are regulated by the master transcription factor GATA-3 (11). These cells are then stimulated by alarmins, which are activating proteins that are released by tissue-resident or lining cells in response to cell injury, death, or degranulation (12). In particular, the alarmins that stimulate ILC2s are IL-33, IL-25, and thymic stromal lymphopoietin (TSLP), leading to ILC2 activation and a subsequent type 2 cytokine response (13–17). Barlow and colleagues (18) demonstrated in 2013 that IL-33 is far more potent than IL-25 and TSLP when it comes to lung ILC2 activation because of its rapid expression and secretion compared with other alarmins. Despite this, the basis of ILC2 heterogeneity relies on the multiple tissues they occupy in mammals and the markers they express.

Most ILC2 populations are tissue resident in their respective organs, and according to the studies by Schneider and collaborators (19), murine ILC2s populate their niches at three different periods: prenatally from the fetal liver, postnatally from the bone marrow (BM) in the period from birth through weaning, and postnatally through adulthood until death (also from the BM). However, it is possible that ILC2 progenitors are already tissue resident through postnatal periods; in other words, they are already occupying their future niche (19). In general, pre-existing ILC2s are very slowly replaced and diluted by newly generated ILC2s in almost every organ except the skin and BM, in which the turnover time is faster. Once ILC2s have seeded their respective peripheral tissues, they activate, expand, and acquire a set of permanent, tissue-specific transcriptomic signatures (19, 20). Most of these involve a variety of receptors that determine the signals to which they will respond.

For example, the gene encoding the IL-33R (Il1rl1) is highly expressed in the lung ILC2s that reside in alveolar tissues, but is also expressed in fat-resident ILC2s associated with the adipose tissue (AT). In contrast, gut ILC2s express low levels of IL-33R (also known as ST2), but are IL-25R^hi and therefore far more responsive to IL-25 stimulation. Skin ILC2s, however, display low expression of both cytokine receptors and, instead, upregulate the IL-18R1 receptor to produce IL-13 in response to IL-18 and independent of IL-33, IL-25, and TSLP (20). Thus, activation patterns of ILC2s are organ specific.

In addition to these tissue-based differential phenotypes, ILC2s also display an obvious sex bias. Female mice have more lung- and gut-resident ILC2s, and thus they are used more often than males in research. In humans, asthma prevalence is two-times higher in women compared with men (21). The reason why males are generally less susceptible to ILC2-mediated allergic lung inflammation seems to be based on sex hormones. Male ILC2 progenitors in the BM express high amounts of the androgen receptor, which negatively controls IL-33–driven expansion and represses differentiation upon binding of its ligand, the male androgen hormone. This possibly occurs through downregulation of the IL-33 signaling machinery (22). Interestingly, androgen receptor signaling also increases expression of the maturity marker KLRG1 in male ILC2s, which inhibits ILC2 function upon binding to E-cadherin (23). In addition, 5α-dihydrotestosterone (precursor molecule of testosterone) decreased IL-13, IL-5, and RORα expression in mice, a key transcription factor for ILC2 differentiation, whereas female sex hormones 17β-estradiol and progesterone increased the levels of GATA-3 (21).

ILC2s were first identified as c-kit⁺ Fc_εR1⁻ cells in mice that were IL-25^−/− and therefore lacking a proper type 2 response against the protease allergen papain (24), hence the importance of IL-25. In addition, IL-2 has proven essential for ILC2 expansion (25), and IL-7 has proven essential for survival in the lungs (26), but both cytokines are ineffective at ILC2 stimulation in the absence of IL-33 or IL-25 (27).

Upon alarmin engagement, ILC2s across most tissues secrete signature type 2 cytokines, such as IL-5, IL-13, and amphiregulin, moderate levels of IL-9, and low levels of IL-4 (27). Interestingly, several studies have identified novel ILC2 activators, such as the neuropeptide NMU, TGF-β1, or several lipid mediators, that also trigger type 2 cytokine release (17, 28–30). IL-5 promotes eosinophil differentiation from the BM and activates them to release granule contents that are toxic to helminths and protozoa but also to host cells in severe cases of eosinophilia (31). IL-13 is directed to a wider range of functions. It can promote smooth muscle contractibility and hyperactivate goblet cells to produce high amounts of protective mucus in the gut or in the airway (32). These effects of IL-5 and IL-13 are observable in RAG mice, demonstrating a T cell–independent role for ILC2s in health and disease (33). However, via IL-13, ILC2s also link innate and adaptive immune responses by stimulating the migration of IL-13R⁺ allergen-activated dendritic cells that have been “licensed” to migrate from the lung to the mediastinal lymph nodes (mLNs), where they will elicit CD4⁺ T_H2 cell proliferation (33). Furthermore, ILC2s are already a proven source of IL-4, which is an important driver of T_H2 differentiation (34) and Ig class switch to IgE, which is a mediator of mast cell degranulation and allergy (31). However, studies using reporter mice stimulated with IL-33 or IL-25 demonstrate very low IL-4 production by ILC2s compared with other type 2 cytokines (15, 16, 35). The primary source of IL-4 has always appeared rather elusive, mostly because of the very low concentrations that a few cells secrete, such as T_H2 cells, mast cells, basophils, NKT cells, or ILC2s (36, 37). Nevertheless, Doherty and colleagues (38) found out that, in vitro, ILC2s produce more IL-4 than usual in response to leukotriene D4 (LTD4). Leukotrienes are lipid mediators generated by a variety of innate immune cells that contribute to lung inflammation and airway remodelling in mice, roles that overlap with those of ILC2s. In fact, lung ILC2s in mice constitutively and stably express the LTD4 receptor CysLT1R, which leads to IL-13, IL-5, and higher IL-4 production upon leukotriene engagement (38). These studies highlight the possibility of targeting CysLT1R as a therapy to attenuate hyperresponsiveness in humans.

Despite being seen as group 2 innate effectors in disease, conventional ILC2s play distinct roles in homeostasis across tissues. IL-25R^hi ST2^lo gut ILC2s, for example, are highly responsive to tuft cell–derived IL-25, leading to IL-13 production. In 2016, von Moltke and colleagues (39) demonstrated a feed-forward circuit in which ILC2-derived IL-13 promotes further differentiation of tuft and goblet cells from epithelial crypt progenitors, highlighting the role of gut ILC2s in tissue homeostasis. Furthermore, production of IL-13 but mainly IL-4 is essential for macrophage polarization into the M2 phenotype, which is a crucial agent in the process of wound healing by promoting epithelial cell turnover, encapsulation and barrier formation, increased luminal fluids, angiogenesis, and myofibroblast activation (40). ILC2s can also participate in wound healing in more direct ways. In response to influenza infection in the lungs, ILC2s present genetic profiles enriched for extracellular matrix genes involved in lung tissue remodelling (41). Upon viral infection, lung ILC2s also produce amphiregulin, which has been reported to be implicated in lung tissue repair (42, 43). In the skin, elevated numbers of ILC2s upon cutaneous inflammation contribute to pathology (13, 44, 45). Yet, an absence of IL-33 in mice and humans presents poor levels of type 2 cytokines, severely impaired re-epithelialization, and deficient wound closure, demonstrating a significant role for the IL-33/ILC2 axis in restoration of the cutaneous barrier (46).

ILC2s also play an important part in AT homeostasis. In mammals, white AT stores excess energy and contributes to fat accumulation, whereas brown AT is involved in dissipating energy through the production of heat and is therefore linked to lower body weight (47). By expressing a certain proprotein convertase and an endopeptidase, IL-33–dependant white AT-resident ILC2s can induce the upregulation of the uncoupling protein 1 (UCP1) in white AT (48). UCP1 is involved in dissipation of energy through the production of heat, so it converts white AT into beige and eventually brown AT. Therefore, the “browning” of AT by ILC2s can prevent obesity (47). Also, by reducing white AT, in which multiple immune cells accumulate, ILC2s prevent chronic low-grade AT inflammation and the systemic insulin resistance that usually evolves into type 2 diabetes mellitus (49). In turn, AT-resident ILC2s are constitutively supported by white AT-resident multipotent stromal cells that act as an IL-33 reservoir and that stimulate activation of ILC2s expressing the integrin LFA-1 in an ICAM-1–mediated manner (50). As with lung tissue, ILC2s that alternatively activate macrophages promote AT remodelling and prevent systemic inflammation. These macrophages are further supported by ILC2-derived IL-4 and competent eosinophils that accumulate in response to ILC2-derived IL-5 (51).

So far, we have outlined what we consider the conventional roles of a member of an already unconventional family of innate lymphocytes. In other words, the role that ILC2s exert based on their “standard” transcriptomic program. Like every other cell, ILC2s respond to environmental stimuli. As a consequence, they rewire their default programming and adapt to changing circumstances in an unexpectedly flexible manner. This is known as cellular plasticity. Given the resemblance between the ILC and the T cell transcriptome, ILC plasticity was, to an extent, expected. The ability of CD4⁺ T cell subsets to interchange identities with each other in response to external cues was already an established phenomenon (52–54). Why would their innate counterparts be any less plastic? In this article, we explore beyond the unconventional, and we gather previous studies into what we consider the current major subdivisions within ILC2s.

Innate effectors of this kind do not need to become an entirely different cell to challenge the current dogma of ILC biology. Martínez-González and her group (55) demonstrated in 2016 the existence of a pool of memory ILC2s (ILC2_m) that respond more vigorously to secondary stimulation after several months, defying the notion of immunological memory being defined by Ag specificity. Upon intranasal injection of IL-33 or papain in mice, there is a rapid expansion in lung ILC2 numbers followed by a contraction phase. However, Martínez-González and colleagues (55) demonstrated that a fraction of the ILC2s can persist in the draining mLNs for up to 160 days after primary challenge. These ILC2s were IL-5⁺ IL-13⁺, but the cytokines were not being secreted; instead, they were stored within the cell (55). This reduced pool that survived the contraction phase was also responding more vigorously to secondary papain challenge and other allergens that they have not previously encountered. ILC2_m secreted more cytokines than other ILC2s that were being stimulated for the first time. Except recognizing an Ag directly, ILC2_m are displaying the same characteristics as a memory T cell response: quick expansion, a contraction phase, and a small, long-lived pool with rapid and enhanced effector functions; plus they expressed a similar gene profile to that of memory T cells (56).

ILC2_m residing in mLNs, as opposed to conventional lung ILC2s, are highly responsive to IL-25 instead of IL-33. Upon primary IL-33 engagement, lung ILC2s upregulate IL-25R. Like gut ILC2s, the cells that survive the contraction phase will remain in the lung and mLN as IL-25R^hi. If mLN ILC2_m can persist up to 5 months, it is possible that they can migrate back to the lung upon secondary challenge to display an enhanced cytokine release (57). This forces us to reconsider ILC2s as a purely tissue-resident cell. Some researchers address this peculiar allergen experience as “trained immunity.” Also, as opposed to macrophages and T cells, ILC2s do not need to undergo differentiation to reach the memory stage; its functions as ILC2_m remain very similar to those of unstimulated ILC2s, meaning that they stay committed as type 2 cytokine secretors (56). The existence of this population of ILC2_m might sensitize individuals to certain allergens and facilitate the development of T cell–independent asthma.

ILC2s serve critical roles in the development of allergic inflammatory reactions such as asthma, atopic dermatitis, or allergic lung inflammation. To prevent these reactions from “spinning” out of control, the immune system employs mechanisms of strict ILC2 regulation (58). The global theory of T_H1 and T_H2 responses counterregulating each other to grant balance and successful homeostasis arose in 1986 and provides the first level of control of ILC2s (59), which are in principle clear representatives of type 2 immunity. Type 1 immunity will therefore inhibit the progression of type 2 responses. Recently, two different groups have demonstrated that ILC2 proliferation and function can be effectively suppressed by the proinflammatory cytokines IFN–α/β, IFN-γ, and IL-27 (58, 60). The most potent suppressor of ILC2 function was IFN-γ, and this was demonstrated using Ifng^−/− mice that displayed strong type 2 immunopathology (58). In fact, single nucleotide polymorphisms in the Ifng gene or the chains encoding its receptor have been associated with asthma in humans (61–63). An immunosuppressive role for IFN-γ in ILC2s implies that they express the IFN-γ receptor (IFN-γR1/2). The IFN-γ receptor IFN-γR1 is detectable in low levels in ILC2s (64), but its upregulation mediated by IL-1β could result in a IFN-γR1^hi ILC2 subset prone to suppression in a proinflammatory environment (65). Our laboratory has detected increased expression of this receptor in ILC2s harvested from the lungs of mice with primary tumors grown from a murine, lung epithelial tumor cell line (P. de Lucía Finkel, I. Saranchova, and W.A. Jefferies, unpublished observations). These results suggest that a systemic proinflammatory environment could attenuate ILC2 type 2 functions.

Even more intriguing is the capacity of self-regulation that ILC2s have. In 2016, Kim and colleagues (66) revealed a non–T/non–B-lymphocyte whose surface markers and tissue residency resembled an ILC. This ILC, however, was a new effector that produced notable amounts of the immunosuppressing cytokine IL-10. It was named ILC10. A year later, this ILC10 was unmasked and revealed an alternatively activated ILC2 subset that produced large amounts of IL-10 (67). This new subset, the ILC2₁₀, was also activated by IL-33 but persisted after a contraction phase when the stimulus was removed and then was easily expanded upon secondary challenge (67). Interestingly, the ILC2₁₀ displays features similar to that of the previously described memory ILC2. Whether they are the same cell or not remains to be elucidated; however, the fact that ILC2_m respond with enhanced capabilities suggests a need for tighter regulation and hence IL-10 production. The frequency of ILC2₁₀ correlated to stronger ILC2 activation and elevated eosinophilia, suggesting an immunosuppressive role that can act on itself or other immune cells (67).

In 2019, Morita and colleagues (68) reported striking, yet compelling, evidence of a population of ILC2s that not only produced high amounts of IL-10, but also upregulated the regulatory T cell markers CTLA-4 and CD25 in response to retinoic acid. Although they retained IL-13 and GATA-3 expression, their genetic profile was so deviated from conventional ILC2s that they termed it ILC_REG (68). Also, they were demonstrated to genetically differ from the previously mentioned ILC2₁₀ subset. Retinoic acid is known to drive regulatory T cell differentiation from naive T cells (69, 70), but it seems to also have a role involving ILCs, in which ILC2s transition from a type 2 identity toward a regulatory phenotype. ILC_REGs were capable of fully suppressing CD4⁺ T cell differentiation (68), and thus, they present themselves as a novel ILC2-derived innate regulatory subset.

An undeniably effective way to define ILC2 plasticity is by bringing together a comprehensive map of ILC2 subsets clustered by tissue residency, or their differences between tissues (20). In this review, we opt for differencing ILC2 subsets based on the acquisition of novel, non–type 2–associated functions and highly unconventional roles. Both approaches, however, benefit from new emerging technologies, such as mass cytometry or single-cell RNA sequencing. ILC1 to ILC3 plasticity has been studied in detail to a point at which researchers have enough criteria to identify a complex spectrum of identities that end in either fully differentiated ILC1s or ILC3s and at which the direction of such differentiation depends on multiple environmental cues from tissues (71–74). Despite this, plasticity between ILC2 and ILC3 subsets remains a large mystery, and so far, the only reported phenotypical shifts are confined to a limited number of interchangeable traits. Notch signaling has proven to be essential both for ILC2 and ILC3 development; however, it seems that the strength of the signaling determines the digression into one lineage or the other (75, 76). Nevertheless, both effectors strictly depend on T cell factor-1 (TCF-1) to differentiate and exert their functions (77). The role of this transcription factor in ILC development can be inferred from the studies that Zhang and colleagues (78) carried out in thymocytes at early heamotopietic stages of T cell development. TCF-1 has the potential to be the main negative regulator of T_H17 differentiation by binding two promoter regions of the IL-17 gene and hence inhibiting the expression of the cytokine that defines T_H17 cell identity (78). It is therefore expected for TCF-1, along with GATA-3 and RORα, to play a major role in progression onto the ILC2 lineage or, in its absence, to promote plasticity toward an ILC3 identity. As both lineages depend on the same crucial transcription factors, ILC2s and ILC3s are anticipated to possess enough substitutable traits under specific environmental conditions to be defined as plastic cells.

As previously mentioned, lung ILC2s are characterized as Lin⁻ IL-25R^lo ST2⁺ cells that will expand after IL-33 engagement. However, Huang and colleagues (79) identified a scarce, lung-resident subset that expanded mostly upon IL-25 stimulation and that expressed high levels of the killer cell lectin receptor KLRG1, described as Lin⁻ ST2^lo KLRG1⁺ ILC2s. Because these cells were usually encountered after IL-25 administration or infection with the intestinal nematode Nippostrongylus brasiliensis, they were called inflammatory ILC2s (iILC2). Interestingly, despite the low levels of ST2 expression on the surface of these unconventional ILC2s, IL-33 has also proven to be a necessary and sufficient cytokine for iILC2 generation in the lungs and mesenteric lymph nodes. A recent study by Flamar and colleagues (80) evidenced that IL-33 significantly induces the expression of the enzyme tryptophan hydroxylase 1 (Tph1), the rate-limiting enzyme in serotonin biosynthesis. Such IL-33–dependent upregulation of Tph1 in mouse lymphocytes seems to be directly associated with iILC2 recruitment and function. A conditional deletion of the Tph1 gene in genetically ablated mice resulted in a depleted iILC2 pool with defective responses and impaired exclusion of N. brasiliensis (80). Thus, it is safe to assume that iILC2 induction, along with conventional ILC2 proliferation, is a relatively recurrent occurrence in mucosal surfaces upon infection or allergy, given their dual responsiveness to IL-25 and IL-33–dependent Thp1.

iILC2s expressed more GATA-3 than conventional ILC2s and secreted similar amounts of IL-13 and IL-5. IL-4 production was, however, elevated in iILC2s (81). Although their exact progenitor has not yet been identified, iILC2s are found in the lung vascular space as opposed to conventional ILC2s, which reside in the alveolar tissue (82). This suggests that iILC2s are migratory, but from where do they migrate? In the small intestine, specifically in the lamina propria, there is a heterogeneous pool of ILC2s, most of which are sensitive to IL-25 rather than IL-33, and a subset of these are KLRG1⁺ (19). Just like their adaptive counterparts, these IL-25–activated KLRG1⁺ ILC2s access the lymphatic vessels in the villi by upregulating S1P receptors and eventually entering the bloodstream (82). Once in the lung, the environmental cues will dictate the fate of these iILC2s.

Experimental evidence has demonstrated that, if cultured under T_H2 conditions or during parasite infection in vivo, iILC2s evolve into ST2⁺ conventional ILC2s that downregulate IL-25R, suggesting that iILC2s could act as transient progenitors (79). Intriguingly, iILC2s also express intermediate amounts of the master regulator RORγt and begin to resemble an initial ILC3-like identity by producing some IL-17 (but also IL-13), which was drastically increased when cultured under T_H17 conditions (83). These iILC2s were also highly effective at fighting Candida albicans infection in mice compared with conventional ILC2s. In vivo, they shut down IL-13 production and upturn IL-17 levels (implying a lack of TCF-1), proving an increased induction of plasticity upon fungal infection (78). Overall, iILC2s are an excellent example of a highly inducible and plastic subset of ILC2s that evolve toward an ILC2- or ILC3-like phenotype depending on its environment.

Studies on novel, iILC2s were furthered by three independent laboratories in 2019 that confirmed induction of ILC2 plasticity toward an ILC3 phenotype mediated by IL-1β, IL-23, and TGF-β (84–86). The presence of RORγt, the downregulation of GATA-3, the coexpression of IL-13 and IL-17A, and the reduction in IL-5 in these cells coincide with the iILC2-derived ILC3-like cells described by Huang and colleagues (84). Bernink and collaborators (86) expanded the knowledge on iILC2-to-ILC3 induction of plasticity by determining the importance of c-Kit expression in ILC2s. Apparently, c-Kit⁺ ILC2s already express RORγt at baseline, and based on their lower IL-13 production after activation, it is safe to assume that they are more prone to transit into an ILC3-like phenotype. In contrast, c-Kit⁻ ILC2s retain a stronger type 2 identity and require larger amounts of IL-1β, IL-23, and TGF-β to acquire ILC3 properties (86), whereas c-Kit⁺ ILC2s only need IL-1β and IL-23. These studies add another level of complexity to iILC2s.

The implications of these levels of plasticity in asthma have been clear since 2010, when researchers identified an IL-17A⁺ T_H2 adaptive effector that contributed to disease severity by promoting T_H2 and T_H17 responses simultaneously. The innate counterpart of this potent effector was unmasked by Cai and colleagues (85), also in 2019, and it was so powerful at promoting lung inflammation via granulocyte recruitment that it was addressed as the ILC2₁₇, a new effector capable of IL-5, IL-13, and IL-17 production in vast quantities. ILC2₁₇s lack RORγt and therefore cannot be identified as iILC2s. Instead, their acquisition of ILC3 identity (retaining type 2 identity) depends on the expression of the aryl hydrocarbon receptor along with sustained IL-33 signaling, as opposed to a single stimulation (85).

The abandonment of the classical type 2 identity by ILC2s toward a more proinflammatory, T_H1-like phenotype has also been reported by several studies (87–90). The phenomenal abilities of ILC2s to acquire ILC3 properties also apply to ILC1 cells. In an attempt to prove this, Ohne and colleagues (87) stimulated human ILC2 cultures with IL-1α and IL-1β, central mediators of the inflammatory response (91). Unexpectedly, and in the presence of IL-2, IL-1β via IL-1R1 induced significant expression of IL-5 and IL-13 just like IL-33 did (87). This situates IL-1β as a perfectly valid ILC2 activator in addition to IL-33, IL-25, and TSLP. Interestingly, our group has also detected important levels of IL-10 and GM-CSF in culture supernatant cytokine profiling analyses. In addition to classical ILC2 activation, important genes involved in adaptive immunity, such as MHC class II (MHC-II), CD80, and CD40L were upregulated. In the presence of IL-1β with IL-12, however, GATA-3 expression is reduced in favor of an increment in T-bet and abundant IFN-γ, confirming attainment of ILC1 identity by ILC2s (87). Soon after, it was clarified that such induction of plasticity is directly mediated by IL-12, but it is IL-1β that is responsible for IL-12Rβ2 upregulation in conventional ILC2s. Furthermore, inclusion of IL-4 in the culture reversed these plastic properties by promoting STAT6, which in turn induces GATA-3 (89, 90, 92).

An innate effector that is heavily involved in mediating immunity against parasites with capacity to acquire ILC1 proinflammatory properties has tremendous repercussions in disease and also in the global ILC paradigm. In lung disorders, such as tobacco smoke-mediated chronic obstructive pulmonary disease or severe influenza infection, a systemic inflammatory environment rich in IL-1β and IL-12 is necessary to mount an adaptive T_H1 response. As previously mentioned, the action of these cytokines on conventional lung ILC2s will result in GATA-3 downregulation, a halt in IL-5 and IL-13 production, increase T-bet expression, and promote IFN-γ secretion (in other words, the backbone of T_H1 immunity). Silver and colleagues (88) have studied this kind of plasticity on the context of disease, and their results demonstrate that there is a clear derivation of fully transformed ILC1 cells from the local ST2⁺ ILC2 pool and that the newly generated ILC1 population dramatically amplifies antiviral immunity in the lungs of mice. Furthermore, these ex-ILC2s gathered around the influenza-infected foci in the lungs to combat viral expansion (88).

Despite the fascinating evidence that unveils a new proinflammatory subset of ILC2s, there is still not a unified nomenclature for it. This is perhaps because the plasticity toward the type 1 phenotype is so polarized that it can be directly addressed as an ILC1. This review proposes the term “type 1 ILC2” or “ILC2₁” for those conventional ILC2s with the potential to acquire ILC1 properties in an intermediate state of plasticity or as a substitute for ILC1 ex-ILC2s.

If plastic ILC2s play a significant role against influenza, it goes without saying that characterizing their response in the context of the novel 2019 coronavirus disease (COVID-19) is a matter of urgent interest and a unique opportunity to pioneer an unexplored field of ILC2 biology. This, indeed, forces us to question the current understanding of ILC2s in the context of respiratory viral infections; however, the general notion stipulates that these cells are not avid antiviral responders. Nevertheless, such notion is novel, and most ILC2 studies in this context have only been associated with influenza, rhinovirus (RV), and the respiratory syncytial virus (RSV), which is known to infect a majority of infants and to increase the risk of asthma later in life (93).

RV, a virus associated with asthma predisposition, infects the airway epithelium and triggers noteworthy ILC2 expansion (94). In six-day-old mice, RV infection leads to an abundance of IL-33, IL-25, and TSLP, which are required for ILC2 activation and further mucous metaplasia and airway hyperresponsiveness (94). Along these lines, Jackson and colleagues (95) have demonstrated that RV infection in a human experimental model leads to abundant induction of IL-33 and subsequent ILC2 activation in the lung, which is thought to be the mechanistic link between respiratory viral infections and asthma exacerbation. In infants and young adults, however, such exacerbation is more likely to occur upon RSV infection. Interestingly, a 2016 study by Stier and collaborators (96) did not identify sufficiently significant IL-33 production to explain the high numbers of IL-13–producing ILC2s that increased disease severity in a murine model. Instead, this was attributed to TSLP secretion by lung epithelial cells infected by RSV (96). The type 2 immunopathogenesis of RSV was further clarified by demonstrating an upregulation of uric acid during RSV lung infection in infants and mice. Uric acid is secreted as a pathogen-associated molecular pattern upon cellular stress, which in turn activates the NLRP3 inflammasome, mostly in neutrophils, to eventually release IL-1β (93), a potent ILC2 activator (87). Furthermore, using an inhibitor for xanthine oxidase, an enzyme involved in uric acid metabolism, Fonseca and colleagues (93) observed decreased levels of IL-33 and TSLP and therefore reduced ILC2-mediated immunopathology upon RSV infection.

These studies suggest that an important trigger for the unfavorable ILC2 response observed in respiratory infections could be the release of uric acid by infected lung epithelial cells. This is very relevant to the current SARS-CoV-2 pandemic, for which several Chinese teams have recently reported a significant percentage of patients displaying elevated uric acid in COVID-19 pneumonia (Z. Li, M. Wu, J. Yao, J. Guo, X. Liao, S. Song, J. Li, G. Duan, Y. Zhou, X. Wu, et al., manuscript posted on medRxiv, DOI: 10.1101/2020.02.08.20021212 and H. Yu, D. Li, Z. Deng, Z. Yang, J. Cai, L. Jiang, K. Wang, J. Wang, W. Zhou, X. Wei, et al., manuscript posted on SSRN, DOI: 10.2139/ssrn.3551289). Such evidence raises the possibility of having disadvantageous ILC2 participation in the COVID-19 response. To further support the notion of TSLP and IL-33–activated ILC2s playing a detrimental role in sustaining lung disease during COVID-19, a recent publication by Zizzo and Cohen (97) argues that the ILC2-derived IL-9 stimulates proliferation and expansion of a particular type of γδ lymphocyte, the Vγ9Vδ2⁺ T cell. This specific cell is of interest because of its dual proinflammatory cytokine profile, which includes a combination of the T_H1 and the T_H17 phenotype (98). Thus, the antiviral response against SARS-CoV-2 will not benefit from this ILC2–γδ T cell proinflammatory axis given its potential to enhance lung inflammation and pneumonia in COVID-19 patients. Furthermore, Vγ9Vδ2⁺ T cells not only express a very powerful effector memory phenotype but also display elevated levels of CXCR3 compared with their αβ counterparts, suggesting that this specific subset of γδ T cell is selectively stimulated in COVID-19 and rapidly recruited to the lungs (97).

As previously mentioned, ILC2-derived IL-13 and IL-4 also favor macrophage M2 polarization and recruitment (99), a phenotype highly responsible for the release of key profibrotic agents that may contribute to the loose interstitial fibrosis that accompanies moderate-to-severe cases of COVID-19 (100). Outside of the lungs, ST2⁻ gut ILC2s in humans express elevated levels of the SARS-CoV-2 entry receptor ACE2 transcript, suggesting their input toward lung immunopathology. If so, these ILC2s would be capable of migrating from the gut to the lungs in response to alarmins released by lining epithelial cells upon SARS-CoV-2 infection (101). It would be of interest not only to determine whether human ACE2^hi ILC2s are actually migratory but also to find a potential S1PR⁺ KLRG1⁺ iILC2 subset among them. The major drawback with ILC2s and SARS-CoV-2 studies, however, is its inability to infect wild-type laboratory mice, given the inefficient interactions between the viral S protein and the mouse ortholog of ACE2 (102). In humans, research involving peripheral blood ILC2s may not accurately represent the tissue-resident populations in the gut or the lungs of COVID-19 patients, which is why a mouse-adapted model of SARS-CoV-2 would be beneficial.

Although ILC2s appear not to contribute toward COVID-19 resolution, their role in influenza infection seems to be beneficial. ILC2s are known to proliferate and accumulate in the lungs of wild-type and Rag1^−/− mice after mouse-adapted H1N1 influenza infection in the presence and absence of adaptive immunity. However, ILC-depleted mice via anti-CD90.2-depleting mAbs showed significantly decreased lung function and epithelial integrity upon H1N1 infection (41). Curiously, this phenotype can be effectively rescued by administering recombinant amphiregulin, a member of the epidermal growth factor family that is highly expressed in lung ST2⁺ ILC2s. This highlights a pivotal set of functions associated with lung tissue repair and homeostasis that ILC2s exert following influenza infection in mice, in particular via IL-5 (103) and amphiregulin production (41).

Given the scope of this review on ILC2 plasticity, whether conventional ILC2s help or worsen viral respiratory infections is not as relevant. So far, the focus in research has been studying ILC2s within the framework of a classical type 2 identity and from there to determine the feasibility of taking these cells into consideration for therapies. As mentioned previously, ILC2s are able to shift their transcriptomic and functional identity into a cell type that, when scrutinized carefully, has no resemblance whatsoever to a conventional ILC2. For this reason, we propose an alternative way to examine the role of ILC2s in respiratory viral infections.

The flagship of ILC2 antiviral immunity is the previously mentioned study by Silver and colleagues (88), in which they demonstrate how severe influenza infection can trigger the transformation of lung tissue-resident ILC2s into fully functional ILC1s with antiviral properties (see Type 1 ILC2s section). As discussed earlier, the immense potential for plasticity and heterogeneity of ILC2s calls for an open mind when carrying out ILC research. It would be interesting to study other lung ILC populations, especially ILC1s, to determine whether these cells with significant antiviral capacity (104, 105) are potential ex-ILC2s with a shifted identity in response to the powerful T_H1 environment triggered by COVID-19 (106, 107). Therefore, we highly recommend taking into account ILC2 plasticity when interrogating the roles of ILC2s in novel respiratory infections. Technologies that allow for precise single-cell trajectory and RNA velocity analyses could help elucidate whether ILC2s respond to COVID-19 by acquiring a T_H1 identity, remain neutral, or enhance their classical T_H2 identity in favor of immunopathology.

Once again, the field of immunology proves that every new discovery entails uncountable levels of complexity and more questions than answers. In this review, we have defined an unconventional group of ILC2s. Their developmental origin qualifies them as lymphoid, and most of their functions allocate them in the innate branch of immunity. For now, not much more can be assured about them. Recently classified as group 2 innate effectors, conventional ILC2s have proven to be fundamental in other noncanonical roles, such as tissue remodelling and wound healing across numerous tissues. In theory, ILC2s are a fixed, homogeneous population, but in practice, they are flexible and possess significant plasticity. Research has already identified memory, regulatory and iILC2s, ILC2s with ILC3-like properties, and type 1 ILC2s involved in antiviral responses. Even without acquisition of plasticity, conventional ILC2s are still immensely heterogeneous across tissues. This review has dived beyond the unconventional and leaves the initial question as open as ever. What do we really know about ILC2s?

The authors have no financial conflicts of interest.