Antiviral Activities of Group I Innate Lymphoid Cells

Abstract

Even before the adaptive immune response initiates, a potent group of innate antiviral cells responds to a broad range of viruses to limit replication and virus-induced pathology. Belonging to a broader family of recently discovered innate lymphoid cells (ILCs), antiviral group I ILCs are composed of conventional natural killer cells (cNK) and tissue-resident ILCs (ILC1s) that can be distinguished based on their location as well as by the expression of key cell surface markers and transcription factors. Functionally, blood-borne cNK cells recirculate throughout the body and are recruited into the tissue at sites of viral infection where they can recognize and lyse virus-infected cells. In contrast, ILC1s are poised in uninfected barrier tissues and respond not through lysis but with the production of antiviral cytokines. From their frontline tissue locations, ILC1s can even induce an antiviral state in uninfected tissue to preempt viral replication. Mounting evidence also suggests that ILC1s may have enhanced secondary responses to viral infection. In this review, we discuss recent findings demonstrating that ILC1s provide several critical layers of innate antiviral immunity and the mechanisms (when known) underlying protection.

Introduction

Innate lymphoid cells (ILCs) are a relatively recently discovered population of immune sentinels that reside in barrier tissues and rapidly respond to invading microbes before the initiation of the adaptive immune response [1]. As their name suggests, ILCs do not require somatically mutated antigen-specific receptors for development or function, distinguishing them from adaptive lymphocytes. Instead, ILCs produce and release cytokines into the extracellular milieu when triggered by homeostatic imbalances, such as infection [2]. Indeed, ILCs have been referred to as “guardians of the tissue” due to their swift response to a broad range of insults.

Despite their ability as a group to respond to a continuum of Ags, subsets of ILCs populate different tissues and protect against different classes of microbes. In both humans and mice, three groups of ILCs have been identified--group 1, group 2, and group 3 ILCs. Each broad ILC group possesses heterogeneity and has been proposed to serve an analogous role to a subset of adaptive lymphocytes. Unlike their adaptive equivalents, ILCs do not express lineage markers. Instead, each ILC group can generally be distinguished by analysis of their complement of cell surface proteins and transcription factors (discussed in more detail below). Group I ILCs are composed of both conventional natural killer (cNK cells) and tissue-resident ILC1s that participate in innate antiviral immune responses. While cytolytic NK cells are considered the innate counterpart of cytotoxic T cells, non-cytolytic ILC1s are thought to function akin to Th1 cells, secreting the antiviral cytokines IFN-γ and TNF-α. ILC2s are most like Th2 cells, producing IL-5 and IL-13 and participating in responses against helminths and during allergic inflammation. ILC3s resemble Th17 cells, release IL-17 and IL-22 and protect against extracellular bacterial infection.

In this review, we focus specifically on the mechanisms underlying innate antiviral immunity provided by canonical ILC1s (summarized in Figure 1), beginning with a discussion of the distinguishing features of ILC1s and NK cells. We summarize recent findings in the field focusing on innate antiviral activity, including evidence for ILC1-mediated protection during primary and secondary viral infection. We close with our perspectives on outstanding questions regarding ILC1 function.

Figure 1. The many functions of ILC1s promoting antiviral immunity.

ILC1s provide antiviral protection by both direct and indirect mechanisms. Even before infection in some tissues, ILC1s can secret low levels of IFN-γ that promote the upregulation of antiviral genes in the epithelium. Though unknown during antiviral responses, ILC1s likely secrete cytokines and proteases that protect the tissue and promote epithelial integrity, preventing viral spread and they can likewise shape the microbiota. After viral infection, ILC1s upregulate production of IFN-γ to directly limit viral replication. Mounting evidence suggests that ILC1s also provide enhanced protection to secondary viral infection; this has been demonstrated in MCMV infection and depends on the viral protein m12. During inflammation, ILC3s and NK cells can also convert to bona fide ILC1s or ILC1-like cells, also contributing to antiviral immunity.

Defining ILC1s

The importance of NK cells in host defense against viruses has been long appreciated (reviewed in [3–5]). In 1995, Delano and Brownstein demonstrated that murine susceptibility to poxvirus infection mapped to a locus containing a high number of NK cell receptors (now known as the NK cell complex) [6, 7]. In humans, primary immunodeficiencies resulting in reduced NK cell number or function predispose individuals to recurrent viral infections [8–10]; this finding has also been demonstrated after viral challenge of NK-deficient mice. Mechanistically, NK cells secret anti-viral cytokines (like ILC1s); however, NK cells can also lyse virally infected cells either directly through recognition of viral proteins, induced self-ligands, or alterations in MHC class I molecules; or indirectly via antibody-dependent cellular cytotoxicity (reviewed in [11]). Due to their wide distribution and presence in the circulation, NK cells may be particularly important for the control of systemic viral infections. Recently, Borst et al. selectively reconstituted interferon-γ (IFN-γ) production in the NKp46⁺ cells of Ifng^−/− mice, which protected mice against intravenous vaccinia virus (VACV) challenge and demonstrated unequivocally that group I ILC-derived IFN-γ alone can protect against a systemic viral infection [12]. For more information on NK-mediated protection against viruses, we point the reader to several excellent reviews ([3, 13–15]).

ILC1s were originally identified in the livers of C57BL/6 mice and were thought to be immature NK cells (as they share many cell surface receptors with cNKs) [16]. From lineage and functional studies, it is now appreciated that NK cells and ILCs have distinct precursors and developmental pathways. Both subsets are broadly defined in C57BL/6 mice as Lin-NKp46⁺ NK1.1⁺ T-bet⁺ cells. Several distinguishing features have been used to separate cNK cells and ILC1s, discussed briefly below and reviewed in [17].

Location.

Conventional NK cells are circulatory, trafficking via the bloodstream through secondary lymphoid organs and into inflamed or infected tissues [18]. In contrast, ILC1s are tissue-resident at steady state, as demonstrated by parabiosis studies [18–20]. We have shown using intravital microscopy that ILC1s in the murine oral mucosa slowly patrol the basal epithelium of uninfected tissue and are maintained in their epithelial niche even after local VACV infection [21] (Figure 2). Interestingly, in our study and unlike our previous studies with antiviral T cells [22]. ILC1s do not accumulate around areas containing VACV-infected cells. Although we did not detect ILC1 migration out of the tissue during infection, a recent study has shown that hapten-sensitizing recruits ILC1s into skin-draining lymph nodes in a CXCR3-dependent manner [23], raising the possibility that ILC1s may also move through via the lymphatics or circulation under some circumstances. With the creation of mouse models allowing genetic separation of NK cells and ILC1s, it will be interesting to examine ILC1 mobility under different circumstances.

Figure 2. ILC1s populate the basal epithelium of the murine labial mucosa.

A) Frozen tissue section of the uninfected labial mucosa (lip) of a T-bet-ZsGreen reporter mouse. Left panel shows staining for cell surfaces using fluorescently conjugated wheat germ agglutinin (WGA), DAPI for nuclei, and ILC1 in green.

B) Frozen tissue section of the uninfected labial mucosa of an Ncr1-gfp mouse stained for collagen IV (white), a component of the basement membrane separating the upper epithelium from the lower lamina propria. ILC1s are shown in green; nuclei in blue. Right panel omits the DAPI signal to show close association of ILC1s with the basement membrane.

Scalebars = μms.

Transcription factors.

ILC subsets are often differentiation based on expression of specific transcription factors. While ILC1s require GATA-3 for maintenance [24], they express low levels of this transcription factors compared to ILC2s, which continuously express high levels of GATA-3 protein [25]. ILC3s also require GATA-3 expression after fate commitment despite low protein expression levels [26]; ILC3s also express RORγt [27, 28]. To separate ILC1s from NK cells, expression of the transcription factor Eomesodermin (Eomes) has often been used, as NK cells require Eomes for their development while most ILC1s do not [29–31]. However, this is not a steadfast rule, as ILC1s in different tissues rely on different transcription factors for development. Salivary gland ILC1s and some liver ILC1s have been shown to express Eomes [20, 32]. Most ILC1s require the expression of T-bet for development and retain expression when mature (although NK cells do as well) [29]. Here, ILC1s contrast to another prominent tissue-resident lymphocyte population involved in antiviral responses, tissue-resident memory T cells (T_RM), which require T-bet downregulation for optimal development or survival in some tissues [33]. Another transcription factor implicated in T_RM development, Hobit, was recently shown to be necessary for liver and salivary gland ILC1 development as Zfp683^−/− (Hobit)-deficient mice are severely reduced in liver ILC1s [34, 35]. Very recently, Hobit was shown to confer effector function in TCF-1^hi lineage-committed ILC1s throughout multiple tissues including the liver and salivary glands [36]. In this study, Friedrich et al. demonstrated that Hobit transcriptionally regulated a decreasing gradient of stem-like genes to establish cytotoxic effector function in ILC1s [36].

Adhesion molecules and chemokine receptors.

Many different cell surface markers have been used to distinguish NK cells from ILC1s, with definite caveats. Like T_RM, ILC1s express many markers necessary for maintenance in the tissue, including CD69, CD103, CD49a, and CXCR6. CD69, a common marker for T cell activation, also prevents T_RM egress from the tissue by reducing responsiveness to sphingosine-1-phosphate (S1P) gradients and is a commonly used marker of T cell tissue residency (though it is not always expressed on tissue-resident cells) [37, 38]. The molecules CD49a and CD49b have been used to separate ILC1s from NK cells, respectively. CD49a (integrin α1 paired with β1) can also be expressed by T_RM, where it regulates motility and maintenance through adherence to collagen IV [39]. Accordingly, CD49a⁺ oral mucosal ILC1s maintain close association with collagen IV present in the epithelial basement membrane (Figure 2) [21]. We have not determined the role of CD49a expression on ILC1 motility. In some instances, such as in the tumor microenvironment, CD49a expression corresponds to downregulation of IFN-γ production; however, this does not appear to be the case in all tissues [21, 40]. Many ILC1s also express CD103 (integrin αE), a receptor for E-cadherin on endothelial cells that is upregulated in response to TGF-β signaling [20, 41].

Chemokine receptors dictate tissue entry and localization. Some chemokine receptors are preferentially expressed by ILC1s. CXCR6, the receptor for CXCL16, is required for T_RM development and maintenance in some tissues, including the skin and lung [42, 43]. CD49a⁺ ILC1s in the liver can also express CXCR6, though CD49b⁺ NK cells do not [23, 44].

Other markers.

The inhibitory receptor CD200R1 has been used to separate NK cells from ILCs in both mice and humans [35,45, 46]. Weizman et al. demonstrated that, although NK cell surface receptors changed in response to viral infection, they did not acquire the expression of CD200R1, making this a suitable marker to separate ILC1s from NK cells [35]. In oral mucosal ILC1s, we found intermediate CD200R1 expression, making it difficult to use as a distinguishing marker [21]. To unravel the remarkable (and confusing) heterogeneity in NK cells and ILC1s, and better separate these two populations, McFarland et al. recently performed a large-scale single-cell RNA sequencing analysis of these cells in multiple mouse tissues [46]. Despite such a large sequencing effort, the authors failed to identify a single, universal marker that could unequivocally denote ILC1s, instead suggesting that different tissues would require different combinations of markers. No doubt the field will continue to refine and expand the possible markers of NK cells and ILC1s during different infections and inflammatory diseases.

ILC1 generation and maintenance

Like other tissue-resident cells, after establishment, ILC1s maintain residency through local expansion rather than predominantly relying on hematogenous replenishment during homeostatic conditions [18]. However, during inflammatory conditions, ILC1 numbers can be increased through cellular conversion of both NK cells and ILC3s.

ILC1 development.

The developmental pathways leading to ILC1 lineage commitment have been the subject of intense study and have been reviewed elsewhere (see [47–49]); thus, they will only be briefly covered in this review. All innate lymphocytes are generated from a common progenitor found in either the fetal liver or the bone marrow that is dependent on NFIL3 for differentiation; ILC progenitor development also requires the transcription factors Id2 and GATA-3 [25, 50–56]. Constantinides et al. first identified a PLZF^hi ILC progenitor cell that serves as the bifurcation between NK and ILC development [57, 58]. These PLZF^hi ILC progenitor cells likely represent a more mature progenitor that stems from a broader common helper ILC-progenitor (CHILP) lineage that only gives rise to helper ILCs and not NK cells [24]. New experiments also suggest that ILC progenitors further differentiate in the tissue, with progeny decisions shaped by the microenvironment [54, 59, 60]. Recently, Batf was also shown to regulate both commitment to an ILC progenitor as well as peripheral ILC functions, including proliferation and cytokine production [61].

Organ-specific development.

The anatomical pathways through which ILC1s seed different organs are not entirely clear. ILC1 development has been most heavily examined in the murine liver, which contains large numbers of resident ILC1s, along with other tissue-resident lymphocytes [29, 44]. Bai et al. found a population of hematopoietic stem cells in the murine fetal liver that were exclusively tissue-resident in parabiosis studies [62]. Genetic deletion of Ifng in NKp46⁺ cells (both ILC1s and cNKS) resulted in a selective deficiency in liver ILC1s, indicating that IFN-γ production by group I ILCs is needed to fully populate the adult liver ILC1 compartment. Therefore, the fetal liver seeds progenitors that locally proliferate to generate ILC1s later in the adult. Other studies have pointed to a role for perinatal ILC1 development [63]; for skin-homing ILC1s, this occurs in fetal thymi [64]. Using parabiotic mice, Wang et al. demonstrated that lymph node ILC1s were likely partially recruited from the blood after hapten sensitization; these cells later migrated to the liver and established residency [23].

Proliferation.

In adult mice, neither ILC1 adoptive transfer nor bone marrow (BM) transplantation leads to complete reconstitution of the liver ILC1 compartment, suggesting that not all ILC1s fully develop from BM progenitors. Indeed, this failure to reconstitute after adoptive transfer, seen for many tissue-resident lymphocytes, has hampered some studies examining ILC1 biology. As mentioned earlier, Bai et al. demonstrated the IFN-γ-dependent proliferation of liver ILC1s [62]. As previous studies by this group had shown that selective deletion of NK cells in Ncr1-cre Eomes^fl/fl mice did not impact ILC1 development [31], the authors conclude that ILC1s can produce IFN-γ that positively regulates their maintenance and leads to local proliferation. It will be interesting to determine whether infections leading to enhanced IFN-γ production (by ILC1s or other infiltrating cells) results in increased numbers or local densities of ILC1s. We have shown during VACV infection that NK cells traffic into the infected tissue and produce IFN-γ, yet most are not recruited into the epithelium in proximity to ILC1s, which are maintained in relatively steady numbers in the oral mucosa for the first five days of infection [21]. This suggests there may be a yet-to-be-appreciated spatial component to this IFN-γ signaling loop.

Age-associated depopulation.

Though not extensively examined, aging may contribute to shifts in ILC1 populations in different tissues. For instance, while the skin of adult mice has few ILC1s [65]. Yang et al. recently demonstrated that neonatal thymi support the development of skin-homing ILC1s, similar to the early development of γδT cells [64]. In contrast, skin ILC1s were generated in both the thymus (albeit to a lesser extent) and the skin-draining lymph nodes, though the lymph nodes preferentially support ILC2/3 development. Thus, ILC1s are eventually replaced in the skin by ILC2s and ILC3s. In the oral mucosa, ILC1s are likely perinatally established and influenced by the microbiota, which changes dramatically from birth to weaning and adulthood [66].

NK conversion to ILC1s.

Inflammation can increase ILC1 numbers through the conversion of other cells into bona fide ILC1s or ILC1-like cells. NK conversion to ILC1s during viral infection has not been well documented; nonetheless, experiments in other murine models suggest this likely occurs during virus-driven inflammation. In a mouse fibrosarcoma tumor model [40], the largest population of group 1 ILCs recovered from the tumor environment were not canonical NK cells or ILC1s, but rather “intermediate” cells expressing both CD49a and CD49b and a distinct transcriptional profile. The same study showed that mature splenic NK cells could repopulate the liver with both ILC1s and intermediate ILC1s, which proliferated more readily than either NK cells or ILC1s. NK cell conversion also occurred after murine infection with Toxoplasma gondii [67]. Although few ILC1s were found in the spleen of uninfected mice, a clear population of ILC1-like cells was present by 21 days post-infection. While these cells expressed markers typical of ILC1s, they also expressed the NK cell activation marker Ly6c and the T_reg activation-associated molecule neuropilin-1. These phenotypic changes in ILC1-like cells were dependent on the downregulation of Eomes. Interestingly, these ILC1-like cells were also circulatory, retained a high level of proliferative capacity and depended on Stat4 for expansion. More data will need to be gathered during infection to unambiguously determine if ILC1-like cells arise from ‘ex-NK’ cells rather than ILC1 expansion. Additionally, it will be important to identify the extent of interconversion during virus-induced inflammation in comparison to other modes of inflammation with prominent NK infiltration.

Though some conflicting data exist in different systems, the cytokine TGF-β appears to serve as a critical regulator of ILC1 development as well NK/ILC1 conversion [20, 68]. The TGF-β-rich fibrosarcoma environment favored NK conversion to non-cytolytic, intermediate ILC1s, which failed to control fibrosarcoma growth [40]. These data suggest that TGF-β might directly signal NK conversion. Interestingly, however, conditional deletion of Smad4, an important TGF-β signal transducer, in NK cells led to an increase in NK expression of ILC1 markers (including CD49a), which similarly impaired melanoma control [20]. This suggests conversion might also occur through a SMAD4-independent TGF-β signaling pathway. Many viruses, including SARS-CoV-2, also modulate TGF-β signaling [69, 70]; thus, some viral infections could also shift the local environment to a cell type less capable of cytolysis, favoring viral replication.

ILC3 conversion to ILC1s.

NK cells are not the only cells to convert to ILC1s. ILC3s depend on RORγt for their development, express NKp46, share many transcriptional patterns with ILC1s [71] and can also convert to ILC1s [72–75]. As ILC3s protect barrier epithelia (such as the small intestine), conversion to an ILC1-like phenotype with IFN-γ production can exacerbate inflammation and tissue pathology [76]. Conversely, this increase in IFN-γ production might enhance viral clearance (as seen during human AIRE deficiency [77]), though this has not yet been demonstrated. Like SMAD4 in NK cells, the transcription factor c-Maf serves as a negative regulator of ILC3 conversion by regulating chromatin accessibility and maintaining a reduced expression of T-bet and ILC1 genes [78]. In isolated human tonsillar ILC3s, conversion relied on transitional cell subsets and increased their expression of ILC1 genes such as T-bet and IFN-γ [79]. The Ikaros family transcription factor, Aiolos, tandemly functioned with T-bet in ILC1-like transitionary cells to suppress ILC3 genes. Interestingly, TGF-β stimulation upregulated Aiolos expression while inhibiting RORγt expression [79]. Thus, TGF-β may be important for the conversion of both ILC3s and NK cells into ILC1s. In a separate report, human tonsillar ILC1s have been shown to convert to ILC3s [75]. More work will need to be performed with viral infection models in areas rich in ILC3s to determine whether conversion is induced by viral infection and whether this promotes or hinders viral clearance.

ILC1-mediated protection during viral infection

As ILC1s are oft described as the innate counterparts of Th1 cells, it follows that they would limit intracellular pathogen replication through the production of IFN-γ. Despite this prevalent view, to date, only a handful of studies have unequivocally demonstrated that ILC1s directly limit viral replication and protect against virus-induced pathology. In a seminal study in 2017, Weizman and colleagues demonstrated ILC1 control of viral replication at initial points of virus entry in the tissue [35]. Intranasal infection with Sendai virus (SeV or murine parainfluenza virus type 1) and intraperitoneal or hydrodynamic injection of murine cytomegalovirus (MCMV) led to an increase in the production of IFN-γ by local ILC1s within 24 hours of infection. To more carefully dissect the contribution of ILC1s to host protection, Weizman et al. identified Hobit as a major transcription factor present in liver-resident ILC1s but not NK cells. Accordingly, Hobit-deficient mice (.Zfp683^−/− mice) lacked liver-resident ILC1s but not cNKs. Either NK1.1-treated or Hobit-deficient mice rapidly succumbed to hydrodynamic injection of MCMV, while WT animals did not. Thus, ILC1s alone can provide sufficient immunity to protect against virus-induced morbidity with this route of infection. As ILC1s provide initial protection against infection, their role may be most profound with high levels of inoculum in which mice quickly succumb to infection, rather than in infections with slow and protracted viral replication in the tissue.

Even before the COVID-19 pandemic, there was interest in the function of ILC1s during respiratory virus infections. Like Weizman’s finding with SeV, intranasal infection with mouse-adapted influenza virus (IAV strain A/PR/8/34) leads to a rapid increase in IFN-γ production by ILC1s that is maintained, in this case, for at least 6 days post-infection [80]. Adoptive transfer of ILC1s into deficient mice resulted in reduced lung titers on day 3 post-infection. Thus, although ILC1s are not the predominant lung-resident innate immune cells, they are still poised to impact respiratory viral infection.

In a different mucosal tissue, our group has recently demonstrated that ILC1s in the mouse’s inner lip (labial mucosa) limit VACV replication by secreting IFN-γ even before viral infection [21]. Tonic production of IFN-γ upregulated numerous antiviral genes, including IRF7 and STAT1, in uninfected cells in the oral mucosa, which then prevented virus spread. This mechanism for curtailing virus replication may be particularly relevant for viruses that encode immunomodulatory proteins, such as IFN decoys or receptors, that can circumvent innate signaling in infected cells. Intriguingly, depletion of NK1.1⁺ cells before infection led to the most dramatic increase in viral titer; by 9 hours there was little effect of depletion. This suggests that environmental signals in the mucosa can drive ILC1 activation and prevent infection without upregulation of IFN-γ production by stimulation with viral products. As VACV has broad tropism, it will be interesting to determine if this is true in multiple tissues and to what extent ILC1s can restrict viral replication in different environments.

We still don’t fully understand the distinct contributions of ILC1s versus NK cells to the antiviral immune response. Weizmann et al. demonstrated that ILC1-IFN-γ production proceeded more rapidly than that in NK cells after SeV infection [35]. A separate study established that ILC1s also secrete higher levels of TNF-α and IL-2 compared to NK cells after intraperitoneal infection with VACV, although the immunological implications of this remain to be determined [29]. We recently showed that ILC1s present in the tissue can produce an antiviral state even in uninfected tissue. Paradoxically, depletion of NK1.1⁺ cells on day 2 post-infection protected tissue, suggesting that in some circumstances, newly recruited NK cells might result in tissue pathology [21]. More studies will need to be performed to unequivocally determine the unique role of each cell type during the antiviral immune response.

Indirectly, ILC1s can modulate the adaptive immune response to impact viral protection, particularly by either enhancing or dampening antiviral CD8⁺ T cells responses. In the liver, ILC1 activation resulted in the production of the potent T and NK cell-attracting chemokine CXCL9 [81]. During adenovirus (Ad) liver infection, reduced ILC1 activation occurring via inhibitory receptor NKG2A signaling limited CXCL9-dependent NK cell recruitment, which led to lower levels of DC activation and antiviral CD8⁺ T cell priming. Similarly, during chronic lymphocytic choriomeningitis virus (LCMV) infection, ILC1s upregulated PD-L1, which reduced the function of hepatic antiviral CD8⁺ T cells [82]. Ad5 infection of the female reproductive tract (FRT) led to IFN-γ production by NK1.1⁺CD3-cells (likely ILC1s) [83]. This, in turn, resulted in monocyte production of CXCL9 and recruitment of antiviral CD8⁺ T cells. Together, these data demonstrate that changes in ILC1 activation or inhibition can cascade to dramatically alter antiviral T cell responses via several different mechanisms.

The mechanisms leading to ILC1 activation after viral infection are still an area of active investigation. During MCMV infection, XCR1⁺ DCs in the liver produced IL-12 in response to infection; IL-12 signaling in ILC1s resulted in IFN-γ upregulation [35]. Interestingly, activated murine invariant NKT cells (iNKTs) generate systemic IL-12 in response to CD1d ligands, which results in IFN-γ production by lung and splenic ILC1s [84]. Accordingly, iNKT-meditated activation of ILC1s before infection with IAV led to decreased viral titers in the lung [85]. As ILC1s do not require IL-18 for IFN-γ production, the threshold for activation by IL-12 may be lower in these tissue-resident ILC1s than in circulating NK cells [86]. We speculate that other potential mechanisms of ILC1 activation may include stimulation from the microbiota or direct activation via recognition of viral proteins. Along these lines, STAT1 signaling was required for maximal IFN-γ⁺ ILC1s after respiratory syncytial virus (RSV) infection, suggesting that infection-induced type I interferons (IFN-I) drive ILC1 activation [87]. As the microbiota can induce tonic IFN-I in the absence of overt viral infection [88], commensals alone may be sufficient to induce IFN-γ–producing ILC1s, particularly in heavily colonized tissues.

Evidence for enhanced ILC1-mediated immunity to secondary infection

While immunologic memory was once thought to be the sole purview of adaptive lymphocytes, it has become evident in recent years that innate lymphocytes can also mount enhanced responses to secondary exposure to Ag [89]. This phenomenon, termed “trained immunity,” involves the long-term reprogramming of innate lymphocytes to respond in an altered way to a secondary challenge. Both epigenetic and metabolic reprogramming can lead to innate trained immune responses [89].

Though mostly studied in the context of other innate leukocytes, group I ILCs also exhibit hallmarks of trained immunity. Enhanced secondary responses by NK cells were first reported over a decade ago and have now been described in mice, non-human primates, and humans (reviewed in [90, 91]). Interestingly, in pioneering studies, Paust et al. demonstrated that adoptive transfer of liver-resident NK cells from virus-vaccinated mice protected recipients during secondary viral challenge with IAV or vesicular stomatitis virus (VSV); this protection was virus-specific as IAV-sensitized NK cells did not protect against VSV and vice versa [92]. In accord with our current understanding of ILC1s versus NK cells, these cells represented CD49a⁺CXCR6⁺ liver-resident ILC1s and required CXCR6 for their enhanced secondary responses, as injection of an anti-CXCR6 Ab abolished protection. Wang et al. later showed that IL-7Ra⁺ ILC1s use CXCR6 to establish residency in the liver, maintaining numbers through IL-7R signaling [23]. They also formally demonstrated that the liver-resident NK1.1⁺ cells mediating enhanced secondary responses (albeit to haptens and not viruses) were indeed ILC1s.

Recently, Weizman et al. reported that the viral protein m12 from MCMV could directly stimulate ILC1 memory responses [93]. After intraperitoneal MCMV infection, liver ILC1s proliferated in response to proinflammatory cytokine signaling with a subset maintaining IL-18R expression for an extended time point post-infection. After MCMV rechallenge, IFN-γ⁺ liver IL-18R⁺ ILC1s increased from ~2% to ~6%, while only a nominal increase was seen during primary challenge or with IL-18R⁻ ILC1s. Challenge experiments in Rag2^−/−-mice demonstrated a reduction (~2.5 fold) in liver viral genomes during viral rechallenge compared to primary infection, although this reduction was not specifically shown to depend on ILC1s. MCMV strains lacking glycoprotein m12 failed to induce enhanced secondary responses. The specific ILC1 receptor recognizing m12 has not yet been identified. More investigation is needed to determine if ILC1s employ other virus-specific receptors to acquire trained immunity or if these receptors detect viral proteins with similar homology shared by viral families.

Memory T cells exhibit transcriptional and epigenetic changes related to their conversion to a memory phenotype [94, 95]. Likewise, IL-18R⁺ ILC1s isolated from MCMV-infected mice had increased transcription of approximately 800 genes compared to ILC1s from naïve mice, including Ifng; the enriched genes aligned more closely with memory T cells than those from naive mice [93]. ATAC-seq in these same cells detected changes in chromatin accessibility for a number of the genes, suggesting stable epigenetic changes in memory-phenotype ILC1s. Thus, these virus-experienced ILC1s possess all the features of adaptive memory: expansion, contraction, altered gene expression and epigenetics, and enhanced ability to reduce viral replication.

The dark side of ILC1s?

While ILC1s limit viral infection, like other immune effector cells, there is also some indication that they may also cause local pathology in some circumstances. Because ILC1 numbers increase dramatically in mucosal tissues during certain diseases, such as irritable bowel syndrome [96], ILC1s were originally ascribed a primarily pathogenic role in the tissue. Experimental evidence for enhanced pathology during viral infection remains limited. Although associations exist between alterations in ILC1s during several human viral infections, current studies remain largely correlative [97–99]. The best current experimental model of the potential role of ILC1s in immune-mediated pathology occurs in the CNS. Infection of neonatal mice with MCMV induces CXCR3-dependent recruitment of NK cells and ILC1s (~40% of NKp46⁺ cells) into the brain [100]. Though the authors did not separate the function of ILC1s and NK cells in this study, infiltration resulted in significant immunopathology in the brain which was ameliorated with IFN-γ-neutralizing Abs. NKp46⁺ ILCs also mediate neuroinflammation during non-viral experimental encephalitis [101]. More studies will need to be done to determine if this extends to encephalitic or other systemic viruses.

Perspectives and future directions

While our understanding of anti-viral ILC1s is becoming much clearer, several outstanding questions remain concerning the precise functions of these cells in the tissue.

Are antiviral TRM and ILC1s redundant?

Both tissue-resident memory T cells (T_RM) and ILC1s reside in barrier tissues and provide frontline protection against viral infection. Based on human data, Vély et al. have posited that ILC1s might be dispensable when human T cell and B cell responses are intact, though this hypothesis has been questioned [102]. In support, humans with severe combined immunodeficiency that received a hematopoietic stem cell transplant did not repopulate with ILC1s, yet also did not exhibit enhanced susceptibility to common diseases (including viral infections). Others have questioned this finding, citing the inability to test most tissues for the presence of ILC1s and infection of the patients with human papillomavirus [103, 104]. Thus, it will be interesting to further elucidate the anti-viral protection afforded by T_RM and ILC1s in both humans and experimental animal models to determine the extent of functional redundancy.

Do ILC1s function in an antiviral capacity in lymph nodes?

The lymph node (LN) is often thought of as the site of initiation of adaptive immune responses, though it also plays an innate role in filtering draining pathogens and preventing them from entering the circulation. It is currently unknown whether ILC1s can contribute to viral clearance in the node, either directly or indirectly through the activation of other cells. Although not a prominent population in the lymph node (LN) at steady state, ILC1s can be recruited to the node under the few inflammatory conditions thus far examined. Wang et al. demonstrated that hapten exposure induced the accumulation and “priming” of IL-7Ra⁺ ILC1s in skin draining lymph nodes (LNs); these LN-resident ILC1s produced increased levels of IFN-γ compared to their naive counterparts [23]. Interestingly, the increased numbers of ILC1s in the draining LN required ILC1 expression of CXCR3, an important receptor for the recruitment of both NK cells and effector T cells to the draining LN. Further, LN-deficient mice failed to mount hapten-induced memory responses. A recent analysis of single-cell RNA-seq from murine inguinal LNs revealed that LN ILC1s exhibited transcriptional similarity to ILC1s present in other organs and likely represented circulating cells [46]. Thus, the LN may represent an important site for the development of trained ILC1 immunity and may also serve as an initial wave of protection against draining virions.

Do ILC1s provide antiviral protection through tissue protection?

ILC1s likely protect against viral infection through tissue maintenance and promotion of epithelial barrier integrity. In the liver, ILC1s protected against drug-induced acute liver injury through the production of IFN-γ, which upregulated the anti-apoptotic protein Bcl-xL in damaged hepatocytes [105]. In mouse models, MCMV infection of the liver also upregulated ILC1-production of IFN-γ to rein in viral replication [35]. Thus, it seems likely that ILC1s would also protect the liver from enhanced viral spread in damaged tissue. Recently, Jowett et al. used intestinal organoids to demonstrate that ILC1s secreted TGF-β and matrix metalloproteinase 9 (MMP9) to drive intestinal remodeling and promote epithelial integrity [106]. Although confirmation is still needed using in vivo models, this suggests that ILC1s may play a prominent role in mucosal epithelial maintenance, where they are often enriched. As epithelial integrity is important to prevent the infiltration and spread of many viruses, ILC1s may confer protection through this indirect function.

How do ILC1s exert their functions in the context of the complex tissue environment?

Relatively little is known regarding the spatial organization and temporal dynamics of ILC1s during viral infection. Using intravital microscopy and reporter mice, we have shown that ILC1s in the oral mucosa slowly patrol the basal layers of the murine oral epithelium at steady state. Intriguingly, after infection, most ILC1s did not accumulate directly around areas of viral infection, suggesting that their primary purpose was not to provide IFN-γ directly to infected cells [21]. In our system, we could not separate NK cells from ILC1s during live imaging. Thus, it will be critical to track the movement of ILC1s specifically during viral infection. Studies of the spatial dynamics of ILC1s will also be very interesting to better understand the role of IFN-γ in local ILC1 proliferation. Will there be pockets of proliferation or is this completely cell-autonomous? In future studies, it will also be important to test whether insights gained in one tissue (in this case – the oral mucosa) translate into other tissues, particularly those with distinctly different environments such as the liver.

Concluding remarks

In the relatively short time since the discovery of ILC1s, scientists have uncovered myriad functions for these remarkable tissue-resident innate lymphocytes. Despite such rapid progress, more experiments are needed to fully understand the role of ILC1s in antiviral protection in each unique tissue environ and against a range of pathogens.

Highlights.

• ILC1s are tissue-resident innate lymphocytes with diverse functions.

• ILC1s are maintained in the tissue during homeostasis and infection.

• ILC1s limit early virus replication and dissemination.

• ILC1s may exhibit enhanced secondary responses to viral infection.

Abbreviations

Ad

Adenovirus

Ag

antigen

CHILPs

common helper lymphoid ILC-progenitor cells

CILPs

common innate lymphoid progenitor cells

cNKs

conventional natural killer cells

Eomes

Eomesodermin

FRT

female reproductive tract

HPV

human papilloma virus

IAV

influenza virus

IFN-γ

interferon-gamma

IRF7

interferon regulatory factor 7

iNKTs

invariant NKT cells

LCMV

lymphocytic choriomeningitis virus

LN

lymph node

MCMV

murine cytomegalovirus

NK

natural killer

SeV

Sendai virus

T_RM

tissue-resident memory T cells

TGF-β

transforming growth factor-beta

VACV

vaccinia virus

VSV

vesicular stomatitis virus