Innate Lymphoid Cells in Graft‐Versus‐Host Disease

Innate lymphoid cells (ILC) are lymphocytes lacking rearranged antigen receptors such as those expressed by T and B cells. ILC are important effector and regulatory cells of the innate immune system, controlling lymphoid organogenesis, tissue inflammation, and homeostasis. The family of ILC consists of cytotoxic NK cells and the more recently described noncytotoxic group 1, 2, and 3 ILC. The classification of noncytotoxic ILC—in many aspects—mirrors that of T helper cells, which is based on the expression of master transcription factors and signature cytokines specific for each subset. The IL‐22 producing RORγt+ ILC3 subset was recently found to be critical in the prevention of intestinal graft‐versus‐host disease (GVHD) following allogeneic hematopoietic cell transplantation (HCT) via strengthening the intestinal mucosal barrier. In this review, we summarize the current view of the immunological functions of human noncytotoxic ILC subsets and discuss the potentially beneficial features of IL‐22 producing ILC3 in improving allo‐HCT efficacy by attenuating susceptibility to GVHD. In addition, we explore the possibility of other ILC subsets playing a role in GVHD.

Innate lymphoid cells (ILC) are lymphocytes lacking rearranged antigen receptors such as those expressed by T and B cells. ILC are important effector and regulatory cells of the innate immune system, controlling lymphoid organogenesis, tissue inflammation, and homeostasis. The family of ILC consists of cytotoxic NK cells and the more recently described noncytotoxic group 1, 2, and 3 ILC. The classification of noncytotoxic ILC-in many aspects-mirrors that of T helper cells, which is based on the expression of master transcription factors and signature cytokines specific for each subset. The IL-22 producing RORgt þ ILC3 subset was recently found to be critical in the prevention of intestinal graft-versus-host disease (GVHD) following allogeneic hematopoietic cell transplantation (HCT) via strengthening the intestinal mucosal barrier. In this review, we summarize the current view of the immunological functions of human noncytotoxic ILC subsets and discuss the potentially beneficial features of IL-22 producing ILC3 in improving allo-HCT efficacy by attenuating susceptibility to GVHD. In addition, we explore the possibility of other ILC subsets playing a role in GVHD.
Abbreviations: Ahr, aryl hydrocarbon receptor; AMP, antimicrobial peptide; GVHD, graft-versus-host disease; HCT, hematopoietic cell transplantation; IFN, interferon; ILC, innate lymphoid cells; ISC, intestinal stem cells; KLRG1, killer cell lectin-like receptor subfamily G member 1; LTi, lymphoid tissue inducer; NCR, natural cytotoxicity receptor; TNF, tumor necrosis factor; TSLP, thymic stromal lymphopoietin  (1), stimulating tissue regeneration (2) and strengthening the epithelial barrier at mucosal surfaces (3). A detailed description of the tissue-specific functions of the human ILC subsets have recently been provided elsewhere (4). The family of ILC includes the cytotoxic natural killer (NK) cells, as known since the 1970s for their ability to kill target cancer cells and produce interferon (IFN)-g. In addition to NK cells, three major subsets of noncytotoxic ILC have been described during the last decade. The classification of these noncytotoxic ILC into group 1, 2, and 3 ILC is based on their expression of specific transcription factors and cytokine secretion profiles largely reflecting the Th1/Th2/ Th17 subsets ( Figure 1) (5). Additionally, ILC subsets are defined by the expression pattern of certain cell surface proteins, which are partly different in humans and in mice (Table 1).
ILC show lymphoid morphology but lack rearranged antigen-specific receptors and lineage markers associated with other lymphocytes such as T and B cells. Alike these latter cell types, ILC develop from the common lymphoid progenitor (CLP). However, in contrast to T and B cells, ILC critically depend on the transcriptional repressor Id2 for commitment to the ILC lineage (6). The Id2-dependent ILC progenitor (ILCp) or common innate lymphoid progenitor (CLIP) gives rise to all ILC subsets including NK cells. Furthermore, as a progenitor of ILC1, ILC2, and ILC3, the common helper-like innate lymphoid progenitor (CHILP) was identified, which depends on the transcription factors PLZF, GATA3, NFIL-3, and Notch (6,7). The development of human ILC and their progenitor(s) are less well characterized. However, recently, the human progenitor of IL-22 producing ILC3 was described as RORgt þ CD34 þ hematopoietic stem cells selectively residing in the tonsil and intestinal lamina propria (8).

ILC1
Group 1 ILC, similarly to NK and CD8 þ T cells, depend on Tbet and produce IFN-g and TNFa in response to IL-12 and IL-18 stimulation.
In humans, based on IL-7Ra (CD127) expression, two ILC1 populations can be distinguished. The intraepithelial CD127 À ILC1, found in the human intestine, express CD56, CD94, NKp44, and CD103. This cytotoxic ILC1 subset exists in mouse as well, where it was found to express CD160 and NKp46 but not CD103. Furthermore, this subset was NFIL-3-, and T-bet-dependent and IL-15responsive (11). Importantly, these cells, although cytotoxic, were shown to be distinct from NK cells as they also developed in il15r À/À mice lacking NK cells.
The second subgroup of ILC1 in humans, the CD127 þ ILC1 was found in the inflamed human gut lamina propria, and do not express CD56, CD94, and NKp44 (12). In the mouse, another CD127 þ T-bet þ NKp46 þ Eomes À non-NK ILC1 lineage was identified (13). This particular ILC1 subset showed poor cytotoxicity, but was the major producer of IFN-g and TNF in response to Toxoplasma gondii intestinal infection.
NCR þ refers to NKp44 þ ILC3 in human and NKp46 þ ILC3 in mouse. Expression level of the cell surface markers is displayed as follows: þ indicates high expression, À indicates no expression, þ/À indicates bimodal expression. ND indicates that expression level is not defined and NA indicates that the marker is not applicable in the respective species.
Konya and Mj€ osberg ILC1 are most probably involved in inflammation by producing IFN-g upon sensing epithelial and myeloid cellderived danger signals. Both CD127 À and CD127 þ ILC1 are highly accumulated in the inflamed ileum of Crohn's disease patients, potentially contributing to the pathogenesis of gut inflammation (11,12).

ILC2
ILC2 secrete type 2 cytokines; IL-4, IL-5, IL-9, and IL-13 in response to IL-33, IL-25, and thymic stromal lymphopoietin (TSLP) (14)(15)(16). Additionally, ILC2 produce amphiregulin, which is crucial in tissue regeneration (2). For the development and function of human and mouse ILC2, the transcription factors GATA3 and RORa are essential (17). In the peripheral blood, the majority of CD127 þ ILC are ILC2, additionally, group 2 ILC have been found in the healthy human lung (2,18) and in the skin (19,20). An updated view on ILC2 function in pulmonary and dermal inflammatory diseases is provided elsewhere (21). In humans, ILC2 are abundant in nasal polyps of patients with chronic rhinosinusitis (CRS) and in bronchoalveolar lavage (BAL) of idiopathic pulmonary fibrosis patients (18,22). In the skin, increased presence of ILC2 is associated with atopic dermatitis both in human and mouse as reviewed elsewhere (21). Furthermore, ILC2 were identified in the mouse adipose tissue regulating fat metabolism (14).

ILC3
Group 3 ILC are characterized by IL-22 and IL-17 secretion in response to IL-23 and IL-1b stimulation (5,23,24). ILC3 express the transcription factor RORgt, and the development of mouse ILC3 depends on RORgt, Notch, and Tcf1, as recently summarized elsewhere (6). Additionally, the transcription factor aryl hydrocarbon receptor (Ahr) is crucial in the maintenance and function of ILC3 (4). The first subset of ILC3 to be described was the lymphoid tissue inducer (LTi) cells, essential for proper lymph node formation during embryogenesis and for the formation of lymphoid follicles in the postnatal intestine, as extensively reviewed elsewhere (25). Human ILC3 can be further divided into two subsets based on expression of the natural cytotoxicity receptor (NCR) NKp44. NCR þ ILC3 are cellular sources of the tissue protective cytokine IL-22 (4). Additionally, engagement of NKp44 triggers TNF release providing a pro-inflammatory function for NCR þ ILC3 (26). The NCR À ILC3 secrete little IL-22 but do produce IL-17, similarly to the LTi-like ILC3 in the mouse (25). In the mouse, the ILC3 populations are differentially defined. Murine a4b7 þ progenitors give rise to CCR6 þ T-bet À ILC3, such as LTi cells, which heterogeneously express CD4. The a4b7 À ILC3 progenitors depend on AhR for their development into CCR6 À T-bet þ ILC3, which may or may not express the NCR NKp46 (5).
NCR À ILC3 make up one of the most frequent ILC subsets in the human peripheral blood, while NCR þ ILC3 are virtually absent in the circulation during homeostasis (20). Importantly, the healthy human intestine hosts dominantly NCR þ ILC3, which have been shown to be essential in maintaining gut homeostasis (1). The signature cytokine of NCR þ ILC3 is IL-22, a cytokine belonging to the IL-10 cytokine family. Exceptionally among cytokines, IL-22 does not target immune cells, but stromal cells and epithelial cells at physical barriers, such as skin, gut, and liver (27). Mouse studies established a crucial role for IL-22 in gut homeostasis, since ILC3-derived IL-22 protected against Citrobacter rodentium-induced colitis (1). Furthermore, IL-22-producing ILC3 in the gut were found to control anatomical containment of commensal bacteria, thereby protecting against intestinal inflammation (28). Intestinal epithelial cells express the IL-22 receptor (IL-22R) and IL-22 directly induces production of the Reg family of antimicrobial proteins (AMPs) (29). Reg3b and Reg3g are important regulators of epithelial barrier function and control the microbial colonization at epithelial surfaces. Further notions on how ILC3 and other ILC subsets may control intestinal homeostasis and inflammation are discussed elsewhere (27).
Interestingly in the skin, accumulation of IL-22 producing NCR þ ILC3 is linked to dermal inflammation in psoriasis patients (21).

Development of GVHD
The only curative treatment for hematologic malignancies nonresponding to induction chemotherapy is allogeneic HCT. However, alloreactive donor T cells can attack vital organs by recognizing recipient antigens, which leads to development of GVHD. The occurrence of GVHD is the major limiting factor in the therapy success of HCT and represents high morbidity and mortality. Prior to HCT, patients with malignant hematological disorders receive conditioning chemotherapy and radiotherapy in order to cytoreduce the malignancy and to deplete the nonmalignant host immune system to avoid rejection of the transplant. However, conditioning chemotherapy affects not only the immune cells but also the rapidly proliferating epithelial cells lining the mucosal barrier of the gastrointestinal tract and the skin (30). The consequence of compromised barrier function is uncontrolled spreading of commensal bacteria and tissue injury, which may further stimulate donor alloreactive T cells, and thereby aggravate tissue damage in the gut, skin, and liver (31). This syndrome is clinically recognized as acute (a)GVHD, which typically occurs within the first 3 months after transplantation. Major characteristics of aGVHD are activated antigen-presenting cells (APC), pro-inflammatory cytokine milieu, and enhanced recruitment and activation of effector T cells. Gastrointestinal aGVHD or mucositis is an especially important component of transplant morbidity and mortality, since it cannot be efficiently controlled by immunosuppressive corticosteroid therapy (32). Notably, the colonic ILC in GVHD mucosal tissue alterations of GVHD patients were reminiscent of chronic inflammatory bowel disease (33). The currently existing therapy of aGVHD focuses exclusively on the donor or graft immune responses against the recipient tissue (31) but the host responses during GVHD and the recipient-specific potential therapeutic target mechanisms are very poorly understood. Additionally, the regulation of recipient GVHD sensitivity has not been addressed yet.

Potential of IL-22 Producing ILC3 in GVHD Therapy
Lessons from mouse studies The critical involvement of ILC in GVHD immunity was evidenced for the first time in a mouse model of acute intestinal GVHD, where IL-22 producing ILC3 were found to attenuate GVHD (9). In particular, host IL-22 deficiency resulted in increased incidence and severity of GVHD with excessive epithelial cell apoptosis and disrupted intestinal mucosal barrier. Furthermore, GVHD caused the most significant damage to the intestinal stem cells (ISC), which was even more pronounced in case of host IL-22 deficiency. ISC are located at the base of the intestinal epithelial crypts between Paneth cells and ISC are able to generate the whole crypt-villus structures in vivo and in vitro (34). The authors identified IL-22R expression on ISC and epithelial progenitor cells, which-together with mature intestinal epithelial cells-produce antimicrobial proteins Reg3b and Reg3g in response to IL-22 (29). In IL-22-deficient host mice, Reg3b and Reg3g mRNA was dramatically reduced which was associated with increased epithelial damage and loss of intestinal stem cells (9). Thus, IL-22 appears as a crucial host-derived factor modulating susceptibility to GVHD (Figure 2). The known cellular sources of IL-22 are T cells and ILC3 and very recently an IL-22 reporter mouse revealed that the major and most stable IL-22-producing cells are the NKp46 À CD4 À ILC3, while only a minority of NKp46 þ ILC3 expressed IL-22 (35).
In line with these recent findings, only IL-22 produced by the IL-23-responsive RORgt þ CCR6 þ NKp46 À ILC3 proved to be essential in the protection against GVHD (9). Importantly, intestinal IL-22 producing ILC3 were of recipient origin, as being radio-resistant they persisted following lethal conditioning radiotherapy, bone marrow transplantation, and even after donor T cell reconstitution in the lamina propria.
The significance of ILC3-derived IL-22 after immunodepletion was also shown in the enhanced regeneration of thymic epithelial cells and in the reconstitution of thymic T cell compartments (36). The identified cellular cascade after thymic injury consisted of upregulation of radioresistant CD103 þ DCs providing IL-23, which in turn stimulated IL-22 release from RORgt þ CCR6 þ NKp46 À LTi cells in the thymus. This supported proliferation and recovery of the epithelial microenvironment and boosted thymopoiesis.

Lessons from human studies
Recently, human peripheral blood ILC recovery was followed longitudinally in acute myeloid leukemia (AML) patients after induction chemotherapy and allogeneic HCT (10). Human ILC were depleted after induction and conditioning chemotherapy and their complete reconstitution post-HCT required more than 6 months. Reconstituted ILC were donor-derived and all ILC subsets showed higher expression of the activation marker CD69 and the proliferation marker Ki-67 as compared to ILC from healthy individuals' blood. Already after induction chemotherapy, but prior to allogeneic HCT, ILC2 and ILC3 subsets showed a dichotomy in CD69 expression. Patients with increased expression of gut-homing markers on CD69 high ILC showed lower incidence of intestinal GHVD. Additionally, NCR þ ILC3 appeared rapidly in the peripheral blood after HCT, while this ILC subset is absent in the circulation under homeostatic conditions. Notably, 12 weeks after HCT, the circulating NCR þ ILC3 count was highest in patients that did not develop GVHD. Further associations were observed between the phenotype of reconstituted ILC and the development of aGVHD; (1) the activated CD69 high ILC subsets expressed more homing markers for gut (a4b7, CCR6) and skin (CLA, CCR10), (2) the appearance of CD69 high ILC, especially NCR þ ILC3, lowered the incidence of aGVHD after HCT (Figure 2). In contrast, patients with a . After HCT, NCR þ ILC3 expressing CD69 and the guthoming markers a4b7, CCR6 appear in the circulation most probably via upregulation of IL-23. The activated NCR þ ILC3 produce IL-22, which is required for the expression of antimicrobial peptides (AMPs) by the ISCs and the epithelial progenitors. AMPs limit the tissue damage caused by conditioning chemotherapy and immune depletion (left panel). Additionally, there are clues for an association between protection against GVHD and ILC2 and NCR À ILC3 subsets, however, in a yet uncharacterized manner. NCR, natural cytotoxicity receptor; ILC, innate lymphoid cells; GVHD, graft-versus-host disease; HCT, hematopoietic cell transplantation.
These findings point toward an important role of IL-22 producing ILC3 in preventing GVHD development in humans and are in line with previous observations in a mouse model of acute intestinal GVHD (9,10). It is very likely that IL-22 producing NCR þ ILC3, as epithelial tissue protecting cells, are crucial in the reintegration of mucosal barriers after immune depletion and allogeneic HCT. Thereby, NCR þ ILC3 suppress further tissue damage, alloreactive T cell activation and the occurrence of GVHD ( Figure 2) (31). Furthermore, IL-22 cytokine-based therapy has already been considered for prevention of GVHD and lung transplantation complications (37). The active status and gut/skin homing receptor expression of ILC, especially the IL-22-producing ILC3, seems to minimize GVHD development following HCT. Therefore, screening patients for activation molecules and homing integrins on ILC3 prior to allogeneic HCT can be critical in anticipating the patient's sensitivity to subsequent GVHD. In case that this association between ILC3 and susceptibility to GVHD is further confirmed in future independent cohorts, supplying acute myeloid leukemia patients with IL-22 or IL-22-producing ILC3 might be a mean to prevent GVHD. IL-22 could be a relatively safe target in enhancing mucosal healing, since it does not directly affect immune cell functions. However, possible adverse effects might occur, for example, due to dysregulated responses of keratinocytes in the skin (20,37). Of note, in an earlier large-scale clinical study investigating human gastrointestinal GVHD risk factors, Reg3a protein was identified as indicator of higher incidence for developing GVHD after allo-HCT (38). As mentioned previously, the antimicrobial Reg3 proteins are downstream of IL-22/IL-22R pathway, mainly produced by epithelial cells and ISC. This finding suggests possible differences in the underlying mechanism of GVHD development in mouse and human.

Possible Roles for Other ILC Subsets in GVHD
In the above mentioned study, in the patients that did not develop GVHD, not only the NCR þ ILC3, but all circulating ILC subsets expressed high levels of the activation marker CD69 and either skin-or gut-homing integrin molecules (10). This suggests that ILC1, ILC2, and the NCR À ILC3 might also be associated with protection against GVHD. Particularly, the CD69 high ILC2 and NCR À ILC3 subsets expressed significantly higher levels of the gut-homing marker a4b7 in patients without GVHD. However, the authors stated that further larger cohorts are necessary for studying the relation between these ILC subsets and GVHD sensitivity. Theoretically, these ILC subsets and their released cytokines and other mediators might also contribute to the GVHD pathology.

ILC2 in GVHD
Mouse studies based on Th2 cells provided evidence for a protective role for type 2 cytokines, such as IL-4, IL-5, IL-9, and IL-13, in GVHD (31). These cytokines are also extensively produced by ILC2. All together, variable effects had been reported about Th2 cytokines. Interestingly, at least in mice, ILC2 produce the EGF-related factor amphiregulin, which was shown to trigger epithelial tissue repair and restore lung tissue homeostasis after influenza virus infection (2). This notion suggests that ILC2 could possibly affect the sensitivity to GVHD via boosting epithelial cell regeneration after the conditioning therapyinduced tissue damage. However, the experimental data that support such a role of ILC2 in GVHD development are still missing.
NCR À ILC3 in GVHD NCR À ILC3 and NCR þ ILC3 produce IL-17. In mouse models of GVHD, IL-17 produced by Th17 cells turned out to be sufficient but not necessary to induce GVHD (31).
Notably, IL-17 contributed to the early development of CD4 þ T cell-mediated GVHD by triggering pro-inflammatory cytokines. Elsewhere, adoptive transfer of Th17 cells resulted in lethal GVHD, which was dependent on TNF-a and IL-17 production. In contrast, selective depletion of Th17 cells by targeting the Th17-specific RORgt transcription factor did not affect the GVHD development in a mouse model. There is no direct evidence for a role of NCR À ILC3derived IL-17, but based on the reports about Th17 cells, IL-17 might not be crucially involved in the GHVD pathology.

Conclusions and Perspectives
Despite advances in our understanding and management of allogeneic hematopoietic cell transplantation immunology field, GVHD remains the major challenge and limitation of allogeneic HCT therapy. Besides mainly focusing on the donor immune system, the role of host defense responses is becoming increasingly recognized. Recent reports described the central importance of recipient sensitivity to developing acute intestinal GVHD that is dictated by the homeostatic status of the gut epithelial tissue. Conditioning chemo-and radiotherapy causes damage to the intestinal epithelial crypts and the epithelial progenitor cells, which are responsible for regenerating the entire gut mucosal lining. Group 3 ILC have recently been identified as sentinels of the intestinal epithelial crypts via production of IL-22, a cytokine that is crucial in maintaining epithelial cell integrity. It has turned out that the appearance of activated human ILC3 in the circulation positively correlates with enhanced protection against GVHD development after HCT. There is a strong reason to believe that the IL-22 producing ILC3 are able to diminish the risk for GVHD by triggering the regeneration of the intestinal mucosa. Future

ILC in GVHD
research should dissect the exact contribution of ILC3 and possibly also other ILC subsets in GVHD immunology. Additionally, larger clinical studies are required for confirming the critical importance of ILC in regulating GVHD susceptibility.
Outstanding Questions: (1) What regulates the appearance of NCR + ILC3 in the blood circulation after conditioning chemotherapy? (2) Besides IL-22 which other NCR + ILC3-derived factors might be important in the prevention against GHVD after allo-HCT? (3) Are ILC1 and ILC2 involved in the development of GVHD?