Inhibitory KIRs are highly polymorphic at the genetic level, heterogeneously expressed by NK cells. Inhibitory KIRs possess two (2D) or three (3D) extracellular immunoglobulin domains, a transmembrane domain and a long (L) cytoplasmic tail containing two immunoreceptor tyrosine-based inhibitory motifs (ITIMs) [4]. Each inhibitory KIR has for ligands a subgroup of HLA class I allotypes. Inhibitory KIRs with 2 extracellular domains recognize HLA-C molecules based on two dimorphisms at position 77 and 80 of the α1 domain of the α chain [4]. KIR2DL1 associates with HLA C2 allotypes [4]. In contrast, KIR2DL2 and three show specificity for HLA C1 allotypes [4]. KIR3DL1 recognizes HLA-A and HLA-B molecules that display the Bw4 motif [5]. Finally, KIR3DL2 solely interacts with HLA-A3 and A11 molecules [4].

The complex interplay between inhibitory KIRs and HLA class I molecules is of paramount importance in the education and activation of NK cells. As the genes for inhibitory KIRs and HLA class I molecules are on different chromosomes and that NK cells express germline-encoded receptors that do not undergo somatic rearrangement during development, NK cells need to undergo an educational process where they learn to distinguish “self” from “non-self” through interactions between inhibitory KIRs and self HLA class I molecules [6]. NK cells that encounter self-HLA class I molecules during their maturation process receive inhibitory signals, ensuring that they do not attack healthy cells displaying self HLA class I molecules (Figure 2). In absence of any interaction between inhibitory KIRs and HLA class I molecules during their development, NK cells become hypo responsive (Figure 2). This process helps to refine NK cell self-tolerance and future reactivity as the number of interactions between inhibitory KIRs on NK cells and HLA class I molecules determines the degree of responsiveness of mature NK cells. Conversely, when these educated NK cells encounter virally infected or transformed cells that lack or display altered HLA class I molecules, the lack of inhibitory signals transmitted by inhibitory KIRs allows for the activation of NK cells by the recognition of missing self (Figure 2).

For a considerable period, the implication of the “missing self” in organ transplantation was overlooked. New advances were made when Colvin and his team observed that when a heart transplant from a parental strain was transplanted to an F1 hybrid strain (between the parental line and a second strain), the graft presented chronic vascular rejection [7]. It is notable that this model demonstrates that NK cells can initiate chronic vascular rejection, but that they are insufficient to trigger chronic vascular lesions when acting alone. For this to occur, they must recruit T cells that are not specific for donor major histocompatibility complex (MHC) and macrophages in particular, through the secretion of interferon gamma (IFN-γ) [7, 8]. In humans, many studies have looked at this issue for a long time without results, probably because of a lack of data to analyze the presence of missing-self correctly [9–13], or because the chosen read out was not the good one [10], or because of confounding factors [14, 15]. A decade ago, in a cohort of 137 kidney transplant patients compatible for HLA-A, HLA-B and HLA-DR (i.e., in whom minimal T- and B cell alloreactivity was expected), Van Bergen et al. observed that missing self was associated with a worse allograft survival [14]. This was one of the first clinical studies to suggest the involvement of missing self after kidney transplantation, but to go deeper into the demonstration, we have to wait for a study published by our group a few years ago [15]. In a cohort of deeply phenotyped kidney transplant patients, we demonstrated that a greater proportion of patients with microvascular inflammation lesions in the absence of DSA (either HLA or non-HLA DSA) exhibited a missing self on their graft susceptible to being sensed by the recipient’s NK cells and this resulted in reduced allograft survival compared to a control cohort (patients with no vascular lesions and no DSA) [15]. This clinical correlation was confirmed in an in vitro model in which we demonstrated that the lack of self HLA class I molecules on endothelial cells can activate NK cells, which in turn induce endothelial cell damage (Figure 2) [15]. An in vivo mouse model of heart transplantation was then developed. Heart transplants were performed from β2 micro KO mice (i.e., mice that do not express MHC class I molecules) to mice of the same genetic background [15]. The grafts developed microvascular inflammation lesions only in the presence of NK cells and a missing self, thereby confirming the existence of missing-self induced NK-mediated rejection [15]. Our results were confirmed in a population-based study which showed that missing self increased microvascular inflammation occurrence after kidney transplantation, independently of DSA [16]. Microvascular inflammation severity increased with the number of missing self [16].

It is widely accepted that NK cells play a pivotal role in the development of chronic vascular rejection in the presence of DSA. Binding of circulating DSA to directly accessible donor HLA molecules expressed by graft endothelial cells can sometimes trigger the classical complement pathway, which accelerates the rejection process [17, 18]. However, this is not a mandatory requirement for the development of chronic AMR [19]. It has been demonstrated that the recruitment of innate immune cells by DSA is sufficient to trigger endothelial cell damage through an ADCC mechanism (Figure 2). In 2012, Colvin and his team have shown that when RAG^−/−C3^−/− mice were transplanted with an allogeneic heart and injected regularly with DSA, they developed chronic vascular lesions in their graft and that these lesions were completely abrogated in the absence of NK cells [20]. In the clinical setting, the key role of NK cells in AMR is supported by transcriptomic analyses of renal graft biopsies [21–23]. Recent data suggesting that afucosylation of HLA-specific IgG1 is directly related to antibody pathogenicity in kidney transplantation also provides indirect evidence for the key role of NK cells in AMR, as fucosylation is known to strongly influence the affinity of IgG for FcγRIIIA, which is expressed by NK cells [24, 25]. Furthermore, the intensity of NK cell infiltration within the graft correlates with graft survival after kidney transplantation [26].

NK cell activation during AMR is thought to be triggered by the interaction between NK’s unique Fcγ receptor, FcγRIIIA and DSA (Figure 2). In the murine heart transplantation model of AMR described above, mice injected with F (ab′)₂ DSA fragments failed to develop chronic vascular lesions in comparison to mice injected with intact DSA, suggesting that the interaction of NK cells with the Fc fragment of DSA is necessary [20]. Transcriptomic analyses conducted on sets of renal allograft biopsies from patients with AMR identified NK cell signaling, including evidence for the FcγRIIIA signaling elements [27, 28] suggestive of an NK cell activation through FcγRIIIA in AMR. In a recent study, thanks to single cell RNA sequencing analyses and multiplexed immunofluorescence, Lamarthée and colleagues have shown an association between FcγRIIIA+ NK cells and the severity of intragraft inflammation in the context of AMR [22]. Like in missing-self induced NK-mediated rejections, NK cells do not act alone and seem to interact in particular with FcγRIIIA+ non-classical monocytes via LGALS9-HAVCR2 to trigger allograft destruction [22]. They also interact to a lesser extent with T cells expressing CD74CXCR4 via the secretion of macrophage migration inhibitory factor (MIF) [22].

A single nucleotide polymorphism exists in the FcGR3A gene, resulting in the expression of two co-dominant alleles coding for either a phenylalanine (F) or a valine (V) at amino acid position 158 in the extracellular domain of the receptor [29]. This single nucleotide polymorphism modulates the binding capacity of FcγRIIIA to the Fc fragment of IgG, which may in turn modulate the severity of lesions and the prognosis of patients with AMR. In kidney transplantation, Arnold et al. observed that patients with DSA and presenting at least one high binding allele (V/V or V/F) exhibited more frequent and severe peritubular capillaritis lesions [30]. Furthermore, the researchers hypothesize that this heightened microvascular inflammation may be due to the secretion of IFN-γ, which facilitates the recruitment of other immune cells in the graft [30]. These findings were corroborated in a cohort of kidney transplant recipients with chronic humoral rejection [31]. The patients with two V/V alleles exhibited a higher glomerulitis score and a reduced graft survival compared to patients of other genotypes [31]. The authors propose that the V-allele is associated with an increased expression of FcγRIIIA by NK cells, which translates into a greater propensity for NK cells to degranulate when they interact with their targets [31].

In addition, a recent study has investigated the impact of different genetic variations present in the genes of different NK cells in a cohort of patients with de novo DSA [32]. In particular, they got interested in a polymorphism in KLRC2 gene which encodes NKG2C, an activating NK cell receptor, which binds to HLA-E and is known to be present on memory like NK cells with more potent effector functions [29]. Heterozygous or homozygous KLRC2 deletion (KLRC2 ^del) is associated with a significantly lower or absent NKG2C expression level. Here, they showed that patients with a KLRC2 ^wt/wt genotype presented more microvascular lesions than KLRC2 ^wt/del patients but failed to show any impact on allograft survival [32].

Finally, our group has recently shown that missing self and DSA can combine to induce chronic vascular rejection in a cohort of kidney transplant patients with AMR (Figure 2) [33]. Notably, the additive effect of missing self on allograft survival was only observed in patients with ADCC-dependent AMR [33]. This may be explained by the fact that allograft loss is too rapid in patients with complement-dependent AMR.

Whether the NK cell initiates rejection in missing self-induced rejection or is recruited by the adaptive immune system in AMR, it seems that a final common pathway involving NK cell is triggered during chronic vascular rejection. This was suggested in particular by a study which showed that the transcriptomic signatures were similar in biopsies of patients with DSA-positive and DSA-negative microvascular inflammation lesions [34]. These patients showed similar upregulation of pathways such as IFNγ-induced pathways and NK cell activation, and similar enrichment of infiltrating leukocytes [34]. A better understanding of the pathways leading to NK activation may therefore be useful in the development of treatments targeting NK cells in chronic vascular rejection, whether or not DSA are present.

Among the various cytokines that may play a role in NK cell priming (IL-2, IL-12, IL-15, IL-18 and IL-21), IL-15 is of particular importance in regulating NK cell homeostasis and activation [35–37]. IL-15 binds to IL-15Rα expressed on antigen presenting cells and is then presented in trans to the IL-2/IL-15Rβγ heterodimer on NK cells. IL-15 induces the phosphorylation of JAK1/JAK3, allowing the recruitment and activation of the transcription factor STAT5 (Figure 3), which then translocates to the nucleus to support NK cell survival [38]. In parallel, it induces the activation of the PI3K-AKT-mTOR signaling pathway which promotes the acquisition of cytolytic potential by NK cells when they will encounter abnormal cells (Figure 3) [38].

In the context of transplantation, we and others have shown that NK cells need to undergo priming to get activated against the graft [15, 39, 40]. In particular, we have shown that prolonged cold ischemia (suggesting significant ischemia-reperfusion injury) and viral infections such as cytomegalovirus can lead to NK cell priming after transplantation, a prerequisite for NK cell activation and the development of chronic vascular rejection in the presence of a missing self [15]. These two types of events favor the induction of IL-15 and its trans presentation by dendritic cells, thereby promoting NK cell cycle entry, proliferation, survival and cytotoxicity [35, 41–44]. In addition, cytomegalovirus also induces a remodeling of the NK phenotype with an increase in NK cell populations expressing educated inhibitory KIRs and expression of markers such as NKG2C and CD57 [45, 46]. This subpopulation of NK cells is considered to be a memory like population and is known to be better at exhibiting cytotoxicity and secreting cytokines such as IFN-γ [46].

Finally, a recent study also showed that IL-21 may help NK cells to differentiate into polyfunctional type 1 activated cytotoxic effectors with a potentially deleterious ability to infiltrate the kidney allograft during AMR and damage the vascular endothelium [47].

Activating NK cell receptors are associated with different adaptor molecules which contain immunoreceptor tyrosine-based activation motifs (ITAMs) on their cytoplasmic domains. FcγRIIIa receptor which is a low affinity receptor for IgG forms a complex with 2 adaptor molecules FcRγ and CD3ζ. Upon ligation of the Fc fragment of IgG by FcγRIIIa, the ITAMs are phosphorylated. This triggers a cascade of downstream events involving the PI3K pathway, the mitogen-activated protein kinase (MAPK) pathway and calcium release conducting to the activation of nuclear factor of activated T-cells (NFAT) (Figure 1) [48–50]. Together, these pathways contribute to the cellular response leading to NK cell activation [51–53]. Unlike activating receptors, inhibitory receptors and in particular inhibitory KIRs have 2 ITIMs in their cytoplasmic tails. When inhibitory KIRs interact with their self-MHC class I ligands, the ITIMs are phosphorylated and recruit the tyrosine phosphatases such as SHP-1 (Src homology containing tyrosine phosphatase-1), which remove phosphate groups from several proteins downstream of activating receptors, thereby preventing NK cell activation (Figure 1) [54]. In the absence of ligation of their self-MHC class I ligands, inhibitory KIRs do not interfere with activating signals, allowing NK cell activation. We have shown that missing cell induced-NK cell activation triggers the mTOR pathway [15]. Whether this is also the case for FcγRIIIa-induced NK cell activation is not clear, but experimental data investigating NK cell activation by other activating receptors suggest that this is certainly the case [55].

Activation of NK cells can lead to killing of the target cell. NK cells exert their cytotoxic activity through two main pathways: the exocytosis of lytic granules containing perforin and granzymes A and B; and the expression of ligands (FAS ligand, TRAIL) of death receptors (FAS, DR4 and DR5) expressed on target cells. In addition, NK cells can secrete pro-inflammatory cytokines (TNF-α and INF-γ) that favor the activation of dendritic cells, T cells and monocytes [56]. They also secrete chemokines such as CCL-2 (MCP-1), CCL3 (MIP-1α), CCL4 (MIP-1β), CCL5 (RANTES), which attract effector lymphocytes and myeloid cells to inflamed tissues [57, 58]. In a murine heart transplantation model of AMR, Lin and colleagues showed that both IFN-γ and contact-dependent cytotoxicity are necessary for NK cells to induce chronic vascular lesions [59]. The same observations have been made in the hybrid resistance model regarding the need for IFN-γ to help NK cells recruit T cells and monocytes to induce the vascular lesions [7, 8]. Several clinical studies have shown that the transcripts involved in IFN-γ pathway and NK cell-mediated cytotoxicity are upregulated in biopsies of patients with HLA positive or negative microvascular inflammation lesions [28, 34].

Transplant patients classically received a two- or three-drug regimen consisting of a calcineurin inhibitor, an antiproliferative agent, and corticosteroids. The aim of these treatments is to inhibit T-cell activation. Theoretically, calcineurin inhibitors could have an effect on NK cell activation because NFAT is expressed by NK cells [50]. In a first in vitro study, human NK cells were treated with cyclosporine and different cytokines (IL-2 or IL-15) for 7 days. This study confirmed that the NFAT pathway was blocked, especially in CD56^dimCD16+inhKIR+ NK cells, and that this resulted in a decrease in NK cell proliferation but an increase in cytotoxicity against different target cells and an increased ability to secrete IFN-γ after a new stimulation with IL-12 and IL-18 [60]. Two more recent studies confirm the conserved cytotoxicity of NK cells exposed to cyclosporine or tacrolimus and activated by the two mechanisms involved in chronic rejection (MS and ADCC) [61, 62]. In vivo, we confirm that cyclosporine does not prevent missing-self-induced NK-mediated rejection in our murine heart transplant model, suggesting that this immunosuppressant may have a limited effect in preventing chronic vascular rejection in patients [15]. The effect of antiproliferative drugs such as azathioprine and mycophenolate mofetil and low-dose corticosteroids on NK cells has been less studied. Some in vitro studies suggest that these drugs may prevent NK cell activation, but these data should be treated with caution due to the lack of in vivo data [61, 62].

Alternatively, transplant recipients may receive mTOR inhibitors (mammalian target of rapamycin) as part of their maintenance regimen. As previously depicted, mTOR pathway plays an important role in NK cell education, priming and activation [38, 55]. In vitro studies have shown that mTOR inhibitors can prevent NK cell cytotoxicity [55, 61]. In vivo, we confirm that rapamycin can efficiently prevent missing-self induced NK-mediated rejection in our murine heart transplantation model [33]. Finally, a recent clinical study in a cohort of lung transplant recipients treated with rapamycin showed that their NK cells had reduced mTOR activity, which was associated with decreased cell proliferation and lower levels of FcRγ, the adaptor molecule of FcγRIIIa, suggesting that they may be less efficient at performing ADCC [63]. A pilot clinical trial (NCT03955172) is currently underway to assess the efficacy of mTOR inhibitors in preventing the formation of chronic lesions in kidney transplant patients presenting a missing-self induced NK-mediated rejection. Although promising, the universality of mTOR inhibitors could be questioned, as this treatment is associated with many side effects, leading to its discontinuation in 30%–50% of patients, and cannot be used late in the course of chronic rejection, as mTOR inhibitors are poorly tolerated in patients with chronic glomerular lesions, as they prevent the adaptation of podocytes to stress [64].

As the cytokine priming process is necessary for all NK cell mechanisms involved in chronic vascular rejection, blocking it offers a unique opportunity to prevent the deleterious effects of NK cells during chronic vascular rejection. As NK cell priming may be the result of multiple cytokines, rather than trying to block one cytokine, blocking the pathways which result from the binding of cytokines on their receptors which share the common γc chain may be more appropriate. Tofacitinib is an oral JAK inhibitor that selectively inhibit intracellular cytokine signaling mediated by JAK3 and/or JAK1. A recent study in a rat model of mixed cellular and antibody-mediated rejection indicates that tofacitinib effectively reduces the infiltration of T cells and NK cells into the graft, thereby limiting the progression of lesions and improving graft and recipient survival [65]. Its use in renal transplant patients has been tested in a clinical randomized phase 2b trial [66]. Patients received either tofacitinib (low or high dose) or cyclosporine in combination with mycophenolic acid and corticosteroids [66]. Tofacitinib was equivalent to cyclosporine in preventing acute rejection and was associated with an improved renal function and less tubular atrophy and interstitial fibrosis [66]. However, it was also responsible for more cancerous and infectious complications in the high dose tofacitinib regimen [66]. This side effect may be the reflect of the efficient blocking of NK cells which are well known for their role in anti-viral and anti-tumoral immunity.

The current consensus for the treatment of AMR associates rapid depletion of circulating DSA with plasmapheresis with a combination of corticosteroids and high-dose intravenous immunoglobulins (IVIg) [67]. This costly and prolonged therapeutic approach has a reported 3-year graft survival of <50% [67]. By preventing complement activation, these treatments slow down the course of rejection and convert it from acute to chronic rejection in which ADCC is the main mechanism of graft destruction.

Few studies have investigated the effect of high-dose corticosteroids on NK cells. In one in vitro study, human purified NK cells and T cells were incubated with elevated doses of hydrocortisone, comparable to drug levels achieved in patients receiving 1 mg/kg to 1 g of methylprednisolone. In contrast to T cells, NK cells were resistant to steroid-induced apoptosis and their cytolytic capacity against MHC class I deficient tumoral cell was not affected [68]. In the same line, although methylprednisolone seems to affect NK cell activation through some activating receptors such as NKp46, NKG2D or 2B4, it does not seem to prevent FcγRIIIA-mediated NK cell cytotoxicity [69]. High dose IVIg may also prevent NK cell activation by saturating all FcγRIIIa receptors present on NK cells, rendering them blind to DSA. However, their impact on NK cells is still debated [67].

To achieve a sustained therapeutic effect in AMR, attempts have been made to target DSA-producing cells with either anti-CD20 monoclonal depleting antibody (rituximab) or a proteasome inhibitor (bortezomib) without efficacy [70, 71]. Recently, an antibody targeting CD38 (felzartamab), a receptor present on plasma cells and NK cells, has been tested in a randomized phase 2 trial study for the treatment of renal transplant recipients diagnosed with late active or chronic-active antibody-mediated rejection after kidney transplantation [72]. This treatment efficiently reduced rejection lesions. This improvement seems to be due to NK cell depletion rather than plasma cells depletion as DSA MFI barely decreased with the treatment. These data need to be confirmed in a larger cohort of patients. A further clinical trial is currently underway to test an alternative anti-CD38 (daratumumab) in the same indication (NCT05913596). If effective, its use can be extended to missing self-induced NK-mediated rejection.