Keywords

11.1 NK Cells Are a Group of Innate Lymphocytes that Secrete Adaptive Immune Cytokines

The innate immune system is constituted with granulocytes, monocytes, macrophages, and dendritic cells that secrete inflammatory cytokines, as well as innate lymphocytes that secrete adaptive cytokines such as IFN-γ, interleukin (IL)-4, and IL-17. NK cells are the prototypic ILCs, and they were first described in 1975 as being able to naturally kill mouse leukemia cells [1]. Since 2008, the concept of ILCs [2] has been expanded and now includes the related subsets of NK, ILC1, ILC2, ILC3, and the lymphoid tissue inducer (LTi) cells [3]. ILCs are characterized as having lymphoid morphology but lack rearranged antigen-specific receptors and myeloid and dendritic cell phenotypical markers. ILCs develop initially from progenitors in the fetal liver [4, 5] and, later, in the adult bone marrow [6,7,8]. They subsequently seed mucosal tissues, where they continue to proliferate and become tissue-resident cells and maintain tissue homeostasis. ILCs and T cells share similar transcription factors that govern their differentiation and produce similar key cytokines [2, 9]. Thus, in analogy to T cells, ILCs are subdivided into cytotoxic (NK) and all other “helper”-like subsets that resembles IFN-γ/Th1-, interleukin 4 (IL-4)/Th2-, and IL-17/Th17-secreting CD4+ T helper cells [10].

11.2 ILCs Are Generated from Progenitors Downstream of the Common Lymphoid Progenitor

All ILCs initially derive from the common lymphoid progenitor (CLP). The transition from CLP to ILC-specific transcriptional program is accompanied with differential expression of over 400 genes [4, 5, 11], with temporal requirements for Nfil3 (nuclear factor interleukin 3, also known as E4bp4), TCF-1(T-cell factor 1, encoded by Tcf7), and ID2 (inhibitor of DNA binding 2). Nfil3 expression is essential for the development of ILC progenitors prior to their commitment, and it is induced by mesenchymal-derived IL-7 [12,13,14]. NFIL3 also directly activates ID2 [14, 15]. TCF-1 represses genes critical for stem cell (Hhex and Lmo2) and pro-B cell (Spib, Irf8, Ly6d) function [11], and its loss affects the differentiation of both NK and other ILC subsets [16,17,18]. ID2 induces a major regulatory shift with broad repression of progenitor cell transcription factor genes and upregulation of critical regulators including Tox (thymocyte selection-associated high-mobility group box) and Gata3 (GATA-binding protein 3) [11]. Thus, immedicably downstream of the CLP, the earliest ILC progenitor (EILP) is TCF-1+ [17], which further becomes ID2hi common helper ILC precursor (CHILP) when NK cell potential is lost [6, 14, 19, 20]. After acquisition of promyelocytic leukemia zinc finger protein (PLZF, encoded by Zbtb16), ILC progenitor (ILCP) loses the capacity to differentiate into LTi cells [5, 6]. Programmed cell death-1 (PD-1) is co-expressed with PLZF and can be used as a cell surface marker to identify ILCP [11] (Fig. 11.1 ).

Fig. 11.1
figure 1

NK and helper ILCs development in mice

T, B, NK and all other helper ILCs develop from common lymphoid progenitor (CLP). NK and ILCs development accompany with sequential differential acquirement of hundreds of transcription factors: Nfil3 and Tcf1 are required for the development and commitment of early ILC progenitors (EILP). Expression of ID2 leads to the commitment of common helper ILC precursor (CHILP), which is not able to further develop into NK cells. When PLZF is expressed, ILC progenitor (ILCP) is formed and its LTi potential is lost. Downstream of EILP, pre-NK progenitor (pre-NKP) develops into re-defined NK progenitor (rNKP) that expresses CD122, which couples with CD132 to form the IL-2/IL-15 receptor, allowing NK cells to respond to IL-15. rNKP then develops through an immature stage (iNK) to become mature NK cells (mNK). Nfil3, Tcf1, Ets1, Id2, Eomes, T-bet and Zeb2 governs NK cell development from EILP to mNK cells

Full size image

11.3 NK Cells Develop Through Immature and Mature Stages

In the adult mouse bone marrow, pre-NK cell progenitor (pre-NKP) downstream of CLP (Lin−Flt3+ CD27+CD244+ CD127+CD122−Ly6D−) has a Lin−Flt3−CD27+CD244+CD127+CD122− surface phenotype, which further develop into rNKP (recently re-defined NK progenitor) that expresses CD122 [21, 22]. CD122 couples with the common γ-chain (CD132) and forms the IL-2/IL-15 receptor, allowing NK cells to respond to IL-15 and activate JAK1/3 and STAT5 [23,24,25]. IL-15 also activates 3′-phosphoinositide-dependent kinase 1 (PDK1)-mTOR and regulates Nfil3 and CD122 expression [26]. rNKP develops through an immature NK cell (iNK) stage to become mature NK (mNK) cells. iNK expresses NK1.1 but does not express CD49b (antigen to DX5). The expression of Ly49 receptors on the developing iNK cells is critical for NK cell education and maturation and for the detection of invading pathogens, such as murine cytomegalovirus (MCMV) [27, 28]. The most iNK-cell-proximal mNK cells are CD27+CD11b−, produce IFN-γ and TNF-α when activated, but are not yet fully cytotoxic effector cells. Cytotoxic capacity improves with NK cell maturation by type I interferons (IFN-α or IFN-β) or proinflammatory cytokines IL-2, IL-15, IL-12, and IL-18, which upregulate CD11b through T-bet and zinc finger E-box-binding homeobox 2 (Zeb2) [29, 30]. Of note, iNK cells in the bone marrow differentiate through four stages sequentially as CD27−CD11b−, CD27+CD11b−, CD27+CD11b+, and CD27−CD11b+ [31, 32]. Besides CD11b and Dx5, mature NK cells also highly express KLRG1, CD62L, and CD43 [32].

Apart from Tcf1 and Nfil3 [8, 13,14,15, 17, 33] required for EILP commitment, Ets1, Id2, Eomes, and T-bet are transcription factors essential for NK cell development. Ets1, required for early NK cell lineage commitment, induces Id2, Tbx21, and Il2rb (CD122) expression [34,35,36]. Id2 suppresses E protein target genes (e.g., Socs3, Tcf7, Cxcr5), and the suppression of Socs3 promotes NK cell response to IL-15 [37, 38]. IL-15 is crucial for NK cell survival through the induction of the anti-apoptotic protein Bcl-2 [39, 40]. Eomesodermin (Eomes) and T-bet are members of the T-box family of transcription factors and are required by iNK and mNK cells [41]. However, tissue-resident NK cells may exhibit different developmental reliance on T-bet and Eomes [42].

NK cell maturation and function are regulated by an additional group of transcription factors. These include the Ets family protein myeloid elf-1-like factor (Mef, also known as ELF4) [43] and PU.1 (encoded by Spi1) [44], which respectively regulate perforin expression and NK cell proliferation in response to IL-2 and IL-12. PR domain zinc finger protein 1 (Blimp1, encoded by Prdm1), induced by IL-15 in a T-bet-dependent manner during early NK cell development, promotes granzyme B expression but inhibits NK cell maturation and proliferation to low concentrations of IL-15 [45]. Tox regulates mNK development partially through the induction of Id2 [46]. The Ikaros family member Aiolos (encoded by Ikzf3) promotes IFN-γ expression; however, its absence enhances the ability of NK cells to control tumor cells [47]. Kruppel-like factor 2 (Klf2) restricts iNK cell proliferation but is required for migration of NK cells toward IL-15-rich microenvironment [48]. IFN regulatory factor 2 (Irf2) is required for NK cell maturation in the periphery and survival in bone marrow. At homeostatic state, Gata3 is required for bone marrow NK cell maturation from CD27+CD11b− stage and for bone marrow egress, liver migration, and IFN-γ expression. In the face of infection, Gata3-deficient NK cells demonstrated inferior control of Listeria monocytogenes burden in the liver [49]. However, Gata3-deficient NK cells exhibited superior activity toward MCMV due to increased CD25 expression [50]. Discrepancies regarding forkhead box protein O1 (Foxo1) exist in the literature. In one report, Foxo1 was shown to be required for iNK cell survival by inducing autophagy that removes damaged mitochondria and intracellular reactive oxygen species (ROS) [51]. In another report, however, Foxo1 inhibited late-stage NK cell maturation and function by downregulating Tbx21 expression [52].

11.4 Tissue-Resident NK Cells Acquire Unique Phenotype and May Have Distinct Developmental Pathways

Tissue-resident NK (trNK) cells often express CD69, CD103 (αE integrin), and CD49a (α1 integrin), which are involved in retaining NK cells in the tissues. CD69 inhibits type I interferon-induced expression of sphingosine-1-phosphate receptor 1 (S1P1). S1P1 and S1P5 on NK cells binds to sphingosine-1-phosphate (S1P), which forms a gradient with the highest concentration in peripheral blood and, thereby, promotes egress of lymphocytes from tissues into the blood [53, 54]. CD103 forms a heterodimer with β7 integrin and binds to E-cadherin on epithelial cells [55]. CD49a forms a heterodimer with β1 integrin and binds to collagen [56]. The expression of both CD103 and CD49a is regulated by transforming growth factor-β (TGF-β1) [57]. Development of trNK cells may be different from conventional blood NK cells. CD49a+ DX5− Trail+ trNK cells in the mouse liver express higher amount of TNF-α and GM-CSF than blood and spleen conventional NK cells, and they develop in a T-bet-dependent manner in the absence of Nfil3 [41, 42]. CD49a+DX5− NK cells that resemble liver trNK cells are also observed in the mouse uterus and skin [42]. In contrast, salivary glands [58] and uterine NK cells [59,60,61] develop require Eomes in the absence of Nfil3. In addition, a population of CD127+ NK cells develop in Gata3- and IL-7-dependent manner independently from T-cell precursors in the mouse thymus, and thymic trNK cells demonstrate reduced granzyme B but increased IFN-γ, GM-CSF, and TNF expression [62, 63].

11.5 NK Cell Diversity and Activity Are Regulated by Variegated Surface Receptors

The activities of NK cells are regulated by various germ line-encoded activating or inhibitory receptors (Table 11.1 ), many of which are expressed in stochastic patterns, resulting in many subsets of functionally distinct NK cells [64,65,66]. The families of NK receptors that recognize MHC class I include the murine Ly49 receptors, the primate killer cell immunoglobulin-like receptors (KIRs), and the CD94-NKG2 receptors in both rodents and primates [65]. Inhibitory receptors in humans and rodents normally contain one or more intracellular immunoreceptor tyrosine-based inhibitory motifs (ITIM) that can activate downstream SHP-1, SHP-2, and SHIP phosphatase [67, 68]. Many of the activating receptors lack intracellular signaling motifs and transduce signals via the association with immunoreceptor tyrosine-based activating motif (ITAM)-containing adapters DAP12, FcεRγ, and CD3ζ, which recruit and activate Syk or ZAP70 tyrosine kinases [69]. NKG2D ligands are self-proteins related to MHC class I molecules. They are generally absent on the cell surface of healthy cells but are frequently upregulated upon cellular stress [70]. NKG2D recruits DAP10 and mediates signaling through the activation of PI3K [71, 72] and ERK [73]. Human KIRs contain either two (KIR2D) or three (KIR3D) extracellular immunoglobulin (Ig)-like domains. They are designated as KIR2DL or KIR3DL, respectively, if they possess a long cytoplasmic domain containing ITIM motif. KIR2DS and KIR3DS have short cytoplasmic domains lacking ITIM but associate through a charged residue in their transmembrane regions with DAP12 or FceRIγ, respectively. KIR2D receptors typically recognize human HLA-C alleles, whereas KIR3D receptors recognize HLA-B or some HLA-A alleles [74, 75]. The NKG2 family contains one inhibitory NKG2A and two activating members NKG2C and NKG2E. The CD94-NKG2 receptors recognize nonclassical MHC-I that is HLA-E in humans and its ortholog Qa-1 in mice [76,77,78]. A subset of human NK cells express KIR-related inhibitory receptor, LILRB1, which recognizes a shared epitope in all human MHC class I proteins [79].

Table 11.1 The activating and inhibiting NK receptors on mouse and human NK cells
Full size table

NK cells also express activating and inhibitory receptors that recognize non-MHC ligands [80]. For example, murine CD244 (2B4) recognizes CD48, an interaction essential for the IL-2-driven expansion and activation of NK cells [81]; human NKR-P1A (CD161) recognizes the lectin-like transcript-1 (LLT1, encoded by Clec2d), which is expressed on activated dendritic cells and B cells and inhibits NK cell cytotoxicity and IFN-γ expression [82, 83]; killer cell lectin-like receptor G1 (KLRG1) recognizes cadherins and mediates ‘missing-self’ education [84]; Gp49B1 recognizes αvβ3 integrin and inhibits IFN-γ expression [85, 86]. The activating DNAX accessory molecule-1 (DNAM-1, also known as CD226) [87,88,89] and the inhibiting T-cell immunoreceptor with Ig and ITIM domains (TIGIT) [90, 91] receptors both recognize poliovirus receptor (PVR, also known as CD155) and poliovirus receptor-related 2 (PVRL2, also known as nectin-2 and CD112), which are frequently expressed on transformed or stressed cells.

During NK cell development, the expression of self-MHC class I-reactive inhibitory receptors ‘licenses’ NK cells. Under physiological conditions, licensed NK cells engage through the Ly49 and KIR inhibitory receptors with MHC class I and prevent NK cells from attacking self, and this self-tolerance is mediated through the recruitment of SHP-1, SHP-2, and SHIP phosphatase [67, 68]. Interestingly, licensed NK cells are more potent in their cytotoxicity toward MHC class I-deficient target cells and secrete more IFN-γ and TNF-a under noninflammatory conditions [92, 93]. During infection, however, inhibitory receptor engagement impairs the ability of licensed NK cells to control cytomegalovirus (CMV) infection [93]. The absence of inhibitory receptors on NK cells can have a beneficiary effect in human leukemia patients receiving irradiation therapy followed by bone marrow transplantation. The absence or mismatch of donor NK inhibitory KIR receptors with recipient MHC-I was associated with better leukemic cell clearance and graft acceptance [94].

Activating receptors have the ability to recognize ‘altered-self’, which is often induced on malignant or stressed cells [95], and trigger NK cells to kill their targets. NK cells mediate target-cell killing by a number of mechanisms, including (1) the secretion of cytokines, (2) exocytosis of cytoplasmic granules containing perforin and enzyme, (3) FAS ligand and TNF-related apoptosis ligand (TRAIL)-mediated induction of apoptosis, and (4) CD16 cross-linking and antibody-dependent cell-mediated cytotoxicity (ADCC) [94]. However, when NK cells are chronically exposed to endogenous, as well as foreign ligands recognized by their activating receptors, they are tolerated through either receptor downregulation or hyporesponsiveness [65]. NK cell tolerance mediated by activating receptors is reversible and can be broken in the presence of inflammatory cytokines or infection. For instance, in C57BL/6 mice receiving MCMV infection, initially both licensed and unlicensed NK cells expressed CD69 and upregulated IFN-γ and granzyme B at similar level, but, subsequently unlicensed NK cells predominated in numbers and were the main mediators of viral clearance. The engagement of the activating Ly49H receptor with MCMV-encoded glycoprotein m157 on infected cells promoted unlicensed NK cells to undergo a proliferative burst, but the inhibitory receptors on licensed NK cells restrained the proliferation through SHP-1 phosphatase signaling [68, 93] (Fig. 11.2 ).

Fig. 11.2
figure 2

NK cell license, activation and inhibition

(a) NK cell license occurs with the expression of self-MHC class I-reactive inhibitory receptors, Ly49 in mice and KIR in man. This prevents NK cells from attacking self. In the absence of inhibitory receptors, chronic exposure of activating receptors with their ligands can also render NK cell hyporesponsive. (b) NK cell activation takes place under instances of human leukemia patients receiving irradiation therapy followed by bone marrow transplantation. The absence or mismatch of donor inhibitory NK receptors with recipient MHC-class I promotes leukemic cell clearance by both licensed and unlicensed NK cells. (c) During viral infection, inhibitory receptors on licensed NK cells inhibit their proliferation burst, and under these circumstances, unlicensed NK cells are the main mediators of viral clearance

Full size image

11.6 NK Cells Participate in Tissue Remodeling in Humans and Undergo Clonal-Like Expansion During Viral Infection

A mouse analog of human NK progenitor has been defined as Lin−CD34+CD38+CD123−CD45RA+CD7+CD10+CD127−, which selectively gives rise to NK cells in vitro and in vivo [96]. Circulating human NK cells are a diverse population. In any given individual, the diversity is generated by the developmentally distinct NK cell subsets, KIR gene content, polymorphisms, and copy number variations [64], with differentiation and reprogramming in response to tissue-specific environment and infections [97]. Transcriptional, telomere length, and transfer of human NK cells into NOD/SCID/γc−/− mice have demonstrated that circulating NK cells in human blood display sequential CD56bright CD62L+, CD56dimCD62L+ CD94high , and CD56dimCD62L− CD94low developing stages [98, 99]. CD56bright CD62L+ NK cells are mostly KIR− NKG2A+CD27dim CD57−CD16+/− but express CD127 and CD117 (also known as KIT and SCFR), which are also hallmarks of non-NK ILCs [2, 100]. Upon stimulation with combinations of IL-12, IL-15, and IL-18, CD56bright CD62L+ and CD56dimCD62L+ NK cells strongly proliferate and produce significantly greater amount of IFN-γ than CD56dimCD62L− NK cells. However, engagement of the activating receptors evokes more prominent chemokine (MIP-1α, MIP-1β and RANTES) and cytokine (IFN-γ) expression and NK cell cytotoxicity in CD56dimCD62L+ and CD56dimCD62L− cells. [98, 101, 102]. CD56dim NK cells can further develop with the sequential loss of NKG2A and the acquisition of KIRs and CD57 [103]. CD56dimCD57+ NK cells have increased cytotoxic capacity than CD56dimCD57− NK when they are activated through CD16 [104].

In parallel to mice, human tissue-resident NK cells also express CD69, CD103, and CD49a, and they may derive directly from progenitors that reside within the tissues [97]. NK cells are found at high frequencies in the endometrium of human uterus and decidua in the first trimester of pregnancy. Throughout the second half of the menstrual cycle, progesterone from the ovaries acts on uterine stromal cells, which in turn secrete IL-15 and support uterine NK cell proliferation [105]. During pregnancy, a key role for CD56bright uNK cells is to promote trophoblast invasion and maternal spiral artery remodeling, which is mediated through the production of IL-8, interferon-inducible protein-10 (IP10), and an array of angiogenic factors including vascular endothelial growth factor A (VEGF-A), VEGF-C, and angiopoietins [99, 106]. Critically, fetal trophoblasts, which come into direct contact with maternal blood and tissues during pregnancy, are exempt from uNK-mediated cell killing. Uterine CD56bright CD49a+CD103+CD9+ NK cells express perforin, granzymes A and B, and the activating receptors NKp30, NKp44, NKp46, NKG2D but are unable to form mature activating synapses and thus are not cytotoxic [107, 108]. Furthermore, the high expression of inhibitory KIRs (KIR2DL1, KIR2DL2, KIR2DL3), the CD94-NKG2A receptor complex, and the LILRB1 inhibit NK cell activation through the recognition of HLA-C, HLA-E, and HLA-G expressed on the extravillous trophoblasts, respectively [107, 109]. Interestingly, primary villous trophoblasts do not express HLA, and extravillous trophoblasts are devoid of HLA-A and HLA-B.

In liver sinusoids, NK cells represent up to 30–40% of all hepatic lymphocytes [110], and CD56bright and CD56dim cells are present in equal proportions [111, 112]. Hepatic CD56bright NK cells express CD69 and are tissue resident [113, 114]. Liver resident macrophages (Kupffer cells) interact with NK cells to keep immune tolerance to nonpathogenic antigens from food and LPS from gut commensal bacteria, but remain alert to infections by pathogens and viruses. In recognition of bacterial cell wall products via TLR2/4 -MyD88, Kupffer cells secrete IL-10 and blunt NK cell activation [115]. However, when DNA or RNA viruses activate the TLR3-TRIF-IRF-3 [115] or TLR8 pathways [116], Kupffer cells elicit potent IFN-γ and TNF expression in CD56bright trNK cells. Intrahepatic NK cells also mediate target-cell killing through the expression of TRAIL, whose expression is correlated with the control of hepatitis C virus (HCV) infection [117]. But during HBV infection, TRAIL also causes liver damage and can eliminate antigen-specific T cells [118, 119].

Clonal-like expansion and memory formation of NK cells have been observed in humans with cytomegalovirus (HCMV) [120,121,122,123], chikungunya virus (CHIKV) [124] and hantavirus [125] infections. Clonal-expanded cells are characterized by the expression of NKG2C, CD57, and activating KIRs (KIR2DS4, KIR2DS2, KIR3DS1), a general lack of the expression of inhibitory NKG2A and KIR3DL1 receptors (in individuals expressing its HLA-Bw4 ligand), and the decreased expression of CD161 (also known as KLRB1), NKp30, NKp46, and CD7. A subset of clonal-expanded NK cells can further acquire adaptive phenotypes that resembles more with cytotoxic CD8+ T lymphocytes than conventional NK cells. The intronic region of ZBTB16 in adaptive NK cells is hypermethylated, which is correlated with the decreased expression of PLZF and its target genes encoding FcεRγ, SYK, and EAT-2. Adaptive PLZF-deficient NK cells are distinct from clonal-expanded NK cells expressing CD57, NKG2C and PLZF, and produce less IFN-γ upon cytokine stimulation with IL-12 and IL-18 [126].

11.7 Conclusion

NK cells are a heterogeneous population of innate lymphocytes that develop from the common lymphoid progenitors. Tissue-resident NK cells may have different developmental origins and are phenotypically distinct from their blood counterparts. NK cells employ both inhibiting and activating receptors for ‘missing-self’ education, activation, and terminal differentiation. In humans, NK cells are critical for the implantation of the embryos and for the control of neoplastic transformation and viral infections, but they may also induce collateral damages to the tissues. Despite lacking rearranged antigen-specific receptors, NK cells can acquire adaptive T-cell features by clonal-like expansion and alteration in their DNA methylation profiles during viral infections.