NK cells originate from common lymphoid progenitors (CLPs) that are derived from hematopoietic stem cells (HSCs) in the bone marrow (BM) (3). Unlike T and B cells, which generate antigen receptors via gene rearrangements, NK cells express a broad range of germline-encoding receptors (4). The development of NK cells is facilitated by cytokine IL-15, which binds to the IL-15 receptor subunit alpha, CD122 (IL-2/IL-15 receptor subunit beta), and the common gamma chain on NK progenitors (NKPs), resulting in the commitment of these cells to the NK cell lineage (5). Although IL-2 promotes NK cell proliferation during ex vivo culture, it is not essential for the development of NK cells in mice (6). NKPs differentiate into immature NK cells (iNKs), which are identified by the expression of NK1.1 marker in mice and the NKp46 activation receptor in both mice and humans (7). In humans, CD56^bright NK cells are abundant in secondary lymphoid tissues and mature into CD56^dim NK cells in PB (8). Usually, CD56^dim NK cells are more cytotoxic and express the CD16 (Fc gamma RIIIA) receptor, while CD56^bright NK cells produce more cytokines and have lower levels of CD16 expression (9). CD56^dim NK cells are generally considered as a more mature form of NK cells (mNK); however, the precise mechanism governing the differentiation of these two subsets is not yet well understood. NK cells can be classified into group 1 innate lymphoid cells (ILCs), as they produce type 1 cytokines, interferon gamma (IFN-γ) and tumor necrosis factor (TNF) (10–13). NK cells also have tissue-specific subsets with distinct functions that potentially modulate immune responses in different contexts (14–16).

NK cell development is regulated by several transcription factors including E4BP4, TBX21, EOMES, GATA3, and ID2 (17). For instance, mice lacking E4bp4 exhibit decreased levels of NK1.1⁺ NKPs and impaired NK cell-mediated cytotoxicity (18). In mice, CD27⁺ NK cells mature into CD27^-CD11b^high NK cells. A deficiency of Aiolos (Ikzf3), a member of the Ikaros family of zinc-finger transcription factors, was found to prevent the maturation of CD27⁺ to CD27^- NK cells, while proliferation of NK1.1⁺CD122⁺ iNKs was enhanced in Aiolos-deficient cells following IL-15 stimulation (19). Likewise, researchers have investigated the gene network involved in NK cell development and the regulation of NK cell function. Detailed markers and the distribution of NK cells during development are shown in Figure 1 .

NK cells eliminate target cells that appear to be missing self or stressed, such as those infected with viruses, cancer, or those undergoing cellular senescence. NK cell killing can be achieved through the activation of receptors such as NKG2D and natural cytotoxic receptors (NCRs) NKp30, NKp44, and NKp46. Ligands for these receptors include major histocompatibility complex class I (MHC-I) chain-related polypeptide A (MICA) and MICB proteins, a family of six cytomegaloviral unique long 16 (UL16)-binding proteins (ULBP 1-6) (20–22), or murine retinoic acid early transcript 1 (RAET1), histocompatibility H60 (H60), and murine UL16-binding protein-like transcript 1 (MULT1), and antigens from pathogens (23–25). DNAX accessary molecule 1 (DNAM1) and signaling lymphocytic activation molecule (SLAM) family receptors, including 2B4, SLAMF7, and NKp80, act as co-receptors to enhance NK cell activity (26). Upon activation, NK cells use an immunoreceptor tyrosine-based activation motif (ITAM) or tyrosine-based signaling motif (YINM) to activate receptors to stimulate protein kinases such as SYK and ZAP70, resulting in the secretion of perforins and granzymes (27). This process leads to apoptosis of the target cells. Furthermore, CD56^dim NK cells in PB can mediate antibody-dependent cellular cytotoxicity (ADCC) via the Fc gamma receptor CD16.

The killer cell immunoglobulin-like receptor (KIR) family and CD94/NKG2A heterodimer are inhibitory receptors on NK cells that recognize MHC Class I molecules (28). Humans have 14 KIR genes specific to each individual (29). KIRs are composed of two or three immunoglobulin-like (Ig-like) domains: KIR2D and KIR3D. KIRs have either activating or inhibitory functions. Generally, KIRs with a short cytoplasmic tail are activated, while those with long cytoplasmic tails are inhibitory. Activating KIRs, such as KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5, and KIR3DS1 associate with the transmembrane adaptor DAP12 to transduce activation signals in NK cells. Inhibitory KIRs, such as KIR2DL1, KIR2DL2, KIR2DL3, KIR2DL5, KIR3DL1, KIR3DL2, and KIR3DL3 have an immunoreceptor tyrosine-based inhibitory motif (ITIM) (30). PB NK cells have a diverse repertoire owing to the random expression of KIR genes in each NK cell, resulting in various NK cell clones with distinct receptor expression patterns. Consequently, a subset of NK cell clones may recognize a unique ligand expressed by a particular tumor cell (31, 32). Inhibitory receptors on NK cells include KLRG1, SIGLEC7, SIGLEC9, leukocyte immunoglobulin-like receptor subfamily B member 1 (LILRB1), CD161, TIGIT, LAG3, TIM3, and PD-1, which recruit protein tyrosine phosphatases, such as SHIP1 (also known as PTPN6) and SHP2 (also known as PTPN11), to eliminate tyrosine phosphorylation (27).

NK cells produce chemokines and express various receptors. Resting CD56^bright NK cells preferentially express chemokine receptors related to BM or lymph node residency, such as CCR2, CCR7, CXCR3, and CXCR4. In contrast, CD56^dim NK cells primarily express CX3CR1, ChemR23 or CXCR1 (33, 34). CXCR1 is a cognate receptor of CXCL8, a senescence-associated secretory phenotype (SASP) factor that can also be released from NK cells (35, 36).

Upon recognition of NK cell activation receptors, the phosphorylation of the adaptor proteins DAP10 and DAP12 initiates downstream signal transduction (37, 38). The activation of phosphoinositide 3-kinase (PI3K) plays a crucial role in NK cells by recruiting AKT (also known as protein kinase B), which promotes cell survival and metabolism (39, 40). The MAPK pathway is also activated, leading to sequential phosphorylation events involving extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38, ultimately resulting in gene transcription and cytokine secretion (41–43). Phospholipase Cγ2 (PLCγ2) generates inositol 1,4,5-triphosphate (IP3) and diacylglycerol (DAG), which trigger calcium ion release from the endoplasmic reticulum and activate protein kinase C (PKC). Increased intracellular calcium levels stimulate the Ca²⁺/calmodulin (CaM)-dependent phosphatase calcineurin, which dephosphorylates nuclear factor of activated T cells (NFAT). Dephosphorylated NFAT translocates into the nucleus, promoting the transcription of genes involved in NK cell cytotoxicity and cytokine production, as well as inducing NK cell development by inducing EOMES and TBX21 upregulation (44).

The mechanistic/mammalian target of rapamycin (mTOR) signaling pathway is upregulated by cytokines or growth factors, and inhibition of mTOR suppresses NK cell activity (45, 46). In the immune suppressive tumor microenvironment, TGF-β1 has been found to inhibit IL-15-promoted NK cell activation via the mTOR pathway (47). Upon activation, NK cells release cytotoxic granules containing perforin and granzymes. Granzymes induce target cell apoptosis and perforin forms pores in the target cell membrane. Downstream signaling pathways of NK cell receptors activate transcription factors, such as nuclear factor-kappa B (NF-κB) (48) and STAT4, promoting IFN-γ gene transcription. IFN-γ is secreted by activated NK cells in response to viral infections or cell transformation. The signaling pathways involved in NK cell activation are summarized in Figure 2 .

Immunosenescence, which typically involves a decrease in the frequency of naïve T and B cells, and an increase in the number of effector/terminally differentiated memory cells, has been examined and discussed in many studies. Although the number of NK cells tends to remain stable or increase with age, their function seems to be impaired (49). Although IL-2 stimulates NK cell responsiveness, cells from older donors are less activated than those from younger donors (50). Notably, NK cell dysfunction has been linked to the development of diseases, particularly an elevated risk of cancer with age (51–53).

The age-related decline in NK cell activity can be attributed to several intrinsic and extrinsic factors. Although changes in perforin levels and KIR diversity have been investigated, other mechanisms may be involved (54, 55). Human NK cells exhibit telomere shortening with age in both the CD56^bright and CD56^dim populations (56). Extrinsically, research has focused on changes in endocrine factors with age, as some appear to be associated with NK cell function (57–59). For example, glucocorticoids, which inhibit NK cell function by reducing perforin and granzyme B levels (60, 61), are upregulated in humans (62). Additionally, the plasma level of TGF-β1 increases as senescent fibroblasts secrete more of it (63).

Recently, the coronavirus disease 2019 (COVID-19) pandemic has highlighted advanced age as a major risk factor for mortality, with 95% of deaths occurring individuals over 50 years of age (64). This suggests that immunosenescence in the elderly population may partially play a role in determining the severity of viral infections (65, 66). Several studies have used clinical samples to investigate these mechanisms. For instance, it has been shown that type I IFN signaling is increased in mild cases of COVID-19, while impaired in severe cases (67). Single-cell RNA sequencing (scRNA seq) revealed a decrease in the expression of the human leukocyte antigen-DR isotype (HLA-DR) on monocytes in patients with moderate or mild COVID-19 compared to healthy controls (68). Notably, reduced numbers and cytolytic activity of NK cells have been associated with severe COVID-19 (69). In a study comparing severely young and old immune cells, it was found that the functional exhaustion of NK cells was more pronounced in older individuals (70).

The level of CD57 on NK cells is a marker commonly used to measure the number of long-lived NK cells in the blood, with higher levels typically observed in older individuals ( Figure 1 ). These populations were also considered less cytotoxic. However, CD57 expression levels can vary depending on infection, chronic diseases, and age. Furthermore, NKG2C⁺CD57⁺ NK cells have been suggested to have strong and rapid antiviral and antitumor effects, indicating an adaptive immune response (71–73). Killer cell lectin-like receptor subfamily G member 1 (KLRG1) is an inhibitory receptor found in terminally differentiated effector lymphocytes, including memory CD8⁺ T and NK cells. It has been shown that an increase of KLRG1⁺CD57⁺ NK cells was induced by COVID-19 infection, particularly in older individuals, indicating NK cell exhaustion (70). Therefore, the significance of CD57⁺ NK cells should be interpreted in the context of disease and their function as long-lived and/or memory NK cell populations.

Hematologic malignancies, including AML, chronic lymphocytic leukemia (CLL), lymphoma, and multiple myeloma (MM), have become targetable diseases by adoptive immune cell therapies (86, 87). These therapies are effective because the administered immune cells can efficiently travel to tumor site in hematopoietic organs (88). In addition, hematopoietic tumor cells often express relatively homogeneous antigens such as CD19 in diffuse large B-cell lymphoma (DLBCL) and B-cell maturation antigen (BCMA) in MM, making it easier to apply CAR or monoclonal antibody (mAb) therapies. Since its approval by the US Food and Drug Administration (FDA) in 2017, six CAR-T cell therapies have been developed. Four of these target CD19, including the first CAR-T cell therapy and the remaining target, BCMA (89, 90). They are used to treat refractory or relapsed B cell ALL, B cell non-Hodgkin’s lymphoma, follicular lymphoma, mantle cell lymphoma, and MM (91). Notably, a phase I/II clinical trial of anti-CD19 CAR-transduced cord blood-derived NK cells showed promising results, with a 73% response rate and long-term survival of infused CAR-NK cells without CRS toxicity (92). CAR-NK cells have the potential to benefit from the intrinsic characteristics of NK cell-activating receptors, which may maintain therapeutic efficacy even after the CAR is lost (93). As a result, CAR-NK cells are being developed for hematological malignancies such as T-cell ALL, and preclinical studies have shown the effectiveness of CAR-NK cells specific for antigens such as CD5 and CD7 (94, 95).

Many countries are becoming increasingly aging societies. All tissues, including those of the immune system, become senescent. Systemic immunosenescence can lead to a decrease in the efficient clearance of harmful cells, resulting in an increase in inflammatory or infectious diseases (155). In this section, we discuss the relationship between age-related diseases and the potential use of NK cell therapies.

Different sources of NK cells have been used to develop NK cell therapies, including peripheral blood mononuclear cells (PBMC) (187), umbilical cord blood (UCB) (92), NK-92 cells (188), and induced pluripotent stem cells (iPSCs) (189). Although NK-92 cells derived from NK lymphoma offer a valuable research tool for exploring the function of CARs and genetically engineered cells, their effectiveness may be limited due to the absence of CD16 and the need for irradiation before being administered to patients. In contrast, primary NK cells derived from UCB or PB contain not only cytotoxic NK cells but also CD34⁺ hematopoietic stem cells, which can be expanded and differentiated into mature NK cells. Both UCB- and PB-derived NK cells exhibit therapeutic efficacy against leukemia following CAR gene transduction. Notably, UCB-derived NK cells demonstrate higher proliferative capacity than PB-derived NK cells (190). Additionally, cytokine-induced memory-like NK cells sustained by IL-12, IL-15, and IL-18 in primary NK cells have shown promising clinical outcomes in patients with myeloid neoplasms (187). However, the isolation of large numbers of primary NK cells and hematopoietic stem cells is challenging, and purifying NK cells to deplete CD3 and CD19 positive cells is difficult during allogeneic NK cell transfer. By contrast, iPSC-derived NK cells are an unlimited source of homogeneous human NK cells. These iPSCs can be genetically modified to express or deplete specific genes, followed by therapeutic NK cell differentiation. However, safety concerns remain owing to the presence of undifferentiated iPSCs and the complexity of the differentiation protocol (185).

CARs enhance immune cell function by combining an intracellular signaling domain with an antigen-specific single-chain variable fragment (scFv) that recognizes a specific antigen. There are four to five generations of CARs with increasing complexity of the intracellular domains, whereas the recognition of antigens is dependent on the ectodomains present in each CAR (191). The second generation, which contains either CD28 or 41BB as a costimulatory domain, is effective and widely used. Second-generation CAR-T cells have been utilized in all FDA-approved CAR-T cell therapies to date. CAR-NK cells are also being developed, with promising results reported in 2020 for anti-CD19 CAR-NK cells for the treatment of certain blood cancers, as previously described (92). At least 30 clinical trials of CAR-NK cells using different sources of NK cells are currently underway worldwide (192). CAR-NK cells are being developed to target different diseases, including hematological malignancies (82–85), solid tumors (78–81), and COVID-19 ( Table 1 ).

Despite the ability of second-generation CARs to stimulate NK cells such as T cells, researchers have been investigating novel intracellular CAR domains designed for NK cell activation. Adaptor proteins such as DAP10 and DAP12, which recruit PI3K in association with activating NK cell receptors ( Figure 2 ), have been investigated as potential replacements for CD3z in CAR-NK cells (193, 194). Furthermore, NKG2D-engineered CAR (NKG2D-DAP10-CD3z)-NK cells, which targets NKG2D ligands in tumor cells, have been tested (195). Notably, iPSC-derived NK cells were used to test various CAR constructs, and the results showed that NKG2D is a transmembrane domain, followed by 2B4 and CD3z in iPSC-NK cells, producing optimized activity (189). Similarly, DNAM1 expression was found to be more effective than that of CD28 as an intracellular domain in anti-GPC3 CAR-NK-92 cells (196). A better understanding of the mechanisms underlying NK cell biology may lead to the development of novel CAR-NK cell therapies.

In addition to CARs, various genes have been investigated to enhance the persistence and function of NK cells in vivo. Expression of the membrane-bound form of IL-15 in human PB-derived NK cells has been shown to improve NK cell survival and in vivo anti-tumor activity (197). Anti-CD19 CAR, IL-15, and inducible caspase-9-based suicide genes were introduced into CB-derived NK cells, resulting in the extended survival of modified NK cells in a mouse model (198). Scalable and uniformly edited NK cells can be generated from iPSCs. Quadruple-gene-modified iPSC-derived NK cells, called FT576, were created by Cichocki et al. (199) for the treatment of MM by incorporating anti-BCMA CAR (200), a non-cleavable version of the CD16a Fc receptor (201, 202), and a membrane-bound IL-15/IL-15R fusion molecule. These modified NK cells can target BCMA-expressing tumor cells and display improved ADCC and in vivo persistence. Additionally, CD38 has been knocked out in NK cells (203), enabling a combination approach with an anti-CD38 mAb for the treatment of relapsed or refractory MM (204).

CRISPR-Cas9 gene editing has been used in many studies to delete negative regulators in NK cells. For example, iPSC-derived NK cells have been edited to remove the cytokine-inducible SH2-containing proteins (CIS), encoded by CISH, resulting in potential IL-15-dependent NK activation (205). Tumor necrosis factor-α (TNF-α)-induced protein-8 like-2 (TIPE2) negatively regulates the immune response of NK cells against tumors. This discovery led to the development of a novel approach involving the use of TIPE2-deficient NK cells, which could potentially enhance the effectiveness of cancer treatment in combination with CISH knockout (206). In animal models, when TIPE2-defecient NK cells are transferred to the host, they activate the IL-15-mTORC1 pathway, leading to improved antitumor activity and providing support for T cell-based immunotherapy (207).

Genetic disruption of NKG2A has been used to modulate inhibitory signaling in NK cells, resulting in superior tumor control in a mouse model (208). In the tumor microenvironment, adenosine A2A receptor is knocked out in CAR-NK cells to counteract the immunosuppressive effects of adenosine (209). Deletion of TGF-β receptor 2 in CB-derived NK cells (210) or downregulation of TGF-β-induced miR-27a-5p has been used to block TGF-β signaling and improve NK cell function (211). In MM, the therapeutic mAb daratumumab targets CD38, which is expressed by both malignant and NK cells, leading to fratricide. To overcome this issue, CRISPR-modified CD38 knockout NK cells have recently shown improved efficacy (212, 213).

Adoptive NK cell therapy has limited efficacy against solid tumors because of the immunosuppressive tumor microenvironment. To improve the targeting of solid tumors, NK cells have been engineered to overexpress chemokine receptors such as CXCR1, CXCR2, CXCR3, CCR7, and CXCR4 in various animal models of cancer (214–218). In addition, the overexpression of a high-affinity dominant-negative TGF-β receptor (TGF-β DNRII) has been used to inhibit TGF-β signaling (219, 220). Clinical trials on patients with prostate cancer have shown the feasibility of expressing TGF-β DNRII and T cells in combination with the antiprostate-specific membrane antigen CAR (221). Another approach involves targeting the inhibitory receptors on NK cells via viral transduction. For instance, human PB NK cells were transduced with a retrovirus to express NKG2A protein expression blocker (PEBL), which abrogates NKG2A expression (208). The PEBL construct contained an anti-NKG2A scFv linked to four different endoplasmic reticulum retention domains, which resulted in enhanced NK cell cytotoxicity and in vivo anti-tumor activity.

There are several ways to enhance the activity of NK cells against tumor cells, including pharmacological interventions or the use of cytokines. Immunomodulatory imide drugs (IMiDs) such as lenalidomide and pomalidomide, which interact with cereblon (CRBN), can activate NK cells via CRBN-dependent or independent mechanisms (246, 247). IMiDs activate ZAP70 and increase granzyme B expression in NK cells (248). In addition, IMiDs make tumor cells more susceptible to NK cell killing by upregulating ligands for NKG2D, DNAM1, MICA, and PVR in tumor cells (249). Bortezomib, a proteasome inhibitor, has been shown to have similar effects on tumor cells, including downregulation of MHC-I expression (250). Other pharmacological interventions that increase the vulnerability of cancer cells to NK cell killing have also been reported. BH3 mimetics enhance NK cell-mediated killing by inducing mitochondrial apoptosis in target tumor cells (251). ONC021, an orally administered antitumor agent, activates TNF-related apoptosis-inducing ligand (TRAIL) and stimulates NK cells at the tumor site, leading to reduced metastasis. Clinical trials for ONC021 are currently underway (NCT02525692) (252).

The direct stimulation of NK cells has also been extensively studied, particularly in the context of the tumor microenvironment, where immune cells are often suppressed. For example, HIF1α expression induced by a hypoxic tumor microenvironment can inhibit NK cell activity, but inhibition of HIF1α or using Hif1α ^-/- NK cells has been shown to potentiate NK cell activity against tumors in in vitro and in vivo experiments (253). Various pharmacological inhibitors against the TGF-β signaling pathway have been tested in preclinical and clinical studies. It has been shown that the TGFβRI kinase inhibitor galunisertib (LY2157299) enhances NK cell-mediated anti-tumor activity (254). On the other hand, ADAM17, a metalloproteinase that cleaves CD16 on the surface of NK cells, can suppress ADCC responses (255). Although ADAM17 inhibitors have been proposed to enhance ADCC in NK cells, the potential advantage of using ADAM17 inhibitors remains a matter of ongoing debate (256).

The cGMP-AMP synthase (cGAS)-stimulator of interferon gene protein (STING) pathway is crucial for immune cell activation via the production of type I IFNs (IFNα and IFNβ) (257). Interestingly, cyclic dinucleotides (CDNs), which are STING agonists, have been shown to activate NK cells (258). NK cell activation appears to be driven by increased IL-15 production in dendritic cells treated with CDNs, in response to increased type I IFNs. Notably, several STING agonists, including CDNs and small molecules, are currently undergoing clinical testing for tumor treatment (259).

Numerous cytokines have been identified as stimulators of NK cell proliferation and activation and cytokine mixtures are being studied to develop adoptive NK cell maturation strategies for cancer treatment. Furthermore, efforts are being made to develop cytokines that can be used as direct cancer treatments, including IL-2 (260), IL-15 (261), IL-12 (262), IL-18, and IL-21 (263). However, most of these cytokines have toxic adverse effects and limited efficacy when used as natural proteins. Consequently, novel approaches are being explored such as combining cytokines with other therapeutics or improving their delivery to target tissues. Among these cytokines, IL-15 potentiates NK and CD8⁺ T cell activity. Clinical trials are currently underway for heterodimeric fusion (hetIL-15) packaged in extracellular vesicles (264) and the superagonists IL-15 and N-803 (223, 265). Delivery methods for IL-12 such as direct intratumoral gene delivery or plasmids in combination with PD-1 inhibitors have also been tested (266). Furthermore, a modified form of the proinflammatory member of the IL-1 cytokine family, IL-18, has been suggested (267).