Targeting natural killer cells: from basic biology to clinical application in hematologic malignancies

Metabolic reprogramming has been widely revealed in cancer cells with the appearance of increased glycolysis, lipid synthesis and amino acids catabolism, which not only serves as crucial determinant in signal pathways for sustaining tumorigenesis, but also has profound implication to immunocytes [47, 48]. Immune cells including NK cells have been subsequently found in engagement in metabolic manipulation due to competition for fuels with malignant cells in TME. Depletion of nutrients, aberrant accumulation of toxic metabolites and intermediates in TME influence NK cell proliferation and effector function. Clinical applications of targeting tumor metabolism have emerged and achieved remarkable progresses over the past decades, such as using metabolomics-based biomarker for early diagnosis and therapeutic approaches that aim at metabolic enzymes or metabolites [49, 50].

Adoptive infusion of NK cells has overcome uncontrolled acute GvHD reaction, the principal barrier of adoptive T cell therapy [102]. The first exploited sources for adoptive NK cell therapy were autologous NK cell as early as 1985. Metastatic cancer patients who had failed in standard therapy were treated with 1.8 to 18.4 × 10¹⁰ autologous NK cells, observing objective regression (more than 50 percent of tumor volume) in 11 of the 25 patients [10]. Hareth Nahi et al. demonstrated the feasibility of infusing autologous NK cells in MM patients [103]. Another phase II clinical trial of adoptive transfer of haploidentical NK cells found no decrease of the cumulative incidence of relapse and no improvement of overall survival (OS) in AML patients, which mainly attributed to the limited persistence of alloreactive donor NK cells [104]. Further research discovered that using activating factors such as IL-2 could assist augmentation of NK therapeutic efficacy. However, IL-2 also added toxicity and complication including elevation of creatinine and bilirubin levels, oliguria, hypotension at the same time [105, 106]. Human umbilical cord blood (UCB) and placenta are rich sources for cytotoxic CD56⁺ NK cell which has high lytic capabilities. It is estimated that around 30% of lymphoid populations in UCB are NK cells, which tend to be younger and have a stronger proliferation potential [107]. Harry Dolstra et al. evaluated the safety and functional effect of NK-cell product derived from HLA partly matched UCB. These NK cells were demonstrated to be well tolerated, and 2 of 4 AML patients became minimal residual disease (MRD) negative in bone marrow after adoptive infusion [108], indicating that immunotherapy based on UCB-derived NK cells has remarkable therapeutic potential. Subsequently, strategies targeting NK cells modification to obtain durable anti-tumor ability experienced great advances. Pluripotent stem cell-derived NK cells engineered with key surface molecules such as high-affinity noncleavable variant of CD16a were demonstrated to improve antibody-dependent cellular cytotoxicity (ADCC) properties of NK cells and contribute to tumor regression in B-cell lymphoma xenograft studies [109, 110]. Clonal NK cell lines containing NK-YS, KHYG-1, NK-92, has become alternative sources for adoptive NK cell therapy [111‐113]. Due to potentially broad applicability against tumors and little risk potential for GvHD complications, NK-92 cell line has become a popular agent in thriving clinic trials [114, 115]. However, application of clonal NK cell lines still faces challenges. For instance, irradiation is required before cell infusion to prevent further hyper-proliferation, which in turn drastically limits cell persistence after infusion [116].

Clinical achievements and current shortcomings in CAR-T cell therapies consisting neurotoxicity and cytokines release syndrome (CRS) force the improvement of alternative approaches. CAR-NK cells refer to engineered genetically to express specific CAR structures which mainly have three domains: I) extracellular region, containing a single-chain variable fragment (scFv), generally derives from antibodies that recognize surface antigens of tumor cells. II) transmembrane region, anchors CAR to NK cell membrane. III) cytoplasmic domain, transmits activating signals then causes downstream processes and facilitates the killing effects [117‐119]. Key components of CAR are shown in Fig. 3. NK cells distinct biology allows them to offer alternative, and perhaps even superior immunotherapeutic strategy in comparison with CAR-T cell therapy. First, CAR-NK cells have reduced risk of GvHD due to a non-HLA-restricted modality [120]. Restricted lifespan of CAR-NK cells in circulation allows no requirement for suicide vectors to prevent excessive expansion [121]. Activated NK cells release GM-CSF and IFN-γ, rather than proinflammatory factors consisting IL-1 and IL-6, implying less likely to occur CRS and neurotoxicity and preferable safety profile [120, 122]. Besides, CAR-NK cells exert more killing modes containing executing cell degranulation, activating apoptotic pathways, and mediating ADCC effects [123]. Cancer-initiating cells (CICs) which play a vital part in tumor recurrence are often characterized by resistant to drugs and irradiation therapy. It has been found that CICs were highly sensitive to NK cells, confirming therapeutic CAR-NK cells could become a strategy for recognition and clearance of CICs and thus prevent tumor recurrence [124]. NK cells have been identified with memory-like function in previous studies, which could allow a more rapid and robust response [125]. Moreover, CAR-NK cells also protect against pathogens such as bacteria and virus, helping to prevent concurrent and secondary infections, which plays an essential role in immunocompromised state in cancer patients.

CAR-NK cell therapies have revealed preliminary potential in a large number of animal experiments [126‐128], which laid a sufficient foundation for human studies. For example, arming cytokine-induced NK cells with a neoepitope-specific CAR significantly enhanced their anti-tumor responses and avoided off-target toxicity in AML models [129]. 30 CAR-NK cell related clinical studies for hematologic malignancies were retrieved on https://​beta.​clinicaltrials.​gov [till Feb. 2023]. Most of them are in early-stage aiming to determine the safety and initial efficacy. In a phase I and II clinical trial performed to assess the safety, relative efficacy and overall response rate (ORR), HLA-mismatched cord blood derived anti-CD19 CAR-NK cells were infused to 11 enrolled patients with relapsed/refractory (R/R) CD19⁺ tumors. One of three dose-regimes (1 × 10⁵, 1 × 10⁶, or 1 × 10⁷ cells per kilogram of body weight) of CAR-NK cells were administered after chemotherapies. GvHD, CRS and neurotoxicity were not found after infusion and therapeutic evaluation result was encouraging. Among them, 8 achieved a response including 7 (3 with CLL and 4 with lymphoma) achieved complete response (CR) [NCT03056339] [130]. The shortcoming of this trial was that different therapeutic interventions were received before and after adoptive infusion of CAR-NK cells, exact conclusions regarding the efficacy cannot be drawn. Nkarta, a clinical-stage biotechnology company, announced positive preliminary results of NKX101 and NKX019 in Apr 2022. NKX101, engineered to target NKG2D ligands on cancer cells, showed striking early single-agent activity and no dose-limiting toxicities in R/R AML or myelodysplastic syndromes (MDS) patient populations [NCT04623944]. NKX019 is another leader CAR product engineered to target B-cell antigen CD19. Evaluating NKX019 in B cell malignancies found 3 of 6 patients treated with higher dose level of three-dose regimens achieved 50% CR. It was also proved with satisfying tolerance, the most frequent higher-grade adverse events (AEs) were myelosuppression [NCT05020678]. Although therapeutic potential and clinical outcomes may materially change as patient enrollment continues, progression of these candidates is worthy expecting. Details of other current clinical trials about CAR-NK cell therapy in hematologic tumors are concluded in Table 1.

Further researches are still needed to conquer existing barriers of CAR-NK cell therapy: I) choose more suitable CAR structures, optimize the distance between epitopes and NK cell surface to enhance their effect [131, 132]. II) seek efficient gene transfer approaches. Traditional method for T cells by viral transfection resulted transgene expression of NK cells in low levels and damped their survival [119, 132]. A variety of small molecular compounds have been employed to reduce the repulsion of NK cells to foreign viral particles via charging cells or colocalizing viruses and cells in close proximity [120, 133]. Novel non-viral methods such as electroporation have been also proved to increase transfection efficiency [119]. III) exogenous cytokines are indispensable for survival and proliferation of infused NK cells, while these cytokines cause undesirable AEs like cross stimulating other subpopulation of immunocytes consisting regulatory T cells, which lead immunosuppressive environment for NK cells [134]. Pharmacological interventions, designed CARs targeting NKG2DL-expressing MDSCs and strategies to disrupt TGF-β signaling show the potential to preserve NK cell therapy efficacy [135, 136].

Immune checkpoint receptors anchored on cell surface could mediate the delivery of either inhibitory or activating signals, the balance of which decides if NK cells remain in a quiescent state or kill target cells [40, 137]. Targeting these immune checkpoint receptors has exploited a prospective therapeutic strategy for hematologic tumors. ICB serves the role of reactivating anti-tumor immunoreaction through blocking inhibitory molecules on the surfaces of tumor-infiltrating lymphocytes [138]. Marketing approved and well-studied ICBs of NK cells in recent years were concluded in Table 2.

Targeting PD-1 and PD-L1 mAbs are one of the first research hotspots entering people’s vision, which have been observed to treat both hematologic and solid tumors [139, 140]. Genetic analyses identified that RS cells of Hodgkin Lymphoma (HL) exhibited frequent amplification of 9p24.1, leading to overexpression of correlative gene products PD-L1 and PD-L2. The amplification also increased JAK/STAT pathway in turn by acting on JAK2 locus and further drove PD-L1 expression [140]. PD-L1 overexpression of malignant cells lays the groundwork for strengthening the anti-tumor functions of NK cell via blocking PD-1/PD-L1. Pembrolizumab is a humanized IgG4 PD-1 mAb with high-affinity. Philippe Armand et al. reported Pembrolizumab treatment in 31 HL patients, showing only 5 grade 3 drug-related AEs. CR rate was 16% (5/31) and partial response (PR) rate was 48% (15/31), most responses sustained at least 24 weeks with a median follow-up of 17 months [NCT01953692] [141]. The results identified considerable therapeutic outcome on account of blocking inhibitory immune checkpoints. Moreover, application of PD-1 blockades has broken new ground for advanced R/R tumors. Patients with advanced Sezary syndrome (SS) and mycosis fungoides (MF) suffer worse disease progress and poor OS. One phase II trial found 38% (9/24) ORR including 2 CR and 7 PR among 24 enrolled patients with advanced SS or MF in Pembrolizumab treatment regime of 2 mg/kg every 3 weeks [142]. Pembrolizumab also delineated satisfactory therapeutic effect in other hematologic malignancies such as MM and lymphoma [143, 144].

As above mentioned, KIRs are generally expressed on NK cells to preclude normal cells from damaging. Early study found that KIR epitope incompatible transplants could achieve higher engraftment rates [145]. Lacking of interaction between KIR-MHC I was concluded to trigger NK cell activity, suggesting that specifically blocking the recognition and binding between them could rejuvenate NK cells [146]. Relevant clinical trials are booming, early safety evaluation test of anti-KIRs ICBs draw desirable results but the efficacy tests were inconsistent [147‐149]. And addition of KIR blockade improved meaninglessly in the efficacy over single agent PD-1 blockade in some clinical studies [150]. Heterogeneous results may approximately due to that study populations were too small to overcome the biological heterogeneity within and across disease types. Thus, additional benefits accruing from the combination regimens still require longer follow-up time to assess.

Application of ICBs and the resultant breakthrough strategy of taking advantage of immune system for tumor therapy have displayed profound superiorities. However, there are still varieties of challenges need to be addressed. The major barrier is drug-resistant, either lacking of initial response to treatment or with initial promising response but developed resistance during therapeutic stage [151]. Published studies have revealed the minimal expression of PD-1 on NK cells in several tumors, which may lead to primary drug-resistant to ICBs [152]. Strategic approaches such as the combination of immune checkpoint treatments with other therapies including angiogenesis inhibitors and oncolytic viruses have been demonstrated to improve the responses to ICBs [153]. A complex of checks and balances were observed in human immune system that afford response or preserve tolerance. ICBs have the capacity for this homeostatic balance perturbation, causing immune related AEs (irAEs). IrAEs refer to inflammatory adverse events due to non-specific stimulation of immune system by ICBs, generally involving endocrine glands, skin, gastrointestinal tract [154]. IrAEs have been described in many clinical trials including nausea, vomiting, diarrhea, bilirubin increase, rash et al. [NCT01822509] [155]. Besides acute clinical toxicities of these agents, chronic irAEs (usually refer to > 12 weeks sustaining after ICBs discontinuation) are more prevalent [156]. Long-term potential chronic toxicity may be ignored because clinical studies tend to pay attention to the most frequent treatment-associated adverse effects. A generally limited life expectancy for patients with metastatic tumors constrains long follow-up time to exhibit chronic irAEs. Endocrinopathies and rheumatological toxicities has become the most frequent chronic irAEs, and pneumonitis, neuropathy, dermatitis et al. are relatively low-prevalence events [157]. The onset of irAEs varies widely and is hard to predict. Terminating ICBs therapy and beginning high-dose corticosteroids treatment are most commonly used for irAEs control. But there still remains drawbacks including irAEs-overlapped drug toxicities, serious infections, and the risk of suppressing tumor immunosurveillance [158]. Deeper exploration of intrinsic and extrinsic factors that impact ICB response and standard managements built on these hallmarks are intensively needed.

Cytokines appear to be critical in many aspects including regulating innate or adaptive immunity, cytogenesis, cell growth, as well as damaged tissue repairment [159]. Earlier results have shown cytokines such as IL-2 could promote the regression of solid tumor models established in animals [160, 161], providing novel insights for augmenting anti-tumor effects of immunocytes containing NK cell by cytokines. Table 3 showed recent applications of cytokine-induced NK cells for treatment of hematological malignancies.

Bi- and tri-specific antibodies are designed to build efficient immunological synapses between immune cells and malignant cells, specifically recruiting immune cells and forming more densely interconnected to tumor cells. Bispecific antibodies targeting CD3 and specific tumor epitopes to recruit T cells developed rapidly over the past decades. Blinatumomab is one FDA approved CD3/CD19 bispecific T cell engager (BiTE) in Dec. 2014 for adult leukemia. Both CAR T cell therapy and BiTE applications are limited by severe toxicities [177], leading partial study focus more and more turn to NK cell engagers. BiKEs or TrikEs are formed by single variable heavy (V_H) and light chain (V_L) of certain antibody, of which a flexible polypeptide linker joined to keep from dissociating [178]. Structures of BiKEs or TrikEs are visualized in Fig. 3. Current NK cell engagers are mainly designed with CD16 and tumor epitopes like CD33 simultaneously, which have several additional advantages compared to mAbs. They are non-immunogenic so that could alleviate many complications of their CAR counterparts [179]. Size of NK cell engagers are small, mainly among 50–80 kDa, which allows efficient tumor penetration and increased biodistribution of the agents. Small size of these fragments allows rapidly elimination through kidneys, contributing to maintain appropriate serum concentration levels and limit associated toxicities [180].

It was previously believed innate immunocytes consisting NK cells lack antigen-specificity and immunologic memory because their receptor genes cannot undergo somatic rearrangements. However, “ML NK cells” were subsequently observed in mice, NK cells possessing expression of Ly49H receptors had ability to drive the expansion during infectious phase. Expanded effector NK cells then established a pool of long-lived antigen-specific cells with a unique transcriptional signature [199, 200]. NK cells pre-activated through cytokines and then adoptive transferred into vivo showed enhanced proliferation and exhibited restimulation responses to cytokines [201]. Effector function and persistence of syngeneic IL-preactivated NK cells were observed, which also markedly reduced the growth of established mouse tumors [202]. A phase I study observed 5 clinical responses of 9 evaluable AML patients adoptively transferred ML NK cells, demonstrating robust responses of these immunocytes to leukemia cells [203]. An innovative proposal to enhance tumor-specific recognition of ML NK cells by modifying with CARs was come up to increase IFN-γ generation, degranulation and cytotoxicity of NK cells. For instance, ML NK cells derived from lymphoma patients engineered with anti-CD19 CAR reduced lymphoma burden and thus obtained survival improvement in human xenograft models [204]. Han Dong et al. found that arming ML NK cells with a neoepitope-specific CAR significantly enhanced anti-tumor response and avoided off-target toxicity in AML, suggesting that ML NK cells represented a promising cellular platform for modified adoptive cell therapy [129]. CAR-ML NK cells offer apparent advantages including inducing response to NK cell-resistant tumor targets through an obvious synergistic cooperation of CAR-mediated effects and “memory”, representing a powerful tumor immunotherapy approach.