The Rise of Allogeneic Natural Killer Cells As a Platform for Cancer Immunotherapy: Recent Innovations and Future Developments

Natural killer (NK) cells are critical immune effector cells in the fight against cancer. As NK cells in cancer patients are highly dysfunctional and reduced in number, adoptive transfer of large numbers of cytolytic NK cells and their potential to induce relevant antitumor responses are widely explored in cancer immunotherapy. Early studies from autologous NK cells have failed to demonstrate significant clinical benefit. In this review, the clinical benefits of adoptively transferred allogeneic NK cells in a transplant and non-transplant setting are compared and discussed in the context of relevant NK cell platforms that are being developed and optimized by various biotech industries with a special focus on augmenting NK cell functions.


NATURAL KiLLeR (NK) CeLLS iN ONCOLOGY
So far T cells have been the mainstay of cancer immunotherapy; however, it is generally recognized that NK cells also play an essential role in antitumor immunity. Certainly in the prevention of metastases through the elimination of circulating cancer stem cells with a high metastatic potential, NK cells are recognized as main immune effector cells (1). Moreover, as solid tumors have a propensity to particularly down-regulate MHC-I, NK cells provide a failsafe mechanism in these circumstances where cytotoxic T cells, which depend on MHC-I for tumor recognition and elimination, are debilitated. NK cells have recently been more intensely explored as a viable therapeutic platform next to T cell-based approaches. This review aims to summarize the latest developments in the clinical translation of adoptive transfer of NK cells in the oncology field.

NK CeLLS AND THeiR ACTivATiNG AND iNHiBiTORY ReCePTORS
Human NK cells are generally categorized by their level of CD56 and CD16 expression into two subsets: CD56 bright CD16 dim and CD56 dim CD16 bright NK cells. Most NK cells in the peripheral blood and spleen are CD56 dim CD16 bright and are cytotoxic against a variety of tumor cells, whereas CD56 bright CD16 dim NK cells are immune regulatory in function and constitute the majority in secondary lymphoid tissues, producing abundant cytokines but exerting weak cytotoxicity compared to CD56 dim CD16 bright

NK CeLL DYSFUNCTiONALiTY iN CANCeR
Natural killer cells can control circulating tumor cells and prevent formation of tumor metastases (11). However, tumors employ different strategies to evade killing by NK cells. Upregulation of inhibitory ligands such as MHC class I molecules (HLA-ABC, HLA-G and HLA-E) has been associated with a stronger inhibitory signal to NK cells (12)(13)(14)(15). Furthermore, increased expression of the inhibitory NKG2A receptor reported in renal cell carcinoma resulted in decreased functionality of tumor infiltrating NK cells (16). On the other hand, downregulation of NK-activating ligands for NKG2D such as MICA and MICB and increased shedding of tumor-derived soluble MIC also impair NKG2D-mediated NK cell tumor recognition (17). Another important necessity for optimal NK cell function is the ability to home and migrate to tumor sites. Several studies have correlated increased homing of NK cells to tumor tissues with improved treatment outcomes in solid tumors (18)(19)(20)(21)(22). However, the immunosuppressive tumor stroma comprising regulatory T cells (T-regs) (23), myeloid-derived suppressor cells (MDSCs) (24), M2 macrophages (25), and immature dendritic cells severely restricts NK cell functionality and their entry into solid tumors. In chronic diseases, such as those associated with human immunodeficiency virus and cytomegalovirus infections, mainly exhausted NK cells with decreased cytokine production and reduced cytolytic activity are observed (26,27). In a study with breast cancer patients, the NK cell expression levels of activating receptors (NKG2D, DNAM, CD16, and NKp30) were decreased, whereas inhibitory receptor (NKG2A) expression levels were increased and this apparent dysfunctionality of NK cells was found to directly affect NK cell cytotoxicity (28). Similarly, the effector subset of NK cells (CD56 dim CD16 + ) from head and neck and breast cancer patients, when tested in vitro, was highly prone to apoptosis, thus pointing to low NK cell activity in these patients (29). Impaired NK cell functionality may result from tumor-imposed suppressive mechanisms and presents a major hurdle for NK cell-targeted immunotherapies. Therefore, approaches to restore or replace impaired NK cell cytotoxicity may prove essential for an effective host defense against cancers.

NK CeLLS iN THe CLiNiC
Novel NK cell-based immunotherapeutic strategies are being developed to overcome the functional limitations of the use of cancer patients' autologous NK cells. To increase the number of functional NK cells even in case of a high tumor load, adoptive transfer of autologous NK cells served as a very feasible approach, as this ruled out the need for immunosuppression, HLA-matching, and prevented the risk of graft versus host disease (GvHD). These advantages sparked the initiation of large-scale expansion protocols and clinical trials using autologous NK cells as a treatment modality for cancer. Though adoptive transfer of autologous NK cells resulted in an increased number of circulating NK cells in peripheral blood, it failed to produce significant therapeutic effects in hematological malignancies, metastatic melanoma, and renal cell carcinoma patients due to the inhibition by self-HLA molecules (30)(31)(32). Moreover, the expansion efficiency and functional status of autologous NK cells were still limited when compared to allogeneic NK cells, as autologous cells were often obtained from heavily pretreated patients (33). In addition to this, it was difficult to track infused autologous NK cells in patients and to study their antitumor effects from peripheral blood analyses due to the inability to differentiate ex vivo manipulated and transferred autologous NK cells from the non-manipulated circulating NK cells. These limitations motivated researchers shifting their focus to allogeneic NK cells to treat cancer.
In patients with leukemia undergoing allogeneic hematopoietic stem cell transplantation (HSCT), NK cells, being the first lymphoid subset to appear after allogeneic HSCT (34), play a crucial role in controlling host defense against infections and residual cancer cells before T cells are reconstituted (35). These donor T cells are prime mediators of GvHD (36), and the lifethreatening complications that arise due to GvHD have completely overshadowed the beneficial effects of alloreactive NK and T cells, fueling efforts to use T cell depleted grafts (37). Further, this led to the development of NK cell-based therapies coupled with T cell depleted HSCs to enhance the graft versus tumor effect (GvT) without causing GvHD. Unlike autologous NK cells, allogeneic NK cells are not restricted by the patient's tumor's HLA expression, which is an added advantage to mount an improved anti-tumor effect (38,39). Current translational efforts that are explored as anticancer therapies include adoptive transfer of ex vivo activated and/or expanded allogeneic NK cells, either alone or in combination with HSCT.

SOURCeS OF ALLOGeNeiC NK CeLLS USeD iN THe CLiNiC
Commonly used allogeneic NK cells are apheresis products collected from haploidentical and unrelated donor PBMC (40). Another source is umbilical cord blood (UCB), where NK cells are generated from CD34+ progenitor cells that undergo expansion and differentiation using cytokines and growth factors and thereby mature into cytolytic NK cells (41). Apart from PBMC and UCB, NK cells have also been obtained from the clonal cell line NK-92, derived from immortalized lymphoma NK cells (42,43).

ALLOGeNeiC NK CeLL THeRAPY iN A TRANSPLANT SeTTiNG
Autologous or allogeneic HSCT serves as a curative regimen by reconstituting the immune system in hematological malignancies. At an earlier stage post HSCT, NK and T cells developing from the graft are immature and less in number with reduced functionality. Under those circumstances, the infusion of purified allogeneic NK cells was explored as a viable option to target minimal residual disease (MRD), prevent graft failure, and relapse. Grafts for allogeneic HSCT and allogeneic NK cell treatments were obtained from HLA matched/mismatched and related/ unrelated donors (38,39).  (49) have shown that NK cells can be safely administered prior to or post HSCT in patients with different types of hematological diseases. Immune suppression is a prerequisite prior to most of the allogeneic HSCT and NK-cell infusions. A non-myeloablative conditioning regimen usually consisting of cyclophosphamide (Cy) and fludarabine (Flu) was found to facilitate NK cell persistence and expansion in vivo (50). High doses of Cy/Flu caused pancytopenia and resulted in high plasma IL-15 levels, which also correlated with the detection of adoptively transferred NK cells up to 14 days after infusion, thus suggesting that excess IL-15 was probably utilized by the NK cells to proliferate and persist longer in vivo (51). A summary of clinical trials with allogeneic NK-cell infusions in a HSCT setting with published data is summarized in Table 1, and selected clinical trials from recent years are reviewed below.
In 2013, Stern et al. treated acute myeloid leukemia (AML), Acute Lymphocytic Leukemia, Hodgkin's lymphoma (HL), and sarcoma patients with allogeneic NK cells (CD3 depleted and CD56 selected) after a haploidentical HSCT, using the same donor as NK cell source. An overall survival (OS) of 25% was achieved during a median follow-up of 5.8 years. And 4/16 patients developed acute GvHD (aGvHD) due to high T-cell impurities present in two NK cell products and two from stem cell grafts, both containing ≥0.5 × 10 5 cells/kg T cells (52). Although this prospective Phase II study reported the safety and feasibility of NK-cell infusion following allo-HSCT, it failed to yield results in support of anti-leukemia effects, raising questions as to whether the NK cell dosage of 0.3-3.8 × 10 7 cells/kg used was sufficient to induce a clinical effect. In the same year, Klingemann et al. published data highlighting the safety and alloreactivity of HLA-mismatched (CD3 depleted) NK cells, transfused after autologous HSCT in multiple myeloma (MM), non-Hodgkin's lymphoma (NHL), and HL patients. In this study, 13 patients were enrolled; 6/13 relapsed and 7/13 were in remission during a follow-up between 144 and 1,158 days following autologous stem cell transplantation. The allogeneic NK cells were well tolerated without GvHD. In addition, this study also demonstrated that NK cells generated and processed at distant centers can be shipped and transfused without significantly affecting the viability and cytotoxicity of the NK cell product (53).
Choi et al. (54) summarized their observations from a study, in which allogeneic ex vivo expanded and activated NK cells derived from the same donor were administered 14 and 21 days post HLA haploidentical HSCT to patients with hematological malignancies (n = 41). The data set from this study was compared with a group of 31 patients, who underwent only HLA haploidentical HSCT. A significantly higher progression-free survival (PFS) was seen in the HSCT + NK group compared to HSCT only group (74% versus 46%). In addition to this, the occurrence of chronic GvHD (cGvHD) (15% versus 10%) and transplant-related mortality (27% versus 19%) was reduced in the HSCT + NK group compared to the HSCT only group (54). In another study by Killig et al. (55), AML patients were treated with haploidentical HSCT followed by NK-cell infusions (CD3 depleted and CD56 selected) from the same donor, on days +1 and +2 post HSCT. aGvHD was highly prevalent in 20/24 patients in this study and histological analysis of skin revealed that GvHD was associated with infiltration by perforin + CD8 + T cells. Allogeneic NK cells contributed to an increased OS in the HSCT + NK group compared to the HSCT only group (37% versus 14%) over a median follow-up of 2 years (55).
Subsequently, Shah et al. published data from a Phase I study treating patients with Ewing sarcoma, rhabdomyosarcoma, and desmoplastic small round cell tumors (n = 9), using donorderived IL-15/4-1BBL-activated allogeneic NK cells (CD3 depleted and CD56 selected) following allogeneic HSCT from the same donor. aGvHD was highly prevalent in the patient group that received stem cells from matched unrelated donors and was directly linked with a faster T cell recovery and higher T cell chimerism from reconstituted HSCT grafts. About 4/9 patients were alive in this study with a median follow-up of 23.1 months (56). Lee et al. reported results from a Phase I study  in patients with AML, myelodysplastic syndromes (MDS), and chronic myelogenous leukemia (CML), in which alloreactive haploidentical NK cells (CD3 depleted and CD56 selected) were administered along with IL-2 injections, followed by thymoglobulin conditioning and allogeneic HSCT. Thymoglobulin was administered to prevent NK cells from hampering engraftment of allogeneic HSCT. Out of 20 evaluable patients, 16 had GvHD (10/16 aGvHD and 6/16 cGvHD) after transplantation. In this study, GvHD was not directly associated with donor T cell or NK cell contents. From this study, also it was concluded that the lack of anti-leukemic effect was mainly due to the low dose of infused NK cells and it was further suggested that thymoglobulin conditioning could also have potentially affected NK cells survival in vivo (57). Later, Choi et al. presented results from a modified treatment protocol of four consecutive infusions of ex vivo activated and expanded haploidentical NK cells after HLA-matched HSCT and compared the outcomes to their previous study, in which they administered two infusions. In the subsequent study, additional donor NK-cell infusions were given on days 6 and 9 (i.e., at days 6, 9, 13, and 20). Out of 51 patients with ALL (n = 6) and AML (n = 45), 24/51 (47%) had four NK infusions. Out of 45 evaluable patients, the 3-year OS rate was 9% in AML and 21% for ALL and 9/45 had aGvHD. Early administration of NK cells after HSCT caused significant toxicities with no improvements in antileukemic effects, compared to the previous study. In this study group, a higher CR rate correlated with higher expression levels of NK activating receptors NKG2D and NCRs (NKp44, NKp46, and NKp30) on donor NK cells. In addition, NKp30 expression was significantly higher than that of NKG2D and other NCRs, thus suggesting a role for NKp30 as a predictive biomarker for anti-leukemic effects of NK cells (58).
In 2017, Shah et al. published data from a Phase I study, treating MM patients with UCB-derived NK cells (day 5) along with autologous-HSCT (day 0), following high-dose chemotherapy and low-dose lenalidomide. Mononuclear cells (MNCs) isolated from UCB units (CD3 depleted) were cultured with K562-based artificial antigen presenting cells (aAPCs) expressing membrane bound IL-21. No treatment related toxicities or GVHD was reported in this study. During a median follow-up of 21 months in 12 patients, 4/12 patients had progressive disease (PD) or relapsed. Stable expression of NKG2D and increased expression of CD16 and NKp30 of UCB-NK cells were observed in six patients. This study further reiterates the safety of NK-cell infusions in high doses; however, due to combinatorial set up with HSCT and lenalidomide, it is difficult to interpret the clinical efficacy of UCB-NK cells alone from this study (59).
Taken together, it is evident from these studies, as well as from many others, that GvHD, which is mainly caused by T cells from transplanted grafts, is a major concern in the field of allogeneic HSCT. Under these circumstances, it is difficult to reliably study the safety of allogeneic NK-cell infusions. The timing of NK-cell infusion, NK cell dosage and NK cell promoting conditioning regimens are critical factors that need to be more extensively studied to assess the safety and efficacy of allogeneic NK-cell infusions.

ADOPTive NK CeLL THeRAPY iN A NON-TRANSPLANT SeTTiNG
To gain a better understanding of the safety and efficacy of allogeneic NK cell transfer, investigators started to study NK cells in a non-transplant setting. Landmark clinical trials were performed by Miller et al. (50) Table 2.
Here, we focus on the latest reports from clinical trials using allogeneic NK cells in a non-transplant setting. Bachanova et al. developed a recombinant cytotoxic protein, i.e., an IL-2/diphteria toxin fusion protein (IL2DT), which functions by selectively depleting the IL-2 receptor CD25 expressing cells, including regulatory T cells (T-regs). In total, 57 AML patients were treated with KIR and HLA-mismatched haploidentical NK cells and 15 of them in cohort 3 received IL2DT, 1 or 2 days prior to NK-cell infusion, to deplete T-regs. In addition to IL2DT treatment, three different processing methods were used, i.e., a CD3-depleted cohort (cohort 1, n = 32), a cohort using CD3 depletion followed by CD56 selection (cohort 2, n = 10), and a cohort using CD3 and CD19 depletion (cohort 3, n = 15). Higher NK cell doses were obtained from cohort 3, and it could possibly be the reason for the observed longer disease-free survival (DFS) (33% versus 5% at 6 months) in cohort 3 compared to cohorts 1 and 2. In this study, endogenous IL-15 serum levels correlated with reduced T-reg levels in patients treated with IL2DT (64).
Szmania et al. investigated the effect of infusing cryopreserved and freshly prepared NK cells (CD3 depleted), either allogeneic or autologous, given after bortezomib with or without lymphodepletion in high risk relapsed MM patients. NK cells were cocultured with K562-mb15-41BBL cells for 8-9 days. Initially, 6 out of 8 NK products were cultured with low dose IL-2 (10 U/ml), however, increasing the dosage of IL-2 to 500 U/ml resulted in enhanced expression of NKp30 and NK cytolytic activity without affecting the T cell content of the NK cell product and, therefore the last 2 patients in this study were treated with high dose IL-2 cultured NK cells. The highest post-transfer number of circulating NK cells was observed in the high dose IL-2 group. Patients treated with fresh NK cells showed a median 21-fold increase in peripheral NK cell rates by day 7, while no in vivo expansion of NK cells was seen in patients treated with cryopreserved NK cells. Overall, the NK-cell infusions were well tolerated and no GvHD was observed. From 7 evaluable patients, 6 had (PD), and 1 had a partial response (PR) for up to 6 months post infusion (65).
In the same year, Kottaridis et al. presented data from a clinical trial in AML, for the first time using tumor-primed NK cells from related haploidentical donors. During a 6-month follow-up period (n = 7), three patients in CR remained in CR, one patient in PR achieved CR, two patients relapsed, and one patient died. Median OS was 468 days post NK-cell infusion (66).
Rubnitz and his team reported on the safety and feasibility of haploidentical NK cell therapy in children with relapsed or    (68) and administered to patients with advanced lymphoma and recurrent solid tumors. Following NK cell-adoptive transfer, an increase in NKG2D expression levels on CD8 + T cells, a reduction in the number of T-regs, and MDSCs followed by a decrease in serum levels of transforming growth factor-beta were noted. An enhanced PFS was noted in KIR-mismatched NK cell recipients. In addition, a KIR B haplotype was associated with a higher incidence of stable disease (SD). This study demonstrated that KIR-ligand mismatched donor NK cells can be safely administered without any sign of GvHD and with a GvT effect. Though an antitumor effect of the adoptively transferred NK cells could be observed, their persistence in vivo was shorter (between 1 and 4 days) in comparison to other clinical trials. This stresses the potential need for an effective NK cell-promoting conditioning regimen, to increase the life span and migration of NK cells in patients (69).
Shaffer et al. published results from a Phase II study in patients with relapsed or progressive AML (n = 6) or MDS (n = 2), treated with allogeneic NK-cell infusions and supported by IL-2 injections in vivo. NK cell donor chimerism was not detected post infusion, and no signs of GvHD were reported from this study. About 3/8 patients achieved PR of which 1/6 patients with AML and 1/2 patients with MDS achieved a CR after treatment but relapsed within 2 months. Of note, both these patients survived for 20.2 months post NK infusion, while the remaining 5 patients without response had a median survival of 5.4 months (70).
Around the same time, Curti et al. published results from a Phase I trial using KIR ligand-mismatched haploidentical NK cells to treat AML patients in CR (n = 16). About 7/16 patients relapsed, while 9/16 remained disease free at a median follow-up of 22.5 months. Overall, 69% (11/16) responded to therapy, with no signs of GvHD. Prolonged DFS was higher in patients with an absolute increase in the number of circulating alloreactive NK cells in this study (71).
Romee et al. investigated the antitumor effects of cytokine induced memory-like NK cells, which were adoptively transferred in relapsed or refractory AML patients after overnight activation with IL-12, IL-15, and IL-18. They were defined as memory-like NK cells based on their enhanced responsiveness upon restimulation with cytokines. NK-cell infusions were safe and well tolerated and no GvHD was reported in this study. Out of nine evaluable patients, four had CR and five showed disease responses (72).
Most recently, a first clinical trial using UCB CD34+ progenitor cell-derived NK cells was published by Dolstra et al. Allogeneic NK cells were generated from UCB CD34+ cells using an ex vivo expansion and differentiation method developed by Glycostem Therapeutics (41). In this study, 10 AML patients in morphologic CR, who were ineligible for HSCT transplantation, received partially HLA-matched (5/10 KIR ligand-ligand mismatched and 7/10 KIR receptor-ligand mismatched) UCB-NK cells (oNKord ® ). Following Cy/Flu conditioning, lymphocytopenia was induced and found to correlate with elevated IL-15 levels, which peaked at day 6 after NK-cell infusion. Out of 10 treated patients, 5 were alive and 4 had a DFS of 60, 52, 22, and 16 months after infusion. About 2/4 patients with very poor prognosis (i.e., with detectable MRD in bone marrow before NK infusion) became MRD negative for 6 months post NK infusion, indicating a potential GvT effect. UCB-NK-cell infusions were safe and well tolerated without signs of GvHD. Interestingly, UCB-NK cells expressing low levels of KIRs and CD16a at the end of the ex vivo culture, underwent further maturation post-transfer in vivo, resulting in the upregulation of KIRs and CD16a, but continued to preserve the activated phenotype denoted by high expression of NKp30, NKp44, NKp46, NKG2D, and DNAM (73).
In addition to NK cells from PBNK and UCB-NK, two clinical studies reported on the use of the NK-92 cell line. A Phase I trial conducted by Arai et al. investigated the safety and feasibility of allogeneic NK-92 cells in advanced renal cancer and melanoma. A total of 12 patients were evaluated and 6/12 had PD, 4/12 had SD, while 1/12 had minor response and 1/12 had mixed response, 4 weeks post infusion (42). Similarly, in another study with NK-92 conducted by Tonn et al., 15 patients were included with advanced solid (n = 13) and hematological malignancies (n = 2). About 1/7 tested patients produced antibodies against the HLA antigens expressed by NK-92 cells, 1/15 showed SD, 1/15 a mixed response, and the rest of the patient group had disease progression, being treated with a maximum tolerated dose of 10 10 cells/ m 2 . NK-92 cells had a very short persistence (48 h) in vivo (43). As NK-92 cells are derived from cancer (lymphoma) cells and require irradiation before infusion, which could hamper their ability to proliferate and home in vivo, potentially limiting their efficacy.
Overall, analyzing the data from adoptive allogeneic NK cell therapy trials in a non-transplant setting, we conclude that such treatments are very safe and well tolerated and efficacious in hematological malignancies, especially in AML, but as yet relatively ineffective in solid tumors. Trials using allogeneic NK cells alone yielded valuable information on the in vivo persistence, donor chimerism, and antitumor potential in different indications. Furthermore, unlike combined approaches with HSCT, the absence of life-threatening GvHD and major treatment-related toxicities makes this method advantageous and provides an opportunity to further enhance the cytotoxic effects of allogeneic NK cells.

APPROACHeS TO AUGMeNT NK CeLL FUNCTiONS: A view ON BiOTeCH iNDUSTRieS
As reviewed above, various clinical trials have been published, mainly initiated by academia, proposing allogeneic NK cells as an effective therapeutic option. As a result of these studies, interest in NK cell-based immunotherapy strategies has been engendered in an increasing number of biotech companies. Clinical trials conducted in academia are often restricted to Phase I or II, as progression of experimental therapies to Phase III clinical trials and further on to commercialization and marketing requires a level of funding that surpasses the capacity of academic institutions. The financing of market enabling studies is coming mainly from industry. Although NK cells can be effective in some types of cancer as a monotherapy, considering their heterogeneity, complex networking, and the inherent adaptability of several tumors to evade killing by immune cells, one believe is that it is necessary to improve on the efficacy of currently available NK cell products. In this respect, it is worthwhile to consider combinatorial approaches of different treatment strategies involving NK cell functions. A summary of biotech companies involved in NK cell research is listed in Table 3 and Figure 1. Here, we review for a selected group of NK cell companies, which develop NK cellspecific treatments, the underlying scientific principles and findings of their product pipelines, revealing highly innovative concepts that herald future clinical applications.

Fc OPTiMiZeD MONOCLONAL ANTiBODieS (mAbs)
The potential of NK cells to mediate ADCC with therapeutic mAbs has been well described over the years (74). However, concerns have been voiced based on results from certain clinical trials, showing that polymorphisms in NK CD16 (V158V, V158F, and F158F) could influence the efficacy of mAb treatment and ADCC (75). To address this issue and limit the variations between different CD16 sequences, Fc glyco-engineered (defucosylated) mAbs with enhanced binding affinities to NK CD16a were developed. The Fc optimized anti-CCR4 mAb mogamulizumab (76) (Kyowa Hakko Kirin) has entered Phase III clinical testing in patients with adult T cell leukemia, emerging as the lead NK cell ADCC product to reach the market soon. Fc-optimized anti-CD20 mAbs Obinutuzumab (Genentech) (77) and Ocaratuzumab (Mentrik Biotech, LLC) (78) are currently tested in patients with chronic lymphocytic leukemia and follicular lymphoma. Similarly, the Fc-optimized anti-EGFR mAb imgatuzumab (Roche Glycart) is tested in Phase I/II clinical trials for head and neck cancer and in KRAS mutant colorectal cancer (79,80). Although Fc-engineered mAbs address NK-mAb-binding issues, reports of serious side effects, like from the imgatuzumab study (81), have made the scientists rethink this strategy and call for the careful study of the advantages and disadvantages of this approach.

BiSPeCiFiC ANTiBODieS
In the last decade, several bispecific and trispecific Ab platforms, simultaneously targeting immune cells and tumor cells, have been developed in the field of cancer immunotherapy (82). To date, the majority of bispecific Abs that has been developed targets T cells, while only a limited number of bispecific approaches targets NK cells (83). Affimed is a clinical stage pharmaceutical company developing bifunctional antibodies that recruit immune cells such as T and NK cells to tumor sites. These bispecifics (TandAbs) are tetravalent in nature, thus offering four binding sites, two aimed at tumor antigens and two aimed at immune cells. Currently, Affimed's AFM13 that targets CD30 on cancer cells and CD16a on NK cells is in clinical Phase II testing in patients with HL. In Phase I studies AFM13 was found to be safe and well tolerated and resulted in an overall response rate of 23%. Furthermore, AFM13 treatment resulted in an increase in NK cell activation and a decrease in soluble CD30 levels in peripheral blood (NCT01221571) (84). Further, two other bispecific CD16a-based tumor targeting antibodies are in preclinical phase development, i.e., AFM22 and AFM24 that bind to EGFRvIII expressed by several solid tumors, including glioblastoma (GBM), and wild-type EGFR, respectively. Another promising NK cell-focused bispecific platform is developed by AvidBiotics to target tumors that evade NK killing via downregulation or shedding of the NKG2D ligand MICA, which is a major limiting step in NK-mediated tumor targeting. To overcome this, AvidBiotics designed MicAbody proteins that bind to the NK cell NKG2D receptor with high affinity. Further, this MicAbody was engineered with an additional binding site to target tumor antigens of interest, thus enabling recruitment of NK cells to tumors (85).

NK CeLL CHeCKPOiNT iNHiBiTORS
Another strategy to increase NK cell functionality is the disruption or blocking of NK inhibitory signals. Innate Pharma is a clinical stage pharmaceutical company focused on developing NK cell checkpoint inhibitors. Lirilumab (IPH2102/BMS 986015) is a fully humanized IgG4 anti-KIR mAb against the inhibitory KIRs KIR2DL1, L2, and L3, which are expressed predominantly on NK cells and on some T cells. Lirilumab induced significant anti-tumor activity of NK cells against HLA-C-expressing tumor cells, contributing to increased survival in lirilumab-treated mice (86). Similar to KIRs, the NK cell inhibitory receptor NKG2A binds to its ligand HLA-E on tumor cells resulting in an inhibition of NK cell function. HLA-E is overexpressed in colon, cervical, and ovarian cancers, thus serving as an escape mechanism for NK killing in these tumors (87,88). The anti-NKG2A mAb monalizumab was developed to block the interaction between NKG2A and HLA-E and is currently under clinical investigation. IPH4102, which targets KIR3DL2, is under Phase I clinical investigation in cutaneous T cell lymphoma (CTCL). Clinical trials testing lirilumab, monalizumab, and IPH4102 are listed in Table 3.

GeNeTiC MODiFiCATiON OF NK CeLLS
In addition to successful expansion, differentiation, and demonstrable anti-tumor effects of NK cells, NK cell tumor targeting can be made more specific by employing chimeric antigen receptors (CARs) as demonstrated for T cell adoptive transfer strategies (89). CARs are recombinant Ab-based molecules that upon expression in immune effector cells bind antigens of interest on target cells, resulting in immune activation and enhanced immune effector cell survival through specific   Besides specific targeting of tumor antigens and strategies to promote ADCC, Nkarta therapeutics developed NKG2D CARs (NKG2D-CD3ζ-DAP10) using NK-92 cells and PBNK cells, which exhibited enhanced cytotoxicity against osteosarcoma and hepatocellular carcinoma when compared to activated and expanded PBNK cells (100,101). mRNA-based genetic engineering has been used to enhance migration of NK cells to tumors.
Apart from gene modification, gene editing is also widely used to overexpress or knock out genes of interest to augment NK cell function. Expression of HLA-A on allogeneic NK cells leads to rejection of allogeneic NK cells by the recipient's T and NK cells. Cooper and colleagues from Ziopharm Oncology used zing finger nuclease (ZFN) technology to remove HLA-A sequences from allogeneic NK cells, thus enabling these immune effector cells to escape rejection from recipient T cells. However, in that case, there is yet a high probability of being attacked by endogenous NK cells targeting HLA-A negative allogeneic cells. This was further addressed by retaining HLA-B and HLA-C genes in donor NK cells (102)(103)(104). To increase NK cell persistence in vivo, scientists at oNKo-innate identified a group of proteins called suppressor of cytokine signaling (CIS, SOCS 1-7), which negatively regulate CIS pathways. SOCS1 and SOCS3 bind to JAK1, JAK2, and TYK2 molecules and inhibit JAK activity. Similarly, CIS protein binds to JAK1 and suppresses IL-15 signaling in NK cells. It became evident from in vivo studies in mice with Cish − / − knockout NK cells that loss of CIS led to prolonged IL-15 signaling, resulting in an increased proliferation, survival, and functionality of NK cells (105).

NK CeLLS FROM iPSCs
In recent years, NK cells generated from induced pluripotent stem cells (iPSC-NK) and human embryonic stem cells (hESC-NK) have been gaining more interest as an NK cell therapeutic product. Fate Therapeutics developed a platform technology to generate NK cells from iPSC. hESC/iPSC were made into aggregates by centrifugation to form so-called embryoid bodies (spin EBs) (106), giving rise to hematopoietic progenitor cells expressing CD34 and CD45, which were then differentiated into mature NK cells using a specific cytokine cocktail. iPSC/hESC-derived NK cells were shown to express common NK cell markers, such as KIRs, CD16, NKp44, NKp46, NKG2D, and TRAIL, and were cytotoxic against several hematological and solid tumor cells in vitro (107,108). In the next stage, iPSC/hESC-derived NK cells were successfully expanded using IL-2 and K562-based aAPCs with membrane-bound IL-21 to generate sufficiently high numbers for clinical applications (109).

NK CeLLS FROM HUMAN UCB CeLLS
Stem cell progenitors from cord blood offer a unique platform to be expanded and differentiated into cytotoxic NK cells. The low immunogenicity of cord blood cells strongly reduces the risk of relapse and GvHD after transplantation (110). Considering the advantages of using cord blood, Glycostem Therapeutics, a clinical stage biotech company, which in the last decade has developed a flexible platform technology to expand and differentiate NK cells from CD34+ cells (111), upgraded this into a large scale GMP UCB-NK platform for clinical implementation (oNKord ® ) (41). UCB-NK cells were infused at up to 30 × 10 6 cells/kg/bodyweight in elderly AML patients, resulting in excellent safety and initial efficacy in a Phase I trial. Infused oNKord ® cells showed active migration to the marrow and further matured in the absence of any exogenous cytokine injections. This confirms previous findings from a preclinical model, showing migration to the bone marrow and upregulation of KIRs and CD16a in vivo as well as antileukemic activity (112). oNKord ® is well characterized and was found to have a similar functionality and gene expression profile as PBNK cells (113). Furthermore, oNKord ® is highly cytotoxic against solid tumor targets such as cervical cancer cells, in which killing was independent of HLA expression levels, tumor histology and HPV types (114), or colorectal cancer cells, in which killing was independent of tumor EGFR levels, and RAS and RAF mutations (115), thus paving the way for oNKord ® as immunotherapy for advanced solid tumors.

CYTOKiNeS TO eNHANCe NK CeLL FUNCTiONS
To improve the antitumor activity of autologous NK cells, systemic administration of clinical grade recombinant IL-2 (rIL-2) and single chain IL-15 (scIL-15) has been used in high doses and this has resulted in severe grade 3/4 toxicities (116)(117)(118). Since then, their safety and efficacy have been tested in low doses following NK cell-adoptive transfer in cancer patients (50,63,119). However, IL-2 resulted in expansion and mobilization of inhibitory T-regs, severely limiting NK cell cytotoxicity (120). This shifted the focus toward the use of IL-15 for clinical trials involving NK cells. Currently more potent and advanced heterodimeric IL-15, which has a longer shelf life than scIL-15, is being tested in several studies (121). IL-15 is known to be more effective in membrane-bound form (i.e., bound to its receptor), engaging target immune cells in a cell contact dependent manner. Campana and his team (from Nkarta Therapeutics) addressed this by stably transducing the membrane bound IL-15 (mbIL-15) gene into proliferating PBNK cells, which were stimulated with K562-mb15-41BBL. mbIL-15 resulted in increased survival, proliferation, and enhanced cytotoxic functions of NK cells (122). Further, Cyto-Sen Therapeutics compared mbIL-15 to K562based aAPCs with mbIL-21. From their findings, it was evident that mbIL-21 NK cells have a significantly higher expansion and proliferation ability compared to mbIL-15 NK cells (123). Cyto-Sen also developed plasma membrane particles (PM21) engineered from K562-mb21-41BBL cells and found that these PM21 particles stimulated efficient NK cell expansion in AML patient's PBMC samples (124).
Compared to IL-2, the use of IL-15 minimizes capillary leak syndromes and has less side effects overall, thus providing a strong rationale to use IL-15 instead of IL-2. However, the use of mammalian recombinant IL-15 in the clinic has been limited due to its short half-life and decreased functional activity in vivo. Altor BioSciences corporation came up with a unique design to overcome these limitations. It developed an IL-15 super agonist known as ALT-803. It consists of a human IL-15 mutant N72D variant, which is stably complexed with a soluble human IL-15Rα sushi-Fc dimer protein. Enhanced biological activity of ALT-803 was reported in several preclinical studies showing durable antitumor activity in various solid and hematological malignancies (125)(126)(127)(128). Furthermore, ALT-803 facilitated expansion of effector and migratory NK cell subsets and significantly decreased the metastatic activity of tumor cells in a murine colon cancer pulmonary metastasis model (129). ALT-803 stimulated primary human NK cells to exhibit increased degranulation, IFNγ production, and ADCC when exposed to B cell lymphoma cell lines coated with IgG1 therapeutic anti-CD20 mAbs (130). Several clinical trials are currently ongoing with ALT-803 as monotherapy in patients with advanced solid tumors, hematological malignancies, and AIDS as summarized in Table 3.

PRiMiNG NK CeLLS TO eNHANCe TUMOR KiLLiNG
Mark Lowdell and his team proposed that for a NK cell to be able to kill tumor cells, it requires a priming and triggering signal. NK cells failing to kill tumor cells, though they are exposed to the triggering signal, remain inactive due to the absence of a priming ligand. To address this, Fortress Biotech (previously known as Coronado Biosciences) developed a technology to increase NK cell tumor killing using cell lysates from the leukemia cell line CTV-1, known as CNDO-109, to prime NK cells. A Phase I/II clinical trial of activated PBNK cells from haploidentical donors coincubated with CNDO-109, infused at doses of up to 3 × 10 6 kg/recipient/body weight was tolerable without any adverse reactions. Out of seven evaluable patients, four remained disease relapse free for more than 1 year (131).
Another NK cell-activating product is ENKASTIM-ev, developed by Multimmune GmbH, which mimics the functions of heat shock protein 70 (Hsp70). ENKASTIM-ev resulted in NK specific activation and actively targeted Hsp70 expressing tumors. Safety of Hsp70-activated autologous NK cells has been documented in a Phase I study in patients with metastatic colorectal and nonsmall cell lung cancer (132).

eNHANCiNG NK CeLL HOMiNG FUNCTiONS
Gamida-cell developed a feeder cell-free NK cell culture and expansion system containing nicotinamide (NAM) to generate NK cells from PBMC apheresis products. Nicotinamide, a derivative of vitamin B3, serves as a potent inhibitor of NAD dependent enzymes. Results from in vivo studies in mice showed that PBNK cells expanded with NAM in feeder free cultures exhibited increased homing potential toward lymphoid organs, with a significant increase in the expression of CD62L (L-selectin) compared to cultures without NAM (133).

TUMOR DiSRUPTive TeCHNOLOGY AiDiNG NK TUMOR ReCOGNiTiON
NOXXON Pharma target chemokine receptor CXCL12, with the aim of increasing the sensitivity of tumor cells to drugs and immune cells. Their product NOX-A12 functions as a CXCL12 inhibitor and enabled the release of CXCL12 from the surface of tumor stromal cells and blocked its interaction with cell surface receptors CXCR4 and CXCR7. This mechanism facilitated the mobilization of CXCR4-expressing tumor cells from their tissue niches to areas, where they become easily accessible by NK cells or T cells (134,135). Using tumor spheroids, increased mobilization of T and NK cells toward tumor cells in the tumor microenvironment was demonstrated. NOX-A12 also enhanced NK killing of obinutuzumab-coated Raji cells in vitro, mediated by ADCC (136).

CONCLUSiON
From this literature review, we conclude that adoptive transfer of allogeneic NK cells in a non-transplant setting is safe and shows early signs of clinical efficacy against hematological and certain solid tumors. Current data are mostly based on Phase I clinical trials, and hence it is still too early to get an overall picture of NK cell alloreactivity in different kinds of cancer. Most of the clinical studies conducted so far have used primary NK cells but with limited efficacy, pointing to the need to improve the functionality of these NK cells after their transfer to patients. The growing opportunities to augment NK cell functions have attracted several biotech companies to invest in NK cell research, spearheading NK therapy development with different innovative approaches. This review also stresses the need for combining adoptive transfer of allogeneic NK cells with NK function-augmenting products to achieve a maximum anti-tumor effect. As NK cells are safe to infuse, the use of CAR-NK cells may be instrumental in providing a much safer but still very effective platform, to bring CAR-based therapies to broader clinical applications. It may also facilitate effective tumor targeting of NK cells. oNKord ® and iPSC-derived NK cells could serve as alternative allogeneic platforms to develop CAR-NK products, besides NK cell lines. In a solid tumor setting, NK cells are challenged by several factors that affect their homing and penetration into the tumor tissues. Moreover, they should achieve and maintain an activated effector state, even in the face of immune suppressive conditions, that are prevalent in patients with cancer. To overcome these bottlenecks in NK therapy of solid tumors, a plethora of creative solutions are being pursued by numerous research labs as well as by biotech companies in clinical or close to clinical phase. Strategies to enhance NK cell functions from leading NK cell products are summarized in Figure 2. With all these exciting developments, NK cells are set to make a considerable impact on the future treatment of patients with hematological as well as with solid tumors.

AUTHOR CONTRiBUTiONS
JV, NK, HVV, TG, and JS wrote the paper. HVV, TG, HV, and JS reviewed the paper.