Current strategies exploiting NK‐cell therapy to treat haematologic malignancies

Natural killer (NK) cells recognize targets that have been changed via malignant transformation or infection. Previously, NK cells were thought to be short‐lived, but we now know that NK cells can be long‐lived and remember past exposures in response to CMV. NK cells use a plethora of activating and inhibitory receptors to recognize these changes and attack targets, but tumour cells often evade NK cells. Therefore, major efforts are being made to hone in on NK cell antitumour properties in immunotherapy. In the clinical setting, haploidentical NK cells can be adoptively transferred to help treat cancer. To expand NK cells in vivo and enhance tumour targeting, IL‐15 is being tested in combination with a glycogen synthase kinase (GSK) 3 inhibitor (CHIR99021), an inhibitor that has been shown to expand mature, highly functional NK cells capable of killing multiple tumour targets. One major limitation to NK cell therapy is lack of specificity. To address this concern, bispecific or trispecific engagers that target NK cells to the tumour and an ADAM17 inhibitor that prevents CD16 shedding after NK cell activation are being tested. Additionally, monoclonal antibodies are being designed to redirect the inhibitory signals that limit NK cell functionality. Further understanding of the biology of NK cells will inform strategies to exploit NK cells for therapeutic purposes.


| INTRODUC TI ON
Natural killer (NK) cells are CD56 + , CD3 − large granular lymphocytes that are a vital component of the innate immune system in killing cancer cells. NK cells were first identified in 1964 when lethally irradiated mice rejected bone marrow allografts without prior sensitization (Cudkowicz & Stimplfing, 1964). Subsequently, NK cells were described in the 1970s as major histocompatibility complex (MHC) unrestricted killers due to their ability to kill tumour targets without prior sensitization to antigens (Herberman, Nunn, Holden, & Lavrin, 1975). Cells may lose MHC class-I molecules under malignant or viral conditions, and this "loss of self" leads to an increased susceptibility to NK cell killing. After the description of the "missing self" hypothesis by Kárre, Ljunggren, Piontek, and Kiessling (1986) to explain this phenomenon, several families of NK cells receptors were identified that recognize MHC and regulate NK-cell activity. Because of the ability of NK cells to lyse cells without previous exposure to tumour antigens, NK cells are of great interest as therapeutic targets to treat cancer and improve the benefits of hematopoietic cell transplantation (HCT) and cancer therapy. Therefore, great strides have been made to manipulate NK cell antitumour properties in the clinic. In this review, we discuss the current status of NK cells in immunotherapy (Table 1). We also consider future directions to stimulate endogenous NK cells using bispecific, trispecific and tetraspecific killer engagers, anti-KIR antibodies, and the expansion of NK cells to enhance NK cell activity and specificity in treating cancers.

| NK CELL B I OLOGY
In humans, NK cells are derived from CD34 + progenitor cells in the bone marrow, and commitment to the NK cell lineages is dependent upon loss of CD34 and gain of the cytotoxicity receptors NKp44 and NKp46 (Freud et al., 2006). Further differentiation is marked by the expression of the surface adhesion marker CD56 and lack of the TCR/CD3 complex. NK cells represent 5%-15% of circulating lymphocytes in humans and can be categorized into two distinct phenotypic and functional subsets based on their surface density of CD56. Approximately 2%-10% of NK cells have high surface density expression of CD56 and are designated as CD56 bright . This population produces high levels of cytokines, proliferates robustly in response to IL-2, lacks expression of CD16 and killer-immunoglobulin-like receptors (KIRs) and is poor mediators of cytotoxicity (Jacobs et al., 2001). This population is found predominantly in secondary lymphoid tissues (Fehniger et al., 2003). In contrast, 90% of NK cells that are found in peripheral blood and are designated as CD56 dim .
This subset demonstrates limited proliferation in response to IL-2, expresses CD16 and KIR, and is potently cytotoxic.
NK cells are distinguished from other innate lymphoid cells by their dependence on IL-15 for development and their intrinsic cytotoxic ability. This cytotoxic activity is mediated by a variety of mechanisms. NK cells express a wide variety of germline receptors, such as NKG2D, which bind to stress-induced ligands found on tumour cells (Smyth et al., 2005). Once bound to tumour cells, NK cells degranulate and release granzyme B and perforin to induce apoptosis in the tumour cell (Voskoboinik, Smyth, & Trapani, 2006). NK-cell degranulation can also occur through antigen-dependent cellular cytotoxicity (ADCC) in which the Fc portion of tumour antibodies binds to the low-affin- Las and TRAIL-receptor expressed on tumour cells (Screpanti, Wallin, Ljunggren, & Grandien, 2001). In addition, NK cells secrete proinflammatory cytokines and chemokines that can have antitumour activity.
This antitumour activity is regulated by receptors found on the NK cells.

| NK-CELL RECEP TOR S
Therapeutic strategies using allogeneic NK cells in treating cancer are based on our understanding of the signalling pathways that regulate the antitumour activity of NK cells.
NK-cell activity is mediated through a balance of activating and inhibitory receptors. The combination of receptors that an individual has ultimately determines the cytotoxic activity of the NK cell.
Therefore, certain human tumours are more amenable to NK cell activity than others. Inhibitory receptors prevent NK cells from killing "self" expressing normal tissue, protecting against autoimmune diseases. Major histocompatibility class-I (MHC-1) molecules provide an inhibitory signal to the NK cell when ligated to an inhibitory receptor, preventing degranulation and cytokine production. These inhibitory receptors include the following: (a) killer-immunoglobulinlike receptors (KIRs); (b) NKG2A/CD94; and (c) leucocyte immunoglobulin-like receptors (LILRs; Lanier, 2008). In addition to inhibitory receptors, NK cells express germline encoded activating receptors, including CD94/NKG2C, the SLAM family receptors and the low-affinity Fc receptor CD16 which mediates antibody-dependent cellular cytotoxicity (ADCC). Great reviews have been published on these activating receptors, and they will not be discussed further (Lanier, 2015;Raulet, Gasser, Gowen, Deng, & Jung, 2013).
Of these receptors, most studies have focused on ways to manipulate NK cells to lessen interactions between inhibitory KIRs and MHC-I ligands on target cells. Inhibitory KIRs are transmembrane molecules that are found on chromosome 19 and interact with HLA-A, HLA-B and HLA-C allotypes. KIR genes are highly polymorphic, and these polymorphisms determine their binding affinities to MHC-I molecules (Hilton & Parham, 2017). In general, KIR genes can be divided into two broad haplotypes, KIR-A and KIR-B. KIR-A contains only one activating receptor, whereas KIR-B contains two or more (Pyo et al., 2010). The effectiveness of NK cell therapy is based on the variations in ligands among individuals.
KIRs became a major focus for NK-cell immunotherapy when the Perugia group showed that KIR ligand mismatch between donors and patients was associated with improved outcomes in myeloid leukaemia after T-cell deplete haploidentical transplantation (Ruggeri et al., 2002).
Since then, KIR ligand incompatibility in hematopoietic cell transplantation (HCT) has been highly studied and much progress has been made in understanding NK-cell function and the immunogenetics of NK-cell receptors. The presence of activating KIR2DS1 has been shown to prevent relapse in AML patients (Venstrom et al., 2012). Additionally, studies have emerged showing that donors with KIR-B haplotypes can protect against relapse in both AML and non-Hodgkin lymphoma in unrelated donor transplants (Bachanova et al., 2016;Cooley et al., 2010Cooley et al., , 2014 ). In addition to relapse protection, recent studies have shown that donor-recipient matching for KIR genotypes can protect against chronic GVHD and that missing inhibitory KIR ligands reduced relapse after unrelated donor transplantation (Faridi et al., 2016). Selection of donors with favourable KIR-B haplotypes is important in HCT, and formal prospective clinical testing is in progress. Strategies for donor selection based on allele level KIR typing are also being contemplated as higher resolution typing methods become available.  Anfossi et al., 2006;Cooley et al., 2007). Therefore, it has been proposed that NK cells can be tuned by the strength of their class I recognizing inhibitory signals.

| DISCOVERY OF VIR AL-INDUCED ADAP TIVE NK CELL S
NK cells were traditionally seen as short-lived lymphocytes that lack antigen specificity and are considered part of the innate immune system. Over the past several years, this view has been challenged as studies have provided evidence that in response to stress, NK cells do possess some adaptive immune traits as a result of their ability to be educated. Recently, the novel concept of NK-cell memory emerged with the identification of NK cells in mice that expand in response to cytomegalovirus (CMV) infection (Sun, Beilke, & Lanier, 2009). In humans, we and others have discovered that NKG2C + NK cells specifically expand in response to human CMV. These NKG2C + NK cells express an inhibitory receptor for self-HLA and progressively acquire CD57, a marker of maturation. Foley, Cooley, Verneris, Pitt, et al., 2012;Lopez-Vergès et al., 2011). Whether these NK cells possess all of the attributes ascribed to classical memory T and B cells or whether they are "memory-like" is still a matter of debate and terminology.
Regardless, the NKG2C + CD57 + NK cells are referred to as "adaptive" We believe that these adaptive NK cells are induced by CMV specifically and are not induced by other pathogens. Although others have suggested that NKG2C + NK cells are associated with the hantavirus in Sweden and the chikungunya virus in Africa, all of the patients participating in these studies were also co-exposed to CMV, suggesting that the NKG2C + NK cells were caused by CMV infection rather than infection with other viruses (Braun et al., 2014;Petitdemange et al., 2016).

| ADOP TIVE NK-CELL THER APY
The first studies in NK cell adoptive transfer used autologous NK

| US E OF C Y TOK INE S TO ENHAN CE AND E XPAND NK CELL S IN ADOP TIVE THER APY
Overcoming the immunosuppressive tumour environment is an attractive therapeutic option for improving NK cell antitumour activity.
Thus, approaches to enhance NK-cell antitumour functionality before adoptive transfer are being explored. The use of IL-2 to expand NK cells showed no clinical advantage as Tregs express a high affinity IL-2 receptor (CD25) which competes for IL-2 and dampens NKcell proliferation (Ghiringhelli et al., 2005). We have since overcome this challenge with the use of IL-2 diphtheria toxic fusions (IL2DT) to deplete Tregs (Bachanova et al., 2014). IL2DT is a recombinant fusion protein that includes sequences for diphtheria toxin followed by truncated amino acid sequences for IL-2 that will deplete all cells expressing IL-2 receptors, including Tregs. After patients received the and the clearance of lung lesions in patients (Conlon et al., 2015).
Other studies have shown that preactivation with IL-12, IL-18 and IL-15 differentiates NK cells in adaptive NK cells, suggesting promise as an immunotherapy strategy (Romee et al., 2016). It is believed that endogenous IL-15 in serum binds to IL-15Rα to form a natural complex. This natural complex interacts with IL-2Rβ on NK cells and CD8 + T cells through a process called IL-15 trans-presentation. This process is thought to overcome the Treg stimulatory effects of IL-2 (Cooper et al., 2002;Oh, Perera, Burke, Waldmann, & Berzofsky, 2004;Wong, Jeng, & Rhode, 2013). The recently designed IL-15 superagonist/IL-15Rα-Sushi-Fc fusion complex (ALT 803) exhibits greater activity than that of native IL-15 in various malignancies (Felices et al., 2017;Kim et al., 2016). The design included a mutant IL-15, the addition of a sushi domain to inhibit complement activation, increased avidity of the molecule to IL-2R β on NK cells and increased half-life and stability by inclusion of the Fc domain (Xu et al., 2013). An additional advantage to the IL-15 complexes, like ALT-803, is that they stabilize IL-15 so it can be given in fewer doses and allow for better homing to lymphoid tissue. The first-in-human studies of ALT-803 show that ALT-803 is well tolerated by patients.
Additionally, ALT-803 was shown to promote NK and CD8 + T-cell expansion in vivo without stimulating Treg cells. IL-15 complexes hold much promise in cancer immunotherapy, and the future will include combining IL-15 with other methods to activate NK cells.

| S TR ATEG IE S TO E XPAND NK CELL S FOR ADOP TIVE TR ANS FER
In addition to cytokines, other avenues are being explored to expand NK cells. One popular approach is to expand NK cells from PBMCs using feeder cells, such as K562 cells modified with membrane-bound IL-15 or IL-21. Fujisaki et al. (2009) showed that the leukaemia cell line K562 was able to be modified with 41BB ligand These results represent a new way to expand NK cells that can be used to treat various cancers.
Although the use of haploidentical NK cells has shown promise form a variety of cancers, limitations still exist. One of the limitations to using NK cells isolated from peripheral blood that are CD3 and CD19 depleted is that the cellular produce still contains a mixture of cells, with only about 30%-50% of the infused cells being NK cells (Koepsell et al., 2013). To produce a homogenous NK cell product, studies have used NK cell derived from induced pluripotent stem cells (iPSCs). These iPSC-NK cells are able to be expanded in vivo and have been shown to be effective against leukaemia and ovarian cancer in xenograft models (Hermanson et al., 2016;Woll, Martin, Miller, & Kaufman, 2005). Recently, a good-manufacturing practicecompatible iPSC source and industry-friendly protocol to produce NK cells from iPSCs has been developed. This protocol makes largescale quantities of "universal" NK cells that express no KIRs, making them unrestricted by HLA genotypes. This overcomes selection of NK cell donors for a particular patient (Zeng, Tang, Toh, & Wang, 2017) and offers a truly off-the-shelf advantage. These iPSCs show promise to be used NK-cell immunotherapy, but there is still much to be learned about their efficacy and viability in clinical trials.

| DE S I G N OF CD16 TARG E T AG ENTS TO ENHAN CE ADOP TIVE TR AN S FER THER APIE S
Targeted immunotherapies are currently a subject of great clinical potential. Recently, a great deal of interest has been placed on genetic manipulations geared towards maximizing NK-cell function, such as the generation of chimeric antigen receptor-expressing NK cells. Although these approaches show great potential for improving NK-cell adoptive therapy, they require a personalized approach that is expensive, time-consuming and difficult to apply on a large scale. Therefore, we have focused on two strategies to augment the ADCC in NK-cell adoptive transfers-improving maintenance of the CD16 receptor and using bispecific, trispecific and tetraspecific killer engagers to improve targeting to tumours (BiKEs, TRiKEs and TetraKEs, respectively; Figure 1).
One strategy to enhance NK cell function is to exploit the ability of NK cells to recognize antibody-coated targets through CD16 and kill tumour targets through ADCC. Recently, we discovered that NK-cell activation leads to a decrease in CD16 expression. Loss of CD16 expression is problematic because, without CD16, ADCC will not occur. This decrease in CD16 expression is due to disintegrin and metalloproteinase-17 (ADAM17) mediated clipping of the CD16 receptor, which leads to CD16 shedding (Gryzwacz, Kataria, & Verneris, 2007;Zhou, Gil-Krzewska, Peruzzi, & Borrego, 2013 (Gleason et al., 2012). We have recently shown that anti-CD16x33 BiKE activation overcomes inhibitory signalling via class-I HLA to potently kill primary cancer targets, as well as targeting CD33 + myeloid-derived suppressor cells (MDSC) in patients with myelodysplastic syndrome. (Gleason et al., 2014;Schmohl, Felices, et al., 2016;Schmohl, Gleason, Dougherty, Miller, & Vallera, 2016). This BiKE has important therapeutic potential due to its ability to target a drug-resistant cell population that is  Schmohl, Gleason, et al., 2016). As shown by these studies, the BiKE and TRiKE platforms are highly flexible, making it possible to target any tumour-specific antigen. One major advantage of the anti-CD16 in this platform is that BiKEs and TRiKEs bind Fc at high affinity compared with the low-affinity binding of Fc portions of whole antibodies. These are continuing to revolutionize NK celltargeted immunotherapy.

| C AR-NK CELL S IN C AN CER IMMUNOTHER APY
Recently, Chimeric antigen receptors (CARs) on NK cells have been used to direct NK-cell antitumour activity. In a recent study, cordblood-derived NK cells were transduced with a retroviral vector incorporating CAR-CD19, IL-15 and inducible caspase-9 suicide (iC9) gene (Liu et al., 2017). This CAR-NK cell incorporates CD19 to redirect NK-cell specificity, IL-15 so that they can sustain their growth, and iC9 so that the CAR-NK cell can be eliminated. This  (Romagné et al., 2009; Figure 1). Phase I clinical trials showed that this antibody is tolerated by AML and MM patients and that it led to enhanced cytotoxic activity (Benson et al., 2011(Benson et al., , 2015Vey et al., 2012). Despite the promise that IPH2101 held, Phase II trials failed to show any clinical efficacy in patients with smouldering multiple myeloma (Carsten et al., 2016;Korde et al., 2014). In a recent commentary, we discussed the potential that IPH2101 dampens NK-cell education though monocyte-drive trogocytosis, a process that would not have been identified when studying IPH2101 outside of the larger context of the immune system . Despite this drawback as a single agent, IPH2101 may be useful in combination with other agents in targeted immunotherapy. These findings suggest that KIR inhibition is context-dependent and that investigating a single immune component out of context may result in misleading information and faulty conclusions.
Similarly, studies blocking CD94:NKG2A, the inhibitory receptor for HLA-E is also being explored. The humanized anti-NKG2A antibody (IPH2201-monalizumab) converted NKG2A + NK cells in effector NK cells that had the ability to kill HLA-E + lymphoma cells and multiple myeloma cells (Ruggeri et al., 2016). The anti-NKG2A antibody is currently in Phase I/II clinical trials for a variety of tumour types, either alone or as combination therapy.

| CON CLUDING REMARK S
Over 60 years of study have led to a vastly increased understanding of NK-cell development and how their innate properties can be honed to improve cancer immunotherapy. The current approaches in NK cell-targeted immunotherapy interrupt NK-cell inhibition, further stimulate NK cells through cytokines or enhance targeting through CD16. Current studies are focused on ways to expand NK cells and identify how adaptive NK cells can be used in the clinic. While acknowledging the importance of these developments, much more work is needed to fully understand and utilize NK cells in cancer immunotherapy. Relapse remains a major problem after HCT. Strategies to exploit favourable donor immunogenetics, NK-cell expansion ex vivo from blood and the potential of adaptive NK cells into successful clinical applications require further study.
A major challenge will be finding ways to activate NK cells endogenously without NK-cell infusion or with the use of off-the-shelf NK-cell products in development. The future of NK-cell immunotherapy lies in using combination therapies. Combining expansion of CMV-induced adaptive NK cells with enhanced CD16 signalling and activation by IL-15/IL-15Ra complexes and creation of NK-cell antigen-specific BiKEs and TRiKEs and CAR-NK cells offer great promise for the success of NK-cell immunotherapies.