NK cells in cancer immunotherapy

NK cells play a crucial role in immune surveillance as their cytotoxic activity was shown to correlate with a generally lower cancer risk (Imai et al., 2000). Moreover, the number or type of tumour-infiltrating NK cells have been shown benefits in therapy response and overall patient survival in cutaneous melanoma (Cursons et al., 2019; Ladanyi, 2015) as well as in other cancers (Barry et al., 2018; Chiossone et al., 2018; Cursons et al., 2019). Nevertheless, the success of NK cells as a tool for immunotherapy against solid tumours belies expectations. One of the possible reasons for the past failure of these therapies may lie in the tumour microenvironment that could dampen NK cell activity (Paul et al., 2016; Terren et al., 2019). Accordingly, to maintain or even boost the NK cell activity in cancer patients, different therapeutic strategies have been developed.

Cytokine-based therapies have been used to increase NK cell cytotoxicity. To this end, IL-2 was the first cytokine with the desired properties that showed initial promising results in cellulo and in animal-model based studies (Grimm et al., 1982; Hoffman et al., 1989; Miller and Lanier, 2019;

Rosenberg et al., 1985). However, the systemic administration of IL-2 alone achieved only limited clinical response and had even toxic and life-threatening effects when applied at higher doses (Hoffman et al., 1989; Miller and Lanier, 2019). These unsatisfactory results by IL-2 supplementation can be, at least partially, explained by the unintended expansion and activation of immunosuppressive regulatory T cells (Tregs) (Miller and Lanier, 2019). Furthermore, the long-lasting administration of IL-2 induces the downregulation of the NK cell IL-2 receptors (Pillet et al., 2011). These difficulties can be circumvented by ex vivo activation as well as the use of other interleukins such as IL-15, IL-12, IL-18 or IL-21 (Nayyar et al., 2019; Zhang et al., 2020). IL-15 was shown to be the most promising therapeutic cytokine. IL-15 not only increased NK cell cytotoxicity more specifically (no cross activation of Tregs) but also enhanced NK cell expansion (Miller and Lanier, 2019). The administration of IL-15 achieved clinical success in haematopoietic cancers (Cooley et al., 2012; Nayyar et al., 2019) and showed also promising results in solid tumours including metastatic melanoma (Conlon et al., 2015). However, the short half-life of IL-15 and the observed toxicities at higher dosages are still limiting the clinical usage (Robinson and Schluns, 2017). The development of super-agonist complexes such as ALT-803 (IL-15N72D/IL-15Rα-FC) further improved the in vivo half-life time and the clinical efficacy (Felices et al., 2017).

Another approach to enhance NK cell cytotoxicity is to reduce the threshold of NK cell activation by lowering inhibitory stimulation. This can be achieved by blocking the engagement of KIR receptors with MHC class I molecules on the target cells. The use of allogeneic NK cells prevents the inhibition by self-MHC molecules and facilitates NK cell activation (Souza-Fonseca-Guimaraes et al., 2019). Unfortunately, this can also induce graft-versus-host effects, so that NK

cell donors with haploidentical HLA phenotypes are preferably used in adoptive cell transfer (ACT) (Souza-Fonseca-Guimaraes et al., 2019). Although, ACT has been successful in the treatment of haemopoietic cancers such as acute myeloid leukemia (AML) (Tanaka and Miller, 2020), so far it fails in the treatment of solid tumours (Paul and Lal, 2017). Also, the direct inhibition of KIR receptors with antibodies such as lirilumab was only effective in haematopoietic malignancies (Konjević et al., 2017; Miller and Lanier, 2019). Another important inhibitory NK cell receptor is CD94-NKG2A, that recognizes HLA-E, and can be blocked by the antibody monalizumab (Li and Sun, 2018). The administration of checkpoint inhibitors has been successful for T-cell activation. Moreover, tumour-infiltrating NK cells were also shown to express PD-1 (Liu et al., 2017b) and can thus benefit from checkpoint inhibitors (Sanseviero et al., 2019). Accordingly, the clinical successes of anti-PD-1 therapy might also be contributed by increased anti-tumour NK cell activity (Hsu et al., 2018; Pesce et al., 2019). Additional checkpoint inhibitors targeting different inhibitory NK cell receptors (TIGIT, IL1R8) require further investigation (Khan et al., 2020; Nayyar et al., 2019).

Another approach to boost anti-tumour NK cell activity is to strengthen the activating stimuli in order to overcome the inhibitory signals. Bi-specific antibodies, known as bi- specific killer engagers (BiKEs) promote the formation of an antigen-specific immunological synapse between NK cells and tumour cells and thereby enhance the NK cell-mediated killing of the antigen expressing target cell (Miller and Lanier, 2019; Shimasaki et al., 2020). One fragment of the BiKE binds to an activating NK cell receptor such as the CD16 receptor that induces ADCC (Felices et al., 2016). The part of the killer engager is directed against tumour associated antigens such as CD30 in Hodgkin lymphoma, CD33 in myelodysplastic syndromes, CD133 in colorectal cancer or human epidermal growth factor receptor 2 (HER2) in breast cancer (Li and Sun, 2018). To increase the NK cell cytotoxicity, tri-specific killer engagers (TriKEs) with additional linkage to IL-15 such as GTB-2550 (anti-CD16 x IL-15 x anti-CD33 TriKE) have recently entered their first clinical trials in haematopoietic malignancies (Miller and Lanier, 2019).

The use of NK cell lines (NKL, NK-92, KYHG-1, YT and NKG) derived from malignant NK cell leukemia or lymphoma has several advantages and opens new possibilities to modify effector cell cytotoxicity (Miller and Lanier, 2019). The most promising NK cell line for clinical application is NK-92. The IL-2 dependent NK-92 cell line is known to express several activating

Administration (FDA) and was already applied in clinical trials including patients with renal carcinoma and malignant melanoma with modest success but good tolerance (Arai et al., 2008).

NK cell lines can be genetically modified to display different receptor profiles. For example, parental NK-92 cells lack CD16 expression and are therefore not able to mediate ADCC. The artificial expression of high-affinity CD16 (haNK) increases their cytotoxicity (Klingemann et al., 2016), a modification that was shown to be very effective in combination with tumour epitope specific monoclonal antibodies in preclinical studies (Williams et al., 2018). Even more advanced are NK cell lines, which have been armed with engineered chimeric antigen receptors (CARs).

CARs targeted to specific tumour epitopes such as CD19, CD7 and CD33 in leukemia and lymphoma have successfully entered their first clinical trials (Kloess et al., 2019; Liu et al., 2020).

Also preclinical studies in solid tumours directed against HER2 or epidermal growth factor receptor (EGFR) showed promising results (Paul and Lal, 2017).

In addition to NK cell lines, NK cells differentiated from human pluripotent stem cells (hPSCs) could be used as ‘off-the-shelf’ immunotherapy (Wang et al., 2019). HPSCs include human embryonic stem cells, hematopoietic stem/progenitor cells from umbilical cord blood and induced pluripotent stem cells (iPSCs) (Hu et al., 2019). These stem cell derived NK cells can be produced on a large scale and genetically modified (Nianias and Themeli, 2019; Shimasaki et al., 2020; Zhu et al., 2018).

One important aspect that might explain the modest clinical success of NK cell therapy in solid tumours so far, in comparison to hematopoietic malignancies are the immunosuppressive effects of the tumour microenvironment (see following chapter). Therefore, increasing number of clinical investigations try to improve the conditions for effector cells in the tumour. For example, several drugs can prevent the shedding of MICA/MICB and promote NK cell cytotoxicity against tumours (Ferrari de Andrade et al., 2018). Alternatively, soluble MICA/MICB can be neutralized by antibodies such as IPH4301 (Li and Sun, 2018; Nayyar et al., 2019). Also the secretion of other soluble factors such as transforming growth factor β (TGF-β) or adenosine in the tumour microenvironment was shown to suppress NK cell proliferation and function (Leone and Emens, 2018; Viel et al., 2016). The TGF-β neutralizing antibody Fresolimumab (GC1008) has been used in a Phase I study for the treatment of advanced malignant melanoma and renal cell carcinoma (Morris et al., 2014). Furthermore, Galunisertib (LY2157299 monohydrate), an inhibitor of TGF-β receptor 1 (TGFbR1) has also been evaluated in clinical trials in solid tumours (Fujiwara et al., 2015; Yingling et al., 2018).