Natural killer cells

The success of current immunotherapies is usually attributed to the improvement of T-cell cytotoxicity against cancer cells. However, natural killer cells (NK cells) are also able to detect and eliminate tumour cells. Natural killer cells develop in the bone marrow and represent up to 15 % of all lymphocytes (Abel et al., 2018). They are characterised as CD3 negative and CD56 positive cells and can be divided into two major NK cell subsets differing in their CD16 and CD56 expression as well as their resting immunological potential (Freud et al., 2017). For example, CD16⁺⁺ CD56^dim NK cells show increased cytotoxicity in comparison to CD16^± CD56^bright NK cells (Arnon et al., 2006). However, these CD56^bright NK cells have important immune regulatory functions by increased cytokine production of e.g. Interferon gamma (IFN-γ), tumour necrosis factor beta (TNF-β) as well as granulocyte-macrophage colony-stimulating factor (GM-CSF), among others (Cooper et al., 2001). After stimulation with interleukin (IL)-2, both NK cell subsets increase their cytotoxic potential dramatically (Poli et al., 2009). In contrast to CTLs, NK cells are part of the innate immune system and are cytotoxic without previous antigen-specific stimulation (Farag and Caligiuri, 2006). Furthermore, NK cells complement the immunosurveillance of T-cells, which are dependent on the engagement of TCR and MHC-I antigens (Freud et al., 2017). MHC-I molecules are often downregulated in tumour cells, so these cells can evade T-cell recognition. However, tumour cells with particularly low MHC-I expression can be detected and eliminated by NK cells (Karre, 2002).

1.4.1 NK cell receptors

NK cells recognise tumour cells by their large repertoire of germ-line encoded receptors, which have different structural and functional properties. Therefore, NK cell receptors can be divided into killer cell immunoglobulin-like receptors (KIR) and killer cell lectin-like receptors (KLR) (Murphy, 2012). Besides their structural properties, they can be further classified into activating and inhibitory receptors.

Inhibitory receptors

Inhibitory receptors have an immunoreceptor tyrosine-based inhibition motif (ITIM), a long cytoplasmic domain with a conserved terminal amino acid sequence (V/I/LxYxxL/V). This ITIM contains tyrosine residues that can be phosphorylated by proto-oncogene tyrosine-protein kinase Src thus initiating a signalling cascade. After receptor-ligand engagement, the signal transduction results in the inhibition of NK cell activity, thereby preventing the elimination of the target cell (Cerwenka and Lanier, 2001; Farag and Caligiuri, 2006; Yokoyama, 2005). One of the best investigated inhibitory ligands are the major histocompatibility complex class I (MHC class I) molecules. As depicted in Figure 3, classical MHC-I molecules (HLA-A, B, C) are detected by KIR2DL (recognizes HLA-C alleles) and KIR3DL (recognizes HLA A, B alleles). Beside these KIR receptors, the lectin like NKG2A/CD94 heterodimer recognises HLA-E, a non-classical MHC-I molecule (Sivori et al., 2019). Although MHC-I molecules are the strongest inhibitory ligands, there are reports that stimulation of non-MHC-I dependent receptor can also dampen NK cell activation (Li et al., 2009; McNerney et al., 2005)

Figure 3. Inhibitory NK cell receptors and their corresponding tumour ligands. 2DL/3DL indicate the number and type of immunoglobulin domains and, L= long cytoplasmic domain; NKG2A, B, are splice variants. ITIM = immunoreceptor tyrosine-based inhibition motif. Figure was adopted from (Cappello, 2015)

KIR-2DL

KIR-3DL

NKG2A,B CD94

HLA A,B,C HLA-E

Tumour cell

NK cell ITIM

Activating receptors

In contrast to the inhibitory receptors, activating receptors need an adapter molecule that carries the conserved terminal sequence YXX[L/I]6-9YXX[L/I]. This sequence is named immunoreceptor tyrosine-based activating motif (ITAM) and it initiates a phosphorylation cascade resulting in the activation of NK cells (Murphy, 2012). With the help of the adapter molecule DAP12, the KIR receptors KIR2DS and KIR3DS recognize HLA-C and other unknown ligands (Cerwenka and Lanier, 2001). The same adapter molecule DAP12 is used by the heterodimer CD94/NKG2C,E that detects HLA-E. These activating receptors compete with their inhibitory counterparts, but with a lower ligand affinity (Bryceson et al., 2006).

The most important example of the activating KLR receptor is NKG2D. This receptor, unlike the other lectin like receptors, is a homodimer that is associated with the adaptor molecule DAP 10. This adaptor molecule carries a phosphatidylinositol-3 kinase (PI3K) binding motif YxxM instead of an ITAM (Bryceson et al., 2006; Kumar, 2018). MIC (MHC class I chain-related genes) A and B as well as UL16-binding proteins 1-6 (ULBPs) are ligands for NKG2D (Casado et al., 2009; Vivier et al., 2012). Essential members of the activating KIR receptors are the three natural cytotoxicity receptors (NCRs): NKp46, NKp30 and NKp44. The latter is de novo expressed after IL-2 stimulation, whereas NKp46 and NKp30 are constitutively expressed on NK cells (Arnon et al., 2006). NCRs also require an adapter molecule for signal transduction. In contrast to NKp44, which associates with the adaptor molecule DAP12, NKp46 and NKp44 use homodimers of CD3 ζ or heterodimers of CD3 ζ -FcεRγ heterodimers to initiate the phosphorylation cascade (Arnon et al., 2006). Although the ligands for NCRs are still not fully understood, the identified ligands are often only expressed on tumour cells or virus-infected cells. All natural cytotoxicity receptors recognize different viral proteins; e.g. hemagglutinins or hemagglutinin neuraminidases (Kruse et al., 2014) as well as heparan sulphate (HS) sequences, which are also shown to be upregulated on tumour cells (Barrow et al., 2019).

Beside heparan sulphates (HS), the complement factor P (CFP) is a known ligand for the NCR NKp46 (Narni-Mancinelli et al., 2017). NKp30a, b is known to bind BCL2-associated athanogene 6 (BAG-6) (Binici et al., 2013) as well as B7-H6, which is only expressed on tumour cells (Bjornsen et al., 2019). Identified activating ligands for NKp44 (NKp44L) are an unusual isoform of MLL5 (mixed lineage leukemia-5), often referred as NKp44L (Rajagopalan and Long,

Figure 4. Activating NK cell receptors and their corresponding tumour ligands. 2DS/3DS indicate the number and type of immunoglobulin domain; S= short cytoplasmic domain. Adaptor proteins and binding motifs are also shown in colour. Figure was adopted from (Cappello, 2015).

Several other activating receptors were also shown to play an important role in tumour recognition. The DNAX Accessory Molecule-1 (DNAM-1), recognizes poliovirus receptor (CD155) and nectin adhesion molecule (CD112) (Chester et al., 2015; Long et al., 2013). The DNAM-1-mediated signalling cascade requires the phosphorylation of conserved tyrosine as well as asparagine residues (Kumar, 2018). The receptor CD244, also known as 2B4, leads to NK cell activation by the engagement to CD48 and the subsequent phosphorylation of Tyr-based motif S/TxYXXL/I, referred to as immunoreceptor Tyr-based switch motifs (ITSM), in their cytosolic tails (Chester et al., 2015; Kumar, 2018).

In addition, CD16 plays a special role among the activating receptors, as it can to induce antibody-dependent cellular cytotoxicity (ADCC). CD16 is an Fεcγ receptor, which can bind antibodies, such as IgG. Therefore, NK cells are able to recognize antibody-coated target cells and act as mediators between innate and adaptive immunity.

1.4.2 NK cell recognition of tumour cells

The ‘missing self’ hypothesis of Klas Kärre was the first attempt to describe the recognition of tumour cells by NK cells. This hypothesis is based on the observation that tumour cells expressing low or no MHC class I molecules are more susceptible to NK cells than healthy (MHC-I positive) cells (Kärre, 1985). Nevertheless, the lack of inhibitory signals alone was shown to be insufficient to trigger NK cell activation and initiate tumour cell lysis (Arnon et al., 2006;

Lanier, 2005). In fact, the balance of signals derived from all activating and inhibitory receptors determines the final NK cell activation status (summarised in Figure 5) (Dustin and Long, 2010).

Although, the weighting of single receptors is not known, the overall signals derived from the

DAP12 ITAM

inhibitor receptors tend to be stronger and overrule the signals from the activating receptors (Long et al., 2013). However, the integration of several activating stimuli can overcome the inhibitory signals, thus allowing the initiation of a cytotoxic signal cascade in the target cell (Cerwenka and Lanier, 2001).

Figure 5. Principles of NK cell-mediated tumour recognition. NK cells have a large repertoire of activating and inhibitory receptors, which allow the recognition of tumour cells. The integration of all arriving stimuli determine the activation status of the NK cell. The engagement of many inhibitory ligands prevents an activation, whereas increasing activating stimuli can overcome NK cell inhibition, leading to the elimination of the target cell. Figure was adopted from (Cappello, 2015).