Stage III: Mature NK Cells, Education and Self-Tolerance

Within this stage, 2 types of mature NK cells emerged. NK cells that express at least one inhibitory NK cell receptor, which have been engaged by self-MCH-I, become educated or license, and NK cells, which expressed self-specific receptors but having not been engaged by self-MHC-I molecules. Both type of NK cells have a mature phenotype, but mature NK cells, which develop in absence of MHC-I are hyporesponsive to activation signals. The process called NK cell education generates NK cells that become tolerant to self and are able to detect MHC-I negative target cells. The site where this process occurs, remain unknown.

What makes NK cells relevant is their ability to distinguish target cells, which differ only in their MHC-I expression. NK cell activity is blocked when inhibitory receptors bind to self-MHC-I, but this inhibitory response is canceled in the absence, or down-regulated of self-MHC-I. The missing self-recognition concept, proposed by Kärre et al., (24) is the recognition and killing of MHC-I down-regulated cells by NK cells. The down-regulation of MHC-I is a way to avoid CD8⁺ T-cells recognition. Yet as a counter-response, the immune system uses the NK cells (96). Thus, NK cells need to be educated to kill MHC-I negative target cells. However, MHC-I is encoded on chromosome 6 and NK cells receptors are encoded on chromosome 19. So NK cells receptors and MHC are inherited independently (32). This means that in the NK cells maturation process; a sub-population of NK cells could arise, which do not express inhibitory receptors for the self-MHC-I, as these receptors are expressed in a stochastic way. So how do these NK cells avoid attacking their host? Until recently, the "at least one" rule assumed that each NK cells had at least one inhibitory KIR specific for a self-MHC (29). But as KIR receptors are clonally expressed and in a stochastic manner, and moreover, MHC shows great allelic

variation, this does not corroborate the previous rule. Later, studies showed that human or mice lacking MHC-I had a normal number of NK cells, but they were hyporesponsive, meaning with a low level of cytokine secretion and / or cytotoxic activity. While it had been presumed that this happened in specific patients (97), healthy individuals or mice also show hyporesponsive NK cells (98, 99). Moreover, to investigate if NK cells are educated or not, only functional tests can be performed. No specific markers exist, which will distinguish between these both NK cells populations (100). Several studies in humans and in mice show that KIR inhibitory receptors expressed by NK cells binding to self-MHC-I render these NK cells more cytotoxic and increase IFN-γ secretion (101, 102, 103, 104)). But recently, Orr et al. (105) demonstrated in a mouse study, that unlicensed NK cells are still able to kill target cells expressing m157. Several possible mechanisms for NK cell education, licensing or arming were proposed (100, 96).

5.3.1. Arming Model

In this model, NK cells become educated with the binding of their inhibitory KIR receptors with their self-MHC-I. The inhibitory receptor "activates" NK cell functionality. This infers that NK cells are hyporesponsive in their initial state. This means an inhibitory signal, which trigger a positive signal (96, 100).

Fig. 12. Illustration of the arming model. This model supposes an hyporesponsive NK cell, which will become responsive through the activating signal from its inhibitory receptor bound to its MHC-I ligand.

5.3.2. Disarming Model

In this model, we find the opposite situation of the arming model. No self-MHC-I specific inhibitory receptor is expressed at the NK cell surface, and it is supposed that all NK cells are initially responsive. Chronic stimulation by activating receptor renders this NK cell anergic. The situation is restored if a self-MHC-I specific inhibitory receptor is expressed. (96, 100).

Fig. 13. Illustration of the disarming model. It is only with the expression of an inhibitory receptor that an NK cell become responsive. In the presence of only an activating receptor, NK cell is hyporesponsive, meaning no reactivity despite its activating receptors.

5.3.3. Rheostat Model

In the case of the arming and disarming model, this implies an on-off signal.

Joncker et al. (106) proposed another model based on the amount of inhibitory and activating receptors, which are engaged with their ligands. They showed that depending on the amount of inhibitory signals; NK cells increase or decrease their response. If an inhibitory signal is absent due to a lack of inhibitory receptors, NK cells will be hyporesponsive or disarming. Yet, if both types of receptors are engaged, NK cells will be educated or arming. (96, 100).

Fig. 14. Illustration of the rheostat model. NK cells education will depend on the binding strength of its inhibitory receptors and its MHC-I ligand.

Finally, two pathways can explain NK cells tolerance. NK cells express inhibitory receptors, which bind to self-MHC-I and become educated or armed, or alternatively, NK cells lack the expression of inhibitory receptors, only activating receptors are engaged and NK cells are hyporesponsive.

In fact, the precise molecular and cellular mechanism, which is responsible for NK cells to become self-tolerant, is still not well understood. It is clear that the majority of NK cells become educated, but between 10 to 20% of human NK cells do not express any inhibitory receptor for self-MHC-I (97). If these NK cells play a specific role in the immune response remain unsolved.

Among the many activating and inhibitory receptors expressed by NK cells, a lot of them possess as ligand the MHC-I molecule, but some receptors, has as ligand a non-MHC-I ligand. Contrary to KIR receptors, which are mainly expressed by a sub-population of NK cells (CD56^dim), these non-MHC-I binding receptors are expressed by all NK cells. These receptors-ligands, such as NKR-P1D-CLR-B, SIGLEC7-sialic acid or gp49B1-αβintegrins are inhibitory receptors. Human 2B4 (CD244) is an activating receptor, whereas mouse 2B4 has inhibitory function. What renders these receptors interesting is that these non-MHC-I binding receptors could keep NK cells tolerant in a MHC-I deficient environment (107).