A major role of natural killer cells is distinguishing between “self” and “non–self”. This is accomplished through their ability to recognize MHC class I molecules and eliminate cells that have lost MHC I expression, a common feature of cancers and virus–infected cells. The interaction between NK receptors and MHC I on another cell is usually inhibitory and when MHC I is absent from target cells inhibition is relieved, leading to killing by NK cells (in a context of activating ligand expression). A subset of NK cells in wild–type mice does not express any MHC I specific inhibitory receptors and NK cells in MHC I–deficient (e.g. β2–microglobulin knockout) animals are never exposed to inhibition through MHC I. In both cases the NK cells do not attack MHC I deficient cells that are otherwise normal and exhibit a hyporesponsive phenotype when stimulated through their activating receptors, as evidenced by lack of inflammatory cytokine production. Work presented in this dissertation addresses the relationship of hyporesponsiveness and self–tolerance and demonstrates, for the first time, that the two phenomena are distinguishable. Additionally, the plasticity of NK cell education, the roles of hematopoietic vs. non–hematopoietic cells in this process, and finally the effects of inflammation on NK cell tolerance are also investigated.

In order to investigate whether responsiveness of NK cells is set once during their development or can be changed in mature NK cells based on their environment we employed an adoptive transfer model. We transferred splenocytes containing mature NK cells from WT mice into β2m–deficient mice, or vice–versa. 8 to 10 days after transfer we assessed the responsiveness of the transferred NK cells to activating receptor crosslinking, as well as the capacity of the cells to reject grafts of β2m^−/−spleen cells. We found that upon transfer into β2m knockout hosts WT NK cells reset their responsiveness downward and lost the capacity to reject MHC I — negative grafts. Conversely, when β2m–deficient NK cells were transferred to MHC I — expressing animals their responsiveness was reset upwards. Interestingly — despite increased responsiveness — these NK cells did not acquire the ability to reject β2m — deficient grafts.

To address the roles or hematopoietic vs. non–hematopoietic cells in NK cell education we generated fetal liver chimeras by reconstituting WT or β2m^−/−hosts with WT, β2m^−/−, or a 1:1 mixture of WT and β2m^−/−fetal liver cells. We found that NK cells that developed in an MHC I–expressing host acquired normal responsiveness whereas those that developed in an MHC I — negative host were hyporesponsive. Interestingly, however, NK cells that developed in the presence of β2m — deficient cells, of either non–hematopoietic or hematopoietic origin, acquired tolerance to β2m^−/−grafts. Therefore, our data show that responsiveness of natural killer cells is mainly set by non–hematopoietic cells, while tolerance can be imposed either by hematopoietic or non–hematopoietic cells. Thus our data show for the first time that responsiveness of NK cells can be (partially) dissociated from tolerance of NK cells. These findings, taken together, lead to the proposal that two mechanisms operate to ensure NK cell self–tolerance, which are imposed by different cell types in vivo.

Finally, we investigated whether inflammation can affect the ability of NK cells to respond to “missing self”. We infected mixed fetal liver chimeras with mouse cytomegalovirus and monitored the maintenance of β2m^−/−donor cells. We found that β2m^−/−donor cells were selectively rejected when the infected host was MHC I — sufficient, whereas no rejection was observed when the host was MHC I–deficient. These data indicate that different mechanisms of self tolerance are differentially regulated by inflammation, and that the form associated with hyporesponsiveness is more stable in the face of inflammation associated with infections.

Altogether, the data presented in this dissertation demonstrate that different interacting cell types play different roles in the process of NK cell education, with non–hematopoietic cells being especially critical for regulating the sensitivity of activating NK receptors. The data also show, however, that under steady state conditions (no infection) responsiveness of mature NK cells can be re–set upward or downward depending on corresponding changes in their MHC environment. This dissertation provides the first evidence that tolerance to MHC I — deficient cells can be decoupled from hyporesponsiveness. Finally, we show that infection can break tolerance of NK cells towards MHC I — negative grafts only a setting where it is accompanied by high responsiveness, whereas tolerance that is accompanied by low responsiveness is more stable and cannon be broken as a result of infection. These findings represent a substantial revision of current notions of NK cell self tolerance and hyporesponsiveness.