Natural killer cell education under conditions of perturbed sensing of MHC : studies in MHC-I mosaic and chimaeric mice

Abstract

Natural killer (NK) cells are innate lymphocytes that play a role in infections, tumours and transplantation. Their activating and inhibitory receptors sense cell surface ligands on potential target cells. Increased levels of stress induced activating ligands and decreased levels of inhibitory MHC class I (MHC-I) ligands cause activation of NK cells. NK cells are “educated” to tolerate self, and the “rheostat model” proposes that this is a reversible and quantitative process, where the degree of responsiveness acquired by each NK cell is proportional to the amount of perceived inhibitory MHC-I signals. Assuming that the NK cell integrates MHC-I input over multiple cell encounters, education may also depend on the ratio between cells with and without a critical MHC-I ligand, e.g. in chimaerism after haematopoietic transplantation. This thesis addressed how NK cell education is affected by perturbed sensing of MHC-I, and how the process is influenced quantitatively by the frequency of cells with a given MHC-I ligand. “Allele specific” as well as “general” missing self reactivity (targets lacking one or all host MHC-I molecules respectively), against normal and tumour target cells, was analysed.

The DL6 mouse strain presents a mosaic expression of an MHC-I transgene such that each cell either expresses it (Dd+) or not (Dd-). To test NK cell killing in vivo of two different targets in the presence of a control cell population in these mice, one aim was to develop an assay to measure rejection of three fluorochrome labelled targets in parallel. Using this assay, all mosaic DL6 mice were completely tolerant towards normal Dd-spleen cells (“allele specific” missing self), even if they had a relatively high proportion of Dd+ vs. Dd-cells. There were varying degrees of impaired reactivity in the mosaic mice upon challenge with MHC-I defective spleen cells (“general” missing self), with a moderate inverse correlation between the number of host Dd+ cells and the survival of these target cells. Tolerance towards Dd-RMA tumour cells representing “allele specific” missing self was not complete; partial rejection was observed in many DL6 mice. There was a weak inverse correlation between the number of Dd+ cells and the survival of these target cells; in particular, there was an efficient rejection in all mice with >83% Dd+ cells. The frequency of Dd+ cells did not influence rejection of MHC-I deficient RMA-S lymphoma cells; all DL6 mice rejected these cells efficiently. In chimaera studies where Dd+ bone marrow was used to reconstitute mosaic DL6 mice, there was again complete tolerance towards normal Dd-spleen cells, indicating that a fraction of ligand deficient nonhaematopoietic cells is sufficient to induce tolerance to normal cells of this phenotype. Again, the results differed when tumour cells were used; there was significant (but reduced) capacity for “allele specific” missing self rejection in these chimaeras. In the reverse setting, mixed bone marrow chimaeras were constructed in which different ratios of Dd+ and Dd-cells were transplanted to mice with normal Dd expression in the non-haematopoietic compartment. Significant but reduced capacity for rejection of Dd-RMA tumour cells was observed, despite apparent tolerance towards normal Dd-cells. Finally, when MHC-I sensing by NK cells in B6 mice was perturbed by treatment with antibodies against inhibitory Ly49I/C receptors, it led to impairment of “general” missing self reactivity towards MHC-I deficient spleen cells, despite increased killing of MHC-I expressing RMA tumour cells. Both changes occurred within 24 hours after inhibitory receptor blockade, and it was not possible to separate them in kinetic studies. As in the DL6 mice, efficient killing of MHC-I defective RMA-S cells and tolerance towards normal MHC-I expressing cells, were preserved.

The results show that tolerance develops robustly towards normal host cells lacking one MHC-I ligand, i.e. the Dd-cells are critical and dominant tutors in the education process even if they are in minority. A model is proposed in which the education sets responsiveness for each NK cell such that tolerance is ensured towards all normal cells in the host. This occurs via multiple cell encounters, and results in a reduction of responsiveness in certain NK cell subsets with inhibitory receptors for Dd molecules, correlating with the proportion of Dd- vs. Dd+ cells in the environment. This Dd+ cell frequency dependent, quantitative educational effect is revealed when using MHC-I deficient spleen cells or Dd-tumour cells as targets. A similar interpretation is offered for the results observed in antibody mediated inhibitory receptor blockade, where the perturbed MHC-I sensing is manifested as a reduced inhibitory signal from all cells to certain NK cell subsets, rather than complete lack of inhibitory signal from a fraction of the host cells. The results are discussed in the context of testable experimental predictions of in vitro NK cell responsiveness in mice with mosaic or chimaeric MHC-I expression, as well as in the context of implications for clinical immunotherapy against cancer.