Mapping of the specificity in MHC class I recognition by natural killer cells

Abstract

Natural killer (NK) cells represent the third major lymphocyte subpopulation. They are distinguishable from B and T lymphocytes by their surface phenotype, cytokine profile and the ability to mediate spontaneous cytotoxicity, without prior sensitization, against certain tumor cells and virally infected cells. NK cells can mediate resistance against tumor growth and metastasis, as well as against certain viral and bacterial infections. They can also reject MHC mismatched bone marrow grafts, and they have recently been implicated in certain autoimmune conditions.

This thesis addresses the molecular specificity of NK cells, which express two functional types of receptors, activating and inhibitory. The activating receptors can recognize a variety of widely distributed cell surface molecules, including MHC class I molecules. Most of the inhibitory receptors characterized to date are specific for MHC class I molecules. Many normal cells express ligands for the activating receptors, and the specificity is therefore often ultimately determined by the inhibitory receptors. This can explain why NK cells kill MHC deficient tumor cells as well as non-autologous bone marrow cells lacking self MHC class I alleles as ligands for inhibitory receptors. The dominating MHC class I specific inhibitory receptors in the mouse are found within the Ly49 family, belonging to the larger superfamily of C-type lectin-like molecules. The Ly49 receptors are type II membrane glycoproteins expressed as disulfide-linked homodimers. At the start of the studies, the first receptor in this group (Ly49A) had just been identified; it was furthermore clear that tumor cells with total loss or incomplete expression of MHC class I molecules in relation to a particular host could be rescued from attack by NK cells of that host by restoring a complete MHC phenotype.

To adress the allelic specificity of this MHC class I mediated protection, a series of mutant and transfected lymphoma cells were tested for the ability to escape rejection in SCID mice of H-2d type, devoid of T and B cells but with functional NK cells. It was not necessary to restore a complete self MHC phenotype in order to achieve protection from NK cells. One syngeneic allele, Dd, protected efficiently while another had no effect at all. Protection could be achieved also with an allogeneic class I molecule.

The next question was whether this allelic specificity could be mapped to a particular domain of the MHC class I molecule. This was addressed by exon shuffling between the Dd gene on one hand, and the Ld or Db gene on the other hand. The resulting chimeric genes were transfected to lymphoma cells, and studied with respect to their ability to confer protection against NK cell mediated rejection in vivo. This demonstrated that the allelic specificity resided in the [alpha]1 /[alpha]2 domains, i.e. the membrane distal part of the MHC class I molecule that builds up the peptide-binding cleft. Further exon shuffling allowed a more precise mapping: the [alpha]2 domain of the Dd molecule was sufficient to confer protection in vivo as well as in vitro to rat NK cells transfected with the gene for the inhibitory Ly49A receptor. It was thus concluded that the allelic specificity (Dd ) of the Ly49A receptor can be explained either by i) exclusive binding to the [alpha]2 domain, ii) simultaneous binding to conserved motifs in the [alpha]1 domain and to allele specific motifs in the [alpha]2 domain iii) binding to the [alpha]1 domain only, but to a conserved motif, strongly dependent on the Dd [alpha]2 domain.

This led to a search for structural motifs that differed between the Db and the Dd [alpha]2 domains, and could be tested for influence on Ly49A recognition. The search was initially based on molecular models, and eventually on the published structure of the Dd crystal. Two motifs, introduced or interfered with by site-directed mutagenesis, were found to markedly impair the Ly49A/Dd interaction: i) a hydrophobic ridge in the floor of the peptide binding cleft (absent in Dd , present in Db ), created by amino acids at positions 73 and 156, ii) a solvent exposed loop (present in both alleles, but creating distinct motifs in each of them), formed by residues 1 02-1 09. The mutations introduced in this loop of the Dd molecule were studied by expression in constructs for MHC class I tetramer production, thus allowing direct assessment of the Ly49A binding to its ligand. The mutations in this loop did not only impair the Ly49A/Dd interaction; they also abrogated binding of the only monoclonal antibody that efficiently can interfere with Ly49A recognition of Dd.

These results are discussed in relation to the current literature on MHC class I ligands for NK cell receptors. This includes other studies on site-directed mutagenesis, on the role of the MHC bound peptide and on the MHC associated glycans, as well as the published cocrystal of the Ly49A/Dd complex, demonstrating two contact sites involving the [alpha]1 as well as the [alpha]2 domain of Dd.