The cytotoxic activity of NK cells is regulated by class I MHC proteins. Although much has been learned about NK recognition of class I autologous targets, the mechanisms of NK self-tolerance are poorly understood. To examine the role of a nonpolymorphic, ubiquitously expressed class Ib Ag, Q9, we expressed it on class I-deficient and NK-sensitive B78H1 melanoma. Presence of this Qa-2 family member on tumor cells partially protected targets from lysis by bulk lymphokine-activated killer (LAK) cells. H-2Kb-expressing B78H1 targets also reduced LAK cell activity, while H-2Db offered no protection. Importantly, blocking with F(ab′)2 specific for Q9 or removal of this GPI-attached molecule by phospholipase C cleavage restored killing to the level of vector-transfected cells. Experiments with LAK cells derived from H2b SCID and B6 mice established that NK1.1+TCR− NK and NK1.1+TCR+ LAK cells were the prevalent cytolytic populations inhibitable by Q9. Treatment of mice with poly(I:C) also resulted in generation of Q9-regulated splenic cytotoxicity. LAK cells from different mouse strains responded to Q9, suggesting that the protective effect of this molecule is not detectably influenced by Ly49 polymorphisms or the presence/absence of Q9 in NK-harboring hosts. We propose that Q9 expressed on melanoma cells serves as a ligand for yet unidentified NK inhibitory receptor(s) expressed on NK1.1+ NK/T cells.

One of the principal mechanisms of tumor immune evasion is alteration of class I MHC expression. We have identified defects contributing to down-regulation of class I MHC expression in the widely studied murine B16 melanoma and its variants B16F1, B16F10, BL6-2, BL6-8 and B78H1. Transcription of the nonclassical class I MHC genes Q8 and Q9 (Qa-2 Ags) has been switched off in the entire panel of melanoma lines, suggesting that this event occurred early during tumor progression. B78H1, unlike B16F1 and B16F10 sublines, is also selectively devoid of TAP2 and low molecular weight protein 7 as well as classical class I MHC Kb and Db transcripts. Cotransfection of B78H1 with TAP2 and class I H chain genes is sufficient to reconstitute surface expression of exogenously delivered class I MHC without concomitant re-expression of endogenous β2-microglobulin-associated class I. The serological absence of endogenous class Ia and Ib at the surface of TAP2-negative as well as TAP2-transfected B78H1 makes this system a suitable model for studying the properties of isolated class I proteins in tumors. We used this system to demonstrate that B78H1 cells genetically manipulated to re-express Q9 Ag have reduced tumor potential in syngeneic B6 mice compared with TAP2-transfected parental melanoma. Both NK cells and CTLs appear to collaborate in restraining growth of Q9-positive tumors. The results implicate Qa-2 in antitumor responses and illustrate the utility of the B78H1 system for identifying in vivo interactions between class I MHC molecules of interest and immune cells of innate and/or adaptive immunity.

Studies of naturally processed peptides eluted from membrane-bound and soluble isoforms of murine class Ib Qa-2 molecules determined several features of these ligands, such as the conserved nonameric length and the preferred usage of specific residues at four to six of nine peptide positions. The structural information derived from these studies proved insufficient to distinguish between two interpretations: 1) that Qa-2 are peptide receptors of higher stringency than ordinary class I molecules, and 2) that Qa-2 molecules, like classical class I Ags, bind diverse arrays of peptides. We have addressed this issue by a systematic analysis of peptide residues involved in the binding of membrane-bound Qa-2 molecule, MQ9b. The optimal binding of synthetic peptides in vitro occurs at neutral pH. Two dominant anchors are required for peptide binding to MQ9b: His at position 7 and a hydrophobic residue, Leu, Ile, or Phe, at position 9. In addition, one or two auxiliary anchors participate in binding. The identity and the position of the auxiliary anchors differ from peptide to peptide, suggesting that the binding motifs defined from pool sequencing are composed of many superimposed alternative motifs present in individual peptides. The number of anchors used by Qa-2 peptides is similar to that found in ligands of classical class I Ags. Consequently, the Qa-2 are predicted to bind large repertoires of self and nonself peptides. In support of this interpretation we demonstrate that MQ9b binds strongly 5 of 17 motif-positive, pathogen-derived synthetic peptides.