A number of lymphocyte surface proteins are anchored in the cell membrane by glycophosphatidyl inositol (known as GPI) linkages instead of hydrophobic protein domains. Treatment of mouse T lymphocytes with antibodies specific for two such proteins, Thy-1 and Ly-6, are known to induce proliferation. We have found that antibodies specific for Qa-2, a GPI-anchored class I histocompatibility antigen, can also activate mouse T cells. To determine whether the GPI-anchor is important for this pathway of cell activation, we produced transgenic mice expressing either normal GPI-anchored Qa-2, or Qa-2 molecules with a membrane-spanning protein domain derived from H-2. Our studies show that only lymphocytes from transgenic mice carrying GPI-anchored forms of Qa-2 can be activated in vitro by Qa-2-specific antibodies. We also show that transgenic mouse T cells expressing a GPI-anchored form of H-2Db can be activated by anti-H-2Db antibodies. These results strongly indicate that the GPI-anchor is critical for this pathway of T cell activation.

The peptide-binding properties of the nonclassical major histocompatibility complex (MHC) class 1b molecule Qa-1 were investigated using a transfected hybrid molecule composed of the alpha 1 and alpha 2 domains of Qa-1b and the alpha 3 domain of H-2Db. This allowed the use of a monoclonal antibody directed against H-2Db whilst retaining the peptide-binding groove of Qa-1b. By comparison with classical MHC class I molecules, intracellular maturation of the chimeric molecule was inefficient with weak intracellular association with beta 2-microglobulin. However, at the cell surface the hybrid molecules were stably associated with beta 2-microglobulin and were recognized by cytotoxic T lymphocyte (CTL) clones specific for the Qa-1b-presented peptide Qdm (AMAPRTLLL). A whole-cell binding assay was used to determine which residues of Qdm were important for binding to Qa-1b and CTL clones served to identify residues important for T cell recognition. Substitutions at position 1 and 5 did not reduce the efficiency of binding and had little effect on CTL recognition. In contrast, substitutions at position 9 resulted in loss of MHC class I binding. Mass spectrometric analysis of peptides eluted from immunopurified Qa-1b/Db molecules indicated that Qdm was the dominant peptide. The closely related peptide, AMVPRTLLL, which is derived from the signal sequence of H-2Dk, was also present, although it was considerably less abundant. The mass profile suggested the presence of additional peptides the majority of which consisted of eight to ten amino acid residues. Finally, the finding that a peptide derived from Klebsiella pneumoniae can bind raises the possibility that this non-classical MHC class I molecule may play a role in the presentation of peptides of microorganisms.

Allelic exclusion of T cell receptor (TCR) genes is incomplete: a significant percentage (10-30%) of normal human and mouse peripheral T cells express two surface TCR alpha chains, and a small percentage of peripheral human T cells have been reported to express two surface TCR beta chains. A proportion of thymocytes in TCR transgenic mice rearrange endogenous T cell receptor genes, and peripheral T cells with two TCR alpha chains, transgenic and endogenous, have been reported. T cell clones with more than a single TCR heterodimer on their surface might be expected to show specificity for more than one cognate antigen: we report here a T cell clone with dual antigen specificity, isolated from an F5 TCR influenza nucleoprotein (NP 366-374/Db)-specific transgenic female mouse which had rejected an H-2-matched male skin graft. It was selected in vitro by stimulation with male H-2b spleen cells in the absence of the NP366-374 peptide but has specificity for both H-Y/Db and NP366-374. This contrasted with the single NP366-374/Db specificity shown by a control clone isolated from a Rag1-/- F5 mouse. The dual antigen specificity was associated with the rearrangement of endogenous TCR genes and cell surface expression of these as well as the TCR transgene.

KT16, like other anti-Thy-1 antibodies, induces T cell aggregation. Protein A-gold labelling shows the antibody to be concentrated along areas of intercellular contacts. Electron micrographs of KT16 treated T cells reveal a consistent type of junction between the cells. We demonstrate that this type of cell junction is Thy-1 specific, is predominantly the property of antibodies directed against a particular epitope, and is distinct from cellular aggregation caused by concanavalin A or anti-CD3 antibodies. The degree of adhesiveness induced by different anti-Thy-1 antibodies is related to their mitogenic capacity.

Fetal mouse NK cells are grossly deficient in the expression of Ly49 molecules yet show a limited ability to distinguish between wild-type and MHC class I-deficient target cells. In this paper we report that during their development in vitro from immature thymic progenitors, a proportion of C57BL/6 fetal NK cells acquires receptors for a soluble form of the nonclassical class I molecule Qa1b associated with the Qdm peptide, but not for soluble forms of the classical class I molecules Kb and Db. The acquisition of these Qa1 receptors occurs in a stochastic manner that is strictly controlled by cytokines, and in particular is strongly inhibited by IL-4. All fetal NK clones tested, including those that lack detectable Qa1 receptors, express mRNA for CD94 and for both inhibitory and noninhibitory members of the NKG2 family. Fetal NK cells lacking receptors for Qa1 (and also for classical class I molecules) cannot distinguish between wild-type and class I-deficient blasts but, surprisingly, distinguish efficiently between certain wild-type and class I-deficient tumor cells. A variant line that lacks several members of the NKG2 family kills both types of tumor cell equally well, suggesting the existence of NKG2-containing inhibitory receptors that recognize as yet undefined nonclassical class I molecules of restricted distribution.