Adapter proteins such as Grb2 play a central role in the formation of signaling complexes through their association with multiple protein binding partners. These interactions are mediated by specialized domains such as the well-characterized Src homology SH2 and SH3 motifs. Using yeast three-hybrid technology, we have identified a novel adapter protein, expressed predominantly in T lymphocytes, that associates with the activated form of the costimulatory receptor, CD28. The protein is a member of the Grb2 family of adapter proteins and contains an SH3-SH2-SH3 domain structure. A unique glutamine/proline-rich domain (insert domain) of unknown function is situated between the SH2 and N-terminal SH3 domains. We term this protein GRID for Grb2-related protein with insert domain. GRID coimmunoprecipitates with CD28 from Jurkat cell lysates following activation of CD28. Using mutants of CD28 and GRID, we demonstrate that interaction between the proteins is dependent on phosphorylation of CD28 at tyrosine 173 and integrity of the GRID SH2 domain, although there are also subsidiary stabilizing contacts between the PXXP motifs of CD28 and the GRID C-terminal SH3 domain. In addition to CD28, GRID interacts with a number of other T cell signaling proteins, including SLP-76 (SH2 domain-containing leukocyte protein of 76 kDa), p62dok, and RACK-1 (receptor for activated protein kinase C-1). These findings suggest that GRID functions as an adapter protein in the CD28-mediated costimulatory pathway in T cells.

CD28 has been demonstrated to provide the major costimulatory signal for CD4-positive T cells. Ligation with its natural ligands CD80 (B7-1) and CD86 (B7-2) leads to signals during activation that are required for the production of interleukin-2, and this process has been implicated in the regulation of T-cell anergy and programmed cell death. This article describes the assay development, assay validation, and primary screening for small molecule antagonists of this interaction, which could be potential drug candidates. The assay uses homogeneous time-resolved fluorescence based on energy transfer from excited europium ions to cross-linked allophycocyanin, which then subsequently emits a fluorescent signal. An "indirect" approach was taken, whereby the cross-linked allophycocyanin (XL665) is covalently linked to an antihuman antibody that binds to a human immunoglobulin (Ig) domain fused to CD28. The CD86 that is expressed as a fusion protein with a rat Ig domain is bound to biotinylated sheep antirat antibody, which is complexed with streptavidin-europium cryptate. This "cassette" format facilitates the development of related assays using CTLA-4 in place of CD28 and/or CD80 in place of CD86, allowing easy determination of the selectivity of active compounds. When the CD28 and CD86 are in close proximity (i.e., bound), there is a specific time-resolved emission at 665 nm that is largely absent in either unbound partner. Experiments to optimize the reagent concentrations, incubation time, solvent effects and quench effects by colored compounds are discussed, as are the results from robustness testing and data from primary screening.

CD80 and CD86 are cell surface glycoproteins expressed on a variety of professional APCs. They have attracted much attention due to their function as potent costimulators of T lymphocyte function through their interaction with CD28 and possibly CTLA4. Because inhibitors of this interaction may have therapeutic relevance in human autoimmune disease, we investigated the properties of linear peptides derived from conserved regions of CTLA4 and CD80 known to be essential for binding. None of these peptides were sufficient to bind ligand, nor did they act as potent competitive inhibitors. Conformationally constrained versions of the CTLA4 motif were also inactive. These results suggested that other parts of the proteins are important in determining binding, so a series of modified CD80 and CD86 molecules were constructed in an attempt to identify other binding determinants. Insertion of two residues between the two Ig domains of CD80 resulted in decreased affinity for CTLA4, but a similar mutation in CD86 was without effect. We also identified another asymmetry between CD80 and CD86 in that the V domain of CD86 but not that of CD80 is sufficient for CTLA4 binding. The CD86-V domain appears to have CTLA4 binding properties equivalent to that of intact CD86. These data illustrate a fundamental difference between these costimulatory molecules and suggest a mechanism by which they may be differentially recognized by receptors on the T cell surface.