820
Our previous studies have demonstrated that vascular endothelial cells have the means, motive and opportunity to act as APC for T cells at sites of inflammation, but direct evidence for endothelial APC function in vivo remains lacking. We devised an experimental system to address this issue. Purified peripheral blood T cells from tetanus toxoid (TT)-sensitized individuals were injected subcutaneously within the pinnae of mice, along with TT and human APCs. DTH-like swelling responses developed within 24 hours, but only if human T cells, human APC and TT were all provided within the same pinnae. Effective APCs included not only autologous macrophages but also human umbilical vein endothelial cells (HUVEC), provided the latter were pre-treated with IFNg to drive expression of costimulation and MHC class II molecules. Both APC types failed to function if they were pre-treated with mAb to MHC class II, but not MHC class I mAbs. In a second series of experiments using similar experimental conditions, mixtures of allogeneic human PBMC caused strong DTH-like swelling responses when injected into murine pinnae. However, mixtures of human T cells plus allogeneic IFNg-treated HUVEC did not. These observations demonstrate that human endothelial cells can function as APC in vivo, and provide a model with which to study these responses. Further, they show that human endothelial cells can readily promote local recall responses to soluble antigens in vivo, but are unable to promote primary allogeneic responses under the same conditions. This suggests that endothelial cells may act in vivo as APC for secondary responses by memory T cells, but are poor APC for primary responses by naive T cells.