Mouse models suggest that the processing of exogenous Ag by dendritic cells can be important for priming the CD8(+) CTL response. To study the situation in humans, we have exploited the CTL response to EBV infection. In this context EBV expresses eight latent proteins, of which EBV-encoded nuclear Ag (EBNA) 3A, 3B, and 3C appear to be immunodominant for CTL responses, whereas another nuclear Ag, EBNA1, which is completely protected from endogenous presentation via the MHC class I pathway, is thought to induce responses rarely, if ever. Here, using EBNA1 peptides and/or EBNA1 protein-loaded dendritic cells as in vitro stimuli, we have identified memory CTL responses to HLA-B*3501, -B7, and -B53-restricted EBNA1 epitopes that can be as strong as those seen in immunodominant epitopes from the "conventionally processed"" EBNA3 Ags. Furthermore, we used HLA-peptide tetramers to show that the primary response to one such EBNA1 epitope constituted up to 5% of the CD8(+) T cells in infectious mononucleosis blood, the strongest latent Ag-specific response yet detected in this setting. We conclude that exogenous protein represents a significant source of Ag for priming the human CTL response.