Model systems to study human responses to potential pathogens in general, and to complex viruses such as poxviruses in particular, are of considerable practical interest. They allow for characterization of immune responses directed against epitopes potentially capable of being recognized in humans in a rigorously controlled experimental situation, without having to rely on samples derived from exposed individuals. These issues are particularly relevant in the case of pathogens of bioterrorism concern, as well as emerging and re-emerging pathogens.
Several alternative approaches have been described to circumvent the need to use samples from infected humans. These include using mice re-colonized with human hematopoietic stem cells, primary in vitro restimulation utilizing peripheral blood mononuclear cells (PBMC) from healthy volunteers, and human leukocyte antigen (HLA) transgenic (Tg) mice. Murine/human chimeric systems appear to be associated with significant experimental variability, and responses of low magnitude and diversity compared to those observed in humans have been reported . Primary in vitro restimulation systems are an attractive alternative, but are associated with limited throughput, high cost, labor intensiveness, and the lack of testing the impact of variables associated with host-pathogen interactions in a living organism.
Several groups have reported the development and validation of HLA Tg mice as a model system to study T cell responses restricted by human major histocompatibility complex (MHC) molecules [2, 3, 4]. It has been established that HLA Tg mice represent a fairly accurate model of human immune responses based on peptide immunizations. For example, Wenthworth and co-workers analyzed the repertoire of influenza-specific T cell responses obtained in HLA Tg mice with that obtained following primary induction of human cytotoxic T lymphocyte (CTL), and observed concordant results for approximately 85% of the peptides assayed .
The question of how suitable HLA Tg mice are for defining responses to complex pathogens and antigens has, however, not been addressed in detail. This issue is relevant because significant differences between humans and HLA Tg mice could exist at several levels. It is possible that co-expression of different MHC molecules might impact the TCR repertoire, in that certain specificities are lost while others are gained, as a function of the MHC alleles present . Differences also exist in the cellular processing machinery operating in mice and humans at the level of the proteasome and TAP transport [6, 7, 8]. Discrepancies might also exist in the pattern of expression of viral antigens in infected cells of human and murine origin. Furthermore, it is currently unknown to what degree these differences might influence the global outcome of immune responses.
In our laboratory, we have characterized the T cell response to VACV observed in infected HLA-A*0201 (A2.1) Tg mice  and in human Dryvax vaccinees expressing A2.1 . This enabled us to compare responses to a complex viral pathogen in the two different systems. We have identified a total of 28 different A2.1-restricted epitopes, and other investigators have described three additional epitopes (J8R11–19, A47L169–177, and B5R5–19) using A2-positive vaccinees [10, 11, 12]. Concordant results were found in some cases, but for the most part, epitopes detected in humans were not detected in mice, and vice versa. In the present study, we have further assessed the issue of limited epitope overlap observed in humans and HLA Tg mice.