Three different vaccination sites were compared for expression vector used. The cytomegalus virus promoter efficiency of immunization with naked DNA. Using the bacdriven pCMV vector was superior to the Moloney murine terial lacZ gene as a model, all three sites of the mouse leukemia virus LTR driven BAG vector. LacZ DNA immuni-(skeletal muscle, dermis of abdominal skin or of the ear zation was also compared with cell-based vaccination with pinna) could express the gene product -gal but varied in lacZ-transfected tumor cells, in which case again the pinna expression time with muscle tissue showing the longest was the best site for inducing strong immune responses. expression. Expression time, however, did not correlateTumor-specific T cell responses could also be well induced with immune response intensity. The ear pinna was by far in the pinna, leading to cytotoxic T lymphocyte induction the most effective and muscle the least effective priming and protective antitumor immunity. Thus, the pinna was site for specific humoral and cytotoxic T cell-mediated found to be a privileged site for induction of antitumor immune responses. Following intra-pinna DNA inoculation, responses and for genetic immunization, an important find--gal expressing cells were detectable around the injection ing of immediate practical and potential future clinical site and in the major draining lymph node. Efficiency of implications. immunization was also dependent on the promoter and