DNA immunization can induce cytotoxic T lymphocytes (CTL), antibodies, and protection against microbial challenge. The underlying mechanisms remain obscure and must be understood to permit rational manipulation and optimization of the technique. We set out to enhance the intracellular degradation of a viral antigen, with the intent of improving antigen entry into, and presentation by, the class I major histocompatibility complex pathway. We achieved this goal by cotranslational ubiquitination of a plasmid-encoded viral antigen, lymphocytic choriomeningitis virus (LCMV) nucleoprotein (NP). We show that native NP is very stable in cell culture, while the ubiquitinated product is so rapidly degraded that it is barely detectable. This rapid degradation leads to more efficient sensitization of target cells in an in vitro cytotoxicity assay, consistent with enhanced antigen presentation, and both degradation and target cell recognition are blocked by a proteasome inhibitor. We have used the plasmid for in vivo studies and find that, remarkably, ubiquitination leads to a complete abrogation of antibody responses, presumably because the encoded protein is so rapidly and completely degraded that insufficient antigen remains to interact appropriately with B cells. In contrast, in vivo CTL induction is improved by ubiquitination of NP. That CTL are induced at all by this rapidly degraded protein may shed light on the mechanism by which CTL are induced by DNA immunization; it has been suggested that CTL induction following intramuscular DNA injection results not from antigen presentation by cells taking up and expressing the DNA but rather from uptake of soluble protein by specialized antigen-presenting cells (APC). It appears to us unlikely that the ubiquitinated protein could function in this manner, since it is so rapidly degraded in vitro and fails to induce antibodies in vivo. Finally, the ubiquitinated protein confers markedly enhanced protection against LCMV challenge. Mice immunized with a plasmid encoding NP show approximately 100-fold reductions in virus titers compared to controls, while mice immunized with a plasmid encoding the ubiquitinated NP show reductions in virus load of at least 5 x 10(4)- to 5 x 10(5)-fold. This is by far the most effective DNA vaccine that we have yet designed. Ubiquitination therefore may improve DNA immunization, but caution is warranted, since immunity to many microbes depends on induction of good humoral immunity, and we show here that this may be prevented by ubiquitination of the encoded protein.