Cellular immune responses against human immunodeficiency virus type 1 (HIV-1) and the related simian immunodeficiency virus (SIV) have been shown to play an important role in controlling HIV-1 and SIV infection and in delaying disease progression. Containment of primary HIV-1 infection in infected individuals correlates with the emergence of virus-specific cytotoxic T-lymphocyte (CTL) responses (4, 14, 26). In chronically infected individuals, a high-frequency CTL response against HIV-1 is also correlated with a low viral load and slow disease progression (24, 25). An HIV-1-specific CTL response has also been demonstrated in certain highly exposed seronegative individuals (2, 15, 32). Also, strong HIV-specific proliferative responses, which may be critical for the maintenance of CTL responses, have been identified in long-term nonprogressors (31, 35). HIV-1 Gag is one of the most conserved viral proteins. Broad, cross-clade CTL responses recognizing conserved epitopes in HIV-1 Gag have been detected in HIV-1-infected people (11, 21), and the development of a safe and effective HIV-1 vaccine may depend on the induction of effective CTL and/or T-helper responses against conserved HIV-1 proteins such as Gag. DNA vaccines have been shown to induce efficient cellular immune responses and protection against a variety of viral, bacterial, and parasitic pathogens in animal models. However, DNA vaccines that could induce potent cellular immune responses against HIV-1 Gag are not yet available. We have recently demonstrated that by destroying inhibitory sequences in the coding region of HIV-1 gag, we could significantly increase Gag protein expression in primate as well as in mouse cells (27, 34, 36, 37) and dramatically enhance immune responses induced by DNA vaccine (27). Since this new Gag expression vector is Rev/RRE-independent and species-independent, it provides a feasible approach to systematically evaluating the strategies that could lead to the maximum induction of cellular immune responses against HIV Gag molecules in animal models. Intramuscular (i.m.) administration of a DNA vaccine represents a simple and effective means of inducing both humoral and cellular immune responses (10). There are three potential pathways responsible for antigen presentation after i.m. injection of DNA. First, muscle cells could take up the DNA, express the encoded protein antigen, and present it to immune cells. Recent data suggest that this pathway is rather unlikely in vivo (40). Second, dendritic cells attracted to the site of injection may take up the DNA, express the encoded protein, and present it to T and B cells. Third, muscle cells may take up the DNA and express the protein antigen, with the antigen then being transmitted to dendritic cells for presentation. If the second possibility is the case, a protein that is synthesized and degraded in the cytoplasm of dendritic cells would be an excellent target for major histocompatibility complex (MHC) class I presentation and induction of CTL responses. Alternatively, if the third scenario were true, a protein synthesized in the muscle cells that could be targeted efficiently to dendritic cells would induce the best CTL response. To distinguish among these different possibilities, we have now constructed and compared three different forms of HIV-1 Gag DNA vaccine vectors for the induction of immune responses. These different forms of Gag include (i) a standard Gag (St-Gag) that assembles into particles, which are efficiently released from cells and become surrounded by host-cell-derived lipid membrane acquired during virus budding; (ii) a cytoplasmic form of Gag (Cy-Gag) that fails to target the plasma membrane and therefore remains in the cytoplasm; and (iii) a secreted form of Gag (Sc-Gag) that is synthesized on the cytoplasmic face of the rough endoplasmic reticulum (ER), transported through the ER and Golgi apparatus, and released as a secreted protein (i.e., not surrounded by a lipid membrane). When these forms of Gag were administered to mice as DNA vaccine, we found that the DNA vector encoding the Sc-Gag generated better primary CTL and T-helper responses than did the DNA vector encoding Cy-Gag. Furthermore, the DNA vector encoding the Sc-Gag also generated a higher level of secondary CTL responses than did the DNA vector encoding Cy-Gag after DNA priming and recombinant vaccinia virus-Gag boost. Vaccinia virus titers were notably reduced in the ovaries of mice immunized with Gag DNA vaccine more than 125 days before infection, as compared to the titer in mice that received only the control DNA vector. These data indicate that CD8+ T-cell memory elicited by DNA vaccination is functionally relevant and provides protective immunity in this system. Again, the DNA vector encoding the Sc-Gag provided better protection against recombinant vaccinia virus-Gag than did the DNA vector encoding Cy-Gag.