Biol. Pharm. Bull. 28(10) 2019—2022 (2005)

their excellent safety profile, despite their low transgene expression efficiency in comparison to that of viral vectors. For their administration, cationic lipid-mediated transfer is a promising approach, because of its low toxicity, ease of preparation, and potential for active targeting. One disadvantage associated with the use of DNA–cationic lipid complexes systemically is the induction of an inflammatory response. To overcome the inflammatory response associated with the cationic lipid-mediated gene transfer, preinjection of free liposomes and coinjection of NFkB decoy oligodeoxyribonucleotides (ODNs) have been attempted. These treatments decrease cytokine production and increase transgene expression. The inflammatory response elicited by the DNA–cationic lipid complexes is thought to be due to unmethylated CpG sequences, because the unmethylated CpG dinucleotides in ODNs effectively induce cytokine production. Additionally, Yew et al. achieved enhanced and long-term transgene expression by eliminating the CpG sequences in plasmid DNA, suggesting the importance of inflammation suppression. They reported an inverse correlation between transgene expression and the number of CpG dinucleotides, which cause an inflammatory response: the CpG-depleted plasmid DNA expressed transgene products more efficiently than the usual CpG-rich plasmid DNA. Recently, we found that an ODN containing N -methyladenine (N -MeA), a bacterium-specific modified base, induced cytokines. This base is post-replicationally modified by the Dam protein in Escherichia coli. When complexed with cationic lipids, plasmid DNA containing N -MeA also induced interleukin-12 (IL-12). Nearly half of the total IL-12 production was due to the N -MeA residues in the plasmid DNA. Thus, plasmid DNAs without N -MeA are expected to express transgenes more efficiently in vivo than those containing N -MeA, because of the lower inflammatory response. Previously, Allamane et al. compared the in vivo levels of transgene expression from plasmid DNAs with and without N -MeA, delivered by direct injection and electroporation in the mouse muscle. They observed that the plasmid DNA containing N -MeA produced the transgene product more efficiently than that without N -MeA, after 24 h. However, they did not measure the cytokine levels after the administration, and thus the effects of cytokine production by N -MeA could not be evaluated. In addition, they amplified plasmid DNAs with and without N -MeA in the E. coli JM109 (dam endA ) and JM110 (dam endA ) strains, respectively. The presence of endonuclease A, encoded by the endA gene, may damage the amplified plasmid DNA in the JM110 strain. Indeed, it was reported that the presence of an endA mutation had a significant effect on both the yield and quality of plasmid DNA. The transgene on the damaged DNA may be transcribed less efficiently and may be degraded by cellular nucleases more rapidly than undamaged DNA. To examine the effects of adenine methylation and its induction of inflammatory cytokines on transgene expression in mice, we amplified plasmid DNA in dam endA and dam endA E. coli strains. The plasmid DNAs, complexed with cationic lipids and a cationic polymer, were intravenously injected into mice. Our results suggest that the use of plasmid DNA isolated from a dam strain does not increase transgene expression in vivo.