The cytosine analog 5-azacytidine kills Escherichia coli cells that carry plasmids expressing EcoRII DNA (cytosine 5)methyltransferase under control of its own promoter. We previously showed that this enzyme binds tightly to azacytidine-containing DNA in vitro and proposed that such binding is lethal in vivo. In support of this proposal, we now show that the enzyme sediments with the nucleoid of azacytidine-treated cells. Azacytidine treatment led to an increase in the amount of enzyme, and this increase required sequences in the ecoRIIM promoter region. Enzyme inducibility correlated with drug sensitivity: plasmids carrying the methyltransferase gene but lacking the wild-type promoter did not confer sensitivity. These results suggested that the ecoRIIM gene was under autogenous control. Transcriptional ecoRJIM'-lacZ fusions in E. coli were, therefore, constructed. They showed that expression from the ecoRIIM promoter was inhibited when EcoRU DNA (cytosine-5)methyltransferase was introduced into the cell in trans and inhibition was reversed by treating the cells with azacytidine. These results provide evidence that the expression of the ecoRIIM gene is under autogenous regulation and that cell death induced by azacytidine is due, in part, to the disruption of autoregulation.5-Azacytidine (azaCyd) is a nucleoside analog that selectively kills Escherichia coli when the bacteria carry plasmids that specify the EcoRII restriction-modification system (7). The killing is dependent on the presence of the methyltransferase (MTase) and not the restriction enzyme. This phenomenon was also demonstrated when other MTases were cloned and expressed in E. coli on multicopy plasmids (2, 16). The effect appeared to be dependent on the amount of MTase present in the cell (7). With the endogenous Dcm MTase there was little effect of the drug. When this MTase was cloned onto a multicopy plasmid and introduced into E. coli, the strain became sensitive to azaCyd (2). A similar but less dramatic effect was also noted when the MspI MTase, cloned in pBR322, was introduced into E. coli (16).Cells deficient in any of a number of different repair enzymes have enhanced sensitivity to azaCyd. These results imply that the drug causes DNA damage. Support for this conclusion includes the observations that the drug is a weak inducer of the SOS system and is a mutagen in E. coli (1, 16). Although the drug is known to undergo ring opening under mild conditions, it has not been shown that DNA strand breaks occur in vivo.We have shown that the drug inhibits DNA (cytosine-5)methyltransferases (5, 6) once the drug is incorporated into DNA in the sequence CC(A/T)GG (9). We proposed that the basis for cell death was the binding of the enzyme to azacytosine-containing DNA (azaC-DNA), which would inhibit the ability of the cell to repair DNA damage caused by azaCyd (13). In support of this suggestion, we demonstrated that DNA containing azaCyd will not undergo recA-mediated strand exchange if the EcoRII DNA (cytosine-5)methyltransferase (M.EcoRII) is boun...