Aberrant de novo methylation of CpG island DNA sequences has been observed in cultured cell lines or upon malignant transformation, but the mechanisms underlying this phenomenon are poorly understood. Using eukaryotic DNA (cytosine-5)-methyltransferase (of both human and murine origin), we have studied the in vitro methylation pattern of three CpG islands. Such sequences are intrinsically poor substrates of the enzyme, yet are efficiently methylated when a small amount of 5-methylcytosine is randomly introduced by the M.SssI prokaryotic DNA (cytosine-5)-methyltransferase prior to in vitro methylation by the eukaryotic enzyme. A stimulation was also found with several other double-stranded DNA substrates, either natural or of synthetic origin, such as poly(dG-dC).poly(dG-dC). An A + T-rich plasmid, pHb beta 1S, showed an initial stimulation, followed by a severe inhibition of the activity of DNA (cytosine-5)-methyltransferase. Methylation of poly(dI-dC).poly(dI-dC) was instead inhibited by pre-existing 5-methylcytosines. The extent of stimulation observed with poly(dG-dC).poly(dG-dC) depends on both the number and the distribution of the 5-methylcytosine residues, which probably must not be too closely spaced for the stimulatory effect to be exerted. The activity of the M.SssI prokaryotic DNA methyltransferase was not stimulated, but was inhibited by pre-methylation on either poly(dG-dC).poly(dG-dC) or poly(dI-dC).poly(dI-dC). The prokaryotic and eukaryotic DNA methyltransferases also differed in sensitivity to poly(dG-m5dC).poly(dG-m5dC), which is highly inhibitory for eukaryotic enzymes and almost ineffective on prokaryotic enzymes.

We have characterized the inhibition exerted by histone H1 on the activity of human placenta DNA (cytosine-5-)-methyltransferase. Our experiments demonstrate that the extent of inhibition depends on the DNA base composition, AT-rich substrates being more severely affected than GC-rich substrates and CpG-rich islands. With bacterial SssI methylase, the effect is completely reversed since its activity on AT-rich substrates undergoes a 4-5-fold stimulation upon the addition of H1. Poly(L-lysine) mimicks H1 effects, suggesting an essential role of lysine residues in both the inhibitory and stimulatory effects of H1. By comparison of the different behaviors of the two enzymes, the inhibitory effect over the eukaryotic enzyme might be accounted for by hypothesizing a competition between minor groove-binding motifs (SPKK-like) present in placenta methylase as well as in histone H1.