Gene silencing and endogenous DNA methylation in mammalian cells

Abstract

It is known that transformed mammalian cells can spontaneously inactivate genes at low frequency by the de novo methylation of promoter sequences. It is usually assumed that this is due to DNA methyl transferase activity, but an alternative possibiity is that 5-methyldCTP is present in these cells and can be directly incorporated into DNA. The ongoing repair of DNA containing 5-methylcytosine will produce 5-methyldeoxycytidine monophosphate (5-methyldCMP), so the question arises whether this can be phosphorylated to 5-methyldCTP. We have tested this using three strains of CHO cells with different levels of 5-methyldCMP deaminase activity. That with the lowest enzyme activity, designated HAM −, has previously been shown to incorporate tritium labelled 5-methyldeoxycytidine into 5-methylcytosine in DNA, with a greater amount of label in thymine. This strain is phenotypically unstable producing cells resistant to bromodeoxyuridine (BrdU) and 6-thioguanine (6-TG) at high frequency. In contrast, the strain with the highest 5-methyldCMP deaminease, designated HAM +, is extremely stable, and the starting strain K1 HAM s1 is intermediate between the HAM − and HAM + phenotypes. We have also shown that human diploid fibroblast strain MRC-5 has a phenotype like HAM +, whereas its SV40 transformed derivative, MRC-5V2 resembles HAM − in having low 5-methyl dCMP deaminase activity, and is phenotypically unstable with regard to 6-TG resistance. It seems that 5-methyldCMP deaminase can be down-regulated in transformed cells, and this can promote de novo methylation by incorporation of 5-methyldCTP derived from 5-methyldCMP.