Cell and Molecular Biology of DNA Methyltransferase 1 - International Review of Cell and Molecular Biology - page 30

mammals. Cells of preimplantation embryos and primordial germ cells are
the logical sites of such demethylation, because the genomes of these cells
are significantly hypomethylated. No orthologs of the ROS1 and DME
DNA glycosylases exist in mammals. Therefore, it is highly unlikely that
an active demethylation process akin to that in Arabidopsis VC and CC exists
in mammals. There are DNA glycosylases in mammalian species that can act
on 5-methyl-cytosine bases in DNA, but their in vitro affinity for
5-methyl-cytosine is so low that they probably do not function in this regard
in tissues of mammalian species. For example, it has been proposed that
mammalian DNA glycosylases such as TDG and MBD4 might catalyze
the removal of 5-methyl-cytosine from DNA, even though they have a
much higher affinity toward mismatched thymine than against
5-methyl-cytosine ( Cort´zar et al., 2007; Dalton and Bellacosa, 2012;
Hardeland et al., 2001; Jiricny and Menigatti, 2008; Ooi and Bestor,
2008; Zhu et al., 2000 ). Finally, it has been reported that DNMT enzymes
themselves can catalyze the reverse of its normal methylation reaction ( Chen
et al., 2013; Kangaspeska et al., 2008; Metivier et al., 2008 ), converting
methylated CpG dinucleotides in DNA to unmethylated CpGs. We con-
clude from this that directed excision of methyl-cytosines in mammals does
not seem to be possible, unless there are novel enzymes yet to be identified
or one or more of the DNMT proteins is truly capable of demethylating
methylated CpGs in the nuclear environment.
Because there is no known DNA glycosylase with high affinity
toward 5-methyl-cytosine in mammalian cells, a number of investigators
have proposed a variety of mechanisms to achieve net replacement of
5-methyl-cytosine with cytosine in the mammalian genome. One such pos-
tulated mechanism begins with an initial 5-methyl-cytosine deamination to
generate thymine and thus thymine-guanosine (T:G) mismatches. Such
mismatches would then be corrected to C:Gmatches in a secondary “repair”
step. Processes such as normal base excision repair (BER) could function
here. The net effect of the two steps (deamination and BER) would be
an active demethylation of 5-methyl-cytosine to cytosine. In support of this
notion, activation-induced deaminase (AID), MBD4, and GADD45A can
cooperate to promote DNA demethylation when genes encoding these pro-
teins are overexpressed in zebrafish ( Rai et al., 2008 ). There is also another
deaminase APOBEC that has been proposed to act on 5-methyl-cytosine in
mammalian cells. GADD45A (growth arrest and DNA damage-inducible
protein 45a) has no known catalytic activity toward DNA ( Nabel et al.,
2012 ). Its postulated role in active demethylation is to coordinate the