Recent evolution of a TET-controlled and DPPA3/STELLA-driven pathway of passive demethylation in mammals

Genome-wide DNA demethylation is a unique feature of mammalian development and naive pluripotent stem cells. So far, it was unclear how mammals specifically achieve global DNA hypomethylation, given the high conservation of the DNA (de-)methylation machinery among vertebrates. We found that DNA demethylation requires TET activity but mostly occurs at sites where TET proteins are not bound suggesting a rather indirect mechanism. Among the few specific genes bound and activated by TET proteins was the naive pluripotency and germline marker Dppa3 (Pgc7, Stella), which undergoes TDG dependent demethylation. The requirement of TET proteins for genome-wide DNA demethylation could be bypassed by ectopic expression of Dppa3. We show that DPPA3 binds and displaces UHRF1 from chromatin and thereby prevents the recruitment and activation of the maintenance DNA methyltransferase DNMT1. We demonstrate that DPPA3 alone can drive global DNA demethylation when transferred to amphibians (Xenopus) and fish (medaka), both species that naturally do not have a Dppa3 gene and exhibit no post-fertilization DNA demethylation. Our results show that TET proteins are responsible for active and - indirectly also for - passive DNA demethylation; while TET proteins initiate local and gene-specific demethylation in vertebrates, the recent emergence of DPPA3 introduced a unique means of genome-wide passive demethylation in mammals and contributed to the evolution of epigenetic regulation during early mammalian development.