Conserved dual-mode gene regulation programs in higher eukaryotes

Abstract

Recent genomic data analyses have revealed important underlying logics in eukaryotic gene regulation, such as CpG islands (CGIs)-dependent dual-mode gene regulation. In mammals, genes lacking CGIs at their promoters are generally regulated by interconversion between euchromatin and heterochromatin, while genes associated with CGIs constitutively remain as euchromatin. Whether a similar mode of gene regulation exists in non-mammalian species has been unknown. Here, through comparative epigenomic analyses, we demonstrate that the dual-mode gene regulation program is common in various eukaryotes, even in the species lacking CGIs. In cases of vertebrates or plants, we find that genes associated with high methylation level promoters are inactivated by forming heterochromatin and expressed in a context-dependent manner. In contrast, the genes with low methylation level promoters are broadly expressed and remain as euchromatin even when repressed by Polycomb proteins. Furthermore, we show that invertebrate animals lacking DNA methylation, such as fruit flies and nematodes, also have divergence in gene types: some genes are regulated by Polycomb proteins, while others are regulated by heterochromatin formation. Altogether, our study establishes gene type divergence and the resulting dual-mode gene regulation as fundamental features shared in a broad range of higher eukaryotic species.

Misexpression of genes lacking CpG islands drives degenerative changes during aging

“…The data described above indicate that not all CGI − genes are simultaneously up-regulated during aging: Some are more commonly misexpressed than others in cells that have disrupted nuclear/chromatin architecture. To investigate what determines this misexpression of some CGI − genes but not others, we tested whether it is affected by chromosomal gene organization, which provides information on the spatial positioning of the genes in the nucleus and their epigenetic regulation ( 8 – 10 , 16 , 17 ). We collected Hi-C chromatin conformation data generated from 28 healthy, young adult mouse tissues and cells ( n = 191; table S5A) and performed principal components analysis (PCA) to classify genomic regions in each sample as heterochromatin or euchromatin [represented by low or high PC1 values, respectively ( 21 )].…”

“…The fact that CGI − genes in heterochromatic domains are rarely activated during aging suggests that loss of heterochromatin is not sufficient to activate all associated genes simultaneously but rather requires additional layers of regulation. As our prior studies demonstrated that CGI − genes are generally regulated by local binding of transcription factors (TFs) ( 8 – 10 ), we further performed TF-binding motif analysis. Notably, known TF-binding motifs were significantly more prevalent in the promoters of CGI − genes in euchromatic domains when compared to those in cHet domains ( P < 2.2 × 10 −16 for both), and this was even more evident in recurrently up-regulated CGI − genes (fig.…”

“…CGI + and CGI − genes were defined as previously described ( 8 – 10 ) with minor modifications. Briefly, we combined two methods of CGI annotation to minimize noise: (i) CGIs defined by sequence characteristics [i.e., GC content, ≥50%; length, >200 bp; and CG observed /CG expected , >0.6; downloaded from the UCSC database ( http://genome.ucsc.edu/ )] ( 61 ) and (ii) experimentally validated CGIs identified by CxxC-affinity purification followed by parallel sequencing ( 62 ).…”

“…In mammals, about 60% of all genes contain CGIs at their promoters (CGI + genes); these genes are broadly expressed throughout the body. The remaining 40% of genes lack CGIs (CGI − genes) and are expressed more specifically in particular tissues ( 8 , 10 , 11 ). CGI + and CGI − genes are regulated by distinct mechanisms ( Fig.…”

“…CGI + and CGI − genes are regulated by distinct mechanisms ( Fig. 1A ) ( 8 – 10 ): CGI − genes generally reside within nuclear lamina–associated heterochromatin when they are transcriptionally inactive and associate with euchromatin when they are transcriptionally active. By contrast, CGI + genes reside in euchromatin regardless of whether they are active or repressed.…”