MBD2/3 lost its methyl-CpG binding ability in multiple families of Holometabola

Abstract

DNA methylation is sparse in insects, compared to vertebrates. This reduction is even more pronounced in Holometabola, where the loss of DNA methyltransferases DNMT1 and DNMT3 has occurred several times. In some Holometabola, DNA methylation is lost entirely. Methyl-CpG-binding domain (MBD) proteins bind methylated CpGs and are therefore important readers of these epigenetic marks. We hypothesize that the evolutionary reduction of genome-wide methylation may be paralleled by changes to MBD proteins. Among insects, only a single MBD family member, MBD2/3, is known. Two isoforms of MBD2/3 have been identified in Bombyx mori, MBD2/3-L and MBD2/3-S. The long isoform MBD2/3-L contains a complete MBD domain spanning the first two exons, whereas the short isoform MBD2/3-S lacks the second exon and therefore half of its MBD domain. It has been reported that only the MBD2/3-L isoform is able to bind methyl-CpGs. In this study, we analyzed transcriptomic and genomic sequence data across holometabolous orders to identify MBD2/3 genes and their isoforms. Our findings reveal that MBD2/3 is highly conserved in sequence and gene structure. While both isoforms are present in most hemimetabolous orders, the long isoform MBD2/3-L, capable of binding methylated CpG, has been lost in multiple holometabolous orders. The results suggest that MBD2/3 has lost its ability to bind methyl-CpG in several insect orders, with several independent losses in Holometabola. This occurred through different changes to the MBD domain, from sequence divergence within the domain to the absence of half of the MBD domain. These losses may be linked to the reduced levels of CpG DNA methylation.