Context
Methyl-CpG-binding domain protein 2 (MBD2) binds methylated CpG in vitro or in vivo, and recruit the NuRD repressive complex. MBD2 is thought to play a role in the transcriptional repression of genes with a methylated promoter in vertebrates. However in a surprising study [1], a quadruple knock-out of MBD2 along with MECP2, MBD1, and MBD4, (i.e. all known proteins with an active Methyl-DNA Binding Domain) lead to no widespread up-regulation of genes with a methylated promoters. This author carried out a meta-analysis of publicly available transcriptome studies after MBD2 perturbation to see how contradictory this unexpected results was considering the literature.
Methods
Twenty-seven transcriptome datasets were identified from 17 different studies in human or mouse, including 26 datasets with MBD2 removal and one dataset with MBD2 over-expression. The number of up-regulated or down-regulated genes was identified for each study using a differentially expressed gene reanalysis with {limma}, using four different null-hypothesis thresholds of fold-change. MBD2 perturbation was confirmed for 19 of the 26 datasets. It should be noted that this author directly contributed to two of the datasets included in this meta-analysis [2,3].
Results
Eleven of the 18 datasets (61%) with MBD2 removal were showing more gene over-expressions than down-regulations, while the other datasets were showing more down-regulations than over-expressions (4 datasets), or balanced profiles (3 datasets). The number of up-regulated genes after MBD2 removal was globally estimated as 3.2 times (CI 95%: 1.4; 7.3) more than the number of down-regulated genes for a fold change threshold of 2. The only datasets with MBD2 over-expression had more down-regulated genes than upregulated genes. There was no obvious association between the role of MBD2 as a gene repressor and the species, the tissues / cell types, its oncogenic status, nor the perturbation method.
Conclusion
This meta-analysis supports a role for MBD2 in gene repression. This author is still confused about results reported by Kaluscha et al., 2022 [1].