In today’s world, where choices are a plenty, alternative is a fashionable choice. When it comes to genomic imprinting, traditionalists swear by DNA cytosine methylation (5mC); however, sometimes imprinting turns to alternative epigenetic marks.
Genomic imprinting is typically driven by 5mC at imprinting control regions (ICRs), where it represses the expression of the marked allele and enables parent-of-origin specific gene profiles. When investigating imprinting, researchers normally employ two common approaches. You can go straight for the nucleus via nucleus transplantation to create embryos with two copies of a single parent’s genome (androgenetic/paternal and gynogenetic/maternal). Alternatively, you can take advantage of underlying single nucleotide polymorphisms (SNPs) that differ between parents for allele-specific assays.
H3K27me3 Liberates Imprinting from Its DNA Methylation Dependence
Intrigued by previous observations of genomic imprinting where certain genes display paternal allele-specific expression without maternal DNA methylation, the lab of Yi Zhang at the Howard Hughes Medical Institute (Massachusetts) set forth to discover the mysterious mark behind non-canonical imprinting.
The team made use of their low-input DNase I-sequencing (liDNase-seq) to identify allele-specific sites of accessible chromatin that regulate gene expression. Using mice, Inoue et al. isolated pronuclei from zygotes and, to study morula embryos, made use of nuclear transplantation, hybrids, and androgenetic/gynogenetic embryos. They subsequently integrated their results with data from RNA-seq, H3K27me3 ChIP-seq, and whole-genome bisulfite sequencing (WGBS).
Here’s what they found:
- Allele-specific DNase I hypersensitive sites (DHSs) in zygotes prime later allele-specific gene expression during zygotic genome activation
- Injecting zygotes with Kdm6b mRNA, a H3K27me3-specific demethylase, turns 78 of 431 (18%) of paternal allele-specific sites bi-allelic
- Notably, these sites exhibit DNA hypomethylation in oocytes
- Turning to morula embryos, they discovered that 76 genes with paternal allele-specific DHSs have maternal allele-specific H3K27me3 without DNA methylation
- These genes show paternal expression in the morula embryo
- The findings were verified by injecting Kdm6b mRNA, where, interestingly, genes imprinted by DNA methylation were not affected
Furthermore, this novel type of imprinting differs from canonical imprinting, since even though some genes retain their imprint in the extra-embryonic cell lineage, it’s typically temporary in preimplantation embryos. Overall, this study demonstrates that not all imprinted genes depend on DNA methylation.
5hmC Compliments 5mC in More than One Way
While we’ve now seen that DNA methylation isn’t always needed for imprinting, the lab of David Monk at the Bellvitge Biomedical Research Institute (IDIBELL) in Spain have now revealed that DNA methylation is needed in more than one way.
By coupling oxidative bisulfite conversion (ox-BS) with the 450k (methylation) array, the team interrogated 5-hydroxymethylcytosine (5hmC) profiles. While standard bisulfite conversion doesn’t differentiate 5hmC from 5mC, ox-BS conversion selectively converts 5hmC into 5-formylcytosine (5fC), which isn’t protected from bisulfite like 5mC. Therefore, by subtracting the ox-BS signal from the standard bisulfite signal, you can analyze 5hmC.
Here’s what went down when the team applied these methods to human placenta and compared to previous data from the brain (cerebellum and frontal cortex):
- There are over 17,000 CpGs with 5hmC in the placenta, although this is about 10 times less than the number of sites in the brain
- Bump-hunting for larger differentially methylated regions (DMRs) revealed an enrichment for 5hmC at genomically imprinted regions in both placenta and brain samples
The team confirmed these findings by making use of the T4 β-glucosyltransferase (T4-BGT) assay, which can distinguish 5hmC from 5mC by adding a glucose to 5hmC to prevent digestion by the MspI restriction enzyme. By coupling this assay to PCR genotyping, the team could determine which allele contained 5hmC.
This revealed that:
- 5hmC often overlaps the allelic region with parent-of-origin specific 5mC, which puts it at a site that represses gene expression.
- Showing off imprinting’s versatility, the assay found tissue-specific monoallelic 5hmC outside of the ICRs and on the actively transcribed alleles of the only two large clusters of imprinted snoRNAs in the brain. The snoRNAs derived from Snrpn-Ube3a and Dlk1-Dio3, which also contains a large cluster of miRNAs, are processed from lncRNA host genes.
In addition to reminding you to think twice when interpreting your standard bisulfite conversion results, this study highlights the complexities of DNA methylation and its oxidative derivatives. Finally, in the spirit of open science, the team has made their data available online.
Taken together, these two studies demonstrate the versatile nature of genomic imprinting and show that while imprinting loves to make use of 5mC, there is always more than meets the eye.
Go learn how to leave your mark over at Nature and Epigenetics, July 2017