Imprinting @ NESCENT I am back from a week long trip to The National Evolutionary Synthesis Center (NESCENT), where I was discussing the evolution of parent of origin dependent gene expression (i.e. genomic imprinting). Genomic imprinting is special to me, as it is one of the amazing things that first excited me about studying evolutionary biology.
The kinship theory
The most widely accepted theory of the evolution of genomic imprinting (articulated by David Haig), is known as the kinship theory. According to the kinship theory, a major force influencing the evolution of imprinting is the selective pressure arising from the fact that individuals may have different degrees of relatedness via maternally and paternally inherited alleles (e.g. my maternally inherited alleles are more likely to be present in my maternal aunt, Alice than in my paternal aunt, Rifka).
Since relatedness plays a major role in the evolution of cooperative behavior, the selectively favored level of cooperation may differ at maternally and paternally inherited loci. This idea is largely born out by considering the phenotypic effect of many imprinted genes: alleles expressed when paternally inherited are often growth promoters, while alleles expressed when maternally inherited are often growth enhancers. I have previously argued that this conflict can generate reproductive isolation between plant species, I have also extended the theory to more complex interactions, and suggested how our an extended kinship theory can be evaluated in light of next-gen transcriptome data.
However, not all imprinted genes work this way, and it was clear from the meeting that not everyone in the field is on board with this theory. Hopefully, this discontent will lead to novel tests of the kinship theory, careful evaluation of alternative ideas (e.g. the coadaptation theory), and new theoretical proposals.
Genomic imprinting in the transcriptomic era
Historically, most imprinted regions were discovered by observing gross phenotypic effects in reciprocal crosses. This phenotypic work was then followed up with careful genetic experiments. This laborious method has taught us much about imprinting, but is a lot of work and leaves us with a biased view of imprinting – only imprinted genes with large and observable phenotypic effects can be easily identified. But now we can identify imprinted genes by quantitative RNA sequencing. This new technology provides the potential to discover all imprinted genes in the genome without bias, across many species, cell types, and development.
However, I there are some hazards associated with these data. Andy Clark pointed out that since most RNA-Seq data is associated with an amplification step, there is an extra round of sampling. Thus asking whether there is a significant difference in read counts of alternative alleles from a single locus (without biological replicates, or numerous independent amplifications) is not great evidence for diferences in allelic expression. It seems that the safest way forward is to conduct numerous replicates and to confirm candidate alleles with other technologies (e.g. pyrosequencing).
I also learned a few other things to be cautious about when studying imprinting. Specifically, Ueli Grossniklaus pointed out that it is not always easy to differentiate between maternally contributed RNAs and imprinted loci. In fact, if you’re using inbred lines, it is not possible to tell the differences between these explanations without having some hunch (or data) suggesting that candidate imprinted alleles are not maternally derived RNAs. In animals, this problem could be adressed by using heterozygous parents; however, this makes it more difficult to uncover imprinted loci. Even worse, using heterozygous maternal plants does not solve this issue because maternal RNAs are derived from the haploid mega-gametophyte, which is genotypically equivalent to maternally derived chromosomes in the embryo and endosperm.