For the past decade, plant research has improved our understanding of
the distribution and function of eukaryotic DNA methylation at a genome-wide scale.
Significant progress has been made in understanding DNA methylation in the model
plant Arabidopsis thali-ana and more recently in maize,
two species that diverged at the monocot-eudicot
split approximately 150 million years ago.
In their recent article, Robert Schmitz and colleagues analyzed whole-genome bisulfite sequencing
data from 34 diverse flowering plant species to
vastly expand our knowledge of plant methylome patterns [1].
Results from their study showed substantial variation
in the extent and distribution of DNA methylation in angiosperms.
Results from the large-scale methylome analysis
by Schmitz and colleagues also demonstrated some surprising patterns.
For example, results showed that maize is not an extreme example of a highly methylated genome.
Beet (Beta vulgaris) has higher methylation levels than any of the other species assayed,
with particularly high CHH methylation,
seemingly driven by a high percentage of genes that contain repetitive elements.
Among repeats, there was substantial interspecies variation
in the amount of CHG and CHH methylation,
and only CHG methylation correlated with genome size across all species.
CHH islands in gene flanking regions were not restricted to maize
and were found in many other species.
Yet, the positive correlation between CHH islands and gene expression in maize was not universal.
It remains unclear if all regions annotated
as CHH islands are comparable across- or even within-species,
owing to the fairly broad definition of CHH islands.
Schmitz and colleagues analyzed multiple species from the same family,
a powerful aspect of the study
that allowed broader phylogenetic conclusions to be drawn.
For example, Arabidopsis has lower CG methylation than any of the other examined species
but that reduced methylation is not restricted to Arabidopsis.
The six examined species of the Brassicaceae family, of which Arabidopsis is a member,
have distinctly lower levels of CHG and CHH methylation compared to other families.
The grasses (Poaceae) have overall low levels of CHH methylation,
particularly in the inner regions of repeats,
but the CHH methylation that is present is concentrated at high levels
in smaller regions of the genome.
What causes interspecies methylation variation?
In some species there may be differences in the activity of,
or mutations in, DNA methylation machinery.
Schmitz and colleagues have shown previously that Eutrema salsugineum,
which has the lowest levels of
CHG methylation and no CG gene body methylation,
lacks a functional CMT3 enzyme [6].
Genome-wide association studies in Arabidopsis have linked methylation variation to CMT2 [7],
which is absent in maize. Another potent contributor to
interspecies methylation variation is likely to be genomic content,
specifically the percentage of repetitive elements.
The study by Schmitz and colleagues provides several intriguing findings
that warrant follow-up study.
Outside of the grasses, multiple dicots (grape, cassava, wild strawberry, and others)
also had low levels of CHH methylation,
independent of genetic relatedness [1]. The authors speculate that low CHH methylation
could be a result of how these species are propagated agriculturally, through clonal production.
This hypothesis is intriguing in light of evidence that (1) CHH methylation
is partially lost during male gametogenesis but is restored in the embryo [8] and (2)
RdDM acts progressively during reproductive development over multiple generations,
at least in genomes that have undergone massive hypome-thylation [9].
A sexual reproductive phase may be essential to reinforce and preserve methylation patterning.
It is not yet clear if there are functional consequences of the differences in methylation and,
if so, what they might be.
However, the data presented by Schmitz and colleagues
generats many hypotheses for future investigation.
The rules of DNA methylation are not yet fully written.