Differentially methylated cytosines and methylation patterns
We first assessed differentially methylated cytosines (DMCs) between
parental environments (enriched vs poor), using logistic regression on
quantitated normalised data with q-value < 0.01 after multiple
testing correction and >20% minimal CpG methylation
difference (|ΔM|), using methylKit. To test whether
the number of DMCs between environments were different from the expected
by random, we generated with 4,000 random combinations of 16 parental
individuals and tested for the number of DMCs for each combination
following the same parameters as the ones described for the original
grouping.
We then analysed whether the DNA methylation patterns (hypomethylated or
hypermethylated) in the parents were maintained in the offspring. For
this, we classified DMCs in two categories (i) environmentally-induced
(differences in methylation patterns between the parents changed in the
offspring depending on the offspring rearing environment) and (ii)
intergenerational (differences in methylation patterns between parental
environments were maintained in the offspring regardless of their
rearing environment) (Fig. S3). We
set up a threshold of ±10% average methylation score value in the
offspring relatively to its parents to consider whether an individual
epiallele methylation pattern maintained the parental methylation state.
For DMCs classified as intergenerational we identified the genomic
location (within gene body, promoter region (≤2 kb upstream of the
transcription start site (TSS)), or intergenic region (≥2kb upstream of
TSS or downstream the gene bodies).
To test whether the methylation patterns of the offspring on the DMCs
classified as potentially intergenerational were significantly
influenced by the parental environment, we analysed the methylation
score of the offspring for each DMC (as a proportion index) as a
function of the parental environment (enriched or poor), the offspring
environment and their interactions using a generalized linear model with
a binomial link, with multiple testing correction.
The annotated regions affected by these DMCs were used for the gene
ontology enrichment analysis using zebrafish (Danio rerio ) gene
orthologs in PANTHER v. 11 (Mi et
al. 2016). We searched for enrichments across biological process and
pathways ontologies curated for zebrafish. Only genes which matched with
the genes names annotated for zebrafish were included in the gene
ontology analysis.