Parental methylation mediates how progeny respond to environments of parents and of progeny themselves

Abstract

Background and aims: Environments experienced by both parents and offspring influence progeny traits, but the epigenetic mechanisms that regulate the balance of parental vs. progeny control of progeny phenotypes are not known. We tested whether DNA methylation in parents and/or progeny mediates responses to environmental cues experienced in both generations.

Methods: Using Arabidopsis thaliana, we manipulated parental and progeny DNA methylation both chemically, via 5-azacytidine, and genetically, via mutants of methyltransferase genes, then measured progeny germination responses to simulated canopy shade in parental and progeny generations.

Key results: We first found that germination of offspring responded to parental but not seed demethylation. We further found that parental demethylation reversed the parental effect of canopy in seeds with low (Cvi-1) to intermediate (Col) dormancy, but it obliterated the parental effect in seeds with high dormancy (Cvi-0). Demethylation did so by either suppressing germination of seeds matured under white-light (Cvi-1) or under canopy (Cvi-0), or by increasing the germination of seeds matured under canopy (Col). Disruption of parental methylation also prevented seeds from responding to their own light environment in one genotype (Cvi-0, most dormant), but it enabled seeds to respond to their own environment in another genotype (Cvi-1, least dormant). Using mutant genotypes, we found that both CG and non-CG DNA methylation were involved in parental effects on seed germination.

Conclusions: Parental methylation state influences seed germination more strongly than does the progeny's own methylation state, and it influences how seeds respond to environments of parents and progeny in a genotype-specific manner.

Keywords: Arabidopsis; DNA methylation; epigenetics; germination; parental effects; phenotypic plasticity.

Fig. 1.

Experiment 1: effects of parental vs. progeny DNA demethylation on seed germination in Col. Seed treatment is indicated on the x-axis, with control (water only) on the left and demethylated treatment (‘5-azacytidine’) on the right. Parental treatment is specified by the fill; see key. Median germination proportion is shown as the black horizontal line within bars, quartiles are shown as bars and the range is shown as vertical lines. The horizontal brackets above the bars indicate the significance of the comparison between parental treatments (black vs. blue bars). There were no significant differences between seed treatments. Significant effects of parental treatment on germination are depicted with ***P < 0.001. See Table 2 for the analysis of variance.

Fig. 2.

Experiment 2: reaction norms of germination proportion in response to parental demethylation treatment, parental canopy and seed canopy. Parental canopy treatment is shown. Mean germination proportions with standard error bars are plotted. Error bars represent the 95 % confidence interval. Each panel shows a separate genotype. Solid lines indicate parental control treatment with water (‘Water’), and dotted lines indicate parental treatment with 5-azacytidine (‘Azacytidine’). Black lines indicate progeny white light, and green lines indicate progeny shade with green filter. Tables 3 and 4 present the ANOVA of the generalized linear model and Tukey’s pairwise comparisons, respectively.

Fig. 3.

Experiment 3: reaction norms of germination response to demethylation treatment via genetic or chemical disruption, parental canopy and seed canopy. Parental light environment is shown on the x-axis. Mean germination proportion is plotted, and error bars represent 95 % confidence intervals. Line type represents seed canopy treatment, with solid lines indicating white light and dashed lines indicating green filter. Line colour indicates genotype: the wildtype control is in black (‘Col’), wildtype parent treated with 5-azacytidine chemical demethylation is in orange (‘Col Aza’), and mutant genotypes in blue (see key). Tables 5 and 6 present results for the ANOVA of the generalized linear model and Tukey’s pairwise comparisons, respectively.