Mild drought in the vegetative stage induces phenotypic, gene expression, and DNA methylation plasticity in Arabidopsis but no transgenerational effects

Abstract

There is renewed interest in whether environmentally induced changes in phenotypes can be heritable. In plants, heritable trait variation can occur without DNA sequence mutations through epigenetic mechanisms involving DNA methylation. However, it remains unknown whether this alternative system of inheritance responds to environmental changes and if it can provide a rapid way for plants to generate adaptive heritable phenotypic variation. To assess potential transgenerational effects induced by the environment, we subjected four natural accessions of Arabidopsis thaliana together with the reference accession Col-0 to mild drought in a multi-generational experiment. As expected, plastic responses to drought were observed in each accession, as well as a number of intergenerational effects of the parental environments. However, after an intervening generation without stress, except for a very few trait-based parental effects, descendants of stressed and non-stressed plants were phenotypically indistinguishable irrespective of whether they were grown in control conditions or under water deficit. In addition, genome-wide analysis of DNA methylation and gene expression in Col-0 demonstrated that, while mild drought induced changes in the DNA methylome of exposed plants, these variants were not inherited. We conclude that mild drought stress does not induce transgenerational epigenetic effects.

Keywords: Arabidopsis; drought; epigenetics; maternal effects; methylation; plasticity; transgenerational effects.

Fig. 1.

Schematic representation of the multigenerational experimental design. G, generation of growing plants; P, phenotyping experiment.

Fig. 2.

Descriptive analysis of growth curves for the projected rosette area (PRA) in the first generation (G1) for the five accessions grown under control or mild drought conditions. Individuals where PRA remained below 1 cm² by the end of the experiment or that died prematurely are not shown. (A) Kinetics of shoot size estimated by daily measurements of PRA. Growth curves of plants in control conditions are in blue, those that experienced mild drought are in red. Black lines represent the means for each group. (B) Box-and-whiskers plots showing PRA on day 29 for the plants in each accession×treatment combination. Representative images of the plants are shown. See Fig. 1 for experimental design.

Fig. 3.

The time-dependent pattern of relative growth rates in P3 and P4 as predicted for the first pot on the Phenoscope (the model accounts for pot order) for accession Col-0, based on generalized additive mixed models (gamms). The age of the plants in days is given on the bottom axis: recording started at 9 d after sowing. The panels with grey lines show the raw data for each treatment combination, together with the predicted trajectories of relative growth rate for each combination of treatments in G1/G2 (Memory) and P3 or P4 (Plasticity). P3/P4 treatments are shown in blue (control) and red (drought). The same predicted trajectories are shown in the panels above and to the right, so that pairwise comparisons can easily be made between the ancestral drought treatment and drought in the phenotyping generation (plasticity). The graphs indicate clear growth plasticity in response to mild drought and that plants managed to compensate for the initial drop in relative growth rate shortly after the mild drought had reached a stable level at day 20. Note that there are very few plants with outlying patterns, and that they have very low growth rates only within a restricted age window. See Fig. 1 for experimental design.

Fig. 4.

Effects of maternal traits in Col-0 and Bur-0 for P3 and P4. The data points of both accessions are shown in grey for each phenotyping experiment. Dependencies of the offspring trait values on maternal trait values are shown for log-transformed rosette compactness (see Methods). Linear regressions are estimated for each ancestral drought environment. Data points are circled in blue for individuals with ancestors under drought in G1/G2, and circled in black for individuals with ancestors under control conditions. See Fig. 1 for experimental design.

Fig. 5.

Characterization of stress-induced local changes in DNA methylation. (A) Number of differentially methylated positions (DMPs) and differentially methylated regions (DMRs) for each of the three types of site (CG, CHG, and CHH). (B) Annotation of DMPs and DMRs in relation to genes, transposable elements (TEs), and intergenic regions. (C) Distribution of local gains and losses of DNA methylation across DMPs and DMRs. (D) Example of CHH-DMRs on a TE. (E) Graphical representation of the 18 TE families that showed more DMRs than the random expectation (P<0.01). (F) Overlap (including 500-bp flanking windows) of DMRs induced by mild drought and DMRs found in mutation accumulation (MA) lines according to Becker et al. (2011) and Schmitz et al. (2011), and DMRs induced by hyperosmotic stress according to Wibowo et al. (2016) (G) Hierarchical clustering based on mean CHH methylation levels in wild-type (wt) and mutants for the RdDM (rdr2, ago4, and drd1), CMT2 (ddm1 and cmt2), and DNA demethylation (rdd) pathways in regions overlapping hypo or hypermethylated CHH-DMRs according to Stroud et al., (2013). (H) Abundance of 24-nt siRNAs in random TEs or those with hypo or hypermethylated CHH-DMRs.