Potential Role of DNA Methylation as a Driver of Plastic Responses to the Environment Across Cells, Organisms, and Populations

Abstract

There is great interest in exploring epigenetic modifications as drivers of adaptive organismal responses to environmental change. Extending this hypothesis to populations, epigenetically driven plasticity could influence phenotypic changes across environments. The canonical model posits that epigenetic modifications alter gene regulation and subsequently impact phenotypes. We first discuss origins of epigenetic variation in nature, which may arise from genetic variation, spontaneous epimutations, epigenetic drift, or variation in epigenetic capacitors. We then review and synthesize literature addressing three facets of the aforementioned model: (i) causal effects of epigenetic modifications on phenotypic plasticity at the organismal level, (ii) divergence of epigenetic patterns in natural populations distributed across environmental gradients, and (iii) the relationship between environmentally induced epigenetic changes and gene expression at the molecular level. We focus on DNA methylation, the most extensively studied epigenetic modification. We find support for environmentally associated epigenetic structure in populations and selection on stable epigenetic variants, and that inhibition of epigenetic enzymes frequently bears causal effects on plasticity. However, there are pervasive confounding issues in the literature. Effects of chromatin-modifying enzymes on phenotype may be independent of epigenetic marks, alternatively resulting from functions and protein interactions extrinsic of epigenetics. Associations between environmentally induced changes in DNA methylation and expression are strong in plants and mammals but notably absent in invertebrates and nonmammalian vertebrates. Given these challenges, we describe emerging approaches to better investigate how epigenetic modifications affect gene regulation, phenotypic plasticity, and divergence among populations.

Keywords: DNA methylation; acclimation; environmental stress; epigenetics; gene expression; phenotypic plasticity.

Fig. 1.

In this review, we examine evidence for or against epigenetic regulation of phenotypically plastic responses across cellular, organismal, and population levels. We explore this hypothesis across levels of biological organization by reviewing literature regarding (cellular) associations between differential methylation and expression induced by environmental variation (left panel), (organismal) causal tests of chromatin-modifying enzymes’ effects on phenotypic plasticity (middle panel), and (population) environmentally associated epigenomic variation across natural populations (right panel). We depict a continuous environmental variable corresponding to gene regulatory and phenotype states. The canonical model posits that an environmental state (for example, one of the two colors) may affect gene transcription via changes in DNA methylation (left panel) which may adjust phenotype toward an optimum in that environment (i.e. specific color phenotype in a specifc color environment). This mechanism can be tested by modifying epigenetic factors in organisms and measuring the extent of their phenotypic change between levels of the environmental variable (middle panel). Assuming this model, associations between environment, methylation, and phenotype should arise across natural landscapes and their populations (right panel).

Fig. 2.

Variance in environmentally induced differential DNA methylation (DM) associated with differential expression (DE) across 23 datasets in plants, invertebrates, and vertebrates. Values either represent overlap in genes exhibiting DE and DM as a proportion of DM sites exhibiting DE or correlation coefficients. Small gray points represent singular pairwise contrasts used to estimate DE and DM (e.g. control vs. stress treatments). Large black points depict means ± 95% CI. Fidelity was given to the data as they were described in publication and thus, genomic methods and reporting statistics are not standardized.