Epigenetic memory is a phenomenon of trans-generational, altered trait inheritance without changes to DNA sequence. Our present ability to predict or direct epigenomic behavior is extremely limited, even though epigenetic factors participate in nearly all aspects of multicellular development, environmental stress response, and disease development. There are critical gaps in our current knowledge of DNA methylation patterning, stable epiallele formation, and the relationship of genome-wide epigenomic behavior to gene expression and phenotype in both plant and animal systems. We have developed a system that will directly address these questions. Our long-term goals are to decode the heritable epigenome, and its relationship to organismal phenotype, particularly in response to stress. What distinguishes our proposed research is the availability of robust biologicals in the model plant Arabidopsis to impose artificial stress, recurrent heritable epigenetic memory, and methylome repatterning. These resources emanate from discovery of the MSH1 gene, disruption of which leads to epigenomic reprogramming. Recent data from this system lead to the overarching hypothesis that stress-induced gene expression recruits methylation machinery to gene networks in a non-stochastic manner. To address this hypothesis, we have developed novel genome-wide methylome analysis procedures for high-resolution identification of gene-associated methylation repatterning. These analyses reveal gene networks that are strikingly consistent with phenotype changes, and display repatterning that is intragenic and often subtle, yet reproducible. We have also identified epigenetic components of the DNA methylation and RdDM pathways that are essential to reprogramming based on msh1 double mutant analysis. Building upon strong preliminary data, we propose to pursue three specific aims to characterize trans-generational epigenomic behavior: (1) To delineate stable, de novo epialleles in the Arabidopsis msh1 model system, exploiting a five-generation memory lineage, (2) to develop a mechanistic understanding of stable epiallele formation in response to stress, implementing machine learning and mutant screening, and (3) to test locus-specific mechanics of epiallele establishment, capitalizing on gene relocation to delimit germane local chromatin features. The proposed research will broadly impact the field by providing the first example of inducible epigenomic reprogramming in a non-stochastic pattern that permits machine learning-based predictive modeling and identification of cis-acting sequence features. The results will be pertinent to mammalian systems and, possibly, to diagnostic strategies for diseases with a strong GxE component.
The proposed research has relevance to public health priorities as we see life style diseases, stress and environmental impacts on health, in utero and throughout development, pose greater challenges to conventional medicine. Environmental stress is a multifaceted influence on health, and epigenomic response is an important, under-investigated factor in cancer, neurodevelopmental, metabolic and toxin-related disorders with high GxE effects. Upon conclusion of the study, we will have defined the architectural features of stable, de novo epiallelic variation, assessed its influence on heritable phenotype, and refined a methodology for high resolution analysis of epigenomic data that is necessary to inform the development of early diagnostics and countermeasures.