Evolutionary and functional genomics of DNA methylation in maize domestication and improvement

ARTICLE https://doi.org/10.1038/s41467-020-19333-4 OPEN Evolutionary and functional genomics of DNA methylation in maize domestication and improvement 1234567890():,; Gen Xu1,2, Jing Lyu1,2, Qing Li3,4, Han Liu5, Dafang Wang6, Mei Zhang5, Nathan M. Springer Jeffrey Ross-Ibarra 7 & Jinliang Yang 1,2 ✉ 3, DNA methylation is a ubiquitous chromatin feature, present in 25% of cytosines in the maize genome, but variation and evolution of the methylation landscape during maize domestication remain largely unknown. Here, we leverage whole-genome sequencing (WGS) and whole-genome bisulfite sequencing (WGBS) data on populations of modern maize, landrace, and teosinte (Zea mays ssp. parviglumis) to estimate epimutation rates and selection coefficients. We find weak evidence for direct selection on DNA methylation in any context, but thousands of differentially methylated regions (DMRs) are identified population-wide that are correlated with recent selection. For two trait-associated DMRs, vgt1-DMR and tb1-DMR, HiChIP data indicate that the interactive loops between DMRs and respective downstream genes are present in B73, a modern maize line, but absent in teosinte. Our results enable a better understanding of the evolutionary forces acting on patterns of DNA methylation and suggest a role of methylation variation in adaptive evolution. 1 Department of Agronomy and Horticulture, University of Nebraska-Lincoln, Lincoln, NE 68583, USA. 2 Center for Plant Science Innovation, University of Nebraska-Lincoln, Lincoln, NE 68583, USA. 3 Department of Plant Biology, Microbial and Plant Genomics Institute, University of Minnesota, Saint Paul, MN 55108, USA. 4 National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan 430070, China. 5 Key Laboratory of Plant Molecular Physiology, Institute of Botany, Chinese Academy of Sciences, Nanxincun 20, Fragrant Hill, Beijing 100093, China. 6 Division of Math and Sciences, Delta State University, Cleveland, MS 38733, USA. 7 Department of Evolution and Ecology, Center for Population Biology and Genome Center, University of California, Davis, CA 95616, USA. ✉email: NATURE COMMUNICATIONS | (2020)11:5539 | https://doi.org/10.1038/s41467-020-19333-4 | www.nature.com/naturecommunications 1 ARTICLE G NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-19333-4 enomic DNA is tightly packed in the nucleus and is functionally modified by various chromatin marks such as DNA methylation of cytosine residues. DNA methylation is a heritable covalent modification prevalent in most species, from bacteria to humans1,2. In mammals, DNA methylation commonly occurs in the symmetric CG context with exceptions of non-CG methylation in specific cell types, such as embryonic stem cells3, but in plants it occurs in all contexts including CG, CHG, and CHH (H stands for A, T, or C). Genome-wide levels of cytosine methylation exhibit substantial variation across angiosperms, largely due to differences in the genomic composition of transposable elements (TE)4,5, but broad patterns of methylation are often conserved within species6,7. Across plant genomes, levels of DNA methylation vary widely from euchromatin to heterochromatin, driven by the different molecular mechanisms for the establishment and maintenance of DNA methylation in CG, CHG, and CHH contexts8,9. DNA methylation is considered essential to suppress the activity of transposons10, to regulate gene expression11, and to maintain genome stability8. Failure to maintain patterns of DNA methylation in many cases can lead to developmental abnormalities and even lethality12–14. Nonetheless, variation in DNA methylation has been detected both in natural plant15 and human populations16. Levels of DNA methylation can be affected by genetic variation and environmental cues17. In addition, heritable de novo epimutation—the stochastic loss or gain of DNA methylation—can occur spontaneously and has functional consequences18,19. Population methylome studies suggest that the spread of DNA methylation from transposons into flanking regions is one of the major sources of epimutation, such that 20% and 50% of the cis-meQTL (methylation quantitative trait loci) are attributable to flanking structural variants in Arabidopsis7 and maize20. In Arabidopsis, a multi-generational epimutation accumulation experiment21 estimated forward (gain of DNA methylation) and backward (loss of methylation) epimutation rates per CG site at about 2.56 × 10−4 and 6.30 × 10−4, respectively. Other than this Arabidopsis experiment, there are no systematic estimates of the epimutation rates in higher plants (but see recently estimates for poplar and dandelion22), making it difficult to understand the extent to which spontaneous epimutations contribute to methylome diversity in a natural population. As the per-base rates of DNA methylation variation are several orders of magnitude larger than DNA point mutation, conventional population genetic models, which assume infinite sites models, seemed inappropriate for epimutation modeling. As an attempt to overcome the obstacle, Charlesworth and Jain23 developed an analytical framework to address evolution questions for epimutations. Leveraging this theoretical framework, Vidalis et al.24 constructed the methylome site frequency spectrum (mSFS) using worldwide Arabidopsis samples, but they failed to find evidence for selection on genic CG epimutation under benign environments. The confounding effect between DNA variation and methylation variation, as well as the high-scaled epimutation rates become obstacles to further dissect the evolutionary forces in shaping the methylation patterns at different timescales under different environments. Maize, a major cereal crop species, was domesticated from its wild ancestor teosinte (Zea mays ssp. parviglumis) near the Balsas River Valley area in Mexico about 9000 years ago. Genetic studies reveal that the dramatic morphological differences between maize and teosinte are largely due to selection of several major effect loci25. As maize spread across the Americas, many additional loci have played an important role in local adaptation26. Flowering time, a trait that directly affects plant fitness, played a major role in this local adaptation process27–29. Previous research, however, 2 has focused almost entirely on DNA variation, and the contributions of methylation variation to maize domestication and adaptation remain largely elusive. In this work, we collect a set of geographically widespread Mexican landraces and a natural population of teosinte near Palmar Chico, Mexico30, from which we generate genome and methylome sequencing data. In addition, we profile the teosinte interactome using the highly integrative chromatin immunoprecipitation (HiChIP) method. Together with the analysis from previously published genome31, transcriptome32, methylome6, and interactome33 datasets, we estimate epimutation rates and selection pressures across diffe (...truncated)