Genetic variation is generally regarded as a prerequisite for evolution. In principle, epigenetic information inherited independently of DNA sequence can also enable evolution, but whether this occurs in natural populations is unknown. Here we show that single-nucleotide and epigenetic gene body DNA methylation (gbM) polymorphisms explain comparable amounts of expression variance in Arabidopsis thaliana populations. We genetically demonstrate that gbM regulates transcription, and we identify and genetically validate many associations between gbM polymorphism and the variation of complex traits: fitness under heat and drought, flowering time and accumulation of diverse minerals. Epigenome-wide association studies pinpoint trait-relevant genes with greater precision than genetic association analyses, probably due to reduced linkage disequilibrium between gbM variants. Finally, we identify numerous associations between gbM epialleles and diverse environmental conditions in native habitats, suggesting that gbM facilitates adaptation. Overall, our results indicate that epigenetic methylation variation fundamentally shapes phenotypic diversity in a natural population.

Results from the high throughput drug screen in the non-isogenic OCCC cell line panel, see also Figure 1

siRNA library gene list for dasatinib resistance screen, see also figure 5

MS Stats output from dasatinib bead proteomics, see also Figure 3

Results from the dasatinib target siRNA screen, see also Figure S6

ARID1A mutation status in OCCC panel determined by whole exome sequencing

MS Stats output from dasatinib bead proteomics, see also Figure 3

Methylation of CG dinucleotides (mCGs), which regulates eukaryotic genome functions, is epigenetically propagated by Dnmt1/MET1 methyltransferases. How mCG is established and transmitted across generations despite imperfect enzyme fidelity is unclear. Whether mCG variation in natural populations is governed by genetic or epigenetic inheritance also remains mysterious. Here, we show that MET1 de novo activity, which is enhanced by existing proximate methylation, seeds and stabilizes mCG in Arabidopsis thaliana genes. MET1 activity is restricted by active demethylation and suppressed by histone variant H2A.Z, producing localized mCG patterns. Based on these observations, we develop a stochastic mathematical model that precisely recapitulates mCG inheritance dynamics and predicts intragenic mCG patterns and their population-scale variation given only CG site spacing. Our results demonstrate that intragenic mCG establishment, inheritance, and variance constitute a unified epigenetic process, revealing that intragenic mCG undergoes large, millennia-long epigenetic fluctuations and can therefore mediate evolution on this timescale.

Pearson's correlation and Z prime values for OCCC panel demonstrating robust reproducibility and dynamic range from the high-throughput drug screen

Additional compounds included in the isogenic high-throughput drug screen

Grass pea (Lathyrus sativus L.) is a rich source of protein cultivated as an insurance crop in Ethiopia, Eritrea, India, Bangladesh, and Nepal. Its resilience to both drought and flooding makes it a promising crop for ensuring food security in a changing climate. The lack of genetic resources and the crop's association with the disease neurolathyrism have limited the cultivation of grass pea. Here, we present an annotated, long read-based assembly of the 6.5 Gbp L. sativus genome. Using this genome sequence, we have elucidated the biosynthetic pathway leading to the formation of the neurotoxin, β-L-oxalyl-2,3-diaminopropionic acid (β-L-ODAP). The final reaction of the pathway depends on an interaction between L. sativus acyl-activating enzyme 3 (LsAAE3) and a BAHD-acyltransferase (LsBOS) that form a metabolon activated by CoA to produce β-L-ODAP. This provides valuable insight into the best approaches for developing varieties which produce substantially less toxin.

Results from the siRNA kinase and tumour suppressor dasatinib resistance screen, see also Figure 5

Pearson's correlation and Z prime values for isogenic cell lines demonstrating robust reproducibility and dynamic range from the high-throughput drug screen, see also Figure 1

Figure legends for Supplementary figures and supplementary methods

Supplementary Figure 1 ARID1A mutations in the panel of OCCC tumour cell line models with corresponding protein expression. Supplementary Figure 2 Un-cropped western blots from main and supplementary figures. Supplementary Figure 3 ARID1A selective effects from the high throughput drug screen. Supplementary Figure 4 Inhibitors of the PI3K/mTOR signalling pathway in the panel of OCCC cell lines and ARID1A selectivity. Supplementary Figure 5 Dasatinib is a synthetic lethal drug in ARID1A mutant OCCC tumour cell line models - see also Figure 2. Supplementary Figure 6 Dasatinib sensitivity in ARID1A mutant isogenic HCT116 colorectal tumour cell line model. Supplementary Figure 7 Dasatinib siRNA screen results. Supplementary Figure 8 Apoptosis assay in four OCCC cell line models, see also Figure 4. Supplementary Figure 9 Dasatinib sensitivity in ARID1A mutant OCCC is dependent upon G1/S checkpoint effectors, see also Figure 5. Supplementary Figure 10 Determining the optimal method of dasatinib delivery in vivo.

Methylation of CG dinucleotides (mCG), which regulates eukaryotic genome functions, is epigenetically propagated by Dnmt1/MET1 methyltransferases. How mCG is established and transmitted across generations despite imperfect enzyme fidelity remains mysterious. Here we show that MET1 de novo activity, which is enhanced by existing proximate methylation, seeds and stabilizes mCG in Arabidopsis thaliana genes. MET1 activity is restricted by active demethylation and suppressed by histone variant H2A.Z. Based on these observations, we develop a mathematical model that precisely recapitulates mCG inheritance dynamics and predicts intragenic mCG patterns and their population-scale variation given only CG site spacing as input. The model reveals that intragenic mCG undergoes large, millennia-long epigenetic fluctuations, and can therefore mediate evolution on this timescale. Our results demonstrate how genic methylation patterns are created, reconcile imperfect mCG maintenance with long-term stability, and establish a quantitative model that unifies the establishment and epigenetic inheritance of mCG. Highlights MET1 mediates a unified process of mCG establishment and maintenance within genes ROS1 and H2A.Z regulate the epigenetic dynamics of genic mCG A mathematical model predicts genic mCG patterns and their population variance Genic mCG undergoes large epigenetic fluctuations that can last thousands of years

Genetic variation is regarded as a prerequisite for evolution. Theoretical models suggest epigenetic information inherited independently of DNA sequence can also enable evolution. However, whether epigenetic inheritance mediates phenotypic evolution in natural populations is unknown. Here we show that natural epigenetic DNA methylation variation in gene bodies regulates genes expression, and thereby influences the natural variation of complex traits in Arabidopsis thaliana . Notably, the effects of methylation variation on phenotypic diversity and gene expression variance are comparable with those of DNA sequence polymorphism. We also identify methylation epialleles in numerous genes associated with environmental conditions in native habitats, suggesting that intragenic methylation facilitates adaptation to fluctuating environments. Our results demonstrate that methylation variation fundamentally shapes phenotypic diversity in natural populations and provides an epigenetic basis for adaptive Darwinian evolution independent of genetic polymorphism.### Competing Interest StatementThe authors have declared no competing interest.

The plant microbiome plays a vital role in determining host health and productivity. However, we lack real-world comparative understanding of the factors which shape assembly of its diverse biota, and crucially relationships between microbiota composition and plant health. Here we investigated landscape scale rhizosphere microbial assembly processes in oilseed rape (OSR), the UK’s third most cultivated crop by area and the world's third largest source of vegetable oil, which suffers from yield decline associated with the frequency it is grown in rotations. By including 37 conventional farmers’ fields with varying OSR rotation frequencies, we present an innovative approach to identify microbial signatures characteristic of microbiomes which are beneficial and harmful to the host. We show that OSR yield decline is linked to rotation frequency in real-world agricultural systems. We demonstrate fundamental differences in the environmental and agronomic drivers of protist, bacterial and fungal communities between root, rhizosphere soil and bulk soil compartments. We further discovered that the assembly of fungi, but neither bacteria nor protists, was influenced by OSR rotation frequency. However, there were individual abundant bacterial OTUs that correlated with either yield or rotation frequency. A variety of fungal and protist pathogens were detected in roots and rhizosphere soil of OSR, and several increased relative abundance in root or rhizosphere compartments as OSR rotation frequency increased. Importantly, the relative abundance of the fungal pathogen Olpidium brassicae both increased with short rotations and was significantly associated with low yield. In contrast, the root endophyte Tetracladium spp. showed the reverse associations with both rotation frequency and yield to O. brassicae, suggesting that they are signatures of a microbiome which benefits the host. We also identified a variety of novel protist and fungal clades which are highly connected within the microbiome and could play a role in determining microbiome composition. We show that at the landscape scale, OSR crop yield is governed by interplay between complex communities of both pathogens and beneficial biota which is modulated by rotation frequency. Our comprehensive study has identified signatures of dysbiosis within the OSR microbiome, grown in real-world agricultural systems, which could be used in strategies to promote crop yield.

SUMMARYThe photosynthetic capacity of mature leaves increases after several days’ exposure to constant or intermittent episodes of high light (HL) and is manifested primarily as changes in chloroplast physiology. How this chloroplast‐level acclimation to HL is initiated and controlled is unknown. From expanded Arabidopsis leaves, we determined HL‐dependent changes in transcript abundance of 3844 genes in a 0–6 h time‐series transcriptomics experiment. It was hypothesized that among such genes were those that contribute to the initiation of HL acclimation. By focusing on differentially expressed transcription (co‐)factor genes and applying dynamic statistical modelling to the temporal transcriptomics data, a regulatory network of 47 predominantly photoreceptor‐regulated transcription (co‐)factor genes was inferred. The most connected gene in this network was B‐BOX DOMAIN CONTAINING PROTEIN32 (BBX32). Plants overexpressing BBX32 were strongly impaired in acclimation to HL and displayed perturbed expression of photosynthesis‐associated genes under LL and after exposure to HL. These observations led to demonstrating that as well as regulation of chloroplast‐level acclimation by BBX32, CRYPTOCHROME1, LONG HYPOCOTYL5, CONSTITUTIVELY PHOTOMORPHOGENIC1 and SUPPRESSOR OF PHYA‐105 are important. In addition, the BBX32‐centric gene regulatory network provides a view of the transcriptional control of acclimation in mature leaves distinct from other photoreceptor‐regulated processes, such as seedling photomorphogenesis.

Cytosine methylation is an epigenetically heritable DNA modification common in plant and animal genes, but the functional and evolutionary significance of gene body methylation (gbM) has remained enigmatic. Here we show that gbM enhances gene expression in Arabidopsis thaliana. We also demonstrate that natural gbM variation influences drought and heat tolerance and flowering time by modulating gene expression, including that of Flowering Locus C (FLC). Notably, epigenetic variation accounts for as much trait heritability in natural populations as DNA sequence polymorphism. Furthermore, we identify gbM variation in numerous genes associated with environmental variables, including a strong association between flowering time, spring atmospheric NO2 – a by-product of fossil fuel burning – and FLC epialleles. Our study demonstrates that gbM is an important modulator of gene expression, and its natural variation fundamentally shapes phenotypic diversity in plant populations. Thus, gbM provides an epigenetic basis for adaptive evolution independent of genetic polymorphism.