Major alleles of CDCA7α shape CG-methylation in Arabidopsis thaliana

Major alleles of CDCA7α shape CG-methylation in Arabidopsis thaliana | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Major alleles of CDCA7α shape CG-methylation in Arabidopsis thaliana View ORCID Profile Pierre Bourguet , Zdravko J Lorković , Darya Kripkiy Casado , Valentin Bapteste , Chung Hyun Cho , Anna Igolkina , View ORCID Profile Cheng-Ruei Lee , Magnus Nordborg , Frédéric Berger , Eriko Sasaki doi: https://doi.org/10.1101/2025.09.03.673934 Pierre Bourguet 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Pierre Bourguet For correspondence: pierre.bourguet{at}gmi.oeaw.ac.at frederic.berger{at}gmi.oeaw.ac.at sasaki.eriko.997{at}m.kyushu-u.ac.jp Zdravko J Lorković 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site Darya Kripkiy Casado 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site Valentin Bapteste 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site Chung Hyun Cho 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site Anna Igolkina 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site Cheng-Ruei Lee 2 Institute of Ecology and Evolutionary Biology, National Taiwan University Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Cheng-Ruei Lee Magnus Nordborg 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site Frédéric Berger 1 Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter (VBC) , Vienna, Austria Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: pierre.bourguet{at}gmi.oeaw.ac.at frederic.berger{at}gmi.oeaw.ac.at sasaki.eriko.997{at}m.kyushu-u.ac.jp Eriko Sasaki 3 Faculty of Science, Kyushu University , Fukuoka, Japan Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: pierre.bourguet{at}gmi.oeaw.ac.at frederic.berger{at}gmi.oeaw.ac.at sasaki.eriko.997{at}m.kyushu-u.ac.jp Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF Abstract DNA methylation is a key epigenetic mark that impacts gene expression and represses transposable elements (TEs) in eukaryotes. Numerous examples of cis -elements targeted by DNA methylation, particularly at CG sites (mCG), have been reported to be under selective pressure in animals and plants. By contrast, there is limited knowledge of trans -regulators of mCG leading to adaptation. Here, using genome-wide association studies, we identify CELL DIVISION CYCLE-ASSOCIATED PROTEIN 7 ALPHA ( CDCA7α ) as a trans -regulator of mCG in natural populations of Arabidopsis thaliana . CDCA7α and its paralog, CDCA7β, directly bind to the chromatin remodeler DECREASE IN DNA METHYLATION 1 (DDM1), which facilitates access of methyltransferases to DNA. CDCA7α/β selectively regulates mCG and minimally impacts other DDM1-dependent processes such as non-CG methylation and histone variant deposition. We identify the cis -regulatory sequence modulating CDCA7α expression in natural populations and determining the degree of mCG and TE silencing. The geographic distribution of CDCA7α alleles suggests that new alleles have repeatedly expanded to novel ecological niches, indicating a potential role in local adaptation. Altogether, our findings provide new insight into how changes in global DNA methylation levels through transcriptional regulation of the epigenetic machinery have the capacity to facilitate local adaptation. Introduction 5-methylcytosine (5mC) is the predominant form of DNA modification in multicellular eukaryotes ( Boulias and Greer, 2022 ; Romero Charria et al ., 2024 ). 5mC regulates gene expression during development and preserves genome integrity by repressing transposable elements (TEs) ( Smith and Meissner, 2013 ; Schmitz, Lewis and Goll, 2019 ). While conserved molecular pathways control the genomic distribution of 5mC ( Law and Jacobsen, 2010 ; Greenberg and Bourc’his, 2019 ), the epigenome varies between individuals due to genetic differences, environmental influences and stochastic catalytic activity ( Kader and Ghai 2017 ; Fraga et al. 2005 ; Jeltsch and Jurkowska 2014 ; Johannes and Schmitz 2019 ; Briffa et al. 2023 ; Richards 2006 ). In many organisms, epigenome variation contributes to phenotypic variation at the cellular, individual and population level ( Becker et al ., 2011 ; Chatterjee et al ., 2015 ; Quadrana and Colot, 2016 ; Zheng et al ., 2017 ; Schmid et al ., 2018 ; Greenberg and Bourc’his, 2019 ; Venney et al ., 2023 ). Key remaining questions are how this epigenome variation was established and whether it contributes to adaptive evolution in nature. Proteins essential for 5mC maintenance are well characterized in plants ( Zhang, Lang and Zhu, 2018 ). Methylation of CG dinucleotides (mCG) is maintained by METHYLTRANSFERASE 1 (MET1), while non-CG methylation (mCH, where H is A, T or C) requires the chromomethyltransferases (CMT) CMT2 and CMT3, and the RNA-directed DNA methylation (RdDM) pathway. MET1 and CMT2/3 rely on the chromatin remodeler DECREASE IN DNA METHYLATION 1 (DDM1), which facilitates methyltransferase access to DNA by remodeling nucleosomes. In wild plant populations, 5mC levels are often associated with local climates (Foust et al. 2016; Platt et al. 2015; Sammarco et al. 2024; Dubin et al. 2015 ; Rodríguez et al. 2022). The 1001 Epigenomes Project revealed 5mC variation for 1107 worldwide accessions of Arabidopsis thaliana , and genome-wide association studies (GWAS) demonstrated that at least some of it had a genetic basis ( Kawakatsu et al ., 2016 ). mCH variation is controlled by genetic variation at trans -acting modifiers, including CMT2 , CMT3 , and NRPE1 , a subunit of RNA polymerase V ( Kawakatsu et al ., 2016 ; Sasaki et al ., 2019 , 2022 ). These regulators influence TE activity and their allelic distribution shapes geographic clines ( Baduel et al ., 2021 ; Sasaki et al ., 2022 ). Trans regulators of mCG variation in gene bodies have been studied ( Zhang et al ., 2020 ; Briffa et al ., 2023 ; Pisupati et al ., 2023 ), although the function of gene body methylation remains debated ( Muyle et al ., 2022 ). In contrast, the genetic basis of mCG variation at TEs remains unknown, despite its critical role in transcriptional silencing and genome stability. Here, we investigated trans regulators of mCG variation at TEs in natural A. thaliana lines. Using GWAS, we identified CDCA7α , an ortholog of the human Cell division cycle-associated protein 7 ( Thijssen et al ., 2015 ), as a major trans regulator of mCG variation. We show that CDCA7α and its paralog, CDCA7β, act as facultative activators of DDM1 and specifically promote its function in depositing mCG. In natural populations, the highly diverged CDCA7α promoter region modulates its expression, thus influencing the degree of mCG and TE silencing genome-wide. New CDCA7α alleles emerged multiple times after speciation and accompanied expansion to new ecological niches, suggesting a role of mCG optimization in local adaptation. Results Natural TE mCG is controlled by genetic variants of CDCA7α To explore the genetic architecture of mCG variation, we conducted GWAS using the 1001 Epigenomes data set ( n =774) ( Kawakatsu et al ., 2016 ). We focused on mCG levels from two domains of chromatin in which TEs are repressed by distinct mechanisms: euchromatic regions, represented by RdDM-targeted TEs, and heterochromatic regions, represented by CMT2-targeted TEs ( Kawakatsu et al. 2016 ; see also methods) ( Fig. 1a , S1a ). Download figure Open in new tab Figure 1. The genetic basis of CG methylation. a, GWAS for mCG on CMT2-targeted TEs. b, Zoom-in figures show the highest peak (left) and the DDM1 region (right). c, Association for mCG on individual TEs. Associations with -log 10 p -value > 6 are shown. Green triangles indicate centromeric regions. d, Simplified phylogenetic tree of CDCA7 in vertebrates and plants. The full tree is shown in Fig. S3. e, Differential mCG levels of TEs associated with chr4:17486863 or chr4:17497441 for loss-of-function mutants. Lower-case letters indicate significant differences as determined using Tukey’s HSD test ( p < 0.05). While no genetic variant was associated with average mCG in RdDM-targeted TEs, several peaks were detected in GWAS for average mCG levels in CMT2-targeted TEs on chromosomes (chr) 4 and 5 ( Fig. 1a , S1a-e, g, Supplementary Table S1 ). The association on the right end of chr 5 (chr5:26657906) was not statistically significant but was located just 4.8 kbp downstream of the coding region of DDM1 and covered the coding to the 3’ region of DDM1 (-log 10 p -value=5.80; Minor Allele Frequency (MAF)=47%) ( Fig. 1b , S1j ). This peak may reflect several causative alleles and one allele is associated with a SNP on chr5:26649074 resulting in an amino acid substitution (V9F) in the first exon of DDM1 . This amino acid is not conserved across Angiosperms ( Fig. S2a ), does not belong to a domain of known function ( Fig. S2b ), and is not represented in available cryo-EM structures. Given the known major role of DDM1 in the control of mCG at CMT2-dependent TEs ( Zemach et al ., 2013 ; Akinmusola, Wilkins and Doughty, 2023 ), a peak in DDM1 is not unexpected, and the association is unlikely to be spurious. The significant peak on chr4 spanned a 15 kbp region from chr4:17482000 to 17497000, and appeared to involve two distinct haplotypes ( Fig. 1b ). One haplotype was associated with the highest peak (chr4:17497441; -log 10 p-value=7.6; MAF=5.3%), and a complete subset of the other haplotype was associated with the other SNP chr4:17486863 (-log 10 p-value=5.47, MAF=30.1%). GWAS accounting for the genetic variants at chr4:17497441 did not affect the peak at DDM1 ; i.e., the associations were independent ( Fig. S1f, h, i ). To gain further insight, we conducted GWAS for mCG levels of 6379 individual TEs commonly found in all 774 lines and containing at least one CG site ( Sasaki et al ., 2019 ). Strong associations were primarily observed in cis (diagonal), suggesting either direct epigenetic inheritance of mCG, or that genetic variation within TEs influences mCG levels, consistent with a previous study on differential mCG variation ( Hüther et al ., 2022 ). In addition, the vertical lines on the graph identified clear trans regulation of mCG levels in TEs distributed around pericentromeric regions associated with the chromosome 4 SNPs discussed above ( Fig. 1c ). Chr4:17486863 and chr4:17497441 showed associations with 229 and 116 TEs (threshold -log 10 p -value>6, MAF>5%), respectively, with a significant overlap (chr4:17486863: 62% and chr4:17497441: 75%) with CMT2-targeted TEs and almost no overlap (5.2% and 0%) with RdDM-targeted TEs. Two haplotypes represented by chr4:17486863 and chr4:17497441 included several genes, but were centered on AT4G37110 and AT4G37120 . AT4G37120 encodes SWELLMAP 2 (SMP2), a step II splicing factor that works redundantly with the homolog SMP1 ( Liu, Wu and Zhang, 2016 ). AT4G37110 contains a zinc finger domain with homology to human Cell Division Cycle Associated 7 (CDCA7) ( Thijssen et al ., 2015 ), which has been shown to recruit the vertebrate ortholog of DDM1, “Helicase, Lymphoid-Specific” (HELLS; ( Jenness et al ., 2018 ; Unoki, 2021 ; Shinkai et al ., 2024 ; Wassing et al ., 2024 ), making it an excellent candidate for being the causal gene ( Fig. 1d ). Further evidence for this hypothesis was provided by bisulfite sequencing of loss-of-function mutants for SMP2 , AT4G37110, and the neighbouring gene AT4G37130 as control ( Fig. 1e ). Unlike the other genes, the at4g37110 mutant showed significantly lower mCG levels in the same TEs identified as targets of the trans modifier using GWAS ( Fig. 1e ). These results strongly suggest that allelic variation of AT4G37110 is responsible for the identified association on chromosome 4 and that AT4G37110 is an ortholog of human CDCA7 . CDCA7α and CDCA7β redundantly maintain mCG and heterochromatin features to enforce silencing of TEs In addition to AT4G37110 , AT2G23530 also contains a zinc finger domain with high sequence homology to metazoan CDCA7 , suggesting that there might be two bona fide CDCA7 orthologs in A. thaliana ( Funabiki et al ., 2023 ). Our study confirms this (see below). Phylogenetic analysis revealed that CDCA7 duplication occurred in Brassicaceae ( Fig. S3a, b ), coinciding with a whole-genome duplication event ( Schranz, Mohammadin and Edger, 2012 ). These two putative A. thaliana CDCA7 orthologs share similar expression patterns throughout development ( Fig. S4 ). Given its higher expression, AT4G37110 was designated CDCA7α , while AT2G23530 was named CDCA7β ( Fig. 2a ). According to ancestral sequence reconstruction, the zinc finger domain of CDCA7β is more similar to the ancestral gene than CDCA7α ( Fig. S3c, d ). We found no evidence for allelic variation affecting mCG in CDCA7β . Download figure Open in new tab Figure 2. CDCA7α and CDCA7β redundantly maintain mCG, heterochromatin marks and transcriptional silencing. a, Protein (left) and gene (right) models for CDCA7α and CDCA7β, with sequence identity scores between domains. Secondary structures (boxes) and low confidence regions (black lines, pLDDT < 70) are shown. Orange triangles mark T-DNA insertions and CRISPR-Cas9 gRNAs. b, 5mC in indicated contexts at TEs. c , Heatmap comparing ΔmCG (mutant - WT) for cdca7 null mutations with the effect of CDCA7α natural alleles on mCG, using the association of the SNP at chr4:17486863 with TE mCG ( p-value ). K-mean clusters were defined based on ΔmCG. The WT value is shown as a reference. d, Metaplots of mean levels of histone marks and variants at heterochromatic TEs. e, TEs upregulated in the indicated mutants relative to WT. f, Transcript levels of TEs upregulated in both cdca7α/β mutants, shown with variance-stabilized counts. Lowercase letters denote statistical groups determined by Tukey’s HSD test ( P < 0.05). To investigate the functions of these CDCA7 paralogs, we generated single and double mutants for cdca7α and cdca7β using T-DNA insertions and CRISPR-Cas9, obtaining two independent lines per mutant to control for potential genetic aberrations ( Fig. 2a , S5a ). We compared them with ddm1 mutants of the same generation (see methods), because the defects caused by ddm1 loss of function gradually worsen over consecutive homozygous generations ( Kakutani et al ., 1996 ; Ito et al ., 2015 ). TEs had reduced mCG levels in the two cdca7α mutant alleles, with a stronger effect in cdca7α-2 ( Fig. 2b ). The T-DNA in cdca7α-2 disrupts the coding sequence, whereas the T-DNA in cdca7α-1 is located in the promoter ( Fig. S5a ). Single mutations in CDCA7β did not affect mCG levels, but cdca7α/β double mutants had strongly decreased mCG compared to cdca7α ( Fig. 2b ), demonstrating that CDCA7α and CDCA7β redundantly maintain mCG levels. The cdca7α and cdca7β mutations had only minimal impact on mCHG and mCHH levels, with a synergistic effect in cdca7α/β ( Fig. 2b ). Compared with cdca7α/β , loss of DDM1 had a stronger impact in all contexts ( Fig. 2b ). Mirroring our GWAS results, the impact of cdca7α/β was strongest at CMT2-dependent TEs ( Fig. S5c ). Importantly, TEs with the strongest mCG decrease in cdca7α and cdca7α/β also had the most significant association of the CDCA7α SNP with mCG variation at CMT2-dependent TEs ( Fig. 2c ), indicating that null and natural alleles of CDCAα impact the same targets. Because DDM1 directly controls the genomic distribution of histone H2A.W and H2A.Z, and affects the H3K9me2 and H3K27me1 histone post-translational modifications ( Soppe et al ., 2002 ; Ikeda et al ., 2017 ; Osakabe et al ., 2021 ; Jamge et al ., 2023 ; Lee et al ., 2023 ), we profiled these chromatin marks in ddm1 and two cdca7α/β mutants using ChIP-seq, focusing on heterochromatic TEs (see methods). Our results showed that cdca7α/β mutants exhibited decreased levels of H3K9me2, H3K27me1 and H2A.W, while H2A.Z levels increased, resembling ddm1 but with milder effect ( Fig. 2d ). Furthermore, we quantified the impact of cdca7 mutations on TE silencing using 3’-Tag RNA-seq. Single mutations in cdca7α or cdca7β had negligible effects, but the double mutants showed over 200 upregulated TEs ( Fig. 2e ). The extent of TE upregulation in cdca7α/β was less than in ddm1 mutants ( Fig. 2f ). To further validate these observations, cdca7α/β-1 was complemented with CDCA7α or CDCA7β , fused with an N or C-terminal tag. Constructs with a C-terminal tag partially complemented silencing defects ( Fig. S5d ), confirming a direct role of CDCA7α / β in TE silencing. Interestingly, the N-terminal tag inhibited CDCA7α / β -mediated silencing, suggesting it interfered with a critical CDCA7 function. We found no developmental defects in cdca7α/β mutants and, in contrast to ddm1 ( Kakutani et al ., 1996 ; Ito et al ., 2015 ), inbreeding homozygous mutants for up to eight generations had no evident effect on the plant phenotype ( Fig. S5e ). In summary, CDCA7α and CDCA7β redundantly maintain heterochromatin marks and TE silencing. CDCA7α and CDCA7β appear to have identical functions, but CDCA7α plays a more prominent role than CDCA7β ( Fig. 2b ), possibly due to its higher expression ( Fig. S4 ). CDCA7α and CDCA7β require DDM1 to maintain CG methylation At the molecular level, cdca7α/β null mutants resembled ddm1 ( Fig. 2 ) , suggesting CDCA7α/β and DDM1 operate in a common pathway. To test this genetically, we combined ddm1 and cdca7α/β mutations and measured their impact on 5mC and TE silencing. The triple cdca7α/β ddm1 mutants showed 5mC levels comparable to ddm1 mutants of the same generation ( Fig. 3a, b ). Additionally, introducing any cdca7 mutation alongside ddm1 did not alter TE transcript levels ( Fig. 3c , S6a ). Therefore, we conclude that CDCA7α/β fully operate through DDM1 to regulate 5mC and TE silencing. Download figure Open in new tab Figure 3. CDCA7α and CDCA7β function through DDM1 to maintain mCG. a , Linear correlation of mCG loss at TEs in ddm1 cdca7α/β-2 compared to ddm1 . Pearson correlation coefficients (R 2 ) and a permutation test-derived P -value are indicated. b, Pearson correlation coefficients (R 2 ) of TE 5mC loss in ddm1 and the indicated mutants. All correlations had P < 1e-10. c , Number of transposable elements upregulated in the indicated mutants relative to WT or ddm1 . Mutant combinations of ddm1 and cdca7 were compared to ddm1 mutants of the corresponding generation. d , e , Pearson correlation coefficients (R 2 ) showing the relationship between ddm1 and the indicated mutants in terms of 5mC loss ( d ) and chromatin mark changes ( e ). All correlations had P < 1e-10. f , Distribution of methylation ratios (mutant / WT) across genotypes and 5mC contexts. The x-axis is capped at 2, excluding at most 3.2% of the data. g , Spearman correlations coefficients (ρ) of methylation context density (number of CN sites per bin) with 5mC loss in mutants, calculated in 100 bp bins. h , Metaplot showing mean gene body methylation in CG context. Only gene-body methylated genes are included (n=9938). In all assays, loss of CDCA7α/β had a milder effect than the loss of DDM1 ( Fig. 2 ), suggesting that CDCA7α/β might specifically target a subset of DDM1-dependent TEs. To test this hypothesis, we compared TEs upregulated in both mutants. Although TEs upregulated in cdca7α/β were a subgroup of those upregulated in ddm1 ( Fig. S6b ), 51.8% of TEs upregulated exclusively in ddm1 had increased transcript levels in cdca7α/β , but did not meet significance thresholds ( Fig. S6c ). Additionally, all TEs silenced by DDM1 showed reduced 5mC in cdca7α/β , irrespective of their expression levels ( Fig. S6d ). This demonstrates that CDCA7α/β is required for the maintenance of 5mC at all TEs silenced by DDM1 , not just a specific subset. We extended the analysis to all TEs with mCG in the WT and found that 91.4% of these lost mCG in both cdca7α/β-1 and cdca7α/β-2 ( Fig. S6d ), demonstrating that CDCA7α/β target nearly all TEs. DDM1 has two distinct functions in heterochromatin maintenance: first, depositing H2A.W while preventing deposition of H2A.Z ( Osakabe et al ., 2021 ; Jamge et al ., 2023 ; Zhou et al ., 2023 ) and second, facilitating access of DNA methyltransferase to nucleosomal DNA through nucleosome remodeling ( Zemach et al ., 2013 ; Lyons and Zilberman, 2017 ). Heterochromatic histone modifications also depend on DDM1 ( Soppe et al ., 2002 ; Ikeda et al ., 2017 ; Osakabe et al ., 2021 ; Lee et al ., 2023 ) through unknown mechanisms. CDCA7α/β might therefore differentially affect specific functions of DDM1. To explore this, we compared the effects of ddm1 and cdca7α/β mutations on 5mC, histone variants and modifications. The effects of the mutations correlated strongly for mCG ( Fig. 3d , S6f , R 2 =0.52-0.57) but weakly for mCH (R 2 =0.19-0.24). Relative to these, correlations were intermediate for H3K9me2 ( Fig. 3e , R 2 =0.30-0.37), likely due to H3K9me2 being reliant on both mCH ( Du et al ., 2015 ) and mCG ( Tariq et al ., 2003 ; Mathieu et al ., 2007 ; Deleris et al ., 2012 ; Choi, Lyons and Zilberman, 2021 ). For H3K27me1 and histone variants H2A.W and H2A.Z, correlations were low ( Fig. 3e ), although overall trends resembled those seen in ddm1 ( Fig. S6g ). Therefore, cdca7α/β mutants primarily resembled ddm1 in their effect on mCG and not on deposition of histone variants or histone post-translational modification. Strikingly, cdca7α/β mutants were disproportionately affected in CG context, whereas all 5mC contexts were equally affected in ddm1 ( Fig. 3f ), highlighting the specific role of CDCA7α/β in regulating mCG. Likewise, GWAS pointed towards a distinct role of CDCA7α/β in mCG regulation ( Fig. S1 ). Finally, and in contrast with the mild effect on other marks, loss of CDCA7α/β accounted for 60% of the mCG changes observed in ddm1 mutants ( Fig. S6h ). Taken together, these data demonstrate that CDCA7α/β are primarily involved in mCG maintenance. In met1 mutants, mCG loss leads to depletion of mCHG, mCHH, and H3K9me2, with a concurrent increase in H2A.Z ( Soppe et al ., 2002 ; Tariq et al ., 2003 ; Zilberman et al ., 2008 ; Stroud et al ., 2013 ; Choi, Lyons and Zilberman, 2021 ). This suggests that the moderate changes in levels of mCH, H2A variants and H3 modifications in cdca7α/β mutants are secondary effects of reduced mCG. In support of this interpretation, we found that the changes of mCHG and mCHH in cdca7α/β mutants was anticorrelated with the density of CG sites ( Fig. 3g ). This differed from ddm1 , where the loss of mCH was best correlated with CHG density. We next examined gene bodies, where mCG is decoupled from mCH ( Muyle et al ., 2022 ). The absence of CDCA7α/β mirrored the impact of ddm1 mutations ( Fig. 3h ), indicating that CDCA7α/β is fully required for DDM1-dependent gene body methylation. This is in accordance with a model in which CDCA7α/β specifically regulates mCG. We conclude that CDCA7α/β primarily regulate DDM1’s function in maintaining mCG without directly affecting its other roles. CDCA7 interacts with DDM1 through conserved regions Next, we explored the function of the three CDCA7α/β domains using a genetic complementation strategy. The N-terminal domain contains a highly conserved motif ( Fig. 4a ), the 4CXXC zinc finger domain binds hemimethylated CG dinucleotides ( Hardikar et al ., 2024 ; Shinkai et al ., 2024 ; Wassing et al ., 2024 ) and the C-terminal part of the protein has lower conservation. Download figure Open in new tab Fig 4. Characterization of CDCA7α/β protein domains and DDM1 interaction. a , Conservation of CDCA7α among 315 orthologs in Viridiplantae, with domains represented to scale, including the DDM1-interacting alpha-helices (yellow). A consensus sequence (bottom, n = 37) highlights residues predicted to contact DDM1. b , TE upregulation in cdca7α/β-3 complemented with various CDCA7α constructs. Each data point is a T1 plant, showing the mean log 2 fold change (relative to WT) of TEs upregulated in cdca7α/β-3 , scaled by transgene expression. c , Predicted Aligned Error (PAE) of the modeled interaction between CDCA7α and DDM1, with protein domains indicated. d , Predicted CDCA7α-DDM1 structure, showing predicted contacts ≤ 3 Å colored by PAE. CDCA7α is colored by local confidence, using predicted local distance difference test (pLDDT). e , In vitro co-immunoprecipitation (Co-IP) of recombinant CDCA7α and CDCA7β with DDM1, detected by Western Blot. f , Co-IP of recombinant DDM1 with truncated CDCA7α constructs. g , Predicted interaction of CDCA7α N-terminal alpha helices with DDM1. Contacts ≤ 3 Å are shown, colored by PAE. h , Co-IP of recombinant wild-type and mutant DDM1 proteins with CDCA7α. Once mCG is lost in met1 or ddm1 mutants, it can largely not be restored by complementation ( Finnegan, Peacock and Dennis, 1996 ; Kakutani et al ., 1999 ; Lippman et al ., 2003 ; Saze, Scheid and Paszkowski, 2003 ), likely because of the semi-conservative nature of mCG maintenance. Accordingly, complementation of cdca7α/β-1 was partial and locus-specific ( Fig. S5d ). To bypass this limitation, we introduced a CDCA7α transgene before the complete loss of endogenous CDCA7α/β ( Fig. S7a, b ). We quantified TE transcripts in primary transformants using transcriptomes and took into account transgene expression, since it is highly variable between primary transformants ( Matzke and Matzke, 1998 ). Untransformed cdca7β-1 and cdca7α/β-3 mutants showed no or few reads mapping to the transgenic tag, reflecting background noise ( Fig. 4b ). As expected, a full length CDCA7α transgene restored TE silencing, while sister plants lacking the transgene had elevated TE transcript levels. CDCA7α(ΔC-ter) complemented silencing defects in individuals with high transgene expression, indicating that the C-terminal domain is dispensable for TE silencing. Constructs with an N-terminal deletion were not expressed, preventing conclusions on the role of the N-terminus in silencing ( Fig. 4b ). To test the function of the zinc finger domain, we mutated the conserved cysteines C183 and C209, which coordinate a zinc ion ( Fig. S8a ). The lack of complementation with this construct ( Fig. 4b ) indicates that the CDCA7α zinc finger is essential for TE silencing. To further explore the function of CDCA7α/β domains, we used AlphaFold-Multimer ( Evans et al ., 2022 ) to model potential interactions with DDM1. High confidence interactions were identified between CDCA7α/β N-termini and DDM1, specifically its lobe 1 and H2A.W binding domain ( Fig. 4c-d , S8b ). In vitro immunoprecipitation assays with purified recombinant proteins showed that both CDCA7α and CDCA7β bind to DDM1 ( Fig. 4e ), and these interactions were maintained under more restrictive salt concentration ( Fig. S8c ). The N-terminal region of CDCA7α alone was sufficient to interact with DDM1 ( Fig. 4f ), whereas neither the Zn finger nor the C-terminal part showed an interaction. Remarkably, the N-terminal residues predicted to interact with DDM1 ( Fig. 4g ) were highly conserved across the green lineage ( Fig. 4a ). CDCA7α R35 and N42 were also conserved in model vertebrates and predicted to interact with HELLS (DDM1 ortholog) ( Fig. S8d ). However, R35A and N42L mutations did not disrupt binding to DDM1 ( Fig. S8e ) or CDCA7α silencing function ( Fig. 4b ), indicating that other CDCA7α residues can mediate DDM1 binding in A. thaliana . DDM1 residues predicted to interact with CDCA7α were also conserved ( Fig. S8f ), and their mutation reduced the interaction with CDCA7α ( Fig. 4h ). Since DDM1 activity is restricted by a N-terminal autoinhibitory coiled-coil domain (autoN) ( Lee et al ., 2023 ; Nartey, Goodarzi and Williams, 2023 ), we considered that CDCA7α binding could stimulate DDM1 by displacing the autoN. However, the position of the autoN was predicted with poor confidence relative to DDM1 lobes ( Fig. S8g ) and was not improved by co-modeling CDCA7a/b or a nucleosome ( Fig. 4c, S8h, i ), providing no evidence that CDCA7α stimulates DDM1 activity through autoN displacement. Further modeling suggested that CDCA7α/β paralogs could not bind to each other or themselves, and that DDM1 can bind only one CDCA7α/β paralog at a time ( Fig. S8j ), consistent with CDCA7α/β acting redundantly ( Fig. 2 ). We conclude that the conserved N-terminal domain of CDCA7α/β directly interacts with DDM1 in vitro . In summary, the C-terminal domain of CDCA7α does not contribute to its silencing function, while the zinc finger is essential. In both CDCA7α and vertebrate CDCA7, the zinc finger binds hemi-methylated CG ( Hardikar et al ., 2024 ; Shinkai et al ., 2024 ; Wassing et al ., 2024 ), implying that CDCA7α recruits DDM1 to hemi-methylated CG sites, facilitating mCG deposition after DNA synthesis. This recruitment is mediated by a conserved N-terminal domain of CDCA7, which also interacts with HELLS in vertebrates ( Wassing et al ., 2024 ). Combined with our previous results, our biochemical data indicate that DDM1 interacts directly with either CDCA7α or CDCA7β, and this interaction is crucial for maintaining mCG and repressing TEs. Genetic variation in the CDCA7 α promoter shapes large mCG variation Next, we investigated the molecular causes of the CDCA7α natural alleles. Although protein sequences showed a cysteine to serine mutation at position 182, this cysteine is not conserved in sister species ( Fig. S9 ), nor in eukaryotes ( Funabiki et al ., 2023 ), and it does not coordinate a zinc atom ( Fig. S8a ). There were no amino acid changes in conserved sites within the zinc finger domain or the DDM1-interacting domain, suggesting that a causal effect on protein activity was unlikely ( Fig. S9 ). Therefore, to explore the impact of natural genetic variation on CDCA7α regulation, we performed GWAS for CDCA7α expression using published transcriptome data ( n = 427 lines; ( Kawakatsu et al ., 2016 ; Kornienko et al ., 2023 )). Remarkably, a strong cis effect was detected for CDCA7α expression, with the most significant SNP identified as chr4:17486863 ( CDCA7α-alt a haplotype) ( p -value=7.8E-43, MAF=0.33), the same allele identified in our GWAS for mCG levels ( Fig. 1a-c , 5a ). Additionally, chr4:17497441 non-ref ( CDCA7α-alt b haplotype) was also present in this large peak ( p -value=2.08E-6, MAF=0.06). Both haplotypes were associated with higher CDCA7α expression, two to three times higher than the reference allele Col-0, as well as elevated mCG levels ( Fig. 5b ). Download figure Open in new tab Figure 5. Cis-polymorphism regulates CDCA7α promoter activity and shapes mCG variation a , GWAS for CDCA7 α expression. SNPs identified in GWAS for mCG are highlighted with orange dots. b , The allelic effects of CDCA7 α on mCG and the expression levels. Violin plots show mean mCG levels of CDCA7α-targeted TEs ( n =774 lines) and CDCA7 α expression for each genotype ( n =461 lines). c , Sequence alignment upstream of CDCA7a in three haplotypes. d , A mediation model shows the genetic effect of CDCA7 α alleles on mCG through CDCA7a expression. Black plot, diamond, and rectangle indicate genotype, gene expression, and mCG phenotype, respectively. Arrows show the regulation. e , Effect of CDCA7α promoters on transgene expression, measured by mRNA-seq in complemented cdca7α/β-3 and cdca7α-2 mutants. Transgenic constructs are shown to the left (to scale, except deletions). Each data point is an individual primary transformant with mean values ± standard error shown in black and significance between groups indicated by lowercase letters (one-way ANOVA with Tukey’s post-hoc tests, p < 0.05). f , g , Correlation of transgene expression with mean TE CG methylation ( f ) or mean TE log 2 (fold change), relative to WT ( g ). Pearson correlation coefficients and p -values (permutation test) are shown. To test for a causal link between DNA methylation and CDCA7α expression, we performed a mediation analysis ( Sasaki, Frommlet and Nordborg, 2018 ). This showed that 65.6% of the CDCA7α genetic effect was mediated by CDCA7α expression, suggesting that overexpression of CDCA7α in CDCA7α-alt a and CDCA7α-alt b haplotypes contributes to the divergence of mCG levels ( Fig. 5d ). Alignments of the CDCA7α promoter region using fully assembled genomes revealed that the region is unusually polymorphic. Two indels distinguished CDCA7α-alt a and CDCA7α-alt b from the reference-type haplotype ( CDCA7α-ref ) ( Fig. 5c ). This suggests that different haplotypes in the CDCA7α promoter region change CDCA7α expression, resulting in diverse mCG levels in wild populations of A. thaliana . To test the regulatory effect of the CDCA7α haplotypes on CDCA7 expression, we cloned the CDCA7α-alt a promoter and fused it to the CDCA7α-ref coding sequence to complement cdca7α and cdca7α/β-3 mutants ( Fig. S7a ). We also generated constructs where the indels present in CDCA7α-alt a and CDCA7α-alt b were deleted. In primary transformants, transgene expression was higher with the CDCA7α-alt a promoter, and this effect depended on the presence of a 26 bp indel but not a 4 bp indel ( Fig. 5e ). Importantly, CDCA7α expression positively correlated with mCG levels at TEs ( Fig. 5f ) and inversely correlated with TE expression ( Fig. 5g ), regardless of the mutant background. These findings demonstrate that the 26 bp indel in the CDCA7α promoter regulates its expression, which in turn controls DNA methylation levels in natural A. thaliana populations. The fact that higher expression of CDCA7α led to increased mCG levels further implies that in nature, CDCA7α is a limiting factor in promoting mCG and TE repression. The construct lacking the 26 bp indel, despite producing CDCA7α transcript levels similar to those of the reference promoter ( Fig. 5e ), was associated with reduced mCG ( Fig. S10a ), suggesting that steady-state CDCA7α transcript levels are not sufficient to predict mCG levels. There are six reported CDCA7α transcript isoforms ( Zhang et al ., 2022 ). Under our experimental conditions, we detected three: CDCA7α.2 , CDCA7α.3 , and CDCA7α.5 ( Fig. S10b ). CDCA7α.2 , the most abundant isoform ( Fig. S10b ), encodes the canonical CDCA7α protein, while CDCA7α.3 has no annotated protein product and CDCA7α.5 produces a protein lacking the DDM1-interacting domain. We found that both CDCA7α.2 and CDCA7α.3 were overexpressed in lines with the CDCA7α-alt a promoter, and this overexpression required the 26 bp indel ( Fig. S10c ), consistent with our quantification of all isoforms ( Fig. 5e ). However, deleting the 26 bp indel had no strong effect on the CDCA7α.5 isoform relative to the intact CDCA7α-alt a promoter ( Fig. S10c ), indicating that this indel regulates isoform usage. CDCA7α.5 , which lacks the DDM1-interacting domain but retains an intact zinc finger, could compete with the canonical CDCA7α for binding to target loci, thereby antagonizing DDM1 activity. We hypothesize that an increased abundance of CDCA7α.5 , along with reduced levels of CDCA7α.2 , would contribute to the observed reduction in mCG levels in transgenic lines with CDCA7α-alt a promoters lacking the 26 bp indel ( Fig. S10a ). In conclusion, the three CDCA7α haplotypes influence DNA methylation and TE repression through different mechanisms. The CDCA7α-alt a and CDCA7α-alt b haplotypes, with their unique promoter sequence, drive higher CDCA7α expression and corresponding mCG levels compared to the reference haplotype. Moreover, the 26 bp indel plays a critical role not only in regulating CDCA7α expression but also in controlling the balance of transcript isoforms, which has downstream effects on mCG levels. While no genetic difference between CDCA7α-alt a and CDCA7α-alt b haplotypes was observed in the tested promoter region, additional structural variants that could explain their expression variation likely exist. The CDCA7α -ref allele is derived and predominantly occurs in Europe We investigated the history of the CDCA7α promoter region to gain insight into the evolution of the epigenome in A. thaliana . To understand which form of the CDCA7α promoter is ancestral, we compared the CDCA7α haplotypes with closely related species, including A. lyrata , A. halleri , and Capsella rubella . These species clearly exhibited higher similarity to the CDCA7α-alt a and CDCA7α-alt b haplotypes than to CDCA7α-ref ( Fig. 6a ). The predicted haplotype network supported the conclusion that the CDCA7α-alt a haplotype, causing higher mCG levels, is ancestral, while CDCA7α-ref is derived, and the promoter indels reflect deletions ( Fig. 6b ). Compared with CDCA7α-alt a , CDCA7α-ref has a 12.3% shorter promoter region ( Fig. 6c ). CDCA7α-alt b , which drives the highest mCG levels, accumulated specific SNPs not shared with relative species, and the haplotype network showed that CDCA7α-alt b was derived from CDCA7α-alt a ( Fig. 6b ). Consequently, the regulation of CDCA7α expression has undergone multiple changes, resulting in variation in mCG levels at TEs in heterochromatic regions. Download figure Open in new tab Figure 6. The evolutionary history of CDCA7 α haplotypes. a, Alignment of CDCA7 α regions and the ancestral state. Blue, orange, and gray regions represent the reference, the alternative, and non- A. thaliana alleles, respectively. b , Haplotype network of a 4 kbp region covering the CDCA7 α and SMP2 coding region. c , Variation in CDCA7 α promoter length. d, Population structures of 1323 natural lines with CDCA7 α alleles and the geographic distribution. The phylogenetic tree is a neighbor-joining tree ( Hsu et al., 2019 ) with populations (outer circle) and the genotype of CDCA7 α haplotypes (at the tips). A global map shows an overview of allele distributions. e , Altitudinal cline of alleles and mCG variation. The local map of the Iberian Peninsula indicates the altitudes of summits with triangles, and scatterplots show associations between altitude and CDCA7 α expression (top) and mCG in CMT2-targeted TEs (bottom), with the allele distribution as density plots. f , The effect of CDCA7 α on seed size. A scatter plot shows the association between altitude and seed size, with the allele distribution shown as density plots (reanalyzed data from Vidigal et al. 2016 ). Dot plots (bottom) show parent-of-origin effects of cdca7α/β-2 and ddm1 on seed size, in F1 crosses to WT. Lowercase letters denote statistical groups determined by Tukey’s HSD test ( P < 0.05). The CDCA7α-alt a haplotype was predominantly found in three regions: (1) Northern Sweden, (2) the Iberian peninsula, and (3) Eastern Asia, specifically the Yangtze River Basin population, which is the most southern-eastern edge of native A. thaliana habitats ( Fig. 6d ). Consistent with CDCA7α-alt a being ancestral, its distribution overlaps with the ancestral-type populations named ‘relict’ observed in Northern Sweden and the Iberian peninsula ( 1001 Genomes Consortium, 2016 ; Lee et al ., 2017 ). Relicts have genomes that diverged from those of the current worldwide population, referred to as ‘non-relicts,’ and are considered to have been dominant in post-glacial Eurasia ( 1001 Genomes Consortium, 2016 ; Lee et al ., 2017 ). Non-relicts comprising over 95% of current populations are likely modern populations that replaced the relicts as they spread to central and western Europe through human activity. Most of the lines carrying the CDCA7α-ref were non-relicts ( Fig. 6d ). Although the Yangtze River Basin population is non-relict, expected to have diverged about 60 thousand years ago ( Hsu, Lo and Lee, 2019 ), CDCA7α-alt a was fixed in this relatively new population ( Fig. 6d ). This is in agreement with studies suggesting that genomic introgression occurred from local relicts to the Yangtze River Basin population ( Hsu, Lo and Lee, 2019 ), and the CDCA7α region appears to have been introgressed as shown in our phylogenetic tree ( Fig. 6d ). Additionally, a prior selection scan using the composite likelihood ratio (CLR) in Zou et al (2017) indicated a signature of positive selection over a 15 kbp region that includes CDCA7α , suggesting that this allele may be involved in local adaptation. Unlike CDCA7α-alt a , CDCA7α-alt b was predominantly found in the Spanish population, particularly in the central mountain region of the Iberian Peninsula ( Fig. 6d, e ). This mountainous area stretches from east to west, exhibiting diverse climates due to its wide range of altitudes, up to 3000 m above sea level ( Vidigal et al ., 2016 ). While CDCA7α-alt b appears to have been derived from CDCA7α-alt a , non-relict Spanish populations carry CDCA7α-alt b ( Fig. 6d ), suggesting that CDCA7α-alt b likely emerged after CDCA7α-alt a was introgressed into non-relicts. Across the Spanish population in the Iberian Peninsula, both CDCA7α-alt a and CDCA7α-alt b were distributed at higher altitudes than CDCA7α-ref , resulting in an altitudinal cline of mCG levels on TEs ( Fig. 6e ). If CDCA7α is directly involved in expansion to new environments, CDCA7α -mediated epigenetic changes must impact life-history traits that provide adaptive features. For example, flowering time and seed size are influenced by DNA methylation ( Adams et al ., 2000 ; Kinoshita, 2004 ; Xiao et al ., 2006 ) and contribute to reproductive success in challenging environments ( Alonso-Blanco et al ., 1999 ; Gaudinier and Blackman, 2020 ). While cdca7α/β mutants did not affect flowering time under laboratory conditions ( Fig. S12 ), a maternal effect was observed in seed size ( Fig. 6f ). This differed from ddm1 , which showed the expected parent-of-origin effect ( Adams et al ., 2000 ; Xiao et al ., 2006 ; FitzGerald et al ., 2008 ). Consistent with the mutant phenotype, seed size variation in the Spanish mountain population showed a clear altitudinal cline and was positively associated with the distribution of CDCA7α alleles along altitude ( Fig. 6f ). While seed size is a complex trait controlled by multiple loci ( Alonso-Blanco et al ., 1999 ; Gnan, Priest and Kover, 2014 ), our results suggest mCG changes through CDCA7α alleles potentially contribute, and may play a role in colonizing new environments. Discussion The establishment and maintenance of the epigenome, particularly mCG, a DNA modification that is transgenerationally inherited, has been much debated ( Richards, 2006 ; Charlesworth, Barton and Charlesworth, 2017 ; Baduel and Sasaki, 2023 ). Our study provides novel molecular mechanisms that explain the epigenome variation observed in nature. We identified CDCA7α as a new trans regulator of mCG in plants and showed that its expression variation has strongly shaped the epigenome of natural populations in A. thaliana . The ancestral CDCA7α promoter allele, present in relict populations, drives higher CDCA7α expression compared to the derived alleles found in non-relict populations across western and central Europe, including the reference strain Col-0. A 26 bp deletion in the derived promoter reduces CDCA7α expression and increases the production of a truncated CDCA7α isoform, with the potential to inhibit DDM1 activity. Together, these changes result in lower mCG levels and reduced TE repression in current worldwide populations compared to the ancestral form. In addition to natural variation in CDCA7α, the retention of two CDCA7 copies following whole genome duplication ( Fig. S3 ) may have influenced mCG dynamics. In jawed vertebrates, CDCA7 is also duplicated ( Funabiki et al ., 2023 ) and the absence of one CDCA7 paralog in mice had no effect on TE silencing ( Vukic et al ., 2024 ), suggesting that CDCA7 paralogs also have redundant functions in animals. In A. thaliana , CDCA7α has a stronger impact on mCG than CDCA7β . This difference is unlikely due to the divergence in their C-terminal domains, as this region was found to be dispensable for TE silencing. Instead, the stronger impact of CDCA7α likely results from its higher expression levels. The redundant roles of CDCA7α and CDCA7β indicate that CDCA7 duplication may have increased CDCA7 protein dosage or enhanced robustness against genetic and environmental challenges ( Kuzmin, Taylor and Boone, 2022 ). How does CDCA7 promote mCG? Studies in Xenopus egg extracts and mouse embryonic stem cells showed that CDCA7 recruits HELLS (the DDM1 ortholog) to chromatin and activates its nucleosome remodeling activity ( Jenness et al ., 2018 ; Hardikar et al ., 2020 ; Shinkai et al ., 2024 ; Wassing et al ., 2024 ). Moreover, the zinc finger of animal CDCA7 and A. thaliana CDCA7α bind hemi-methylated CG dinucleotides, which are formed during DNA replication ( Hardikar et al ., 2024 ; Shinkai et al ., 2024 ; Wassing et al ., 2024 ). Therefore, CDCA7 likely functions by recruiting DDM1/HELLS to hemi-methylated mCG sites ( Wassing et al ., 2024 ) and DDM1/HELLS would then remodel nucleosomes to provide methyltransferases with access to DNA ( Zemach et al ., 2013 ; Lyons and Zilberman, 2017 ). This model is compatible with our observations, implying that this CDCA7 function is conserved in plants and metazoans. CDCA7α/β act as facultative activators of DDM1, since they act through DDM1, but their absence had a weaker impact on mCG than DDM1 loss. This suggests that DDM1 may function on its own or with other factors than CDCA7α/β. DDM1 alone can remodel nucleosomes in vitro ( Brzeski and Jerzmanowski, 2003 ; Lee et al ., 2023 ; Osakabe et al ., 2024 ), but it is not known whether DDM1’s intrinsic activity is sufficient in vivo or whether protein partners like CDCA7α/β are required to stimulate nucleosome remodeling. In contrast, HELLS requires CDCA7 for in vitro activity ( Jenness et al ., 2018 ; Nartey, Goodarzi and Williams, 2023 ) and animals have limited amounts of non-CG methylation ( Ziller et al ., 2011 ). This distinction suggests that DDM1’s intrinsic activity may have evolved to maintain non-CG methylation independently of CDCA7α/β, which fits with our observation that CDCA7α/β had a specific effect on mCG. Beyond mCG maintenance, DDM1/HELLS regulates histone variant exchange in plants and animals ( Ni et al ., 2020 ; Ni and Muegge, 2021 ; Osakabe et al ., 2021 ; Jamge et al ., 2023 ; Lee et al ., 2023 ), but the contribution of CDCA7 to this function had not been previously investigated. Our results indicate that CDCA7α/β act specifically on mCG regulation, with minimal impact on other DDM1 functions. This specificity aligns with the fact that CDCA7 has been lost in most eukaryotes lacking DNA methylation, while DDM1/HELLS has been retained ( Funabiki et al ., 2023 ). Accordingly, the absence of one CDCA7 paralog affects mCG but not H3K9me3 in mice ( Vukic et al ., 2024 ). Therefore, the mild reduction of heterochromatic marks other than mCG in cdca7α/β mutants is likely an indirect effect of decreased mCG. This raises the question of how CDCA7α/β selectively contributes to a single DDM1 function, given that DDM1-dependent heterochromatin marks generally co-exist in the genome. We propose a model in which histone variant incorporation by DDM1 occurs independently of CDCA7α/β after DNA synthesis. Subsequently, CDCA7α/β recruit DDM1 at hemi-mCG sites to stimulate nucleosome remodeling and allow mCG deposition. This model aligns with kinetics of nucleosome assembly being faster than mCG deposition after DNA replication ( Stewart-Morgan, Petryk and Groth, 2020 ; Stewart-Morgan et al ., 2023 ), and is further supported by observations that CDCA7 preferentially binds hemi-mCG sites located in nucleosomal rather than linker DNA ( Wassing et al ., 2024 ). Further studies will be essential to pinpoint the specific timing and coordination of CDCA7 in DDM1/HELLS functions. We provide evidence that a trans -factor, CDCA7α , causes changes in DNA methylation that in turn, have the capacity to influence natural populations. Changes in expression levels of this limiting factor perturbed mCG levels on TEs and potentially affected seed size, an adaptive trait. Our results support the hypothesis that evolutionary changes in the dosage of epigenetic regulators have caused global changes in epigenetic profiles( Baduel and Sasaki, 2023 ). To date, we have identified several other trans -regulators controlling mCH, including CMT2 , CMT3 , MIRNA823 (which targets CMT3 transcripts), JMJ26 , RNA polymerase V ( Dubin et al. 2015 ; Kawakatsu et al. 2016 ; Sasaki et al. 2019 ; Sasaki et al. 2022 ). To this list, we now add CDCA7α and DDM1 , affecting CG, which are directly inherited. As combinations of epigenetic mutants often exhibit epistasis ( Cao and Jacobsen 2002 ; He et al. 2022 ; Shimada et al. 2024 ), these allelic interactions—spanning multiple pathways and all DNA methylation contexts—considerably elevate epigenetic diversity and have the potential to modify gene regulatory networks and TE silencing. Such genetically inherited epigenetic variation can become a target of natural selection ( Baduel and Sasaki, 2023 ). Our study has provided insights into epigenome variation historically observed in nature at the molecular level. Further studies, including analysis of genetic and epigenetic architecture in other species, will enhance our understanding of the complex ecological significance of the epigenome in the tree of life. Materials and Methods Plant material Seeds sown on soil were stratified at 4°C for two days in the dark and grown at 21°C under long-day conditions (16 h light and 8 h dark) in a climatic chamber. Mutants from stock centers and previous studies cdca7α-1 (SALK_100123C), cdca7α-2 (SALKseq_124190.3), ddm1-2 ( Vongs et al ., 1993 ) mutant and control plants were in the Col-0 genetic background. To reset the effect of the transgenerational aggravation of the ddm1-2 mutant phenotype ( Kakutani et al ., 1996 ), we backcrossed a ddm1-2 homozygous mutant six times to Col-0, keeping ddm1-2 heterozygous throughout, and subsequently isolated homozygous ddm1-2 mutants which were considered first generation for this study. CRISPR-Cas9 mutagenesis CRISPR mutants were generated by transforming Col-0, ddm1-2 second-generation mutants, or cdca7α-2 mutants. To mutate the CDCA7α/β genes, we used a CRISPR-Cas9 system based on an intronized Cas9 gene controlled by the ubiquitous RPS5 promoter ( Grützner et al ., 2021 ). Guide RNAs (listed in Supplementary Table 2) controlled by a U6 promoter were cloned into pICH47751 or pICH47761, and binary plasmids were assembled using golden gate assembly with one or two guide RNAs, the intronized Cas9 gene, a linker for plasmid assembly, and a seed coat fluorescent reporter using mVenus ( Bensmihen et al ., 2004 ) or a Basta resistance gene. Transgenic lines were generated with Agrobacterium tumefaciens GV3101 by floral dipping. T1 transformed seeds were selected by Basta resistance or seed coat fluorescence. Mutations were identified by PCR and Sanger sequencing. We isolated T2 seeds devoid of the transgene and selected mutations. We obtained stable homozygous mutant lines in T2 or T3, and used their progeny for this study. Details about primers, guide RNAs and CRISPR mutations used in this study can be found in Supplementary Tables 2 and 3. Complementation CDCA7α and CDCA7β transgenes were cloned from genomic DNA to include endogenous regulatory sequences in promoters, 5’UTRs, and introns. We used the Greengate cloning system ( Lampropoulos et al ., 2013 ) to assemble constructs into a pGGSun backbone ( Incarbone et al ., 2023 ). Promoters included 5’ UTRs and were cloned from the first base before the ATG start codon to 792 bp upstream for CDCA7α and 1349 bp upstream for CDCA7β . The alternative CDCA7α a promoter was cloned from A. thaliana accession 6966 using the same primers as for CDCA7α ref promoter, resulting in an 831 bp fragment. Coding sequences contained all introns and exons, except the 3’ UTR and the start and stop codon, according to the tag position. Endogenous CDCA7β BsaI sites were removed by mutagenesis, introducing silent mutations (GTC to GTT and GGT to GGA). N-terminal tags had a start codon with 3xcMyc fused with mTurquoise2, C-terminal tags had mTurquoise2 fused with 3xcMyc and a stop codon. All constructs were cloned with the terminator of the UBQ10 gene (628 bp) and a mVenus seed coat reporter gene. Deletions and point mutations were introduced in constructs using in vivo cloning ( García-Nafría et al. 2016 ). Transgenic lines were generated with A. tumefaciens GV3101 by floral dipping and transformed seeds were selected by seed coat fluorescence ( Bensmihen et al ., 2004 ). Phylogenetic Analysis of CDCA7 To reconstruct the evolutionary history of CDCA7, we constructed orthologous gene clusters (i.e., orthogroups) from two different proteome datasets using sequential, dense taxon sampling across two levels: Archaeplastida (50 species) and Viridiplantae (66 species, enriched for Brassicales) (Supplementary Table 4). This method enabled us to estimate more precisely the timing of the CDCA7α/β duplication event. Orthofinder v2.5.2 ( Emms and Kelly, 2019 ) was used to cluster genes in a non-biased manner by comparing each gene to the entire proteome dataset, and DIAMOND ( Buchfink, Reuter and Drost, 2021 ) for homology search (‘-S diamond_ultra_sens’). For the Archaeaplastida-level survey, we included previously-identified metazoan CDCA7s ( Funabiki et al ., 2023 ) as an outgroup. MAFFT v7.310 ( Katoh and Standley, 2013 ) was used to align protein sequences based on orthogroups. Individual maximum likelihood gene trees were built with IQ-TREE v2.1.2 ( Minh et al ., 2020 ), which used model selection (‘-m MFP’) and an ultrafast bootstrap approximation approach (1,000 replicates). The phylogenetic trees were visualized using iToL v6.7 ( Letunic and Bork, 2024 ). We used ancestral sequence reconstruction to reconstruct the CDCA7 protein sequence prior to the Brassicaceae CDCA7α/β duplication event. Based on the phylogenetic positions, we selected 48 protein sequences from Brassicaceae CDCA7 homologs and Amborella CDCA7 as an outgroup and aligned them using MAFFT. Following the multiple sequence alignment, IQ-TREE was used to rebuild a phylogenetic tree that will be used as input for the initial ASR analysis. First, we converted the protein sequence alignment to an absence/presence matrix, then used IQ-TREE’s ancestral sequence reconstruction (’-asr’) option with the rooted phylogenetic tree. The absence/presence ancestral state output was used to trim residues from the protein sequence alignment based on the calculated posterior probability of each residue. The final ancestral sequence reconstruction used the trimmed alignment to generate the ancestral Brassicaceae CDCA7 protein sequence. Measuring flowering time and seed size Flowering time was quantified by counting the number of mature rosette leaves at the anthesis of the first flower. Plants were grown in a randomized layout under short-day (8h light/16h dark) or long-day (16h light/8h dark) conditions at 21°C in a climatic chamber. The cdca7α/β mutants were in the 5th generation of inbreeding post-CRISPR transformation, while ddm1-2 mutants were in the 3rd generation. For seed size measurements, plants were grown in a randomized pattern from seed batches with reduced heterogeneity. Specifically, seeds were sorted on size to exclude the 5% of biggest and smallest seeds, using a Boxeed seed phenotyping robot (Labdeers). Four crosses were performed for each inflorescence, yielding 5 to 47 seeds per silique (mean = 31.08, n = 72). We used either the main floral stem, or the first axillary branch as a female inflorescence. A linear model showed that the mean seed surface area per silique was unaffected by the inflorescence type (P = 0.64) or the number of seeds per silique (P = 0.89), allowing comparisons to be made without accounting for these variables. Seed surface was measured using scanners and ImageJ scripts. Sample collection for epigenomics Plants were grown following a randomized pot pattern. For methylomes and transcriptomes of cdca7 and ddm1 cdca7 mutants ( Fig. 2 , 3 ), we pooled rosette leaves from three 31-day-old plants (two rosette leaves each), ground the tissues in liquid nitrogen, and split the ground tissues for DNA and RNA extraction. For ChIP-seq, we pooled at least 5 plants using 5 rosette leaves each up to 1.5 g per sample. For methylomes and transcriptomes of complemented cdca7 mutants ( Fig. 4 , 5 ), we used 28-day-old plants: one rosette leaf (2-3 cm long) was flash frozen first for RNA extraction, then all remaining rosette leaves except cotyledons and leaves 1 and 2 were collected for DNA extraction. All generated libraries are listed in Supplementary Table 5, with the number of inbreeding generations for each mutant. Whole genome bisulfite sequencing (WGBS) Library construction and sequencing To measure DNA methylation levels in cdca7α-2, smp2, and AT4G37130 mutants ( Fig. 1e ), genomic DNA was extracted from rosette leaves at 9-true-leaf stage using the GeneJET Plant Genomic DNA Purification Kit (Thermo Scientific) and sheared with an E220 Focused-ultrasonicator (Covaris) to achieve an average fragment size of approximately 350 bp. Sequencing libraries were prepared using the NEBNext Ultra II DNA Library Prep Kit (New England BioLabs) with methylated adapters (New England BioLabs). The adapter-ligated DNA underwent bisulfite conversion using the EZ-96 DNA Methylation-Gold MagPrep Kit (Zymo Research). Bisulfite-treated samples were amplified using EpiMark Hot Start Taq DNA Polymerase and indexed with NEBNext Multiplex Oligos for Illumina (New England BioLabs). All libraries were sequenced on either an Illumina NextSeq 550 or HiSeq 2500 platform. Estimation of DNA methylation levels All reads were mapped on TAIR10 reference genome using a Methylpy pipeline v1.2 ( https://github.com/yupenghe/methylpy ). DNA methylation levels were estimated as weighted methylation levels for each transposon defined in Araport11 annotation. CMT2- and RdDM-targeted transposons were defined as having differential levels of methylation (> 0.1) between wild-type and cmt2 or drm1drm2 in Col-0 as previously described ( Kawakatsu et al ., 2016 ). For each line, average DNA methylation was calculated using all transposons for which at least one read was mapped. Tagmentation-based WGBS (T-WGBS) Library construction and sequencing Genomic DNA was extracted from rosette leaves with the DNeasy Plant Pro kit (Qiagen, reference 69206), following the manufacturer’s instructions, using 1 to 5 biological replicates per genotype (Supplementary Table 5). Illumina libraries of sodium bisulfite-converted DNA were prepared with a tagmentation-based approach (Wang et al. 2013) with modifications (Montgomery and Berger, 2023, bioRxiv). Briefly, we tagmented 50 ng of DNA with Tn5 (Molecular Biology Services, IMP, Vienna, Austria). Tagmented DNA was purified with the Zymo DNA Clean and Concentrator kit (Zymo Research) instead of SPRI beads. After oligonucleotide replacement, gap repair, and purification, we used the entirety of the eluted DNA for bisulfite conversion, which was performed with the EZ DNA Methylation-Gold Kit from Zymo (reference D5006). The rest of the protocol was performed as indicated by Wang et al (2013). Directional libraries were amplified with Nextera-compatible unique dual indices and sequenced on a NovaSeq 6000 instrument using an S4 flow cell to generate paired-end 150 bp reads with around 25 million reads per sample. T-WGBS analysis Libraries for cdca7α/β and ddm1 mutants ( Fig. 2 , 3 ) were randomly subsampled to 30 million reads to improve depth homogeneity, since there were up to 2-fold more reads in some samples. The data were analyzed with the nfcore methylseq pipeline v2.3.0 ( https://nf-co.re/methylseq ) ( Ewels et al ., 2020 ) with Nextflow (v22.10.7). Cytosine positions covered by fewer than 3 reads (--meth_cutoff 3) were excluded. Tagmentation creates gaps in the double-stranded DNA which are later filled by unmethylated cytosines during gap repair. These nucleotides were excluded from the analysis by trimming reads with the "clip" and "ignore" options, using M-bias plots to ensure the removal of unmethylated fragment ends. The parameters were as follow: --meth_cutoff 4 --comprehensive --cytosine_report --clip_r1 14 --three_prime_clip_r1 4 --three_prime_clip_r2 2 --ignore_3prime_r2 17. Note that the unmethylated cytosines are incorporated at the 5’ end of read 2, while the options mentioned above suggest we ignored the 3’ end of read 2 instead, but methylseq v2.3.0 had an issue which made it clip the wrong end of read 2 (see https://github.com/nf-core/methylseq/issues/299 ). Using M-bias plots, we noticed a 5’ bias on read 1 that resembled the Tn5 sequence bias ( Wolpe, Martins and Guertin, 2023 ), and removed it by clipping 14 nucleotides at the 5’ end of read 1. We also excluded additional small methylation biases at the 3’ end of read 1 and read 2. To average biological replicates, we computed the mean methylation level for individual cytosine positions, including all replicates that met the minimum coverage threshold. Positions covered in only one replicate were retained to maximize data inclusion. To compute the average methylation values over annotation sets or 100 bp bins, we used the bigWigAverageOverBed script from UCSC tools ( https://github.com/ucscGenomeBrowser/kent-core/tree/master ). Only TEs covered in all samples from a common batch were retained. To calculate average values in bins (metaplots), annotations were scaled to an arbitrary size of 1000 bp, and the methylation value was calculated across annotation length by 25 bp bins, and up to 500 bp upstream and downstream. This was done with the computeMatrix scale-regions --binSize 25 and plotProfile functions from deepTools v3.3.1. To analyze 5mC changes in 100 bp bins, we segmented the genome in non-overlapping 100 bp bins, counted the number of each methylation context per bin using a custom python script. Average 5mC per bins was calculated with bigWigAverageOverBed , and only bins that were covered in all samples were retained for further analysis. To compute Spearman correlations, we kept bins which had at least 10% mCG, 6% mCHG and 5% mCHH in WT. Gene body methylation analysis was restricted to protein-coding genes covered in all samples which had more than 10% mCG, less than 10% mCHG and 10% mCHH in WT (n = 9938). To analyze DNA methylation at TEs, we excluded TEs unmethylated in the WT, using a threshold of 5% methylation for mCG, 3% for mCHG and 2% for mCHH, after which sample size were of 18052, 16739 and 18037, respectively. 3’ tag-sequencing of messenger RNAs Library construction and sequencing Total RNA was extracted using the RNA isolation kit provided by the VBC core facilities. The kit uses a lysis step based on guanidine thiocyanate (adapted from ( Boom et al ., 1990 ) and carboxylate-modified Sera-Mag Speed beads and was applied using the KingFisher instrument (Thermo). Libraries were prepared in 96-well plates using a custom-protocol described previously ( Clavel et al ., 2024 ). The oligo d(T) primers used for reverse transcription contain 8-bp unique molecular identifiers (UMIs), 7-bp barcodes for multiplexing, and a Truseq adapter. After reverse transcription, samples are pooled and RNA-cDNA duplexes tagmented with Tn5. Using primers matching the Truseq and Tn5 adapters, the 3’ ends of mRNA transcripts are subsequently amplified by PCR, while P5 and P7 sequences and plate-specific dual indices are added for Illumina sequencing. We started from 80 ng of total RNA in 4 µl, adding 1 µl of 0.2 µM primers for reverse transcription. RNAs were denatured 3 min at 72°C and kept on ice. We mixed 1 µl of the SMARTScribe reverse transcriptase (Takara) in a final volume of 10 µl with 1x SMARTScribe buffer, 1.75 mM dNTPs, 2 mM DTT, 3 mM MnCl 2 for 1h at 42°C followed by inactivation for 15 min at 70°C. All samples were pooled together and purified using 1.2x volume of SPRI beads (Molecular Biology Services, IMP, Vienna, Austria). After mixing and 2 min at room temperature, beads were collected on a magnet, washed two times with 70% ethanol, dried, and eluted in 60 µl of 10 mM Tris-HCl pH 8.0. To assemble transposomes, we mixed Tn5 (0.38 mg / ml) with 18 µM of annealed adapters, 50% glycerol and 1x Tn5 loading buffer (6x buffer: 300 mM Hepes KOH pH 7.2, 600 mM NaCl, 0.5 mM EDTA, 6 mM DTT, 0.6% Triton-X-100) and incubated for 1h at 37°C. For tagmentation, 16 µl of pooled RNA-cDNA was mixed with 16 µl of transposome and 32 µl of 2x tagmentation buffer (20 mM Tris (hydroxymethyl)aminomethane, 10mM MgCl2, adjusted to pH 7.6 and 20% (vol/vol) dimethylformamide). After 8 min at 55°C, we added 16 µl of 0.2% SDS and incubated 5 min at room temperature. Nucleic acids were purified with 2x SPRI bead volume as indicated above with a 48 µl elution. For PCR, we pre-incubated at 95°C the KAPA HiFi HotStart 2X ReadyMix for 90 s to activate the polymerase and amplified libraries in 100 µl final volume using 1 µM of primers and 46 µl of purified cDNA-RNA duplexes with the following program: 72°C for 3 min, 95°C for 30 sec, (95°C for 20 sec, 65°C for 30 sec, 72°C for 30 sec) for 12 cycles, 72°C for 5 min. Libraries were purified with 0.8x volume of SPRI beads as indicated above and eluted in 62 µl. After controlling for appropriate fragment size distribution on a 5200 Fragment Analyzer System (Agilent), libraries were sequenced as paired-end 150 bp reads on a NovaSeq S4 ( Fig. 2 , 3 ) or a NovaseqX 10B ( Fig. 4 , 5 ) flow cell. Mapping and differential expression Reads were demultiplexed, trimmed, collapsed to remove duplicates, mapped, and quantified. The code and documentation are available at https://github.com/pierre-bourguet/3-tag-rna-seq , but we provide the most important details below. Read1, which starts directly in the transcript at the site of Tn5 cutting, was used for mapping. Read2, which contains the sample-specific index, UMI, polyA tail, and eventually the 3’ end of the transcript, was used for demultiplexing and UMI extraction. Specifically, we used the first 7 bases of read2 to identify indices specific to each well in the plate, and the next 8 bases to extract UMIs. For read1, the 3’ end was trimmed for polyA and adapter contamination and reads shorter than 50bp were discarded. UMIs were prepended to read1 and these chimeric reads were used to collapse duplicates into a single read with clumpify (bbmap v38.26) ( Bushnell, 2014 ), with no mismatches allowed in the UMI and two mismatches allowed in the transcript sequence. Mapping was performed with STAR v2.7.11a ( Dobin et al ., 2013 ) and mapped reads were counted with Salmon v1.10.1 in alignment mode ( Patro et al ., 2017 ) in sense and antisense. We used the TAIR10 genome and AtRTD3 transcriptome reference ( Zhang et al ., 2022 ). Because the AtRTD3 GFF and fasta files were sometimes discordant, we regenerated a fasta file from a fixed GFF file using AGAT v1.0.0. We observed that the 3’-end of transcripts were often located outside of transposable element gene annotations, likely because they do not include 3’ UTRs, but within TAIR10 TE annotations. For this reason, and to avoid overlapping annotations, we removed “transposable element gene” annotations from AtRTD3 and added the 31189 transposable element annotations from TAIR10 instead. We used DESeq2 v1.42.1 ( Love, Huber and Anders, 2014 ) for a quantitative analysis of read counts. For a given feature, counts in the sense or antisense orientation were considered as independent features. We ran the DESeq2 model in single-strand mode, removing features which did not have at least 10 reads in a minimum of 3 samples. Features were considered differentially expressed at a FDR of 10% (adjusted p-value ≤ 0.1) and an absolute log 2 fold change relative to the WT equal or greater than 1. To avoid confounding effects of overlapping annotations, we removed TEs misregulated in sense and antisense if they overlap (1 bp or more) a protein-coding gene in the same or reverse orientation, respectively. We use DESeq2 to normalize counts with the median ratio method and apply a variance-stabilizing transformation (VST). To quantify transgene expression, the transgene sequence and annotations were added as an additional chromosome in the fasta and GTF files. The C-terminal mTurq2:3xcMyc tag allowed us to distinguish transgenic reads from endogenous CDCA7α reads. For Fig. 5e, a two-way ANOVA showed a significant effect of the promoter haplotype ( p- value = 5e-6) on transgene expression levels, but no effect of the genetic background ( p- value = 0.46) and no interaction ( p-value = 0.48) and therefore, we merged the data from the two different backgrounds. Chromatin Immunoprecipitation followed by sequencing Library construction and sequencing 1.5 g of rosette leaves from 31-d-old plants were fixed in 37 ml 1% formaldehyde in PBS by vacuum-infiltration for 10 min. The cross-linking reaction was quenched by adding glycine to a final concentration of 125 mM. Tissues were frozen in liquid nitrogen and ground in a 5-ml grinding jar on a Retsch Mixer Mill MM 200 at 30 Hz for 45 s. Tubes were kept at room temperature until the powder showed signs of thawing. We added 35 ml of nuclei isolation buffer (NIB, 10 mM MES-KOH pH 5.3, 250 mM sucrose, 10 mM NaCl, 10 mM KCl, 2.5 mM EDTA, 0.1 mM spermine, 0.1 mM spermidine, 2.5 mM ß-mercaptoethanol, 0.3% Triton X-100 and protease inhibitors) (Lorkovic 2004 MBC), followed by vortexing until homogeneous. The extract was filtered through two layers of Miracloth, which were washed with 10 ml of NIB and spun at 3,400 rpm at 4°C for 5 min. Pellets were resuspended in 25 ml of NIB, vortexed and incubated for 10 min on ice to fully dissolve chlorophyll content. Tubes were spun as above and pellets were carefully washed with 5 ml of shearing buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 0.1% SDS, and protease inhibitors) without resuspending and spun again. This washing step was repeated and pellets were resuspended in 0.9 ml of shearing buffer. One ml of the extracted chromatin was transferred to Covaris glass tubes and sonication was conducted at 4°C for 900 seconds per sample (peak power 140.0; duty factor: 5.0; Cycles/Burst: 200). Sheared chromatin was spun for 5 min at 4°C at 15,000 rpm. The supernatant was collected and spun again, and the supernatant was transferred to a 5 ml tube. Fifteen µl of chromatin were kept to control for sonication efficiency. Chromatin was diluted to 3.5 ml to reduce SDS concentration with a ChIP dilution buffer (16.7 mM Tris-HCl pH 8.0, 167 mM NaCl, 1.2 mM EDTA, 1.1% Triton X-100, 0.01% SDS and protease inhibitors). For preclearing, 200 µl of Dynabeads protein A (Thermofisher, reference 1001D) were added to the chromatin and rotated 1 h at 4°C. Beads were discarded by collecting supernatants after two centrifugations at maximum speed for 30 s. A hundred µl of pre-cleared chromatin was kept at -20°C for input control, and the remaining chromatin was aliquoted with 500 µl per immunoprecipitation (IP) and incubated with 5 µg of anti-H3 (Abcam, ab1791), anti-H2A.W.6/7, anti-H2A.Z.9/11, anti-H3K27me1 (Millipore, 17-643) or anti-H3K9me2 (Abcam, ab1220) antibodies at 4 °C overnight with rotation. After incubation, samples were mixed with 30 µl of protein A magnetic beads, rotated at 4°C for 3 h, and washed two times with a low-salt buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100 and 0.1% SDS), once with a high-salt buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 2 mM EDTA, 1% Triton X-100 and 0.1% SDS), once with a LiCl buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 0.25 M LiCl, 1% IGEPAL CA-630 and 0.1% deoxycholic acid) and twice with TE buffer (10 mM Tris-HCl pH 8.0 and 1 mM EDTA). Elution was done with 200 µl of 0.1 M NaHCO3 and 1% SDS, incubated at 65°C for 15 min. We added 20.4 µl of reverse cross-link buffer (0.4 M Tris-HCl pH 8.0, 2 M NaCl, 10 mM EDTA, 0.6 mg ml−1 proteinase K (Thermo Fisher Scientific)) and incubated at 45°C for 3 h and 65°C for 16 h. RNA was subsequently degraded for 30 min at room temperature with 10 µg of RNase A (Thermo Fisher Scientific), and DNA was purified with the ChIP DNA Clean & Concentrator kit (Zymo, reference D5205). Libraries were prepared with the NEBNext Ultra II DNA library prep kit for Illumina (New England Biolabs), following the manufacturer’s instructions. Fragment size distribution was analyzed on a 5200 Fragment Analyzer System (Agilent), and size selection with SPRI beads (Molecular Biology Services, IMP, Vienna, Austria) was applied to remove large fragments and self-ligated adapters, when necessary. To maximize comparability, if only one sample had undesired fragments, we size-selected all samples for a given antibody. For size selection, briefly, DNA was diluted to 100 µl, we mixed in 60 µl of magnetic SPRI beads, collected the supernatant, and added 40 µl of magnetic SPRI beads, washed two times with 80% ethanol, and eluted DNA. Sequencing was carried out on a NovaSeq 6000 instrument using an S4 flow cell to generate paired-end 150 bp reads with around 25 million reads per sample. ChIP-seq analysis The data were analyzed using the nfcore/chipseq pipeline v2.0.0 ( https://nf-co.re/chipseq/2.0.0/ ) ( Ewels et al ., 2020 ) with Nextflow (v22.10.7). We used the TAIR10 genome, custom arguments for read length (150 bp), fragment size (275), and blacklisted mitochondrial and chloroplast genomes. To get the average value for each annotation, we used the bigWigAverageOverBed script from UCSC tools ( https://github.com/ucscGenomeBrowser/kent-core/tree/master ). We further analyzed the data with deepTools v3.3.1 ( Ramírez et al ., 2016 ). To normalize by H3 or by the WT, we used bigwigCompare with a bin size of 10 bp, ignoring non-covered regions with skipNonCoveredRegions . To produce average profile plots (metaplots), annotations were scaled to an arbitrary size of 1000 bp and the enrichment value was calculated by 50 bp bins, using computeMatrix scale-regions . RT-qPCR We used 21-d-old and 27-d-old rosette leaves to perform experiments in Fig. S5d and Fig. S7b , respectively. RNA extraction, transcript quantification, and normalization were performed as previously reported ( Bourguet et al ., 2020 ), except that reactions were run on a Roche TM LightCycler® 96 system with the following program: 10 min at 45°C, 2 min at 95°C, and 40 cycles of 5 s at 95°C, 10 s at 60°C, and 5 s at 72°C. Primers are listed in Supplementary Table 2. Protein structure prediction We used Alphafold3 ( Abramson et al ., 2024 ) to model protein structures and interactions, with the random seed set to 1. The top-ranking prediction was selected for representation. The predicted structure of CDCA7α with three Zn 2+ ions was superimposed to H. sapiens CDCA7 in a complex with non-B-form DNA containing 5mC (PDB ID 8TLK) ( Hardikar et al ., 2024 ). Structures of DDM1 with a nucleosome ( Fig. S8 ) were predicted with the H2A.W.6 and H3.1 histone variants, using the exact sequences from a cryo-EM structure of DDM1 (PDB ID 8J90) ( Osakabe et al ., 2024 ). Precisely, protein chains were from the HTA6, HTR2, AT2G28740 (H4) and HTB9 proteins, with a Widom 601 DNA sequence. The best ranked prediction was in good agreement with the 8J90 structure, as root mean square deviation (RMSD) between 190 pruned atom pairs was 1.043 angstroms. Unstructured regions were removed from the visualization for clarity. To find interacting residues between CDCA7α/β and DDM1 ( Fig. 4d, g , S8d ), we used AlphaFold-Multimer ( Evans et al ., 2022 ) using a local implementation of Colabfold ( Mirdita et al ., 2022 ) and the top-ranking prediction. Graphical representations and analyses of structures were done with ChimeraX ( Meng et al ., 2023 ). Recombinant protein expression Cloning CDCA7α and CDCA7β cDNAs were amplified by RT-PCR with primers indicated in Supplementary Table 2, using RNA purified from WT flowers. We used the canonical CDCA7α.2 transcript variant according to AtRTD3 ( Zhang et al. 2022 ). cDNAs were cloned into pGEX-4T-1 (Cytiva) using EcoRI / SalI . DDM1 cDNA into pET15b was previously described ( Osakabe et al. 2021 ). Expression and purification of recombinant proteins and in vitro pull-down experiments BL21 (DE3) RIL E. coli cells transformed with plasmids for protein expression were grown at 37°C overnight in 100 ml LB. Cultures were diluted in 1 L of LB and grown for three hours at 20°C and then induced for 7 hours at 20°C with 1 mM IPTG. For GST-tagged proteins cell pellets were resuspended in 20 ml of extraction buffer (50 mM Tris-HCl pH 8.0, 1 M NaCl, 1 mM DTT, 0.1% Triton X-100) containing protease inhibitors (Roche), 10 µl of benzonase (1 mg/ml) and 50 mg of lysozyme. After sonication (Bioruptor, Diagenode) for 10 min at high intensity (5’’ on / 5’’ off) and 5 min at medium intensity (5’’ on / 5’’ off), extracts were centrifuged for 15 min at 4°C at 40,000 × g . Extracts were incubated with 500 µl of glutathione Sepharose 4 fast flow (Cytiva) at RT for one hour and then transferred to disposable columns and washed with 5 column volumes of extraction buffer. Proteins were eluted with six 300 µl elution steps with 50 mM Tris-HCl pH 8.0, 500 mM NaCl buffer containing 20 mM reduced glutathione and 1 mM DTT. For purification of His 6 - or His 6 SUMO-tagged DDM1, cell pellets from 2 L cultures were resuspended in 25 ml of extraction buffer (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 2 mM DTT, 0.05% NP-40) containing protease inhibitors (Roche), 10 µl of benzonase (1 mg/ml) and 50 mg of lysozyme and processed as for GST-tagged proteins. Extracts were incubated with 500 µl of Ni-NTA beads (Qiagen) for one hour at RT and then transferred to disposable columns and washed with 5 column volumes of extraction buffer containing 5 mM imidazole but without benzonase and lysozyme. Proteins were eluted with six 400 µl elution steps with elution buffer (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 2 mM DTT, 300 mM imidazole). In all purifications, elution fractions were analyzed on 10-12% SDS-PAGE, pooled, and buffer was exchanged into 50 mM Tris-HCl pH 7.5, 100 mM NaCl, 1 mM DTT by centrifugation over Amicon Ultra-15 30 kDa and 50 kDa cut-off centrifugal filters (Millipore) for GST-tagged CDCA7 and His-tagged DDM1, respectively. For pull-down, equimolar amounts of GST-tagged proteins along with GST alone were mixed with 5 µg of His 6 SUMO-DDM1 and incubated with 10 µl of magnetic glutathione beads (Thermo Fisher Scientific) for 90 min at RT in binding buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM DTT, 0.1% Triton X-100). Beads were washed 6 times with binding buffer, denatured in 30 µl of 1 × SDS-PAGE loading buffer and 10 µl were loaded on 10% SDS-PAGE and analyzed by western blotting with anti-DDM1 antibody. Input lanes were loaded with 1/20 of protein used for pull-down. Genome-wide association studies GWAS were conducted using LIMIX ( Lippert et al ., 2014 ) version 3.0.4 with full genome SNPs from the 1001 genome project (10,709,949 SNPs for mCG levels and 3,686,507 SNPs for gene expression) and the following linear mixed model (LMM), where Y is the vector of a phenotype, fixed terms, L and X , are n × 1 vectors corresponding to a cofactor and a genotype to be tested (SNP) with the parameters α and β , respectively. and are random terms, including the identity-by-state (IBS) kinship K matrix representing the genetic relatedness ( Yu et al ., 2006 ; Kang et al ., 2008 ) and the residuals, respectively. The vectors Y , L , and X were z-scored. Models without a cofactor take α = 0. SNPs with MAF ≥ 5% were used for association studies. Bonferroni correction was applied for multiple-testing correction after excluding all SNPs with MAF < 5%. Mediation analysis and heritability The fraction of genetic effects on mCG variation mediated by CDCA7α expression was estimated using mediation analysis with the published program ( Sasaki, Frommlet and Nordborg, 2018 ). Two alleles, CDCA7α a and CDCA7α b , were fitted into one model together as a single genotype vector: CDCA7α Ref :0, CDCA7α a :1, and CDCA7α b :2 . Local genetic structure Genome alignment of CDCA7α region and haplotype network construction Fully assembled genome sequences of 27 A. thaliana lines ( Igolkina et al ., 2024 ), A. lyrata v1.0, A. halleri v2.1.0, and Capsella rubella v1.0 ( Slotte et al ., 2013 ) were used for the alignments of diverged CDCA7α promoter region. A. lyrata and C. rubella genomes were downloaded from NCBI, and A. halleri genome was downloaded from Phytozome ( https://phytozome-next.jgi.doe.gov/ ). We extracted a 4 kb region covering the CDCA7α and SMP2 coding sequences based on homology (chr4: 17,484,000–17,488,000 in the A. thaliana Col-0 genome, TAIR10) and made the alignment using CLC Sequence Viewer 8 with a gap open cost of 10.0 and a gap extension cost of 1.0. After removing all gaps, the alignment was applied to haplotype network construction by minimum spanning network ( Paradis, 2018 ) implemented in popart ( Leigh and Bryant, 2015 ). The length of the intergenic region between CDCA7α and SMP2 was estimated as the distance of one bp outside the start codons of CDCA7α.1 and SMP2.1 (chr4: 17,486,198–17,486,678; 481 bp in Col-0). Local genetic structure of DDM1 The local genetic structure around the DDM1 locus was analyzed using principal component analysis (PCA) according to a previous study ( Sasaki et al ., 2021 ). Using the prcomp() function in R, we analyzed 475 SNPs from 1,135 lines in the 1001 Genomes Project ( Kawakatsu et al. 2016 ), covering the region from chr5:26649000 to 26658200. Clusters represented by PC1 and PC2 were selected as components indicating the major haplotypes. Population structure and geographic distribution We used the population structure information as genetic groups for 1135 lines from the 1001 genome dataset. The information for the additional 188 lines, including Tanzania and the Yangtze river populations, was extracted from a previous study ( Hsu, Lo and Lee, 2019 ). The pairwise genetic distance matrix calculated by Hsu et al. (2019) was used to construct the neighbor-joining tree. Latitude and longitude coordinates for each line were collected from the previous studies ( 1001 Genomes Consortium, 2016 ; Zou et al ., 2017 ), and altitude information was estimated using the R package elevatr ( https://github.com/USEPA/elevatr ). Local maps (Northern Sweden, The Iberian peninsula, and Yangtze Liver) and the elevation profile view were generated using QGIS 3.38.2 ( https://www.qgis.org/ ). Public data analysis Molecular phenotypes in natural populations DNA methylation data used in this study was previously described ( Kawakatsu et al ., 2016 ; Sasaki et al ., 2019 ). Briefly, bisulfite sequencing data for 774 lines in leaf tissue grown under ambient temperature was used for this analysis. Reads were mapped on each pseudogenome, using a Methylpy pipeline ( https://github.com/yupenghe/methylpy ). mCG levels were calculated as weighted methylation levels ( Schmitz and Ecker, 2012 ) for individual TEs. We used TAIR10 annotation to define TE regions and 6,378 TEs having mapped reads in the region in all lines were used for all analyses. CMT2 and RdDM-targeted TEs were defined based on differential methylation level (>0.1) between wild-type and drm1drm2 or cmt2 in Col-0 ( Stroud et al ., 2013 ). Expression data for 461 lines, collected in the 1001 Genomes Project under the same conditions as the DNA methylation data, were used in this study ( Kawakatsu et al ., 2016 ; Kornienko et al ., 2023 ). As described in ( Kornienko et al ., 2023 ), raw RNA-seq data were aligned to the TAIR10 genome using STAR v.2.9.6 ( Dobin et al. 2013 ) and gene expression levels were quantified using featurecounts (Liao et al. 2014). Seed size phenotypes in natural populations Seed size and altitude information in the Spanish local population were collected from ( Vidigal et al ., 2016 ). After extracting lines sequenced in the 1001 genome project, we analyzed only genetically Spanish group to exclude the global population structure. CDCA7α and CDCA7β expression profiles To analyze transcript levels of CDCA7α and CDCA7β across tissues and developmental stages ( Hofmann, Schon and Nodine, 2019 ), we kept only tissues from the original dataset ( Klepikova et al ., 2016 ) and from ( Hofmann, Schon and Nodine, 2019 ), to limit overrepresentation of embryonic samples, and filtered out senescent tissues. Data processing, plots and statistical analysis All analyses were performed using R Statistical Software (R v4.3.2; https://www.R-project.org/ ) with the following packages: tidyverse 2.0.0, ggplot2 3.5.1 ( Wickham, 2016 ; Wickham et al ., 2019 ), patchwork 1.2.0, agricolae 1.3-7, ggbeeswarm 0.7.2, MASS 7.3-60 ( https://patchwork.data-imaginist.com , https://cran.r-project.org/package=agricolae , https://cran.r-project.org/package=ggbeeswarm , https://cran.r-project.org/package=MASS ). Tukey’s HSD tests were performed using one-way ANOVA with the aov() function, followed by the TukeyHSD() function. Heatmaps were built with the ComplexHeatmap package ( Gu, Eils and Schlesner, 2016 ). Clustering was performed with k-means, using the consensus clustering from 1000 repeats. Data availability The data described in this publication have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE284119 (T-WGBS), GSE283989 (WGBS), GSE283987 (ChIP-seq), GSE284067 (3’ tag mRNA-seq). Authors contributions P.B., Z.L, D.C.K. and V.B. performed the experiments. P.B., E.S., C.H.C, A.I., and C.R.L. analyzed the data. P.B., E.S., F.B. and M.N. supervised the study and designed the experiments. P.B., E.S., F.B. and M.N. wrote the manuscript. Acknowledgments The authors express their gratitude to Akihisa Osakabe and Hironori Funabiki for the initial suggestion; Aleksandra E Kornienko and Almudena Molla Morales for providing DNA and RNA from A. thaliana accessions; Yoav Voichek for the 3’ tag-seq protocol; Vikas Shukla for the original tag-seq script; Marion Clavel, Nenad Grujic and Yasin Dagdas for GreenGate and CRISPR constructs; Tal Dahan-Meir and Thomas James Ellis for help with seed size measurements; Tatsuo Kanno for ddm1 inbred lines; Jurah Alej for the local AlphaFold Multimer implementation; Elizaveta Grigoreva and Pierre Baduel for sharing genotype information of A. lyrata and A. halleri ; Yoav Voichek, Olivier Mathieu, Leandro Quadrana for fruitful discussions; the Berger and Nordborg group for constructive feedback. For their support, we thank the staff at the Next Generation Sequencing and PlantS facilities at the Vienna BioCenter Core Facilities (VBCF), and the Molecular Biology Services from the Institute for Molecular Pathology (IMP). The computational results presented were obtained using the CLIP cluster ( https://clip.science ). This research was supported by FWF grants to F.B. P32054, P33380, PAT6138924; European Union’s Framework Programme for Research, Innovation Horizon 2020 (2014–2020) under the Marie Curie Skłodowska Grant Agreement no. 847548 (VIP2) to P.B, and Japan Society for the Promotion of Science to E.S. (JSPS 20 K22671 and 21H02538). Funder Information Declared FWF Austrian Science Fund, https://ror.org/013tf3c58 , P32054 , P33380 , PAT6138924 European Union, https://ror.org/019w4f821 , 847548 Japan Society for the Promotion of Science , 20K22671 , 21H02538 References ↵ 1001 Genomes Consortium ( 2016 ) ‘1,135 Genomes Reveal the Global Pattern of Polymorphism in Arabidopsis thaliana’ , Cell , 166 ( 2 ), pp. 481 – 491 . OpenUrl CrossRef PubMed ↵ Abramson , J. et al. ( 2024 ) ‘ Accurate structure prediction of biomolecular interactions with AlphaFold 3 ’, Nature , 630 ( 8016 ), pp. 493 – 500 . OpenUrl CrossRef PubMed ↵ Adams , S. et al. ( 2000 ) ‘Parent-of-origin effects on seed development in Arabidopsis thaliana require DNA methylation’, Development (Cambridge , England ) , 127 ( 11 ), pp. 2493 – 2502 . OpenUrl ↵ Akinmusola , R.Y. , Wilkins , C.-A. and Doughty , J . ( 2023 ) ‘ DDM1-mediated TE silencing in plants ’, Plants , 12 ( 3 ), p. 437 . OpenUrl PubMed ↵ Alonso-Blanco , C. et al. ( 1999 ) ‘ Natural allelic variation at seed size loci in relation to other life history traits of Arabidopsis thaliana ’, Proceedings of the National Academy of Sciences of the United States of America , 96 ( 8 ), pp. 4710 – 4717 . OpenUrl Abstract / FREE Full Text ↵ Baduel , P. et al. ( 2021 ) ‘ Genetic and environmental modulation of transposition shapes the evolutionary potential of Arabidopsis thaliana ’, Genome biology , 22 ( 1 ), pp. 1 – 26 . OpenUrl CrossRef PubMed ↵ Baduel , P. and Sasaki , E . ( 2023 ) ‘ The genetic basis of epigenetic variation and its consequences for adaptation ’, Current opinion in plant biology , 75 , p. 102409 . OpenUrl PubMed ↵ Becker , C. et al. ( 2011 ) ‘ Spontaneous epigenetic variation in the Arabidopsis thaliana methylome ’, Nature , 480 ( 7376 ), pp. 245 – 249 . OpenUrl CrossRef PubMed Web of Science ↵ Bensmihen , S. et al. ( 2004 ) ‘ Analysis of an activated ABI5 allele using a new selection method for transgenic Arabidopsis seeds ’, FEBS letters , 561 ( 1-3 ), pp. 127 – 131 . OpenUrl CrossRef PubMed Web of Science ↵ Boom , R. et al. ( 1990 ) ‘ Rapid and simple method for purification of nucleic acids ’, Journal of clinical microbiology , 28 ( 3 ), pp. 495 – 503 . OpenUrl Abstract / FREE Full Text ↵ Boulias , K. and Greer , E.L . ( 2022 ) ‘ Means, mechanisms and consequences of adenine methylation in DNA ’, Nature reviews. Genetics , 23 ( 7 ), pp. 411 – 428 . OpenUrl CrossRef PubMed ↵ Bourguet , P. et al. ( 2020 ) ‘ DNA polymerase epsilon is required for heterochromatin maintenance in Arabidopsis ’, Genome biology , 21 ( 1 ), p. 283 . OpenUrl PubMed ↵ Briffa , A. et al. ( 2023 ) ‘ Millennia-long epigenetic fluctuations generate intragenic DNA methylation variance in Arabidopsis populations ’, Cell systems , 14 ( 11 ), pp. 953 – 967.e17 . OpenUrl PubMed ↵ Brzeski , J. and Jerzmanowski , A . ( 2003 ) ‘ Deficient in DNA methylation 1 (DDM1) defines a novel family of chromatin-remodeling factors ’, The Journal of biological chemistry , 278 ( 2 ), pp. 823 – 828 . OpenUrl Abstract / FREE Full Text ↵ Buchfink , B. , Reuter , K. and Drost , H.-G . ( 2021 ) ‘ Sensitive protein alignments at tree-of-life scale using DIAMOND ’, Nature methods , 18 ( 4 ), pp. 366 – 368 . OpenUrl PubMed ↵ Bushnell , B. ( 2014 ) BBMap: A fast, accurate, splice-aware aligner . Available at: https://www.osti.gov/servlets/purl/1241166 . (Accessed: 8 November 2024 ). ↵ Cao , X. and Jacobsen , S.E . ( 2002 ) ‘ Locus-specific control of asymmetric and CpNpG methylation by the DRM and CMT3 methyltransferase genes ’, Proceedings of the National Academy of Sciences of the United States of America , 99 Suppl 4( Suppl 4 ), pp. 16491 – 16498 . OpenUrl Abstract / FREE Full Text ↵ Charlesworth , D. , Barton , N.H. and Charlesworth , B . ( 2017 ) ‘ The sources of adaptive variation ’, Proceedings. Biological sciences / The Royal Society , 284 ( 1855 ). Available at : doi: 10.1098/rspb.2016.2864 . OpenUrl CrossRef PubMed ↵ Chatterjee , A. et al. ( 2015 ) ‘ Genome-wide DNA methylation map of human neutrophils reveals widespread inter-individual epigenetic variation ’, Scientific reports , 5 ( 1 ), p. 17328 . OpenUrl PubMed ↵ Choi , J. , Lyons , D.B. and Zilberman , D . ( 2021 ) ‘ Histone H1 prevents non-CG methylation-mediated small RNA biogenesis in Arabidopsis heterochromatin ’, eLife , 10 , p. e72676 . OpenUrl CrossRef PubMed ↵ Clavel , M. et al. ( 2024 ) ‘ Metabolic enzymes moonlight as selective autophagy receptors to protect plants against viral-induced cellular damage ’, bioRxiv. Available at : doi: 10.1101/2024.05.06.590709 . OpenUrl Abstract / FREE Full Text ↵ Deleris , A. et al. ( 2012 ) ‘ Loss of the DNA methyltransferase MET1 Induces H3K9 hypermethylation at PcG target genes and redistribution of H3K27 trimethylation to transposons in Arabidopsis thaliana ’, PLoS genetics , 8 ( 11 ), p. e1003062 . OpenUrl ↵ Dobin , A. et al. ( 2013 ) ‘ STAR: ultrafast universal RNA-seq aligner ’, Bioinformatics , 29 ( 1 ), pp. 15 – 21 . OpenUrl CrossRef PubMed Web of Science ↵ Dubin , M.J. et al. ( 2015 ) ‘ DNA methylation in Arabidopsis has a genetic basis and shows evidence of local adaptation ’, eLife , 4 , p. e05255 . OpenUrl CrossRef PubMed ↵ Du , J. et al. ( 2015 ) ‘ DNA methylation pathways and their crosstalk with histone methylation ’, Nature reviews. Molecular cell biology , 16 ( 9 ), pp. 519 – 532 . OpenUrl CrossRef PubMed ↵ Emms , D.M. and Kelly , S . ( 2019 ) ‘ OrthoFinder: phylogenetic orthology inference for comparative genomics ’, Genome biology , 20 ( 1 ), p. 238 . OpenUrl CrossRef PubMed ↵ Evans , R. et al. ( 2022 ) ‘ Protein complex prediction with AlphaFold-Multimer ’, bioRxiv. Available at : doi: 10.1101/2021.10.04.463034 . OpenUrl Abstract / FREE Full Text ↵ Ewels , P.A. et al. ( 2020 ) ‘ The nf-core framework for community-curated bioinformatics pipelines ’, Nature biotechnology , 38 ( 3 ), pp. 276 – 278 . OpenUrl CrossRef PubMed ↵ Finnegan , E.J. , Peacock , W.J. and Dennis , E.S . ( 1996 ) ‘ Reduced DNA methylation in Arabidopsis thaliana results in abnormal plant development ’, Proceedings of the National Academy of Sciences of the United States of America , 93 ( 16 ), pp. 8449 – 8454 . OpenUrl Abstract / FREE Full Text ↵ FitzGerald , J. et al. ( 2008 ) ‘ DNA methylation causes predominant maternal controls of plant embryo growth ’, PloS one , 3 ( 5 ), p. e2298 . OpenUrl CrossRef PubMed ↵ Fraga , M.F. et al. ( 2005 ) ‘ Epigenetic differences arise during the lifetime of monozygotic twins ’, Proceedings of the National Academy of Sciences of the United States of America , 102 ( 30 ), pp. 10604 – 10609 . OpenUrl Abstract / FREE Full Text ↵ Funabiki , H. et al. ( 2023 ) ‘ Coevolution of the CDCA7-HELLS ICF-related nucleosome remodeling complex and DNA methyltransferases ’, eLife . Available at : doi: 10.7554/elife.86721 . OpenUrl CrossRef ↵ García-Nafría , J. , Watson , J.F. and Greger , I.H . ( 2016 ) ‘ IVA cloning: A single-tube universal cloning system exploiting bacterial In Vivo Assembly ’, Scientific reports , 6 ( 1 ), p. 27459 . OpenUrl PubMed ↵ Gaudinier , A. and Blackman , B.K . ( 2020 ) ‘ Evolutionary processes from the perspective of flowering time diversity ’, The new phytologist , 225 ( 5 ), pp. 1883 – 1898 . OpenUrl PubMed ↵ Gnan , S. , Priest , A. and Kover , P.X . ( 2014 ) ‘ The genetic basis of natural variation in seed size and seed number and their trade-off using Arabidopsis thaliana MAGIC lines ’, Genetics , 198 ( 4 ), pp. 1751 – 1758 . OpenUrl Abstract / FREE Full Text ↵ Greenberg , M.V.C. and Bourc’his , D . ( 2019 ) ‘ The diverse roles of DNA methylation in mammalian development and disease ’, Nature reviews. Molecular cell biology , 20 ( 10 ), pp. 590 – 607 . OpenUrl CrossRef PubMed ↵ Grützner , R. et al. ( 2021 ) ‘High-efficiency genome editing in plants mediated by a Cas9 gene containing multiple introns’ , Plant Communications , 2 ( 2 ), p. 100135 . OpenUrl PubMed ↵ Gu , Z. , Eils , R. and Schlesner , M . ( 2016 ) ‘Complex heatmaps reveal patterns and correlations in multidimensional genomic data’, Bioinformatics (Oxford , England ) , 32 ( 18 ), pp. 2847 – 2849 . OpenUrl ↵ Hardikar , S. et al. ( 2020 ) ‘ The ZBTB24-CDCA7 axis regulates HELLS enrichment at centromeric satellite repeats to facilitate DNA methylation ’, Protein & cell , 11 ( 3 ), pp. 214 – 218 . OpenUrl CrossRef PubMed ↵ Hardikar , S. et al. ( 2024 ) ‘ The ICF syndrome protein CDCA7 harbors a unique DNA binding domain that recognizes a Cp G dyad in the context of a non-B DNA ’, Science advances , 10 ( 34 ), p. eadr0036 . OpenUrl PubMed ↵ He , L. et al. ( 2022 ) ‘ DNA methylation-free Arabidopsis reveals crucial roles of DNA methylation in regulating gene expression and development ’, Nature communications , 13 ( 1 ), pp. 1 – 16 . OpenUrl PubMed ↵ Hofmann , F. , Schon , M.A. and Nodine , M.D . ( 2019 ) ‘ The embryonic transcriptome of Arabidopsis thaliana ’, Plant reproduction , 32 ( 1 ), pp. 77 – 91 . OpenUrl CrossRef PubMed ↵ Hsu , C.-W. , Lo , C.-Y. and Lee , C.-R . ( 2019 ) ‘ On the postglacial spread of human commensal Arabidopsis thaliana: journey to the East ’, The New phytologist , 222 ( 3 ), pp. 1447 – 1457 . OpenUrl PubMed ↵ Hüther , P. et al. ( 2022 ) ‘MethylScore, a pipeline for accurate and context-aware identification of differentially methylated regions from population-scale plant whole-genome bisulfite sequencing data’ , Quantitative Plant Biology , 3 , p. e19 . OpenUrl ↵ Igolkina , A.A. et al. ( 2024 ) ‘ Towards an unbiased characterization of genetic polymorphism ’, bioRxiv . Available at : doi: 10.1101/2024.05.30.596703 . OpenUrl Abstract / FREE Full Text ↵ Ikeda , Y. et al. ( 2017 ) ‘ Arabidopsis proteins with a transposon-related domain act in gene silencing ’, Nature communications , 8 ( 1 ), pp. 1 – 10 . OpenUrl PubMed ↵ Incarbone , M. et al. ( 2023 ) ‘ Salicylic acid and RNA interference mediate antiviral immunity of plant stem cells ’, Proceedings of the National Academy of Sciences of the United States of America , 120 ( 42 ), p. e2302069120 . OpenUrl CrossRef PubMed ↵ Ito , T. et al. ( 2015 ) ‘ Genome-wide negative feedback drives transgenerational DNA methylation dynamics in Arabidopsis ’, PLoS genetics , 11 ( 4 ), p. e1005154 . OpenUrl PubMed ↵ Jamge , B. et al. ( 2023 ) ‘ Histone variants shape chromatin states in Arabidopsis ’, eLife , 12 . Available at : doi: 10.7554/eLife.87714 . OpenUrl CrossRef PubMed ↵ Jeltsch , A. and Jurkowska , R.Z . ( 2014 ) ‘ New concepts in DNA methylation ’, Trends in biochemical sciences , 39 ( 7 ), pp. 310 – 318 . OpenUrl CrossRef PubMed Web of Science ↵ Jenness , C. et al. ( 2018 ) ‘ HELLS and CDCA7 comprise a bipartite nucleosome remodeling complex defective in ICF syndrome ’, Proceedings of the National Academy of Sciences of the United States of America , 115 ( 5 ), pp. E876 – E885 . OpenUrl Abstract / FREE Full Text ↵ Johannes , F. and Schmitz , R.J . ( 2019 ) ‘ Spontaneous epimutations in plants ’, The new phytologist , 221 ( 3 ), pp. 1253 – 1259 . OpenUrl PubMed ↵ Kader , F. and Ghai , M . ( 2017 ) ‘ DNA methylation-based variation between human populations ’, Molecular genetics and genomics: MGG , 292 ( 1 ), pp. 5 – 35 . OpenUrl ↵ Kakutani , T. et al. ( 1996 ) ‘ Developmental abnormalities and epimutations associated with DNA hypomethylation mutations ’, Proceedings of the National Academy of Sciences of the United States of America , 93 ( 22 ), pp. 12406 – 12411 . OpenUrl Abstract / FREE Full Text ↵ Kakutani , T. et al. ( 1999 ) ‘ Meiotically and mitotically stable inheritance of DNA hypomethylation induced by ddm1 mutation of Arabidopsis thaliana ’, Genetics , 151 ( 2 ), pp. 831 – 838 . OpenUrl Abstract / FREE Full Text ↵ Kang , H.M. et al. ( 2008 ) ‘ Efficient control of population structure in model organism association mapping ’, Genetics , 178 ( 3 ), pp. 1709 – 1723 . OpenUrl Abstract / FREE Full Text ↵ Katoh , K. and Standley , D.M . ( 2013 ) ‘ MAFFT multiple sequence alignment software version 7: improvements in performance and usability ’, Molecular biology and evolution , 30 ( 4 ), pp. 772 – 780 . OpenUrl CrossRef PubMed Web of Science ↵ Kawakatsu , T. et al. ( 2016 ) ‘ Epigenomic Diversity in a Global Collection of Arabidopsis thaliana Accessions ’, Cell , 166 ( 2 ), pp. 492 – 505 . OpenUrl CrossRef PubMed ↵ Kinoshita , T . ( 2004 ) ‘ One-Way Control of FWA Imprinting in Arabidopsis Endosperm by DNA Methylation ’, Science , pp. 521 – 523 . Available at : doi: 10.1126/science.1089835 . OpenUrl Abstract / FREE Full Text ↵ Klepikova , A.V. et al. ( 2016 ) ‘ A high resolution map of the Arabidopsis thaliana developmental transcriptome based on RNA-seq profiling ’, The Plant journal: for cell and molecular biology , 88 ( 6 ), pp. 1058 – 1070 . OpenUrl ↵ Kornienko , A.E. et al. ( 2023 ) ‘Population-level annotation of lncRNAs in Arabidopsis reveals extensive expression variation associated with transposable element–like silencing’ , The Plant Cell , koad233 . Available at: https://academic.oup.com/plcell/advance-article-abstract/doi/10.1093/plcell/koad233/7264822 . ↵ Kuzmin , E. , Taylor , J.S. and Boone , C . ( 2022 ) ‘ Retention of duplicated genes in evolution ’, Trends in genetics: TIG , 38 ( 1 ), pp. 59 – 72 . OpenUrl PubMed ↵ Lampropoulos , A. et al. ( 2013 ) ‘ GreenGate---a novel, versatile, and efficient cloning system for plant transgenesis ’, PloS one , 8 ( 12 ), p. e83043 . OpenUrl CrossRef PubMed ↵ Law , J.A. and Jacobsen , S.E . ( 2010 ) ‘ Establishing, maintaining and modifying DNA methylation patterns in plants and animals ’, Nature reviews. Genetics , 11 ( 3 ), pp. 204 – 220 . OpenUrl CrossRef PubMed Web of Science ↵ Lee , C.-R. et al. ( 2017 ) ‘ On the post-glacial spread of human commensal Arabidopsis thaliana ’, Nature communications , 8 , p. 14458 . OpenUrl PubMed ↵ Lee , S.C. et al. ( 2023 ) ‘ Chromatin remodeling of histone H3 variants by DDM1 underlies epigenetic inheritance of DNA methylation ’, Cell , 186 ( 19 ), pp. 4100 – 4116.e15 . OpenUrl CrossRef PubMed ↵ Leigh , J.W. and Bryant , D . ( 2015 ) ‘ Popart: Full-feature software for haplotype network construction ’, Methods in ecology and evolution , 6 ( 9 ), pp. 1110 – 1116 . OpenUrl ↵ Letunic , I. and Bork , P . ( 2024 ) ‘ Interactive Tree of Life (iTOL) v6: recent updates to the phylogenetic tree display and annotation tool ’, Nucleic acids research , 52 ( W1 ), pp. W78 – W82 . OpenUrl CrossRef PubMed ↵ Lippert , C. et al. ( 2014 ) ‘LIMIX: genetic analysis of multiple traits’ , bioRxiv . Available at: https://edoc.mdc-berlin.de/16584/ (Accessed: 25 June 2019 ). ↵ Lippman , Z. et al. ( 2003 ) ‘ Distinct mechanisms determine transposon inheritance and methylation via small interfering RNA and histone modification ’, PLoS biology , 1 ( 3 ), p. E67 . OpenUrl CrossRef PubMed ↵ Liu , L. , Wu , F. and Zhang , W . ( 2016 ) ‘ A Conserved Interaction between SKIP and SMP1/2 Aids in Recruiting the Second-Step Splicing Factors to the Spliceosome in Arabidopsis ’, Molecular plant , 9 ( 12 ), pp. 1660 – 1663 . OpenUrl PubMed ↵ Love , M.I. , Huber , W. and Anders , S . ( 2014 ) ‘ Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 ’, Genome biology , 15 ( 12 ), p. 550 . OpenUrl CrossRef PubMed ↵ Lyons , D.B. and Zilberman , D . ( 2017 ) ‘ DDM1 and Lsh remodelers allow methylation of DNA wrapped in nucleosomes ’, eLife , 6 . Available at : doi: 10.7554/eLife.30674 . OpenUrl CrossRef PubMed ↵ Mathieu , O. et al. ( 2007 ) ‘ Transgenerational stability of the Arabidopsis epigenome is coordinated by CG methylation ’, Cell , 130 ( 5 ), pp. 851 – 862 . OpenUrl CrossRef PubMed Web of Science ↵ Matzke , A.J. and Matzke , M.A . ( 1998 ) ‘ Position effects and epigenetic silencing of plant transgenes ’, Current opinion in plant biology , 1 ( 2 ), pp. 142 – 148 . OpenUrl CrossRef PubMed Web of Science ↵ Meng , E.C. et al. ( 2023 ) ‘ UCSF ChimeraX: Tools for structure building and analysis ’, Protein science: a publication of the Protein Society , 32 ( 11 ), p. e4792 . OpenUrl CrossRef ↵ Minh , B.Q. et al. ( 2020 ) ‘ IQ-TREE 2: New models and efficient methods for phylogenetic inference in the genomic era ’, Molecular biology and evolution , 37 ( 5 ), pp. 1530 – 1534 . OpenUrl CrossRef PubMed ↵ Mirdita , M. et al. ( 2022 ) ‘ ColabFold: making protein folding accessible to all ’, Nature methods , 19 ( 6 ), pp. 679 – 682 . OpenUrl PubMed ↵ Muyle , A.M. et al. ( 2022 ) ‘ Gene body methylation in plants: Mechanisms, functions, and important implications for understanding evolutionary processes ’, Genome biology and evolution , 14 ( 4 ). Available at : doi: 10.1093/gbe/evac038 . OpenUrl CrossRef PubMed ↵ Nartey , W. , Goodarzi , A.A. and Williams , G.J . ( 2023 ) ‘ Cryo-EM structure of DDM1-HELLS chimera bound to nucleosome reveals a mechanism of chromatin remodeling and disease regulation ’, bioRxiv . Available at : doi: 10.1101/2023.08.09.551721 . OpenUrl Abstract / FREE Full Text ↵ Ni , K. et al. ( 2020 ) ‘ LSH mediates gene repression through macroH2A deposition ’, Nature communications , 11 ( 1 ), p. 5647 . OpenUrl PubMed ↵ Ni , K. and Muegge , K . ( 2021 ) ‘ LSH catalyzes ATP-driven exchange of histone variants macroH2A1 and macroH2A2 ’, Nucleic acids research , 49 ( 14 ), pp. 8024 – 8036 . OpenUrl CrossRef PubMed ↵ Osakabe , A. et al. ( 2021 ) ‘ The chromatin remodeler DDM1 prevents transposon mobility through deposition of histone variant H2A.W ’, Nature cell biology , 23 ( 4 ), pp. 391 – 400 . OpenUrl CrossRef PubMed ↵ Osakabe , A. et al. ( 2024 ) ‘ Molecular and structural basis of the chromatin remodeling activity by Arabidopsis DDM1 ’, Nature communications , 15 ( 1 ), p. 5187 . OpenUrl PubMed ↵ Paradis , E . ( 2018 ) ‘ Analysis of haplotype networks: The randomized minimum spanning tree method ’, Methods in ecology and evolution , 9 ( 5 ), pp. 1308 – 1317 . OpenUrl ↵ Patro , R. et al. ( 2017 ) ‘ Salmon provides fast and bias-aware quantification of transcript expression ’, Nature methods , 14 ( 4 ), pp. 417 – 419 . OpenUrl PubMed ↵ Pisupati , R. et al. ( 2023 ) ‘ On the causes of gene-body methylation variation in Arabidopsis thaliana ’, PLoS genetics , 19 ( 5 ), p. e1010728 . OpenUrl PubMed ↵ Quadrana , L. and Colot , V . ( 2016 ) ‘ Plant transgenerational epigenetics ’, Annual review of genetics , 50 ( 1 ), pp. 467 – 491 . OpenUrl CrossRef PubMed ↵ Ramírez , F. et al. ( 2016 ) ‘ deepTools2: a next generation web server for deep-sequencing data analysis ’, Nucleic acids research , 44 ( W1 ), pp. W160 – 5 . OpenUrl CrossRef PubMed ↵ Richards , E.J . ( 2006 ) ‘ Inherited epigenetic variation--revisiting soft inheritance ’, Nature reviews. Genetics , 7 ( 5 ), pp. 395 – 401 . OpenUrl CrossRef PubMed Web of Science ↵ Romero Charria , P., et al. ( 2024 ) ‘ Adenine DNA methylation associated to transcription is widespread across eukaryotes ’, bioRxiv . Available at : doi: 10.1101/2024.10.28.620566 . OpenUrl Abstract / FREE Full Text ↵ Sasaki , E. et al. ( 2019 ) ‘ Common alleles of CMT2 and NRPE1 are major determinants of CHH methylation variation in Arabidopsis thaliana ’, PLoS genetics , 15 ( 12 ), p. e1008492 . OpenUrl PubMed ↵ Sasaki , E. et al. ( 2021 ) ‘ Revisiting a GWAS peak in Arabidopsis thaliana reveals possible confounding by genetic heterogeneity ’, Heredity [Preprint]. Available at : doi: 10.1038/s41437-021-00456-3 . OpenUrl CrossRef PubMed ↵ Sasaki , E. et al. ( 2022 ) ‘ Conditional GWAS of non-CG transposon methylation in Arabidopsis thaliana reveals major polymorphisms in five genes ’, PLoS genetics , 18 ( 9 ), p. e1010345 . OpenUrl PubMed ↵ Sasaki , E. , Frommlet , F. and Nordborg , M . ( 2018 ) ‘ GWAS with Heterogeneous Data: Estimating the Fraction of Phenotypic Variation Mediated by Gene Expression Data ’, G3 , 8 ( 9 ), pp. 3059 – 3068 . OpenUrl Abstract / FREE Full Text ↵ Saze , H. , Scheid , O.M. and Paszkowski , J . ( 2003 ) ‘ Maintenance of CpG methylation is essential for epigenetic inheritance during plant gametogenesis ’, Nature genetics , 34 ( 1 ), pp. 65 – 69 . OpenUrl CrossRef PubMed Web of Science ↵ Schmid , M.W. et al. ( 2018 ) ‘ Contribution of epigenetic variation to adaptation in Arabidopsis ’, Nature communications , 9 ( 1 ), p. 4446 . OpenUrl PubMed ↵ Schmitz , R.J. and Ecker , J.R . ( 2012 ) ‘ Epigenetic and epigenomic variation in Arabidopsis thaliana ’, Trends in plant science , 17 ( 3 ), pp. 149 – 154 . OpenUrl CrossRef PubMed Web of Science ↵ Schmitz , R.J. , Lewis , Z.A. and Goll , M.G . ( 2019 ) ‘ DNA Methylation: Shared and Divergent Features across Eukaryotes ’, Trends in genetics: TIG , 35 ( 11 ), pp. 818 – 827 . OpenUrl PubMed ↵ Schranz , M.E. , Mohammadin , S. and Edger , P.P . ( 2012 ) ‘ Ancient whole genome duplications, novelty and diversification: the WGD Radiation Lag-Time Model ’, Current opinion in plant biology , 15 ( 2 ), pp. 147 – 153 . OpenUrl CrossRef PubMed ↵ Shimada , A. et al. ( 2024 ) ‘ Retrotransposon addiction promotes centromere function via epigenetically activated small RNAs ’, Nature plants , pp. 1 – 13 . ↵ Shinkai , A. et al. ( 2024 ) ‘ The C-terminal 4CXXC-type zinc finger domain of CDCA7 recognizes hemimethylated DNA and modulates activities of chromatin remodeling enzyme HELLS ’, Nucleic acids research , 52 ( 17 ), pp. 10194 – 10219 . OpenUrl PubMed ↵ Slotte , T. et al. ( 2013 ) ‘ The Capsella rubella genome and the genomic consequences of rapid mating system evolution ’, Nature genetics , 45 ( 7 ), pp. 831 – 835 . OpenUrl CrossRef PubMed ↵ Smith , Z.D. and Meissner , A . ( 2013 ) ‘ DNA methylation: roles in mammalian development ’, Nature reviews. Genetics , 14 ( 3 ), pp. 204 – 220 . OpenUrl CrossRef PubMed ↵ Soppe , W.J.J. et al. ( 2002 ) ‘ DNA methylation controls histone H3 lysine 9 methylation and heterochromatin assembly in Arabidopsis ’, The EMBO journal , 21 ( 23 ), pp. 6549 – 6559 . OpenUrl Abstract / FREE Full Text ↵ Stewart-Morgan , K.R. et al. ( 2023 ) ‘ Quantifying propagation of DNA methylation and hydroxymethylation with iDEMS ’, Nature cell biology , 25 ( 1 ), pp. 183 – 193 . OpenUrl CrossRef PubMed ↵ Stewart-Morgan , K.R. , Petryk , N. and Groth , A . ( 2020 ) ‘ Chromatin replication and epigenetic cell memory ’, Nature cell biology , 22 ( 4 ), pp. 361 – 371 . OpenUrl CrossRef PubMed ↵ Stroud , H. et al. ( 2013 ) ‘ Comprehensive analysis of silencing mutants reveals complex regulation of the Arabidopsis methylome ’, Cell , 152 ( 1-2 ), pp. 352 – 364 . OpenUrl CrossRef PubMed Web of Science ↵ Tariq , M. et al. ( 2003 ) ‘ Erasure of CpG methylation in Arabidopsis alters patterns of histone H3 methylation in heterochromatin ’, Proceedings of the National Academy of Sciences of the United States of America , 100 ( 15 ), pp. 8823 – 8827 . OpenUrl Abstract / FREE Full Text ↵ Thijssen , P.E. et al. ( 2015 ) ‘ Mutations in CDCA7 and HELLS cause immunodeficiency-centromeric instability-facial anomalies syndrome ’, Nature communications , 6 , p. 7870 . OpenUrl PubMed ↵ Unoki , M . ( 2021 ) ‘ Chromatin remodeling in replication-uncoupled maintenance DNA methylation and chromosome stability: Insights from ICF syndrome studies ’, Genes to cells: devoted to molecular & cellular mechanisms , 26 ( 6 ), pp. 349 – 359 . OpenUrl PubMed ↵ Venney , C.J. et al. ( 2023 ) ‘ The evolutionary complexities of DNA methylation in animals: From plasticity to genetic evolution ’, Genome biology and evolution , 15 ( 12 ). Available at : doi: 10.1093/gbe/evad216 . OpenUrl CrossRef ↵ Vidigal , D.S. et al. ( 2016 ) ‘ Altitudinal and climatic associations of seed dormancy and flowering traits evidence adaptation of annual life cycle timing in Arabidopsis thaliana ’, Plant, cell & environment , 39 ( 8 ), pp. 1737 – 1748 . OpenUrl CrossRef ↵ Vongs , A. et al. ( 1993 ) ‘Arabidopsis thaliana DNA methylation mutants’, Science (New York , N.Y .) , 260 ( 5116 ), pp. 1926 – 1928 . OpenUrl ↵ Vukic , M. et al. ( 2024 ) ‘ CDCA7-associated global aberrant DNA hypomethylation translates to localized, tissue-specific transcriptional responses ’, Science advances , 10 ( 6 ), p. eadk3384 . OpenUrl CrossRef PubMed ↵ Wassing , I.E. et al. ( 2024 ) ‘ CDCA7 is an evolutionarily conserved hemimethylated DNA sensor in eukaryotes ’, Science advances , 10 ( 34 ), p. eadp5753 . OpenUrl PubMed ↵ Wickham , H . ( 2016 ) Ggplot2 [PDF] . 2nd edn . Basel, Switzerland : Springer International Publishing (Use R!). ↵ Wickham , H. et al. ( 2019 ) ‘ Welcome to the tidyverse ’, Journal of open source software , 4 ( 43 ), p. 1686 . OpenUrl ↵ Wolpe , J.B. , Martins , A.L. and Guertin , M.J . ( 2023 ) ‘ Correction of transposase sequence bias in ATAC-seq data with rule ensemble modeling ’, NAR genomics and bioinformatics , 5 ( 2 ), p. lqad054 . OpenUrl ↵ Xiao , W. et al. ( 2006 ) ‘ Regulation of seed size by hypomethylation of maternal and paternal genomes ’, Plant physiology , 142 ( 3 ), pp. 1160 – 1168 . OpenUrl Abstract / FREE Full Text ↵ Yu , J. et al. ( 2006 ) ‘ A unified mixed-model method for association mapping that accounts for multiple levels of relatedness ’, Nature genetics , 38 ( 2 ), pp. 203 – 208 . OpenUrl CrossRef PubMed Web of Science ↵ Zemach , A. et al. ( 2013 ) ‘ The Arabidopsis nucleosome remodeler DDM1 allows DNA methyltransferases to access H1-containing heterochromatin ’, Cell , 153 ( 1 ), pp. 193 – 205 . OpenUrl CrossRef PubMed Web of Science ↵ Zhang , H. , Lang , Z. and Zhu , J.-K. ( 2018 ) ‘Dynamics and function of DNA methylation in plants’ , Nature reviews. Molecular cell biology , 19 ( 8 ), pp. 489 – 506 . OpenUrl CrossRef PubMed ↵ Zhang , R. et al. ( 2022 ) ‘ A high-resolution single-molecule sequencing-based Arabidopsis transcriptome using novel methods of Iso-seq analysis ’, Genome biology , 23 ( 1 ), p. 149 . OpenUrl CrossRef PubMed ↵ Zhang , Y. et al. ( 2020 ) ‘ Natural variation in DNA methylation homeostasis and the emergence of epialleles ’, Proceedings of the National Academy of Sciences of the United States of America , 117 ( 9 ), pp. 4874 – 4884 . OpenUrl Abstract / FREE Full Text ↵ Zheng , X. et al. ( 2017 ) ‘Transgenerational epimutations induced by multi-generation drought imposition mediate rice plant’s adaptation to drought condition’ , Scientific reports , 7 , p. 39843 . OpenUrl PubMed ↵ Zhou , J. et al. ( 2023 ) ‘ DDM1-mediated R-loop resolution and H2A.Z exclusion facilitates heterochromatin formation in Arabidopsis ’, Science advances , 9 ( 32 ), p. eadg2699 . OpenUrl PubMed ↵ Zilberman , D. et al. ( 2008 ) ‘ Histone H2A.Z and DNA methylation are mutually antagonistic chromatin marks ’, Nature , 456 ( 7218 ), pp. 125 – 129 . OpenUrl CrossRef PubMed Web of Science ↵ Ziller , M.J. et al. ( 2011 ) ‘ Genomic distribution and inter-sample variation of non-CpG methylation across human cell types ’, PLoS genetics , 7 ( 12 ), p. e1002389 . OpenUrl PubMed ↵ Zou , Y.-P. et al. ( 2017 ) ‘ Adaptation of Arabidopsis thaliana to the Yangtze River basin ’, Genome biology , 18 ( 1 ). Available at : doi: 10.1186/s13059-017-1378-9 . OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted September 07, 2025. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Major alleles of CDCA7α shape CG-methylation in Arabidopsis thaliana Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Major alleles of CDCA7α shape CG-methylation in Arabidopsis thaliana Pierre Bourguet , Zdravko J Lorković , Darya Kripkiy Casado , Valentin Bapteste , Chung Hyun Cho , Anna Igolkina , Cheng-Ruei Lee , Magnus Nordborg , Frédéric Berger , Eriko Sasaki bioRxiv 2025.09.03.673934; doi: https://doi.org/10.1101/2025.09.03.673934 Share This Article: Copy Citation Tools Major alleles of CDCA7α shape CG-methylation in Arabidopsis thaliana Pierre Bourguet , Zdravko J Lorković , Darya Kripkiy Casado , Valentin Bapteste , Chung Hyun Cho , Anna Igolkina , Cheng-Ruei Lee , Magnus Nordborg , Frédéric Berger , Eriko Sasaki bioRxiv 2025.09.03.673934; doi: https://doi.org/10.1101/2025.09.03.673934 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Plant Biology Subject Areas All Articles Animal Behavior and Cognition (7618) Biochemistry (17636) Bioengineering (13861) Bioinformatics (41848) Biophysics (21401) Cancer Biology (18538) Cell Biology (25424) Clinical Trials (138) Developmental Biology (13354) Ecology (19861) Epidemiology (2067) Evolutionary Biology (24287) Genetics (15585) Genomics (22463) Immunology (17701) Microbiology (40300) Molecular Biology (17141) Neuroscience (88437) Paleontology (666) Pathology (2825) Pharmacology and Toxicology (4813) Physiology (7633) Plant Biology (15107) Scientific Communication and Education (2042) Synthetic Biology (4285) Systems Biology (9808) Zoology (2268)