RNA m6A regulates the transcription and heterochromatin state of retrotransposons in Arabidopsis

Song, Peizhe; Cai, Zhihe; Tayier, Subiding; Tian, Enlin; Chen, Zixin; Yu, Kemiao; Liu, Lixiang; Jia, Guifang

doi:10.1038/s41477-025-02137-z

Download PDF

Article
Open access
Published: 24 October 2025

RNA m⁶A regulates the transcription and heterochromatin state of retrotransposons in Arabidopsis

Nature Plants volume 11, pages 2300–2318 (2025)Cite this article

10k Accesses
3 Citations
32 Altmetric
Metrics details

Subjects

Abstract

Chromatin and transcription states are crucial for establishing and maintaining gene regulation and genome integrity, governed by a complex network of epigenetic modifications. Despite significant advancements, the interplay between epitranscriptomic and epigenetic modifications in transcriptional regulation and chromatin state remains elusive in plants. Here we profile a comprehensive spectrum of N⁶-methyladenosine (m⁶A) on chromatin-associated RNAs, especially retrotransposon transcripts, in Arabidopsis. We show that m⁶A writer subunits FIP37 and VIR mediate the methylation of retrotransposon RNAs, which are recognized by the nuclear m⁶A reader proteins CPSF30-L and ECT12, leading to a closed heterochromatin state and transcription suppression of m⁶A retrotransposon RNAs. Through monitoring nuclear RNA stability and transcription rate upon the depletion of either writer or reader, we provide evidence consistent with a model in which CPSF30-L associates with SUVH4/5/6 and ATXR5/6 and contributes to histone H3 K9 dimethylation and H3 K27 monomethylation at m⁶A-marked retrotransposons. Our findings elucidate a regulatory mechanism involving RNA m⁶A formation and recognition, coupled with histone modifications, thereby highlighting the intricate epitranscriptomic and epigenetic interplay in plant physiological and biological processes.

Quantitative profiling of m⁶A at single base resolution across the life cycle of rice and Arabidopsis

Article Open access 07 June 2024

Biological roles of adenine methylation in RNA

Article 19 October 2022

RNA m⁶A regulates transcription via DNA demethylation and chromatin accessibility

Article 07 September 2022

Main

Over 170 types of RNA modifications have been identified across all kingdoms of life. N⁶-methyladenosine (m⁶A) is the most widespread and highly conserved modification in mRNAs and noncoding RNAs (ncRNAs)¹. In Arabidopsis, m⁶A is co-transcriptionally incorporated into RNA transcripts by the MTA–MTB complex and several accessory subunits^2,3,4, erased by demethylases ALKBH9B/10B^5,6 and recognized by readers such as YTH-domain-containing proteins^{7,8,9,10,11,12,13}. The m⁶A modification regulates many critical post-transcriptional processes in the cytoplasm and nucleus in both plants and mammals, such as mRNA stability, RNA–protein granule formation, alternative polyadenylation and the resolution of R-loops^{8,9,10,11,13,14}. Additionally, chromatin states and transcription can be modulated by m⁶A methylation in critical nuclear RNAs, especially chromatin-associated regulatory RNAs, in mammalian cells^{15,16,17,18,19}. These studies suggest that m⁶A modification in different RNA types plays distinctive roles in the epitranscriptomic regulation of gene expression.

Transposons are DNA molecules capable of moving from one location to another and are widespread in most eukaryotic genomes²⁰. Retrotransposons, classified as class I transposons, proliferate through reverse transcription of RNA intermediates and comprise over 40% and 70% of the human and maize genomes, respectively^21,22,23. Due to their potentially adverse effects on host genomes, most transposons are extensively repressed by epigenetic marks, including DNA methylation and histone modifications. Plants and mammals exhibit striking divergence in heterochromatin-associated histone modifications: histone H3 K9 dimethylation and H3 K27 monomethylation serve as heterochromatin marks in plants, whereas mammals primarily use H3 K9 trimethylation for heterochromatin assembly, with H3 K9 dimethylation predominantly marking euchromatic regions^24,25. Recent independent studies have revealed that METTL3-mediated m⁶A methylation on retrotransposons orchestrates with H3 K9 trimethylation to maintain heterochromatin formation, thereby regulating retrotransposon repression and preserving embryonic stem cell identity^18,26. FTO-mediated m⁶A demethylation on LINE1 RNA regulates its abundance and the local chromatin state during mouse oocyte and embryonic development¹⁶. Moreover, the expression of human FTO in rice demethylates m⁶A and promotes chromatin accessibility and transcriptional activation²⁷, suggesting that plant endogenous m⁶A modification might play a role in regulating chromatin state and transcription. However, it remains unknown whether and how m⁶A regulates chromatin state and transcription in plants.

Here we optimized m⁶A antibody immunoprecipitation sequencing (m⁶A-Seq) for plant chromatin-associated RNA (caRNA) and profiled the methylome landscape of retrotransposons in Arabidopsis. m⁶A methylation on retrotransposon RNAs, deposited by m⁶A writers, is recognized by the nuclear m⁶A reader protein CPSF30-L. CPSF30-L physically interacts with histone methylation writers SUVH4/5/6 and ATXR5/6 to increase the repressive histone marks H3 K9 dimethylation and H3 K27 monomethylation in the local heterochromatin region, leading to closed chromatin accessibility and transcription suppression. Moreover, ECT12 is characterized as another nuclear m⁶A reader protein, which interacts with CPSF30-L and redundantly regulates retrotransposon RNA transcription and the local heterochromatic histone levels. Our results uncover a layer of epigenetic regulation in which RNA methylation couples with histone modifications to ensure silenced transcription and heterochromatin integrity in plants.

Results

Landscape of m⁶A retrotransposon RNAs revealed via caRNA m⁶A-Seq in Arabidopsis

To unveil the transcriptional spectrum of m⁶A-methylated RNA molecules, we optimized m⁶A-Seq suitable for plant caRNA, referred to as caRNA m⁶A-Seq (Fig. 1a). This method was developed on the basis of the efficient isolation of non-DNA-contaminated and non-ribosomal caRNAs followed by m⁶A immunoprecipitation (m⁶A-IP) (Fig. 1b and Supplementary Fig. 1a,b). We also incorporated m⁶A spike-in and External RNA Controls Consortium (ERCC) spike-ins to calibrate the efficiency of m⁶A-IP and the variability of expression levels, respectively.

Fig. 1: m⁶A methylome profiling on retrotransposon RNAs in *Arabidopsis.*

To compare the m⁶A profiles of transcriptional transcriptomes and steady-state transcriptomes, we conducted parallel extractions of caRNAs and processed mRNAs (poly(A)⁺ RNAs) from two biological replicates of six-day-old wild-type (WT) seedlings and performed caRNA m⁶A-Seq and canonical m⁶A-Seq. Correlation analysis of paired input and m⁶A-IP sets between caRNAs and poly(A)⁺ RNAs was conducted independently for each set (Supplementary Fig. 1c). Considering that newly transcribed transcripts contain processing intermediates, we classified the reads into exonic, intronic, exon/intron junction and intergenic categories. Notably, caRNA sequencing (caRNA-Seq) revealed a more than fivefold higher occupancy of reads in intronic, exon/intron junction and intergenic regions compared with poly(A)⁺ RNA (Fig. 1c and Supplementary Fig. 1d), suggesting that caRNA m⁶A-Seq can specifically capture nascent transcripts. On the basis of the analysis criteria (false discovery rate < 0.05), 11,466 and 11,227 m⁶A peaks were exclusively detected in caRNAs and poly(A)⁺ RNAs, respectively (Supplementary Fig. 1e and Supplementary Tables 1 and 2).

Comparing identified m⁶A sites between caRNAs and poly(A)⁺ RNAs, we observed an increased number of m⁶A peaks in exonic regions in caRNA m⁶A-Seq (Fig. 1d). This difference may be attributed to variations in transcript composition, particularly the enrichment of nascent RNAs in which RNA Polymerase II (Pol II) has not yet completed transcription of the full 3′ untranslated region (UTR), as well as insufficient m⁶A demethylation during transcription. Moreover, we observed that a high number of m⁶A peaks were enriched in transposon RNAs and other noncoding RNA regions (Fig. 1d and Supplementary Fig. 1f). We computationally interrogated the m⁶A levels of these three caRNA types with spike-ins for calibration. Intriguingly, the m⁶A levels of transposon RNAs are higher than those of pre-mRNAs and ncRNAs (Fig. 1e).

Retrotransposons and DNA transposons are the two primary classes of transposable elements. Through caRNA m⁶A-Seq analysis, we identified 242 m⁶A peaks on retrotransposon RNAs and 67 on DNA transposon RNAs (Supplementary Fig. 1g). We also observed that retrotransposons are transcribed more actively (Supplementary Fig. 1h), underscoring the preferential involvement of m⁶A in retrotransposon RNA regulation. Interrogating a comprehensive methylome profiling of m⁶A in poly(A)⁺ RNAs and caRNAs, we identified a total of 139 annotated m⁶A-modified retrotransposon RNAs (referred to as m⁶A retrotransposon RNAs) due to the low expression of retrotransposon RNAs (Supplementary Table 3). These m⁶A retrotransposon RNAs are predominantly localized at pericentromeric heterochromatin regions (Supplementary Fig. 1i). Unlike the typical accumulation of m⁶A in the 3′ UTR of mRNA, m⁶A was distributed across the entire transcripts, with particularly high enrichment at the 5′ and 3′ ends of retrotransposon transcripts (Fig. 1f and Supplementary Fig. 1j,k). Motif analysis clustering all m⁶A peaks in retrotransposon RNA identified the WAC (W indicates A or U) motif, consistent with the canonical RAC (R indicates A or G) motif found in mRNAs (Supplementary Fig. 1l). Additionally, m⁶A retrotransposon RNAs exhibited GAG or SAU (S indicates G or C) motifs (Supplementary Fig. 1l), resembling the GAAGG and GGAUU non-canonical motifs found in mammalian retrotransposon RNAs²⁶.

Retrotransposons include long-terminal-repeat (for example, COPIA and GYPSY) and non-long-terminal-repeat retrotransposons (for example, LINE and SINE)²⁸. We observed that COPIA, GYPSY and LINE retrotransposons harbour 108, 88 and 43 m⁶A peaks, corresponding to 48, 61 and 24 transcripts, respectively, whereas SINE displays fewer m⁶A peaks (Fig. 1g, Supplementary Fig. 1m,n and Supplementary Table 3). m⁶A-IP followed by quantitative PCR (m⁶A-IP–qPCR) using COPIA62 and COPIA66 as representative examples confirmed authentic m⁶A enrichment in these identified peak regions but not elsewhere (Fig. 1h,i). The high detectability of methylation on nascent retrotransposon RNAs enables us to explore the role and mechanism of m⁶A in transcriptional regulation.

m⁶A is deposited on retrotransposon RNA transcribed by Pol II and Pol V

Since m⁶A is deposited co-transcriptionally during RNA polymerase transcription, we first investigated which polymerase is responsible for retrotransposon transcription. Unlike mammals, which possess only Pol I, Pol II and Pol III, plants have uniquely evolved two additional RNA polymerases, Pol IV and Pol V²⁹. Pol I is involved in rRNA transcription; Pol II transcribes mRNAs and certain ncRNAs; Pol III synthesizes 5S rRNA, tRNA and other small RNAs; Pol IV produces ~30–40-nucleotide transcripts, primarily from repetitive regions; and Pol V generates noncoding transcripts, mainly from heterochromatic regions^30,31,32. We performed caRNA-Seq using six-day-old seedlings of the WT line, a line carrying the weak Pol II mutant allele nrpb2-3 (ref. ³³) and the Pol V mutant nrpe1-11 (ref. ³⁴) to examine whether the expression levels of retrotransposon RNAs are controlled by Pol II and Pol V. We found that the expression levels of certain retrotransposon RNAs, regardless of m⁶A modification, were decreased in both nrpb2-3 and nrpe1-11 mutants, while some expression levels were specifically reduced in nrpb2-3 or nrpe1-11 (Supplementary Fig. 2a–d and Supplementary Table 4). Our results revealed that retrotransposon transcription is finely regulated by Pol II and Pol V occupancy, with both polymerases coordinately and specifically modulating the transcription of these RNAs.

Considering the distinct RNA products transcribed by each polymerase, we analysed Pol II and Pol V occupancy levels across m⁶A and non-m⁶A retrotransposon RNAs using published data from chromatin immunoprecipitation followed by sequencing (ChIP–Seq)^35,36. We observed that both Pol II and Pol V exhibit higher binding abundance on m⁶A retrotransposons than on non-m⁶A retrotransposons, which aligns with the higher expression levels of m⁶A retrotransposon RNAs than those of non-m⁶A retrotransposons (Supplementary Fig. 2e,f), indicating that m⁶A is deposited on highly expressed transcripts. Furthermore, to examine whether m⁶A is deposited on transcripts transcribed by Pol II and Pol V, we performed formaldehyde-crosslinking-assisted RNA immunoprecipitation (FA-RIP) and quantified m⁶A levels on Pol-II- and Pol-V-bound transcripts using Pol II antibody in six-day-old WT seedlings and for Pol V using anti-GFP beads in six-day-old NRPE1::NRPE1-GFP nrpe1-11 seedlings. The results showed that m⁶A was significantly enriched in Pol II- and Pol V-bound transcripts compared with the input and immunoglobulin G (IgG) control portions, providing direct evidence that m⁶A is deposited on both Pol-II and Pol-V-transcribed transcripts (Supplementary Fig. 2g).

The m⁶A modification on retrotransposon RNAs deposited by the MTA–MTB writer complex represses the expression of retrotransposon RNAs

To investigate the m⁶A-mediated regulatory functions of retrotransposons in Arabidopsis, we employed the subunit mutants of the m⁶A MTA–MTB writer complex: fip37-4 LEC1::FIP37 and vir³⁷, which cause a nearly 90% reduction in m⁶A levels in poly(A)⁺ RNA (Supplementary Fig. 3a). We performed caRNA m⁶A-Seq on six-day-old fip37-4 LEC1::FIP37 and vir mutants with two biological replicates, ensuring high replicability between the replicates (Supplementary Fig. 3b and Supplementary Tables 5 and 6). Compared with WT plants, the disruption of m⁶A writer function in either fip37-4 LEC1::FIP37 or vir globally reduced total m⁶A levels and m⁶A peaks in pre-mRNAs, retrotransposon RNAs and other ncRNAs (Fig. 2a and Supplementary Fig. 3c–e), suggesting that the MTA–MTB writer complex is responsible for m⁶A installation on these RNAs. Almost all m⁶A modifications on retrotransposon RNAs were dramatically decreased in both fip37-4 LEC1::FIP37 and vir (Supplementary Fig. 3f). Furthermore, expression level analysis revealed that over 80% of m⁶A retrotransposon RNAs were upregulated in fip37-4 LEC1::FIP37 and vir (Supplementary Fig. 3g). In contrast to the minimal changes in m⁶A DNA transposon RNA expression, the abundance of m⁶A-modified retrotransposon RNAs, but not non-m⁶A ones, was significantly increased upon the depletion of either FIP37 or VIR (Fig. 2b,c, Supplementary Fig. 3h–k and Supplementary Table 7). To confirm the reliability of m⁶A-writer-mediated regulation of retrotransposon RNAs, we independently conducted caRNA-Seq on six-day-old WT, fip37-4 LEC1::FIP37 and vir seedlings with two biological replicates, which showed consistent results (Supplementary Fig. 3l–n).

**Fig. 2: m⁶A writers regulate retrotransposon m⁶A methylation and retrotransposon repression.**

We further examined the correlation between m⁶A and expression levels across the classified types of retrotransposon RNAs. Retrotransposon RNAs in the COPIA, GYPSY and LINE families, as well as their transcriptionally active subfamilies, displayed lower m⁶A methylation levels and higher expression levels in either fip37-4 LEC1::FIP37 or vir than in WT plants (Fig. 2d). The elevated expression levels of these retrotransposon RNAs were positively correlated with the decreased m⁶A methylation level caused by either FIP37 or VIR depletion (Fig. 2e). Moreover, m⁶A-IP–qPCR and qPCR assays were performed to validate the m⁶A-mediated negative regulation of COPIA22 and ATRE1 transcripts (Fig. 2f–h). Taken together, these results suggest that the MTA–MTB writer complex is required for retrotransposon m⁶A deposition, and m⁶A methylation suppresses retrotransposon RNA abundance.

CPSF30-L binds and suppresses the expression of m⁶A retrotransposon RNAs

The mechanism by which m⁶A guides and decides RNA fate is determined by its reader proteins^38,39. In Arabidopsis, CPSF30-L has been characterized as the nuclear m⁶A reader^9,10. Its co-expression patterns with FIP37 and VIR, along with its nuclear localization, suggest a potential role in binding m⁶A retrotransposon RNAs (Supplementary Fig. 4a). To determine whether the m⁶A-binding function of CPSF30-L regulates the expression of these RNAs, we performed two biological replicates of caRNA-Seq using six-day-old seedlings from WT, cpsf30-l and two complementation lines respectively expressing WT CPSF30-L (CPSF30::CPSF30-L cpsf30-l, termed CPSF30-L) and the CPSF30-L(W259A/W310A) mutant with disrupted m⁶A-binding capability (CPSF30::CPSF30-Lm cpsf30-l, termed CPSF30-Lm) in the cpsf30-l background¹⁰. The pairwise correlation between the WT and CPSF30-L, as well as between cpsf30-l and CPSF30-Lm, showed high reproducibility (Supplementary Fig. 4b). In line with the results obtained when depleting m⁶A methyltransferase subunits (Fig. 2b,c), we observed a substantial increase in the abundance of m⁶A retrotransposon RNAs and typical retrotransposon subfamilies in the cpsf30-l mutant (Fig. 3a–c). Importantly, the alterations observed in cpsf30-l were rescued by expressing CPSF30-L but not the m⁶A-binding-function-abolished CPSF30-Lm (Fig. 3a–c, Supplementary Fig. 4c,d and Supplementary Table 8). We also compared the expression levels of m⁶A and non-m⁶A retrotransposon RNAs, revealing that the expression levels of m⁶A-modified transcripts, but not non-m⁶A RNAs, were significantly higher in both cpsf30-l and CPSF30-Lm than in WT and CPSF30-L plants (Supplementary Fig. 4e). Furthermore, the increased expression levels of m⁶A retrotransposon RNAs in cpsf30-l and CPSF30-Lm plants exhibited a positive correlation with their m⁶A methylation levels (Fig. 3d). These results suggest that the m⁶A-binding function of CPSF30-L suppresses the expression levels of m⁶A retrotransposon RNAs.

Given the nuclear subcellular localization of CPSF30-L, we exploited nuclear enrichment in combination with formaldehyde cross-linking and immunoprecipitation (nuclear FA-CLIP) in six-day-old CPSF30-L and CPSF30-Lm seedlings (Supplementary Fig. 4f). Compared with standard FA-CLIP, nuclear FA-CLIP demonstrated more effective enrichment of m⁶A-modified RNA in the CPSF30-L-IP fraction (Supplementary Fig. 4g). Clustering analysis of two biological replicates confirmed the reproducibility of our nuclear FA-CLIP sequencing data (Supplementary Fig. 4h). CPSF30-L displayed significantly higher binding intensity to m⁶A-modified retrotransposon RNAs than to non-m⁶A RNAs (Fig. 3e). As illustrated in nuclear FA-CLIP and caRNA-Seq, CPSF30-L specifically bound to the m⁶A modification on COPIA22 and ATRE1 and mediated their expression (Fig. 3f). FA-RIP-qPCR and qPCR assays further confirmed that CPSF30-L directly binds to retrotransposon RNAs and suppresses their abundance in an m⁶A-dependent manner (Fig. 3g,h).

We also overlapped the upregulated m⁶A retrotransposon RNAs in m⁶A writer mutants (fip37-4 LEC1::FIP37 and vir) with those in m⁶A reader mutants (cpsf30-l and CPSF30-Lm). The results showed that the depletion of m⁶A writer subunits FIP37 and VIR, and m⁶A reader CPSF30-L, increases the expression of the majority of m⁶A retrotransposon RNAs (Fig. 3i). This further supports the notion that m⁶A methylation on retrotransposon RNAs, recognized by CPSF30-L, contributes to suppressing the expression of these RNAs.

m⁶A has less effect on the stability of retrotransposon RNAs

The increased abundance of m⁶A retrotransposon RNAs may result from m⁶A-meditated transcriptional and/or post-transcriptional regulation. Given its major effect on the lifetime of chromatin-associated regulatory RNA in mammals^15,19, we sought to determine whether m⁶A could destabilize retrotransposon RNAs. We performed nuclear RNA decay assays using six-day-old WT, vir and cpsf30-l seedlings to sequence retrotransposon RNA levels after transcription inhibition with cordycepin, and then calculated the half-life (Supplementary Tables 9 and 10). The half-life of m⁶A retrotransposon RNAs was comparable to that of non-m⁶A retrotransposon RNAs in both vir and cpsf30-l (Fig. 4a and Supplementary Fig. 5a). We also conducted nuclear RNA decay assays and measured the half-life of COPIA22 following transcriptional inhibition with either cordycepin or actinomycin D. The results showed that neither m⁶A removal nor the loss of m⁶A recognition affected the half-life of COPIA22 (Fig. 4b and Supplementary Fig. 5b). Given that the mammalian m⁶A reader YTHDC1 associates with components of the nuclear exosome targeting complex and is involved in retrotransposon RNA degradation¹⁵, we performed yeast two-hybrid (Y2H) assays and found that CPSF30-L does not directly interact with the nuclear exosome targeting components ZCCHC8A, ZCCHC8B and RBM7 (Supplementary Fig. 5c). These results suggest that m⁶A on retrotransposon RNAs barely affects their stability and probably functions at the chromatin level.

**Fig. 4: m⁶A functions in transcriptional repression, but not stability, of retrotransposon RNAs.**

m⁶A silences the transcription of retrotransposon RNAs

The m⁶A modification has been shown to affect chromatin-based transcriptional regulation^15,16,18. To explore this further, we performed time-course RNA sequencing of nascent transcripts in six-day-old WT and cpsf30-l seedlings (Supplementary Table 11). Global analysis revealed that the transcription rates of m⁶A retrotransposon RNAs were significantly upregulated in the cpsf30-l mutant relative to non-m⁶A retrotransposon RNAs (Fig. 4c). Additionally, nuclear run-on assays in six-day-old WT, fip37-4 LEC1::FIP37 and vir seedlings showed that the loss of m⁶A caused a significant upregulation in the transcription rate of retrotransposon RNAs, such as COPIA22 and ATRE1 (Fig. 4d). The increased transcription rates of COPIA22 and ATRE1 were consistently observed in the cpsf30-l mutant, which was restored upon expressing CPSF30-L but not CPSF30-Lm (Fig. 4e). Together, these findings suggest that m⁶A recognized by CPSF30-L suppresses the transcriptional activity of m⁶A-modified retrotransposon RNAs in Arabidopsis.

m⁶A on retrotransposon RNAs maintains local closed heterochromatin by enhancing the histone repressive modifications H3 K9 dimethylation and H3 K27 monomethylation

Transcriptional regulation is a complex and dynamic process in which histone modifications are a key component⁴⁰. We first investigated the association of m⁶A retrotransposon RNAs with heterochromatin and euchromatin histone modifications^41,42. The calculated overlapping ratios (Jaccard statistics) showed that most m⁶A retrotransposon RNAs were associated with two repressive heterochromatin modifications: dimethylation at H3 K9 and monomethylation at H3 K27, with rare associations with other repressive marks or euchromatin modifications (Supplementary Fig. 6a). Moreover, both dimethylated H3 K9 (H3K9me2) and monomethylated H3 K27 (H3K27me1) were highly enriched at the m⁶A position on m⁶A-modified retrotransposons (Fig. 5a). The methylation levels of H3 K9 and H3 K27 are significantly higher at the loci of m⁶A retrotransposon RNAs than at non-m⁶A ones (Supplementary Fig. 6b). The average m⁶A peak densities are positively correlated with H3K9me2 and H3K27me1 levels on m⁶A retrotransposon RNAs (Fig. 5b). These observations suggest that m⁶A may play a role in heterochromatin in Arabidopsis. Immunostaining H3K9me2 and H3K27me1 in isolated nuclei from six-day-old seedlings revealed a slight but clear reduction in their levels in fip37-4 LEC1::FIP37, vir and cpsf30-l (Fig. 5c,d). The reduced H3K9me2 and H3K27me1 signals in cpsf30-l could be restored by expressing CPSF30-L but not the m⁶A-binding-function-abolished CPSF30-Lm (Fig. 5c,d), indicating that the m⁶A binding function of CPSF30-L regulates H3K9me2 and H3K27me1 levels.

**Fig. 5: m⁶A affects heterochromatin state through CPSF30-L-mediated SUVH4/5/6-dependent H3K9me2 and ATXR5/6-dependent H3K27me1 deposition.**

In plants, constitutive heterochromatin forms over repetitive elements, including retrotransposons, which are heavily decorated by H3K9me2 and H3K27me1. These repressive histone modifications prominently maintain heterochromatin integrity and silence transposons in Arabidopsis²⁴. To further investigate whether and how m⁶A contributes to the specialized establishment of heterochromatin states, we conducted genome-wide H3K9me2 with two biological replicates and H3K27me1 ChIP–Seq in six-day-old WT, vir and cpsf30-l seedlings (Supplementary Fig. 6c and Supplementary Tables 12 and 13). Compared with the reduced levels of H3K9me2 and H3K27me1 in their respective writer mutants, the loss of m⁶A writer subunit VIR or m⁶A reader CPSF30-L partially decreased these two histone modification levels (Supplementary Fig. 6d,e), consistent with the immunostaining results (Fig. 5c,d). Notably, H3K9me2 and H3K27me1 levels were significantly reduced on m⁶A-modified retrotransposons and their subfamilies (Fig. 5e and Supplementary Fig. 6f), whereas no significant differences were observed in non-m⁶A retrotransposons between vir and WT or cpsf30-l and WT plants (Supplementary Fig. 6g). To confirm these findings, we also performed ChIP-qPCR assays to examine the H3K9me2 and H3K27me1 levels on COPIA22 and ATRE1 in the genotypic plants. The results demonstrated that m⁶A removal in m⁶A writer mutants or the loss of m⁶A recognition in cpsf30-l decreases H3K9me2 and H3K27me1 levels on COPIA22 and ATRE1 (Fig. 5f and Supplementary Fig. 6h). These results thus suggest that m⁶A modulates H3K9me2 and H3K27me1 deposition at specific m⁶A-modified retrotransposon regions.

Given that both repressive histone modifications, H3K9me2 and H3K27me1, are concurrently present on transposons⁴³, we speculated that they might synergistically contribute to the transcriptional silencing of m⁶A retrotransposon RNAs. To address this, we analysed ChIP–Seq data in the WT and observed substantial overlap between H3K9me2 and H3K27me1 peaks (Supplementary Fig. 6i). Over 60% of m⁶A retrotransposon RNAs harboured both modifications (termed H3K9me2 & H3K27me1 & m⁶A retrotransposon RNAs; Supplementary Fig. 6j). Strikingly, vir and cpsf30-l mutants exhibited pronounced reductions in H3K9me2 and H3K27me1 levels at H3K9me2 & H3K27me1 & m⁶A retrotransposon RNAs (Supplementary Fig. 6k). Moreover, lacking functions of either m⁶A methylation or m⁶A recognition leads to a greater increase in the expression of m⁶A retrotransposon RNAs carrying both histone modifications compared with those marked by individual histone modifications (Supplementary Fig. 6l,m).

Considering that the decrease in repressive histone modifications potentially opens the chromatin state and influences nearby gene expression⁴⁴, we categorized the distance of genes to the nearest m⁶A retrotransposon RNAs, ranging from 0 to 1 Mb, to explore the effect of m⁶A regulation on neighbouring gene expression. The results showed that the expression of genes adjacent to m⁶A retrotransposon RNAs tends to get lower as the distance increases in both vir and cpsf30-l mutants (Supplementary Fig. 7a), accompanied by gradually increased H3K9me2 and H3K27me1 levels (Supplementary Fig. 7b,c). However, these trends remained largely unchanged for non-m⁶A retrotransposon RNAs (Supplementary Fig. 7d–f). These results indicate that the loss of deposition and recognition of m⁶A compromises local and adjacent heterochromatin integrity and transcription.

The m⁶A reader protein CPSF30-L associates with SUVH4/5/6 and ATXR5/6 and is required for H3K9me2/H3K27me1 levels at m⁶A retrotransposons

We next investigated the mechanism underlying histone modifications and heterochromatin repression mediated by m⁶A. Since H3K9me2 is methylated by SUVH4 (also known as KYP), SUVH5 and SUVH6 methyltransferases^45,46,47,48, and H3K27me1 is methylated by ATXR5 and ATXR6 (ref. ⁴⁹), we analysed publicly available ChIP–Seq data for H3K9me2 and H3K27me1 in suvh4 suvh5 suvh6 and atxr5 atxr6 mutants^50,51. Our analysis revealed a substantial reduction in H3K9me2 and H3K27me1 levels at the regions of m⁶A-modified retrotransposons in suvh4 suvh5 suvh6 and atxr5 atxr6 (Supplementary Fig. 8a,b), confirming the roles of these methyltransferases in depositing H3K9me2 and H3K27me1 at the local regions of m⁶A retrotransposon RNAs. Accordingly, we classified H3K9me2 peaks into two categories: SUVH4/5/6-dependent (decreased in suvh4 suvh5 suvh6 mutants) and SUVH4/5/6-independent (unaltered in suvh4 suvh5 suvh6 mutants). Similarly, H3K27me1 peaks were classified into ATXR5/6-dependent (decreased in atxr5 atxr6 mutants) and ATXR5/6-independent (unaltered in atxr5 atxr6 mutants). The depletion of VIR or CPSF30-L partially decreases SUVH4/5/6-dependent H3K9me2 and ATXR5/6-dependent H3K27me1 levels but has little effect on SUVH4/5/6-independent or ATXR5/6-independent peaks (Fig. 5g and Supplementary Fig. 8c), indicating that m⁶A partially modulates H3K9me2 and H3K27me1 deposition via their respective methyltransferases.

As the deposition of H3K9me2 and H3K27me1 on retrotransposons requires heterochromatin targeting, we proceeded to investigate how m⁶A spatially associates with heterochromatin to facilitate specialized methylation. Although no obvious differences in the expression levels of SUVH4/5/6 and ATXR5/6 were observed across WT, fip37-4 LEC1::FIP37, vir and cpsf30-l seedlings (Supplementary Fig. 8d), we observed that the expression levels of CPSF30-L were positively correlated with those of SUVH4/5 and ATXR5/6 (Supplementary Fig. 8e). Given that the depletion of m⁶A writers and the m⁶A reader CPSF30-L reduces H3K9me2 and H3K27me1 levels and enhances the transcription of m⁶A-modified retrotransposons (Figs. 4c and 5e), we speculated that CPSF30-L, as an m⁶A reader, might be responsible for m⁶A-mediated transcriptional regulation through associating with SUVH4/5/6 and ATXR5/6 at specific loci for H3K9me2 and H3K27me1 deposition. To verify this hypothesis, we performed Y2H assays, revealing that CPSF30-L independently interacts with SUVH4/5/6 and ATXR5/6 (Fig. 5h). Bimolecular fluorescence complementation (BiFC) assays conducted in Nicotiana benthamiana leaves showed strong reconstituted YFP signal in the nucleus or nuclear foci among five protein pairs (Fig. 5i and Supplementary Fig. 8f). Moreover, semi-in vivo pull-down assays revealed that CPSF30-L–FLAG specifically binds glutathione S-transferase (GST)-targeted SUVH4/5/6 and ATXR5/6, but not GST alone (Fig. 5j). These interactions were not observed with m⁶A writers (Supplementary Fig. 8g). Altogether, our findings provide evidence consistent with CPSF30-L acting as an intermediary linking m⁶A to H3K9me2 and H3K27me1 deposition, thereby maintaining heterochromatin integrity in m⁶A-modified retrotransposon RNA regions.

ECT12, as a nuclear m⁶A reader, interacts with CPSF30-L and enhances the m⁶A-binding function

Although disturbing either the deposition or recognition of m⁶A impacts retrotransposon transcription, we noticed that partial m⁶A hypomethylation on retrotransposons in fip37-4 LEC1::FIP37 and vir mutants led to more pronounced expression increases in m⁶A retrotransposon RNAs than in the cpsf30-l mutant (Supplementary Fig. 9a). These imbalanced effects caused by m⁶A writers and an m⁶A reader imply that other m⁶A reader proteins may redundantly participate in this regulation. To identify the potential m⁶A reader responsible for retrotransposon transcription in Arabidopsis, we conducted phylogenetic analysis comparing the identified metazoan YTHDC1 with its plant counterparts. This analysis revealed two putative homologues of YTHDC1 in Arabidopsis: CPSF30-L and ECT12 (AT4G11970; Supplementary Fig. 9b). Unlike other YTH-containing proteins, both CPSF30-L and ECT12 are exclusively localized in the nucleus (Supplementary Fig. 9c). Moreover, the mRNA expression level pattern of ECT12 strongly correlated with that of CPSF30-L (Supplementary Fig. 9d), indicating that they might have synergistic roles in biological and molecular process.

Given that ECT12 has been identified as a potential m⁶A reader⁵², we proceeded to elucidate its m⁶A-binding function by performing an electrophoretic mobility shift assay with recombinant GST-tagged ECT12 (GST–ECT12) and a synthetic fluorescein-amidite-labelled RNA probe containing either m⁶A or A. The electrophoretic mobility shift assay results revealed that ECT12 specifically binds m⁶A-modified RNA but not unmethylated RNA (Supplementary Fig. 9e). Structural analysis of ECT12 identified conserved tryptophan residues at positions 94 and 144 as critical for m⁶A recognition in the YTH domain. These residues form a hydrophobic aromatic cage, similar to those observed in other YTH domain proteins, such as YTHDC1 and CPSF30-L (Supplementary Fig. 9f). To verify that ECT12 depends on its YTH domain to recognize m⁶A, we performed in vitro RIP followed by liquid chromatography–tandem mass spectrometry (LC–MS/MS) detection using the recombinant GST–ECT12, GST–ECT12/YTH and GST–ECT12/YTHm harbouring a putative non-m⁶A binding alteration (W144A) with isolated Arabidopsis poly(A)⁺ RNA. The LC–MS/MS results showed that m⁶A was highly enriched in the immunoprecipitated fraction of GST–ECT12 and GST–ECT12/YTH, but not in the flow-through or the bound fraction of GST–ECT12/YTHm (Supplementary Fig. 9g), suggesting that the YTH domain is responsible for m⁶A binding in ECT12. To further confirm that ECT12 indeed serves as an m⁶A-binding protein in planta, we generated ECT12::gECT12 and ECT12::gECT12m transgenic plants (Supplementary Fig. 9h), where ECT12m harbours the m⁶A-binding-function-abolished alteration W144A, and then performed an in vivo RIP–LC–MS/MS assay. The results showed that ECT12 pulled down significantly more m⁶A-modified RNAs than the control IgG-IP and ECT12m-IP (Fig. 6a). These results establish ECT12 as an m⁶A reader protein in Arabidopsis. To investigate the potential spatial linkage between ECT12 and CPSF30-L, we examined transgenic Arabidopsis plants (ECT12::gECT12-eGFP) and found that ECT12 is localized in the nucleus (Supplementary Fig. 9h,i). We also performed Y2H and BiFC assays, confirming that ECT12 physically interacts with CPSF30-L, but not with m⁶A writers (Fig. 6b,c and Supplementary Fig. 9j). Semi-in vivo pull-down assays also validated the interaction between CPSF30-L and ECT12 (Fig. 6d). To assess the regulatory role of the interaction between CPSF30-L and ECT12 in m⁶A-modified RNAs, we conducted an in vitro RIP assay using purified recombinant CPSF30-L and ECT12 proteins along with poly(A)⁺ RNAs. The results showed that CPSF30-L and ECT12 together bind significantly more m⁶A-modified poly(A)⁺ RNAs than the same amount of single CPSF30-L-IP and ECT12-IP (Fig. 6e). This indicates that the interaction between CPSF30-L and ECT12 enhances the m⁶A-binding ability more effectively than the individual actions of CPSF30-L or ECT12 alone.

**Fig. 6: CPSF30-L and ECT12 redundantly mediate H3K9me2 and H3K27me1 deposition.**

CPSF30-L and ECT12 redundantly mediate H3K9me2 and H3K27me1 formation and suppress the transcription of m⁶A retrotransposon RNAs

Considering that CPSF30-L and ECT12 interact with each other as nuclear m⁶A reader proteins, we proceeded to assess whether they redundantly regulate heterochromatic histone modifications and transcription of m⁶A retrotransposon RNAs. To address this, we first characterized a homozygous transfer-DNA insertion mutant, ect12-3 (SALK_071944), which almost completely disrupts the expression of full-length ECT12 transcript (Supplementary Fig. 10a). We also generated cpsf30-l ect12 double mutants using the CRISPR–Cas9 genome editing system, incorporating two guide RNAs targeting the second exonic region of ECT12 in the cpsf30-l background (Supplementary Fig. 10a). We performed H3K9me2 and H3K27me1 immunostaining on these mutants to investigate whether ECT12 and CPSF30-L redundantly affect the methylation levels of these histone modifications. The results showed that the decrease in these two histone modification levels was less pronounced in ect12-3 than in cpsf30-l, while the cpsf30-l ect12 double mutants showed a more substantial reduction in these histone marks than either single mutant (Fig. 6f,g), consistent with genetic redundancy.

Consistent with these observations, ChIP–Seq analysis for H3K9me2 and H3K27me1 in six-day-old WT, cpsf30-l, ect12-3 and cpsf30-l ect12 seedlings also indicated redundant regulation of heterochromatic histone modifications (Supplementary Tables 12 and 13), particularly on m⁶A retrotransposon RNAs, but not on non-m⁶A retrotransposon RNAs (Fig. 6h and Supplementary Fig. 10b–f). These findings were corroborated by ChIP-qPCR targeting selected loci (Supplementary Fig. 10g). Consistently, the reduction in the levels of SUVH4/5/6-dependent H3K9me2 and ATXR5/6-dependent H3K27me1 in cpsf30-l was more pronounced in the cpsf30-l ect12 double mutants (Fig. 6i and Supplementary Fig. 10h). Furthermore, no significant differences in the expression levels of SUVH4/5/6 and ATXR5/6 were detected across WT, cpsf30-l, ect12-3 and cpsf30-l ect12 seedlings, and no physical interaction was observed between ECT12 and SUVH4/5/6 or ATXR5/6 (Supplementary Fig. 10i,j). These results confirm that CPSF30-L and ECT12 form a complex to redundantly mediate H3K9me2 and H3K27me1 deposition.

To further elucidate how CPSF30-L and ECT12 redundantly regulate the expression of m⁶A retrotransposon RNAs, we performed parallel caRNA-Seq on six-day-old WT, cpsf30-l, ect12-3 and cpsf30-l ect12 seedlings with two biological replicates (Supplementary Fig. 11a and Supplementary Table 14). Reflecting their shared involvement in regulating heterochromatic histone modifications, CPSF30-L and ECT12 collectively modulate the abundance of m⁶A retrotransposon RNAs and specific retrotransposon subfamilies, displaying genetic redundancy (Fig. 7a and Supplementary Fig. 11b,c). The ect12-3 and cpsf30-l mutants exhibited significant increases in m⁶A retrotransposon RNA expression, and cpsf30-l ect12 double mutants showed more pronounced upregulation than either single mutant (Fig. 7a and Supplementary Fig. 11b,c). Overlapping the upregulated m⁶A retrotransposon RNAs in cpsf30-l, ect12-3 and cpsf30-l ect12 demonstrated that virtually all of them were coordinately modulated by CPSF30-L and ECT12 (Fig. 7b). Further supporting this, we confirmed the significant elevation in the transcription and expression levels of COPIA22 and ATRE1 in cpsf30-l ect12 mutants, with a twofold to tenfold increase compared with cpsf30-l or ect12-3 mutants (Fig. 7c,d and Supplementary Fig. 11d). We also found that the changes in expression levels primarily followed a redundant trend, accompanied by a concurrent decrease in both H3K9me2 and H3K27me1 levels at H3K9me2 & H3K27me1 & m⁶A retrotransposon RNA regions (Supplementary Fig. 11e,f). The disruption of m⁶A nuclear readers led to a greater increase in the expression of m⁶A retrotransposon RNAs carrying both histone modifications compared with those marked by a single modification (Supplementary Fig. 11f). Altogether, these results emphasize that CPSF30-L and ECT12 redundantly regulate H3K9me2 and H3K27me1 deposition, thereby suppressing the transcription of m⁶A-modified retrotransposon RNAs.

**Fig. 7: CPSF30-L and ECT12 cooperatively suppress the transcription and abundance of m⁶A retrotransposon RNAs.**

Discussion

Our study unveils a comprehensive landscape of the RNA modification m⁶A on caRNAs, especially in retrotransposon transcripts in Arabidopsis, revealing their transcriptional regulation via m⁶A deposition and recognition with m⁶A writer and reader proteins. Specifically, CPSF30-L and ECT12 form a complex to enhance m⁶A binding affinity. CPSF30-L interacts with SUVH4/5/6 and ATXR5/6, together with ECT12, to facilitate H3K9me2 and H3K27me1 deposition at designated retrotransposons, achieving largely closed chromatin and repressed retrotransposons (Fig. 7e). These results highlight a new layer of epitranscriptomic regulation that modulates retrotransposon RNA transcription and maintains heterochromatin integrity.

It has been shown that m⁶A promotes the decay of a subset of chromatin-associated regulatory RNAs, such as LINE1 in the nucleus of mouse embryonic stem cells¹⁵. Unlike the regulatory mechanism in mammals, our data show that m⁶A modifications do not affect the stability of retrotransposon RNAs (Fig. 4a,b). Additionally, the lack of interaction between nuclear m⁶A readers and nuclear exosome targeting components suggests that the m⁶A-mediated regulatory mechanism primarily operates at the transcriptional level in plants (Supplementary Fig. 5c). This was further supported by increased transcription rates of m⁶A retrotransposon RNAs in m⁶A writer and reader mutants (Fig. 4c–e), implicating both evolutionarily conserved and divergent mechanisms of m⁶A in transcriptional regulation.

We discovered that m⁶A modifications are prevalent across various retrotransposon families and are particularly enriched at the 5′ and 3′ ends of transcripts (Fig. 1f). Disruption of the m⁶A writer subunit in fip37-4 LEC1::FIP37 and vir mutants significantly reduced m⁶A levels on retrotransposon RNAs, leading to their increased abundance (Fig. 2a,b). This inverse relationship between m⁶A methylation and RNA levels suggests that m⁶A modification is crucial for suppressing retrotransposon expression. The investigation into the m⁶A reader CPSF30-L revealed its essential role in binding m⁶A-marked retrotransposons and suppressing their expression through m⁶A recognition (Fig. 3).

We noticed that partial m⁶A hypomethylation on retrotransposons in fip37-4 LEC1::FIP37 and vir mutants led to more pronounced expression increases in m⁶A retrotransposon RNAs than in the cpsf30-l mutant (Supplementary Fig. 9a). This disparity in effects, attributed to the mechanisms of m⁶A writer and reader molecules, hints at the participation of additional m⁶A reader proteins in a redundant regulatory capacity. Through phylogenetic analysis and molecular assays, we identified ECT12 as a nuclear m⁶A reader that interacts with CPSF30-L and enhances its m⁶A-binding function (Fig. 6a–e). The synergistic action between CPSF30-L and ECT12 underscores the evolutionary significance of m⁶A-mediated regulation, where multiple m⁶A readers ensure robust transcriptional repression of retrotransposons.

Heterochromatic histone marks differ between mammals and plants. Although m⁶A has been reported to regulate H3K9me3 deposition and heterochromatin states in mouse embryonic stem cells^16,18,26, it is necessary to investigate whether and how m⁶A regulates heterochromatin state in plants. We observed that m⁶A-modified retrotransposons were associated with heterochromatic marks H3K9me2 and H3K27me1 (Fig. 5a). Immunostaining results showed that the reduced H3K9me2 and H3K27me1 signals in cpsf30-l can be restored by expressing CPSF30-L but not by the m⁶A-binding-function-abolished CPSF30-Lm (Fig. 5c,d), indicating that the m⁶A binding function of CPSF30-L regulates H3K9me2 and H3K27me1 levels. In m⁶A writer and reader mutants, we noted a notable reduction in H3K9me2 and H3K27me1 levels at m⁶A retrotransposon loci (Fig. 5e). Compared with the reduced levels of H3K9me2 and H3K27me1 in their respective writer mutants (suvh4 suvh5 suvh6 and atxr5 atxr6), the loss of m⁶A writer subunit VIR or m⁶A reader CPSF30-L only partially decreased these histone modifications (Supplementary Fig. 6d,e). This suggests that m⁶A acts as a regulatory layer in modulating H3K9me2 and H3K27me1 deposition, specifically at m⁶A-modified retrotransposon RNA regions. Mechanistically, CPSF30-L interacts with SUVH4/5/6 and ATXR5/6, the enzymes responsible for H3K9me2 and H3K27me1 deposition (Fig. 5h–j). This interaction links m⁶A methylation with histone modification, maintaining a closed chromatin state and transcriptional repression of m⁶A retrotransposon RNAs.

Besides histone modification, DNA methylation serves as another gene-silencing epigenetic mark. Heterochromatic non-CG methylation and H3 K9 dimethylation are interconnected through a cyclic reinforcement loop between the writers of non-CG DNA methylation and H3 K9 dimethylation^53,54. In addition, DNA methylation is guided by small RNAs that target specific genomic sequences in a process known as RNA-directed DNA methylation, where Pol IV is required for short interfering RNA (siRNA) biogenesis and Pol V provides scaffold RNAs that guide siRNA-directed DNA methylation⁵⁵. ChIP–Seq and caRNA-Seq analyses reveal that retrotransposon transcription is tightly regulated by the occupancy of Pol II and Pol V (Supplementary Fig. 2), suggesting a potential role for siRNAs and the involvement of Pol IV and Pol V in retrotransposon regulation. Furthermore, recent mammalian studies found that the m⁶A reader FXR1 recruits DNA 5mC dioxygenase TET1 for 5mC demethylation in nearby genomic loci in normal and cancer cells, and the m⁶A writer METTL3–METTL14 recruits DNA methyltransferase DNMT1 to chromatin for gene-body methylation^56,57, suggesting that m⁶A could directly regulate DNA 5mC deposition. We therefore hypothesize that m⁶A modification may also influence DNA methylation directly and siRNA-directed DNA methylation at retrotransposons apart from the H3K9me2-mediated DNA methylation pathway.

Retrotransposons account for over 40% of the human and mouse genomes, and their transcriptional regulation is essential for mammalian zygotic genome activation, embryonic stem cell identity and embryonic development^16,18,58. Similarly, retrotransposons comprise an even more significant portion of crop plant genomes and can be transcriptionally activated by environmental challenges, promoting genetic diversity and adaptive evolutionary changes^59,60. Here we found that the RNA modification m⁶A is installed on retrotransposon RNAs to suppress their transcription, thereby securing genome integrity. It is worth investigating whether environmental challenges affect the installation of RNA modifications on retrotransposon RNAs. Besides m⁶A, it is also important to explore whether other RNA modifications regulate retrotransposon RNAs in plants. Recently, the expression of FTO in rice was shown to increase field yield and biomass, accompanied by m⁶A demethylation on repeat RNAs, enhanced chromatin accessibility and transcriptional activation²⁷. This suggests that epitranscriptomic regulation in transcription represents a promising new avenue for plant breeding. Understanding how endogenous m⁶A in plants modulates retrotransposon transcription and heterochromatin states could provide valuable insights for future crop improvement strategies.

Overall, our work demonstrates that m⁶A co-transcriptionally directs heterochromatic histone modifications through its reader complexes, enabling a direct flow of RNA modification information from RNA to chromatin to ensure silenced transcription and heterochromatin integrity at m⁶A-modified retrotransposon RNA regions. This regulatory interplay between the epitranscriptome and the epigenome provides deeper insights into m⁶A functions as well as the mechanisms of retrotransposon regulation and heterochromatin maintenance in plant evolution and adaptation.

Methods

Plant materials and growth conditions

The WT Arabidopsis thaliana plants used in this study were of the Col-0 ecotype. nrpe1-11, nrpb2-3, fip37-4 LEC1::FIP37, vir, cpsf30-l, CPSF30::CPSF30-L cpsf30-l (CPSF30-L) and CPSF30::CPSF30-Lm cpsf30-l (CPSF30-Lm) have been described previously^3,4,10,33,34. The ect12-3 (SALK_071944) mutant lines were obtained from the Arabidopsis Biological Resource Center. Loss-of-function mutants of cpsf30-l ect12 were generated using the CRISPR–Cas9 system. Specifically, two single guide RNA sequences targeting the ECT12 gene were amplified via PCR, using the pDT1T2 vector as a template. The resulting products were purified, digested with BsaI and then ligated into the binary vector pHEE401E. Subsequently, the constructed plasmid was introduced into the cpsf30-l background through GV3101-mediated floral dipping. Positive seedlings were selected from 1/2× Murashige and Skoog plates containing 50 mg l⁻¹ hygromycin B and confirmed via direct Sanger sequencing. All seeds of the WT and mutants were sterilized in 75% ethanol for 10 min followed by immersion in 20% bleaching solution for an additional 10 min, and immediately rinsed at least four times with sterile water. The sterilized seeds were stratified at 4 °C in darkness for three days and grown on 1/2× Murashige and Skoog nutrient agar plates. All plant germination and growth were under long-day conditions (16 h light/8 h dark at 22 °C with a light intensity of 90 to 120 μmol m⁻² s⁻¹).

RT-qPCR

Isolated total RNAs and caRNAs were reverse-transcribed into first-strand cDNA using SuperScript III (Thermo Fisher Scientific) with random primers. The transcribed cDNAs were then appropriately diluted and used as templates for PCR reactions, employing Hieff qPCR SYBR Green Master Mix (Low Rox) from Yeasen. These reaction systems were subsequently analysed on a ViiA 7 instrument (Applied Biosystems) following the provided instructions. To ensure the accuracy of the results, each independent sample comprised at least three biological replicates and two technical replicates. A complete list of the primers used is provided in Supplementary Table 15.

Extraction of caRNAs

caRNAs were extracted using a previously reported method⁶¹. Briefly, 2 g of six-day-old seedlings were ground into a fine powder using liquid nitrogen. The powdered material was transferred into a 50-ml tube containing 10 ml of ice-cold Honda buffer (0.44 M sucrose, 1.25% (w/v) Ficoll, 2.5% (w/v) dextran T40, 20 mM HEPES-KOH pH 7.4, 10 mM MgCl₂, 0.5% (w/v) Triton X-100, 1 mM dithiothreitol (DTT), 1× protease inhibitor (Roche), 0.4 U μl⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific) and 100 ng μl⁻¹ yeast tRNA (Invitrogen)). After homogenization via rotation, the homogenate was filtered through two layers of Miracloth (Merck). After centrifugation at 4 °C and 2,000 g for 5 min, the pellet was resuspended and washed with 5 ml of Honda buffer three or four times until the pellet appeared white. The washed pellet was resuspended in 1 ml of Honda buffer and centrifuged at 4 °C and 8,000 g for 1 min. After the supernatant was removed, the nucleus pellet was weighed. One volume of nucleus resuspension buffer (50% (v/v) glycerol, 0.5 mM EDTA, 1 mM DTT, 25 mM Tris-HCl pH 7.5, 100 mM NaCl and 0.4 U μl⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific)) was added, and the pellet was stirred to mix with a pipette tip. Two volumes of washing buffer (25 mM Tris-HCl pH 7.5, 300 mM NaCl, 1 M urea, 0.5 mM EDTA, 1 mM DTT, 1% (v/v) Tween-20 and 0.4 U μl⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific)) were then added, and the pellet was washed by pipetting up and down 30 times and centrifuged at 4 °C and 8,000 g for 1 min. This washing procedure was repeated three times, and the pellet was retained as the chromatin-associated fraction. For RNA extraction, the pellet was resuspended in 1 ml of TRIzol (Invitrogen), and the extraction procedures were performed using the Direct-zol RNA Miniprep Plus kit with DNase digestion (ZYMO).

caRNA m⁶A-Seq and m⁶A-Seq

The caRNAs were concentrated with a ZYMO RNA Clean & Concentrator-5 kit (ZYMO) and processed for rRNA depletion according to the manufacturer’s instructions (RiboPOOL). Then, 1 μl of 1:1,000 diluted m⁶A spike-in from an EpiMark N⁶-Methyladenosine Enrichment Kit (NEB) and 1:100 diluted ERCC spike-ins (Ambion) were added to 1 μg of rRNA-depleted caRNA, followed by fragmentation into 100–150 nucleotides using the Magnesium RNA Fragmentation Module (NEB). m⁶A-IP was performed following the manufacturer’s protocols. Library preparation was performed using a NEBNext Ultra II Directional RNA Library Prep Kit (NEB) according to the manufacturer’s protocols. Sequencing was performed on an Illumina Novaseq machine in paired-end mode with 150 bp per read (Genewiz).

Poly(A)⁺ RNA from six-day-old WT seedlings was applied to profile the m⁶A methylome. 1 μg of poly(A)⁺ RNA was fragmented into 100–150 nucleotides using the Magnesium RNA Fragmentation Module (NEB). m⁶A-IP and library preparation were performed using the EpiMark N⁶-Methyladenosine Enrichment Kit (NEB). Input and RNA eluted from m⁶A-IP were used to prepare libraries with a NEBNext Ultra II Directional RNA Library Prep Kit (NEB). Sequencing was performed on an Illumina Novaseq machine in paired-end mode with 150 bp per read (Genewiz).

m⁶A-IP–qPCR

For each biological replicate, 20 µg of caRNAs extracted from six-day-old Arabidopsis seedlings were supplemented with 1 µl of 1:100 diluted m⁶A and non-m⁶A spike-ins from the EpiMark N⁶-Methyladenosine Enrichment Kit (NEB). The RNA samples were then fragmented using the Magnesium RNA Fragmentation Module (NEB), and the resulting fragments were subjected to IP with m⁶A-conjugated antibodies bound to Protein G Dynabeads (Thermo Fisher Scientific), following the manufacturer’s instructions. RNA was extracted from both the IP and input samples and analysed via RT-qPCR. The spike-ins served as normalization controls for m⁶A level quantification. The primers used for qPCR are listed in Supplementary Table 15.

Nuclear FA-CLIP

For each biological replicate, 2 g of six-day-old seedlings were harvested and crosslinked with 1% (v/v) formaldehyde under vacuum. The crosslinking reaction was stopped after 15 min by adding glycine to a final concentration of 0.125 M, followed by an additional 5 min under vacuum. Nuclei were isolated from frozen, ground plant tissue using ice-cold Honda buffer (0.44 M sucrose, 1.25% (w/v) Ficoll, 2.5% (w/v) dextran T40, 20 mM HEPES-KOH pH 7.4, 10 mM MgCl₂, 0.5% (w/v) Triton X-100, 1 mM DTT, 1× protease inhibitor (Roche) and 0.4 U μl⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific)) and collected via centrifugation at 2,000 g for 5 min at 4 °C. The nuclei were washed twice with Honda buffer and resuspended in 1 ml of nuclear lysis buffer (150 mM KCl, 50 mM HEPES, pH 7.5, 2 mM EDTA, 1% (v/v) NP-40, 0.5 mM DTT, 1× protease inhibitor (Roche) and 40 U ml⁻¹ RNase inhibitor (Thermo Fisher Scientific)). Sonication was performed using a Bioruptor (Diagenode) for four cycles (low power, 30 s pulse-on, 60 s pulse-off). After centrifugation, the supernatant was treated with 1 U μl⁻¹ RNase T1 (Thermo Fisher Scientific) at 22 °C for 10 min and incubated with Anti-Flag M2 Magnetic beads (Sigma) with rotation at 4 °C for 3 h. The beads were washed three times with low-salt washing buffer (150 mM KCl, 50 mM HEPES, pH 7.5, 0.05% (w/v) NP-40, 0.5 mM DTT and 1× protease inhibitor (Roche)), followed by RNase T1 treatment and a second incubation at 22 °C for 10 min. After three additional washes with high-salt washing buffer (500 mM KCl, 50 mM HEPES, pH 7.5, 0.05% (w/v) NP-40, 0.5 mM DTT and 1× protease inhibitor (Roche)), the RNA underwent end repair and proteinase K digestion on beads. RNA was then recovered using the RNA Clean & Concentrator-5 Kit (Zymo), and libraries were prepared using the NEBNext Small RNA Library Prep Set (NEB). Sequencing was performed on the MGISEQ-2000 platform using a paired-end model (PE100).

In vivo FA-RIP-qPCR

Six-day-old CPSF30::CPSF30-L-Flag cpsf30-l and CPSF30::CPSF30-Lm-Flag cpsf30-l seedlings were separately fixed with 1% (v/v) formaldehyde solution. The fixed plant materials were ground into powder and incubated with lysis buffer (150 mM KCl, 50 mM HEPES, pH 7.5, 2 mM EDTA, 1% (v/v) NP-40, 1× protease inhibitor (Roche) and 40 U ml⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific)) in a head-over-tail rotation for 30 min at 4 °C. After full lysis and centrifugation, the lysates were collected and then immunoprecipitated with Anti-Flag M2 beads (Sigma-Aldrich) or a control IgG (CST) conjugated with protein G Dynabeads (Thermo Fisher Scientific) on a rotating wheel for 2 h at 4 °C. After washing, proteinase K digestion (NEB) and ethanol precipitation, the recovered RNA fractions were reverse-transcribed into cDNA to calculate the relative enrichment fold via RT-qPCR. AT2G07689 was used as the internal control.

Nuclear RNA lifetime sequencing

Six-day-old WT and mutant seedlings were placed in 3 ml of incubation buffer (15 mM sucrose, 1 mM KCl, 1 mM PIPES, pH 6.25 and 1 mM sodium citrate) in six-well plates (Corning Costar) and rotated at 600 g for 15 min. The initial (T₀) samples were promptly frozen, while the remaining samples underwent transcriptional inhibition by adding 3 ml of fresh buffer with 1 mM cordycepin (Macklin). Subsequently, the samples were subjected to three additional rounds of 1-min vacuum infiltration with 1-min decompression intervals. Samples were collected at various time points (15, 30, 60 and 120 min) after the third vacuum infiltration and frozen in liquid nitrogen. Approximately 0.2 g of seedlings per sample at each time point was used. The nuclear fraction was isolated, and total RNA was purified using Trizol (Invitrogen). Prior to purifying non-ribosomal RNA with the rRNA Depletion Kit (RiboPOOL), an equal amount of ERCC RNA spike-in control (Ambion) was added to the total nuclear RNA samples as internal controls. Library preparation was carried out using the NEBNext Ultra II Directional RNA Library Prep Kit (NEB). Sequencing was conducted on an Illumina Novaseq machine in paired-end mode with 150 bp per read (Genewiz).

Nuclear RNA lifetime measurement via qPCR

Six-day-old WT and mutant seedlings were placed in six-well plates containing 1 mM cordycepin (Macklin) or 200 μM Actinomycin D (Selleck). After three rounds of 1-min vacuum infiltration, samples were collected at 15, 30, 60 and 120 min. The nuclear fraction was isolated, and total RNA was extracted using Trizol (Invitrogen). Before RNA quantification via RT-qPCR, 1 μl of 1:100 diluted m⁶A and non-m⁶A spike-in from the EpiMark N⁶-Methyladenosine Enrichment Kit (NEB) was introduced. The RNA degradation rate (k) was determined by plotting N_t/N₀ against time and fitting the equation N_t/N₀ = e^−kt, where t represents transcription inhibition time, and N_t and N₀ indicate RNA quantities at time t and time 0. The RNA lifetime (t_1/2) was calculated as t_1/2 = (ln2)/k.

Time-course RNA sequencing for transcription rate measurement

Six-day-old WT and cpsf30-l seedlings were placed in six-well plates containing 0.5 mM 5-ethynyl uridine (Jena Bioscience) and were collected at 30 min, 60 min and 120 min. Total RNA was purified using Trizol followed by rRNA depletion; nascent rRNA-depleted RNA was captured using a Click-iT Nascent RNA Capture Kit (Invitrogen). Before constructing libraries with a NEBNext Ultra II Directional RNA Library Prep Kit (NEB), an equal amount of ERCC RNA spike-in control (Ambion) was added to each sample. Sequencing was conducted on an Illumina Novaseq machine in paired-end mode with 150 bp per read (Genewiz).

Nuclear run-on assay

Six-day-old WT and mutant seedlings were ground into a fine powder in liquid N₂ and resuspended in 5 ml of pre-cooled nuclease-free Honda buffer. After filtering through two layers of Miracloth and centrifuging at 2,000 g for 10 min at 4 °C, the nuclei were washed two or three times with Honda buffer and resuspended in 50 µl of nucleus storage buffer (50 mM Tris-HCl, pH 7.8, 1 mM DTT, 20% (v/v) glycerol, 5 mM MgCl₂ and 0.44 M sucrose). The run-on assay was performed in a reaction containing 10 µl of 10× transcription assay buffer (500 mM Tris-HCl, pH 7.5, 50 mM MgCl₂, 1.5 M KCl, 1% (v/v) sarkosyl, 20 U ml⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific) and 100 mM DTT), 50 µl of nuclei in storage buffer, 5 µl of NTP mixture with BrUTP (100 mM ATP, 100 mM CTP, 100 mM GTP and 100 mM BrUTP (Sigma)) and 35 µl of H₂O. The run-on reaction was performed at 30 °C for 30 min. The reaction was stopped by adding 900 µl of TRIzol reagent (Invitrogen), and RNA was extracted using the Direct-zol RNA Miniprep Plus kit (ZYMO). The purified RNA was diluted in 500 µl of incubation buffer (20 mM Tris-HCl pH 7.5, 4 mM MgCl₂ and 0.2% (v/v) NP-40) and incubated with 60 μl of anti-BrdU beads (Santa Cruz) at 4 °C for 2 h. The precipitated RNA was extracted with TRIzol reagent (Invitrogen) and used for RT-qPCR analysis.

ChIP–Seq and ChIP-qPCR

The protocol for ChIP–Seq was adapted from a previously published method⁶². Briefly, 2 g of six-day-old WT, cpsf30-l, ect12-3, cpsf30-l ect12-1, cpsf30-l ect12-2 and vir seedlings were ground with liquid nitrogen into fine powder and immediately fixed with 1% (v/v) formaldehyde. Nuclei were extracted and mononucleosomes were generated with micrococcal nuclease (MNase, NEB). Subsequently, MNase digestion was carried out and stopped with the addition of EDTA and EGTA to a final concentration of 10 mM. After centrifugation, the supernatant was diluted for IP. IP was conducted with anti-H3K9me2 (Abcam) and anti-H3K27me1 (Millipore) antibodies. After incubation overnight, Protein G Dynabeads (Thermo Fisher Scientific) were added to collect the immunocomplexes. After washing, elution, reverse crosslinking and DNA purification, recovered DNA and input were subjected to qPCR and library preparation with the NEXTflex Rapid DNA-Seq Kit for Illumina (PerkinElmer) according to the manufacturer’s instructions and sequenced with the MGISEQ-2000 platform using a paired-end model (PE150).

Immunostaining

For immunostaining, 0.5 g of six-day-old seedlings were fixed with 4% paraformaldehyde for 20 min under vacuum. The fixed seedlings were then chopped in a petri dish containing 3 ml of ice-cold NEB1 buffer (10 mM Tris-HCl pH 9.5, 10 mM KCl, 500 mM sucrose and 0.1% Triton-X 100) to obtain a homogenate. The homogenate was filtered with a 40-μm cell strainer and centrifuged at 600 g for 3 min at 4 °C. The pellet was gently resuspended in 300 μl of NEB2 buffer (10 mM Tris-HCl, pH 9.5, 10 mM KCl, 125 mM sucrose and 0.1% Triton-X 100) and layered onto 600 μl of NEB3 buffer (10 mM Tris-HCl, pH 9.5, 10 mM KCl, 850 mM sucrose and 0.1% Triton-X 100). Nuclei were pelleted via centrifugation at maximum speed for 30 min at 4 °C, resuspended in 40 μl of NEB1 buffer and spread onto slides. The slides were air-dried, fixed with 4% paraformaldehyde for 30 min at room temperature and washed with PBST. Blocking was performed with 5% BSA for 1 h at 37 °C, followed by overnight incubation at room temperature with primary antibodies (1:100 anti-H3K27me1 (Abcam) and anti-H3K9me2 (Abcam)). After the PBST washes, the slides were incubated with secondary antibodies (1:100 Goat Anti-Mouse IgG H&L (Alexa Fluor 488, Abcam) and 1:100 Goat Anti-Rabbit IgG H&L (Alexa Fluor 555, Abcam)) for 1 h at 37 °C. Finally, the slides were washed with PBST, mounted with DAPI Fluoromount-G and imaged using a Zeiss LSM900 confocal microscope.

Y2H assay

The full-length coding sequences of ZCCHC8A, ZCCHC8B, RBM7, SUVH4, SUVH5, SUVH6, ATXR5, ATXR6, CPSF30-L and ECT12 were fused in frame with the GAL4 activation domain in pGADT7 as prey. The full-length coding sequences of CPSF30-L, ECT12, MTA, FIP37 and VIR with amino acid residues 1 to 1,470 (VIR^N) and 1,456 to 2,138 (VIR^C) were fused in frame with pGBKT7 as bait. Yeast strain Y2HGold cells were co-transformed with the bait and prey constructs. The empty vectors pGADT7 and pGBKT7 were co-transformed as the negative controls. Co-transformed yeast strain was spotted onto selective medium (minus His, Ade, Leu and Trp) and supplemented with the inhibitor of histidine synthesis 3-amino-1,2,4-triazole (Sigma). Images were taken after incubation for two to three days at 28 °C.

BiFC assay

The full-length coding sequences of SUVH4, SUVH5, SUVH6, ATXR5, ATXR6, CPSF30-L and ECT12 were cloned into the binary pBI121–YFP^C or pBI121–YFP^N vector to generate SUVH4–YFP^C, SUVH5–YFP^C, SUVH6–YFP^C, ATXR5–YFP^C, ATXR6–YFP^C, ECT12–YFP^C and CPSF30-L–YFP^N constructs. The constructs with the indicated combinations were transiently co-expressed in N. benthamiana leaves via GV3101-mediated infiltration. The plants were grown in dark conditions for another two or three days, and the YFP fluorescence signal for each combination was detected using a confocal laser scanning microscope (ZESIS). Specifically, a 488-nm-wavelength laser was used to excite the YFP, with the emission signal collected from 500 to 550 nm.

Semi-in vivo pull-down assay

Total proteins were extracted from Arabidopsis seedlings expressing CPSF30-L–FLAG with IP buffer. 2 mg of total protein extracts were mixed with 20 μg of purified proteins infused with GST tag or GST alone. The mixture was incubated at 4 °C with gentle rotation for 2 h. After incubation, pre-washed Pierce Glutathione Magnetic Agarose Beads (Thermo Fisher Scientific) were added and incubated at 4 °C with gentle rotating for another 2 h. Proteins bound to the beads were eluted in elution buffer (125 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM DTT and 50 mM reduced glutathione) and were analysed via protein immunoblot with GST antibody (CST) and FLAG antibody (MCE).

Protein expression and purification

Plasmids containing GST–SUVH4, GST–SUVH5, GST–SUVH6, GST–ATXR5, GST–ATXR6, GST–ECT12, GST–ECT12/YTH, GST-ECT12/YTHm and MBP–CPSF30-L were transfected into Escherichia coli strain BL-21 Rosetta competent cells. Protein expression was induced at 18 °C with 1 mM isopropyl β-D-1-thiogalactopyranoside for 16 h. Cells were collected and resuspended in lysis buffer (10 mM Tris-HCl, pH 8.0, 500 mM NaCl, 1 mM PMSF, 3 mM DTT and 5% (v/v) glycerol), then lysed via sonication and centrifuged. The soluble proteins were first purified with either a GST-affinity column (GE Healthcare) or Amylose Resin (NEB) and eluted using the corresponding elution buffer. The crude eluted fractions were further purified on a Superdex 75 exclusion column (GE Healthcare). The purified proteins were stored in storage buffer (10 mM Tris-HCl, pH 7.5, 200 mM NaCl, 3 mM DTT and 20% (v/v) glycerol) at −80 °C.

Electrophoretic mobility shift assay

The recombinant proteins GST–ECT12 were separately diluted to a series of concentrations ranging from 0 to 2 μM in binding buffer (10 mM HEPES-KOH, pH 8.0, 50 mM KCl, 1 mM EDTA, 0.05% (v/v) Triton X-100, 5% (v/v) glycerol, 10 μg ml⁻¹ salmon DNA, 1 mM DTT and 40 U ml⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific)). Fluorescein-amidite-labelled RNA probe (4 nM final concentration) and 1 μl of protein were mixed in 10 μl of binding buffer and incubated on ice for 30 min. The entire RNA–protein mixture was loaded into a Novex 4–20% TBE gel (Thermo Fisher Scientific) for electrophoresis separation at 4 °C. The gel was visualized using ChemiDoc (Bio-Rad).

In vivo nuclear FA-RIP–LC–MS/MS

Nuclear RIP was conducted following a protocol similar to that for nuclear FA-CLIP, omitting the second RNase T1 digestion. Crosslinked six-day-old seedlings of CPSF30::CPSF30-L cpsf30-l, CPSF30::CPSF30-Lm cpsf30-l, ECT12::gECT12 ect12-3 and ECT12::gECT12m ect12-3 transgenic lines were subjected to IP using FLAG M2 Magnetic Beads (Sigma) or normal rabbit IgG (CST) bound to Protein G (Invitrogen). Crosslinked six-day-old WT seedlings were subjected to RIP using anti-Pol II antibody (Abcam) or IgG (CST) as a control. For GFP-tagged material, crosslinked NRPE1::NRPE1-GFP/nrpe1 seedlings were immunoprecipitated using GFP-Trap Dynabeads (ChromoTek) or IgG (CST) as a control. After washing and RNA recovery, the m⁶A levels in the RNA fractions were quantified via LC–MS/MS.

In vitro RIP–LC–MS/MS

For RIP–LC–MS/MS, 0.8 μg of poly(A)⁺ RNA and the recombinant protein GST–ECT12 or MBP–CPSF30-L (final concentration 500 nM) were incubated in 200 μl of IPP buffer (150 mM NaCl, 0.1% (v/v) NP-40, 10 mM Tris-HCl, pH 7.5, 40 U ml⁻¹ Ribolock RNase inhibitor (Thermo Fisher Scientific) and 0.5 mM DTT) with rotation at 4 °C for 2 h. Pre-equilibrated 10 μl of GST-affinity magnetic beads (Thermo Fisher Scientific) or amylose magnetic beads (NEB) with IPP buffer were added to each solution and incubated for an additional 2 h at 4 °C. The aqueous phase was harvested, recovered via ethanol precipitation and termed the flow-through. GST beads were washed four times with 200 μl of IPP buffer. The RNA on the beads was extracted with TRIzol reagent (Invitrogen) and saved as the recombinant protein-IP sample. LC–MS/MS was used to measure the ratio of m⁶A/C in input, flow-through and recombinant protein-IP samples.

Quantification analysis via LC–MS/MS

LC–MS/MS analysis was performed as follows: RNA (around 100 to 200 ng) was digested with Nucleoside Digestion Mix (NEB) in 20 μl at 37 °C for 2 h. After centrifugation at maximum speed for 30 min, the aqueous phase containing nucleosides was separated using a LC pump (Shimadzu) equipped with a ZORBAX SB-Aq column (Agilent) and analysed via MS/MS on a Triple Quad 5500 mass spectrometer (AB SCIEX) in positive ion mode with multiple reaction monitoring. Nucleosides were quantified using the following mass transitions: 282.0 to 150.1 (m⁶A), 244.0 to 112.0 (C). Quantification was based on a standard curve generated from pure nucleoside standards run alongside the samples. The ratios of m⁶A/C were calculated using the calibrated concentrations.

Data analysis for caRNA-Seq and RNA-Seq

The sequencing data from the caRNA-Seq and RNA-Seq libraries were processed by trimming adapters with cutadapt (v.4.4)⁶³, followed by alignment to the TAIR10 reference genome via hisat2 (v.2.2.1)⁶⁴ with the following parameters: rna-strandness, FR; k, 10; pen-noncansplice, 1000000. We eliminated PCR duplicates using SAMtools in conjunction with Picard (v.21.0.1; https://broadinstitute.github.io/picard/)⁶⁵. To enhance the authenticity and the repeatability of the results, we eliminated the transcript components of rRNA and tRNA, as well as those belonging to chloroplasts and mitochondria. Gene expression quantification was conducted using featureCounts (v.2.0.1)⁶⁶ and edgeR (v.3.36.0)⁶⁷ to compute counts, counts per million and differential gene expression. Therein, external ERCC spike-ins were also used to normalize the counts of each gene by using the R package RUVseq (v.1.28.0)⁶⁸. The Pearson correlation was calculated using the R package cor (v.4.3.2)⁶⁹.

Data analysis for caRNA m⁶A-Seq and m⁶A-Seq

The analysis for the input group in caRNA m⁶A-Seq or m⁶A-Seq was conducted following the pipeline of RNA-Seq above. We employed cutadapt (v4.4)⁶³ for adapter trimming and size selection, retaining fragments that were 30 nucleotides or longer. These sequences were subsequently mapped to the TAIR10 reference genome using hisat2 (v.2.2.1)⁶⁴ with the default parameters. PCR duplicates were eliminated using SAMtools (v.1.13)⁶⁵ and Picard (v.21.0.1; https://broadinstitute.github.io/picard/). After obtaining the BAM files avoiding the reads classified to rRNA, tRNA, chloroplasts and mitochondria, we conducted a m⁶A region enrichment analysis using MACS2 callpeak (v.2.2.7.1)⁷⁰ with the following settings: nomodel; shift, 25; extsize, 50; slocal, 200; llocal, 500; f, BAM; q, 0.05; g, 1.3e + 8. We considered regions exhibiting a false discovery rate below 0.05 for the analysis of m⁶A peaks. Referring to the strategy and parameters used for m⁶A peak calibration with IVT RNAs⁷¹, we applied the same elimination rate criteria. If m⁶A peaks appeared on retrotransposon RNAs in caRNA m⁶A-Seq and did not overlap by more than 50% with those in m⁶A-Seq, we considered these peaks as belonging to retrotransposon RNAs.

The m⁶A level was calculated as previously described¹⁶. Specifically, to determine the m⁶A level for each transcript in the WT, we first calculated the CPM using edgeR (v.3.36.0)⁶⁷. The m⁶A level was then derived using the formula (CPM_IP + 0.01)/(CPM_Input + 0.01). To account for differences in IP efficiency between WT and mutant samples, an equal amount of m⁶A-modified spike-in RNA from the EpiMark N⁶-Methyladenosine Enrichment Kit (NEB) was added to the rRNA-depleted caRNA samples of both the WT and mutants as an external control. For normalization across samples, the m⁶A ratio of the spike-in (r_spike-in) was calculated as (CPM_IP-spike-in + 0.01)/(CPM_Input-spike-in + 0.01). The m⁶A normalization factor (n.f.) for each mutant was defined as the ratio of r_spike-in in each mutant to the average r_spike-in across all WT samples. This factor corrects for differences in IP efficiency, and the final m⁶A level for each mutant was determined using the formula (CPM_IP + 0.01)/(CPM_Input + 0.01)/n.f. The decrease in m⁶A level was determined using the formula (m⁶A level in WT − m⁶A level in mutant)/(m⁶A level in WT).

Data analysis for nuclear RNA stability assay

The data processing procedure to generate the BAM files mirrors the procedure we described above. Subsequent to this, gene expression quantification was performed using featureCounts (v.2.0.1)⁶⁶ and edgeR (v.3.36.0)⁶⁷ to compute counts normalized by ERCC and CPM at each time point. Setting the expression level at the initial time to 1, we fitted all degradation curves using the equation y = exp(−A × x), where y is the relative expression level of a certain gene and x is the processing time. Subsequently, we eliminated genes with a value of A less than 0, derived from the fitting equation, from all sets of sequencing data. The half-lives of other genes were computed using the formula t_1/2 = (ln2)/A.

Data analysis for time-course RNA sequencing

The procedure to generate the BAM files for each sample at different time points aligns with the pipeline we outlined above. Following this, gene expression was finished using featureCounts (v.2.0.1)⁶⁶ and edgeR (v.3.36.0)⁶⁷ to calculate gene counts, normalized by ERCC and CPM at each respective time point. We fit all gene expression levels using a linear model represented as y = A × x + B, where y signifies the expression level of a specific gene and x indicates the processing time. We then excluded genes with an A value less than 0, derived from the fitting equation, from all sequencing datasets. The transcription rate of a gene was represented by A.

Data analysis for ChIP–Seq

The alignment process was similar to that for other sequencing data including input groups and H3K9me2/H3K27me1 IP groups. H3K9me2/H3K27me1 peaks were identified using broad-peak mode in MACS2 (v.2.2.7.1)⁷⁰ with false discovery rate < 0.05. To assess the H3K9me2/H3K27me1 levels, heat maps and profile plots were generated using deepTools (v.3.5.1)⁷². The enrichment levels were normalized using reads per kilobase per million mapped reads to account for the differences between the IP and input datasets.

Statistics and reproducibility

The statistical tests performed on experimental data and the sample sizes are noted in the figure legends. All data points are derived from biological replicates.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The raw sequence data from caRNA m⁶A-Seq, m⁶A-Seq, caRNA-Seq, nuclear RNA lifetime sequencing, time-course RNA sequencing and ChIP–Seq reported in this paper have been deposited in the Genome Sequence Archive at the National Genomics Data Center, Beijing Institute of Genomics, Chinese Academy of Sciences, under accession number PRJCA037507; they are publicly accessible at https://bigd.big.ac.cn/gsa. All remaining data are provided in the main paper or the supplementary materials. Source data are provided with this paper.

References

Deng, X. et al. RNA N⁶-methyladenosine modification in cancers: current status and perspectives. Cell Res. 28, 507–517 (2018).
Article PubMed PubMed Central Google Scholar
Zhong, S. et al. MTA is an Arabidopsis messenger RNA adenosine methylase and interacts with a homolog of a sex-specific splicing factor. Plant Cell 20, 1278–1288 (2008).
Article PubMed PubMed Central Google Scholar
Shen, L. et al. N⁶-Methyladenosine RNA modification regulates shoot stem cell fate in Arabidopsis. Dev. Cell 38, 186–200 (2016).
Article PubMed PubMed Central Google Scholar
Růžička, K. et al. Identification of factors required for m⁶A mRNA methylation in Arabidopsis reveals a role for the conserved E3 ubiquitin ligase HAKAI. N. Phytol. 215, 157–172 (2017).
Article Google Scholar
Martínez-Pérez, M. et al. Arabidopsis m⁶A demethylase activity modulates viral infection of a plant virus and the m⁶A abundance in its genomic RNAs. Proc. Natl Acad. Sci. USA 114, 10755–10760 (2017).
Article PubMed PubMed Central Google Scholar
Duan, H. C. et al. ALKBH10B is an RNA N⁶-methyladenosine demethylase affecting Arabidopsis floral transition. Plant Cell 29, 2995–3011 (2017).
Article PubMed PubMed Central Google Scholar
Arribas-Hernández, L. et al. An m⁶A–YTH module controls developmental timing and morphogenesis in Arabidopsis. Plant Cell 30, 952–967 (2018).
Article PubMed PubMed Central Google Scholar
Wei, L. H. et al. The m⁶A reader ECT2 controls trichome morphology by affecting mRNA stability in Arabidopsis. Plant Cell 30, 968–985 (2018).
Article PubMed PubMed Central Google Scholar
Hou, Y. et al. CPSF30-L-mediated recognition of mRNA m⁶A modification controls alternative polyadenylation of nitrate signaling-related gene transcripts in Arabidopsis. Mol. Plant 14, 688–699 (2021).
Article PubMed Google Scholar
Song, P. et al. Arabidopsis N⁶-methyladenosine reader CPSF30-L recognizes FUE signals to control polyadenylation site choice in liquid-like nuclear bodies. Mol. Plant 14, 571–587 (2021).
Article PubMed Google Scholar
Wu, X. et al. N⁶-Methyladenosine-mediated feedback regulation of abscisic acid perception via phase-separated ECT8 condensates in Arabidopsis. Nat. Plants 10, 469–482 (2024).
Article PubMed Google Scholar
Cai, Z. et al. The m⁶A reader ECT8 is an abiotic stress sensor that accelerates mRNA decay in Arabidopsis. Plant Cell 36, 2908–2926 (2024).
Article PubMed PubMed Central Google Scholar
Song, P. et al. m⁶A readers ECT2/ECT3/ECT4 enhance mRNA stability through direct recruitment of the poly(A) binding proteins in Arabidopsis. Genome Biol. 24, 103 (2023).
Article PubMed PubMed Central Google Scholar
Xu, C. et al. R-loop resolution promotes co-transcriptional chromatin silencing. Nat. Commun. 12, 1790 (2021).
Article PubMed PubMed Central Google Scholar
Liu, J. et al. N⁶-Methyladenosine of chromosome-associated regulatory RNA regulates chromatin state and transcription. Science 367, 580–586 (2020).
Article PubMed PubMed Central Google Scholar
Wei, J. et al. FTO mediates LINE1 m⁶A demethylation and chromatin regulation in mESCs and mouse development. Science 376, 968–973 (2022).
Article PubMed PubMed Central Google Scholar
Li, Y. et al. N⁶-Methyladenosine co-transcriptionally directs the demethylation of histone H3K9me2. Nat. Genet. 52, 870–877 (2020).
Article PubMed Google Scholar
Liu, J. et al. The RNA m⁶A reader YTHDC1 silences retrotransposons and guards ES cell identity. Nature 591, 322–326 (2021).
Article PubMed Google Scholar
Chelmicki, T. et al. m⁶A RNA methylation regulates the fate of endogenous retroviruses. Nature 591, 312–316 (2021).
Article PubMed Google Scholar
Feschotte, C. Transposable elements and the evolution of regulatory networks. Nat. Rev. Genet. 9, 397–405 (2008).
Article PubMed PubMed Central Google Scholar
Lander, E. S. et al. Initial sequencing and analysis of the human genome. Nature 409, 860–921 (2001).
Article PubMed Google Scholar
Lanciano, S. & Cristofari, G. Measuring and interpreting transposable element expression. Nat. Rev. Genet. 21, 721–736 (2020).
Article PubMed Google Scholar
Jiao, Y. et al. Improved maize reference genome with single-molecule technologies. Nature 546, 524–527 (2017).
Article PubMed PubMed Central Google Scholar
Liu, C., Lu, F., Cui, X. & Cao, X. Histone methylation in higher plants. Annu. Rev. Plant Biol. 61, 395–420 (2010).
Article PubMed Google Scholar
Jambhekar, A., Dhall, A. & Shi, Y. Roles and regulation of histone methylation in animal development. Nat. Rev. Mol. Cell Biol. 20, 625–641 (2019).
Article PubMed PubMed Central Google Scholar
Xu, W. et al. METTL3 regulates heterochromatin in mouse embryonic stem cells. Nature 591, 317–321 (2021).
Article PubMed Google Scholar
Yu, Q. et al. RNA demethylation increases the yield and biomass of rice and potato plants in field trials. Nat. Biotechnol. 39, 1581–1588 (2021).
Article PubMed Google Scholar
Quesneville, H. Twenty years of transposable element analysis in the Arabidopsis thaliana genome. Mob. DNA 11, 28 (2020).
Article PubMed PubMed Central Google Scholar
Ream, T. S. et al. Subunit compositions of the RNA-silencing enzymes Pol IV and Pol V reveal their origins as specialized forms of RNA polymerase II. Mol. Cell 33, 192–203 (2009).
Article PubMed Google Scholar
Wierzbicki, A. T., Haag, J. R. & Pikaard, C. S. Noncoding transcription by RNA polymerase Pol IVb/Pol V mediates transcriptional silencing of overlapping and adjacent genes. Cell 135, 635–648 (2008).
Article PubMed PubMed Central Google Scholar
Böhmdorfer, G. et al. Long non-coding RNA produced by RNA polymerase V determines boundaries of heterochromatin. eLife 5, e19092 (2016).
Article PubMed PubMed Central Google Scholar
Zhai, J. et al. A one precursor one siRNA model for Pol IV-dependent siRNA biogenesis. Cell 163, 445–455 (2015).
Article PubMed PubMed Central Google Scholar
Zheng, B. et al. Intergenic transcription by RNA polymerase II coordinates Pol IV and Pol V in siRNA-directed transcriptional gene silencing in Arabidopsis. Genes Dev. 23, 2850–2860 (2009).
Article PubMed PubMed Central Google Scholar
Pontes, O. et al. The Arabidopsis chromatin-modifying nuclear siRNA pathway involves a nucleolar RNA processing center. Cell 126, 79–92 (2006).
Article PubMed Google Scholar
Liu, C. et al. Arabidopsis ARGONAUTE 1 binds chromatin to promote gene transcription in response to hormones and stresses. Dev. Cell 44, 348–361.e347 (2018).
Article PubMed Google Scholar
Liu, Y. et al. A conserved Pol II elongator SPT6L mediates Pol V transcription to regulate RNA-directed DNA methylation in Arabidopsis. Nat. Commun. 15, 4460 (2024).
Article PubMed PubMed Central Google Scholar
Shen, L. Functional interdependence of N⁶-methyladenosine methyltransferase complex subunits in Arabidopsis. Plant Cell 35, 1901–1916 (2023).
Article PubMed PubMed Central Google Scholar
Wang, X. et al. N⁶-Methyladenosine-dependent regulation of messenger RNA stability. Nature 505, 117–120 (2014).
Article PubMed Google Scholar
Wang, X. et al. N⁶-Methyladenosine modulates messenger RNA translation efficiency. Cell 161, 1388–1399 (2015).
Article PubMed PubMed Central Google Scholar
Zaib, S., Rana, N. & Khan, I. Histone modifications and their role in epigenetics of cancer. Curr. Med. Chem. 29, 2399–2411 (2022).
Article PubMed Google Scholar
Luo, C., Durgin, B. G., Watanabe, N. & Lam, E. Defining the functional network of epigenetic regulators in Arabidopsis thaliana. Mol. Plant 2, 661–674 (2009).
Article PubMed Google Scholar
Antunez-Sanchez, J. et al. A new role for histone demethylases in the maintenance of plant genome integrity. eLife 9, e58533 (2020).
Article PubMed PubMed Central Google Scholar
Roudier, F. et al. Integrative epigenomic mapping defines four main chromatin states in Arabidopsis. EMBO J. 30, 1928–1938 (2011).
Article PubMed PubMed Central Google Scholar
Allshire, R. C. & Madhani, H. D. Ten principles of heterochromatin formation and function. Nat. Rev. Mol. Cell Biol. 19, 229–244 (2018).
Article PubMed Google Scholar
Du, J. et al. Mechanism of DNA methylation-directed histone methylation by KRYPTONITE. Mol. Cell 55, 495–504 (2014).
Article PubMed PubMed Central Google Scholar
Ebbs, M. L. & Bender, J. Locus-specific control of DNA methylation by the Arabidopsis SUVH5 histone methyltransferase. Plant Cell 18, 1166–1176 (2006).
Article PubMed PubMed Central Google Scholar
Ebbs, M. L., Bartee, L. & Bender, J. H3 lysine 9 methylation is maintained on a transcribed inverted repeat by combined action of SUVH6 and SUVH4 methyltransferases. Mol. Cell. Biol. 25, 10507–10515 (2005).
Article PubMed PubMed Central Google Scholar
Jackson, J. P., Lindroth, A. M., Cao, X. & Jacobsen, S. E. Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 416, 556–560 (2002).
Article PubMed Google Scholar
Jacob, Y. et al. ATXR5 and ATXR6 are H3K27 monomethyltransferases required for chromatin structure and gene silencing. Nat. Struct. Mol. Biol. 16, 763–768 (2009).
Article PubMed PubMed Central Google Scholar
Li, X. et al. Mechanistic insights into plant SUVH family H3K9 methyltransferases and their binding to context-biased non-CG DNA methylation. Proc. Natl Acad. Sci. USA 115, E8793–E8802 (2018).
PubMed PubMed Central Google Scholar
Ma, Z. et al. Arabidopsis Serrate coordinates histone methyltransferases ATXR5/6 and RNA processing factor RDR6 to regulate transposon expression. Dev. Cell 45, 769–784 (2018).
Article PubMed Google Scholar
Amara, U., Hu, J., Park, S. J. & Kang, H. ECT12, an YTH-domain protein, is a potential mRNA m⁶A reader that affects abiotic stress responses by modulating mRNA stability in Arabidopsis. Plant Physiol. Biochem. 206, 108255 (2024).
Article PubMed Google Scholar
Stroud, H. et al. Non-CG methylation patterns shape the epigenetic landscape in Arabidopsis. Nat. Struct. Mol. Biol. 21, 64–72 (2014).
Article PubMed Google Scholar
Law, J. A. & Jacobsen, S. E. Establishing, maintaining and modifying DNA methylation patterns in plants and animals. Nat. Rev. Genet. 11, 204–220 (2010).
Article PubMed PubMed Central Google Scholar
Matzke, M. A. & Mosher, R. A. RNA-directed DNA methylation: an epigenetic pathway of increasing complexity. Nat. Rev. Genet. 15, 394–408 (2014).
Article PubMed Google Scholar
Deng, S. et al. RNA m⁶A regulates transcription via DNA demethylation and chromatin accessibility. Nat. Genet. 54, 1427–1437 (2022).
Article PubMed Google Scholar
Quarto, G. et al. Fine-tuning of gene expression through the Mettl3–Mettl14–Dnmt1 axis controls ESC differentiation. Cell 188, 998–1018.e1026 (2025).
Article PubMed Google Scholar
Li, X. et al. LINE-1 transcription activates long-range gene expression. Nat. Genet. 56, 1494–1502 (2024).
Article PubMed Google Scholar
Lisch, D. How important are transposons for plant evolution? Nat. Rev. Genet. 14, 49–61 (2013).
Article PubMed Google Scholar
Ito, H. et al. An siRNA pathway prevents transgenerational retrotransposition in plants subjected to stress. Nature 472, 115–119 (2011).
Article PubMed Google Scholar
Long, Y., Jia, J., Mo, W., Jin, X. & Zhai, J. FLEP-seq: simultaneous detection of RNA polymerase II position, splicing status, polyadenylation site and poly(A) tail length at genome-wide scale by single-molecule nascent RNA sequencing. Nat. Protoc. 16, 4355–4381 (2021).
Article PubMed Google Scholar
Xue, M. et al. The INO80 chromatin remodeling complex promotes thermomorphogenesis by connecting H2A.Z eviction and active transcription in Arabidopsis. Mol. Plant 14, 1799–1813 (2021).
Article PubMed Google Scholar
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10–12 (2011).
Article Google Scholar
Kim, D., Langmead, B. & Salzberg, S. L. HISAT: a fast spliced aligner with low memory requirements. Nat. Methods 12, 357–360 (2015).
Article PubMed PubMed Central Google Scholar
Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).
Article PubMed PubMed Central Google Scholar
Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014).
Article PubMed Google Scholar
Chen, Y., Lun, A. T. & Smyth, G. K. From reads to genes to pathways: differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline. F1000Res. 5, 1438 (2016).
PubMed PubMed Central Google Scholar
Risso, D., Ngai, J., Speed, T. P. & Dudoit, S. Normalization of RNA-seq data using factor analysis of control genes or samples. Nat. Biotechnol. 32, 896–902 (2014).
Article PubMed PubMed Central Google Scholar
R Core Team. R: a language and environment for statistical computing. R package version 4.3.2. https://www.R-project.org/ (2025).
Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008).
Article PubMed PubMed Central Google Scholar
Zhang, Z. et al. Systematic calibration of epitranscriptomic maps using a synthetic modification-free RNA library. Nat. Methods 18, 1213–1222 (2021).
Article PubMed Google Scholar
Ramírez, F., Dündar, F., Diehl, S., Grüning, B. A. & Manke, T. DeepTools: a flexible platform for exploring deep-sequencing data. Nucleic Acids Res. 42, W187–W191 (2014).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank H. Yu (National University of Singapore), Y. Qi (Tsinghua University), B. Zheng (Fudan University) and C. Chen (Chinese Academy of Sciences) for providing the fip37-4 LEC1::FIP37, nrpe1-11, nrpb2-3 and NRPE1::NRPE1-GFP nrpe1 lines. We also thank the National Center for Protein Sciences at Peking University in Beijing, China, for assistance with sequencing, and G. Li for help with the MGI2000 experiment. This work was supported by the Biological Breeding-National Science and Technology Major Project of China (grant no. 2023ZD04073, G.J.), the National Key R&D Program of China (grant no. 2024YFF1000302, G.J.) and the National Natural Science Foundation of China (grant nos. 22225704, 22477005 and 22321005, G.J.).

Author information

These authors contributed equally: Peizhe Song, Zhihe Cai, Subiding Tayier.

Authors and Affiliations

Synthetic and Functional Biomolecules Center, Beijing National Laboratory for Molecular Sciences, Key Laboratory of Bioorganic Chemistry and Molecular Engineering of Ministry of Education, College of Chemistry and Molecular Engineering, Peking University, Beijing, China
Peizhe Song, Zhihe Cai, Subiding Tayier, Enlin Tian, Zixin Chen, Lixiang Liu & Guifang Jia
Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, China
Kemiao Yu & Guifang Jia
Beijing Advanced Center of RNA Biology (BEACON), Peking University, Beijing, China
Guifang Jia

Authors

Peizhe Song
View author publications
Search author on:PubMed Google Scholar
Zhihe Cai
View author publications
Search author on:PubMed Google Scholar
Subiding Tayier
View author publications
Search author on:PubMed Google Scholar
Enlin Tian
View author publications
Search author on:PubMed Google Scholar
Zixin Chen
View author publications
Search author on:PubMed Google Scholar
Kemiao Yu
View author publications
Search author on:PubMed Google Scholar
Lixiang Liu
View author publications
Search author on:PubMed Google Scholar
Guifang Jia
View author publications
Search author on:PubMed Google Scholar

Contributions

G.J. and P.S. conceived of the project. P.S. performed the experiments with the help of S.T., E.T., K.Y., L.L. and Z. Cai. Z. Cai, Z. Chen and P.S. analysed the sequencing data. G.J. and P.S. designed the experiments, interpreted the results and wrote the paper. All authors commented on the paper.

Corresponding author

Correspondence to Guifang Jia.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Plants thanks Leandro Quadrana, Jixian Zhai and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary Figs. 1–12.

Reporting Summary

Supplementary Table 1

Statistics of m⁶A peaks from caRNA m⁶A-Seq in 6-day-old WT seedlings.

Supplementary Table 2

Statistics of m⁶A peaks from m⁶A-Seq in 6-day-old WT seedlings.

Supplementary Table 3

Statistics of m⁶A peaks on retrotransposon RNAs in 6-day-old WT seedlings.

Supplementary Table 4

Statistics of expression levels of retrotransposon RNAs in 6-day-old WT, nrpb2-3 and nrpe1-11 seedlings.

Supplementary Table 5

Statistics of m⁶A peaks from caRNA m⁶A-Seq in 6-day-old fip37-4 LEC1::FIP37 seedlings.

Supplementary Table 6

Statistics of m⁶A peaks from caRNA m⁶A-Seq in 6-day-old vir seedlings.

Supplementary Table 7

Statistics of relative expression levels of m⁶A retrotransposon RNAs in 6-day-old WT, fip37-4 LEC1::FIP37 and vir seedlings.

Supplementary Table 8

Statistics of relative expression levels of m⁶A retrotransposon RNAs in 6-day-old WT, cpsf30-l, CPSF30-L and CPSF30-Lm seedlings.

Supplementary Table 9

Statistics of half-lives of m⁶A retrotransposon RNAs in 6-day-old WT and vir seedlings.

Supplementary Table 10

Statistics of half-lives of m⁶A retrotransposon RNAs in 6-day-old WT and cpsf30-l seedlings.

Supplementary Table 11

Statistics of transcription rates of m⁶A retrotransposon RNAs in 6-day-old WT and cpsf30-l seedlings.

Supplementary Table 12

Statistics of H3K9me2 levels of m⁶A retrotransposons in 6-day-old WT, vir, cpsf30-l, ect12-3 and cpsf30-l/ect12 seedlings.

Supplementary Table 13

Statistics of H3K27me1 levels of m⁶A retrotransposons in 6-day-old WT, vir, cpsf30-l, ect12-3 and cpsf30-l/ect12 seedlings.

Supplementary Table 14

Statistics of relative expression levels of m⁶A retrotransposons in 6-day-old WT, cpsf30-l, ect12-3 and cpsf30-l/ect12 seedlings.

Supplementary Table 15

List of primers used in this study.

Supplementary Data 1

Source data for Supplementary Figs. 1–5 and 8–10.

Source data

Source Data Fig. 1

Unprocessed western blots.

Source Data Fig. 5

Unprocessed western blots.

Source Data Fig. 6

Unprocessed western blots.

Source Data Figs. 1–7

Statistical source data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Song, P., Cai, Z., Tayier, S. et al. RNA m⁶A regulates the transcription and heterochromatin state of retrotransposons in Arabidopsis. Nat. Plants 11, 2300–2318 (2025). https://doi.org/10.1038/s41477-025-02137-z

Download citation

Received: 19 July 2024
Accepted: 21 September 2025
Published: 24 October 2025
Version of record: 24 October 2025
Issue date: November 2025
DOI: https://doi.org/10.1038/s41477-025-02137-z

Subjects

Abstract

Similar content being viewed by others

Main

Results

Landscape of m6A retrotransposon RNAs revealed via caRNA m6A-Seq in Arabidopsis

m6A is deposited on retrotransposon RNA transcribed by Pol II and Pol V

The m6A modification on retrotransposon RNAs deposited by the MTA–MTB writer complex represses the expression of retrotransposon RNAs

CPSF30-L binds and suppresses the expression of m6A retrotransposon RNAs

m6A has less effect on the stability of retrotransposon RNAs

m6A silences the transcription of retrotransposon RNAs

m6A on retrotransposon RNAs maintains local closed heterochromatin by enhancing the histone repressive modifications H3 K9 dimethylation and H3 K27 monomethylation

The m6A reader protein CPSF30-L associates with SUVH4/5/6 and ATXR5/6 and is required for H3K9me2/H3K27me1 levels at m6A retrotransposons

ECT12, as a nuclear m6A reader, interacts with CPSF30-L and enhances the m6A-binding function

CPSF30-L and ECT12 redundantly mediate H3K9me2 and H3K27me1 formation and suppress the transcription of m6A retrotransposon RNAs

Discussion

Methods

Plant materials and growth conditions

RT-qPCR

Extraction of caRNAs

caRNA m6A-Seq and m6A-Seq

m6A-IP–qPCR

Nuclear FA-CLIP

In vivo FA-RIP-qPCR

Nuclear RNA lifetime sequencing

Nuclear RNA lifetime measurement via qPCR

Time-course RNA sequencing for transcription rate measurement

Nuclear run-on assay

ChIP–Seq and ChIP-qPCR

Immunostaining

Y2H assay

BiFC assay

Semi-in vivo pull-down assay

Protein expression and purification

Electrophoretic mobility shift assay

In vivo nuclear FA-RIP–LC–MS/MS

In vitro RIP–LC–MS/MS

Quantification analysis via LC–MS/MS

Data analysis for caRNA-Seq and RNA-Seq

Data analysis for caRNA m6A-Seq and m6A-Seq

Data analysis for nuclear RNA stability assay

Data analysis for time-course RNA sequencing

Data analysis for ChIP–Seq

Statistics and reproducibility

Reporting summary

Data availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding author

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

Search

Quick links

Landscape of m⁶A retrotransposon RNAs revealed via caRNA m⁶A-Seq in Arabidopsis

m⁶A is deposited on retrotransposon RNA transcribed by Pol II and Pol V

The m⁶A modification on retrotransposon RNAs deposited by the MTA–MTB writer complex represses the expression of retrotransposon RNAs

CPSF30-L binds and suppresses the expression of m⁶A retrotransposon RNAs

m⁶A has less effect on the stability of retrotransposon RNAs

m⁶A silences the transcription of retrotransposon RNAs

m⁶A on retrotransposon RNAs maintains local closed heterochromatin by enhancing the histone repressive modifications H3 K9 dimethylation and H3 K27 monomethylation

The m⁶A reader protein CPSF30-L associates with SUVH4/5/6 and ATXR5/6 and is required for H3K9me2/H3K27me1 levels at m⁶A retrotransposons

ECT12, as a nuclear m⁶A reader, interacts with CPSF30-L and enhances the m⁶A-binding function

CPSF30-L and ECT12 redundantly mediate H3K9me2 and H3K27me1 formation and suppress the transcription of m⁶A retrotransposon RNAs

caRNA m⁶A-Seq and m⁶A-Seq

m⁶A-IP–qPCR

Data analysis for caRNA m⁶A-Seq and m⁶A-Seq