The CARBON CATABOLITE REPRESSION 4A-mediated RNA deadenylation pathway acts on the transposon RNAs that are not regulated by small RNAs

siRNAs. This indicates that the cytoplasmic RNA quality control mechanism targets the TEs that are poorly recognized by the previously well-characterized RDR6-mediated pathway, and thereby augments the host genome stability. (cid:1) Our study suggests a hitherto unknown mechanism for transposon repression mediated by RNA deadenylation and unveils a complex nature of the host’s strategy to maintain the genome integrity.


Introduction
Transposable elements (TEs) are mobile genetic elements that pose a significant threat to the host genome stability and integrity.It is well documented that transposons are subject to epigenetic silencing that is mediated by a so-called RNA-directed DNA methylation (RdDM; Matzke & Mosher, 2014).TEs that escape such transcriptional suppression or are newly introduced to the host genome, thus are not yet epigenetically silenced, are recognized by the RNA-DEPENDENT RNA POLYMERASE 6 (RDR6)-SUPPRESSOR OF GENE SILENCING 3 (SGS3) complex, which generates 21/22-nucleotide (nt) small interfering (si) RNAs and initiates epigenetic silencing (Nuthikattu et al., 2013;Creasey et al., 2014;Panda et al., 2016;Lee et al., 2020).Accumulating evidence suggests that the incompleteness of mRNA (e.g.tail-less or truncated mRNA) is crucial for specific targeting of RDR6 (Luo & Chen, 2007;Creasey et al., 2014;Baeg et al., 2017).In our previous study, we showed that the suboptimal codon usage of transposons causes ribosome stalling and RNA cleavage, which accounts for their frequent targeting to the RDR6-mediated siRNA biogenesis pathway (Kim et al., 2021).In addition, the ribosome-stalled transcripts are preferentially guided to cytoplasmic compartments where SGS3 and RDR6 are localized (Kim et al., 2021;Han et al., 2023;Tan et al., 2023).It is also worth noting that the 21/22-nt siRNAs are associated with only around one-third of active and transcribed TEs in Arabidopsis, and the posttranscriptional suppression of transposon that is independent of siRNAs is largely unknown.
Despite the inherent aberrancy of transposon transcripts, their regulation by the cellular RNA surveillance system has been poorly reported.In Drosophila, for instance, it was suggested that the mutants defective in CCR4 accumulate TE transcripts in the chromatin-associated RNA fraction, and the CCR4-NOT complex interacts with the piRNA pathway components in the nucleus, indicating a co-transcriptional suppression of transposons (Kordyukova et al., 2020).In addition, a CCR4-NOT complex component NOT1 was identified in a genetic screen for RdDM regulators in Arabidopsis (Zhou et al., 2020).However, it is important to note that it is still uncertain whether TE suppression by the CCR4-NOT complex requires the catalytic activity of RNA deadenylation.
In this study, we investigated the mutants for RNA deadenylases in Arabidopsis and assessed the transposon RNA levels.Intriguingly, we found that RNA deadenylases suppress a set of transposons that are not usually regulated by the RDR6-mediated pathway.Oxford Nanopore direct RNA sequencing (ONT-DRS) revealed that CCR4a shortens the poly(A) tails, destabilizes the transcripts, and reduces the steady-state mRNA levels of transposons.Moreover, we also carried out whole-genome resequencing and droplet digital PCR (ddPCR) experiments to interrogate the mobilization of TEs and observed an increased mobility of transposons in the deadenylase mutants.Our study unveils a previously unknown cellular mechanism that degrades transposon RNAs through an evolutionarily conserved RNA surveillance system.

Plant materials and growth condition
All Arabidopsis (Arabidopsis thaliana L. Heynh) plants used in this study are in the Col-0 background.The ccr4a-1 (SAIL_802_A10), ccr4b-1 (SALK_151541C), caf1a-1 (SALK_070336), and caf1b-3 (SALK_044043) mutants were obtained from the Arabidopsis Biological Resources Center (https://abrc.osu.edu/).The caf1a-1 caf1b-3 double mutant was identified from the F2 segregation population derived from crosses.De novo ddm1 mutants were generated by the CRISPR-Cas9 system containing three sgRNAs.The sgRNA sequences were designed by an online web tool (https://chopchop.cbu.uib.no/), and the sgRNA secondary structure was predicted in the UNAFold web server (www.unafold.org/mfold/).The synthesized oligonucleotides were annealed and inserted into the digested entry vectors pENTR_L4_R1, pENTR_L1_L2, and pENTR_R2_L3 at the BbsI (NEB) sites.The entry vectors were subsequently transferred to a destination vector pFG7m34GW (Shimada et al., 2010), carrying the Fast-Green fluorescent seed selection marker and proUBQ10-driven Cas9 cassettes, using the Gateway LR reaction (Thermo Fisher Scientific, Waltham, MA, USA).Agrobacterium tumefaciensmediated floral dip method was used to transform the DDM1-targeting pFG7m34GW vector into the indicated mutant background.Editing events were confirmed by Sanger sequencing.We were unable to identify any mutations at the sgRNA3-targeted regions, and the editing events at the regions targeted by sgRNA1 and sgRNA2 are summarized in Supporting Information Fig. S1.T-DNA was segregated out at T3 generation, and unless otherwise stated, plants at T4 generation were used in this study.Sequences of sgRNAs are listed in Table S1.

RT-qPCR
Total RNA was isolated from 10-d-old seedlings using the TRIzol extraction method (Tiangen, Beijing, China) and reversetranscribed using ReverTra Ace qPCR RT Master Mix with gDNA Remover (Toyobo, Osaka, Japan).To quantify the relative abundance of transcripts, quantitative PCR was carried out using a ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad).Actin2 (AT3G18780) was used as an internal control for normalization.Gene expression levels were determined by the DDC t method.Sequences of primers are listed in Table S1.

RNA-Seq
For RNA-Seq library construction, total RNA was isolated from 10-d-old seedlings using the TRIzol Reagent (Invitrogen), and poly(A)-RNA was purified from 3 lg of total RNA using poly-T oligo-attached magnetic beads.Library was prepared using the NEBNext Ultra Directional RNA Library Prep Kit (NEB) following the manufacturer's instructions.Sequencing was performed on an Illumina NovaSeq 6000 platform, and 150-bp paired-end (PE150) reads were generated.RNA-Seq dataset is summarized in Table S2.
Oxford Nanopore direct RNA sequencing (ONT-DRS) Total RNA was isolated from 10-d-old seedlings by TRIzol (Qiagen), and poly(A)-RNA was purified using Dynabeads mRNA Purification Kit (Invitrogen) following the manufacturer's instructions.The quality and quantity of mRNA were assessed using the Nano-Drop 2000 spectrophotometer and Qubit.Library was prepared using direct RNA sequencing kit (SQK-RNA002; Oxford Nanopore Technologies, Oxord, UK), loaded onto an R9.4 Flow Cell (Flow cell type FLO-MIN106), and sequenced on a GridION device for 72 h.ONT-DRS dataset is summarized in Table S3.
The raw nanopore signals were converted to base sequences by GUPPY (v.6.1.5)using the high-accuracy basecalling model.Since transposons are not properly annotated in the reference assembly of Arabidopsis and therefore often omitted in the downstream analysis, we generated a custom transcript assembly by merging the reference transcript assembly and all the de novo assembled transcripts derived from the RNA-Seq data generated in this study using STRINGTIE (v.2.1.7;Pertea et al., 2015).Then, the nanopore reads with a mean quality score > 7 were mapped to the custom transcriptome using MINIMAP2 (v.2.24-r1122; Li, 2018) with the following parameters: -ax map-ont -L -p 0 -N 10.Poly(A) tail length was detected by NANOPOLISH (v.0.13.3;Workman et al., 2019).Transcripts with > 15 reads were used to obtain the median poly(A) tail length.The reads with poly(A) tail were re-aligned to TAIR10 genome with the following parameters: -ax splice -k14 -uf and visualized by the python genome package BUGV (Jia et al., 2022).

mRNA half-life
Four-day-old etiolated Arabidopsis seedlings were immersed in cordycepin solution (1 mM PIPES (pH 6.25), 15 mM sucrose, 1 mM KCl, 1 mM sodium citrate, and 1 lM cordycepin) and harvested at 0, 0.25, 0.5, 1, 2, and 4 h for three biological replicates.RNA extraction and RNA-Seq were performed as described previously.mRNA half-lives were calculated as follows: decay rate K i = Àlog e (F i /F 0 )/T i , in which F i is the FPKM at time i, and T i is the time of cordycepin treatment.K i was calculated from each time point, and the half-life is log e (2)/K a , in which K a is the average decay rate measured for all time points.

Whole-genome resequencing
Genomic DNA was extracted using the CTAB method.One microgram genomic DNA was randomly fragmented by ultrasonicator (Covaris, Woburn, MA, USA).An average size of 200-400-bp DNA fragments was selected by Agencourt AMPure XP-Medium kit.The fragments were then end-repaired, 3 0 adenylated and ligated with adaptors.The purified double-stranded products were heat denatured to single-stranded DNA and then circularized.The single-stranded circular DNA was sequenced by a DNBSEQ-T7 generating 150-bp paired-end reads.Wholegenome resequencing dataset is summarized in Table S4.
Paired-end short-read whole-genome sequencing data were mapped to TAIR10 and processed following the SPLITREADER pipeline (Baduel et al., 2021).Briefly, discordantly mapping and nonmapping reads were recovered, and then, the reads were remapped to the TE pools and the genome.Insertions supported by at least three reads (DP filter = 3) were filtered and only nonreference insertions were considered.

Droplet digital PCR
Droplet digital PCR was performed on TargetingOne ® Digital PCR System (TargetingOne, Beijing, China) following the manufacturer's instruction.Briefly, Genomic DNA was extracted using a N96 DNAsecure Plant Kit (Tiangen).One hundred nanogram of genomic DNA was digested using AluI (NEB) for 4 h at 37°C.The digested DNA was quantified using the Qubit4 DNA quantification system (Thermo Fisher Scientific) and diluted to 0.15 ng ll À1 .The reaction mixture containing 29 ddPCR Supermix (Bio-Rad), 0.8 lM primer, 0.25 lM probe, and 0.6 ng of cleaved sample DNA was thoroughly mixed and added into the droplet generation chip.Then, 180 ll of droplet generation oil was added to the mixture in the reaction mix inlet.Subsequently, the generated droplets were transferred into an 8strip PCR tube and used for PCR reaction that was performed on a PTC-200 Thermal Cycler.FAM (488 nm) and VIC (532 nm) fluorescence signals were detected through the separate channels on the Chip Reader.Finally, the data were subjected to Poisson distribution analysis using the Chip Reader R1 software to obtain the target DNA copy numbers.Sequences of primers and probes are listed in Table S1.

mRNA deadenylases suppress transposons
We previously showed that Arabidopsis TE RNAs often undergo ribosome stalling and RNA cleavage, which are required for the RDR6-mediated siRNA biogenesis (Kim et al., 2021).However, a substantial fraction of TEs with signatures of ribosome stalling and RNA cleavage is not associated with siRNAs.Since such aberrancy of RNA is monitored and resolved by RNA surveillance and decay pathways (Harigaya & Parker, 2010;Chen & Shyu, 2011;Graille & S eraphin, 2012;Shoemaker & Green, 2012;D'Orazio & Green, 2021), we reasoned that transposon RNAs might be controlled by the RNA degradation pathways.To test this possibility, we first identified the Arabidopsis mutants for RNA deadenylases (ccr4a-1 and ccr4b-1 single mutants, and caf1a-1 caf1b-3 double mutant) and induced de novo mutations in DDM1 using CRISPR-Cas9 to release transposons from epigenetic silencing (Fig. S1).It is worth noting that pre-existing ddm1 mutants contain many newly inserted transposons that could be unevenly segregated in genetic crosses with other mutants and therefore may lead to erroneous assessment of transposon expression.For this reason, we generated de novo mutants of DDM1 and used the plant materials collected in the same generation (See Materials and Methods section).RNA-Seq was then carried out in two independent ddm1 mutant alleles of each RNA deadenylase mutant (Fig. S2).Our transcriptome analysis identified hundreds of genes that are up-or downregulated; however, transposons exhibited a strikingly different pattern that most differentially expressed transposons are upregulated in the deadenylase mutants (Fig. 1a-c).We then compared the upregulated transposons in these double and triple mutants and found that a large fraction of TEs is commonly upregulated, while CCR4a displays the greatest impact on transposon RNA levels (Figs 1d,e, S3).These data imply that the mRNA deadenylation pathway is involved in transposon repression.

Differential TE control by RDR6 and CCR4a
It is well documented that some transposons give rise to 21/22-nt siRNAs that can target transposon RNAs for cleavage (Nuthikattu et al., 2013;Creasey et al., 2014).This specific class of siRNAs is also known as epigenetically activated siRNAs (easiRNAs) and is generated by the RDR6-DICER LIKE 2 and 4 (DCL2/4) module (Gasciolli et al., 2005;Nuthikattu et al., 2013).Since the cleaved transcript products are eliminated by the RNA decay pathways, we suspected that the observed derepression of transposons in the deadenylase mutants might be merely a consequence of compromised RNA decay of the easiRNA-cleaved transcripts.However, the transposons regulated by CCR4a marginally overlapped with those targeted by RDR6 (Fig. 2a), suggesting that the reactivation of these TEs is more strongly associated with the loss of CCR4a than RDR6.Transposon classification analysis further supports this conclusion; the RDR6-regulated transposons are strongly enriched with the LTR/Gypsy family (hypergeometric test, P = 2.57eÀ36), and in the ccr4a mutants, DNA/MuDR DNA transposon family is strongly overrepresented (hypergeometric test, P = 5.34eÀ12; Fig. 2b).To further confirm the divergence of the RDR6-and CCR4a-regulated transposons, we compared the 21/22-and 24-nt siRNA levels.As shown in Fig. 2(c), the 21/22-nt siRNAs of RDR6-controlled TEs were greatly increased in ddm1, whereas CCR4a-regulated transposons exhibited a significant reduction in both classes of siRNAs in ddm1.In addition, the transposons regulated by RDR6 and CCR4a were mapped across the Arabidopsis chromosomes.The RDR6 target transposons were mostly found in the centromeric region, and the transposons regulated by CCR4a were also mapped to the pericentromeric and euchromatic regions in addition to centromeres (Figs 2d, S4).Collectively, loss of mRNA deadenylases is associated with increased RNA levels of TEs that are largely independent to easiRNA and RDR6 control.

CCR4a shortens poly(A) length and destabilizes TE RNAs
We next wanted to assess the poly(A) tail lengths of TE RNAs in the deadenylase mutant.For this, we took advantage of ONT-DRS, which allows for the tail length measurement of native RNA.Transposon transcripts identified by ONT-DRS reproducibly showed a strongly increased levels in the ccr4a mutant (Fig. S5), verifying our observation shown in Fig. 1.The ONT-DRS data from ddm1-L2 revealed that the tail lengths peak at 20, 40-50, and 70-80 nt, which are distanced by c. 25 nt (Fig. 3a,b).A similar pattern was also observed in previous studies (Parker et al., 2020;Jia et al., 2022), suggesting a robust estimation of poly(A) tail length by ONT-DRS.Importantly, the ccr4a-1 ddm1-L2 mutant displayed a longer tail length distribution compared with ddm1-L2 in both genes and transposons (Fig. 3a,b), confirming that CCR4a is a key cellular factor shortening the poly(A) tail.We then retrieved the transposon transcripts from our ONT-DRS dataset and analyzed their tail lengths.As shown in Fig. 3(c,d), TE RNAs possess longer poly(A) tails compared with non-TE transcripts, which is consistent with a previous study (Li et al., 2021), and the loss of CCR4a led to a lengthening of their mRNA tails as did in genic transcripts.
It has been previously reported that highly expressed and stable genes are featured with short steady-state poly(A) tail length (Jia et al., 2022;Passmore & Coller, 2022).Nonetheless, lengthening of poly(A) tail contributes to active translation and RNA stability in humans and plants (Suzuki et al., 2015;Eisen et al., 2020).For example, the poly(A) tail length of a CACTA-like transposon   c, d) in ddm1-L2 and ccr4a-1 ddm1-L2 of Arabidopsis thaliana, shown as heatmap (a, c; color key shows density of distribution) and density plot (b, d).Poly(A) length was measured by Oxford Nanopore direct RNA sequencing.P-value was obtained by the one-sided Wilcoxon rank sum test.(e, f) A CACTA-like transposable element (TE) exhibiting higher RNA levels (e) and longer tail length (f) in ccr4a-1 ddm1-L2 double mutant.In (e), each line represents individual transcript detected by Oxford Nanopore direct RNA sequencing, and poly(A) tail is shown in red line.In (f), mRNA tail lengths of the CACTA-like element in ddm1-L2 and ccr4a-1 ddm1-L2 are compared.White circle, median; black rectangle, upper and lower quartile.P-value was obtained by the one-sided Wilcoxon rank sum test.(g) Fold changes of CCR4a-regulated transposons (n = 48, log 2 -fold change in ddm1-L2 vs ccr4a-1 ddm1-L2) compared with randomly chosen transposons (n = 48).CCR4a-regulated transposons are those with longer tails by at least 10 nt in the ccr4a-1 ddm1-L2 mutant.P-value was obtained by the one-sided Wilcoxon rank sum test.(h) Poly(A) tail length difference in transcripts stabilized by the loss of CCR4a in ddm1-L2 and ccr4a-1 ddm1-L2.mRNA half-lives were determined for ddm1-L2 and ccr4a-1 ddm1-L2 by the transcription arrest assay followed by RNA-Seq.Transcripts with longer half-lives in ccr4a-1 ddm1-L2 by at least 0.5 h were selected (n = 97) and compared against randomly chosen transcripts (n = 100).Tail length difference was calculated by subtracting the tail lengths in ddm1-L2 from those in ccr4a-1 ddm1-L2.P-value was obtained by the one-sided Wilcoxon rank sum test.

Research
New Phytologist became longer, and its RNA level was increased in the ccr4a-1 ddm1-L2 mutant compared with ddm1-L2 (Fig. 3e,f).We further tested other TE transcripts that have longer tails in the ccr4a-1 ddm1-L2 mutant for their levels and found that almost 90% of these TEs are increased in transcript levels in ccr4a-1 ddm1-L2 (Fig. 3g).Moreover, a transcription arrest RNA-Seq was carried out to determine the RNA stability in ddm1-L2 and ccr4a-1 ddm1-L2.For this, seedlings were treated with cordycepin, a transcription elongation inhibitor, and then serially harvested at different time points for RNA-Seq, and mRNA half-lives were determined (See Materials and Methods section for details).Genes that became more stabilized in ccr4a-1 ddm1-L2 as compared to ddm1-L2 (increased half-lives by at least 0.5 h) exhibited longer tail lengths when CCR4a is mutated (Fig. 3h).These data together suggest that CCR4a shortens the poly(A) tail and destabilizes transposon RNAs.
Given that shortening of poly(A) tail is often coupled to translation repression (Tang et al., 2019;Passmore & Coller, 2022) and weak translation leads to RNA localization to cytoplasmic RNA granules, which contain RNA deadenylases and degrading enzymes (Wheeler et al., 2017;Chantarachot & Bailey-Serres, 2018;Arae et al., 2019;Kim et al., 2021), the TE transcripts controlled by RNA deadenylases might be more strongly enriched in RNA granules and actively degraded.Indeed, in the comparison between the RDR6-and CCR4a-regulated transposons, we were able to observe that the transcripts controlled by CCR4a exhibit weaker translational activity and stronger RNA granule enrichment (Fig. S6).Overall, TE transcripts regulated by CCR4a might be guided to the RNA turnover pathway in specific cytoplasmic compartments, further differentiating them from the RDR6-regulated transposons.

CCR4a suppresses transposon mobilization
We have so far demonstrated that loss of RNA deadenylases results in the increased RNA levels of transposons.This led us to test whether transposons mobilize more strongly in the deadenylase mutants.To test this idea, we carried out a wholegenome resequencing experiment to interrogate transposon proliferation.Ten individual plants from each genotype (ddm1, ccr4a-1 ddm1, ccr4b-1 ddm1, and caf1a-1 caf1b-3 ddm1) were randomly chosen and analyzed for nonreference and neoinsertions of transposons using the SPLITREADER pipeline (Baduel et al., 2021).Intriguingly, all three RNA deadenylase mutants showed an increased number of transposon insertions compared with the ddm1 single mutant (Fig. 4a).New insertions were observed for TEs that were previously shown for mobility in natural Arabidopsis population (Quadrana et al., 2016) and ddm1 mutant (Tsukahara et al., 2009).We also found that the transposons that exhibited mobilization are of largely different types from those transcriptionally activated in the RNA deadenylase mutants; for instance, the LTR/Gypsy type was among the most actively induced in ddm1, but it was the LTR/Copia family that was most proliferative in the ddm1 mutant (Fig. 4b).This indicates that an additional layer of regulation exists to control transposon mobilization.Moreover, a ddPCR experiment was carried out to validate the transposition of a representative LTR/Copia element in the Arabidopsis genome known as Evade.ddPCR is an experimental method that can quantitatively measure DNA copies and is particularly useful for assessing transposon copy number (Fan & Cho, 2021;Fan et al., 2022).In this experiment, we used ddm1 and ccr4a-1 ddm1 at T4 and T5 generations.As shown in Fig. 4(c), the copy number of Evade was greatly increased in ccr4a-1 ddm1 as compared with ddm1, further supporting our conclusion that RNA deadenylation represses transposition.In short, the RNA deadenylases destabilize transposon RNAs and inhibit the subsequent step of mobilization in Arabidopsis.

Discussion
In this study, we showed that RNA deadenylation is a critical cellular mechanism that augments the host's general suppression of TEs distinct from and in addition to the easiRNA-mediated pathway.This suggests a previously unknown complexity of transposon control, which complementarily contributes to the The easiRNA-mediated TE repression of the host can be seen as an efficient and persistent way of controlling transposons because it can switch on the epigenetic silencing that can be maintained through cell divisions and generations and target other TE transcripts with similar sequences (Nuthikattu et al., 2013;Creasey et al., 2014).On the contrary, RNA decay is merely degeneration of transcripts and does not generate any signals or biomolecules that can be amplified and transmitted.This partly accounts for why the RDR6-mediated pathway primarily acts on young transposons, particularly those that are structurally intact and long (Panda et al., 2016), while non-TE transcripts are predominantly controlled by RNA decay (Gazzani et al., 2004;Thran et al., 2012;Branscheid et al., 2015;De Alba et al., 2015).In this regard, the CCR4a-targeted TEs might be older in age compared with those regulated by RDR6 and have likely undergone more evolutionary sequence degeneration, which makes them less harmful to the host genome and thus less demanded for the easiRNA-mediated epigenetic silencing.Further investigations into the sequence and structural features determining the target specificity of RDR6 and CCR4a will be a compelling follow-up study.
Our work mainly focused on the cytoplasmic role of RNA deadenylases; however, a previous report suggested a nuclear role of the CCR4-NOT complex in Arabidopsis as one of the essential elements for RdDM (Zhou et al., 2020).This is reminiscent of what is known in Drosophila that CCR4 co-transcriptionally represses TEs in association with Piwi (Kordyukova et al., 2020).These together suggest that the RNA deadenylase complex controls transposons in the nucleus; however, it has not yet been elucidated if the nuclear function of the CCR4-NOT complex requires the RNA deadenylation activity.In this study, we directly demonstrated using ONT-DRS that CCR4a shortens the poly(A) tail lengths of active TEs in Arabidopsis.
In summary, the shortening of TE RNA poly(A) tail length by RNA deadenylases and thereby RNA destabilization is a critical cytoplasmic mechanism suppressing transposon activity.This work unveils a hidden complexity of transposon regulation, which helps broaden our understanding of the host's defense against endogenous parasitic DNA.Table S1 Sequences of primers used in this study.
Table S2 Summary of RNA-Seq.
Table S3 Summary of Oxford Nanopore direct RNA sequencing.
Table S4 Summary of whole-genome resequencing.

Fig. 1
Fig. 1 Loss of mRNA deadenylases leads to transposon derepression.(a-c) Volcano plots shown for ccr4a-1 ddm1 (a), ccr4b-1 ddm1 (b), and caf1a-1 caf1b-3 ddm1 (c) in comparison with the ddm1 single mutant of Arabidopsis thaliana.Mutations for DDM1 were generated de novo by CRISPR-Cas9 and two independent ddm1 mutant lines were used.Differential expression was defined by the log 2 -fold change > 1 or < À1 and FDR values < 0.05.Grey dots and red triangles represent genes and transposons, respectively.Numbers indicate differentially expressed genes and transposons, and up-or downregulation was expressed by arrows.(d) Overlap of transposons upregulated by the mutations of CCR4a, CCR4b, and both CAF1a and CAF1b.(e) Genome browser snapshots for representative transposon loci showing the increased RNA levels in the mRNA deadenylase mutants.Numbers in parentheses indicate read coverage and two independent ddm1 mutant lines are displayed in separate tracks.

Fig. 2
Fig. 2 CCR4a regulates distinct set of transposons from those controlled by RNA Polymerase 6 (RDR6).(a) Overlap of transposons regulated by RDR6 and CCR4a.RDR6-regulated transposons were retrieved from the previous study (Kim et al., 2021) and identified by the reduced 21/22-nt siRNA levels in the rdr6 ddm1 double mutant as compared to the ddm1 single mutant of Arabidopsis thaliana.CCR4a-regulated transposons are those that are upregulated in ccr4a-1 ddm1 compared with ddm1 with the log 2 -fold change > 1 and FDR values < 0.05.(b) Fraction of transposon families in all annotated transposons, derepressed in ddm1, regulated by RDR6, and upregulated in the ccr4a mutant.Reactivated transposons in the ddm1 mutant were identified from a public dataset (GSE52952) by filtering those with the log 2 -fold change > 1 and FDR < 0.05.(c) Levels of 21/22-and 24-nt sRNAs in the transposons regulated by RDR6 and CCR4a.White circle, median; black rectangle, upper and lower quartile.The sequencing datasets were obtained from GSE52952.(d) Chromosomal distribution of RDR6-and CCR4a-regulated transposons.Numbers of transposable elements (TEs) in 500-kb overlapping windows sliding in steps of 100 kb are shown.Regions overrepresented in the CCR4a-regulated transposons are marked by asterisks, and representative transposon families corresponding to the region are also indicated.Pericentromeric regions are expressed as grey boxes.

Fig. 3
Fig. 3 Longer transposon RNA tails are associated with increased expression.(a-d) mRNA tail lengths of all transcripts (a, b) and transposon RNAs (c, d) in ddm1-L2 and ccr4a-1 ddm1-L2 of Arabidopsis thaliana, shown as heatmap (a, c; color key shows density of distribution) and density plot (b, d).Poly(A) length was measured by Oxford Nanopore direct RNA sequencing.P-value was obtained by the one-sided Wilcoxon rank sum test.(e, f) A CACTA-like transposable element (TE) exhibiting higher RNA levels (e) and longer tail length (f) in ccr4a-1 ddm1-L2 double mutant.In (e), each line represents individual transcript detected by Oxford Nanopore direct RNA sequencing, and poly(A) tail is shown in red line.In (f), mRNA tail lengths of the CACTA-like element in ddm1-L2 and ccr4a-1 ddm1-L2 are compared.White circle, median; black rectangle, upper and lower quartile.P-value was obtained by the one-sided Wilcoxon rank sum test.(g) Fold changes of CCR4a-regulated transposons (n = 48, log 2 -fold change in ddm1-L2 vs ccr4a-1 ddm1-L2) compared with randomly chosen transposons (n = 48).CCR4a-regulated transposons are those with longer tails by at least 10 nt in the ccr4a-1 ddm1-L2 mutant.P-value was obtained by the one-sided Wilcoxon rank sum test.(h) Poly(A) tail length difference in transcripts stabilized by the loss of CCR4a in ddm1-L2 and ccr4a-1 ddm1-L2.mRNA half-lives were determined for ddm1-L2 and ccr4a-1 ddm1-L2 by the transcription arrest assay followed by RNA-Seq.Transcripts with longer half-lives in ccr4a-1 ddm1-L2 by at least 0.5 h were selected (n = 97) and compared against randomly chosen transcripts (n = 100).Tail length difference was calculated by subtracting the tail lengths in ddm1-L2 from those in ccr4a-1 ddm1-L2.P-value was obtained by the one-sided Wilcoxon rank sum test.

Fig. 4
Fig. 4 CCR4a suppresses transposon mobilization.(a) Number of new insertions of transposable elements (TEs) detected in ddm1, ccr4a-1 ddm1, ccr4b-1 ddm1, and caf1a-1 caf1b-3 ddm1 of Arabidopsis thaliana.Ten individual plants were randomly chosen, and whole-genome resequencing was performed for each individual plant independently.Data are presented in mean AE SD.P-value was obtained by the one-sided Student's t-test.(b) Percentage of TE families that were detected for neo-insertions.(c) Droplet digital PCR experiment determining the copy number of Evade retroelement in ddm1 and ccr4a-1 ddm1.Plants were randomly chosen from a pool of selfed plants of an identical genotype and extracted for DNA individually.The experiment was performed at T4 and T5 generations of the DDM1-targeting CRISPR-Cas9 transformation.Data are presented in mean AE SD.

Fig
Fig. S4 TE fractions in euchromatic and heterochromatin regions.
Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors.Any queries (other than missing material) should be directed to the New Phytologist Central Office.New Phytologist (2023) www.newphytologist.comÓ 2023 The Authors New Phytologist Ó 2023 New Phytologist Foundation ResearchNew Phytologist10 14698137, 0, Downloaded from https://nph.onlinelibrary.wiley.com/doi/10.1111/nph.19435by Test, Wiley Online Library on [29/11/2023].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License