The CCR4‐NOT Complex Component NOT1 Regulates RNA‐Directed DNA

The Plant Journal (2020) doi: 10.1111/tpj.14818 The CCR4-NOT complex component NOT1 regulates RNA-directed DNA methylation and transcriptional silencing by facilitating Pol IV-dependent siRNA production

Hao-Ran Zhou1, Rong-Nan Lin1, Huan-Wei Huang1, Lin Li1, Tao Cai1, Jian-Kang Zhu2, She Chen1 and Xin-Jian He1,3,* 1National Institute of Biological Sciences, Beijing 102206, China, 2Shanghai Center for Plant Stress Biology and Center of Excellence for Molecular Plant Sciences, Chinese Academy of Sciences, Shanghai 201602, China, and 3Tsinghua Institute of Multidisciplinary Biomedical Research, Tsinghua University, Beijing, China

Received 3 March 2020; revised 29 April 2020; accepted 7 May 2020. *For correspondence (e-mail [email protected]).

SUMMARY Small interfering RNAs (siRNAs) are responsible for establishing and maintaining DNA methylation through the RNA-directed DNA methylation (RdDM) pathway in plants. Although siRNA biogenesis is well known, it is relatively unclear about how the process is regulated. By a forward genetic screen in Arabidopsis thaliana, we identiﬁed a mutant defective in NOT1 and demonstrated that NOT1 is required for transcriptional silencing at RdDM target genomic loci. We demonstrated that NOT1 is required for Pol IV-dependent siRNA accumulation and DNA methylation at a subset of RdDM target genomic loci. Furthermore, we revealed that NOT1 is a constituent of a multi-subunit CCR4-NOT deadenylase complex by immunoprecipitation combined with mass spectrometry and demonstrated that the CCR4-NOT components can function as a whole to mediate chromatin silencing. Therefore, our work establishes that the CCR4-NOT complex regulates the biogenesis of Pol IV-dependent siRNAs, and hence facilitates DNA methylation and transcriptional silencing in Arabidopsis.

Keywords: NOT1, CCR4, pol IV, RNA, DNA methylation, transcriptional silencing.

coding RNAs, which are then converted to double-stranded INTRODUCTION RNAs (dsRNAs) by RNA-DIRECTED RNA POLYMERASE 2 As an important epigenetic mark, DNA methylation at cyto- (RDR2) (Haag et al., 2012). Then, these dsRNAs are cleaved sine is required for transposon silencing, genome stability into 24-nucleotide (24-nt) small interfering RNAs (siRNAs) and regulation of gene expression in eukaryotes (He et al., by DICER LIKE 3 (DCL3) and loaded on to ARGONAUTE 4 2011; Borges and Martienssen, 2013). In plants, DNA (AGO4) (Xie et al., 2004; Pontes et al., 2006; Ye et al., 2012). methylation occurs in the symmetric CG and CHG sites It is believed that the interactions among AGO4-siRNA and the asymmetric CHH sites, where H represents A, T or complex, Pol V transcripts and KTF1 recruit the DNA C (Law and Jacobsen, 2010; Matzke and Mosher, 2014; methyltransferase DRM2 to the target loci, by which DNA Zhang et al., 2018). CG, CHG and CHH methylation are cat- methylation is ﬁnally established (Bohmdorfer et al., 2014; alyzed and maintained by METHYLTRANSFERASE1 Zhong et al., 2014). (MET1), CHROMOMETHYLASE3 (CMT3) and CMT2, respec- Given that Pol IV-dependent siRNAs are responsible for tively (Stroud et al., 2013; Zemach et al., 2013; Du et al., the establishment of DNA methylation, investigation of 2015). The initiation and maintenance of DNA methylation whether and how the production of Pol IV-dependent siR- at CHH sites and to a smaller degree at CG and CHG sites NAs is regulated could contribute towards understanding is also mediated via the RNA-directed DNA methylation the mechanism of the establishment of DNA methylation. (RdDM) pathway (Law and Jacobsen, 2010; He et al., 2014; SHH1/DTF1 interacts with Pol IV and is responsible for the Matzke and Mosher, 2014), in which two plant-speciﬁc recruitment of Pol IV to genomic loci with histone H3K9 RNA polymerases, Pol IV and Pol V, are involved (Herr methylation; the production of Pol IV-dependent siRNAs is et al., 2005; Zhang et al., 2007; Wierzbicki et al., 2008; Haag strikingly reduced in the shh1/dtf1 mutant (Law et al., 2013; and Pikaard, 2011). Pol IV generates single-stranded non- Zhang et al., 2013b). The chromatin-remodeling regulators

CLSY1-4 interact with Pol IV and facilitate siRNA produc- silencing (Braun et al., 2011; Mathys et al., 2014). For tion and DNA methylation in a locus-specific manner (Yang instance, GW182, a component of the miRNA-induced et al., 2018; Zhou et al., 2018). Reduction of Pol IV-depen- silencing complex, interacts with Negative on TATA-less 1 dent siRNAs was also observed in the Pol V mutant nrpe1 (NOT1) and functions to recruit the CCR4-NOT complex to (Mosher et al., 2008), which is indirectly attributed to a its target genomic loci in animal cells (Braun et al., 2011). self-reinforcing loop between Pol IV-dependent siRNA pro- In Arabidopsis, NOT2 was shown to regulate the produc- duction and DNA methylation in the RdDM pathway (Law tion of primary miRNA transcripts and the location of et al., 2013; Zhang et al., 2013b; Johnson et al., 2014; Liu DCL1, thereby facilitating miRNA-induced silencing (Wang et al., 2014). RNA processing factors, including some pre- et al., 2013). These studies reveal the function of the mRNA splicing factors and the exosome subunit RRP6L1, CCR4-NOT complex in miRNA-induced silencing at the were reported to regulate siRNA production, Pol V tran- post-transcriptional level. However, it is unknown whether scription, DNA methylation and transcriptional silencing at the CCR4-NOT complex affects chromatin modification RdDM target loci in Arabidopsis (Ausin et al., 2012; Dou and transcriptional silencing. et al., 2013; Zhang et al., 2013a; Zhang et al., 2014; Ye Here, we identified NOT1 involved in chromatin silenc- et al., 2016). Involvement of pre-mRNA splicing factors and ing in Arabidopsis by a forward genetic screen. Mutation exosome regulators in siRNA-mediated chromatin silenc- of NOT1 released the silencing state of a transgene and ing was also observed in other eukaryotes (Bayne et al., several endogenous RdDM target loci. Pol IV-dependent 2008; Tabach et al., 2013; Yamanaka et al., 2013), suggest- 24-nt siRNA production and DNA methylation were ing that different eukaryotic organisms regulate siRNA reduced in the not1 mutant, suggesting that NOT1 regu- accumulation and chromatin silencing through conserved lates chromatin silencing through RNA-directed DNA mechanisms. However, it is unknown whether there are methylation. Moreover, components of the Arabidopsis any other conserved mechanisms involved in the regula- CCR4-NOT complex were deciphered in our study. We tion of siRNA-mediated chromatin silencing. demonstrate that components of the CCR4-NOT complex The CCR4-NOT complex is a conserved multi-subunit function as a whole to mediate transcriptional silencing. protein complex in yeast and animals (Nasertorabi et al., 2011). Not1 is the largest subunit in the CCR4-NOT com- RESULTS plex and integrates the other components of the complex Identification of not1 as a mutant with a released as a scaffold protein (Collart and Panasenko, 2012). The transcriptional silencing phenotype CCR4-NOT complex is involved in mRNA deadenylation as it contains the deadenylases Ccr4 and Caf1 (Collart and We performed a genetic screen to discover new compo- Panasenko, 2012; Wahle and Winkler, 2013). Deadenyla- nents taking part in chromatin silencing. In this genetic tion of mRNAs is responsible for initiation of mRNA decay screening system, the wild-type plants (C24 ecotype) har- (van Hoof and Parker, 2002; Chen and Shyu, 2011). Addi- bor a luciferase transgene driven by the RD29A promoter tionally, the CCR4-NOT complex regulates cellular pro- (RD29A-LUC) and an NPTII transgene driven by the 35S cesses such as transcriptional initiation and elongation, promoter (35S-NPTII) that enable the plants, respectively, RNA export, translation repression and protein ubiquitina- to emit strong luminescence in stress conditions and to tion (Collart and Panasenko, 2012; Inada and Makino, grow normally on medium with kanamycin (Figure 1a). 2014). Components of the CCR4-NOT complex are also Both RD29A-LUC and 35S-NPTII were silenced in a null known to participate in microRNA (miRNA)-induced gene mutant of ROS1, which is a 5-mC glycosylase initiating

(a) (b) Figure 1. Silencing of RD29A-LUC transgene and Luminescence endogenous loci is affected by not1. (a) Expression of RD29A-LUC and 35S-NPTII transgenes in wild-type (WT), ros1, ros1/#88 and ros1nr- AtGP1 pd1 was detected by luminescence imaging and by growing them on MS medium with 150 mg/L kana- solo LTR mycin, respectively. (b) Transcript levels of endogenous loci in WT and SDC the indicated mutants were determined by semi- MS + Kanamycin quantitative reverse transcription–polymerase chain ERT7 reaction. ACTIN7 gene was ampliﬁed as an internal control. WT ros1 ERT14 ACTIN7 ros1/ ros1/ #88 nrpd1 no RT

DNA demethylation (Figure 1a) (Gong et al., 2002). In responsible for the defect in the silencing of both the screening for ros1 suppressors, we have previously identi- RD29A-LUC transgene and the endogenous RdDM target fied several mutants involved in DNA methylation and loci. transcriptional silencing (Dataset S1) (Liu et al., 2011; Dou NOT1 is involved in silencing at the transcriptional level et al., 2013; Zhou et al., 2013; Zhang et al., 2013a; Han et al., 2014). Here, we identified a previously uncharacter- In yeast and animals, the CCR4-NOT complex was reported ized ros1 suppressor mutant, #88, in which the silencing of to act at different cellular processes to regulate gene RD29A-LUC but not 35S-NPTII is released (Figure 1a). The expression, including transcription, mRNA degradation different effects on the silencing of RD29A-LUC and 35S- and protein quality control (Collart and Panasenko, 2012; NPTII were also detected in the mutants of RdDM compo- Miller and Reese, 2012). As we know, transcription occurs nents (He et al., 2009a; Liu et al., 2011), indicating that the in the nucleus while mRNA degradation and protein qual- #88 mutant affects transcriptional silencing in a similar ity control occur in the cytoplasm, which means that the mechanism with mutants defective in canonical RdDM CCR4-NOT complex exists in both the nucleus and cyto- components. plasm. To understand the function of the Arabidopsis We further determined whether the #88 mutation NOT1, we need to figure out its cellular localization. We changes the silencing state of endogenous RdDM target separated the nuclear and cytoplasmic fractions of the genomic loci. As reported by previous studies, RdDM tar- NOT1-FLAG transgenic plants and control plants (without get genomic loci, including AtGP1, solo LTR, SDC, ERT7 the NOT1-FLAG transgene) and then examined the NOT1- and ERT14, were silenced in wild-type plants but this FLAG fusion protein in these fractions by immunoblotting silencing was released by mutations of canonical RdDM (Figure 2a). The cytosolic protein UDP-glucose pyrophos- components (Henderson and Jacobsen, 2008; He et al., phorylase (UGPase) and the nuclear protein histone H3 2009b; Blevins et al., 2014). We showed that the expression were also examined in the same fractions (Figure 2a). As levels of these RdDM target loci were increased by the #88 expected, we found UGPase in total proteins and the cyto- mutation as well as by the RdDM mutations, ago4 and plasmic fraction but not in the nuclear fraction and nrpd1 (Figure 1b and Figure S1), suggesting that the detected the histone H3 in total proteins and the nuclear #88 mutation influences the silencing state of these RdDM fraction but not in the cytoplasmic fraction (Figure 2a), targets. which indicated that the nuclear and cytoplasmic fractions To identify the mutated gene in the #88 mutant, the were well separated. The NOT1-FLAG fusion protein was mutant was crossed to the ros1 mutant in the Columbia detected both in the nuclear and cytoplasmic fractions,

(Col-0) background and the F2 population emitting strong indicating that NOT1 located in both the nucleus and cyto- luminescence was selected for map-based cloning. With plasm in Arabidopsis (Figure 2a). simple sequence-length polymorphism markers, we deter- To determine whether the effects of the not1 mutation mined the mutated site in a region at approximately on gene expression occurred at the transcriptional level or 110 kb on chromosome 1 (Figure S2a). A G to A mutation the post-transcriptional level or both, several experiments was found in this region by whole-genome sequencing, were performed. First, we extracted RNA from the nuclear which disrupted the splicing donor site of the 25th intron fraction of the wild-type, not1 and ago4 mutant plants and of AT1G02080 (Figure S2b). Our reverse transcription–poly- then measured the transcript levels of AtGP1, solo LTR, merase chain reaction (RT-PCR) experiment followed by SDC, ETR7 and ERT14 in the nuclear fraction. As shown in sequencing indicated that the mutation of the splicing Figure 2b, the transcripts of these loci increased in the donor site disrupts the normal transcript of this gene (Fig- nucleus once NOT1 and AGO4 mutated (Figure 2b). This ure S3a,b). This gene encodes a putative transcription reg- result indicated that the accumulation of these transcripts ulator whose protein sequence shows high similarity with occurred before being transported to the cytoplasm, sug- NOT1 in yeast as well as in flies and humans (Figure S4) gesting that the effect of the not1 mutation on the silencing and is hereafter referred to as NOT1. of these loci is likely caused by the defect of its function in To confirm that the released transcriptional silencing the nucleus. Secondly, the premature mRNA (pre-mRNA) phenotype of the #88 mutant is caused by the NOT1 muta- of the intron-containing not1-upregulated gene SDC was tion, a complementation assay was conducted by introduc- detected in the not1 mutant and the wild type, revealing ing a native promoter-driven NOT1 into the #88 mutant. As that the pre-mRNA of SDC significantly increased in not1 indicated by luminescence imaging (Figure S5a), the NOT1 compared with wild type (Figure 2c). Considering that pre- transgene restored the silencing of RD29A-LUC in the #88 mRNA splicing is mostly co-transcriptional (Neugebauer, mutant. Besides, the silencing of the endogenous RdDM 2002; Reddy et al., 2013; Naftelberg et al., 2015), we believe target loci AtGP1, solo LTR and ERT7 was also restored that NOT1 represses gene expression at the transcriptional when the NOT1 transgene was introduced into the #88 level. Thirdly, we treated the not1 and wild-type plants mutant (Figure S5b). Thus, the not1 mutation is with a-amanitin, a Pol II inhibitor, and observed that the

(a) (b) (c) (d)

Figure 2. NOT1 mediates transcript silencing at the transcriptional level. (a) Localization of NOT1 in different protein extracts from Arabidopsis seedlings. CP, cytoplasmic protein; NP, nuclear protein; TP, total protein. (b) Relative expression levels of the identiﬁed not1-upregulated loci in the nucleus were determined by reverse transcription–polymerase chain reaction. Expres- sion level of the actin gene ACT7 is shown as an internal control. WT, wild type. (c) Precursor mRNA (pre-mRNA) of SDC was detected by quantitative polymerase chain reaction in not1 and WT using primers ﬂanking the splicing site. (d) Decay rate of the SDC transcript in the WT and not1 mutant. a-amanitin treatment was performed for 0, 3, 6, 9, 12 and 18 h. Expression level of SDC transcript was normalized to ACT7.

SDC transcript level was quickly declined in both not1 and CHH DMRs caused by not1 overlapped with that caused by wild type, and the upregulation caused by not1 was almost nrpd1, respectively (Figure 3a and Figure S6), implying blocked after treatment with a-amanitin in hours (Fig- that NOT1 regulates DNA methylation mainly at CHH sites ure 2d), implying that the deficiency of NOT1 affects the of RdDM target loci. production of the SDC pre-mRNA rather than the degrada- To understand the function of NOT1 in CHH methylation tion of its mRNA. Together, these results support the better, we divided the nrpd1 CHH hypo-DMRs into two notion that NOT1 affects silencing at the transcriptional groups (not1-dependent DMRs and the remaining nrpd1- level in the nucleus. dependent DMRs) and evaluated their CHH methylation levels in nrpd1, not1 and wild type (Figure 3a,b). The CHH NOT1 is necessary for CHH methylation at a subset of methylation levels of the not1-dependent DMRs were sig- RdDM target loci nificantly decreased in both nrpd1 and not1 in comparison To figure out whether NOT1 functions in DNA methylation, with wild type (Figure 3b), indicating that NOT1 is indis- whole-genome DNA methylation levels of not1 and wild- pensable for the CHH methylation of these RdDM target type plants were detected by bisulfite sequencing. Given loci. The CHH methylation levels of the remaining nrpd1- that NRPD1, the largest subunit of Pol IV, plays a critical dependent DMRs were also reduced in not1 relative to wild role in the canonical RdDM pathway (Haag and Pikaard, type, even though the reduction was markedly weaker than 2011; Matzke and Mosher, 2014), we also analyzed the that in nrpd1 (Figure 3b), which implies that NOT1 has a effect of nrpd1 on whole-genome DNA methylation as a broader effect on the CHH methylation of RdDM target loci. control. Differentially methylation regions (DMRs) were Furthermore, the not1 CHH hypo-DMRs located in pericen- determined in both not1 and nrpd1 compared with wild- tromeric regions as well as in the chromosome arms; the type plants (Figure 3a and Figure S6). Our results indicated distribution pattern of the not1 CHH hypo-DMRs was simi- that 1308 CG DMRs, 2236 CHG DMRs and 6007 CHH DMRs lar to that of the nrpd1 CHH hypo-DMRs (Figure 3c and Fig- were significantly hypomethylated in nrpd1 (Figure 3a and ure S7). Moreover, both the not1 and nrpd1 hypo-DMRs Figure S6), which reflects previous reports demonstrating are primarily composed of transposons and gene promot- that RdDM components are required for all three types of ers (Figure 3d). Considering that the RdDM pathway is DNA methylation, particularly the CHH methylation (Stroud involved in establishment of DNA methylation, particularly et al., 2013; Zemach et al., 2013). In the not1 mutant, the at CHH sites of short transposable elements and gene pro- numbers of significantly hypomethylated CG, CHG and moters (Stroud et al., 2013), we suggest that NOT1 simi- CHH were 628, 71 and 1490, respectively (Figure 3a and larly regulates CHH methylation with the RdDM pathway. Figure S6), which suggests that NOT1 is more inclined to NOT1 reinforces the production of Pol IV-dependent affect CHH methylation but has little effect on CHG methy- siRNAs lation. Furthermore, approximately 35% (218 of 628) of hypomethylated CG DMRs, a half (37 of 71) of hypomethy- Considering that NOT1 is involved in DNA methylation and lated CHG DMRs and >75% (1142 of 1490) hypomethylated transcriptional silencing of RdDM target loci, we

(a) nrpd1 CHH hypo-DMRs (c) CHH hypo-DMRs

(6007, >10%) nrpd1

not1 CHH hypo-DMRs (1490, >10%) 348 1142 4865 not1

No. of DMRs No. of DMRs per 100 kb Chr.1

(b) not1-dependent Remaining nrpd1-dependent (d) Gene Intergenic region DMRs (1142) DMRs (4865) Promoter Transposon 100% 0.6 ** **

0.4 80% 0.4 60% 0.2 mCHH 0.2 40% Percentage 0 0 20% 1 2 1 2 1 2 1 2 1 2 1 2 0% nrpd1 not1 nrpd1 not1 WT nrpd1 not1 All nrpd1 not1 WT regions CHH hypo-DMRs

Figure 3. Mutation of NOT1 causes reduction of CHH methylation. (a) Venn diagram shows the overlap between not1 CHH hypo-differentially methylation regions (DMRs) and nrpd1 CHH hypo-DMRs. Significance of overlap is determined by hypergeometric test. (b) Box plots show the CHH methylation (mCHH) levels at two indicated groups of nrpd1 CHH hypo-DMRs in not1, nrpd1 and wild type. **P < 0.01 as determined by Student’s t-test with two tails. WT, wild type. (c) Distributions of nrpd1 CHH hypo-DMRs (red columns) and not1 CHH hypo-DMRs (green columns) on chromosome 1 (Chr.1). Scales on the left indicate the numbers of DMRs per 100-kb region. Black bar and gray bars indicate the pericentromeric region and chromosome arms, respectively. (d) Percentage of protein-coding gene, intergenic region, promoter and transposon on all DMRs, nrpd1 CHH hypo-DMRs and not1 CHH hypo-DMRs, respectively. investigated whether it affects the production of Pol IV-de- NRPD1-dependent 24-nt siRNA regions located at chromo- pendent siRNAs. Small RNAs were detected by deep some arms (Figure 4d and Figure S9), suggesting that sequencing in wild-type, not1 and nrpd1 mutant plants. In NOT1 reinforces the production of Pol IV-dependent siR- agreement with previous reports (Zhang et al., 2007; NAs through a mechanism that is different from Pol V. Mosher et al., 2008), the production of 24-nt siRNAs was To establish the link between the decrease of DNA almost abolished in the Pol IV mutant, nrpd1 (Figure 4a,b). methylation and the reduction of 24-nt siRNA production The Pol IV-dependent siRNA production was significantly in the not1 mutant, we analyzed the abundance of 24-nt dropped in the not1 mutant (Figure 4a,b), suggesting that siRNAs at the not1 hypo-DMRs and the remaining nrpd1 NOT1 reinforces the production of Pol IV-dependent hypo-DMRs as mentioned above in the wild-type, nrpd1 siRNAs. Furthermore, the distribution of Pol IV-dependent and not1 mutants. In these regions, the enrichment of 24- 24-nt siRNA was evaluated over the five Arabidopsis chro- nt siRNAs was markedly low in nrpd1 and to a lesser mosomes in wild type, not1 and nrpd1. The 24-nt siRNAs extent in not1 compared with the wild type, which is con- were highly enriched particularly at pericentromeric sistent with the different effects of nrpd1 and not1 on CHH regions in wild-type plants, which were almost eliminated methylation (Figures 3b and 4e). Besides, the not1 hypo- by nrpd1 and were reduced moderately but clearly by not1 DMRs showed a slightly lower 24-nt siRNA abundance (Figure 4b and Figure S8). We identified 6016 NRPD1-de- than the remaining nrpd1-dependent DMRs in the not1 pendent siRNA regions where 24-nt siRNAs remarkably mutant, which is also consistent with the lower CHH decreased (>4-fold reduction) in the nrpd1 mutant (Fig- methylation levels of these loci in the not1 mutant (Fig- ure 4c). In addition, we identified 1862 NOT1-dependent ures 3b and 4e). Moreover, the 24-nt siRNA accumulation siRNA regions where 24-nt siRNAs were at least moder- and the CHH methylation at three RdDM target loci, includ- ately reduced (>2-fold reduction) in the not1 mutant (Fig- ing RD29A transgene, ERT14, and SDC (Henderson and ure 4c). Over 93% (1739 of 1862) of the NOT1-dependent Jacobsen, 2008; He et al., 2009b; Blevins et al., 2014), were siRNA regions were overlapped with the NRPD1-depen- visualized by snapshots of genome browsers. As expected, dent siRNA regions (Figure 4c), suggesting that NOT1 is the 24-nt siRNAs at these loci were abolished and the CHH involved in the production of a subset of Pol IV-dependent methylation was reduced correspondingly in the nrpd1 siRNAs. As shown by a previous study (Mosher et al., mutant (Figure 4f). In accordance with the genome-wide 2008), Pol V reinforces the production of 24-nt siRNAs par- results, the not1 mutation led to the reduction of both the ticularly at chromosome arms. Our results showed that 24-nt siRNA accumulation and the CHH methylation at comparable ratios (approximately 50%) of NOT1- and these loci even though the impact of not1 is generally

(a) (b) (c) Small RNA distribution Chr.1

nrpd1 123 4277 1739 not1 siRNA (RPKM)

Normalized 24-ntNormalized Ratio of small RNAs Ratio of small

not1-dependent Remaining nrpd1- (d) Chromosome arm (e) DMRs (1142) dependent DMRs (4865) Pericentromere 100% ** ** 80% 60% 40% Percentage

20% nt SiRNA (RPM) - 0 5 10 15 20 20 15 10 5 0 0 5 10 15 20 20 15 10 5 0 0% 24 nrpd1- not1 1 2 1 2 1 2 nrpd1-depdep not1-dep-dep 1 2 1 2 1 2 WT nrpd1 not1 WT nrpd1 not1 (f) RD29A ERT14 SDC

rep1 [0-20] [0-20] [0-20] WT rep2 [0-20] [0-20] [0-20]

rep1 [0-20] [0-20] [0-20] not1 rep2 [0-20] [0-20] [0-20] rep1 [0-20] [0-20] [0-20] nrpd1 24-nt siRNAs (RPM) (RPM) 24-nt siRNAs rep2 [0-20] [0-20] [0-20]

rep1 [0-1] [0-1] [0-1] WT rep2 [0-1] [0-1] [0-1] rep1 not1 [0-1] [0-1] [0-1] rep2 [0-1] [0-1] [0-1] rep1 nrpd1 [0-1] [0-1] [0-1]

CHH methylation rep2 [0-1] [0-1] [0-1]

Figure 4. Effects of not1 and nrpd1 on small RNA accumulation and DNA methylation. (a) Percentages of 18–30 bp small RNAs in not1, nrpd1 and wild type (WT) were proﬁled by length based on two replicates of small RNA deep sequencing. (b) Distribution of 24-nt small interfering RNAs (siRNAs) throughout the chromosome 1 (Chr.1) in not1, nrpd1 and WT according to the two replicates. (c) Venn diagram showing the relationship between nrpd1-dependent 24-nt siRNA regions and not1-dependent 24-nt siRNA regions. Red circle and orange circle represent regions in which 24-nt siRNAs were at least four-fold less in nrpd1 and at least two-fold less in not1 than that in WT, respectively. Signiﬁcance of overlap is determined by hypergeometric test. (d) Showing the proportions of nrpd1-dependent (nrpd1-dep) and not1-dependent (not1-dep) 24-nt siRNA regions on chromosome arms and pericentromeric regions. (e) Box plots show the 24-nt siRNA abundance in reads per million (RPM) at indicated loci in WT, nrpd1 and not1. 1 and 2 represent two replicates of small RNA deep sequencing results. **P < 0.01 as determined by Student’s t-test with two tails. DMRs, differentially methylation regions. (f) Browser views of the 24-nt siRNA occupancies and the CHH methylation levels at RD29A (left), ETR14 (middle) and SDC (right) in two replicates (rep1 and rep2) of WT (blue bar), not1 (green bar) and nrpd1 (red bar), respectively. Vertical bars indicate 24-nt siRNA in RPM (upper panel) or the percentage of CHH methylation (lower panel).

© 2020 Society for Experimental Biology and John Wiley & Sons Ltd, The Plant Journal, (2020), doi: 10.1111/tpj.14818 NOT1 regulates RNA-directed DNA methylation 7 weaker than that of nrpd1 (Figure 4f). Therefore, NOT1 is were co-purified with NOT1-FLAG (Table 1). Although 11 responsible for reinforcing the Pol IV-dependent siRNA CAF1s were previously identified in Arabidopsis based on production and thereby facilitating DNA methylation and sequence similarity (Walley et al., 2010), only three CAF1s transcriptional silencing at the RdDM target loci. (CAF1h, CAF1i and CAF1k) that belong to clade III CAF1s were co-purified with the NOT1-FLAG protein in our study Composition of the CCR4-NOT complex in Arabidopsis (Table 1). Coincidentally, the same CAF1 isoforms were Recently, several components of the CCR4-NOT complex identified by the previous FLAG-tagged CCR4b co-immuno- have been identified in the Arabidopsis protoplast by precipitation assay (Arae et al., 2019). This may result from immunoprecipitation combined with mass spectrometry the different expression patterns or levels of these CAF1 using a transiently overexpressed FLAG-tagged CCR4b genes (Walley et al., 2010; Arae et al., 2019). In Arabidop- (Arae et al., 2019). To elucidate the composition of the Ara- sis, there are three CAF40/CNOT9-like proteins AT3G20800, bidopsis CCR4-NOT complex in plants, we used the NOT1- AT5G12980 and AT2G32550, which we termed NOT9a, FLAG transgenic plants for NOT1 affinity purification and NOT9b and NOT9c, respectively (Table 1 and Figure S10b). identified the co-purified proteins with mass spectrometry. NOT9a and NOT9b, which share higher similarity with Proteins corresponding to core components of the yeast or each other and with the yeast Caf40 but not with NOT9c human CCR4-NOT complex were co-purified with the (Figure S10b), were present in the co-purified proteins NOT1-FLAG fusion protein (Table 1 and Dataset S2). (Table 1). Among the co-purified proteins, AT1G07705 and The tetratricopeptide repeat-containing protein AT5G59710, two orthologs of NOT2 in other eukaryotes, AT5G35430 and the unknown protein AT5G18420 in the were referred to as NOT2a and NOT2b in a previous study NOT1-FLAG co-purified proteins (Table 1) are orthologs of (Wang et al., 2013). AT5G18230, a homolog of the yeast CNOT10 and CNOT11 subunits of the human CCR4-NOT Not3 and Not5, was hereafter referred to as NOT3. complex, respectively (Table 1). Orthologs of CNOT10 and In Arabidopsis, there are three closely related RNA-bind- CNOT11 were not found in yeast, suggesting that they may ing (RRM/RBD/RNP motifs) proteins AT2G28540, specifically function in higher eukaryotes as subunits of AT3G45630 and AT5G60170 showing homologous to the CCR4-NOT complex. Besides, three DEAD-box ATP-de- NOT4 of other eukaryotes, which we referred to as NOT4a, pendent RNA helicases, AT4G00660, AT3G61240 and NOT4b and NOT4c, respectively (Figure S10a). However, AT2G45810, were also identified (Table 1 and Dataset S2). only NOT4a was identified in our affinity purification of These RNA helicases correspond to DHH1/DXX6 subunits NOT1-FLAG (Table 1), which may hint that the association in other eukaryotes and thus were named DHH1a, DHH1b of NOT4a with the Arabidopsis CCR4-NOT complex is and DHH1c, respectively (Table 1 and Figure S10c). more stable than NOT4b and NOT4c. CCR4a (AT3G58560) To validate the composition of the Arabidopsis CCR4- and CCR4b (AT5G58580), two homologs of the yeast CCR4, NOT complex as determined by affinity purification of

Table 1 Identiﬁcation of proteins co-puriﬁed with NOT1-FLAG and NOT3-FLAG by IP-MS

NOT1-FLAG NOT3-FLAG Ortholog(s)

Mascot Matched Matched Mascot Matched Matched Annotation Protein score queries peptides score queries peptides Yeast Human

AT1G02080 NOT1 16169 364 119 8921 232 111 Not1 CNOT1 AT1G07705 NOT2a 660 16 9 925 21 12 Not2 CNOT2 AT5G59710 NOT2b 594 17 11 1087 23 14 AT5G18230 NOT3 2616 58 33 3548 95 44 Not3; Not5 CNOT3 AT2G28540 NOT4a 107 2 2 89 3 3 Not4 CNOT4 AT3G20800 NOT9a 797 19 12 752 20 11 Caf40 CAF40/CNOT9 AT5G12980 NOT9b 470 11 8 344 9 8 AT5G35430 NOT10 3500 76 38 2669 61 32 — CNOT10 AT5G18420 NOT11 2342 49 25 1990 40 22 — CNOT11 AT3G58560 CCR4a 924 25 16 426 10 10 Ccr4 CNOT6; CNOT6L AT3G58580 CCR4b 695 19 12 387 10 10 AT1G15920 CAF1h 232 5 4 239 6 5 Caf1/Pop2 CNOT7; CNOT8 AT2G32070 CAF1k 167 5 5 52 1 1 AT5G10960 CAF1i 65 1 1 116 4 4 AT4G00660 DHH1a 880 19 14 222 7 7 Dhh1 DDX6 AT3G61240 DHH1b 802 18 11 289 11 10 AT2G45810 DHH1c 867 18 12 286 8 7

NOT1-FLAG, we performed another affinity purification fol- transcriptional silencing of RdDM target loci (Figure 5a–f). lowed by mass spectrometry using transgenic plants har- The transcript levels of SDC, AT2TE01740, AT2TE18240 boring a NOT3-FLAG fusion transgene and found that all and AT4TE09845 were previously found to be increased in the proteins co-purified with NOT1-FLAG mentioned above the mutants defective in the RdDM pathway components were also detected as NOT3-FLAG co-purified proteins (Henderson and Jacobsen, 2008; Liu et al., 2016), suggest- (Table 1 and Dataset S3). Together, these results estab- ing that they are RdDM target loci. Our RNA analyses indi- lished the composition of the Arabidopsis CCR4-NOT com- cated that the transcription levels of most of these RdDM plex. target loci were increased not only in the not1 mutant but also in all the RNAi lines tested in the experiment (Fig- CCR4-NOT complex functions in transcriptional silencing ures 1b and 5a–f), which indicates that the components of To determine whether other components of the Arabidop- CCR4-NOT complex as a whole are involved in transcrip- sis CCR4-NOT complex have a similar function with NOT1 tional silencing. in transcriptional silencing, we obtained T-DNA insertion DISCUSSION mutants of NOT2a, NOT2b, NOT3, NOT9a, NOT10 and NOT11 (Figure S11). We failed to identify homozygotes of Our study shows that NOT1 is required to produce Pol the not3, not10 and not11 mutants, which suggest that IV-dependent siRNAs at a subset of RdDM target loci. Pol these mutants are likely to be either gametophytically or IV-dependent siRNAs are predominantly located in pericen- embryonically lethal. Moreover, we found that the not2a tromeric regions (Mosher et al., 2008; Wierzbicki et al., not2b double mutant was unable to survive as previously 2012; Dou et al., 2013). In addition to Pol IV, RDR2 and described (Wang et al., 2013). Therefore, we knocked down DCL3 are required for the biogenesis of Pol IV-dependent NOT2b, NOT3, NOT9b, NOT10 and NOT11 by RNA interfer- siRNAs in the canonical RdDM pathway (Blevins et al., ence (RNAi), among which the knockdown of NOT2b and 2015; Zhai et al., 2015). However, relatively little is known NOT9b were performed in the not2a and not9a back- about how the production of Pol IV-dependent siRNAs is grounds, respectively. Our quantitative RT-PCR results regulated. In this study, we show that NOT1 engages in showed that the expression levels of NOT2b, NOT3, the production of a subset of Pol IV-dependent siRNAs, NOT9b, NOT10 and NOT11 were markedly reduced in the and NOT1-dependent siRNAs show a similar pericen- corresponding RNAi lines, suggesting that the knockdown tromeric pattern with Pol IV-dependent siRNAs. Pol V act- experiments were successful. In these RNAi lines, the ing downstream of the RdDM pathway was also reported NOT2b and NOT3 RNAi lines showed severe developmen- to enhance the biogenesis of Pol IV-dependent siRNAs at a tal defects, whereas the other RNAi lines had no visible subset of RdDM target loci (Dou et al., 2013; Law et al., developmental defects. We selected NOT2b, NOT3, NOT9b, 2013; Zhang et al., 2013b; Liu et al., 2014). However, the NOT10 and NOT11 RNAi lines to determine their effects on function of Pol V in the production of Pol IV-dependent

(a) (b) (c) Figure 5. Subunits of CCR4-NOT complex are involved in chromatin silencing. (a–f) Average transcript levels of RNA-directed DNA methylation target loci in the (a) not1, (b) not2a; NOT2b-KD, (c) not19a;NOT9b-KD, (d) NOT3-KD, (e) NOT10-KD, and (f) NOT11-KD lines and wild-type (WT) plants in three biological replicates. KD, knockdown. Transcripts levels of NOT2b, NOT9b, NOT3, NOT10 and NOT11 in the corresponding KD lines are indicated. Values are mean SD of three replicates. ns, P > 0.05; *P < 0.05; **P < 0.01 as determined by Student’s t-test.

(d) (e) (f)

© 2020 Society for Experimental Biology and John Wiley & Sons Ltd, The Plant Journal, (2020), doi: 10.1111/tpj.14818 NOT1 regulates RNA-directed DNA methylation 9 siRNAs is likely to be indirectly caused by its function in previous study showed that a weak Pol II mutant showed a DNA methylation (Law et al., 2013; Zhang et al., 2013b; reduction of 24-nt siRNAs at the RdDM target loci (Zheng Johnson et al., 2014; Liu et al., 2014). Different from the et al., 2009), supporting the notion that Pol II transcripts pericentromeric distribution patterns of Pol IV- and NOT1- may function as precursors of 24-nt siRNAs. Thus, dead- dependent siRNAs, Pol V-dependent siRNAs are abundant enylated mRNAs mediated by the CCR4-NOT complex may in both pericentromeric regions and chromosome arms not only be subjected to mRNA degradation but also be (Lee et al., 2012; Wierzbicki et al., 2012). Thus, the function channeled to 24-nt siRNA production at RdDM target loci. of NOT1 in the production of Pol IV-dependent siRNAs is Alternatively, the complex may participate in the process- unlikely to result from alteration of DNA methylation. ing of Pol IV-produced RNAs. Pol IV-produced RNAs are Therefore, NOT1 is most likely to regulate Pol IV-depen- approximately 30–40 nt as reported by several different dent siRNA production directly and thereby to affect DNA groups (Blevins et al., 2015; Yang et al., 2015; Zhai et al., methylation. 2015; Ye et al., 2016). The short Pol IV-dependent RNAs In yeast and mammals, the orthologs of NOT1 act as a were subjected to RRP6L, a 30-50 exonuclease for matura- scaffold integrating other subunits into a whole complex tion (Ye et al., 2016). However, it is unknown whether the (Collart et al., 2013; Shirai et al., 2014; Temme et al., 2014). 30–40-nt RNAs are primary RNAs. The participation of the However, little is known about the constituents of the CCR4-NOT complex in the production of Pol IV-dependent CCR4-NOT complex in plants. In this study, we comprehen- siRNAs suggests that the complex may be responsible for sively identified components of the Arabidopsis CCR4-NOT deadenylation of Pol IV-produced primary RNAs. Pol IV- complex and confirmed the composition of the complex. produced siRNA precursors were previously reported to be The Arabidopsis CCR4-NOT complex identified in this non-polyadenylated (Li et al., 2015). It is possible that Pol study possesses counterparts of NOT10 and NOT11 in the IV-dependent transcripts will be polyadenylated when they metazoans but lacks a homolog of the yeast-specific sub- are produced and these transcripts are immediately dead- unit Caf130 (Table 1), suggesting that the CCR4-NOT com- enylated by the CCR4-NOT complex followed by cleavage plex in plants is more similar to that in metazoans than in by Dicer-like proteins. Considering that the NOT2 proteins yeast. In Arabidopsis, CCR4a and CCR4b are involved in were reported to interact with the conserved Piwi/Ago/ deadenylation of granule-bound starch synthase 1 (GBSS1) Zwille domain of the Dicer-like protein DCL1 (Wang et al., transcript and thus modulate sucrose and starch metabo- 2013), we predict that the CCR4-NOT complex may also lism (Wang et al., 2013; Suzuki et al., 2015). The Arabidop- cooperate with the Dicer-like protein DCL3, thereby facili- sis NOT2a and NOT2b affect Pol II-dependent transcription tating the coupling of RNA deadenylation and cleavage. and interact with key factors in miRNA biogenesis; inacti- Our study sheds new light on the mechanism of Pol IV-de- vation of both NOT2a and NOT2b causes severe defects in pendent siRNA production in eukaryotes. Owing to the male gametophytes (Wang et al., 2013). In addition, NOT1 conservation of the CCR4-NOT complex, the involvement is essential for proper pollen development and germina- of the complex in chromatin silencing may represent a tion capacity (Motomura et al., 2020), implying that the conserved mechanism from plants to animals. Arabidopsis CCR4-NOT complex as a whole plays a role in the development of male gametophytes. The identification of the composition of the Arabidopsis CCR4-NOT complex EXPERIMENTAL PROCEDURES may expand our knowledge from the previously studied Plant materials, mutant identification and cloning components to the other components. Our study reveals the involvement of NOT1 in Pol IV-dependent siRNA pro- The wild-type and ros1 mutant plants harboring the RD29A-LUC and 35S-NPTII transgenes are in the C24 ecotype of Arabidopsis duction, DNA methylation and transcriptional silencing. thaliana. Suppressors of ros1 were screened from the ethyl Considering that several other components of the CCR4- methanesulfonate-mutagenized library in the ros1 background as NOT complex also function in transcriptional silencing, we previously described (Liu et al., 2011). Positive mutants were predict that the CCR4-NOT complex as a whole is involved selected based on luminescence and were then crossed to the in these processes. Further studies are required to clarify ros1 mutant (Salk_045303) in the Col-0 ecotype for map-based cloning. High-throughput sequencing (Illumina) was carried out to how different components of the CCR4-NOT complex determine the mutation sites in the mapped genomic interval. cooperate during the diverse biological processes in The full-length genomic sequence of NOT1 was inserted into eukaryotes. the modified pRI909 binary vector whose 30-end in frame with the The CCR4-NOT complex is engaged in deadenylation of 39 FLAG epitope and transformed into ros1not1 for complement mRNAs followed by mRNA degradation mediated by 30-50 testing and affinity purification. The full-length NOT3 sequence exonucleases (Garneau et al., 2007). In light of the contri- was cloned into the modified pCAMBIA1300 vector followed by the 39 FLAG tag sequence and transformed to the Col-0 wild type bution of the CCR4-NOT complex to the production of 24- for affinity purification. The cDNA fragments of NOT2b, NOT3, nt siRNAs, it is possible that deadenylated mRNAs will NOT9b, NOT10 and NOT11, respectively, were inserted into the function as 24-nt siRNA precursors at RdDM target loci. A pFGC5941 vector to generate RNAi transgenic lines.

The not1, ago4 and nrpd1 single mutants used in this study not1 relative to wild type were termed as not1-dependent siRNA were obtained by backcrossing the ros1not1, ros1ago4 and ros1n- regions. rpd1 to the C24 wild type. The T-DNA insertion mutants not2a (GK-104B08-012183), not3 (SALK_025624), not9a (SALK_100945), Affinity purification and mass spectrometry not10 (SALK_025832) and not11 (SALK_128382) in the Col-0 back- Three grams of flower tissues were collected and ground into ground were ordered from the Arabidopsis Biological Resource powder in liquid nitrogen. The powder was then homogenized in Center (ABRC; http://abrc.osu.edu) and the European Arabidopsis 15 ml of lysis buffer (50 mM Tris [pH 7.6], 150 mM NaCl, 5 mM Stock Centre (NASC; http://www.arabidopsis.info). MgCl2, 10% glycerol, 0.1% NP-40, 0.5 mM DTT, 1 mM phenyl- Analysis of RNA transcript levels methylsulfonyl fluoride, and one protease inhibitor cocktail tablet [Roche] per 50 ml) at 4°C. After debris removal by centrifugation, Semi-quantitative RT-PCR and quantitative RT-PCR were con- the supernatant was incubated with 100 ll ANTI-FLAG M2 Affinity ducted to determine the transcript levels of protein-coding genes Agarose Gel at 4°C for 2.5 h (A2220; Sigma). Subsequently, the and transposable elements. First, total RNA was isolated from 2- ANTI-FLAG beads were subjected to wash with 1 ml of lysis buffer week-old seedlings with TRIzol Reagent (Thermo Fisher Scientific) for 5 min at 4°C for five times. Proteins on the beads were eluted â and was subjected to RT by using ReverTra Ace quantitative PCR with 39 FLAG peptides (F4799; Sigma). Finally, the purified pro- RT Master Mix with gDNA Remover (FSQ-301; Toyobo). The semi- teins were separated in a sodium dodecyl sulfate–polyacrylamide quantitative RT-PCR results were examined on ethidium bromide- gel electrophoresis gel and detected by silver staining with the stained agarose gels, while quantitative RT-PCR was run in an ProteoSilver Silver Stain Kit (PROT-SIL1; Sigma) before mass Applied Biosystems 7500 Fast real-time PCR system with KAPA spectrometry analysis. â SYBR FAST quantitative PCR Kit Master Mix (29) Universal Mass spectrometry analyses were conducted as previously (KR0389; Kapa Biosystems). ACTIN7 was an internal control for explained (Zhang et al., 2012; Zhang et al., 2013a). Protein identifi- transcripts analysis. Primers used for amplifying the transcripts cation was performed by searching the identified peptides in the were listed in Dataset S4. International Protein Index Arabidopsis protein database on the Mascot server (Matrix Science Ltd, UK) and then calculating the Whole-genome bisulfite sequencing and data analysis mapped peptides. The whole-genome DNA methylation analysis was conducted as Nuclear-cytoplasmic fractionation and immunoblotting previously described (Han et al., 2016). In brief, genomic DNA of the 2-week-old seedlings was extracted and converted by bisulfite and The separation of the nuclear and cytoplasmic fractions was car- the converted genomic DNA was amplified and used for high- ried out as previously described (Wang et al., 2011). Two-week-old throughput sequencing with HiSeq 2000 (Illumina) in the Beijing seedlings (0.5 g) were collected and ground in liquid nitrogen. Genomics Institute. Using Bismark, unique reads from bisulfite The powder was then transferred to 4 ml lysis buffer (20 mM Tris- sequencing were mapped to the modified TAIR10 reference gen- HCl [pH 7.5], 20 mM KCl, 2 mM EDTA, 2.5 mM MgCl2, 25% glycerol, ome, which allows for one mismatch. The percentage of reads num- 250 mM sucrose, 5 mM DTT and one protease inhibitor cocktail ber reporting a C against the total number of reads reporting a C or T tablet [Roche] per 50 ml), and then homogenized at 4°C for was calculated as the methylation level for each cytosine site. Only 10 min. The homogenate was filtered with two layers of Miracloth sites that have at least five-fold coverage were considered. To iden- and the flow-through was centrifuged at 1500 g for 10 min at 4 tify DMRs, sequencing data of two biological replicates of each sam- Next, the supernatant was centrifuged twice at maximum speed ple were combined and the DNA methylation levels of CG, CHG and for 10 min at 4°C and the final supernatant was collected as the CHH sites in 200-bp bins were separately evaluated. DMRs are bins cytoplasmic fraction. The pellet from the first centrifugation was in which the absolute change in DNA methylation was greater than subjected to wash for four times with 5 ml of NRBT buffer (20 mM

40%, 20% and 10% between the mutant and the wild type for CG, Tris-HCl [pH 7.4], 25% glycerol, 2.5 mM MgCl2 and 0.2% Triton CHG and CHH methylation, respectively. X-100) at 4°C before it was resuspended with 500 ll of NRB2 (20 mM Tris-HCl [pH 7.5], 0.25 M sucrose, 10 mM MgCl , 0.5% Triton Small RNA deep sequencing and data analysis 2 X-100, 5 mM b-mercaptoethanol and one protease inhibitor cocktail Small RNAs were separated by size from total RNA isolated from tablet [Roche] per 50 ml). The suspension was laid carefully over l 2-week-old seedlings using TRIzol Reagent (Thermo Fisher Scien- 500 l NRB3 (20 mM Tris-HCl [pH 7.5], 1.7 M sucrose, 10 mM MgCl2, b tific). Library preparation and deep sequencing were accom- 0.5% Triton X-100, 5 mM -mercaptoethanol and one protease inhi- plished in BIONOVA (Beijing). For the analysis of small RNA data, bitor cocktail tablet [Roche] per 50 ml), and then centrifuged at ° raw reads were filtered by removing adapter sequences and low- maximum speed for 45 min at 4 C Lastly, the final pellet was l quality (q < 15) reads, then clean reads with a length of 18–30 nt resuspended in 400 L of lysis buffer as the nuclear fraction. were mapped to the reference nuclear genome (C24 genome) Proteins were extracted from the total, cytoplasmic, and nuclear using Bowtie (Langmead et al., 2009), and only reads perfectly fractions and run on sodium dodecyl sulfate–polyacrylamide gel matched (v = 0) to the genome were kept for further analysis. The electrophoresis gels for immunoblotting. NOT1-FLAG was reference genome was divided into 200-bp bins, then the counts detected using the antibody against FLAG (F1804; Sigma). As of 24-nt reads in every bin were normalized to reads per million quality controls for the fractionation, the unclear histone H3 and by the total clean reads number. The profile of small RNAs in each the cytosolic protein UGPase were also detected in these fractions genotype over chromosomes was plotted using 100-kb bins. To using anti-histone H3 (ab1791; Abcam) and anti-UGPase determine 24-nt siRNA regions, sequencing data of two biological (AS05086; Agrisera) antibodies. replicates of each sample were combined and the 24-nt siRNA Ten-day-old seedlings of not1 and wild type grew under normal levels in 200-bp bins were evaluated. The nrpd1-dependent 24-nt conditions, and then were collected and transferred to liquid MS siRNA regions were regions displaying at least a four-fold reduc- medium containing 5 lM a-amanitin (Sigma) for 0, 6, 9, 12 and tion of 24-nt siRNAs in nrpd1 compared with the wild type. The 18 h. Total RNAs were extracted from the seedlings collected at regions showing at least two-fold decrease of 24-nt siRNAs in each time point and were subjected to RT as described above.

Then, the transcript levels of SDC were evaluated by quantitative Ausin, I., Greenberg, M.V., Li, C.F. and Jacobsen, S.E. (2012) The splicing RT-PCR. factor SR45 affects the RNA-directed DNA methylation pathway in Ara- bidopsis. Epigenetics. 7,29–33. ACKNOWLEDGEMENTS Bayne, E.H., Portoso, M., Kagansky, A., Kos-Braun, I.C., Urano, T., Ekwall, K., Alves, F., Rappsilber, J. and Allshire, R.C. (2008) Splicing factors facili- This work was supported by the National Key Research and Devel- tate RNAi-directed silencing in fission yeast. Science. 322, 602–606. opment Program of China (2016YFA0500801) from the Chinese Blevins, T., Podicheti, R., Mishra, V., Marasco, M., Tang, H. and Pikaard, Ministry of Science and Technology (to XJH). C.S. (2015) Identification of Pol IV and RDR2-dependent precursors of 24 nt siRNAs guiding de novo DNA methylation in Arabidopsis. eLife. 4, AUTHOR CONTRIBUTIONS e09591. Blevins, T., Pontvianne, F., Cocklin, R. et al. (2014) A two-step process for XJH, HRZ and JKZ designed the research. HRZ performed epigenetic inheritance in Arabidopsis. Mol. Cell. 54,30–42. most of the experiments. RNL, HWH and TC performed Bohmdorfer, G., Rowley, M.J., Kucinski, J., Zhu, Y., Amies, I. and Wierz- bicki, A.T. (2014) RNA-directed DNA methylation requires stepwise bind- bioinformatic analysis, LL and SC provided technical sup- ing of silencing factors to long non-coding RNA. Plant J. 79, 181–191. port for mass spectrometric analysis. HRZ and XJH wrote Borges, F. and Martienssen, R.A. (2013) Establishing epigenetic variation the manuscript. All authors read and approved the final during genome reprogramming. RNA Biol. 10, 490–494. Braun, J.E., Huntzinger, E., Fauser, M. and Izaurralde, E. (2011) GW182 pro- manuscript. teins directly recruit cytoplasmic deadenylase complexes to miRNA targets. Mol. Cell. 44, 120–133. CONFLICT OF INTEREST Chen, C.Y. and Shyu, A.B. (2011) Mechanisms of deadenylation-dependent decay. Wiley Interdiscipl. Rev. RNA. 2, 167–183. The authors declare no conflict of interest. Collart, M.A. and Panasenko, O.O. (2012) The Ccr4–not complex. Gene, 492, 42–53. DATA AVAILABILITY STATEMENT Collart, M.A., Panasenko, O.O. and Nikolaev, S.I. (2013) The Not3/5 subunit of the Ccr4-Not complex: a central regulator of gene expression that inte- Raw small RNA-seq and bisulfite-seq data have been grates signals between the cytoplasm and the nucleus in eukaryotic cells. deposited in the Gene Expression Omnibus (GEO) data- Cell. Signal. 25, 743–751. base with the accession no. GSE148430. Dou, K., Huang, C.F., Ma, Z.Y., Zhang, C.J., Zhou, J.X., Huang, H.W., Cai, T., Tang, K., Zhu, J.K. and He, X.J. (2013) The PRP6-like splicing factor SUPPORTING INFORMATION STA1 is involved in RNA-directed DNA methylation by facilitating the production of Pol V-dependent scaffold RNAs. Nucl. Acids Res. 41, Additional Supporting Information may be found in the online ver- 8489–8502. sion of this article. Du, J., Johnson, L.M., Jacobsen, S.E. and Patel, D.J. (2015) DNA methylation pathways and their crosstalk with histone methylation. Nat. Rev. – Figure S1. Determination of the effect of not1 on transcriptional Mol. Cell Biol. 16, 519 532. silencing by quantitative RT-PCR. Garneau, N.L., Wilusz, J. and Wilusz, C.J. (2007) The highways and byways of mRNA decay. Nat. Rev. Mol. Cell Biol. 8, 113–126. Figure S2. Identification and characterization of NOT1. Gong, Z., Morales-Ruiz, T., Ariza, R.R., Roldan-Arjona, T., David, L. and Zhu, Figure S3. Identification of abnormal NOT1 transcript variants in J.K. (2002) ROS1, a repressor of transcriptional gene silencing in Ara- ros1not1. bidopsis, encodes a DNA glycosylase/lyase. Cell, 111, 803–814. Figure S4. Alignment analysis of NOT1 proteins. Haag, J.R. and Pikaard, C.S. (2011) Multisubunit RNA polymerases IV and V: purveyors of non-coding RNA for plant gene silencing. Nat. Rev. Mol. Figure S5. Complementation assay for the not1 mutant. Cell Biol. 12, 483–492. Figure S6. Effects of not1 and nrpd1 on CG and CHG methylation Haag, J.R., Ream, T.S., Marasco, M., Nicora, C.D., Norbeck, A.D., Pasa-Tolic, at the whole-genome level. L. and Pikaard, C.S. (2012) In vitro transcription activities of Pol IV, Pol V, and RDR2 reveal coupling of Pol IV and RDR2 for dsRNA synthesis in Figure S7. Distributions of not1 CHH hypo-DMRs and nrpd1 CHH plant RNA silencing. Mol. Cell. 48, 811–818. hypo-DMRs. Han, Y.F., Dou, K., Ma, Z.Y., Zhang, S.W., Huang, H.W., Li, L., Cai, T., Chen, Figure S8. Genome-wide effects of not1 and nrpd1 on accumula- S., Zhu, J.K. and He, X.J. (2014) SUVR2 is involved in transcriptional tion of 24-nt siRNAs. gene silencing by associating with SNF2-related chromatin-remodeling Figure S9. Locations of the reduced 24-nt siRNA regions caused proteins in Arabidopsis. Cell Res. 24, 1445–1465. by nrpd1 and not1 on chromosomes. Han, Y.F., Zhao, Q.Y., Dang, L.L. et al. (2016) The SUMO E3 ligase-like proteins PIAL1 and PIAL2 interact with MOM1 and form a novel complex Figure S10. Neighbor-joining trees of NOT4, CAF40/NOT9 and required for transcriptional silencing. Plant Cell, 28, 1215–1229. DHH1. He, X.J., Chen, T. and Zhu, J.K. (2011) Regulation and function of DNA Figure S11. Diagrams of NOT2a, NOT3, NOT9a, NOT10 and methylation in plants and animals. Cell Res. 21, 442–465. NOT11, and positions of T-DNA insertions. He, X.J., Hsu, Y.F., Pontes, O., Zhu, J., Lu, J., Bressan, R.A., Pikaard, C., Dataset S1. List of ros1 suppressors identified by map-based clon- Wang, C.S. and Zhu, J.K. (2009a) NRPD4, a protein related to the RPB4 subunit of RNA polymerase II, is a component of RNA polymerases IV ing in the study. and V and is required for RNA-directed DNA methylation. Genes Dev. 23, Dataset S2. List of copurified proteins with NOT1-FLAG. 318–330. Dataset S3. List of copurified proteins with NOT3-FLAG. He, X.J., Hsu, Y.F., Zhu, S., Wierzbicki, A.T., Pontes, O., Pikaard, C.S., Liu, H.L., Wang, C.S., Jin, H. and Zhu, J.K. (2009b) An effector of RNA-di- Dataset S4. Primers used in this study. rected DNA methylation in arabidopsis is an ARGONAUTE 4- and RNA- – REFERENCES binding protein. Cell, 137, 498 508. He, X.J., Ma, Z.Y. and Liu, Z.W. (2014) Non-coding RNA transcription and – Arae, T., Morita, K., Imahori, R., Suzuki, Y., Yasuda, S., Sato, T., Yamaguchi, RNA-directed DNA methylation in Arabidopsis. Mol. Plant. 7, 1406 1414. J. and Chiba, Y. (2019) Identification of Arabidopsis CCR4-NOT com- Henderson, I.R. and Jacobsen, S.E. (2008) Tandem repeats upstream of the plexes with pumilio RNA-binding proteins, APUM5 and APUM2. Plant Arabidopsis endogene SDC recruit non-CG DNA methylation and initiate – Cell Physiol. 60, 2015–2025. siRNA spreading. Genes Dev. 22, 1597 1606.

Herr, A.J., Jensen, M.B., Dalmay, T. and Baulcombe, D.C. (2005) RNA poly- Stroud, H., Greenberg, M.V., Feng, S., Bernatavichute, Y.V. and Jacobsen, merase IV directs silencing of endogenous DNA. Science, 308, 118–120. S.E. (2013) Comprehensive analysis of silencing mutants reveals com- Inada, T. and Makino, S. (2014) Novel roles of the multi-functional CCR4- plex regulation of the Arabidopsis methylome. Cell, 152, 352–364. NOT complex in post-transcriptional regulation. Front. Genet. 5, 135. Suzuki, Y., Arae, T., Green, P.J., Yamaguchi, J. and Chiba, Y. (2015) AtCCR4a Johnson, L.M., Du, J., Hale, C.J. et al. (2014) SRA- and SET-domain-contain- and AtCCR4b are involved in determining the Poly(A) length of granule- ing proteins link RNA polymerase V occupancy to DNA methylation. Nat- bound starch synthase 1 transcript and modulating sucrose and starch ure, 507, 124–128. metabolism in Arabidopsis thaliana. Plant Cell Physiol. 56, 863–874. Langmead, B., Trapnell, C., Pop, M. and Salzberg, S.L. (2009) Ultrafast and Tabach, Y., Billi, A.C., Hayes, G.D. et al. (2013) Identification of small RNA memory-efficient alignment of short DNA sequences to the human gen- pathway genes using patterns of phylogenetic conservation and diver- ome. Genome Biol. 10, R25. gence. Nature, 493, 694–698. Law, J.A., Du, J., Hale, C.J., Feng, S., Krajewski, K., Palanca, A.M., Strahl, Temme, C., Simonelig, M. and Wahle, E. (2014) Deadenylation of mRNA by B.D., Patel, D.J. and Jacobsen, S.E. (2013) Polymerase IV occupancy at the CCR4-NOT complex in Drosophila: molecular and developmental RNA-directed DNA methylation sites requires SHH1. Nature, 498, 385– aspects. Front. Genet. 5, 143. 389. van Hoof, A. and Parker, R. (2002) Messenger RNA degradation: beginning Law, J.A. and Jacobsen, S.E. (2010) Establishing, maintaining and modify- at the end. Curr. Biol. 12, R285–R287. ing DNA methylation patterns in plants and animals. Nat. Rev. Genet. 11, Wahle, E. and Winkler, G.S. (2013) RNA decay machines: deadenylation by 204–220. the Ccr4-not and Pan2-Pan3 complexes. Biochem. Biophys. Acta. 1829, Lee, T.F., Gurazada, S.G., Zhai, J., Li, S., Simon, S.A., Matzke, M.A., Chen, 561–570. X. and Meyers, B.C. (2012) RNA polymerase V-dependent small RNAs in Walley, J.W., Kelley, D.R., Nestorova, G., Hirschberg, D.L. and Dehesh, K. Arabidopsis originate from small, intergenic loci including most SINE (2010) Arabidopsis deadenylases AtCAF1a and AtCAF1b play overlapping repeats. Epigenetics, 7, 781–795. and distinct roles in mediating environmental stress responses. Plant Li, S., Vandivier, L.E., Tu, B., Gao, L., Won, S.Y., Li, S., Zheng, B., Gregory, Physiol. 152, 866–875. B.D. and Chen, X. (2015) Detection of Pol IV/RDR2-dependent transcripts Wang, L., Song, X., Gu, L., Li, X., Cao, S., Chu, C., Cui, X., Chen, X. and Cao, at the genomic scale in Arabidopsis reveals features and regulation of X. (2013) NOT2 Proteins promote polymerase II-dependent transcription siRNA biogenesis. Genome Res. 25, 235–245. and interact with multiple microRNA biogenesis factors in Arabidopsis. Liu, J., Bai, G., Zhang, C., Chen, W., Zhou, J., Zhang, S., Chen, Q., Deng, X., Plant Cell. 25, 715–727. He, X.J. and Zhu, J.K. (2011) An atypical component of RNA-directed Wang, W., Ye, R., Xin, Y., Fang, X., Li, C., Shi, H., Zhou, X. and Qi, Y. (2011) DNA methylation machinery has both DNA methylation-dependent and - An importin beta protein negatively regulates MicroRNA activity in Ara- independent roles in locus-specific transcriptional gene silencing. Cell bidopsis. Plant Cell. 23, 3565–3576. Res. 21, 1691–1700. Wierzbicki, A.T., Cocklin, R., Mayampurath, A., Lister, R., Rowley, M.J., Gre- Liu, Z.W., Shao, C.R., Zhang, C.J., Zhou, J.X., Zhang, S.W., Li, L., Chen, S., gory, B.D., Ecker, J.R., Tang, H. and Pikaard, C.S. (2012) Spatial and func- Huang, H.W., Cai, T. and He, X.J. (2014) The SET domain proteins SUVH2 tional relationships among Pol V-associated loci, Pol IV-dependent and SUVH9 are required for Pol V occupancy at RNA-directed DNA siRNAs, and cytosine methylation in the Arabidopsis epigenome. Genes methylation loci. PLoS Genet. 10, e1003948. Dev. 26, 1825–1836. Liu, Z.W., Zhou, J.X., Huang, H.W., Li, Y.Q., Shao, C.R., Li, L., Cai, T., Chen, Wierzbicki, A.T., Haag, J.R. and Pikaard, C.S. (2008) Noncoding transcription S. and He, X.J. (2016) Two Components of the RNA-Directed DNA methy- by RNA polymerase Pol IVb/Pol V mediates transcriptional silencing of lation pathway associate with MORC6 and silence loci targeted by overlapping and adjacent genes. Cell, 135, 635–648. MORC6 in Arabidopsis. PLoS Genet. 12, e1006026. Xie, Z., Johansen, L.K., Gustafson, A.M., Kasschau, K.D., Lellis, A.D., Zilber- Mathys, H., Basquin, J., Ozgur, S., Czarnocki-Cieciura, M., Bonneau, F., man, D., Jacobsen, S.E. and Carrington, J.C. (2004) Genetic and func- Aartse, A., Dziembowski, A., Nowotny, M., Conti, E. and Filipowicz, W. tional diversification of small RNA pathways in plants. PLoS Biol. 2, (2014) Structural and biochemical insights to the role of the CCR4-NOT E104. complex and DDX6 ATPase in microRNA repression. Mol. Cell. 54, 751– Yamanaka, S., Mehta, S., Reyes-Turcu, F.E., Zhuang, F., Fuchs, R.T., Rong, 765. Y., Robb, G.B. and Grewal, S.I. (2013) RNAi triggered by specialized Matzke, M.A. and Mosher, R.A. (2014) RNA-directed DNA methylation: an machinery silences developmental genes and retrotransposons. Nature, epigenetic pathway of increasing complexity. Nat. Rev. Genet. 15, 394– 493, 557–560. 408. Yang, D.L., Zhang, G., Tang, K., Li, J., Yang, L., Huang, H., Zhang, H. and Miller, J.E. and Reese, J.C. (2012) Ccr4-Not complex: the control freak of Zhu, J.K. (2015) Dicer-independent RNA-directed DNA methylation in eukaryotic cells. Crit. Rev. Biochem. Mol. Biol. 47, 315–333. Arabidopsis. Cell Res. 26,66–82. Mosher, R.A., Schwach, F., Studholme, D. and Baulcombe, D.C. (2008) Yang, D.L., Zhang, G., Wang, L. et al. (2018) Four putative SWI2/SNF2 chro- PolIVb influences RNA-directed DNA methylation independently of its matin remodelers have dual roles in regulating DNA methylation in Ara- role in siRNA biogenesis. Proc. Natl Acad. Sci. USA. 105, 3145–3150. bidopsis. Cell Discov. 4, 55. Motomura, K., Arae, T., Araki-Uramoto, H. et al. (2020) AtNOT1 Is a novel Ye, R., Chen, Z., Lian, B., Rowley, M.J., Xia, N., Chai, J., Li, Y., He, X.J., regulator of gene expression during pollen development. Plant Cell Phys- Wierzbicki, A.T. and Qi, Y. (2016) A dicer-independent route for biogene- iol. 61, 712–721. sis of siRNAs that direct DNA methylation in Arabidopsis. Mol. Cell. 61, Naftelberg, S., Schor, I.E., Ast, G. and Kornblihtt, A.R. (2015) Regulation of 222–235. alternative splicing through coupling with transcription and chromatin Ye, R., Wang, W., Iki, T., Liu, C., Wu, Y., Ishikawa, M., Zhou, X. and Qi, Y. structure. Annu. Rev. Biochem. 84, 165–198. (2012) Cytoplasmic assembly and selective nuclear import of Arabidopsis Nasertorabi, F., Batisse, C., Diepholz, M., Suck, D. and Bottcher, B. (2011) Argonaute4/siRNA complexes. Mol. Cell. 46, 859–870. Insights into the structure of the CCR4-NOT complex by electron micro- Zemach, A., Kim, M.Y., Hsieh, P.H., Coleman-Derr, D., Eshed-Williams, L., scopy. FEBS Lett. 585, 2182–2186. Thao, K., Harmer, S.L. and Zilberman, D. (2013) The Arabidopsis nucleo- Neugebauer, K.M. (2002) On the importance of being co-transcriptional. J. some remodeler DDM1 allows DNA methyltransferases to access H1- Cell Sci. 115, 3865–3871. containing heterochromatin. Cell, 153, 193–205. Pontes, O., Li, C.F., Costa Nunes, P., Haag, J., Ream, T., Vitins, A., Jacob- Zhai, J., Bischof, S., Wang, H. et al. (2015) A one precursor one siRNA sen, S.E. and Pikaard, C.S. (2006) The Arabidopsis chromatin-modifying model for Pol IV-dependent siRNA biogenesis. Cell, 163, 445–455. nuclear siRNA pathway involves a nucleolar RNA processing center. Cell, Zhang, C.J., Ning, Y.Q., Zhang, S.W., Chen, Q., Shao, C.R., Guo, Y.W., Zhou, 126,79–92. J.X., Li, L., Chen, S. and He, X.J. (2012) IDN2 and its paralogs form a Reddy, A.S., Marquez, Y., Kalyna, M. and Barta, A. (2013) Complexity of the complex required for RNA-directed DNA methylation. PLoS Genet. 8, alternative splicing landscape in plants. Plant Cell. 25, 3657–3683. e1002693. Shirai, Y.T., Suzuki, T., Morita, M., Takahashi, A. and Yamamoto, T. (2014) Zhang, C.J., Zhou, J.X., Liu, J. et al. (2013a) The splicing machinery pro- Multifunctional roles of the mammalian CCR4-NOT complex in physio- motes RNA-directed DNA methylation and transcriptional silencing in logical phenomena. Front. Genet. 5, 286. Arabidopsis. EMBO J. 32, 1128–1140.

Zhang, H., Lang, Z. and Zhu, J.K. (2018) Dynamics and function of DNA siRNA-directed transcriptional gene silencing in Arabidopsis. Genes Dev. methylation in plants. Nat. Rev. Mol. Cell Biol. 19, 489–506. 23, 2850–2860. Zhang, H., Ma, Z.Y., Zeng, L. et al. (2013b) DTF1 is a core component of Zhong, X., Du, J., Hale, C.J., Gallego-Bartolome, J., Feng, S., Vashisht, A.A., RNA-directed DNA methylation and may assist in the recruitment of Pol Chory, J., Wohlschlegel, J.A., Patel, D.J. and Jacobsen, S.E. (2014) IV. Proc. Natl Acad. Sci. USA. 110, 8290–8295. Molecular mechanism of action of plant DRM de novo DNA methyltrans- Zhang, H., Tang, K., Qian, W. et al. (2014) An Rrp6-like protein positively ferases. Cell, 157, 1050–1060. regulates noncoding RNA levels and DNA methylation in Arabidopsis. Zhou, H.R., Zhang, F.F., Ma, Z.Y., Huang, H.W., Jiang, L., Cai, T., Zhu, J.K., Mol. Cell. 54, 418–430. Zhang, C. and He, X.J. (2013) Folate polyglutamylation is involved in Zhang, X., Henderson, I.R., Lu, C., Green, P.J. and Jacobsen, S.E. (2007) chromatin silencing by maintaining global DNA methylation and histone Role of RNA polymerase IV in plant small RNA metabolism. Proc. Natl H3K9 dimethylation in Arabidopsis. Plant Cell. 25, 2545–2559. Acad. Sci. USA. 104, 4536–4541. Zhou, M., Palanca, A.M.S. and Law, J.A. (2018) Locus-speciﬁc control of the Zheng, B., Wang, Z., Li, S., Yu, B., Liu, J.Y. and Chen, X. (2009) Intergenic de novo DNA methylation pathway in Arabidopsis by the CLASSY fam- transcription by RNA polymerase II coordinates Pol IV and Pol V in ily. Nat. Genet. 50, 865–873.