2018年5月21日，Nature Reviews Molecular Cell Biology在线发表了中国科学院上海植物逆境生物学研究中心朱健康研究员、张惠明研究员与郎曌博研究员共同完成的题为“Dynamics and function of DNA methylation in plants”的综述文章。本博客将持续解读该文章，本文为连载第二期，包括第一章节DNA甲基化动态的第一部分内容：RdDM介导的DNA甲基化建立。

Dynamics and function of DNA methylation in plants

First author: Huiming Zhang; Affiliations: Shanghai Center for Plant Stress Biology, Chinese Academy of Sciences (中国科学院上海植物逆境生物学研究中心): Shanghai, China

Corresponding author: Jian- Kang Zhu

• DNA methylation dynamics

A specific DNA methylation state reflects the outcome of the dynamic regulation of establishment, maintenance and active-removal activities. These activities are catalysed by various enzymes that are targeted to specific genomic regions by distinct pathways. Plant DNA methylation occurs in all cytosine sequence contexts: CG, CHG and CHH (H represents A, T or C)12,13. In A. thaliana, genome-wide DNA methylation is characterized by heavy methylation in heterochromatin, which is enriched with transposable elements (transposons) and other repetitive DNA sequences12,14. Interspersed transposon-associated DNA methylation also exists in euchromatic chromosome arms12.

一个特异性DNA甲基化状态反映了甲基化建立、维持和主动去除的动态调控的结果。这些过程受到多个酶的催化，并且由不同通路靶向到特异性基因组区域。植物DNA甲基化发生在所有的胞嘧啶序列上，包括三种类型，即CG、CHG和CHH，其中H代表了A、T或C。在拟南芥中，全基因组范围的DNA甲基化特征主要是异染色质中的高度甲基化，主要富集在转座子元件和其他类型的DNA重复序列上。散在分布转座子相关的DNA甲基化同样在真核生物染色体臂上存在。

Establishment of DNA methylation by the RNA-directed DNA methylation pathway

In plants, de novo DNA methylation is mediated through the RNA-directed DNA methylation (RdDM) pathway, which involves small interfering RNAs (siRNAs) and scaffold RNAs in addition to an array of proteins (Fig. 1). According to the current understanding of canonical RdDM in A. thaliana7,11,15,16, the production of 24-nucleotide siRNAs is initiated through transcription by RNA POLYMERASE IV (POL IV), which is followed by RNA-DEPENDENT RNA POLYMERASE 2 (RDRP2; also known as RDR2)-dependent copying of the transcript to generate a double-stranded RNA (dsRNA) and by DICER-LIKE PROTEIN 3 (DCL3)-dependent cleavage of the dsRNA into siRNAs. The siRNAs are loaded onto ARGONAUTE (AGO) proteins, mainly AGO4 and AGO6, and pair with complementary scaffold RNAs, which are nascent (初生的) transcripts produced by POL V. AGO4 interacts with the DNA methyltransferase DOMAINS REARRANGED METHYLASE 2 (DRM2)¹⁷, which catalyses de novo DNA methylation in a sequence-independent manner. This reaction may be assisted by RNA-DIRECTED DNA METHYLATION 1 (RDM1), which associates with both AGO4 and DRM2 and may bind single-stranded methylated DNA18 (Fig. 1).

在植物中，从头DNA甲基化是由RNA指导的DNA甲基化通路（RdDM）所介导的，除了一系列的蛋白还涉及到了小干扰RNA（siRNA）和支架RNA（图1）。根据目前对于拟南芥中经典的RdDM通路理解，24核苷酸siRNA的产生是由RNA聚合酶POL IV通过转录起始的，起始后由RNA依赖性的RNA聚合酶RDRP2（也叫RDR2）进行转录本拷贝以形成双链的RNA（dsRNA），再由类DICER蛋白DCL3将dsRNA剪切成siRNA。这些siRNA会装载到AGO蛋白上，主要是AGO4和AGO6，然后通过与支架RNA互补配对，这些支架RNA是由POL V. AGO4与DNA甲基转移酶DRM2互作产生的，进而通过一种不依赖于序列的方式催化从头甲基化。该反应可能由RDM1所协助，该蛋白可同时关联AGO4和DRM2，还可能能够结合单链的被甲基化的DNA（图1）。

In addition to the sequence-specific pairing between siRNAs and scaffold RNAs, protein interactions between AGO4 and the AGO hook-containing proteins DNA-DIRECTED POL V SUBUNIT 1 (also known as NRPE1) and RDM3 are also important for RdDM. NRPE1 is the largest subunit of POL V, and RDM3 is a POL V-associated putative transcription elongation factor19,20. POL V-transcribed ncRNAs must remain on the chromatin to function as scaffold RNAs; this association seems to be facilitated by RRP6-LIKE 1 (RRP6L1), which is a homologue of the yeast and mammalian ribosomal RNA-processing 6 (RRP6) proteins that can function in RNA retention21. In addition, the siRNA-scaffold RNA pairing may be stabilized by the INVOLVED IN DE NOVO 2 (IDN2)-IDN2 PARALOGUE (IDP) complex, which binds RNA and interacts with the SWI/SNF chromatin-remodelling complex that contains SWI/SNF COMPLEX SUBUNIT SWI3B and participates in POL V-mediated transcriptional silencing by altering nucleosome positioning22-28.

除了siRNA和支架RNA之间序列特异性的配对外，AGO4和NRPE1及RDM之间的蛋白互作对于RdDM也是非常重要的。NRPE1是POL V最大的亚基，而RDM3是POL V相关的转录延伸因子。POL V转录的ncRNA肯定保留在染色质上，作为支架RNA发挥功能，这个过程可能受到了RRP6L1的促进，RRP6L1蛋白是酵母和哺乳动物核糖体RNA加工RRP6蛋白的同源物，主要作用于RNA保留。另外，IDP复合物能够稳定siRNA与支架RNA之间的配对，IDP结合RNA，并与SWI/SNF染色质重塑复合物互作，该复合物主要包含SWI/SNF复合物亚基SWI3B，并参与POL V介导的核小体位置改变引起的转录沉默。

The recruitment of POL IV and POL V to RdDM target loci can be facilitated by pre-existing chromatin modifications. POL IV is recruited by SAWADEE HOMEODOMAIN HOMOLOGUE 1 (SHH1), which binds dimethylated histone H3 lysine 9 (H3K9me2) through its Tudor domain29,30. SHH1 also interacts with SNF2 DOMAIN-CONTAINING PROTEIN CLASSY 1 (CLSY1), which is a chromatin-remodelling protein associated with POL IV and is required for POL IV-dependent siRNA production30,31 (Fig. 1). The association of POL V with chromatin for scaffold-RNA production requires the chromatin-remodelling DDR, which comprises the chromatin-remodelling protein DEFECTIVE IN RNADIRECTED DNA METHYLATION 1 (DRD1), the putative structural maintenance of chromosomes protein DEFECTIVE IN MERISTEM SILENCING 3 and RDM118,32-35 (Fig. 1). The DDR complex physically interacts with SUPPRESSOR OF VARIEGATION 3-9 HOMOLOG PROTEIN 2 (SUVH2) and SUVH9, which are SUPPRESSOR OF VARIEGATION (SU(VAR)) 3-9 histone methyltransferase family proteins but lack histone methyltransferase activity36,37 (Fig. 1). SUVH2 and SUVH9 recognize methylated cytosine through theirSET and RING finger-associated (SRA) domains and are required for the proper genome-wide chromatin occupancy of POL V, and were therefore proposed to recruit POL V through pre-existing DNA methylation37. However, the tethering (用绳子拴住) of SUVH9 by a zinc-finger to unmethylated DNA is sufficient to recruit POL V and to establish DNA methylation and gene silencing37.

预先存在的染色质修饰可以促进招募POL IV和POL V到RdDM的靶向位点。POL IV由SHH1蛋白招募，该蛋白通过本身的Tudor结构域结合二甲基化的组蛋白H3赖氨酸9（H3K9me2）。SHH1还能够与CLSY1互作，而CLSY1蛋白是与POL IV相关的染色质重塑蛋白，并且对于POL IV-依赖性的siRNA合成是必要的（图1）。POL V与染色质之间的关联生成支架RNA这个过程需要染色质重塑DDR复合物，主要由染色质重塑蛋白DRD1、染色体结构维持蛋白DMS3和RDM1构成（图1）。DDR复合物物理上能够与SUVH2和SUVH9互作，这两个蛋白是属于SU(VAR)3-9组蛋白甲基转移酶家族蛋白的成员，但缺失足蛋白甲基转移酶活性（图1）。SUVH2和SUVH9通SRA结构域识别甲基化的胞嘧啶，对于POL V在全基因组范围染色质上的正确分布是必需的，因此被认为通过预先存在的DNA甲基化来招募POL V。然而，通过锌指将SUVH9栓到未甲基化的DNA上已经能够招募POL V，并建立起DNA甲基化和基因沉默。

Given that POL V can generate ncRNAs with different 5ʹ ends from the same locus, it seems to initiate transcription independently of promoters38. Genomewide POL V or POL IV chromatin occupancy mapping did not reveal consensus promoter motifs29,39. Some POL V-transcribed ncRNAs have 7-methylguanosine caps at the 5ʹ ends38, indicating that POL V-generated transcripts can be subjected to certain RNA-processing activities that are known to modify POL II-transcribed mRNAs. Nevertheless, POL V-generated transcripts are devoid (缺乏) of polyadenylation at their 3ʹ ends and thus are different from mRNAs38. Unlike POL V transcripts, which are long enough to be detected by regular PCR38, POL IV-transcribed ncRNAs (P4 RNAs) are mostly 26–50 nucleotides in length and were thus identified only recently by small-RNA deep sequencing in A. thaliana with mutant dcl2, dcl3and dcl4(dcltriple mutant) and in A. thalianawith mutant dcl1, dcl2, dcl3and dcl4 (dclquadruple mutant)40-44, in which the cleavage of POL IV-dependent and RDR2-dependent dsRNAs into 24-nucleotide siRNAs is presumably blocked. P4 RNAs accumulate in dcltriple mutants and can be processed into 24-nucleotide siRNAs by exogenous DCL3. Because P4 RNAs are small, each 24-nucleotide siRNA could be produced from one slicing of a precursor P4 RNA42.

既然POL V能够在同一位点产生带有不同5ʹ端的ncRNA，貌似起始转录的过程并不依赖于启动子。全基因范围的POL V或POL IV染色质分布图谱并未发现一致的启动子基序。一些POL V转录的ncRNA在5ʹ端具有7-甲基鸟苷的帽子，显示POL V产生的转录本可以适用于一些已知的用于修饰POL II转录的mRNA加工过程。然而，POL V产生的转录本在3ʹ端缺少多聚腺苷酸，因此与mRNA不同。与POL V转录本足够长能够被常规的PCR检测到不一样，POL IV转录的ncRNA（P4 RNA）大多只有46-50个核苷酸，因此只有近期对拟南芥的dcl三突和四突植株采用小RNA深度测序检测到了这一类ncRNA，这两个突变体植株中POL IV依赖性和RDR2依赖性的dsRNA剪切成24核苷酸siRNA可能失效了。P4 RNA在dcl三突植株中积累，而且可以被外源的DCL3处理成24核苷酸siRNA。因为P4 RNA非常小，每一个24核苷酸siRNA都可能来自于一个P4 RNA前体的切割。

In addition to the canonical POL IV–RDR2–DCL3 pathway that generates 24-nucleotide siRNAs, paralogues of these proteins can also produce siRNAs that trigger non-canonical RdDM (Fig. 1). POL II-mediated transcription can not only generate 24-nucleotide siRNAs and scaffold RNAs but also recruits POL IV and POL V to promote siRNA production at some RdDM target loci45. POL II also has spatially distinct associations with different AGO proteins when compared with POL V46. At Trans- acting siRNA genes and at some regions of transcriptionally active transposons, RdDM depends on POL II and RDR6 rather than on POL IV and RDR247-49. RDR6-dependent RdDM can be mediated either through 21-nucleotide or 22-nucleotide siRNAs, which are produced by DCL2 and DCL4, or through 24-nucleotide siRNAs produced by DCL349,50.

除了经典的POL IV–RDR2–DCL3通路产生24核苷酸siRNA，这些蛋白的旁系同源物同样可以通过非典型RdDM通路产生siRNA（图1）。POL II介导的转录不仅可以产生24核苷酸siRN和支架RNA，同时还能在某些RdDM靶向位点招募POL IV和POL V促进siRNA的产生。POL II与POL V相比，能够与不同的AGO蛋白有空间上不同的联系。在反式激活siRNA基因和一些转录激活转座子区域中，相比于POL IV和RDR2，RdDM更加依赖于POL II和RDR6。由DCL2和DCL4产生的21核苷酸或22核苷酸或者由DCL2产生的24核苷酸可以影响RDR6依赖性RdDM。

Genome wide, most siRNAs in A. thalianaare 24-nucleotide siRNAs, which disappear almost completely in the dclquadruple mutant; however, DNA methylation at approximately two-thirds of RdDM target regions still remains43,44, indicating the existence of DCL-independent RdDM that may be mediated by some DCL-independent siRNAs or directly by P4 RNAs (Fig. 1). RNase III enzymes other than DCL proteins could dice dsRNAs51, and in wild-type plants, they may work with DCLs in processing POL II, POL IV or POL V transcripts into siRNAs.

拟南芥中的siRNA大多数是24核苷酸的，在dcl四突变体植株中几乎完全消失；然而，RdDM靶区域大约三分之二的DNA甲基化仍然保留，表明存在不依赖于DCL的RdDM存在，可能由一些不依赖于DCL的siRNA或直接由P4 RNA所介导（图1）。除了DCL蛋白外的RNase III酶能够切割dsRNA，并且在野生型植株中，RNase III酶可能与DCL蛋白一同发挥作用，将POL II、POL IV或POL V转录本加工成siRNA。

Genetic screens have identified some pre-mRNA splicing factors whose mutations reduce the levels of POL IV-dependent siRNAs to varying degrees52-55,although it remains largely unclear how these splicing factors affect siRNA levels. Similarly, mutations in two splicing factors, STABILIZED 1 and RDM16, reduce the accumulation of POL V-dependent scaffold RNAs53,54. Presumably, some of the splicing factors that normally bind pre-mRNAs may interact with the non-coding transcripts generated by POL IV and POL V and affect their processing or stability, thus influencing siRNA or scaffold RNA abundance.

遗传筛选已经鉴定了一些mRNA前体剪切因子，这些因子突变掉会不同程度地降低POL IV依赖性siRNA的水平，而这些剪切因子是如何影响siRNA的水平暂时还不清楚。比如说，STABILIZED 1和RDM16这两个剪切因子的突变会降低POL V依赖性支架RNA的积累。据推测，一些能够结合mRNA前体的剪切因子可能会跟由POL IV和POL V产生的非编码转录本结合，并影响它们的加工或是稳定性，进而影响siRNA或支架RNA的丰度。

doi: https://doi.org/10.1038/s41580-018-0016-z

Journal: Nature Reviews Molecular Cell Biology

Published date: 21 May, 2018