Cotranscriptional RNA processing and modification in plants

Abstract

The activities of RNA polymerases shape the epigenetic landscape of genomes with profound consequences for genome integrity and gene expression. A fundamental event during the regulation of eukaryotic gene expression is the coordination between transcription and RNA processing. Most primary RNAs mature through various RNA processing and modification events to become fully functional. While pioneering results positioned RNA maturation steps after transcription ends, the coupling between the maturation of diverse RNA species and their transcription is becoming increasingly evident in plants. In this review, we discuss recent advances in our understanding of the crosstalk between RNA Polymerase II, IV, and V transcription and nascent RNA processing of both coding and noncoding RNAs.

Figure 1

Arabidopsis promoter-proximal noncoding transcription. A, Divergent noncoding transcription at the At5g55390 locus. IGV genome browser screenshot of Arabidopsis transcriptomics and epigenomics data. Transcriptome annotation of TAIR10 and from the data-driven annotation tool TranscriptomeReconstructoR (Ivanov et al., 2021) are shown. TranscriptomeReconstructoR identifies MC_gene_1448 as a divergent noncoding transcript respective to At5g55390. Note that the MC_gene_1448 overlaps with the noncoding read-through tail transcript (thin pink line) of At5g55400. TSS sequencing (TSS-seq) data in Columbia-0 (Col-0) wild-type (WT) (Nielsen et al., 2019) identified the At5g55390 TSS transcribed on the crick strand (orange bar facing down). TSS-seq data in the hen2-2 mutant (Thomas et al., 2020) identified the TSS of MC_gene_1448 (blue bar facing up). Plant native elongating transcript sequencing (plaNET-seq) data for WT NRPB2 and the NRPB2-Y732F mutant acceleration RNAII transcription are given below (Leng et al., 2020b). plaNET-seq signal for the mRNA genes encoded on the crick strand is shown (gray). plaNET-seq signal for MC_gene_1448 transcribed on the Watson strand (black) indicated strong divergent noncoding transcription. DNAseI-seq data to map accessible chromatin regions (blue) and MNase-seq to map nucleosomes (blue) are displayed below (Zhang et al., 2016). These data are consistent with the interpretation of the At5g55390 mRNA and the MC_gene_1448 divergent noncoding RNA as a divergent transcript pair from a shared promoter NDR. The positions of the flanking nucleosomes are indicated as gray cylinders. TIF-seq data for Col-0 (WT) and the hen2-2 mutant are given below (Thomas et al., 2020). TIF-seq data reveals RNA isoforms overlapping with the MC_gene_1448 transcriptome annotation. Here, TIF-seq in the hen2-2 mutant revealed increased levels. Divergent noncoding transcription from the promoter NDR of the At5g55390 mRNA is thus supported by data-driven transcriptome annotation, plaNET-seq tracks, TSS-seq and TIF-seq data. B, sppRNAs of mRNA at the At1g33810 locus. Data described in (A) to offer insight into transcript isoforms at the At1g33810 locus. A schematic distinguishing the sppRNA isoform (purple) from the mRNA (black) is included. mRNA and sppRNA start from the same TSS (TSS-seq, blue bar facing up), but they are distinguished by their 3′-end. Note that TIF-seq data in the hen2-2 mutant (bottom) reveals sppRNAs as populations of RNAs ending near or shortly after initiation at the TSS, indicated by the purple TIF-seq signal (here: 3′-ends) overlapping the red signal (here 5′-ends). The sppRNA 3′-ends are positioned near the center of the +1 nucleosome indicated by the MNase-seq data (indicated by a gray cylinder). The At1g33810 locus also illustrates a separate point that reflects reduced Pol II stalling peaks in the plaNET-seq tracks of NRPB2-Y732F compared to in WT NRPB2 (black signal). Differences are most clear at the mRNA boundaries where plaNET-seq in WT NRPB2 accumulates.

Figure 2

Main alternative splicing outcomes. Cassette exons, an exon is included/excluded in/from the mature RNA; alternative 5′ss, differential recognition and usage of distinct alternative 5′ss, mature RNAs include an exon with a 3′-end longer or shorter; alternative 3′ss, differential competitive usage of 3′ss generating isoform with a 5′-end of exons extended/shortened; intron retention is the permanence of an intron in the mature RNA. Boxes represent exons, horizontal lines specify introns, discontinuous lines indicate the recognized and used alternative splicing sites.

Figure 3

Impact of RNA polymerase II transcription on alternative splicing decisions. The Recruitment Model (top) is based on the recruitment of splicing factors mediated by interactions with the Pol II CTD. Binding of SR proteins could favor the usage of different splice sites in competition, modulating alternative splicing outcomes. In the Kinetic Model (bottom), the regulation of alternative splicing is modulated by the rate of Pol II elongation (de la Mata and Kornblihtt, 2006). Slow elongation (left) gives more time for recognition of weak splice sites, while fast elongation results in the usage of the stronger sites as a result of the competition (right). The rate of elongation can be influenced, among other things, by the chromatin state that can present obstacles to Pol II movement and by the level of phosphorylation of different residues in the repeats of the heptapeptide unit Y1S2P3T4S5P6S7 (42 in Arabidopsis; Dietrich et al., 1990) of the large Pol II-subunit CTD. Boxes represent exons, lines represent introns.

Figure 4

Proposed model of RdDM initiation. Unmethylated loci containing TEs can be transcribed by Pol II to produce siRNAs from highly complementary dsRNA hairpin transcripts derived from inverted repeated sequences (Sasaki et al., 2014) (a.i) or easiRNAs after the initial cut of the TE transcripts by an miRNA-loaded AGO1 and conversion to dsRNA by RDR6 (Creasey et al., 2014) (a.ii). A nuclear pool of hetsiRNA is then preferentially loaded into AGO4 or AGO6 (b). AGO-loaded siRNAs will recognize unmethylated target loci either by binding Pol II transcripts (Sigman et al., 2021), scarce transcripts produced by surveilling Pol V (Tsuzuki et al., 2020), or even perhaps by direct AGO–DNA interaction (Lahmy et al., 2016) (c). AGO4/6 then act as a scaffold recruiting Pol V (Sigman et al., 2021) (d), which in turn initiates de novo methylation of the targeted loci (e). Methylated DNA will promote the recruitment of Pol IV to the locus (f) triggering the production of RDR2/DCL3-derived siRNAs and initiating the feed-forward loops of RdDM that will keep the locus silenced (g).

Figure 5

A “Processing Priming” hypothetical model for the recognition of MIRNA loci and initiation of cotranscriptional processing of pri-miRNAs. In a first round of Pol II-transcription, the dsRNA region of pri-miRNAs is recognized by the microprocessor (represented here by its core components DCL1 and HYL1) to initiate processing (a). DCL1 is then able to interact with the Elongation complex recruiting miRNA biogenesis-specific factors, such as HASTY, to the loci (b). The Elongator and Mediator complexes, along with HASTY and other accessory proteins, will then act as scaffolds promoting the recruitment of DCL1, and the microprocessor, to the primed MIRNA loci (c). This association enables a quick interaction of the microprocessor to nascent pri-miRNAs and their cotranscriptional processing (d) in a process that once primed is likely to perpetuate (e).