Epigenetic differences in an identical genetic background modulate alternative splicing in A. thaliana

How stable and temperature-dependent variations in DNA methylation and nucleosome occupancy influence alternative splicing (AS) remains poorly understood in plants. To answer this, we generated transcriptome, whole-genome bisulfite, and MNase sequencing data for an epigenetic Recombinant Inbred Line (epiRIL) of A. thaliana at normal and cold temperature. Our transcriptome data revealed that differential DNA methylation and nucleosome occupancy modulate expression levels of many genes and AS in response to cold. Collectively, DNA methylation and nucleosome levels exhibit characteristic patterns around intron-exon boundaries at normal and cold conditions, and any perturbation in them, in an identical genetic background is sufficient to modulate AS in Arabidopsis.

Furthermore, the AS analysis showed that epigenetic differences between Col-0 and epiRIL-368 induced fewer but contrasting changes under similar temperature conditions (22 C and 4 C) (one-way ANOVA p-value = 1.7492e-09 for DAS genes). For instance, the number of identified DAS genes between Col-0 versus epiRIL-368 was 305 and 311 at 22 C and 4 C, respectively (Fig. 1A a,b).

Intrestingly, there is no overlap between DEGs and DAS genes between Col-0 and epiRIL-368 at 22 C (hypergeometric test p-value =0.198) and 4 C (hypergeometric test p-value = 2.520e-05) (Fig. 1B a-b). Whereas, this number significantly increases (Col-0 at 22 C versus epiRIL-368 at 4 C; 7.1%; hypergeometric test p-value 0.011) when epigenetic variations and cold stress adds on together (Fig. 1B c).

Further, we performed gene functional enrichment analysis for DEGs and DAS genes (Supplementary File S1) for all three gene ontology (GO) terms i.e. Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) (Fig. S2). Among DEGs, the most significant (FDR <0.05) terms involved transcription regulation, Pol II processivity, cold and other such as response to abscisic acids, are highly enriched (Fig. S2 A) in different contrasting groups.

Therefore, we reasoned those variations at the methylation and nucleosome (epigenetic) levels may affect AS because of epigenetic differences between Col-0 and epiRIL-368 ecotypes under different temperatures. To further investigate these variations, we performed WGBS for epiRIL-368 plants grown at 22 C and 4 C. We identified high confidence differentially methylated regions (hc_DMRs) (Fisher's exact test, p-value <0.01) in comparison to 54 Columbia (Col) lines of Arabidopsis using the hc_DMR caller pipeline developed by the Jacobsen group at the University of California [60].

The high number of hypomethylated regions suggests relatively lower methylation levels in epiRIL-368 compared to Col ecotype (Fig. 2A). Among all hcDMRs, 22,968 were hypomethylated regions in the epiRIL-368 line as compared to Col lines (Fig. 2A).

We first calculated AS event inclusion levels (PSI) for a total of 43,953 AS events identified in the reference annotation file of Arabidopsis, followed by the differences in their inclusion (PSI) among different contrast groups. Differential AS events analysis suggests that epiRIL-368 display significant (p-value <0.05) differences in 474 and 516 AS events compared to Col-0 at 22 C and 4 C, respectively (Fig. 2C; Supplementary File S4).

Different types of AS events detected in our analysis show an overall similar distribution observed previously in Arabidopsis [9,11] where intron retention (IR) events are the most prevalent, followed by usage of the alternative acceptor (A3'SS) and alternative donor (A5'SS) sites, and exon skipping (ES) (Fig. 2C; Supplementary File S4).

The CpG, CHG, and CHH methylation levels generated using methylpy [65] were plotted around the donor sites (exon-intron; 5'SS), the acceptor sites (intron-exon; 3'SS), and the exons for epiRIL-368 at 22 C and 4 C (Fig. 3A; Fig. S3). We observed a sharp drop in CpG methylation at both splice sites (5'SS and 3'SS; Fig. 3A a-b) suggesting its role in transcription and splicing dynamics by affecting Pol II processing around 5'SS and 3'SS as compared to flanking regions. DNA methylation around exons always shows a higher methylation level and can be differentiated from their flanking regions (introns, especially splice sites) (Fig. 3A; Fig. S3A). Regardless of temperature treatment, we also found the level of methylated CpG dinucleotides (mCpG) is higher in exons as compared to flanking regions including introns and splice sites (Fig. 3A).

Next, we looked at the function of the genes with high confidence differentially methylated regions (hcDMRs), differential nucleosome positioning (DNPs), and differential splice junctions (DSJs) in our dataset. We first identified the genes with hcDMRs, and DNPs in addition to genes with DSJs for the contrasting groups Col-0 at 22 C versus epiRIL-368 at 22 C, and Col-0 at 22 C versus epiRIL-368 at 4 C. We divided genes into three groups including genes with hcDMRs (hcDMR genes), genes with significant DNPs (DNP genes) and genes with significant DSJs (DSJ genes). Finally, significantly overlapping (Fig. 4A) genes between hcDMR, DNP, and DSJ genes for the contrasting groups Col-0 at 22 C versus epiRIL-368 at 22 C, and Col-0 at 22 C versus epiRIL-368 at 4 C were selected for gene ontology (GO) functional enrichment analysis.

Interactions of gene expression, alternative splicing, and DNA methylation in determining nodule identity

In the present study, we profiled gene expression changes, alternative splicing events, and DNA methylation patterns during nodule formation, development, and senescence. Alternative splicing analysis uncovered a total of 2323 genes that undergo alternative splicing events in at least one nodule developmental stage, with activation of exon skipping and repression of intron retention being the most common splicing events in nodules compared to roots. Approximately 40% of the differentially spliced genes were also differentially expressed at the same nodule developmental stage, implying a substantial association between gene expression and alternative splicing. Genome-wide-DNA methylation analysis revealed dynamic changes in nodule methylomes that were specific to each nodule stage, occurred in a sequence-specific manner, and impacted the expression of 1864 genes. An attractive hypothesis raised by our data is that increased DNA methylation may contribute to the efficiency of alternative splicing.

We next explored our RNA-seq data for differentially spliced events using the junctionseq package (Hartley and Mullikin, 2016). Using a FDR cut-off of 0.05, we identified 5564, 1099, and 542 differentially spliced events in the nodules compared to the root tissues at 12, 22, and 36 dpi, respectively (Figure 4a-c, Data S9-S11). These differentially spliced events were classified into two main groups: differentially used exons (DUEs) and intron retention (IR) events.

Analysis of the DUEs at each nodule developmental stage revealed unique and common splicing variants. For example, we identified 142 DUEs common to the 12- and 22-day-old nodules; 236 common to the 22- and 36-day-old nodules; and 96 common to the 12- and 36-day-old nodules.

We examined whether gene expression impacts alternative splicing events in the nodules. To this end, we compared the lists of DEGs and DSGs identified at each nodule developmental stage for potential overlaps. As shown in Figure 5a-c, 32.5%, 59.1%, and 55.2% of the DSGs were also differentially expressed in the 12-, 22-, and 36-day-old nodules, implying that gene transcription may contribute to alternative splicing in the developing nodules.

To provide further insights into the functional significance of alternative splicing during nodule development, the amino acid sequences of the DUEs were retrieved and used to identify annotated protein domains and families using the Pfam database (https://pfam.xfam.org) .

Our RNA-seq data revealed that key genes involved in DNA methylation (MET1, DRM2, DRM3, CMT1, CMT3, DDM1, and VIM1) and active demethylation (ROS1 and DME) are significantly differentially expressed between nodules and root tissues (Figure S11), suggesting active reprogramming of DNA methylation in the developing nodules. Therefore, we used bisulphite sequencing to profile genome-wide DNA methylation at a single-base resolution in the three nodule developmental stages and the corresponding root tissues (Table S2) using the same tissues utilized for RNA-seq library preparation.

Comparing the DEGs with those associated with DMRs at each nodule developmental stage revealed that more than one-third of the DMR-overlapping genes were significantly differentially expressed, resulting in a unique set of 1864 differentially expressed/differentially methylated genes (Figure 8a).

Given recent studies indicating that DNA methylation may contribute to the regulation of alternative splicing in maize and rice (Regulski et al., 2013; Wang et al., 2016), we examined this possibility by comparing the DSGs and gene body DMGs at each developmental stage. Interestingly, 159 unique genes common to these sets were identified, indicating that DNA methylation may contribute to the regulation of alternative splicing in the developing nodules. Remarkably, the large majority (89.3%) of the gene body DMRs overlapping with the 159 DSGs was hypermethylated, suggesting that gain but not loss of DNA methylation may influence alternative splicing efficiency. Our finding also implies that the impact of DNA methylation on alternative splicing efficiency is independent of DNA methylation sequence context because gene body methylation of these 159 DSGs was found in CG (28.3%), CHG (34.4%), and CHH (37.3%) contexts.

DNA Methylation Affects Gene Alternative Splicing in Plants: An Example from Rice

We previously characterized a null mutant of a major CG methyltransferase in rice (OsMet1-2), which showed genome-wide loss of mCGs by approximately 76% compared with its isogenic wild-type (WT) (Hu et al., 2014). The mutant also showed reduction of mCHG and mCHH, but to a much lesser extent (Hu et al., 2014). Our high-quality transcriptome (RNA-seq) and single-base resolution methylome (BS-seq) data of this rice mutant and its isogenic WT of the same tissue (seedlings) (Hu et al., 2014) provide a robust resource to assay the genome-wide impact of loss of cytosine methylation in gene-body regions on AS changes for the first time in plants.

We next tested the possibility that loss of mCG may have affected expression of AS factor-coding genes, which in turn caused changes of AS in the mutant. We identified a total of 169 AS factor-coding genes, of which only 32 were differentially expressed between mutant and WT, but 113 contained AS events (Supplemental Table 4, binomial test, p < 0.05).

Indeed, we found a substantial proportion of AS factor genes (159, 94.1%) that showed altered gene-body methylation in the mutant relative to WT (Supplemental Table 4).

First, we divided all expressed genes (15 973) that contain multiple exons into two groups: differential methylation (DM) genes (13 434) and non-differential methylation (NDM) genes (2539) (Figure 1C). There were 1266 and 160 DAS genes in the DM group and NDM group, respectively. Likewise, there were 6456 and 1142 genes containing differential spliced junctions (DSJs) in the DM group and NDM group, respectively. Thus, both DAS and DSJ genes were significantly enriched in the DM group as comparison with the NDM group (Fisher's exact test, p < 0.05, Figure 1C), strongly suggesting that DNA methylation affects AS in rice.

Second, it has been established in animal cells that a generally higher mCG level in exon than intron of a given exon-intron junction may serve as a marker of exons to facilitate splicing (Gelfman et al., 2013, Yearim et al., 2015). In light of this, we compared mCG levels around the donor (exon-and-intron) and acceptor (intron-and-exon) regions of all splicing junctions in WT and mutant. We found that the mCG level in WT was markedly higher in exons than in their adjacent introns (Figure 1D, upper panel), consistent with previous findings in human and mouse cells (Gelfman et al., 2013, Yearim et al., 2015).

The maize methylome influences mRNA splice sites and reveals widespread paramutation-like switches guided by small RNA

Here, we present the genome-wide map of cytosine methylation for two maize inbred lines, B73 and Mo17. Correlations with methylation patterns suggest that CG methylation in exons (8%) may deter insertion of Mutator transposon insertion, while CHG methylation at splice acceptor sites may inhibit RNA splicing.

Like in Arabidopsis, CHH and CHG methylation was largely excluded from gene bodies (Hollister et al. 2011; Yang et al. 2011), but a notable exception lay at the intron-exon junctions. The donor and acceptor site consensus sequences in plants are AAG^GTAAG and TTTGCAG^GT, respectively (Reddy 2007), with rare nonconsensus U12-splice donor sites CAG^GCAAG (Sheth et al. 2006). Cytosines at these sites were sometimes methylated on the sense and the antisense strands (Fig. 4A,B, respectively). We hypothesized that this methylation might influence splicing efficiency and/or alternate splicing.

To assess the influence of DNA methylation on splicing, we first collected donor and acceptor sites genome-wide and divided them into low and high methylation categories. Methylation was measured at cytosines on both strands in each methylation context within the donor GT (or GC) and acceptor AG dinucleotides (Methods). Next, we scored the number of RNA-seq reads that corresponded to spliced and unspliced transcripts at each site (Supplemental Table S7). Strikingly, we found that acceptor sites with high levels of CHG methylation were much less efficiently spliced than sites with low levels of methylation. Next, we collected those genes in the genome that were alternatively spliced in at least one tissue (from ZmB73 4a.53 WGS; http://ftp.maizesequence.org/release-4a.53/working-set/), for which we had sufficient RNA-seq data to examine splicing, and for which the alternate exons displayed differing levels of methylation at their acceptor sites. Although only 28 genes met these criteria, 23 of them preferentially used the acceptor site with reduced CHG methylation (Fig. 4C). In contrast, CHH methylation at the donor site did not correlate with splicing efficiency.

Exon methylation in human cells has been proposed to influence splicing by recruiting the DNA binding protein CTCF and slowing down transcription, favoring splicing of adjacent exons (Shukla et al. 2011). However, plants do not possess CTCF, and the CHG methylation we report here does not occur in mammals. While the mechanism by which methylation influences splicing in maize is unclear, CHG methylation in plants can be guided by small RNA and by histone H3K9 methylation, and we speculate that splice site methylation might be mediated by these mechanisms.

Relationship between nucleosome positioning and DNA methylation

Here we report a genome-wide nucleosome positioning analysis of Arabidopsis thaliana using massively parallel sequencing of mononucleosomes. By combining this data with profiles of DNA methylation at single base resolution, we identified 10-base periodicities in the DNA methylation status of nucleosome-bound DNA and found that nucleosomal DNA was more highly methylated than flanking DNA. Finally, as has been observed in animals, nucleosomes were highly enriched on exons, and preferentially positioned at intron-exon and exon-intron boundaries.

Because nucleosomes present a barrier to RNA Pol II transcription, we tested for Pol II occupancy in exons using a chromatin immunoprecipitation microarray approach. We observed significant enrichment of Pol II in exons relative to introns, consistent with the hypothesis that Pol II is paused on exonic DNA (Fig. 4). One possibility is that Pol II stalling on exons could enhance accurate splicing of upstream introns, thus reducing exon skipping and aiding in the fidelity of exon definition20. This is consistent with the finding of Pol II enrichment on human exons and indicates that Pol II enrichment on exons might be a common eukaryotic feature20. Furthermore, particular histone modifications have been recently shown to recruit splicing regulators, providing additional possible mechanisms for the regulation of splicing by nucleosome positioning25.

Maize DNA Methylation in Response to Drought Stress Is Involved in Target Gene Expression and Alternative Splicing

To understand DNA methylation dynamics in maize roots under water stress (WS), we reanalyzed DNA methylation sequencing data to profile DNA methylation and the gene expression landscape of two inbred lines with different drought sensitivities, as well as two of their derived recombination inbred lines (RILs). Gene body DNA methylation showed a negative correlation with gene expression but a positive correlation with exon splicing events.

First, we depicted the DNA methylation coverage around the exon junction, but did not find any evident signals (Figure S9A). Next, we chose genes with equal expression at random as negative controls (Figure S9B) and compared the DNA methylation level and AS number. The DNA methylation level of genes with more AS events was significantly higher than that of genes without AS events (Student's t test, p < 0.001, Figure 4C). Moreover, by comparing the DNA methylation level of genes with different expression levels and determining the correlation, we found a significant positive correlation between DNA methylation and the expression level of exons (Spearman's rank correlation test, p < 0.001, r = 0.18, Figure S9C).

To elucidate whether the ability of DNA methylation in regulating gene expression and AS was used in response to drought stress, we compared the change in the gene expression level, splicing pattern and DNA methylation under WS. To study the association between the change in DNA methylation and AS under WS, we compared the AS patterns and DMRs. Genes with altered DNA methylation under WS had a significantly higher proportion of differential AS (DAS, 2.40% in the promoter, 1.31% in the gene body, and 2.45% in the downstream region) than genes without DMR (0.89%, x2 test, p < 0.001). Furthermore, genes with DMR always had a higher ratio of DAS than genes with stable DNA methylation under WS. As shown in Figure 5E, MSTRG.42149 simultaneously undergoes AS and significant DNA methylation responses under WS.

For differential splicing analysis, Salmo (ver. 0.9.1) [55] was used for transcript quantification and SUPPA2 (ver. 2.3, download date: 7 February 2018, Juan L. Trincado, E08003, Barcelona, Spain) [56] was used for AS event identification with default settings.

DNA methylation/hydroxymethylation regulate gene expression and alternative splicing during terminal granulopoiesis

Using whole genome bisulfite sequenc- ing, reduced representation hydroxymethylation profiling and mRNA sequencing, we compare changes in DNA methylation, DNA hydroxymethylation, gene expression and alternative splicing in mouse promye- locytes and granulocytes.

By comparing the DNA methylation levels within these included and excluded exons, we found that included exons have significantly higher methylation levels than excluded exons (p = 6.3e-07, paired t-test, Figure 4A), consistent with a previous report in IMR90 and HCT116 cell lines [9]. We further examined whether changes in exonic DNA methylation during differentiation would alter the inclusion of exons. We investigated all exons with >20% change in inclusion during the transition from promyelocytes to granulocytes. These changes account for approximately top 10% differential splicing events during granulopoiesis. Intriguingly, our WGBS data did not show a significant correlation between exonic DNA methylation and exon inclusion ratio (R2 = -0.18, p = 0.1, Pearson correlation test, Figure 4B).

Epigenetic features are significantly associated with alternative splicing

Here, we investigate whether the level of genomic CpG dinucleotides (termed CG) and the percentage of methylated CpG dinucleotides (termed mCG) are different in between ASE and CNE. The mCG data is obtained from a study of DNA bisulfite sequencing (BS-seq) of H1 hESC and IMR90 cells [11]. The nucleosome occupancy dataset is obtained from a study using high-throughput sequencing after MNase digestion in resting CD4+ cells [24].

AS events can be generally classified into the following categories: cassette exon (CE), exon skipping (ES), mutually exclusive exon (ME), alternative 5' splice site selection (A5SS), alternative 3' splice site selection (A3SS) and intron retention (IR) [1,2].

We find that ME and ES have significantly lower level of both CG and mCG in the exonic regions (Figure 1a-b). In contrast, IR has significantly higher level of CG in both the exonic and intronic regions; yet, its mCG level in the exonic region is significantly lower while in the intronic region is significantly higher. For both A3SS and A5SS, their associations with CG and mCG are generally not as significant as that with the other types of ASE, though the association is significant in some regions (See Additional file 2 for details).

From human.

Genome-wide association between DNA methylation and alternative splicing in an invertebrate

In order to investigate the relationship between DNA methylation and transcription across the entire honeybee genome, we generated genome-wide transcription data using RNA-seq (pooled RNA from 20 individuals) and analyzed them in conjunction with available genome-wide DNA methylation data from honeybees (pooled DNA from 50 individuals) [18].

We calculated the distribution of DNA methylation across the start and end sites of exons that were either included or skipped during transcription. We found that, overall, exons included in the gene transcript contained significantly more DNA methylation than skipped exons just after the exon start site and before the exon end site (Figure (Figure1A1A and and1B,1B, Additional file 1: Figure S2A and Additional file 1: S2B).

To test whether methylated genes are enriched for splice variants when compared to unmethylated genes we assembled a honeybee transcriptome that included alternatively spliced transcripts using a de novo gene prediction from genome-wide RNA-seq data. We found that alternative transcripts occurred significantly more often in methylated genes as compared to unmethylated genes (Fisher's exact test, P < 1e-10, Figure2), over several different expression thresholds to ensure RNA-seq data quality.

Maternal methionine supplementation during gestation alters alternative splicing and DNA methylation in bovine skeletal muscle

For each gene under evaluation, we calculated the methylation level as the number of differentially methylated cytosines divided by the number of cytosines evaluated in each genomic region of interest, including exons, introns, or the promoter region. Notably, among the genes with at least one differentially used exon and methylation data (1,216 out of 1,290), methylation level in the significant exons was significantly different than the methylation level in the non-significant exons (Kolmogorov-Smirnov test, P-value < 0.001, Fig. 3). Similarly, methylation level in the introns that were contiguous to significant exons was significantly different than the methylation level in the non-contiguous introns (Kolmogorov-Smirnov test, P-value < 0.001, Fig. 3). On the other hand, among genes that had at least one differentially expressed isoform and had methylation data (125 out of 175), there were no differences in methylation levels among exons.

To further visualize the relationship between DNA methylation and alternative splicing, we evaluated differential exon usage and differential DNA methylation at a single-gene level. As an illustration, Fig. 4 shows the exon expression level and DNA methylation status of pre-mRNA-processing factor 40 homolog A (PRPF40A), a gene that is highly involved in mRNA splicing and mRNA processing. Among the 30 exons and 26 introns annotated in the reference annotation, exon E007 was identified as differentially used between maternal diets. Notably, marked differences in methylation proportion were observed between this (significant) exon and the rest of the (non-significant) exons.

Global Co-transcriptional Splicing in Arabidopsis and the Correlation with Splicing Regulation in Mature RNAs

RNA splicing and spliceosome assembly in eukaryotes occur mainly during transcription. However, co-transcriptional splicing has not yet been explored in plants. Here, we built transcriptomes of nascent chromatin RNAs in Arabidopsis thaliana and showed that nearly all introns undergo co-transcriptional splicing, which occurs with higher efficiency for introns in protein-coding genes than for those in noncoding RNAs.

This paper does not investigate splicing and epigenetics, but the hypothesis of methylation affecting splicing is currently grounded in the idea that plants undergo co-transcriptional splicing.

The pattern of DNA methylation alteration, and its association with the changes of gene expression and alternative splicing during phosphate starvation in tomato

Here, we performed integrative methylome and transcriptome analyses of tomato seedlings under Pi-deficient and sufficient conditions. We found Pi caused a slight increase in the overall methylation level, with millions of differentially methylated cytosines (DmCs) and a few hundred differentially methylated regions (DMRs). We also identified thousands of differentially expressed (DE) and differential AS (DAS) genes induced by Pi-, and found that DmCs were more abundant in non-expressed genes than in DE or DAS genes. Moreover, DNA methylation alterations weakly correlated with transcription changes but not with DAS events, and hyper-CHH-DMRs overlapping with transposable elements (TEs) were enriched in a subset of Pi starvation response (PSR) genes.

To assess the influence of Pi- on AS in tomato, we analyzed the RNA-Seq data using rmats (Shen et al., 2014).

To investigate the preferential localizations of methylation changes and their possible relationships with gene expression or AS, we compared the DmC distribution densities at gene body and flanking regions of DE or DAS genes with those of randomly selected genes. Based on this analysis, we found the biased distribution of different types of DmCs at different genes (Figure 5a; Table S3). Generally, the densities of hyper- or hypo-CG-DmCs at DE genes were slightly higher than those of randomly selected genes in either gene body or flanking regions. By contrast, DAS genes in roots and shoots both had significantly fewer CG-DmCs (the patterns of hyper- and hypo-DmC are similar, if not specified thereafter) in the gene bodies as compared with randomly selected genes.

We found that DAS genes with few DmCs tended to be highly expressed, with longer length and more exons, whereas DmC-rich non-expressed genes were mostly short and contained few exons (Figure S7).

To further assess whether and how DMRs relate to alterations of gene expression and AS upon Pi-, we evaluated the enrichment of different types of DMRs around (inside or in the flanking 10-kb regions of) DE and DAS genes using Fisher's exact test (Table S5). We detected preferential localization of CHH-DMRs around DE genes in both roots and shoots, and CNN-DMRs around shoot DE genes. In comparison, many fewer DMRs were seen close to DAS genes.

Does co-transcriptional regulation of alternative splicing mediate plant stress responses?

Review paper 2019

Plants exhibit extensive variation in DNA methylation and gene expression under different developmental and stress conditions (104-108). In eukaryotes, DNA methylation occurs in symmetric CG and CHG (H = A, T or C) and asymmetric CHH contexts (109). However, DNA methylation is largely dependent on the CpG context in plants. In the Arabidopsis genome, 24% of CG sites are methylated, compared with only 6.7% of CHG and 1.7% of CHH sites (110,111). Interestingly, nucleosomal DNA is highly methylated, and exons rather than the introns are marked at the DNA level by high occupancy of nucleosomes. These are preferentially positioned at intron-exon and exon-intron boundaries in both mammals and Arabidopsis (42,77,112,113). Additionally, nucleosome occupancy is also lower in alternatively spliced exons compared with constitutively spliced exons (77-79,114,115). Since DNA is packaged into nucleosomes, pol II elongation rate is inherently subject to frequent pausing at constitutively spliced exons with high GC levels (116,117), and regions of high nucleosome density slow down pol II to facilitate the recruitment of SFs to weaker upstream splice sites (24,28,79,114).

An example of this is found in the honeybee, in which DNA methylation is almost exclusively found in exons with a strong correlation between methylation patterns on alternative exons and splicing patterns of these exons in workers and queens (73). Intriguingly, a reduction in methylation of the dnmt3 gene encoding a methyltransferase via RNAi results in widespread changes in AS in honeybee fat tissues (118). Additionally, a DNA-binding protein, CCCTC-binding factor (CTCF), promotes inclusion of weak upstream exons in the CD45 gene by causing local pol II pausing in mammals. Methylation of exon 5 abolished CTCF binding and resulted in the complete loss of exon 5 from CD45 transcripts (28). Interestingly, a direct link was very recently unveiled between DNA methylation and AS in humans by perturbing DNA methylation patterns of alternatively spliced exons. In this study, the authors used CRISPR-dCas9 proteins (for details, see the 'Engineering splicing variation' section below) and methylating/demethylating enzyme fusions (119). This work clearly demonstrates that changes in the methylation pattern of alternatively spliced exons mediates their inclusion, but has no effect on introns or constitutively spliced exons (119).

Recent work in plants demonstrated abundant DNA methylation and splicing variation under different growth and stress conditions, and during different developmental stages. For example, quantification of AS in wild-type (WT) and OsMet1-2 (CG methyltransferase mutant) rice lines revealed widespread differences in splicing variation (120). Consistent with the metazoan data (120), CG methylation was found to be higher in WT exons compared with adjacent introns, and was not solely dependent on the CG composition of exons and introns (120). Further evidence from cotton showed similar CG methylation levels in constitutive and alternative exons, but variable patterns during different fiber development stages (121). By contrast, CG methylation was higher in alternative introns than constitutive introns. Furthermore, differential CG methylation has a strong influence on nucleosome formation since constitutive exons displayed higher nucleosome occupancy than alternative exons. However, alternative exons exhibited higher nucleosome density than constitutive introns (121). These findings clearly demonstrate that the relationship between DNA methylation and nucleosome occupancy is conserved between animals and plants, and AS is also predominantly regulated at the chromatin level in plants (42,82,92).

The Crosstalk Between Epigenetic Mechanisms and Alternative RNA Processing Regulation

Review paper 2020

DNA methylation, resulting in 5' methylation of cytosine (5mC), is a conserved and heritable DNA modification that affects gene expression in a genome-wide manner (Li and Zhang, 2014). The impact of DNA methylation on gene expression varies depending on its genomic contexts. The role of promoter DNA methylation in gene expression has been well investigated, which is widely believed to cause transcriptional inhibition of downstream genes (Law and Jacobsen, 2010). Interestingly, recent studies in model plant Arabidopsis revealed that two SU(VAR)3-9 homologs, SUVH1 and SUVH3, bind to methylated DNA and recruit the DNAJ proteins to enhance proximal gene expression, thereby counteracting the repressive effects of transposon insertion near genes (Harris et al., 2018; Xiao et al., 2019; Zhao et al., 2019). Compared to DNA methylation in promoter regions, the function of genic DNA methylation remains elusive (Ball et al., 2009). During the last decade, several studies indicate that genic DNA methylation has a positive effect on the expression of associated genes and prevents spurious transcription initiation, and it is present within a number of cancer-related genes and has been regarded as a hallmark of human cancer (Baylin and Jones, 2011; Yang et al., 2014; Neri et al., 2017).

Recent studies reveal a strong correlation between DNA methylation and alternative splicing. Yang et al. (2014) showed that gene body DNA demethylation mediated by DNA methyltransferase inhibitor 5-aza-2'-deoxycytidine results in reduced efficiencies of transcription elongation or splicing. In human cells, Shukla et al. (2011) reported that a DNA-binding protein, called CCCTC-binding factor (CTCF), can promote inclusion of weak upstream exons by mediating local RNA polymerase II pausing. In this case, DNA methylation inhibits CTCF binding to CD45 exon 5, which enables Pol II to transcribe more rapidly, giving rise to an exon 5 exclusion (Ong and Corces, 2014). More recently, Nanavaty et al. (2020) further revealed that CTCF is a bifunctional regulator which influences both alternative splicing and alternative polyadenylation. Removal of DNA methylation enables CTCF binding and recruitment of the cohesin complex, which in turn form chromatin loops to promote proximal polyadenylation site usage. These works clearly demonstrate that DNA methylation has an important participation in RNA processing regulation. While, limited information is currently available regarding how DNA binding proteins disturb the elongation of Pol II. It reminded us that there maybe are other factors influencing Pol II elongation in CTCF-mediated AS regulation, like the cohesin complex.

Unlike CTCF protein which binds to unmethylated DNA, a growing number of studies have shown that the methyl cytosine-guanine dinucleotide (CpG) binding protein 2 (MeCP2) binds to methylated regions to influence AS. MeCP2 is the earliest reported multifunctional protein that contains both methyl-CpG-binding domains and transcriptional repressor domains (Nan et al., 1997). Acting as a chromatin adaptor, MeCP2 is attracted to 5mC on alternative exons, triggering its interaction with histone deacetylases (HDACs), which modulate alternative splicing (Maunakea et al., 2013). As we delve deeper into the function of MeCP2, it is becoming clear that MeCP2 recruiting splicing factors to regulate mRNA splicing is also a nearly ubiquitous mechanism in animals (Cheng et al., 2017; Wong et al., 2017).

In plants, the available information regarding whether gene body DNA methylation affects AS and the extent of this mediation is currently limited. The first study of DNA methylation-related functions in splicing was reported in maize (Regulski et al., 2013). More recently, the cytosine methyltransferase OsMET1 was found to affect global AS events in rice, in which a total of 6319 more events were identified with the met1 mutant compared with those associated with the wild-type strain (Wang et al., 2016). However, deeper research combining DNA methylation and AS/APA in plant is lacking. Whether it has the similar regulatory mechanism with mammals needs to be further elucidated.

Unlike animals, plants display a high degree of plasticity during growth and development. In plants, to overcome the constant challenge from a rapidly changing environment, specific adaptation mechanisms have been evolved, among which alternative RNA processing is an important strategy (Chaudhary et al., 2019). Recent work has indicated that the role of epigenetic modifications in regulating AS/APA under stress is emerging (Jabre et al., 2019). Temperature is one of the environmental signals that strongly affects plant development. An recent study indicated that temperature variation is memorized by chromatin via H3K36me3 modification, resulting in a specific splicing pattern, which enables a feasible adaptation to stress conditions (Pajoro et al., 2017). Another study showed that genes which are quickly activated under cold stress and differentially expressed at the splicing level, were found to be modified by H3K27me3 in non-stress conditions (Vyse et al., 2020). These reports suggest a dynamic regulation of temperature stress-responsive genes by alternative RNA processing and histone modification. In Arabidopsis, the Nuclear speckle RNA binding proteins (NSRs) have been known as regulators of AS functioning in auxin-associated developmental processes such as lateral root formation (Bazin et al., 2018). These proteins were shown to interact with specific alternatively spliced mRNA targets and at least with one structured lncRNA named ASCO (Bardou et al., 2014). The specific interaction of NSR with the ASCO is able to modulate AS patterns of a subset of NSR target genes, thereby impacting auxin response (Bazin et al., 2018). In other plants, specific association between epigenetic regulators and RNA processing factors under stress conditions has also been found. A maize SWI3D protein, ZmCHB101, has been found to impact alternative splicing contexts of a subset of osmotic stress-responsive genes on genome-wide level (Yu et al., 2019). In turn, alternative RNA processing of pivotal regulatory genes confers plants quick response to the changing climate conditions through alteration of reversible epigenetic marks. While, most of the current researches only focus on one aspect of how plants respond to changeable environment. That means, alternative RNA processing impacts the transcriptome of responsive genes or environment change leads to dynamic alterations of diverse epigenetic modifications (Rataj and Simpson, 2014; Calixto et al., 2018; Li et al., 2018). The mechanistic insights into the detailed interplay between epigenetic regulation and AS/APA in changing environment remains largely limited. In addition, the complicated regulatory mechanisms controlling mRNA isoform ratios in a tissue- or condition-specific manner still remain unclear.