Research On RNA M6A Methylated Modification Of Rice Based On MeRIP-seq Data

RNA modifications, especially methylation of the N6 position of adenosine (A)-m6A, represent an emerging research territory in RNA biology. m6A was the most common modification in eukaryotic mRNAs, which was dynamically and reversibly regulated both by m6A:RNA methyltransferase (METTL3, METTL14, WTAP, etc) and demethylases (FTO, ALKBH5, etc). m6A modification mainly occurs at the consensus motif sequence of ’RRACH’(R=purine, H=A, C, or U), but we cannot exclude the probabilities of other motifs. The post-transcriptional RNA modifications play important roles, and they need some ’reader’ proteins to specifically recognize them and make them function in processes of mRNA export, transport, translation, degradation, alternative splicing, etc. With the development of MeRIP-seq (m6A-seq) technology based on combining high-throughout sequencing and immunoprecipitation, whole-genome wide and in-depth study of m6A distribution and function becomes feasible, but methods and software for MeRIP-seq data analysis were relatively scarce. Our studies in this paper were composed of two aspects:firstly, we developed and packaged a user-friendly data analysis pipeline MeRIP-PF; secondly, we took the lead in carrying out studies of construction and comparison of RNA m6A profiles in rice various tissues.MeRIP-PF is a special analysis pipeline for MeRIP-seq data, which can quickly identify m6A modification peaks. The basal principle of m6A peak identification by MeRIP-PF is searching for m6A signals by comparing distributions between control and IP data in genome/transcripts statistically. Firstly, reads from control and IP sequencing data are aligned to reference genome sequences, and reads within every split window (Bin=25bp) are counted; then MeRIP-PF provides a statistical p-value and FDR value for each identified m6A region based on the difference of read distribution when compared to the control data and FDR≤0.05 is set as a cut off to differentiate reliable m6A regions from the background. MeRIP-PF can also provide annotation information in the levels both of gene and peak, including size, genomic coordinates and enrichment score of m6A peaks. And MeRIP-PF produces some outputs in graphical format to show distribution characteristics of m6A in transcriptome-wide, distribution characteristics of transciptomic data in different genomic regions, examples of m6A modification in one individual gene, and correlations between m6A peak enrichment and modified gene abundances, MeRIP-PF is implemented in Perl and is freely available at http://software.big.ac.cn/MeRIP-PF.html.Based on MeRIP-seq technology, we for the first time carried out studies of RNA m6A profiles in rice various tissues (callus, leaf, and 4 embryos from different time points during seed germination). Firstly, we explored rice m6A distribution features in levels of mRNA, genes and chromosomes, respectively:(1) the distribution of m6A along mRNA was similar to that in mammals (human, mouse, etc), which was enriched in regions of CDS and 3’UTR; (3) the numbers of m6A-modified sites were varied widely among individual genes, with average 2 peaks in one gene. Correlation analysis showed that this unequal distribution might be associated with gene structure characteristics (exon number and the length of introns and genes) and gene abundance; (3) the distribution in chromosomes showed that m6A preferred appearing in telomeric regions, and became sparse towards to centromeric regions, which we called ’2-terminal hot’. Also, m6A distribution in different chromosomes showed different densities, and we inferred that this might be related to state and conformation of chromatins. Secondly, by comparing m6A profiles of rice callus and leaf tissues, we identified 626 and 5,509 SMGs (Selectively m6A-Modified Gene) respectively and explored the possible mechanism. Functional analysis showed that SMGs in callus and leaf were mainly involved in terms related to transcriptional regulation and photosynthesis, respectively. By comparison to known motifs of RBPs, we further found that the conserved motif -UGUAMM (M=A|C), which was very similar to PUM-binding motif, was significantly over-represented in SMGs of leaf. And gene expression analysis showed that six members of PUM-family genes were all down-regulated or nearly non-expressed in leaf tissue when compared to those in callus. Therefore, we proposed a ’competitive-binding’ model as below. Some cellular expressed RBPs (for example, PUM) may compete with m6A RNA methyltransferase, therefore hindering generation of m6A. However, if these RBPs remained in extremely low or non-expressed state in some special tissues or tissues of special stages, RNA methyltransferase would access to m6A sites without barriers, therefore leading to existence of SMGs. Finally, by combining analysis of rice embryos’ transcriptome and m6A methylome data in different germination stages, we found that many of regulation-related genes in early germination (genes related to hormone metabolism, genes encoding storage proteins and LEA proteins) had significantly changed m6A levels during germination, which suggested that m6A modification might be involved in regulation of rice early germination.In summary, in this study we first developed an easy-to-use package called MeRIP-PF for rapidly analyzing high-throughput m6A sequencing data, and equipped with MeRIP-seq technology we carried out studies of construction and comparison of RNA m6A profiles in rice various tissues, which fills gaps in m6A studies of plants, especially of rice. This work laid the foundation for the study of rice RNA epigenetic modification.