Population-level annotation of lncRNA transcription in Arabidopsis reveals extensive variation associated with transposable element-like silencing

Abstract

Background Long non-coding RNAs (lncRNAs) are under-studied and under-annotated in plants. In mammals, lncRNA loci are nearly as ubiquitous as protein-coding genes, and their expression has been shown to be highly variable between individuals of the same species. Using A. thaliana as a model, we aimed to understand the true scope of lncRNA transcription across plants from different regions and study its natural expression variability. Results Using RNA-seq data spanning hundreds of natural lines and several developmental stages to create a more comprehensive annotation of lncRNAs, we found over 10,000 new loci - three times as many as in the current public annotation. While lncRNA loci are ubiquitous in the genome, most appear to be actively silenced and their expression and repressive chromatin levels are extremely variable between natural lines. This was particularly common for intergenic lncRNAs, where pieces of transposable elements (TEs) present in 50% of the loci are associated with increased silencing and variation, and such lincRNAs tend to be targeted by TE silencing machinery. Conclusion lncRNAs are ubiquitous in the A. thaliana genome but largely silenced, and their expression is highly variable between different lines. This high expression variability is largely caused by high epigenetic variability of non-coding loci, especially those containing pieces of transposable elements. We create the most comprehensive A. thaliana lncRNA annotation to date and improve our understanding of plant lncRNA biology.

Long non-coding RNAs contribute to DNA damage resistance inArabidopsis thaliana

“…Analyzing expression data from multiple accessions (Kawakatsu et al, 2016; Kornienko et al, 2023) generated from soil-grown plants without genotoxic stress indicated absence of transcripts of all three lncRNAs in the reference accession Col-0 seedlings (Supplemental Figure 5 A) but expression in seedlings from some other accessions under the same conditions. This is rare for lncRNA B but most common for lncRNA C (Figure 5 D).…”

“…To find the gene copies, the blastn results were filtered for sequences with >80% sequence identity and >80% length match to the TAIR sequence, allowing for insertions of up to 1.5 kb to account for possible TE insertions in the non-reference accession genomes. For gene expression calculation, we used RNA-seq data from mature leaves (Kawakatsu et al, 2016, GEO accession number GSE80744) and RNA-seq data from 7-day-old seedlings, 9-leaf rosettes, flowers (with flower buds), and pollen (Kornienko et al, 2023, GEO accession number GSE226691). Raw RNA-seq reads were mapped to the TAIR10 genome using STAR (v.2.7.1) and exonic read counts were calculated using feature counts software from the subread package (v.2.0.0).…”

“…The best-fit models of evolution were obtained by ModelFinder as implemented in IQ-TREE (Kalyaanamoorthy et al, 2017) using BIC. (D) Percent of A. thaliana natural accessions that express (TPM>0.5) lncRNAs B, C or D. The ratios were calculated from RNA-seq data from seedlings, rosettes at the 9-leaf stage from 25 accessions, flowers and pollen from 23 accessions (Kornienko et al, 2023), and leaves from mature pre-bolting rosettes from 461 accessions (Kawakatsu et al, 2016).…”

Alternative silencing states of Transposable Elements in Arabidopsis

“…1A and EV1A). To determine their abundance in the genome, we defined four classes of TEs based on the length coverage of H3K27me3 marks and CG methylation, using these datasets and previously published methylomes 37 (Fig. 1B and EV1B).…”

“…H3K27me3 and H2AZ (GSE103361) 40 , ref6 elf6 jmj13 H3K27me3 ChIP-seq (GSE106942) 43 and Ecotypes related data (GSE224761) 37 .…”