Population-level annotation of lncRNAs in Arabidopsis thaliana reveals extensive expression and epigenetic variability associated with TE-like silencing [ChIP-Seq]

Background Long non-coding RNAs (lncRNAs) are under-studied and under-annotated in plants. In mammals, lncRNA expression has been shown to be reaching the extent of protein-coding expression and be highly variable between individuals of the same species. Using A. thaliana as a model plant organism, we aimed to understand the true scope of lncRNA transcription across plants from different regions, characterize lncRNA natural expression variability, and study the causes of this variability. Results Using RNA-seq data spanning 499 natural lines and 4 different developmental stages to create a more comprehensive annotation of lncRNAs in A. thaliana, we found over 10,000 novel loci — three times as many as in the current public annotation. We showed that, 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. It was particularly prominent in intergenic lncRNAs, where TE-like sequences 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 their expression is highly variable between different lines and tissues. This high expression variability is largely caused by high structural and 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. We performed this ChIP-seq to analyze epignetic patterns and silencing of lncRNAs in Arabidopsis thaliana. plants from many different A.thaliana natural accessions (https://1001genomes.org/accessions.html) were grown at the growth chambers of GMI, Vienna, Austria. 14 accessions were used for ChIP-sequencing with replicates corresponding to different growing batches, or in the case of not enough material - pooling samples from two-three batches. The growing conditions were 21 degrees Celcius, long-day (16-hour light, 8-hour dark). Mature leaves (not the oldest and not the youngest) from 2-3 individual plants were collected for each accession and snap frozen in liquid nitrogen. Leaves were collected when the plants reached 14-16-leaf stage prior to the start of bolting. Frozen tissue was ground with a Retsch machine using steel beads. Chromatin immunoprecipitation was performed with a protocol adapted from [Yelagandula R et al, Cell 2014]. Briefly, 1-2 grams of ground frozen leaf tissue was fixated with formaldehyde at 4oC for 5 min, then nuclei were isolated and lysed, and chromatin was fragmented using Covaris E220 Focused-ultrasonicator. The input sample was then taken out and frozen at -20oC and 5 antibody reactions were processed together. The antibodies used were: H1 (Agrisera) - 3 microgram per reaction, H3K4me3 (Abcam) - 3 microgram per reaction , H3K9me2 (Abcam) - 4 microgram per reaction, H3K27me3 (Millipore) - 4 microgram per reaction, H3K36me3 (Abcam) - 4 microgram per reaction. The immunoprecipitation was performed using pre-washed Dynabeads Protein A magnetic beads (Invitrogen) and ran overnight. Afterwards the samples were washed, followed by elution, overnight reverse-crosslinking, RNAse A (Fermentas) treatment for 30 min at room temperature, and DNA isolation using Qiagen PCR purification kit Qiaquick with 0.3M Sodium Acetate. Next, ChIP-seq libraries were prepared from a half of the resulting sample (due to very low amounts we did not measure the DNA concentrations) with the NEBNext Ultra II DNA kit (New England Biolabs) according to the manufacturer’s protocol and sequenced with 100bp single-end read mode on Illumina NovaSeq 6000 machine.