Maize transposable elements contribute to long non-coding RNAs that are regulatory hubs for abiotic stress response

Yuanda Lv,Fengqin Hu,Yongfeng Zhou,Feilong Wu,Brandon S Gaut

doi:10.1186/s12864-019-6245-5

Abstract

BackgroundSeveral studies have mined short-read RNA sequencing datasets to identify long non-coding RNAs (lncRNAs), and others have focused on the function of individual lncRNAs in abiotic stress response. However, our understanding of the complement, function and origin of lncRNAs – and especially transposon derived lncRNAs (TE-lncRNAs) - in response to abiotic stress is still in its infancy.ResultsWe utilized a dataset of 127 RNA sequencing samples that included total RNA datasets and PacBio fl-cDNA data to discover lncRNAs in maize. Overall, we identified 23,309 candidate lncRNAs from polyA+ and total RNA samples, with a strong discovery bias within total RNA. The majority (65%) of the 23,309 lncRNAs had sequence similarity to transposable elements (TEs). Most had similarity to long-terminal-repeat retrotransposons from the Copia and Gypsy superfamilies, reflecting a high proportion of these elements in the genome. However, DNA transposons were enriched for lncRNAs relative to their genomic representation by ~ 2-fold. By assessing the fraction of lncRNAs that respond to abiotic stresses like heat, cold, salt and drought, we identified 1077 differentially expressed lncRNA transcripts, including 509 TE-lncRNAs. In general, the expression of these lncRNAs was significantly correlated with their nearest gene. By inferring co-expression networks across our large dataset, we found that 39 lncRNAs are as major hubs in co-expression networks that respond to abiotic stress, and 18 appear to be derived from TEs.ConclusionsOur results show that lncRNAs are enriched in total RNA samples, that most (65%) are derived from TEs, that at least 1077 are differentially expressed during abiotic stress, and that 39 are hubs in co-expression networks, including a small number that are evolutionary conserved. These results suggest that lncRNAs, including TE-lncRNAs, may play key regulatory roles in moderating abiotic responses.

Highlights

Several studies have mined short-read RNA sequencing datasets to identify long non-coding RNAs, and others have focused on the function of individual lncRNAs in abiotic stress response
Bringing these diverse analyses together, we identify several lncRNAs that are hubs in co-expression networks that respond to abiotic stress and show that several of these hubs are lncRNAs derived from Transposable element (TE)
Among the 23,309 lncRNA candidates, 59.3% were identified from polyadenylated RNAseq samples, and the remaining 40.7% were from total RNA samples, representing potential polyA- transcripts (Table 1; hereafter we refer to lncRNAs from total RNAs as polyA- for simplicity)

Summary

Introduction

Several studies have mined short-read RNA sequencing datasets to identify long non-coding RNAs (lncRNAs), and others have focused on the function of individual lncRNAs in abiotic stress response. Much of the additional function is encoded by RNAs, which vary in size from small RNAs (sRNAs) of< 25 nucleotides (nt) in length, to tRNAs of 70 to ~ 90 nt in length, to an even larger class of long non-coding RNAs (lncRNAs). Microarrays have been used to detect 6480 lncRNAs from Arabidopsis thaliana [4]; single-stranded RNA sequence data have led to the identification of 2224 lncRNA transcripts in rice (Oryza sativa) [5]; and total RNAseq data have been employed to detect 7245 lncRNAs in maize Microarrays have been used to detect 6480 lncRNAs from Arabidopsis thaliana [4]; single-stranded RNA sequence data have led to the identification of 2224 lncRNA transcripts in rice (Oryza sativa) [5]; and total RNAseq data have been employed to detect 7245 lncRNAs in maize (Zea mays ssp. mays) [6]

Methods

Results

Conclusion