Dissecting The Genetic Architecture Of Long Non-coding RNA In Maize

Long non-coding RNAs(lncRNAs)are the transcripts that are longer than 200 bp and can’t translate into any protein.LncRNAs act functionally as RNAs,rather than encoding protein,to regulate the transcription and various steps in the post-transcriptional processing of mRNAs by recruiting chromatin remodeling complexes or RNA-bingding protein.LncRNAs play critical roles in the signal transduction,stress response and development by regulating the expression of protein coding genes.However,the genetic architecture of lncRNAs and the regulatory network with genes is poorly understood.Here,we performed genome-wide association analysis to identify QTLs controlling lncRNA expression(referred to as lncQTLs)using the published RNA-seq data of 503 inbred lines.By integrating the information of genes that may be regulated by lncRNAs,we analyzed the function and regulatoion mechanism of lncRNAs.Besides,to dissect the role of lncRNAs during maize domestication,we analyzed the genome-wide cis-regulatory difference between maize and teosinte using RNA-seq data of F1 hybrids and parents for three tissues(leaf,ear and stem).The major results are as follows:1.Totally,187,849 transcripts were assembled,45,982 lncRNAs were identified and 40,737(88.59%)were novel transcripts.Based on their genomic positions relative to protein-coding genes,lncRNAs were classified into four main types,lncluding antisense lncRNAs(NATs),intergenic lncRNA(lincRNAs),overlapping lncRNAs(OLP)and intron contained lncRNAs(INT).LincRNAs and NATs were the predominant types.LncRNAs were relatively equally distributed along genome,and they were shorter in length and contained less exon numbers than protein-coding genes.Compared to protein-coding genes,lncRNAs exhibited lower expression level and showed stronger expression variation between genotypes.2.After quality control,263,199 SNPs were identified.Using the mixed linear model in the TASSLE,13,794 significantly associated eQTLs(lncQTL)for 7,367 lncRNAs were identified through GWAS.The averge number of lncQTL mapped for one lncRNAs is 1.87,indicating that lncRNAs exhibited a relatively simple genetic architecture.Besides,local-lncQTL explained more expression variation than distant-lncQTL.3.LncRNA tended to be co-mapped with its neighboring gene,same of these lncRNA maybe the regulator of neighboring gene.Consistent with this,35(94.60%)eRNAs were co-mapped with their nearby protein coding genes.In addition to regulating the expression of neighboring genes,lncRNAs are co-mapped with local-eQTL clusters and distant-eQTL clusters.These eQTL-clusters genes take part in growth and development,biotic and abiotic response.4.To dissect the role of lncRNAs during maize domestication,we analyzed the genome-wide cis-regulatory difference between maize and teosinte using RNA-seq data of F1 hybrids and parents for leaf,ear and stem tissues.We found that maize alleles were more highly expressed than teosinte alleles,suggesting domestication more frequently favored up-regulated lncRNAs.Furthermore,we showed that cis-regulated lncRNAs tended to be resided in regions under selection.These results indicated that the regulatory differences in lncRNA might play an important role in driving the transcriptome divergence during maize domestication.5.We analyzed the gene expression variation of a large maize-teosinte experimental population.We proved that the mapping bias can be corrected by using a higher mismatch parameter for alignment and constructing a shared transcriptome between parents.We further analyzed the distribution additive effects of eQTL for domestication and improvement genes,and found these genes showed a higher expression divergence than the whole genome gene.Next,we performed QTL mapping for gene expression variance(evQTLs).Interestingly,the overlap between evQTLs and previous eQTLs is only 34.29%,suggesting that gene expression variance appeared to be under relatively independent genetic control from gene expression level.By analyzing the parental contribution of evQTLs,we found that,-70%local-evQTL,teosinte allele showed higher expression variance,implying that selection more frequently favored stably expressed alleles during domestication.