Genome-wide prediction and identification of cis-natural antisense transcripts in Arabidopsis thaliana

Background: Natural antisense transcripts (NAT) are a class of endogenous coding or non-protein-coding RNAs with sequence complementarity to other transcripts. Several lines of evidence have shown that cis- and trans-NATs may participate in a broad range of gene regulatory events. Genome-wide identification of cis-NATs in human, mouse and rice has revealed their widespread occurrence in eukaryotes. However, little is known about cis-NATs in the model plant Arabidopsis thaliana. Results: We developed a new computational method to predict and identify cis- encoded NATs in Arabidopsis and found 1,340 potential NAT pairs. The expression of both sense and antisense transcripts of 957 NAT pairs was confirmed using Arabidopsis full-length cDNAs and public massively parallel signature sequencing (MPSS) data. Three known or putative Arabidopsis imprinted genes have cis- antisense transcripts. Sequences and the genomic arrangement of two Arabidopsis NAT pairs are conserved in rice. Conclusion: We combined information from full-length cDNAs and Arabidopsis genome annotation in our NAT prediction work and reported cis- NAT pairs that could not otherwise be identified by using one of the two datasets only. Analysis of MPSS data suggested that for most Arabidopsis cis- NAT pairs, there is predominant expression of one of the two transcripts in a tissue-specific manner.