Abstract
Long noncoding RNAs (lncRNAs) have been characterized extensively in animals and are involved in several processes, including homeobox gene expression and X-chromosome inactivation. In comparison, there has been much less detailed characterization of plant lncRNAs, and the number of distinct lncRNAs encoded in plant genomes and their regulation by developmental and epigenetic mechanisms remain largely unknown. Here, we analyzed transcriptome data from Asian rice (Oryza sativa) and identified 6,309 long intergenic noncoding RNAs (lincRNAs), focusing on their expression in reproductive tissues and organs. Most O. sativa lincRNAs were expressed in a highly tissue-specific manner, with an unexpectedly high fraction specifically expressed in male gametes. Mutation of a component of the Polycomb Repressive Complex2 (PRC2) resulted in derepression of another large class of lincRNAs, whose expression is correlated with H3K27 trimethylation in developing panicles. Overlap with the sperm cell-specific lincRNAs suggests that epigenetic repression of lincRNAs in the panicles was partially relieved in the male germline. Expression of a subset of lincRNAs also showed modulation by drought in reproductive tissues. Comparison with other cereal genomes showed that the lincRNAs generally have low levels of conservation at both the sequence and structural levels. Use of a novelty detection support vector machine model enabled the detection of nucleotide sequence and structural homology in ∼10% and ∼4% of the lincRNAs in genomes of purple false brome (Brachypodium distachyon) and maize (Zea mays), respectively. This is the first study to report on a large number of lncRNAs that are targets of repression by PRC2 rather than mediating regulation via PRC2. That the vast majority of the lincRNAs reported here do not overlap with those of other rice studies indicates that these are a significant addition to the known lincRNAs in rice.