IC4R004-lncRNA-2016-26860696

Project Title

Functional analysis of long intergenic non-coding RNAs in phosphate-starved rice using competing endogenous RNA network

The Background of This Project

• Long intergenic non-coding RNAs (lincRNAs) may play widespread roles in gene regulation and other biological processes, however, a systematic examination of the functions of lincRNAs in the biological responses of rice to phosphate (Pi) starvation has not been performed. Here, we used a computational method to predict the functions of lincRNAs in Pi-starved rice. Overall, 3,170 lincRNA loci were identified using RNA sequencing data from the roots and shoots of control and Pi-starved rice.
• Inorganic phosphate (Pi) is essential for the growth and productivity of plants; however, those in agricultural environments can be exposed to Pi starvation 1 . Understanding the biological responses of plants to Pi starvation is vital for improving the efficiency of Pi use and maintaining an acceptable yield 2 . A number of studies have attempted to investigate the complex mechanisms regulating Pi homeostasis in rice, and have reported regulation at the transcript level 3–6 . Long integrate non-coding RNAs (lincRNAs) exist in both mammalian and plants and may play widespread roles in gene regulation and other biological processes 7–9 , however, the function of lincRNAs that response to Pi starvation are poorly understood.
• The competing endogenous RNA (ceRNA) theory has been proved and is now acknowledged widely 10,11 . This theory states that ceRNAs, including mRNA, lincRNAs, pseudogenes, and other microRNAs (miRNA) sponges, share common miRNA binding sites and can act as molecular sponges because the amount of a given miRNAs is limited 11 . LincRNAs compete with other miRNA sponges to play important roles in both plants and animals 9,12–15 . In addition, ceRNA networks are useful for studying cancer biology and other biological problems 16–19 . However, to our knowledge, ceRNA networks have not yet been used to study the functions of lincRNAs in plants such as Arabidopsis and rice.

Figure 1 The basic characteristics of lincRNAs in rice.

Figure 2 The ceRNA network of the rice root and shoot.

Research Findings

• The pipeline shown in Figure. 1a was used to identify lincRNAs from the RNA-seq data of rice undergoing Pi starvation 6 . In brief, if a longer-than-200 nt transcript with no coding capability is located in the intergenic regions and is not similar to known protein-coding genes, it is identified as a candidate lincRNA. The details of the pipeline are shown as follow.

• We compared the genomic features of the identified lincRNAs with those of protein-coding genes in rice. The mean exon length of the lincRNA was larger than that of the mRNA (Figure. 1b), while more than 70% of the lincRNAs, but less than 10% of the mRNAs, contained only one exon (Fig. 1c). In the meanwhile, lincRNAs in rice have fewer, but longer, exons than mRNAs 9 . The GC content of the lincRNAs was also lower than that of the mRNAs (Figure. 1d).
• Based on the miRNA-gene and miRNA-lincRNA relationships, a hypergeometric cumulative distribution function test was used to identify RNA pairs that may compete with each other for binding to the limited number of miRNA. For the rice root samples, Spearman correlation analysis was used to select ceRNA pairs in 27 Pi-starved samples; the resulting ceRNA network contained 31,794 ceRNA pairs. The network comprised 4,847 nodes (511 lincRNAs) with an average degree of 13.12, indicating that the network was very dense (Figure. 2a). The denseness of the network indicated the ceRNA phenomenon may be common in rice roots undergoing Pi starvation. In addition, the degrees of the nodes fit the power law distribution well, with a correlation of 0.91 and an R-squared value of 0.88 (Figure. 2b). A ceRNA network of the shoot was also generated from the RNA-seq data; this network comprised 4,979 nodes (376 lincRNAs) and 63,660 edges (Figure. 2c). The average degree of the nodes was 25.57, indicating that the ceRNA network of the shoot is denser than that of the root. As observed for the root, the degrees of the nodes in the shoot network fit the power law distribution well, with a correlation of 0.87 and an R-squared value of 0.87 (Figure. 2d). Taken together, these results indicate that the ceRNA networks for the two tissues were scale-free and had similar topologies; therefore, we were able to use the topological components, such as the communities and hubs, to investigate the biological significance of the networks.

Labs working on this Project

• National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, Wuhan 430070, P.R. China
• College of Informatics, Agricultural Bioinformatics Key Laboratory of Hubei Province, Huazhong Agricultural University, Wuhan 430070, P.R. China

Corresponding Author

• LingLing Chen (llchen@mail.hzau.edu.cn)