A bioinformatics approach to distinguish plant parasite and host transcriptomes in interface tissue by classifying RNA-Seq reads

The genus Cuscuta is a group of parasitic plants that are distributed world-wide. The process of parasitization starts with a Cuscuta plant coiling around the host stem, and then haustorial organs establish a permanent vascular connection allowing for access to the phloem content. The host and the parasite form new cellular connections, suggesting coordination of developmental and biochemical processes. Simultaneous monitoring of gene expression in the parasite’s and host’s tissues may shed light on the complex events occurring between the parasitic and host cells and may help to overcome experimental limitation (i.e. how to separate host tissue from Cuscuta tissue at the haustorial connection). An alternative approach is to use bioinformatic analysis to classify sequencing reads as either belonging to the host or to the parasite and to characterize the expression patterns. Owing to the lack of a comprehensive genomic dataset from Cuscuta spp., such a classification has not been performed previously. We first performed dissection of RNA-Seq reads from an interface region between a non-model parasitic plant and a non-model host plant, Cuscuta japonica and Impatiens balsamina. Without established reference sequences, we classified reads into either of the plants by stepwise similarity search against de novo assembled transcript set of themselves, unigene sets of the same genus, and cDNA sequences of the same family, and then assembled de novo transcriptomes. We assessed the quality of the classification by mapping a given read to contigs of both plants, achieving a misclassification rate low enough (0.22-0.39%) to be used reliably for the analysis of differential gene expression. We then applied the read dissection method to RNA-Seq data from the interface between a non-model parasitic plant and a model host plant, C. japonica and Glycine max. Analysis of gene expression profiles at 5 parasitizing stages revealed differentially expressed genes from both C. japonica and G. max, and uncovered coordination of cellular processes between the two plants. We demonstrated that reliable identification of differentially expressed transcripts in undissected interface region of the parasite-host association is feasible and informative with respect to differential-expression patterns.