Next generation sequencing for studying viruses and RNA silencing-based antiviral defense in crop plants

The main objectives of this work have been to use next generation sequencing (NGS) and develop bioinformatics tools for plant virus diagnostics and genome reconstruction as well as for investigation of RNA silencing-based antiviral defense. In virus-infected plants, the host Dicer-like (DCL) enzymes process viral double-stranded RNAs into 21-24 nucleotide (nt) short interfering RNAs (siRNAs) which can potentially associate with Argonaute (AGO) proteins and guide the resulting RNA-induce silencing complexes (RISCs) to target complementary viral RNA for post-transcriptional silencing and, in the case of DNA viruses, complementary viral DNA for transcriptional silencing. In the pioneering work, Kreuze et al. (2009) have demonstrated that an RNA virus genome can be reconstructed from multiple siRNA contigs generated by the short sequencing read assembler Velvet.
In this PhD study, we developed a bioinformatics pipeline to analyze viral siRNA populations in various model and crop plants experimentally infected with known viruses and naturally infected with unknown viruses. First, we developed a bioinformatics tool (MISIS) to view and analyze maps of small RNAs derived from viruses and genomic loci that generate multiple small RNAs (Seguin et al. 2014b). Using MISIS, we discovered that viral siRNAs cover the entire genomes of RNA and DNA viruses as well as viroids in both sense and antisense orientation without gaps (Aregger et al. 2012; Seguin et al. 2014a; Rajeswaran et al. 2014a, 2014b), thus allowing for de novo reconstruction of any plant virus or viroid from siRNAs. Then, we developed a de novo assembly pipeline to reconstruct complete viral genomes as single contigs of viral siRNAs, in which Velvet was used in combination with other assemblers: Metavelvet or Oases to generate contigs from viral redundant or non-redundant siRNA reads and Seqman to merge the contigs. Furthermore, we employed the mapping tool BWA and the map viewing tool IGV to verify the reconstructed genomes and identify a consensus master genome and its variants present in the virus quasispecies. The approach combining deep siRNA sequencing with the bioinformatics tools and algorithms, which enabled us to reconstruct consensus master genomes of RNA and DNA viruses, was named siRNA omics (siRomics) (Seguin et al. 2014a).
We utilized siRomics to reconstruct a DNA virus and two viroids associated with an emerging grapevine red leaf disease and generate an infectious wild type genome clone of oilseed rape mosaic virus (Seguin et al. 2014a). Furthermore, siRomics was used to investigate siRNA-based antiviral defense in banana plants persistently infected with six distinct banana streak pararetroviruses (Rajeswaran et al. 2014a) and rice plants infected with rice tungro bacilliform pararetrovirus (Rajeswaran et al. 2014b). Our results revealed that multiple host DCLs generate abundant and diverse populations of 21-, 22- and 24-nt viral siRNAs that can potentially associate with multiple AGO proteins to target viral genes for post-transcriptional and transcriptional silencing. However, pararetroviruses appear to have evolved silencing evasion mechanisms such as overexpression of decoy dsRNA from a short non-coding region of the virus genome to engage the silencing machinery in massive siRNA production and thereby protect other regions of the virus genome from repressive action of viral siRNAs (Rajeswaran et al. 2014b). Furthermore, despite massive production of 24-nt siRNAs, the circular viral DNA remains unmethylated and therefore transcriptionally active, while the host genome is extensively methylated (Rajeswaran et al. 2014b). These findings shed new light at the siRNA generating machinery of economically-important crop plants. Our analysis of plant small RNAs in banana and rice revealed a novel class of highly abundant 20-nt small RNAs with 5'-terminal guanidine (5'G), which has not been identified in dicot plants. Interestingly, the 20-nt 5'G-RNA-generating pathway does not target the pararetroviruses, which correlates with silencing evasion (Rajeswaran et al. 2014a, 2014b).
This thesis work is a part of the European Cooperation in Science and Technology (COST) action that aims develop an RNA-based vaccine to immunize crop plants against viral infection. Our analysis of viral siRNA profiles in various virus-infected plants allowed to identify the regions in the viral genomes that generate low-abundance siRNAs, which are the candidate regions to be targeted by RNA interference (RNAi). Our analysis of RNAi transgenic tomato plants confirmed that targeting of the low-abundance siRNA region of Tomato yellow leaf curl virus (TYLCV) by transgene-derived siRNAs renders immunity to TYLCV disease, one of the major constraints for tomato cultivation worldwide.