Plants produce a vast array of specialized metabolites, many of which are used as pharmaceuticals, flavors, fragrances, and other high-value fine chemicals. However, most of these compounds occur in non-model plants for which genomic sequence information is not yet available. The production of a large amount of nucleotide sequence data using next-generation technologies is now relatively fast and cost-effective, especially when using the latest Roche-454 and Illumina sequencers with enhanced base-calling accuracy. To investigate specialized metabolite biosynthesis in non-model plants we have established a data-mining framework, employing next-generation sequencing and computational algorithms, to construct and analyze the transcriptomes of 75 non-model plants that produce compounds of interest for biotechnological applications. After sequence assembly an extensive annotation approach was applied to assign functional information to over 800,000 putative transcripts. The annotation is based on direct searches against public databases, including RefSeq and InterPro. Gene Ontology (GO), Enzyme Commission (EC) annotations and associated Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway maps are also collected. As a proof-of-concept, the selection of biosynthetic gene candidates associated with six specialized metabolic pathways is described. A web-based BLAST server has been established to allow public access to assembled transcriptome databases for all 75 plant species of the PhytoMetaSyn Project (www.phytometasyn.ca).

Background: Papaver somniferum (opium poppy) is the source for several pharmaceutical benzylisoquinoline alkaloids including morphine, the codeine and sanguinarine. In response to treatment with a fungal elicitor, the biosynthesis and accumulation of sanguinarine is induced along with other plant defense responses in opium poppy cell cultures. The transcriptional induction of alkaloid metabolism in cultured cells provides an opportunity to identify components of this process via the integration of deep transcriptome and proteome databases generated using next-generation technologies. Results: A cDNA library was prepared for opium poppy cell cultures treated with a fungal elicitor for 10 h. Using 454 GS-FLX Titanium pyrosequencing, 427,369 expressed sequence tags (ESTs) with an average length of 462 bp were generated. Assembly of these sequences yielded 93,723 unigenes, of which 23,753 were assigned Gene Ontology annotations. Transcripts encoding all known sanguinarine biosynthetic enzymes were identified in the EST database, 5 of which were represented among the 50 most abundant transcripts. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) of total protein extracts from cell cultures treated with a fungal elicitor for 50 h facilitated the identification of 1,004 proteins. Proteins were fractionated by one-dimensional SDS-PAGE and digested with trypsin prior to LC-MS/MS analysis. Query of an opium poppy-specific EST database substantially enhanced peptide identification. Eight out of 10 known sanguinarine biosynthetic enzymes and many relevant primary metabolic enzymes were represented in the peptide database. Conclusions: The integration of deep transcriptome and proteome analyses provides an effective platform to catalogue the components of secondary metabolism, and to identify genes encoding uncharacterized enzymes. The establishment of corresponding transcript and protein databases generated by next-generation technologies in a system with a well-defined metabolite profile facilitates an improved linkage between genes, enzymes, and pathway components. The proteome database represents the most relevant alkaloid-producing enzymes, compared with the much deeper and more complete transcriptome library. The transcript database contained full-length mRNAs encoding most alkaloid biosynthetic enzymes, which is a key requirement for the functional characterization of novel gene candidates.