Proteomic Profiles Are Greatly Expanded Using de novo RNA-Seq Data: the Proteome of the Apogamous Fern Dryopteris affinis ssp. affinis

Performing proteomic studies on non-model organisms with little or no genomic information is still difficult. However, many specific processes and biochemical pathways occur only in species that are poorly characterized at the genomic level. For example, many plants can reproduce both sexually and asexually, the former allowing the generation of new genotypes and the latter their fixation. Thus, both modes of reproduction are of great agronomic value. However, the molecular basis of asexual reproduction is not understood in any plant. In ferns, it combines the production of unreduced spores (diplospory) and the formation of sporophytes from somatic cells (apogamy). To set the basis to study these processes, we performed transcriptomics by next- generation sequencing (NGS)1 and shotgun proteomics by tandem mass spectrometry in the apogamous fern D. affinis ssp. affinis. For protein identification we used the public viridiplantae database (VPDB) to identify orthologous proteins from other plant species and new transcriptomics data to generate a “species-specific transcriptome database” (SSTDB). In total 943 protein clusters with 6552 unique peptide sequences were identified (protFDR< 0.02). Such a proteogenomics approach, searching against an orthologue database concatenated to an SSTDB, revealed that, of all unique identified peptide sequences, >70% exclusively matched the SSTDB, while only ~19% exclusively matched the VPDB. The intersection of peptides identified in both databases is ~10%. At the protein cluster level, >73% (696 clusters) were exclusively identified in the SSTDB, while only 247 (26%) would be identified if one searched only against VPDB. With the increasing availability of genomic data from non-model species, this approach will improve the sensitivity in protein identification for species only distantly related to models. The benefit of using a SSTDB when working with non-model organisms is demonstrated here, as almost four-times more peptide sequences were identified with high confidence than when using only publically available data.