Harnessing the genetic diversity engendered by alternative gene splicing

Our original objectives were to assess the unexplored dimension of alternative splicing as a source of genetic variation. In particular, we sought to initially establish an alternative splicing database for Arabidopsis, the only plant for which a near-complete genome has been assembled. Our goal was to then use the database, in part, to advance plant gene prediction programs that are currently a limiting factor in annotating genomic sequence data and thus will facilitate the exploitation of the ever increasing quantity of raw genomic data accumulating for plants. Additionally, the database was to be used to generate probes for establishing high-throughput alternative transcriptome analysis in the form of a splicing-specific oligonucleotide microarray. We achieved the first goal and established a database and web site termed Alternative Splicing In Plants (ASIP, http://www.plantgdb.org/ASIP/). We also thoroughly reviewed the extent of alternative splicing in plants (Arabidopsis and rice) and proposed mechanisms for transcript processing. We noted that the repertoire of plant alternative splicing differs from that encountered in animals. For example, intron retention turned out to be the major type. This surprising development was proven by direct RNA isolation techniques. We further analyzed EST databases available from many plants and developed a process to assess their alternative splicing rate. Our results show that the lager genome-sized plant species have enhanced rates of alternative splicing. We did advance gene prediction accuracy in plants by incorporating scoring for non-canonical introns. Our data and programs are now being used in the continuing annotation of plant genomes of agronomic importance, including corn, soybean, and tomato. Based on the gene annotation data developed in the early part of the project, it turned out that specific probes for different exons could not be scaled up to a large array because no uniform hybridization conditions could be found. Therefore, we modified our original objective to design and produce an oligonucleotide microarray for probing alternative splicing and realized that it may be reasonable to investigate the extent of alternative splicing using novel commercial whole genome arrays. This possibility was directly examined by establishing algorithms for the analysis of such arrays. The predictive value of the algorithms was then shown by isolation and verification of alternative splicing predictions from the published whole genome array databases. The BARD-funded work provides a significant advance in understanding the extent and possible roles of alternative splicing in plants as well as a foundation for advances in computational gene prediction.