Large-scale phenotyping of transgenic tobacco plants (Nicotiana tabacum) to identify essential leaf functions

Two of the major challenges in functional genomics are to identify genes that play a key role in biological processes, and to elucidate the biological role of the large numbers of genes whose function is poorly characterized or still completely unknown. In this study, a combination of large-scale expressed sequence tag sequencing, high-throughput gene silencing and visual phenotyping was used to identify genes in which partial inhibition of expression leads to marked phenotypic changes, mostly on leaves. Three normalized tobacco (Nicotiana tabacum) cDNA libraries were prepared directly in a binary vector using different tissues of tobacco as an RNA source, randomly sequenced and clustered. The Agrobacterium-tobacco leaf disc transformation system was used to generate sets of antisense or co-suppression transgenic tobacco plants for over 20 000 randomly chosen clones, each representing an independent cluster. After transfer to the glasshouse, transgenic plants were scored visually after 10-14 days for changes in growth, leaf form and chlorosis or necrosis. Putative hits were validated by repeating the transformation. This procedure is more stringent than the analysis of knockout mutants, because it requires that even a partial decrease in expression generates a phenotype. This procedure identified 88 validated gene/phenotype relations. These included several previously characterized gene/phenotype relationships, demonstrating the validity of the approach. For about one-third, a function could be inferred, but a loss-of-function phenotype had not been described previously. Strikingly, almost one-half of the validated genes were poorly annotated, or had no known function. For 77 of these tobacco sequences, a single or small number of potential orthologues were identified in Arabidopsis. The genes for which orthologues were identified in Arabidopsis included about one-half of the genes whose function was completely unknown. Comparison with published gene/phenotype relations for Arabidopsis knockout mutants revealed surprisingly little overlap with the present study. Our results indicate that partial gene silencing identifies novel gene/phenotype relationships, which are distinct from those uncovered by knockout screens. They also show that it is possible to perform these analyses in a crop species in which full genome sequence information is lacking, and subsequently to transfer the information to a reference species in which functional studies can be performed more effectively.