High-throughput experimental verification of predicted tissue-specific and tumor-specific splice isoforms

Alternative splicing of transcripts may lead to different mRNA species and therefore to potentially different proteins. Any failure or error in the splicing control mechanism can be involved in a number of pathological processes, such as cancer. Splice isoforms that are disease specific could therefore serve as excellent diagnostic markers, which are easily identifiable by PCR.

Computational prediction of alternative splice variants has been highly facilitating the identification of novel splice isoforms. Our prediction strategy is based on the genomic mapping (SpliceNest) of EST consensus sequences and library annotation provided in the GeneNest database. This revealed 427 genes with at least one tissue-specific transcript as well as 1120 genes showing tumor-specific isoforms. Out of these genes, a subset of predicted isoforms was experimentally verified by an RT-PCR screening approach. We have set up an experimental strategy that allows us to screen expression of genes in up to 112 different human tissues of multiple developmental stages and cell lines. Within this project, the electrophoretic separation of RT-PCR products turned out to be the bottleneck impeding the switch from a medium-throughput to a high-throughput strategy. To circumvent the limitations of DNA slab gel analysis, a laboratory prototype of an automated on-chip electrophoresis system that allows high-throughput analysis of DNA fragments was implemented in the workflow. In our experimental set-up, we analyzed RT-PCR samples on 4 × 96-well plates within a defined sequence of consecutive one-on-one measurements. The high-throughput experimental verification of computationally predicted tissue specific isoforms revealed a high success rate in confirming their expression in the respective tissue. However, low expression levels of the respective transcript and the limited sensitivity of the experimental method can explain failed detection of the restricted expression pattern.

The combination of computational prediction of alternative splicing events with high-throughput experimental verification facilitates the efficient detection of tissue-specific and tumor-specific transcripts.