Abstract 3905: Identification of new tumor-specific splice variants in NSCL cancers by whole genome exon arrays

Although lung cancer is one of the major causes of cancer death in men and women worldwide, the molecular pathogenesis of this disease remains elusive. The contribution of alternative splicing (AS) to cancer pathogenesis and progression is emerging as an area of considerable interest. Cancer-specific splice variants seem to play a key role in disease mechanism and etiology. In this study, we analyzed 20 matched pairs of tumor specimens of Adenocarcinomas (AC), Squamous Cell Carcinomas (SCC) and adjacent normal tissue using Affymetrix GeneChip Exon microarrays. With this technology, it is possible to profile simultaneously gene expression and alternative splicing patterns for the entire genome using a single array. At the exon level, multiple probes for each exon enable the discrimination among different isoforms of a gene. Total RNA from tumors or normal tissue samples were amplified using the Ambion whole transcript expression kit and were hybridized onto GeneChip Human Exon 1.0 ST Array®. Data generated were normalized using RMA and were subjected to several layers of filtering (DABG filter, Multiple mRNAs filter). The delta splice index (ΔSI) was calculated, and alternative splicing events were detected using MIDAS and Rank Product method. Probesets were ranked based on their SI p-values. By comparing tumor samples to the corresponding normal subgroups, we generated a list of 1689 splice variant candidates for SCC and 1010 for AC showing significant differential expression between normal and tumor tissues. Top candidates in each tumor subtype were selected based on their ΔSI for manual inspection and further bioinformatics analysis. Study of alternatively spliced candidates revealed that approximately 30 % of the genes were detected in both types of cancer. Gene function analysis combining text mining and knowledge-based approaches indicated that the largest subset (70 %) of these genes was related to cancer. Within this group, 30% of genes were related to lung cancers. Interestingly the major part (85%) of variant candidates was not previously associated with AS events in cancer. We focused our attention on this latter group and, through validation by RT/PCR, confirmed the existence of new specific alternative splicing events occurring in each cancer type. In addition, we identified biological pathways and gene networks that are specifically altered in AC or SCC. The cancer-specific AS variants and pathways identified in our study are promising biomark