Multiple genomic alterations on 21q22 predict various TMPRSS2/ERG fusion transcripts in human prostate cancers

A number of TMPRSS2/ERG fusion transcripts have been reported since the discovery that recurrent genomic rearrangements result in the fusion of TMPRSS2 and ETS family member genes. In this article we present evidence demonstrating that multiple genomic alterations contribute to the formation of various TMPRSS2/ERG transcripts. Using allele-specific analysis of the data generated from the GeneChip 500K SNP array we observed both hemizygous and homozygous deletions occurring at different locations between and within TMPRSS2 and ERG in prostate cancers. The 500K SNP array enabled us to fine map the start and end of each deletion to specific introns of these two genes, and to predict a variety of fusion transcripts, including a new form which was confirmed by sequence analysis of the fusion transcripts in various tumors. We also inferred that translocation is an additional mechanism of fusion for these two genes in some tumors, based on largely diploid genomic DNA between TMPRSS and ERG, and different fusion transcripts produced in these tumors. Using a bioinformatics approach, we then uncovered the consensus sequences in the regions harboring the breakpoints of the deletions. These consensus sequences were homologous to the human Alu-Sq and Alu-Sp subfamily consensus sequences, with more than 80% homology. The presence/absence of Alu family consensus sequence in the introns of TMPRSS2 and ERG correlates with the presence/absence of fusion transcripts of theses two genes, indicating that these consensus sequences may contribute to genomic deletions and the fusion of TMPRSS2 and ERG in prostate cancer.