Identifying putative breast cancer-associated long intergenic non-coding RNA loci by high density SNP array analysis

Recent high-throughput transcript discoveries have yielded a growing recognition of long intergenic non-coding RNAs (lincRNAs), a class of arbitrarily defined transcripts (>200 nt) that are primarily produced from the intergenic space. lincRNAs have been increasingly acknowledged for their expressional dynamics and likely functional associations with cancers. However, differential gene dosage of lincRNA genes between cancer genomes is less studied. By using the high-density Human Omni5-Quad BeadChips (Illumina), we investigated genomic copy number aberrations in a set of seven tumor-normal paired primary human mammary epithelial cells (HMECs) established from patients with invasive ductal carcinoma. This Beadchip platform includes a total of 2,435,915 SNP loci dispersed at an average interval of ~700 nt throughout the intergenic region of the human genome. We mapped annotated or putative lincRNA genes to a subset of 332,539 SNP loci, which were included in our analysis for lincRNA-associated copy number variations (CNV). We have identified 122 lincRNAs, which were affected by somatic CNV with overlapped aberrations ranging from 0.14% to 100% in length. lincRNA-associated aberrations were detected predominantly with copy number losses and preferential clustering to the ends of chromosomes. Interestingly, lincRNA genes appear to be less susceptible to CNV in comparison to both protein-coding and intergenic regions (CNV affected segments in percentage: 1.8%, 37.5%, and 60.6%, respectively). In summary, our study established a novel approach utilizing high-resolution SNP array to identify lincRNA candidates, which could functionally link to tumorigenesis, and provide new strategies for the diagnosis and treatment of breast cancer.