Abstract 1660: Lack of correlation between in silico projection of function for CYP19A1 and NFκB1 SNPs and qRT-PCR-determined gene expression levels in colon tissue

Abstract

Abstract The ability to link functional genetic variants with disease risk leads to advances in diagnostics and therapeutics. There are an estimated 10 million single nucleotide polymorphisms (SNPs), thus computerized projection programs that predict SNP-based alterations in functionality are used for genotyping experiments. We explored if projection programs could accurately predict functionality in colon samples. TagSNPs were identified for the CYP19A1 and NFκB1 genes, and normal colon tissue samples were genotyped for the tagSNPs and phenotyped using Taqman gene expression assays. Associations between SNPs and gene expression were noted and compared with projected outcomes from FastSNP (Academia Sinica). Of 25 CYP19A1 SNPs chosen for analysis, 10 were predicted to have no effect, 14 to alter transcription factor (TF) binding sites, and 1 to have a sense/synonymous coding change. Of those that were predicted to have no effect, one (rs17523880) showed significant changes in CYP19A1 expression levels (p=0.05 when adjusted for age, sex, and race). Two of the SNPs predicted to alter TF binding (rs1902584, rs6493497) showed a change in expression levels (p<=0.01), and there was no expression change in the SNP with synonymous coding change. In a case-control study exploring genetic associations in colon cancer and utilizing the same SNPs, three of the SNPs that were predicted to have no change (rs1961177, rs17523880, rs3751591) were found to be significantly associated with colon cancer (p<0.01, p=0.015, p<0.01). Of those, one (rs17523880) was associated with change in expression in our samples (p=0.05). In the case-control study, significant interactions between CYP19A1 and NFκB1 SNPs and colon cancer were reported. Thus, we also examined 11 NFκB1 SNPs, five of which were predicted to have no functional effect, and six of which were predicted to alter TF binding sites. Of the SNPs with no predicted functional consequences, three (rs3821958, rs1801, rs3774964) were associated with significant changes in gene expression (p<0.01 to p=0.04), and of those with projected TF binding site alterations, two were associated with a significant change in expression levels (rs4648090, p<0.01; rs3755867, p=0.04). Further, of two NFκB1 SNPs that interacted with CYP19A1 SNPs in the case control study, both were projected to have no effect on function, but were significantly associated with expression in our samples ((rs3821958, rs3774964; p<0.01). Future studies will include protein expression and enzymatic activity in these samples, and exploration in other tissues. In summary, while prediction programs are useful for choosing candidate SNPs for large scale assays, they cannot be relied on for accuracy, and functional effects of significant SNPs in disease association studies should be confirmed empirically in order to design treatment and prevention strategies. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 1660. doi:1538-7445.AM2012-1660