A Noncoding Polymorphism In A Gata-Containing Enhancer Element Drives The Association Of Lmo1 With Neuroblastoma

Background: Our ongoing genome-wide association study (GWAS) in neuroblastoma has identified multiple common and rare polymorphisms highly associated with disease susceptibility and phenotype. SNPs at the LMO1 locus showed one of the most robust associations and were associated with LMO1 overexpression and oncogenicity in preclinical models (Nature, 2011). Here we seek to define the DNA polymorphism(s) responsible for the GWAS association and determine the mechanism by which common variation determines tumor aggressiveness. Methods: Using 1000 Genomes data, we imputed all known variants within a 500-kilobase window around the LMO1 gene in a discovery set of 2,101 cases and 4,202 controls of European ancestry using the IMPUTE2 algorithm and performed association testing with SNPTEST. We further prioritized significant variants by evolutionary conservation based on comparative genome analysis of 46 mammalian species. Identification of non-coding variant regions with regulatory potential was performed through analysis of neuroblastoma-specific DNase I hypersensitivity mapping and chromatin immunoprecipitation (ChIP) sequencing data generated through the ENCODE consortium. Identification of canonical transcription factor binding motifs was performed using the MATCH-TM algorithm and JASPAR motif database. Independent replication of the top candidate causal SNP was performed by PCR-based genotyping in case series from Italy (420 cases, 751 controls) and the United Kingdom (369 cases, 1,108 controls). For statistical testing, homozygous and heterozygous odds ratios (OR) were computed from a comparison of (TT vs. GG) and (GT vs. GG) genotypes, respectively, and a single chi-squared p-value was computed from the 2x3 contingency table of phenotype vs. genotype. Results: The most significantly associated SNP at the LMO1 locus (rs2168101, homozygous OR=0.55, heterozygous OR=0.69, p-value=6.18x10-21) resides within an active enhancer region inferred by DNase I sequencing and histone modification ChIP-seq profiling in the SKNSH neuroblastoma cell line. Furthermore, rs2168101 lies within a GATA motif with perfect evolutionary conservation (5′-AGATAA-3′, phastCons score=100% across 46 mammalian species) that robustly binds to GATA2 and GATA3 transcription factors by ChIP-seq profiling. Consistent with an enhancer model, the risk “G” allele preserves the GATA motif and is associated with higher levels of LMO1 expression across a panel of 24 neuroblastoma cell lines (t-test p=0.047). Additionally, primary tumors with heterozygous G/T genotypes exhibit a greater degree of allele-specific expression imbalance relative to homozygous tumors by RNA-sequencing (t-test p=4.42x10-4), consistent with an effect in cis. In follow-up to imputation-based analysis, robust replication of the rs2168101 neuroblastoma association was observed by direct genotyping in an independent Italian cohort (homozygous OR=0.40, heterozygous OR=0.57, chi-squared p=9.94x10-6) and UK cohort (homozygous OR=0.31, heterozygous OR=0.51, chi-squared p=2.44x10-9). Furthermore, rs2168101 genotype was also associated with high-risk neuroblastoma and outcome (homozygous OR=0.67, heterozygous OR=0.74, p-value=2.16x10-4). Conclusions: Integrative genomic analysis provides mechanistic insight into how germline variation within non-coding DNA can impact neuroblastoma susceptibility and prognosis, supporting the hypothesis that the risk “G” allele at rs2168101 promotes oncogenic LMO1 overexpression through increased GATA transcription factor binding within an active enhancer region. Functional validation experiments are currently underway, including ChIP-PCR of the enhancer locus and targeted genome editing of the GATA binding site. Citation Format: Derek A. Oldridge, Andrew Wood, Cindy Winter, Maura Diamond, Ian Crimmins, Shile Zhang, Jun Wei, Javed Khan, Mario Capasso, Nazneen Rahman, Sharon J. Diskin, John M. Maris. A noncoding polymorphism in a GATA-containing enhancer element drives the association of LMO1 with neuroblastoma. [abstract]. In: Proceedings of the AACR Special Conference on Pediatric Cancer at the Crossroads: Translating Discovery into Improved Outcomes; Nov 3-6, 2013; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2013;74(20 Suppl):Abstract nr A27.