Abstract
Abstract 821
Fetal hemoglobin (HbF) is the major modulator of sickle cell anemia (SCA, homozygosity for HBB glu6val) severity. In a genome-wide association study in African Americans with SCA we sought to identify single nucleotide polymorphisms (SNPs) associated with HbF concentrations. A discovery sample of 848 African American subjects and a primary replication study of 305 subjects were examined. DNA was genotyped with the Illumina Human610-Quad SNP; some replication set samples were genotyped with the Sentrix HumanCNV370 or the 317K array. For quality control we excluded SNPs with a call rate less than 95%; we excluded subjects with a call rate less than 93%; identity by descent measurements were computed to identify related individuals who were removed from analysis; we inferred gender using chromosome X SNPs removing subjects with gender mismatches; a genome-wide principal components analysis found no association between the phenotype and the first 10 principal components, indicating that the results were not affected by population substructure. The association between HbF and the genotype for each SNP was tested with a multiple linear regression analysis adjusting for sex and assuming an additive model using the software PLINK. SCA is a rare disease in developed countries and assembling large data sets is not feasible. Therefore, true associations with limited effect sizes might not reach “genome-wide” significance of 10-08. To identify genes enriched with moderately strong associations, we developed a SNP set enrichment analysis (SSEA) that computes the probability that a set of SNPs is selected as significant by chance and scores each gene by this probability. Two SNPs exceeded the strict genome-wide significance: SNP rs5006884 in a novel region on chromosome 11 upstream of the β-globin gene cluster locus control region (LCR) containing the olfactory receptor (OR) genes OR51B5 and OR51B6; SNP rs766432 in BCL11A, previously found to be associated with HbF in several different populations. Data for SNPs common to the discovery and replication sets were combined and analyzed jointly. Similarity of the regression beta coefficients across datasets and increased significance of the p-values compared with those observed in the analyses of individual datasets provide additional evidence that the associations were consistent in the both datasets. The Q-Q plot and a genomic inflation factor of 1.003 both suggest that the test statistics are not inflated and are distributed appropriately. SSEA identified 2 OR genes (OR51B5, OR51B6) and BCL11A as enriched in both the discovery and replication sets. The most significant SNP in the OR region (rs5006884) and BCL11A (rs766432) explained 15.6% of the variability in HbF. Also, in the interval Xp 22.2-22.3 we found moderate, but not “genome-wide” significance for 1 SNP in Xp22.2. Phylogenetic conservation of some OR genes and their flanking sequences suggests that this region might also have a role in controlling expression within the β-globin gene-like complex. Low linkage disequilibrium between SNPs in the β-globin locus and the OR genes suggests that one or more variants in the OR genes independently regulate HbF. The top SNP in the OR51B5/OR51B6 locus, rs5006884, was still associated with HbF (p = 1.5E-05) in a model adjusting both for sex and rs2071348, a SNP in tight LD with the HBG2 5' -158 C-T SNP, giving further evidence that the OR region provides important information in addition to the SNPs in the β-globin gene-like complex. Polymorphisms in the upstream OR region might conceivably modulate HbF levels by altering chromatin structure within the β-globin gene cluster. Conserved binding sites for the transcription factor CTCF flank the β-globin gene cluster and evidence suggests that these sites function as insulators. Polymorphisms in this region might affect the actions of enhancers, possibly through their effects on CTCF binding its receptors, thereby affecting the interaction of the globin genes with enhancers in the OR regions.
Disclosures:
No relevant conflicts of interest to declare.
Expression of Human Globin Genes in Transgenic Mice Carrying the ?-Globin Gene Cluster with a Mutation CausingG??+Hereditary Persistence of Fetal Hemoglobin