Fetal Hemoglobin (HbF) in Sickle Cell Anemia: Genome-Wide Association Studies Using Pooled DNA Samples Can Reveal Genetic Associations with HbF Concentration.

The complex regulatory environment controlling HbF levels provides ample opportunity for modulating expression of this key modifier of sickle cell anemia (HbSS). Identifing genetic modulators by single gene or biochemical pathway-based candidate gene association studies is limited by the a priori hypothesis that they are responsible for modulation of HbF expression and thus, fail to capture the totality of possible modulators. In contrast, genome-wide association (GWA) studies using hundreds of thousands of SNPs across the entire genome can capture an unbiased assessment of gene association or linkage disequilibrium (LD) with a phenotype. A disadvantage of GWA is the cost when analyzing thousands of subjects. We developed a GWA screening strategy using pooled DNA samples followed by analysis using the Sentrix® HumanHap300 genotyping beadchip that includes over 317,000 tag SNPs, many derived from the Phase I of the HapMap project. These SNPs are within 10kb of a gene, or in regions that are evolutionary conserved, and incorporate the majority of the variation in regions of the genome with high LD by capturing approximately 80% of the loci genotyped in HapMap Phase I and II. DNA concentration for pooling was measured by RNAase-P assays and each pool was run in duplicate. Pools of approximately 60 DNA samples each were analyzed; pools included age, sex and β-globin gene haplotype-matched HbSS patients with the 1st and 4th quartiles of HbF levels. Raw data were processed to compute initial estimates of the allele frequencies. Estimates were normalized and SNPs with a minor allele frequency >0.15 were analyzed for allelic association using a Bayesian approach. Our analysis integrated the measure of association based on the allele frequencies, estimated from the pooled DNA samples, with the strength of LD between consecutive SNPs, information on minor allele frequencies derived from the HapMap project and an estimate of the variability between pools based on preliminary experimental data. By leveraging on this exogenous information we could reduce false positive associations without imposing unnecessarily stringent corrections for multiple comparisons that may lead to substantial loss of power. We identified more than 100 novel genes potentially involved with regulation of HbF and with roles in signal transduction, cell adhesion, and regulation of transcription. These genes are distributed unevenly among the whole genome, and chromosome 8 appears to be the most "enriched" for genes associated with regulation of HbF. Consistently with previous findings, our analysis confirms an involvement of genes in 6q22, 8q12 and 11p15.5 in the regulation of HbF. For further validation, we compared allele frequencies based on pooled DNA samples with those from individual genotyping of several SNPs. The high degree of agreement suggests that GWA based on pooled DNA samples can provide a cost-effective approach to the identification of the full array of genes responsible for HbF expression and can identify new candidates that must be further examined in follow-up genetic and biological studies.