Abstract
In the Arab-Indian (AI) β-globin gene (HBB) haplotype of sickle cell anemia, fetal hemoglobin (HbF) levels are higher and the disease phenotype milder than in African HBB haplotypes. To study the genetic basis of HbF regulation in the AI haplotype, whole genome sequencing (WGS) was done in 14 unrelated Saudi AI haplotype HbS homozygotes; 7 selected for low HbF (7.9±1.9%) and 7 for high HbF (18.5±3.2%). WGS was also done in 3 Indians with the AI haplotype (HbF 26.0±4.5%), 3 African Americans with the Benin haplotype selected because of the uncommonly high HbF for this haplotype (19.8±0.4%) and 1 African American homozygote with the Senegal haplotype (HbF 16.0%). Association of SNPs with HbF was tested using simple linear regression with an additive genetic model, and the Sequence Kernel Association Test using sliding windows of 5 SNPs. We found 465 SNPs that were significantly associated with HbF in a single SNP analysis (p≤10-5). Following annotation using SNPper and ENCODE RegulomeDB, PLINK was employed to remove all SNPs in high linkage disequilibrium (r2>0.8) ending with 128 SNPs for follow-up analysis. These prioritized variants were further investigated using functional evaluation tools such as SCAN, Hembase, ErythronDB and the results of WGS were followed by genome-wide SNP analysis, imputation of genotypes to 1000genomes, and targeted direct genotyping in the following cohorts with sickle cell anemia: 110 AI haplotype cases (HbF 18.0±7.0%); 71 Saudi Benin haplotype cases (HbF 11.4±4.1%); 44 Indian AI haplotype homozygotes (HbF 23.0±4.8%); 894 African Americans with diverse HBB haplotypes (HbF 5.2±5.6%). All cases were examined when the HbF levels had stabilized. None of the patients were using hydroxyurea when HbF was measured. Based on WGS, 2 SNPs, rs4527238 and rs35685045 (D'=1) in intron 9 of ANTXR1 (anthrax toxin receptor 1, 2p13.1) remained associated with HbF after correction for multiple comparisons. This result was replicated in the cohort of 110 additional AI haplotype Saudi patients but not in cohorts with other HBB haplotypes. The alternative alleles of these 2 SNPs had nearly equal frequencies in Saudis with the Saudi AI haplotype and lower frequencies in Indians, Saudi Benin haplotype and African Americans with sickle cell anemia. The T allele of rs4527238 and the C allele of rs35685045 are associated with high HbF. ANTXR1 SNPs explained 10% of the HbF variability compared with 8% for BCL11A; these genes had independent, additive effects on HbF and explained about 15% of HbF variability. RNA sequencing and gene expression microarray analysis at 3 time points during the erythroid differentiation of CD34+ and iPSCs isolated from the same subjects showed that only the long splice variant of ANTXR1 was expressed. Expression increased over time in CD34+ erythroid cells and decreased over time in iPSC-derived erythroid progenitors in a pattern similar to BCL11A expression, with increasing expression as CD34+ cells matured and HbF expression fell and declining expression as iPSCs differentiated and HbF expression increased. Immunofluorescence co-staining for ANTXR1 and HbF during the erythroid differentiation of sickle iPSCs showed that reduced expression of ANTXR1 was paralleled by increased expression of HbF. By GTEx analysis, eQTLs at rs4527238 and rs35685045 showed a trend for increased expression according to genotype with rs4527238 T/T having the lowest level of expression and T/C and C/C genotypes, higher expression levels (p= 0.08) and for rs35685045, C/C the lowest expression levels and C/T and T/T higher expression levels (p= 0.07); rs4527238 and rs35685045 had no effect on the expression of BCL11A, a repressor of HbF expression located at 2p16 (p=0.8). STRING 9.1 predicted protein interactions of ANTXR1 directly with LRP6, HDAC2, BCLX and BRAC1 with the highest level of confidence (0.95) with LRP6. The HbF phenotype in AI haplotype sickle cell anemia is unique and ANTXR1 is the first trans-acting element associated with HbF whose effects appear to be limited to the Saudi AI haplotype. ANTXR1 might have dual effects on HbF, indirectly affecting hematopoiesis via interactions with LRP6 and Wnt signaling, and directly modulating HBG expression through its effects on HDAC2. A dual effect on HbF expression has been suggested for MYB. Whether targeting ANTXR1 is a therapeutic option will require further studies.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.