Fetal hemoglobin in sickle cell anemia: Saudi patients from the Southwestern province have similar <i>HBB</i> haplotypes but higher HbF levels than African Americans

Abstract

Patients with sickle cell disease (SCD) from the Southwestern (SW) Province of Saudi Arabia have variable fetal hemoglobin (HbF) levels and have HBB gene cluster haplotypes of African origin. We studied 77 patients, aged 17.7 ± 10 (range 4–46) years (69% HbS homozygotes and 31% HbS-β0 thalassemia), to determine the associations of known HbF quantitative trait loci (QTL) with HbF concentration. HBB gene cluster haplotypes were 74% Benin, 22% Bantu, and 4% others. Genotyping Single nucleotide polymorphism (SNPs) in BCL11A, HBS1L-MYB, and OR51B5/6 showed that BCL11A was the sole QTL associated with HbF level. We compared these findings with two studies of African American with SCD. After adjusting for the BCL11A genotype, Saudi cases from the SW Province had HbF levels almost twice that of African Americans (P < 0.0001). When we examined the genetic population structure of the African Americans and Saudi patients using genome-wide data, we found that African Americans were similar to Yoruban, Mandenka, and Bantu Africans while Saudi patients resembled Arab populations. The commonality of HBB haplotypes coupled with the genetic distance between these populations suggests that genetic modifiers remote from the HBB cluster or unknown environmental influences are likely to account for the higher HbF in these Saudi patients. Fetal hemoglobin (HbF) is a known genetic modifier of sickle cell disease (SCD). High HbF levels decrease the incidence and severity of vaso-occlusive acute painful episodes, acute chest syndrome, and osteonecrosis [1]. Moreover, HbF is associated with increased survival [2]. HbF levels vary among patients with SCD and its expression is influenced by genetic [3, 4] and epigenetic factors [5, 6]. Three quantitative trait loci (QTL) play a major role in determining HbF expression and account for about 30% of the variability in HbF [7]. BCL11A and HBS1L-MYB are associated with HbF expression in African Americans with SCD [8, 9]. In addition, a region on chromosome 11 that harbors olfactory receptor gene cluster was recently shown to be a potential modifier of HBG expression in SCD [10]. SCD is common in Saudi Arabia with the highest incidence in the Eastern and Southwestern (SW) provinces [11, 12]. Patients from the SW Province have many complications of this disease and have phenotypic differences from African Americans that include a high prevalence of splenomegaly, rare CNS disease, and absence of leg ulcers [13-15]. In contrast, Eastern Province patients have a milder phenotype that is related to a nearly uniformly high HbF level that is associated with the Saudi–Indian haplotype of the HBB gene-like cluster [16]. SW Province patients have variable HbF levels and have HBB haplotypes similar to patients of African descent. Since the effect of known HbF QTLs in Saudi SW patients is unclear, we tested the associations of known HbF QTL with HbF concentration. As, Saudi SW and African American patients have nearly identical distribution of HBB gene cluster haplotypes, we compared HbF levels and the genetic population structure of these two different ethnic groups. Seventy-seven Saudi patients, aged ≥4 years, with SCD were studied. Fifty-three were HbS homozygotes and 24 had HbS-β0 thalassemia. Their age was 17.7 ± 10 years (range 4–46 years) and 38 (49%) were male. Mean hemoglobin level was 8.5 ± 1.7 g/dL (range 3.4–14.5 g/dL), mean corpuscular volume was 82 ± 11 fL (range 62–118 fL), and mean HbF was 11.9 ± 7.1% (range 1.7–34.6%). HbF in HbS homozygotes was 12.5 ± 6.4% (range 1.7–28.7%) and in HbS-β0 thalassemia, 10.5 ± 8.5% (range 2.4–34.6%) These differences were not significant (P = 0.24). Forty-five percent of Saudi patients took hydroxyurea. HBB gene cluster haplotypes were 57 (74%) Benin, 17 (22%) Bantu, 1 (1.3%) Senegal/Bantu, 1 (1.3%) Senegal/Benin, and 1 (1.3%) Senegal. Genotypes of HbF QTL SNPs are summarized in Table I. Minor allele frequency (MAF) for BCL11A (rs4671393, rs766432) was 0.35; however, MAF was less frequent (0.05–0.17%) for SNPs in HBS1L-MYB (rs28384513, rs9399137, rs4895441) and OR51B5/6 (rs5006884). BCL11A was the sole QTL associated with HbF whereas the HBS1L-MYB and OR51B5/6 loci had no effect. All HbF QTLs showed a similar trend in their effects on HbF level when patients on hydroxyurea were excluded from analysis (data not shown). Compared with African Americans with similar HBB haplotypes, and after adjusting for the BCL11A genotype (rs766432), Saudi cases with HbSS from SW Province, who were not on hydroxyurea, had HbF levels approximately twice that of African Americans in both the Cooperative Study of Sickle Cell Disease (CSSCD) and the Multicenter Study of Hydroxyurea (MSH) (Table II). Figure 1 shows the principal component analysis (PCA) and reports the Fst statistic, a measure of genetic distance, between the Saudi and African Americans patients and populations worldwide from the Human Genome Diversity Project. SW Province patients resembled Arab populations (Fst = 0.011–0.019) while African American cases were similar to Yoruban, Mandenka, and Bantu (Fst = 0.006–0.010) African populations. Despite similar HBB haplotypes, the differentiation between the Saudi SW and the African American patients was substantial (Fst = 0.096). Genetic population structure of Saudi SW patients. A. Top two principal components (PC) from the analysis of the genetic population substructure. Each point represents a subject and is colored by population membership. The Saudi SW populations are in close proximity to the Bedouin, Druze, and Palestinian populations while the CSSCD are closer to the African populations. B. The Fst statistic, a measure of genetic distance with higher values indicating higher differentiation, between the Saudi SW, African Americans, and populations worldwide. HbF is the major genetic modulator of SCD [17, 18]. Saudi patients from the SW Province have variable HbF levels and heterogeneities of their SCD phenotypes like most other ethnicities with SCD [1, 14, 15, 19]. Genetic modifiers of HbF level in Saudi SW patients have not been reported, perhaps because of the focus on studying Saudi SCD patients from the Eastern Province who carry the Saudi–Indian HBB haplotype and have much higher HbF levels and a milder SCA phenotype [15, 16, 20, 21]. The HBG2 C-T Xmn1 restriction site polymorphism site at position −158 5′ to the Gγ-globin gene (rs7482144) that was associated with increased HbF in patients with the Senegal haplotype is present in Eastern Province patients but not in typical Saudi SW patients [22]. However, the HbF in Eastern Province patients with the Saudi–Indian haplotype is far higher than in patients with the Senegal haplotype [23, 24]. HbF levels in Saudi SW patients are variable and we identified BCL11A as an important HbF QTL associated with HbF level in this population. Similar findings were reported in African Americans with SCD [8, 9]. However, contrary to African Americans, HBS1L-MYB and OR51B5/6 had no influence on HBG expression in Saudi SW patients [8-10]. This may be related in part to the lower MAF of SNPs in these genes and insufficient power to detect an effect on HbF level. Saudi SW patients have baseline HbF level approximately twice that of African Americans suggesting that other genetic modifiers or unknown environmental factors might modulate their HbF level [13-15]. We determined the genetic population structure of Saudi SW patients and African Americans with similar HBB haplotypes. The similarity of HBB haplotypes and the geographic proximity between Saudi SW region and Africa could indicate that these populations are closely related. However, there is a high level of differentiation between the Saudi SW and African American patients, and the Saudi SW individuals are similar to other Arab groups. Millennia ago, the HbS mutation had several independent origins in Africa, the Middle East, and Asia and spread to other populations by gene flow from wars, slave trading, and population migration [25-29]. Gene flow from Africa to Saudi Arabia likely introduced the HbS mutation into the SW Province [25]. The HbS mutation on typical African-derived haplotypes in Saudi SW Arabs suggests that following its introduction, the HbS gene was established locally because of the selective pressure of Plasmodium falciparum malaria and during many generations, admixture of HbS carriers with the indigenous Arabs took place. As a result of this population admixture, over time, perhaps, alternative means of regulating HBG expression on African haplotypes became established. As in most other populations, BCL11A is an important HbF QTL that influences HbF level in Saudi SW patients. These individuals have approximately twice the level of HbF compared with African American patients. The commonality of HBB haplotypes in Saudi and African Americans coupled with the genetic distance between these populations suggests that genetic modifiers remote from the HBB cluster or unknown environmental influences are likely to account for the higher HbF in these Saudi patients. HbF levels were determined using capillary electrophoresis in patients diagnosed with homozygosity for HBB glu6val (sickle cell anemia, HbSS) and HbS-β0 thalassemia, aged more than 4 years, and with HBB gene cluster haplotypes commonly observed in SW Province (i.e., Benin, Bantu, and Senegal). HBB gene cluster haplotype was determined using restriction fragment length polymorphism technique as previously described [30]. We used the following enzymes: XmnI, HincII, HindIII, and HinfI. We further confirmed the absence of the CT polymorphism (rs7482144), at position –158 of HBG2 (XmnI site), except in Senegal haplotype, using Taqman genotyping assay. We used the lowest HbF level after age 4 years, the age where HbF levels stabilized, for analysis. At the time of blood collection patients were untransfused and had no HbA in their blood. The study was approved by our institutional review board (No. 09-688). For HbF level comparison, we selected HbSS cases not taking hydroxyurea from two studies of African American with SCD—Cooperative Study of Sickle Cell Disease (CSSCD; N = 848) and Multicenter Study of Hydroxyurea (MSH; N = 224). We compared these cases with Saudi patients with HbSS homozygotes not on hydroxyurea (N = 29). DNA samples from Saudi patients were genotyped with TaqMan SNP genotyping assays, and included SNPs in BCL11A (rs4671393, rs766432), HBS1L-MYB (rs28384513, rs9399137, rs4895441), and OR51B5/6 (rs5006884). For each SNP analysis: 10 μl of PCR Mix containing TaqMan Universal PCR Master Mix (Applied Biosystem), the 6-FAM, and VIC probes assay mix (Applied Biosystem) and 20 ng of DNA were loaded on a 96-well plates. Genotyping results were visualized on TaqMan 7000. Allelic discrimination was determined using SDS v 1.2 software. A negative control with no DNA was added for each SNP analyses. Given that Saudi and African American patients had a nearly identical distribution of HBB gene cluster haplotypes, we examined the genetic population structure of Saudi patients in families that were originally from the SW Province excluding recent African immigrants and African American patients. Using a genome-wide set of SNPs, we performed a PCA with a subset of the Saudi SW patients (N= 8) and also with four Saudi SW individuals without disease. The Saudi groups had similar findings. Saudi sickle cell anemia patients from the Eastern Province; African Americans with sickle cell anemia; and African, Arab, Asian, and European populations from the Human Genome Diversity Project (HGDP) were also examined [31]. We estimated the Fst statistic between these populations, which is defined as the proportion of genetic diversity due to allele frequency differences among populations and can be interpreted as the distance between populations [32]. MSH samples were genotyped using the Illumina 370K array; CSSCD samples were genotyped on the Illumina 610K array; HGDP samples were downloaded from the Illumina iControlDB database and were genotyped on the Illumina 650K array; European and African populations from the Illumina repository were genotyped on the Illumina 650K array; Saudi SW were genotyped on the Affymetrix GeneChip® Human 250K Nsp array; Saudi patients from the Eastern Province were genotyped on the Illumina 610k array. The association between HbF and each SNP was tested using linear regression with a genotypic genetic model. To compare HbF levels between the Saudi and African Americans (CSSCD and MSH), we used a linear regression model and adjusted for the rs766432 genotype. In all regression models, we used the natural logarithm transformation of HbF to satisfy the normality assumption. To study the genetic population structure of the Saudi, African Americans, and HGDP populations, we conducted a PCA using a genome-wide set of SNPs. We used a total of 21,903 SNPs that were common among all array types and had a call rate greater than 95%. All subjects had a call rate greater than 93%. We estimated the Fst statistic between these populations, which is defined as the proportion of genetic diversity due to allele frequency differences among populations, and can be interpreted as the distance between populations. The PCA and the Fst statistics were computed using the software smartpca [33]. Abdulrahman Alsultan*, Nadia Solovieff , Aamer Aleem , Farjah H. AlGahtani , Ali Al-Shehri§, Mohamed Elfaki Osman§, Kadijah Kurban§, Hasan Bahakim§, Abdul Kareem Al-Momen , Clinton T. Baldwin¶, David H.K. Chui¶, Martin H. Steinberg¶, * Sickle Cell Disease Research Center and Department of Pediatrics, College of Medicine, King Saud University, Riyadh, Saudi Arabia, Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, Department of Medicine, College of Medicine, King Saud University, Riyadh, Saudi Arabia, § Department of Pediatrics, College of Medicine, King Saud University, Riyadh, Saudi Arabia, ¶ Center of Excellence in Sickle Cell Disease, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts.