Whole genome sequence-based haplotypes reveal a single origin of the 1393 bp HBB deletion

Abstract

Background

Mutations of HBB give rise to two prevalent haemoglobin disorders—sickle cell disease (SCD) and β-thalassaemia. While SCD is caused by a single base substitution, nearly 300 mutations that downregulate expression of HBB have been described. The vast majority of β-thalassaemia alleles are point mutations or small insertion/deletions within the HBB gene; deletions causing β-thalassaemia are very rare. We have identified three individuals with haemoglobin Sβ⁰-thalassaemia in which the β⁰-thalassaemia mutation is caused by a large deletion.

Objective

To use whole genome sequence data to determine whether these deletions arose from a single origin.

Methods

We used two approaches to confirm unrelatedness: pairwise comparison of SNPs and identity by descent analysis. eagle, V.2.4, was used to generate phased haplotypes for the 683 individuals. The Neighbor-Net method implemented in Splitstree V.4.13.1 was used to construct the network of haplotypes.

Results

All three deletions involved 1393 bp, encompassing the β-promoter, exons 1 and 2, and part of intron 2, with identical breakpoints. The cases were confirmed to be unrelated. Haplotypes based on 29 SNPs in the HBB cluster showed that the three individuals harboured different β^S haplotypes. In contrast, the haplotype harbouring the 1393 bp deletion was the same in all three individuals.

Conclusion

We suggest that all the reported cases of the 1393 bp HBB deletion, including the three cases here, are likely to be of the same ancestral origin.

INTRODUCTION

Inherited disorders of the β-globin gene (HBB)—β-thalassaemia and sickle cell disease (SCD)—rank among the most clinically significant human monogenic disorders.¹ β-Thalassaemia is caused by a quantitative deficiency of β-globin chains that are structurally normal. In contrast, SCD is caused by an abnormal haemoglobin (Hb) variant (Hb S, βGlu6Val) that results from a point mutation in the HBB gene. Heterozygotes (carriers) for β-thalassaemia and Hb S (Hb AS) have a substantial survival advantage in malaria-endemic regions, leading to an overlap in the regions of prevalence of both disorders.² Hence, it is not unusual to encounter individuals with SCD who are compound heterozygotes for β-thalassaemia and Hb S (Hb Sβ-thalassaemia).³

β-Globin is encoded by a structural gene found in a cluster with the other β-like genes on chromosome 11, arranged in order of developmental expression.⁴ Upstream of the complex is the β-locus control region (β-LCR), which plays a pivotal role in expression of the downstream globin genes via direct physical interaction with the globin promoters, mediated through binding of erythroid-specific and ubiquitous transcription factors.

Unlike SCD, β-thalassaemia encompasses a spectrum of over 300 described mutations (http://www.ithanet.eu/db/ithagenes; http://globin.bx.psu.edu/hbvar), all leading to a quantitative deficiency of β-globin chains.⁴ The vast majority of β-thalassaemia mutations are point mutations or small insertions/deletions (indels) within the HBB gene and its flanking sequences. These β-thalassaemia alleles can be inherited heterozygously, homozygously or co-inherited with structural Hb variants, leading to SCD (Hb Sβ-thalassaemia) or thalassaemias (Hb Eβ-thalassaemia). A hallmark of heterozygous β-thalassaemia is elevated Hb A₂ level (3.5%–5.5%), used as a common criterion for β-thalassaemia screening accompanied by hypochromia and microcytosis. Hb F can also be variably elevated in β-thalassaemia carriers, but the level is typically <2.5% of the total Hb.

Deletions downregulating HBB expression are rare, limited to HBB gene itself or comprise extensive deletions of the whole β-globin cluster.⁴ Eighteen deletions restricted to the HBB gene have been described, 14 of which (from 290 bp to >67 kb) remove in common a region in the β promoter which includes the CACCC, CCAAT and TATA elements. These deletions are extremely rare but of particular clinical interest because they are associated with unusually high levels of Hb A₂ (>6.5%) and Hb F (>2%) levels in heterozygotes. Although these deletions abolish β-globin production, the increase in Hb F is adequate to compensate for the complete absence of Hb A in homozygotes and are thought to confer milder clinical phenotypes when co-inherited with Hb S, unlike typical Hb Sβ⁰-thalassaemia genotypes. Here, we report the identification and phenotype of three unrelated Hb Sβ⁰-thalassaemia cases with a rare 1393 bp HBB deletion, along with whole genome sequencing (WGS)-based haplotype analysis that reveals a common ancestral origin.

METHODS

Study population

The study population is from an observational cohort of patients with SCD. WGS was performed in 683 subjects with SCD. Variant analysis identified 546 Hb SS, 90 Hb SC, 2 Hb SD, 1 Hb SOArab and 44 Hb AS subjects. Further analyses of the so-called ‘Hb AS’ group revealed a diagnosis of Hb Sβ-thalassaemia in all subjects, with the β-thalassaemia allele caused by several point mutations and short indels resulting in 12 Hb Sβ⁰-thalassaemia and 29 Hb S β⁺-thalassaemia. As large deletions co-inherited with rs334 could be misidentified as Hb SS, we manually reviewed the HBB gene region using the software Integrative Genomics Viewer (MIT, Broad Institute) and identified three identical deletions of HBB, 1393 bp in length, spanning the β promoter, exon 1 and 2 and part of intron 2. Using previously published primer sequences and PCR conditions, gap-PCR and sequence analyses confirmed the deleted sequence on chr11: 5247393–5248785 (GRch37/hg19). The deletions were heterozygous, with β^S (rs344) identified on the other allele, confirming the genotype of Hb Sβ⁰-thalassaemia for these three individuals.

Relatedness and identity by descent

We confirmed that the three identified individuals were unrelated using two approaches. First, we used pairwise comparison of SNPs in the whole cohort of 683 individuals to estimate the relatedness of samples. We employed the bcftools (V.1.9)⁵ --isec option to evaluate union and intersection of variants between pairs. Second, we estimated genome-wide identity by descent (IBD) sharing⁶ within the cohort. We used plink⁷ to compute genome-wide estimates of identity by descent (IBD). We first carried out linkage disequilibrium (LD)-based pruning of our SNP set using the --maf 0.01 --indep-pairwise 50 5 0.2 options to prune SNPs in a window size of 50 SNPs, and leave SNPs with MAF at least 1%, with no pairs remaining with r² >0.2. This step reduces correlation among SNPs so that the remaining SNPs are roughly independent. Then, we used the plink -- genome option and used only the LD-pruned set of SNPs from our cohort for IBD sharing estimation.

Haplotype analysis

Phased haplotypes were estimated for the 683 individuals. Haplotypes were based on rs334 (at which the derived allele is the Hb S allele), rs33930165 (at which the derived allele is the Hb C allele) and 27 polymorphisms in linkage disequilibrium with rs334.⁸ Phasing was performed using Eagle, V.2.4.⁹ Due to ascertainment for SCD, the genotype distributions were mismatched with external, population-based reference panels. Consequently, phasing was performed without a reference panel. The Neighbor-Net method implemented in SplitsTree V.4.13.1 ¹⁰ was used to construct the network of haplotypes.

Results

Case descriptions

Patient 1 is an African-American man aged 24 years with Hb Sβ⁰-thalassaemia, otherwise healthy. He experiences infrequent pain crises that are typically managed at home with non-steroidal anti-inflammatory medications. He occasionally presents to the emergency department (ED) for acute pain but never requires hospitalisations and has no chronic pain. He had a splenectomy for splenomegaly at age 17. He reports receiving less than four lifetime blood transfusions but notably has a ferritin of 1795 μg/L. He has no evidence of cardiomyopathy or pulmonary hypertension by transthoracic echocardiogram, no other sickle-related complications (ie, acute chest syndrome (ACS), stroke, nephropathy, avascular necrosis, leg ulcers or priapism), and is not on hydroxyurea. Demographics and details of his laboratory evaluation are shown in table 1. His Hb F level is elevated at 11.8% and Hb A₂ at 7.9%.

Table 1.

Demographics, laboratory data and DNA analysis of three individuals with Hb Sβ⁰-thalassaemia carrying the 1393 bp deletion

	Patient 1	Patient 2	Patient 3
Age (years)	24	36	22
Gender	Male	Male	Male
Ethnicity	African-American	African (Sierra Leone)	Cuban-Native, American-African, American
WBC count (K/μL)	12.40	5.38	11.40
RBC count (M/μL)	3.82	5.35	4.00
Haemoglobin (g/L)	95	137	111
Haematocrit (%)	28.6	40.7	32.2
Mean Cell Volume (fL)	74.9	76.1	80.5
Mean Cell Haemoglobin (pg)	–	25.6	27.8
Platelet count (K/μL)	293	76	427
Absolute reticulocyte count (K/μL)	441.0	185.6	683.0
Creatinine (mg/dL)	0.7	1.05	0.76
Total bilirubin (mg/dL)	2.8	0.8	2.3
Direct bilirubin (mg/dL)	0.5	0.3	0.3
LDH (U/L)	411	219	299
Ferritin (μg/L)	1795	62	1159
Hb S (%)	76.6	74.7	77.6
Hb F (%)	11.8	18.8	15.2
Hb A2 (%)	7.9	6.5	7.2
βS Haplotype	Senegal (HAP107)	Senegal (HAP124)	Benin (HAP51)
rs7482144 (Xmn-I)	C/T	C/T	C/C
Hydroxyurea use	No	No	Yes
Clinical severity of SCD	Mild	Mild	Severe

rs7482144 (Xmn-I) C/C is wild-type, while C/T is heterozygous for the Xmn-I polymorphism associated with elevated Hb F levels.

Hb, haemoglobin; LDH, lactate dehydrogenase; RBC, red blood cell; SCD, sickle cell disease; WBC, white blood cell.

Patient 2 is a man aged 36 years originally from Sierra Leone with Hb Sβ⁰-thalassaemia, schizophrenia, hypertension and chronic hepatitis B virus infection (HBV; positive HBV core antigen). His SCD is complicated by frequent pain crises in joints requiring occasional emergency care but no hospitalisations, and he does not have chronic pain. He has never been transfused and again, has minimal sickle-related complications. He has no cardiomyopathy by echocardiogram; tricuspid regurgitation velocities (TRV) have historically varied (2.4, 2.6 and 3.0 m/s), and most recent TRV is 2.4 m/s. In addition to antihypertensives and antipsychotics, he had previously been treated with hydroxyurea in the distant past but was not taking hydroxyurea at time of entry into the study. Physical examination is notable for obesity (body mass index 33.5 kg/m²) and splenomegaly 10 cm below costal margin. His Hb F level is 18.8% and Hb A₂ is 6.5% (table 1).

Patient 3 is a man aged 22 years of mixed ethnicity with Hb Sβ⁰-thalassaemia complicated by severe pain crises requiring multiple hospitalisations and ED visits annually, as well as chronic pain, recurrent episodes of ACS, acute splenic sequestration requiring splenectomy and cholecystectomy at age 3, neurological complications of transient ischaemic attack and brain MRI documented infarcts, transfusional iron overload (ferritin 1159 μg/L), asthma and sleep paralysis. He was treated with chronic monthly transfusion therapy from age 14 to 17 years and started hydroxyurea therapy at age 18 years. Echocardiogram showed TRV of 2.5 m/s but no cardiomyopathy. Physical examination is notable only for digital clubbing. Laboratory results are again notable for the unusually elevated Hb F and A₂ levels of 15.2% and 7.2%, respectively (table 1).

Haplotype analysis

Phasing yielded 123 distinct haplotypes (figure 1 and online supplementary table). The 1393 bp HBB deletion removed rs334 and rs33930165. The haplotype carrying the 1393 bp HBB deletion was identical in the three carriers; at all 27 remaining positions, the haplotype carried the ancestral state. The other haplotype clustered with Senegal β^S haplotypes for patients 1 and 2 (HAP107 and HAP124, respectively), and Benin β^S haplotypes (HAP51) for patient 3. Patients 1 and 2 were also heterozygous for the Xmn-I C→T polymorphism in association with the Senegal β^S haplotype.

Figure 1.

Split decomposition networks of haplotypes carrying β-globin mutations. phased haplotypes were estimated for the 683 individuals. Haplotypes were based on rs334 (at which the derived allele is the haemoglobin S allele), rs33930165 (at which the derived allele is the haemoglobin C allele) and 27 polymorphisms in linkage disequilibrium with rs334. The red dot in the middle on the right represents the ancestral, wild-type haplotype. The blue dots represent the β^S haplotypes in the three individuals with Hb Sβ⁰-thalassaemia. The left cluster includes the classical Benin β^S haplotype, the middle cluster includes the classical Central African Republic/Bantu β^S haplotype and the right cluster includes the classical Senegal β^S haplotype. The scale bar represents 0.1 mutations/site. The 1393 bp HBB deletion removed rs334 and rs33930165 from the set of 29 polymorphisms. Based on haplotype frequencies, the deletion most probably occurred on the ancestral (wild-type) haplotype.

DISCUSSION

We have identified three unrelated Hb Sβ⁰-thalassaemia patients heterozygous for the same 1393 bp HBB deletion with a shared ancestral haplotype. The first case of this HBB deletion was likely reported in 1984¹¹ as ~1.4 kb. Subsequently, more case reports of this deletion followed, and DNA sequence analyses confirmed breakpoint sequences with a size of 1393 bp. Similar to the other deletions that include the promoter sequences of HBB, heterozygotes have unusually elevated levels of Hb A₂ and variable increases of Hb F.^11–15 It has been proposed that deletion of the β promoter removes competition for the upstream β-LCR and limiting transcription factors, allowing greater interaction of the LCR with the cis δ and γ genes, thus enhancing their expression. Indeed, studies of an individual heterozygous for the 1.39 kb β-thalassaemia deletion and a δ-chain variant showed that there is a disproportionate increase of Hb A₂ derived from the δ gene in-cis to the β gene deletion.¹⁶ The β-promoter can also be inactivated by point mutations, and again, carriers have unusually high Hb A₂ and Hb F levels.¹⁷ While the increase in Hb F is typically variable and modest in heterozygotes for such mutations, Hb F can be sufficiently increased to partially compensate for the complete absence of β globin in homozygotes.^18–20

This deletion has been confirmed only once previously in combination with the β^S mutation in a pair of African-American siblings, whose mild SCD phenotypes were attributed to the elevated Hb F levels.¹⁵ Although all three of our cases had similarly elevated Hb F levels, clinical severity was variable. Patients 1 and 2 had a mild SCD phenotype (manifested as mainly acute pain crises not requiring hospitalisation) and were not on hydroxyurea therapy, while patient 3 had a more severe phenotype, requiring chronic transfusion and hydroxyurea therapy. Of note, Hb F levels ranged from 11.8% in patient 1 to 18.8% in patient 2. Despite hydroxyurea therapy, the MCV in patient 3 was relatively normal at 80.5 fL, which is likely to be related to the co-inherited β⁰-thalassaemia allele. Unfortunately, we do not have an MCV value before starting hydroxyurea. The Hb S allele in patients 1 and 2 were associated with the Senegal β^S haplotype that has been associated with milder SCD phenotypes, while patient 3 had the most severe phenotype among the three individuals, in keeping with the Benin β^S haplotype (table 1). The Senegal β^S haplotype carries the Xmn-I C→T polymorphism, which is known to increase Hb F production and less severe disease. However, the difference in severity among patients 1, 2 and 3 appeared to be independent of Hb F levels, highlighting the complexity of predicting genotype-phenotype correlation in SCD.

The 1393 bp HBB deletion removed rs334 and rs33930165 from the set of 29 polymorphisms. Therefore, there are three possible origins of the haplotype carrying the deletion. In decreasing order of probability based on haplotype frequencies, the deletion could have occurred on (1) the ancestral (wild-type) haplotype; (2) the modal Central African Republic/Bantu haplotype, which carries the derived sickle allele on the background of ancestral alleles at all other positions or (3) the haplotype carrying the Hb C allele. Formally, a fourth possibility is that the deletion occurred on a haplotype carrying both Hb S and C alleles, but such a haplotype has not been observed to date.

In all three individuals in our cohort, the 1393 bp deletion shared an ancestral origin. We suggest that all of the reported cases of the 1393 bp HBB deletion, including the three cases reported here, are likely to be of the same ancestral origin. The handful of cases identified to date are of African ancestry with the exception of one Anglo-Saxon-Dutch subject.¹⁴ African ancestry is also reported in all the three cases in our cohort, suggesting that the ancestral haplotype likely originated in Africa.