Multiplex ligation-dependent probe amplification screening of isolated increased HbF levels revealed three cases of novel rearrangements/deletions in the β-globin gene cluster


Chang-Seok Ki, MD, PhD, Department of Laboratory Medicine and Genetics, Samsung Medical Centre, Sungkyunkwan University School of Medicine, 50 Irwon-Dong, Gangnam-Gu, Seoul 135-710 Korea. E-mail:


Investigations of naturally occurring mutations, such as the deletional thalassaemias and hereditary persistence of fetal haemoglobins (HPFHs), have brought many insights into human globin switching, but limited data have been reported so far. We selected 15 individuals with elevated fetal haemoglobin (HbF) levels (>5%) from a previous screening of 27 006 Korean individuals and analysed dosage changes of the globin gene cluster using multiplex ligation-dependent probe amplification (MLPA). Dosage changes detected by the MLPA probes were followed up with gap-polymerase chain reaction and sequence analysis. Three subjects were found to have deletions in the globin gene cluster, including a β-thalassaemia due to deletion of HBB (β-globin gene), an HPFH due to deletions of HBD (δ-globin gene) and HBB, and an HPFH due to a novel HBG2–HBG1 fusion gene consisting of exons 1 and 2 of HBG2 (Gγ-globin gene) and exon 3 of HBG1 (Aγ-globin gene). The case with the HBG2–HBG1 fusion suggested the existence of another mechanism for the reactivation of HBG2 and HBG1. The IVS2 of HBG2 and HBG1might play a role in HbF regulation, and combinations of specific polymorphisms could influence the reactivation of these genes in adults.

Inherited disorders of haemoglobin (Hb), such as sickle cell disease and thalassaemias, are common worldwide and create major healthcare problems in many countries. The anaemias caused by these haemoglobinopathies are often intractable and require novel therapeutic approaches. Increasing fetal haemoglobin (HbF) production with inducing agents or through genetic manipulation represents one potentially promising therapeutic option because higher postnatal HbF levels have an ameliorating effect on the clinical manifestations of these haemoglobinopathies (Thein, 2005; Manca & Masala, 2008). Accordingly, globin gene regulation has become the focus of intense investigation to identify new molecular targets for the induction of HbF.

Thus far, many insights into human globin switching have been gained by studying naturally-occurring mutations within the globin gene cluster. The human globin gene structure differs from that of mice, and so mouse models may not exactly reflect human mechanisms (Di Marzo et al, 2005; Manca & Masala, 2008). Investigations on deletional thalassaemias and the hereditary persistence of fetal haemoglobins (HPFHs) have revealed a locus control region (LCR) upstream of the globin gene cluster, enhancer-like elements downstream of HBB (β-globin gene), and silencing sequences upstream of HBD (δ-globin gene) (Forget, 1998; Bank et al, 2005; Chakalova et al, 2005; Thein, 2005). Nonetheless, the mechanism of globin expression and switching is not fully understood because reports on such informative cases remain limited. In particular, studies on cases with the deletion of a small genomic region are very rare but essentially needed, because they can help demonstrate the functional significance of a specific region, as shown by Corfu δβ-thalassaemia (Chakalova et al, 2005).

In this study, we screened dosage changes of the globin gene cluster in individuals with elevated levels of HbF and found three different deletions in the globin gene cluster, including a β-thalassaemia due to deletion of HBB, an HPFH due to deletions of HBD and HBB, and a HBG2–HBG1 fusion gene consisting of exons 1 and 2 of HBG2 (Gγ-globin gene) and exon 3 of HBG1 (Aγ-globin gene). The HBG2–HBG1 fusion gene does not fit into existing hypotheses based on juxtapositioning of enhancers, deletion of silencers, or competition of globin genes, suggesting the existence of another mechanism for HbF upregulation in adults.

Materials and methods

Subjects and measurement of HbF levels

Study subjects were recruited from our previous study (Lee et al, 2006), which screened HbF levels in 27 006 Korean individuals between May 2004 and October 2004. HbF levels were measured with the Bio-Rad Variant II Turbo high performance liquid chromatography system (Bio-Rad, Hercules, CA, USA), and 15 individuals with elevated HbF levels representing more than 5% of total Hbs were selected for further genetic analysis. Clinical and haematological data including complete blood counts (CBC) and peripheral blood smear findings were obtained for each subject. Patients with haematological malignancies or young infants under 2 years of age were excluded from the study. In all 15 patients, the presence of point mutations in HBB, HBG2 and HBG1 were ruled out by sequencing analyses in the previous study.

Sample collection and DNA extraction

Whole blood specimens were collected from each individual using EDTA tubes, and genomic DNA was isolated from peripheral blood leucocytes using the Wizard Genomic DNA Purification kit according to the manufacturer’s instructions (Promega, Madison, WI, USA).

Multiplex ligation-dependent probe amplification (MLPA) analysis

Gene dosage analysis of the LCR region and HBE, HBG2, HBG1, HBD and HBB on the β-globin gene cluster was performed using the SALSA MLPA KIT P102 HBB kit (MRC Holland, Amsterdam, the Netherlands) according to the manufacturer’s instructions. The MLPA samples consisted of approximately 200 ng of genomic DNA. Ligation and amplification were carried out on an ABI 9600 Thermal Cycler (Model 9600; Applied Biosystems, Foster City, CA, USA). The polymerase chain reaction (PCR) conditions included 33 cycles of 94°C for 30 s, 60°C for 30 s and 72°C for 60 s, followed by a final extension at 72°C for 20 min. All amplified fragments were separated using capillary electrophoresis on an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems). The area under the peak for each amplified fragment was measured and normalized in comparison with the peak areas of normal control individuals using GeneMarker software v.1.8 (SoftGenetics, State College, PA, USA).

Gap-PCR and determination of breakpoints

For patients found to have dosage changes in the MLPA analysis, we performed subsequent gap-PCR and sequencing analyses to confirm the results and determine the deletional breakpoints. Gap-PCR was done using forward- and reverse-primer pairs designed to encompass the regions where the MLPA probes showed decreased dosage, followed by sequence analysis of the amplified products. The PCR was initially performed with a thermal cycler with 32 cycles of denaturation at 94°C for 30 s, annealing at 60°C for 30 s and extension at 72°C for 60 s. The amplicon (5 l) was treated with 10 U shrimp alkaline phosphatase and 2 U exonuclease I (USB Corp., Cleveland, OH, USA) at 37°C for 15 min followed by incubation at 80°C for 15 min to achieve enzyme inactivation. Cycle sequencing was performed on an ABI PRISM 3130 Genetic Analyzer with the BigDye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems).


MLPA analysis and gap-PCR

Of the 15 individuals with elevated HbF (>5%), three were found to have reduced dosages (approximately half that of the normal controls) in different regions of the β-globin gene cluster, based on the MLPA analysis. Demographics and haematological data of these three patients are summarized in Table I.

Table I.   Haematological parameters and mutational states in 15 patients with high HbF levels.
PatientSex/age (years)Hb (g/l)MCV (fL)MCH (pg)MCHC (g/l)HbF (%)HbA2 (%)PhenotypeHBG2 XmnI −158C>TMutation analysisHGVS nomenclature
  1. HGVS, Human Genome Variation Society.

 1F/489969·521·83136·43·7β-thalassaemia−/−HBB deletionHBB:c.-77_*66delinsGT
 2F/6711984·827·432219·52·7HPFH−/−HBD and HBB deletionHBD:c.48-?_HBB:c.*645 + ?del
 3M/61132103·735·13385·92·5HPFH−/+HBG2-HBG1-fusionHBG2:c.315 + 573_HBG1:c.315 + 572del

Patient 1, a 48-year-old female, had an enduring microcytic and hypochromic anaemia with increased HbF (∼6·4%) and HbA2 (∼3·7%) levels typical of β-thalassaemia minor. MLPA analysis showed decreased dosages for six consecutive MLPA probes whose targets ranged from the 5′-untranslated region (UTR) to the 3′-UTR of HBB, suggesting a deletion of all coding exons of HBB (Fig 1). Gap-PCR and sequencing analysis using a primer pair designed to cover the removed region (forward: 5′-ACTCCTAAGCCAGTGCCAGA-3′; reverse: 5′-GATTCCGGGTCACTGTGAGT-3′) revealed a novel deletion of a 1568-bp genomic region ranging from −27 bp upstream of the transcription start site to 66 bp downstream of the TAA stop codon of HBB, linked via two orphan nucleotides (HBB:c.-77_*66del1568insGT; Fig 2A).

Figure 1.

 MLPA analysis showed half dosages for six probes from 5′-UTR to 3′-UTR of HBB in Patient 1, all probes located in 3′ of HBD in Patient 2 and two probes located in the HBG2 and HBG1 region in Patient 3.

Figure 2.

 (A) Schematic of MLPA probes with decreased dosages (marked with *) and the deleted regions in Patient 1. Sequence analysis revealed a novel 1568-bp deletion from −27 bp upstream of the transcription start site to 66 bp downstream of the TAA stop codon of HBB, linked via two orphan nucleotides. (B) Schematic of the location of MLPA probes with decreased dosages in Patient 2 and the inferred deleted regions. Further gap-PCR was not done because the 3′ margin was inestimable.

Patient 2 was a 67-year-old female who showed markedly high levels of HbF (∼19·5%) without other clinical or haematological abnormalities, consistent with HPFH. In the MLPA analysis, dosages for all MLPA probes located 3′ from exon 1 of HBD were reduced to half of those of the normal controls, indicating deletional δβ-HPFH (Fig 1). We were unable to perform gap-PCR in this patient because the 3′ margin of the removed region was located beyond the coverage of the MLPA probes and was therefore inestimable (Fig 2B).

Patient 3, a 61-year-old male with mild alcoholic liver disease, displayed HbF levels representing about 6% of his total Hbs. MLPA analysis showed half dosages for two probes targeted for exon 3 of HBG2 and a genomic region between HBG2 and HBG1 (Fig 1). Gap-PCR was performed using a forward primer specific for the HBG2 promoter (5′-CCAATGCTTACTAAATGAGACTAAGACG-3′) and a reverse primer specific for the HBG1 downstream region (5′-ATAAATGAGGAGCATGCACACAC-3′), resulting in the amplification of an abnormally shortened PCR product (∼2·2 kb; Fig 3A). The complete sequence of the amplified product was reconstructed through sequence analysis using several internal primers, and a ∼4·9 kb deletion between the HBG2 intervening sequence (IVS) 2 and the HBG1 IVS2 (Fig 3B,C) was identified. Although we could not pinpoint the exact breakpoint due to the high homology between the HBG2 and HBG1, the fusion appeared to have occurred just before the polymorphic (TG)n(CG)m repeats in the middle of the IVS2, considering that the TG13 observed in the patient was a HBG1-specific polymorphism (Zertal-Zidani et al, 2002). The HBG2−158C>T polymorphism, also known as the XmnI polymorphism, was found in the mutant allele. The mutant allele also had novel sequence variations located 5–6 bp after the stop codon (HBG1:c.*5A>C and c.*6C>T). These variations seemed to be polymorphisms rather than mutations because they were observed in normal Korean controls with high frequency (data not shown). The patient had one son and one daughter, both with normal HbF levels, and further genetic study indicated that neither had inherited the mutant allele.

Figure 3.

 (A) Gap PCR using a HBG2-specific forward primer and HBG1-specific reverse primer yielded an abnormally shortened PCR product (∼2·2 kb) in Patient 3 (lane P), while the patient’s son and daughter (lanes S and D, respectively) with normal HbF levels showed normally expected 7·1 kb bands. (B) Schematic of a ∼4·9 kb deletion between HBG2 IVS2 and HBG1 IVS2 creating a novel HBG2–HBG1 fusion gene. The polymorphic (TG)n(CG)m repeats in IVS2 are indicated by filled triangles. (C) Complete sequences of the HBG2–HBG1 fusion gene in Patient 3. The HBG2-specific sequences are highlighted in blue, the HBG1-specific sequences are in red. The −158C>T variation on the promoter and the polymorphic (TG)n(CG)m repeats in the middle of IVS2 are underlined.


Different genetic alterations have been demonstrated to increase HbF in adult life. They include point mutations in HBG2, HBG1 or HBB and large deletions involving the globin gene clusters. In the preceding sequence analysis, we failed to find any point mutations in HBG2, HBG1 or HBB. We found a high frequency (10/15) of only the HBG2−158C>T polymorphism, which is known to be associated with increased HbF levels in certain patient subsets, especially those with erythoid stress conditions (Gilman & Huisman, 1985; Garner et al, 2000). The near absence of point mutations in our subjects may reflect a specific genetic background in Korean ethnicity, but it should be taken into account that our study was designed to select individuals with >5% HbF levels and therefore could have missed those with the borderline HbF levels (2∼5%) that are frequently seen in non-deletional HPFHs or β-thalassaemias. Rather, we found large deletions in three individuals with >5% HbF levels; they included one β-thalassaemia and two HPFHs. For the screening of deletional mutations, an MLPA method was used instead of traditional Southern blot analysis. Concurrent gap-PCR analysis confirmed that the MLPA method is highly reliable and could be a substitute for the laborious Southern blot analysis. We also compared the haematological parameters between individuals with deletions and those without, but found no notable differences.

For HBG2 and HBG1 activation in deletional β-thalassaemia and HPFH, several different, but not mutually exclusive, hypotheses have been proposed. The competition model holds that the HBG2 (or HBG1) and HBB genes are in competition for access to regulatory elements (most probably the LCR), which could account for many β-thalassaemia cases (Hanscombe et al, 1991; Peterson & Stamatoyannopoulos, 1993). The juxtaposition model suggests that the deletions juxtapose an enhancer located in the neighbourhood of the deletional breakpoints to the vicinity of HBG2 or HBG1 and thereby activate it (Anagnou et al, 1995; Forget, 1998; Katsantoni et al, 2003). The silencer model proposes that loss of silencing sequences residing within a region upstream of HBD play an important role in γ reactivation (Calzolari et al, 1999; Bank, 2005; Bank et al, 2005).

Patient 1 showed typical genetic and phenotypic features of the deletional β-thalassaemias. Entire coding regions of HBB were removed and thus did not function. Therefore, the increase of HbF in this patient could be attributed to the loss of competition between HBB and HBG2 or HBG1 and relative activation of the latter genes. Patient 2 had normal haematological parameters except for a relatively high HbF level. The deletion in this patient was shown to span from the 5′ part of HBD to the 3′ part of HBB. From the MLPA probe information, we deduced that the 5′ breakpoint lay between HBG1 and HBD while the 3′ margin was inestimable. Like all deletional HPFHs hitherto reported, the deletion in this patient also removed the region upstream of HBD, and this suggested that the silencer model may be attributable. The 3′ breakpoint in this patient was expected to be located around the enhancers of the downstream olfactory nerve receptor genes, as in other deletional HPFHs.

Patient 3, with a HPFH phenotype, could not be easily accounted for by the previously suggested models; he had an intact silencer region, and the deleted region was not large enough to juxtapose enhancer elements. One possible explanation may be an interaction of the HBG2 promoter, especially the −158C>T polymorphism, with the HBG1-derived IVS2 or 3′UTR of the recombinant gene. Only one previous report has genetically characterized the HBG2–HBG1 fusion (Sukumaran et al, 1983), which investigated a newborn with a HBG2–HBG1 fusion gene as well as his father who carried the mutant allele and yet showed normal HbF levels. That fusion gene had a HBG2 promoter and a HBG1 3′UTR, but was different from our case in that the breakpoint was estimated to be upstream of exon 2, because the Ile75Thr variation observed in that case was a HBG1-specific polymorphism. Consequently, one of the essential differences of our case from the previous case may lie in IVS2, and this observation suggests that differences in IVS2 of the HBG2 or HBG1 genes might have an effect on HbF regulation in adults. Some authors demonstrated that IVS2, particularly its polymorphic microsatellite (TG)n(CG)m, could be important in the expression of HBG2 and HBG1 (Donovan-Peluso et al, 1987, 1991). The (TG)13 repeat, which is exclusively found in HBG1, has received particular attention because some individuals with this polymorphism have a tendency to have increased HbF levels (Lapoumeroulie et al, 1999; Zertal-Zidani et al, 2002). When the functional impact of IVS2 containing the (TG)13 repeat was investigated by transfecting it into K562 cells and measuring the reporter gene expression (Lapoumeroulie et al, 1999), the 5′ part of IVS2 had silencing activity and the 3′ part of IVS2 with the (TG)13 repeat possessed enhancing activity. The entire IVS2 had almost no influence on the reporter gene expression, however, probably due to counteraction of the 5′ and 3′ IVS2 segments. Given that our case had a recombinant IVS2 consisting of a HBG2-derived 5′ part and a HBG1-derived 3′ part, we hypothesized that some HBG2-specific sequences in the 5′ part of IVS2 could influence such counteracting activity.

Interestingly, this patient must have had a mild γ-thalassaemic phenotype in fetal life, because he had only three copies of the genes encoding the γ-globulin. The HBG2–HBG1 fusion that causes γ-thalassaemia has been observed with substantial frequencies in different populations (0·8–1·5%) in the screening of neonatal bloods by high-performance liquid chromatography (Shimizu et al, 1986). This could be because HBG2 and HBG1 are highly homologous and thus prone to recombinational events (Slightom et al, 1980). Their breakpoints are expected to be very heterogeneous, although the data are absent thus far. It is possible that more cases resembling that of our patient exist, and the measurement of HbF levels in adult life for those HBG2–HBG1 patients and their genetic characterization could help to identify HbF-regulating sequences.

In summary, we investigated genomic changes in patients with increased HbF levels and presented two cases with deletional β-thalassaemia or HPFH that fit the traditional models for HbF upregulation and a case of HBG2–HBG1 fusion with an uncertain mechanism. The last case indicates the existence of a novel mechanism that may underlie reactivation of HBG2 and HBG1. The IVS2 of HBG2 and HBG1 may play a role in HbF regulation, and combinations of specific polymorphisms could influence reactivation of HBG2 and HBG1 in adult life.


This work was supported by Samsung Biomedical Research Institute grant, # SBRI C-A8-205-1.

Conflict of interest disclosure

The authors declare they have no conflicts of interest related to this study.