The molecular basis of beta-thalassemia intermedia in southern China: genotypic heterogeneity and phenotypic diversity

Background The clinical syndrome of thalassemia intermedia (TI) results from the β-globin genotypes in combination with factors to produce fetal haemoglobin (HbF) and/or co-inheritance of α-thalassemia. However, very little is currently known of the molecular basis of Chinese TI patients. Methods We systematically analyzed and characterized β-globin genotypes, α-thalassemia determinants, and known primary genetic modifiers linked to the production of HbF and the aggravation of α/β imbalance in 117 Chinese TI patients. Genotype-phenotype correlations were analyzed based on retrospective clinical observations. Results A total of 117 TI patients were divided into two major groups, namely heterozygous β-thalassemia (n = 20) in which 14 were characterized as having a mild TI with the Hb levels of 68-95 g/L except for five co-inherited αααanti-3.7 triplication and one carried a dominant mutation; and β-thalassemia homozygotes or compound heterozygotes for β-thalassemia and other β-globin defects in which the β+-thalassemia mutation was the most common (49/97), hemoglobin E (HbE) variants was second (27/97), and deletional hereditary persistence of fetal hemoglobin (HPFH) or δβ-thalassemia was third (11/97). Two novel mutations, Term CD+32(A→C) and Cap+39(C→T), have been detected. Conclusions Chinese TI patients showed considerable heterogeneity, both phenotypically and genotypically. The clinical outcomes of our TI patients were mostly explained by the genotypes linked to the β- and α-globin gene cluster. However, for a group of 14 patients (13 β0/βN and 1 β+/βN) with known heterozygous mutations of β-thalassemia and three with homozygous β-thalassemia (β0/β0), the existence of other causative genetic determinants is remaining to be molecularly defined.

Background β-thalassemia is one of the most common monogenic disorders in the world. The incidence for this disease is high in tropical and subtropical areas including southern China. In southern China, the carrier rate of β-thalassemia is 2.54% in Guangdong [1] and 6.78% in Guangxi [2] where two provinces were thalassemia occurred most frequently. The hemoglobin E (HbE) is one of the common β-thalassemia variants with carrier rates of 0.09-0.13% [1,3], and 0.19-0.21% [ [2], unpuplished report from our laboratory] , respectively, in these two regions. According to the clinical phenotypes, β-thalassemia can be divided into three main types: thalassemia major (TM), thalassemia trait (TT) and thalassemia intermedia (TI). TM is a severe form that requires transfusions from infancy for survival, whereas TT is usually asymptomatic. TI is used to indicate a clinical condition of intermediate gravity between TT and TM, which encompasses a wide phenotypic spectrum spanning from mild anemia to more severe anemia and these patients require only occasional blood transfusions, if any [4,5].
Corresponding to the phenotypic diversity, the molecular basis of TI is also variable. The major genetic modifiers of β-thalassemia are genotypes of βand αglobin and expression of γ-globin [6,7]. Some genotypic factors have been reported to affect synthesis of the γ-globin chain, such as the 3'HS1 (+179 C!T) polymorphism [8], the (AT)xNy(AT)z motif in the 5'HS2 site [9], and the (AT)x(T)y motif in the -540 region of the β-globin gene [10]. Variation of rs11886868 (T!C) in the BCL11A gene has also been shown to correlate with increased HbF in European TI patients [11]. In addition, those factors that can moderate globin imbalances indirectly or cause the β-thalassemia-like phenotype, such as GATA-1 [12], alpha hemoglobin stabilizing protein (AHSP) [13,14], and heme-regulated initiation factor 2 alpha kinase (HRI) [15], are also thought to contribute to the phenotypic diversity of TI. Researchers have described molecular characterization of TI in Iranian [16], Indian [17], Italian [4,18], Israelis [19] and other populations [20]. However, to date, the genetic basis of TI in Chinese patients is still poorly understood [21]. In this study, a comprehensive analysis of the molecular basis underlying TI was performed in southern China. Genotypes of β-globin and other known modifiers linked to α/β imbalance were investigated in 117 patients with β-thalassemia intermedia phenotypes. Their clinical, hematological, and molecular data were analyzed systematically with the aim of creating a genotype-phenotype correlation. During the analysis, two novel mutations, Term CD+32(A!C) and Cap +39(C!T), were found. These two mutations were named according to the standard nomenclature rules of human hemoglobin mutation http://globin.cse.psu.edu/. Our findings provide genetic insights of TI occurrence in southern China and are useful in genetic counselling, treatment and management.

Subjects
A total of 117 patients from 109 families with TI phenotypes were recruited for this study and consisted of the following: 63 males and 54 females, 73 children and 44 adults. 78 patients (66.7%) came from Guangxi, 37 (31.6%) came from Guangdong, and 2 (1.7%) came from either Guizhou or Zhejiang. Their ethnic backgrounds were as follows: 71 Han, 45 Zhuang and 1 Miao. We recruited the TI patients in this study according to previously described criteria [4,5,20], in whom the classical clinical diagnosis of TI patients, such as the steady state Hb level of 60-105 g/L, age at diagnosis over two years old, and transfusion independence were emphasized for all our patients. The protocol for this study was approved by medical ethics committee of Southern Medical University. Informed consent was obtained from each individual or parents of individuals younger than 18 years old.

Hematological analysis and Clinical data collection
Hematological data: complete blood counts and red cell indices were determined by automated cell counting (Model Sysmex F-820; Sysmex Co Ltd, Kobe, Japan); the levels of HbA, A 2 and F were analyzed on the Bio-Rad Variant II HPLC system (HPLC, VARIAN™, Bio-Rad, Hercules, CA, USA).
Clinical data: the information about blood transfusions, thalassemia appearance, and age at diagnosis, hepatosplenomegaly and splenectomy was obtained by retrospective clinical data.

RNA analysis
Total cellular RNA was isolated from fresh peripheral blood using QIAamp RNA Blood Mini Kit (Qiagen, Germany). The cDNA synthesis was performed using the ExScript RT reagent Kit (TaKaRa Biotechnology, China). The expression levels of β-globin (target gene) and β-actin (control gene) were measured by SYBR Green-based relative quantitative RT-PCR assays described previously by Yipeng [28]. Five heterozygous subjects carrying the Cap+39(C!T) mutation and three heterozygous subjects carrying the Term CD+32(A!C) mutation were selected respectively, as two patient groups, while six normal subjects were used as the control group.

Statistical methods
Statistical analysis was performed using SPSS software (Version 13.0, SPSS inc, USA). The difference of the relative mean mRNA concentration between mutation carriers and normal individuals were analyzed by the independent samples t-test. A p-value < 0.05 was considered as statistically significant.

Molecular basis of 117 TI patients in southern China
In this study, 117 Chinese individuals between 2 and 60 years of age were enrolled to characterize the molecular basis of β-thalassemia intermedia in southern China. Their clinical, hematological and molecular data are summarized in table 2. According to their genotype of β-globin, we divided them into two major types: Type I β-thalassemia homozygotes or compound heterozygotes for β-thalassemia and other β-globin defects (Hb E, HPFH or δβ-thalassemia) (n = 97) who inherited two deficient β-globin alleles and Type II β-thalassemia heterozygotes (n = 20) who had only a single β-thalassemia allele. Statistical comparison showed that their steadystate Hb level of Type II (86.7 ± 10.8 g/L) was significantly higher than that of Type I (75.7 ± 11.4 g/L) (Mann-Whitney test, p < 0.05). In contrast, their HbF level of Type I (57.9 ± 27.2%) was much higher than that of Type II (4.0 ± 3.8%, p = 0.000). Significant differences in hematological parameters such as MCV (66.3 ± 8.7 fL in Type I vs. 58.7 ± 6.9 fL in Type II, p = 0.004) and MCH (20.2 ± 2.5 pg in Type I vs. 18.5 ± 2.0 pg in Type II, p = 0.000) were also observed. In total, we detected 18 β-thalassemia alterations including two novel ones which were termed as Term CD +32(A!C) and Cap+39(C!T), β-globin variant HbE, the two deletional mutations which can lead to SEA-HPFH or (δβ) 0thalassemia Chinese G γ + ( A γδβ) 0 (table 3).
Analysis of genotypic modifiers in 3 b 0 /b 0 and 10 b 0 /b N or b + /b N TI cases Usually, thalassemia patients with β 0 /β 0 genotype and normal α-globin gene had the TM phenotypes, but we detected 3 β 0 /β 0 patients with TI phenotypes whose Hb levels are 74-99 g/L (table 4). We examined the existence of known genetic modifiers, which related to increasing HbF levels in these three cases. The clinical, hematological and molecular results observed from the group of patients are listed in table 4.
Fourteen patients who displayed the relative mild TI phenotypes with steady Hb levels of 68-95 g/L were identified to be heterozygotes with one known common mutation for β-thalassemia and normal α-globin gene. Analysis of other known genetic modifiers that mainly focused on several factors of aggravate α/β imbalance were done in ten patients (blood samples from other four patients were not available). The overall findings are summarized in table 5.

T/T) (T/T) (T/C) (T/T) (T/C) (T/C) (T/C) (T/T) (T/C) (T/T) (C/C) (C/C) ND ND ND ND
normal control group after a normalization procedure using linear regression equations. Statistical analysis showed that there was a significant difference (p = 0.021) of mean relative β-globin mRNA concentration between the Term CD+32(A!C) group and the control group. The decreased level of β-globin mRNA in the patient group compared with that of the normal control group was calculated to be 17.8%. In contrast, no significant difference (p = 0.270) was found between the Cap+39(C!T) group and the control group.

Discussion
Provision of appropriate treatment and management for TI patients is still a challenge because its characteristic features, phenotypic diversity, and genotypic heterogeneity, the clinical severity of thalassemia intermedia cases are very difficult to predict from their genotypic data. The molecular mechanisms underlying different populations in Europe [18,20], Israel [19], India [17] and Iran [16] had been reported. However, the molecular basis of Chinese TI patients was still uncertain, which brought about much trouble for Chinese doctors in offering genetic counselling, treatment and management. The present phenotype-genotype correlation data observed from 117 Chinese TI patients could provide solid basis for improving the understanding this problem (table 2). On the basis of the clinical pictures and statistical results of hematological parameters, type I patients were more clinically severe than type II patients. This is a patient-based study in which the in-patient samples were recruited during the period of about 2 years from Guangxi and Guangdong Provinces, where the thalassemia are most endemic in southern China. We believe that the molecular basis of TI obtained from this study should be generalizable to the Chinese TI patients of the various regions in southern China because the differences between regions in southern China for the ß-thalassemia mutation spectrum in the southern Chinese are not statistically significant according to previously observations [1][2][3]5]. However, there were some milder forms of TI patient's samples who were outpatient we might have missed and most of heterozygotes (10/14) whose negative results could not be explained were detected in Guangxi, two limitations of our study should be noted: (1) Maybe there was a broad spectrum of genetic defects in the Chinese TI patients, such as no the HbE/β + TI patients could be identified in our cohort. (2) The finding of heterozygotes with TI phenotype might not provide representitive profile in the Chinese population. According to our molecular analysis, three main characteristic features of Chinese TI patients were found. First, the homozygous or compound heterozygous forms with minor β-thalassemia mutations (the β + -thalassemia alleles) were a common pathogenetic mechanism of disease in Chinese TI patients, in which 49 out of 59 patients was found to having β-globin genotypes of β + /β + (11 cases) and β + /β 0 (38 cases), among them 41 with co-existing normal α-globin genes and 8 with mutant αglobin allele (table 2). In other words, the β + -thalassemia mutation is a major contribution of genes to the Chinese TI patients. In addition to this, additional an α-thalassemia defects may reduce the severity of homozygous β-thalassemia. We observed 14 such cases (23.7%, 14/ 59) with co-inheritance of β-thalassemia (β + /β + or β + /β 0 ) and α-thalassemia defects in the patient group. Second, the affect of an XmnI polymorphism in increasing HbF levels was not as important as it was in other populations. The ratio of XmnI [(+/+), (+/-), (-/-) and (NA, not available)] pattern was detected to be [1:30:75:11] in 117 cases, while the same ratio in 52 Iranian TI patients was [20:20:10:2] [16]. In India, among 73 TI patients, 20 of them (27.4%) were confirmed to be homozygous for the XmnI polymorphism [17]. Third, the molecular mechanism of more than half of the thalassemia heterozygote cases (14/20) was unexplained. Twenty individuals carried only one β-thalassemia allele: one (5%) was dominantly inherited; five (25%) had coinherited triplicated α-globin genes; and 14 (70%) had normal α-globin genes. In contrast, the α-globin gene triplication was determined to be the predominant factor of Indian heterozygous thalassemia patients (14/23 cases) [17]. In Ho's research, among 22 heterozygous β-thalassemia intermedia cases, 10 patients had α triplication [20]. Furthermore, no HbE/β + heterozygote was observed in our cases while 9/45 TI patients were HbE/ β + heterozygotes in Thailand [29], suggesting that HbE/ β + heterozygosity perhaps displayed a more milder form of TI in Chinese people, which would help doctors in providing appropriate genetic counselling. HbE/β-thalassemia results in a varied clinical expression ranging from severe transfusion dependence to a complete lack of symptoms [29,30]. In this study, there are 27 cases HbE/β 0 TI patients whose clinical phenotypes including Hb levels, transfusion, splenectomy, age at diagnosis, and so on, varies greatly (table 2). It is very difficult to predict the clinical phenotype of the HbE/β-thalassemia although it was reported that a system for classifying disease severity of HbE/β-thalassemia had been constructed [29,31]. The possible genetic mechanisms for phenotypic diversity of HbE/βthalassemia have been analyzed in some literatures [29,30,32]. We also found two patients that were compounds of β-thalassemia, δβthalassemia and α-thalassemia genes. This was rarely reported in other ethnicities.
Three β 0 /β 0 TI patients with normal α-globin gene remained unexplained despite extensive effort to examine the existence of known genetic modifiers linked to increase HbF levels (table 4). The non-deletional forms of HPFH were excluded in sequencing the promoter region of the G γand A γ-globin genes. Two cases were +/-of XmnI, which partially, but not completely explains the high level of HbF. The 3'HS1 (+179 C!T) and rs11886868 (T!C) variations were shown to correlate with increased HbF in European TI patients [8,11]. The three patients all had C/C genotypes of both at +179 of 3'HS1 and rs11886868. However, we could not consider that C/C genotypes of rs11886868 contributed to the high HbF level based on two facts. One is that the frequency of C/C genotypes was 0.867 in Chinese people from the HapMap data http://www.hapmap.org/. The second is that an additional 115 samples were analyzed, including 30 TI patients (25 β 0 /β + , 5 β + /β + ) and 85 normal individuals, in which all were found to be the C/C genotype (data not shown). The (AT) 8 N 12 (AT) 11 motif in the 5'HS2 site and the (AT) 9 (T) 5 motif in the -540 region had been previously found to be associated with elevated HbF [9,10], but we did not find them in the 3 cases.
We found 14 patients with heterozygous for β-thalassemia among our 117 TI patient cohort, in whom the βglobin gene was found to be structurally intact by sequence analysis and the second α-globin gene triplication was excluded. They mostly manifested a relatively mild form of TI because of the Hb levels being 68-95 g/ L, the average age at diagnosis being 7.5 ± 5.1 (years), no blood transfusion or occasionally received (1/14) and mild hepatosplenomegaly (5/14). As for the underlying defective targets, we focused on primary genetic modifiers that could modulate the imbalance of α/β for investigation in ten TI patients (table 5). We evaluated the effects of some potential modifiers involving one linked to the β-globin gene cluster, the LCR resides in the 5'HS2 and 5'HS3 regions [33], and three ones not to be linked to the β-globin cluster, AHSP [14], GATA-1 [12] and HRI. The later is a kinase that can be inactivated by hemin, which is very recently reported to modify the phenotypic severity of murine models of erythropoietic protoporphyria and β-thalassemia [15]. Sequence analysis of the core regions of 5'HS2 and 5'HS3 showed wild type sequences, thus excluded the possibility that mutations in these control regions reducing the expression of β-globin gene. No mutations in the cDNA sequences of both AHSP and GATA-1 have been found. In the case of HRI, the polymorphism of rs2639 and rs2640 were detected. The rs2639 (A!G) is a synonymous mutation and the rs2640 (A!G) is a missense mutation resulting in a K558R substitution. However, the affect on the function of this substitution is unknown since Lys (K) and Arg (R) are similar in amino acid properties. These results indicate the existence of causative genetic determinants have not been defined molecularly, in agreement with the similar conclusion from Rosatelli's experiment [34] although there were a few differences between the candidate targets for these two studies.
In this study, we reported two novel mutations in the UTR regions of the human β-globin gene ( Figure.1). The Term CD+32(A!C) mutation slightly reduced βglobin mRNA levels to 17.8% less than the normal level. The reduced β-globin mRNA levels were not detected in the individuals with Cap+39(C!T) mutation. The detailed mechanisms of two novel mutations are not clear at present. Functional experiments, such as analysis of mRNA stability and trafficking in erythrocytic cells should be performed in the future.

Conclusions
Chinese TI patients showed a high degree of heterogeneity in both phenotypic and genotypic aspects. Some cases could not be explained according to the present data. A more detailed analysis of genetic modifiers modulating the imbalance of α/β will provide more insight into the recognition of thalassemia intermedia. Moreover, other potential genetic determinants which have been demonstrated in cancers and single gene disorders including αand β-thalassemia, such as a gain-of-function regulatory SNP in a nongenic region [35], loss of heterozygosity in the β-globin gene [36], segmental duplications involving the α-globin gene cluster [37], uniparental disomy event [38], DNA methylation and chromatin alterations, should also be considered.