Brazilian Journal of Medical and Biological Research Online Provisional Version

Fetal hemoglobin (HbF), encoded by the HBG2 and HBG1 genes, is the best-known genetic modulator of sickle cell anemia, varying dramatically in concentration in the blood of these patients. This variation is partially associated with polymorphisms located in the promoter region of the HBG2 and HBG1 genes. In order to explore known and unknown polymorphisms in these genes, the sequences of their promoter regions were screened in sickle cell anemia patients and correlated with both their HbF levels and their betaS-globin haplotypes. Additionally, the sequences were compared with genes from 2 healthy groups, a reference one (N = 104) and an Afro-descendant one (N = 98), to identify polymorphisms linked to the ethnic background.The reference group was composed by healthy individuals from the general population. Four polymorphisms were identified in the promoter region of HBG2 and 8 in the promoter region of HBG1 among the studied groups. Four novel single nucleotide polymorphisms (SNP) located at positions -324, -317, -309 and -307 were identified in the reference group. A deletion located between -396 and -391 in the HBG2 promoter region and the SNP -271 C-->T in the HBG1 promoter region were associated with the Central African Republic betaS-globin haplotype. In contrast, the -369 C-->G and 309 A-->G SNPs in the HBG2 promoter region were correlated to the Benin haplotype. The polymorphisms -396_-391 del HBG2, -369 SNP HBG2 and -271 SNP HBG1 correlated with HbF levels. Hence, we suggest an important role of HBG2 and HBG1 gene polymorphisms on the HbF synthesis.


Introduction
Clinical heterogeneity is an important characteristic of sickle cell anemia (SCA) and other hemoglobinopathies (1).The molecular data regarding diseases comprise the target gene and a large number of known and unknown DNA changes.These alterations concomitantly modify the expression of several genes, including the gamma globin genes (HBG2 and HBG1), responsible for the synthesis of the γ chains present in fetal hemoglobin (HbF) (2)(3)(4).Gene expression is modulated by a large number of polymorphisms located in the β-globin gene cluster.Of these, the HBG2 -158 C→T variation is consistently linked with an increase of HBG2 expression in cis, which are DNA sequences in the vicinity of the structural portion of a gene that are required for gene expression (5).Thus, there are cis-acting sequences that behave both positively and negatively in transcriptional regulation.For this reason, many studies have focused on learning about the

O N L I N E P R O V I S I O N A L
reactivation of the dormant HBG genes during adulthood and in understanding which cis-acting elements govern HbF synthesis.The regulatory sequences 5' of HBG2 and HBG1 harbor at least eight polymorphic markers.These include the 4-bp deletion (AGCA) at positions -225 to -222 upstream of HBG1 (2) and the HBG2 -157 C→T single nucleotide polymorphism (SNP), which has been reported to be associated with high HbF levels (6).Moreover, some of the HBG2 and HBG1 polymorphisms are associated with the major β S -globin gene haplotypes, which are named Benin (Ben), Central African Republic (CAR), Senegal (Sen), Arab-Indian and Cameroon (Cam), according to their geographical origin and the ethnic groups in which they are frequently found (1,7).For example, there is a T→A variation in the HBG1 promoter region (-499) that is associated with the Ben haplotype among Sicilian and North American individuals (8).Because of the wide variability in clinical expression, and the consistent differences in HbF levels among SCA patients with different haplotypes, we have initiated sequence analyses of the HBG2 and HBG1 promoter regions in order to identify differences that may be important for understanding the variations in γ-chain synthesis.The 5' promoter regions were chosen because of their potential regulatory role in gene expression.These regions are known to contain both positive (enhancer) and negative (silencer) regulatory elements upstream of the consensus promoter sequences (9).We report here sequence data for these regions of both HBG2 and HBG1 from different Brazilian groups, i.e., SCA patients and two healthy groups, a reference one and an Afro-descendant one.

Patients and control subjects
Blood samples were collected from 152 SCA patients, 104 apparently healthy individuals randomly chosen from the general population (reference group) and 98 Afro-descendant individuals.The SCA patients have been followed at the Hematology and Hemotherapy Bahia Foundation and were selected between 2004 and 2005.The median age (± SD) of this group of 73 women and 79 men was 10.7 ± 11.2 years.
The reference group consisted of volunteers (employed or their children) with a median age of 9.7 ± 12.2 years, 54 men and 50 women who went to the Faculty of Pharmacy of the Federal University of Bahia for laboratory analyses and who showed an AA hemoglobin profile.The SCA patient group and the reference group are from Bahia, Northeast region of Brazil, which has both the highest frequency of abnormal hemoglobin and the highest rate of race admixture, mainly of individuals of African, European and Amerindian origin (10).
The Afro-descendant group, from the Saracura and Arapema communities, is located in Pará, Brazil.This group consisted of 51 men and 47 women with a median age of 66 ± 14.1 years.In contrast to the SCA and reference groups, the Afro-descendant group derives from a slave community located in Pará, Northern Brazil, where the race admixture is lower, composed basically of individuals of African origin (11).
Individuals who reported any disease were excluded from the reference and the Afro-descendant groups.
The study was approved by the Human Subject Research Ethics Committee of the Oswaldo Cruz Research Foundation in Bahia (process No. 0025.0.0225.000-07)and Amazonia (process No. 209/03).Informed consent was obtained from the parents or guardians.

DNA analyses
Genomic DNA was isolated from peripheral blood leukocytes using the QIAMP DNA Blood Kit (Qiagen, USA) according to manufacturer protocols.The HBG2 and HBG1 promoter regions were sequenced (12) with an ABI Prism 3100 DNA sequencer using Kit BigDye 03 Terminator TM Sequencing Standards (Applied Biosystems, USA) and specific sequencing primers: 5'-TGA AAC TgT ggT CTT TAT gAA AAT Tg-3' forward HBG1, 5'-gCA CTg AAA CTg AAA CTg TTg CTT TAT Agg AT-3' forward HBG2, and 5'-ggC gTC Tgg ACT Agg AgC TTA TTg-3' reverse for both genes.The β S -globin gene haplotypes were identified in the SCA patients by the polymerase chain reaction and by restriction fragment length polymorphism techniques (13).

Statistical analyses
Descriptive analyses included gene and allelic frequencies, odds ratio as an estimate of relative risk and the 95% confidence interval (CI).The chi-square test was used to associate independent categorical variables and to test for the Hardy-Weinberg equilibrium.The level of significance was set at P < 0.05.All analyses were carried out using the Epi-Info software, version 6.04.

HBG2 and HBG1 promoter region polymorphisms
Table 1 shows the frequencies of the HBG2 and HBG1 promoter region polymorphisms in the two target groups and in the reference group.Each polymorphism was either a deletion or an SNP, and their frequencies were consistent with the Hardy-Weinberg equilibrium (Table 1) except the -396_-391 del, -369 C→G and -157 T→C of the Afro-descendant group, which was not calculated because there was only one genotype for each polymorphism.We identified four polymorphisms in the HBG2 promoter, which were located at positions -396 to -391, -369, -309, and -157.Additionally, we identified eight polymorphisms in the HBG1 promoter, located at -398, -369, -225 to -222, -271, -324, -317, -309, and -307.All polymorphisms showed a more homogeneous distribution pattern in the Afro-descendant group compared with the SCA and reference groups.
When we compared the frequencies of these polymorphisms between the SCA and reference groups, who are supposed to have similar genetic and environmental backgrounds, we found some interesting differences (Table 2).The 6-bp deletion located between -396 and -391 of the HBG2 promoter region was three times more frequent among SCA patients than in the reference group.Moreover, the other HBG2 promoter polymorphisms showed a higher frequency of mutant alleles among SCA patients, which were the SNPs -369 C→G, -309 A→G and -157 T→C.
The frequencies of the 4-bp deletion located at -225 to -221 and of the SNPs -398 G→A and -369 C→G of HBG1 were similar for the SCA and reference groups.On the other hand, the T allele for the 271 SNP was more frequent in the SCA group.

New polymorphisms identified in the reference group
We identified four new polymorphisms in the HBG1 promoter region, which were SNPs located at positions -324 (C→T), -317 (A→G), -309 (A→G), and -307 (G→A) (GeneBank Bankit #1186662).Interestingly, they were found together in a heterozygous pattern within three individuals from the reference group.

The polymorphisms -396_-391 del and -369 SNP in the HBG2 promoter are associated with the -271 HBG1 SNP and the -309 SNP HBG2, respectively
The chance of finding the mutant allele of the -271 HBG1 SNP in patients who had the HBG2 -396_-391 deletion was 20 times higher than in patients who did not have it.The same was true when we combined the -369 and -309 HBG2 SNPs, but the chance decreased approximately 6-fold in this case.The SCA patients who had the -396_-391 deletion within HBG2 showed a lower chance of having the associated -369 or -309 HBG2 SNP, with an odds ratio of 0.07 (95%CI = 0.03-0.17)or 0.14 (95%CI = 0.06-0.31),respectively.Similarly, the possibility of finding the -309 HBG2 and the -271 HBG1 SNPs together was dramatically lower than the possibility of finding each SNP alone (Table 3).

The polymorphisms HBG2 -396_-391 del and -271 HBG1 SNP are associated with CAR β S -globin gene haplotype
The frequencies of the β S -globin gene haplotypes were analyzed in 144 patients, and we found 58 (40.3%) with genotype CAR/Ben, 43 (30%) with Ben/Ben, 32 (22.2%) with CAR/CAR, 4 (2.8%) with Ben/Sen, 4 (2.8%) with Ben/ Atypical, 2 (1.4%) with Ben/Cam, and 1 (0.7%) with CAR/Cam.The CAR haplotype was strongly associated with the HBG2 -396_-391 deletion and with the mutant allele for the -271 HBG1 SNP (Table 4).However, one CAR/CAR patient did not have the mutant allele for the -271 HBG1 SNP.Among the heterozygous haplotypes, there were only one CAR/ Cam and 27 CAR/Ben with the HBG2 -396_-391 deletion.The mutant allele for the -271 HBG1 SNP was present in 21 CAR/Ben patients, but not in other heterozygous haplotypes.

The polymorphisms -369 and -309 SNP in HBG2 are associated with the Ben β S -globin gene haplotype
The mutant alleles for the -369 and -309 HBG2 gene SNP were found in patients with the Ben haplotype.Table 4 3 shows Ben chromosomes in association with these SNP.Among subjects who had the mutant allele for the -369 HBG2 SNP, 41 were Ben/Ben, 27 CAR/Ben, 2 Ben/Sen, 2 Ben/Atypical, and 1 CAR/Cam.In addition, the mutant allele for the -309 HBG2 SNP was present in 43 Ben/Ben, 35 CAR/Ben, 4 Ben/Atypical, 3 Ben/Sen, and 2 Ben/Cam subjects.On the other hand, 1 CAR/CAR patient was heterozygous for the -369 HBG2 SNP.

O N L I N E P R O V I S I O N A L
The polymorphisms -396_-391 del HBG2, -369 SNP HBG2 and -271 SNP HBG1 are associated with HbF levels In order to determine a possible relationship of the HBG2 and HBG1 polymorphisms with HbF levels, we classified the patients into four groups according to their HbF levels (Table 5).Most of the patients who did not have the -396_-391 HBG2 deletion showed HbF >5.0%.The presence of the wild-type allele for the -271 HBG1 SNP and the mutant allele for the -369 HBG2 SNP was associated with patients who had HbF levels >5.0%.The -309 HBG1 SNP and the remaining polymorphisms identified did not show any association with HbF levels (data not shown).

Discussion
The present study has identified both previously described and new HBG2 and HBG1 polymorphisms in three different population groups.It was believed that four of these polymorphisms may be important to SCA patients because of their apparent association with HbF levels and with the β S -globin gene haplotypes.Moreover, all polymorphisms were found to have a more homogeneous pattern of distribution in the Afro-descendant group, which consisted of individuals with a lower rate of racial admixture.This group was formed by descendants of two African slave populations living in Northern Brazil, a different region from that of the other two groups.Historical data suggest that about 90% of the slaves brought to Northern Brazil were from Angola, Congo and Mozambique (14).Both the SCA group and the reference group were from Bahia, Brazil, and both were composed of individuals with high rates of racial admixture.An estimated rate of 97% interethnic panmixia has been shown in Bahia, possibly due to the intensive African slave trade during which more than five million people from several African countries were brought to Brazil between 1550 and 1850, mainly between the 18th and 19th centuries (15).Moreover, Bahia was the only Brazilian State that restricted slave traffic through Ecuador, a fact that explains the correlation between genotype frequencies found in Bahia and Western Africa, principally the Bay of Benin region (10).These facts may explain the different frequencies of all polymorphisms found in the present study.
When we compared the polymorphisms of the HBG2 and HBG1 genes between the SCA group and the reference groups we found five of them with a different distribution in their genotypes.Since both groups have had similar genetic and environmental backgrounds, it was believed that these differences could be related to the β S chromosomes.Thus, the -396_-391 deletion and the -369 C→G and -309 A→G SNPs in the HBG2 promoter might be associated with the polymorphic sites linked to the β S haplotypes before the occurrence of this mutation.Although the age of the β S -globin gene mutation is not known with certainty, expansion of the mutation probably occurred in parallel to malaria, a strong selective factor, becoming endemic on the African continent (16).The -157 HBG2 SNP was recently described in SCA patients from the same region of Brazil, and it was suggested that the C allele at this position may be related to HbF levels (6).In contrast to the report of Adorno et al. (6), we found that 100% of healthy Afro-descendant individuals and about 90% of the SCA and reference group subjects had the variant allele for the -157 HBG2 SNP.Hence, this SNP seems to be simply an ethnic marker, and may not be related to HbF levels.
The -271 C→T SNP in the HBG1 promoter has been described in previous reports (2,16) to be exclusively located in β S chromosomes.However, in the present study, this SNP was also present among the reference and Afro-descendant groups.All 5 Afro-descendant individuals who had the T allele (mutant) were heterozygous for the β S mutation.On the other hand, 19 β A individuals from the reference group had this mutant allele instead of the C, which may be a result of a crossing over event between two heterozygous individuals for the β S mutation.Additionally, the SNP in HBG1 may be considered a gene conversion of a β S CAR haplotype because of HBG2-like sequences between bases -306 and -271 and extending to bases 25-1107, which have replaced the corresponding HBG1 sequence (17).
The four new SNPs in the HBG1 promoter (-324 C→T, -317 A→G, -309 A→G and -307 G→A) identified in the reference group may have origins in Europe or among the Amerindians since they were not found in the two other groups studied.Studies are necessary to confirm this explanation and to determine the genetic impact of these polymorphisms on HBG1 expression.
When the odds of finding the -396_-391 deletion in HBG2 together with the -271 HBG2 SNP were measured, a strong association between these two polymorphisms was identified.There was also an important association between the -369 and -309 SNPs in HBG2.These facts may indicate a similar origin of these closely associated polymorphisms or a possible association with the β S -globin haplotypes.
The major sequence differences in the HBG2 promoter region are in the β S chromosomes with the haplotypes CAR (-396_-391 deletion) and Ben (-369 and -309 SNPs).In addition, the -271 SNP in the HBG1 promoter was also strongly associated with CAR chromosomes.Only one CAR/CAR patient did not show this association, and our results are different from those obtained in previous studies (2,16) in which the investigators did not detect any non-CAR haplotype associated with the mutant allele for the -271 SNP in HBG1.This difference may be explained by multiple recombinations between heterozygous CAR and Ben haplotypes.
Finally, we stratified the SCA patients into two groups according to HbF levels to investigate the correlation between

O N L I N E P R O V I S I O N A L
these levels and the polymorphisms -396_-391 del HBG2, -369 SNP HBG2, -309 SNP HBG2 and -271 SNP HBG1, as shown in Table 5.Most of patients who did not have the -396_-391 del in the HBG2 gene had HbF levels >5.0 (P < 0.0001).The absence of the mutant allele (T) for the -271 SNP in the HBG1 gene also correlated with higher HbF levels (P < 0.0001).Conversely, the presence of the mutant allele (G) for the -369 SNP in the HBG2 gene correlated with higher levels of HbF (P = 0.006).The other polymorphisms did not show any relationship with HbF levels.These results are different from those reported by Bouhassira et al. (17) and by Lanclos et al. ( 2), but both groups of patients previously studied were African-American who had a genetic background different from the Brazilian SCA patients.Moreover, these gene polymorphisms may be part of other molecular events that act synergistically on HbF expression.Five patients were excluded from these analyses because they were younger than 6 years and this fact may have influenced the results.At birth, HbF represents 80-90% of the total hemoglobin synthesized and gradually decreases to 1% by 10 months in normal infants, but not in SCA children, who may have higher HbF levels up to the age of 6 years (18).
Further studies will be necessary to clarify the complex interaction between these specific sequences and certain trans-activating factors within specific developmental stages of erythroid cells and its possible impact on the clinical outcome of SCA patients.

Table 1 .
Frequencies of polymorphisms in the HBG2 and HBG1 promoter regions of patients with sickle cell anemia (SCA), of individuals from the reference group and of Brazilian Afro-descendants.

Table 2 .
Differences of HBG2 and HBG1 polymorphisms between patients with sickle cell anemia, and the reference group.

Table 3 .
Association of polymorphisms in the HBG2 and HBG1 promoter region of patients with sickle cell anemia (SCA, N = 152).