Significance of heme oxygenase-1(HMOX1) gene on fetal hemoglobin induction in sickle cell anemia patients

Though the patients with sickle cell anemia (SCA) inherit same genetic mutation, they show considerable phenotypic heterogeneity. It has been observed that patients with elevated fetal hemoglobin (HbF) levels have a relatively mild clinical course. There is sparse literature on the association of higher HbF levels leading to reduction in the oxidative stress in SCA patients. Hence in this study, the significance between the HMOX1 gene polymorphisms and the HbF levels has been studied. Preliminary screening was carried out. Genotyping of 3 variants in the HMOX1 gene was performed in 90 SCA patients and 50 healthy controls by PCR–RFLP, GeneScan and direct DNA sequencing. It was observed that SCA patients with higher HbF levels, showed improved hematological indices with an inverse effect on HbS levels. The TT genotype of rs2071746 (A→T) polymorphism was found to be associated with elevated HbF levels (P: 0.012). Also, the long form (> 25 GT repeats) of rs3074372 (GT)n repeats was found to be linked with increased HbF levels. We could not find any association of rs2071749 (A→G) polymorphism with the HbF levels. As, the sickle cell anemia patients show significant oxidative stress due to hemolysis, the study of polymorphisms in the HMOX1 gene may act as a potential independent marker for elevated HbF levels.

in oxidative stress pathway and which also plays a role in modulating the HbF levels will be of importance as this may assist in selecting patients with milder or severe clinical disease.
Thus in this study we have screened for the 3 polymorphisms in the HMOX1 gene and have evaluated their association of genotypes with the HbF levels in sickle cell homozygous patients. These 3 polymorphisms in the HMOX1 gene rs3074372 (GT)n repeats, − 413: rs2071746 (A→T) and an intronic variation rs2071749 (A→G) have been earlier studied in disease conditions like Chronic Obstructive Pulmonary Disease (COPD), Parkinson's disease, HIV-induced CNS neuroinflammation, Cardiovascular disease, with their effect on developing a pathologic condition. Further recent study in the Brazilian population, showed that, the HMOX1 rs2071746 genotype frequencies were 24.3% (AA), 48.6% (AT) and 27.0% (TT) in the SCA patient group and 28.2% (AA), 52.7% (AT), 19.1% (TT) in the control group. Similarly, for the (GT)n repeats, genotype frequencies of the long form repeats were higher 56.0% (LL), 37.3% (SL) as compared to the shorter repeats 6.7% (SS) in the SCA patient group. Conversely in control group, the frequencies were found to be: 53.1% (LL), 40.0% (SL) and 6.9% (SS) 11 . rs2071749 (A→G) polymorphism has not been earlier studied in SCA patients.
As there is a sparse documentation of these polymorphisms in Sickle cell anemia (SCA) patients, we selected these 3 polymorphisms, to understand their prevalence and association with HbF levels in Indian SCA patients.

Materials and methods
Patient selection. 90 sickle cell anemia patients referred to the Department of Hematogenetics, National Institute of Immunohaematology and 50 healthy controls were enrolled in the study. The study was approved by the National Institute of Immunohaematology-Institutional Ethics Committee and the samples were obtained after a written informed consent from the patients. For patients below the age of 18 years we obtained the informed consent from parents/Legal guardians. All methods were carried out in accordance with relevant guidelines and regulations (declaration of helsinki). Since the HbF levels are influenced with hydroxyurea therapy, the patients on hydroxyurea therapy were excluded from this study.
Preliminary screening. 5 ml of peripheral blood EDTA samples were collected from the patients and the healthy controls. Red cell indices were measured on an automated blood cell counter (Sysmex KX-21). HbA 2 , HbF and HbS levels were measured using cation exchange HPLC on the Variant Hemoglobin Testing System (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

Molecular analysis.
Genomic DNA was isolated from peripheral blood leucocytes using the QIAamp Blood Mini Kit. The HbS confirmation was carried out using Covalent Reverse Dot Blot technique 12 . 3 polymorphisms in the HMOX1 gene based on their presumptive functional effects and on the basis of the literature survey were selected. The rs3074372 (GT)n repeats were detected using fluorescently labelled primers with primer sequence as followed:F'FAM:5′AGA GCC TGC AGC TTCT3′ and R:5′ACA AAG TCT GGC CAT AGG AC 3′ and PCR was conducted 13 . The PCR products were loaded along with a size standard LIZ (− 250) (Applied Biosystems) onto ABI genetic analyser 3130 XL, under fragment analysis protocol and the fragment size was analysed by GeneMapper v4.1 (Applied Biosystems) software. The rs2071746 (A→T) polymorphism was detected by direct DNA sequencing, using BigDye terminator mix (Applied Biosystems) with already described primers and protocol 14 . The rs2071749 (A→G) polymorphism was detected by using MluCI restriction enzyme digestion (New England BioLabs) 13 . Statistical analysis. The genotypic and allelic frequency was calculated. Allele frequency was calculated as the number of occurrences of the test allele divided by the total number of alleles in a given group. The differences among the HbF levels within the genotypes of the each polymorphism were calculated by using non parametric test (Mann-Whitney within 2 groups) or ANOVA (more than 2 groups were involved), using GraphPad Prism version 6.0 software (GraphPad Software, Inc., San Diego, CA). The linear regression analysis was used to construct the genetic models to understand the association of the single nucleotide polymorphisms (SNPs) with the response factor (HbF levels) using SNPStats software (https ://bioin fo.iconc ologi a.net/SNPst ats) 15 . Further, SNP-SNP interaction analysis was performed by using generalised multifactor dimensionality reduction [GMDRv1.0] software, where in the most effective SNP associated with HbF level was determined.

Results
A total of 90 Sickle cell homozygous patients with Arab-Indian haplotype, not on hydroxyurea therapy (Age range: 1 to 41 years, mean age: 14 years) and 50 healthy controls (Age range: 5 to 52 years, mean age: 18 years) were recruited for the study. The hematological analysis was carried out. Table 1 shows the details of hematological analysis of the patient and the control group. In order to understand the effect of elevated HbF levels on the hematological indices, the patients were classified into 2 groups considering the median HbF levels of 18.1%.It was observed that the hematological indices were significantly improved in patients with higher HbF levels as compared to the other group. It was also observed that the HbS levels showed a significant negative correlation with the HbF levels [Pearson Correlation coefficient (R): − 0.422, P < 0.0001]. (Fig. 1).
The genotype distribution for the three polymorphisms was consistent with the Hardy-Weinberg equilibrium. The genotype and the allelic distribution of the 3 polymorphisms [rs2071749 (A→G), rs2071746 (A→T) and rs3074372 (GT)n repeats] in both healthy and the patient group was calculated. Only the genotypic frequency of the long form of the (GT)n repeats was found to be significantly higher in the patient group [P < 0.002]. The genotypic and allelic frequencies of the long form of the (GT)n repeats were also significantly higher in SCA patients with HbF levels > 18.1 as compared to the other SCA patient group with HbF levels < 18.1% (Table 2).
Scientific Reports | (2020) 10:18506 | https://doi.org/10.1038/s41598-020-75555-y www.nature.com/scientificreports/ The genotypic variations were then compared with the response (HbF) level in the overall patient group. The genotypes of the 2 polymorphisms: rs2071749 (A→G) and rs3074372 (GT)n repeats did not show any significant association with the HbF levels (P value: 0.64 and 0.10 respectively, using Mann Whitney non parametric test). However the mutant genotype 'TT' of rs2071746 polymorphism was found to be significantly associated with the HbF levels (P: 0.012, Mann Whitney non parametric test) (Fig. 2). Further, generalised multifactor dimensionality reduction [GMDR] analysis was carried out to understand the interaction of the covariates (SNPs) in elevating the HbF levels. It was observed that the presence of mutant allele T of rs2071746 A→T polymorphism was higher in SCA patients in raised HbF group (Fig. 3). Multivariate analysis using different genetic models indicated that the T allele of rs2071746 A→T variant was significant for the Co dominant and dominant genetic model with the TT genotype resulting in significant induction of HbF levels (Table 3).
Thus these observations are in concordance with the univariate analysis of this SNP with the HbF levels, thus suggesting its association in elevating the HbF levels in SCA patients.

Discussion
Sickle cell anemia is a monogenic disorder caused by the homozygosity of single β globin gene mutation at codon 6 position. The deoxygenated sickle hemoglobin has differed physiological properties and the sickle RBCs show altered rheological effects. Polymerization of HbS, precipitates in RBCs leading to disease related crises such as anemia, vaso-occlusion, acute chest syndrome, ischemia reperfusion, secondary bacterial infection and fever 16 . Though, SCA follows a simple mendelian inheritance, the disease shows varied clinical phenotypes. This is attributed to various genetic modifiers such as coexisting α-thalassemia and the innate ability to produce elevated HbF levels 16 . Many studies have shown that there is strong association between elevated HbF levels and amelioration of the disease severity in SCA patients, as elevated HbF level delays the polymerisation of the sickle hemoglobin 17 .
Thus the effect of HbF levels on the hematological parameters and the HbS levels was first analysed. In this study, it was observed that the patients with higher HbF levels showed significantly improved hemoglobin (P: 0.01) as compared to the other group with lower HbF levels. A similar result was observed in Nigerian SCA children, where in the elevated HbF levels showed a direct relationship with hemoglobin concentration. Thus determination of HbF and its induction may be useful to maintain optimal hematological state of patients with sickle cell anemia 18 . In this study it was also observed that the sickle hemoglobin showed a significant inverse correlation with the HbF levels.
The complication of SCA involves generation of excessive free radicals, due to an electron transfer between the heme iron and oxygen, thus causing an oxidative stress 2 . The pro-oxidative and proinflammatory environment created by free heme, activates an antioxidant gene heme oxygenase 1 (HMOX1) to protect cellular mechanisms. Further recent a study have shown that low doses of carbon monoxide (the by product of heme metabolism) plays a key role in modulating the Nrf2 pathway that is linked to HbF regulation 19 . . The left bar represents the SCA cases with higher HbF levels and the right bar represents SCA cases with lower HbF levels. The dark grey tiles represent the genotypes associated with higher HbF levels. Among the 3 SNPs , it could be observed that the presence of mutant allele T of rs2071746 (A→T) polymorphism may be associated with higher HbF levels. www.nature.com/scientificreports/ Hence, we intended to analyse 3 polymorphisms located in HMOX1 gene, and their direct association with HbF levels, as this may be of importance in predicting the disease severity of SCA patients. Study by Nacoulma et al., showed that increase in HbF levels in SCD patients, reduce the clinical manifestation of the disease by lowering the oxidative stress 20 .
The most significant SNP associated with higher HbF levels in SCA patients was found to be − 413A/T (rs2071746) polymorphism. Our study demonstrated that the mutant allele T in the heterozygous state and the homozygous form was found to be higher in both normal and SCA patients, as compared to the wild type AA genotype of − 413A/T (rs2071746) SNP. A similar frequency of the mutant allele was observed Bakr et al., in the sickle cell disease patients, where in the mutant genotype AT and TT was found to be at a higher frequency in both patients (53% and 39% respectively) and unaffected control groups (44.4% and 41.4% respectively) as compared to the wild genotype (AA) (8% and 14.1% respectively) 21 . However, the A (wild type allele) formed the major allele in the Iranian SCD patients and the normal control 22 . In the Brazilian cohort, AT genotype, was found to be the most common genotype in both patients and control group (44% and 62% respectively) 14 .
In connection to the association with the HbF levels, the mutant allele T of rs2071746 SNP, was found to be associated with elevated HbF levels in SCA patients. A similar result was observed by Gil et al., where the patients carrying the TT genotype had significantly higher levels of HbF when compared with other genotypes (P value: 0.0131). However in this contrast, in the study by Bakr et al., the TT genotype did not show any association with the HbF levels, but was related to improved clinical conditions with significant reduction in vaso-occlusive crisis (VOC), stroke, hospitalization rate , and better response to hydroxyurea treatment 21 .
The wild allele ' A' is reported to be associated with expression of the HMOX1I gene, however the mutant allele T of rs2071746 (A→T) diminishes the gene expression thus reflecting its functional significance. This polymorphism is also been well studied in the association with other diseases. In a study by Ono K et al., they observed that the AA genotype reduced the incidence of ischemic heart diseases, possibly due to high expression level of HMOX1 23 . Similarly, Ono et al., have showed that AA genotype is associated with increased incidence of hypertension in women 24 .
The second polymorphism studied was the intronic variant rs2071749 (A→G). In this study, though the patients with AA genotype, showed higher HbF levels, they could not achieve significance. This polymorphism has not been studied in SCA patients, nonetheless it is found to be associated with its gene expression and the lung function. Study by Park et al., showed that A allele of this polymorphism, had a protective effect on lung function in paediatric cystic fibrosis patients 25 . In another study by Jiménez-Osorio et al., the carriers for AA genotype, were predisposed to obesity as they showed lower BMI values. The rs2071749 SNP is located within an intron of the HMOX1 gene and, therefore, does not have an obvious functional impact 26 . However, this polymorphism is found to be in strong linkage disequilibrium with another promoter polymorphism (rs3761439) which influences the HMOX1 gene expression, thus indirectly gaining the favourable importance 25 .
The HMOX1 (GT)n promoter microsatellite has been evaluated for its role in a number of clinical outcomes, including cardiovascular and pulmonary diseases 27,28 . In this study, we tried to evaluate the role (GT)n promoter polymorphism with the HbF levels in SCA patients. We observed that, the longer repeats were significantly higher in SCA patients as compared to the normal control (P: 0.002). Also, the allelic frequency of longer repeats, was significantly higher in SCA patients with higher HbF levels (P: 0.03). SCA patients with longer repeats showed elevated HbF levels, however could not achieve the significance. The (GT)n repeat alleles were first studied by Bean C et al., to determine the association of this polymorphism with the clinical phenotype in SCD patients. They witnessed, that the shorter HMOX1 (GT)n alleles showed a protective effect and the patients showed significant reduction in hospitalization and in acute chest syndrome 29 .
Considering the functional effect of these polymorphisms, the long form of (GT) n repeat and the T allele of rs2071746 (A→T) SNP both located in the promoter region, have been reported to reduce the HMOX1 gene expression leading to a decrease in the protein formation 30 . Belcher et al., highlighted the role of hemoxygenase I (HO l) in the transgenic SCD mice and showed that, the elevated or reduced expression of the HO1 enzyme may be inversely associated with hypoxia/reoxygenation-induced vascular stasis. These results suggest the encouraging role of HO in inhibiting the vascular inflammation and vaso-occlusion in SCA patients 30 .
Studies have shown that the T allele of rs2071746 (A→T) polymorphism and the long form of (GT)n repeats lead to the reduction in the HMOX1 gene expression. Thus, a possible hypothesis for elevated HbF levels in SCA patients with T allele and long form of (GT)n repeats could be due to disrupted stress hematopoiesis of stem cells and progenitors. This leads to premature release of erythroblasts into the circulation, which in turn elevates the HbF levels 30 .Our hypothesis is supported by the review of Mabaera R et al., that the presence of a cellular stress, may activate integrated stress response, p38 mitogen-activated protein kinase (MAPK) and cyclic adenosine monophosphate (cAMP) signaling pathways which ultimately augment in indirect activation of HBG gene and increase the HbF levels 31 . Also an interlinked role of low concentration of carbon monoxide in activating the NRF2 gene transcription was described by Wang et al., as a therapeutic agent in ischemia-reperfusion brain injury 19 . NRF2 transcription factor a master regulator of cellular oxidative stress response has been shown to positively regulate the γ-globin transcription with an associated increase in HbF levels 32 . Thus, our findings, support the synergetic role of these factors in elevating the HbF levels.
HbF levels are hereditable and the phenomenon of raised HbF in SCA patients where the HMOX1 gene polymorphism associated with HbF levels is absent, may be due to the other SNPs linked to HbF regulation like mutations within the β-globin gene cluster which includes XmnI polymorphism (− 158 C→T) or in the γ globin promoter region 33 . In Indian sickle cell anemia patients, the Arab-Indian haplotype which is linked to XmnI polymorphism, is found to be the most prevalent haplotype. However though the patients inherit the Arab Indian haplotype a wide range of HbF variability is seen. Thus, this study highlights, a presence of alternative mechanisms for HbF induction. To date, the studies have shown that increased HbF level involves 2 mechanisms. The first mechanism states that there is direct effect on the γ-globin gene expression, either by activation of the Scientific Reports | (2020) 10:18506 | https://doi.org/10.1038/s41598-020-75555-y www.nature.com/scientificreports/ γ-globin gene transcription or by the inhibition of γ-globin gene repression. Some quantitative trait loci (QTL) have been identified as important determinants for increase in HbF levels in SCD patients such as HMIP locus on chromosome 6, the BCL11A locus on chromosome 2, the Xp22.2 region of the X chromosome and the 8q region on chromosome 8 and more recently KLF1 locus 34 . Second possible mechanism is alteration of kinetics of erythroid maturation and differentiation, with release of more erythroid progenitors which predominantly have high HbF concentration 24 . With our current data, it could be speculated that HMOX1 gene polymorphisms may promote HbF induction, through this alternative secondary mechanism independent of the γ-globin gene regulation pathway.

Conclusions
True understanding of how the genetic variants in HMOX1 gene functions, requires an integrative study of the expression patterns and their impact on other erythroid genes. Thus studying these SNPs will help in understanding the mechanism of HbF induction in hemoglobinopathy patients. In future the cumulative study of the genetic modifiers of fetal hemoglobin may help in forecasting of disease severity.