miRNA Expression Associated with HbF in Saudi Sickle Cell Anemia

Background and Objectives: Sickle cell anemia (SCA) is a hereditary monogenic disease due to a single β-globin gene mutation that codes for the production of sickle hemoglobin. Its phenotype is modulated by fetal hemoglobin (HbF), a product of γ-globin genes. Exploring the molecules that regulate γ-globin genes at both transcriptional and translational levels, including microRNA (miRNA), might help identify alternative therapeutic targets. Materials and Methods: Using next-generation sequencing we identified pre-miRNAs and mature miRNA expression signatures associated with different HbF levels in patients homozygous for the sickle hemoglobin gene. The involvement of identified miRNAs in potential SCD-related pathways was investigated with the DIANA TOOL and miRWalk 2.0 database. Results: miR-184 were most highly upregulated in reticulocytes. miR-3609 and miR-483-5p were most highly downregulated in sickle cell anemia with high HbF. miR-370-3p that regulates LIN28A, and miR-451a which is effective in modulating α- and β- globin levels were also significantly upregulated. miRNA targeted gene pathway interaction identified BCL7A, BCL2L1, LIN28A, KLF6, GATA6, solute carrier family genes and ZNF genes associated with erythropoiesis, cell cycle regulation, glycosphingolipid biosynthesis, cAMP, cGMP-PKG, mTOR, MAPK and PI3K-AKT signaling pathways and cancer pathways. Conclusions: miRNA signatures and their target genes identified novel miRNAs that could regulate fetal hemoglobin production and might be exploited therapeutically.


Introduction
Sickle cell anemia (SCA) is a result of homozygosity for a mutation in codon 6 of the β-globin gene (HBB; GAG→GTG) that leads to the production of sickle hemoglobin (HbS; glu7val; α2βS2). Fetal hemoglobin (HbF; α2γ2) encoded by HBG2 and HBG1 (HBG1/2) prevents the polymerization of deoxyHbS, the proximate cause of the phenotype of SCA, and therefore the major modulator of this disease [1,2].
HbF levels vary 2 to 4-fold among the five common HBB haplotypes associated with the HbS gene. The Arab-Indian (AI) haplotype is associated with HbF that averages~17% in adults compared with 5 to 7% in the Bantu, Benin and Cameroon haplotypes and 10% in the Senegal haplotype [2,[13][14][15]. Even within the AI haplotype there is variation of HbF levels. The causes of high HbF in AI haplotype SCA are not fully explained [16]. To determine whether miRNAs have an association with HbF levels in AI haplotype SCA we explored the complete hairpin and mature miRNA expression profile of HbS homozygotes with "high" and "low" HbF levels.

Study Population and Samples
Participants aged from 3 to 42 years, 29 males (49%) and 30 females (51%), were clinically diagnosed by high performance liquid chromatography (HPLC) of hemoglobin. HbS homozygotes who were not having an acute vaso-occlusive episode or who were not currently on hydroxyurea, and normal hemoglobin controls were recruited from Qatif Central Hospital, Al-Qatif, Saudi Arabia. Blood samples (8-10 mL) were collected into EDTA tubes from 59 individuals that included 31 SCA patients and 28 healthy individuals without SCA. Blood counts were carried out by an automated clinical hematology analyzer (Sysmex, K-1000, Sysmex Corporation, Kobe, Japan), the percentage of HbS, HbA2 and HbF were measured by HPLC on the Bio-Rad Variant II Hb Testing System (Bio-Rad Laboratories, Hercules, CA, USA).
Ethical approval for the study was obtained from the Institutional Review Board Committee (IRB-2019-01-107) at Imam Abdulrahman bin Faisal University. The study was conducted according to the ethical principles of the Declaration of Helsinki and Good Clinical Practice Guidelines.

Magnetic Separation of CD71+ and CD235+ Cells
Peripheral blood mononuclear cells (PBMC) were isolated using density gradient Ficoll-Histopaque (Merck, St. Louis, MO, USA). PBMC were labeled with CD71 Mi-croBeads (Miltenyi Biotec, Bergisch Gladbach, Germany), which were used for the positive selection of human reticulocytes. The cell suspension was loaded onto a MACS ® Column which was placed in the magnetic field of a superMACS separator. The magnetically labeled CD71+ cells were retained on the column. After removal of the column from the magnetic field, the magnetically retained CD71+ cells were eluted as the positively selected cell fraction. Similarly, Glycophorin A+ (CD235+) cells were magnetically separated after labeling with Glycophorin-A MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). CD235a (Glycophorin A) MicroBeads were used for the positive selection of human erythroid cells as single-pass transmembrane glycoprotein is expressed on mature erythrocytes. These isolated cells were fluorescently stained with CD71-FITC and CD235-APC antibodies and confirmed by flow cytometry (Guava, Millipore, Burlington, MA, USA). Both these cell types were further processed for miRNA isolation.

RNA Extraction and Quantification
Total RNA was isolated using miRNeasy Mini kit (Qiagen, Hilden, Germany) in accordance with the manufacturer's instructions to obtain 50 µL of extracted sample. The RNA isolation protocol utilizes QIAzol Lysis reagent, chloroform and ethanol. The final elution was with 50µL of RNase-free water and was immediately quantified using the Qubit ® microRNA Assay Kits and the Qubit ® 4.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). The quantified RNA was stored at −80 • C until library preparation.

miRNA Library Construction and Sequencing
Aliquots of 100ng of miRNAs, derived from the above-mentioned procedure, were used for the preparation of miRNA libraries. Sequencing-ready cDNA libraries using the Small RNASeq Library Prep kit for Illumina (Lexogen, Austria) were prepared according to the manufacturer's instructions. Multiplexing indices were introduced during the PCR amplification step to distinguish each sample after the pooling phase, allowing multiplexing of up to 96 libraries [17]. The library product was cleaned and concentrated with a magnetic bead-based purification protocol [18]. Lastly, the sequencing step was performed with NGS technologies using Illumina platform NextSeq 500 and the NextSeq 500/550 High Output v2.5 kit (150 cycles).

Bioinformatic Analysis of the Raw Sequencing Data
FASTQ files were generated via Illumina bcl2fastq2 (Version 2.17.1.14-http://support. illumina.Com/downloads/bcl-2fastq-conversion-software-v217.html accessed on 16 June 2021) starting from the raw sequencing reads produced by Illumina NextSeq sequencer. MiRNAs were mapped on miRbase hairpins using SHRiMP software. Differential expression analysis for miRNAs was performed with the R package DESeq2 (Ver 1.18.1) (R software, Vienna, Austria). MiRNAs with more than 5 counts were retained for further analysis. The miRNA expression is considered differentially significant if the log2(disease sample/healthy control) ≥ 1 and a False Discovery Rate (FDR) ≤ 0.1. Hence, a minimum |Log2FC| of 1 and an FDR lower than 0.1 as thresholds to differentially expressed genes to maximize the sensitivity of this analysis and to perform a massive screening and identification of candidate miRNAs. For comparative analysis, we dichotomized HbF levels into low (6.35 ± 3.57%) and high (21.31 ± 6.73%) groups [15,19].

Target Gene and Interaction Analysis
The involvement of identified miRNAs in potential SCD-related pathways (predicted target genes and predicted pathways) was investigated with the DIANA TOOL and miR-Walk 2.0 database. Significantly differentially expressed miRNAs in SCA patients were considered for the analyses conducted on the mirWalk ver-2.0 to predict potential targets of these miRNAs [20]. Analysis was preferred for only validated results, using miRTar-Base filter option and were compared using other popular algorithms, such as miRDB or Target Scan. DIANA-mirPath was used to hierarchical targeted genes clusters-based pathway identification utilizing the exact significance (p-value threshold set at 0.05) after FDR correction (Benjamini and Hochberg) and exclusion of pathways with few targeted genes nodes.
Similarly, miR-451a was upregulated in reticulocytes (p = 7.10 × 10 −04 ). Expression of miRNA-451 is elevated during erythroid cell differentiation. Fang and Bartel [25] reported that miR-451, which derives from a hairpin with a suboptimal terminal loop and a suboptimal stem length, accumulates 40-fold higher levels when clustered with a helper hairpin. miR-451 was upregulated in reticulocytes by approximately 2.5-fold in the high HbF
The zinc finger proteins, ZNF802 and ZNF410 are HbF gene repressors. Induced deficiency of ZNF802 resulted in an increase in HbF to 35.0 ± 3.5% [27]. Knockout of ZNF410 reduced Chromodomain Helicase DNA Binding Protein 4 (CHD4) levels by 60%, enough to substantially derepress HbF [28]. ZNF154, ZNF706, ZNF655, ZNF789, ZNF394 and ZNF618 were among the miRNA targets of the present study. Similarly, other than the well-studied BCL11A, BCL7A and BCL2L1 were also among the target genes in the present study. BCL2L1 has been associated with HbF gene expression in patients with SCA, where overexpression of BCL2L1 resulted in increased HBG2/1 expression fourfold and F-cells by 13% [29]. Likewise, other than the known KLF1, an interaction between KLF6, a new potential regulator of erythropoiesis and miR-2355-5p was identified, and its downregulation significantly raised γ-globin mRNA [30]. miR-181c-5p targeted KLF6, and the significantly downregulated miR-3609 targeted the binding site of KLF6 [31].
Hematopoietic stem cell differentiation is associated with activation of multiple signaling pathways including MAPK and PI3K [32][33][34]. We found novel miRNAs, miR-5787, miR-3609 and miR-483-5p associated with these signaling pathways might be potentially exploited as therapeutic targets. Histone deacetylase inhibitors like Apicidin and Tricostatin, increase H3 and H4 acetylation in the region adjacent to the HBG2/1 promoters [35] and induce γ-globin gene expression [36] by activating MAPK signaling pathway.

Conclusions
miRNA signatures and their target gene-pathway analysis identified novel miRNAs that could regulate fetal hemoglobin production and might be exploited therapeutically. Additional studies in erythroid cell lines and primary erythroid progenitor cells are needed to define how any of the identified miRNAs might modulate HBG2/1 expression and the accumulation of HbF and if some of the HbF-associated miRNAs can account for the high levels of HbF in AI haplotype SCA.