Enhanced induction of fetal hemoglobin by the combination of decitabine with RN-1 in β-thalassemia/HbE erythroid progenitor cells

Abstract

Background

Fetal hemoglobin (HbF; α₂γ₂) induction is a well-established approach for β-hemoglobinopathies, including sickle cell disease (SCD) and β-thalassemia. Decitabine, a DNA methyltransferase 1 (DNMT1) inhibitor, has been shown to effectively induce HbF production with a favorable safety profile. However, more potent therapeutic strategies are needed, particularly for β-thalassemia/HbE patients.

Methods

We evaluated the HbF-inducing efficacy of ten DNMT1 inhibitors in erythroid progenitor cells derived from β-thalassemia/HbE patients. To further enhance HbF induction, we investigated a combination treatment with decitabine and RN-1, a lysine-specific demethylase 1 (LSD1) inhibitor. HbF expression, cell viability, erythroid differentiation, and proliferation were assessed. Additionally, we investigated the association between treatment response and well-characterized single-nucleotide polymorphisms (SNPs) previously linked to HbF expression.

Results

Of the ten DNMT1 inhibitors tested, SGI-110, a dinucleotide analog of decitabine, exhibited similar HbF-inducing efficacy and toxicity profiles as decitabine at equivalent molar dose. The combination treatment with decitabine and RN-1 resulted in a robust additive increase in HbF expression in β-thalassemia/HbE erythroid progenitor cells, albeit with a slight reduction in cell viability. Additionally, the combination treatment improved the delayed differentiation phenotype in β-thalassemia/HbE erythroid cells, accompanied by a reduction in cell proliferation. Interestingly, individual variability in response to RN-1 and the combination treatments was observed, with major responders exhibiting significantly greater increases in HbF compared to minor responders. We identified two SNPs in the BCL11A gene (rs766432 and rs1427407) that were potentially associated with a higher likelihood of major response to treatments.

Conclusions

Our findings highlight the potential of targeting two distinct epigenetic corepressors within the γ-globin repressor complex to achieve robust HbF induction. The combination of decitabine and RN-1 represents a promising therapeutic strategy for β-thalassemia, warranting further investigation into the molecular mechanisms underlying individual response variability.

Graphical Abstract

- Decitabine and its analog SGI-110 have similar effects in inducing HbF.

- Combined decitabine and RN-1 treatment resulted in an additive increase in HbF.

- Responses to the RN-1 and combined treatments were divided into major and minor responders.

- Specific SNPs in the BCL11A gene may contribute to the observed response variability.

Supplementary Information

The online version contains supplementary material available at 10.1186/s10020-026-01421-8.

Introduction 

β-Thalassemia, an inherited hemoglobinopathy, has been recognized by the World Health Organization (WHO) as a significant global health burden (Weatherall and Clegg 2001). The disease arises from genetic mutations in the β-globin gene, resulting in a reduction or absence of β-globin chains and adult hemoglobin (HbA; α₂β₂) production. The pathophysiology of β-thalassemia primarily involves the accumulation of excess unpaired α-globin chains in erythroid cells, leading to ineffective erythropoiesis and red blood cell (RBC) hemolysis, the hallmarks of the disease (Taher et al. 2021). Clinically, patients present with chronic severe anemia, growth retardation, skeletal deformities, hepatosplenomegaly, and if left untreated, death (Taher et al. 2021). Coinheritance of the β-thalassemia allele and the hemoglobin E (HbE, HBB:c.79G > A) results in β-thalassemia/HbE disease, a prevalent severe form of β-thalassemia globally (Fucharoen and Weatherall 2012). Current standard treatments for β-thalassemia include lifelong regular blood transfusions and iron chelation therapy. While allogeneic hematopoietic stem cell transplantation (HSCT) is an available curative option, its application is limited by the availability of human leukocyte antigen (HLA)-matched donors and the risk of immunological complications (Taher et al. 2021). Recently, ex vivo gene therapies for β-hemoglobinopathies, including lentiviral β-globin gene addition (Kwiatkowski et al. 2024; Locatelli et al. 2022) and CRISPR-Cas9 mediated genome editing to induce fetal hemoglobin (HbF; α₂γ₂) (Locatelli et al. 2024), have been approved by the U.S. Food and Drug Administration (FDA) and have shown promising outcomes, with nearly all patients achieving transfusion independence. Although gene therapy holds curative potential for transfusion-dependent β-thalassemia (TDT) patients, its application remains restricted to those with favorable clinical characteristics (e.g., adequate stem cell reserves and no evidence of organ damage (Baronciani et al. 2021). Moreover, there are concerns regarding the long-term safety and efficacy of these therapies, including the potential for insertional oncogenesis with lentiviral vectors and off-target effects associated with gene editing (Hardouin et al. 2025). These advanced therapies also require specialized facilities, which may not be widely accessible in clinical settings due to their limited availability. Therefore, a drug-based therapy, which is a readily accessible and economically feasible therapeutic option, is highly desirable (Njeim et al. 2024).

Elevating HbF levels has been shown to ameliorate the clinical severity of β-thalassemia by reducing the burden of excess α-globin chains (Musallam et al. 2012; Nuinoon et al. 2010). Therefore, induction of HbF expression has been established as a successful therapeutic approach for β-thalassemia (Steinberg 2022). Hydroxyurea, currently the only FDA-approved HbF inducer for sickle cell disease (SCD), is also used off-label for β-thalassemia. However, its efficacy varies widely among patients with SCD and is more restricted in β-thalassemia (Chuncharunee et al. 2017; Fucharoen et al. 1996; Musallam et al. 2013). Specific genetic variations, such as XmnI and BCL11A polymorphisms, may explain the variability in hydroxyurea response among patients (Musallam et al. 2013; Sales et al. 2021). Additionally, hydroxyurea is associated with adverse effects, including myelosuppression and the potential for long-term carcinogenesis (Musallam et al. 2013). Therefore, more consistently effective and less toxic pharmacological HbF inducers or combination drug regimens are greatly desired.

Recent studies have identified several transcriptional repressors of the γ-globin genes, including testicular nuclear receptors (TR2/TR4), B-cell lymphoma 11a (BCL11A), and lymphoma-related factor (LRF, also known as ZBTB7A). These transcription factors recruit common epigenetic-modifying enzymes, such as DNA methyltransferase 1 (DNMT1), histone deacetylases (HDAC), lysine-specific demethylase 1 (LSD1), euchromatin histone lysine methyltransferases 1/2 (EHMT1/2), and protein arginine N-methyltransferase 5 (PRMT5), to cooperatively mediate γ-globin gene silencing (Hariharan and Nadkarni 2021). Small molecules inhibiting the activity of specific epigenetic corepressors have been extensively investigated for their potential to reactivate HbF expression (Lavelle et al. 2018; Yu et al. 2020). Some of these inhibitors include decitabine, a DNMT1 inhibitor (Olivieri et al. 2011); HQK-1001 (Fucharoen et al. 2013) and vorinostat (Mettananda et al. 2019), both HDAC inhibitors; tranylcypromine (Shi et al. 2013) and RN-1 (Cui et al. 2015; Kaewsakulthong et al. 2021), LSD1 inhibitors; and UNC0638, an EHMT1/2 inhibitor (Krivega et al. 2015; Nualkaew et al. 2020; Renneville et al. 2015).

Among these inhibitors, decitabine (5-aza-2’-deoxycytidine, DAC), a deoxycytidine analog, has been shown to increase HbF levels with a favorable safety profile in both SCD (Lavelle et al. 2018; Saunthararajah et al. 2008) and β-thalassemia patients (Olivieri et al. 2011). However, its clinical efficacy has been limited, and DAC is not currently used in general clinical practice. DAC reactivates γ-globin and HbF likely through the inhibition of DNMT1, leading to DNA demethylation at CpG sites within γ-globin promoters (Lavelle et al. 2007, 2018; Saunthararajah et al. 2003). The clinical use of DAC is further constrained by rapid degradation through cytidine deaminase (CDA), which significantly reduces the plasma half-life, tissue distribution, and oral bioavailability of DAC. To address this limitation, co-administration of DAC with tetrahydrouridine (THU), a competitive CDA inhibitor, has been shown to improve DAC stability and bioavailability, thereby enhancing HbF induction efficiency (Lavelle et al. 2012; Molokie et al. 2017). Importantly, this DAC–THU regimen is currently being evaluated in the ASCENT-1 Phase 2 clinical trial (NCT05405114) for SCD, underscoring its translational relevance to hemoglobinopathies.

Development of DNMT1 inhibitors with superior biological and pharmacokinetic properties compared to DAC could further enhance HbF production. Moreover, recent studies have shown that combining DAC with other HbF-inducing agents, such as pomalidomide and the EHMT1/2 inhibitor UNC0638 enhances the expression of γ-globin and HbF in β-thalassemia/HbE erythroid progenitor cells (Khamphikham et al. 2020; Nualkaew et al. 2020). Encouragingly, the combination of DAC and LSD1 inhibitor RN-1 has led to significant increases in HbF and F-cells in both healthy non-anemic and anemic baboons (Ibanez et al. 2023a, b). Collectively, these findings suggest that combinations of HbF-inducing agents, which act with different epigenetic corepressors in the γ-globin repression complexes or different molecular targets, may offer greater therapeutic benefits by inducing higher levels of HbF expression.

In this study, we aimed to identify the most effective DNMT1 inhibitors for inducing HbF and to evaluate the therapeutic potential of combining a DNMT1 inhibitor with the LSD1 inhibitor RN-1. To achieve this, we assessed the HbF-inducing activities of ten DNMT1-targeting small molecules, as well as the combination of DNMT1 and LSD1 inhibitors, using ex vivo cultures of erythroid progenitor cells derived from β-thalassemia/HbE patients. Additionally, we investigated whether patient-specific genetic variants associated with HbF regulation could influence the response to these treatments.

Materials and methods 

Chemical compounds and dosage

Chemical compounds used in this study are listed in Supplementary Table S1. Epigallocatechin gallate (EGCG; HY-13653), fisetin (HY-N0182), lomeguatrib (HY-13668), procainamide (HY-A0084), RG-108 (HY-13642), SGI-1027 (HY-13962), SGI-110 (HY-13542), zebularine (HY-13420), 6-thioguanine (HY-13765), and hydralazine (HY-B0464) were purchased from MedChemExpress (Monmouth Junction, NJ). Decitabine (DAC; A3656) was obtained from Sigma-Aldrich (St. Louis, MO), and RN-1 (489479) from Merck (Darmstadt, Germany). All compounds, except for hydralazine, were dissolved in dimethylsulfoxide (DMSO; Sigma-Aldrich), while hydralazine was dissolved in Milli-Q water. DMSO (0.1% v/v) was used as a negative control. Detailed information on stock concentrations, working stock preparation, and final treatment concentrations for each compound is provided in Supplementary Table S2.

Erythroid progenitor cell culture and compound treatment

This study was conducted in accordance with the Declaration of Helsinki and was approved by Mahidol University Central Institutional Review Board (MU-CIRB 2020/037.1103). Written informed consent was obtained from all participants. Human CD34⁺ hematopoietic stem/progenitor cells (HSPCs) were isolated from peripheral blood of severe β-thalassemia/HbE patients and cultured in a 3-phase liquid culture system as described previously (Khamphikham et al. 2020; Nualkaew et al. 2020). Genotypes and hematological parameters of the donors at the time of blood collection were displayed in Supplementary Table S3. Briefly, CD34⁺ HSPCs were purified using immunomagnetic selection with CD34 MicroBeads Kit and cell separation columns as per the manufacturer’s protocols (Miltenyi Biotec, Gladbach, Germany). The basal erythroid medium was composed of Iscove’s Modified Dulbecco’s medium (Biochrom GmbH, Berlin, Germany) supplemented with 20% fetal bovine serum (Merck, Temecula, CA), 300 µg/mL holo-transferrin (Prospec, Rehovot, Israel) and 1% penicillin/streptomycin (Gibco, Grand Island, NY). The purified CD34⁺ HSPCs were cultured for 14 days in three different phases. During days 0–4 (phase I), the cells were cultured in the basal erythroid medium supplemented with 10 ng/mL human interleukin 3 (IL-3, Miltenyi Biotec), 50 ng/mL human stem cell factor (SCF, Miltenyi Biotec), and 2 U/mL erythropoietin (EPO, Janssen-Cilag, Bangkok, Thailand). During days 4–8 (phase II), the cells were cultured in the basal erythroid medium supplemented with 10 ng/mL SCF and 2 U/mL EPO. Finally, during days 8–14 (phase III), cells were cultured in the basal erythroid medium supplemented with 4 U/mL EPO. Cells were incubated at 37 °C, 5% CO₂, and 100% humidified atmosphere. On day 8 of culture, 1 × 10⁶ erythroid progenitor cells were treated with DNMT1 inhibitors at final concentrations of 0.1, 1, or 4 µM; DAC at 0.1 or 0.25 µM; RN-1 at 0.02 or 0.1 µM; or a combination of DAC and RN-1, depending on the experiment. Treatments were maintained until day 14. Cell viability and proliferation were assessed by trypan blue dye exclusion (0.4%), and the number of viable cells was counted using a hemocytometer.

Flow cytometry analysis of erythroid differentiation

Differentiation patterns of erythroid cells were determined at day 10 and day 12 of culture by flow cytometry with an Accuri C6 Plus (BD Biosciences, San Jose, CA) using a phycoerythrin (PE)-conjugated anti-human CD71 (clone CY1G4; Biolegend, San Diego, CA) and an allophycocyanin (APC)-conjugated anti human-CD235a (clone GA-R2; BD Biosciences). Flow cytometry data were analyzed with FlowJo version 10.3.0 (FlowJo LLC, Ashland, OR). Cells were first gated to exclude doublets and debris, followed by identification of erythroid subpopulations. Detailed gating strategy and representative plots are provided in Supplementary Figure S1.

Analysis of gene expression by reverse transcription-quantitative real-time PCR (RT-qPCR)

RNA samples were isolated from erythroid cells at day 10 of culture using TRIzol reagent (Ambion, Carlsbad, CA) according to the manufacturer’s procedure. Total RNA treated with DNase I (Thermo Fisher Scientific, Waltham, MA) was subjected to cDNA synthesis using RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) as per the manufacturer’s procedure. Expression levels of globin mRNAs were analyzed by RT-qPCR using gene-specific primers as described previously (Khamphikham et al. 2020; Nualkaew et al. 2020), and FastStart Essential DNA Green Master (Roche, Mannheim, Germany) as per the manufacturer’s procedure on the CFX96 Real-Time System (BioRad, Hercules, CA). Melt curve analysis was performed for testing the specificity of primers. Relative expression levels of α-globin (HBA), β-globin (HBB), and γ-globin (HBG) mRNAs were calculated using 2^-ΔΔCt method by normalizing to β-Actin (ACTB). Fold change of globin gene expression for the treatment groups was defined as the relative expression, compared with the DMSO control group.

Hemoglobin analysis by high-performance liquid chromatography (HPLC)

The analysis of the percentage of HbF (%HbF) in erythroid cells was conducted using the VARIANT^™ II system with β-thalassemia Short Program (BioRad). The Lyphocheck Hemoglobin A2 control (BioRad) was used for normalization. The percentage of HbF was reported relative to total Hb (HbF + HbE) in β-thalassemia/HbE cultured erythroid cells. The increase in HbF percentage after treatment (Δ%HbF) was calculated by subtracting the %HbF (DMSO control) from the %HbF (compound treatment).

Genotyping of SNPs related to HbF expression

Genomic DNA was extracted from peripheral blood mononuclear cells using Gentra Puregene Blood kit (Qiagen, Hilden, Germany). Ten widely-known genetic variants were selected from the three major HbF modifier loci, including four SNPs in BCL11A gene on chromosome 2p16 (rs766432, rs1427407, rs7569946 and rs7606173), three SNPs in the HBS1L-MYB intergenic region on chromosome 6q23 (rs4895441, rs9376092 and rs9399137), and three SNPs in the β-globin gene cluster on chromosome 11p15 (HBBP1 rs2071348, HBE1 rs72872548, and XmnI-HBG2 rs7482144) (Menzel et al. 2007; Munkongdee et al. 2021; Nuinoon et al. 2010). The − 158 (C-T) ^Gγ XmnI polymorphism (rs7482144) was genotyped by the PCR-restriction fragment length polymorphism (PCR-RFLP) method as described previously (Fucharoen et al. 1990). Genotyping of other SNP genes was performed by Sanger DNA sequencing. Briefly, SNP-located regions were amplified by PCR using primers as listed in Supplementary Table S4 and Phusion High-Fidelity DNA Polymerase (Thermo Fisher Scientific) as per the manufacturer’s protocol. The PCR amplicons were then purified using GenepHlow^™ Gel/PCR kit (Geneaid, New Taipei City, Taiwan) according to the manufacturer’s protocol and subjected to Sanger DNA sequencing.

Statistical analysis

Data were analyzed and plotted using GraphPad Prism version 8.2.0 (GraphPad Software, San Diego, CA). All results are represented as mean ± standard deviation (SD). The statistical significances were considered at P-value less than 0.05 (P < 0.05) after employing Student’s t test and one-way ANOVA with Tukey’s multiple comparison test. To assess the association of SNPs to treatment responses, patients were initially classified as major responders (MaR) or minor responders (MiR) based on their HbF induction levels. Differences of genotype frequencies for each SNP between MaR and MiR groups were analyzed using Fisher’s exact test. Relative risk ratio (RR) and 95% confidence intervals (95% CI) were calculated using the Koopman asymptotic score method (Koopman 1984) to compare risk genotypes between MaR and MiR groups. A P-value < 0.05 is considered statistically significant.

Results 

Screening of DNMT1 inhibitors as novel HbF inducers in β-thalassemia/HbE erythroid progenitor cells

We initially searched for DNMT1 inhibitors that were more effective at inducing HbF and less toxic than decitabine (DAC). We evaluated ten small molecules targeting DNMT1, including epigallocatechin gallate (EGCG), fisetin, hydralazine, lomeguatrib, procainamide, RG108, SGI-1027, SGI-110, zebularine, and 6-thioguanine, for their ability to induce HbF in β-thalassemia/HbE erythroid progenitor cells derived from CD34⁺ hematopoietic stem/progenitor cells (HSPCs). The DNMT1 inhibitors used in this study were chosen based on their higher potency (lower half-maximal inhibitory concentration, or IC50) compared to DAC, or their potential for clinical development (Supplementary Table S1). Cells were exposed to the compounds at concentrations of 4, 1, and 0.1 µM on day 8 of culture, and these concentrations were maintained throughout the rest of the experiment. DMSO was used as a negative control, while DAC at a concentration of 0.1 µM served as a positive control. Among the DNMT1 inhibitors examined, only SGI-110 at 0.1 µM and 6-thioguanine at 1 µM exhibited the ability to induce HbF expression. In contrast, no HbF induction was observed in cell treated with epigallocatechin gallate (EGCG), fisetin, hydralazine, lomeguatrib, procainamide, RG108, SGI-1027, or zebularine at concentrations of 4, 1, or 0.1 µM (Fig. 1A-B, Supplementary Figure S2, and Supplementary Table S5). Treatment with SGI-110 at 0.1 µM led to an increase in the HbF level from 43.57% in DMSO-treated cells to 56.07%, which was comparable to the level induced by DAC (58.78%). However, SGI-110 at 1 and 4 µM caused a significant reduction in cell viability. Although 6-thioguanine at 1 µM induced a higher level of HbF (66.99%) compared to DAC, it significantly reduced cell viability (Supplementary Figure S3). Therefore, only SGI-110 was further evaluated for its therapeutic potential as a HbF inducer in β-thalassemia/HbE erythroid progenitor cells.

Fig. 1.

Screening of DNMT1 inhibitors as novel HbF inducers in β-thalassemia/HbE erythroid progenitor cells. β-thalassemia/HbE erythroid progenitor cells at day 8 of an ex vivo culture were treated with DNMT1 inhibitors at a concentration of 0.1 µM. A HPLC chromatograms displaying Hb composition and B the percentage of HbF levels relative to total Hb (%HbF + %HbE) assessed by HPLC on day 14 of culture. C The percentage of relative HbF levels assessed by HPLC at day 14 of erythroid cells treated with DAC and the candidate HbF inducer SGI-110 (n = 7–8 independent donors). D Cell viability during erythroid differentiation assessed by trypan blue staining. Data are presented as mean ± standard deviation (SD). Statistical significance was determined using Student’s t-test, with ^∗P < 0.05, ^∗∗P < 0.01, or ^∗∗∗P < 0.001 compared with DMSO control. ^#P < 0.05 compared with 0.1 µM DAC treatment.

We next confirmed the HbF-inducing activity of SGI-110 in erythroid progenitor cells derived from CD34⁺ HSPCs isolated from the peripheral blood of eight β-thalassemia/HbE patients. The results showed that SGI-110 at 0.1 µM had a similar HbF-inducing ability to DAC. Specifically, SGI-110 and DAC at 0.1 µM increased HbF levels from 39.55 ± 7.60% to 51.39 ± 13.07% (Δ%HbF = 11.84 ± 13.73%) and 52.30 ± 9.60% (Δ%HbF = 12.75 ± 9.14%), respectively (Fig. 1C). Both DAC and SGI-110 at 0.1 µM did not significantly affect cell viability compared to the DMSO-treated control (Fig. 1D). However, SGI-110 at 1 µM induced a higher level of HbF (58.21 ± 7.34%), but at the cost of significantly reduced cell viability. SGI-110 (guadecitabine), a dinucleotide analog of decitabine, exhibits resistance to cytidine deaminase (Yoo et al. 2007) and has greater pharmacokinetics and metabolic stability. Nonetheless, our findings suggest that SGI-110 has similar profiles to DAC in terms of HbF-inducing activity and cytotoxicity in the ex vivo erythroid culture system.

Combined DAC and LSD1 inhibitor RN-1 reveals an additive effect

Studies have shown that using HbF inducers that have different mechanisms of action can effectively enhance HbF induction. In this study, we explored the possibility of using DAC in combination with the LSD1 inhibitor RN-1 to increase HbF levels in erythroid culture cells derived from individuals with β-thalassemia/HbE. To determine the optimal doses of DAC and RN-1 for the combination treatment that would most effectively induce HbF without causing significant cytotoxicity, we identified effective doses of DAC (0.1 and 0.25 µM) and RN-1 (0.02 and 0.1 µM) through dose titration studies (Supplementary Figure S4 and Kaewsakulthong et al. 2021). On day 8 of the culture, we applied these doses of DAC alone, RN-1 alone, or their combination to cells, maintaining the treatments throughout the experiment to assess their combined effects. We observed that all combination treatments resulted in a slight decrease in cell viability; nonetheless, viability remained > 85% (Fig. 2A). Notably, all combination treatments significantly reduced cell proliferation compared to the DMSO and the single-treatment controls (Fig. 2B). HPLC analysis of hemoglobin composition revealed that all combination regimens produced similar levels of HbF induction (Fig. 2C-D). The combination of DAC and RN-1 appeared to have additive effects on HbF induction, with HbF levels significantly increasing from 40.23 ± 9.25% in DMSO-treated cells to approximately 69.52 ± 1.20% to 72.44 ± 4.25% in cells treated with DAC and RN-1 combinations. Flow-cytometric analysis of erythroid differentiation showed that RN-1 alone and DAC + RN-1 combinations reduced proportions of R2 (CD71^high/CD235a⁺) cells and increased proportions of R3 (CD71^medium/CD235a⁺) cells on days 10 and 12, relative to DMSO controls (Fig. 2E-F). These findings suggest an improvement in delayed erythroid differentiation, a characteristic feature of β-thalassemia. In addition, the combination of SGI-110 (0.1 µM) + RN-1 (0.02 µM) produced outcomes comparable to those of DAC (0.1 µM) + RN-1 (0.02 µM) in terms of HbF induction, cell viability, proliferation, and erythroid maturation (Supplementary Figure S5). Based on these results, the combination of DAC at 0.1 µM and RN-1 at 0.02 µM was selected for further experiments, as it exhibited the least cytotoxicity (Fig. 2A-B).

Fig. 2.

Effect of decitabine (DAC), RN-1, and their combinations on cell viability, proliferation, HbF expression, and erythroid differentiation in β-thalassemia/HbE erythroid progenitor cells. β-thalassemia/HbE erythroid progenitor cells at day 8 of an ex vivo culture were treated with DAC, RN-1, or their combinations at the indicated concentrations. A Cell viability and B proliferation assessed by trypan blue staining at day 10 of culture. The fold change of cell proliferation indicates the ratio of cell number in the treatment groups relative to DMSO controls. C Representative HPLC chromatograms showing Hb composition at day 14 of culture. D The percentage of HbF levels relative to total Hb (%HbF + %HbE) assessed by HPLC at day 14 of culture. E Representative flow-cytometry dot plots (day 10 of culture) and F quantitative analysis (days 10 and 12) of erythroid subpopulations. Erythroid cells were gated based on CD71 and CD235a into four populations: R1 (CD71^high/CD235a^⁻), R2 (CD71^high/CD235a⁺), R3 (CD71^medium/CD235a⁺), and R4 (CD71^low/CD235a⁺). Data are presented as mean ± standard deviation (SD); n = 3 independent donors. ^∗P < 0.05, ^∗∗P < 0.01, or ^∗∗∗P < 0.001 compared with DMSO control. ^#P < 0.05, or ^##P < 0.01 compared with single compound treatments. The condition marked with “–” for both DAC and RN-1 represents the DMSO vehicle control.

We next evaluated the therapeutic potential of the optimized combination regimen of 0.1 µM DAC and 0.02 µM RN-1 in erythroid progenitor cells from ten β-thalassemia/HbE patients with different β-thalassemic mutations (Supplementary Table S3), including two cases of codon17 (A > T), one case of IVS1-1 (G > T), five cases of codon41/42 (-TTCT), and two cases of IVS2-654 (C > T). The data showed that while the combination treatment caused a slight decrease in cell viability, it remained > 85% (Fig. 3A), consistent with the previous experiment (Fig. 2A). Additionally, a significant decrease in cell proliferation was observed following the combination treatment. On day 10 of culture, the fold change in cell proliferation with the combination treatment reduced from 1.00 in the DMSO control to 0.856 ± 0.18. Furthermore, cell proliferation on day 12 of culture had decreased to 0.82 ± 0.18 with DAC alone, 0.76 ± 0.14 with RN-1 alone, and 0.37 ± 0.13 with combined treatment (Fig. 3B). The significant reduction in cell proliferation observed in cells treated with RN-1 alone and with the combination of DAC and RN-1 was correlated with accelerated erythroid differentiation, as evidenced by a lower proportion of R2 (CD71^high/CD235a⁺) and a higher proportion of R3 (CD71^medium/CD235a⁺) populations on days 10 and 12 of culture when compared to controls (Fig. 3C).

Fig. 3.

Enhancement of HbF induction in β-thalassemia/HbE erythroid progenitor cells by the combination of DAC and RN-1. A Cell viability and B proliferation assessed by trypan blue staining at day 10 and 12 of β-thalassemia/HbE erythroid progenitor culture in the presence of 0.1 µM DAC, 0.02 µM RN-1, or the combination of 0.1 µM DAC and 0.02 µM RN-1 (n = 10 independent donors). C Quantitation of erythroid subpopulations assessed by flow cytometry according to the expression of cell surface markers (CD71 and CD235a) at days 10 and 12 of culture (n = 4 independent donors). R2, CD71^high/CD235a⁺; R3, CD71^medium/CD235a⁺. D Histograms showing the percentage of HbF levels relative to total Hb (%HbF + %HbE) assessed by HPLC at day 14 of culture (n = 10 independent donors). E Relative fold change of α-globin (HBA), β-globin (HBB), and γ-globin (HBG) mRNA expression normalized to β-actin (ACTB) was assessed by RT-qPCR at day 10 of culture (n = 4 independent donors). Data are presented as mean ± standard deviation (SD). Statistical significance was determined using an unpaired student’s t-test: ^∗P < 0.05, ^∗∗P < 0.01, ^∗∗∗P < 0.001, or ^∗∗∗∗P < 0.0001 compared with DMSO control. ^#P < 0.05, ^##P < 0.01, ^###P < 0.001, or ^####P < 0.0001 compared with single compound treatments.

Hemoglobin analysis by HPLC revealed that the percentage of HbF increased from 38.45 ± 8.19% in the DMSO control to 51.78 ± 9.17% after culturing in DAC alone, 58.60 ± 11.14% in RN-1 alone, and 67.21 ± 8.24% in the combination treatment (Fig. 3D). These results suggest significant increases in HbF levels when cells were exposed to either single agents or the combination of DAC and RN-1. Consistent with these findings, the combinatorial treatment significantly increased γ-globin (HBG) mRNA expression, resulting in a 1.88 ± 0.52-fold increase compared to DMSO-treated cells (Fig. 3E). This increase was more pronounced than the increases observed in cells that were treated with DAC alone (1.44 ± 0.68-fold increase) and RN-1 alone (1.34 ± 0.49-fold increase). Moreover, the increase in γ-globin mRNA expression in the combination treatment was accompanied by a decrease in β-globin (HBB) mRNA expression (0.60 ± 0.16-fold compared to DMSO control), while no significant changes were observed in α-globin (HBA) mRNA expression (Fig. 3E).

Importantly, the increase in HbF levels induced by the combination (Δ%HbF = 28.76 ± 10.16%) was additive compared to either compound alone (Δ%HbF of DAC alone and RN-1 alone = 13.33 ± 11.01% and 20.14 ± 9.42%, respectively) (Fig. 4A). Taken together, the combined DAC and RN-1 treatment additively induced HbF expression in β-thalassemia/HbE erythroid progenitor cells. However, the combination caused a slight decrease in cell viability and a significant reduction in cell proliferation due to accelerated erythroid differentiation. These findings suggest that this combination could be an effective strategy for enhancing HbF induction.

Fig. 4.

Association of HbF induction response to the DAC and RN-1 combination treatment with ten HbF-related SNPs. A Violin plot illustrating the distribution of HbF increment (Δ%HbF), measured by HPLC, following treatments of DAC alone, RN-1 alone, and their combination. Two distinct response groups, major responder (MaR) and minor responder (MiR), were observed for the RN-1 and combination treatments. Responder thresholds were defined empirically from the bimodal distribution of Δ%HbF values, using the group mean to separate major from minor responders (20% for RN-1; 28% for DAC + RN-1). B Box plot depicting the increase in HbF percentage after treatment (Δ%HbF = %HbF [compound treatment] - %HbF [DMSO control]), as quantified by HPLC. The line represents the median, while the + symbol represents the mean. C-E Bar graph displaying the distribution of SNP genotypes in the MaR and MiR groups in response to the DAC and RN-1 combination treatment: C four SNPs in the BCL11A gene on chromosome 2p16, D three SNPs in the HBS1L-MYB intergenic region on chromosome 6q23, and E three SNPs in the β-globin gene cluster on chromosome 11p15 are shown (n = 10 independent donors). One-way ANOVA with Tukey’s multiple comparison test was used to compare between means of all groups: ^∗P < 0.05, or ^∗∗∗P < 0.001.

Responses to the combination of DAC and RN-1

Interestingly, β-thalassemia/HbE erythroid progenitor cells exhibited two distinct responses when treated with RN-1 alone and the combination of DAC and RN-1, as illustrated in the violin plot (Fig. 4A), suggesting significant individual variability in response to both treatments. In contrast, no such variability was observed in cells treated with DAC alone. Responders were therefore categorized into two groups based on the extent of increase in HbF percentage (Δ%HbF) after treatment, as determined by HPLC quantification. The cutoff criteria for defining responder groups were determined empirically from the distribution of HbF induction responses. The mean of each distribution was used as the threshold to separate these populations—20% (mean Δ%HbF = 20.14%) for RN-1 and 28% (mean Δ%HbF = 28.76%) for the DAC + RN-1 combination (Fig. 4A).

Accordingly, individuals with an increase in HbF exceeding 20% following RN-1 treatment were classified as major responders (MaR; n = 4) while those with an increase of less than 20% were designated as minor responders (MiR; n = 6). Major responders to RN-1 treatment displayed an average Δ%HbF of 30.32 ± 4.59%, whereas minor responders exhibited an average Δ%HbF of 13.36 ± 2.62% (Fig. 4A-B) (Supplementary Table S6). In the case of the combined treatment with DAC and RN-1, responders who achieved an elevation of HbF over 28% were classified as major responders (MaR; n = 6) while individuals with an increase of less than 28% were classified as minor responders (MiR; n = 4) (Table 1). The combination therapy elicited a more pronounced response, with major responders exhibiting a mean Δ%HbF of 35.70 ± 6.17%, compared to 18.35 ± 2.28% in the minor responders (Fig. 4A-B). Notably, the variability in response to RN-1 and the combination therapy was not associated with specific α-globin and β-globin genotypes of individuals, nor with the baseline levels of HbF in peripheral blood samples or in the culture system, as no clear patterns were observed across different genotypes or varying HbF baseline levels (Supplementary Table S3 and S5). In addition, donor age did not appear to influence HbF induction. Correlation analyses showed no statistically significant association between age and Δ%HbF following treatment with DAC, RN-1, or the DAC + RN-1 combination (Supplementary Figure S6), indicating that age is unlikely to account for the observed interindividual variability. Together, these findings suggest that other molecular mechanisms may underlie the differential response to HbF induction by RN-1 and the combination of DAC and RN-1.

Table 1.

Genotypes of single nucleotide polymorphisms (SNPs) associated with HbF expression in major (MaR) and minor (MiR) responders to the combined DAC and RN-1 treatment

Response to combination treatment	Sample name	Genotypes∗	Δ%HbF			BCL11A gene				HBS1L-MYB intergenic region			β-globin gene cluster
			DAC	RN-1	DAC + RN-1	rs766432	rs1427407	rs7569946	rs7606173	rs4895441	rs9376092	rs9399137	rs2071348	rs7482144	rs72872548
Major Responder (MaR)	BE13	CD17(A > T): CD26(G > A)	5.63	25.61	47.22	AA	GG	GG	GG	AA	CC	TT	TG	TC	AC
	BE10	CD41/42(-TTCT): CD26(G > A)	13.90	11.20	35.67	AC	TG	GG	GC	AA	CC	TT	TG	TC	AC
	BE15	CD17(A > T): CD26(G > A)	16.15	35.13	35.41	AC	TG	GG	GG	AG	AC	TC	TT	CC	CC
	BE11	CD41/42(-TTCT): CD26(G > A)	15.05	26.51	34.33	AA	GG	GG	GG	AG	AC	TC	TG	TC	AC
	BE1	IVS I-1(G > T): CD26(G > A)	15.22	34.03	32.71	AA	GG	GG	GG	AA	CC	TT	GG	TT	AA
	BE14	IVS II-654(C > T): CD26(G > A)	37.19	11.92	28.85	AA	GG	GG	GG	AA	CC	TT	TG	TC	AC
Minor Responder (MiR)	BE8	CD41/42(-TTCT): CD26(G > A)	16.00	10.64	21.51	AC	TG	AG	GC	AA	CC	TT	TG	TC	AC
	BE16	IVS II-654(C > T): CD26(G > A)	9.87	17.48	17.91	AC	TG	GG	GG	AG	AC	TC	TG	TC	AC
	BE12	CD41/42(-TTCT): CD26(G > A)	11.59	15.08	17.88	AC	TG	GG	GG	AA	CC	TT	TG	TC	AC
	BE9	CD41/42(-TTCT): CD26(G > A)	−7.28	13.82	16.09	AA	GG	GG	GG	AA	CC	TT	GG	TT	AA

BE1, BE8-16 participants were included in a combination treatment experiment of 0.1 µM decitabine and 0.02 µM RN-1 (Fig. 3) and an SNP analysis (Fig. 4). CD17 (A > T), (HBB: c.52 A > T); CD26 (G > A), (HBB: c.79 G > A); IVS1-1 (G > T), (c.92 + 1G > T); CD41/42 (-TTCT), (HBB: c.126_129delCTTT); IVS2-654 (C > T), (c.316–197 C > T). ^∗No deletions or mutations in the α-globin genes were detected in all cases. Δ%HbF, the increase in HbF percentage after treatment (Δ%HbF = %HbF [compound treatment] - %HbF [DMSO control])

To assess whether genetic modifiers influence the variability in response to RN-1 treatment alone and the combination treatment of DAC and RN-1 in individuals with β-thalassemia/HbE, we analyzed the genotypes of ten well-characterized SNPs within key loci associated with HbF expression in both MaR and MiR groups. These included SNPs in the BCL11A gene (rs766432, rs1427407, rs7569946, rs7606173), the HBS1L-MYB intergenic region (rs4895441, rs9376092, rs9399137), and the β-globin gene cluster (HBBP1; rs2071348, HBE1; rs72872548, XmnI-HBG2; rs7482144). These SNPs were selected for the analysis based on their established correlation with HbF levels in previous studies (Menzel et al. 2007; Munkongdee et al. 2021; Nuinoon et al. 2010). Our analysis did not identify a statistically significant correlation between any of the analyzed SNPs and the response to either RN-1 or the combination of DAC and RN-1. Interestingly, despite the lack of statistical significance, the data revealed suggestive trends for two SNPs within the BCL11A gene: rs766432 and rs1427407. Both SNPs exhibited a higher frequency of specific genotypes, AA for rs766432 and GG for rs1427407, in major responders compared to minor responders in both the RN-1 treatment and the combination therapy groups. In the combination treatment group, 66.7% of major responders carried the AA genotype of rs766432 and the GG genotype of rs1427407, compared to only 25% of minor responders (Fig. 4C and Supplementary Table S7). Similarly, in the RN-1 treatment group, 75% of major responders carried these genotypes, compared to 33.3% of minor responders (Supplementary Figure S7 and Supplementary Table S7). In contrast, individuals with the AC genotype of rs766432 and the TG genotype of rs1427407 were more frequently observed in the minor responder groups. The distribution of other SNPs, including those in the BCL11A gene (rs7569946 and rs7606173), in the HBS1L-MYB intergenic region (rs4895441, rs9376092, and rs9399137), or in the β-globin gene cluster (rs2071348, rs72872548, and rs7482144) did not show any differences between major and minor responders to the RN-1 and combination treatments (Fig. 4C-E, Supplementary Figure S7, and Supplementary Table S7).

The Koopman asymptotic score method was employed to estimate the relative risks (RRs) and 95% confidence intervals (95% CIs) for the association between SNPs and drug response. Individuals with the AA genotype of rs766432 and the GG genotype of rs1427407 were found to have a higher likelihood of being major responders, with an observed relative risk (RR) of 2.00 (95% CI: 0.6909 to 7.112) for the combination treatment, and 3.00 (95% CI: 0.6194 to 17.91) for RN-1 treatment alone. These findings suggest that the presence of the AA genotype of rs766432 and the GG genotype of the rs1427407 allele may be associated with an increased likelihood of a major response to both the combination and RN-1 treatments, compared to the AC and the TG genotypes, respectively (Supplementary Table S7). Furthermore, the variability in response to the combination treatment appears to be primarily driven by the response of RN-1, as a more uniform response was observed in cells treated with DAC alone (Fig. 4A).

Regardless of the response group, individuals with AA genotypes of rs766432 and the GG genotype of rs1427407 exhibited a greater increase in HbF following the combination and RN-1 treatments, compared to those with AC and TG genotypes, respectively (AA/GG; 31.84 ± 11.19% vs. AC/TG; 25.68 ± 9.125% of the combination treatment, and AA/GG; 22.38 ± 9.299% vs. AC/TG; 17.91 ± 10.03% of RN-1 treatment) (Supplementary Figure S8). We did not observe any trends toward the association between the HbF response and other analyzed SNPs in the BCL11A gene, the HBS1L-MYB intergenic region, or β-globin gene cluster (Fig. 4B-D and Supplementary Table S7). It is currently uncertain how individual patients might respond to RN-1 treatment and the combination treatment of DAC and RN-1. Our findings indicate that BCL11A SNPs rs766432 and rs1427407 may be associated with the response to this treatment. However, further studies with larger cohorts are necessary to validate this potential association due to the wide confidence intervals and the small sample size.

Discussion

Our study provides the systematic evaluation of combined DNMT1 and LSD1 inhibition in erythroid progenitors derived from β-thalassemia/HbE patients—a genotype highly prevalent in Southeast Asia and characterized by high baseline HbF, profound ineffective erythropoiesis, and distinctive responses to HbF-modifying therapies. By directly testing these compounds in patient-derived cultures, we offer new insights into their therapeutic potential in this clinically important population. Notably, we demonstrate that DAC and RN-1 exert additive HbF-inducing effects, improve erythroid maturation, and exhibit genotype-associated variability in response. These findings highlight the relevance of pharmacologic HbF induction specifically for β-thalassemia/HbE and provide translational evidence not reflected in studies based on healthy donor cells, SCD models, or non β-thalassemia/HbE genotypes.

In this study, we evaluated the HbF-inducing activity and cytotoxicity of ten DNA methyltransferase 1 (DNMT1) inhibitors, including nucleotide analogs, DNA binders, and S-adenosylmethionine (SAM) competitors as potential novel HbF inducers in cultured β-thalassemia/HbE erythroid progenitor cells. Our screening showed that only SGI-110 and 6-thioguanine were capable of stimulating HbF expression comparable to decitabine at non-toxic doses (Fig. 1A-B, Supplementary Figure S2, and Supplementary Table S5). However, a significant cytotoxicity of 6-thioguanine limits its potential for therapeutic application (Supplementary Figure S3). Other DNMT1 inhibitors tested were not as effective as decitabine at non-toxic doses. The lack of significant HbF induction by other DNMT1 inhibitors tested, such as RG108, SGI-1027, and zebularine, may be due to differences in their cellular uptake, DNMT1 selectivity, or insufficient demethylation of the γ-globin promoters in these erythroid progenitor cells. Interestingly, fisetin appeared to slightly increase HbE/HbA₂ while reducing HbF levels compared with the control. The mechanism underlying this observation remains unclear; however, fisetin has been reported to modulate HIF-1α signaling in a context-dependent manner (Chen et al. 2015; Wang et al. 2022), which may indirectly influence γ-globin gene expression (Feng et al. 2022). While some of these compounds have shown promise in other cell types or disease models, their utility in β-thalassemia appears to be limited based on the results shown here.

SGI-110 (guadecitabine), a dinucleotide analog of decitabine, demonstrated similar efficacy and toxicity profiles to DAC at equivalent molar doses in our ex vivo culture system (Fig. 1C-D). These findings are consistent with a previous report showing comparable effects of SGI-110 and DAC on HbF induction and DNA methylation of the γ-globin promoters in human erythroid progenitors from healthy donors and in anemic baboons (Lavelle et al. 2010). SGI-110 was developed as a more stable and less toxic hypomethylating agent for leukemia and cancer therapy due to its resistance to cytidine deaminase (Chuang et al. 2010). However, it failed to provide a significant advantage over DAC in vivo due to its rapid conversion into DAC (Lavelle et al. 2010). Thus, DAC remains of considerable interest as a therapeutically potent DNMT1 inhibitor for HbF induction.

Given that β-thalassemia patients require a higher level of HbF induction than those with sickle cell disease (SCD), combinatorial drug regimens targeting multiple key regulators of γ-globin repressors may offer a more effective strategy for maximizing therapeutic outcomes while minimizing drug-related toxicity. Our findings revealed that combining DAC and RN-1 to target two epigenetic-modifying enzymes in the γ-globin repressor complex, DNMT1 and lysine-specific demethylase 1 (LSD1), exerts an additive effect on HbF induction in β-thalassemia/HbE erythroid progenitors (Figs. 2 and 3). The combination treatment led to a marked increase in HbF levels, with an increase in γ-globin (HBG) mRNA expression and a concomitant decrease in β-globin (HBB) mRNA expression (Fig. 3E). This additive effect contrasts with previous studies, where DAC and RN-1 exhibited a synergistic effect in SCD mice and anemic baboons (Ibanez et al. 2023a, b; Jagadeeswaran et al. 2015). Additionally, the combination treatment of DAC and LSD1 inhibitor tranylcypromine (TCP) produced a greater-than-additive effect on HbF induction in primary human erythroid cells derived from healthy donors (Shi et al. 2013). The observed additive effect in our study may be attributed to the relatively high baseline level of HbF (~ 40%) in cultured β-thalassemia/HbE erythroid progenitors, likely influenced by patient phenotypes and culture conditions that constrained further HbF induction. Similar findings were observed in a recent study, where the combination of DAC and RN-1 induced HbF to a lesser extent in non-anemic baboons with high baseline HbF levels compared to those with low HbF levels (Ibanez et al. 2023a, b). The combination of DAC and RN-1 caused a slight reduction in cell viability, possibly due to the combined cytotoxic effects of both agents. Thus, the dosing of DAC and RN-1 in combination regimens should be optimized to minimize cytotoxicity while maintaining therapeutic efficacy. Furthermore, the significant reduction in cell proliferation observed with this combination correlated with the improvement of delayed erythroid differentiation, a hallmark of β-thalassemia pathology. This suggests that the elevated HbF levels induced by the combination treatment compensate for deficient HbA, promoting more effective erythropoiesis, and phenotypic improvement of β-thalassemia/HbE erythroid cells. Additionally, the combination of SGI-110 and RN-1 demonstrated therapeutic effects comparable to those of the DAC and RN-1 regimen, suggesting that the observed therapeutic effects may also extend to other DNMT1 inhibitors with improved pharmacological profiles (Supplementary Figure S5).

Previous genotype-phenotype studies in β-thalassemia/HbE have shown that patients with higher endogenous HbF levels, particularly those with milder β⁰-thalassemia/HbE phenotypes after excluding α-thalassemia coinheritance, show significantly improved erythropoiesis and clinical outcomes (Nuinoon et al. 2010). Additionally, several studies have demonstrated that individuals with β-thalassemia/HbE experience greater hematologic benefit from increases in HbF compared to other β-thalassemia genotypes (Fucharoen et al. 1996; Yasara et al. 2022). These observations indicate that increasing HbF can have a substantial clinical impact in β-thalassemia/HbE, where even modest HbF increments can improve ineffective erythropoiesis and reduce the transfusion burden. Thus, the additive increases in HbF observed with the DAC + RN-1 combination in our study may hold particular therapeutic significance for this highly prevalent Southeast Asian genotype.

The mechanism of HbF induction by several compounds involves disrupting the recruitment of transcriptional repressors and corepressors to γ-globin promoters or enhancers, resulting in altered epigenetic modifications and gene expression (Lavelle et al. 2018; Yu et al. 2020). Reactivation of γ-globin gene expression by DAC primarily results from DNMT1 depletion, leading to DNA hypomethylation of CpG sites within γ-globin promoters and the accumulation of acetylated histone H3 (H3ac) and trimethylated histone H3 lysine 4 (H3K4me3), both of which are active histone marks at γ-globin promoters (Akpan et al. 2010; Chin et al. 2009). Proteomic studies further show that DAC induces HbF induction in thalassemic erythroid progenitor cultures through downregulation of transcriptional repressors and chromatin modifications. In addition, DAC activates genes involved in oxidative stress pathways, promoting the survival of immature erythroblasts and enhancing their resistance to stress (Theodorou et al. 2020). Together, these findings suggest that DAC-mediated γ-globin reactivation involves both epigenetic modifications and alterations of erythroid differentiation kinetics.

Inhibition of LSD1 by RN-1 induces γ-globin expression by promoting the accumulation of active histone marks, including H3K4me2, H3K4me3, and H3K9ac, and reducing a repressive histone mark H3K9me2 at γ-globin promoters (Cui et al. 2015; Rivers et al. 2016). In addition to these chromatin changes, RN-1 modulates the expression of several genes involved in erythropoiesis and globin gene regulation, potentially contributing to HbF upregulation. We previously demonstrated that RN-1 downregulates γ-globin corepressors, such as NCOR1 and SOX6, in erythroid cultures derived from β⁰-thalassemia/HbE patients (Kaewsakulthong et al. 2021). Moreover, RN-1 has been shown to upregulate PGC-1α, a transcriptional coactivator that supports erythroid maturation and globin gene expression, in SCD mice (Cui et al. 2015). It also increases the expression of GATA2, Gfi1B, and LYN, genes involved in erythroid differentiation, which may contribute to delayed maturation (Ibanez et al. 2023a, b). This inhibition of erythroid progression could prolong early developmental stages that are permissive to γ-globin expression, thereby enhancing HbF production in response to LSD1 inhibition. However, our findings suggest that RN-1-mediated HbF induction improves delayed erythroid differentiation in β-thalassemia/HbE cells, highlighting the therapeutic potential of HbF induction by RN-1.

Therefore, DAC and RN-1 act in concert to activate γ-globin gene expression through epigenetic modifications (DNA and histone demethylation) and alterations in erythroid growth and differentiation kinetics. These combined effects may establish a chromatin landscape that is epigenetically permissive to γ-globin transcription. Nevertheless, further investigations, including genome-wide DNA methylation profiling, histone modification analysis, and transcriptomic and proteomic approaches, are required to elucidate the underlying mechanisms and evaluate the long-term safety and efficacy of this combination in vivo.

Interestingly, we observed two distinct response patterns, categorized as major and minor responders, to RN-1 and the combination of DAC and RN-1 in β-thalassemia/HbE erythroid progenitor cells (Fig. 4A-B), highlighting significant variability in HbF induction among individuals. We demonstrated that this variability may be linked to specific BCL11A SNPs (rs766432 and rs1427407) (Fig. 4C), which are well-characterized and strongly associated with HbF levels (Menzel et al. 2007; Munkongdee et al. 2021; Nuinoon et al. 2010). Although the results suggested that these SNPs may play a role in enhancing HbF induction by RN-1 and the combination treatment, it is possible that the effect may be driven by variants in linkage disequilibrium with these SNPs rather than the SNPs themselves. This finding has significant implications for clinical response to HbF enhancers in β-thalassemia and SCD. Specifically, we identified the higher prevalence of the AA genotype of rs766432 and GG genotype of rs1427407 among major responders to both RN-1 and combination treatments (Fig. 4C and Supplementary Figure S7A). This finding contrasts with previous reports where the BCL11A rs766432 C allele and rs1427407 T allele were associated with elevated HbF levels and milder clinical phenotypes in patients with β-hemoglobinopathies (Bhanushali et al. 2015; Munkongdee et al. 2021; Nuinoon et al. 2010; Sedgewick et al. 2008). Furthermore, the BCL11A rs766432 C allele has been shown to correlate strongly with a favorable response to hydroxyurea treatment (Friedrisch et al. 2016). The discrepancy with previous studies may stem from patient heterogeneity, as our study focused specifically on severe β-thalassemia/HbE cases. It is also possible that the regulatory mechanisms governing baseline HbF expression might be different from those driving pharmacological HbF induction. Additionally, different compounds may activate distinct pathways of pharmacogenetic modulation. Further investigation is necessary to clarify the relationship between BCL11A SNPs and the variability in response to RN-1 and DAC treatment. Larger-scale studies with broader genetic analyses are required to confirm these findings and elucidate the underlying genetic and epigenetic mechanisms. Importantly, it remains essential to evaluate whether the results from our ex vivo study can reliably predict clinical responsiveness to these treatments.

The findings described here emphasize the critical role of genetic factors in determining response to pharmacological treatments, underscoring the importance of having a range of HbF inducers available to account for individual variability. Genetic variability in response to HbF inducers has been reported for other drugs such as hydroxyurea, thalidomide, and rapamycin (Biswas et al. 2019; Friedrisch et al. 2016; Yang et al. 2020; Zuccato et al. 2023). Therefore, it is important to characterize the SNPs involved in HbF induction response, as it will help to predict treatment responses and thus guide early treatment options for each individual.

In recent years, the therapeutic landscape for β-hemoglobinopathies has advanced significantly with the emergence of gene therapy and gene-editing strategies, including lentiviral β-globin addition (Kwiatkowski et al. 2024; Locatelli et al. 2022) and CRISPR/Cas9-mediated disruption of the BCL11A enhancer (Locatelli et al. 2024). Although these approaches offer curative potential, their availability is currently limited by high costs, specialized infrastructure requirements, and eligibility constraints, particularly in regions where β-thalassemia/HbE is endemic. Therefore, drug-based HbF induction remains an essential and complementary therapeutic approach, particularly for patients who may not have access to gene therapy or who are medically unsuitable for transplantation or conditioning regimens (Baronciani et al. 2021). Pharmacologic HbF induction provides important advantages, including lower cost, wider accessibility, adjustable dosing, and reversibility, making it feasible across diverse healthcare settings (Njeim et al. 2024). Our findings support continued development of pharmacologic HbF-inducing strategies, which may serve as effective standalone therapies or as adjuncts that enhance clinical outcomes before or after gene therapy.

Conclusions

In summary, this study revealed that the combined DAC and RN-1 treatment resulted in an increased expression of HbF in β-thalassemia/HbE erythroid progenitor and improved erythroid differentiation, offering a promising strategy for enhancing HbF expression. The variability in response to this treatment is potentially due to genetic factors in the BCL11A region; however, a comprehensive study would be required to understand the pharmacogenomics of this regimen further. Our results clearly showed that the HbF induction by the combined treatment of DNMT1 and LSD1 inhibitors holds therapeutic potential for β-thalassemia/HbE and encourages further exploration of combinatorial approaches.

Abbreviations

APC

Allophycocyanin

BCL11A

B-cell lymphoma 11a

CDA

cytidine deaminase

DAC

Decitabine

DMSO

Dimethylsulfoxide

DNMT1

DNA methyltransferase 1

EGCG

Epigallocatechin gallate

EHMT1/2

Euchromatin histone lysine methyltransferases 1/2

EPO

Erythropoietin

FDA

Food and Drug Administration

H3ac

Acetylated histone H3

H3K4me2

Dimethylated histone H3 lysine 4

H3K9me2

Dimethylated histone H3 lysine 9

HbA

Adult hemoglobin

HbE

Hemoglobin E

HbF

Fetal hemoglobin

HDAC

Histone deacetylases

HLA

Human leukocyte antigen

HPLC

High-performance liquid chromatography

HSCT

Hematopoietic stem cell transplantation

HSPCs

Hematopoietic stem/progenitor cells

IC50

Half-maximal inhibitory concentration

IL-3

Interleukin 3

LRF

Lymphoma-related factor

LSD1

Lysine-specific demethylase 1

MaR

Major responders

MiR

Minor responders

PE

Phycoerythrin

PRMT5

Protein arginine N-methyltransferase 5

RBC

Red blood cell

RR

Relative risk

RT-qPCR

Reverse transcription-quantitative real-time PCR

SAM

S-adenosylmethionine

SCD

Sickle cell disease

SCF

Stem cell factor

SNPs

Single nucleotide polymorphisms

TCP

Tranylcypromine

THU

Tetrahydrouridine

TR2/TR4

Testicular nuclear receptors

WHO

World Health Organization

Authors’ contributions

Contribution: T.N., P.P., O.S., and N.J. designed the research; T.N., P.P., T.M., N.B., and N.J. performed experiments; T.N., P.P., and N.J. analyzed data and contributed to criticism and conclusions; K.P., S.H., and S.F. provided samples and resources; T.N., P.P., and N.J. wrote the manuscript; J.D.E., S.H., S.F., and N.J. conceptualized the idea and supervised the project; and all authors read and approved the final manuscript.

Funding

Open access funding provided by Mahidol University. This work was supported by grants from Mahidol University (Basic Research Fund: fiscal year 2021 and MRC-IM 02/2565) to N.J. and S.H. P.P. was supported by the Siriraj Graduate Scholarship.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki. Approval was granted by the Mahidol University Central Institutional Review Board (MU-CIRB 2020/037.1103). Written informed consent was obtained from all participants included in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.