A natural DNMT1 mutation elevates the fetal hemoglobin level via epigenetic derepression of the γ-globin gene in β-thalassemia

For greater insight into the mechanism of hematopoietic diseases, finding genetic modifiers that account for the clinical variability of β-hemoglobinopathies, including hemoglobin E (HbE)/β-thalassemia, is of crucial importance.^1,2  A group of genetic modifiers involved in the regulation of fetal hemoglobin (HbF) has been shown to be responsible for the phenotypic heterogeneity of β-hemoglobinopathies.^1,3-6  DNA methyltransferase 1 (DNMT1) is one such genetic modifier that regulates human γ-globin silencing epigenetically by maintaining promoter CpG methylation.^7-9 

The repressive effect of DNMT1 is functionally mediated by its N-terminal domain, which facilitates its interaction with numerous γ-globin regulators and is indispensable for the recruitment of DNMT1 to HBG promoters.^9-11  Moreover, phosphorylation of serine residues within the N-terminal domain is an important factor in the catalytic activity and protein stability of DNMT1.^12,13  Although clinical studies evaluating DMNT1 depletion have shown efficacy in upregulating HbF levels in patients with β-hemoglobinopathies,^14-16  the association of DNMT1 mutations with β-thalassemia phenotypes and the underlying epigenetic mechanisms have yet to be demonstrated.

Here, we identified a novel missense mutation (c.2633G>A, S878F) in DNMT1 in patients with HbE/β-thalassemia, and uncovered a novel mechanism of this mutation in a DNMT1 germline mutant in direct epigenetic derepression of HBG, which in turn ameliorates the clinical severity of β-thalassemia.

A previously established cohort of 1142 Chinese β-thalassemia patients (supplemental Table 1, available on the Blood Web site) was recruited for molecular genetic analysis using genomic DNA.¹⁷  Target capture-based next-generation sequencing (NGS) was used to identify variants in a panel of candidate genetic modifiers associated with increased HbF, and the mutations were confirmed by Sanger sequencing. To uncover the mechanistic role of the novel S878F DNMT1 mutation in regulating the reactivation of γ-globin expression, the function of this mutation was analyzed through various in vitro assays using CD34⁺ cells from patients and HuDEP-2 cells, with the S878F mutation introduced by clustered regularly interspaced short palindromic repeats (CRISPR)–CRISPR-associated protein 9 (Cas9). The F-cell pattern in patients’ blood and cultured erythroid cells was examined using immunofluorescence assays and flow cytometry analysis. Hematology assays, genetic tests, and DNA methylation assays were performed as previously described.^5,17  Detailed protocols and information are described in supplemental Methods. Written informed consent was obtained from all participants and/or their family members.

Using a target capture NGS approach for screening variants associated with altered HbF levels (supplemental Figures 1-2; supplemental Tables 2-5), we identified 10 potential HbF-regulating variants in IKZF2, CHD3, RAD21, and DNMT1 from a cohort of 1142 β-thalassemia patients. After further analysis, including prediction of well-known BCL11A variants influencing HbF levels, only a novel heterozygous missense mutation (c.2633G>A, p.Ser878Phe) in DNMT1 was found to be a positive modifier in 3 cases (supplemental Figure 2; supplemental Table 3). ¹⁷  A precedent for this mutation could not be found in the International HapMap Project (https://www.genome.gov/10001688/international-hapmap-project) or ChinaMAP (http://www.mbiobank.com/) databases (Figure 1A). The variant was further confirmed by Sanger sequencing, and the effect of the DNMT1 mutation in increasing HbF in individuals carrying the mutation from 2 Chinese pedigrees was determined. Among them, 3 patients (with β-thalassemia compound heterozygous mutations and this coinherited missense mutation) had high HbF (49.5%, 34.7%, and 32.5%, respectively) (Figure 1B; supplemental Table 6). Multiple alignment analysis suggested that this missense variant occurred at a highly conserved codon, changing a serine to phenylalanine at position 878 (S878F) in the first bromo-adjacent homology domain (BAH1) of DNMT1^18,19  (Figure 1C), with a predicted loss of 2 hydrogen bonds as essential structural elements (Figure 1D; supplemental Figure 3).

Cox proportional hazards analysis demonstrated that S878F mutation is associated with a stronger effect on the clinical severity of β-thalassemia (hazard ratio = 0.230, P = .012; supplemental Table 7), measured by the age at first transfusion in 1142 β-thalassemia subjects. All 3 patients with S878F mutation were diagnosed with HbE/β⁰-thalassemia and exhibited significantly milder phenotypes than control group subjects without this mutation, as determined by case-control univariate analysis (Figure 1E-G; supplemental Table 8). Thus, the DNMT1 variant represents a novel genetic modifier, as evidenced by its association with phenotypes in β-thalassemia.

To investigate the precise mechanistic role of the S878F mutation in the reactivation of γ-globin, we performed functional analysis of this mutation in the HuDEP-2 cell line and primary CD34⁺ erythroid cells from 4 patients with HbE/β⁰-thalassemia (II-1 and II-3 in family 1, II-4 and II-5 in family 2). We first characterized the effect of the S878F mutation on interactions with coregulatory proteins, including a group of well-known erythroid transcription factors (BCL11A, SP1, GATA1, LRF, MYB, and TR2/TR4) and epigenetic coregulators (histone deacetylase 1/2 [HDAC1/2]) involved in macromolecular complexes that mediate HbF silencing.^9-12,19,20  S878F mutation dramatically decreased the interaction of DNMT1 with BCL11A, GATA1, and HDAC1/2, as examined by coimmunoprecipitation using either target protein-specific antibodies (Figure 2A) or anti-DNMT1 antibody (Figure 2B) in CD34⁺ erythroid cells with mutated compared with wild-type (WT) DNMT1. This analysis suggests that the mutated BAH1 domain obstructs the formation of macromolecular complexes via protein-protein interactions.^19,20  These results are consistent with those observed in HuDEP-2 cells (supplemental Figure 4A-B). Collectively, our results suggest that the structure of the mutant S878F protein disturbs the protein-protein interactions between DNMT1 and the coregulators.

Using a specific antibody against the phosphorylated form of Ser878 in DNMT1, we demonstrated abrogation of DNMT1 phosphorylation at position 878 of the mutant S878F protein in HuDEP-2 cells (Figure 2C, left panel) and CD34⁺ erythroid cells (Figure 2C, right panel). We also observed that the mutant S878F protein displayed a faster degradation rate than its WT counterpart (Figure 2D), and was correlated with significantly decreased methyltransferase activity of DNMT1 in HuDEP-2 cells (Figure 2E). These results suggest that DNMT1 mutation leads to impairment of both its structural stability and catalytic activity, which is similar to results from previous studies of phosphorylation at Ser154 of DNMT1.¹³  However, these observations were not replicated in the HEK-293T human renal epithelial cell line when transfected with either WT DNMT1 or mutated DNMT1 (S878F) (supplemental Figure 5), suggesting that phosphorylation of DNMT1 at residue Ser878 or BAH1 domain–mediated interactions may be regulated in a tissue- or developmental-stage restricted manner.

We further evaluated the impact of S878F mutation on HBG promoter methylation and γ-globin expression both in CD34⁺ cells from patients and CRISPR/Cas9-edited S878F mutant HuDEP-2 cells (supplemental Figure 6; supplemental Table 10). Chromatin immunoprecipitation (ChIP) quantitative polymerase chain reaction (qPCR) showed that S878F mutation significantly reduced the accumulation of DNMT1 at the HBG promoter,^7,21  but no changes were observed in the enrichment of GATA1 and HDAC2 between the WT and mutated CD34⁺ cells (Figure 2F; supplemental Figure 7), indicating the involvement of DNMT1 in this regulation. We next investigated the DNA methylation status at the HBG promoter, and the S878F mutant exhibited a high degree of hypomethylation at 6 CpG sites (−162, −53, −50, +6, +17, and +50) in HBG promoter regions (Figure 2G; supplemental Figure 8), but no changes were observed in global methylation levels between the WT and mutated proteins in either CD34⁺ or HuDEP-2 cells (supplemental Figure 9), suggesting a narrower set of demethylation changes confined to the γ-globin genes and their active participation in the reactivation of γ-globin. We also determined that γ-globin expression was increased at both the messenger RNA (mRNA) and protein levels in CD34⁺ and HuDEP-2 cells (Figure 2H-K; supplemental Figure 10), and the effect of S878F mutation was heterocellular, as a result of an increased number of HbF⁺ cells (F cells) (supplemental Figure 11). Furthermore, significant changes were not observed in erythroid differentiation kinetics in CD34⁺ cells from patients with HbE/β⁰-thalassemia (supplemental Figure 12).

Our results not only suggest that γ-globin gene methylation is reduced either by decreased DNMT1 binding to HBG promoters (Figure 2F; supplemental Figure 7) or impaired catalytic activity (Figure 2D-E), but also that S878F mutation attenuated DNMT1 interactions with BCL11A, GATA1, and HDAC1/2 (Figure 2A-B). Taken together, we propose that the reactivation of γ-globin induced by DNMT1 S87F mutation constitutes a mechanism for reduced γ-globin gene methylation in combination with reduced recruitment of a corepressor complex.

In summary, we identified a DNMT1 variant (S878F) as a novel genetic modifier responsible for high HbF production by epigenetically derepressing γ-globin expression and thereby ameliorating β-thalassemia severity. Our findings suggest that the BAH1 domain of DNMT1 harboring a serine at position 878 maintains DNA methylation. This mutation occurs naturally in a benign genetic condition, which could potentially be recapitulated in primary erythroid cells by Cas9 or base-editor technology²²  to treat β-hemoglobinopathies.

The authors thank all those who participated in this study. They are particularly grateful to Dali Li from East China Normal University, and Xingxu Huang Shanghai Tech University for their timely help with editing erythroid cells. The authors also thank Wei Liao from the 923rd Hospital of the People’s Liberation Army for assistance in collecting patient samples; Fu Xiong, Qiyang Li, Jiaqiong Lin, Jie Bai, Zhongju Wang, and Xuan Shang from Southern Medical University and Peng Xu from Soochow University for valuable advice and comments on this work; and Mingli Xu and Zhunqiang Zhong from Southern Medical University for assistance in conducting technical work.

This work was supported by grants from the National Key Research and Development Program of China (2018YFA0507800 and 2018YFA0507803) and the National Natural Science Foundation of China (31671314 and 31871265).

Contribution: Y.G., Y.Y., and X.X. designed the study; Y.G., Y.Z., J.H., Liren Wang, W.Y., and C.S. performed the experiments and analyzed the data; Y.L., T.Y., and Li Wang collected samples and clinical data; X.Z. performed clinical classifications and molecular diagnosis and managed β-thalassemia patients; Q.Z., Y.Y., and J.Z. analyzed the data; Y.G., Q.Z., and X.X. wrote the manuscript; X.X. supervised the study; and all the authors read and approved the contents of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Xiangmin Xu, Department of Medical Genetics, School of Basic Medical Science, Southern Medical University, Guangzhou, 510515, Guangdong, China; e-mail: gzxuzm@pub.guangzhou.gd.cn or xixm@smu.edu.cn.