MDS/AML-associated DDX41 helicase facilitates homologous recombination repair by potentially resolving R-loops
Aanchal Aggarwal, Shizhuo Yang, Lacey Winstone, Sohaumn Mondal, Harmony Grainger, Ravi Shankar Singh, Ananna Bhadra Arna, Franco J Vizeacoumar, Yuliang Wu
TL;DR
This study shows that the DDX41 protein helps repair DNA by resolving R-loops, and its malfunction may lead to blood cancers like MDS and AML.
Contribution
The study reveals a novel role for DDX41 in resolving R-loops during homologous recombination repair, linking its dysfunction to genomic instability in MDS/AML.
Findings
DDX41 deficiency increases genomic instability and DNA damage in cells.
DDX41 wild-type resolves R-loops, but the R525H mutant fails to do so.
DDX41 supports homologous recombination repair by resolving R-loops.
Abstract
DDX41 mutations are associated with myeloid neoplasms, including myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), and missense mutant R525H is seen in 67% of patients; however, its molecular pathogenesis is unknown. Using DDX41 knockout (KO) cells, we found that these cells were sensitive to bleomycin, camptothecin, and UV. DDX41 deficiency led to increased genomic instability, indicated by elevated DNA double-strand breaks (DSBs) and comet tails. We found that R-loop formation increased in DDX41–KO cells. DDX41 wild-type (WT) protein resolved DNA:RNA hybrid of R-loops in vitro, but the mutant R525H failed. DDX41–R525H expressing cells were sensitive to DNA damage agents and had significantly more R-loops than DDX41–WT expressing cells. Interestingly, DDX41 colocalized with DSB marker γH2AX and R-loop marker S9.6, and knockdown of DDX41 in the U2OS GFP reporter cells…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8- —CIHR10.13039/501100000024
- —NSERC10.13039/501100000038
- —Cancer Research Society10.13039/100009326
- —Leukemia & Lymphoma Society of Canada10.13039/100009447
- —Saskatchewan Health Research Foundation10.13039/501100000106
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsDNA Repair Mechanisms · Acute Myeloid Leukemia Research · Genetic factors in colorectal cancer
Introduction
Myelodysplastic syndrome (MDS) is a group of clonal hematopoietic disorders characterized by ineffective hematopoiesis, peripheral blood cytopenia, and an increased risk of progression to acute myeloid leukemia (AML). Inherited hematologic malignancies have been discovered in MDS/AML. So far, at least 11 genes (RUNX1, CEBPA, TERC, TERT, GATA2, SRP72, ANKRD26, ACD, ETV6, ATG2B, and DDX41) representing germline heterozygous mutations have been reported to be associated with familial MDS/AML, of which DDX41 is one of the most recent identified [1, 2]. Most of these genes are involved in chromatin modification, cohesion regulation, DNA methylation, transcription, splicing, signal transduction, and DNA repair, indicating their roles in the pathogenesis of MDS and AML [3, 4].
DEAD-box helicase 41 (DDX41) is a member of helicase superfamily 2. Its mutations have been identified, both as germline and acquired somatic mutations, in the families of patients with late-onset MDS and AML. In 2012, somatic DDX41 mutations were reported in a study of sporadic AML syndrome [5]. In 2015, a familial AML syndrome was first identified and characterized by long latency and germline mutations in the DDX41 gene [6]. In 2016, five novel mutations, including missense and splicing mutations, were reported from the families with MDS and AML [7–9]. The connection between DDX41 mutations and MDS/AML was further confirmed in 2017 when two cases of donor cell leukemia were reported due to DDX41 germline mutations [10, 11]. In 2019, it was confirmed that germline DDX41 mutations were relatively common in adult MDS/AML [12–14]. In recent years, increasing evidence has suggested that DDX41 is a predisposition gene to myeloid neoplasms [15–34]. Typically, patients have a germline mutation in one allele and acquire a somatic mutation in another allele. Recently, a patient with biallelic germline DDX41 variants was also reported [35]. The most frequently identified somatic mutation in DDX41 is c.1574G > A, p.R525H [6, 7, 36, 37], which is located in a highly conserved helicase motif VI [38]. Our recent work showed that the R525H mutation causes defects in ATP binding, thus impacting the ATP hydrolysis and unwinding activities of DDX41 protein [39], although its biological consequences are not completely known.
R-loop is a three-stranded nucleic acid structure consisting of a DNA:RNA hybrid and a displaced single-stranded (ss)DNA. Generally, R-loops are categorized into physiological and pathological R-loops [40, 41]. The physiological R-loops are programmed, whereas the pathological R-loops are nonprogrammed. Pathological R-loops pose a significant threat to genomic stability in different ways, such as generating transcription–replication collisions (TRCs), ssDNA breaks (SSBs), and double strand breaks (DSBs) [42, 43]. Many diseases, including several neurodegenerative disorders [44, 45] and various cancers [46], are associated with increased R-loops. On the other hand, the physiological functions of R-loops comprise immunoglobulin class switching of B cells in vertebrates [47], gene editing using CRISPR–Cas9 [48], mitochondrial DNA replication [49], specific regulatory steps in transcription [50], DSB repair [51], and maintaining telomere homeostasis [52].
DSBs are a particularly dangerous threat to genome stability, and they are repaired by two major pathways: non-homologous end joining (NHEJ) and homologous recombination (HR). R-loop is a double-edged sword at DSB sites. It has been reported that R-loops can facilitate DSB repair. For example, RNA serves as a repair template [53], RNA recruits repair proteins RAD52 and BRCA2 [54, 55], and R-loop triggers HR [51, 56]. On the other hand, R-loop may affect DSB repair. The foremost reason is the exposed ssDNA regions in the R-loop structure, which is subject to ROS (reactive oxygen species) [57], APOBEC cytosine deaminases [58], nucleases [59], and adopt secondary structures, including G-quadruplexes (G4s) and hairpins, which become obstacles to DNA replication [60]. Importantly, resolution of R-loops is mandatory for the completion of HR [61, 62].
Numerous proteins regulate the homeostasis of R-loops. Nuclease RNase H1 and H2 remove R-loops by degrading the RNA in the DNA:RNA hybrid [63–65]. DNA endonuclease XPG can degrade the ssDNA in the hybrid [61]. Some helicases, such as SETX (Sen1 in yeast) [62], DDX1 [66], DDX5 [67], and UAP56/DDX39B [68], can separate the DNA:RNA hybrid. DDX18 inhibits R-loop accumulation [69]. Also, BRCA2 suppresses R-loops by promoting the release of paused RNA Pol II at a promoter region [70, 71]. In contrast, several proteins promote R-loop formation. DDX1 separates G4 structures to form DNA: RNA hybrids in the IgH S-region and promotes class switch recombination [72]. Replication protein A (RPA) binds RNA and promotes R-loop formation [73]. DHX9 is reported to promote [74] and inhibit [75, 76] R-loop formation by different labs. Nevertheless, it is unclear how these various factors regulate the equilibrium between the formation and removal of R-loops inside cells.
Genome instability due to DNA damage and alteration in the damage response pathway have been implicated in the pathogenesis of MDS and AML [3, 4, 77]. DDX41 has been identified as an R-loop-associated protein using R-loop antibody S9.6 pulldown [78, 79] and the top candidate for R-loop binding/resolving by an R-loop proximity proteomics approach [80]. Work on the zebrafish model suggested that DDX41 triggers an R-loop-mediated sterile inflammatory cascade [81]. R-loops accumulate in CD34⁺ cells from DDX41-mutated MDS patients and in DDX41-deficient K562 cells [82]. Therefore, we investigated DDX41’s potential role in DNA repair, particularly in R-loop metabolism. Here, we demonstrate that DDX41 facilitates HR of DSB repair process by resolving R-loops.
Materials and methods
Cell lines
HeLa (CCL-2) and HT1080 (CCL-121, both from ATCC) cells were grown in Dulbecco’s modified Eagle medium (DMEM, 11995065, Gibco) supplemented with 10% fetal bovine serum (F1051, Sigma) and penicillin/streptomycin (100 U/ml each, P4333, Sigma). U2OS (HTB-96, ATCC) and U2OS GFP reporter cells [83] (a gift from Jeremy Stark, City of Hope) were grown in Mccoy 5A media (16600082, Thermo Fisher) supplemented with 10% fetal bovine serum. CRISPR–Cas9 DDX41 knockout (KO) HeLa cell lines, and lentivirus overexpression of DDX41–WT and DDX41–R525H genes in wildtype and DDX41 KO background cell lines have been generated previously [39].
Primary antibodies
DDX41 mouse antibody (MABF1107) for some immunofluorescence imaging and β-actin (A5441) were from Sigma. DDX41 rabbit antibody (15076) for both immunofluorescence imaging and Western blotting, γH2AX (Ser139, 9718), phosphor-BRCA1 (Ser1524, 9009), and 53BP1 (88439), Rad51 (8875) antibodies were from NEB. The S9.6 antibody (ENH001) was from Kerafast. The double-stranded DNA (ab27156), RPA32/RPA2 (ab2175), nucleolin (ab22758), and I-SceI (ab216263) antibodies were from Abcam.
Clonogenic survival assays
Cells were seeded in six-well plates (or 35-mm Petri dishes) with 2 × 10^3^ cells per well for 3 h prior to treatment. For UV, the medium was removed completely, cells were washed with PBS, overlaid with a minimal volume of PBS to avoid drying, and exposed to UV (0–100 J/m^2^, Analytik Jena). For drug treatment, the cells were incubated with bleomycin (BLM, 0–50 μg/ml for 24 h; BLE011, BioShop), camptothecin (CPT, 1–100 nM for 24 h; CAM222, BioShop), and Olaparib (0–8 μM for 24 h; S1060, Selleck Chemicals). After treatments, the cells were washed and maintained in regular media until the surviving cells formed colonies. The clones were stained with 0.5% crystal violet (in 70% methanol) for visualization, and clones with a diameter of >1 mm were counted. The survival percentage is presented as the relative seeding efficiency of UV radiation or drug-treated versus control untreated cultures.
Confocal microscopy
Cells were washed twice with PBS and fixed with 100% methanol at −20°C for 30 min, then washed with PBS and blocked with blocking buffer (5% BSA in PBS) at room temperature for 1 h. The immunostaining was performed by first incubating cells with primary antibodies (1:1000) overnight at 4°C. After washing with PBS, the cells were incubated with secondary antibodies, Alexa Fluor 488 goat anti-rabbit IgG (1:1000, A32731, Invitrogen), and/or Alexa Fluor 594 goat anti-mouse IgG (1:1000, A11032, Invitrogen) for 1 h at room temperature. Cells were then washed with PBS and mounted with Prolong Diamond antifade reagent containing 4',6-diamidino-2-phenylindole (DAPI, P36962, Invitrogen) and cured at room temperature in the dark for 24 h. Immunofluorescence was performed on a Zeiss LSM 700 or a Zeiss LSM900 Airyscan 2 laser scan microscope with a Plan-Apochromat ×63/1.4 oil DIC objective. Images were captured with a CCD camera and analyzed using LSM Browser software ZEN (Zeiss). Confocal images were acquired as Z-stacks with a total depth of 20 µm and an interval of 0.5 µm. ImageJ was used to count foci (e.g. DDX41, γH2AX, 53BP1, S9.6, RPA, RAD51, and PLA signals), and the JACoP plugin in ImageJ was used to quantify colocalization (Pearson’s correlation coefficient) between two fluorescent channels for at least 30 cells.
For GFP-tagged protein, 24 h after transfection with plasmid DNA, cells were fixed with methanol, mounted with DAPI, and then directly observed under a LSM 700 microscope.
Neutral comet assay
Neutral comet assay was performed using a CometAssay kit (4250-050-K, R&D systems) according to the manufacturer’s instructions. Briefly, cells were seeded in six-well plates, and on the next day, they were treated with bleomycin (30 µg/ml for 1 h), washed, cultured with fresh DMEM, and collected at 4 h post treatment. Cells were washed with 1× PBS and centrifuged, the pellet was suspended at 1 × 10^5^ cells/ml in ice-cold 1× PBS, and 50 µl of cells were mixed with 500 µl of molten low-melting-point agarose at 37°C. The mixture (around 50 µl) was applied onto microscope slides that were precoated with agarose and incubated at 4°C for 30 min in the dark. To unwind the supercoiled DNA, cells were lysed at 4°C for 1 h in the dark and exposed to a neutral electrophoresis buffer (pH 9) at 4°C for 30 min. Next, the electrophoresis at 21 volts for 30 min was performed in the cold room, where the DNA-strand breaks present in the cells migrated in the direction of the anode. Finally, the slides were kept with DNA precipitation buffer for 30 min at room temperature and fixed with 70% ethanol for 5 min and later stained with GelRed (41 003, Biotium) for 30 min at room temperature in the dark. Images were taken with a confocal microscope (Zeiss LSM 700) at 20× magnification and numerical aperture 0.8. Tail moment (combining tail length and tail intensity) of at least 30 cells from three biological repeats were measured by using the “Open Comet” plugin in ImageJ.
Western blotting
Depending on the target protein’s size, proteins were separated on 6, 8, 10, 12, or 15% SDS–polyacrylamide gels (homemade) using Tris-glycine buffer (with SDS). After electrophoresis, the proteins were transferred to the polyvinylidene fluoride (PVDF) membrane (1620 177, Bio-Rad) at 100 V for 2 h at 4°C. The membrane was incubated with blocking buffer (5% BSA in PBS with 0.1% Tween 20) for 1 h at room temperature, then incubated with respective primary antibodies (1:1000) in PBST with 1% BSA at 4°C overnight. After washing with PBST five times (5 min each), the membrane was incubated with HRP conjugated secondary antibody (1:5000, SC2004 or SC2005, Santa Cruz) in PBST with 1% BSA for 1 h at room temperature; then washed five times with PBST (5 min each). The membrane was then treated with Clarity Western ECL Substrate (1705061, Bio-Rad) and visualized using a ChemiDoc MP Imaging System (Bio-Rad). Quantification of the bands was carried out by optical densitometry and analyzed using the ImageJ software from one or three independent experiments. The expression of each protein was normalized with β-actin.
Dot blot assays
On day 1, cells (5 million) were collected and lysed in 1 ml of lysis buffer, composed of 500 µl of 2 × lysis buffer (0.5% SDS, 40 mM Tris–HCl pH 7.5, 150 mM NaCl, and 5 mM EDTA), 500 µl of 2× TE (100 mM Tris–HCl pH 8, 10 mM EDTA pH 8), and 10 µl of proteinase K (20 mg/ml, P6556, Sigma), at 37°C overnight. The next day, the genomic DNA was extracted using DNeasy Blood & Tissue Kit (69506, Qiagen) and extracted DNA (∼50 µg each) were fragmented using a restriction enzyme cocktail of 1 µl of HindIII (20 U/µl), 1 µl of EcoRI (20 U/µl), 2 µl of BsrGI (10 U/µl), 1 µl of XbaI (20 U/µl), and 4 µl of SspI (5 U/µl, all from NEB) in NEB Buffer 2 (volume = 300 µl) at 37°C for overnight. On day 3, the mixture was treated with 2 U of RNase III (M0245, NEB) and 4 U of RNase T1 (FEREN0541, ThermoFisher) for 2 h. The digested fragments were divided into two parts. One was treated with 4 U of RNase H (M0297, NEB) for 2 h, and another was untreated. The fragmented DNA samples were repurified using the DNeasy Blood & Tissue Kit and dissolved in 100 µl of 5 mM Tris–HCl pH 8.5. The extracted DNA was quantified using Nanodrop (ThermoFisher). Each sample was diluted to 250 ng in 200 µl of TE, and 50 µl of each DNA sample was spotted on two duplicated nylon membranes (LC2003, ThermoFisher) and crosslinked by UV treatment (1200 J/m^2^). Both membranes were blocked with 5% milk in PBST, one membrane was incubated with S9.6 antibody (1:1000 dilution in 3% BSA) and another with dsDNA antibody (1:1000 dilution in 3% BSA) overnight at 4°C. Both membranes were washed five times with PBST and incubated with goat anti-mouse secondary antibody (1:5000, Santa Cruz) for 1 h. Later, the membrane was treated with ECL reagent (Bio-Rad) and visualized using a ChemiDoc MP Imaging System (Bio-Rad). The ImageJ was used to quantify the dot intensity of three independent experiments.
For co-immunoprecipitation (co-IP), HT1080 cells were treated with or without bleomycin, lysed in NP40-containing lysis buffer (0.5% NP40, 100 mM NaCl, 10% glycerol, 1 mM EDTA and 50 mM Hepes pH 7.5, 1 mM NaVO_4_, and protease inhibitors), pre-cleared and immunoprecipitated with a S9.6 antibody, a DDX41 antibody or control mouse IgG. Immunoprecipitated protein-nucleic acid complex were washed with lysis buffer, eluted, spotted on two duplicated nylon membranes, or separated on SDS–PAGE gel followed by transferring to a PVDF membrane, and detected by Western blotting using DDX41 and S9.6 antibodies, respectively.
Nucleic acid substrates
PAGE-purified oligonucleotides were purchased from IDT: RNA 30 mer: 5′-GAGCTACCAGCTACCCCGTATGTCAGAGAG-3′ and DNA 30 mer comp + 15T: 5′-CTCTCTGACATACGGGGTAGCTGGTAGCTCTTTTT TTTTTTTTTTTTTTTTTTTT-3′. The substrates were prepared as previously described [84]. Briefly, a single oligonucleotide was 5′-end-labeled with [γ-^32^P] ATP (BLU502A, PerkinElmer) using T4 polynucleotide kinase (M0201, NEB) at 37°C for 1 h. Unincorporated radionucleotides were removed by a G25 chromatography column (28918007, GE Healthcare). ssDNA or RNA substrates were kept at 4°C and ready to use. For the double-stranded substrates, a [γ-^32^P] ATP-labeled oligonucleotide was annealed to a 2.5-fold excess of the unlabeled complementary strands in annealing buffer (10 mM MOPS, pH 6.5, 1 mM EDTA, and 50 mM KCl) by heating at 95°C for 5 min and then cooling slowly to room temperature. The DNA:RNA hybrid substrate was purified by PAGE isolation and its concentration was determined by a liquid scintillation counter (MicroBeta TriLux, PerkinElmer) before use.
Helicase assays
Helicase assay reaction mixtures (20 μl) contained 40 mM Tris (pH 8.0), 0.5 mM MgCl_2_, 15 mM NaCl, 0.01% Nonidet P-40, 0.1 mM dithiothreitol (DTT), 1 mg/ml bovine serum albumin, an equimolar mixture of 2 mM ATP and MgCl_2_, 0.5 nM of the specified duplex substrate, and the indicated concentrations of DDX41 protein as previously described [85]. Helicase reactions were initiated by adding DDX41 and were then incubated at 37°C for 15 min unless otherwise indicated. Reactions were quenched with the addition of 20 μl of 2 × Stop buffer (17.5 mM EDTA, 0.3% SDS, 12.5% glycerol, 0.02% bromophenol blue, and 0.02% xylene cyanol). For duplex substrates, a 10-fold excess of unlabeled oligonucleotide (cold oligo) with the same sequence as the labeled strand was included in the quench to prevent reannealing. The products of the helicase reactions were resolved on nondenaturing 15% (19:1 acrylamide:bisacrylamide) polyacrylamide gels. Radiolabeled DNA or RNA species in polyacrylamide gels were visualized using a Typhoon FLA 7000 and analyzed with ImageQuant 5.2 software (GE Healthcare). The percent helicase substrate unwound was calculated using the following formula: % unwinding = 100 × (P/(S + P)), where P is the product and S is the substrate. The values of P and S were corrected after subtracting background values in the no enzyme and heat-denatured substrate controls, respectively.
Strand annealing assays
Annealed DNA:RNA hybrid was denatured at 95°C for 5 min and kept on ice, followed by the helicase assays described above in the presence or absence of ATP (2 mM), and reactions were performed at room temperature for 30 min.
siRNA and GFP reporter assays
ON-TARGETplus human DDX41 (51428) siRNA was purchased from Horizon Discovery (cat #LQ-010394-00-0002). The efficiency of knocking down was analyzed in each experiment by Western blotting. The control siRNA (cat#D-001810-01-50) does not recognize human mRNA. U2OS reporter cell lines (DR–GFP, SA–GFP, EJ2–GFP, and EJ5–GFP) and I-SceI expression construct were obtained from Jeremy Stark [83]. Briefly, U2OS cells were seeded in a six-well plate, and siRNAs targeting DDX41, and siRNA control were transfected using Lipofectamine 3000 (L3000150, Invitrogen). Cells were maintained at 37°C in a humidified atmosphere containing 5% CO_2_. After 24 h, cells were re-transfected with the I-SceI expression plasmid using Lipofectamine 3000. Forty-eight hours after I-SceI transfection (72 h after siRNA), cells were harvested and subjected to Western blot with an I-SceI antibody and flow cytometry analysis (CytoFLEX, Beckman) with a minimum of 20 000 events acquired. The percentage of GFP + cells was estimated using FlowJo (BD).
Cell cycle analysis
Cells were fixed with 75% ethanol at 4°C overnight. After being washed twice with 1× PBS, the cells were suspended in a solution containing 50 μg/ml of propidium iodide (P4170, Sigma) and 100 μg/ml of RNase A (R5503, Sigma) and incubated at room temperature in the dark for 30 min. The stained cells were then analyzed using a CytoFLEX Flow Cytometer and the FlowJo software.
Proximity ligation assay
The colocalization of DDX41 with R-loops or proteins was assessed via proximity ligation assay (PLA) using the Duolink In Situ Red Starter Kit Mouse/Rabbit (DUO92101, Sigma) according to manufacturer’s instructions. In brief, cells were cultured on glass coverslips, fixed with 4% formaldehyde for 15 min, and permeabilized with 0.1% Triton X-100 for 20 min at room temperature. After blocking with Duolink blocking solution for 1 h at 37°C, samples were incubated with primary antibodies (1:250) overnight at 4°C. Following washing with buffers provided by the kit, species-specific PLA probes (anti-rabbit PLUS or anti-mouse MINUS) were applied and incubated for 1 h at 37°C. Ligation and amplification steps were carried out using the Duolink ligation and polymerase regents, respectively, with incubation at 37°C for 30 min for ligation and 100 min for amplification. The resulting fluorescence PLA signals were visualized by a confocal microscope (Zeiss LSM 900). Nuclei were counterstained with DAPI, and images were analyzed by ImageJ.
Cytogenetic analysis
HT1080 (WT and DDX41–KO) cells were treated with or without 1 μM cisplatin (232120, Sigma) for 24 h or 10 ng/ml mitomycin C (MMC, M0503, Sigma) for 48 h, followed by colcemid (0.02 µg/ml, 50-591-851, GIBCO) for 3 h at 37°C to arrest cells in metaphase. Cells were then trypsinized, harvested, and incubated in 75 mM KCl for 15 min at 37°C to induce hypotonic swelling. The swollen cells were fixed in Carnoy’s fixative (3:1 methanol: acetic acid) with centrifuge and repeat fixation three times to ensure optimal chromosomal preservation. Fixed cells (in Carnoy’s fixative) were dropped onto pre-cleaned glass slides from a height of 10 cm to facilitate the spread of chromosome and were allowed to air-dry. For visualization, slides were stained with freshly prepared 5% Giemsa solution (10092013, GIBCO) in phosphate buffer (pH 6.8) for 10 min, followed by rising in distilled water and air-dried completely. After drying, the stained slides were mounted with a drop of Permount mounting medium (SP15-100, Thermo Fisher), covered with glass coverslip and left for overnight in room temperature. The slides were examined under the white field of Zeiss LSM 900 with a ×63/1.4 objective.
Statistical analysis
All statistical analyses were performed in GraphPad Prism 8 or Microsoft Excel. Results are reported as mean ± s.e.m. or s.d. of at least three independent experiments. Comparisons were analyzed using an unpaired Student’s t-test or one-way analysis of variance (ANOVA) (NS, P ≥ 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001), as indicated in the individual figures.
Results
DDX41 loss alters cell survival against genotoxic stress
Since mutations of DDX41 cause MDS/AML, and the dysfunction of DNA damage signaling and repair is one of the molecular pathogenesis of MDS/AML [77], we sought to explore the potential role of DDX41 in DNA damage response. We have established a DDX41 CRISPR–Cas9 knockout (KO) HeLa cell line [39], and we used the same sgRNA to knockout DDX41 in HT1080 cell line, which was from a patient who had not undergone radiotherapy or chemotherapy [86], making it less likely has unwanted mutations and widely used in DNA damage related research. Western blotting confirmed the gene disruption of DDX41 in HT1080 (Fig. 1A) and HeLa (Supplementary Fig. S1A). We treated wild-type (WT) and DDX41–KO cells with DNA damage reagents/treatments: bleomycin (BLM), camptothecin (CPT), and UV. Results showed that loss of DDX41 sensitized cells to BLM (Fig. 1B), CPT (Fig. 1C), and UV (Fig. 1D) exposure. We further examined endogenous and exogenous GFP-tagged DDX41 in HT1080 cells and found that DDX41 was predominantly present in the nucleus and foci formation appeared at 30 min post-BLM treatment, peaked at 2–4 h, and decreased until 24 h (Fig. 1E and F). Similar results were observed in HeLa cells under different genotoxic stress (Supplementary Fig. S1B–F). These results suggest that DDX41 loss impacts cell survivability against genotoxic stress, likely due to DNA damage response.
*DDX41 loss alters cell survival against genotoxic stress. (A) Western blot assays of WT and DDX41–KO HT1080 cell lines with a DDX41 antibody. β-Actin serves as a loading control. (B–D) Cell survival assays of WT and DDX41–KO HT1080 cells treated with bleomycin (B), camptothecin (C), or UV (D). (E and F) Immunofluorescence of endogenous DDX41 (E) and GFP tagged DDX41 (F) in HT1080 cells post-bleomycin treatment (30 μg/ml for 1 h). Quantification of DDX41 foci is shown at the bottom. The number of foci was calculated manually using 30 cells per individual panel from three independent experiments. UT, untreated with DNA damage. Data represent the mean ± SEM of three independent experiments; P < 0.05.
DDX41 loss leads to increased DSBs and delayed repair
Using γH2AX and 53BP1 as DNA DSB markers, we found that there were increased DSBs in DDX41–KO HT1080 cells compared with WT cells after BLM treatment (Fig. 2A and B). On average, there were 46 γH2AX foci/cell and 38 53BP1 foci/cell in DDX41–KO cells compared to 26 γH2AX foci/cell and 18 53BP1 foci/cell in WT cells. In addition, we performed neutral comet assays and found that DDX41–KO cells exhibited a mean tail moment of 375, compared to 198 in WT cells following BLM treatment (Fig. 2C), suggesting a substantial accumulation of fragmented DNA due to DDX41 loss. Based on above, we examined several key molecules’ expression in the DNA repair pathway by Western blotting. As expected, there was significant increased and/or prolonged γH2AX expression in DDX41–KO cells compared to WT cells after BLM treatment (Fig. 2D). Namely, γH2AX lasted until 4 h in WT compared to 24 h in DDX41–KO cells. In addition, reduced and delayed phosphorylation of BRCA1 (serine 1524) [87] was observed in DDX41–KO cells compared to WT cells. Similar results were obtained in HT1080 cells (Supplementary Fig. S2) and HeLa cells (Supplementary Fig. S3) treated with UV, hydroxyurea (HU), or IR. Taken together, our results suggest that DDX41 is required for DSB repair.
*Increased DSBs in DDX41–KO cells. (A and B) Fluorescent microscopy analysis of HT1080 cells (WT and DDX41–KO) stained with DDX41 and γH2AX (A) or 53BP1 antibody (B), and DAPI (blue) at 4 h post-bleomycin treatment (30 μg/ml for 1 h). Quantification of γH2AX or 53BP1 foci is shown on the right. The number of foci were calculated using 30 cells per individual panel in ImageJ. (C) Neutral comet assays of HT1080 cells (WT and DDX41–KO) at 4 h post-bleomycin treatment (30 μg/ml for 1 h). Quantification of comet moments is shown on the right. The value of comet moments were calculated using 30 cells per individual panel by ImageJ. (D) Western blot assays of HT1080 cell lines (WT and DDX41–KO) after bleomycin treatment (30 μg/ml for 1 h) with indicated antibodies. β-Actin serves as a loading control. Quantification of the relative level of γH2AX, p-BRCA1, and DDX41 is shown below. Data represent the mean ± SEM of three independent experiments; **P < 0.001. UT, untreated with DNA damage.
DDX41 loss leads to increased R-loops
R-loops have been associated with DSBs and genome instability [88], and DDX41 is involved in R-loop metabolism [80–82]. To determine how DDX41 affects R-loop formation, we checked the status of R-loops in DDX41-proficient and -deficient HT1080 cells. We collected cells at different time points after BLM treatment, and isolated nucleic acids from WT and DDX41–KO cells and treated with or without RNase H, which specifically degraded the RNA strand in DNA:RNA hybrids. Dot blot analysis with R-loop-specific-antibody S9.6 revealed increased R-loops in DDX41–KO cells compared with WT cells (Fig. 3A). Correspondingly, immunofluorescence analysis also showed increased R-loops in DDX41–KO cells after DNA damage (Fig. 3B). Increased R-loops were also found in DDX41–KO HT1080 cells under UV and HU treatments (Supplementary Fig. S4). These results suggest that DDX41 is required for R-loop resolution.
*Increased R-loops in DDX41–KO cells. (A) Dot blot analysis of nucleic acids with S9.6 antibody (top) or dsDNA antibody (bottom) with or without RNase H treatment after HT1080 cells (WT and DDX41–KO) treated with bleomycin (30 μg/ml for 1 h). Quantification of S9.6/dsDNA is shown on the right. UT, untreated with DNA damage. (B) Immunofluorescence staining with DDX41 and S9.6 antibodies with or without RNase H treatment in HT1080 cells (WT and DDX41–KO) post BLM treatment (30 μg/ml for 1 h). Quantification of S9.6 foci is shown on the right. Data represent the mean ± SD of three independent experiments for (A) and (B). ns, not significant (P ≥ 0.05); *P < 0.05, **P < 0.01, and **P < 0.001.
DDX41 interacts with R-loops and colocalizes with DSB marker γH2AX and 53BP1
After observing that DDX41 loss alters R-loop levels, we investigated whether DDX41 physically interacts with R-loops. Immunofluorescence analysis revealed partial colocalization between DDX41 and R-loops, as detected by the S9.6 antibody (Fig. 3B). However, previous studies have shown that S9.6 exhibits non-specific binding to dsRNA [89, 90]. Consistent with this, we observed that S9.6 also stained nucleoli in HT1080 cells, suggesting recognition beyond R-loops (Supplementary Fig. S5). To further assess the interaction between DDX41 and R-loops, we performed reciprocal co-IP assays with and without BLM treatment. We found that DDX41 co-precipitated with S9.6 (Fig. 4A), and conversely, S9.6 co-precipitated with DDX41 (Fig. 4B), with their association increasing upon DNA damage. To validate this interaction in situ, we conducted PLAs, which confirmed the association between DDX41 and R-loops (S9.6) and similarly showed enhanced interaction under DNA damage conditions (Fig. 4C). The mean DDX41–S9.6 PLA dots/nucleus in untreated cells was 1 and increased to 19 in BLM-treated cells. Notably, we observed that some DDX41–S9.6 PLA signals were located outside the nucleus, in contrast to their predominantly nuclear localization observed by immunofluorescence (Fig. 3B). Given that PLA is up to 1000-fold more sensitive than conventional immunofluorescence, which likely enabled detection of low-abundance interactions between DDX41 and RNA:DNA hybrids. Also at certain conditions such as BLM treatment, DDX41 may be sequestered in the cytoplasm due to the accumulation of cytoplasmic RNA:DNA hybrids, consistent with previous reports of cytoplasmic localization of both R-loops [91] and DDX41 [92–94].
*DDX41 interacts with R-loops and colocalizes with γH2AX. (A) Dot blot analysis of nucleic acids with the S9.6 antibody after HT1080 cells (WT and DDX41–KO) treated with or without bleomycin (30 μg/ml for 1 h) and immunoprecipitated with a DDX41 antibody or normal mouse IgG; IB, immunoblotting. (B) Co-IP of DDX41 protein by the S9.6 antibody in HT1080 cells (WT and DDX41–KO) that were treated with or without bleomycin (30 μg/ml for 1 h) and blotted with a DDX41 antibody. Normal mouse IgG was used as a control; IP, immunoprecipitation. (C) Representative images of PLA between DDX41 and S9.6 in HT1080 cells (WT and DDX41–KO) that were treated with or without bleomycin (30 μg/ml for 1 h) and quantification of PLA signal (right panel). Thirty cells were counted in each PLA by ImageJ. (D) Immunofluorescence staining of WT HT1080 cells with DDX41 (green), γH2AX (red), and DAPI (blue) at 4 h after bleomycin treatment (30 μg/ml for 1 h) or untreated (UT). Quantification of colocalization with Pearson’s correlation coefficient is shown on the right. The Pearson’s correlation coefficient was obtained using the JACoP plugin in ImageJ for at least 30 cells. (E) Representative images of PLA between DDX41 and γH2AX or 53BP1 in HT1080 cells (WT and DDX41–KO) that were treated with bleomycin (30 μg/ml for 1 h) and quantification of PLA signal (right panel). Data represent the mean ± SEM of three independent experiments. ns, not significant (P ≥ 0.05), **P < 0.001.
To further investigate whether DDX41 is directly involved in DNA repair, we determined the colocalization of DDX41 and γH2AX. Intriguingly, we found partial colocalization of DDX41 and γH2AX in HT1080 cells after BLM treatment, with Pearson’s correlation coefficient of 0.55 after BLM treatment (Fig. 4D). The dynamic colocalization of DDX41 and γH2AX was observed during 24 h post-BLM treatment (Supplementary Fig. S6A). Exogenous GFP tagged DDX41 also colocalized with γH2AX after BLM treatment (Supplementary Fig. S6B). Using 53BP1 as a DSB marker, similar results were obtained in HT1080 cells and HeLa cells (Supplementary Fig. S7). Additionally, this colocalization experiment also confirmed that BLM-induced γH2AX/53BP1 foci were higher and prolonged (up to 24 h) in DDX41–KO cells compared with the WT cells. Moreover, we conducted PLA and confirmed the association between DDX41 and γH2AX/53BP1, with two PLA dots/nucleus for DDX41–γH2AX interaction and four PLA dots/nucleus for DDX41–53BP1 interaction (Fig. 4E). Collectively, these results suggest that DDX41 localizes at DSB sites and might participate in the DNA repair process in the cells.
DDX41 loss diminishes homologous recombination repair
DSBs are usually repaired by NHEJ, HR, and single strand annealing (SSA) repair pathways. Given that DDX41 partially colocalizes with γH2AX and 53BP1, we next examined the role of DDX41 in DSB repair. We used the U2OS GFP reporter cell lines, where I-SceI endonuclease generates a DSB in the cassette of the dysfunctional GFP gene, which can be repaired via classical NHEJ (detected by EJ5–GFP), alternative end joining (alt-EJ by EJ2–GFP), HR (DR–GFP), or SSA (SA–GFP) [83]. Four siRNAs targeting DDX41 were tested, and knockdown was detected by Western blot (Supplementary Fig. S8). We combined siRNA 2 and 4 and caused about 70% knockdown of DDX41 protein in the four cell lines, while I-SceI was expressed at similar levels (Fig. 5A). FACS analysis revealed that DDX41 knockdown by siRNA led to significant reduction in the proportion of cells that repaired by HR to become GFP positive, but not by NHEJ, Alt-EJ, and SSA repair (Fig. 5B and Supplementary Fig. S9). We observed that cell cycle was not affected by DDX41 siRNA knockdown in the test cells (Supplementary Fig. S10), suggesting that the impact of HR is not due to G1 arrest. Moreover, U2OS cells with siRNA knockdown DDX41 were also sensitive to DNA damage (Supplementary Fig. S11), which are consistent with DDX41–KO HT1080 and HeLa cells used earlier.
*DDX41 loss diminishes HR repair. (A) Western blot assays of DDX41 siRNA and siRNA control in U2OS GFP reporter cell lines with indicated antibodies. β-Actin serves as a loading control. Quantification of relative protein level for DDX41 and I-SecI is shown on the right. (B) Percentages of GFP positive cells as assessed by flow cytometry 36 h after U2OS GFP reporter cell lines treated with siRNA control, siRNA control + I-SceI plasmid, or DDX41 siRNA + I-SceI plasmid. (C) Cell survival assays of WT and DDX41–KO HT1080 cells treated with Olaparib. (D and E) Immunofluorescence staining of WT and DDX41–KO HT1080 cells with γH2AX, RPA32 (D) or RAD51 (E), and DAPI without bleomycin treatment (UT) or 4 h post BLM treatment (30 μg/ml for 1 h). Quantification of RPA32 and RAD51 foci is shown in the middle. Data represent the mean ± SEM of three independent experiments. *P < 0.05, ***P < 0.001, and ***P < 0.0001.
HR deficiency is known to confer sensitivity to PARP inhibition [95]. To investigate this, we performed cell survival assays using the PARP inhibitor Olaparib in HT1080 cells and observed that DDX41–KO cells exhibited increased sensitivity to Olaparib (Fig. 5C). Additionally, HR defects are associated with Fanconi anemia (FA), a disorder characterized by chromosomal instability and a high frequency of radial structures following exposure to DNA interstrand crosslinking agents [96]. To assess whether DDX41 is involved in FA-related pathways, we treated cells with cisplatin or MMC and conducted cytogenetic analyses. DDX41–KO cells did not display significant radial figures, chromatid breaks, or complex chromosomal rearrangements (Supplementary Fig. S12). Although further investigation is warranted, these findings suggest that DDX41 is unlikely to be part of the expanding group of 23 known FA genes [97].
In HR repair, DSBs are processed by nucleases to create ssDNA that are coated by RPA, which will be replaced by RAD51 to form a nucleoprotein filament on the ssDNA. Therefore, next we examined the RPA and RAD51 foci in DDX41-deficient and proficient cells with or without BLM treatment. Intriguingly, DDX41–KO cells exhibited increased RPA foci and reduced RAD51 foci compared with WT cells following DNA damage (Fig. 5D and E). To further examine the temporal dynamics of these responses, we assessed RPA and RAD51 foci formation over a 24-h period after BLM treatment. Consistently, DDX41 KO cells showed elevated RPA foci (Supplementary Fig. S13A) and diminished RAD51 foci relative to WT cells (Supplementary Fig. S13B). Notably, RPA foci peaked at 2 h in WT cells but were delayed to 4 h in DDX41–KO cells, whereas RAD51 foci reached maximal levels at 4 h in both genotypes. These findings indicate that the transition from RPA-coated ssDNA to RAD51 filament assembly is both delayed and reduced in DDX41-deficient cells. Taken together, these results suggest that DDX41 is required for DSB repair through HR, though its exact function remains to be determined.
MDS/AML mutant R525H fails to unwind DNA:RNA hybrids in vitro
Since R-loops are increased in DDX41–KO cells (Fig. 3 and Supplementary Fig. S4), we asked whether DDX41–R525H mutant proteins can suppress R-loop formation directly. Using DDX41–WT and DDX41–R525H mutant proteins [39], helicase assays showed a significant decrease in the unwinding activity in the mutant (Fig. 6A), indicating that the mutant lost its helicase activity. Further, we compared the annealing activities of both WT and mutant; there was no significant difference (Fig. 6B). Interestingly, when the same annealing assay reaction was performed in the presence of ATP, DDX41–WT unwound the DNA:RNA hybrid, whereas DDX41–R525H mutant failed to do so (Fig. 6C). In summary, these data suggest that DDX41–WT resolves DNA:RNA hybrids, whereas R525H tends to accumulate them, indicating that this mutation is associated with loss of function.
DDX41 protein resolves R-loop structures in vitro but not R525H mutant. (A) Representative image of helicase reactions performed by incubating 0.5 nM 3′ tailed 30-bp DNA:RNA hybrid substrate with increasing protein concentration (0–300 nM) of DDX41–WT (left) and DDX41–R525H (right) at 37°C for 15 min. (B and C) Representative images of strand annealing reactions performed by incubating 0.5 nM of 30-mer ssRNA (32P labeled) and 45-mer ssDNA (unlabeled) substrate with increasing protein concentration (0–300 nM) of DDX41–WT (left) and DDX41–R525H (right) at room temperature for 30 min, without ATP (B) and with ATP (2 μM, C). Quantification for panels (A–C) is shown on the right. Data represent the mean ± SD of three independent experiments. DNA is indicated with a black line, RNA with a gray line. NE, no enzyme; filled triangle, heated; no cold, no unlabeled complementary ssDNA.
MDS/AML patient mutant R525H accumulates R-loops and attenuates HR in cells
To investigate further, DDX41–WT or DDX41–R525H gene was re-expressed in DDX41–KO HeLa cell lines [39]. After DNA damage treatment, we determined the DNA damage response-related proteins’ expression by Western blot and R-loop status by dot blot (Fig. 7A and B). Overexpressed R525H mutant showed significantly higher γH2AX expression along with delayed and diminished DNA repair protein’s expression, such as pBRCA1, after BLM (Fig. 7A), UV (Supplementary Fig. S14A), or HU (Supplementary Fig. S14B) treatment. Correspondingly, dot blot assays showed that R525H had elevated and persistent RNA: DNA hybrids compared to WT after BLM (Fig. 7B), UV (Supplementary Fig. S14B), or HU (Supplementary Fig. S14C) treatment. Cell survival assays revealed that DDX41–R525H-expressing cells were sensitive to BLM, CPT, UV, and Olaparib (Fig. 7C). We also observed increased RPA foci and reduced RAD51 foci in DDX41–R525H-expressing cells compared with DDX41–WT-expressing cells following DNA damage (Fig. 7D and E), resembling DDX41 KO cells (Fig. 5D and E). Similarly, cells expressing the R525H mutant displayed increased RPA foci (Supplementary Fig. S15A) and reduced RAD51 foci compared with WT–expressing cells (Supplementary Fig. S15B). Interestingly, RPA foci reached their peak at 2 h in WT-expressing cells but were delayed until 4 h in R525H-expressing cells, whereas RAD51 foci peaked at 4 h in both groups. The delay in RPA-to-RAD51 transition in R525H-expressing cells is analogous to that observed in DDX41 KO cells (Supplementary Fig. S13). We conclude that the patient-derived R525H mutation impairs the resolution of R-loops, particularly those formed at DSB sites, leading to dysregulation of the DNA damage response and subsequent genomic instability, which might underline the molecular pathogenesis in patients with MDS and AML.
*MDS/AML patient mutant R525H delays DSB repair and accumulates R-loops when re-expressed in DDX41–KO cells. (A) Western blot assays of indicated genotype HeLa cell lines with indicated antibodies after BLM treatment (30 μg/ml for 1 h). β-Actin serves as a loading control. Quantification of relative proteon level is shown on the right. (B) Dot blot analysis of nucleic acids with S9.6 antibody (top) or dsDNA antibody (bottom) with or without RNase H treatment after bleomycin treatment (30 μg/ml for 1 h) in indicated genotype HeLa cell lines. Quantification of S9.6/dsDNA is shown on the right. (C) Cell survival assays of DDX41–KO HeLa cells re-expressed with DDX41–WT or DDX41–R525H gene, and treated with bleomycin, camptothecin, UV, or Olaparib. (D and E) Immunofluorescence staining of DDX41 KO HeLa cells re-expressed with DDX41–WT or DDX41–R525H gene with γH2AX, RPA32 (D) or RAD51 (E), and DAPI without bleomycin treatment (UT) or 4 h post BLM treatment (30 μg/ml for 1 h). Quantification of RPA32 and RAD51 foci is shown on the right. Data represent the mean ± SEM of three independent experiments. ns, not significant (P ≥ 0.05); *P < 0.05, **P < 0.01,***P < 0.001, and ***P < 0.0001.
Discussion
DDX41 has been shown to act in conventional pre-mRNA splicing [6, 98, 99] and other RNA processing, such as pre-rRNA processing [100] and small nucleolar RNA processing [101]. DDX41 is also known as a host intracellular DNA sensor against DNA virus infection [39, 94]. The connection between DDX41 and R-loop was first found in a zebrafish ddx41 mutant model [81] and confirmed in DDX41 mutated human cells [82] and liver cancer [102]. A proximity proteomics approach provided direct evidence that DDX41 is an R-loop binding partner and resolver [80]. In this study, we found that DDX41 plays a critical role in R-loop resolution and DSB repair, particularly HR. Therefore, DDX41 is a multifaceted protein involved in RNA processing, innate immunity, and DNA repair.
DEAD-box helicases are usually involved in RNA metabolism, such as mRNA splicing, translation initiation, ribosome biogenesis, and RNA degradation [103]. However, more than a dozen DEAD-box helicases have been identified in DNA repair process. For example, DDX5, also known as p68, localizes to DNA damage sites [104]. DHX9, also known as RNA helicase A, is directly associated with the DNA repair marker γH2AX after actinomycin D treatment in HeLa cells [105, 106]. DDX1 is found to facilitate RNA removal and HR at DSBs [66]. The Bowman lab found that DDX41 is essential for erythropoiesis in zebrafish [107]. They found that the deficiency and mutations in DDX41 lead to ATM- and ATR-mediated cell arrest triggered by DNA damage, genomic stress, misexpression of genes, and alternative splicing of genes related to the cell cycle. Our data support their findings and prove that DDX41 is another DEAD-box protein that functions in DNA repair.
The novel function of DDX41 concerning R-loop structures is just emerging. The Bowman group found excess R-loops formed in DDX41 mutant zebrafish models, where excess R-loops disequilibrate the hemogenic endothelium, hematopoietic stem and progenitor cell populations required for cellular fitness [108]. The Beli lab reported that DDX41 siRNA knockdown leads to replication stress, DNA damage, inflammatory signaling, and the accumulation of R-loops [80]. However, this accumulation is opposed by the presence of DDX41 through unwinding DNA:RNA hybrids at the gene promoter region. Correspondingly, we showed that DDX41 protein unwinds DNA:RNA hybrids in vitro, suggesting its function in removing and regulating R-loops. We also performed in vitro annealing assays using DDX41–WT and R525H mutant proteins and deciphered that both could anneal the DNA:RNA hybrids without any significant difference. Excitingly, a significant difference was observed when this annealing assay was performed in the presence of ATP; when the DDX41 R525H mutant showed an accumulation of R-loop structures, and WT was able to unwind the DNA:RNA hybrids. Herein, we conclude that DDX41 is required for resolving and regulating R-loops.
Compelling evidence has implicated several other DEAD-box proteins in suppressing co-transcriptional R-loop structures, including DDX5, DDX18, DDX19, DDX21, DDX23, and DDX47. DDX5 counteracts DSB-related DNA deletions and repair defects by resolving persistent R-loops [67]. DDX5 is also reported to promote HR repair by acting as an R-loop-resolvase at the sites of DSBs in U2OS cells [109]. DDX18 associates with PARP-1 and prevents R-loop accumulation [69]. DDX19 was discovered to utilize its helicase activity to resolve R-loops in vitro [110]. In a similar study, the phosphorylation of DDX23 by a serine/arginine protein kinase 2 (SRPK2) was found to initiate a signal transduction cascade to help DDX23 resolve R-loops [111]. DDX21 was discovered to cooperate with SIRT7 to resolve R-loops and safeguard genome stability [112]. DDX47 was added to the list as it interacts with FA protein, FANCD2, during mild replication stress to decrease the number of R-loops by reducing TRCs [113]. DDX21 and H3K27me3 demethylase JMJD3 were shown to resolve aberrant R-loops [114]. It is notable that some helicases have been reported to promote R-loop formation, including DHX9 [74] and UPF1 [52]. Thus, a fundamental question arises: Why do cells need so many R-loop helicases, even in the same DEAD-box family, and with opposite functions? One possibility is that various DEAD-box helicases act distinctly on R-loops based on (i) type of R-loop, (ii) different genome sites, (iii) cell- or tissue-specificity, and (iv) unknown context-dependent inhibiting and promoting of R-loops. Nevertheless, further studies are required to resolve this mystery.
Our working model supplements the exciting possibility of DDX41 that is associated with R-loop, genome stability, and diseases (Fig. 8). According to the recent findings by the Beli group [80], DDX41–WT co-transcriptionally resolves any unwanted R-loop structures during transcription in the gene promoter region, whereas the DDX41–R525H mutant fails to do that, hence accumulating R-loops result in TRCs, DSBs, genome instability, inflammatory response, and possibly MDS and AML (Fig. 8A). DSBs are formed as a consequence of DNA damage (from any sources) or TRCs in the cells, and there is a tendency for R-loop accumulation at the sites of DSB. DDX41–WT resolves R-loops, whereas the DDX41–R525H mutant fails, resulting in R-loop accumulation. Elevated R-loop levels generate persistent ssDNA that is subsequently coated by RPA, providing a mechanistic explanation for the increased and prolonged RPA foci observed in DDX41–KO and DDX41–R525H-expressing cells (Figs 5D and 7D). Consistent with our findings, recent studies have reported impaired RAD51 recruitment in DDX41–KO and knock-in (Y259C or R525H) K562 cells, as well as in CD34⁺ cells from DDX41-mutated MDS patients [82]. Together, these findings suggest that loss of DDX41 function—such as through the R525H mutation—leads to R-loop accumulation, increased ssDNA exposure, elevated DSBs, prolonged RPA binding, and reduced RAD51 recruitment, thereby compromising RAD51 loading and delaying HR repair kinetics. The HR defects contribute to genomic instability and potentially drive the development of MDS/AML (Fig. 8B). Despite increased RPA accumulation indicating that DNA end resection is initiated in DDX41–deficient cells, the concomitant reduction in RAD51 foci suggests a defect in the transition from RPA-coated ssDNA to RAD51 filament formation, consistent with reduced HR (Fig. 5B). This phenotype—characterized by reduced HR efficiency, increased and prolonged RPA foci, and diminished RAD51 focus formation—has been reported previously in cells deficient in other HR factors, including BRCA1 [115], BRCA2 [116], PALB2 [117], and RAD51 paralogs (RAD51B/C/D, XRCC2/3) [118]. Nevertheless, the exact role of DDX41 in the RPA → RAD51 transition step and whether DDX41 directly participates in end resection and Holliday junction dissolution remains to be determined.
Proposed the roles of DDX41 in R-loop mediated genome instability and MDS/AML. (A) Based on the report by the Beli group [72], DDX41 associates with R-loops in promoters of active genes and balances R-loop levels by unwinding RNA:DNA hybrids. Pathogenic DDX41 mutant R525H displays impaired RNA:DNA hybrid unwinding activity, leading to the accumulation of R-loops at promoters, resulting in increased replication stress, DSBs, inflammatory signaling, and MDS/AML. (B) A proposed model of DDX41 resolves R-loops (DNA:RNA hybrids) during HR process at the DSB site. DSBs form by TRC from panel (A) or exogenous sources and R-loops form at DSB sites. DDX41–WT resolves R-loops and facilitates DNA repair; in contrast, DDX41–R525H fails to resolve R-loops and delays DNA repair, such as RPA coating, RAD51 filament formation and Holliday junction dissolution. Therefore, defective HR and additional DNA damage lead to genome instability and ultimately to MDS/AML. For simplicity, only a simplified HR pathway and a subset of relevant proteins are depicted. The later step, either the double-strand break repair (DSBR), synthesis-dependent strand annealing (SDSA), break-induced replication (BIR), or single-strand annealing (SSA) pathway is not shown. Whether DDX41 directly participates in end resection, RAD51 loading or Holliday junction dissolution remains unknown. RNAP, RNA polymerase; MRN, MRE11, RAD50, and NBS1; CtIP, CtBP-interacting protein; EXO1, exonuclease 1.
Recently, single-cell transcriptome analysis revealed that R525H expressing MDS patient cells have increased DNA damage and R-loops [119]. It is not clear how accumulated R-loops lead to further DSB or delay HR. One possibility is that the displaced ssDNA is more susceptible to nuclease attacks, such as XPF, XPG, FEN1, and MRE11 [120], leading to SSBs and DSBs. This is supported by a study from the Wong’s lab, who found that cell cycle inhibitors cause accumulation of fragmented DNAs in the cytosol that activates the DDX41–STING pathway [121]. Our Western blot results showed the levels of 53BP1 and pBRCA1 largely mirrored each other, suggesting that DDX41 might be involved in HR-mediated DNA repair. Indeed, using the U2OS DR–GFP cell line, we found that knockdown of DDX41 results in reduced HR efficiency. Therefore, we believe that the role of R-loop in DSB might be context-dependent (e.g. cell cycle) and regulated on time. For instance, prolonged R-loops at DSB sites make the ssDNA vulnerable and delay the DNA repair process.
Several helicases have been implicated in the HR process. BLM promotes end resection and Holliday junction dissolution (via the BTR complex: BLM–TOP3A–RMI1/2) [122]. WRN participates in end resection and resolution of recombination intermediates [123]. RTEL1 disrupts displacement loops (D-loops) formed during strand invasion [124]. HelQ facilitates RAD51 loading and strand invasion [125]. Moreover, some helicases are involved in R-loop-mediated HR. For example, DHX9 [75, 76] and HELZ [126] resolve R-loops and facilitate HR. DDX5 is recruited by BRCA2 to DSB sites for resolving R-loops [127]. DDX1 resolves DNA:RNA hybrids at DSB sites [66]. Interestingly, DDX1 facilitates R-loop formation by resolving G4 [72]; in contrast, DDX41 inhibits R-loop formation and resolves G4 [128]. Whether DDX41 functions redundantly or synergistically with these helicases in R-loop-mediated HR remains to be determined.
While it is compelling to apply the DDX41 model to the pathogenesis of MDS and AML, further investigation is needed to elucidate how additional pathways interact and influence one another in this context. Our study provides new insights into the interplay between DDX41-mediated regulation of the R-loop formation and DNA damage response, highlighting their potential relevance to the development of MDS, AML, and possibly other diseases. Moreover, we found that DDX41 knockout and R525H expressing cells exhibit sensitivity to the PARP inhibitor Olaparib (Figs 5C and 7C), suggesting a potential therapeutic avenue. This finding raises the possibility that MDS/AML patients with DDX41 mutations could benefit from PARP inhibitor therapy, similar to BRCA1/2-mutated breast, ovarian, prostate, and pancreatic cancers.
Supplementary Material
gkag219_Supplemental_File
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Tawana K, Fitzgibbon J. Inherited DDX 41 mutations: 11 genes and counting. Blood. 2016;127:960–1. 10.1182/blood-2016-01-690909.26917736 · doi ↗ · pubmed ↗
- 2Truong P, Pimanda JE. DDX 41: the poster child for familial MDS/AML grows up. Blood. 2023;141:447–9. 10.1182/blood.2022018787.36729548 · doi ↗ · pubmed ↗
- 3Ogawa S . Genetics of MDS. Blood. 2019;133:1049–59. 10.1182/blood-2018-10-844621.30670442 PMC 6587668 · doi ↗ · pubmed ↗
- 4Menssen AJ, Walter MJ. Genetics of progression from MDS to secondary leukemia. Blood. 2020;136:50–60. 10.1182/blood.2019000942.32430504 PMC 7332895 · doi ↗ · pubmed ↗
- 5Ding L, Ley TJ, Larson DE et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. Nature. 2012;481:506–10. 10.1038/nature 10738.22237025 PMC 3267864 · doi ↗ · pubmed ↗
- 6Polprasert C, Schulze I, Sekeres MA et al. Inherited and somatic defects in DDX 41 in myeloid neoplasms. Cancer Cell. 2015;27:658–70. 10.1016/j.ccell.2015.03.017.25920683 PMC 8713504 · doi ↗ · pubmed ↗
- 7Lewinsohn M, Brown AL, Weinel LM et al. Novel germ line DDX 41 mutations define families with a lower age of MDS/AML onset and lymphoid malignancies. Blood. 2016;127:1017–23. 10.1182/blood-2015-10-676098.26712909 PMC 4968341 · doi ↗ · pubmed ↗
- 8Cardoso SR, Ryan G, Walne AJ et al. Germline heterozygous DDX 41 variants in a subset of familial myelodysplasia and acute myeloid leukemia. Leukemia. 2016;30:2083–6. 10.1038/leu.2016.124.27133828 PMC 5008455 · doi ↗ · pubmed ↗