MYBL2 Supports DNA Double Strand Break Repair in Hematopoietic Stem Cells

Myelodysplastic syndrome (MDS) is an age-associated hematopoietic malignancy, characterized by abnormal blood cell maturation and a high propensity for leukemic transformation. It is a clonal disease thought to originate in the hematopoietic stem cell (HSC; ref. 1). Effective management and treatment of MDS is important, as early identification of patients who are likely to progress to malignant disease allows for an optimal therapeutic regime to be implemented. Genetic alterations are often present in MDS and a frequent chromosome abnormality is del20q, whose common deleted region only contains 5 genes expressed in HSCs, one of which is MYBL2 (2, 3). The MYBL2 gene encodes a ubiquitously expressed protein belonging to the MYB family of transcription factors and has been shown to form part of different protein complexes such as the Myb-MuvB/DREAM complex (4–7), Myb–Clafi complex (8), and the MRN complex (9), through which it exerts its vital role in cell-cycle regulation and maintenance of genome stability (10–14). Analysis of publicly available global gene expression data from CD34⁺-MDS patient cells (15) have confirmed that downregulation of MYBL2 expression correlates with poor prognosis; even in patients with a normal karyotype (2, 3). This suggests that changes in MYBL2 expression could have significant consequences with regards to disease pathogenesis. Furthermore, it has been demonstrated that mice with low levels of Mybl2 develop hematologic disorders during ageing that closely resemble the human disease, implying that MYBL2 functions as a haploinsufficient tumor suppressor gene (2, 3). However, how low MYBL2 expression contributes to MDS during ageing remains unknown.

Given that HSCs must last for the entire lifetime of an individual to guarantee continuous blood cell production, this increases the dependence of these cells on DNA repair to maintain genomic integrity. Because HSCs are predominantly quiescent, they are thought to primarily utilize nonhomologous end joining (NHEJ) rather than homologous recombination (HR) to repair DNA double-strand breaks (DSB; refs. 16, 17). Although mainly error-free, NHEJ-dependent DSB repair can also result in the generation of small genomic deletions at the repaired break site, leading to the hypothesis that HSCs accumulate somatic mutations over time. This is thought to precede the appearance of blood disorders such as MDS and acute myeloid leukemia (AML; refs. 18–20), although no direct link between DNA repair and the pathogenesis of MDS has been reported.

Because recent studies have shown that MDS originates from HSCs and that MYBL2 is known to play a role in maintaining genome stability (1, 21, 22), we hypothesized that low levels of MYBL2 may compromise the DNA repair capacity of the cell, resulting in the accumulation of genetic alterations to a sufficient level to induce HSC transformation. To test this, we used ionizing radiation (IR) in vitro to induce DNA damage in MDS patient's stem cells. Following treatment, we determined the ability of these cells to repair their DNA and correlated this with expression levels of MYBL2. To further study the role of MYBL2 in DNA repair, we utilized a Mybl2 haploinsufficient mouse model, which is susceptible to MDS development. Our findings uncover a novel role for MYBL2 in regulating DSB repair in the HSC population.

To assess differential gene expression and pathway enrichment between MDS samples displaying higher and lower levels of MYBL2 expression, we used previously published microarray data (15) deposited in the NCBI Geo DataSets repository (GSE19429), and the BROAD Institute Gene Set Enrichment Analysis (GSEA) software (23, 24). Differential expression was assessed over 1,000 permutations and ranked according to signal-to-noise ratio. A weighted enrichment statistic was applied. Gene sets comprising less than 15 genes were excluded from the analysis (the list of gene set used in the analysis is presented as Supplementary Tables S1 and S2). The adjustment of the FAB composition was done using a method of random sample removal. To balance the composition of the MYBL2hi and MYBL2lo sets, samples of specific diagnosis (for example RA or RAEB) were randomly removed when they were found to be over-represented in a set. Four different permutations were performed to verify that the results were not affected by the methodology.

Peripheral blood samples from patients with MDS were obtained in heparin-coated vacutainers. Peripheral blood mononuclear cells were isolated using Ficoll-Paque (GE Healthcare) and stored at −80°C. Cells were thawed and cultured for 8 days in expansion medium as described previously (25), with the exception that the base medium was StemSpan H3000 (Stem Cell Technologies). Medium was refreshed on day 3 and 6. On day 8, cells were harvested and CD34⁺ cells purified using microbeads (Miltenyi Biotec).

For human gene expression assays, qRT-PCR for MYBL2 (Hs00942543_m1 MYBL2, Applied Biosystems) was carried out using TaqMan PCR Master Mix (Applied Biosystems) and qRT- PCR for β-glucuronidase (HsGusB QT00046046, Quantitect primer assay, Qiagen) was carried out using SYBRGreen Master Mix (Thermo Fisher Scientific). For murine gene expression assays, qRT-PCR for p21 (Mm01303209_m1), Puma (Mm00519268_m1), Noxa (Mm00451763_m1), Bax (Mm00432051_m1), and β-2-microglobulin (Mm00437762_m1) were carried out using TaqMan PCR Master Mix (Applied Biosystems). Reactions were carried out in a Stratagene Mx3000P machine and samples were run in duplicate. Relative gene expression was calculated as 2^−ΔΔC_t values relative to control genes (β-glucuronidase for human samples and β-2-microglobulin for murine samples).

Mice were maintained on a C57/BL6 background and genotyped by Transnetyx. For mouse studies, no specific randomization or blinding protocol was used during experimental protocols. Mice of both genders were used. Age- and gender-matched mice were used per experiment. Seventy-week-old healthy mice were chosen to perform aging studies. Disease-free status of these animals was assessed on the basis of behavior and physical appearance of the mice, normal values of white blood cell, red blood cell, and platelets obtained from peripheral blood counts, and by internal organ examination after dissection, in particular no signs of splenomegaly.

Inhibitors were dissolved in DMSO; KU60019 (10 μmol/L) was used for inhibition of ATM and NU7441 (1 μmol/L) was used for inhibition of DNA-dependent protein kinase (Tocris Bioscience).

Single-cell suspensions of bone marrow were prepared using standard techniques and red blood cells were depleted by ACK lysis (0.15 mol/L NH₄Cl, 1 mmol/L KHCO₃, 0.1 mmol/L EDTA, pH 7.4). Nonspecific antibody binding to Fc receptors was blocked using anti-CD16/CD32 (93, eBioscience) and the cells were stained with a combination of fluorochrome-conjugated antibodies including anti-mouse lineage; CD5, CD8a, CD11b, Gr-1, Ter119, B220 (APC or FITC), cKit (PeCy5 or e780; 2B8, eBioscience), Sca-1 PeCy7 (D7, eBioscience), Flk2 PE (A2F10, eBioscience), CD48 APC (HM48-1, eBioscience), and CD150 PEcy7 (TC15-12F12.2, BioLegend), to allow identification of Flk2⁻ HSCs (lineage⁻cKit⁺ Sca-1⁺Flk2⁻) and long-term HSCs (SLAM staining; lineage⁻cKit⁺Sca-1⁺CD48⁻CD150⁺). Some cells were analyzed directly by flow cytometry using a CyAn ADP Analyzer (Beckman Coulter) and some were sorted using a Cytomation XDP MoFlo machine (Beckman Coulter). In both cases, data were analyzed using either Summit software (Dako) or FlowJo software (FLOWJO, LLC). When cells were required for sorting, a cKit⁺ enrichment using streptavidin microbeads (130-048-101, Miltenyi Biotec) and columns (130-042-201, Miltenyi Biotec) was performed prior staining.

For proliferation assays in vivo using BrdU, mice were given an intraperitoneal injection of 2 mg BrdU in PBS and 24 hours later animals were sacrificed. cKit-enriched bone marrow cells were isolated using anti-mouse cKit biotin (eBioscience) and streptavidin microbeads (Miltenyi Biotec) as per the manufacturer's instructions. For proliferation assays in vitro using BrdU, expanded CD34⁺ cells were labeled with 10 μmol/L BrdU for 3.5 hours. Cells were stained using the BrdU Flow Kit (8811-6600, BD Biosciences) according to the manufacturer's instructions. For proliferation assays using Ki67, cKit-enriched bone marrow cells were stained using the Ki67 flow kit (BD Biosciences) according to the manufacturer's instructions. For colony-forming assays, purified HSCs (lineage⁻cKit⁺Sca-1⁺Flk2⁻) were obtained by sorting. Five-hundred HSCs were plated in Methocult (Stem Cell Technologies) supplemented with penicillin/streptomycin (Invitrogen) in 35-mm petri dishes and incubated for 6 days at 37°C in an atmosphere containing 5% CO_2. Colonies were counted using a standard light microscope with ×10 objective. For G₂–M checkpoint studies, cKit-enriched bone marrow cells were isolated as described previously. A total of 1–2 × 10⁶ cKit-enriched cells were cultured for 18 hours in Iscove's modified Dulbecco's medium (IMDM) containing 10% heat-inactivated FBS (HIFBS), 3% penicillin/streptomycin, 1 mmol/L sodium pyruvate, 2 mmol/L l-glutamine, 0.1 mmol/L nonessential amino acids, 50 μmol/L 2-mercaptoethanol, 25 ng/mL SCF, 10 ng/mL IL3, 25 ng/mL IL11, 25 ng/mL TPO, 4 U/mL EPO, 10 ng/mL GM-CSF, and 25 ng/mL FLT3L (complete cytokine medium). Medium was replaced and the cells were cultured for 1 hour with DMSO or 10 μmol/L KU60019 prior to irradiation (IR) in vitro (2 Gy). Cells were cultured for a further 5 hours and stained with (i) cell surface marker antibodies to identify subpopulations; (ii) mouse anti-phospho histone H3 Ser10 (2-hour staining; clone 6G3, Cell Signaling Technology) and goat anti-mouse Alexa 488 (30-minute staining; A-11001, Molecular Probes); and (iii) Vybrant DyeCycle (Molecular Probes). Fixation and permeabilization were carried out using buffers from the BrdU Flow Kit (BD Biosciences) according to the manufacturer's instructions. For apoptosis assays, whole bone marrow cells were stained with cell surface marker antibodies to identify subpopulations as well as mouse anti-cleaved PARP (Asp214; clone F21-852, BD Biosciences).

Purified murine cells (Flk2⁻ HSCs or SLAM HSC) or human CD34⁺ cells were cytospun onto microscope slides for 5 minutes at 800 rpm. For 53BP1, γH2AX, and MRE11 staining, cells were treated with CSK buffers (buffer 1 for 4 minutes at room temperature, buffer 2 for 1 minute at room temperature), washed in PBS, and fixed in 4% PFA/PBS for 15 minutes at room temperature. Fixation was quenched with 50 mmol/L ammonium chloride for 5 minutes at room temperature and the cells were permeabilized with 0.3% Triton-X 100/PBS for 5 minutes at room temperature. For pKap1, cells were fixed in 4% PFA/PBS for 10 minutes at room temperature, washed in PBS, and permeabilized with ice-cold methanol for 10 minutes at room temperature. Cells were blocked with 3% BSA/10% HIFBS/1% goat serum/0.3% Triton-X 100/PBS (blocking buffer) for 1 hour at room temperature. Cells were incubated with primary antibodies to 53BP1 (NB-100-904, Novus Biologicals), pKAP1 (S824;A300-767A, Bethyl Laboratories Inc), γH2AX (JBW301, Merck), MRE11 (4895S, Cell Signaling Technology), mouse IgG control (G3A1, Cell Signaling Technology), or rabbit IgG control (sc-2027, Santa Cruz Biotechnology) and diluted in blocking buffer overnight at 4°C. Cells were washed three times in 0.1% Tween 20/PBS and incubated with goat anti-rabbit Alexa 488 secondary antibody diluted in blocking buffer for 1 hour at room temperature. Cells were washed three times in 0.1% Tween 20/PBS, dipped in Milli-Q water, and mounted with ProLong Gold AntiFade Reagent containing DAPI (Invitrogen). Microscopy imaging was performed using a Zeiss LSM 510 Meta confocal microscope (100× objective NA 1.4 lens) and the images were analyzed using ImageJ software. For all immunofluorescence staining, 30–50 cells were scored for each independent experiment. Experiments were performed at least three times and results represent a minimum of 3 animals. For 53BP1 staining on CD34⁺ cells from patients with MDS, a minimum of 25 cells were scored per patient per condition.

HSCs (Lin⁻/cKit⁺/Sca1⁺/Flk2⁻) were purified 0, 1, 5, and 24 hours after IR in vivo (2 Gy). Alkaline comet assays were performed as described previously (26). Cells were stained with SYBR Safe (Invitrogen) for 1 hour and imaged using a Leica DM6000 microscope (20× objective). Analysis was performed using ImageJ software using the Open Comet plugin and the head finding was selected to the brightest region. Olive moment values were automatically generated by the software. Statistical significance was calculated using GraphPad Prism software utilizing the Mann–Whitney test.

Purified HSCs were obtained by cell sorting, exposed to ionizing radiation in vitro (2 Gy), and cultured for further 7 days in methylcellulose semi-solid medium containing cytokines (M3434). Colonies were dissociated and cultured with 100 ng/mL colcemid for 3 hours at 37°C to arrest cells in metaphase. Peptide nucleic acid staining was performed as described previously (27). Briefly, cells were exposed to hypotonic solution (0.56% KCl) for 16 minutes at 37°C and then fixed in methanol:acetic acid (3:1). After three changes of fixative solution, the cells were dropped on to slides that had been pretreated with 1 N HCl, followed by 100% ethanol and finally fixative solution. FISH with FITC-labeled (CCCTAA)₃ peptide nucleic acid (F1009-5, Panagene) was performed followed by mounting with ProLong Gold AntiFade Reagent containing DAPI (Invitrogen). Microscopy imaging was performed using a Leica DM6000 microscope (100× objective) and images were analyzed blindly using ImageJ software. An average of 20–30 metaphases were scored per condition.

All data shown are presented as mean ± SEM. When comparing datasets between Mybl2^+/+ and Mybl2^+/Δ animals, two-tailed unpaired Student t test was applied using GraphPad Prism software, unless indicated. No statistical method was used to estimate the sample size. No specific randomization or blinding protocol was used. N indicates the numbers of independent experiments performed and was chosen to ensure adequate statistical power. P ≤ 0.05 was considered statistically significant. Significance tests were performed on all samples and therefore graphs lacking P values indicate results were not statistically significant.

All animal experiments were performed under an animal project license in accordance with UK legislation. Human patients with MDS were recruited from the clinic held at the Centre for Clinical Haematology, University Hospital Birmingham NHS Foundation Trust. All the subjects have read the patient information sheet and signed the consent form. The study was conducted as according to Good Clinical Practice guidelines, consistent with the principles that have their origin in the Declaration of Helsinki. The study was approved by the West Midlands – Solihull Research Ethics Committee (10/H1206//58).

Given the role of MYBL2 in transcription and in maintaining genome stability (10–14) and recent studies showing increased DNA damage in MDS (28, 29), we decided to determine whether MYBL2 levels were associated with altered expression of DNA repair genes in patients with MDS. To do this, we first subdivided the patients into MYBL2lo and MYBL2hi populations based on a differential gene expression analysis on HSCs from patients with MDS (15), using the 25th and 75th percentiles of MYBL2 expression to define each group. Comparing these two groups, genes were then ranked on the basis of the signal-to-noise ratio for differential gene expression (DGE; Supplementary Table S1A). This DGE scoring was then analyzed against a collection of reactome gene sets (Supplementary Table S1B), to assess their individual enrichment, which encompassed pathways with which MYBL2 has previously been associated (cell cycle, DNA replication) and pathways relevant to this study (DNA damage, DNA repair, chromosome maintenance, and apoptosis). Consistent with previously published observations, the analysis (Supplementary Table S1C) confirmed the previously published association between MYBL2 levels and cell-cycle progression in the context of MDS (2). Remarkably, it also revealed a significant enrichment (NOM P = 0.00065, FWER P = 0.006) for the reactome gene set "DNA damage pathway" (R-HAS-73894), indicating an overall downregulation of DNA repair pathways components in the MYBL2lo MDS cases (Fig. 1A). Upon further inspection of the two populations, and confirming our previous results (2), we found that the MYBL2lo population was significantly enriched in MDS with excess blasts type II (MDS-EB2) cases with worse diagnosis (Fig. 1B, top table). To exclude the possibility that the observed link could primarily reflect an association between the DNA damage pathway and advanced disease states rather than MYBL2 levels, we recurated the compared subsets to display balanced diagnosis composition (Fig. 1B, bottom table), and repeated the gene expression analysis (Fig. 1C). This unbiased analysis confirmed that expression of DNA repair genes correlates with MYBL2 levels in MDS (Fig. 1C; Supplementary Table S2), independently of the disease state.

To determine whether the differential expression of DNA repair genes (Fig. 1) had a functional consequence, we next assessed the DNA repair kinetics of CD34⁺ cells, a multipotent stem and progenitor cell population, from 5 patients with MDS with differing prognoses (Table 1), without prior knowledge of MYBL2 expression levels. CD34⁺ cells were irradiated in vitro (2 Gy) and the prevalence of p53-binding protein 1 (53BP1) foci, a robust marker for DSBs, was measured at 1, 3, and 5 hours post-irradiation (Fig. 2A and B). Cells from these patients showed a striking difference in their ability to repair IR-induced DSBs, as evidence by their differing ability to resolve 53BP1 foci. After MYBL2 expression levels were measured (Fig. 2C) in the same patients, it was evident that DSB repair kinetics strongly correlated with MYBL2 mRNA expression (Fig. 2D; R² = 0.83, P = 0.0114). For example, a patient (patient 5) expressing similar MYBL2 levels to cells isolated from a healthy individual could efficiently repair DSBs within 5 hours, whereas in contrast, patients expressing around 50% normal levels of MYBL2 (patients 3 and 4), or even lower (patients 1 and 2) exhibited defective clearance of 53BP1 foci, suggesting the persistence of unresolved DSBs. We also investigated whether proliferation rates correlated with MYBL2 expression in MDS patient CD34⁺ cells by performing BrdU incorporation assays. This revealed that MYBL2 expression in these patients did not correlate with proliferation, (Supplementary Fig. S1A and S1B), nor did this correlate with the cell's DNA repair capacity (Fig. 2E). These data suggest that reduced expression of DNA damage genes in patients with compromised MYBL2 expression has a functional impact on the ability of these cells to repair genetic damage. Furthermore, we propose that MYBL2 mRNA expression levels may also be used as a potential biomarker predicting the cellular response to DNA damage, which could be of use for patient stratification.

Following the observation that MYBL2 expression levels correlated with DSB repair kinetics in human MDS stem cells, we wanted to further investigate the involvement of MYBL2 in regulating the DNA damage response. To do this, we used a Mybl2 haploinsufficient mouse model (Mybl2^+/Δ), known to be susceptible to MDS with ageing (30). We decided to focus our studies on the HSC population, the population in which MDS originates (1). Importantly, these animals did not show any major differences in the numbers of HSC/progenitor cells when compared with wild-type mice prior to treatment (Supplementary Fig. S2). To investigate the DNA damage response, we assessed the clearance of 53BP1 foci, as a measure of repair kinetics, in both wild-type and Mybl2^+/Δ HSCs [defined as (Lin⁻/cKit⁺/Sca1⁺/Flk2⁻)] followed by irradiation of the mice with 2 Gy of IR (Fig. 3A; Supplementary Fig. S3A). While we did not observe any difference in the initial recruitment of 53BP1 1 hour post-irradiation between wild-type and Mybl2^+/Δ HSCs, there were notable differences in the kinetics of 53BP1 foci resolution over time between the two genotypes (Fig. 3B and C). Wild-type HSCs showed a significant reduction in cells positive for 53BP1 foci at 5 hours post-irradiation, whereas in contrast, more than 50% of Mybl2^+/Δ HSCs still retained a significant number of 53BP1 foci–positive cells at this time point. Moreover, the absolute number of 53BP1 foci per cell was also increased in the Mybl2^+/Δ HSCs (Supplementary Fig. S3B and S3C). To extend these observations to a more refined HSC population, we repeated this experiment with HSCs purified from young animals (7 weeks) using SLAM staining (KSL CD48⁻CD150⁺). This revealed that the retention of 53BP1 foci was also apparent in Mybl2^+/Δ CD150⁺ HSCs, but not in their wild-type counterparts (Fig. 3D). Moreover, to determine whether this effect was dose dependent, we measured the percentage of 53BP1 foci 5 hours after 1 Gy of IR. This revealed that at a lower dose of IR, Mybl2^+/Δ HSCs exhibited a "wild-type" 53BP1 response (Supplementary Fig. S3D), suggesting that there is an insufficient quantity of DNA repair proteins present in the Mybl2^+/Δ HSCs required to cope with repairing high-level DNA damage. Because aged haploinsufficient Mybl2 mice develop an MDS-like disease, we investigated how aging affected DSB repair capacity of wild-type and Mybl2^+/Δ HSCs from young (7 weeks) and old mice (70 weeks). Unsurprisingly, 70-week-old HSCs exhibited a higher percentage of 53BP1-positive cells than 7-week-old HSCs. However, in keeping with a role for MYBL2 in regulating DNA repair, Mybl2 haploinsufficiency exacerbated the age-associated decrease in genome stability (Fig. 3B and C; Supplementary Fig. S3C). Interestingly, levels of 53BP1-positive cells in young Mybl2^+/Δ HSCs were comparable with aged wild-type HSCs, suggesting that these cells may demonstrate a premature aging phenotype. Together, these data indicate that Mybl2 haploinsufficiency is associated with defective repair of IR-induced DSBs in HSCs.

Because our data are only indicative of unrepaired DNA DSBs, we performed comet assays to directly measure the total amount of DNA damage remaining in these cells at different times post-irradiation. In keeping with a failure to properly repair IR-induced damage, Mybl2^+/Δ HSCs displayed an increase in the olive tail moment 5 hours post-irradiation when compared with wild-type HSCs (Fig. 3E). These differences in DNA repair kinetics were not the consequence of changes in cell-cycle profile between wild-type and Mybl2^+/Δ HSCs, as HSCs from both genotypes showed a similar percentage of cells in G₀–G₁ prior to irradiation (Supplementary Fig. S4), and the same percentage of cells in S-phase measured by in vivo BrdU incorporation (Supplementary Fig. S5A and S5B). Importantly, we did not observe any changes in the absolute numbers of HSCs after in vivo IR (Supplementary Fig. S5C), nor apoptosis, measured either by PARP1 cleavage (Supplementary Fig. S5D) or the induction of p53-dependent apoptotic genes (Supplementary Fig. S5E), which could account for our observations using the comet assay. Overall, these data demonstrate that Mybl2 haploinsufficient mice display a defect in the kinetics of DSB repair in response to IR, which is heightened during aging, but that has no impact on HSC survival.

To gain a mechanistic understanding of the DNA repair defect in Mybl2^+/Δ HSCs, we used small-molecule inhibitors to investigate the relationship of Mybl2 haploinsufficiency with two key proteins involved in the DSB response, namely DNA-dependent protein kinase (DNA-PK) and ATM. Treatment of wild-type HSCs with the DNA-PK inhibitor NU7441 (Fig. 4A; ref. 31) induced an increase in the percentage of cells positive for 53BP1 foci, in line with an expected defect in DSB repair due to inhibition of the NHEJ pathway (Fig. 4B–D). In contrast, Mybl2^+/Δ HSCs treated with the same inhibitor demonstrated a dramatic loss of 53BP1 foci formation at 5 hours after IR (Fig. 4B–D). These findings were recapitulated by analysis of the formation/retention of γH2AX foci, a pan-DNA damage marker (Fig. 4E–G). Nonetheless, 53BP1 foci were detected at 1 hour post-irradiation in Mybl2^+/Δ HSCs (Fig. 4B–D), suggesting that our observations did not reflect a global inability to form 53BP1 foci. Furthermore, Mybl2^+/Δ HSCs treated with DNA-PK inhibitor were proficient at sensing DNA damage, because MRE11 foci formed 1 and 5 hours after IR (Fig. 4H and I). Moreover, the absence of 53BP1 and γH2AX foci 5 hours post-irradiation was not because DNA repair had been completed, as Mybl2^+/Δ HSCs exhibited increased olive tail moments by comet assay (Fig. 4J). However, these breaks were eventually repaired, as by 24-hour levels of DNA damage in Mybl2^+/Δ HSCs was equal to that of wild-type cells (Fig. 4J). Interestingly, it has been previously shown that DNA-PK and ATM are both required for efficient H2AX phosphorylation and 53BP1 recruitment to DSBs and that ATM^−/− B cells are completely dependent on DNA-PK to sustain the phosphorylation of H2AX after gamma irradiation (32). On the basis of this, these data suggest that reliance that Mybl2^+/Δ HSCs have on DNA-PK to mediate DSB signaling is indicative of an underlying defect in the ATM-dependent DNA damage response.

While previously published data have been suggestive of a link between MYBL2 and ATM signaling (9), we wanted to specifically determine whether Mybl2^+/Δ HSCs exhibit defective ATM-dependent signaling in response to DSBs. To address this, we treated wild-type and Mybl2^+/Δ HSCs with the ATM inhibitor KU60019 (33), and analyzed 53BP1 foci clearance (Fig. 5A). These analyses revealed that treatment with KU60019 delayed the clearance of 53BP1 foci in wild-type cells, with ATM inhibition causing a >2-fold increase in the number of cells still displaying 53BP1 foci 5 hours post-irradiation (Fig. 5B–D), in line with the known requirement for ATM in DSB repair. In contrast, ATM inhibition in Mybl2^+/Δ HSCs had little effect on 53BP1 clearance after IR (Fig. 5C; Supplementary Fig. S6A). These data further support the prediction that the altered DSB repair kinetics observed in Mybl2^+/Δ HSCs are potentially due to a defect in ATM-dependent signaling.

Given these findings, it is tempting to speculate that because ATM inhibition in wild-type HSCs mimics the 53BP1 foci clearance defect observed in Mybl2^+/Δ HSCs, a similar treatment would also lead defective cell survival as seen in Mybl2^+/Δ cells. However, wild-type HSCs transiently treated with an ATM inhibitor did not display the same characteristics as untreated Mybl2^+/Δ HSCs when assessed by a colony-forming assay (Supplementary Fig. S6B), indicating that either short-term inhibition of ATM pathway does not have the same overall effect as Mybl2 haploinsufficiency or that Mybl2^+/Δ HSCs display additional defects that are not mimicked by ATM inhibitor.

To further investigate the defective ATM signaling in Mybl2^+/Δ HSCs, we next assessed the phosphorylation of KAP1. It has been previously reported that approximately 10%–15% of DSBs require ATM signaling to be repaired in G₀–G₁ phases of the cell cycle (34) and that this repair requires the phosphorylation of Kap1, which is known to be completely dependent on ATM (35). Therefore, we examined Kap1 phosphorylation after exposure to IR in HSCs (Fig. 6A). In line with previous reports, wild-type HSCs displayed robust pan-nuclear p-KAP1 staining within minutes following IR, which could be distinguished as discrete foci by 3 hours post-irradiation (Fig. 6B and C; ref. 35). Interestingly, although Mybl2^+/Δ HSCs also exhibited rapid KAP1 phosphorylation immediately post-irradiation, they were unable to maintain this phosphorylation at later time points (Fig. 6B and C). This further reinforced our findings that partial Mybl2 loss leads to defective ATM signaling, and also suggests that MYBL2 may be required to maintain rather than initiate ATM-dependent signaling during late-stage repair. To confirm this prediction, we next stimulated HSCs to enter the cell cycle, and analyzed activation and maintenance of the ATM-dependent G₂–M cell-cycle checkpoint after IR. Interestingly, Mybl2^+/Δ HSCs retained the ability to activate this checkpoint following exposure to IR (Supplementary Fig. S7A–S7C), suggesting that these cells are not completely defective in ATM function. This suggests that MYBL2 is required to maintain, but not initiate, activation of a specific subset of ATM-dependent signaling pathways.

One of the characteristics of loss of ATM function is telomere instability (36). We therefore postulated that lower MYBL2 levels in HSCs might lead to telomere fragility as a result of defective ATM signaling. To examine this possibility, HSCs from young and old mice were irradiated and cultured for 7 days in a colony-forming assay. Metaphase spreads were prepared from these cells and stained with telomere probes. These investigations revealed that telomere instability (defined as sister chromatid fusion or loss of telomere signal) was twice as frequent in the progeny derived from young Mybl2^+/Δ HSCs compared with controls (Fig. 6D). In fact, this percentage was similar between young Mybl2^+/Δ HSCs and old wild-type HSCs, in line with our earlier suggestion that Mybl2^+/Δ HSCs display an aging phenotype that could lead to neoplastic lesions.

In conclusion, we demonstrate that Mybl2^+/Δ HSCs are defective in the maintenance of ATM-dependent DNA damage signaling at the sites of DSBs, leading to slower DSB repair kinetics and a higher dependency on DNA-PK in the surviving cells. Overall, these data suggest that correct MYBL2 protein levels are required for a proper DNA damage response and appropriate DSB repair in the HSC compartment. Deregulation of these levels leads to defective DSB repair, telomere instability, and likely contributes to the accumulation of genetic alterations in MDS.

HSCs are the life-long pillars of continuous blood cell production. Maintenance of their genetic integrity is paramount to avert the accumulation of mutations that can contribute to the development of blood disorders such as MDS during the aging process. Our work demonstrates a previously undescribed role for MYBL2 in promoting efficient DSB repair in HSCs, possibly via regulation of the ATM kinase. Furthermore, our findings suggest that MYBL2 levels may be used as a biological biomarker to determine the DSB repair capacity of CD34⁺ cells from patients with MDS. Furthermore, MYBL2 levels could also be used as a clinical biomarker to inform decisions regarding patient selection for transplantation or treatments that target DNA repair, highlighting the translational importance of this work.

In line with a role for MYBL2 in regulating ATM signaling, we have shown that Mybl2 haploinsufficient HSCs display a delay in 53BP1 clearance after DNA damage induced by IR, which is exacerbated during ageing. In these cells, defective ATM signaling leads to loss of sustained KAP1 phosphorylation and telomere instability. This renders these cells reliant on other DNA repair pathways prevalent in noncycling cells. As a result, Mybl2^+/Δ HSCs are highly dependent on the NHEJ regulator DNA-PK for DSB signaling, because inhibition of this kinase leads to a failure to maintain γH2AX and 53BP1 at sites of damage. Moreover, because ATM-dependent phosphorylation of KAP1 has been suggested to be required for chromatin relaxation and the repair of DSBs within heterochromatin regions (35, 37), our data suggests a requirement for MYBL2 in repairing a subset of DSBs associated with heterochromatic chromosomal regions.

In agreement with our conclusions, studies using human cell lines have recently demonstrated that MYBL2 interacts with Nbs1, which is required for the activation of ATM in response to DSBs and also ATM-dependent heterochromatic DSB repair in G₀–G₁ (9). While on face value, this may explain our observations Mybl2^+/Δ HSCs, in stark contrast to the work of Henrich and colleagues, we were unable to detect a G₂–M checkpoint defect in our Mybl2-haploinsufficient HSCs, suggesting that at least some ATM-dependent signaling is intact in these cells. Moreover, Henrich and colleagues failed to observe any defects in DNA repair, leading them to conclude that MYBL2 does not have an essential role in the DNA repair response. In contrast, we have shown that MYBL2 does have a role for DSB repair in HSCs, and without sufficient MYBL2 expression cells show defective repair kinetics of IR-induced DSBs. These differences could be due to cell type differences, such as primary cells versus cell lines, or the use of different DNA-damaging agents to induce DSBs (IR vs. UV). Alternatively, the highly quiescent nature of HSCs in vivo may also account for these discrepancies, as this may make any defects in fast DSB repair by NHEJ more pronounced as they cannot utilize repair by HR.

The importance of an appropriate DNA damage response for the maintenance and protection of the HSC pool against functional decline during ageing has been well reported (38–41). Quiescent HSCs cannot use the HR pathway and thus rely on NHEJ-dependent mechanisms to repair their DNA (16, 17). A failure to repair DSBs by the canonical DNA repair pathways can be detrimental to the cell, as alternative pathways may allow the potential for genome instability (42, 43). It has also been reported that reduction in or mutation of splicing factors in MDS leads to altered splicing of DNA repair and telomere maintenance genes (44). This then perturbs myeloid differentiation and contributes to disease development via a mechanism not necessarily involving chromosomal rearrangements. Equally, abnormal DNA damage signaling in Mybl2^+/Δ HSCs could increase the mutational burden by facilitating the use of alternative, more error-prone signaling pathways. Indeed, after inducing irradiated Mybl2^+/Δ HSCs to proliferate, we found evidence of telomere instability in their progeny. In vivo, this prospect is likely to have a severe impact, and may ultimately lead to HSC malfunction and the accumulation of cells that are primed for the development of blood disorders such as MDS. Recent work in the field of MDS suggests that telomere dysfunction is a potent driver of the disease phenotype (44) and telomere elongation using danazol treatment has been shown to improve hematologic responses including reducing transfusion dependency (45).

Importantly, we show that the association between low MYBL2 levels and impaired DNA repair also holds true in patients with MDS. Thus, by directly measuring the DNA repair kinetics in CD34⁺ MDS cells, our data shows a correlation between MYBL2 levels and functional DSB repair. Moreover, in CD34⁺ cells from these patients, low MYBL2 levels largely associate with low expression of DNA repair genes. Together, these data provide a molecular rationale for the accumulation of genetic anomalies in patients deficient for MYBL2, which could play a role in the progression of their disease. Importantly, this data represents the first direct study of repair kinetics in MDS patient CD34⁺ cells. Current treatment options for patients with MDS are mostly based on cytotoxic agents prior to autologous transplant, and are challenged by the occurrence of clonogenic relapse, increased resistance to therapy and, in some cases, leukemic transformation. Relapse of disease in these patients is typically driven by additional genetic lesions. Of note, delq20 clones are not uncommon in patients following cytotoxic chemotherapy, with more than 20% of these patients having a therapy-related myeloid neoplasm, conferring a high mortality (46). Patients with therapy-related myeloid neoplasms more frequently have clonal hematopoiesis with the originating mutation being present prior to chemotherapy treatment (47). We therefore hypothesize that these clones are susceptible to DNA-damaging agents and that DNA repair defects may be involved in the etiology of their subsequent myeloid neoplasm. Moreover, our work supports the notion that Mybl2 haploinsufficiency results in changes in DNA repair kinetics and defective ATM signaling. Both ATM signaling and NHEJ are known to be activated to repair DSBs induced by doxorubicin (48, 49), indicating that our findings with IR are also likely to be applicable to the use of anthracyclines in chemotherapy.

In light of our findings, we propose that compromising the MYBL2-dependent DNA damage response in HSCs can facilitate MDS development by allowing inefficient repair of physiologic DSBs, and promoting telomere instability, two processes known to contribute to the generation of oncogenic transformation in the surviving HSC population. Furthermore, we suggest that sustained low levels of the MYBL2 protein in the premalignant cells may confer a susceptibility to disease progression through the accumulation of incorrectly repaired DNA lesions.

R. Almaghrabi reports receiving a commercial research grant from The Saudi Embassy. M. Raghavan has received speakers’ bureau honoraria from Celgene Ltd and Pfizer Ltd. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. Bayley, M.R. Higgs, E. Petermann, P. García

Development of methodology: R. Bayley, D. Blakemore, L. Cancian, P. Garcia

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Cancian, G. Volpe, N. Reeve, M. Raghavan, P. García

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Bayley, D. Blakemore, L. Cancian, S. Dumon, G. Volpe, C. Ward, P. García

Writing, review, and/or revision of the manuscript: R. Bayley, L. Cancian, S. Dumon, G. Volpe, C. Ward, M. Raghavan, M.R. Higgs, GS Stewart, E. Petermann, P. García

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Bayley, P. García

Study supervision: P. García

Others (contributed to the experiments): R. Almaghrabi

Others (technical assistance to perform the experiments): J. Gujar

Others (advised with experiments and contributed to writing/revision of the manuscript): M.R. Higgs

The authors wish to thank the members of the Petermann, Stewart, Higgs, García, and Frampton laboratories for advice and constructive criticisms, and Professor Frampton for covering the experiments through his animal license. The authors also wish to thank the animal facility and cell sorter facility at the University of Birmingham, and Donna Walsh for collecting blood samples from myelodysplastic syndrome patients at the Centre for Clinical Haematology (Queen Elizabeth Hospital, Birmingham). R. Bayley and L. Cancian were supported by MRC grant (MR/K01076X/1). D. Blakemore was supported by a MRC PhD studentship (1632704). This work was funded by an MRC New Investigator Research Grant (MR/K01076X/1) and MRC Proximity to Discovery: Industry Engagement fund (MC_PC_14123; to P. García).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.