BRCA1 Promoter Methylation Is Linked to Defective Homologous Recombination Repair and Elevated miR-155 to Disrupt Myeloid Differentiation in Myeloid Malignancies

Myeloid malignancies are a heterogeneous collection of clonal hematologic disorders ranging from the myeloproliferative neoplasms (MPN) to more aggressive conditions including myelodysplastic syndrome (MDS), overlap mixed MDS/MPNs and acute myeloid leukemias (AML; ref. 1). The majority of these conditions affect individuals above 60 years, hampering the use of potential curative treatments such as intensive chemotherapy and bone marrow transplantation (2). With the exception of BCR-ABL chronic myelogenous leukemia (CML), many of these disorders continue to be diagnosed by morphologic features and quantitative differences in myeloid components, despite significant progress in identifying key molecular events during disease pathogenesis. This highlights the urgent need to elucidate novel disease pathways amenable to a new generation of targeted agents, without which patients continue to be at significant risk of bone marrow failure and leukemic transformation (3). Given the common morphologic, cytogenetic, and genetic abnormalities, we hypothesized that distinct pathways are similarly disrupted across the spectrum of myeloid neoplasms (4, 5). If identified, therapies designed to target these shared functional defects could be selected for these diverse diseases.

In myeloid disorders, unbalanced chromosome aberrations are reported in a significant proportion of patients and increase in frequency with disease progression, suggesting a role for defects in the repair of double-strand DNA breaks (DSB) carried out by the nonhomologous end joining (NHEJ) and homologous recombination (HR) pathways (4, 6–8). Increased error-prone NHEJ activity has been demonstrated in myeloid leukemias, which could be a compensatory mechanism for defective HR (9, 10). However, direct analysis of the high fidelity, error-free HR pathway in myeloid malignancies is not well characterized, unlike in breast and ovarian cancers (11, 12). This lack of knowledge has therapeutic significance, given the interest in HR repair defects with the approval of olaparib, a PARP inhibitor (PARPi), for BRCA1/2-mutated ovarian cancers (13). Because PARPi is exquisitely toxic to HR defective malignant cells, a comprehensive study of HR repair and the associated molecular aberrations in myeloid malignancies could determine whether patients are potential candidates for this therapy (11, 14). In a recent phase I clinical trial in patients with multiple myeloid neoplasms, we reported increased sensitivity to topotecan, carboplatin with the PARP inhibitor veliparib, and showed reduced colony formation in primary mononuclear cells exposed to PARPi (15). Another phase I study in advanced myeloid malignancies found that a combination of veliparib and temozolomide was well tolerated and exhibited clinical activity (16). Although encouraging, the limited assessment of BRCA1 promoter methylation and the inability to identify genetic biomarkers underlying defective HR necessitates a detailed examination into HR repair and its underlying molecular mechanisms in myeloid malignancies. Finally, functional characterization of the association between DNA repair gene epigenetic silencing and HR defects can inform biomarker discovery and the development of targeted therapeutics.

In this study, we extensively studied an independent set of myeloid neoplasm patients, including some with sequential sampling, to discern the prevalence of HR defects using functional assays and then examined whether epigenetic changes could underlie HR deficiency. In the panel of HR repair genes that was examined, we observed increased incidence of patient samples with BRCA1 promoter methylation, leading us to investigate novel mechanisms of BRCA1 silencing in leukemogenesis. This led us to uncover an HR-independent mechanism in which epigenetic silencing of BRCA1 may drive myeloid disease progression.

Peripheral blood or bone marrow aspirates from patients with myeloid neoplasms were collected under IRB approved protocols at the Johns Hopkins Hospital (Baltimore, MD). All patients provided informed consent in accordance with the Declaration of Helsinki. Patient characteristics are provided in Table 1 and Supplementary Table S3. Mononuclear leukocytes were isolated by density centrifugation with Ficoll–Paque Plus (GE Healthcare). The ZR-Duet DNA/RNA MiniPrep Kit (Zymo Research) was used to obtain genomic DNA and RNA. Samples from earlier studies on myeloid and lymphocytic neoplasms were used for the initial screening of HR gene methylation (17, 18).

HR status was evaluated as described by Patel and colleagues with modifications (19). Primary mononuclear cells, resuspended in RPMI media with 10% (v/v) FBS, were radiated with 10Gy using a Gammacell 1000A ¹³⁷Cs source (Atomic Energy of Canada) at a rate of 2.3 Gy/minute and placed in 37°C, 5% CO₂ in humidified atmosphere for 6 hours. The cells were washed with PBS and fixed with 2% formaldehyde. 250,000 cells were cytospun onto poly-lysine–coated slides, permeabilized with 0.25% (v/v) Triton X-100 in PBS and incubated in blocking buffer [PBS with 1% (v/v) glycerol, 0.1% (w/v) fish skin gelatin, 0.1% (w/v) BSA, 5% (v/v) goat serum and 0.4% (w/v) sodium azide] for 1 hour. Slides were incubated overnight at 4°C with RAD51 rabbit polyclonal (ActiveMotif) and phospho-Ser139 H2AX mouse monoclonal (Millipore) antibodies diluted 1:500 in blocking buffer. Cells were then incubated for 1 hour with secondary Alexa Fluor 488 goat anti-mouse and Alexa Fluor 555 goat anti-rabbit antibodies (Invitrogen) diluted 1:1,000. Slides were washed, counterstained with 1 μg/mL Hoechst 33258 for 5 minutes and mounted with Prolong Gold antifade reagent (Life Technologies).

Confocal images were obtained with a Nikon C1si confocal laser-scanning microscope with a 100X.1.4 N.A. oil-immersion objective at excitation wavelengths of 408, 488, and 561 nm. Maximum projection images were generated from optical sections of 0.5 μm increments and processed in ImageJ using the PZFociEZ macro. All image analysis parameters were kept constant for each pair of mock and irradiated sample. As with previous studies, cells positive for RAD51 and phospho-H2AX foci are defined as having ≥5 foci per nucleus (20–22). HR-competent samples are defined as having greater than two-fold increase in percentage of RAD51-foci positive cells after radiation (20–22). More than 100 cells from at least three fields were analyzed for each sample.

MSP reactions were performed as described previously with bisulfite-treated genomic DNA treated with the EZ DNA Methylation Kit (Zymo Research; ref. 23). qMSP for BRCA1 promoter methylation, normalized to β-actin levels, was carried out using the iTaq SYBR Green. qMSP and MSP primer information are listed in Supplementary Table S1. BRCA1 qMSP primers are adapted from Esteller and colleagues and cover −175 to +9 relative to the TSS (24). The controls for unmethylated and methylated templates were bisulfite-treated normal lymphocyte DNA and CpG Methylated Jurkat genomic DNA (NEB), respectively. Methylated samples are defined as amplicons with melting temperatures matching that of the methylated control.

For mRNA, cDNA was synthesized using the iScript Reverse Transcription Supermix (Bio-Rad) and transcript levels were measured using the SsoAdvanced SYBR Green mix (Bio-Rad). Primer sequences are listed in Supplementary Table S1. BRCA1 expression was quantified using the 2^−ΔΔC_t method after normalizing to GAPDH. To detect expression of miRNAs, RNA was reverse-transcribed using the qScript microRNA cDNA Synthesis Kit (Quanta BioSciences) and real-time quantitative analysis was performed according to the manufacturer's instructions. miR-155 levels were normalized to RNU6B and quantified using the 2^−ΔΔC_t method.

OCI-AML3 cells were purchased from The German Collection of Microorganisms and Cell Culture (DSMZ, cat #582) and grown in RPMI media supplemented with 20% FBS, 100 U/mL penicillin and 100 μg/mL streptomycin. HEK293 cells were cultured in DMEM with 10% FBS. All cells were maintained at 37°C, 5% CO₂ in humidified atmosphere. For HDACi treatment, cells were treated with suberoylanilide hydroxamic acid (SAHA) for 72 hours.

HEK293 cells were transfected with BRCA1-specific (sh34: NM_007294.2-3465s21c1, sh49: NM_007294.2-498s21c1) or nontargeting shRNA (Sigma SHC016), together with psPAX2 and pMD2.G plasmids. Concentrated lentivirus stock and 8 μg/mL polybrene were added to cells and spun at 2,400 rpm for 60min. After 48 hours, transduced cells were selected with 2 μg/mL puromycin and repression confirmed by qRT-PCR.

OCI-AML3 cells were treated with PARP inhibitor ABT-888 (ApexBio) at 0.01% final DMSO concentration. Cell viability was determined using the RealTime-Glo MT assay (Promega).

All statistical analyses were performed using Graphpad Prism 6.

In response to DNA DSBs, the histone variant H2AX is phosphorylated at the site of damage and recruits components of the HR pathway to generate single-stranded DNA ends that are in turn bound by RAD51, an established marker of HR repair, to initiate homology search and repair (20–22, 25). This phenomenon has been utilized extensively as a tool to characterize HR repair in a variety of malignancies (20–22, 25). However, previously reported studies of the HR pathway in myeloid malignancies require extended treatment (24 hours) with DNA-damaging agents, limiting investigation to a small number of transformed cases due to the stress of prolonged ex vivo culture, while failing to take into account the effects of drug uptake and metabolism (26, 27).

To minimize the confounding factors linked with lengthy drug incubations, we exposed primary cells to ionizing radiation, which has been shown to acutely induce similar levels of DSBs (19). To ensure ideal viability of primary cells, we performed this assay only on samples that can be processed within 6 hours of collection. Previously, we have characterized impaired RAD51 foci induction in a small cohort of MPN samples that was associated with PARP inhibitor sensitivity (28). Here, we conducted a comprehensive investigation of peripheral blood or bone marrow mononuclear cells from myeloid neoplasm patients (14 MDS/MPNs, 5 MPNs, 1 sAML, 1 therapy-related MDS; Supplementary Table S3). For each pair of mock and radiated samples, we performed confocal microscopy for γH2AX and RAD51 foci as markers of DSB and HR respectively 6 hours postradiation (Fig. 1A; Supplementary Fig. S1). We measured the fold change in the percentage of γH2AX- and RAD51-foci positive cells (≥5 foci) in mock and radiated fractions (Fig. 1B). All irradiated samples demonstrated a greater than 2-fold increase in percentage of γH2AX-foci–positive cells, confirming that irradiation generated reproducible and robust DNA DSBs. HR-defective cells are defined as samples that failed to achieve a 2-fold induction of RAD51-positve cells, a criterion used in previous studies that correlates with HR defects and PARP inhibitor sensitivity (21, 22). This approach reduces subjectivity and variation during sample and image processing. We observed compromised RAD51 foci induction in 9 of 21 (43%) samples indicating a high prevalence of impaired HR (Fig. 1C).

Given previous reports of aberrant gross cytogenetic abnormalities and increased NHEJ activity that suggest deficient HR repair in myeloid neoplasms, it is of interest to investigate the molecular alterations responsible for these defects (4, 6–8). Inactivation of tumor suppressor genes, such as those involved in HR repair, commonly involves genetic alterations, loss of chromosome regions, and abnormal promoter hypermethylation (29). However, mutations in HR genes have been reported infrequently and have not been shown to be associated with functional disruption of HR in myeloid malignancies (27, 30). Using a previous cohort of 144 MPN samples with SNP array data, we found no significant association for genes involved in the HR pathway (GO:0000724 double-strand break repair via homologous recombination, P = 0.121; refs. 28, 31). Although this analysis does not exclude all genetic changes, it eliminates recurrent genomic deletions involving HR genes as the cause of HR deficiency. Because hematologic disorders frequently display alterations in CpG methylation, we explored whether HR defects can result from aberrant epigenetic regulation (29, 32–34). In treatment related AML, a high incidence of BRCA1 promoter methylation was reported, whereas healthy bone marrow and peripheral blood samples were unmethylated, reflecting a molecular change that has been argued to predict sensitivity to PARP inhibitors in breast cancers (35, 36). As a first approach to determine whether HR genes might be altered by promoter methylation, we examined a large series of previously collected MPN, AML, and acute lymphoid leukemia (ALL) samples (17, 18). As reported in Supplementary Table S2, we observed CpG island promoter methylation of BRCA1 (MPN: 6/66, 9.09%; AML: 2/26, 7.69%), FANC-C (AML: 1/30, 3.33%), and FANC-L (AML: 1/30, 3.33%). These epigenetic events were absent in 11 B-cell, nine T-cell ALL samples (Supplementary Table S2) and normal lymphocytes examined. We did not detect promoter methylation in other HR genes examined (BRCA2, ATM, ATR, and FANC-A) (Supplementary Table S2). Given previous data linking BRCA1 promoter hypermethylation to impaired ionizing radiation-induced HR response, we next examined whether HR defects in myeloid neoplasia are associated with this epigenetic change.

We examined 96 samples from 51 unique myeloid neoplasm patients and with a focused methylation analysis of BRCA1, FANC-C, and FANC-L based upon our initial screen (Fig. 2). None of the patient samples had promoter methylation of FANC-C and FANC-L, suggesting that epigenetic silencing of these genes is rare in myeloid neoplasms and could not account for the frequency of HR deficiency (data not shown). We further refined promoter methylation detection for BRCA1, using a quantitative MSP (qMSP) assay with melt curve analysis that compares the melting temperature of each amplicon to that of the in vitro methylated positive control to exclude amplification of unconverted DNA. With this approach, we detected BRCA1 methylation in 22.9% (22/96) of the samples, corresponding to 12 of 51 (23.5%) patients (Fig. 2A). Among patients with sequential samples, promoter methylation was unchanged over time in most patients, with persistent absence of methylation in 14 patients and retained methylation in four cases (Pt. 10, 12, 28, 33). There was loss of BRCA1 methylation in subsequent samples for 2 patients (Pt. 19, 23), while in two cases (Pt. 1, 17), we observed acquisition of BRCA1 methylation in samples collected at later time points (Fig. 2A). A review of the patient history revealed that patient 23 had received 5′azacytidine, a demethylating agent, after the first sample was collected, potentially explaining the loss of methylation over time.

To determine whether BRCA1 promoter methylation was associated with transcriptional repression, we examined BRCA1 expression in 88 myeloid neoplasm patient samples with available RNA. We found lower expression in cases with BRCA1 promoter hypermethylation compared with those samples lacking methylation (P < 0.05, unpaired t test; Fig. 2B), although some samples with reduced expression did not have detectable methylation, which is suggestive of other mechanisms of transcriptional repression. To determine whether there was an association between HR repair capability and this epigenetic alteration, we compared samples with known HR status to BRCA1 methylation, finding a statistically significant association between BRCA1 methylation and HR defect (P < 0.05, Fisher exact test; Table 2). Finally, to delineate the interplay between promoter methylation, gene expression and HR repair, we classified BRCA1 expression according to HR and promoter methylation status (Fig. 2C). HR-defective samples with BRCA1 CpG island methylation exhibited significantly lower BRCA1 gene expression compared with HR-competent cases with unmethylated BRCA1 (P < 0.05, unpaired t test). Of note, all four BRCA1 methylated samples were defective for HR, strongly suggesting that this epigenetic silencing event disrupted HR repair (Fig. 2C). The varying levels of BRCA1 expression in unmethylated BRCA1 samples defective for HR points to dysregulation in other HR genes or BRCA1 repression through alternative mechanisms (chromatin repression) as the cause of HR defect. In Pt. 17, the acquisition of hypermethylation of BRCA1 promoter at the later time point was associated with disruption of HR repair, while the initial sample from this patient was unmethylated and had intact HR repair (Supplementary Fig. S2). The combined data suggest that promoter methylation and silencing of BRCA1 was associated with HR defects, showing for the first time the consequence of BRCA1 methylation and associated gene repression, on disrupting HR repair in myeloid malignancies.

To determine whether BRCA1 inactivation with HR defects can be used as a target for therapy, we examined in vitro sensitivity to the PARP inhibitor ABT-888 (veliparib). BRCA1 inactivation is associated with increased sensitivity to PARP inhibition in breast and ovarian cancer due to synthetic lethality in HR-defective cancer cells (11). We stably expressed two independent short hairpin RNAs targeting BRCA1 in OCI-AML3 cells previously characterized as HR-competent and expressing BRCA1 with absence of DNA methylation (27). BRCA1-repressed clones exhibited impaired RAD51 foci induction, recapitulating the association observed in primary samples (Fig. 3A and B). Cell viability after 72-hour exposure to ABT-888 was significantly decreased with BRCA1 knockdown (IC₅₀ values scrambled control: 16.2 μmol/L; BRCA1 sh34: 7.5 μmol/L; BRCA1 sh49: 5.72 μmol/L; P < 0.0001, F test), indicative of increased sensitivity to PARP inhibition (Fig. 3C). This 2- to 3-fold reduction in IC₅₀ is relevant for clinical activity and is predicted to increase exposure to therapeutic levels, because the recommended dose of olaparib (300 mg) generates C_max plasma levels of 16.6 μmol/L (NCT01894256). These results demonstrate that loss of BRCA1 expression impairs HR and leads to increased sensitivity to PARP inhibition in myeloid leukemia cells.

Although our studies suggested an important role for BRCA1 inactivation leading to HR, the lack of inactivation of other HR genes was notable. We hypothesized that there may be additional consequences of BRCA1 inactivation in myeloid neoplastic cells that confer transformed cells a growth advantage in addition to creating HR deficiency. In breast cancer, BRCA1 directly represses miR-155 by recruiting a repressive complex that includes histone deacetylase 2 (HDAC2) to the miR-155 promoter (37). This would suggest that loss of BRCA1 could be associated with increased miR-155 levels, which is of relevance for myeloid malignancies because previous studies have identified miR-155 as an oncomiR frequently overexpressed that promotes myeloid lineage expansion of hematopoietic stem cells (38, 39). Because an association between miR-155 and BRCA1 loss has not been reported in myeloid malignancies, we examined miR-155 expression in our study cohort. We found a significant inverse correlation between miR-155 and BRCA1 expression (Spearman correlation coefficient r: −0.245, P < 0.05; Fig. 4A). Although this imperfect correlation suggests additional factors may regulate miR-155, we tested whether BRCA1 expression was responsible for miR-155 repression. Indeed, when we examined the effect of BRCA1-repression in OCI-AML3 cells, we found a >1.5-fold increase in miR-155 expression (Fig. 4B). To confirm that this was epigenetically mediated, we treated OCI-AML3 cells with SAHA, a pan-HDAC inhibitor. This produced a dose-dependent increase in miR-155 levels, which at highest dose exceeded that seen by BRCA1 knockdown, implicating HDAC mediated repression as the mechanism by which BRCA1 regulates miR-155 expression (Fig. 4C).

As an oncogenic microRNA, multiple miR-155 targets have been proposed, including PU.1 and SHIP1 (40–44). Both proteins have well-established roles in myeloid differentiation and have been reported to be frequently downregulated in myeloid malignancies (42, 45, 46). We therefore examined whether increased miR-155 expression was associated with altered expression of these target genes. Indeed, we found a strong inverse correlation of both PU.1 and SHIP1 transcripts (Fig. 4D and E) with miR-155 expression (PU.1: Spearman correlation r: −0.264, P < 0.05; SHIP1: Spearman correlation r: −0.364, P < 0.005). This suggests a mechanism by which the loss of BRCA1 drives myeloid transformation via miR-155 upregulation, that in turn downregulates key mediators of myeloid differentiation including PU.1 and SHIP1.

A major influence underlying the 2016 update to the WHO classification of myeloid malignancies was the need of an integrated morphologic and molecular approach to diagnosis (47). This is because myeloid malignancies often present with significant clinical heterogeneity, presenting a challenge for diagnosis and identification of effective therapies (48). Recent clinical data in advanced myeloid neoplasms suggest clinical activity when PARP inhibitors are incorporated with existing chemotherapies, which motivated us to study the fidelity of the HR pathway given the favorable activity of PARP inhibitors in tumors with HR deficiencies (15, 16). PARP inhibitors represent promising therapies for patients with myeloid malignancies, given their synthetic lethal killing of HR-defective neoplastic cells while sparing normal tissues. The identification of cancers with impaired HR repair may therefore allow selection for favorable responders. Unfortunately, previous assays for HR defects in hematological samples required long-term ex vivo drug treatment of patient mononuclear cells that present significant challenges for implementation into clinical practice. In this study, we investigated acute leukemia and a spectrum of myeloid malignancies (MPN, MDS, mixed MDS/MPN) to determine the prevalence of HR functional defects. To address the lack of a short-term HR assay for hematological samples, we developed an ionizing radiation-based assay and determined HR repair function by measuring RAD51 foci recruitment to DSBs. This modification has allowed us to comprehensively and objectively query HR competency, and conclude that a high proportion (43%) of the myeloid malignancies we examined exhibited HR defects.

Increasingly, it is acknowledged that HR repair defects can arise in sporadic cancers from molecular aberrations aside from germline mutations in BRCA1/2 (20, 49). To identify the molecular underpinning of this HR defect, we considered the various mechanisms of disrupting HR genes, including mutation, loss of heterozygosity, and promoter methylation. Multiple genetic studies have described germline mutations in HR repair genes in hematologic neoplasms (ATM, BLM, Fanconi anemia), but these events are infrequent in sporadic myeloid disorders (27, 30, 50). Our analysis of genomic regions lost in MPN also failed to highlight a HR pathway that was significantly disrupted, similar to a previous report (15). Although BRCA1 promoter methylation was previously identified in treatment related AML patients that correlated with complex karyotypes, HR capacity was not evaluated (36). Here, we describe BRCA1 CpG island methylation in a significant proportion of patient samples and link BRCA1 promoter methylation with gene repression and HR defects in primary samples. This epigenetic change could be a promising biomarker for HR defective myeloid neoplasm with increased sensitivity to PARP inhibition. Because it may not be feasible to obtain patient samples in the timely manner required to perform this HR assay and gene expression analyses, the ability to detect stable epigenetic changes in a simple and sensitive manner favors promoter hypermethylation as a clinical biomarker. Although the sample size is inadequate to draw a definitive conclusion, we have presented sufficient data to provide the rationale for more extensive longitudinal studies with larger sample sets focusing on the HR repair pathway in myeloid malignancies.

By establishing the link between HR functionality and BRCA1 methylation, we postulate that DNA methylation assays could be developed into valuable clinical companion tools for decision-making regarding the use of PARP inhibitors. The recent approval of olaparib for BRCA1/2 mutated ovarian cancer based on improved progression free survival illustrates the clinical benefit of this agent for tumors that lack HR, a strategy that is actively pursued in multiple malignancies (13, 51, 52). The ABT-888 IC₅₀ values of BRCA1-repressed cells reported in this study are below the single-dose C_max readings recorded in preclinical animal models and human studies, highlighting a therapeutic window to apply this agent in this population (53). Our results build upon the published findings from two clinical trials at the Johns Hopkins Hospital to evaluate the effects of adding veliparib in combination with chemotherapeutic agents for hematologic disorders (15, 16). Nieborowska-Skorska and colleagues have recently also observed significant variations of PARP inhibitor sensitivity in leukemic cell lines and patient specimens (54). In the same study, gene expression profiling showed that leukemias expressing the BCR-ABL1 and AML1-ETO oncogenes downregulate genes in the BRCA1 pathway. Others have showed that treatment of MPN cells with the JAK2 inhibitor ruxolitinib reduced expression levels of HR and D-NHEJ genes, particularly BRCA1 and RAD51. Follow-on studies of ruxolitinib, hydroxyurea (ribonucleotide reductase inhibitor), and PARPi showed synergistic cell killing in selected sensitive cells in both in vitro and in vivo models (55).

Despite the numerous genes involved in HR, the prevalence of BRCA1 inactivation in myeloid neoplasms suggests the potential for additional roles in which loss of BRCA1 drives disease progression besides DSB repair. The recent finding that BRCA1 represses oncogenic miR-155 provides a possible role, given that miR-155 is frequently upregulated in myeloid malignancies and ectopic overexpression of miR-155 in mouse models results in a myeloproliferative-like phenotype (37–41). The inverse correlation of BRCA1 and miR-155 transcripts in patient samples, increased miR-155 levels in the BRCA1-silenced cells, and observation that this is repression mediated by HDACs are suggestive of a transcriptional repressor role for BRCA1 on miR-155 (37). Given the prominent role of miR-155 in normal and pathogenic hematological systems, and the direct targeting of PU.1 and SHIP1 by this oncogenic miRNA, our results suggest a HR-independent mechanism by which BRCA1 loss contributes to myeloid malignancies. While the data presented is correlative, this mechanistic role of BRCA1 warrants greater investigation.

In conclusion, we describe a subset of MPNs that exhibit HR defects and are susceptible to the synthetic lethal killing effects of PARP inhibitors. Our observation that BRCA1 promoter methylation associates with HR defect in patient samples provides a plausible explanation for epigenetic aberrations driving this phenotype, and the lack of genetic alterations and chromosome aberrations previously examined as the cause of HR defects. This study supports the clinical relevance of further explorations of the PARPi pathway in myeloid diseases, and the critical need for biomarkers of sensitivity and understanding of BRCA1 biology.

K.W. Pratz is a consultant/advisory board member for Abbvie, Astellas, and Boston Biomedical. M.A. McDevitt is an employee of Abbvie. No potential conflicts of interest were disclosed by the other authors.

Conception and design: W. Poh, K.W. Pratz, M.A. McDevitt, J.G. Herman

Development of methodology: W. Poh, K.W. Pratz, J.G. Herman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Poh, R.L. Dilley, A.R. Moliterno, K.W. Pratz, M.A. McDevitt

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Poh, A.R. Moliterno, K.W. Pratz, M.A. McDevitt, J.G. Herman

Writing, review, and/or revision of the manuscript: W. Poh, R.L. Dilley, A.R. Moliterno, J.P. Maciejewski, K.W. Pratz, M.A. McDevitt, J.G. Herman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Poh

Study supervision: M.A. McDevitt, J.G. Herman

We thank James Eshleman and Fred Bunz for insightful discussion, and Lillian Dasko-Vincent for assistance with confocal imaging.

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.