Rare germline alterations of myeloperoxidase predispose to myeloid neoplasms

Abstract

Myeloperoxidase (MPO) gene alterations with variable clinical penetrance have been found in hereditary MPO deficiency, but their leukemia association in patients and carriers has not been established. Germline MPO alterations were found to be significantly enriched in myeloid neoplasms: 28 pathogenic/likely pathogenic variants were identified in 100 patients. The most common alterations were c.2031-2 A > C, R569W, M519fs* and Y173C accounting for about half of the cases. While functional experiments showed that the marrow stem cell pool of Mpo^−/− mice was not increased, using competitive repopulation demonstrated that Mpo^−/− grafts gained growth advantage over MPO wild type cells. This finding also correlated with increased clonogenic potential after serial replating in the setting of H₂O₂-induced oxidative stress. Furthermore, we demonstrated that H₂O₂-induced DNA damage and activation of error-prone DNA repair may result in secondary genetic damage potentially predisposing to leukemia leukemic evolution. In conclusion, our study for the first time demonstrates that germline MPO variants may constitute risk alleles for MN evolution.

INTRODUCTION

Identification of risk alleles for development of myeloid neoplasia (MN) is an important objective of hematology research. To this end, various strategies have been deployed, including analysis of informative kindreds, unbiased genome association studies, and targeted screens. Previous results indicate different possible pathogenetic mechanisms: i) direct dominant or recessive traits presenting mostly in childhood; ii) heterozygous variants of dominant traits serving as risk alleles with lower penetrance and longer disease anticipation; iii) germline mutations responsible for bone marrow failure (BMF) or other marrow conditions serving as indirect risk alleles for later leukemia (e.g., ELANE in severe neutropenia) [1]. Some of the germline variants exhibit several of these features, making difficult the recognition of the phenotype/genotype association through a direct pathogenic mechanism. For instance, genetic events leading to stem cell expansion, genetic immunodeficiency, or marrow failure may serve as an accelerator for acquisition of secondary events either through stem/progenitor pool expansion or by increased cycling due to stress hematopoiesis.

Our previous analysis of allelic burden for a select pool of genes involved mechanistically in hematopoiesis has identified a set of genes harboring risk alleles for development of MN [2]. Myeloperoxidase (MPO), the subject of this study, is one such gene. The MPO gene, located on chromosome 17q22-23, consists of 12 exons and 11 introns [3]. It encodes a heme-containing enzyme, found primarily in azurophilic granules of neutrophils [4], which plays a role in oxygen-dependent microbacterial killing. In particular, reactive oxygen species (ROS) are generated by NADPH oxidase and molecular oxygen. ROS such as superoxide (O₂⁻) are then catalyzed into hydrogen peroxide (H₂O₂) by superoxide dismutase (SOD). The H₂O₂ and halides (Cl⁻, Br⁻, I⁻) or pseudohalide (SCN⁻) ions are assembled and catalyzed by MPO, a myeloid-specific enzyme, which is released by activated neutrophils to produce more potent bactericidal agents and hypohalous acids such as hypochlorous acid (HOCl), hypobromous acid (HOBr), hypoiodous acid (HOI), and hypothiocyanous acid (HOSCN). These agents play important roles in MPO-mediated innate immune response against invading microbes [5]. However, MPO is also involved in suppression of adaptive immune responses. Indeed, Mpo^−/− mice immunized with ovalbumin/complete Freund’s adjuvant have higher T cell activation and proliferation compared to wild type mice [6]. T-cells from Mpo^−/− mice also produce higher levels of IFN-γ and IL-17A when simulated by ovalbumin in vitro. These enhanced adaptive T cell immune responses are also responsible for T cell-mediated skin delayed-type hypersensitivity and antigen-induced arthritis in Mpo^−/− mice and may thus have clinical implications [6].

MPO deficiency is the most common autosomal recessive disorder of phagocytes [7]. The majority of patients with hereditary MPO deficiency are asymptomatic, possibly as a result of MPO-independent compensatory mechanisms. However, various clinical phenotypes are present paralleling the multiplicity of germline variants in MPO gene identified in primary MPO deficiency. The most common pathogenic missense MPO mutation (R569W) causes defects in posttranslational processing, while other previously reported variants such as splice-site mutations (c.2031-2 A > C) lead to a truncated MPO protein resulting in lack of peroxidase activity [8]. Conflicting results have been reported with regard to the consequences of MPO deficiency in various diseases. For instance, MPO deficiency is associated with increased inflammation [9-11], a condition often correlated with cancer pathogenesis. High MPO levels generate MPO-derived oxidants, which contribute to tissue damage in cardiovascular, rheumatologic, and kidney disease [12-14]. However, a strong correlation between MPO deficiency and cancer has not been yet established [15].

Herein, we have investigated the clinical features of MPO carriers who developed MN, and studied the potential mechanisms explaining the increased risk of leukemia in carriers of hypomorphic MPO alleles.

MATERIAL AND METHODS

Patients

A total of 3280 patients with MN (AML n = 1661, MDS n = 1138, MDS/MPN n = 160, MPN n = 100) and BMF syndromes (AA/PNH = 221) were enrolled in this study from our own and historical cohorts (Table S1). Patients with germline MPO mutations were detected by whole exome (WES), targeted next generation (NGS) and Sanger sequencing. Tumor specimens from peripheral blood or BM were collected for discrimination of somatic variants. Germline DNA was obtained from CD3-positive T cells. Clinical data including age, gender, disease type, comorbidities, treatments, cytogenetics, risk stratification, blood and blast counts, follow-up were collected. BM biopsy and aspirate analysis were reviewed by expert Cleveland Clinic hematopathologists. Dysplasia was defined according to the 2016 World Health Organization criteria. MDS patients were classified according to the Revised International Prognostic Scoring System (IPSS-R) with ≤3.5 vs >3.5 as lower-risk (LR) vs higher risk (HR) of transformation to secondary AML [16]. MDS/MPN patients were grouped by WHO 2016 classification; CMML-1, MDS/MPN-U and RARS-T are LRs, CMML-2 were HR. All secondary AML patients derived from MDS or MDS/MPN belonged to the HR group. Patients were treated according to institutional review-approved protocols and in accordance with the Declaration of Helsinki.

Whole exome sequencing

WES was performed as previously described [17, 18]. Whole-exome capture was accomplished through hybridization of sonicated genomic DNA to a bait cDNA library synthesized on magnetic beads (SureSelect Human All Exon 50 Mb or V4 kit, Agilent Technologies). Captured targets were subjected to massively parallel sequencing using HiSeq 2000 (Illumina) according to the standard protocol for 100 bp paired-end reads. All raw read files were first converted to fastq format, then aligned to human genome build hg19 by a Burrows-Wheeler aligner (http://bio-bwa.sourceforge.net/). Aligned reads were processed using Genome Analysis Toolkit (GATK) best practices for variant calling and to remove sequencing errors.

Deep targeted sequencing

Targeted sequencing was performed using a TruSeq Custom Amplicon (Illumina) or a custom cRNA bait library (SureSelect, Agilent Technology) as previously described [19, 20]. For somatic mutations, patients underwent deep targeted sequencing for 36 myeloid panel genes (Table S2). Sequencing libraries were generated according to an Illumina paired-end library protocol. Variants were annotated using Annovar [21]. A bioanalytic pipeline, devised in-house, was used to identify somatic mutations by comparison with sequenced controls and mutational databases, such as dbSNP138, 1000 Genomes Project or ESP 6500 database Exome Aggregation Consortium (ExAC) and genome aggregation database (gnomAD).

Direct genomic sequencing

Genomic regions of MPO gene were amplified using Sanger sequencing by standard techniques as previous described [22] to detect MPO mutations (R569W, c.2031-2 A > C, I642L). The list of primers is provided in Table S3.

Competitive repopulation assay

Mpo mutant mice (Stock# 004265) were procured from Jackson Laboratory. Recipient mice (ROSA26^{(tdTomato-EGFP)}) received two doses of 4.5 Gy total body irradiation delivered 4 hours apart (9 Gy total); a mixture of 2 million mononuclear cells from donor mice BM (50% Mpo^−/− cell, CD45.2; and 50% Mpo^+/+, PEPBoy, CD45.1) were inoculated i.v. into recipient mice via tail vein after the 2nd irradiation. Mice were maintained on antibiotic-containing chow for 4 weeks. Peripheral blood was collected (retro-orbital) for flow cytometry analysis using a BD FACSVerse every 2 weeks for a total of 10 weeks. Antibodies used in this study are listed in Table S4. A replicate experiment was performed using PEPBoy mice as recipients and ROSA26^{(tdTomato-EGFP)} vs. Mpo mutant mice as donors. Animal care and procedures were conducted under a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at Cleveland Clinic.

Flow cytometry for progenitor cells

BM cells from Mpo mutant mice and C57BL/6 J mice were harvested and then lysed with red blood cell lysis buffer on ice. Then 2 μg/mL of primary labeled antibodies were added. Cells were incubated for 20 min at RT in the dark before being washed 3 times and suspended in 500 μL PBS. Flow cytometry was performed on BD FACSVerse. Results were analyzed by FlowJo software (FlowJo LLC). The list of antibodies used is presented in Table S4.

Colony forming assays for murine bone marrow cells

Methylcellulose-based medium with recombinant cytokines for mouse cells (Methocult^™, M3434; STEMCELL Technologies) was supplemented with 200 μM hydrogen peroxide before cell seeding. Mononuclear cells derived from mouse BM were seeded at 50,000 cells/ mL. Colonies were scored on day 10. Cells were harvested for second and third plating at the same density. Cell number and viability were assessed by trypan blue staining using The Vi-CELL^™ XR Cell Viability Analyzer before each replating.

Western blotting

Cell pellets were lysed in RIPA buffer (Thermo, 89900) supplemented with 1X protease inhibitor cocktail (Thermo, A32965), 5 mM EDTA and 1X phosphatase inhibitor cocktail 2&3 (SIGMA, P0044 & 103 P5726) on ice for 20 min. Lysate was spun 20,000 g for 15 min. Supernatant was collected for western blotting analyses. Supernatants were mixed with Laemmli sample buffer 4X plus reducing agent, then boiled at 95 °C for 5 min. Samples were run using NuPAGE^™ 4 to 12%, Bis-Tris (Invitrogen^™, NP0321PK2). Proteins were transferred onto blotting membranes. Membranes were placed into blocking solution for 2 hr at RT, then incubated overnight at 4 °C with an anti-MPO antibody. An anti-actin antibody was used as loading control.

Human myeloperoxidase concentration

Cell lysate samples from human leukemia cell lines (HL-60, SIGM-5, NB-4, MOLM-13, THP-1, K562) were used for quantitative determination of human MPO by Quantikine^® ELISA kit (R&D Systems, DMYE00B) according to the manufacturer’s protocol. Briefly, 50 μL of 5 × 10⁶ cell/mL were mixed with 100 μL of assay diluent RD1-27 and then incubated for 2 hours at RT on a microplate shaker. Aspirate supernatants were then washed with 400 μL washing buffer. Next, 200μL of human MPO conjugate was added to the plate and incubated for 2 h at RT on microplate shaker. Supernatants were measured at emission of 450 nm (excitation 530 nm) using a Synergy H1 Hybrid Multi-Mode Microplate Reader (BioTek). Results were reported in ng/mL.

Hydrogen peroxide/peroxidase activity

Hydrogen peroxide levels released from cell lines and peroxidase activity were detected using Amplex^™ Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen^™, A22188). In total 2 × 10⁶ cell/mL of HL-60 and K562 were harvested. Cells were maintained in RPMI 1640 Medium (Gibco^™) supplemented with 10% FBS (Invitrogen) and 100 U/mL of penicillin and 100 μg/mL of streptomycin at 37 °C with 5% CO2 before collection of supernatants to measure peroxidase activity. 50 μL of peroxidase-containing samples (both controls and experimental samples) were incubated with 100 μM of Amplex Red reagent and 2 mM H₂O₂. The mixture was incubated at RT for 30 min. Resorufin fluorescence in the supernatant was measured at emission of 590 nm (excitation 530 nm) using Synergy H1 Hybrid Multi-Mode Microplate Reader. Background was corrected by subtracting the value derived from no-horseradish peroxidase (HRP) control. Results were reported in raw fluorescence units.

Extracellular H₂O₂ was measured using Amplex Red Hydrogen Peroxide/Peroxidase assay kit (Life Technologies) as described by manufacturer’s protocol. Briefly, 5 × 10⁶ cell/mL of HL-60 and K562 were incubated with 50 μL of 10 mM Amplex Red reagent and 100 μL of 10 U/mL HRP in Krebs-Ringer phosphate buffer for 1 h at 37 °C. Cells were removed by centrifugation and resorufin fluorescence in the supernatant was measured at 0 min (basal point) and then every 10 min until 1 h and every 1 h until 3 h, at emission of 590 nm (excitation 530 nm) using Synergy H1 Hybrid Multi-Mode Microplate Reader.

To mimic MPO deficiency, we used as experimental groups: K562 and HL-60, pretreated with 400 nM of AZD5904 (MedChem Express; HY-111341) overnight, and as control groups: K562 and HL-60 maintained in RPMI 1640 medium supplemented with 10% FBS (Invitrogen). All cell lines were grown in presence of 100 U/mL penicillin and 100 μg/mL streptomycin and were incubated at 37 °C with 5% CO₂. Cells were harvested and washed 3 times with PBS. Both groups were divided in groups receiving 200 μM of H₂O₂ and controls. Cells were collected at pre-H₂O₂ exposure (basal point) and then after H₂O₂ exposure (0.5, 2, and 4 h).

γ-H2AX assay to detect DNA damage after H₂O₂ exposure

All the samples were washed 3 times with PBS after 200 μM hydrogen peroxide exposure before the addition of pre-chilled 70% ethanol drop by drop. Cells were fixed with 100% methanol and then suspended in PBS at 1 × 10⁶ cells/mL. Immunofluorescent staining was performed for flow cytometry using an anti-H2A.X Phospho (Ser139) antibody (BioLegend^®, 613404) FITC-conjugated. An anti-H2A.X phospho (Ser139) antibody and a mouse IgG1, κ Isotype Ctrl (ICFC) antibody were used as controls.

Assessment of viability and apoptosis

Early apoptosis was measured by staining cells with FITC-conjugated annexin V in combination with propidium iodide (BD Pharmingen^™; FITC Annexin V Apoptosis Detection Kit I) following the manufacturer’s instructions. Twenty-four hours after H₂O₂ exposure, HL-60 and K562 were harvested and then washed with cold PBS and suspended in 1X binding buffer at a concentration of 1 × 10⁶ cells/mL. Cell suspension (100 μL) were incubated with 5 μL of annexin V-FITC and 5 μL of PI. After staining, 50,000 cells were analyzed by flow cytometry within one hour, and results were reported as percent cells positive for annexin V and negative for PI.

Cell viability was assessed by CellTiter-Blue^® Cell Viability Assay (Promega, G8081). HL-60 and K562 were harvested and washed with PBS 3 times. Cells were treated with 500 μM of H₂O₂ for 15 min then washed with PBS. Ninety-six well plates containing HL-60 and K562 in culture medium were prepared. Samples, including no cells (negative controls), no H₂O₂ exposure (vehicle controls), and H₂O₂ exposure with or without 400 nM of MPOi were added to wells for a final volume of 100 μL. Cells were incubated at 37 °C with 5% CO₂ for 48 h. After 48 h, 20 μL/per well of CellTiter-Blue^® Reagent was added. Samples were incubated for 2 h. Fluorescence was measured at 560/590 nm.

RESULTS

Germline mutant MPO allelic burden and genomic characterization

Recently, we demonstrated that likely hypomorphic germline variants in MPO may serve as risk alleles for MN based on a high allelic burden identified from WES of germline DNA [2]. Consistent with this report, our expanded analysis of 3280 patients with MN including MDS, MDS/MPN, AML and BMF, demonstrated that MPO mutations are strongly associated with MN evolution (Fig. 1A). MPO variant annotation and filtering was performed as previously described [2]. The paired normal and tumor samples were sequenced in order to confirm germline (GL) origin in 17% of patients for all the described variants (Fig. S1). The variant calling strategy is described in Fig. S2. In total, 38 different GL MPO mutations were identified in 143 cases. We found that GL MPO alterations were enriched in MN compared to the control population (294 vs. 125 per 10⁴ individuals; P < 0.001). Indeed, the carriers had significantly increased risk for AML (OR 2, 95%CI = 1.6 −2.5, P < 0.001) and MDS (1.8, 95%CI = 1.4–2.4, P < 0.001) (Fig. 1B). This enrichment of MPO variants was not observed in BMF patients (OR 1.1), used as a hematologic disease control for myeloid neoplasia. For further analysis, we focused on 28 pathogenic variants from 100 MN patients (Table S5, S6), for assignment algorithm see Fig. S3). The most common recurrent mutation was c.2031-2 A > C (46%; 46/100) (3’ splice site of intron 11) followed by R569W (13%; 13/100), M519fs (13%; 13/100) and Y173C (6%; 6/100) (Fig. 1C). Of note is that all variants mentioned above were described in patient autosomal recessive total/partial MPO deficiency [8, 23-25]. Interestingly, we did not observe any biallelic mutations of MPO.

Fig. 1. Allelic imbalance and characteristics of germline MPO alterations among myeloid malignancies and bone marrow failure syndromes.

A Manhattan plot showing pathogenic/likely pathogenic germline variants of potential susceptibility gene in myeloid malignancies. Each dot represents the P-value (−Log₁₀) of allelic imbalance between myeloid malignancies compared to general populations (data from databases of normal populations: gnomAD). Germline MPO alterations including c.2031-2 A > C, R569W and Y173C are the largest pathogenic/likely pathogenic variant burden among rare variants in myeloid malignancy susceptibility genes. B Frequencies of all germline MPO variants among MN and bone marrow failure syndromes vs. control (left), only pathogenic/likely variants were considered (right). Bars in navy color represent allelic frequencies among disease group and bars in grey color represent control populations. (NS; non significance, ****P < 0.0001) C Schematic diagram of germline pathogenic/likely pathogenic MPO mutations on chromosome 17q22-23. Red and grey circles represent MPO pathogenic and likely pathogenic, respectively. Blue stars represent germline MPO mutations that are reported in MPO deficiency syndrome. D Genetic landscape of associated somatic mutations identified in our cohort. Diagnosis, karyotype, and position of mutations were indicated. DNMT3A, TET2, and SF3B1 were the most common associated mutated genes. E Forest plot showing odd ratios of clinical and genetic associations in MPO^MT and MPO^WT. Circle and whiskers indicate odds ratios (OR) and 95% confidence interval. LR-MDS, FLT3, NRAS and NPM1 mutations were associated with MPO^MT. The significance was determined by Fisher’s exact test; *P < 0.05.

Clinical and molecular landscape of patients with germline MPO mutations

MPO alterations were most frequently observed in AML followed by MDS, MPN and MDS/MPN (3.5%, 3.3%, 2%, 1.3%, respectively). Median age among MPO mutation carriers and wild type MPO cases did not differ. MPO^MT were enriched in patients with LR-MDS (76% vs. 57%; P = 0.04) and thereby had lower BM blast counts (10% vs. 19%; P = 0.009) compared to MPO^WT. Among AML patients, MPO^MT were significantly enriched in AML with NPM1 mutations while no differences were noticed in other AML subtypes (e.g., core-binding factor) (Fig. S4). Overall, patients with MPO alterations showed less megakaryocytic and erythroid dysplasia (Fig. S5). (33% vs. 75%, P = .003 and 42% vs. 70%, P = 0.05, respectively) and did not differ in regard to myeloid dysplasia (42 vs. 55%, P = 0.40). Family history of cancer was found in 11% of patients with MPO mutations, including plasma cell disorders and non-Hodgkin’s lymphoma.

When looking at the co-occurring myeloid lesions, somatic mutations in DNMT3A (22%), TET2 (18%), SF3B1 (18%), NRAS (17%) and NPM1 (15%) (Fig. 1D) were commonly observed in MPO variant carriers. Normal karyotype was found in 46% of MPO^MT, who harbored less trisomy 8 compared to MPO^WT (2% vs. 12%; P = 0.008). Interestingly, only 1/89 MPO^MT harbored del(17p) associated with complex cytogenetics compared to 140/2708 (5%) of MPO^WT. While in univariate analysis LR-MDS, FLT3 NRAS and NPM1 mutations were significantly enriched among carriers of MPO variants (Fig. 1E), multivariate analysis showed association only for FLT3 and NRAS mutations (FLT3; CI 1-3.94, P = 0.02, NRAS; CI = 1–5.8, P = 0.03). There was no survival difference observed between MPO^MT vs. MPO^WT (Median survival 64 vs. 60 mo, P = 0.90) (Fig. S6).

Loss-of-function Mpo mutation provides proliferative advantage over normal bone marrow in murine models

To clarify mechanisms of partial MPO deficiency on potential proleukemogenic propensity of MPO variant carriers, we postulated that we would be able to detect a growth advantage of MPO deficient over normal marrow in competitive repopulation assays. ROSA26^{(tdTomato-EGFP)} mice were sublethally irradiated and then inoculated with donor cells containing 50% of WT PepBoy (CD45.1, Mpo^+/+) and 50% Mpo^−/−(CD45.2) BM cells to distinguish recipient vs. transplant and MPO-deficient vs MPO WT hematopoiesis, respectively (Fig. 2A). Ratios of CD45.2/CD45.1-red fluorescence protein (RFP) negative cells were monitored for 4 weeks after transplantation and then every 2 weeks until the end of the study. Mpo^−/− BM cells gained a proliferative advantage over normal murine BM cells. To confirm this effect, a crossover repopulation assay was done. ROSA26^{(tdTomato-EGFP)} mice and Mpo^−/− BM cells were mixed 50:50 and transplanted into wild type PepBoy mice. Outgrowth of Mpo^−/− were again observed, consistent with the previous result. This prompted us to investigate whether the quantity or function of Mpo^−/− BM cells played a part in this outcome (Fig. 2B). The percentage of progenitor cells including Lineage⁻ c-Kit⁺ Sca1⁺ (LSK), common myeloid progenitors (CMPs), granulocyte–macrophage progenitor (CD34⁺CD16/32⁺Lineage⁻Sca⁻1⁻Kit⁺, GMPs) and megakaryocyte–erythroid progenitor (Lin, Sca-1⁻, c-Kit⁺, CD34⁻, CD16/32⁻, MEPs) of Mpo^−/− and Mpo^+/+ was quantified by flow cytometry. However, the number of each compartment of stem cells was similar in both groups. (1.6% vs. 1.7%; 0.7% vs. 1.0%; 11% vs. 11%, and 32% vs. 33%, respectively)

Fig. 2. Loss of function Mpo mutations gain proliferative advantage over wild-type competitors and increase clonogenic potential after exposure to peroxide treatment.

A i – Schematic diagram of the in vivo competitive repopulation assay. Bone marrow cells from donor mice (CD45.1 + ; PEPBoy mice) and competitor mice (CD45.2 + ; Mpo^−/− mice) were mixed in a 1:1 ratio and injected into the tail vein of irradiated recipient mice (ROSA26^{(tdTomato-EGFP)}). The cross over experiment was carried out. 7 mice per group were included. The experiments were repeated twice. ii and vi- The efficacy of bone marrow (BM) reconstitution was determined in peripheral blood (PB) at 4-, 6-, 8- and 10-weeks post-transplantation by flow cytometry. The recipient cells were excluded by gating out RFP + cells (ii) or FITC-CD45.1 (vi). iii and iv-Representative flow cytometric analysis from recipient mice after BM transplantation. CD45.1FITC + cells and CD45.2 APC + cells from MPOKO mice represent blood cells from Mpo^+/+ and Mpo^−/−cells, respectively(iii). CD45.1+ cells and CD45.2 from C57BL/6 cells were used as controls vii and viii- Reverse competitive repopulation assay between CD45.2+RFP + cells and CD45.2 RFP− cells were done. CD45.2+RFP− represent Mpo^−/−cells while CD45.2+ RFP + cells represent Mpo^+/+cells. C57BL/6 (CD45.2; Mpo^+/+) vs PEP boy mice (45.1; Mpo^+/+) were used as controls v and ix- Mpo^−/− mice gain proliferative advantage over WT competitors. The cross over experiment yielded the same results. B Percentage of Lineage⁻ c-Kit⁺ Sca1⁺ (LSK), common myeloid progenitors (CMPs), granulocyte–macrophage progenitor (CD34 + CD16/32+Lineage-Sca-1−Kit + , GMPs) and megakaryocyte–erythroid progenitor (Lin⁻, Sca-1⁻, c-Kit⁺, CD34⁻, CD16/32⁻, MEPs) from Mpo^−/− mice vs Mpo^+/+. Each dot represents % of total cells from each mouse. (N = 4, NS = no statistical significance, independent sample t-test) C Replating clonogenic assays of Mpo^−/− cells vs Mpo^+/+ with or without hydrogen peroxide exposure. Colony numbers and total cell numbers were quantified. Each dot represents the quantification of each independent experiments. The colony and total cell numbers at second replating were significantly higher in Mpo^−/− cells compared to Mpo^+/+ cells after exposure to peroxide treatment. (N = 4 mice), *P < 0.05, independent sample t-test).

Loss-of-function Mpo mutation increases clonogenic potential after exposure to peroxide

Since MPO produces hypoclorous acid from H₂O₂, we also examined the proliferative/clonogenic consequences of MPO deficiency in the setting of oxidative stress induced by H₂O₂. BM cells derived from Mpo^+/+ and Mpo^−/− mice were treated with H₂O₂ and serially replated. After the second replating, Mpo^−/− cells increased their clonogenic potential upon exposure to peroxide treatment. The colony numbers at the second replating was higher in Mpo^−/− compared to Mpo^+/+ (41 ± 2 vs. 26 ± 7, P = .01), total cell numbers / 5 × 10⁴ parental cells were 2.85 × 10⁶ ± 0.1 vs. 1.73 ± 0.04, P < .001. This finding was even more pronounced at third replating (22 ± 3 vs. 13 ± 4, P = .002), when total cell numbers/ 5 × 10⁴ parental cells were 2.1 × 10⁶ ± 0.08 vs. 1.4 ± 0.09, P < .001) (Fig. 2C).

To identify appropriate models, we analyzed MPO protein expression levels among human leukemia cell lines. Western blot analyses revealed high MPO protein expression in HL-60, SIGM-5 and NB-4 cell lines and absent expression in MOLM-13, THP-1 and K562 (Fig. 3A). To quantify MPO activity, MPO from lysates was measured by ELISA. Values for HL-60 lysate were 36.7 ng/mL(high-MPO) while were 4.6 ng/mL, 1.8 ng/mL and 0 ng/mL for MOLM-13, THP-1 and K562 lysates (low-MPO), respectively (Fig. 3B). We observed that MPO protein levels correlated with MPO activity (Fig. S7). Indeed, HL-60 had high MPO activity, which was attenuated by MPO inhibitor AZD5904, while low MPO activity was found in MOLM-13, THP1 and K562 (Fig. 3C-E). We then hypothesized that partial or total MPO deficiency might result in decreasing intracellular hydrogen peroxide clearance. Therefore, we measured the hydrogen peroxide level in high- and low-MPO cell lines. The low-MPO cell line (K562) released greater amounts of hydrogen peroxide compared to high-MPO cell line (HL-60) over time. (Fig. 3F). Addition of MPOi to HL-60 caused an incremental release of hydrogen peroxide (Fig. 3G). We then explored the consequences of peroxide exposure to the survival/proliferation of leukemic cell lines. Because we did not observe an impact on cell viability at 24 h after H₂O₂ exposure, we studied early-stage apoptotic cells using Annexin V and PI staining detection by flow cytometry. Addition of MPO inhibitor decreased early-stage apoptotic cells in both HL-60 and K562 (HL-60; 16.7 ± 0.2 vs. 11 ± 0.5, P < 0.001 and K562; 30 ± 1 vs. 23 ± 0.6, P < .001, respectively; Fig. 4A). The cell viability changes at 48 h after peroxide exposure showed that the presence of MPO resulted in inhibitor-improved viability following H₂O₂ addition (85 ± 10 vs. 59 ± 8, P < .001), while the cell viability of K562 with or without MPOi were not significantly different (84 ± 2 vs. 89 ± 10, P = 0.70; Fig. 4B).

Fig. 3. Human myeloperoxidase levels, activities and H₂O₂ release from human leukemia cells.

A Western blot analysis of MPO level in various myeloid cell lines. Actin was used as a positive control. B Human MPO concentrations (ng/mL) of myeloid leukemia cell lines. (N = 4). C Myeloperoxidase activity of HL-60 and K562 cell lines. Horseradish peroxidase (HRP) were used to generate peroxidase standard curve. D, E Myeloperoxidase activity of various myeloid leukemia cell lines. HL-60 had high MPO activity while MOLM-13, THP-1 and K562 were nearly absent MPO activity. Adding MPOi reduced MPO activity in HL-60. (N = 4) (N = 4, *P < 0.05, independent sample t-test). F Detection of H₂O₂ release. MPO level cell lines (K562) released higher amount of H₂O₂ compared to high MPO cell line (HL-60). (N = 4). G Adding MPOi to decrease MPO activity increased H₂O₂ release from cells (N = 4, ***P < 0.001, independent sample t-test).

Fig. 4. Apoptosis assay of HL-60 and K562, treated with H₂O₂ 500 μM and with/without MPOi for 24 h (A) and 48 h (B) before analysis.

A Assessment of early apoptosis cells using annexin V-FITC/propidium iodide staining detected by flow cytometry 24 hour after H₂O₂ exposure. (i)& (iii) Bar graphs showing percentage of early apoptosis cells of HL-60 (i) and K562 (iii). (N = 4, NS = no statistical significance, independent sample t-test) (ii)& (iv) Annexin-V flow cytometry fluorescence intensity distributions of HL-60 (ii) and K562 (iv) treated with hydrogen peroxide and incubated with (light blue curve) or without (red curve) MPOi. (v)& (vi) Scatter plots showing early apoptotic cells, propidium iodide (y-axis) versus annexin V (x-axis). Adding MPOi decreased the percentage of early apoptotic cells after H2O2 exposure. B Assessment of cell apoptosis using cell viability test. Bar graphs showing percentage of cell viability of HL-60 (left) and K562 (right) after 48 h following the exposure to H₂O₂ 500 μM with/without MPOi. In High MPO cell line, High MPO cell lines with MPOi had higher cell viability compared to without MPOi. (N = 4).

Effects of MPO activity on DNA damage

The pre-leukemogenic effect of MPO deficiency may also be related to its effects on genomic instability via ROS production. We postulated that an increase in hydrogen peroxide was capable of causing minute DNA damage over a period of time should the appropriate repairing mechanisms not work properly. HL-60 and K562 cell lines (models for MPO-high and -low) were treated with 200 μM hydrogen peroxide. DNA damage was assessed by using ɣ-H2AX flow cytometric assay. Following H₂O₂ exposure, phosphorylated ɣ-H2AX levels were increased in K562 compared to their baseline level (0.9 ± 0.4 vs. 76 ± 0.3; P < .001), while phosphorylated ɣ-H2AX was similar in HL-60 (3 ± 0.2 vs. 3.5 ± 0.5; P = 0.90). Interestingly, phosphorylated ɣ-H2AX levels were remarkably increased when adding MPOi to HL-60 (3.5 ± 0.5 vs. 24 ± 0.6; P < .001) while adding MPOi to K562 had minimal effects on DNA damage (76 ± 0.7 vs. 87 ± 6.0; P = .034) (Fig. 5A, B).

Fig. 5. Phospho-ɣ-H2AX, the DNA damage/repair marker, detection by flow cytometry. HL-60 and K562 were treated with H₂O₂ 200 μM and with/without MPOi.

A Scatter plots showing APC-phospho-ɣ-H2AX positive cells of HL-60 (i) and K562 (ii) at 4 h after H₂O₂ exposure including control, following H₂O₂ exposure and H₂O₂ co-culture with MPOi, respectively. (N = 4) B Line graph showing percentage of APC-phospho-ɣ-H2AX positive cells of HL-60 with/without MPOi (left) and K562 with/without MPOi (Right) at baseline levels, 30 min, 2 h and 4 h, respectively. Blue lines represented percentage of APC-phospho-γ-H2AX positive cells with MPOi while redline represented percentage of APC-phospho-ɣ-H2AX positive cells without MPOi. (N = 4).

DISCUSSION

The identification of germline alterations contributing to predisposition to MN has been the domain of pediatric hematology, whereas the genotype-phenotype association and causality behind disease-prone mutations and leukemia are less pronounced in adults. For some recessive and dominant nonmalignant marrow failure states, the carrier status may not directly translate into obligatory leukemia evolution. Nevertheless, some variants may represent the genomic underpinnings of a non-negligible etiologic fraction of cases. Our results suggest that heterozygous germline MPO alterations may serve as predisposition factors for MN in adults. In particular, we were able to provide several lines of evidence to support this theory, including 1) findings from population genetics established within a large cohort of patients vs. control groups; 2) study of invariant genomic features private to MPO variants carriers; 3) functional effects on hematopoiesis in murine models; 4) biochemical consequences of MPO defects. Altogether, these results may argue for the role of MPO variants as contributors of a “mutator phenotype”, generating an increased HSC turnover and thereby stochastically amplifying the likelihood of clonal events.

Increased incidence of cancer in patients with MPO deficiency has been reported [26-28]. Our study showed that heterozygous MPO mutants had an approximately 2-fold higher risk for developing MN than the general population. This moderate increase is comparable to that described for many other established germline defects; e.g. telomerase complex mutations [29]. Indeed, the risk for leukemia in monoallelic A1062T TERT mutations carriers was estimated at ~3-fold. However, the exact value is not clear. For instance, odds ratio of ~4.4 for developing pediatric AML were derived for heterozygous carriers of FANC gene variants [30], while others found that the risk of cancer among FA heterozygotes was not increased [31]. Inherited leukemia traits are genetically and phenotypically heterogeneous, have variable clinical penetrance, and environmental exposures such as lifetime infections and inflammation burden, and thus carrier status of MPO alterations may not have a uniform impact on leukemia evolution. Generally, an exact calculation of germline contribution to leukemia risk may be very complex, and phenotypes may be overlapping among many factors which preclude recognition of less dramatic players via traditional approaches.

Based on biochemical experiments coupled with in vivo characterization, we demonstrated that loss of MPO leads to proliferative advantage to BM cells in competitive repopulation assays. This advantage has been shown to be primarily driven by higher self-replication and the genetic loss of MPO was recapitulated by a small molecule MPO inhibitor. However, this advantage was not due to an increased HSC compartment as demonstrated by similar numbers of LSK, GMP, CMP, and MEP cells. On the contrary, the clonogenic potential of Mpo knockout cells under Mpo deficient stage was significantly higher.

ELANE mutations in severe congenital neutropenia (SCN) are thought to cause nonsense-mediated decay or truncations of neutrophil elastase, depending on the exon affected [32, 33]. The defective misfolded protein induces endoplasmic reticulum stress, increased ROS production [34] and ultimately results in oxidative damage to DNA. The ROS accumulation possibly responsible for a stressed hematopoiesis induces second genetic hits such as CSF3R/RUNX1 mutations and MDS/AML evolution in SCN patients [35, 36]. The pathophysiologic scenario in MPO deficiency may be analogous. In fact, MPO is the key regulator of ROS production. This peroxidase utilizes H₂O₂ as a substrate in a Fenton reaction to generate hypochlorous acid (HOCl) and other ROS products. [6, 37, 38] In addition to clearance of extracellular pathogens, H₂O₂ plays important roles in cellular signaling that mediate cell proliferation, differentiation, and survival [39-43]. Differences in spatiotemporal change of H₂O₂ exposure induce diverse cellular stress responses from physiologic oxidative eustress to oxidative distress [44, 45]. Indeed, basal H₂O₂ concentration is essential for biological fundamental processes while higher H₂O₂ concentrations lead to oxidative stress and eventually cell death [45]. Studies have repeatedly shown that high H₂O₂ beyond physiologic levels activate the DNA damage response by single- and double-strand breaks [46-48]. Ultimately, MPO defects could lead to increased hematopoietic progenitor cell turnover and stressed hematopoiesis with a development of secondary MN.

We were able to demonstrate that pharmacological inhibition of MPO in HL-60 cells with higher induced H₂O₂ production mimics genetic disruption of MPO. Such increased H₂O₂ caused by reduction of MPO activity led to genotoxicity, evidenced by higher amounts of γ-H2AX, which is the genetic footprint activating DNA repair pathway rather than apoptosis.

In sum, our study identified that genetic loss of MPO in progenitor and stem cells leads to higher accumulation of H₂O₂ resulting in DNA damage and activation of error-prone non-homologous end-joining mediated by PARP. Thus, the oncogenic evolution of MPO deficient cells is in part driven by higher levels of an error-prone repair process, while wild type cells will either utilize homologous end joining and create error-free progeny or kill the cell via apoptosis. On the other hand, increased levels of DNA repair induced in MPO mutant and MPO deficient cells might lead to genomic instability and higher mutational burden among the surviving clones. Therefore, loss of MPO aids neoplastic evolution and drives pathogenesis of MN.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41375-022-01630-0.