Germ line ERG haploinsufficiency defines a new syndrome with cytopenia and hematological malignancy predisposition

Abstract

The genomics era has facilitated the discovery of new genes that predispose individuals to bone marrow failure (BMF) and hematological malignancy (HM). We report the discovery of ETS-related gene (ERG), a novel, autosomal dominant BMF/HM predisposition gene. ERG is a highly constrained transcription factor that is critical for definitive hematopoiesis, stem cell function, and platelet maintenance. ERG colocalizes with other transcription factors, including RUNX family transcription factor 1 (RUNX1) and GATA binding protein 2 (GATA2), on promoters or enhancers of genes that orchestrate hematopoiesis. We identified a rare heterozygous ERG missense variant in 3 individuals with thrombocytopenia from 1 family and 14 additional ERG variants in unrelated individuals with BMF/HM, including 2 de novo cases and 3 truncating variants. Phenotypes associated with pathogenic germ line ERG variants included cytopenias (thrombocytopenia, neutropenia, and pancytopenia) and HMs (acute myeloid leukemia, myelodysplastic syndrome, and acute lymphoblastic leukemia) with onset before 40 years. Twenty ERG variants (19 missense and 1 truncating), including 3 missense population variants, were functionally characterized. Thirteen potentially pathogenic erythroblast transformation specific (ETS) domain missense variants displayed loss-of-function (LOF) characteristics, thereby disrupting transcriptional transactivation, DNA binding, and/or nuclear localization. Selected variants overexpressed in mouse fetal liver cells failed to drive myeloid differentiation and cytokine-independent growth in culture and to promote acute erythroleukemia when transplanted into mice, concordant with these being LOF variants. Four individuals displayed somatic genetic rescue by copy neutral loss of heterozygosity. Identification of predisposing germ line ERG variants has clinical implications for patient and family diagnoses, counseling, surveillance, and treatment strategies, including selection of bone marrow donors and cell or gene therapy.

Graphical abstract

Figure 1.

Germ line ERG variant identified in a family with hematological conditions. (A) Pedigree of family 1 containing the ERG (Y373C) variant that segregates with thrombocytopenia and AML. Age of onset (years). (B) History of platelet and neutrophil counts of affected family members (patient 15, 16, and 17) (I-1, II-1, and II-2). Absolute neutrophil or platelet counts per microliter of blood from complete blood examinations plotted with age (years). Normal lower limit for neutrophils and platelets is marked with a dotted red line. (C) Log R ratio and B-allele plots of single-nucleotide polymorphism–array analysis. All individuals show a cnLOH event that encompasses the entire ERG gene (yellow highlight). Individual II-1 (patient 16) had a second cnLOH event (blue highlight). Log R ratios (top panels) show no loss of copy number, and B-allele frequencies (bottom panels) show regions of LOH, together demonstrating cnLOH. Samples used: patient I-2, BM cytogenetic pellet; II-1 and II-2, PB mononuclear cells. d, died; dx, diagnosed.

Figure 2.

Functional characterization of ERG variants. (A) ERG variants residing in the ETS domain reduce transactivation. K562 cells were transfected with pcDNA3 empty vector (EV) or pcDNA3-ERG (WT or variants). All constructs were cotransfected with a luciferase reporter plasmid driven by an ITGA2B promoter (quadruplicate replicates, repeated 3 times). Fold change (mean ± standard error of the mean [SEM]) in comparison with the WT is plotted. Pairwise comparisons are shown (∗P < .05 in comparison with WT). (B) ETS domain variants reduce ERG DNA binding affinity. EMSA of WT ERG and variants. Transfected HEK293 whole cell lysates were prepared and bound to an oligonucleotide containing an ETS DNA consensus sequence. Probes were visualized using chemiluminescence. Pairwise comparisons are shown (∗P < .05 compared with WT). (C) ERG ETS domain variants alter subcellular localization. A Myc-tag was added to WT ERG and variants. COS-7 cells were transfected with pcDNA3 EV, pcDNA3-ERG-Myc (WT), and pcDNA3-ERG-Myc variants. Cells were stained for a Myc-tag and DAPI (4′,6-diamidino-2-phenylindole). The nuclear to cytoplasmic ratio of each variant was quantified and the fold change (mean ± SEM) in comparison with the WT was plotted. Pairwise comparisons are shown (∗P < .05 compared to WT). Nuclear to cytoplasmic ratio was obtained from 3 independent experiments. In all comparisons, a Student t test was used (∗P < .05; ∗∗P < .01; ∗∗∗∗P < .001 in comparison with WT).

Figure 3.

ERG LOF variants fail to drive megakaryocytic differentiation and cytokine independence. (A) In vitro differentiation of FLCs that overexpress WT ERG and variants. Immature stem or progenitor cell population with megakaryocytic features were stained using antibodies against cKIT⁺ and CD41⁺ with the percentage of double positive cells measured by flow cytometry (6 independent experiments). Pairwise comparisons are shown (∗∗∗∗P < .001 in comparison with WT). In all comparisons, a Student t test was used. (B) Example of gating strategy for cKIT⁺ and CD41⁺ expression on ERG WT– or variant-transduced FLCs. Cells were first gated for viability and mCherry expression. (C) Cell viability of FLC culture after cytokine removal measured over time.

Figure 4.

ERG variants are LOF in an in vivo leukemia model driven by ERG overexpression. (A) Enforced expression of ERG WT and WT-like variants (P116R, M219I) in mice led to the development of erythro-megakaryocytic leukemia within 220 days. (B) mCherry engraftment over time in the PB of mice. Numbering refers to 3 ERG WT mice that succumbed to disease (Figure 5A). (C) Infiltration of ERG WT/mCherry⁺ transplanted FLCs into recipient mouse BM, spleen, and PB as measured by fluorescence-activated cell sorter (FACS) analysis. Recipient mouse number 3 shown is representative of all ERG WT and WT-like (P116R, M219I) mice. (D) ERG WT and WT-like variant mice developed mCherry⁺ erythro-megakaryocytic leukemia with cKit⁺ and CD71⁺ expression. FACS plots of a representative ERG WT leukemia are shown. Nonleukemic (ie, mCherry^–) BM and spleen cells are shown as comparison.

Figure 5.

ACMG classification of germ line ERG variants.ERG variants were classified using the ACMG and the Association for Molecular Pathology criteria for cytopenias, HMs, and/or lymphedema (#). Based on functional assay criteria, variants with complete LOF in at least 1 functional assay were classified as PS3_Strong, variants with hypomorphic activity (<50%) in 1 or more functional assays were classified as PS3_Moderate, and variants with hypomorphic activity (>50%) in 1 or more assays were classified as PS3_Supporting. The PS3 and BS3 criteria were not applied when variants showed no change in functional assays or were not tested. PNT, pointed domain.

Figure 6.

Functional consequences of rare ERG variants. Rare ERG variants from similar phenotypic groups, including BMF and/or HM and lymphedema, among population variants (gnomAD >200), a germ line thrombocytopenic mouse variant, a paralogous ETV6 pathogenic variant (thrombocytopenia), and Catalogue Of Somatic Mutations In Cancer mutation (somatic) are mapped onto the ERG protein (isoform, NP_891548.1; transcript, NM_182918.4). Functional characterization of each variant via transactivation, DNA binding, subcellular localization, FLC myeloid differentiation, FLC cytokine independence, and leukemogenesis assays are displayed.

Figure 7.

Proposed model for ERG deficiency syndrome–associated phenotypes showing adaptive/maladaptive events in ERG heterozygous carriers. It is proposed that ERG carriers harbor a threshold of activity that is insufficient under certain physiological and environmental stressors, thus leading to a range of phenotypic outcomes (top panel). Malignant cells (yellow cells). Maladaptive and adaptive events observed in our study (arrow). Other potential (maladaptive/adaptive) somatic events seen in other BMF and/or HM predisposition genes, , (dotted arrow) or hypothetical events (question mark preceding dotted arrow). Game of clones (bottom panel). The development of different physiological outcomes in patients is the consequence of a game of clones in which context-dependent competitive fitness of the clone(s) determines progression and the observed outcome. Notably, adaptive and maladaptive events and subsequent clonal selection may occur simultaneously. The clonal output shown is based on mature blood cells, reflecting collectively the impact on mature and stem and progenitor cells.