CBFA2T3::GLIS2 pediatric acute megakaryoblastic leukemia is sensitive to BCL-XL inhibition by navitoclax and DT2216

Abstract

Acute megakaryoblastic leukemia (AMKL) is a rare, developmentally restricted, and highly lethal cancer of early childhood. The paucity and hypocellularity (due to myelofibrosis) of primary patient samples hamper the discovery of cell- and genotype-specific treatments. AMKL is driven by mutually exclusive chimeric fusion oncogenes in two-thirds of the cases, with CBFA2T3::GLIS2 (CG2) and NUP98 fusions (NUP98r) representing the highest-fatality subgroups. We established CD34+ cord blood-derived CG2 models (n = 6) that sustain serial transplantation and recapitulate human leukemia regarding immunophenotype, leukemia-initiating cell frequencies, comutational landscape, and gene expression signature, with distinct upregulation of the prosurvival factor B-cell lymphoma 2 (BCL2). Cell membrane proteomic analyses highlighted CG2 surface markers preferentially expressed on leukemic cells compared with CD34+ cells (eg, NCAM1 and CD151). AMKL differentiation block in the mega-erythroid progenitor space was confirmed by single-cell profiling. Although CG2 cells were rather resistant to BCL2 genetic knockdown or selective pharmacological inhibition with venetoclax, they were vulnerable to strategies that target the megakaryocytic prosurvival factor BCL-XL (BCL2L1), including in vitro and in vivo treatment with BCL2/BCL-XL/BCL-W inhibitor navitoclax and DT2216, a selective BCL-XL proteolysis-targeting chimera degrader developed to limit thrombocytopenia in patients. NUP98r AMKL were also sensitive to BCL-XL inhibition but not the NUP98r monocytic leukemia, pointing to a lineage-specific dependency. Navitoclax or DT2216 treatment in combination with low-dose cytarabine further reduced leukemic burden in mice. This work extends the cellular and molecular diversity set of human AMKL models and uncovers BCL-XL as a therapeutic vulnerability in CG2 and NUP98r AMKL.

Graphical abstract

Figure 1.

Generation of human models of CG2 leukemia. (A) Experimental procedure used to establish xenograft models of CG2 AMKL (mCG2 AMKL) using independent lentiviral transduction in CB CD34⁺ cells (CB CD34⁺, pool of 6 CB units) and transplantation in recipient NSG mice. (B) Schematic representation of 6 CG2 leukemia models (mCG2) describing initial gene transfer (GT; %GFP) at the time of transplantation and leukemia latency in primary recipient mice. Mouse identification shown in brackets; 1 of 10 mice was not available for analysis. (C) Detection of CG2 fusion transcript expression by reverse transcription polymerase chain reaction (RT-PCR) with RNA isolated from leukemic blasts, as indicated. M07e and normal lineage-depleted CB (CB LIN⁻) cells were used as the positive and negative control, respectively. ABL1 was used as housekeeping gene. (D) Percentage of infiltrating human blasts (hCD45⁺ GFP⁺) and (E) CD41⁺ cells (of hCD45⁺ GFP⁺ population) in the BM and spleen of CG2 primary recipient mice (color code indicates distinct primary mice). (F) Hematoxylin phloxine saffron–stained longitudinal sections of tibia bones harvested from primary recipient mice that received transplantation with control CB CD34⁺ cells transduced with empty vector (top panel, 47 weeks after transplantation) and from a CG2-1 AMKL model (secondary recipient mice that received transplantation with 1 × 10⁶ CG2-1 AMKL xenograft cells and were euthanized 11.1 weeks after transplantation) (bottom). (G) Survival curves of primary recipient mice that received transplantation with CB CD34⁺ cells transduced with CG2 (black line) or empty vector (gray dashes), and mice that serially received transplantation up to 3 times with CG2-1 AMKL (red lines). Giemsa-stained cytospin and flow cytometry profiles of leukemic BM cells from representative (H-I) primary (1^ary) and (J-K) secondary (2^ary) CG2-1 AMKL recipient mice (2^ary recipient mice that received transplantation with 1.4 × 10⁶ CG2-1 AMKL xenograft cells and that were euthanized 6.9 weeks after transplantation). Detailed characteristics of other CG2 leukemia xenografts in recipient mice that serially received transplantation are described in supplemental Table 5 and supplemental Figure 1. (L) Survival curves of recipient NSG mice that received transplantation with CG2-1 AMKL xenograft cells in a limiting-dilution cell transplantation assay and (M) estimation of LIC frequency and 95% confidence interval (red dashes) using the extreme limiting dilution analysis software (http://bioinf.wehi.edu.au/software/elda/).

Figure 1.

Generation of human models of CG2 leukemia. (A) Experimental procedure used to establish xenograft models of CG2 AMKL (mCG2 AMKL) using independent lentiviral transduction in CB CD34⁺ cells (CB CD34⁺, pool of 6 CB units) and transplantation in recipient NSG mice. (B) Schematic representation of 6 CG2 leukemia models (mCG2) describing initial gene transfer (GT; %GFP) at the time of transplantation and leukemia latency in primary recipient mice. Mouse identification shown in brackets; 1 of 10 mice was not available for analysis. (C) Detection of CG2 fusion transcript expression by reverse transcription polymerase chain reaction (RT-PCR) with RNA isolated from leukemic blasts, as indicated. M07e and normal lineage-depleted CB (CB LIN⁻) cells were used as the positive and negative control, respectively. ABL1 was used as housekeeping gene. (D) Percentage of infiltrating human blasts (hCD45⁺ GFP⁺) and (E) CD41⁺ cells (of hCD45⁺ GFP⁺ population) in the BM and spleen of CG2 primary recipient mice (color code indicates distinct primary mice). (F) Hematoxylin phloxine saffron–stained longitudinal sections of tibia bones harvested from primary recipient mice that received transplantation with control CB CD34⁺ cells transduced with empty vector (top panel, 47 weeks after transplantation) and from a CG2-1 AMKL model (secondary recipient mice that received transplantation with 1 × 10⁶ CG2-1 AMKL xenograft cells and were euthanized 11.1 weeks after transplantation) (bottom). (G) Survival curves of primary recipient mice that received transplantation with CB CD34⁺ cells transduced with CG2 (black line) or empty vector (gray dashes), and mice that serially received transplantation up to 3 times with CG2-1 AMKL (red lines). Giemsa-stained cytospin and flow cytometry profiles of leukemic BM cells from representative (H-I) primary (1^ary) and (J-K) secondary (2^ary) CG2-1 AMKL recipient mice (2^ary recipient mice that received transplantation with 1.4 × 10⁶ CG2-1 AMKL xenograft cells and that were euthanized 6.9 weeks after transplantation). Detailed characteristics of other CG2 leukemia xenografts in recipient mice that serially received transplantation are described in supplemental Table 5 and supplemental Figure 1. (L) Survival curves of recipient NSG mice that received transplantation with CG2-1 AMKL xenograft cells in a limiting-dilution cell transplantation assay and (M) estimation of LIC frequency and 95% confidence interval (red dashes) using the extreme limiting dilution analysis software (http://bioinf.wehi.edu.au/software/elda/).

Figure 2.

Gene expression in CG2 leukemia models correlate with pediatric disease. (A) (Left) Correlation of differential gene expression (log2 fold change [L2FC]) in CG2 AMKL models and patients from our institutional data set (Centre hospitalier universitaire Sainte-Justine [CHUSJ]) compared with a validation data set of pediatric CG2 AMKL (St Jude). Differentially expressed genes (DEGs), defined as |L2FC > 1| and false discovery rate (FDR) q value < .05, common to both data sets are indicated in blue or red and define the CG2 signature (supplemental Table 7). Institutional (Inst) data set: CG2 AMKL models (n = 10) and patient-derived CG2 samples (n = 2) compared with N5A AMKL models (n = 5), patient-derived N5A samples (n = 2), and normal CB CD34⁺ cells (n = 4). Validation (Val) data set: CG2 AMKL (n = 12) vs other genetic subtypes of AMKL (n = 61) from pediatric patients at diagnosis. (Right) Upset plots showing DEGs that are jointly overexpressed (n = 399) or underexpressed (n = 330) in CG2 leukemias, corresponding to blue and red dots in panel A. (B) Hierarchical clustering using the 729 DEGs of the CG2 gene expression signature. (C) Heat map showing protein (log2[mass spectrometry [MS] values, left panels) and messenger RNA (RNAseq; fragments per kilobase of transcript per million mapped reads [FPKM] values, right panels) expression of cell surface markers associated to CG2 leukemia (high expression in mCG2-1 and mCG2-2 leukemia models log2[MS values] ≥ 16 and RNAseq FPKM ≥ 5) and weak/no expression in normal CB CD34⁺ cells (log2[MS values] < 16 and RNAseq FPKM < 5). Samples were analyzed in triplicates and represented as mean expression (biological triplicates for RNAseq and technical triplicates for proteomic data). For comparison, values for mN5A and pdxNTF are shown alongside mCG2 samples. The St Jude Val cohort RNAseq expression is presented as a separate column. (D) Star plot presenting the adjusted P values of DEGs and differentially expressed proteins (FDR < 0.05) (transcriptome: CHUSJ CG2 vs N5A AMKL and normal CB CD34⁺; surfaceome: CG2 vs NUP98r and CB CD34⁺). The scores were calculated by multiplying the algebraic sign (+ or −) of the log2FC, surfaceome or transcriptome, by the corresponding log10(adjusted P value). Significantly upregulated CG2 AMKL–specific surface markers intersecting both data sets are labeled. (E) Validation by flow cytometry of PCDH19 surface expression on CG2 AMKL cells from patient (pCG2-1), model mCG2-1 and M07e cell line. PCDH19 is not expressed on lineage-depleted human CB cells (CB CD34⁺). Samples were costained with NCAM1 (CD56). (F) Scatterplot representations showing the best pairwise correlations of the 2 sets of cell surface marker genes that select for CG2 genotype in a validation data set of pediatric AMKL (of 8 cell surface marker combinations, see supplemental Figure 13). Values, from top to bottom, represent the global and subtype-specific Kendall rank correlation coefficients. P values: ∗P < .05 and ∗∗P < .005.

Figure 3.

CG2 leukemia consists of immature megaerythroid stem and progenitor lineages. (A) Uniform manifold approximation and projection (UMAP) of normal BM lineages. The stem-megaerythroid compartment is circled, relevant populations are labeled, and differentiation trajectories are highlighted with arrows. (B) DEGs between cells of the megakaryocytic lineage (CD34⁺ MEPs, CD34⁺ MKPs, and platelets) and all other cell types in normal BM (supplemental Table 11). Genes relevant to the megakaryocytic differentiation are highlighted in red, and genes of the mitochondrial apoptotic pathway are highlighted in blue. (C, top) Percentage positive cells for each gene in mCG2s, pCG2s, other subtypes of AML and in specific populations of the normal BM (HSCs, MEPs, MKPs, and PLTs). The horizontal line represents the median of the group, the patient and models are color coded as in panel D, and the normal samples are color coded based on donor as in supplemental Figure 14D. (Bottom) ITGA2B (CD41) expression per cell on a UMAP representation of mCG2-1, pCG2-1, and normal BM (extended samples in supplemental Figures 15 and 16). (D) Radar plot (top) of lineage composition assessed by single-cell RNAseq in CG2 AMKL models and pediatric patients, as compared with diverse phenotypic and genetic subtypes of adult AML. The detailed proportion of each cell type is presented as stacked bar plots (bottom). Color coding of populations is as depicted in panel A. (E) Representation as in panel C of terminal differentiation gene expression (extended samples in supplemental Figures 15 and 16). (F) DEGs in the CD34⁺ MKPs from the mCG2 and pCG2s compared with those in normal CD34⁺ MKPs (supplemental Table 12). Selected markers are highlighted in red, and genes of the mitochondrial apoptotic pathway are highlighted in blue. (G) Representation as in panel C of BCL2 and BCL2L1 (BCL-X_L) expression. ERP, erythroid progenitors; HSCs, hematopoietic stem cells; PLT, platelet.

Figure 3.

CG2 leukemia consists of immature megaerythroid stem and progenitor lineages. (A) Uniform manifold approximation and projection (UMAP) of normal BM lineages. The stem-megaerythroid compartment is circled, relevant populations are labeled, and differentiation trajectories are highlighted with arrows. (B) DEGs between cells of the megakaryocytic lineage (CD34⁺ MEPs, CD34⁺ MKPs, and platelets) and all other cell types in normal BM (supplemental Table 11). Genes relevant to the megakaryocytic differentiation are highlighted in red, and genes of the mitochondrial apoptotic pathway are highlighted in blue. (C, top) Percentage positive cells for each gene in mCG2s, pCG2s, other subtypes of AML and in specific populations of the normal BM (HSCs, MEPs, MKPs, and PLTs). The horizontal line represents the median of the group, the patient and models are color coded as in panel D, and the normal samples are color coded based on donor as in supplemental Figure 14D. (Bottom) ITGA2B (CD41) expression per cell on a UMAP representation of mCG2-1, pCG2-1, and normal BM (extended samples in supplemental Figures 15 and 16). (D) Radar plot (top) of lineage composition assessed by single-cell RNAseq in CG2 AMKL models and pediatric patients, as compared with diverse phenotypic and genetic subtypes of adult AML. The detailed proportion of each cell type is presented as stacked bar plots (bottom). Color coding of populations is as depicted in panel A. (E) Representation as in panel C of terminal differentiation gene expression (extended samples in supplemental Figures 15 and 16). (F) DEGs in the CD34⁺ MKPs from the mCG2 and pCG2s compared with those in normal CD34⁺ MKPs (supplemental Table 12). Selected markers are highlighted in red, and genes of the mitochondrial apoptotic pathway are highlighted in blue. (G) Representation as in panel C of BCL2 and BCL2L1 (BCL-X_L) expression. ERP, erythroid progenitors; HSCs, hematopoietic stem cells; PLT, platelet.

Figure 4.

CG2 and NUP98r AMKL xenografts are sensitive to the induction of the intrinsic apoptotic pathway. (A-B) Dose-response curves and IC₅₀s (supplemental Table 13) determined for each indicated sample of AMKL or AML, submitted to a viability assay in presence of venetoclax or navitoclax. (Cell-Titer Glo, 6-day incubation, 4 replicates). Viability readout was normalized to dimethyl sulfoxide (DMSO) controls for each sample. (C-D) Apoptosis assessed by annexin V staining and flow cytometry of AKML xenografts (mCG2, pdxNTF, and mN5A; n = 2 biological replicates for each cell type), ML2 (AML cell line, n = 2), or normal CB CD34⁺ cells (n = 3) treated for 72 hours in culture as triplicates with the indicated BH3 mimetics (123.5 nM, 370 nM, and 10 μM) or DMSO. P values: ∗∗P < .005; ∗∗∗P < .001; ∗∗∗∗P < .0001. (E-F) Mitochondrial superoxide production was assessed by MitoSox staining and flow cytometry of AMKL xenografts or normal CB CD34⁺ cells, treated with the indicated BH3 mimetics or DMSO for 72 hours in culture in duplicates. Staurosporine (STS) was used as positive control at 1 μM. mAMKL, synthetic xenograft of AMKL; mAML, synthetic xenograft of AML; pdx, patient-derived xenograft; CB CD34⁺, CB CD34⁺ cells.

Figure 5.

AMKL xenografts depend on prosurvival protein BCL-X_L for survival. (A) RNAseq expression values (log2[FPKM]) of selected genes implicated in intrinsic apoptosis pathway in model and patient leukemias from our institutional data set. (B) Crossvalidation of gene expression in a large and published data set of pediatric AMKL. (C) Protein levels of BCL-X_L (left) and BCL2 (right) in AMKL xenograft cells or normal CB CD34⁺ cells assessed by intracellular flow cytometry. (D) Apoptosis, assessed by annexin V staining and flow cytometry, (E) relative gene expression detected by quantitative reverse transcription polymerase chain reaction, and (F) prosurvival protein levels quantified by flow cytometry, after short hairpin RNA (shRNA)-mediated KD of prosurvival BCL2 proteins (BCL-X_L, BCL2, and BCL-W) in CG2 xenografts (72 hours after infection, 2 selected shRNA per gene). Samples were compared with nontransduced (NT) or cells infected with control shRNA against Renilla (shRenilla). See supplemental Figure 18 for KD studies in mCG2-2 cells. (G) Dose response curves and IC₅₀ values were determined after incubation of AMKL or AML xenografts with DT2216 (BCL-X_L proteolysis-targeting chimera) or DMSO for 6 days, followed by viability readout with Cell-Titer Glo. Viability readout was normalized against DMSO for each sample. (H) Amount of apoptosis assessed by annexin V staining and flow cytometry in AMKL xenografts or normal CB CD34⁺ cells after 72 hours of incubation with DT2216 at 100 nM or 1 μM in comparison with cells treated with DMSO. GATA1s, GATA1 truncation; HOXr, HOX rearrangement; KMT2Ar, KMT2A rearrangement; MFI, mean fluorescence intensity; RBM-MKL, RBM15::MKL1; VHL, von Hippel-Lindau; xAMKL, synthetic model of AMKL; xAML, synthetic model of AML. P values: ∗P < .05; ∗∗P < .005; ∗∗∗∗P < .0001.

Figure 6.

CG2 AMKL is impaired by navitoclax and DT2216 treatment in vivo. (A) Workflow of experimental procedures to assess the in vivo activity of navitoclax or DT2216 in CBFA2T3::GLS2 (mCG2-1) AMKL xenografts. On the day of euthanizing, (B) spleen weights, (C) percentage of leukemic blasts in the peripheral blood (% GFP⁺hCD45⁺), and (D) infiltration of the BM and spleen (% GFP⁺ cells) were assessed in mice that received transplantation after treatment with either vehicle only (n = 5) or navitoclax (n = 6) for 3 weeks. For comparison, spleen weights of healthy mice that did not receive transplantation (n = 6) were recorded for data shown in panel B. (E) Hematoxylin and eosin–stained longitudinal sections of the tibia and spleen collected from mice on the day of euthanizing. Conditions were as follows: transplantation with CG2 but not treated (leukemic), transplantation and treated with vehicle only (Vehicle), no transplantation and untreated, age-matched littermates of treatment groups (not transplanted), or transplantation and treated with navitoclax (Navitoclax). (F) Leukemic burden (% GFP hCD45⁺) was monitored during treatment in the blood of Vehicle- and navitoclax–treated mice. (G) Kaplan-Meier survival curves of mice that received transplantation with mCG2-1 treated either with vehicle or navitoclax. Log-rank Mantel-Cox test was used to determine survival benefit. (H) Leukemic burden (GFP⁺hCD45⁺) was monitored by bleeding in mCG2-1 vehicle-treated mice vs DT2216-treated mice. (I) Kaplan-Meier survival curves of mCG2-1 AMKL model treated either with vehicle or DT2216. Survival benefit was determined with log-rank Mantel-Cox test. P values: ∗P < .05; ∗∗P < .005; ∗∗∗P < .001; ∗∗∗∗P < .0001.

Figure 7.

Combinatorial use of BCL-X_L inhibitors and cytarabine demonstrates greater reduction of leukemic burden in vivo than single-agent treatment. (A) Dose-response curves and IC₅₀ values (supplemental Table 13) determined for all 6 samples of CG2 AMKL, submitted to a viability assay in presence of cytarabine. (Cell-Titer Glo, 6-day incubation, 4 replicates). Viability readout was normalized to DMSO controls for each sample. (B) Schematic overview of experimental design of combinatory treatments with navitoclax and cytarabine (AraC) of mice that received xenotransplantation. (C) Percentage of leukemic blasts in the peripheral blood (% GFP⁺hCD45⁺, left graph), infiltration of the BM (middle graph) and spleen (right graph) was assessed in mice that received transplantation, after 3 weeks of indicated treatments. (D) Percentage infiltration (% GFP⁺hCD45⁺) in the BM of mice that received xenotransplantation and that were treated was compared at the end point between matched Vehicle controls and mice either treated with DT2216 or navitoclax (Navito) (E) as well as their respective combinations with cytarabine (AraC).