Figure 1.

Divergent IL3Rα and βc expression preferentially occurs at the apex of the hematopoietic cell hierarchy and is linked to poor patient survival in AML. A, IL3Rα and βc gene expression ratio in 451 de novo normal karyotype patients with AML, as shown in reads per kilobase per million mapped reads (RPKM; Beat AML; ref. 26). Av, average. B, Kaplan–Meier overall survival curve for normal karyotype patients with AML comparing those with higher or lower than 1.1 ratio of IL3Rα/βc expression in The Cancer Genome Atlas cohort (n = 184; ref. 25). C, Single-cell gene expression of IL3Rα and βc across 11,641 single AML cells from 12 patients (35), with cell types annotated from (32). Mean gene expression is depicted for each AML cell type. D, Hierarchical clustering of primary AML patient samples with high (n = 5) and low (n = 5) IL3Rα/βc ratio [RNA sequencing (RNA-seq)] combining transcriptional (*; RNA-seq) and immunophenotypic (flow cytometry) profiling in the Toronto cohort. E, IL3Rα and βc gene expression in sorted and functionally validated immunophenotypic cellular fractions from CB (36, 37). CMP, common myeloid progenitors; GMP, granulocyte-monocyte progenitors; Gr, granulocytes; HSC, hematopoietic stem cells; MEP, megakaryocyte-erythroid progenitors; MLP, multilymphoid progenitors; Mono, monocytes; MPP, multipotent progenitors. F, Flow cytometric analysis for %IL3Rα^hiβc^lo population of high (n = 5, blue) and low (n = 5, red) IL3Rα/βc transcript ratio AML patient samples (CD3⁻CD19⁻CD45⁺) in the Toronto cohort based on cell-surface IL3Rα and βc protein expression profiles. CB mononuclear cells (n = 3) and mobilized peripheral blood (mPB) mononuclear cells (n = 1) served as controls (ctrl). ns, not significant. G, Correlation of relative abundance of cells with quiescent and primed combined transcriptional phenotypes with %IL3Rα^hi/βc^lo population in CD3⁻CD19⁻CD45⁺ cells with high and low IL3Rα/βc ratio (by RNA-seq) in the Toronto cohort by Pearson analysis. H, Spearman correlation between gene expression and LSC frequency from 88 AML fractions in which specific LSC frequencies were calculated by limiting dilution analysis (LDA) in xenograft assays (20, 38). IL3Rα (IL3RA) and βc (CSF2RB) are highlighted. I, IL3Rα/βc transcript ratio from 138 LSC⁺ (engrafting) and 82 LSC⁻ (nonengrafting) fractions (20). J, Sorting gates for xenotransplanted high/medium (med)/low IL3Rα/βc ratio (CD123 vs. CD131) fractions for patient sample AML#140005 and AML#130578 are shown pregated for viable (SytoxBlue⁻) and CD45⁺CD3⁻CD19⁻ cells and in relation to an in parallel stained G-CSF mobilized peripheral blood control sample (mPB ctrl) from a healthy donor. K, LSC frequencies of high/med/low IL3Rα/βc ratio fractions from AML#140005 and AML#130578 xenotransplanted in limiting dilution into NSG-SGM3 and NSG mice as estimated from CD45⁺CD33⁺ engraftment (>0.1%) at 7 to 8 weeks after transplantation (see also Supplementary Fig. SIQ). Due to low cell numbers retrieved, the low-ratio fraction of AML#130578 was transplanted only into NSG-SGM3. Open triangle and dotted error bar line indicate an estimated 1/LSC frequency and upper limit of 1. Upper and lower estimated limit (error bars) and P values as calculated by ELDA.