CD34⁺ and CD34⁻ LSCs in CD34⁻ AML are not hierarchically arranged and share near-identical RNA-seq profiles. (A) Mean percentage of cells with immunophenotypes indicated (as a percentage of Lin⁻CD34⁻ MNCs) in engrafting and nonengrafting populations (total of 28 samples injected). The only populations significantly more abundant in engrafting samples were Lin⁻CD34⁻117⁺ and Lin⁻CD34⁻244⁺117⁺. *, P < 0.05; Student’s t test. Error bars represent SEM. (B) Primary engraftment (eight patients). Each symbol represents mean engraftment in one to six mice from the indicated population from one patient. Y axis, mean percentage of human (h)CD45⁺CD33⁺CD19⁻ cell engraftment/total live MNCs. X axis, injected cell fraction. Red dashed line, engraftment threshold. Leukemic engraftment was confirmed by genetic analysis in each case. (C) Primary engraftment experiments where different numbers of cells were injected from sorted patient subpopulations annotated as engrafting (Eng) or nonengrafting (Non-eng). At least one engrafting subpopulation in each patient sample propagates leukemia at cell numbers lower than nonengrafting subpopulations. Each data point represents one injected mouse. Asterisks indicate cell doses where multiple mice were injected with the same cell numbers. The number of asterisks indicates the number of different mice injected. (D) LSC frequency in sorted subpopulations calculated from limit dilution transplant assays. Error bars indicate the calculated 95% confidence interval. Data points without error bars indicate cell fractions where the threshold nonengrafting cell number was not achieved. Here, lowest injected cell numbers are shown (see Table S2 D). (E) Two points are made here: first, the patient’s CD34⁺ and CD34⁻ populations when injected into mice produce both CD34⁺ and CD34⁻ populations; second, the immunophenotype of the patient’s AML is recapitulated in the mouse. Two representative patient samples, #1037 and #875, from experiments with five different patient samples, show the immunophenotype in the patient sample. Blue arrows indicate CD244 and CD117 expression on CD34⁻ AML cells. CD34⁺ (red boxes) and CD34⁻ (blue boxes) subpopulations are flow sorted with purity for injection into mice. (Middle and bottom) Representative FACS analysis plots of engrafted human cells in mice. Mice injected with CD34⁺ cells or CD34⁻ cells. In both cases, engrafting leukemic cells contain both CD34⁺ and CD34⁻ populations and recapitulate the immunophenotype of the patient’s AML (see Table S2 E). (F) Box-whisker plots show percentage of human CD34⁺ leukemic cells in patient samples or primary recipients injected (Inj) with either CD34⁺ (samples from five patients injected into five mice) or CD34⁻ populations (samples from five patients injected into 12 mice). The middle line is the median. Box limits are first and third quartile. Bars indicate minimum and maximum values. (G) Serial engraftment shows no hierarchy between CD34⁺ and CD34⁻ LSCs. X axis, subpopulations from patient samples injected into primary mice; Y axis, mean percentage of human CD45⁺CD33⁺CD19⁻ leukemia engraftment of total live MNCs in secondary mice when populations indicated on the left were injected. Each data point represents engraftment in one mouse (see Table S2 E). (H) RNA-seq data showing log2-fold change (logFC) in expression of all expressed genes (5 CD34⁺ and14 CD34⁻ AML LSC samples on the y axis) plotted against mean log2 cpm (x axis). Statistically significant differentially expressed genes (P < 0.05) are annotated (red, up-regulated in CD34⁺ LSCs; blue, up-regulated in CD34⁻ LSCs). (I) CD34⁺ and CD34⁻ LSCs in CD34⁻ AML generate both CD34⁺ and CD34⁻ populations in primary transplants, each with leukemia-propagating activity in secondary transplants. n, number of patient samples studied per population.