Mutant H3 histones drive human pre-leukemic hematopoietic stem cell expansion and promote leukemic aggressiveness

Abstract

Our ability to manage acute myeloid leukemia (AML) is limited by our incomplete understanding of the epigenetic disruption central to leukemogenesis, including improper histone methylation. Here we examine 16 histone H3 genes in 434 primary AML samples and identify Q69H, A26P, R2Q, R8H and K27M/I mutations (1.6%), with higher incidence in secondary AML (9%). These mutations occur in pre-leukemic hematopoietic stem cells (HSCs) and exist in the major leukemic clones in patients. They increase the frequency of functional HSCs, alter differentiation, and amplify leukemic aggressiveness. These effects are dependent on the specific mutation. H3K27 mutation increases the expression of genes involved in erythrocyte and myeloid differentiation with altered H3K27 tri-methylation and K27 acetylation. The functional impact of histone mutations is independent of RUNX1 mutation, although they at times co-occur. This study establishes that H3 mutations are drivers of human pre-cancerous stem cell expansion and important early events in leukemogenesis.

Fig. 1

H3.1 mutations alter HSC frequency and hematopoietic differentiation in vitro and in vivo. Number of a granulocyte-macrophage colonies and b macrophage colonies (CFU-M) from CD34⁺ sorted human cord blood transduced with HIST1H3H WT or Q69H. c, d Number of (c) erythroid (BFU-E) and d granulocytic (CFU-G) colonies from CD34⁺ sorted human cord blood transduced with HIST1H3H WT/K27M or HIST1H3F WT/K27I. For a–d n = 4 and is representative of three independent experiments. e Circulating haemocytes in WT larvae (control), larvae expressing H3.3WT and H3.3K27M counted by Neubauer haemocytometer (n = 40). Data represent mean ± standard error of the mean. f Design of in vivo xenotransplantation of transduced CB cells injected into sublethally-irradiated NSG mice. g–n Flow cytometry analysis of populations from the bone marrow of the injected femur of mice xenotransplanted with CD34⁺CD38^- human cord blood cells transduced with HIST1H3H WT/K27M or HIST1H3F WT/K27I after 12–14 weeks. Data is representative of two independent experiments. g Frequency of CD34⁺CD38^- HSPCs in the CD45⁺GFP⁺ population (n = 6). h Frequency of HSC1 (CD45RA⁻CD90⁺CD49f⁺) and i HSC2 (CD45RA^-CD90⁻CD49f⁺) in the CD34⁺CD38⁻ population. j Frequency of CMP (CD135⁺CD45RA⁻), k MEP (CD135⁻CD45RA⁻) and l GMP (CD135⁺CD45RA⁺) in the CD34⁺CD38⁺CD7⁻CD10⁻ population. Data for h–l represents pooled pairs of samples; n = 3. m Frequency of granulocytes (CD33^dim, SSC^high) in the CD45⁺ population and n ratio of CD71⁺ erythroid cells to CD71- erythroid cells in the CD45⁻GlyA⁺ populations (n = 6). o Schematic depicting the changes in frequency of HSCs and the block of differentiation in the erythroid lineage with H3.1 K27M/I mutations. See Supplementary Figs. 1 and 3 for gating strategy used. Data represents mean ± standard deviation. Statistical analysis was performed by two-way Student’s t-tests. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001

Fig. 2

H3.1 K27M/I mutations increase the proliferation of AML cells in vitro and in vivo. a In vitro cell proliferation assay of TEX cells transduced with HIST1H3H WT (blue), HIST1H3H K27M (red), HIST1H3F WT (green), HIST1H3F K27I (pink), and Luc2 control (black) (n = 3, representative of two independent experiments). b Colony formation unit assay of TEX cells transduced with HIST1H3H WT, HIST1H3H K27M, HIST1H3F WT, HIST1H3F K27I, and Luc2 control (n = 4, representative of three independent experiments). c Design of in vivo xenotransplantation of transduced TEX cells mixed equally with untransduced TEX cells into sublethally-irradiated NSG-S mice. d–g Flow cytometry analysis of mice xenotransplanted with TEX cells transduced with HIST1H3H WT/K27M, HIST1H3F WT/K27I, or Luc2 control for 5 weeks. Data is representative of two independent experiments. d Percent GFP⁺ transduced cells in the human (CD45⁺) population 5 weeks post-injection of TEX cells. e–g Percent GFP⁺CD45⁺ engraftment of e the injected femur, f the spleen, and g the contralateral femur 5 weeks post-injection of TEX cells. h Representative image of injected femur and surrounding tissue of mice intrafemorally injected with TEX cells overexpressing HIST1H3H WT (top panel) or HIST1H3H K27M (bottom panel). Flow cytometry analysis indicates the solid mass consists of TEX cells (95.4% human CD45⁺ transduced cells) (n = 5). Data represents mean ± standard deviation. Statistical analysis was performed by two-way Student’s t-tests. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001

Fig. 3

H3.1 K27M/I mutations modify H3K27me3 and H3K27ac marks and alter gene expression in leukemic cells. a Western blot analysis of H3K27me3 levels from total histone protein from TEX cells transduced with HIST1H3H WT/K27M, HIST1H3F WT/K27I, Luc2 control, or untransduced (UT) cells. b Volcano plot depicting genes showing differential expression in HIST1H3H K27M relative to WT. H3K27me3 change of K27M/WT is overlaid as a heatmap, with red and blue representing loss and gain, respectively. Dashed gates indicate genes called as being significantly down- or upregulated, using threshold of |z-score| >1.5, p-value <0.05. c Volcano plot depicting genes showing differential expression in HIST1H3F K27I relative to WT, as in b. d Venn diagram showing overlap of significantly upregulated genes observed in K27M and K27I expressing TEX cells. e Box-whisker plot showing change of promoter-specific H3K27me3 in the TEX cells, comparing all annotated promoters and the subset that showed up-regulation in both K27M and K27I. The lower and upper whisker represents the minimum and maximum, respectively, after removing outliers where the upper whisker = min(max(x), Q3+1.5 × IQR) and lower whisker = max(min(x), Q1−1.5 × IQR), where IQR = Q3-Q1. The two ‘hinges’ are versions of the first and third quartile. The notches extend to ± 1.58 IQR/sqrt(n) representing a confidence interval. IQR stands for interquartile range. Center line indicates median. f, g two-dimensional scatterplot depicting the change of H3K27me3 and H3K27ac at all annotated promoters in f K27M and g K27I overexpressing TEX cells. Corresponding K27M/WT RNA-seq z-score is overlaid as a heatmap, with red and blue representing up- and down-regulation, respectively. h, i Genome browser snapshot of the h GATA1 locus and i LIF locus with RNA-seq and Rx-normalized H3K27me3 and H3K27ac data