Sequential CRISPR gene editing in human iPSCs charts the clonal evolution of myeloid leukemia and identifies early disease targets

Abstract

Human cancers arise through the sequential acquisition of somatic mutations that create successive clonal populations. Human cancer evolution models could help illuminate this process and inform therapeutic intervention at an early disease stage, but their creation has faced significant challenges. Here, we combined induced pluripotent stem cell (iPSC) and CRISPR-Cas9 technologies to develop a model of the clonal evolution of acute myeloid leukemia (AML). Through the stepwise introduction of three driver mutations, we generated iPSC lines that, upon hematopoietic differentiation, capture distinct premalignant stages, including clonal hematopoiesis (CH) and myelodysplastic syndrome (MDS), culminating in a transplantable leukemia, and recapitulate transcriptional and chromatin accessibility signatures of primary human MDS and AML. By mapping dynamic changes in transcriptomes and chromatin landscapes, we characterize transcriptional programs driving specific transitions between disease stages. We identify cell-autonomous dysregulation of inflammatory signaling as an early and persistent event in leukemogenesis and a promising early therapeutic target.

Keywords: AML; IRAK1 inhibitor; IRAK4 inhibitor; UBE2N inhibitor; clonal evolution; gene editing; hematopoietic stem/progenitor cells; inflammatory response; innate immunity; leukemogenesis.

Figure 1.. HSPCs generated from gene-edited iPSCs phenotypically mimic the clonal evolution of AML

(A) Schematic overview of the generation and phenotypic characterization in vitro and in vivo of isogenic clonal iPSC lines generated through CRISPR-Cas9-mediated gene editing. Stepwise introduction of mutations was performed to generate isogenic single-, double-, triple-, and quadruple-mutant iPSCs. Three independent lines per genotype were established after detailed genetic characterization (see also Table S1, Figure S1, and STAR methods for details) and subjected to hematopoietic differentiation for phenotypic characterization and engraftment assays in NSG and NSG-3GS mice. (B) Fraction of CD34⁻/CD45⁺ cells (i.e., hematopoietic cells that have lost CD34 expression upon maturation) on day 14 of hematopoietic differentiation. Mean and SEM of values from 5–12 independent differentiation experiments with two different iPSC lines from each genotype are shown (see also Figure S2A). (C) CD34⁺/CD45⁺ fraction of all CD45⁺ cells on day 14 of hematopoietic differentiation. Mean and SEM of values from 5–12 independent differentiation experiments with two different iPSC lines from each genotype are shown (see also Figures S2B and S2C). (D) Fraction of CD11b+ cells (i.e., more mature myeloid cells) on day 14 of hematopoietic differentiation. Mean and SEM of values from 3 or 4 independent differentiation experiments from one line per genotype are shown. (E) Wright-Giemsa staining of representative cytospin preparations of hematopoietic cells derived from the indicated iPSC lines after 14 days of hematopoietic differentiation culture. The P, A, and SA panels show mainly differentiated myeloid cells (granulocytes and monocytes), whereas the SAR panel shows immature myeloid progenitors with blast morphology. Scale bars, 25 μm. (F) Number of colonies obtained from 5,000 cells seeded in methylcellulose assays on day 14 of hematopoietic differentiation. Mean and SEM of 2 to 4 independent methylcellulose experiments using two different iPSC lines from each genotype are shown (see also Figure S2D). (G) Competitive growth assay. The cells were mixed 1:1 at the onset of hematopoietic differentiation with an iPSC line stably expressing GFP derived from the parental line. On days 2–12 of differentiation, the relative population size was estimated as the percentage of GFP⁻ cells (calculated by flow cytometry) at each time point relative to the population size on day 2. Mean and SEM of 1–4 independent experiments using two different iPSC lines from each genotype are shown (see also Figure S2E). (H) Cell counts of HSPCs at the indicated days of liquid hematopoietic differentiation culture. Mean and SEM of 2 or 3 independent differentiation experiments with one or two different iPSC lines from each genotype are shown. (I) Levels of human engraftment in the BM of NSG and NSG-3GS mice 13–15 weeks after transplantation with HSPCs derived from one iPSC line per genotype (P, N-2.12; A, A-1; SA, SA-2; SAR, SAR-1). Error bars show the mean and SEM of individual mice. For SAR, values from 26 NSG and 5 NSG-3GS mice from 5 independent transplantation experiments are included. (J) Representative flow cytometry panels assessing human cell engraftment and lineage markers in the BM of recipient mice 13–15 weeks post-transplantation. (K) Wright-Giemsa-stained cytospin of BM cells of a recipient mouse transplanted with SAR hematopoietic cells. Scale bar, 25 μm. (L) Schematic summary of all in vitro and in vivo phenotypic analyses P, parental; A, ASXL1^C-truncation; SA, SRSF2^P95L/ASXL1^C-truncation; SAR, SRSF2^P95L/ASXL1^C-truncation/NRAS^G12D. See also Figures S1–S3 and Tables S1 and S2.

Figure 2.. Transcriptional and chromatin accessibility landscapes of gene-edited iPSC-HSPCs capture those of primary human cells and show leukemic features

(A) Principal-component analysis (PCA) of RNA-seq data. (B) PCA of ATAC-seq data. (C) Heatmap showing expression of genes preferentially expressed in primary AML HSPCs relative to normal HSPCs from healthy BM (van Galen et al., 2019), showing marked upregulation in SAR and SARF stages. (D) Heatmap showing expression of a set of myeloid differentiation genes derived from normal primary cells (van Galen et al., 2019), showing marked downregulation in SAR and SARF. (E) ssGSEA analysis of data from individual patients with “chromatin-spliceosome” AML against gene expression signatures from A, SA, and SAR iPSC-HSPCs, as indicated. (F) ssGSEA analysis of data from individual patients with splicing factor mutant MDS against gene expression signatures from A, SA, and SAR iPSC-HSPCs, as indicated. (G) Heatmap showing Pearson correlation values of normalized read counts for ATAC-seq peaks that overlap between our clonal evolution stages (P, A, SA, SAR, SARF) and primary normal hematopoietic cell subpopulations (hematopoietic stem cell [HSC], multi-potent progenitor [MPP], common myeloid progenitor [CMP], and granulocyte-monocyte progenitor [GMP]) from Corces et al. (2016). See also Figure S4.

Figure 3.. Dynamic gene expression and chromatin accessibility changes during clonal evolution

(A) Bar plot showing the number of differentially expressed genes (DEGs; log₂ fold change [log₂FC] > 1, FDR-adjusted p < 0.05) in the indicated comparisons. Shaded areas represent DEGs found to also be differentially expressed in a previous stage. (B) Bar plot showing the number of differentially accessible peaks (DAPs; FDR-adjusted p < 0.05) in the indicated comparisons. Shaded areas represent DAPs found to also be differentially accessible in a previous stage. (C) Scatterplots of gene expression changes (log₂FC) showing correlation between the two comparisons indicated in the x and y axes of each plot. Significant changes (FDR-adjusted p < 0.05) in the x axis comparisons are highlighted in blue. The ρ values represent Spearman correlation values calculated from the log₂FC of genes with significant changes in the x axis comparison. (D) Scatterplots of chromatin accessibility changes (log₂FC) showing correlation between the two comparisons indicated in the x and y axes of each plot. Peaks with a significant change (FDR-adjusted p < 0.05) in the x axis comparisons are highlighted in red. The ρ values are Spearman correlation values calculated from the log₂FC of peaks with significant changes in the x axis comparison. (E) Venn diagrams showing overlap of up- and downregulated genes in the indicated comparisons with selected genes annotated. (F) Heatmap showing the expression of the 10 upregulated genes that are common in all comparisons (A versus P, SA versus P, SAR versus P, and SARF versus P, shown in the Venn diagram in E, left panel) across stages. Marked in red are genes involved in inflammatory responses and innate immunity. (G) Venn diagrams showing overlap of opening or closing peaks in the indicated comparisons. Selected genes associated with the indicated groups of peaks are annotated. (H–L) Heatmaps showing expression of the indicated gene sets (adhesion molecules, genes involved in Rho GTPase signaling, interleukin signaling, MHC class II genes, and RUNX1 target genes) across stages. (M) Summary dynamics of transcriptional changes in selected gene sets, as indicated. The bar corresponding to each stage is colored by the median log-transformed gene expression level in each gene set. See also Figures S5 and S6.

Figure 4.. Characterization of transcriptional programs underlying distinct stage transitions

(A) TF motif enrichment in the indicated stage transitions. Opening or closing peaks in each stage were compared with the total atlas, and TF motif enrichments were calculated using the one-sided Kolmogorov-Smirnov (KS) test. The KS test effect size is shown on the y axis, and the percentage of peaks associated with the TF motif is plotted on the x axis. The dashed lines indicate TF motifs with a KS test effect size ≥ 0.05. The top six and bottom six motifs are marked in blue and red, respectively. (B) TF motif enrichment dynamics across stages. Differentially accessible peaks in each stage compared with P were compared with the total atlas and TF motif enrichments were calculated using the one-sided KS test. The KS test effect size is shown on the y axis. (C) Six stage-specific groups of ATAC-seq peaks defined on the basis of their patterns of chromatin accessibility across the four stages P, A, SA, and SAR. The number of peaks for each group is annotated. Left panel: schematic representation. Right panel: tornado plots of the annotated stage-specific groups of ATAC-seq peaks in the indicated stages. (D) TF motif enrichment in the peaks of the four indicated stage-specific groups from A, containing numbers of peaks sufficient for motif enrichment analysis. The hypergeometric test was used to compare the enrichment of proportions of TF motifs for a stage-specific group (foreground ratio) versus those for the total atlas (background ratio). The top 20 most highly enriched TFs are shown. The red dashed vertical line indicates a Bonferroni-corrected p value threshold of 0.05. AP-1 TF family motifs are marked in light blue, NF-κB family motifs in dark blue, CEBP family motifs in orange, RUNX family motifs in purple, GATA family motifs in red, and MECOM motifs in green. (E) Summary of TF motif enrichment for the four stage-specific groups indicated (containing numbers of peaks sufficient for motif enrichment analysis), on the basis of the data shown in (D). Genes inferred to be putative relevant targets of these TFs (on the basis of the analyses presented in Figures 3, 4, S5, S6, and S7A–S7C) are shown underneath, connected with arrows to the putative TFs. (A dashed arrow connects RUNX to ribosome biogenesis genes, as this link is inferred from the literature rather than from our data, which could not document enrichment for RUNX motifs in this gene set.)

Figure 5.. Identifying early AML targets

(A) Breakdown of all peaks accessible at the SAR stage: A-SA-SAR peaks first become accessible at the A stage. SA-SAR peaks first become accessible at the SA stage. SAR-specific peaks only become accessible at the SAR stage. (B) Diamond plots showing the accessibility change (log₂FC) of the peaks belonging to the SA-SAR group together with the expression change (log₂FC) of genes associated with them. Each row represents one gene and all peaks associated with it across all comparisons (A versus P, SA versus P, SAR versus P, and SARF versus P), allowing backward and forward tracking of changes in successive stages. (C) Strategy for identification of “AML early” genes. From the 2,247 chromatin regions (ATAC-seq peaks) that were accessible in SAR, we selected the 258 that first became accessible at the SA stage and remained accessible in SAR (SA-SAR peaks). One hundred fifty-three genes could be associated with these 258 SA-SAR peaks. Of those, we selected 13 genes whose expression (RNA-seq) was upregulated in both SA and SAR stages compared with P (p_adj < 0.1). The 13 genes are listed ranked by accessibility change in SAR versus P (see also Table S4). Genes involved in inflammatory responses are shown in red font. (D–F) Boxplots of ATP6V0A2 expression (DESeq2-normalized counts) in PB and BM of patients from the BeatAML cohort (Tyner et al., 2018) grouped in de novo, secondary and “chromatin-spliceosome” AML, compared with that in healthy individuals. p values were calculated using a Wilcoxon rank-sum test. (G) Left panel: viability determined by cell counts of P, SA, and SAR iPSC-HSPCs treated with AraC (200 nM) compared with untreated cells. Right panel: viability determined by cell counts of P, SA, and SAR iPSC-HSPCs treated with the V-ATPase inhibitor bafilomycin A1 (Baf A1; 10 nM) compared with untreated cells. All groups were tested in triplicate. See also Figure S7 and Tables S3 and S4.

Figure 6.. ATP6V0A2 inhibition in primary AML cells

(A) BM blasts from a “chromatin-spliceosome” AML patient (AML-32) were used to derive iPSC lines capturing both an early AML clone (AML-32.18, harboring only IDH2, ASXL1, and SRSF2 mutations) and a late AML clone (AML-32.13, harboring a CEBPA mutation in addition to the IDH2, ASXL1, and SRSF2 mutations). (B) Treatment of HSPCs derived from the iPSC lines shown in (A) with 200 nM AraC (left) or 10 nM bafilomycin A1 (right). AraC preferentially kills the later clone, while bafilomycin A1 has comparable efficacy in both clones. (C) Viability of primary AML blasts from patient AML-32, treated with AraC or bafilomycin A1, determined by cell counts, relative to DMSO-treated cells. (D) Changes in the clonal composition of the primary AML-32 patient cells after treatment with AraC or bafilomycin A1, assessed by measurement of the change in VAFs of selected mutations present in the sample. Left: decrease in the VAF of the clonal SRSF2 P95H mutation. Right: decrease in the VAF of the subclonal CSF3R T618I mutation. (E) BM blasts from a “chromatin-spliceosome” AML patient (AML-34) were treated with AraC (250 nM), bafilomycin A1 (100 nM), or DMSO for 4 days. (F) Viability of primary AML blasts from patient AML-34, treated with AraC or bafilomycin A1, determined by cell counts, relative to DMSO-treated cells. (G) Decrease in the VAF of the clonal SRSF2 P95H mutation (left) and of the subclonal CSF3R T618I mutation (right) in primary AML blasts from patient AML-34 after treatment with AraC or bafilomycin A1, as indicated. See also Figure S7.

Figure 7.. Inhibition of inflammatory signaling targets both early and late disease clones

(A) Relative expression of selected NF-κB target genes in A, SA, and SAR, compared with P, iPSC-HSPCs, after stimulation with 100 ng/mL LPS for 3 h (upper panels) or 10 ng/mL IL-1β for 30 min (lower panels). Values shown correspond to expression in LPS- or IL-1β-treated relative to untreated (UT) cells. Mean and SEM of three to five independent differentiation experiments with one line per genotype are shown. (B) Schematic of experimental design to test the effects of the IRAK1/4 inhibitor NCGC1481 and UBE2N inhibitor UC764865 in iPSC-HSPCs. (C and D) Number of methylcellulose colonies from P, A, SA, and SAR iPSC-HSPCs treated with the IRAK1/4 inhibitor NCGC1481 or the UBE2N inhibitor UC764865 at the indicated doses (0.1, 0.5, 1, and 2 μM), relative to untreated (PBS) cells. Mean and SEM of three or four (for concentrations 0.5, 1, and 2 μM) or one or two (for 0.1 μM) independent experiments with one line per genotype are shown. See also Figure S7.