RUNX1-ETO and RUNX1-EVI1 Differentially Reprogram the Chromatin Landscape in t(8;21) and t(3;21) AML

Abstract

Acute myeloid leukemia (AML) is a heterogeneous disease caused by mutations in transcriptional regulator genes, but how different mutant regulators shape the chromatin landscape is unclear. Here, we compared the transcriptional networks of two types of AML with chromosomal translocations of the RUNX1 locus that fuse the RUNX1 DNA-binding domain to different regulators, the t(8;21) expressing RUNX1-ETO and the t(3;21) expressing RUNX1-EVI1. Despite containing the same DNA-binding domain, the two fusion proteins display distinct binding patterns, show differences in gene expression and chromatin landscape, and are dependent on different transcription factors. RUNX1-EVI1 directs a stem cell-like transcriptional network reliant on GATA2, whereas that of RUNX1-ETO-expressing cells is more mature and depends on RUNX1. However, both types of AML are dependent on the continuous expression of the fusion proteins. Our data provide a molecular explanation for the differences in clinical prognosis for these types of AML.

Keywords: RUNX1-ETO; RUNX1-EVI1; acute myeloid leukemia; chromatin; epigenetic landscape; transcriptional network.

Graphical abstract

Figure 1

t(3;21) and t(8;21) Are Epigenetically Distinct Types of AML (A) Structure of RUNX1, RUNX1-ETO, and RUNX1-EVI1 with their interacting partners. AA, amino acids; RHD, Runt homology domain; TA, transactivation domain; NHR, nervy homology region; SET, Su(var)3-9 and “Enhancer of zeste”; ZF, zinc finger domain; RD, proline-rich repressive domain; CBP, CREB-binding protein; HDAC, histone deacetylase; CtBP, C-terminal-binding protein; N-CoR, nuclear receptor co-repressor. (B) UCSC genome browser screenshot of DNase-seq and corresponding RNA-seq in two patients with t(3;21) AML, two patients with t(8;21) AML, t(3;21) cell line, and normal CD34⁺ PBSCs at the MEIS1 locus. An enhancer (Xiang et al., 2014) denoted at +140 kb is accessible in t(3;21) AML and normal CD34⁺ PBSCs, but not in t(8;21) AML. (C) Clustering based on the strength of correlation between samples of DNase-seq data from cells of two patients with t(3;21), two patients with t(8;21), two independent CD34⁺ PBSCs, and the t(3;21) SKH-1 cell line. (D) Correlation clustering of RNA-seq data (as in C) from two t(3;21) patients and the SKH-1 cell line with two t(8;21) patients and two normal CD34⁺ PBSCs. (E) DNase-seq profiles spanning 4-kb windows for t(3;21) patient 2, t(8;21) patient 1, and SKH-1 cells. Peaks are ranked from top to bottom in order of increasing relative DNA sequence tag count for peaks identified in t(8;21) patient 1 relative to t(3;21) patient 2. The heatmaps to the right depict the relative expression of genes nearest to each DHS calculated as the ratio of FPKM values for t(3;21) patient 2 (P2) divided by values for t(8;21) patient 1 (P1) or patient 2 (P2). (F) Venn diagram showing the overlap of DNase-seq peaks between t(3;21) patients (both patients combined) and t(8;21) patients (both patients combined). (G) De novo motif discovery in distal DHSs unique to t(3;21) as compared to t(8;21) patients and distal DHSs unique to t(8;21) compared to t(3;21) patients (as shown in F). See also Figure S1.

Figure 2

RUNX1-ETO and RUNX1-EVI1 Bind to Different Sites in Each Type of AML (A) UCSC genome browser screen shot of aligned reads at GATA2 and MEIS1 from ChIP-seq experiments showing binding of RUNX1 (C-terminal antibody) and RUNX1-EVI1 from t(3;21) SKH-1 cells, RUNX1 and RUNX1-ETO from t(8;21) Kasumi-1 cells, and RUNX1 from normal CD34⁺ PBSCs. The boxed element indicates the GATA2 enhancer. (B) Matrix-based depiction of the correlation between ChIP-seq experiments followed by hierarchical clustering. (C) Venn diagram of peak overlap between ChIP-seq for RUNX1-EVI1 in t(3;21) SKH-1 versus RUNX1-ETO in t(8;21) Kasumi-1. Tables depict de novo motif analyses of distal sites bound uniquely by each fusion protein. Gray highlights motifs found uniquely in either RUNX1-EVI1- or RUNX1-ETO-bound sites. (D) Motif enrichment analysis in RUNX1-ETO and RUNX1-EVI1 shared peaks. (E) RUNX1-EVI1 ChIP peaks were ranked according to tag count, and RUNX1-ETO peaks were plotted alongside, together with motifs for the indicated transcription factors. See also Figure S2.

Figure 3

RUNX1 Fusion Proteins Form Part of a Gene Regulatory Network Unique to Each Leukemia (A) UCSC genome browser screen shot of Spi1 showing ChIP-seq data of RUNX1 and RUNX1-EVI1 from t(3;21) SKH-1 cells, RUNX1 and RUNX1-ETO from t(8;21) Kasumi-1 cells, and RUNX1 from normal CD34⁺ PBSCs. (B) GATA2, RUNX1-EVI1, and RUNX1 ChIP-seq in t(3;21) SKH-1 cells. Venn diagram depicting the overlap between GATA2, RUNX1-EVI1, and RUNX1 peaks and the numbers of peaks in each group. White: overlap of GATA2- and RUNX1-EVI1-bound sites; purple: overlap of GATA2 and RUNX1 bound sites; yellow: overlap of RUNX1- and RUNX1-EVI1-bound sites; gray: number of sites bound by all three transcription factors. (C) Transcription-factor-binding motifs enriched in the shared peaks from (B). (D) Hierarchical clustering of enriched motifs discovered in a pairwise comparison between RUNX1 and RUNX1 fusion ChIP-seq peaks between t(3;21) and t(8;21) cells identifying unique peaks for each type of AML. Enrichment score was calculated by the level of motif enrichment in the unique peaks as compared to union of peaks in the pair of experiments. The heatmap depicts the degree of motif enrichment. Two specific sets of enriched motifs unique to each ChIP seq experiment are highlighted: the blue box highlights specifically enriched motifs in RUNX1-bound sites, the green box highlights enriched motifs specific for t(3;21) but not other peaks. (E and F) Bootstrapping analysis of footprinted motifs at RUNX1-EVI1- or RUNX1-ETO-binding sites in patient cells. RUNX1-EVI1-binding sites from the t(3;21) SKH-1 cell line (E) and RUNX1-ETO-binding sites from the t(8;21) Kasumi-1 cell line (F) mapped onto footprints generated from DNase I data of either t(3;21) patient 2 or t(8;21) patient 1, respectively. The heatmap shows the significance of co-localizing footprinted motifs at RUNX1 fusion protein-binding sites for each AML as compared to sampling by chance alone. (G and H) Percentage of Annexin-V-positive cells after 5 days of treatment with a control siRNA (siMM) or with siRNAs specific for RUNX1 and GATA2, respectively, in Kasumi-1 cells (G) and SKH1 cells (H). Each experiment was done at least in triplicate as indicated, and error bars represent SEM. ^∗p < 0.05 and ^∗∗p < 0.01 by unpaired t test. n.s. not significant. See also Figure S3.

Figure 4

Knockdown of RUNX1-EVI1 Results in Loss of the Stem Cell Gene Program (A) Experimental scheme for the siRNA transfection. (B and C) RUNX1-EVI1 siRNA treatment in SKH-1 cells results in reduction in CD34 surface expression. SKH-1 cells after 14 days of either RUNX1-EVI1 or control siRNA transfection were stained with CD34-PE and CD117-APC. (B) Percentage of CD34⁺CD117⁺ cells. (C) Representative flow cytometry plot. Mean of six independent experiments is shown. Error bars represent SEM. ^∗p < 0.05 by paired t test. (D) Growth rates of SKH-1 (dashed lines) and K562 cells (solid lines) treated with RUNX1-EVI1 siRNA relative to treatment with control siRNA. The graph shows mean and SEM values from at least three independent experiments. (E) Hierarchical clustering of gene expression changes as determined by RNA-seq at different time points of treatment. Unsupervised clustering of expression values of genes changing expression 1.5-fold after RUNX1-EVI1 siRNA transfection as compared to control siRNA. Average of two independent replicates. The heatmap color is related to the degree of differential expression (fold change [FC]) between RUNX1-EVI1 siRNA and control siRNA treatment. (F) Percentage and number, respectively (on top of bars), of differentially expressed genes as measured by RNA-seq that are RUNX1-EVI1 ChIP-seq targets. Differentially expressed genes are those with an at least 1.5-fold change in gene expression between RUNX1-EVI1 siRNA as compared to control siRNA treatment. See also Figure S4.

Figure 5

Knockdown of RUNX1-EVI1 Results in Loss of the Expression of Stem Cell Genes and the Upregulation of Myeloid Genes (A–H) RT-PCR analysis of mRNA levels of the indicated genes relative to GAPDH and normalized to untreated cells in SKH-1 cells after RUNX1-EVI1 siRNA as compared to control siRNA transfection. GATA2 (A), MEIS1 (B), HOXA9 (C), CEBPB (D), CSF1R (E), CTSG (F), MPO (G), and CEBPA (H). The graph shows mean and SEM of four independent experiments. n.s., not significant; ^∗p < 0.05; ^∗∗p < 0.01 (unpaired t test). See also Table S1. (I) KEGG pathways highlighting genes and pathways that are upregulated after RUNX1-EVI1 knockdown. (J) KEGG pathways highlighting pathways associated with genes with shared binding of GATA2, RUNX1, and RUNX1-EVI1 that are upregulated after RUNX1-EVI1 knockdown. See also Figure S5.

Figure 6

RUNX1-EVI1 Knockdown Results in Genome-wide Reprogramming of the Epigenome (A and B) UCSC genome browser screen shots showing aligned reads at MPO (A) and MEIS1 (B) depicting DNase-seq and RUNX1, RUNX1-EVI1, GATA2, and C/EBPα ChIP-seq data from SKH-1 after either control siRNA or RUNX1-EVI1 siRNA treatment. (C) Genomic distribution of the indicated factors after transfection with either control or RUNX1-EVI1 siRNA. (D) Average profiles of RUNX1-EVI1, RUNX1, C/EBPα, and GATA2 ChIP-seq reads centered on RUNX1-EVI1 peaks within a 4-kb window after transfection with either control or RUNX1-EVI1 siRNA. (E) Profiles of the DNase-seq and ChIP-seq tag density for the indicated factors in ±4-kb windows centered on DHS for SKH-1 treated with either control or RUNX1-EVI1 siRNA. All peaks were ranked according to the fold change in DNase-seq tag counts between control siRNA and RUNX1-EVI1 siRNA-treated SKH-1 cells. (F) Densities of the indicated motifs underlying the same coordinates plotted within ±1-kb windows around the DHS marked with a blue arrow. See also Figure S6.

Figure 7

C/EBPα DNA Binding Ability Is Critical for the Effects of RUNX1-EVI1 Knockdown (A) Western blot analysis of whole-cell lysates of untreated t(3;21) SKH-1 cells and empty or DNCEBP vector transduced SKH-1 cells, transfected with either control or RUNX1-EVI1 siRNA after 14 days. Sizes (in kDa) are indicated. The blot was probed with EVI1, FLAG, or GAPDH antibodies (loading control) as indicated. (B and C) Flow cytometry of empty and DNCEBP vector transduced SKH-1 cells, untreated or after 14 days of mock, control, or RUNX1-EVI1 siRNA treatment, stained with CD34-APC. (B) Representative histogram with overlay of different treatment conditions. (C) Graph of median fluorescence intensity (MFI) (median) of CD34-APC staining. Bars represent different treatment conditions. Mean of three independent experiments is shown, and error bars represent SEM. n.s., not significant; ^∗p < 0.05 (paired t test). (D and E) RT-PCR analysis of mRNA levels of the indicated genes with and without RUNX1-EVI1 knockdown in the presence and absence of DNCEBP. (D) MPO. (E) HOXA9. mRNA levels relative to GAPDH in either empty vector or DNCEBP vector transduced SKH-1 after either control or RUNX1-EVI1 siRNA transfection (4, 10, or 14 days of treatment). The graph shows mean and SEM of three independent experiments. n.s., not significant; ^∗p < 0.05; ^∗∗p < 0.01 (unpaired t test). (F and G) ChIP-qPCR with chromatin from empty or DNCEBP vector transduced SKH-1 10 days after either control or RUNX1-EVI1 siRNA transfection, using amplicons corresponding to the MPO and SIGLEC1 enhancers, TREM1 and CTSG promoters, the SPI1 3′ upstream regulatory element (URE) enhancer as a positive control, and chromosome 18 as a negative control. (F) C/EBPα ChIP. (G) RUNX1 ChIP. Enrichment was calculated relative to input and IVL. Mean of three independent experiments is shown, and error bars represent SEM. n.s., not significant; ^∗p < 0.05; ^∗∗p < 0.01 (unpaired t test). (H) DNase I accessibility measurement using qPCR validation at MPO and SIGLEC1 enhancer and TREM1 and CTSG promoter. DNase I digestion was performed on empty vector and DNCEBP vector transduced SKH-1 following either control or RUNX1-EVI1 siRNA transfection. The ACTB amplicon was used as negative control. Enrichment was calculated relative to chromosome 18, which is a gene-free region that is DNase I inaccessible and used for normalization. A second independent experiment is shown in Figure S7D. (I) Model depicting the binding sites and transcription factors interacting with RUNX1-ETO and RUNX1-EVI1, respectively, and their independent transcriptional networks that maintain the expression of stem cell/precursor genes but also block the expression of CEBPA. See also Figure S7.