NUP98 fusion proteins and KMT2A-MENIN antagonize PRC1.1 to drive gene expression in AML

Abstract

Control of stem cell-associated genes by Trithorax group (TrxG) and Polycomb group (PcG) proteins is frequently misregulated in cancer. In leukemia, oncogenic fusion proteins hijack the TrxG homolog KMT2A and disrupt PcG activity to maintain pro-leukemogenic gene expression, though the mechanisms by which oncofusion proteins antagonize PcG proteins remain unclear. Here, we define the relationship between NUP98 oncofusion proteins and the non-canonical polycomb repressive complex 1.1 (PRC1.1) in leukemia using Menin-KMT2A inhibitors and targeted degradation of NUP98 fusion proteins. Eviction of the NUP98 fusion-Menin-KMT2A complex from chromatin is not sufficient to silence pro-leukemogenic genes. In the absence of PRC1.1, key oncogenes remain transcriptionally active. Transition to a repressed chromatin state requires the accumulation of PRC1.1 and repressive histone modifications. We show that PRC1.1 loss leads to resistance to small-molecule Menin-KMT2A inhibitors in vivo. Therefore, a critical function of oncofusion proteins that hijack Menin-KMT2A activity is antagonizing repressive chromatin complexes.

Keywords: CP: Cancer; CP: Molecular biology; acute myeloid leukemia; chromatin complex; lysine methyltransferase 2A; menin; nascent transcriptomics; non-canonical polycomb repressive complex 1.1; nucleoporin 98-rearrangement; oncogenic fusion protein.

Figure 1.. A CRISPR epigenetic screen reveals that loss of non-canonical PRC1.1 leads to positive selection of NUP98-r leukemia cells

(A) CRISPR screening strategy in NUP98-HOXA9 leukemia cells. (B) Dot plot demonstrating rank-ordered CRISPR screen results. Components of PRC1.1 are highlighted in green, and molecular dependencies are highlighted in red. Full results of the screen are available in Table S1. (C) Heatmap of beta scores for each PRC1.1 complex member for the CRISPR screen described in (A) and (B). (D) Schematic depicting the hypothesized role for epigenetic repression by PRC1.1 upon Menin-KMT2A inhibition. (E) CRISPR competition assay performed in NUP98-HOXA9 isogenic CRISPR cell lines over 14 days (x axis). Ratios of RFP+ sgRNA-expressing cells to RFP− cells are depicted on the y axis. (F) Dose-response curves for VTP50469-treated NUP98-HOXA9 isogenic CRISPR cell lines (top). Images on the bottom show expression of the myeloid differentiation marker CD11b. Error bars represent standard error of the mean (SEM). Data are representative of 3 individual experiments. ns (non-significant) p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 (one-way ANOVA with Dunnett’s post-test for multiple comparisons).

Figure 2.. Inhibition of the Menin-Kmt2A interaction results in the accumulation of PRC1.1 and repressive histone modifications at select loci

(A) ChIP-seq tracks at select genes from NUP98-HOXA9 leukemia cells treated with DMSO or 2 μM VTP50469 for 96 h. Kmt2a and Menin ChIP-seq data from a previously published study (GEO: GSE175596) were used. (B) Gene expression determined by qPCR in NUP98-HOXA9 CRISPR cell lines treated with DMSO or 2 μM VTP50469 for 72 h. (C) Gene expression determined by qPCR in NUP98-HOXA9 CRISPR cell lines expressing sgRNA targeting Pcgf1, treated with a dose curve of VTP54069 for 72 h. Data are representative of 3 individual experiments. Error bars represent SEM. ns (non-significant) p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 (one-way ANOVA with Sidak’s post-test for multiple comparisons).

Figure 3.. NUP98-fps are necessary for maintaining a transcriptionally active chromatin landscape at key pro-leukemogenic loci

(A) Immunoblot of whole-cell lysates from murine leukemia cells immortalized by HA-tagged, FKBP-tagged NUP98-HOXA9. Cells were treated with DMSO or 500 nM dTAG-13 over the indicated time course. (B) Intracellular staining and flow cytometry to detect HA-tagged, FKBP-tagged NUP98-HOXA9 fusion protein in mouse leukemia cells, treated with 500 nM dTAG-13. (C) Volcano plots depicting log2 fold changes in nascent transcription as measured by SLAM-seq in leukemia cell lines immortalized by HA-tagged, FKBP-tagged NUP98-fps: NUP98-HOXA9 (left) and NUP98-JARID1A (right). (D–G) ChIP-seq and ChIP-qPCR from NUP98-HOXA9 cells treated with DMSO or 500 nM dTAG-13 for 24 h. (D) Genome-wide average signal plots (top) and heatmap/tornado plots of transcription start sites ±3 kb (bottom) for each ChIP target. (E) Gene tracks at selected NUP98-HOXA9 target genes for ChIP-seq targets as described above for 24 h. (F) Hockey stick plots depict log2 fold ratio of chromatin occupancy for each ChIP target with genes rank ordered by chromatin occupancy. Select NUP98-fp targets are highlighted. (G) Schematic of the Hoxa9 locus with location of ChIP-qPCR primers (top left). Bar graphs depict ChIP-qPCR data in cells treated with DMSO or dTAG-13 for 24 h. Data are representative of 3 individual experiments. Error bars represent SEM. **p < 0.01 and ****p < 0.0001 (unpaired, two-tailed t test).

Figure 4.. Degradation of NUP98-fp results in the accumulation of PRC1.1 and repressive histone modifications

(A–D) ChIP-seq data in NUP98-HOXA9 leukemia cells treated with DMSO or 500 nM dTAG-13 for 24 or 96 h. (A) Genome-wide average signal plots (top) and heatmap/tornado plots of transcription start sites ±3 kb (bottom). (B and C) Hockey stick plots depict log2 fold ratio of chromatin occupancy for each ChIP target at 96 h with genes rank ordered by chromatin occupancy of PRC1.1 complex members (B) and histone modifications (C). Select NUP98-fp targets are highlighted. (D) Gene tracks at select target genes in NUP98-HOXA9 leukemia cells treated with DMSO or 500 nM dTAG-13 for 96 h. Data are representative of 3 individual experiments.

Figure 5.. PRC1.1 is not required for the initial transcriptional downregulation of pro-leukemogenic genes

(A) qPCR using total mRNA to determine gene expression changes in NUP98-HOXA9 CRISPR cell lines treated with DMSO or 500 nM dTAG-13 for 72 h. (B) qPCR using primers spanning intron-exon junctions to detect nascently transcribed mRNA in NUP98-HOXA9 CRISPR cell lines expressing sgRNA targeting Pcgf1 or the luciferase control treated with DMSO or 500 nM dTAG-13. (C) Volcano plots depicting the log2 fold changes in nascent transcription measured by PRO-seq in NUP98-HOXA9 CRISPR cell lines treated with 500 nM dTAG-3 for 60 min. Select NUP98-fp targets are highlighted. (D) Metagene plots of PRO-seq read density after dTAG-13 (500 nM) treatment for 60 min. (E) qPCR using primers spanning intron-exon junctions to detect nascently transcribed mRNA in NUP98-HOXA9 CRISPR cell lines treated with DMSO or 500 nM dTAG-13 over a time course of 15 min–96 h. Error bars represent SEM. Data are representative of 3 individual experiments. ns (non-significant) p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 (one-way ANOVA with Sidak’s post-test for multiple comparisons).

Figure 6.. Loss of Pcgf1 results in a failure to accumulate PRC1.1, failure to deposit repressive histone modifications, and failure to silence key pro-leukemogenic genes

ChIP-seq data in NUP98-HOXA9 CRISPR cell lines expressing sgRNA targeting Pcgf1 or the luciferase control. Meis1 gene track for cells treated with DMSO or 500 nM dTAG-13 for 96 h (A) or 2 μM VTP50469 for 96 h (B). Data are representative of 3 individual experiments.

Figure 7.. Loss of PRC1.1 leads to resistance to Menin-KMT2A inhibition in a human PDX model of NUP98-r leukemia in vivo

(A) Experimental design for in vivo CRISPR-Cas9 editing of the BCOR locus in a NUP98-JARID1A PDX. (B) Insertion or deletion (indel)-type analysis for PDX cells edited using CRISPR-Cas9 electroporation pre-transplantation and at the time of sacrifice for mice harboring non-targeting (NT) or BCOR sgRNA and treated with control or 0.1% SNDX chow. Red shading indicates +1 indel events, and multicolor coding represents diversification and expansion of indel events, whereby each individual color represents a unique sequencing read and its respective proportion of the whole. (C) Kaplan-Meier survival analysis of NOG mice harboring NUP98-JARID1A PDX electroporated with Cas9 and a NT sgRNA or BCOR sgRNA as described in (A), n = 4–5 mice per group. Shaded yellow box indicates duration of treatment with 0.1% SNDX-5613 chow. Log rank Mantel-Cox test was used to determine significance (p < 0.0001). For pairwise comparisons, the Gehan-Breslow-Wilcoxon test was used to calculate p values using the Bonferroni alpha correction of 0.0083 as the cutoff for significance (NT control chow versus NT SNDX p = 0.0039, NT control chow versus sgBCOR control chow p = 0.0039, NT control chow versus sgBCOR SNDX p = 0.0039, NT SNDX versus sgBCOR control chow p = 0.0039, NT SNDX versus sgBCOR SNDX p = 0.0039, sgBCOR control chow versus sgBCOR SNDX p = 0.2088). (D–G) Burden of disease experiment at day 48 post-injection of a NUP98-JARID1A PDX edited using CRISPR as above (A). Graphs depict disease burden as measured by human CD45 expression in the bone marrow (D) or spleen (E) for each cohort (n = 3 mice per group), as well as expression of the human megakaryocytic markers CD41 (F) and CD61 (G) in the bone marrow of mice. ns (non-significant) p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 (Kaplan-Meier survival analysis; one-way ANOVA with Sidak’s post-test for multiple comparisons).