IKAROS and MENIN coordinate therapeutically actionable leukemogenic gene expression in MLL-r acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) remains difficult to treat and requires new therapeutic approaches. Potent inhibitors of the chromatin-associated protein MENIN have recently entered human clinical trials, opening new therapeutic opportunities for some genetic subtypes of this disease. Using genome-scale functional genetic screens, we identified IKAROS (encoded by IKZF1) as an essential transcription factor in KMT2A (MLL1)-rearranged (MLL-r) AML that maintains leukemogenic gene expression while also repressing pathways for tumor suppression, immune regulation and cellular differentiation. Furthermore, IKAROS displays an unexpected functional cooperativity and extensive chromatin co-occupancy with mixed lineage leukemia (MLL)1-MENIN and the regulator MEIS1 and an extensive hematopoietic transcriptional complex involving homeobox (HOX)A10, MEIS1 and IKAROS. This dependency could be therapeutically exploited by inducing IKAROS protein degradation with immunomodulatory imide drugs (IMiDs). Finally, we demonstrate that combined IKAROS degradation and MENIN inhibition effectively disrupts leukemogenic transcriptional networks, resulting in synergistic killing of leukemia cells and providing a paradigm for improved drug targeting of transcription and an opportunity for rapid clinical translation.

Extended Data Fig. 1. Chromatin remodeling complexes modulate the cellular response to therapeutic targeting of MLL-r driven gene expression.

a, Proliferation assay conducted in real-time during the functional genomic screen showing cell number over time. Data represent mean+/−SEM (n=4). b, Volcano plots depicting Wald p-value and beta value calculated using MAGeCK MLE for VTP-50469 and EPZ-5676, comparing the vehicle-treated and drug-treated state on Day 14. c, Gene set testing comparing the behaviour of group 2 (resistance) and group 4 (synthetic lethal) genetic hits between each screen. Family-wise error rate p-value determined by GSEA computational method. d, Expanded panels of CRISPR/Cas9-based competition assays targeting either IKZF1 or MTA2 monitoring sgRNA-RFP expression over time in Cas9-expressing in 4 MLL-r human AML cell lines (MOLM13, MV4;11, OCI-AML2 and THP-1), with and without concurrent treatment with the MENIN inhibitor, VTP-50469. e, CRISPR/Cas9-based competition assays targeting IKZF1 monitoring sgRNA-RFP expression over time in 4 non-MLL-fusion human AML cell lines (IMSM2, OCI-AML3, U937 and HL60) expressing Cas9. IMSM2 and OCI-AML3 carry the NPM1c mutation.

Extended Data Fig. 2. Using CRISPR/Cas9 to genetically target IKZF1 and MTA2 and validate genomic screen findings.

a, Combined analysis of sgRNA depletion across samples, depicted in Extended Data Fig. 1d, showing enhanced sgRNA depletion in the presence of the MENIN-inhibitor, VTP-50469, despite the disadvantage of drug-induced cell cycle arrest. Data represent mean+/−SEM with p-value by two-tail t-test evaluated for n=4 cell lines and n=6 sgRNAs targeting IKZF1 or n=4 sgRNAs targeting MTA2. b, Western blot analysis for MTA2 protein in sgRNA-RFP sorted cells comparing MTA2-targeted bulk cell populations with non-targeting control and ACTIN used as a loading control. c, Western blot analysis for IKAROS protein in sgRNA-RFP sorted cells comparing IKZF1-targeted bulk cell populations with non-targeting control and ACTIN used as loading control. d, Analysis of annexin V staining (apoptosis) and CD11b/CD14 (monocytic differentiation) using flow cytometry 7 days following CRISPR/Cas9-mediated deletion of either IKZF1 or MTA2. Representative experiment is shown. e, Colony forming assay comparing MOLM13 cells with sgRNA targeting Luciferase (Non-Targeting) control versus two different sgRNAs targeting IKZF1 (n =4 per sgRNA), representative colony morphology is shown. f, Relative IC₅₀ for proliferation assays, pertaining to Fig. 1j, comparing MOLM13 cells with sgRNA targeting Luciferase (Non-Targeting control) and two sgRNAs targeting IKZF1 (n=3 per sgRNA) upon treatment with MENIN inhibition. g, Apoptosis assay in MOLM13 and MV4;11 cell lines testing the impact of sgRNAs targeting Luciferase (NT control) versus two sgRNAs targeting IKZF1 on the response to MENIN inhibition with VTP-50469, using doses indicated. Apoptosis was assessed by DAPI exclusion (viability) and annexin V staining with 50,000 cells analysed per sample. Representative example of multiple independent experiments.

Extended Data Fig. 3. IMiDs effectively target IKAROS protein for degradation in human MLL-r AML.

a, Western blot analysis for IKAROS, CK1α and MENIN protein following treatment of MV4;11 human MLL-r AML cell line for 5 hours with increasing doses of THAL, LEN, POM and CC220, using ACTIN as a loading control. b, Western blot analysis for IKAROS and CK1α following treatment of OCI-AML2 human MLL-r AML cell line for 5 hours with increasing doses of THAL, LEN, POM and CC220, using ACTIN as a loading control. c, LEN and CC220 dose-response curves on: Day 9 of treatment for the NPM1-mutant (NPM1c) IMSM2 human AML cell line cell (Absolute IC₅₀ 267.0 nM [95% CI 189.6 to 381.0] for LEN, 17.3 nM [95% CI 9.3 to 37.8] for CC220) and Day 18 of treatment for the NPM1-mutant (NPM1c) OCI-AML3 human AML cell line (Absolute IC₅₀ 4626 nM [95% CI could not be calculated] for LEN, 57.7 nM [95% CI 33.1 to 101.5] for CC220). Data represent mean+/−SEM with absolute IC50 indicated (n=3). d, Scatterplot for MS determination of IMiD substrates 5 hours after drug treatment in the MV4;11 and OCI-AML2 cell lines. Data represent the log-fold change in abundance and log10(p-value). p-value determined by moderated t-test as implemented by the Bioconductor Limma package.

Extended Data Fig. 4. IKAROS degradation perturbs diverse cellular pathways.

a, RNA-seq dot-plot with linear regression showing z-score for differential gene expression correlating LEN (5 μM) and CC220 (1 μM) treatment for 3 days. b, RNA-seq volcano plot for MV4;11 treated with LEN for 6 days. c, GSEA results for “oncogenic signatures” for RNA-seq data on Day 3. Dot plot is shown of the log10 false discovery rate (FDR) q-value and normalized enrichment score (NES) as determined by GSEA computational method. d, GSEA results for “Hallmarks” signatures. Dot plot is shown of the log10 false discovery rate (FDR) q-value and normalized enrichment score (NES) as determined by the GSEA computational method. e, Bar code plots created using the GSEA tool for putative MLL-fusion target genes with selected genes from leading edge analysis indicated with adj-p-value<0.05 in MOLM13 or MV4;11. Normalized enrichment score and family-wise error rate p-value as determined by GSEA computational method. f, Gene expression z-score heatmap for selected genes within immune regulatory and inflammatory pathways under treatment with CC220 (n=3) compared to DMSO-treated control (n=3). g, Venn diagram for genes displaying > 2-fold change in expression following CC-220 treatment comparing cell lines. Gene number and percentage overlap is indicated. Selected genes deregulated in both cell lines are indicated. h, Correlation of IKAROS-bound gene TSSes, determined by ChIP-seq, with RNA-seq changes. Bar code plot showing gene expression changes following treatment with CC220 (3 days) for the top 500 genes bound by IKAROS (by peak enrichment over background). Selected genes from leading edge analysis indicated. Normalized enrichment score and family-wise error rate p-value determined by GSEA computational method. i, Correlation of IKAROS-bound enhancers, determined by ChIP-seq and the ABC prediction tool, with RNA-seq gene expression changes. Bar code plots display gene expression changes following treatment with CC-220 (3 days) for the top 500 genes predicted to be regulated by IKAROS-bound enhancers (by peak enrichment over background at enhancers). Normalized enrichment score and family-wise error rate p-value determined by GSEA computational method. Venn diagram for overlap between gene expression change and genes predicted to be regulated by the top 1000 IKAROS-bound enhancers.

Extended Data Fig. 5. IKAROS CUT&RUN motif foot printing.

a, IKZF1 DNA-binding motifs. b, Foot printing for CTCF protein over the CTCF motif for each cell line; used as a positive control for motif detection. Result for CUT&RUNTools de novo DREME motif detection is shown with highly significant discovery of the known CTCF motif. MEME and E-value is indicated. c, CUT&RUNTools foot printing for IKAROS and IgG control over the IKZF1 motif in the MOLM13 cell line. Cut site probability distribution as determined by CUT&RUNTools. d, CUT&RUNTools foot printing for IKAROS over the JUN:FOS motif (centrally bound), CEBPα motif (centrally bound), SPI/Pu.1 motif (centrally bound) and the RUNX motif (non-centrally bound), as indicated, in the MOLM13 cell line. Cut site probability distribution as determined by CUT&RUNTools.

Extended Data Fig. 6. Combined targeting of MENIN and IKAROS.

a, Annexin V staining (apoptosis) by flow cytometry following 6 days treatment with VTP-50469, LEN, CC220 and combination (MOLM13). b, Cell cycle analysis: histogram for DAPI stain and bar graphs displaying cell cycle stage (%) are shown for each condition. Representative experiment shown. (MOLM13) c, CD11b/CD14 expression measured by flow cytometry 6 days following treatment with VTP-50469, LEN, CC220 and combination (MOLM13). Representative experiment shown. d, Dose-response curves for LEN and CC220, with and without VTP-50469 (8 nM), compared to VTP-50469 alone (dotted line) measured by Cell-Titer Glo in MV4;11 for synergy studies. e, Chou-Talalay synergy analysis for VTP-50469 in combination with LEN or CC-220 (MV4;11). Chou-Talalay Combination Index (CI) plots (left panel) and normalized isobolograms (right panel) are shown for each IMiD-VTP-50469 combination (day 6). Line of additivity is shown (red). f, CC-220 and MENIN inhibitor dose response curves (day 6) with and without overexpression of non-degradable IKAROS (Q146H) in MOLM13. Data represent mean+/−SD with non-linear regression curve fit shown. g, IMiD and MENIN inhibitor apoptosis after 6 days treatment, with and without overexpression of a wild-type IKAROS or non-degradable IKAROS mutant (Q146H) in MOLM13. h, Scatterplots for mass spectrometry (MS) results on MOLM13, MV4;11 and OCI-AML2 cell lines after 5 days pre-treatment with VTP-50469 and then treatment with THAL, LEN, POM and CC-220 (iberdomide), for 5 h. Data represent the log-fold change in abundance and log10(p-value). p-value determined by moderated t-test as implemented by the Bioconductor Limma package. i, BH3-profiling in MV4;11 under drug treatments indicated. Cytochrome C (Cyt C) release was measured. Data represent mean+/−SEM (n=3). j, Western blot analysis for BCL2, MCL1 and BCL-XL protein following treatment of OCI-AML2 human MLL-r AML cell line for 72 hours with LEN, EPZ-5676, VTP-50469 and each combination, using GAPDH as a loading control.

Extended Data Fig. 7. MENIN and IKAROS display overlapping functions in regulation of gene expression.

a, Number of differentially expressed genes (greater than 2-fold change and adjusted p-value <0.05) from RNA-seq data for MOLM13 and MV4;11 cell lines treated with CC220 1 μM (n=3), VTP-50469 50 nM (n=3) and the combination (n=3) for 3 days. b, Heatmap showing DESeq2 statistical z-score for all genes deregulated under treatment with VTP-50469 alone across all other treatment groups. c, Heatmap of DESeq2 statistical z-score for MENIN and IKAROS co-regulated gene expression. Showing genes with shared or additive regulation. Selected genes are indicated. d, Cell surface expression of FLT3 as measured by flow cytometry in the MOLM13 cell line. Cells were pre-treated for 2 days with CC220 or DMSO control and then VTP-50469 was added for 24 hours. e, CRISPR screen beta scores from MAGeCK MLE comparing sgRNA representation at day 0 compared to day 14 in the DMSO-treated control samples for the MOLM13 cell line. Negative beta scores for MYC, FLT3, RPA3 (common essential gene), RPS14, and BCL2 are indicated. f, Gene expression for IKZF1 as determined by RNA-seq in MOLM13 and MV4;11, indicating log-2-fold change value and adjusted p-value determined using DESeq2. g, Western blot analysis for IKAROS protein in the MOLM13 cell line following 5 days treatment with either VTP-50469 or iberdomide/CC220, followed by rescue treatment with bortezomib (BTZ) for 6 hours, using GAPDH as a loading control. Performed in the presence of the broad-spectrum caspase inhibitor, QV-D-OPH, to prevent cell death. h, Tornado plots depicting global IKAROS chromatin binding at TSSes, as determined by CUT&RUN in the MOLM13 cell line.

Extended Data Fig. 8. IKAROS protein interactome in MLL-r AML.

a, Binary overlap percentage between all IKAROS peaks enriched 5-fold over background, genome-wide, with each of the other 3 factors, MEIS1 (at 15FE), MENIN (at 5FE) and MLL1 (at 5FE) according ChIP-seq data. b, IGV tracks depicting binding of IKAROS, MENIN, MLL1 and MEIS1, as determined by ChIP-seq, in the region of the TNF gene. Promoters and enhancers, predicted from the ABC tool, are indicated with genomic location indicated (Hg19). c, IGV tracks depicting binding of IKAROS, MENIN, MLL1 and MEIS1 at the MYC and FLT3 genes as determined by ChIP-seq. Promoters and selected enhancers, predicted from the ABC tool, are indicated with genomic location indicated (Hg19). d, Schematic diagrams of protein domains and experimental workflow of BioID system. IKAROS, MEIS1, HOXA10 and ZNF692 proteins were tagged with the biotin ligase BirA*, the left side of the protein schematic diagram denotes N-terminal tagging and the right-side, C-terminal. MV4;11 and MOLM13 cells expressing BioID constructs, 8 total cell lines, were cultured separately. ZF=Zinc finger domain, HBD=homeobox domain, LFQ=label free quantitation, LC-MS/MS=liquid chromatography coupled tandem mass spectrometry. e, Western Blot analysis of the expression of BirA* and HA tagged fusion proteins in MV4;11 and MOLM13 cells, using total Histone H3 as a loading control. f, Venn diagram displaying the number of proteins identified in label free quantitation analysis of the MOLM13 IKAROS, MEIS1 and HOXA10 BioID LC-MS/MS data. All proteins included in Venn diagram have a >2-fold enrichment over the ZNF692 control. g, List of proteins common to IKAROS, MEIS1, HOXA10 BioID identified in LFQ analysis having >2-fold enrichment over the ZNF692 control in the MV4;11 and MOLM13 cell lines.

Extended Data Fig. 9. IKAROS and MENIN co-reside within an isolatable protein complex.

a, IGV ChIP-seq tracks for IKAROS, MENIN, MEIS1 and MLL1 at selected gene TSSes. Tracks are derived from the same ChIP-seq experiments depicted in Fig. 6 and Extended Data Fig. 8. b, Direct protein Co-IP using IKAROS as the bait and probed for MENIN, HOXA9, MEIS1, MTA2 and IKAROS in the MV4;11, OCI-AML2 and MOLM13 cell lines using high salt nuclear extraction and detergent wash. c, Direct protein Co-IP using MLL1 and MENIN as the bait and probed for MENIN and IKAROS, in the MV4;11, OCI-AML2 and MOLM13 cell lines using high salt nuclear extraction and detergent wash. d, Direct protein Co-IP using IKAROS as the bait and probed for MENIN, MLL1, IKAROS and MTA2 in the MV4;11 and THP-1 cell lines using benzonase-treated nuclear extract and high glycerol wash. e, Direct protein Co-IP using MLL1 (rabbit antibody) and MENIN (rabbit antibody) as the bait and probed for MLL1, MENIN and rabbit IKAROS antibody (rIKAROS), in the MV4;11 and THP-1 cell lines using benzonase-treated nuclear extract and high glycerol wash. Rabbit immunoglobulin is seen overlying the IKAROS Western blot bands (indicated in the figure) due to the use of an anti-rabbit secondary antibody. f, Direct protein Co-IP using IKAROS as the bait, with and without prior VTP-50469 treatment (250 nM for 48 hr), and probed for MENIN, MTA2, and IKAROS, in the THP-1 cell line.

Extended Data Fig. 10. Combination therapy with VTP-50469 and LEN results in additive anti-leukemic activity in vivo.

a, Immunohistochemistry for human CD45 antigen on bone marrow sections (sternum) on PDX/CBAM-68552 following 3 weeks treatment with vehicle, LEN 50 mg/kg daily, VTP-50469 0.1% rodent diet and the drug combination. b, Analysis of differentiation in circulating leukaemia cells using CD11b and CD14 expression assessed by flow cytometry in CBAM-68552 after 3 weeks drug treatment. Percentage of double-positive cells is indicated. Representative examples from individual mice are shown. c, Assessment of apoptosis in human CD45-positive cells from bone marrow of drug-treated mice using the CBSK-17D model following 2 weeks drug treatment. The percentage of total viable, human cells are indicated. Representative examples from individual mice are shown. d, Quantitation of apoptotic cells from bone marrow of mice transplanted with the CBSK-17D PDX model following 2 weeks drug treatment. Data represent mean+/−SEM with p-value by unpaired two-tail t-test. e, RNA-seq heatmap from PDX/CBAM-68552 following 3 weeks treatment with vehicle, LEN 50 mg/kg daily, VTP-50469 0.1% rodent diet and the drug combination. Heatmap displaying RNA-Seq DESeq2 statistical z-score for the co-regulated gene network between VTP-50469 and LEN f, RNA-seq heatmap displaying DESeq2 statistical z-score for genes detected in the “granulocyte” pathway as additively deregulated using QIAGEN Ingenuity Pathway Analysis (IPA) analysis. g, Bar code plots using the gene set for HOXA9 regulated genes under each drug treatment from RNA-seq in vivo using the PDX/CBAM-68552 model. Normalized enrichment score and family-wise error rate p-value determined by the GSEA computational method. h, Gene set testing for reported MLL-fusion gene targets for LEN and VTP-50469. Family-wise p-value and normalized enrichment score (NES) determined by the GSEA computational method. i, IPA upstream regulator analysis. Graph displays selected activation z-scores for detected pathways with p-value <0.05. j, Measurement of peripheral blood circulating human CD45 positive cells from PDX mice transplanted with the DFAM-16835 PDX model (NPM1c) after two weeks of drug treatment in vivo. Data represent mean+/−SEM with p-value determined using unpaired, two-tail t-test.

Figure 1.. IKZF1/IKAROS and MTA2 exhibit synthetic lethal interaction with pharmacologic inhibition of MENIN and DOT1L.

a, Functional genetic screen schematic: cells treated with EPZ-5676 (1 μM), VTP-50469 (500 nM) or vehicle. b, Drug targeting of MENIN (VTP-50469) and DOT1L (EPZ-5676) disrupts MLL-fusion complex function. c, Volcano plot depicting Wald p-value and beta value calculated using MAGeCK MLE for the VTP-50469 and EPZ-5676 combined analysis comparing vehicle-treated (n=2) and drug-treated states (n=4) on Day 14. d, MAGeCKFlute 9-square correlation plot using the beta value calculated between Day 0 and Day 14 for vehicle-treatment as compared to drug treatment (VTP-50469 and EPZ-5676 screens combined). Regions corresponding to pathway (group1), resistance (group2) and synthetic lethal (group4) genetic hits are indicated. e, Leading edge analysis from GSEA showing overlapping genetic hits for each screen considered separately. f, CRISPR/Cas9-based competition assays targeting either IKZF1 or MTA2 monitoring sgRNA-RFP expression over time in Cas9-expressing MOLM13 AML cell line +/− VTP-50469. g, Proliferation assay on sorted, sgRNA-RFP-expressing cells +/− VTP-50469 or EPZ-5676. Data represent mean+/−SEM (n=3) with p-value calculated using paired, two-tail t-test. h, Cell morphology assessed using light microscopy (100X) 7 days following CRISPR/Cas9-mediated deletion of IKZF1 or MTA2, compared to non-targeting sgRNA control i, Colony forming assay comparing MOLM13 cells with sgRNA targeting Luciferase (Non-Targeting) control versus two different sgRNAs targeting IKZF1 (n =4 per sgRNA). Data represent mean +/− SEM. p-value determined by unpaired, two-tailed t-test. j, Proliferation assay comparing MOLM13 cells with sgRNA targeting Luciferase (NT control) and two sgRNAs targeting IKZF1 (n=3 per sgRNA) upon treatment with VTP-50469 for 5 days. Data represent mean +/− SEM. k, Schematic for in vivo, MOLM13 xenograft experiment: 200,000 MOLM13 cells with sgRNA targeting Luciferase (NT control) versus two different sgRNAs targeting IKZF1 (n = 5 per group) were injected via tail vein at Day 0, treatment began with VTP-50469 chow at Day 9, and mice were sacrificed at Day 18 when they became ill. l, MOLM13 xenograft experiment, burden of disease and differentiation markers in spleen and bone marrow after 9 days of VTP-50469 treatment. Data represent mean +/− SEM. p-value determined by unpaired, two-tailed t-test.

Figure 2.. IMiDs effectively target IKAROS protein for degradation and show therapeutic efficacy in MLL-r AML.

a, Western blot analysis for IKAROS protein following treatment of the MOLM13 cell line for 5 h with increasing doses of THAL, LEN, POM and CC220, using GAPDH as a loading control. b, Proliferation assay measuring cell number over time for the MV4;11 and MOLM13 cell lines treated with vehicle-control (DMSO), THAL, LEN, POM and CC220. Data represent mean+/−SD (n=3). c, Dose response curves for MV4;11 and MOLM13 cell lines using THAL, LEN, POM and CC220. Data represent mean+/−SD (n=6) with non-linear regression curve fit shown. Percentage indicated as a proportion. d, Calculated absolute IC50 and 95% confidence intervals for MOLM13, OCI-AML2 and MV4;11 AML cell lines, assessed by proliferation assay at Day 18. e, Scatterplot for MS determination of IMiD substrates 5 h after indicated drug treatment in the MOLM13 cell line. Data represent the Log-fold change in abundance and log10(p-value). p-value determined by moderated t-test as implemented by the Bioconductor Limma package. f, Schematic depicting generation of MOLM13 cell lines with IKAROS overexpression constructs treated with DMSO, LEN (5 uM), and CC220 (1 uM) for 5 hours. g, Western Blot showing IKAROS overexpression constructs, h, Proliferation assay showing rescue of antiproliferative effects with overexpressed, non-degradable IKAROS (Q146H). Data represent mean +/− SEM. p-value determined by paired, two-tail t-test.

Figure 3.. IKAROS is a core transcriptional regulator in MLL-r AML.

a, Volcano plots displaying RNA-seq data for MOLM13 (n=3) and MV4;11 (n=3) cell lines treated with iberdomide/CC220 1 μM for 3 days. Selected genes are indicated. b, Bar code plots depicting GSEA results for HOXA9-regulated genes, stem cell genes and myeloid differentiation in the MOLM13 cell line. Normalized enrichment score and family-wise error rate p-value determined by GSEA computational method. c, Heatmap depicting gene expression z-scores for canonical MLL1-fusion target genes and selected HOXA9-regulated genes following 3 days treatment with CC220 compared to DMSO-control in MOLM13 cell line. d, Percentage of ChIP-seq IKAROS peaks with 5-fold enrichment over background at regulatory regions, as indicated. e, Venn diagram depicting overlap between all TSS-proximal IKAROS peaks with genes within the MLL-fusion and HOXA/MEIS1 network. f, CUT&RUN de novo DREME motif detection results for the MOLM13 and MV4;11 cell lines. Top 6 motifs detected for IKAROS, in order of statistical significance are shown. Motif, statistical E-value and predicted motif identity are indicated.

Figure 4.. Combined targeting of MENIN and IKAROS results in synergistic induction of apoptosis and cooperative deregulation of gene expression.

a, Proliferation assay measuring cell number over time for the MV4;11 and MOLM13 cell lines following drug treatment. Data represent mean+/−SD (n=3). b, Analysis of annexin V staining (apoptosis) following drug treatment. Viable cell percentage indicated. Representative experiment shown. c, Proliferation assay in MOLM13 cells treated with VTP-50469 combined with either LEN or CC220. Synergy assessment by isobolgrams (left) and Chou-Talalay (right) d, Proliferation assay after 6 day treatment in human CD34 progenitor cells that were non-transformed versus CD34 progenitor cells with MLL-AF9 transformation. Data represent mean+/−SEM (n=3) relative to DMSO control. e, BH3-profiling in MOLM13 cells following drug treatments. Cytochrome C (Cyt C) release was measured. Data represent mean+/−SEM (n=3). f, Volcano plots displaying RNA-seq data for MOLM13 and MV4;11 cells following 3 days of drug treatment (n=3 per condition). g, Venn diagram for overlapping transcriptional changes between CC220, VTP-50469 and the combination (total number of genes indicated with >2-fold change and p-adj<0.05 as determined by DESeq2 package). h, GSEA results querying the Broad “oncogenic signatures” comparing combination treatment with VTP-50469 monotherapy. Results are displayed as dot plot of the log10 false discovery rate (FDR) q-value and normalized enrichment score (NES). i, Bar code plots depicting selected GSEA results in MOLM13 cells from (h). Depicts additive effect of the combination compared to VTP-50469 therapy alone. Normalized enrichment score and family-wise error rate p-value determined by GSEA computational method. j, BCL2 and MYC gene expression assessed by RT-qPCR following treatment with CC220 1 μM, VTP-50469 250 nM and the combination. Data represent mean+/−SEM for combined anlaysis of 3 AML cell lines (MOLM13, MV4;11 and OCI-AML2) with n=3 replicates each (N=3; n=9). p-value determined by paired, two-tailed t-test. k, Western blot for BCL2 protein following treatment of MOLM13 and MV4;11 cells for 72 h with CC220 1 μM, VTP-50469 50 nM and in combination, using ACTIN as a loading control. l, Western blot for MYC and GATA2 protein following drug treatment of MOLM13 cells for 72 hours. Histone H3 as a loading control.

Figure 5.. Inhibition of the MENIN-MLL1 protein-protein interaction leads to proteasomal degradation of IKAROS protein.

a, Bar code plots depicting the effect of MENIN-inhibition on the set of genes deregulated following IKAROS degradation by CC220 according to RNA-seq data on Day 3. Normalized enrichment score and family-wise error rate p-value determined by the GSEA computational method. b, Western blot analysis for IKAROS protein following treatment of the MV4;11 and MOLM13 cell lines for 1, 4, 24 and 72 hours, using ACTIN as a loading control. c, Western blot analysis for IKAROS protein following treatment of MOLM13 human MLL-r AML cell line for 72 hours with LEN, VTP-50469 and in combination, using GAPDH as a loading control. d, IKAROS protein chromatin binding at the IKZF1 motif displayed as the cut-site probability +/−100bp around the IKZF1 motif under each drug treatment, as indicated.

Figure 6.. Chromatin co-occupancy and proximal protein-protein interactions of IKAROS and the MLL-fusion network.

a, Circos plots displaying co-occupancy between IKAROS, MEIS1, MENIN and MLL1 determined by ChIP-seq at all TSSes and active enhancers bound by at least one factor, genome wide. b, Tornado plots depicting chromatin occupancy as determined by ChIP-seq from the MOLM13 cell line at the indicated regulatory regions of interest: IKAROS-bound enhancers and MENIN-bound TSSes. c, IGV tracks depicting binding of IKAROS, MENIN, MLL1 and MEIS1 at the BCL2 gene. The promoter and a selected intragenic enhancer are indicated (genomic location indicated in Hg19). The predicted enhancer-promoter connection is indicated (red line). d, Dot plots displaying PSM rank values from spectral counting on proteins identified in IKAROS, MEIS1 and HOXA10 BioID in MV4;11 vs. MOLM13 cell lines by LC-MS/MS. Notable proteins ranked in the top 50 are circled and labelled on the left of each plot in order of lowest rank to highest (n=2). e, Dot plots displaying the Log₂ abundance ratios of IKAROS/ZNF692, MEIS1/ZNF692 and HOXA10/ZNF692 BioID in MV4;11 vs. MOLM13 cell lines derived from label free quantitation analysis of LC-MS/MS data (n=2). f, Venn diagram displaying the number of proteins identified in label free quantitation analysis of the MV4;11 IKAROS, MEIS1 and HOXA10 BioID LC-MS/MS data. All proteins included in Venn diagram have a >2-fold enrichment over the ZNF692 control. g, Direct protein Co-IP using IKAROS as the bait and probed for MLL1, MENIN, MEIS1, MTA2 and LAMINB1 for a negative control, in the MV4;11 cell line. h, Direct protein Co-IP using MLL1 and MENIN as the bait and probed for MENIN, MLL1 and IKAROS, in the MV4;11 cell line.

Figure 7.. Combination therapy with VTP-50469 and LEN results in synergistic anti-leukemic activity in vivo.

a, Four PDX models used indicating PDX identification number (ID), MLL-fusion present and additional molecular features. b, Peripheral blood circulating human leukaemia cells indicated by human CD45 positive cells using flow cytometry during treatment with vehicle, LEN 50 mg/kg daily, VTP-50469 continuously and the drug combination in CBAM-68552. Data represent mean+/−SEM with pair-wise p-value calculated using the two-tail t-test. c, Analysis of bone marrow from transplanted mice following 3 weeks (CBAM-68552), 2 weeks (CBSK-17D) or 1 week (CPCT-0007) of drug treatment for leukaemia burden (human CD45 +ve), as assessed by flow cytometry. Data represent mean+/−SEM with p-value calculated using the two-tail t-test (n=4–5 mice in each group). d, Volcano plot depicting RNA-seq results from human leukaemia cells isolated from CBAM-68552 mouse bone marrow following in vivo treatment with LEN (n=4) compared to vehicle control (n=5). Selected genes are indicated. e, Kaplan-Meier Survival curves for CBAM-68552 following 4 weeks treatment with vehicle, LEN 50 mg/kg/day, VTP-50469 0.1% continuously; p-value by Log-Rank (Mantel-Cox) test. f, Kaplan-Meier Survival curves for CPCT-0007 following 2 weeks treatment with vehicle, LEN 50 mg/kg/day, LEN 100 mg/kg/day, VTP-50469 0.033% continuously, or the combination of LEN 50 mg/kg/day with VTP-50469 0.033%; p-value by Log-Rank (Mantel-Cox) test. g, Kaplan-Meier Survival curves for CBSK-17D following 2 weeks treatment with vehicle, LEN 50 mg/kg/day, VTP-50469 0.033% continuously; p-value by Log-Rank (Mantel-Cox) test. 3 combination-treated mice were sacrificed after 350 days without evidence for leukemia by flow cytometry of bone marrow (†) h, Kaplan-Meier Survival curves for DFAM-16835 following 2 weeks treatment with vehicle, LEN 50 mg/kg/day, VTP-50469 0.033% continuously; p-value by Log-Rank (Mantel-Cox) test.