SUPPLEMENTAL INFORMATION

A KMT2A-AFF1 regulatory network highlights the role of core factors and reveals the regulatory logic of key downstream target . Joe R. Harman1,†, Ross Thorne1,†, Max Jamilly2, Marta Tapia1,3, Nicholas T. Crump1, Siobhan Rice1,4, Ryan Beveridge1,5, Edward Morrissey6, Marella F.T.R. de Bruijn1, Irene Roberts4,8, Anindita Roy4,8, Tudor A. Fulga2,7, Thomas A. Milne1,8*

Supplemental items: Supplemental Fig S1 (Page 2). KMT2A-N binding profiles and input tracks in SEM and patient data. Supplemental Fig S2 (Page 3). Comparing the SEM KMT2A-AFF1 GRN with RS4;11 KMT2A-AFF1, and SEM DOT1L and BRD4 networks. Supplemental Fig S3 (Page 4). Patient sub-network analysis highlights core transcription factors. Supplemental Fig S4 (Page 5). KMT2A-AFF1 and RUNX1 binding shows similarities in KMT2A-AFF1 ALL and KMT2A-MLLT3 AML models. Supplemental Fig S5 (Page 6). Comparison of KMT2A-AFF1 GRN nodes with published CRISPR essentiality screens. Supplemental Fig S6 (Page 7). KMT2A-AFF1 cooperates with RUNX1 in FFL and cascade circuits to regulate downstream targets. Supplemental Fig S7 (Page 8). CRISPR screen in combination with venetoclax treatment to test GRN predicted circuits. Supplemental Fig S8 (Page 9). KMT2A-AFF1 and RUNX1 cooperate to regulate CASP9 in a cascade motif. Supplemental Table S1 (Page 10). List of antibodies. Supplemental Table S2 (Page 11). List of primers. Supplemental Table S3 (Page 12). GEO accession numbers for previously published sequencing experiments. Supplemental Data S1. KMT2A-AFF1 and RUNX1 network annotated edge and node tables. Supplemental Data S2. Patient sub-network node clusters. Supplemental Data S3. Nascent RNA-seq results and PANTHER pathway enrichment. Supplemental Data S4. KMT2A-AFF1 – RUNX1 regulatory FFL and cascade circuits. Supplemental Data S5. Results of CRISPR screen in combination with venetoclax treatment in THP-1 cells. Supplemental Code. Jupyter Notebook files of custom scripts used in the study.

1 Figure S1

KMT2A-N ChIP-seq Log reads at TSS i. Gene loci with peak calls “specific” to patient #1 A e B 8 spearman=0.73 spearman=0.74 20 kb hg19 Chr12 12.74 Patient #1 KMT2A-N

6 12.74 Patient #1 Input

KMT2A-N SEM 4 12.74 Patient #2 KMT2A-N

4 5 6 4 5 6 7 12.74 Patient #2 Input KMT2A-N patient #1 KMT2A-N patient #2 BAZ2A C 10 kb hg19 Chr5 12.56 Patient #1 KMT2A-N KMT2A-N ChIP-seq 12.56 Patient #1 Input SEM Patient #1 Patient #2 12.56 Patient #2 KMT2A-N

12.56 Patient #2 Input

i. MAT2B

ii. Gene loci with peak calls “specific” to patient #2 10 kb hg19 Chr1 14.5Patient #1 KMT2A-N

14.5Patient #1 Input

14.5Patient #2 KMT2A-N

ii. 14.5Patient #2 Input

GFI1 5 kb hg19 Chr5 9.27Patient #1 KMT2A-N

9.27Patient #1 Input -3 0 3 -3 0 3 -3 0 3 Distance from peak center (Kb) 9.27Patient #2 KMT2A-N

9.27Patient #2 Input 0 100 200 300 i). Peaks specific to patient #1 ii). Peaks specific to patient #2 MRPL36

D 100 kb hg19 Chr7 67.29 100 kb hg19 Chr6 25.29 100 kb hg19 Chr9 411.01 RUNX1 ChIP-seq RUNX1 ChIP-seq RUNX1 ChIP-seq

95.02 89.14 12.46 KMT2A-N ChIP-seq KMT2A-N ChIP-seq KMT2A-N ChIP-seq

67.29 25.29 12.46 RUNX1/KMT2A-N Input RUNX1/KMT2A-N Input RUNX1/KMT2A-N Input

2.55 3.68 2.55 RUNX1/KMT2A-N Input RUNX1/KMT2A-N Input RUNX1/KMT2A-N Input

1589.84 1125.55 1547.64 ATAC-seq ATAC-seq ATAC-seq

ARID1B GNAQ LAT2 CLIP2 RFC2

Supplemental Fig S1. KMT2A-N binding profiles and input tracks in SEM and patient data. (A) Scatter plot showing Loge KMT2A-N reads at TSS in SEM cells compared with patient samples #1 and #2. (B) ChIP-seq tracks generated in hg19 for KMT2A-N, reads normalized to 1×107 reads. Loci are KMT2A- AFF1 bound genes unique to patient #1 (i) or patient #2 (ii), as defined by peak calling as shown in Fig. 1B. (C) Heatmaps of KMT2A-N ChIP-seq signal in SEM cells, patient #1 and patient #2. Heatmaps generated over KMT2A-N peaks with peak calls unique to patient #1 (i) or patient #2 (ii). (D) ChIP-seq tracks generated in hg19 for KMT2A-N, RUNX1, and input . Input tracks are shown scaled to maximum signal at the , or matching track height in RUNX1 or KMT2A-N (whichever is lowest). ATAC-seq is also included to show accessible regions. ChIP-seq and ATAC-seq data normalized to 1×107 reads. Displayed loci are KMT2A-AFF1 and RUNX1 targets.

2 Figure S2

A B CAPN15 RS4;11 (siMARS) SEM (siMA6) RS4;11 KMT2A-AFF1 network 1.00 NT

KD ARRDC4 GFI1 ETS1 0.75 SPIB HMGA2 FOXP1 GABPB1 0.50 NFKB1 RUNX1 ELF1 FOXO1 KMT2A-AFF1 0.25 LMO2 SMAD9 % Input relative to NT JUN ZEB2 NR3C1 0.00 Degree LEF1 MAZ +23 RUNX1 CDK6 +23 RUNX1 CDK6 centrality 100 200 MYB KMT2A-AFF1 enhancer KMT2A-AFF1 peak peak

C D E RS4;11 siMARS DOT1L network BRD4 network KMT2A MAZ MIR186 ETS2 NFYB ELF1 STAT1 YY1 MYC SNAI3 MYC 3644 202 128 NFYA MAZ JUNB SNAI1 BCL6 MXI1 1000 ID1 YY1 SREBF1 SPI1 RS411 SP1 SPI1 DOT1L ELK4 ELF4 ARNT NFYC MYC MYB BRD4 RUNX1 E2F6 PATZ1 RUNX1 SREBF1 NFE2L1 ELF2 SEM ARNT LMO2 MAZ 500 GTF2I EGR1 ATF5 ELF4 CLOCK LMO2 Non-differential after IRF7 KMT2A-AFF1 KD TFDP1 NRF1 Differentially expressed after KMT2A-AFF1 KD 0 Degree Degree centrality centrality Degree connectivity (SEM KMT2A-AFF1 GRN) 1000 2000 3000 2000 4000 6000

F Inhibitor treatment G KMT2A-AFF1 DOT1L H IBET-151 DOT1Li KMT2A 300 MAZ ELF1 2965 1363 1493 MYC 1000 NFYA 200 alue) of v

RUNX1 KMT2A-AFF1 (P 100 BRD4 10 LMO2 500 ARNT network overlap −log

2576 5596 0

0 935 Degree connectivity (SEM KMT2A-AFF1 GRN) Non-differential KMT2A-AFF1 - BRD4 Differentially expressed KMT2A-AFF1 - DOT1L

Supplemental Fig S2. Comparing the SEM KMT2A-AFF1 GRN with RS4;11 KMT2A-AFF1, and SEM DOT1L and BRD4 networks. (A) KMT2A-N ChIP qPCR data in RS4;11 and SEM cells after KMT2A-AFF1 KD, normalized to input chromatin and displayed relative to NT control, at the RUNX1 +23 enhancer and an KMT2A-AFF1 peak at the CDK6 locus. siMA6 and siMARS refers to SEM and RS4;11 specific MLL-AF4 siRNA, respectively. (B) Top 20 genes of the RS4;11 KMT2A-AFF1 GRN by degree centrality. Lines indicate predicted interaction from to gene locus, with arrowheads pointing towards downstream nodes. (C) Overlap of nodes between SEM and RS4;11 KMT2A-AFF1 GRN models. (D) DEGs after KMT2A-AFF1 KD in RS4;11 cells (n=3, FDR <0.05), plotted against degree centrality in SEM KMT2A-AFF1 GRN. (E) Top 20 genes of the DOT1L (left) and BRD4 (right) GRN models by degree centrality. Lines indicate predicted interaction from protein to gene locus, with arrowheads pointing downstream. (F) DEGs upon 1.5 hours’ IBET or 7 days’ EPZ-5676 treatment, plotted against degree centrality in SEM KMT2A-AFF1 GRN (n=3). (G) Overlap of nodes between KMT2A-AFF1, DOT1L and BRD4 GRNs. (H) Statistical significance of overlaps shown in (G). Significance calculated using a Fisher’s exact test and displayed as -log10(P value).

3 Figure S3

A B C

600000 AML histogram GO: Biological process NFYA KMT2A enrichment

3.75 log2 TPM 400000 MYC reg. hematopoietic differentiation RUNX1 ELF1MAZ cellular response to stress RBPJ leukocyte mediated immunity 200000 Frequency neutrophil mediated immunity response to cytokine

0 ARNT IL-1−mediated signaling pathway

0 5 10 15 ZNF143 reg. of HSC differentiation cellular response to hypoxia ALL histogram cytokine−mediated signaling pathway 600000 SEM and patient samples IKZF2 S phase 1.97 log2 TPM translational initiation KMT2A-AFF1 target gene overlap

400000 proteasomal protein catabolic process 0 500 1000 M phase Frequency Degree connectivity of mitotic M phase 200000 nearest annotated cellular response to DNA damage organization 0 regulation of cell cycle 0 5 10 15 All cell cycle SEM & patient sample #1 signal transduction

400000 FBM histogram cell adhesion 1.65 log TPM SEM & patient sample #2 2 cell junction organization SEM specific 1 2 3 4 5

200000 Patient subgraph clusters Frequency

Fold Enrichment -log10 P-Value

0 1 2 3 4 5 2 3 4 5 0 5 10 15

Log2(TPM)

D Clust 1 Clust 4 1.0

0.5 expression Mean binary 0.0 HSC LMPP PreProB Bcell MPP ELP ProB

Supplemental Fig S3. Patient sub-network analysis highlights core transcription factors. (A) Histograms of log2 TPM expression across AML, ALL and FBM datasets. Red line indicates mean log2 TPM for each dataset, the threshold used to define active expression as used in generating patient- specific sub-networks (see Figure 2). (B) Overlap of SEM KMT2A-AFF1 bound genes with genes bound in patient #1 and #2, as in Figure 1B, against connectivity within the SEM KMT2A-AFF1 GRN. (C) GO Biological process enrichment for patient subgraph clusters. Size of points represents -log10 FDR of enrichment, while point color represents fold enrichment over expected. (D) Bar plots of FBM cell populations showing mean binary expression in clusters 1 and 4.

4 Figure S4

A B KMT2A-N ChIP-seq target genes 10 kb hg19 Chr6 11.57 SEM KMT2A-N

11.57 THP1 KMT2A-N

5611 2359 1190 NFYA 5 kb hg19 Chr16 9.79 SEM KMT2A-N

9.79 THP1 SEM THP1 KMT2A-N KMT2A-FP target in SEM only C Degree vs KMT2A-FP target status in MAZ THP1 and SEM cells KMT2A 100 kb hg19 Chr21 ELF1 MAZ 90 SEM KMT2A-N

MYC 90 NFYA THP1 KMT2A-N ality

r 1000 RUNX1 10 kb hg19 23.13 SEM KMT2A-N RUNX1 LMO2

Degree cent ARNT 500 23.13 CLOCK EGR1 THP1 KMT2A-N STAT1 MYC 50 kb hg19 Chr13 15.48 SEM KMT2A-N 0 15.48 Both None SEM onlyTHP1 only KMT2A-FP target in both SEM & THP1 THP1 KMT2A-N

ELF1 D RUNX1 target overlaps E

5 kb hg19 Chr16 32.41 SEM RUNX1 12345 6604 1254 32.41 THP1 RUNX1 in SEM only RUNX1 target MAZ SEM genes THP1 genes

50 kb hg19 Chr1 F GRN Degree vs RUNX1 target 32.38 SEM RUNX1 overlaps THP1 vs SEM KMT2A 32.38 ELF1 MAZ THP1 RUNX1 MYC GFI1 EVI5 NFYA 100 kb hg19 Chr21 ality 1000 r 90 SEM RUNX1

90 THP1 RUNX1 RUNX1 LMO2 RUNX1 Degree cent 500 ARNT GTF2I RUNX1 target in both SEM & THP1 EGR1 STAT1 CLOCK

0 Both None SEM onlyTHP1 only

Supplemental Fig S4. KMT2A-FP and RUNX1 binding show similarities in KMT2A-AFF1 ALL and KMT2A-MLLT3 AML models. (A) Overlap of KMT2A-N bound genes in THP-1 and SEM cells. (B) ChIP- seq tracks generated in hg19 for KMT2A-N in SEM and THP-1 cells. Reads normalized to 1×107 reads. Displayed loci are KMT2A-FP bound genes unique to SEM, or common to both SEM and THP-1. (C) Association of Log2 degree centrality with KMT2A-N bound gene overlaps in THP-1 and SEM cells as shown in (A). (D) Overlap of RUNX1 bound genes in THP-1 and SEM cells. (E) ChIP-seq tracks generated in hg19 for RUNX1 in SEM and THP-1 cells. Reads normalized to 1×107 reads. Displayed loci are RUNX1 bound genes unique to SEM, or common to both SEM and THP-1. Dashed box highlights an SEM specific RUNX1 peak. (F) Association of Log2 degree centrality with RUNX1 target overlaps in THP-1 and SEM cells as shown in (D).

5 Figure S5

A Project Score - Sanger Institute ELF1 essentiality MAZ essentiality Sarcoma Hematopoietic & Lymphoid Colon/Colorectal Cancer DiFi Brain Cancer

0 1 2 3 0 1 2 3 MYC essentiality RUNX1 essentiality Sarcoma RKN Hematopoietic & Lymphoid OPM2 Colon/Colorectal Cancer

Brain Cancer GI-1 −4 −2 0 2 0 1 2 3 - Bayesian score (Hit < 0)

B DepMap - Broad Institute ELF1 essentiality MAZ essentiality Sarcoma Hematopoietic & Lymphoid Colon/Colorectal Cancer

Brain Cancer

−0.5 0.0 0.5 −1.0 −0.5 0.0 0.5 MYC essentiality RUNX1 essentiality Sarcoma Hematopoietic & Lymphoid Colon/Colorectal Cancer SEM

Brain Cancer

−3 −2 −1 0 −1 0 Dependency score (Hit < 0)

Supplemental Fig S5. Comparison of KMT2A-AFF1 GRN nodes with published CRISPR essentiality screens. (A) and (B) CRISPR essentiality screen for RUNX1, MYC, MAZ and ELF1 from (A) the Project Score database of the Sanger Institute Cancer Dependency Map (Behan et al. 2019), and (B) the Avana 21Q1 dataset from the Broad Institute Cancer Dependency Map (Meyers et al. 2017; Doench et al. 2016). For Project Score, inverted Bayesian show essential hits below 0, and for the Avana dataset CERES scores show essential hits below 0.5. Each datapoint represents a different cell line, with essentiality highlighted in red. Cell lines were categorized into the type of cancer that they model.

6 Figure S6

A B C Reactome pathways 3 1. NT RUNX1 RUNX1 3. 2. siRNA siRNA siRNA 200 5 2 4. 5. 6. 7. (FDR) P-value 10 10 8. 100 logFC 0 1 9. 10. -log −log

Fold enrichment −log (FDR) NT 10 1.5 1.5 4.0 0 siRNA 0 2.5 0 5 0 4 8 12 0 1 2 3 4 logFC Mean logCPM Fold enrichment

1. DNA 4. Activation of HOX genes during diff. 7. Epigenetic regulation of gene exprs. 9. RUNX1 regulates transcription of 2. PRC2 methylates histones and DNA 5. Cell Cycle, Mitotic 8. RNA Polymerase I genes involved in differentiation of HSCs 3. DNA Damage/Telomere Stress 6. Cellular Senescence Transcription 10. Telomere Maintenance Induced Senescence

D E F KMT2A-N ChIP-seq 10 kb hg19 Chr1 at KMT2A-AFF1 peaks KMT2A-AFF1 KD 7.4 48h RUNX1-NT 4 ● 6 Direct ● ● RUNX1 ●

● ● 7.4 Indirect ● ChIP-seq ● ● 48h RUNX1-KD KD) 2 ● ● ● ● ● ● ● ● 5 ● log FC 0 9.3 48h RUNX1-NT RUNX1- (

● ● ● ● ● ● KMT2A-N Log2 read count 4 ● ● ● ● ● ● ● −2 ● ● 9.3 ● ChIP-seq 48h RUNX1-KD ● spearman=0.89 ●

4 5 6 GFI1 Log2 read count (RUNX1-NT) RUNX1 KD

● 5.0 ● G RNA-seq - BCL2 RNA-seq - CEBPG ●

● ● ● ● ● ● ● ● 2.5 ● ● ● ● 600 ● 200 ●

log FC 0.0 400 CPM ● 100 ● ● ● ● ● ● ● −2.5 ● 200 ●

● ● 0 0 NT KD NT KD NT KD NT KD KMT2A-AFF1 KD RUNX1 KD KMT2A-AFF1 KD RUNX1 KD EPZ-5676 FDR = 3.34 × 10-7 FDR = 1.62 × 10-6 FDR = 0.003 FDR = 0.001

● ● ● ● ● H RNA-seq - BID ● 2.5 ● 125 55 RUNX1 10 kb 100

log FC 20 0.0 75 KMT2A-N

CPM 50 20 AFF1-C

−2.5 ● 25 25 H3K27ac 0 NT KD NT KD KMT2A-AFF1 KD RUNX1 KD BID (C3-FFL) FDR = 1.64 × 10-9 FDR = 0.016

Supplemental Fig S6. KMT2A-AFF1 cooperates with RUNX1 in FFL and cascade circuits to regulate downstream targets. (A) Volcano plot showing the relationship between -log10 P value and |logFC| in RUNX1 KD nascent RNA-seq. Orange points represent genes |logFC| ≥ 1 and FDR ≥ 0.05; red represents |logFC| < 1 and FDR < 0.05; green represents |logFC| ≥ 1 and FDR < 0.05. (B) MA plot showing the relationship between logFC and mean log2 CPM in RUNX1 KD nascent RNA-seq. DEGs (FDR < 0.05) are highlighted in red. (C) Pathway enrichment (Reactome) for overlap between KMT2A-AFF1 and RUNX1 KD DEGs (Figure 5B). Size of points represents -log10 FDR of enrichment, while point color represents fold enrichment over expected number of genes. (D) Violin and boxplots showing logFC expression response to KMT2A-AFF1 KD, RUNX1 KD, or EPZ-5676 . Data split between genes in the direct KMT2A-AFF1 GRN (genes directly bound by KMT2A-AFF1) and the indirect GRN (genes not bound by KMT2A-AFF1). (E) Reference-normalized ChIP-seq tracks generated in hg19 for KMT2A-N and RUNX1 after 48 hours’ RUNX1 KD. Reads normalized to 1×107 reads. (F) Scatter plot showing KMT2A-N reads at KMT2A-AFF1 peaks following 48 hours’ treatment with NT or RUNX1 siRNA. (G) Expression of BCL2 and CEBPG from nascent RNA-seq data following KMT2A-AFF1 or RUNX1 KD. Expression normalized as CPM. (H) Left - Expression of BID from nascent RNA-seq data following KMT2A-AFF1 or RUNX1 KD. Expression normalized as CPM. Right - ChIP-seq tracks generated in hg19 for KMT2A-N, AFF1-C, RUNX1 and H3K27ac at BID locus. Reads normalized to 1×107 reads.

7 Figure S7

A B C necrotic late

Apoptosis gene LentiCRISPRv2–TKOv3 PI

plasmid library SSC-A FSC-H

live early AF1 P FSC-A BCL2L1 BID CASP10 TF CASP8 BCL2 CASP6 TP53 CASP7 CASP3 A CASP9 T -6 Virally transduce & AnnexinV–FITC select THP-1 cells RUNX1 KMT2A-AFF1 ELF1 T 0 + 7.4 μM venetoclax MYC 100 MYB MEF2C E2F6 T 18 80 IKZF2 60 No interaction Harvest gDNA, library prep, and sequence TF - gene interaction 40 THP-1 IC50 = 7.4 μM

20 % live cells (Annexin-V/PI) relative to DMSO-only control 0 100 101 102 103 104 105 [VEN] nM

D E TKOv3 library CRISPR validation + Venetoclax 100 T0 2 80 R =0.80 Dropouts Enriched genes

aligned reads 14M 60 coverage 200x 40 3 replicate 2 undetected sg 0.4% 20 skew ratio 3.5 0 2 density 100 VEN 2 80 R =0.69

60 relative to AAVS1 1

40 replicate 2 0 200 400 600 800 Viability over DMSO (CTG) read count 20 0 0

0 20 40 60 80 100 VS1 BAX XIAP A CUL2 MUL1

replicate 1 LGMN A CASP9 KMT2A PSMD2 *CASP3 TKO-BAX

Supplemental Fig S7. CRISPR screen in combination with venetoclax treatment to test GRN predicted circuits. (A) Example interaction matrix showing GRN predicted TF regulation of the apoptosis pathway. Red squares indicate a predicted regulatory interaction between the TF and apoptosis gene. (B) Experimental outline of CRISPR screen performed in combination with 7.4 µM venetoclax treatment in THP-1 cells. Cells harvested for sgRNA counting and analysis at T0 (after transduction and selection) and T18 (18 days treatment with venetoclax). (C) Above – Flow cytometry gating for Annexin-V/PI staining of THP-1 cells to assay viability with venetoclax treatment. Below – THP-1 cells were cultured for 48 hours with different concentrations of venetoclax, and assayed for viability. PI: propidium iodide; early/late: early/late apoptotic. Dots show mean; error bars show SD. Lines show unconstrained four parameter sigmoidal least-squares regression best fit; shaded region shows 99% confidence interval. For all curves, R2 > 0.99. (n=5). (D) Left – Sequencing validation of amplified sgRNA pool showing sgRNA distribution as a histogram and boxplot. Right – Scatter plot of sgRNA counts for biological replicates of T0 and T18+VEN samples. Pearson correlation between replicates shown as R2, and histograms distributions shown along the axis. (E) Functional validation of CRISPR screen hits by individual gene knockouts using sgRNA from the Brunello library. sgRNA were cloned into lentiCRISPRv2 constructs and transduced into THP-1 cells, followed by 48 hours’ treatment with DMSO or 20 µM venetoclax. CellTiter Glo (CTG) was used to assay viability, and is normalized as Venetoclax/DMSO viability, relative to AAVS1 silent control. Genes marked * are FDR > 0.1 in the CRISPR screen. Dots show biological replicates (n=3); bars show mean.

8 Figure S8

A C ParentalA2 A6 B3 B6 C3 D2 F2 F3 H2 ParentalA2 A6 B3 B6 C3 D2 F2 F3 H2 caspase 9 RUNX1 * - KO Vinculin # - Partial KO Vinculin # * * # # * * #

B D RUNX1 expression E 25 7.4 µM venetoclax 200 nM venetoclax RUNX1 10 kb * 2.0 ns 1.0 *** ns * 30 ns 1.5 KMT2A-N

1.0 0.5 30 AFF1-C

0.5er parental SEM cells v o 25 H3K27ac 0.0 0.0 old CTG Luminescence / DMSO WT Partial KO WT Partial KO F WT Partial KO CASP9 F Nascent RNAseq G 48h RUNX1 siRNA KD (RS4;11) H 1.25 20 NT KD NT KD * 1.00 BCL2 enh. BCL2 gene 15

CPM 0.75 10 kb 100 kb 19 13 MAZ xpression e to NT

10 e v 0.50 **

CASP9 16 128 MYB

5 relati 0.25 mRNA 0.00 165 37 0 RUNX1 KMT2A- RUNX1 KD AFF1 KD FDR = 0.025 12 30 UNX1 MLL-N FDR = 3.04×10-6 R CASP9 I J K BCL2 RUNX1 mRNA MYB mRNA MAZ mRNA MYC gene MYC super enh. 10 kb 22 50 kb

e to NT 1.0 61 MAZ v 26 3.8 MYB 0.5 210 14 RUNX1 0.0

Expression relati 23 17 MLL-N NT NT NT MAZ MYB MYB MYB MAZ MAZ RUNX1& & & & MYC

RUNX1RUNX1 RUNX1 RUNX1 siRNA siRNA siRNA

L siRNA RUNX1 RUNX1 NT RUNX1 & MAZ NT RUNX1 NT & MYC

RUNX1 Vinculin

Supplemental Fig S8. KMT2A-AFF1 and RUNX1 cooperate to regulate CASP9 in a cascade motif. (A) Western blot for caspase 9 in CASP9 knockout clones in SEM cells. KO lines indicated with a *, and partial KO with #. Parental refers to original SEM line without clonal selection. (B) CTG viability assay of CASP9 KO clones following 48 hours’ treatment with 7.4 µM venetoclax (left) or 200 nM venetoclax (IC50 in SEM cells (Benito et al. 2015) (right). CTG luminescence displayed as a ratio relative to DMSO control. Separate clones are used as biological replicates (WT n=5, KO and partial KO n=2). (C) Western blot for RUNX1 in CASP9 KO clones, with vinculin as loading control. CASP9 KO lines indicated with a *, and partial KO with #. (D) qRT-PCR results showing RUNX1 expression relative to NT control in CASP9 KO cells. (E) ChIP-seq tracks generated in hg19 for KMT2A-N, AFF1-C, RUNX1 and H3K27ac at CASP9 locus. Reads normalized to 1×107 reads. (F) Expression of CASP9 from nascent RNA-seq data following 96 hours’ KMT2A-AFF1 KD. Expression normalized as CPM. (G) qRT- PCR results showing RUNX1 and CASP9 expression relative to NT control after 48 hours’ RUNX1 KD in RS4;11 cells. (H) ChIP-seq tracks generated in hg19 for MAZ, MYB, RUNX1 and KMT2A-N. Visualized loci include and BCL2 and MYC loci, as well as the BCL2 enhancer (218 Kb downstream of promoter) and MYC super enhancer (1.8 Mb downstream of promoter). Reads normalized to 1×107 reads. (I - K) qRT-PCR analysis for RUNX1 (I) and MYB (J) and BCL2 (K) after 96 hours’ siRNA treatment targeting genes as indicated (n=3, n=5 for RUNX1 KD). Expression normalized to mature GAPDH mRNA levels, and shown relative to NT control. (L) Western blot for RUNX1 after 96 hours’ siRNA treatment targeting genes as indicated. Error bars represent standard error of the mean; * P < 0.05, ** P < 0.01.

9

Catalogue Target Application Dilution Company number RUNX1 Western blotting 1/5,000 4334S Cell Signaling caspase 9 Western blotting 1/10,000 ab202068 Abcam

GAPDH Western blotting 1/10,000 A300-641A Bethyl

Vinculin Western blotting 1/50,000 ab129002 Abcam

RUNX1 ChIP / ChIP-seq 1/500 ab23980 Abcam

KMT2A-N ChIP / ChIP-seq 1/500 A300-086A Bethyl

AFF1-C ChIP-seq 1/500 ab31812 Abcam

MAZ ChIP-seq 1/500 A301-652A Bethyl

Supplementary Table S1. List of antibodies used in this study.

10

Target Primer/probe Note

GAPDH Hs03929097_g1 TaqMan probe, qRT-PCR

RUNX1 Hs00231079_m1 TaqMan probe, qRT-PCR

CASP9 Hs00609647_m1 TaqMan probe, qRT-PCR

BCL2 Hs00608023_m1 TaqMan probe, qRT-PCR

MYC Hs0015348_m1 TaqMan probe, qRT-PCR F - CGATAACGCCTGCCATCTAA BCL2 (intronic) SYBR, qRT-PCR (pre-mRNA) R - CCACCACATCCTACTGGATTAC MYC (intronic) F - AAGGGAGGCGAGGATGTGTCC SYBR, qRT-PCR (pre-mRNA) R - GGCTGGGTGCGGAGATTCG KMT2A-AFF1 F - AGGTCCAGAGCAGAGCAAAC SYBR, qRT-PCR (SEM) R - CGGCCATGAATGGGTCATTTC KMT2A-AFF1 F - TCAGCACTCTCTCCAATGGCAATAG SYBR, qRT-PCR (RS4;11) R - GGGGTTTGTTCACTGTCACTGTCC Negative control F - GGCTCCTGTAACCAACCACTACC SYBR, ChIP-qPCR locus R - CCTCTGGGCTGGCTTCATTC +23 RUNX1 F - TGCGAGAGCGAGAAAACCACAG SYBR, ChIP-qPCR enhancer R - GCAGAAAGCAACAGCCAGAAACG CDK6 KMT2A- F - TCGAAGCGAAGTCCTCAACA SYBR, ChIP-qPCR AFF1 peak R - GCTTGGGCAGAGGCTATGTA

Supplementary Table S2. List of primers used in this study.

11

Experiment Cell line/tissue Antibody Treatment Accession Number ChIP-seq SEM KMT2A-N GSE74812

ChIP-seq SEM AFF1-C GSE74812

ChIP-seq SEM RUNX1 GSE42075

ChIP-seq SEM Input GSE42075

ChIP-seq SEM ELF1 GSE117865

ChIP-seq SEM H3K79me3 GSE117865

ChIP-seq SEM BRD4 GSE83671

ChIP-seq SEM H3K27ac GSE74812

ChIP-seq SEM MYB GSE117865

ATAC-seq SEM GSE74812

ChIP-seq Primograft KMT2A-N GSE83671

ChIP-seq Primograft AFF1-C GSE83671 KMT2A-AFF1 Nascent RNA-seq SEM GSE85988 siRNA Nascent RNA-seq SEM EPZ-5676 GSE83671

Nascent RNA-seq SEM IBET GSE139437 Supplementary Table S3. GEO accession numbers for previously published sequencing experiments

12