Inhibitors of Histone Acetyltransferases KAT6A/B Induce Senescence and Arrest Tumour Growth Jonathan B

Letter https://doi.org/10.1038/s41586-018-0387-5

Inhibitors of histone acetyltransferases KAT6A/B induce senescence and arrest tumour growth Jonathan B. Baell1,2*, David J. Leaver1, Stefan J. Hermans3, Gemma L. Kelly4,5, Margs S. Brennan4,5, Natalie L. Downer4, Nghi Nguyen1, Johannes Wichmann4,5, Helen M. McRae4,5, Yuqing Yang4,5, Ben Cleary1, H. Rachel Lagiakos1,6, Stephen Mieruszynski4,5, Guido Pacini4, Hannah K. Vanyai4,5, Maria I. Bergamasco4,5, Rose E. May4, Bethany K. Davey4,6, Kimberly J. Morgan4,5, Andrew J. Sealey4,5, Beinan Wang4,5,7, Natasha Zamudio4,5, Stephen Wilcox4,5, Alexandra L. Garnham4,5, Bilal N. Sheikh4,5, Brandon J. Aubrey4,5, Karen Doggett4,5, Matthew C. Chung3, Melanie de Silva4,6, John Bentley8, Pat Pilling8, Meghan Hattarki8, Olan Dolezal8, Matthew L. Dennis8, Hendrik Falk4,5,6, Bin Ren8, Susan A. Charman9, Karen L. White9, Jai Rautela4,5, Andrea Newbold10, Edwin D. Hawkins4,5, Ricky W. Johnstone10, Nicholas D. Huntington4,5, Thomas S. Peat8, Joan K. Heath4,5, Andreas Strasser4,5, Michael W. Parker3,11, Gordon K. Smyth4,12, Ian P. Street4,5,6, Brendon J. Monahan4,5,6, Anne K. Voss4,5,13* & Tim Thomas4,5,13*

Acetylation of histones by lysine acetyltransferases (KATs) is and KAT6B (IC50 8 nM and 28 nM, respectively), and is more than 1 essential for chromatin organization and function . Among tenfold less active against KAT7 and KAT5 (IC50 342 nM and 224 nM, the genes coding for the MYST family of KATs (KAT5–KAT8) respectively; Fig. 1b, Supplementary Table 1). Kinetic binding curves are the oncogenes KAT6A (also known as MOZ) and KAT6B obtained from SPR demonstrated that the interaction of this class (also known as MORF and QKF)2,3. KAT6A has essential roles in of compounds with immobilized proteins was fully reversible and normal haematopoietic stem cells4–6 and is the target of recurrent consistent with a single-site binding interaction. The interaction of chromosomal translocations, causing acute myeloid leukaemia7,8. WM-8014 with KAT6A and KAT7, although relatively strong, was Similarly, chromosomal translocations in KAT6B have been in both cases driven by fast association kinetics (association rate 8 6 −1 −1 identified in diverse cancers . KAT6A suppresses cellular senescence constant (ka) >1 × 10 M s ), whereas the dissociation kinetics 9,10 −2 −2 −1 through the regulation of suppressors of the CDKN2A locus , a (dissociation rate constant (kd) ~ 4 × 10 for KAT6A and 17 × 10 s function that requires its KAT activity10. Loss of one allele of KAT6A for KAT7) were indicative of a relatively short lifespan of the binary extends the median survival of mice with MYC-induced lymphoma complex (Extended Data Fig. 1). WM-8014 displayed an order of mag- 11 from 105 to 413 days . These findings suggest that inhibition of nitude weaker binding to KAT7 than to KAT6A (KD 52 nM versus KAT6A and KAT6B may provide a therapeutic benefit in cancer. 5.1 nM, respectively) (Fig. 1b, Extended Data Fig. 1). We also gener- Here we present highly potent, selective inhibitors of KAT6A ated an inactive analogue, WM-2474 (Fig. 1a, Supplementary Table 1). and KAT6B, denoted WM-8014 and WM-1119. Biochemical and Notably, these compounds were almost inactive against KAT8, and structural studies demonstrate that these compounds are reversible no inhibition was observed for the more distantly related lysine competitors of acetyl coenzyme A and inhibit MYST-catalysed acetyltransferases KAT2A, KAT2B, KAT3A and KAT3B (Fig. 1b, c, histone acetylation. WM-8014 and WM-1119 induce cell cycle exit Supplementary Table 1). and cellular senescence without causing DNA damage. Senescence WM-8014 has desirable, drug-like physicochemical properties is INK4A/ARF-dependent and is accompanied by changes in gene (Supplementary Table 2). It is completely stable in cell culture medium expression that are typical of loss of KAT6A function. WM-8014 (10% fetal calf serum); however, relatively high protein binding potentiates oncogene-induced senescence in vitro and in a zebrafish (97.5%) in this medium reduces its free concentration. Although model of hepatocellular carcinoma. WM-1119, which has increased WM-8014 has relatively low solubility in water (8–16 μM), it could bioavailability, arrests the progression of lymphoma in mice. We readily permeate Caco-2 cells (apparent permeability coefficient (Papp) anticipate that this class of inhibitors will help to accelerate the 78 ± 13 × 10−6 cm s−1). Testing of WM-8014 at 1 µM and 10 µM development of therapeutics that target gene transcription regulated revealed no notable affinity for a pharmacological panel of 158 diverse by histone acetylation. biological targets; only eight targets were affected by more than 50% In a screen of 243,000 diverse small-molecule compounds12, we (Supplementary Table 3). obtained the acylsulfonylhydrazide compound CTx-0124143, a We solved the crystal structures of a modified MYST histone competitive KAT6A inhibitor (half-maximal inhibitory concentra- acetyltransferase domain (MYSTCryst) in complex with WM-8014 12 tion (IC50) 0.49 µM) in biochemical assays . Medicinal chemistry (1.85 Å resolution, Fig. 1d–f, Extended Data Fig. 1, Supplementary optimization yielded the compound WM-8014 with an IC50 value of Table 4) or acetyl coenzyme A (acetyl-CoA; 1.95 Å resolution, Fig. 1g). 8 nM (Fig. 1a, Supplementary Table 1), representing a 60-fold increase The WM-8014 molecule occupies the acetyl-CoA-binding site on in inhibitory activity towards KAT6A. This was consistent with the MYSTCryst, being partially enclosed between the α-helix formed by binding affinity measured by surface plasmon resonance (SPR; equilib- residues D685 to R704 and the loop extending from Q654 to G657. The Cryst rium dissociation constant (KD) 5 nM; Fig. 1a, Extended Data Fig. 1). MYST –acetyl-CoA complex adopts a globular fold (Fig. 1g), as seen WM-8014 inhibits predominantly the closely related proteins KAT6A in previously reported structures13, with a root mean square deviation

1Medicinal Chemistry Theme, Monash Institute of Pharmaceutical Sciences, Monash University, Parkville, Victoria, Australia. 2School of Pharmaceutical Sciences, Nanjing Tech University, Nanjing, China. 3ACRF Rational Drug Discovery Centre, St Vincent’s Institute of Medical Research, Fitzroy, Victoria, Australia. 4The Walter and Eliza Hall Institute of Medical Research, Parkville, Melbourne, Victoria, Australia. 5Department of Medical Biology, University of Melbourne, Parkville, Victoria, Australia. 6Cancer Therapeutics CRC, Parkville, Victoria, Australia. 7School of Pharmaceutical Sciences, Tsinghua University, Beijing, China. 8Commonwealth Scientific and Industrial Research Organisation (CSIRO), Biomedical Program, Parkville, Victoria, Australia. 9Centre for Drug Candidate Optimisation, Monash Institute of Pharmaceutical Sciences, Monash University, Parkville, Victoria, Australia. 10The Peter MacCallum Cancer Centre, Melbourne, Victoria, Australia. 11Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology Institute, University of Melbourne, Parkville, Victoria, Australia. 12Department of Mathematics 13 and Statistics, University of Melbourne, Parkville, Victoria, Australia. These authors jointly supervised this work: Anne K. Voss, Tim Thomas. *e-mail: [email protected]; [email protected]; [email protected]

IC = 0.49 M 9,10 a 50 μ Because KAT6A suppresses senescence , we examined the ability of K = 0.38 M D μ WM-8014 to induce cell cycle arrest in embryonic day (E)14.5 mouse O H embryonic fibroblasts (MEFs). Cells treated with WM-8014 failed to S N N proliferate after 10 days of treatment (Fig. 2a; IC50 2.4 µM), with similar O H O F kinetics to Cre-recombinase Kat6a recombination (Fig. 2b). Higher CTx-0124143 N N doses of WM-8014 (up to 40 µM) did not accelerate growth arrest, IC = 0.008 M IC > 125 M 50 μ 50 μ which after 8 days of treatment was irreversible (Extended Data Fig. 2). K = 0.005 μM K SPR: no binding D D The inactive compound WM-2474 did not affect cell proliferation. Cell O O cycle analysis showed an increase in the proportion of cells in G0/G1 H H S N S N after 6 days of treatment and a corresponding reduction in cells in G2/M N N 14 O H O H and S phases, both in Fucci cells and in 5-bromo-2′-deoxyuridine O F O F (BrdU) incorporation assays (Fig. 2c, Extended Data Fig. 2). WM-8014 WM-2474 (inactive control) RNA sequencing (RNA-seq) of MEFs treated with WM-8014

b IC50 (μM) c KAT6A revealed a signature of cellular senescence, including upregulated KAT6A 100 KAT6B expression of Cdkn2a mRNA and decreased expression of Cdc6, which KAT6B 10 KAT7 9 15 KAT7 1 is a KAT6A target gene and a regulator of DNA replication (Fig. 2d; KAT8 0.1 KAT8 day 10: false discovery rate (FDR) < 10−6). A substantial increase KAT5 0.01 KAT5 KAT2A KAT2A in β-galactosidase activity—a marker of senescent cells—was also KAT2B KAT2B KAT3A observed (Fig. 2e), accompanied by morphological changes typical of KAT3B KAT3A senescence (Extended Data Fig. 2). WM-8014 caused a concentration- KAT3B CTx-143 WM-2474 dependent reduction in the level of E2f2 mRNA (adjusted (adj.) WM-8014 R2 = 0.73; P < 0.0005) and Cdc6 mRNA (adj. R2 = 0.5; P = 0.002), d e accompanied by upregulation of both splice products of the Cdkn2a locus, Ink4a and Arf (day 10: P < 0.0005 and P = 0.005, respectively; Extended Data Fig. 3). Notably, MEFs treated for 4 days or 10 days with 10 µM WM-8014, the control compound WM-2474 or DMSO vehicle control showed no change in the levels of γH2A.X (Extended f WM-8014 g Acetyl-CoA Ser690 Data Fig. 4), which suggests that cell cycle arrest was not a conse- quence of DNA damage. No increase in apoptosis or necrosis was Gly659 seen (Extended Data Fig. 4). Treatment of either Trp53-null MEFs −/− −/− −/− Arg655 Arg660 (Trp53 ) or Cdkn2a-null (Ink4a Arf ) MEFs with WM-8014 had Gly657 a minor effect and no effect on cell proliferation, respectively (Fig. 2f, Fig. 1 | Development of an inhibitor of the MYST family of lysine Extended Data Fig. 2). These results show that WM-8014 acts through INK4A ARF acetyltransferases. a, Schematic summary of the medicinal chemistry the p16 –p19 pathway, causing irreversible cell cycle exit leading optimization of high-throughput screening hit CTx-0124143, which to senescence, and does not have a general cytotoxic effect. 16 resulted in WM-8014 and the inactive compound WM-2474. The IC50 KAT7 is essential for global histone 3 lysine 14 (H3K14) acetylation . values (determined by biochemical assays) and equilibrium dissociation By contrast, KAT6A regulates H3K9 acetylation only at target loci17,18. constants (KD, determined by SPR) are shown for KAT6A. b, Histone We determined the effects of WM-8014 on global levels of acetylation acetyltransferase inhibition assay (competition of compound with acetyl- at H3K9 and H3K14 by western blot after 5 days of treatment. CoA) of CTx-0124143, WM-8014 and WM-2474. The areas of the circles Treatment with 10 µM WM-8014 caused a 49% decrease in the global reflect the IC values as indicated, assayed at the Michaelis constant 50 levels of H3K14ac but, as expected on the basis of the locus-specific (Km) of acetyl-CoA for each KAT tested. c, Dendrogram showing the 17,18 relationship between major KAT families based on sequence differences roles of KAT6A , did not significantly affect the global levels of in the acetyltransferase domain. d–g, Crystal structures of WM-8014 H3K9ac (Fig. 3a, b; all gel source data in Supplementary Fig. 1). The and acetyl-CoA bound to the MYST lysine acetyltransferase domain effects of WM-8014 on global H3K14ac levels were concentration- (MYSTCryst; see Extended Data Fig. 1). PDB codes: 6BA2 and 6BA4, dependent (Fig. 3b; H3K14ac/H4 ratio regressed on log concentration 2 respectively. d, Space-filling model showing WM-8014 in the acetyl- of WM-8014; adj. R = 0.76, P < 0.001; IC50 1.2 µM). RNA-seq showed Cryst Cryst CoA-binding pocket of MYST . e, Ribbon diagram of MYST a strong correlation between the changes in gene expression seen in (blue) showing WM-8014 (yellow, with element colouring) bound to the −/− +/+ Cryst Kat6a MEFs compared with Kat6a MEFs and the genes differ- acetyl-CoA-binding site. f, Ribbon diagram of MYST showing key entially expressed after WM-8014 treatment (WM-8014 compared with amino acids interacting with WM-8014 (yellow, with element colouring). Hydrogen bonds are shown as dashed lines. g, Ribbon diagram showing inactive WM-2474), with a 2.6-fold enrichment in upregulated genes acetyl-CoA (yellow, with element colouring) bound to MYSTCryst. (FDR = 0.0001; Fig. 3c) and a 2.1-fold enrichment in downregulated = Means of two experiments are shown for the IC50 values in a and b. SPR genes (FDR 0.0001; Fig. 3c), and gene expression signatures charac- experiments in a were repeated four times. teristic of cellular senescence (Extended Data Fig. 5). Loss of KAT6A results in the downregulation of E2f2, Ezh2 and Melk9. Similarly, treatment with WM-8014 caused significant downregulation of Ezh2, Melk (r.m.s.d.) of 0.6 Å, and is nearly identical to the MYSTCryst–WM-8014 and E2f2 mRNA levels compared with controls (Fig. 3d), as determined complex (r.m.s.d. of 0.3 Å for all aligned atoms). Accordingly, the core by RNA-seq (Extended Data Fig. 5) and confirmed by quantitative acylsulfonylhydrazide moiety of WM-8014 makes similar hydrogen reverse-transcription PCR (RT–qPCR) (Extended Data Fig. 3). After bonds to MYSTCryst as does the diphosphate group of acetyl-CoA treatment with WM-8014, there was a reduction of H3K9ac at the tran- (Fig. 1f, g). This includes hydrogen bonds to the main-chain atoms of scription start sites of these genes (Fig. 3e). Therefore, the treatment R655, G657 and R660—identical to acetyl-CoA—as well as additional of cells with high concentrations of WM-8014 directly inhibits global hydrogen bonds to G659 and S690 (Extended Data Fig. 1). The biphe- H3K14 acetylation catalysed by KAT7, as well as KAT6A-specific H3K9 nyl group of WM-8014 extends further into the acetyl-CoA-binding acetylation at transcription start sites. pocket, which enables van der Waals interactions with residues L601, Because WM-8014 induced cellular senescence, we reasoned that I647, I649, S684 and L686 of MYSTCryst (Extended Data Fig. 1). it might exacerbate oncogenic RAS-induced senescence. Accordingly, WM-8014 therefore competes directly with acetyl-CoA in the MEFs that express HRASG12V, a constitutively active form of RAS, substrate-binding domain. were more sensitive to the induction of cell cycle arrest by WM-8014

M DMSO Vehicle a b c G2 G1 10 μM 2474 Tamoxifen P < 0.0005 S 10 μM 8014 P < 100 20 0.0005 120 25 P < 0.0005 ) ) 7 100 5 P = 80 DMSO 10 16 20 WM-2474 P =

10 0.009 × P < Cells plated 20 μM 2474 80 × 60 0.001 12 0.0005 4 days treatment 15 20 μM 8014 60 8 days treatment P = P < P < 0.001 40 8 11 days treatment 10 0.0005 P < 0.0005 0.0005 40 P = centage of cells P < 0.0005 P ≤ 15 days treatment 20 4 5 0.001

20 Per Cell number ( 0.04 Cell number ( 0 0 0 0 S 0 24710 14 18 WM-8014

Number of cells (% DMSO) 0.1 110 04711 d Early re Days after tamoxifen treatment Early Days of treatment Concentration (μM) G1, DN Mid–late n G1, yellow ee gr Mid-S–G2/M –/– d e DMSO f Arf DMSO P ≤ FDR = 10 μM 2474 500 500 –/– DMSO FDR –/– 10 μM 2474 2.5 ) 0.019 10 μM 2474 0.001 6 –6 10 μM 8014 ) p53 ) <10 FDR P = 0.008 6 NS 10 μM 8014 400 4 10 M 8014 10 μ –6 12 Tr 10 Ink4a 2.0 <10 ×

10 50 50 P = 0.01 × 300 × 10 1.5 DMSO FDR = DMSO 10 μM 2474 –6 8 5 10 μM 2474 5 200 1.8 × 10 1.0 10 μM 8014 P < Wild type RPKM RPKM 10 μM 8014

6 P = 0.01 Wild type P < 0.0005 100 0.5 0.5 0.5 4 P = 0.033 0.0005 Number of cells ( 0 0 2 Number of cells ( Day 4Day 10 Day 4Day 10 0.05 0.05

-Galactosidase MFI ( 0 14810 14 02471014

Cdc6 β 0 Cdkn2a Day 4Day 10 Days of treatment Days of treatment Fig. 2 | Treatment of MEFs with WM-8014 leads to cellular senescence. and 10 days with 10 µM WM-8014 or 10 µM WM-2474 control, assessed a, Left, effects of WM-8014 compared with the inactive compound by RNA-seq. RPKM, reads per kilobase per million reads. e, Flow- WM-2474 or DMSO vehicle control on cell growth of MEFs grown in cytometry assessment (mean ± s.e.m. of median fluorescence intensity 3% O2. Right, effects of the dose of WM-8014 and the duration of (MFI)) of senescence-associated β-galactosidase activity in MEFs after treatment. b, Effects of acute genetic deletion of Kat6a on the growth of 4 and 10 days of treatment with 10 µM WM-8014, 10 µM WM-2474 or MEFs. Loss of KAT6A function was induced by nuclear translocation of DMSO vehicle control. f, Growth of MEFs lacking p16INK4A and p19ARF Cre-recombinase using tamoxifen on MEFs isolated from Kat6alox/lox (left) and of MEFs lacking p53 (right) compared with wild type after RosaCreERT2 and control RosaCreERT2 embryos. c, Left, epifluorescence treatment with WM-8014, DMSO vehicle control or WM-2474. n = 3 phase-contrast images of Fucci MEFs after 6 days of treatment with 20 µM independent MEF isolates per treatment group and genotype. Data are WM-2474 (top) and 20 µM WM-8014 (bottom). Right, the percentage of mean ± s.e.m. Data were analysed by one-way ANOVA followed by Fucci MEFs in each stage of the cell cycle after 6 days of treatment with Bonferroni post hoc test (a (left), b, c, e), non-linear regression curve 10 µM WM-8014, 10 µM WM-2474 or DMSO vehicle control, as fit (a, right) or two-way ANOVA (f) with treatment and with or without quantified by flow-cytometry analysis. DN, double negative. d, mRNA treatment duration as the independent factors. RNA-seq data (d) were levels of Cdkn2a (coding for cell cycle regulators p16INK4A and p19ARF) analysed as described in the Supplementary Methods. (left) and the KAT6A target gene Cdc6 (right) in MEFs treated for 4 days

4 b 1.4 NS 1.2 1.2 P = 0.008 WM- Adj. R2 = 0.76 M 8014 M 2474 1.2 2474 WM-8014 P = 0.001 a μ μ DMSO M: 10 1 2.5 5 10 DMSO 1.0 1.0 10 10 1.0 P = 0.011 μ DMSO H3K9ac 0.8 DMSO (~17 kDa) H3K9ac 0.8 0.8 1 μM 8014 0.6 10 μM 2474 (~17 kDa) 2.5 μM 8014 H3K14ac 0.6 (~17 kDa) 0.4 10 μM 8014 0.6 5 μM 8014 Pan-H4 0.2 H3K14ac 10 μM 8014 (~12 kDa) 0.4 0.4 H3K9ac levels 0 (~17 kDa) H3K14ac levels H3K9ac H3K14ac c Levels normalized to pan-H 0.2 0.2 FDR = 0.0001 Pan-H4 2.6 Upregulated in (~12 kDa) 0 –/– +/+ 0 Kat6a vs Kat6a MEFs H3K14ac H3K9ac

Enrichment 0 d 10 μM 2474 10 μM 8014 6.7 FDR = Weight 20 0.6 e

–5 –5 0 6 × 10 NS FDR = 0.4 –5 FDR –4

0.5 3 10 10 × 10

–6 × Up Down 15 <10 FDR = × –4 DMSO 3 × 10 FDR 0.4 0.3 –6 = 3 0 <10–6 0.3 = 3 10 μM 8014 Weight 10 P –6.7 P = 10

RPKM 0.2 RPKM 10 μM 2474

0 0.2 P 5 ChIP/input Downregulated in FDR = 0.0001 0.1 0.1 –/– +/+ 2.1 Kat6a vs Kat6a MEFs 0 0 Day 4Day 10 Day 4Day 10 Day 4Day 10 0.0

Enrichment Melk E2f2 Ezh2

–4.87 Ezh2 Melk E2f2 4.63 2.64 1.59 0.77 0.04 –0.71 –1.57 –2.78 39.37 –35.47 t-statistics Fig. 3 | Treatment of cells with WM-8014 leads to a reduction in MEF isolates from individual E12.5 embryos, namely from n = 3 Kat6a−/− acetylation of specific histone lysine residues and changes in gene and 2 Kat6a+/+ embryos, as well as 3 MEF isolates from 3 wild-type expression that resemble the genetic loss of KAT6A. a, Western embryos treated with either WM-8014 or WM-2474. d, Ezh2, Melk and blot detection of H3K14ac or H3K9ac in MEFs treated with 10 µM E2f2 mRNA levels measured by RNA-seq in MEFs treated for 4 days and WM-8014, 10 µm WM-2474 or DMSO for 5 days. The densitometric 10 days with 10 µM WM-8014 or 10 µM control WM-2474 (n = 3 MEF analysis is presented on the right. n = 6 (H3K14ac) and n = 9 (H3K9ac) isolates from 3 wild-type embryos treated with either WM-8014 or independent cultures per treatment group. b, Western blot of MEFs WM-2474). e, Anti-H3K9ac chromatin immunoprecipitation followed by treated with increasing doses of WM-8014 and controls as indicated. qPCR detection of transcription start sites of genes after treatment with Densitometric analysis is presented on the right. n = 3 independent DMSO 10 µM WM-8014 or WM-2474 for 3 days. The results of one experiments. Histone acetylation levels were regressed on the log10 of the of four experiments are shown; total n = 16 cultures per treatment WM-8014 concentration. H3K14ac and H3K9ac levels were normalized group in 4 experiments. Data are mean ± s.e.m. (with the exception to pan-H4 levels and DMSO treatment. c, Barcode plot in which genes of e, mean ± s.d.) and were analysed by one-way ANOVA followed that are differentially up- or downregulated in Kat6a−/− versus Kat6a+/+ by Bonferroni post hoc test (a), by regression analysis (b) or by t-test MEFs (that is, after genetic deletion of KAT6A) are compared with genes comparing WM-8014 to WM-02474 (e). The RNA-seq analysis (c, d) is differentially expressed in MEFs treated with WM-8014 versus WM-2474. described in the Supplementary Methods. Combined results of day 4 and day 10 treatment, ROAST P = 0.0001;

(Extended Data Fig. 6). We then examined the effects of WM-8014 in a a 19 G12V zebrafish model of KRAS -driven hepatocellular overproliferation. N We observed a significant, concentration-dependent reduction in liver F volume in response to treatment with WM-8014, and a substantial O H O reduction in hepatocytes in S phase (Extended Data Fig. 6). Notably, S N H N S N O N F WM-8014 did not impair the growth of the normal liver, demonstrating H O O F H O that the inhibitory effects of WM-8014 were specific to hepatocytes WM-8014 WM-1119 K (KAT6A) K (KAT6A) that express oncogenic RAS. Treatment with WM-8014 was found to D = 0.005 μM=D 0.002 μM KD(KAT5) = 0.9 μM KD(KAT5) = 2.2 μM robustly upregulate the cell cycle regulators Cdkn2a and Cdkn1a in KD(KAT7) = 0.09 μM KD(KAT7) = 0.5 μM G12V hepatocytes that express oncogenic KRAS , but not control hepato- b c 100 WM-1119 Day 3 after lymphoma cell transplant, before treatment cytes. Therefore, WM-8014 potentiates oncogene-induced senescence, WM-8014 30 Baseline before treatment ×103 but it does not affect normal hepatocyte growth. 25 photons s–1 50 –2 –1 The progression of lymphoma is highly dependent on KAT6A, as 20 cm sr Day 3 15 10 Colour scale Kat6a heterozygous mice are protected from early-onset MYC-driven 3 Inhibition (% ) 0 Min. = 2.31 ×10 11 5 4 lymphoma . However, the high levels of plasma-protein binding exhib- Max. = 3.19 ×10 1 10 Vehicle 4× per day WM-1119 4× per day 0.01 0.1 100 ited by WM-8014 (Supplementary Table 2) precluded in vivo studies in 0.001 1.2 After treatment 3 –1 Concentration (μM) 1.0 ×10 photons s mice. Development of derivatives of WM-8014 resulted in WM-1119, cm–2 sr–1 0.8 which has reduced plasma-protein binding (Fig. 4a; Supplementary 1119 8014 0.6 IC50 0.25 μM 2.3 μM Day 14 0.4 Colour scale Table 2). The interaction of WM-1119 with KAT6A is similar to 2 Min. = 7.36 106 R 0.91 0.86 0.2 × that of WM-8014: it is characterized by strong reversible binding Max. = 1.23 ×108 (KD 2 nM, compared with 5 nM for WM-8014; Extended Data Fig. 7) d e f Vehicle 4 per day 0.7 ) × that is competitive with acetyl-CoA, and driven by fast association 3 8 6 −1 −1 0.6 kinetics (ka > 1 × 10 M s ; Extended Data Fig. 7). The structure 6 Cryst 0.5 P = 0.046 4 of MYST in complex with WM-1119 was solved (Extended Data 0.4 P = 0.0001 Fig. 7, Supplementary Table 5) and was found to be almost identical to 2 0.3 P = 0.07 P < 0.0005 to baseline ( × 10 Cryst otal ux normalized T P = 0.2 that of MYST –WM-8014, with an r.m.s.d. for aligned main-chain 0 WM-1119 4× per day 0.2

3710 12 14 Spleen weight (g ) atoms of 0.2 Å. There are two key differences between the complexes: an Days after lymphoma 0.1 transplant 0 additional hydrogen bond is formed between the WM-1119 pyridine Vehicle 1119 1119 Normal 1119 Non-responder n = 1 n = 15 3× per day 4× per day spleen Non- Vehicle n = 15 nitrogen and the main chain at I649 that is not present in the complex i.p. i.p. n = 4 responder WM-1119 3× per day n = 6 n = 6 n = 8 n = 1 with WM-8014 (Extended Data Fig. 7), and the hydrophobic inter- WM-1119 4× per day n = 8 action that exists between the meta-methyl of the biphenyl group of Vehicle n = 3 WM-1119 n = 3 g Vehicle 4× per day WM-1119 4× per day

WM-8014 and I663 is not present in the complex with WM-1119. + – + + + – + + 105 CD19 IgM CD19 IgM 105 CD19 IgM CD19 IgM 50 60 P = 13 ± 9 40 ± 2 0.5 ± 0.02 38 ± 3 0.0002 CD1 9 WM-1119 is 1,100-fold and 250-fold more active against KAT6A than CD19 40 104 104 cells (% ) cells (% ) – + 40 against KAT5 or KAT7, respectively (Fig. 4a, Extended Data Fig. 7), 30 P = IgM 103 103 IgM

P = +

+ 0.0002 20 20 and so shows greater specificity for KAT6A than does WM-8014. The 0.007 P = 0 0 testing of WM-1119 at 1 µM and 10 µM against a pharmacological 10 0.0001 –103 –103 0 0 –103 0103 104 105 –103 0103 104 105 Live CD1 9 panel of 159 diverse biological targets revealed no affinity Live CD1 9 BM BM IgM (Supplementary Table 6). Treatment of MEFs with WM-1119 resulted IgM Spleen PWBC Spleen PWBC in cell cycle arrest in G1 and a senescence phenotype similar to that Fig. 4 | Treatment with WM-1119 arrests lymphoma growth. a, Medicinal seen upon treatment with WM-8014 (Extended Data Fig. 8). Notably, chemistry optimization of WM-8014 resulted in compound WM-1119. the activity of WM-1119 in this cell-based assay is an order of mag- The binding data (obtained by SPR) for the interaction of WM-1119 with nitude greater than WM-8014 and WM-1119 is able to induce cell immobilized KAT6A, KAT7 and KAT5 are compared with the interaction µ µ data for WM-8014. b, Growth inhibition assays of E -Myc lymphoma cycle arrest at 1 M. cell line EMRK1184 treated with WM-1119 and WM-8014 at the doses To test inhibitors of KAT6A in a cancer model, we investigated the indicated. c, Bioluminescence images of EMRK1184 lymphoma cells effect of WM-1119 and WM-8014 on the proliferation of lymphoma expressing luciferase before (day 3) and after (day 14) 11 days of treatment cells. We selected the B cell lymphoma cell line EMRK1184, which with WM-1119 (50 mg kg−1 four times per day) or PEG400 vehicle control. was isolated from mice with a tumour resulting from the expression The red boxes show the regions used for quantification (imaging at days 7, of Myc under the control of the IgH enhancer20, because it expressed 10 and 12 in Extended Data Fig. 10). d, Quantification of the signals the Cdnk2a-locus-encoded ARF and wild-type p53 (Extended Data measured in all experiments: two cohorts of mice treated with WM-1119 Fig. 9). Treatment with WM-8014 or WM-1119 inhibited the prolif- three times per day, combined n = 6; two cohorts of mice treated with eration of the EMRK1184 lymphoma cells in vitro (Fig. 4b); RNA- WM-1119 four times per day, combined n = 9; vehicle controls, n = 15. One seq and western blot analysis showed that treatment with WM-1119 mouse did not respond to WM-1119 treatment, shown in grey. e, Dissected spleens obtained after imaging on day 14, taken from the mice shown in c. resulted in increased levels of Cdkn2a and Cdkn2b mRNA and p16INK4a ARF f, Spleen weights of mice treated with WM-1119 or vehicle. n values as and p19 proteins, as well as a delayed increase in Cdkn1a mRNA stated in d. i.p., intraperitoneal. g, Flow-cytometry analysis of spleen cells (Extended Data Fig. 9). WM-1119 (IC50 0.25 µM) was ninefold more from vehicle-treated mice and mice treated with WM-1119 (four times per + − active than WM-8014 (IC50 2.3 µM; Fig. 4b), as expected on the basis day). The tumour cells were CD19 IgM , and normal splenic B cells were of reduced protein binding (Supplementary Table 2). CD19+IgM+. Quantification of flow-cytometry analysis in bone marrow We tested the effectiveness of KAT6 inhibitors in the treatment of (BM), spleen and peripheral white blood cells (PWBC). n = 4 independent lymphoma in mice. Male C57BL/6-albino (B6(Cg)-Tyrc-2J/J) mice experiments for WM-1119 and 2 for WM-8014 in b, and number of were injected intravenously with 100,000 EMRK1184 cells transfected mice as indicated in d, f, g in three independent experiments. Data are ± with a luciferase-expression construct. Lymphoma growth was moni- mean s.e.m. and were analysed by nonlinear regression dose–response curve fit, least squares fit, inhibitor versus response, variable slope (b); tored using the IVIS imaging system. Three days after the lymphoma- one-way ANOVA followed by Bonferroni post hoc test with treatment as the cell transplant, all mice showed luciferase activity (Fig. 4c), which independent factor (d, g), or two-tailed t-tests (f). indicated the expansion of lymphoma cells. Mice were then divided randomly into WM-1119-treatment and vehicle-control groups. (Extended Data Fig. 9), cohorts of mice were injected every 8 h (three Because WM-1119 is rapidly cleared after intraperitoneal injection, times per day, two cohorts of three mice per treatment group) or every 6 h with the plasma concentration decreasing to below 1 µM after 4–6 h (four times per day, two cohorts of three and six mice per treatment

256 | NATURE | VOL 560 | 9 AUGUST 2018 © 2018 Springer Nature Limited. All rights reserved. Letter RESEARCH group; Fig. 4d). Mice were imaged five times over the course of these 5. Thomas, T. et al. Monocytic leukemia zinc fnger protein is essential for the development of long-term reconstituting hematopoietic stem cells. Genes Dev. experiments to monitor the growth of lymphoma. No significant differ- 20, 1175–1186 (2006). ence between the treatment and control groups was seen before day 10 6. Sheikh, B. N. et al. MOZ (KAT6A) is essential for the maintenance of classically (Fig. 4d, Extended Data Fig. 10), which was expected as the inhibition defned adult hematopoietic stem cells. Blood 128, 2307–2318 (2016). 7. Borrow, J. et al. The translocation t(8;16)(p11;p13) of acute myeloid leukaemia of cell proliferation in vitro took approximately seven days. However, by fuses a putative acetyltransferase to the CREB-binding protein. Nat. Genet. 14, day 14, the cohorts that were treated four times per day with WM-1119 33–41 (1996). had arrested tumour growth (Fig. 4c, Extended Data Fig. 10), with the 8. Huang, F., Abmayr, S. M. & Workman, J. L. Regulation of KAT6 acetyltransferases exception of one mouse that did not respond (Fig. 4d). Spleen weights and their roles in cell cycle progression, stem cell maintenance, and human disease. Mol. Cell. Biol. 36, 1900–1907 (2016). in the WM-1119-treatment group (treated four times per day) were 9. Sheikh, B. N. et al. MOZ (MYST3, KAT6A) inhibits senescence via the INK4A–ARF substantially lower than spleen weights in the vehicle-treated group, pathway. Oncogene 34, 5807–5820 (2015). INK4a and not significantly different from those of tumour-free eight-week- 10. Perez-Campo, F. M. et al. MOZ-mediated repression of p16 is critical for the self-renewal of neural and hematopoietic stem cells. Stem Cells 32, 1591–1601 old mice (P < 0.0005 and P = 0.2, respectively; Fig. 4e, f). Treatment (2014). with WM-1119 three times per day led to a significant reduction in 11. Sheikh, B. N. et al. MOZ regulates B-cell progenitors and, consequently, Moz tumour burden and spleen weight, but was not as effective as treat- haploinsufciency dramatically retards MYC-induced lymphoma development. Blood 125, 1910–1921 (2015). ment four times per day (Fig. 4d, f). WM-1119 was well-tolerated; mice 12. Falk, H. et al. An efcient high-throughput screening method for MYST family showed no generalized ill effects and weight loss was not observed acetyltransferases, a new class of epigenetic drug targets. J. Biomol. Screen. 16, (Extended Data Fig. 10). WM-1119 treatment had no effect on haemat- 1196–1205 (2011). 13. Holbert, M. A. et al. The human monocytic leukemia zinc fnger histone ocrit, erythrocytes or platelet numbers, but there was overall leukopenia acetyltransferase domain contains DNA-binding activity implicated in (Extended Data Fig. 10). The proportion and overall number of tumour chromatin targeting. J. Biol. Chem. 282, 36603–36613 (2007). cells was substantially reduced by WM-1119 treatment (four times per 14. Sakaue-Sawano, A. et al. Visualizing spatiotemporal dynamics of multicellular cell-cycle progression. Cell 132, 487–498 (2008). day; Fig. 4g). Analysis by intracellular flow cytometry demonstrated a 15. Yan, Z. et al. Cdc6 is regulated by E2F and is essential for DNA replication in reduction in H3K9ac in tumour cells (P = 0.03; Extended Data Fig. 10). mammalian cells. Proc. Natl Acad. Sci. USA 95, 3603–3608 (1998). These results demonstrate that WM-1119 is effective in treating 16. Kueh, A. J., Dixon, M. P., Voss, A. K. & Thomas, T. HBO1 is required for H3K14 lymphoma in vivo. acetylation and normal transcriptional activity during embryonic development. Mol. Cell. Biol. 31, 845–860 (2011). In summary, using high-throughput screening followed by medicinal 17. Voss, A. K., Collin, C., Dixon, M. P. & Thomas, T. Moz and retinoic acid chemistry optimization, in-cell assays, biochemical assessment of coordinately regulate H3K9 acetylation, Hox gene expression, and segment target engagement and tumour models in mice and fish, we have devel- identity. Dev. Cell 17, 674–686 (2009). 18. Voss, A. K. et al. MOZ regulates the Tbx1 locus, and Moz mutation partially oped a novel class of inhibitors for a hitherto unexplored category of phenocopies DiGeorge syndrome. Dev. Cell 23, 652–663 (2012). epigenetic regulators. These inhibitors engage the MYST family of 19. Chew, T. W. et al. Crosstalk of Ras and Rho: activation of RhoA abates lysine acetyltransferases in primary cells, specifically induce cell cycle Kras-induced liver tumorigenesis in transgenic zebrafsh models. Oncogene 33, 2717–2727 (2014). exit and senescence, and are effective in preventing the progression of 20. Adams, J. M. et al. The c-myc oncogene driven by immunoglobulin enhancers lymphoma in mice. induces lymphoid malignancy in transgenic mice. Nature 318, 533–538 (1985).

Reporting summary Acknowledgements We thank F. Dabrowski, C. D’Alessandro, WEHI Bioservices, Further information on experimental design is available in the Nature Research the WEHI FACS laboratory, the MX2 beamline staff at the Australian Synchrotron Reporting Summary linked to this paper. for their expert help and Z. Gong for the two transgenic zebrafish lines. This work was funded by the Australian Government through NHMRC project grants 1030704, 1080146, Research Fellowships (T.T., A.K.V., G.K.S., J.K.H., M.W.P. and Data availability J.B.), the NHMRC IRIISS and the Cancer Therapeutics Cooperative Research The RNA-seq data of MEFs treated with WM-8014, WM-2474 and DMSO, Centre. The Victorian State Government OIS Grants to WEHI, Monash and of MEFs from Kat6a−/− and wild-type embryos and of lymphoma cell line St Vincent’s Institute are gratefully acknowledged. EMRK1184 treated with vehicle and WM-1119 have been submitted to the Gene Reviewer information Nature thanks P. Adams, R. Marmorstein and the other Expression Omnibus (GEO) database under accession number GSE108244. The anonymous reviewer(s) for their contribution to the peer review of this work. crystal structure data for the MYST domain in complex with WM-8014, acetyl- CoA and WM-1119 have been submitted to the Protein Data Bank (PDB) under Author contributions T.T. was responsible for initiating the project. T.T. and A.K.V. accession numbers 6BA2, 6BA4 and 6CT2, respectively. Source Data for all graphs supervised the project, performed experiments and wrote the manuscript. Medicinal chemistry: supervised by J.B.B., team: D.J.L., N.N., B.C. and H.R.L. are provided. Structural biology: S.J.H., M.C.C., B.R., T.S.P. and M.W.P. Chemical screening, protein biochemistry and assays: M.d.S., J.B., P.P., M.H., O.D., M.L.D., H.F., I.P.S. Online content and B.J.M. Pharmacology: S.A.C. and K.L.W. Bioinformatics: G.P., A.L.G. and Any Methods, including any statements of data availability and Nature Research G.K.S. Cell-based assays, molecular biology and biochemistry: N.L.D., J.W., H.M.M., Y.Y., H.K.V., M.I.B., R.E.M., B.K.D., B.W., N.Z., S.W., B.N.S. and B.J.A. reporting summaries, along with any additional references and Source Data files, Zebrafish model: S.M., K.J.M., A.J.S., K.D. and J.K.H. Mouse cancer models: are available in the online version of the paper at https://doi.org/10.1038/s41586- G.L.K., M.S.B., J.R., A.N., E.D.H., R.W.J., N.D.H. and A.S. 018-0387-5 Competing interests The authors declare no competing interests. Received: 10 December 2016; Accepted: 21 June 2018; Published online 1 August 2018. Additional information Extended data is available for this paper at https://doi.org/10.1038/s41586- 018-0387-5. 1. Lee, K. K. & Workman, J. L. Histone acetyltransferase complexes: one size Supplementary information is available for this paper at https://doi.org/ doesn’t ft all. Nat. Rev. Mol. Cell Biol. 8, 284–295 (2007). 10.1038/s41586-018-0387-5. 2. Allis, C. D. et al. New nomenclature for chromatin-modifying enzymes. Cell 131, Reprints and permissions information is available at http://www.nature.com/ 633–636 (2007). reprints. 3. Voss, A. K. & Thomas, T. MYST family histone acetyltransferases take center Correspondence and requests for materials should be addressed to J.B.B., stage in stem cells and development. BioEssays 31, 1050–1061 (2009). A.K.V. or T.T. 4. Katsumoto, T. et al. MOZ is essential for maintenance of hematopoietic stem Publisher’s note: Springer Nature remains neutral with regard to jurisdictional cells. Genes Dev. 20, 1321–1330 (2006). claims in published maps and institutional affiliations.

Extended Data Fig. 1 | See next page for caption.

Extended Data Fig. 1 | Binding characteristics of the MYST domain– WM-8014 and amino acids within the acetyl-CoA-binding pocket of WM-8014 protein–ligand interaction and comparison of MYST the MYST domain. The amino acids that differ between MYST family family histone acetyltransferase domains. a, SPR binding data for the members are indicated. Data collection and refinement statistics of interaction of WM-8014 with immobilized KAT6A and KAT7 MYST the WM-8014 and acetyl-CoA co-crystal structures can be found in domains. Injected concentrations of WM-8014 are indicated. Binding Supplementary Table 4. The overall structure of WM-8014 bound to responses (data; black sensorgrams) are overlaid, fitted curves of a 1:1 MYSTCryst is nearly identical to the MYSTCryst–acetyl-CoA complex. kinetic interaction model that included mass transport component The pantothenate arm of acetyl-CoA adopts an identical position to (coloured lines), as well as derived kinetic rate constants (ka, kd) and published MYST HAT domain structures; as observed previously, there are 13 equilibrium dissociation (KD) constant. One of at least two experiments differing positions for the 3′-phosphate ADP . Autoacetylation of K604 is shown. b, WM-8014 bound to MYSTCryst, with the WM-8014 OMIT was observed, as expected22. Gol denotes glycerol. f, Comparison of the electron density map contoured to 3σ shown in green. c, Acetyl-CoA conserved MYST domain between MYST family proteins. MYSTCryst is a bound to MYSTCryst, with the acetyl-CoA OMIT map contoured to 3σ MYST domain modified to improve solubility and used in crystallization shown in green. d, Ribbon diagram showing WM-8014 and acetyl-CoA studies. Numbering as in KAT6A sequence, NP_006757.2; amino acids superimposed. e, Protein–ligand interactions (LIGPLOT)21 between interacting with WM-8014 (depicted in the LIGPLOT) are shown in red.

21. Wallace, A. C., Laskowski, R. A. & Thornton, J. M. LIGPLOT: a program to generate schematic diagrams of protein–ligand interactions. Protein Eng. 8, 127–134 (1995). 22. Yuan, H. et al. MYST protein acetyltransferase activity requires active site lysine autoacetylation. EMBO J. 31, 58–70 (2012).

Extended Data Fig. 2 | Time course of MEF growth inhibition upon are double-positive yellow in early S phase and double-negative in early G1. treatment with WM-8014, and requirement for INK4A/ARF and e, Cell cycle analysis of Cdkn2a null (Ink4a−/−Arf−/−) and littermate p53 for WM-8014-induced cell cycle arrest. a, MEF proliferation after control cells after treatment for 8 days with WM-8014, vehicle and the treatment with three high concentrations of WM-8014. MEFs were treated inactive compound WM-2474. MEFs were exposed to BrdU for 1 h before either continuously for 15 days, or treatment was discontinued after 1, 2, 4 flow-cytometry analysis of BrdU incorporation during DNA synthesis or 8 days to determine whether cells could re-enter the cell cycle. b, Phase- (S phase) and DNA content of 2N (G0/G1) compared with 4N (G2/M) contrast images of MEFs after 15 days of treatment with 10 µM WM-8014 using 7-AAD. f, Senescence-associated β-galactosidase activity in or 10 µM WM-2474. Note cells with senescence morphology; that is, Cdkn2a−/− and control MEFs after treatment for 15 days with 10 μM large nuclei indicating endoreplication without cell division and extensive WM-8014, 10 µM WM-2474 or DMSO vehicle control. g, Cell cycle cytoplasm (WM-8014 panel). c, Flow-cytometry gating strategy for the analysis of Trp53 null MEFs (Trp53−/−) and littermate control cells after cell cycle analysis using incorporation of the nucleotide analogue BrdU to treatment with WM-8014, vehicle and inactive compound WM-2474, as mark cells in S phase and 7-aminoactinomycin D (7-AAD) to determine in c. n = 3 MEF isolates per genotype (a–e). Data are mean ± s.e.m., and 2N (G0/G1) and 4N (G2/M) DNA content. d, Flow-cytometry gating were analysed by two-way ANOVA within duration of treatment with strategy for the cell cycle analysis of transgenic Fucci cells that express concentration and days of culture as the independent factors (a), or by Azami Green in mid-S phase, G2 and M, Kusabira Orange in mid–late G1, one-way ANOVA followed by Bonferroni post hoc test (e–g).

Extended Data Fig. 3 | The effect of WM-8014 on cell proliferation 10 µM WM-8014, 10 µM control WM-2474 or DMSO. d, Dose–response is mediated through the cell cycle regulators p16INK4A and p19ARF. plots of WM-8014-dependent reduction in E2f2 and Cdc6 mRNA levels in a, RT–qPCR analysis of expression levels of cell cycle regulators Ink4a and MEFs. e, Levels of mRNA coding for MYST-family proteins after treatment Arf (alternative splice products of the Cdkn2a locus), Ink4b (also known as of MEFs for 4 days or 10 days with WM-8014, vehicle or the inactive Cdkn2b) and Cdkn1a (encoding p21WAF1/CIP1) mRNA in MEFs treated for compound WM-2474. n = 3 MEF isolates treated with WM-8014, 4 days and 10 days with 10 µM WM-8014 or 10 µM control WM-2474. WM-2474 or vehicle (a–e). Data are mean ± s.e.m. and are analysed by b, Dose–response plots of WM-8014 induction of Ink4a mRNA expression one-way ANOVA followed by Bonferroni post hoc test (a–c, e) and by in MEFs. c, RT–qPCR analysis of expression changes in the KAT6A target regression analysis (d). mRNA levels normalized to housekeeping genes gene detected by RNA-seq. MEFs were treated for 4 days and 10 days with (HK) were regressed on the log(concentration) of WM-8014 (d).

Extended Data Fig. 4 | WM-8014 causes cell cycle exit and senescence cells, respectively) are shown in the left panels. Annexin V marks in MEFs, but not DNA damage or cell death. a, Assessment of DNA phosphatidylserine externalization on cells undergoing apoptosis, damage using flow cytometry to detect γH2A.X. Top, exposure of MEFs propidium iodine (PI) uptake marks cells undergoing other forms of cell to ultraviolet light (positive control). Bottom, experimental samples. death, annexin V/PI double-positive staining marks cells in late-stage Quantification is displayed in the bar graph. b, Flow-cytometry gating apoptosis. n = 3 cultures (a, b). Data are mean ± s.e.m. and were analysed strategy for cell death analysis and representative experimental samples. by one-way ANOVA with treatment as the independent factor. Negative and positive controls (untreated and ultraviolet-light-irradiated

Extended Data Fig. 5 | See next page for caption.

Extended Data Fig. 5 | WM-8014 treatment induces a gene signature of MEFs with Kat6a+/+ control MEFs. Differentially expressed genes are cellular senescence. a, Multidimensional scaling plot (log2 fold changes) highlighted in red or blue as indicated (FDR < 0.05). g, Genes typical of showing clustering of MEF expression profiles after treatment with cycling cells23 and E2F3 target genes24 are downregulated in MEFs treated WM-8014 or control WM-2474. MEFs were isolated from 3 different with WM-8014 versus WM-2474 (combined day 4 and day 10 treatment; embryos, numbered 5, 6 and 7 and treated for 4 days (96 h, red) or ROAST gene set tests, P = 0.0001). h, Genes downregulated during 10 days (240 h, green). b, Scatter plot showing gene-wise t-statistics for p53-induced cellular senescence25 were downregulated in MEFs treated differentially expressed (DE) genes (FDR < 0.05) between the compounds with WM-8014 versus WM-2474 (combined day 4 and day 10 treatment; at day 4 and day 10. Most genes were equally affected by 4 days or ROAST P = 0.0001). Differentially expressed genes in cellular senescence26 10 days of treatment (green). Genes differentially expressed at day 10 are strongly correlated with differentially expressed genes in MEFs treated only are highlighted blue, those differentially expressed at day 4 only are with WM-8014 versus WM-2474 (ROAST P = 0.0039). i, DAVID27 was highlighted red. c, Mean-difference plot of treatment, log2 fold changes used to test for functional enrichment in genes downregulated after versus average log2 expression. Treatment effects at 4 days and 10 days treatment with WM-8014 versus WM-2474, with FDR < 0.05. Cell cycle have been averaged. Differentially expressed genes are highlighted in red regulation was the top enriched pathway (FDR = 1.58 × 10−16), with 85% or blue as indicated (FDR < 0.05). d, Number of differentially expressed of the genes downregulated after 10 days of treatment with WM-8014. genes for MEFs treated with WM-8014 versus WM-2474 (FDR < 0.05). Schematic drawing is based on mmu04110: cell cycle28. Downregulated e, Mean-difference plot of treatment log2 fold changes versus average log2 genes are shaded blue; unchanged, green; upregulated genes (Ink4a, Arf, expression comparing WM-2474 to vehicle DMSO. The four differentially Ink4b and p21) are shaded red. Data were collected from n = 3 MEFs expressed genes (FDR < 0.05) are marked in red. f, Mean-difference plot isolates from 3 different embryos per treatment group, WM-8014 or −/− of log2 fold changes versus average log2 expression comparing Kat6a WM-2474 treatment, for 96 h or 240 h.

23. Chang, H. Y. et al. Gene expression signature of fbroblast serum response 26. Lujambio, A. et al. Non-cell-autonomous tumor suppression by p53. Cell 153, predicts human cancer progression: similarities between tumors and wounds. 449–460 (2013). PLoS Biol. 2, e7 (2004). 27. Huang, W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative analysis 24. Kong, L. J., Chang, J. T., Bild, A. H. & Nevins, J. R. Compensation and specifcity of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57 of function within the E2F family. Oncogene 26, 321–327 (2007). (2009). 25. Tang, X., Milyavsky, M., Goldfnger, N. & Rotter, V. Amyloid-β precursor-like protein 28. Kanehisa, M., Sato, Y., Kawashima, M., Furumichi, M. & Tanabe, M. KEGG as a APLP1 is a novel p53 transcriptional target gene that augments neuroblastoma reference resource for gene and protein annotation. Nucleic Acids Res. 44, cell death upon genotoxic stress. Oncogene 26, 7302–7312 (2007). D457–D462 (2016).

Extended Data Fig. 6 | WM-8014 potentiates oncogene-induced transgenic zebrafish Tg(lfabp10:RFP;elaA:eGFP) expressing only RFP senescence. a, Growth curves of MEFs expressing empty vector control (red). c, Quantification of liver volume. d, Incorporation of the nucleotide (pBABE) or oncogenic29 HRASG12V treated with increasing concentrations analogue 5-ethynyl-2′-deoxyuridine (EdU) after treatment of transgenic of WM-8014 as indicated or DMSO vehicle control. All experiments were zebrafish expressing KRASG12V or control zebrafish with WM-8014 or performed in 3% O2. b, The effects of WM-8014 treatment in a zebrafish control compound WM-2474. e, RT–qPCR determination of Cdkn2a model of hepatocellular carcinoma19. Doxycycline-inducible, liver- (Ink4a) and Cdkn1a (encoding p21WAF1/CIP1) mRNA levels in transgenic specific expression of a GFP-krasG12V transgene leads to the accumulation zebrafish Tg(TO-krasG12V) treated as described in b. n = 6 independent of a constitutively active, GFP-tagged form of KRAS in hepatocytes. cultures (a), 20 zebrafish (b, c), 10–12 zebrafish (d) and 4–5 zebrafish (e). TO-krasG12V transgenic embryos were treated with doxycycline at Data are mean ± s.e.m. and were analysed by two-way ANOVA (a) or one- 2 days post fertilization (dpf) and 5 dpf to initiate KRASG12V-driven way ANOVA (d, e) followed by Bonferroni post hoc test with treatment hepatocyte proliferation. The size of the liver was measured by two- and with or without treatment duration as the independent factors photon microscopy. Representative three-dimensional reconstructions or by linear regression analysis regressing liver volume on WM-8014 of whole livers from image stacks after treatment of transgenic zebrafish concentration (c). Tg(TO-krasG12V) expressing KRASGV12 and GFP (green) in the liver or

29. Serrano, M., Lin, A. W., McCurrach, M. E., Beach, D. & Lowe, S. W. Oncogenic ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 88, 593–602 (1997).

Extended Data Fig. 7 | Medicinal chemistry optimization of WM-8014, Supplementary Table 5. PDB: 6CT2. c, Space-filling model showing designed to reduce plasma-protein binding, resulted in compound WM-1119 in the acetyl-CoA-binding pocket of MYSTCryst. d, WM-1119 WM-1119. a, SPR binding data for the interaction of WM-1119 with bound to MYSTCryst with the OMIT electron density map contoured to immobilized KAT6A, KAT7 and KAT5, compared with the interaction of 3σ shown in green. e, Ribbon diagram of MYSTCryst showing key amino WM-8014. b, Crystal structure of WM-1119 bound to the MYST lysine acids interacting with WM-1119, in stick fashion with element colouring. acetyltransferase domain (MYSTCryst). Ribbon diagram of MYSTCryst Hydrogen bonds are shown as dashed lines. f, Schematic diagram of (blue) showing WM-1119 (yellow, with element colouring) bound to protein–ligand interactions (LIGPLOT)21 showing interactions between the acetyl-CoA-binding site. Data collection and refinement statistics the compound WM-1119 and amino acids within the acetyl-CoA-binding of the WM-1119 co-crystal structures (2.13 Å resolution) are listed in pocket of the MYST domain derived from the crystal structure.

Extended Data Fig. 8 | WM-1119 causes retention of cells in the G1 and double-negative (DN, early G1). Dot plots are shown for DMSO and phase of the cell cycle. a, WM-1119 causes cell cycle arrest in MEFs grown 10 µM WM-2474 control treatment groups, and after treatment with in 3% O2. Epifluorescence phase contrast images of Fucci MEFs after 1 µM and 2.5 µM active compound WM1119. d, Percentage of cells in 8 days of treatment with 10 µM WM-1119 compared to 10 µM control each phase of the cell cycle, quantified for all treatment groups. A higher WM-2474-treated cells. b, WM-1119 was tested at concentrations from proportion of WM-1119-treated cells is in mid–late G1. n = 3 independent 1 to 10 µM, compared to DMSO or 10 µM inactive compound WM-2474. MEF isolates. Data are mean ± s.e.m. and were analysed by two-way The cell number under each condition was assessed at passage. c, Flow- ANOVA (b) or one-way ANOVA followed by Bonferroni post hoc test (d) cytometry analysis of Azami Green (mAG1; mid-S, G2, M), Kusabira with treatment and with or without time as the independent factors. Orange (mKO2; mid–late G1), double-positive yellow (early S)

Extended Data Fig. 9 | Characterization of WM-1119 and lymphoma gene expression profiles. d, Mean-difference plot of treatment log2 fold cell line EMRK1184. a, Pharmacokinetic parameters for WM-1119 changes versus average log2 expression for gene expression changes in in mice following intraperitoneal injection. Note that the plasma the EMRK1184 lymphoma cell line after treatment for 3 days and 6 days concentration falls below 1 μM after 4 h. Data of n = 2 mice are shown. with WM-1119 or DMSO vehicle control. Differentially expressed genes b, Characterization of the Eµ-Myc lymphoma cell line EMRK1184. Left, are highlighted (FDR < 0.05). e, mRNA levels assessed by RNA-seq of Western blot of p53 and p19ARF. The negative control cell line EMRK1263 EMRK1184 cells treated with WM-1119 or vehicle. mRNA levels for lacks the ARF (p19ARF) band. Upregulation of p53 protein levels in Cdkn2a (coding for p16INK4A/p19ARF), Cdkn2b and Cdkn1a are shown. positive control cell line EMRK1172 indicates non-functional p53 f, Western blot and densitometry analysis showing p16INK4A and p19ARF (commonly mutations in the DNA-binding domain). Right, EMRK1184 protein in EMRK1184 treated with WM-1119 or vehicle for 3 days. Each cells were sensitive to nutlin-3a-induced cell death, indicating intact lane represents one independent culture, a total of 6 lanes (= 6 cultures) p53. By contrast, EMRK1172 cells are insensitive to nutlin-3a. p53 exon are shown. Data are mean ± confidence interval (a) or ± s.e.m. (e, f). sequencing of EMRK1184 using the MiSeq system (Illumina) confirmed Data in b were derived from three (EMRK1172) and two (EMRK1184) wild-type p53 exon sequences. c, Multidimensional scaling plot showing independent cell culture experiments, reflected by the individual data two-dimensional clustering of the EMRK1184 lymphoma cell culture points. Data in c–e were derived from three independent cultures per expression profiles. EMRK1184 lymphoma cells were treated for either treatment group and analysed as described under RNA-seq in the 3 days or 6 days, in triplicate, with WM-1119 or vehicle before RNA-seq. Supplementary Methods. Data in f were analysed by one-way ANOVA Distances on the plot corresponding to leading log2 fold change between followed by Bonferroni post hoc test.

Extended Data Fig. 10 | WM-1119 is effective in inhibiting tumour normal splenic B cell populations, which are CD19+IgD+. e, Intracellular progression. a, Tumour development monitored by luciferase activity flow-cytometry analysis of H3K9ac in tumour cells. Left, the histogram and bioluminescence imaging. Lateral images of mice treated four times shows H3K9ac levels in the remaining tumour cells (CD19+IgM−) in per day with either vehicle or WM-1119 between day 7 and day 14 after spleens of the WM-1119-treated mice (red profile) compared to the injection with tumour cells. Baseline tumour burden is shown at higher vehicle-treated mice (blue profile). The shift in the red (WM-1119-treated) sensitivity setting for day 3 (before treatment) in Fig. 4. Here, images at profile compared to the blue (vehicle-treated) profile indicates a reduction days 7, 10, 12 and 14 after tumour cell transplant are shown on the same, in signal. Right, the median fluorescence intensity (mean ± s.e.m) is less-sensitive scale. Mice are imaged in the same order. Red boxes indicate shown in the bar graph. f, Peripheral blood analysis of vehicle- or the area used for quantification. b, Mouse body weights are not affected WM-1119-treated mice. The cohort of mice that was treated three times by treatment three or four times per day. c, Concentration of WM-1119 per day is compared to the cohort that was treated four times per day. in peripheral blood and spleen 6 h after the final injection (four times per Images representative of n = 9 mice per treatment group in the four-times- day; n = 6 mice per treatment group). d, Flow-cytometry analysis of total per-day treatment regime (a). n = 3 mice per treatment group (b, d–f) and spleen cells from vehicle- or WM-1119-treated groups (four times per n = 6 mice per treatment group in (c). Data are mean ± s.e.m. and were day; analysis of spleens assayed in a to identify tumour cells independently analysed by one-way ANOVA with treatment as the independent factor of luciferase expression). The lymphoma cell line EMRK1184 has a cell followed by Bonferroni post hoc test (b), or two sided t-test (c, d, f) or surface phenotype of CD19+IgM−IgD−. Flow cytometry was used to one-sided t-test (e). quantify the CD19+IgD− population; this can be distinguished from

Corresponding author(s): Tim Thomas, Anne Kathrin Voss, Jonathan Baell

Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

Statistical parameters When statistical analyses are reported, confirm that the following items are present in the relevant location (e.g. figure legend, table legend, main text, or Methods section). n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement An indication of whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistics including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Clearly defined error bars State explicitly what error bars represent (e.g. SD, SE, CI)

Our web collection on statistics for biologists may be useful.

Software and code Policy information about availability of computer code Data collection Mass spectrometry acquisition was performed using the Agilent Mass Hunter Data Acquisition software ver. B.05.00 Build 5.0.5042.2 and analysis was performed using Mass Hunter Qualitative Analysis ver. B.05.00 Build 5.0.519.13. X-ray diffraction data for all crystals were collected at the Australian Synchrotron (Clayton, Victoria, Australia) using Blue-ice software.

Data analysis All software used is commercially or publicly available and described in the Methods section. Data displayed in graphs were generally analyzed using Stata/IC 12.1 (StataCorp) or for dose-response curves using GraphPad Prism 7b (GraphPadSoftware Inc.).

RNA-seq data, crystallography data and medicinal chemistry data were analyzed with commercially or publicly available software, also as described in the Method section: SPR. Scrubber software (version 2c) was utilized for data processing and analysis (www.biologic.com.au). Crystallization studies, as described in the methods section: the structures were determined by molecular replacement with PHASER using search models. Structures were refined through cycles of manual building using COOT and refinement April 2018 using PHENIX Refine. Restraint libraries were developed through PHENIX eLBOW32 Structure validation was monitored with MolProbity. RNA-seq: Rsubread was used to map reads to the mouse genome. Differential expression analyses were undertaken using the edgeR and limma software packages. Molecular signatures were analyzed using ROAST. Differentially expressed genes were prioritized using TREAT.

For more details see methods section.

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

1 Data nature research | reporting summary Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

The RNA sequencing data of MEFs treated with WM-8014, WM-2474 and DMSO, of MEFs from Kat6a–/– and wild type embryos and of lymphoma cell line EMRK1184 with vehicle and WM-1119 have been submitted to the GEO database under accession number GSE108244. The crystal structure data for the MYST domain in complex with WM-8014, acetyl-CoA and WM-1119 have been submitted under accession numbers PDB accession codes 6BA2, 6BA4 and 6CT2, respectively. Source data for all graphs are provided in Supplementary File “Baell Thomas source data of all graphs April 2018.xlsx”.

Field-specific reporting Please select the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/authors/policies/ReportingSummary-flat.pdf

Life sciences study design All studies must disclose on these points even when the disclosure is negative. Sample size We used up to three technical replicates of three to twenty biological replicates per experimental group per experiment. Each biological replicate represents an individual animal or a primary cell isolate from an individual animal. In compliance with animal ethics regulations, the numbers were the minimum numbers that allowed us to obtain a statistically meaningful results. They were smaller where large differences between experimental groups and small variances within experimental groups were observed.

Data exclusions No data were excluded.

Replication Experiments were replicated at least twice, as indicated in the figure legends. Experiment using animals or cells used a minimum of 3 independent biological replicates and, in the case of cell based assays 3 technical replicates.

Randomization Age and sex matched samples were randomly assigned to experimental and control groups.

Blinding Experimental results were obtained by automated methods (cell counter; flow cytometer; IVIS bioluminescence imaging; NextSeq sequencing; RT-qPCR; SPR).

Reporting for specific materials, systems and methods

Materials & experimental systems Methods n/a Involved in the study n/a Involved in the study Unique biological materials ChIP-seq Antibodies Flow cytometry Eukaryotic cell lines MRI-based neuroimaging Palaeontology Animals and other organisms Human research participants April 2018 Unique biological materials Policy information about availability of materials Obtaining unique materials All unique materials are readily available from the authors.

2 Antibodies nature research | reporting summary Antibodies used The primary antibodies used are: • rabbit monoclonal anti-H3K9ac (Cell signaling Technology, Danvers Massachusetts: C5B11) Lot : 11. Dilution for western blot:1:5000 Dilution for ChIP: 5 ul per 40 ul chromatin. Dilution for FACS: 1/500 • rabbit monoclonal anti-H3K14ac (Cell signaling Technology, Danvers Massachusetts: D4B9, #7627) Lot: 4. Dilution for western blot 1:5000 Dilution, for FACS: 1/500. • rabbit polyclonal anti- H4 (Millipore, Temecula California: 05-858; 1:30,000). • anti-γH2A.X (serine 139; clone: JBW301) biotin-conjugated antibody (1:400, Merck, Kenilworth, NJ, USA, cat: 16-193) • anti-BrdU Flow Kit (BD Biosciences, 552598, diluted according to manufacturer's instructions) • CD19-PeCy7 (clone 1D3, BD Biosciences # 561739, used at 1:200) • B220-A700 (clone 16A, conjugated at WEHI, used at 1:200) • IgM-A647 (clone II/41, conjugated at WEHI, used at 1:200) • IgD-PE (clone 11-26c, conjugated at WEHI, used at 1:200) • isotype control IgG antibody (clone DA1E, Rabbit mAb IgG XP® Isotype Control,#3900, Cell Signaling, lot49, Lot 29, Dilution for FACS: 1:500)

Validation High-quality commercial antibodies tested by the source companies were used. Wherever possible monoclonal antibodies were used to ensure reproducibility. • rabbit monoclonal anti-H3K9ac (Cell signaling Technology, Danvers Massachusetts: C5B11). Specificity determined by western blot. • rabbit monoclonal anti-H3K14ac (Cell signaling Technology, Danvers Massachusetts: D4B9, #7627) specificity determined by western blot. • rabbit polyclonal anti- H4 (Millipore, Temecula California: 05-858). Manufactures report: "Routinely evaluated by western blot: This lot detected unmodified recombinant histone H4 protein (Catalog # 14-412) and histone H4 in acid extracts from both untreated HeLa cells and cells treated with sodium butyrate (hyper-acetylated) or colcemid (hyper-phosphorylated)". • anti-γH2A.X (serine 139; clone: JBW301) biotin-conjugated antibody (Merck, Kenilworth, NJ, USA, cat: 16-193). Manufactures report: "Anti-phospho-Histone H2A.X (Ser139), clone JBW301 is a Mouse Monoclonal Antibody validated in ChIP, ICC, IF, WB. This purified mAb is highly specific for phospho-Histone H2A.X (Ser139) also known as H2AXS139p".

Flow cytometry to identify blood cell populations via cell surface markers used validated monoclonal antibodies either from commercial suppliers or produced from hybridomas in house. • CD19-PeCy7 (clone 1D3, BD Biosciences #552854), • B220-A700 (clone 16A, conjugated at WEHI) • IgM-A647 (clone II/41, conjugated at WEHI) • IgD-PE (clone 11-26c, conjugated at WEHI) • isotype control IgG antibody (clone DA1E, Rabbit mAb IgG XP® Isotype Control,#3900, Cell Signaling)

Eukaryotic cell lines Policy information about cell lines Cell line source(s) Mouse primary cell isolates (in some cases propagated) were used exclusively in this study. Mouse embryonic fibroblasts and lymphoma cells from mice carrying from Myc-driven lymphoma were isolated for this study and used.

Authentication Mouse primary cells were isolated from mice and used at an early passage number. RNA-sequencing experiment confirmed mouse mRNA exclusively in the samples. Cell isolated from genetically modified mice were genotyped by PCR.

Mycoplasma contamination All cell isolates tested negative for mycoplasma.

Commonly misidentified lines No commonly misidentified cell lines were used. (See ICLAC register)

Animals and other organisms Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research Laboratory animals Laboratory mice of the strains: C57BL/6, B6(Cg)-Tyrc-2J/J (B6-albino) were 6 to 8 weeks old. Mid-gestation embryos were used as the source of cells from the following strains of mice Moz-lox/lox;Rosa26-CreERT2, Eμ-Myc transgenic mice, Ink4a-Arf–/–, and p53–/– . Two day old Zebrafish Tg(TO-krasG12V) and Tg(lfabp10:RFP;elaA:EGFP) were used, as described in the Methods section, main text and figure legends.

Wild animals No wild animals were used. April 2018

Field-collected samples No field collected samples were used.

3 Flow Cytometry nature research | reporting summary Plots Confirm that: The axis labels state the marker and fluorochrome used (e.g. CD4-FITC). The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers). All plots are contour plots with outliers or pseudocolor plots. A numerical value for number of cells or percentage (with statistics) is provided. Methodology Sample preparation Tissues were processed and erythrocytes lysed as described in reference 11. WBC counts were obtained using a Countess® Automated Cell Counter (ThermoFisher) for spleen and bone marrow and blood counts using the Advia 3120 analyzer (Bayer) .

Instrument BD Biosciences Fortessa and LSRII were used.

Software Data were analysed using FlowJo 10.3 (Treestar Inc.).

Cell population abundance Cell populations were not sorted. Flow cytometric analysis was performed as described in the methods section.

Gating strategy The gating strategy used in flow cytometry is shown in the Supplementary Data Section. Fluorescence minus one (FMO) controls, compared to unstained cells, were used to define negative cell populations and set gates for analysis. Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information. April 2018