Durable tumor regression in genetically altered malignant rhabdoid tumors by inhibition of methyltransferase EZH2 Sarah K. Knutson1, Natalie M. Warholic1, Tim J. Wigle, Christine R. Klaus, Christina J. Allain, Alejandra Raimondi, Margaret Porter Scott, Richard Chesworth, Mikel P. Moyer, Robert A. Copeland, Victoria M. Richon2, Roy M. Pollock, Kevin W. Kuntz, and Heike Keilhack3 Epizyme, Inc., Cambridge, MA 02139 Edited* by Stuart H. Orkin, Children’s Hospital and the Dana Farber Cancer Institute, Harvard Medical School and Howard Hughes Medical Institute, Boston, MA, and approved March 28, 2013 (received for review February 28, 2013) Inactivation of the switch/sucrose nonfermentable complex com- and functional studies in cell culture demonstrated that SMARCB1 ponent SMARCB1 is extremely prevalent in pediatric malignant loss leads to decreased expression of cell cycle inhibitors (11), rhabdoid tumors (MRTs) or atypical teratoid rhabdoid tumors. This tumor suppressors like BIN1 (12), and genes of neuronal dif- alteration is hypothesized to confer oncogenic dependency on EZH2 ferentiation (13, 14), while hedgehog and MYC pathway genes in these cancers. We report the discovery of a potent, selective, and were up-regulated (14, 15). Homozygous SMARCB1 knockout orally bioavailable small-molecule inhibitor of EZH2 enzymatic activ- mice are embryonically lethal, but SMARCB1-heterozygous mice ity, (N-((4,6-dimethyl-2-oxo-1,2-dihydropyridin-3-yl)methyl)-5-(ethyl are viable and develop sarcomas that closely resemble human (tetrahydro-2H-pyran-4-yl)amino)-4-methyl-4′-(morpholinomethyl)- MRTs with the second allele of SMARCB1 spontaneously lost [1,1′-biphenyl]-3-carboxamide). The compound induces apoptosis (16). SMARCB1-conditional inactivation in T cells leads to fully and differentiation specifically in SMARCB1-deleted MRT cells. Treat- penetrant T-cell lymphomas at a median age of onset of 11 wk ment of xenograft-bearing mice with (N-((4,6-dimethyl-2-oxo-1,2- (17). Interestingly, tumorigenesis can be completely suppressed by dihydropyridin-3-yl)methyl)-5-(ethyl(tetrahydro-2H-pyran-4-yl)amino)- tissue-specific codeletion of EZH2, suggesting an antagonistic in- PHARMACOLOGY 4-methyl-4′-(morpholinomethyl)-[1,1′-biphenyl]-3-carboxamide) leads teraction between PRC2 and SWI/SNF. Indeed, EZH2 expres- to dose-dependent regression of MRTs with correlative diminution sion is elevated in primary SMARCB1-deficient tumors, and of intratumoral trimethylation levels of lysine 27 on histone H3, and polycomb target genes are also broadly repressed in such tumors prevention of tumor regrowth after dosing cessation. These data as well as in SMARCB1 knockout mouse embryonic fibroblasts. demonstrate the dependency of SMARCB1 mutant MRTs on EZH2 Through iterative medicinal chemistry, we have developed a enzymatic activity and portend the utility of EZH2-targeted drugs selective EZH2 inhibitor with favorable pharmacological prop- fi for the treatment of these genetically de ned cancers. erties. The compound was used to test whether SMARCB1- deleted MRTs are selectively sensitive to EZH2 inhibition in vitro epigenetic cancer therapy | EZH2 inhibitor and in vivo, which would suggest new treatment modalities for such genetically defined cancer types. osttranslational modifications of core histone proteins of Pchromatin play important roles in controlling the fidelity of Results gene transcription patterns in cells (1). Paramount among these EPZ-6438 Is a Potent and Selective Inhibitor of EZH2. High-throughput transcription-controlling modifications is methylation events at screening afforded a pyridone-containing EZH2 inhibitor se- lysine and arginine residues, catalyzed by histone methyltrans- ries (5), and through iterative medicinal chemistry, we developed ferases (HMTs) (2). EZH2 is the catalytic subunit of the multi- (N-((4,6-dimethyl-2-oxo-1,2-dihydropyridin-3-yl)methyl)-5-(ethyl protein HMT complex known as polycomb repressive complex 2 (tetrahydro-2H-pyran-4-yl)amino)-4-methyl-4′-(morpholinomethyl)- (PRC2), catalyzing the methylation of lysine 27 of histone H3 [1,1′-biphenyl]-3-carboxamide) (EPZ-6438) (Fig. 1A), a compound (H3K27); trimethylation of H3K27 leads to repression of gene with superior potency and pharmacokinetic properties relative to our expression (3). EZH2 has been implicated in several cancer types previously described tool compound 1-cyclopentyl-N-((4,6-dimethyl- by mutation, amplification, and/or overexpression (4). For in- 2-oxo-1,2-dihydropyridin-3-yl)methyl)-6-(4-(morpholinomethyl) stance, heterozygous EZH2 mutations at residues within the cat- phenyl)-1H-indazole-4-carboxamide (5) (EPZ005687). EPZ-6438 alytic (Su[var]3-9, enhancer of zeste, trithorax) (SET) domain inhibited the activity of human PRC2-containing wild-type EZH2 have been observed in 10% of non-Hodgkin lymphomas and can with an inhibition constant (Ki) value of 2.5 ± 0.5 nM, and similar drive H3K27 hypertrimethylation, abnormal gene expression, and potency was observed for EZH2 proteins bearing all known lymphomagenesis. We and others previously reported that selec- tive inhibition of EZH2 results in selective killing of lymphoma EZH2 cells bearing mutations, suggesting that EZH2 enzymatic Author contributions: S.K.K., R.C., M.P.M., V.M.R., K.W.K., and H.K. designed research; activity is a required driver of proliferation in the mutant-bearing S.K.K., N.M.W., T.J.W., and H.K. performed research; C.R.K., C.J.A., A.R., and K.W.K. con- cells (5–7). tributed new reagents/analytic tools; S.K.K., N.M.W., T.J.W., M.P.S., R.C., M.P.M., R.A.C., Gene expression is also regulated by remodeling of nucleo- V.M.R., R.M.P., K.W.K., and H.K. analyzed data; and S.K.K., R.A.C., R.M.P., and H.K. wrote somes in an ATP-dependent manner (8). Of the ATP-dependent the paper. fl chromatin remodelers, switch/sucrose nonfermentable (SWI/ Con ict of interest statement: All authors except V.M.R. are employees of Epizyme, Inc. SNF) complexes are emerging as bona fide tumor suppressors, as *This Direct Submission article had a prearranged editor. specific inactivating mutations in several SWI/SNF subunits are Freely available online through the PNAS open access option. found in human cancers (9). For instance, the SMARCB1 sub- 1S.K.K. and N.M.W. contributed equally to this work. unit (also known as SNF5, INI1, or BAF47) is inactivated via 2Present address: Department of Oncology, Sanofi, Cambridge, MA 02139. biallelic mutations in nearly all malignant rhabdoid tumors (MRTs) 3To whom correspondence should be addressed. E-mail: [email protected]. and atypical teratoid rhabdoid tumors (ATRTs), aggressive cancers This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. of young children with no effective therapy (10). Gene expression 1073/pnas.1303800110/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1303800110 PNAS Early Edition | 1of6 A B O 1100 nM N naive naive DMSO 0.006 nM EPZ-6438 N O G401 H3K27Me3 SMARCB1 deleted H3 O HN O H3K27Me3 RD HN SMARCB1 WT H3 C D G401, Days 0-7 µM G401, Days 7-14 1.E+07 EPZ-6438 1.E+07 1.E+06 1.E+06 10 1.E+05 1.E+05 1.E+04 1.E+04 2.5 1.E+03 Luminescence 01234567 Luminescence 7 8 9 1011121314 Days of incubaon 0.625 Days of incubaon 0.16 RD, Days 0-7 0.04 RD, Days 7-14 1.E+07 1.E+07 1.E+06 0.010 1.E+06 1.E+05 0.002 1.E+05 1.E+04 1.E+04 Luminescence 01234567 0.0006 Luminescence 7 8 9 1011121314 Days of incubaon DMSO Days of incubaon Fig. 1. Effects of EPZ-6438 on cellular global histone methylation and cell viability. (A) Chemical structure of EPZ-6438. (B) Concentration-dependent in- hibition of cellular H3K27Me3 levels in G401 and RD cells. (C) Selective inhibition of proliferation of SMARCB1-deleted G401 cells by EPZ-6438 in vitro (measured by ATP content). G401 and RD cells were replated at the original seeding densities on day 7. Each point represents the mean for each concentration (n = 3). (D) Proliferation IC50 values (day 14) for SMARCB1 wild-type and mutant cells. Compound incubations for each experiment were performed in triplicate, and symbols represent the mean of two experiments for all cell lines. The horizontal lines represent the median. Mean calculation of duplicate experiments was not possible for RD and SJCRH30 cells; individual values are shown (Table S2). lymphoma change-of-function mutations (Table S1). As previously is, EPZ-6438 binding is not mutually exclusive or synergistic with described for EPZ005687 (5), steady-state analysis of the enzyme binding of the peptide/nucleosome substrate of the enzymatic kinetics revealed that the present compound, EPZ-6438, inhibits reactions) (Fig. S1B) (18). We also assessed inhibition by EPZ- EZH2 in a manner competitive with the substrate S-adenosylme- 6438 against a panel of HMTs other than EZH2 encompassing thionine (SAM) (that is, EPZ-6438 and SAM binding to EZH2 is both lysine and arginine HMTs. EPZ-6438 displayed a 35-fold mutually exclusive, such that only one or the other ligand can bind selectivity versus EZH1 and >4,500-fold selectivity relative to 14 but both cannot bind to the enzyme simultaneously) (Fig. S1A), other HMTs tested (Table S1). These in vitro properties (i.e., and noncompetitive with the peptide or nucleosome substrate (that mechanism of action, specificity, cellular potency) of EPZ-6438 Fig. 2. EPZ-6438 induces G1 arrest and apoptosis in SMARCB1-deleted MRT cells. Cell cycle analysis (by flow cytometry) and determination of apoptosis (by TUNEL assay) in RD (A) or G401 cells (B) during incubation with either vehicle or 1 μM EPZ-6438 for up to 14 d. G1 arrest was observed as of day 7, and apoptosis was induced as of day 11.