Development of a selective CDK2-E inhibitor in CCNE-aberrant cancers 1279

Yoon J. Choi1, Steve Wenglowsky1 , Victoria Brown1 , Neil Bifulco1, Yeon S. Choi1, Jian Guo1, Megan Hatlen1, Joseph Kim1, Tim LaBranche1, Riadh Lobbardi1, Emanuele Perola1, Emily Rozsahegyi1, Michelle Maynard1, Phil Ramsden1, Grace Silva1, Faith Stevison1, Richard Vargas1, Ruduan Wang1, Doug Wilson1, Rich Woessner1, Dean Zhang1, Rob Meissner1, Klaus Hoeflich1, Marion Dorsch1

1Blueprint Medicines Corporation, Cambridge, Massachusetts, USA

Figure 5: Selective CDK2 inhibitor, BLU0298 selectively inhibits (A) Figure 7: Selective CDK2 inhibitor achieves similar efficacy to 1296P Background Results proliferation in CCNE1-amplified cells by (B) arresting cells in G1/S PF-06873600, a CDK2/4/6 inhibitor, or to chemotherapy, but maintains stable body weight • Cyclin E1 and E2 are core cell cycle regulators, that when bound activate CDK2 Figure 3: CDK2 catalytic activity is essential in CCNE1-amplified A. Proliferation B. Cell cycle profile in OVCAR-3 cells 1 A. In vivo efficacy in OVCAR-3 B. BLU1954-treated mice maintained healthy BW (cyclin-dependent kinase 2), driving G1/S progression of the cell cycle (Figure 1A) cell lines [nM] G1 S G2/M • A broad range of aggressive cancers overexpress and/or harbor CCNE gene OVCAR-3 CCNE1-AMP amplifications, thus CDK2 is a potentially impactful therapeutic target A. Genetic validation of dependency B. Catalytic dependency 100 TOV-21G CCNE1 WT 25 4000 Vehicle 15 Vehicle PO BID (28 Days) – In subsets of gynecological, breast, gastric, and other cancers, CCNE amplification 1. TOV21G 4. HCC1569 7. KURAMOCHIa BLU1954 100 mg/kg QD BLU1954 100 mg/kg PO QD (28 Days) correlates with overexpression (Figure 1B)2 and has been associated with poor survival in 2. COV644 5. OVCAR3 8. FUOV1 100 PF-06873600 30 mg/kg BID PF-3600 30 mg/kg PO BID (28 Days) SD 3500 3 3. MDA-MB-157 6. OVISE 9. ONCODG1 Cisplatin 2.5 mg/kg QW Cisplatin 2.5 mg/kg IP QW (4 Weeks) ovarian cancer (Figure 1C) 250 10

50 BLU0298

125 ), Cyclin E amplification/overexpression has been reported as a potential mechanism of 100 3 3000

– CCNE1 WT 500 SD 4,5 SD resistance to CDK4/6 therapies in ER-positive HER2-negative breast cancer ± Mean 100 Inhibition (%), 2500 Dosing

• Deriving a highly selective CDK2 inhibitor, sparing CDK family members has been 0 5

SEM SEM historically challenging; however, it is critical to limit off-target CDK-driven toxicities 25 75 2000 • We report preclinical data supporting a potential best-in-class CDK2 inhibitor exhibiting CCNE1-amplified 0.0001 0.01 1 100

50 BLU0298 (µM) 50 0 Mean ± Mean Mean ± Mean 1500 single digit cellular nanomolar potency and selectivity against CDK 1, 4, 6, 7, 9 with 50 single agent anti-tumor activity 100

100 Tumor (mm volume 1000 25 -06873600 250 ꟷ5 Figure 1: Aberrant cyclin E levels inappropriately activate CDK2 and PF

SD 500 500 Relative body body weightRelative (%), change

1 3 (%), cellsMean ± Viable correlate with poor survival – 0 (%), cellsMean ± Viable 0 50 1 2 3 4 5 6 7 8 9 Control shCDK2 shCDK2 shCDK2 0 ꟷ10 A. + + P P CDK2 (CRISPR deleted gene) KD WT ± Mean 0 7 14 21 28 35 42 49 7 14 21 28

Inhibition (%), DMSO CDK2 CDK2 0 Days Days Cyclin E P Rb E2Fs (A) CRISPR screen in ovarian and breast cell lines. CDK2 was knocked-out in Cas9-expressing breast (3 and 4) regulatory 0.0001 0.01 1 100 0 50 100 subunit and ovarian cell lines (1,2, 5–9). The percentage of viable cells was determined after 21 days. (B) Proliferation in CDK2 Cyclin E CDK2 S-phase genes Cancer cell – OVCAR-3 cells after CDK2 knockdown and rescue with WT or KD mutant CDK2. Empty vector, WT or PF-06873600 (µM) % of cells C. Pharmacodynamic inhibition of pRb in OVCAR-3 catalytic complex ON proliferation T160A/D145N KD CDK2 was expressed in OVCAR-3 cells, followed by induction of CDK2-targeting shRNA. subunit (A) Dose response curves in ovarian cell lines. Ovarian cells were treated with BLU0298 or PF-06873600 for 5 days. BLU1954 PF-06873600 The percentage of growth is normalized to control at 14 days. Proliferation was assessed by CyQuant. (B) Cell cycle profile of OVCAR-3 cells. OVCAR-3 cells were treated with Vehicle 100 mg/kg QD 30 mg/kg BID CRISPR, clustered regularly interspaced short palindromic repeats; KD, kinase dead; SD, standard deviation; shCDK2, shRNA against CDK2; sh, short BLU0298 or PF-06873600 for 24 h. Cell cycle profile assessed by EdU incorporation (2 h) combined with DNA content 2 h 2 h 8 h 24 h 2 h a hairpin; WT, wild-type (i.e., non-amplified/gain). Kuramochi copy number gain. (FxCycle). Cyclin E P Rb E2Fs regulatory DMSO, dimethyl sulfoxide; Dox, doxycycline; EdU, 5-ethynyl-2'-deoxyuridine; h, hour; SD, standard deviation; WT, wild-type (i.e., non-amplified). pRbT821 subunit Figure 4: CDK2 knockdown induces pRb inhibition and irreversible CDK2 Cyclin E–CDK2 S-phase genes Blocked cancer catalytic complex OFF cell proliferation arrest in CCNE1-amplified cell lines subunit Figure 6. Selective CDK2 inhibition leads to sustained anti-tumor growth Total Rb A. Efficient shCDK2 and pRb inhibition B. Senescence induction in vivo B. CCNE amplification correlates C. CCNE amplification correlates in CCNE1-amplified models Actin shCDK2 with expressiona,2 with poor survival (ovarian)b,3 Control shCDK2 OVCAR-3 Kura COV318 TOV21G OVCAR-3 xenograft post 3-days of dosing 200.0 100 A. OVCAR-3 HCC1569 Dox – + – + – + – + (Ovarian model) (Breast cancer model) 100.0 CDK2 2200 1600 (A) Anti-tumor activity in OVCAR-3 xenograft. Mice were inoculated as in Figure 6. (B) Body weight measurement. 50.0 80 Mice were monitored over 28 days and body weight measurements were taken twice a week. (C) Pharmacodynamic 2000 inhibition. Tumor lysates were assessed by Western blot at the indicated time points (vehicle, 2 h; BLU1945 2, 8, and 20.0 pRb T821 1400 24 h; PF-06873600, 2 h) post 3-days of treatment. 60 1800

10.0 ),

Non-amplification ), BID, twice daily; BW, body weight; h, hour; IP, intraperitoneal; pRB, phosphorylated retinoblastoma protein; PO, orally; QD, once daily; QW, once a week; 3 Total Rb 3 1200 1600 Dosing SEM, standard error of mean. 5.0 40 Dosing TCGA ovarian serous P<0.001

2.0 cystadenocarcinoma (n=417) (%) Survival Actin –Dox +Dox 1400 1000 SEM 1.0 20 CCNE1-amplified CCNE1 WT OVCAR-3 SEM 1200 800

CCNE1 expression (FPKM) expression CCNE1 Amplification Conclusions and future directions 0.5 1000

0 C. Irreversible arrest Mean ± Mean 1 2 5 10 20 50 0 12 24 36 48 60 72 84 96 108120132 ± Mean 800 600 ControlTemporal KD Constant KD

CCNE1 DNA copy number Months (A) Dox inducible knockdown of CDK2. Lysates 600 Preclinical data indicate that aberrant CCNE is a predictor of response to Tumor (mm volume 125 Tumor (mm volume 400 • E2F, transcription factor; FPKM, fragments per kilobase of transcript per million; P, phosphorylation; Rb, retinoblastoma protein; TGCA; were probed by Western blot as annotated. a 400 CDK2 inhibition The Cancer Genome Atlas. The results shown here are in whole or part based upon data generated by the TCGA Research Network: (B) SA- -Gal staining in OVCAR-3 cells. Cells https://www.cancer.gov/tcga2. bModified from Etemadmoghadam et al. 20103. β 100 200 were treated with or without Dox for 6 days and 200 Targeting CDK2 in CCNE-aberrant cancer cell lines induces markers of SD • stained SA-β-Gal. 75 Figure 2: CCNE1 amplification across broad cancer types predicts (C) Confluency after CDK2 knockdown. Cells ± 0 0 senescence and irreversible arrest at G1/S 0 4 8 12 16 20 24 28 4 response to CDK2 inhibition2,6 were re-plated after 6 days of dox induction with 50 0 8 12 16 20 24 28 or without sustained knockdown (+/- dox). Days BLU0298, BLU2256, and BLU1954 are selective, potent CDK2 Mean Days •

A. CCNE1 amplification in patient samples B. Validated in broad cell line screens (%), cells Viable 25 compounds that lead to tumor growth inhibition 35 (GISTIC assessment)a,2 ~400 cell linesb,6 Dox, doxycycline; KD, knockdown; SD, standard deviation; WT, wild-type (i.e., non-amplified). 0 MKN1 Vehicle QD CDK2 inhibitors show promise as monotherapy and are under evaluation 0 OVCAR-3 Kuramochi COV318 (Gastric cancer model) • 30 UCS: Uterine carcinosarcoma BLU2256 5 mg/kg BID in combination with CDK4/6 targeted therapies or chemotherapy OV: Ovarian serous cystadenocarcinoma 900 PF-06873600 30 mg/kg BID UCEC: Uterine corpus endometrial –2 Table 1: BLU0298, BLU1954, and BLU2256 are selective, potent Taken together these results provide scientific rationale for advancing 25 carcinoma 800 • STAD: Stomach adenocarcinoma CDK2 inhibitors sparing CDK off-targets including CDK1 SARC: Sarcoma this class of compounds toward clinical development in 20 4 700 ESCA: Esophageal carcinoma – ), b c 3 (A) BLU2256 shows anti-tumor activity in CCNE-aberrant cancers BLCA: Bladder urothelial carcinoma Cellular activity IC50 (nM) Enzyme activity IC50 (nM) Dosing LUSC: Lung squamous cell carcinoma 600 xenograft models. Mice inoculated SC with 15 6 CHOL: Cholangiocarcinoma –6 Kinome S pRb OVCAR-3 (6x10 cells), HCC1569 or MKN-1 p‟X” 7 BRCA: Breast invasive carcinoma Compound (10)a (CDK2 cell) (CDK1 cell) CDK2 CDK1 CDK4 CDK6 CDK7 CDK9 SEM 500 (10x10 cells). Drug treatment (indicated by References 10 ACC: Adrenocortical carcinoma CDK2 essential double-ended arrows) was initiated when CCNE1 amplification=red MESO: Mesothelioma 400 tumors reached ~150–250mm3. The regrowth 1. Malumbres M et al. Genome Biol. 2014;15:122; 2. TCGA Research Network: https://www.cancer.gov/tcga; last accessed March 23, 2021. –8 BLU0298 0.045 4.2 380.2 2.6 233.6 377.4 275.2 6941.2 6115.1 3. Etemadmoghadam D et al. Plos One. 2010; 5(11):e15498. 4. Knudsen ES et al. ASCO Educational Book. 2020;40:115–126; 5. Ding L et al. Int J Mol

5 of the remaining tumors was monitored in the Sci. 2020;21:1960; 6. McDonald RE et al. Cell 2017;170:577–592. Avg scoresensitivity CKD2 Avg Mean ± Mean 300 absence of drug treatment. BLU1954 0.055 1.6 127.7 0.2 110.1 114.5 190.6 3928.9 849.1

0 –10 Frequency of (%) CCNE1amplificationFrequency Tumor (mm volume Acknowledgments

OV 200 BID, twice daily; QD, once daily; SC, subcutaneous;

LGG

UCS

ACC

UVM

GBM

LIHC

KIRP

KIRC

KICH

LAML

BLCA

DLBC

LUAD

LUSC

STAD

THCA

PAAD

ESCA

CHOL

TGCT

PRAD

READ

CESC SARC BRCA BLU2256 0.040 2.6 133.7 0.1 152.9 116.9 393.2 4826.4 9063.2

THYM SEM, standard error of mean.

HNSC

UCEC

PCPG

SKCM COAD MESO Medical writing support was provided by Deborah Cantu, PhD, and editorial support was provided by Michelle Seddon, Dip Psych all of Paragon, 100 Knutsford, UK, supported by Blueprint Medicines Corporation, Cambridge, MA, according to Good Publication Practice guidelines. (A) CCNE1 GISTIC data across tumor types. CCNE1 amplification frequency represented as percentage of total PF-06873600 0.094 1.9 42.8 0.4 31.8 1.5 5.7 3312.9 2481.1 patient samples. (B) CDK2 essentiality scores replotted from Project DRIVE. CCNE1-amplified cell lines (red bars) 0 Disclosures and non-amplified lines (gray bars) plotted against CDK2 essentiality score. Blue dotted line represents cut-off for aKinome S(10): fraction of kinases with <10 percentage of control at 3 uM among all the kinases tested, measured by KINOMEscan platform against 468 kinases. bpRb protein was assessed in synchronized OVCAR-3 cells to reflect CDK2 cellular potency; p‟X” a validated CDK1 specific substrate was 0 3 6 9 12 15 18 21 This research was funded by Blueprint Medicines Corporation. Blueprint Medicines Corporation reviewed and provided feedback on the poster. The essentiality. c assessed in asynchronous OVCAR-3 cells. Enzyme activities IC50 were measured at 1 mM ATP using canonical CDK/Cyclin pairs: CDK2/Cyclin E1; authors had full editorial control of the poster and provided their final approval of all content. YJC, SW, VB, NB, YSC, JG, MH, JK, TLaB, RL, EP, ER, MM, GISTIC, Genomic Identification of Significant Targets in Cancer. aThe results shown here are in whole or part based upon data generated by the TCGA CDK1/Cyclin B1; CDK4/Cyclin D1; CDK6/Cyclin D3; CDK7/Cyclin H1/MNAT1; CDK9/Cyclin T1. ATP, adenosine triphosphate; CDK, cyclin-dependent Days PR, GS, FS, RV, RW, DW, RiW, DZ, RM, MD are employees of and hold equity interest in Blueprint Medicines Corporation. KH was an employee of 2 b 6 Research Network: https://www.cancer.gov/tcga . Modified Source data: McDonald et al. Project DRIVE, Cell 2017 . kinases; IC50, half-maximal inhibitory concentration; pRB, phosphorylated retinoblastoma protein. Blueprint Medicines Corporation at the time of the work.

Presented at AACR 2021, April 9–14, 2021, Virtual Format. Please contact [email protected] for permission to reprint and/or distribute