Characterization of a highly selective inhibitor of the Aurora kinases

Bioorganic & Medicinal Chemistry Letters 27 (2017) 4405–4408

Contents lists available at ScienceDirect

Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl

Characterization of a highly selective inhibitor of the Aurora kinases Fleur M. Ferguson a,h, Zainab M. Doctor a,h, Apirat Chaikuad b,d, Taebo Sim e,f, Nam Doo Kim g, Stefan Knapp b,c,d, Nathanael S. Gray a,⇑ a

Department of Cancer Biology, Dana-Farber Cancer Institute, Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA Structural Genomics Consortium, University of Oxford, Old Road Campus Research Building, Roosevelt Drive, Oxford OX3 7DQ, United Kingdom c German Cancer Consortium (DKTK), Frankfurt site, Germany d Institute for Pharmaceutical Chemistry and Buchmann Institute for Life Sciences, Goethe University, Max-von Laue Str. 9, 60438 Frankfurt am Main, Germany e Chemical Kinomics Research Center, Korea Institute of Science and Technology, Republic of Korea f KU-KIST Graduate School of Converging Science and Technology, Korea University, Republic of Korea g Daegu-Gyeongbuk Medical Innovation Foundation, Republic of Korea b

a r t i c l e

i n f o

Article history: Received 20 July 2017 Revised 8 August 2017 Accepted 9 August 2017 Available online 10 August 2017 Keywords: Aurora kinase Selective kinase inhibitor Pan-Aurora inhibitor Mitosis Cancer

a b s t r a c t Aurora kinases play an essential role in mitosis and cell cycle regulation. In recent years Aurora kinases have proved popular cancer targets and many inhibitors have been developed. The majority of these clinical candidates are multi-targeted, rendering them inappropriate as tools for studying Aurora kinase mediated signaling. Here we report discovery of a highly selective inhibitor of Aurora kinases A, B and C, with potent cellular activity and minimal off-target activity (PLK4). The X-ray co-crystal structure of Aurora A in complex with compound 2 is reported, and provides insights into the structural determinants of ligand binding and selectivity. Ó 2017 Elsevier Ltd. All rights reserved.

The Aurora kinases are a family of cell-cycle regulated serine/ threonine kinases which are primarily active during mitosis.1 These homologous kinases effect distinct processes via differential expression, localization and interaction partners. Aurora A localizes at the centrosome during interphase, and localizes at mitotic poles and to the spindle throughout mitosis.2 Aurora A regulates progression of mitosis and promotes centrosome maturation.1 Aurora B localizes to centromeres during metaphase as part of the chromosomal passenger complex (CPC) and remains associated with the central mitotic spindle during anaphase.3 The CPC regulates chromosome condensation, via phosphorylation of histone H3, the spindle assembly checkpoint (SAC)4 and cytokinesis.5 Unlike Aurora A and B, Aurora C is not expressed in all dividing

Abbreviations: PLK4, Polo-like kinase 4; FACS, fluorescence activated cell sorting; DMSO, dimethylsulfoxide; AML, acute myeloid leukemia; T-ALL, T-cell acute lymphoblastic leukemia; ABL1, Abelson Tyrosine-Protein Kinase 1; LRRK2, Leucine Rich Repeat Kinase 2; BRD4, Bromodomain-Containing Protein 4; ERK5, Mitogen-activated protein kinase 7; LC-MS/MS, Liquid Chromatography–Mass Spectrometry/Mass Spectrometry. ⇑ Corresponding author. E-mail address: [email protected] (N.S. Gray). h These authors contributed equally. http://dx.doi.org/10.1016/j.bmcl.2017.08.016 0960-894X/Ó 2017 Elsevier Ltd. All rights reserved.

cells, as its primary role is in male meiosis during spermatogenesis.2,6 The Aurora kinases are implicated in a variety of hematological and solid cancers. Aurora A and B are frequently overexpressed in cancer, and have been associated with aneuploidy and poor prognosis.7,8 Aurora C has also been show to function as an oncogene and overexpression has been reported in thyroid cancer tissues.6 The Aurora kinases have therefore become attractive drug targets with more than ten Aurora inhibitors undergoing clinical trials.9 In solid tumors, on-target bone marrow toxicity has precluded the use of Aurora inhibitors in cancer therapy. This is hypothesized to be due to the slower proliferation rate of cells in solid tumors relative to those in the bone marrow, and the requirement for drug exposure through several cell cycles before the maximal cytotoxic effects are realized.9 Acute hematological tumors, such as Acute Myeloid Leukaemia (AML), have higher proliferation rates and have shown more promising response rates to Aurora kinase inhibitors in the clinic.10–12 Many Aurora kinase inhibitors developed to date display polypharmacology, targeting numerous kinases.9 Whilst this may contribute to clinical efficacy, it does not facilitate detailed study of Aurora kinase mediated signaling and cellular processes. Recently, Carry et al. reported an exquisitely selective pan-Aurora inhibitor,

4406

F.M. Ferguson et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4405–4408

SAR156497. This compound displayed efficacy but had a narrow therapeutic window in colon adenocarcinoma xenograft studies.13 SAR156497 was used as a control compound in our experiments, is one of the most selective cell permeable pan-Aurora inhibitors described to date. Here we report discovery of highly selective pan-Aurora kinase inhibitors through phenotypic screening. The co-crystal structure of compound 2 bound in the ATP binding site of Aurora A is described, and provides insight into the possible structural basis for the selectivity of the interaction. To identify anti-leukemic compounds, a library of pyrimidobenzodiazepinones was tested for anti-proliferative activity in Jurkat cells, a T-cell acute lymphocytic leukemia (T-ALL) cell line. We previously used this phenotypic screening and profiling approach to identify PI3K-d/c inhibitors.14 Compound 1 was identified as a cytotoxic molecule with an IC50 of 370 nM. Kinase profiling against a panel of 468 human wild-type and mutant kinases using KINOMEscanÒ at 1 lM revealed that compound 1 is a highly selective pan-Aurora inhibitor with exclusively on-target activity detected across the panel at 1 lM compound concentration (Fig. 1). The Aurora A, B and C activity was confirmed using biochemical kinase assays (Fig. 2C). The Aurora kinases do not contain cysteine residues proximal to their ATP binding sites, therefore we hypothesized that the interaction of compound 1 with Aurora kinases is reversible, and that the acrylamide functionality could be removed without loss of potency. To confirm this we incubated compound 1 at 10-fold excess, with recombinant purified Aurora A protein, and observed no labeling by LC-MS/MS analysis. Compound 2 was synthesized

XMD12-1

and exhibited similar potency and selectivity to compound 1. Both compounds have moderate off-target activity on PLK4 (IC50 compound 1 = 43 nM, IC50 compound 2 = 82 nM) but otherwise display excellent kinome selectivity. The ABL1 Y253 F activity of compound 2 from the kinome screen was not reproduced in a biochemical kinase assay and is likely to be a false positive (IC50 > 10,000 nM). Inhibitors based on this scaffold have been reported to bind to LRRK2 and BRD4, proteins not covered in the selectivity panel. Compounds 1 and 2 were tested in biochemical assays against these targets and gratifyingly both molecules were inactive (Supporting Table 1). We have previously described pan-Aurora inhibitors based upon the same chemical scaffold (XMD12-1).15 However, compounds 1 and 2 have significantly improved kinome selectivity (Fig. 1), making them some of the most selective pan-Aurora kinase inhibitors reported to date. The cellular activity of compounds 1 and 2 were assessed by their ability to inhibit the phosphorylation of biomarkers of Aurora A and Aurora B by western blot in HCT116 cells. Compounds 1 and 2 were cell permeable and inhibited Aurora A auto-phosphorylation and Aurora B mediated histone H3 S10 phosphorylation at concentrations below 40 nM (Fig. 2A, C, Supporting Fig. 1). These compounds showed improved cellular inhibition of Aurora A auto-phosphorylation and histone H3 S10 phosphorylation compared to the control compound SAR156497.13 To investigate the effects of compounds 1 and 2 on the cell cycle, we performed FACS cell cycle analysis in HCT116 cells after 48 h of 500 nM compound treatment (Fig. 2B, Supporting Fig. 2). Treatment with compounds 1, 2, or the control pan-Aurora inhibitor

Compound 1

Compound 2

Fig. 1. A) Chemical structure and kinome-wide selectivity plot of compounds 1 and 2 compared to previously reported benzodiazepinone inhibitor XMD12-1. Assay performed by DiscovRX at 1 lM compound concentration. On target activity (Aurora A/B/C) represented by blue circles, off target activity represented by red circles. Chemical modifications to the core (red), and aniline (blue), are highlighted.

4407

F.M. Ferguson et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4405–4408

B

pH3 S10

H3

H3

tubulin

tubulin

pAurora A

pAurora A

Aurora A

Aurora A

vinculin

vinculin

500 nM

250 nM

125 nM

62.5 nM

31.2 nM

15.6 nM

DMSO

pH3 S10

7.8 nM

Compound 2 500 nM

250 nM

125 nM

62.5 nM

15.6 nM

7.8 nM

DMSO

500 nM

250 nM

125 nM

62.5 nM

31.2 nM

15.6 nM

31.2 nM

Compound 1

SAR156497 7.8 nM

DMSO

A

C Biochemicala Compound

Cellularb

Aurora A

Aurora B

Aurora C

Aurora A

Aurora B

IC50 (nM)

IC50 (nM)

IC50 (nM)

EC50 (nM)

EC50 (nM)

SAR156497

75 ± 26

34 ± 4.6

91 ± 17

106

27

1

44 ± 4.5

29 ± 2.0

63 ± 8.1

5.8

7.1

2

52 ± 3.0

41 ± 4.0

53 ± 2.4

38

11

Fig. 2. A) Western blot showing dose dependent inhibition of pAurora A and histone H3 pS10 marks in HCT116 cells. B) FACS analysis after 48 h treatment with 500 nM compound concentration in HCT116 cells. Data shown is average of three replicates ± standard error. C) Biochemical IC50 and cellular EC50 values of pan-Aurora inhibitors. a Z’LYTE kinase assay, measurements performed at km ATM concentration for each kinase. Measurements are shown ± standard error. bCellular EC50 based on 4 h treatment followed by western blot analysis of pAurora A and histone H3 pS10, representative western blot of three replicate experiments shown in this Figure, dose-response curves shown in Supporting Fig. 1.

SAR156497 led to failed cytokinesis and accumulation of cells that contain 4N DNA (polyploidy), consistent with cellular Aurora B inhibition.16 To assess the cytotoxicity of lead compound 1, we treated a panel of human cancer cell lines in dose response, and assessed cell viability after 72 h (Table 1, Supporting Fig. 3). The HeLa cell line is highly sensitive to pan-Aurora inhibition, the IC50 of compounds 1, 2 and control compound SAR156497 are below 10 nM in this cell line. In HCT116, PC3 and HEK293 cells, compound 1 is more potent than SAR156497, consistent with it’s improved cellular inhibition of the Aurora kinases signaling seen in Fig. 2A, C.16 When compound 2 was tested in the cell line panel, it was highly toxic in HeLa cells. In other cells lines, cell viability loss plateaued at 50%, when tested at concentrations up to 33.3 lM, indicating possible cell cycle arrest (Supporting Fig. 3). It is known that the time from mitotic arrest to apoptotic cell death is highly variable between cancer cell lines, which may account for some of these differences.17 Compound 1 is more active against Aurora A (cellular), and PLK4 (biochemical) than compound 2, which may account for it’s additional toxicity in this assay. Furthermore, compound 1 and SAR156497 may have additional uncharacterized off-target activity that contributes to their toxicity. To test the compounds metabolic stability, we measured the half-life of compounds 1 and 2 in mouse hepatic microsomes (1 mg/mL) and found them to have modest stability, with half-lives

of 7.5 and 2.9 min respectively, indicating these compounds may be unsuitable for in vivo studies. In order to gain structural insights into the mode of binding and to rationalize the selectivity of the molecules, we solved the cocrystal structure of compound 2 in complex with Aurora A at a resolution of 2.6 Å (Fig. 3, PDB: 5ONE). The structure of the complex of compound 2 with Aurora A revealed the expected ATP competitive binding mode of pyrimido-benzodiazepinone inhibitors, comprising two hydrogen bonds with the main chain amide and carbonyl group of the hinge residue A213. A similar binding mode was observed in docking studies for compound 1 (Supporting Fig. 4). The 4-propanoylaminobenzamid moiety is rests over the aD helix, with the propyl group oriented towards the solvent exposed area. The kink created by the benzodiazepinone ring inserts the aromatic toluene portion of this core ring system into a deep hydrophobic pocket created by L263 and an unusual inactive conformation of the activation segment which assumed a DFG ‘‘in” conformation with residues C-terminal to the DFG motif packing against the aromatic ring of the benzodiazepinone moiety. The methyl group addition to the scaffold cannot be accommodated in this region by many kinases, including ERK5, which contributes to the selectivity of compounds 1 and 2.18 Comparison of the binding mode of compound 2 with that of the Aurora A specific inhibitor Alisertib (PDB: 2X81) revealed that both inhibitors share the same hinge binding interactions. However, the different coupling of the aromatic ring systems to

4408

F.M. Ferguson et al. / Bioorganic & Medicinal Chemistry Letters 27 (2017) 4405–4408

Table 1 Anti-proliferative activity of pan-Aurora inhibitors. Compound

HCT116 IC50 (nM)

HeLa IC50 (nM)

HEK293 IC50 (nM)

PC3 IC50 (nM)

SAR156497 1 2

3700 ± 2000 840 ± 240 100a

6.0 ± 1.5 <1.0 <1.0

5900 ± 1800 250 ± 48 100a

>10,000 990 ± 1500 100a

Cell Titer Glo cell viability assay, measurements were performed in triplicate and are shown ± standard error. a Loss of cell viability plateaued at 50%.

A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.bmcl.2017.08. 016. References

Fig. 3. X-ray co-crystal structure of Aurora A in complex with compound 2, refined at 2.6 Å resolution (PDB: 5ONE). Compound 2 is shown in ball and stick representation with yellow carbon atoms. Hydrogen bonds are indicated by dotted lines. Main interacting residues are shown and labeled.

the azepine ring results in orientation of the aliserib benzazepine towards the P-loop while the fluoro-methoxyphenyl ring is oriented towards the DFG motif (Supporting Fig. 5).19 The pyrimido-benzodiazepinone scaffold has also been utilized for the development of a selective ERK5 inhibitor by our laboratory.20,21 Superimposition of the structure of 2 with Aurora A with the complex of compound 25 with ERK5 (PDB: 4B99) showed that insertion of a proline residue (P214) in the hinge region results in a considerable re-orientation of 2 in Aurora A. In addition, the introduction of a cyclopentyl moiety to the nitrogen of the benzodiazepinone ring in compound 25 most likely results in steric exclusion of the ERK5 selective inhibitor from the Aurora A ATP binding site (Supporting Fig. 5). We have previously described multi-targeted Aurora inhibitors based upon a pyrimido-benzodiazepinone scaffold,15 in this work we characterize advanced analogs from this scaffold series, with vastly improved selectivity. We envision these molecules will be invaluable for studying Aurora kinase signaling and the therapeutic potential of pan-Aurora inhibition in cellular studies. We also use structural studies to identify key selectivity hot-spots of the Aurora A binding pocket, providing a blueprint for improving selectivity of Aurora kinase targeting inhibitors. Author contributions All authors have given approval to the final version of the manuscript. Acknowledgments We would like to acknowledge funding from NIH R01 U54 HL127365, the Linde center for chemical biology and the KU-KIST Graduate School of Converging Science and Technology Program.

1. Goldenson B, Crispino JD. The aurora kinases in cell cycle and leukemia. Oncogene. 2015;34:537–545. 2. Carmena M, Earnshaw WC. The cellular geography of aurora kinases. Nat Rev Mol Cell Biol. 2003;4:842–854. 3. Carmena M, Wheelock M, Funabiki H, Earnshaw WC. The chromosomal passenger complex (CPC): from easy rider to the godfather of mitosis. Nat Rev Mol Cell Biol. 2012;13:789–803. 4. Sampath SC, Ohi R, Leismann O, Salic A, Pozniakovski A, Funabiki H. The chromosomal passenger complex is required for chromatin-induced microtubule stabilization and spindle assembly. Cell. 2004;118:187–202. 5. Terada Y, Tatsuka M, Suzuki F, Yasuda Y, Fujita S, Otsu M. AIM-1: a mammalian midbody-associated protein required for cytokinesis. EMBO J. 1998;17:667–676. 6. Ulisse S, Delcros JG, Baldini E, et al. Expression of Aurora kinases in human thyroid carcinoma cell lines and tissues. Int J Cancer. 2006;119:275–282. 7. Lassus H, Staff S, Leminen A, Isola J, Butzow R. Aurora-A overexpression and aneuploidy predict poor outcome in serous ovarian carcinoma. Gynecol Oncol. 2011;120:11–17. 8. Takeshita M, Koga T, Takayama K, et al. Aurora-B overexpression is correlated with aneuploidy and poor prognosis in non-small cell lung cancer. Lung Cancer. 2013;80:85–90. 9. Bavetsias V, Linardopoulos S. Aurora kinase inhibitors: current status and outlook. Front Oncol. 2015;5:278. 10. Lowenberg B, Muus P, Ossenkoppele G, et al. Phase 1/2 study to assess the safety, efficacy, and pharmacokinetics of barasertib (AZD1152) in patients with advanced acute myeloid leukemia. Blood. 2011;118:6030–6036. 11. Kantarjian HM, Martinelli G, Jabbour EJ, et al. Stage I of a phase 2 study assessing the efficacy, safety, and tolerability of barasertib (AZD1152) versus low-dose cytosine arabinoside in elderly patients with acute myeloid leukemia. Cancer. 2013;119:2611–2619. 12. Choudary I, Barr PM, Friedberg J. Recent advances in the development of Aurora kinases inhibitors in hematological malignancies. Ther Adv Hematol. 2015;6:282–294. 13. Carry JC, Clerc F, Minoux H, et al. SAR156497, an exquisitely selective inhibitor of aurora kinases. J Med Chem. 2015;58:362–375. 14. Ferguson FM, Ni J, Zhang T, et al. Discovery of a series of 5,11-dihydro-6Hbenzo[e]pyrimido[5,4-b][1,4]diazepin-6-ones as selective PI3K-d/c Inhibitors. ACS Med Chem Lett. 2016;7:908–912. 15. Kwiatkowski N, Deng X, Wang J, et al. Selective aurora kinase inhibitors identified using a taxol-induced checkpoint sensitivity screen. ACS Chem Biol. 2012;7:185–196. 16. Payton M, Bush TL, Chung G, et al. Preclinical evaluation of AMG 900, a novel potent and highly selective pan-aurora kinase inhibitor with activity in taxaneresistant tumor cell lines. Cancer Res. 2010;70:9846–9854. 17. Shi J, Mitchison TJ. Cell death response to anti-mitotic drug treatment in cell culture, mouse tumor model and the clinic. Endocr Relat Cancer. 2017. 18. Deng X, Yang Q, Kwiatkowski N, et al. Discovery of a benzo[e]pyrimido-[5,4-b] [1,4]diazepin-6(11H)-one as a potent and selective inhibitor of big MAP kinase 1. ACS Med Chem Lett. 2011;2:195–200. 19. Sloane DA, Trikic MZ, Chu MLH, et al. Drug-resistant aurora a mutants for cellular target validation of the small molecule kinase inhibitors MLN8054 and MLN8237. ACS Chem Biol. 2010;5:563–576. 20. Elkins JM, Wang J, Deng X, et al. X-ray crystal structure of ERK5 (MAPK7) in complex with a specific inhibitor. J Med Chem. 2013;56:4413–4421. 21. Deng X, Elkins JM, Zhang J, et al. Structural determinants for ERK5 (MAPK7) and leucine rich repeat kinase 2 activities of benzo[e]pyrimido-[5,4-b]diazepine-6 (11H)-ones. Eur J Med Chem. 2013;70:758–767.