University of Groningen Kinome directed target discovery and validation in unique ovarian clear cell carcinoma models Caumanns, Joost

(1)

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Caumanns, J. (2019). Kinome directed target discovery and validation in unique ovarian clear cell carcinoma models. University of Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

526956-L-sub01-bw-Caumanns 526956-L-sub01-bw-Caumanns 526956-L-sub01-bw-Caumanns 526956-L-sub01-bw-Caumanns Processed on: 7-12-2018 Processed on: 7-12-2018 Processed on: 7-12-2018

Processed on: 7-12-2018 PDF page: 1PDF page: 1PDF page: 1PDF page: 1

Kinome directed target

discovery and validation in

unique ovarian clear cell

carcinoma models

(3)

(4)

Kinome directed target discovery and

validation in unique ovarian clear cell

carcinoma models

Proefschrift

ter verkrijging van de graad van doctor aan de

Rijksuniversiteit Groningen

op gezag van de

UHFWRUPDJQL¿FXVSURIGUE. Sterken

en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op

16 januari 2019 om 16:15 uur

door

Joseph Johannes Caumanns

geboren op 16 November 1990

te Hengelo (O)

(5)

(6)

TABLE OF CONTENTS

Chapter 1 Introduction and thesis outline 7

Chapter 2 ARID1A mutant ovarian clear cell carcinoma: A clear target for synthetic lethal strategies

Biochim Biophys Acta, 2018

17

Chapter 3 ARID1A mutation sensitizes most ovarian clear cell carcinomas to BET inhibitors

Oncogene, 2018

35

Chapter 4 ,QWHJUDWLYHNLQRPHSUR¿OLQJLGHQWL¿HVP725&LQKLELWLRQ as treatment strategy in ovarian clear cell carcinoma

Clinical Cancer Research, 2018

61 Chapter 5 /RZGRVHWULSOHGUXJFRPELQDWLRQWDUJHWLQJWKH3,. $.7P725SDWKZD\DQGWKH0$3.SDWKZD\LVDKLJKO\ HႇHFWLYHDSSURDFKLQRYDULDQFOHDUFHOOFDUFLQRPD Submitted 97

Chapter 6 Establishment and characterization of ovarian clear cell carcinoma patient-derived xenograft models

Manuscript in preparation

117

Chapter 7 Summary and discussion 137

Appendix Nederlandse samenvatting List of abbreviations Biography

List of publications Dankwoord

(7)

(8)

CHAPTER 1

(9)

CHAPTER 1 8

Epithelial ovarian cancer (EOC) forms the vast majority of ovarian cancer cases and is now recognized as a heterogeneous disease divided into the histological subtypes low-grade serous ovarian carcinoma (LGSOC), high-grade serous ovarian carcinoma (HGSOC), ovarian clear cell carcinoma (OCCC), endometrioid ovarian carcinoma (ENOC), mucinous ovarian carcinoma (MOC) and mixed ovarian carcinoma (4). HGSOC accounts for most of the EOCs and is followed by OCCC, ENOC, MOC, LGSOC and mixed ovarian carcinoma. Remarkably, OCCC has a higher prevalence in Japan (15-25%) compared to Europe and North America (1-12%) (5).

The six epithelial ovarian cancer subtypes are not only histologically GLVWLQFW EXW DOVR GLႇHU LQ PROHFXODU alterations and originating cells. LGSOC is proposed to derive from epithelial cells of the fallopian tube and is characterized by a high mutation frequency in the mitogen activated protein kinase (MAPK) pathway genes KRAS (20-70%) and BRAF (14-33%) (6, 7). HGSOC arises from fallopian tube epithelial cells as well, but typically is mutated in TP53 (>90%) and has germline or somatic mutations in BRCA1 or BRCA2 (10% each). Other mutations in HGSOC are rare but single nucleotide polymorphism (SNP) array analysis demonstrated a high prevalence of copy number alterations (CNAs) (8). MOC alterations include mutations in KRAS and overexpression of the receptor tyrosine kinase ERBB2,

be found in premalignant endometriosis lesions (10-12). The heterogeneous alteration spectrum found in OCCC is discussed in the next paragraph.

EOC is frequently diagnosed in DGYDQFHGVWDJH),*2VWDJH,,,,9ZLWKD 5-year survival rate of only 20-25%. Since EOCs have historically been considered one entity, all subtypes are still treated by optimal cytoreductive surgery and platinum based chemotherapy. Survival rates have not increased much after the introduction of platinum based treatments about 40 years ago, indicating that a PROHFXODUDOWHUDWLRQDQGVXEW\SHVSHFL¿F treatment approach is warranted (3, 4). Molecular characterization of OCCC Multiple genomic alterations have been reported in small OCCC patient sets in recent years. Loss of function mutations in DNA binding AT-rich interactive domain 1A (ARID1A) were reported in 40-57% of OCCC, the highest percentage found among all cancer types. ARID1A LV D NH\ FRPSRQHQW RI WKH 6:,61) chromatin remodeling complex, an epigenetic regulator from which its loss FDQDႇHFWWKHH[SUHVVLRQRIPDQ\JHQHV (13-15). Activating mutations have been described in the PI3K catalytic subunit PIK3CA in 30-40% of OCCC. PI3K drives proliferation and survival WKURXJK 3,.$.7P725 VLJQDOLQJ D major pathway implicated in cancer (14, 16). Moreover, expression of the PI3K negative regulator PTEN was lost in 40% of OCCC and activating mutations in the

(10)

INTRODUCTION AND THESIS OUTLINE

1

9

MAPK pathway signaling node KRAS are found in 5-14% of OCCC (17, 18). The receptor tyrosine kinases ERBB2 and EGFR ZKLFK FDQ DFWLYDWH 3,.$.7 mTOR and MAPK pathways upstream, ZHUHDPSOL¿HGLQDVPDOOSHUFHQWDJHRI OCCC patients (19, 20). Alterations in the tumor suppressor gene TP53 have an incidence of 10-15% in OCCC, which is less abundant compared to HGSOC (21). Finally, OCCCs are characterized by almost ubiquitous overexpression RI +1)ȕ 7KLV SURWHLQ LV LQYROYHG in glycogen metabolism and aerobic glycolysis, also known as the Warburg HႇHFW DQG FDQ SURPRWH VXUYLYDO LQ hypoxic environments (22-25). Loss of ARID1A expression and overexpression RI(*)5DQG+1)ȕKDYHDOOEHHQIRXQG in endometriosis lesions, supporting the assumption that endometriosis is a precursor of OCCC (20, 22, 26, 27). An overview of pathways implicated in OCCC is presented in Figure 1.

OCCC survival and chemotherapy response

The majority of OCCC are diagnosed LQ),*2VWDJH,,,DQGKDYHD

favorable prognosis compared to stage matched HGSOC. In contrast, advanced VWDJH ),*2 ,,,,9 GLDJQRVHG 2&&& patients perform worse than advanced stage HGSOC (3, 28). OCCC patients with related endometriosis are diagnosed at a younger age and with early stage disease but their overall survival is similar compared to non-endometriosis related OCCC (29, 30).

Retrospective analysis of FIGO stage III EOC patients treated with primary surgery and adjuvant platinum-based (cisplatin and paclitaxel) chemotherapy

LGHQWL¿HG D VLJQL¿FDQW ZRUVH

progression-free survival and overall survival for OCCC patients compared to HGSOC patients (31). Moreover, in 2010 a meta-analysis was published containing survival data of over 8000 ),*2 VWDJH ,,,,9 (2& SDWLHQWV IURP seven clinical trials with platinum-based FKHPRWKHUDS\ DV ¿UVWOLQH WUHDWPHQW (32). Over 200 OCCCs were included and their median overall survival was estimated at 21 months compared to 41 months for HGSOC. Other retrospective VHULHVUHSRUWHG¿UVWOLQHSODWLQXPEDVHG chemotherapy response rates of 22-56% for OCCC compared to 70% in HGSOC

Figure 1 | Hallmarks of OCCC. Genomic alterations that discriminate OCCC from other ovarian cancer

VXEW\SHVKDYHEHHQFKDUDFWHUL]HGLQWKHODVWGHFDGH$FWLYDWLQJPXWDWLRQVLQ3,.$.7P725DQG0$3. pathway subunits promotes tumor cell proliferation and survival and alterations in upstream receptor tyro-VLQHNLQDVHVFDQIXUWKHUDFWLYDWHERWKSDWKZD\V'\VUHJXODWHG6:,61)IXQFWLRQDVDUHVXOWRI$5,'$ ORVVFDQLQÀXHQFHWKHH[SUHVVLRQRIPDQ\JHQHVDQGVHUYHVPDOLJQDQWWUDQVIRUPDWLRQ8ELTXLWRXVRYHU-H[SUHVVLRQRI+1)ȕVXSSRUWVWXPRUVXUYLYDOLQK\SR[LFHQYLURQPHQWVE\SURPRWLQJDHURELFJO\FRO\VLV

Dysregulated gene expression SWI/SNF

SWI/SNF ARID1A

Proliferation and survival PI3K Pathway MAPK Pathway Receptor Tyrosine Kinases Aerobic glycolysis +1)ȕ Glycogen

(11)

CHAPTER 1 10

VWDJH ,,,,9 2&&& SDWLHQWV These disappointing results stress WKH LGHQWL¿FDWLRQ RI QRYHO WKHUDSHXWLF targets and new combinations with chemotherapy in order to improve survival of OCCC patients.

Targeted therapies for the treatment of OCCC and experimental models

New OCCC treatments should be GLUHFWHG DW WDUJHWV VSHFL¿FDOO\ SUHVHQW in OCCC cancer cells and not in other cell types in the human body. In contrast to chemotherapy regimens that are detrimental to all proliferating tissues such as the colon epithelium and hair follicles, targeted therapies can be aimed DWPROHFXODUPHFKDQLVPVWKDWVSHFL¿FDOO\ underlie oncogenesis. Kinases are proteins with phosphorylation capacity and are involved in activation of many signaling pathways in cancer. Inhibitors have been developed to target over half of the approximately 500 human protein kinases (41). The heterogeneous genomic alterations (i.e. ARID1A, PIK3CA, PTEN and KRAS) in OCCC can promote the activation of multiple signaling pathways involving kinases, making OCCC an excellent candidate to exploit kinase targeted therapies. The availability of a large number of in vitro and in vivo preclinical models in order to capture the full OCCC mutation and CNA heterogeneity will therefore be crucial.

There are now between 10 and 20 OCCC cell lines available for in vitro screening and target validation. Most

xenografts, PDX models are proposed to have more clinical predictive value as they better represent genomic heterogeneity of the patient tumor. A comprehensive characterization of these OCCC PDX models is lacking. Besides, a number of genetically engineered mouse models that form de-novo OCCC tumors have been developed. In one study OCCC-like tumors were formed in mouse ovarian surface epithelium using conditional knockout models of both ARID1A and PTEN and knockout of ARID1A in a background of defective PTEN and APC was used in a second study (55, 56). A third study demonstrated the formation of OCCC tumors in mice with coexistent oncogenic activation of PIK3CA and inactivation of ARID1A, the two most frequently mutated genes in OCCC (57). In conclusion, an increasing number of in vitro and in vivo models are available for the evaluation of new therapeutic targets for treatment of OCCC.

AIM OF THE THESIS

Research presented in this thesis aimed to reveal new kinase therapeutic targets for the treatment of OCCC. To this HQG ZH VHDUFKHG IRU VSHFL¿F NLQDVH vulnerabilities in OCCCs with and without deleterious mutations in ARID1A.

CHAPTER OUTLINE

6:,61) FKURPDWLQ UHPRGHOLQJ

complexes are epigenetic regulators of chromatin structure and gene

(12)

1

11

kinome (>500 kinases) and additional cancer related genes were determined by means of kinome sequencing and SNP array analysis. We subsequently assessed the druggability of downstream DႇHFWHGSDWKZD\VLQ2&&&FHOOOLQHVDQG validated our most promising therapeutic target in OCCC PDX models. Because OCCC is heterogeneously mutated DFURVV 3,.$.7P725 DQG 0$3. proliferation pathways, in chapter

5 we combined kinase inhibitors of

P725& 3,. DQG 0(. DW FHOO OLQH VSHFL¿F ,&₂₀ concentrations. With this approach we aimed to identify an optimal inhibitor combination in a panel of seven OCCC cell lines with distinct genetic alterations. We determined if combinations of inhibitors acted synergistically. To prove the clinical utility of this treatment strategy, two OCCC PDX models were treated with low-dose triple inhibitor combinations.

7UDQVODWLRQDO UHVHDUFK FDQ EHQH¿W from the use of PDX models for novel drug testing whereas these models better represent patient characteristics compared to xenograft models. Therefore, in chapter 6 we described the establishment and growth characteristics of PDX models of OCCC. Furthermore, immunohistochemistry, sequencing and SNP array analysis were used to determine the patient tumor resemblance of established PDX models.

Finally, in chapter 7 the results obtained in this thesis are summarized and open questions are discussed. REFERENCES

1. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015 Mar 1;136(5):E359-86.

2. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin. 2015 Jan-Feb;65(1):5-29. 3. Oliver KE, Brady WE, Birrer M, Gershenson DM, Fleming G, Copeland LJ, et al. An evaluation of

progression free survival and overall survival of ovarian cancer patients with clear cell carcinoma YHUVXVVHURXVFDUFLQRPDWUHDWHGZLWKSODWLQXPWKHUDS\$Q15*2QFRORJ\*\QHFRORJLF2QFRORJ\ Group experience. Gynecol Oncol. 2017 Nov;147(2):243-9.

WUDQVFULSWLRQ $ PXOWLWXGH RI 6:,61) complex members are mutated in FDQFHU 7KH 6:,61) '1$ WDUJHWLQJ subunit ARID1A is the most frequently PXWDWHG 6:,61) PHPEHU ZLWK WKH highest incidence found in OCCC. ARID1A mutant cancers are attractive targets for synthetic lethality screening considering that ARID1A is deleteriously mutated. The concept of synthetic lethality describes a relation between two genes in which cells are viable after loss of either one gene but loss of both genes will result in a lethal phenotype. In chapter 2, we reviewed synthetic lethality screens performed in an ARID1A mutant background and evaluated advantages and drawbacks of these studies and the clinical relevance RILGHQWL¿HGWDUJHWV,Qchapter 3, a novel strategy to target ARID1A mutant OCCC was explored by performing kinome-centered lethality screens in a panel of 14 OCCC cell lines. The synthetic lethal hit BRD2 was validated in vitro by genetic and chemical inhibition of BRD2 using shRNAs, CRISPR-knockouts and BET bromodomain inhibitors. We provided a mechanism for the observed synthetic lethality between loss of both ARID1A and BRD2. Ultimately, ARID1A mutant OCCC xenograft and PDX models were treated with BET bromodomain inhibition to support clinical translation of the therapeutic target BRD2 in OCCC.

In chapter 4, we searched for new kinase mutations and CNA in a large set of OCCC patients and cell lines. Mutations and CNA in the human

(13)

CHAPTER 1 12

9. Banerjee S, Kaye SB. New strategies in the treatment of ovarian cancer: current clinical perspectives and future potential. Clin Cancer Res. 2013 Mar 1;19(5):961-8.

10. Cochrane DR, Tessier-Cloutier B, Lawrence KM, Nazeran T, Karnezis AN, Salamanca C, et al. Clear FHOODQGHQGRPHWULRLGFDUFLQRPDVDUHWKHLUGLႇHUHQFHVDWWULEXWDEOHWRGLVWLQFWFHOOVRIRULJLQ"-3DWKRO 2017 Sep;243(1):26-36.

11. Gadducci A, Lanfredini N, Tana R. Novel insights on the malignant transformation of endometriosis into ovarian carcinoma. Gynecol Endocrinol. 2014 Jun 6:1-6.

+XDQJ 5< &KHQ *% 0DWVXPXUD 1 /DL +& 0RUL 6 /L - HW DO +LVWRW\SHVSHFL¿F FRS\QXPEHU alterations in ovarian cancer. BMC Med Genomics. 2012 Oct 18;5:47,8794-5-47.

13. Wiegand KC, Shah SP, Al-Agha OM, Zhao Y, Tse K, Zeng T, et al. ARID1A mutations in endometriosis-associated ovarian carcinomas. N Engl J Med. 2010 Oct 14;363(16):1532-43.

14. Jones S, Wang TL, Shih I, Mao TL, Nakayama K, Roden R, et al. Frequent mutations of chromatin remodeling gene ARID1A in ovarian clear cell carcinoma. Science. 2010 Oct 8;330(6001):228-31. *XDQ%:DQJ7/6KLK,$5,'$DIDFWRUWKDWSURPRWHVIRUPDWLRQRI6:,61)PHGLDWHGFKURPDWLQ

remodeling, is a tumor suppressor in gynecologic cancers. Cancer Res. 2011 Nov 1;71(21):6718-27. .XR.70DR7/-RQHV69HUDV($\KDQ$:DQJ7/HWDO)UHTXHQWDFWLYDWLQJPXWDWLRQVRI3,.&$

in ovarian clear cell carcinoma. Am J Pathol. 2009 May;174(5):1597-601.

17. Hashiguchi Y, Tsuda H, Inoue T, Berkowitz RS, Mok SC. PTEN expression in clear cell adenocarcinoma of the ovary. Gynecol Oncol. 2006 Apr;101(1):71-5.

18. Zannoni GF, Improta G, Chiarello G, Pettinato A, Petrillo M, Scollo P, et al. Mutational status of KRAS, 15$6DQG%5$)LQSULPDU\FOHDUFHOORYDULDQFDUFLQRPD9LUFKRZV$UFK$XJ 19. Tan DS, Iravani M, McCluggage WG, Lambros MB, Milanezi F, Mackay A, et al. Genomic analysis

reveals the molecular heterogeneity of ovarian clear cell carcinomas. Clin Cancer Res. 2011 Mar 15;17(6):1521-34.

20. Okamoto A, Sehouli J, Yanaihara N, Hirata Y, Braicu I, Kim BG, et al. Somatic copy number alterations associated with Japanese or endometriosis in ovarian clear cell adenocarcinoma. PLoS One. 2015 Feb 6;10(2):e0116977.

21. Ho ES, Lai CR, Hsieh YT, Chen JT, Lin AJ, Hung MH, et al. P53 Mutation is Infrequent in Clear Cell Carcinoma of the Ovary. Gynecol Oncol. 2001 Feb;80(2):189-93.

22. Shen H, Fridley BL, Song H, Lawrenson K, Cunningham JM, Ramus SJ, et al. Epigenetic analysis OHDGV WR LGHQWL¿FDWLRQ RI +1)% DV D VXEW\SHVSHFL¿F VXVFHSWLELOLW\ JHQH IRU RYDULDQ FDQFHU 1DW Commun. 2013;4:1628.

23. Okamoto T, Mandai M, Matsumura N, Yamaguchi K, Kondoh H, Amano Y, et al. Hepatocyte nuclear factor-1beta (HNF-1beta) promotes glucose uptake and glycolytic activity in ovarian clear cell carcinoma. Mol Carcinog. 2013 Sep 17.

<DPDJXFKL.0DQGDL02XUD70DWVXPXUD1+DPDQLVKL-%DED7HWDO,GHQWL¿FDWLRQRIDQRYDULDQ FOHDU FHOO FDUFLQRPD JHQH VLJQDWXUH WKDW UHÀHFWV LQKHUHQW GLVHDVH ELRORJ\ DQG WKH FDUFLQRJHQLF processes. Oncogene. 2010 Mar 25;29(12):1741-52.

(14)

1

13

caused by HNF1beta expression in ovarian clear cell carcinoma contribute to cell survival. Oncotarget. 2015 Sep 22;6(28):26002-17.

26. Yamamoto S, Tsuda H, Takano M, Tamai S, Matsubara O. Loss of ARID1A protein expression occurs as an early event in ovarian clear-cell carcinoma development and frequently coexists with PIK3CA mutations. Mod Pathol. 2012 Apr;25(4):615-24.

;LDR:$ZDGDOODK$;LQ:/RVVRI$5,'$%$)DH[SUHVVLRQLQRYDULDQHQGRPHWULRVLVDQGFOHDU cell carcinoma. Int J Clin Exp Pathol. 2012;5(7):642-50.

$QJOHVLR 06 &DUH\ 06 .REHO 0 0DFND\ + +XQWVPDQ '* 9DQFRXYHU 2YDULDQ &OHDU &HOO 6\PSRVLXP6SHDNHUV&OHDUFHOOFDUFLQRPDRIWKHRYDU\DUHSRUWIURPWKH¿UVW2YDULDQ&OHDU&HOO Symposium, June 24th, 2010. Gynecol Oncol. 2011 May 1;121(2):407-15.

29. Davis M, Rauh-Hain JA, Andrade C, Boruta DM,2nd, Schorge JO, Horowitz NS, et al. Comparison of clinical outcomes of patients with clear cell and endometrioid ovarian cancer associated with endometriosis to papillary serous carcinoma of the ovary. Gynecol Oncol. 2014 Mar;132(3):760-6. 30. Paik ES, Kim TJ, Choi CH, Kim BG, Bae DS, Lee JW. Clinical outcomes of patients with clear cell and

endometrioid ovarian cancer arising from endometriosis. J Gynecol Oncol. 2018 Mar;29(2):e18. 31. Winter WE,3rd, Maxwell GL, Tian C, Carlson JW, Ozols RF, Rose PG, et al. Prognostic factors for

stage III epithelial ovarian cancer: a Gynecologic Oncology Group Study. J Clin Oncol. 2007 Aug 20;25(24):3621-7.

32. Mackay HJ, Brady MF, Oza AM, Reuss A, Pujade-Lauraine E, Swart AM, et al. Prognostic relevance RIXQFRPPRQRYDULDQKLVWRORJ\LQZRPHQZLWKVWDJH,,,,9HSLWKHOLDORYDULDQFDQFHU,QW-*\QHFRO Cancer. 2010 Aug;20(6):945-52.

33. Sugiyama T, Kamura T, Kigawa J, Terakawa N, Kikuchi Y, Kita T, et al. Clinical characteristics of clear cell carcinoma of the ovary: a distinct histologic type with poor prognosis and resistance to platinum-based chemotherapy. Cancer. 2000 Jun 1;88(11):2584-9.

34. Takano M, Kikuchi Y, Yaegashi N, Kuzuya K, Ueki M, Tsuda H, et al. Clear cell carcinoma of the ovary: a retrospective multicentre experience of 254 patients with complete surgical staging. Br J Cancer. 2006 May 22;94(10):1369-74.

35. Ho CM, Huang YJ, Chen TC, Huang SH, Liu FS, Chang Chien CC, et al. Pure-type clear cell carcinoma of the ovary as a distinct histological type and improved survival in patients treated with paclitaxel-platinum-based chemotherapy in pure-type advanced disease. Gynecol Oncol. 2004 Jul;94(1):197-203.

36. Pectasides D, Fountzilas G, Aravantinos G, Kalofonos C, Efstathiou H, Farmakis D, et al. Advanced stage clear-cell epithelial ovarian cancer: the Hellenic Cooperative Oncology Group experience. Gynecol Oncol. 2006 Aug;102(2):285-91.

37. Utsunomiya H, Akahira J, Tanno S, Moriya T, Toyoshima M, Niikura H, et al. Paclitaxel-platinum combination chemotherapy for advanced or recurrent ovarian clear cell adenocarcinoma: a multicenter trial. Int J Gynecol Cancer. 2006 Jan-Feb;16(1):52-6.

38. Sugiyama T, Okamoto A, Enomoto T, Hamano T, Aotani E, Terao Y, et al. Randomized Phase III Trial of Irinotecan Plus Cisplatin Compared With Paclitaxel Plus Carboplatin As First-Line Chemotherapy IRU2YDULDQ&OHDU&HOO&DUFLQRPD-*2**&,*7ULDO-&OLQ2QFRO$XJ 39. Takakura S, Takano M, Takahashi F, Saito T, Aoki D, Inaba N, et al. Randomized phase II trial of

SDFOLWD[HOSOXVFDUERSODWLQWKHUDS\YHUVXVLULQRWHFDQSOXVFLVSODWLQWKHUDS\DV¿UVWOLQHFKHPRWKHUDS\ for clear cell adenocarcinoma of the ovary: a JGOG study. Int J Gynecol Cancer. 2010 Feb;20(2):240-7.

40. Farley JH, Brady WE, Fujiwara K, Nomura H, Yunokawa M, Tokunaga H, et al. A phase II evaluation of temsirolimus in combination with carboplatin and paclitaxel followed by temsirolimus consolidation DV¿UVWOLQHWKHUDS\LQWKHWUHDWPHQWRIVWDJH,,,,9FOHDUFHOOFDUFLQRPDRIWKHRYDU\-&2

41. Hu Y, Furtmann N, Bajorath J. Current compound coverage of the kinome. J Med Chem. 2015 Jan 8;58(1):30-40.

(15)

CHAPTER 1 14

2LVKL7,WDPRFKL+.XGRK$1RQDND0.DWR01LVKLPXUD0HWDO7KH3,.P725GXDOLQKLELWRU 193%(=UHGXFHVWKHJURZWKRIRYDULDQFOHDUFHOOFDUFLQRPD2QFRO5HS$XJ 0DEXFKL6.DZDVH&$OWRPDUH'$0RULVKLJH.+D\DVKL06DZDGD.HWDO9DVFXODUHQGRWKHOLDO

growth factor is a promising therapeutic target for the treatment of clear cell carcinoma of the ovary. Mol Cancer Ther. 2010 Aug;9(8):2411-22.

49. Hisamatsu T, Mabuchi S, Matsumoto Y, Kawano M, Sasano T, Takahashi R, et al. Potential role of mTORC2 as a therapeutic target in clear cell carcinoma of the ovary. Mol Cancer Ther. 2013 Jul;12(7):1367-77.

50. Williamson CT, Miller R, Pemberton HN, Jones SE, Campbell J, Konde A, et al. ATR inhibitors as a V\QWKHWLFOHWKDOWKHUDS\IRUWXPRXUVGH¿FLHQWLQ$5,'$1DW&RPPXQ'HF 51. Weroha SJ, Becker MA, Enderica-Gonzalez S, Harrington SC, Oberg AL, Maurer MJ, et al. Tumorgrafts

as in vivo surrogates for women with ovarian cancer. Clin Cancer Res. 2014 Mar 1;20(5):1288-97. 52. Alkema NG, Tomar T, Duiker EW, Jan Meersma G, Klip H, van der Zee AG, et al. Biobanking of patient

DQGSDWLHQWGHULYHG[HQRJUDIWRYDULDQWXPRXUWLVVXHHႈFLHQWSUHVHUYDWLRQZLWKORZDQGKLJKIHWDOFDOI serum based methods. Sci Rep. 2015 Oct 6;5:14495.

53. Eoh KJ, Chung YS, Lee SH, Park SA, Kim HJ, Yang W, et al. Comparison of Clinical Features and Outcomes in Epithelial Ovarian Cancer according to Tumorigenicity in Patient-Derived Xenograft Models. Cancer Res Treat. 2017 Oct 17.

54. Heo EJ, Cho YJ, Cho WC, Hong JE, Jeon HK, Oh DY, et al. Patient-Derived Xenograft Models of Epithelial Ovarian Cancer for Preclinical Studies. Cancer Res Treat. 2017 Oct;49(4):915-26. 55. Guan B, Rahmanto YS, Wu RC, Wang Y, Wang Z, Wang TL, et al. Roles of deletion of Arid1a, a tumor

VXSSUHVVRU LQ PRXVH RYDULDQ WXPRULJHQHVLV - 1DWO &DQFHU ,QVW -XQ MQFL dju146. Print 2014 Jul.

56. Zhai Y, Kuick R, Tipton C, Wu R, Sessine M, Wang Z, et al. Arid1a inactivation in an Apc- and Pten-GHIHFWLYH PRXVH RYDULDQ FDQFHU PRGHO HQKDQFHV HSLWKHOLDO GLႇHUHQWLDWLRQ DQG SURORQJV VXUYLYDO - Pathol. 2016 Jan;238(1):21-30.

57. Chandler RL, Damrauer JS, Raab JR, Schisler JC, Wilkerson MD, Didion JP, et al. Coexistent ARID1A-3,.&$ PXWDWLRQV SURPRWH RYDULDQ FOHDUFHOO WXPRULJHQHVLV WKURXJK SURWXPRULJHQLF LQÀDPPDWRU\ cytokine signalling. Nat Commun. 2015 Jan 27;6:6118.

(16)

(17)

(18)

CHAPTER 2

ARID1A mutant ovarian clear cell

carcinoma: A clear target for synthetic

lethal strategies

Joseph J. Caumanns

1

_{, G. Bea A. Wisman}

1

_{, Katrien Berns}

3

_{Ate G.J.}

van der Zee

1

_{and Steven de Jong}

2

1_{Department of Gynecologic Oncology and}2_{Department of Medical Oncology,} Cancer Research Centre Groningen, University Medical Center Groningen, University of Groningen, Hanzeplein 1, 9713 GZ Groningen, the Netherlands. 3_{Division of Molecular Carcinogenesis, the Netherlands Cancer Institute,} Plesmanlaan 121, 1066 CX Amsterdam, the Netherlands.

(19)

CHAPTER 2 18

INTRODUCTION

ATP dependent chromatin remodeling is an epigenetic process regulating gene transcription in which chromatin structure can be coordinated through mobilization of nucleosomes. The evolutionary conserved ATP dependent chromatin remodelers consist of four subclasses. These are Imitation SWI (ISWI), INO80, nucleosome remodeling and histone GHDFHW\ODVH0L&+'185'0L&+' DQGVZLWFKVXFURVHQRQIHUPHQWLQJ6:, SNF) and are involved in diverse cellular SURFHVVHVVXFKDVWLVVXHGLႇHUHQWLDWLRQ proliferation and DNA repair (1). While all chromatin remodelers contain ATPase GRPDLQV DGGLWLRQDO VXEXQLWV GLႇHU SHU subclass and are important for modulation of ATPase activity and chromatin FRPSOH[UHFUXLWPHQWRQWRWLVVXHVSHFL¿F JHQRPLF ORFL 6:,61) FKURPDWLQ remodeling complexes are implicated in many stages of pluripotency and WLVVXH GLႇHUHQWLDWLRQ DORQJ PDPPDOLDQ GHYHORSPHQW &RPSDUHG WR WKH 6:,

SNF subclass, the role of ISWI, INO80 DQG 0L&+' FKURPDWLQ UHPRGHOLQJ complexes in these processes is limited EXW VXSSOHPHQWDU\ WR 6:,61) IRU D review see Hota et al %$) 6:, SNF chromatin remodeling complexes contain multiple subunits with mutual exclusive characteristics, which include the DNA targeting subunits ARID1A and ARID1B, and the ATPase subunits SMARCA2 and SMARCA4 (Fig. 1). These mutual exclusive members GLVFULPLQDWH %$) 6:,61) FRPSOH[HV IURP FRXQWHUSDUW 3%$) 6:,61) complexes that contain SMARCA4 but not the other three proteins. BAF and 3%$)6:,61)FRPSOH[HVKDYHGLVWLQFW regulatory roles in lineage development. )RU LQVWDQFH %$) VSHFL¿F 60$5&$ plays a key role in smooth muscle IRUPDWLRQ ZKLOH WKH 3%$) VSHFL¿F subunit ARID2 is important for coronary

PRUSKRJHQHVLV 0RUH 6:,61)

subunits have been shown to coordinate VSHFL¿F GHYHORSPHQWDO SURFHVVHV (2). Therefore, it is conceivable that

of deleterious mutations. Current advances have reported synthetic lethal interactions with the loss of ARID1A in several cancer types. In this review, we discuss targets that are only important for tumor growth in an ARID1A mutant context. We focus on synthetic lethal strategies with ARID1A loss in ovarian clear cell carcinoma, a cancer with the highest ARID1A mutation incidence (46-57%). ARID1A directed lethal strategies that can be exploited clinically include targeting of the DNA repair proteins PARP and ATR, and the epigenetic factors EZH2, HDAC2, HDAC6 and BRD2.

(20)

SYNTHETIC LETHAL STRATEGIES FOR ARID1A MUTANT OCCC

2

19

describes a relation between two genes in which cells remain viable after loss of either gene alone but loss of both genes will result in a lethal phenotype (16). Synthetic lethal targets, including proteins involved in DNA repair and epigenetic regulation are reviewed, IRFXVLQJRQWDUJHWVUHFHQWO\LGHQWL¿HGLQ ARID1A mutant OCCC.

2. Role of ARID1A in ovarian clear cell carcinoma

Mutations in ARID1A are often heterozygous nonsense or frameshifts that are not enriched in hotspot sites (9, 10). ARID1A mutations, homo- or heterozygous, coincides with loss of ARID1A protein expression in OCCC. 7KLV VXJJHVWV KDSORLQVXႈFLHQF\ RU WKH occurrence of other mechanisms that are responsible for complete loss of protein expression, such as epigenetic silencing, mutations in non-coding regions or post transcriptional mechanisms (9, 17-20). Additionally, two studies described loss of ARID1A protein expression in precursor lesions of OCCC, i.e. ovarian endometriosis, indicating that ARID1A loss is an early event in progression to OCCC (21, 22). Two studies described D VSHFL¿F VXEXQLW PXWDWLRQ FDXVHV

FKDQJHVLQ6:,61)FKURPDWLQELQGLQJ at typical transcriptional enhancer sites, which ultimately results in loss of GLႇHUHQWLDWLRQ SRWHQWLDO DQG GULYHV FHOOV into a proliferative state (3). Over 20% of cancers contain a mutation in at least one of the members of the 15-subunit %$) 6:,61) FRPSOH[ KHUHLQDIWHU UHIHUUHG WR DV 6:,61) (VSHFLDOO\ ARID1A and ARID1B and SMARCA2 and SMARCA4 have high mutation frequencies and are suggested to be driver mutations in multiple cancers 7KH PXWDWLRQ LQFLGHQFH LQ 6:, SNF subunits varies per tumor type, LQGLFDWLYH IRU GLVWLQFW UROHV RI 6:, SNF complexes in human tissues. For LQVWDQFH $5,'$ GH¿FLHQW PLFH DUH resistant to hepatocellular carcinoma initiating agents while mice with established hepatocellular carcinoma showed enhanced metastasis upon ARID1A loss (7). Conversely, ARID1A mutations promote colorectal tumor formation, indicating the context GHSHQGHQW HႇHFW RI $5,'$ ORVV ARID1A is the most frequently mutated 6:,61)FRPSOH[PHPEHU7KHKLJKHVW alteration incidence, 46-57%, is found in ovarian clear cell carcinoma (OCCC) (9-11). Other tumor types harboring ARID1A mutations comprise uterine endometrioid carcinoma (47-60%), ovarian endometrioid carcinoma (30%), gastric cancer (29%), colorectal cancer (5-10%) and pancreatic cancer (3-5%) (6, 12-14). Tumors with mutations in ARID1B include OCCC (18%), colorectal cancer and gastric cancer (5-10%) (Fig. 2). Mutations in SMARCA2 are found in colorectal cancer, lung cancer (3-5%) and OCCC (2%). Mutations in SMARCA4 are found in melanoma, lung cancer (5-10%) and OCCC (5%) (6, 15) (Fig. 3).

In this review, we discuss recent developments in strategies that take advantage of cellular dependencies VSHFL¿F IRU ARID1A mutant cancers. This concept, named synthetic lethality,

Figure 1 | SWI/SNF complex members. ARID1A

and ARID1B are mutual exclusive core compo-QHQWV LQ WKH PDPPDOLDQ %$) 6:,61) FKURPD-tin remodeling complex. Other core members are shown in turquoise and exchangeable components in yellow. SMARCD1/2/3 SMARCD1/2/3 ACTL6 ACTL6 SMARCE1 SMARCMARCE1 DPF2DPF2 BRD9 BRD9 SS18SS18 SMARCA2/4 SMARCA2/MARCA2/4 SMARCC1

S ARCCMAR 1 SMARCC2SMARCCMAR 2 SMARCB1 SMARCB1C BCL11A/BC / BCL11A/B ARID1A/BRID1A/B/ A

(21)

CHAPTER 2 20

SURPRWHG HSLWKHOLDO GLႇHUHQWLDWLRQ DQG metastasis (30). Moreover, Chandler et al. demonstrated that ovarian tumor formation only occurred in mice with concurrent homozygous knockout of ARID1A and PIK3CAH1047R_missense mutation. These tumors manifested an OCCC like histopathology and gene expression signature. Furthermore, a PHFKDQLVWLFOLQNZDVLGHQWL¿HGEHWZHHQ ARID1A loss and PIK3CA activation with the induction of cytokine expression, LQFOXGLQJ,/YLD1)ڡ%VLJQDOLQJ In vitro and in vivo suppression of IL-6 reduced the tumor cell proliferation, presenting IL-6 blocking as potential synthetic lethal strategy in an ARID1A and PIK3CA mutant OCCC background (32).

3. Mutual exclusive roles for ARID1A and ARID1B

7KH 6:,61) VXEXQLW HQFRGLQJ JHQHV ARID1A and ARID1B share 60% sequence identity but are thought to have disparate roles in cell cycle regulation (33). ARID1B mutations are less prevalent in cancer with fewer a worse progression-free survival in

$5,'$ GH¿FLHQW 2&&& SDWLHQWV However, literature consistently showed that overall survival in OCCC is not predicted by ARID1A status (11, 23-25).

ARID1A_{or ARID1A}_knockout causes lethality in early embryonic mouse development (26). In contrast, depletion of ARID1A expression induced proliferation in ovarian surface epithelium cells. Another study showed that knockdown of ARID1A provided

phenotypic changes associated

with neoplastic transformation in an immortalized endometriosis cell line (27). Upon ARID1A knockdown the expression of 99 genes was up or downregulated. Many of these genes are also dysregulated in OCCC tumors (28). 6WLOO$5,'$ ORVV LWVHOI LV QRW VXႈFLHQW to induce tumorigenesis in the ovaries of conditional Cre-Lox mouse models. Only mice with combined ARID1A and PTEN loss developed ovarian hyperplasia, which progressed to tumors in 59% (29). Another study showed that additional ARID1A_{or ARID1A}_{knockout in} APC and PTEN null mice actually delayed ovarian tumor formation but

Adenoid Cystic Carcinoma Neuroendocrine Prostate CancerCholangiocarcinoma Stomach Adenocarcinoma 0 10 20 30 Mutation Deletion Amplification Alteration frequency (%) 40

(22)

2

21

cell lines (36). Subsequently, Helming et al. demonstrated a reduced proliferation in two ARID1A mutant OCCC cell lines after shRNA induced ARID1B loss. They found a destabilization of full size 6:,61) FRPSOH[HV DIWHU ORVV RI ERWK proteins, suggesting that depletion of LQWDFW6:,61)FRPSOH[HVLVUHVSRQVLEOH for the observed synthetic lethality. Co-mutations in ARID1A and ARID1B do occur but never biallelic in both genes, indicating some ARID1 function is essential for cell survival of both cancer and normal cells (36).

4. ARID1A-directed synthetic lethality Several studies have shown that ARID1A directed synthetic lethality can be attained through diverse molecular mechanisms. These synthetic lethal interactions with ARID1A mutations go beyond direct targeting of ARID1B. 4.1 Lethal relationship between ARID1A loss and inhibition of the DNA damage response

Besides ARID1A’s regulatory role in deleterious mutations as compared to

ARID1A (Fig. 2). In 2007, Nagl et al. reported opposing roles for ARID1A and ARID1B in regulating proliferation in osteoblasts. They found that ARID1B had a more proliferation inducing role via MYC and Cyclin E activation as compared ZLWK$5,'$$5,'$ORVVPRGL¿HV chromatin accessibility in colorectal cancer cells and this accessibility is IXUWKHUPRGL¿HGE\ORVVRI$5,'%,QWKH presence of wild-type ARID1A, ARID1B ORVVGLGQRWKDYHDQHႇHFWRQFKURPDWLQ DFFHVVLELOLW\LQGLFDWLQJDGRPLQDQWHႇHFW of ARID1A. It was reported that ARID1B in the absence of ARID1A facilitates expression of proliferative genes involved LQ3,.$.7P725DQG(5%%UHFHSWRU tyrosine kinase signaling in colorectal carcinoma and OCCC cells (35). These ¿QGLQJV K\SRWKHVL]H WKDW UHGXQGDQW LQFRUSRUDWLRQRI$5,'%LQUHVLGXDO6:, 61) FRPSOH[HV LQ $5,'$ GH¿FLHQW cells could stimulate proliferation. The consecutive discovery that ARID1A PXWDQWWXPRUVDUHVSHFL¿FDOO\YXOQHUDEOH to ARID1B loss was based on the Achilles project, a loss of function genetic screen database, which include over 200 cancer

SMARCA2

SMARCA4

Bladder Urothelial Carcinoma Breast cancer patient xenograftsNon-Small Cell lung Cancer Ovarian Serous CystadenocarcinomaBladder Urothelial Carcinoma Lung Adenocarcinoma Uterine Corpus Endometrial CarcinomaCancer Cell Line Encyclopedia Ovarian Serous CystadenocarcinomaNCI-60 Cell Lines Uterine Corpus Endometrial CarcinomaPre-treatment metastatic melanoma Uterine Carcinosarcoma Neuroendocrine Prostate Cancer

0 10 20 30 Multiple alterations Mutation Deletion Amplification Alteration frequency (%) 40

TCGA data for Esophagus-Stomach CancersStomach Adenocarcinoma Stomach Adenocarcinoma Adenoid Cystic Carcinoma Bladder Urothelial Carcinoma Bladder Urothelial Carcinoma Cancer Cell Line Encyclopedia Paired-exome sequencing of acral melanomaBreast cancer patient xenografts NCI-60 Cell Lines Neuroendocrine Prostate Cancer Malignant Peripheral Nerve Sheath Tumor

0 10 20 30 Alteration frequency (%) Multiple alterations Mutation Deletion Amplification 40 Figure 3 | SMARCA2 and

SMARCA4 alterations

in cancer. Alteration

fre-quencies of SMARCA2 and

SMARCA4 in published

and provisional datasets incorporated in cBioPortal are depicted with a 10% FXWRႇ KWWSFELRSRUWDORUJ downloaded 3-10-2017). Of note, SMARCA2 and

SMARCA4 mutations are

commonly found in several cancer types, but not al-ways displayed due to the FXWRႇ

(23)

CHAPTER 2 22

cells to DSB inducing treatments such as oxaliplatin and cisplatin (38, 40). Two studies have investigated synthetic lethality approaches based on targeting of DDR proteins in ARID1A mutant cancers. Shen et alLGHQWL¿HGDUROHIRU ARID1A in DSB repair as binding partner of ATR, a DDR central regulator involved in DSB and single strand DNA breaks (39). With chromatin immunoprecipitation assays they found ARID1A to be enriched at chromatin regions close to DSBs, a recruitment that was lost upon ATR inhibition. Phenotypically, ARID1A stopcodon introduction in HCT116 colorectal carcinoma cells resulted in LPSDLUHG *0 FKHFNSRLQW FRQWURO XSRQ irradiation-induced DSBs. Cells lacking ARID1A more frequently re-entered the cell cycle after irradiation as compared with control cells. The phosphorylation of ATR and CHK1, another DDR protein 6:,61) FRPSOH[ ELQGLQJ WR '1$ LW

is also involved in DNA double-strand EUHDN '6% UHSDLU ,QLWLDOO\ WKH 6:, SNF ATPase subunits SMARCA2 and SMARCA4 were described to interact ZLWK WKH '6% ORFDOL]LQJ SURWHLQ Ȗ+$; upon DSB formation, suggesting a role for 6:,61)LQWKH'1$GDPDJHUHVSRQVH (DDR) (37). Recently, ARID1A and ARID1B were found to localize to DSBs and facilitate non-homologous end joining (NHEJ) and ARID1A was assigned a role in homologous recombination (HR) (38, 39). NHEJ is the more error prone DSBs repair mechanism compared to HR. ARID1A and ARID1B recruit the NHEJ proteins KU70 and KU80 to the site of DNA damage and are thus involved in this type of DSB repair at an early stage. Downregulation of ARID1A or ARID1B sensitized osteosarcoma cells and immortalized pancreatic ductal epithelial

ROS reduction ROS HDAC6 Apoptosis regulation P53 P53 Ac

Figure 4 | ARID1A implicated mechanisms with synthetic lethal opportunities after ARID1A loss.

$5,'$LWVHOIRULQFRUSRUDWHGLQWKH6:,61)FKURPDWLQUHPRGHOLQJFRPSOH[LVLQYROYHGLQPXOWLSOHRQ-cogenic suppressive processes. Therapeutic targeting of these indicated mechanisms were found to be lethal in an ARID1A mutant background.

(24)

2

23

of ARID1APXWDQW729*2&&&FHOOV In this study, it was also demonstrated that ARID1A loss in HCT116 cells resulted in accumulation of cells in *0 ,Q OLQH ZLWK WKLV UHVXOW ARID1A mutant HCT116 xenografts showed reduced tumor growth. In the presence of ATR inhibitors, they found reduced accumulation of ARID1A mutant HCT116 FHOOVLQ*0DQGPXOWLSOHLQGLFDWLRQVRI chromosomal instability and apoptosis induction in these cells (41).

Previously, it was reported that SMARCA4 deprived mouse embryonic stem cells have decreased DNA localization of DNA topoisomerase 2-alpha (TOP2A), an enzyme important in DNA transcription and translation (42). Elaborating on this study, Williamson et al. found ARID1A mutant HCT116 cells to have decreased TOP2A DNA localization as well, indicating that 6:,61) FRPSOH[HV DUH LPSRUWDQW IRU stable function of TOP2A (41). Since ARID1A status did not determine 723$ H[SUHVVLRQ PRUH OLNHO\ 6:, SNF complexes directly bind to TOP2A via ARID1A, which is important for the localization of TOP2A to at least 12.000 genomic loci, as demonstrated earlier (Fig. 4) (42).

4.2 Epigenetic targeting and ARID1A synthetic lethality

Synthetic lethality between ARID1A loss and inhibition of multiple distinct epigenetic proteins was recently found in OCCC. The lethal interaction between ARID1A mutation and inhibition of polycomb repressive complex 2 (PRC2) catalytic subunit EZH2 resulted in the ¿UVW WKHUDSHXWLFDOO\ GUXJJDEOH WDUJHW IRU ARID1A mutant cancers (43). In contrast WR6:,61)FRPSOH[HVWKDWJHQHUDWHD more open chromatin structure, PRC2 closes chromatin by methylation of histone 3 lysine 27 (H3K27me3), which is associated with gene repression (44). Disruption of epigenetic chromatin LPSRUWDQW LQ *0 FKHFNSRLQW FRQWURO

was reduced in ARID1A mutant cells, indicating that ARID1A is involved LQ *0 SURJUHVVLRQ )XUWKHUPRUH ARID1A loss reduced localization of WKH ''5 DGDSWRU SURWHLQV Ȗ+$; DQG 53BP1 at the site of DSBs. Collectively, they reveal an important role for ARID1A to conduct ATR-mediated DDR signaling required for HR (Fig. 4). It is unclear if ARID1A alone or in complex with other 6:,61) PHPEHUV ORFDOL]HV WR '6%V and activates ATR and subsequent DDR. The involvement of ARID1A in DSB repair led Shen et al. to test for synthetic lethality of PARP inhibitors in ARID1A mutant cells as these inhibitors are known to be lethal in cells with DSB UHSDLU GH¿FLHQFLHV VXFK DV BRCA1/2 mutations. Multiple inhibitors of PARP provided toxicity in (BRCA wild-type) isogenic pairs of ARID1A-depleted non-transformed breast epithelial, breast epithelial carcinoma and colorectal carcinoma cells and a panel of ovarian carcinoma cell lines, showing more than three-fold reduced colony formation in ARID1A-depleted cell lines. Additionally, 3$53 WUHDWPHQW HႈFDF\ ZDV GHWHFWHG in (BRCA wild-type) ARID1A-depleted xenograft models of breast and colorectal cancer (39).

In another study by Williamson et al 51$L VFUHHQLQJ OLEUDULHV LGHQWL¿HG ARID1A loss as a genetic determinant of sensitivity to ATR inhibition in the triple negative (ER, PR and ERBB2 negative) breast cancer cell line HCC1143 and immortalized normal mammary epithelial cell line MCF12A (41). ARID1A loss in FRPELQDWLRQ ZLWK$75 LQKLELWLRQ E\ 9; ZDV FRQ¿UPHG WR EH RYHU WKUHH IROG PRUH HႇHFWLYH LQ PXOWLSOH FDQFHU lineages, including several OCCC cell OLQHV DQG DQ LVRJHQLF$5,'$GH¿FLHQW pair of the colorectal cancer cell line HCT116. Sensitivity to ATR inhibition was observed in xenograft models of the isogenic HCT116 cancer cell line pair and to a lesser extent in the xenograft model

(25)

CHAPTER 2 24

To this end, EZH2 inhibition was found to be lethal in lung, adrenal gland and renal carcinoma cell lines with mutations LQ WKH 6:,61) FRPSRQHQWV ARID1A, PBRM1 and SMARCA4 (47). Kim et al. SURSRVH WKDW 6:,61) PXWDQW WXPRUV could generally depend on EZH2 activity. Their data suggests that EZH2 inhibition is only lethal when the EZH2-PRC2 complex interaction is destabilized, an HႇHFW ZKLFK *6. GLG QRW DFKLHYH in all cell lines. Based on the Achilles project they additionally showed RAS mutations to predict resistance to EZH2 LQKLELWLRQLQ6:,61)PXWDQWFDQFHUFHOO lines. Future inhibitors of EZH2 with the ability to disrupt EZH2-PRC2 complex interaction can presumably be applied in ARID1A, PBRM1 and SMARCA4 mutant cancers with wild-type RAS.

In other work by Bitler et al. shRNA interference of 11 histone deacetylases (HDACs) in ARID1A mutant OCCC led to the discovery of a lethal relationship between ARID1A loss and inhibition of HDAC6, an epigenetic protein known to deacetylate numerous substrates (48). Chemical inhibition of HDAC6 ZLWK$&< ZDV RQO\ HႇHFWLYH LQ DQ ARID1A mutant background in a panel of four ARID1A mutant versus four ARID1A wild-type OCCC cell lines (10-fold lower IC₅₀) and in orthotopically transplanted ARID1A mutant xenografts. They further LGHQWL¿HG$5,'$DVGLUHFWWUDQVFULSWLRQDO repressor of HDAC6. ARID1A loss loss of the proliferation marker Ki67

and increased apoptosis. Similar results were found in xenografts. Approximately two-fold increase in GSK126 sensitivity was observed in ARID1A mutant (n=4) versus ARID1A wild-type (n=3) OCCC 3D cell line models. Gene expression SUR¿OLQJRIARID1A mutant cells treated ZLWK *6. LGHQWL¿HG PIK3IP1, a QHJDWLYH UHJXODWRU RI 3,.$.7P725 signaling, as a mechanistic link for the synthetic lethality between ARID1A loss and EZH2 inhibition (45). PIK3IP1 was found to be downregulated upon ARID1A loss. After ARID1A loss EZH2 methyltransferase induced H3K27Me3 methylation of the PIK3IP1 gene thus preventing expression of PIK3IP1, which LVWKHQIROORZHGE\LQGXFWLRQRI3,.$.7 mTOR signaling and proliferation (Fig. 4). GSK126-induced loss of the PIK3IP1 promoter H3K27Me3 mark resulted in PIK3IP1 expression and lethality, demonstrating the addiction of ARID1A mutant ovarian cancer cells to low PIK3IP1 levels (43). Consecutive work indicated that ARID1A mutant OCCC is selectively susceptible to inhibition of HDAC2, a known binding partner of the EZH2-containing PRC2 complex (46). As it turns out, HDAC2 only interacts with this complex in the absence of ARID1A. In line with the observations on EZH2 inhibition, knockdown of HDAC2 or chemical inhibition of HDAC2, using the QRQVSHFL¿F +'$& LQKLELWRU YRULQRVWDW

(26)

2

25

previously described mutual exclusive relation between ARID1A and ARID1B can at least partially explain the lethal interaction between BET inhibitor mediated BRD2 inhibition and ARID1A loss (36). In addition to ARID1B UHGXFWLRQ PRUH 6:,61) PHPEHUV were downregulated at the mRNA level, potentially augmenting to the lethal SKHQRW\SHE\IXUWKHUUHGXFLQJWKH6:, 61)IXQFWLRQ7KLVVWXG\SURYLGHVD¿UVW opportunity to chemically inhibit ARID1B expression and utilize the ARID1A and ARID1B mutual exclusive properties in ARID1A mutant OCCC.

4.3 Synthetic lethality of ROS induction with ARID1A

Previous observations indicated that 6:,61) IXQFWLRQ LV UHTXLUHG IRU oxidative stress resistance in the model organisms Saccharomyces cerevisiae and Caenorhabditis elegans (52, 53). A comparison of about 140 drug sensitivities in ARID1A mutant versus wild-type human cancer cell lines in the “genomics of drug sensitivity in cancer” database (cancerrxgene.org), encompassing over 700 cancer cell lines, revealed the HSP90 inhibitor and reactive oxygen species (ROS) inducing DJHQWHOHVFORPRODVWKHPRVWGLႇHUHQWLDO sensitive drug (54). Only cell lines with ARID1A frameshift or nonsense mutations were retained in the analysis, potentially generating a bias considering the proportion of cancer patients with this type of ARID1A mutations (9). Subsequently, Kwan et al. proved elesclomol sensitivity to be higher in ARID1A mutants in a panel of 11 ovarian cancer cell lines, including four OCCC cell lines, and three endometrial cancer FHOO OLQHV (OHVFORPRO ZDV ¿YH WR VL[IROGPRUHHႇHFWLYHLQARID1A mutant cells compared with ARID1A wild-type cells and increased ROS levels and apoptosis. Re-expression of ARID1A induced resistance to elesclomol. led to re-expression of HDAC6, which

VSHFL¿FDOO\GHDFHW\ODWHSO\VLQHD residue known to regulate p53-mediated apoptosis (Fig. 4). Knockdown of TP53 reverted ACY1215 mediated apoptosis induction and proliferation inhibition in ARID1A mutant cells, illustrating that ARID1APXWDQW2&&&VSHFL¿F$&< sensitivity is p53 dependent. Additional data demonstrated that acetylated p53-lysine120 localizes to mitochondria and destabilizes mitochondrial membrane potential, presenting a mechanism for the lethal relationship between ARID1A loss and HDAC6 inhibition.

Recently, a fourth epigenetic determinant of ARID1A mutation dependent synthetic lethality was LGHQWL¿HGE\VK51$PHGLDWHGVFUHHQLQJ against all human kinases. In a panel of nine ARID1A PXWDQW YHUVXV ¿YH ARID1A wild-type OCCC cell lines, knockdown of the BET bromodomain member BRD2 HVWDEOLVKHG VSHFL¿F lethality in ARID1A mutant lines (49). BET bromodomains bind acetylated lysine histone tails and are involved in transcriptional regulation, but have also been reported to act as kinases (50, 51). As a result of high homology between WKH%(7PHPEHUV%5'DQG%5'7 only inhibitors that target all four proteins are available to date. The BET inhibitors -4DQGL%(7GHPRQVWUDWHGVSHFL¿F sensitivity in ARID1A mutant OCCC FHOOVZKLFKZDVYHUL¿HGLQWZRARID1A GH¿FLHQW LVRJHQLF 2&&& FHOO OLQH SDLUV in ARID1A mutant OCCC xenografts and patient-derived xenograft models. Explicitly, ARID1A mutant cells had over two-fold stronger growth reduction DIWHU-4WUHDWPHQWDVREVHUYHGLQWKH panel of nine ARID1APXWDQWYHUVXV¿YH ARID1AZLOGW\SH2&&&FHOOOLQHV-4 or shRNA mediated inhibition of BRD2 reduced the expression of ARID1B. Chromatin immunoprecipitation assays indicated direct transcriptional regulation of ARID1B expression by BRD2 at the ARID1B promoter. Therefore, the

(27)

CHAPTER 2 26

other dasatinib targets.

4.5 Targeting of PI3K/AKT signaling in ARID1A mutant cancers

Results from the Achilles project demonstrated PIK3CA to be the second best hit for synthetic lethality with ARID1A loss (36). Accordingly, another analysis from the drug sensitivity in cancer database from Kwan et al. showed $=' DQ P725& LQKLELWRU WKDW WDUJHWV3,.$.7VLJQDOLQJGRZQVWUHDP as second best hit in ARID1A mutant cells in their screen (54). Both studies suggest a link between ARID1A loss DQG 3,.$.7P725 DFWLYDWLRQ 7KH co-existence of ARID1A mutations and DFWLYDWLRQ RI 3,.$.7P725 VLJQDOLQJ has been described in multiple cancer W\SHV6LJQL¿FDQWHQULFKPHQWRIPIK3CA activating mutations and PTEN loss were detected in ARID1A mutant endometrial cancer. Moreover, knockdown of ARID1A induced phosphorylation of the PI3K downstream target AKT (58). 3,.$.7 SDWKZD\ DFWLYDWLRQ RFFXUUHG after ARID1A depletion in MCF7 breast cancer cells. These cells gained AKT phosphorylation, and enhanced sensitivity to AKT and PI3K inhibitors upon ARID1A knockdown (59). In OCCC, concurrent PIK3CA activating mutations DQG 37(1 ORVV ZHUH VLJQL¿FDQWO\ associated with ARID1A mutations, but ARID1A knockdown did not induce is not supported by in vivo data and the

mechanisms underlying ROS induction after ARID1A loss are inconclusive (Fig. 4).

4.4 ARID1A synthetic lethality approaches using dasatinib

In another drug screening study, 68 clinically approved or late-stage clinically developed inhibitors were screened on three ARID1A wild-type and eight ARID1A mutant OCCC cell lines (55). The SRC, ABL and C-KIT LQKLELWRUGDVDWLQLEJDYHWKHPRVWVSHFL¿F LQKLELWRU\HႇHFWLQARID1A mutant OCCC cells, demonstrating a more than two-fold increased sensitivity compared to ARID1A wild-type OCCC cells. Dasatinib synthetic lethality was consistently found after ARID1A knockdown in two OCCC cell lines, one breast cancer cell line and ARID1A knockout in the cell line HCT116. Using mass spectrometry analysis of DFWLYH NLQDVHV ¿YH WDUJHWV RI GDVDWLQLE were shown to be upregulated in ARID1A mutant OCCC cells. The dasatinib target and SRC family protein YES1 was most VHOHFWLYH IRU $5,'$ GH¿FLHQW 2&&& cell lines. Dasatinib treatment increased G1 cell cycle arrest and caspase activity in ARID1A mutant lines, indicating a FHOOF\FOH DUUHVWDSRSWRWLF SKHQRW\SH Using siRNA screening, Miller et al. LGHQWL¿HG H[SUHVVLRQ RI S &,3 WAF1) and its downstream target RB1

(28)

2

27

AKT phosphorylation in OCCC cell lines (60, 61). One study described ARID1A mutant OCCC lines to have lower IC₅₀ for AKT inhibitors. However, no induction RI3,.$.7SDWKZD\DFWLYLW\ZDVIRXQG in these cells after ARID1A knockdown (59). Moreover, ARID1A status could not discriminate between IC₅₀ for PI3K and mTOR inhibitors in a large panel of OCCC cell lines and PDX models (11).

Some mechanistic links between $5,'$ ORVV DQG 3,.$.7 SDWKZD\ activation have been established. As stated earlier, ARID1A was found to LQKLELW 3,.$.7 DFWLYLW\ E\ UHJXODWLQJ expression of the PI3K suppressor PIK3IP1 in ovarian cancer (43). In EUHDVW FDQFHU $5,'$ ZDV LGHQWL¿HG as negative transcriptional regulator of ANXA1, a membrane bound protein and activator of AKT (Fig 4.) (62). These two studies suggest that ARID1A loss may indirectly activate PI3K in some cancer types. Given that PI3K and mTOR inhibitor sensitivity is not only dependent on ARID1A mutational status in OCCC,

WDUJHWLQJWKH3,.$.7SDWKZD\VKRXOG not be regarded as a synthetic lethal strategy for ARID1A mutant OCCC. 5. Clinical development of agents with $5,'$VSHFL¿FOHWKDOLW\

Multiple agents showing synthetic lethality in an ARID1A mutant context are in clinical development (Table 1). From the epigenetic targets with ARID1A PXWDWLRQ VSHFL¿F OHWKDOLW\ LQ 2&&& WKH ¿UVW JHQHUDWLRQ (=+ LQKLELWRUV (DZNep) gave toxicity in vivo. Two novel EZH2 inhibitors are now in clinical trials (63). An alternative for EZH2 targeting agents is HDAC2 inhibition using the clinically applicable broad HDAC inhibitor vorinostat. The HDAC6 inhibitor ACY1215 proved to be well tolerated in myeloma patients, supporting clinical applicability of HDAC6 inhibition for WUHDWLQJ$5,'$GH¿FLHQW2&&&SDWLHQWV in the future (64). BET bromodomain inhibition has attracted great interest for the treatment of cancer. Many BET

Table 1 | Clinical development of agents with ARID1APXWDWLRQVSHFL¿FOHWKDOLW\DQGWKHLULQGXVWU\ QDPHVIURP&OLQLFDO7ULDOVJRYKWWSFOLQLFDOWULDOVJRY

Molecular target

Most advanced clinicalphase

Compounds in clinical development, most advanced are depicted in bold

EZH2 Phase II EPZ-6438, CPI-1205

HDAC6 Phase II ACY-125, KA2507

BRD2 Phase II I-BET-762, GSK2820151, RO6870810, CPI-0610,

(BET) *6,1&%,1&%$%99

BMS-986158, FT-1101, PLX-51107, SF1126, CC90010, ZEN003694, BI894999, N-methyl pyrrolidone, ODM-207

YES1 3KDVH,9 BMS-354825, AZD0530, SKI-606

ATR Phase II VX-970, BAY1895344, AZD6738

PARP1/2 3KDVH,9 AZD2281, ABT-888, BMN673, AG-014699, MK-4827,

SHR3162, SC10914, SOMCL-9112, BGB-290

HSP90 Phase II AT13387, TAS-116, SNX-2112, XL888, PEN-866

3,.Įȕį Phase III BKM120, RP6530, BYL719, PKI-587, AZD8186,

BAY80-6946, TGR-1202, GSK2636771, SF1126, KA2237, GDC0032, INK1197, GDC-0077, GDZ173, GS-1101

(29)

CHAPTER 2 28

trials for diverse genetic backgrounds LQFDQFHUXWPRVWLQSKDVH,,,,9DQG,, respectively. Considering the advanced clinical development of PI3K and PARP targeting compounds, including trials in ovarian cancer patients, these agents may provide direct therapeutic opportunities for ARID1A mutant cancer patients including ARID1A mutant 2&&&7KRXJKHႈFDF\DJDLQVWARID1A wild-type OCCC should not be excluded. 6. SMARCA2 and SMARCA4-directed synthetic lethal strategies

Aside from synthetic lethal targeting strategies in the context of ARID1A PXWDWLRQV WKH 6:,61) FRPSOH[ members SMARCA2 and SMARCA4 have been exploited for this purpose as well. Having SMARCA2 and SMARCA4 mutations in approximately 2% and 5% RIWKHWXPRUV2&&&FRXOGEHQH¿WIURP such strategies (15). In 2013, Oike et al. described susceptibility for SMARCA2 LQKLELWLRQ LQ 60$5&$ GH¿FLHQW QRQ small cell lung cancer. Treatment with SMARCA2 siRNA induced markers of senescence in SMARCA4 mutant cell lines (66). An epigenome directed shRNA screen further supported the ¿QGLQJWKDW60$5&$GH¿FLHQWFDQFHUV GHSHQG RQ 60$5&$ 60$5&$ SMARCA4 synthetic lethality was later demonstrated in multiple other cancer types (67). SMARCA4 loss resulted

cancer types. Somehow these cancers have overcome dependency on either RQHRIWKH6:,61)$73DVH0RUHRYHU restoration of SMARCA2 or SMARCA4 inhibited proliferation (69). These results VXJJHVW WKDW GHSHQGHQF\ RQ 6:, SNF ATPases is lost in SCCOHT. In line with these observations, another VWXG\ LGHQWL¿HG VHYHUDO SMARCA4 mutant cancers cell lines with absent or low expression of SMARCA2. Inhibition of the PRC2 subunit EZH2 re-expressed SMARCA2 only in the EZH2 sensitive SMARCA4 mutant subset of FHOO OLQHV (=+ LQKLELWLRQ VSHFL¿FDOO\ induced apoptosis in one (the ovarian HQGRPHWULRLGFDQFHUFHOOOLQH729' of the four EZH2 sensitive SMARCA4 mutant lines (70). It is unknown whether WKH 6:,61) IXQFWLRQ LV UHWDLQHG DIWHU loss of SMARCA2 and SMARCA4 or that these tumors are more dependent on other chromatin remodelers. Chemical inhibitors of SMARCA2 and SMARCA4 are currently under development. The ¿UVW 60$5&$60$5&$ VSHFL¿F inhibitor PFI-3 did not show anti-SUROLIHUDWLYHHႇHFWVLQOXQJFDQFHUFHOOV SRVVLEO\EHFDXVHRILQVXႈFLHQWLQKLELWLRQ of SMARCA2 binding to DNA (71). 7R XWLOL]H WKH 60$5&$60$5&$ dependency in a clinical setting, development of inhibitors that target either SMARCA2 or SMARCA4 might EHEHQH¿FLDOWRDFKLHYHWXPRUVHOHFWLYH lethality.

(30)

2

29

7. Conclusion and future considerations The abundance of ARID1A loss of function mutations across cancer types designates mutant ARID1A as an attractive target for synthetic lethal approaches. Mechanistic insight into inhibitor induced synthetic lethality with ARID1A loss is at least partially revealed (Fig. 4). EZH2, HDAC2, HDAC6, BRD2 and YES1 inhibition were found to be VSHFL¿FDOO\ OHWKDO LQ ARID1A mutant OCCC. Additional lethal targets that KDYH EHHQ YHUL¿HG DFURVV PXOWLSOH other ARID1A mutant cancer lineages, included PARP, ATR and HSP90 (Fig. 5).

Inhibition of the synthetic lethal targets HDAC2 and HDAC6 demonstrated a ODUJH GLႇHUHQFH LQ VHQVLWLYLW\ EHWZHHQ ARID1A mutant and wild-type OCCC cell lines. Since both targets were tested in a limited number of cell lines evaluation in a larger cell line panel would be required to verify the robustness of HDAC2 and HDAC6 inhibition. Synthetic lethality of drugs targeting EZH2 and BRD2 in ARID1A mutant OCCC was less pronounced. Though, the advantage of these last two studies was the use of 3D in vitro models for the assessment of sensitivity to EZH2 inhibition and the use of patient-derived xenotransplants in addition to a large panel of OCCC cell lines to determine sensitivity to BRD2 LQKLELWLRQ 7KH GLVWLQFW IXQFWLRQ RI 6:, SNF complexes among tissues indicates that molecular dependencies in ARID1A PXWDQW FDQFHUV DUH OLQHDJH VSHFL¿F thus requiring testing of ATR and HSP90 inhibitors in the context of OCCC. For example, PARP inhibitor potency has been evaluated in OCCC cell lines but did not seem to provide selective sensitivity in ARID1A mutant OCCC cells (11, 72). We found no enrichment for alterations in DNA repair genes in ARID1A mutant versus wild-type OCCC tumors and cell lines, which may provide an explanation for this observation (11). However, genetic evidence using isogenic OCCC

models still needs to be provided. Besides, ARID1A was recently assigned an important role in mismatch repair as it recruits the mismatch repair gene MSH2 to chromatin during DNA replication. ARID1A loss correlated ZLWK PLVPDWFK UHSDLU GH¿FLHQF\ KLJK mutational load and increased numbers RI WXPRULQ¿OWUDWLQJ O\PSKRF\WHV across many human cancer types (73). These data propose that ARID1A loss induces microsatellite instability, which subsequently provides a vulnerability to immunotherapy. In ARID1A mutant syngeneic mouse ovarian and colorectal cancer models high susceptibility to immune checkpoint blockade of PD-L1 compared to isogenic ARID1A wild-type models was already shown (73). A small study in OCCC patients found high PD-/H[SUHVVLRQDQGPRUHWXPRULQ¿OWUDWLQJ lymphocytes in microsatellite instable cancers, but the ARID1A mutation status was not determined (74). However, the frequency of microsatellite instability in OCCC is relatively low (10-14%), implying that the percentage of OCCC

Figure 5 | Synthetic lethal therapies identi-¿HGLQARID1A mutant OCCC and other tumor types. Inhibition of EZH2, HDAC2, HDAC6, BRD2

and YES1 were found to be synthetic lethal in ARI-'$GH¿FLHQW2&&&ARID1A mutation dependent (driven) synthetic lethality was observed in other tumor types with ROS induction (via HSP90 inhibi-tion) and inhibition of ATR, PARP, PI3K and PD-L1. Whether these synthetic lethal therapies can be applied to OCCC remains to be proven.

OCCC Applicable to OCCC? $75 3$53 526 3,. 3'/ ,QKLELWRU ,QKLELWRU ,QGXFHU ,QKLELWRU ,QKLELWRU (=+ +'$& +'$& %5' <(6 ,QKLELWRU ,QKLELWRU ,QKLELWRU ,QKLELWRU ,QKLELWRU SWI/SNF ARID1A

University of Groningen Kinome directed target discovery and validation in unique ovarian clear cell carcinoma models Caumanns, Joost

Kinome directed target

discovery and validation in

unique ovarian clear cell

carcinoma models

Kinome directed target discovery and

validation in unique ovarian clear cell

carcinoma models

Proefschrift

ter verkrijging van de graad van doctor aan de

Rijksuniversiteit Groningen

op gezag van de

UHFWRUPDJQL¿FXVSURIGUE. Sterken

en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op

16 januari 2019 om 16:15 uur

door

Joseph Johannes Caumanns

geboren op 16 November 1990

te Hengelo (O)

CHAPTER 1

1

1

1

CHAPTER 2

ARID1A mutant ovarian clear cell

carcinoma: A clear target for synthetic

lethal strategies

Joseph J. Caumanns

, G. Bea A. Wisman

, Katrien Berns

Ate G.J.

van der Zee

and Steven de Jong

2

2

2

2

2

2

UHFWRUPDJQL¿FXVSURIGUE. Sterken

_{, G. Bea A. Wisman}

_{, Katrien Berns}

_{Ate G.J.}

_{and Steven de Jong}