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

(1)

University of Groningen

Kinome directed target discovery and validation in unique ovarian clear cell carcinoma models

Caumanns, Joost

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: 117PDF page: 117PDF page: 117PDF page: 117

CHAPTER 6

Establishment and characterization of

ovarian clear cell carcinoma

patient-derived xenograft models

Joseph J. Caumanns

1

_{, Shang Li}

2

_{, Tushar Tomar}

1

_{, Jolanda A.L.}

Liefers-Visser

2

_{, Gert J. Meersma}

1

_{, Ate G.J. van der Zee}

1

_{, G. Bea A.}

Wisman

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)

CHAPTER 6

6

118

Besides mutations in TP53 in 90% of the tumors, other mutations in high-grade serous ovarian carcinoma (HGSOC) are mainly found in BRCA1 and BRCA2 (2). In contrast, a heterogeneous mutation pattern is found in OCCC. The most frequently altered genes are ARID1A, HQFRGLQJ D 6:,61) FKURPDWLQ remodeling subunit (40-57%), PIK3CA encoding the PI3K catalytic subunit (30-40%) and KRAS encoding a MAPK

INTRODUCTION

Ovarian clear cell carcinoma (OCCC) is the second most common epithelial ovarian cancer subtype. Initial chemoresistance attributes to a poor prognosis for advanced stage OCCC patients compared to their stage matched high-grade serous ovarian carcinoma, indicating that novel therapies are needed (1).

Establishment and characterization of ovarian clear cell

carcinoma patient-derived xenograft models

Ovarian clear cell carcinoma (OCCC) is a subtype of ovarian cancer with high chemoresistance and a poor prognosis in advanced stage, which emphasizes that novel therapies for OCCC are warranted. Patient-derived xenograft (PDX) models are an in vivo platform of increasing interest to test therapy strategies as they are supposed to more accurately represent patient tumors compared to conventional in vivo models. In this study, we established seven OCCC PDX models and compared histopathology, mutation status, and copy number SURÀOHVEHWZHHQSDLUHGSDWLHQWDQG3';2&&&WXPRUVWRGHWHUPLQHWKHOHYHO of similarity. To establish PDX models of OCCC, tumor tissues from 14 OCCC patients who underwent surgery were implanted subcutaneously in NOD scid gamma (NSG) mice. PDXs and paired patient tumors were characterized by sequencing frequent OCCC mutations. Morphological characteristics were assessed with immunohistochemistry and SNP array analysis was used to determine copy number alterations. Successful engraftment of OCCC patient tumors was obtained for seven patients (50%). Primary implantation (F1) VKRZHGDKLJKHUHQJUDIWPHQWZLWKIUHVKSDWLHQWWXPRUWLVVXHÀYHVHYHQ YHUVXVYLWULÀHGWXPRUWLVVXHWZRVHYHQ6XFFHVVUDWHRILPSODQWHGWXPRU pieces in F2 was higher than those in F1. In addition, latency time was 50% shorter and, in agreement with Ki67 staining results, tumor growth rate was faster in F2. Mutations in the OCCC-related genes ARID1A, PIK3CA, PTEN,

ATM and BRCA1 were retained during engraftment. Morphological features

and tumor copy number alterations were also comparable between paired tumor and F2 PDXs. Furthermore, pathway analysis using copy number alterations revealed that amongst others VEGF signalling, ribosome and MAPK signalling pathways were enriched both in paired patient tumors and F2 PDXs. In conclusion, seven OCCC PDX models were established that largely mimicked matched patient tumors at the histopathological and genomic level. Accordingly, these PDXs can serve as relevant preclinical models for future translational research in OCCC.

Joseph J. Caumanns, Shang Li, Tushar Tomar, Jolanda A.L. Liefers-Visser, Gert J. Meersma, Ate G.J. van der Zee, G. Bea A. Wisman and Steven de Jong

(4)

ESTABLISHMENT AND CHARACTERIZATION OF OCCC PDXS

6

119

alterations (CNAs) and mutational signatures of the matching patient WXPRU 0RUHRYHU HႈFDF\ RI platinum based treatment in ovarian cancer PDX models correlated with platinum responses in their equivalent patients (18-20). Our group previously reported biobanking of primarily serous ovarian cancer and characterized the preservation of pathological and genomic features in corresponding PDX models (17).

In the present study, we described the establishment of seven OCCC PDX models and made a comparison between PDXs and their matched patient tumors using histopathology and genomic features, such as mutational status and CNAs.

RESULTS

Establishment of PDX models from OCCC patients

pathway signaling node (5-14%) (3-7). Cell line based xenografts of many tumor types, including OCCC, have been broadly applied for in vivo drug testing, but their clinical predictive value is being argued (8, 9). An alternative for xenografts are genetically engineered mouse models that spontaneously develop cancer. For example mice with typical OCCC mutations can develop ovarian tumors with clear cell characteristics (10-12). Patient-derived xenograft (PDX) models are of increasing interest for pre-clinical drug evaluation, as they more accurately represent patient tumors in terms of genomic heterogeneity compared with cancer cell lines (13). However, clonal selection can be observed throughout passaging in mice, indicating that these models should be used at early passages (14).

Ovarian cancer PDX models of high-grade serous subtype were shown to preserve histology, copy number

PDX model Surgery type FIGO stage Chemotherapy Survival PDX.155 PDX.180 PDX.228 PDX.232 PDX.237 PDX.245 PDX.247 Laparoscopy Interval debulking Interval debulking Interval debulking Incomplete debulking Complete debulking Complete debulking IIIC IIIC IIIC IIIC IIIC IA IIB Carbo/pacl Neo-adjuvant carbo/pacl Neo-adjuvant carbo/pacl Neo-adjuvant carbo/pacl Carbo Carbo Carbo/pacl 9 months 10 months 12 months >37 months 12 months >24 months >29 months Table 1. Clinicopathological characteristics of OCCC tumors implanted as PDX.

Complete debulking surgery was defined as < 2 cm of remaining residual disease. Abbreviations; FIGO, International Federation of Gynaecology and Obstetrics; carbo, carboplatin; pacl, paclitaxel.

Type of Implantation Fresh Fresh Vitrified Vitrified Fresh Fresh Fresh Engraftment Successful Successful Successful Successful Successful Successful Successful PDX.73 PDX.80 PDX.220 PDX.261 PDX.266 PDX.309 PDX.349 Laparoscopy Laparoscopy Laparoscopy Interval debulking Laparoscopy Complete debulking Incomplete debulking IIIB IIIC IIIC IV IIIC IIIB IIIC Carbo/pacl Carbo/pacl Carbo/pacl Neo-adjuvant carbo/pacl Carbo/pacl Carbo/pacl Carbo/pacl >52 months >1 month >1 month 17 months >3 months >22 months 10 months Fresh Fresh Vitrified Vitrified Vitrified Vitrified Vitrified Failed Failed Failed Failed Failed Failed Failed

(5)

CHAPTER 6

6

120

IIB and above) who received debulking surgery, except for PDX 245 (FIGO stage IA). Three of these advanced FIGO stage OCCC patients had received neo-DGMXYDQW FDUERSODWLQSDFOLWD[HO 7DEOH 1). PDXs were successfully established from freshly implanted patient tumors and with a lower success rate from YLWUL¿HGSDWLHQWWXPRUV¿YHVHYHQYHUVXV WZRVHYHQ HVWDEOLVKHG 3'; PRGHOV OCCC tumor specimens from 14 patients

were collected during surgery and implanted in mice between November 2013 and April 2017 according to previously described methods (17). Seven OCCC PDX models were established, resulting in a success rate of 50%. The successfully engrafted PDX models primarily originated from patients with advanced FIGO stage OCCC (FIGO

PDX.155 F1 #1 F1 #2 F2 #2 F2 #1 PDX.180 PDX.228 PDX.232 PDX.237 PDX.245 PDX.247 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 500 1000 1500 2000 Tu m or v ol um e ( m m 3) 0 50 100 150 200 Time (days) 0 50 100 150 200 Time (days) 0 50 100 150 200 Time (days) 0 50 100 150 200 Time (days) 0 100 200 300

Time (days) 0 100Time (days)200 300 0 50 100 150 200 Time (days) F1 #1 F1 #2 F1 #3 F2 #1 F2 #2 F2 #3 F1.Vh #1 F2 #1 F2 #2 F2 #3 F1.Vh #1 F1.Vh #2 F1.Vh #3 F2 #1 F2 #2 F1 #1 F1 #2 F1 #3 F2 #1 F2 #2 F1 #1 F1 #2 F1 #3 F2 #1 F2 #2 F1 #1 F1 #2 F1 #3 F2 #1 F2 #2

Figure 1 | Establishment of seven OCCC PDX models. Growth characteristics of fresh implanted or

vitri-¿HGLPSODQWHG9KWXPRUVIURPLQGLYLGXDOPLFHLQ3';)DQG)'RWWHGOLQHVUHSUHVHQWJURZLQJWXPRUV in F1 that were used for implantation in F2.

Succes rate of PDX Median latency in days F1 (range) 7/14 (50%) Freshly implanted 74 (34-131) 5/7 (71%) 2/7 (29%) Total Vitrified implanted 74 (47-131) * (34-126)

Table 2. Growth characteristics of PDX models.

Latency time defined as time from implantation until first observed tumor size above 200 mm3. * no median given, only 2 datapoints available Median latency in days F2 (range) 34 (29-105) 35 (33-105) * (29-34) Growing pieces in F1 Growing pieces in F2 15/36 (42%) 29/38 (76%) 13/28 (46%) 21/28 (75%) 2/8 (25%) 8/10 (80%)

(6)

6

121

mutated genes in OCCC, were found in two models. Model 155 contained a homozygous ARID1A stop-gain mutation DD 4 ZKLOH PRGHO ZDV heterozygous mutated in PIK3CA (aa K111E), an alteration that is recurrently found in cancer (COSM13570) and led to increased downstream AKT phosphorylation (21). Both mutations were retained during engraftment in F1 and F2. Additional mutations in patients, F1 and F2 were found in ATM, PTEN,

BRCA1 and CTNNB1. Within this set of

40 genes de novo mutations were not found in F1 PDXs or F2 PDXs (Table 3). The patient tumor from model 232 was mutant for TP53 and BRCA1. The patient tumors from models 228, 237 and 247 were wild-type for all the sequenced genes (Table 4). F1 and F2 of these four models were not sequenced.

The morphological resemblance of PDX generations with their patient tumors was assessed with hematoxylin & eosin (H&E) staining (Fig. 2A). Histological characteristics of model 155, 232, 245 and 247 were preserved in F1 and F2 tumors compared with their patient tumors, except that the amount of stromal tissue in model 232 and model 245 decreased over generations. The tumor cell percentage in patient tissue of model 180, 228 and model 237 was relatively low. As a measure of proliferative capacity of the tumor cells, levels of the proliferation marker Ki67 were determined (Fig. 2B). Ki67 staining However, the growth latency time of the

engrafted tumors was similar between both groups. PDX median growth latency was shorter in F2 compared to F1 (Table 2). Only for PDX 155 and PDX 237, latency remained stable across F1 and F2 (Fig. 1). In addition, we calculated how many tumor pieces were engrafting in F1 and F2. In the group of PDX models established from freshly implanted tumors, a higher percentage of tumor pieces were successfully engrafted in F2 (75%) compared to F1 (46%) (p = 0.05) (Table 2). Although only two PDX PRGHOV ZHUH HVWDEOLVKHG IURP YLWUL¿HG patient material, a similar improvement in engraftment of tumor pieces was observed between F2 and F1 (80% vs 25%, p = 0.05) (Table 2). These results suggest that implantation of fresh patient tumor material is preferred to establish PDX models. However, once established LQ PLFH QR GLႇHUHQFHV DUH REVHUYHG during expansion of the tumors in F2.

Mutation and histopathological analysis

Six of the seven established models (patient tumors as well as F1 and F2 PDX tumors) were sequenced for mutations in a panel of commonly mutated kinases and other cancer related genes in OCCC, as described before (Supplementary Table 1) (7). PDX model 245, derived from an early FIGO stage (IA) tumor, was not included. Mutations in ARID1A and PIK3CA, the two most frequently

ARID1A

PDX.155 PDX.180

PIK3CA TP53 ATM PTEN BRCA1 CTNNB1

wt K111E wt wt wt T1020I H93R wt V356A, 1533* wt wt D32Y 1148* wt

Table 3. Mutations in patient and PDX in 40 frequently mutated genes in OCCC.

From both PDX models the patient tumor, F1 and F2 tumor were sequenced. Abbreviations; *, stop-gain mutation. ARID1A Patient.228 Patient.232 Patient.237 Patient.247

PIK3CA TP53 ATM PTEN BRCA1 CTNNB1 wt wt wt wt wt R273H wt wt wt wt wt wt wt wt wt wt wt 895fs wt wt wt wt wt wt wt wt wt wt

Table 4. Patient tumor mutations in 40 frequently mutated genes in OCCC.

Patient tumors from 4 established PDX models were sequenced. Abbreviations; fs, frameshift mutation.

(7)

CHAPTER 6

6

122 H&E Ki67 A B F2 F1 Patient PDX.180 F2 F1 PDX.155 PDX.247 PDX.245 PDX.237 PDX.232 PDX.228 100 μm Patient 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm

TP53 mutation (aa R273H) resulting in

accumulation of p53, had the highest p53 staining intensity together with model 247 (Fig. 2C) (22). ARID1A was only stained on patient tumors and showed high intensity in the patient tumors from model 180, 232, 245 and 247. Low ARID1A staining intensity was observed in the ARID1A homozygous mutated (aa 4 SDWLHQWWXPRUDQGLQSDWLHQW tumor 228 and 237 (Fig. 2D).

in model 155 was similar in F1 and F2 compared to the patient tumor, which is in agreement with the similar growth rate of PDX 155 in F1 and F2. All other models showed increased Ki67 staining in F1 and F2 relative to their patient tumor with the highest Ki67 positivity in F2.

All models showed p53 positive staining, which varied in intensity across patient tumors, F1 and F2. Model 232, that carried a known gain of function

(8)

6

123 TP53 PDX.180 PDX.155 PDX.247 PDX.245 PDX.237 PDX.232 PDX.228 ARID1A C D F2 F1 Patient Patient 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm 100 μm

Figure 2 | Expression

anal-ysis. (A) H&E, (B) Ki67, (C) TP53, (D) ARID1A, immuno-histochemical analysis of pa-tient tumors, PDX F1 and F2 tumors.

in the F2 PDX tumor. CNAs in patient tumors model 180 and 247 were more abundantly present in F2, while patient tumor 232 had multiple CNAs that were lost in F2. Overall, the displayed copy number gains and losses in paired patient tumor and F2 PDX tumors matched previously described CNAs in OCCC, including gains in chromosome 3, 8 and 20 and losses in chromosome 4, 13 and 7KHTXDQWL¿HGSDLUZLVH&1$ comparison between patient tumors

Copy number alterations

To compare patient tumors and PDX genomes, paired patient and F2 PDX tumors from all seven established models were subjected to genome wide CNA analysis (Fig. 3A). Model 228, 237 and 245 did not have considerable chromosome wide gains and losses in both patient and F2 PDX tumors. Patient tumor 155 had CNAs along all chromosomes that were retained

(9)

CHAPTER 6

6

124 g Patient.245 PDX. 245. F 2 Patient.247 PDX. 247. F 2 Patient.155 PD X .155 .F 2 Patient.180 P D X .1 80.F2 P at ient.228 PDX. 228. F 2 Patient.232 PDX. 232. F 2 Patient.237 PDX. 237. F 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1819202122 Chromosome A

(10)

6

125

than the percentage of genes with non-concurrent CNAs (2.9%) in four models or more. Additionally, nearly half of the concurrent CNAs existed in more than six models, while 601 out of 686 genes with non-concurrent CNAs were found in maximally four models, suggesting that the majority of CNAs were maintained during engraftment.

Hierarchical clustering using thresholded CNAs of all genes (n=23109) showed consistently that most F2 PDX tumors clustered to their patient tumors. Based on Pearson correlation, the seven patient tumors and F2 tumors (14 samples in total) were separated into three clusters (Fig. 4A). Patient tumor and F2 tumor of model 155 and patient tumor of model 232, all with a large amount of CNAs, clustered together, separated from the other samples. Another cluster consisted of patient tumor 180 and 245 and PDXs matched the genome-wide

CNA illustration (Fig. 3A-B). Both the CNA Gscore ranges and the average of absolute Gscore of model 228, 237 and 247 were comparable in both patient and PDX tumors. Since most CNAs were retained in F2 in model 155, 180 and 247 with a higher magnitude, these three PDX models had a wider range of CNA Gscores and a higher average absolute Gscore compared to their patient tumor (Fig. 3B). In model 232, both range and average value of the Gscore decreased in F2. Next, we determined if CNAs between patient tumors and F2 tumors were concordant in OCCC models. Concurrent and non-concurrent CNAs of genes in the majority of the PDX models (four models or more) are summarized in Table 5. In these models 67.2% of the genes had a concurrent CNA status, which was more than 20 times higher

Patient.245 PDX.245.F2 Patient.247 PDX.247.F2 Patient.155 PDX.155.F2 Patient.180 PDX.180.F2 Patient.228 PDX.228.F2 Patient.232 PDX.232.F2 Patient.237 PDX.237.F2 -1 0 1 2 3 G score PDX.F2 Patient Tumor

B _{Figure 3 | Copy number}

analysis. (A) CNAs in patient

tumors and PDX F2 tumors across 22 chromosomes. Blue indicates copy num-ber gain, red indicates copy number loss. (B) Scatter plot of CNA Gscore for Patient tu-mors and PDX F2 tutu-mors. The grey dashed line indicates the default threshold (0.1) used in GISTIC procedure, black sol-id lines indicates the average absolute Gscore of each sam-ple (also indicated in text).

(11)

CHAPTER 6

6

126 4 models Amplification Deletion Stable Sum Gained amp Gained del

5 models 6 models 7 models Sum % of genome

1513 266 2818 4597 9 55 2425 194 1672 4291 4 14 1969 56 279 2304 0 0 15693 686 67.91 2.97 1452 395 2654 4501 296 226

Table 5. CNA gene status in 4 PDX models or more.

Lost amp Lost del Concurrent Non-concurrent 53 26 3 0 0 0 0 0 Inverted Sum 1 601 0 67 0 18 0 0

and F2 tumor 180 and 247, whereas the tumors with low CNAs were in the last cluster (F2 tumors of model 245, 228, 232 and 237 and patient tumors 228, 237 and 247). Clustering analysis was also performed on patient tumors and F2 tumors separately. The cluster trees of the patient tumors (Fig. 4B) and F2 tumors (Fig. 4C) were organized mainly in a similar order, but with a stronger Pearson correlation between the F2 tumors. Overall, hierarchical clustering indicated that OCCC PDXs in F2 can represent their corresponding patient tumors in terms of CNAs.

*HQRPLF ,GHQWL¿FDWLRQ RI 6LJQL¿FDQW Targets in Cancer (GISTIC) analysis of the patient tumors and F2 tumors XQFRYHUHG VLJQL¿FDQWO\ DPSOL¿HG DQG GHOHWHG JHQHV )LJ 6LJQL¿FDQWO\ DPSOL¿HGJHQHVZLWKD)'5ERWKLQ the patient tumor set and F2 tumor set included GLI2, MEOX2, AQP1, SGK223,

EXPH5 and BAIAP3.

To investigate commonly enriched SDWKZD\V DIWHU HQJUDIWPHQW '$9,' pathway enrichment analysis was performed using the concurrently DPSOL¿HG RU GHOHWHG JHQHV LQ SDWLHQW and F2 tumors in four models or more from Table 5. Among the pathways ZLWK DPSOL¿HG DQG GHOHWHG JHQHV WKDW surpassed FDR<0.25 (Fig. 6A), we IRXQG WKH 9(*) VLJQDOLQJ SDWKZD\ MAPK signaling pathway and the PI3K-Akt signaling pathway. The two pathways with deleted genes were arrhythmogenic

right ventricular cardiomyopathy and the axon guidance signaling pathway. *HQHV WKDW ZHUH FRQFXUUHQWO\ DPSOL¿HG or deleted in all seven models were indicated in the pathway illustration (Fig. 6B).

DISCUSSION

Here, we report the establishment of VHYHQ 2&&& 3'; PRGHOV DQG GH¿QHG their genetic and histopathological characteristics. OCCC patient tumors had a 50% engraftment success and fresh implantation was more successful FRPSDUHG WR LPSODQWDWLRQ IURP YLWUL¿HG tumors. Typical OCCC mutations were found in our PDX panel, such as ARID1A loss of function and an activating PIK3CA mutation. The majority of the OCCC PDX models retained their patients’ tumor morphology and CNAs signature. Our PDX panel can therefore be used as representative models of OCCC for preclinical testing of therapy strategies. 7KH HQJUDIWPHQW HႈFDF\ RI 2&&& PDXs in our study was comparable to previous work where OCCC tumor specimens were implanted either VXEFXWDQHRXV ¿YH HQJUDIWPHQWV IURP 12 tumors) or subrenal (two from four) (24, 25). Another study established PDXs by intraperitoneal inoculation of minced patient tumors, which could explain the high engraftment success (11 from 12) (18). Besides PDX 155, containing an ARID1A deleterious

(12)

6

127 PDX.232.F2 PDX.245.F2 PDX.237.F2 Patient.247 Patient.228 PDX.228.F2 Patient.237 PDX.180.F2 PDX.247.F2 Patient.245 Patient.180 PDX.155.F2 Patient.155 Patient.232 PDX.232.F2

PDX.245.F2 PDX.237.F2 Patient.247 Patient.228 PDX.228.F2 Patient.237 PDX.180.F2 PDX.247.F2 Patient.245 Patient.180 PDX.155.F2 Patient.155 Patient.232 A 0.4 0.6 0.8 1 Pearson correlation B C

Patient1.55 Patient.232 Patient.180 Patient.247 Patient.228 Patient.237 Patient.245

Patient.247 Patient.228 Patient.237 Patient.245 Patient.180 Patient.232 Patient.155 PDX.155.F2 PDX.180.F2 PDX.247.F2 PDX.228.F2 PDX.245.F2 PDX.232.F2 PDX.237.F2 PDX.232.F2 PDX.245.F2 PDX.228.F2 PDX.237.F2 PDX.180.F2 PDX.247.F2 PDX.155.F2 0.4 0.6 0.8 1 Pearson correlation 0.4 0.6 0.8 1 Pearson correlation

Figure 4 | CNA based clustering. Hierarchical clustering of thresholded CNAs of all genes from GISTIC

analysis. Pearson correlation is indicated for all tumors (A), patient tumors (B) and PDX F2 tumors (C). mutation, and PDX 237 for which tumor

growth remained similar in F1 and F2, OCCC PDX engraftment latency is generally shorter in F2 compared to F1. This is in agreement with previous work on ovarian cancer PDX models (17).

In contrast to previous studies, we performed a full characterization of OCCC PDX models including immunohistochemistry, CNA and mutation analysis. In our PDX panel, mutations in several genes known

to be frequently mutated in OCCC, were represented in the PDX models. However, the genes with the highest mutation incidence in OCCC patients,

ARID1A (40-57%) and PIK3CA

(30-40%) respectively, were each found in only one model (3-5). A TP53 mutation was found in one model (14%) in line with the mutation frequency found in OCCC patients (11%). The low ARID1A mutation frequency in our PDX models (14%) can be explained by the relatively

(13)

CHAPTER 6

6

128 4p14 7p21.2 7p14.3 8p23.1 8q13.1 11q22.3 13q21.2 16p13.3 2q14.2 GLI2 WDR19 MEOX2 AQP1 SGK223 RRS1 EXPH5 PCDH20 BAIAP3 0.25 10í₁₀í ₁₀í ₁₀í ₁₀í 0.27 0.1 0.2 0.8 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18 20 21 22 0.25 10í₁₀í ₁₀í ₁₀í ₁₀í 0.23 0.1 0.8 2q14.2 7p21.2 8p23.1 11q22.3 16p13.3 7p14.3 SGK223 GLI2 MEOX2 AQP1 EXPH5 BAIAP3 17q21.33 EPN3 SPATA20 0.25 10í ₁₀í 0.1 0.8 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18 20 21 22 13q13.1 13q14.2 13q14.3 13q31.2 PDS5B LPAR6 hsa-mir-92a-1 hsa-mir-622 SLITRK5 LINC00410 MIR17 MIR18A MIR19A MIR19B1 MIR20A MIR92A1 MIR17HG MIR4500HG MIR622 MIR4500 hsa-mir-1297 PRR20A PRR20B PRR20C PRR20D PRR20E MIR1297 0.25 10í ₁₀í ₁₀í 0.64 0.4 0.8 1q42.13 4p16.3 11p15.1 18q21.2 NUCB2 POLI SNORA37 ZNF678 ZNF847P hsa-mir-571 Primary tumor PDX.F2 Primary tumor PDX.F2 Gscore Chr FDR Gscore Chr FDR Amplification Deletion 4q35.1 SURWHLQFRGLQJJHQHV QRQFRGLQJJHQHV CCDC110 4q21.23 )LJXUH_*,67,&SORWRIDPSOL¿HGDQGGHOHWHGJHQHV*,67,&SORWZLWKVLJQL¿FDQWO\DPSOL¿HGJHQHV LQUHGDQGVLJQL¿FDQWO\GHOHWHGJHQHVLQEOXH$PSOL¿HGJHQHVZLWK)'5DQGGHOHWHGJHQHVZLWK FDR<0.1 are indicated. The FDR 0.25 threshold, indicated by the green line, and G-score are shown along the horizontal axis.

(14)

6

129

Huntington's_disease

CYC1 POLR2F SDHA NDUFB10 NDUFA3 CREBBP RELB LILRB1 LILRA1 FOSB NCF4 _LILRA5 LILRB5 CYBA LILRA3 MRPS12 RPL8 RPS2 POLR2C COX6B1 COX6A2 BBC3 BAX POLR2I LILRB3 LILRA4 LILRB4 TYROBP MRPL20 MRPL10 MRPL14 RPL18 Osteoclast differentiation LILRA2 LILRB2 OSCAR RPL27 RPS9 RPL28 MRPL28 MRPS18A RPS11 Ribosome LILRA6 RPL13A RPL3L RPS21 RPS16 NR4A1 SH2B2 GNB1 GNG13 GNG8 PHKG2 MAP3K14 CACNA1H CACNA1G MAPK8IP3 _PPP5D1 CACNA1S MAPK1 GYS1 FGF21 PDPK1 TSC2 G6PC _MLX HSP90AB1 EIF4B MLST8 LAMA5 ITGB7 NR1H2 IL2RB PI3K-Akt_signaling_pathway CD19

IL4R SLC27A5 Insulin_resistanceSTAT3 CALM3 CRKL VEGFA Choline_metabolism in_cancer PRKCZ PRKAR1B Insulin_signaling pathway PDGFA MAPK3 CBLC ITGA2B AKT2

CACNG8MAPK_signalingpathway

PPP5C TAOK2 RASGRP4 Arrhythmogenic_right ventricular_cardiomyopathy CTNNA3 CACNG2 CACNG6 CACNG7 MAP4K1 NTF4 LRRC4C Axon_guidance CTU1 CTU2 Sulfur_relay system MPST TST MBOAT7 Glycerophospholipid metabolism CDIPT PLA2G6 CALML6 VASPPRKD2 ITGAL_LAT

SKAP1 Rap1_signaling pathway RAC2 PRKCG ITGAM RRAS GP6 ARHGAP35 Platelet activation PTGIR GP9 GP1BB PLD3 PLA2G4C SPHK2 VEGF_signaling pathway SLC17A7 NAPA AP2S1 AP2A1 ATP6V0C ATP6V0A1 STX1B CLTCL1 Synaptic_vesicle cycle GALNS NAGLU AP1S1 NAGPA COX6B2 GGA1 GNPTG HAP1 NAPSA CLN3 Lysosome Amplified Genes

Amplified Pathways Deleted GenesDeleted Pathways

Pathway Enriched Genes Pathway Size P-Value FDR Alteration Type

VEGF signaling pathway 33 61 1.256E-04 1.672E-03 Amplification

Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Amplification Deletion Deletion

Ribosome 60 136 3.218E-04 4.279E-03

MAPK signaling pathway 100 255 6.288E-04 8.345E-03

Rap1 signaling pathway 81 210 3.676E-03 0.048

Insulin signaling pathway 56 138 5.072E-03 0.065

Osteoclast differentiation 53 131 6.933E-03 0.089

Insulin resistance 45 108 7.282E-03 0.093

Synaptic vesicle cycle 29 63 7.717E-03 0.098

Sulfur relay system 8 10 9.329E-03 0.117

Glycerophospholipid metabolism 40 95 9.609E-03 0.121

Choline metabolism in cancer 42 101 0.010 0.126

Lysosome 48 121 0.015 0.187

Platelet activation 51 130 0.016 0.189

Huntington's disease 71 192 0.019 0.229

PI3K-Akt signaling pathway 120 345 0.020 0.237

Axon guidance 16 127 8.582E-06 1.083E-04

KEGGID hsa04370 hsa03010 hsa04010 hsa04015 hsa04910 hsa04380 hsa04931 hsa04721 hsa04122 hsa00564 hsa05231 hsa04142 hsa04611 hsa05016 hsa04151 hsa04360

hsa05412 Arrhythmogenic right ventricular cardiomyopathy 8 71 6.449E-03 0.078

A

B

Figure 6 | Pathway enrichment.

(A)3DWKZD\VIURP'$9,'SDWKZD\DQDO\VLVRIFRDPSOL¿HGDQGFRGH-leted genes in four or more PDX models with a p<0.05 and FDR<0.25 are indicated. (B) Illustration of WKHVHSDWKZD\VDQGUHODWHGFRDPSOL¿HGDQGFRGHOHWHGJHQHVLQDOOVHYHQ3';PRGHOV7KHQRGHVL]H of genes were based on the number of pathways they are linked to.

(15)

CHAPTER 6

6

130

low ARID1A mutation frequency

(28%) found in Dutch OCCC patients (7). However, we cannot exclude the possibility that the engraftment success rate of OCCC patient tumors harboring these ARID1A mutations is generally low.

In the present study, CNA based clustering demonstrated that CNAs were retained in OCCC PDX models. Loss of human stromal tissue in F1 and F2 could be accountable for the cases where CNAs were more abundant in F2 tumors compared to paired patient tumors. Overall, CNAs in OCCC patient tumors were retained in matched PDXs, in line with previous studies ovarian cancer PDX studies (17, 18, 20). The TP53 and

BRCA1 mutant model 232 was a notable

outlier with less CNAs in the matched PDX F2 tumor, possibly a genomically stable clone was selected during propagation. Pathway enrichment analysis using DPSOL¿HGJHQHVXQFRYHUHGSDWKZD\V in all seven PDX models, suggesting enhanced activity of these cancer-related pathways. Additionally, several well-known cancer genes such as AKT2,

MAPK1 (ERK2) and MAPK3 (ERK1)

ZHUH QRGHV DPSOL¿HG LQ VHYHUDO WXPRU growth related signaling pathways.

Extensive passaging of PDXs can result in the accumulation of undetected CNAs that are caused by positive selection of preexisting clones. Moreover, it has been described that CNAs acquired during passaging in PDX models are distinct from those acquired along tumor progression in patients (14). However, up to F2 we only found <3% non-concurrent genes with CNA, while approximately 70% was concurrent indicating very few clonal selection. Though, for drug testing, it is recommended to use PDX models at low-passages to mimic patient tumors most closely. Targeted therapies have been tested in small numbers of OCCC 3'; PRGHOV (*)5 DQG F0HW VSHFL¿F inhibitors, and the multi targeting receptor

tyrosine kinase inhibitor sunitinib, have been evaluated in subrenal implanted 3';V P725& LQKLELWLRQ and BET bromodomain inhibition were tested in subcutaneous implanted PDXs (PIK3CA mutant versus type and ARID1A mutant versus wild-type respectively) and their responses correlated with PDX mutation status DQG LQ YLWUR HႈFDF\ 7KHVH results demonstrate that novel therapy strategies can thus be evaluated using a panel of PDX models that resembles the genetic heterogeneity of OCCC. Clinical studies using biomarkers established in 2&&& 3'; VWXGLHV KDYH WR GH¿QH WKH value of these models to guide treatment strategies in OCCC patients.

In conclusion, we describe seven OCCC PDX models that generally retain patient tumor characteristics in terms of genomic alterations and histology. Accordingly, these PDXs can serve as in vivo models to guide future translational research in OCCC.

METHODS

Establishment and propagation of PDX models

All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Groningen (Groningen, The Netherlands) and carried out in accordance with the approved guideline “code of practice: animal experiments in cancer research” (Netherlands Inspectorate for Health 3URWHFWLRQ&RPPRGLWLHVDQG9HWHULQDU\ Public Health, 1999). Before surgery, all patients from whom tumor samples were obtained gave written informed consent. Surgical tissues were implanted and propagated to generation 1 (F1) and JHQHUDWLRQ ) LQ WKH ÀDQNV RI WR ZHHN ROG 12'&%3UNGFVFLG NCrHsd (NSG) mice (internal breed, Central Animal Facility, University Medical Center Groningen) according to previously described methods (17).

(16)

6

131

genotyping of the models was performed with HumanOmniExpressExome-8BeadChip (Illumina, USA). Raw data

was pre-processed as described earlier and analyzed using Nexus Copy Number software (7).

3DLUEDVHG GLႇHUHQFHV RI &1$ OHYHO ZHUH TXDQWL¿HG DQG LOOXVWUDWHG E\ WKH 5 SDFNDJH JJSORW KWWS PRGHUQJUDSKLFVSEZRUNVFRPI ggplot2-Book09hWickham.pdf). The average absolute Gscore was calculated per sample with the formula below:

where GscoreN represented the negative Gscores and GscoreP represented the positive Gscores for sample i and k was the total number of genes (23109).

Clustering of PDX models

7R GHWHUPLQH VLJQL¿FDQW &1$V WKH segmented copy number values were analyzed with GISTIC, version 2.6.2 on the Genepattern server from the Broad Institute, USA) using standard settings with Hg19 as reference genome (7). CNAs data of all patient tumors and PDX F2 tumors were subjected to hierarchical clustering based on the gene list thresholded copy numbers (-2 = deep deletion, -1 = deletion, 0 = neutral, 1 = FRS\ QXPEHU JDLQ DPSOL¿FDWLRQ using the R package gplots, version KWWSV&5$15SURMHFWRUJ package=gplots).

Pathway enrichment

All genes in the genome were categorized into eight classes based on their CNA status between paired patient WXPRU DQG ) WXPRU 7KH ¿UVW WKUHH FODVVHV DPSOL¿FDWLRQ GHOHWLRQ VWDEOH consisted of the genes with same CNA status between paired models were formed into a main class “concurrent”. 7KHRWKHUFODVVHVJDLQHGDPSOL¿FDWLRQ JDLQHG GHOHWLRQ ORVW DPSOL¿FDWLRQ ORVW 7XPRUJURZWKZDVTXDQWL¿HGZHHNO\E\

caliper measurements according to the formula (width2_{[ OHQJWK )RU IXUWKHU}

analysis the tumors were resected and VQDSIUR]HQDW&DVZHOODVSDUDႈQ HPEHGGHG 'LႇHUHQFH LQ WXPRU JURZWK between F1 and F2 tumors was assessed using a two-sided Fisher’s exact test.

Immunohistochemical analysis

3'; WXPRUV VOLFHV ȝP WKLFN ZHUH FXW IURP SDUDႈQHPEHGGHG WLVVXH using a microtome and placed on 3-aminopropyltriethoxysilane-coated glass slides. Heat-induced antigen retrieval was performed in 10 mM citrate EXႇHUXVLQJD:URWDU\PLFURZDYH Endogenous peroxidase was blocked by 30 minutes incubation with 0.3% H₂O₂ in PBS for all stainings except H&E. (QGRJHQRXV DYLGLQELRWLQ DFWLYLW\ ZDV blocked using a commercially available EORFNLQJ NLW 9HFWRU /DERUDWRULHV 86$ for all stainings except H&E. Slides were incubated with primary antibodies detecting human ARID1A (Sigma HPA005456), Ki67 (DAKO M7240, USA) 1:350 dilution and TP53 (DAKO M7001, USA). Staining was visualized after incubation with biotinylated or peroxidase-bound secondary antibodies 'DNR 86$ XVLQJ VWUHSWDYLGLQELRWLQ horseradish peroxidase complex (Dako, USA) and 3,3’-diaminobenzidine (Sigma-Aldrich, USA). Hematoxylin counterstaining was applied routinely, and hematoxylin & eosin (H&E) staining was used to analyze tissue viability and morphology. Photographs were acquired by digitalized scanning of slides using the NanoZoomer 2.0-HT multi-slide scanner (Hamamatsu, Japan).

PDX genotyping

The models were sequenced for 40 genes (supplementary table 1), including genes with a high mutation frequency in OCCC, ARID1A and other cancer-related genes using Haloplex custom kit (Agilent technologies®, USA) (7). Additional SNP

(17)

CHAPTER 6

6

132

deletion, inverted alterations) contained genes that lost, gained or inverted their DOWHUDWLRQVIURPDPSOL¿FDWLRQWRGHOHWLRQ or vice versa) during engraftment and comprised the other main class “non-concurrent”. To circumvent overlap, only the genes that retained the same status or changed consistently in more than four paired patient tumor and PDX models were taken into account.

7KH JHQHV FRQFXUUHQWO\ DPSOL¿HG (7359) and deleted (911) in at least

four PDX models were used for '$9,' SDWKZD\ HQULFKPHQW YHUVLRQ VHSDUDWHO\ 7KH VLJQL¿FDQFH WKUHVKROGZDVVHWDVSYDOXHDQG IDOVHGLVFRYHU\ UDWH )'5 1H[W the enrichment result was illustrated by Cytoscape 3.6.1 (30), where the co-DPSOL¿HG DQG FRGHOHWHG JHQHV ZHUH SUHVHQWHG WRJHWKHU ZLWK WKH VLJQL¿FDQW pathways that they are located in.

REFERENCES

1. Oliver KE, Brady WE, Birrer M, Gershenson DM, Fleming G, Copeland LJ, et al. An evaluation of pro-gression free survival and overall survival of ovarian cancer patients with clear cell carcinoma versus VHURXV FDUFLQRPD WUHDWHG ZLWK SODWLQXP WKHUDS\$Q 15* 2QFRORJ\*\QHFRORJLF 2QFRORJ\ *URXS experience. Gynecol Oncol. 2017 Nov;147(2):243-9.

2. Cancer Genome Atlas Research Network. Integrated genomic analyses of ovarian carcinoma. Na-ture. 2011 Jun 29;474(7353):609-15.

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

4. 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. .XR.70DR7/-RQHV69HUDV($\KDQ$:DQJ7/HWDO)UHTXHQWDFWLYDWLQJPXWDWLRQVRI3,.&$

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

6. 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 7. Caumanns JJ, Berns K, Wisman GBA, Fehrmann RSN, Tomar T, Klip H, et al. Integrative kinome

SUR¿OLQJ LGHQWL¿HV P725& LQKLELWLRQ DV WUHDWPHQW VWUDWHJ\ LQ RYDULDQ FOHDU FHOO FDUFLQRPD &OLQ Cancer Res. 2018 Apr 23.

6DVDQR70DEXFKL6.XURGD+.DZDQR00DWVXPRWR<7DNDKDVKL5HWDO3UHFOLQLFDO(ႈFDF\IRU AKT Targeting in Clear Cell Carcinoma of the Ovary. Mol Cancer Res. 2014 Dec 17.

9. Miller RE, Brough R, Bajrami I, Williamson CT, McDade S, Campbell J, et al. Synthetic Lethal Targeting of ARID1A-Mutant Ovarian Clear Cell Tumors with Dasatinib. Mol Cancer Ther. 2016 Jul;15(7):1472-84.

10. 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.

11. Chandler RL, Damrauer JS, Raab JR, Schisler JC, Wilkerson MD, Didion JP, et al. Coexistent ARI- '$3,.&$PXWDWLRQVSURPRWHRYDULDQFOHDUFHOOWXPRULJHQHVLVWKURXJKSURWXPRULJHQLFLQÀDPPD-tory cytokine signalling. Nat Commun. 2015 Jan 27;6:6118.

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

+LGDOJR0$PDQW)%LDQNLQ$9%XGLQVND(%\UQH$7&DOGDV&HWDO3DWLHQWGHULYHG[HQRJUDIW models: an emerging platform for translational cancer research. Cancer Discov. 2014 Sep;4(9):998-1013.

(18)

6

133

PRXVHVSHFL¿FWXPRUHYROXWLRQ1DW*HQHW1RY

5LFFL ) %L]]DUR ) &HVFD 0 *XႇDQWL ) *DQ]LQHOOL 0 'HFLR$ HW DO 3DWLHQWGHULYHG RYDULDQ WX-mor xenografts recapitulate human clinicopathology and genetic alterations. Cancer Res. 2014 Dec 1;74(23):6980-90.

16. Colombo PE, du Manoir S, Orsett B, Bras-Goncalves R, Lambros MB, MacKay A, et al. Ovarian carci-noma patient derived xenografts reproduce their tumor of origin and preserve an oligoclonal structure. Oncotarget. 2015 Sep 29;6(29):28327-40.

17. 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.

18. Weroha SJ, Becker MA, Enderica-Gonzalez S, Harrington SC, Oberg AL, Maurer MJ, et al. Tumor-grafts as in vivo surrogates for women with ovarian cancer. Clin Cancer Res. 2014 Mar 1;20(5):1288-97.

7RSS0'+DUWOH\/&RRN0+HRQJ9%RHKP(0F6KDQH/HWDO0ROHFXODUFRUUHODWHVRISODWLQXP response in human high-grade serous ovarian cancer patient-derived xenografts. Mol Oncol. 2014 May;8(3):656-68.

/LX-)3DODNXUWKL6=HQJ4=KRX6,YDQRYD(+XDQJ:HWDO(VWDEOLVKPHQWRI3DWLHQW'HULYHG7X-mor Xenograft Models of Epithelial Ovarian Cancer for Preclinical Evaluation of Novel Therapeutics. Clin Cancer Res. 2017 Mar 1;23(5):1263-73.

21. Rudd ML, Price JC, Fogoros S, Godwin AK, Sgroi DC, Merino MJ, et al. A unique spectrum of somatic PIK3CA (p110alpha) mutations within primary endometrial carcinomas. Clin Cancer Res. 2011 Mar 15;17(6):1331-40.

22. Olive KP, Tuveson DA, Ruhe ZC, Yin B, Willis NA, Bronson RT, et al. Mutant p53 gain of function in two mouse models of Li-Fraumeni syndrome. Cell. 2004 Dec 17;119(6):847-60.

23. 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.

24. 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.

25. 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.

6WDQ\039DWKLSDGLHNDO92]EXQ/6WRQH5/0RN6&;XH+HWDO,GHQWL¿FDWLRQRIQRYHOWKHUDSHX-tic targets in microdissected clear cell ovarian cancers. PLoS One. 2011;6(7):e21121.

27. Kim HJ, Yoon A, Ryu JY, Cho YJ, Choi JJ, Song SY, et al. c-MET as a Potential Therapeutic Target in Ovarian Clear Cell Carcinoma. Sci Rep. 2016 Dec 5;6:38502.

28. Berns K, Caumanns JJ, Hijmans EM, Gennissen AMC, Severson TM, Evers B, et al. ARID1A mutation sensitizes most ovarian clear cell carcinomas to BET inhibitors. Oncogene. 2018 May 15.

29. Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the compre-hensive functional analysis of large gene lists. Nucleic Acids Res. 2009 Jan;37(1):1-13.

30. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environ-ment for integrated models of biomolecular interaction networks. Genome Res. 2003 Nov;13(11):2498-504.

(19)

CHAPTER 6

6

134

SUPPLEMENTARY DATA

Gene ID ENSMBL Gene ID ENSMBL 1 ARID1A ENST00000324856 21 BRD2 ENST00000374825 2 SMARCA4 ENST00000590574 22 PKN1 ENST00000242783 3 PIK3CA ENST00000263967 23 PRKCQ ENST00000263125 4 PIK3R1 ENST00000521381 24 PRPF4B ENST00000337659 5 PTEN ENST00000371953 25 MYO3B ENST00000408978 6 AKT1 ENST00000555528 26 BRCA1 ENST00000357654 7 KRAS ENST00000311936 27 BRCA2 ENST00000544455 8 ERBB3 ENST00000267101 28 TRRAP ENST00000359863 9 EGFR ENST00000275493 29 NRAS ENST00000369535 10 TP53 ENST00000269305 30 BRAF ENST00000288602 11 ATM ENST00000278616 31 ERBB2 ENST00000269571 12 ATR ENST00000350721 32 EIF2AK4 ENST00000263791 13 CHEK2 ENST00000382580 33 LRRK2 ENST00000298910 14 SGK223 ENST00000622241 34 WNK2 ENST00000297954 15 FES ENST00000328850 35 TAF1 ENST00000423759 16 MYO3A ENST00000265944 36 ERBB4 ENST00000342788 17 PRKDC ENST00000314191 37 CTNNB1 ENST00000396183 18 MAST4 ENST00000403625 38 SMAD4 ENST00000398417 19 FBXW7 ENST00000281708 39 STK11 ENST00000326873 20 CAMK2B ENST00000395749 40 MET ENST00000318493

(20)

(21)