University of Groningen Barrett's Esophagus and Esophageal Adenocarcinoma

University of Groningen

Barrett's esophagus and esophageal adenocarcinoma: transcription factors and biomarkers Pavlov, Kirill

DOI: 10.1016/j.dld.2014.09.014

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: 2015

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): Pavlov, K. (2015). Barrett's esophagus and esophageal adenocarcinoma: transcription factors and biomarkers. University of Groningen. https://doi.org/10.1016/j.dld.2014.09.014

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).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.

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.

Download date: 07-10-2021 Barrett’s esophagus and esophageal adenocarcinoma: transcription factors and biomarkers

Kirill Viktorovich Pavlov

Proefschrift_Kirill.indb 1 10-05-15 23:47 Pavlov, K. V. Barrett’s esophagus and esophageal adenocarcinoma: transcription factors and biomarkers

Thesis, University of Gronigen, Groningen, the Netherlands

Copyright 2015 Kirill Pavlov, the Netherlands All rights reserved. No part of this thesis may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronically, mechanically, or by photocopying, recording, or otherwise, without prior permission of the author

The research presented in this thesis was financially supported by: the Junior Scientific Masterclass (JSM), the Van der Meer-Boerema foundation and the Jan Kornelis de Cock foundation.

Printing of this thesis was financially supported by: The University of Groningen, Groningen, the Netherlands, the Graduate School of Medical Sciences (GSMS), Stichting Werkgroep Interne Oncologie and Greiner Bio-One B.V.

ISBN: 978-90-367-7916-6

Layout and design: Niels Hoekstra Printed by: Ipskamp Drukkers B.V., Enschede, the Netherlands

Proefschrift_Kirill.indb 2 10-05-15 23:47 Barrett’s esophagus and esophageal adenocarcinoma: transcription factors and biomarkers

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen op gezag van de rector magnificus prof. dr. E. Sterken en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op woensdag 10 juni 2015 om 11.00 uur

door

Kirill Viktorovich Pavlov

geboren op 8 juli 1986 te Moskou, Rusland

Proefschrift_Kirill.indb 3 10-05-15 23:47 Promotores Prof. dr. J.H. Kleibeuker Prof. dr. F.A.E. Kruyt Prof. dr. J.H.M. van den Berg

Copromotores Dr. J. Meijer Dr. F.T.M. Peters

Beoordelingscommissie Prof. dr. K.K. Krishnadath Prof. dr. H. Hollema Prof. dr. S. de Jong

Proefschrift_Kirill.indb 4 10-05-15 23:47 Paranimfen Arne R.M. van der Bilt Nicolette G. Alkema

Proefschrift_Kirill.indb 5 10-05-15 23:47 Proefschrift_Kirill.indb 6 10-05-15 23:47 Table of Contents

Preface 9

Chapter 1 Embryological signaling pathways in Barrett’s metaplasia 17 development and malignant transformation; mechanisms and therapeutic opportunities

Chapter 2 New models of neoplastic progression in Barrett’s oesophagus 43

Chapter 3 Development and Characterization of an Organotypic Model of 59 Barrett’s Esophagus

Chapter 4 Transcription factors associated with the development of Barrett’s 77 esophagus; in vivo expression and modulation by Retinoic Acid

Chapter 5 GATA6 expression in Barrett’s oesophagus and oesophageal 93 adenocarcinoma

Chapter 6 Circulating miRNA’s in Barrett’s esophagus, high-grade dysplasia 111 and esophageal adenocarcinoma

Chapter 7 Loss of CD44 and SOX2 expression is correlated with a poor 127 prognosis in esophageal adenocarcinoma patients

Chapter 8 CD44, SHH and SOX2 as novel biomarkers in esophageal cancer 147 patients treated with neoadjuvant chemoradiotherapy

Summary, Discussion and Future perspectives 173

Nederlandstalige samenvatting (Summary in Dutch) 185

Dankwoord (Acknowledgements) 193

Proefschrift_Kirill.indb 7 10-05-15 23:47 Proefschrift_Kirill.indb 8 10-05-15 23:47 Preface

Proefschrift_Kirill.indb 9 10-05-15 23:47 Proefschrift_Kirill.indb 10 10-05-15 23:47 Preface

Esophageal cancer has a poor survival rate and its incidence is increasing. The 5-year survival of patients eligible for curative treatment is 50% (1), but late diagnosis in many patients decreases the overall 5-year survival to a mere 20% (2). Esophageal cancer is divided into two main histological categories: esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC). Despite similar treatment options for both subtypes, EAC and ESCC differ considerably in etiology and tumor biology. ESCC occurs mostly in the proximal and middle part of the esophagus and is associated with smoking and alcohol consumption (3,4). EAC occurs preferentially in the distal part of the esophagus and persistent gastro-esophageal reflux and obesity (in particular abdominal obesity) are the main risk factors for EAC (5-7). Historically ESCC was more prevalent, but in the last decades the incidence of EAC showed a dramatic surge, especially in North- West Europe and North America, surpassing ESCC in incidence (8). The reasons for this increase are poorly understood, highlighting the need for a better understanding of EAC development. The study of EAC carcinogenesis is greatly aided by the fact that EAC has a well-defined and easily accessible precursor lesion, Barrett’s metaplasia of the esophagus (BE). BE is characterized by the presence of columnar epithelium in the distal end of the esophagus. Like EAC, persistent gastro-esophageal reflux disease (GERD) and obesity are the main risk factors for the development of BE (9). The prevalence of BE in the general population is estimated around 1,5% (10,11); since the prevalence of GERD is around 40% in the general population (11,12), only a fraction of patients with GERD develop BE. The clinical relevance of BE detection lies in its potential for malignant transformation through a metaplasia-dysplasia-adenocarcinoma sequence. Despite the fact that presence of BE is the most important risk factor for EAC development, a meta-analysis of BE surveillance studies suggests that BE patients only have a 0,3% risk of malignant transformation per patient year, but this increases to around 13% once dysplasia is present (9,13,14). While the absolute risk of BE malignant transformation is low, it is currently impossible to stratify BE patients according to the risk of malignant progression, and therefore all patients with BE are offered periodic endoscopic surveillance. The current management of BE and EAC has been made possible by considerable progress that has been made in the understanding and clinical management of BE and EAC since the surgeon Norman Barrett first described BE in 1950 (1). The last decades acid-suppressant medical therapy has been introduced to control GERD. In addition, novel diagnostic and therapeutic endoscopic possibilities were developed, as well as new treatment regimens for EAC such as neoadjuvant chemoradiotherapy (nCRT). Despite these encouraging developments, several important challenges remain in the management of BE and EAC.

Proefschrift_Kirill.indb 11 10-05-15 23:47 First, the current understanding of the molecular mechanisms underlying BE development is limited, and this hampers the development of effective pharmacological treatment modalities. Second, the current surveillance protocols are of limited efficacy, since the majority of EAC patients are not diagnosed at the stage of BE/early EAC, but instead present with advanced disease. Third, despite optimal treatment consisting of nCRT and surgery, the 5-year survival of EAC patients treated with curative intent is disappointing, and better insight into EAC biology is required to improve this. The studies presented in this thesis aim to address each of these three challenges and are discussed in more detail in the following section.

Scope of the thesis

Understanding Barrett’s esophagus development Aberrant activity of embryological signaling pathways has been implicated in the development and malignant progression of BE, but the relevant research data are fragmented across different disciplines of medicine and developmental biology. Therefore, in chapter 1 we provide a synthesis of the embryological signaling pathways and transcription factors involved in BE development and its malignant transformation, and offer a perspective on new therapeutic targets. Testing novel therapeutic targets requires representative pre-clinical models. Several approaches have been used to study BE and EAC including cell lines, animal and in silico models. Chapter 2 provides a discussion of the different models and the strengths and limitations of each, and offers suggestions for further development of novel pre-clinical models of BE and EAC. One of the relevant embryological signaling pathways involved in BE development is the retinoic acid signaling pathway. Aberrant re-activation of the retinoic acid signaling pathway has been proposed to induce columnar differentiation of squamous epithelium (15), but the mechanisms through which retinoic acid exerts its effect on differentiation of squamous esophageal epithelium are poorly understood. In chapter 3 we studied the effect of retinoic acid on esophageal tissue morphology and differentiation in an organotypic in vitro model. The aim of chapter 4 was to study the expression of a panel of squamous and columnar transcription factors in normal squamous epithelium and BE in vivo and examine the effect of retinoic acid treatment on the expression of squamous and columnar transcription factors in vitro. The transcription factor GATA6 is a known target gene of the retinoic acid signaling pathway. GATA6 expression contributes to the development of intestinal epithelium (16,17), but the expression pattern of GATA6 during BE development and malignant transformation is unknown. GATA6 gene amplification was associated with a poor survival outcome in EAC patients (18), and the aim of chapter 5 was to study its expression in

Proefschrift_Kirill.indb 12 10-05-15 23:47 different stages BE development and malignant transformation and determine the prognostic value of GATA6 protein expression in EAC.

Improving detection of Barrett’s esophagus malignant transformation The majority of patients that present with EAC were not previously included in BE surveillance programs, suggesting that most BE cases in the general population are not diagnosed (19). Early detection of premalignant esophageal lesions could drastically improve overall survival, but the absence of biomarkers indicative of metaplastic and/ or malignant transformation in the esophagus hampers screening of populations at risk to develop BE and EAC. A surveillance strategy based on a blood-based biomarker seems an attractive approach, but so far no reliable circulating biomarkers have been identified. MiRNA’s are small non-coding regulatory RNA species that are characterized by high tissue specificity and stability in serum. Tumor-specific serum miRNA profiles have been identified in various malignancies, including colon, lung, prostate and breast cancer (reviewed in (20)). In chapter 6 we set out to investigate the potential of circulating serum miRNA’s as biomarkers of BE, dysplasia and EAC by comparing serum miRNA expression profiles from patients with BE, dysplasia and EAC with that of control patients.

Correlating esophageal adenocarcinoma survival with biomarker expression Subpopulations of tumor cells with characteristics traditionally associated with stem cells (known as “Cancer Stem Cells”, CSC), have an increased resistance to DNA damaging agents and increased metastatic potential (reviewed in (21)). Several proteins have been proposed as markers of CSC in various solid tumors, but little is known regarding the prognostic value of stemness-associated proteins in EAC. The aim of chapter 7 was to examine expression and prognostic value of a panel of potential CSC biomarkers in EAC patients undergoing primary surgical resection. Neoadjuvant chemoradiotherapy (nCRT) reduces tumor volume and increases the chance of a complete tumor resection (1,22,23). Despite its success, further optimization of nCRT in EAC is hampered by the lack of predictive and prognostic biomarkers. Previous in vitro research suggested that increased activity of the Hedgehog signaling pathway was associated with resistance to chemoradiotherapy (24). In chapter 8 we studied the predictive and prognostic value of Sonic Hedgehog, a ligand of the Hedgehog pathway and the prognostic biomarkers identified in chapter 7 in a cohort of patients treated with both nCRT and surgery using pretreatment biopsies and post-treatment resection specimens. Finally, chapter 9 provides a discussion of the data described in this thesis, and explores future perspectives regarding the opportunities to further improve BE and EAC management through a better understanding of the underlying biology.

Proefschrift_Kirill.indb 13 10-05-15 23:47 References

(1) van Hagen P, Hulshof MC, van Lanschot JJ, Steyerberg EW, van Berge Henegouwen MI, Wijnhoven BP, et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012 May 31;366(22):2074-2084. (2) Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin 2010 Sep-Oct;60(5):277-300. (3) Lubin JH, Cook MB, Pandeya N, Vaughan TL, Abnet CC, Giffen C, et al. The importance of exposure rate on odds ratios by cigarette smoking and alcohol consumption for esophageal adenocarcinoma and squamous cell carcinoma in the Barrett’s Esophagus and Esophageal Adenocarcinoma Consortium. Cancer Epidemiol 2012 Jun;36(3):306-316. (4) Islami F, Fedirko V, Tramacere I, Bagnardi V, Jenab M, Scotti L, et al. Alcohol drinking and esophageal squamous cell carcinoma with focus on light-drinkers and never-smokers: a systematic review and meta-analysis. Int J Cancer 2011 Nov 15;129(10):2473-2484. (5) Turati F, Tramacere I, La Vecchia C, Negri E. A meta-analysis of body mass index and esophageal and gastric cardia adenocarcinoma. Ann Oncol 2013 Mar;24(3):609-617. (6) Navarro Silvera SA, Mayne ST, Gammon MD, Vaughan TL, Chow WH, Dubin JA, et al. Diet and lifestyle factors and risk of subtypes of esophageal and gastric cancers: classification tree analysis. Ann Epidemiol 2014 Jan;24(1):50-57. (7) O’Doherty MG, Freedman ND, Hollenbeck AR, Schatzkin A, Abnet CC. A prospective cohort study of obesity and risk of oesophageal and gastric adenocarcinoma in the NIH-AARP Diet and Health Study. Gut 2012 Sep;61(9):1261-1268. (8) Castro C, Bosetti C, Malvezzi M, Bertuccio P, Levi F, Negri E, et al. Patterns and trends in esophageal cancer mortality and incidence in Europe (1980-2011) and predictions to 2015. Ann Oncol 2014 Jan;25(1):283-290. (9) Fitzgerald RC, di Pietro M, Ragunath K, Ang Y, Kang JY, Watson P, et al. British Society of Gastroenterology guidelines on the diagnosis and management of Barrett’s oesophagus. Gut 2014 Jan;63(1):7-42. (10) Ronkainen J, Aro P, Storskrubb T, Johansson SE, Lind T, Bolling-Sternevald E, et al. Prevalence of Barrett’s esophagus in the general population: an endoscopic study. Gastroenterology 2005 Dec;129(6):1825-1831. (11) Zagari RM, Fuccio L, Wallander MA, Johansson S, Fiocca R, Casanova S, et al. Gastro-oesophageal reflux symptoms, oesophagitis and Barrett’s oesophagus in the general population: the Loiano- Monghidoro study. Gut 2008 Oct;57(10):1354-1359. (12) Rosaida MS, Goh KL. Gastro-oesophageal reflux disease, reflux oesophagitis and non-erosive reflux disease in a multiracial Asian population: a prospective, endoscopy based study. Eur J Gastroenterol Hepatol 2004 May;16(5):495-501. (13) Desai TK, Krishnan K, Samala N, Singh J, Cluley J, Perla S, et al. The incidence of oesophageal adenocarcinoma in non-dysplastic Barrett’s oesophagus: a meta-analysis. Gut 2012 Jul;61(7):970- 976. (14) Curvers WL, ten Kate FJ, Krishnadath KK, Visser M, Elzer B, Baak LC, et al. Low-grade dysplasia in Barrett’s esophagus: overdiagnosed and underestimated. Am J Gastroenterol 2010 Jul;105(7):1523-1530. (15) Chang CL, Lao-Sirieix P, Save V, De La Cueva Mendez G, Laskey R, Fitzgerald RC. Retinoic acid- induced glandular differentiation of the oesophagus. Gut 2007 Jul;56(7):906-917.

Proefschrift_Kirill.indb 14 10-05-15 23:47 (16) Benahmed F, Gross I, Gaunt SJ, Beck F, Jehan F, Domon-Dell C, et al. Multiple regulatory regions control the complex expression pattern of the mouse Cdx2 homeobox gene. Gastroenterology 2008 Oct;135(4):1238-1247, 1247.e1-3. (17) Mauney JR, Ramachandran A, Yu RN, Daley GQ, Adam RM, Estrada CR. All-trans retinoic acid directs urothelial specification of murine embryonic stem cells via GATA4/6 signaling mechanisms. PLoS One 2010 Jul 13;5(7):e11513. (18) Lin L, Bass AJ, Lockwood WW, Wang Z, Silvers AL, Thomas DG, et al. Activation of GATA binding protein 6 (GATA6) sustains oncogenic lineage-survival in esophageal adenocarcinoma. Proc Natl Acad Sci U S A 2012 Mar 13;109(11):4251-4256. (19) Dulai GS, Guha S, Kahn KL, Gornbein J, Weinstein WM. Preoperative prevalence of Barrett’s esophagus in esophageal adenocarcinoma: a systematic review. Gastroenterology 2002 Jan;122(1):26-33. (20) Madhavan D, Cuk K, Burwinkel B, Yang R. Cancer diagnosis and prognosis decoded by blood- based circulating microRNA signatures. Front Genet 2013 Jun 21;4:116. (21) Visvader JE, Lindeman GJ. Cancer stem cells: current status and evolving complexities. Cell Stem Cell 2012 Jun 14;10(6):717-728. (22) Oppedijk V, van der Gaast A, van Lanschot JJ, van Hagen P, van Os R, van Rij CM, et al. Patterns of recurrence after surgery alone versus preoperative chemoradiotherapy and surgery in the CROSS trials. J Clin Oncol 2014 Feb 10;32(5):385-391. (23) Donahue JM, Nichols FC, Li Z, Schomas DA, Allen MS, Cassivi SD, et al. Complete pathologic response after neoadjuvant chemoradiotherapy for esophageal cancer is associated with enhanced survival. Ann Thorac Surg 2009 Feb;87(2):392-8; discussion 398-9. (24) Sims-Mourtada J, Izzo JG, Apisarnthanarax S, Wu TT, Malhotra U, Luthra R, et al. Hedgehog: an attribute to tumor regrowth after chemoradiotherapy and a target to improve radiation response. Clin Cancer Res 2006 Nov 1;12(21):6565-6572.

Proefschrift_Kirill.indb 15 10-05-15 23:47 Proefschrift_Kirill.indb 16 10-05-15 23:47 Chapter 1 Embryological signaling pathways in Barrett’s metaplasia development and malignant transformation; mechanisms and therapeutic opportunities Authors: Pavlov K.1, Meijer C.2, van den Berg A.3, Peters F.T.M.1, Kruyt F.A.E.2, Kleibeuker J.H.1

Affiliations: 1Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 3Department of Pathology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.

Crit Rev Oncol Hematol. 2014 Oct;92(1):25-37

Proefschrift_Kirill.indb 17 10-05-15 23:47 Abstract

Barrett’s metaplasia of the esophagus (BE) is the precursor lesion of esophageal adenocarcinoma (EAC), a deadly disease with a 5-year overall survival of less than 20%. The molecular mechanisms of BE development and its transformation to EAC are poorly understood and current surveillance and treatment strategies are of limited efficacy. Increasing evidence suggests that aberrant signaling through pathways active in the embryological development of the esophagus contributes to BE development and progression to EAC. We discuss the role that the Bone Morphogenetic Protein, Hedgehog, Wingless-Type MMTV Integration Site Family (WNT) and Retinoic Acid signaling pathways play during embryological development of the esophagus and their contribution to BE development and malignant transformation. Modulation of these pathways provides new therapeutic opportunities. By integrating findings in developmental biology with those from translational research and clinical trials, this review provides a platform for future studies aimed at improving current management of BE and EAC.

Proefschrift_Kirill.indb 18 10-05-15 23:47 Introduction

Barrett’s metaplasia of the esophagus (BE) is characterized by the presence of columnar epithelium in the distal part of the esophagus that is normally lined with squamous 1 epithelium. The causative factors underlying the development of BE are debated, but persistent gastroesophageal reflux disease (GERD) is the main risk factor. The reported percentages of BE in GERD patients vary between 5 and 25% (1-4). BE is the single known precursor lesion of esophageal adenocarcinoma (EAC) and can progress through a multistep process from metaplastic to low-grade, then high-grade dysplasia and eventually Esophageal Adenocarcinoma (EAC) (5). The management of BE and EAC is hampered by the limited effectiveness of surveillance strategies and of current therapeutic regimens in EAC treatment. While the presence of BE increases the risk of EAC development by more than 10-fold, recent large cohort studies estimate the annual progression rate at less than 0,5% (6-8). This low absolute rate of malignant transformation was the reason to question the relevance of current surveillance guidelines that recommend a gastroscopy every 3-5 years in BE patients (2,6,8,9). Despite some promising results of acid suppressant medication and non-steroidal anti-inflammatory drugs (NSAIDs), no pharmacological interventions have been proven effective in eradicating established BE or preventing its progression towards malignancy (10-14). Data from animal experiments and several cohort and case- control studies indicate that non-steroidal anti-inflammatory drugs (NSAIDs) may reduce the rate of BE progression towards malignancy and the risk of EAC (15-22). However, these findings are disputed by others (23) and the only available randomized controlled trial failed to show any benefit of Celecoxib in reducing progression towards malignancy (24). The therapeutic options for patients with EAC are also limited. Despite neoadjuvant chemoradiation and surgery the overall 5-year survival of EAC patients is less than 20%, highlighting the need for new therapeutic targets (25,26). Aberrant activity of embryological signaling pathways is often implicated in the development of metaplasia and cancer. Experimental data suggests that signaling pathways active during esophageal development can provide a new perspective on BE and EAC development and thus provide potential novel therapeutic targets. The Bone Morphogenetic protein (BMP), Hedgehog (HH), Wingless-Type MMTV Integration Site Family (WNT) and Retinoic acid (RA) signaling pathways have a key role during embryological development of the esophagus and alterations in these pathways have been observed in both BE and EAC. In this review, we summarize the current knowledge regarding the role of BMP, WNT, HH and RA signaling pathways in esophageal embryology, their role in BE development and malignant transformation and discuss their potential as therapeutic targets in EAC treatment.

Proefschrift_Kirill.indb 19 10-05-15 23:47 Signaling pathways in foregut embryology

The esophagus is derived from the embryological foregut. During embryological development the foregut lumen divides along the sagittal axis. The ventral half will become the trachea, lined with columnar epithelium while the dorsal half will become the esophagus, lined with squamous epithelium (27). Most of the understanding of the transcription factors and signaling pathways active in foregut separation and patterning comes from transgenic mouse models (Table 1). These models suggest that differentiation of foregut epithelium towards a squamous or columnar phenotype is regulated by the expression of three key transcription factors: NKX2.1, SOX2 and p63. NKX2.1 induces columnar differentiation of the foregut epithelium, while SOX2 and p63 expression is required for squamous differentiation. Knockout models showed that mice lacking NKX2.1 had impaired foregut separation and a common lumen lined with squamous epithelium (28), while in mice lacking SOX2 or p63 the esophagus was lined with columnar epithelium (28-30). Further experiments identified four main signaling pathways active in the embryological foregut: the BMP, HH, WNT and RA signaling pathways. These pathways are required for proper development of the trachea and esophagus and influence differentiation by regulating the expression of NKX2.1, SOX2 and p63 (31-34) (Figure 1). Signaling through the BMP pathway contributes to a columnar differentiation of foregut epithelium. BMP ligands bind to type 1 and type 2 transmembrane BMP receptors (BMPR1 and BMPR2) that activate the SMAD transcription factors. The BMP4 ligand plays the most important role in foregut embryology. BMP4 is produced in the mesenchyme of the future tracheal domain where it stimulates columnar differentiation of foregut epithelium. High levels of ectopic BMP signaling in the esophageal foregut domain block the development of squamous epithelium and promote columnar differentiation (35), while abrogation of BMP signaling through inactivation of BMPR1 leads to ectopic expression of the squamous transcription factors SOX2 and p63 in the tracheal domain of the foregut (36). To prevent BMP-induced columnar differentiation of the esophageal foregut domain, BMP signaling in the esophageal foregut domain is repressed by the expression of Noggin, a BMP-antagonist. In mice with a homozygous deletion of the Noggin gene Nog the foregut fails to separate properly and the esophageal domain is covered with columnar epithelium, underlining the importance of proper spatial restriction of BMP signaling (37). HH signaling is essential for proper foregut separation. Sonic Hedgehog (SHH) is one of three ligands of the HH signaling pathway. SHH binds to the membrane receptor Patched (PTCH). In an inactive state, PTCH represses Smoothened (SMO). Upon binding of SHH to PTCH, SMO repression is relieved and SMO activates the GLI transcription factors. Disrupted or absent HH signaling caused impaired foregut separation in mice (38-40), and alterations in downstream proteins of the HH signaling cascade such as GLI2,

Proefschrift_Kirill.indb 20 10-05-15 23:47 SHH

BARX1 WNT RA

Frizzled PTCH 1 BMP4 SMO

Noggin Beta-catenin stabilization GLI2 BMPR GLI3 RAR/ SMAD 4 RXR SMAD 1/5/8

NKX2.1 NKX2.1 SOX2 SOX2 p63 p63

Columnar differentiation Squamous differentiation

Fig. 1: Signaling pathways in foregut differentiation. Schematic representation of a foregut epithelial cell in which the BMP and WNT pathways contribute to columnar differentiation while RA signaling contributes to squamous differentiation. HH signaling plays a key role in foregut separation, but its effect on differentiation is currently unclear. BMP4: Bone Morphogenetic protein 4; BMPR: Bone Morphogenetic protein receptor; WNT: Wingless-Type MMTV Integration Site Family ligand; SMO: Smoothened; PTCH: Patched; SHH: Sonic Hedgehog; RAR: Retinoic Acid Receptor; RXR: Retinoid X Receptor; NKX2.1: NK2 homeobox 1; SOX2: Sex determining region Y-box 2

GLI3 and FOXF1 yielded a similar phenotype (41,42). While HH signaling is essential for proper foregut separation, the effects of HH signaling on foregut differentiation are less clear. The esophageal domain of Shh-/- mice failed to develop squamous epithelium and was lined with columnar epithelium instead, suggesting that HH signaling contributes to the development of squamous epithelium (40). Moreover, SHH is present in normal adult squamous epithelium, where it stimulates proliferation of epithelial precursor cells (43). HH signaling is also a known inducer of the BMP pathway. During embryological development of the small intestine HH signaling induced BMP4 expression and was required for proper development of the columnar intestinal epithelium (44). Studies in mouse embryos showed that SHH is initially expressed in the tracheal domain, followed by a shift in expression to the esophageal domain (31,38). This ventral-dorsal shift is required for proper development of both trachea and esophagus, and a location-specific effect of SHH may explain the dual role of SHH with regard to the induction of squamous or columnar differentiation.

Proefschrift_Kirill.indb 21 10-05-15 23:47 WNT signaling is important at various stages of embryonic development and contributes to a columnar differentiation of foregut epithelium. WNT ligands are secreted glycoproteins that bind to the Frizzled membrane receptors. Signaling induced by WNT ligands is classified as either canonical (beta-Catenin dependent) or non-canonical (beta- Catenin independent) (45,46) of which canonical WNT signaling is the best-studied signaling pathway in foregut embryology. Canonical WNT signaling was found to induce columnar differentiation of the tracheal foregut domain by inducing the expression of the columnar transcription factor NKX2.1 and repression of squamous transcription factors SOX2 and p63 (47). Mice models showed that abrogation of canonical WNT signaling through inactivation of beta-Catenin correlates with SOX2 expression in the tracheal foregut domain, while constitutive activation of WNT signaling via overexpression of beta-Catenin caused loss of SOX2 and reciprocal upregulation of NKX2.1 in the esophageal domain (48). One way in which WNT signaling could contribute to a columnar differentiation is through activation of BMP signaling. Deletion of theWNT ligands WNT2 and WNT2b in a mouse model reduced BMP4 signaling and caused loss of NKX2.1 in the tracheal foregut domain (47). This suggests that WNT signaling may be positioned upstream of BMP signaling in specifying the columnar differentiation of the tracheal foregut domain. Similarly to BMP-signaling, spatial restriction of WNT signaling in the foregut is important to prevent columnar differentiation of the esophageal foregut domain. During normal embryological development the esophageal foregut domain is shielded from WNT signaling by the expression of BARX1 (49). BARX1 is expressed in the mesoderm of the esophageal foregut domain where it stimulates the expression of the secreted Frizzle-related proteins 1 and 2 (sFRP1 and sFRP2). sFRPs are secreted polypeptides that prevent the binding of WNT ligands with the Frizzled receptor, and thereby antagonize WNT signaling (49,50). RA signaling is essential for proper development of the esophagus and squamous differentiation. RA directs gene transcription by binding to nuclear Retinoic Acid Receptors (RAR-alpha, RAR-beta and RAR-gamma) and Retinoid X Receptors (RXR-alpha, RXR- beta and RXR-gamma) that act as transcription factors by binding to retinoic acid response elements of target genes. Knockout experiments generating Rar α-/-β-/- mice resulted in an undivided common foregut lumen lined with columnar epithelium (51). One way RA could contribute to a squamous differentiation of the foregut is by antagonizing HH and BMP signaling, since administration of RA inhibited HH and BMP signaling in hindgut development in mice (52). However, it is currently unknown whether a similar mechanism is active in foregut development.

Proefschrift_Kirill.indb 22 10-05-15 23:47 NSAID Monoclonal antibodies targeting WNT ligands Small inhibitory peptides RA (?) 1 Cyclopamine RA suppletion SMO-antagonists RAMBA GLI-antagonists Reflux

HH HH WNT RA SOX9

BMP4 Fibroblasts

Squamous epithelium Barrett’s Barrett’s Esophageal metaplasia dysplasia adenocarcinoma

Fig. 2: Dysregulation of embryological signaling pathways in BE and EAC, and potential therapeutic targets. Activation of the HH-BMP signaling axis contributes to the development of a columnar metaplastic epithelium, through direct effects on the epithelial cells and by upregulation of SOX9. Progression towards malignancy of BE and established EAC is accompanied by a further increase in HH signaling, activation of WNT signaling and a decrease in RA levels.

Embryological signaling pathways in Barrett’s metaplasia development and progression towards malignancy

Besides having a role in the embryological foregut development, alterations in the BMP, WNT, HH and RA embryological signaling pathways have also been implicated in the development of BE and its progression towards EAC (Figure 2). The BMP pathway is not active in normal squamous epithelium but is activated when squamous epithelium becomes inflamed due to GERD, even before the presence of detectable BE (53). Stromal BMP4 expression and activated SMAD proteins in squamous epithelium were observed in both human biopsies and animal models of BE (53,54). In in vitro experiments primary human esophageal cells treated with components of gastroesophageal refluxate showed an increase in BMP4 expression, suggesting additional paracrine BMP signaling (55). Also, incubation of primary human squamous cells with BMP4 induced a shift in the gene expression profile towards that of BE and expression of columnar cytokeratins CK7 and CK20 (53), suggesting that BMP4 contributes to a columnar

Proefschrift_Kirill.indb 23 10-05-15 23:47 transdifferentiation of squamous esophageal cells. While the evidence implicating BMP signaling in the development of BE seems solid, the role of BMP signaling in the malignant transformation of BE requires further research. A single available study that looked at this found that a decrease in BMP4 protein expression in EAC compared to BE, but did not provide evidence of a corresponding decrease at mRNA level (56). SHH expression in the normal esophagus is disputed, but it is likely that HH signaling contributes to BE development and subsequent malignant transformation. Studies reported either absent (57-60) or low-level SHH expression in the basal layer of the esophagus (43,61). However, expression of SHH and its receptor PTCH was increased in biopsies of both BE and EAC (60,62,63) and in a mouse model of BE (54,59,64). Also, an increased expression level of the GLI transcription factors was found in biopsies of EAC compared with biopsies of BE or squamous epithelium (65). HH signaling was found to contribute to BE development in two ways. First, HH signaling can activate the BMP4 promoter leading to BMP signaling (66). Second, HH signaling can induce epithelial SOX9 expression (67). SOX9 is a transcription factor associated with intestinal stem cells (68-70). Overexpression of SOX9 in an organotypic model of BE induced columnar differentiation of squamous cells and expression of the columnar cytokeratin CK8 (71). While data from patient material suggest that HH signaling is further increased during malignant transformation of BE, it is currently less clear how HH signaling contributes to the development of EAC. Possible explanations include a stimulatory effect of HH signaling on cell survival and proliferation (63,72), but this requires further study. The relationship between HH and BMP signaling is another area requiring further study. One study reported that BMP4 protein expression was lower in EAC compared to BE, but a corresponding decrease of mRNA was not found. This is remarkable, as HH signaling is an inducer of BMP signaling in embryonic development and BE development (44). While WNT signaling does not appear to play a role in the development of BE, a progressive increase in WNT signaling is observed during malignant transformation of BE. Squamous epithelium lacks nuclear beta-Catenin indicating that canonical WNT signaling is not active (73,74). Refluxate components can activate WNT signaling in vitro (75,76) and WNT signaling activated SOX9 expression in the esophagus in a mouse model (77). However, nuclear beta-Catenin was not detected in patient biopsies of BE (73,77) suggesting that canonical WNT signaling is not involved in the development of BE metaplasia. In line with this notion, reflux increased the expression of the WNT-antagonist Dickkopf (78,79) that may prevent canonical WNT activation. Moreover, constitutive activation of beta- Catenin in normal squamous epithelium cultured in an organotypic model failed to induce a columnar phenotype. Instead, it yielded a thickened, hyperproliferative squamous epithelium (77,80), suggesting a role for WNT in regulating cellular proliferation. The fact that WNT signaling plays no role in the development of BE is remarkable, given its role

Proefschrift_Kirill.indb 24 10-05-15 23:47 in inducing columnar differentiation during foregut development. On the other hand, WNT signaling does appear to be an important driver of BE malignant transformation. A progressive increase in WNT signaling (defined by nuclear beta-Catenin staining) was observed in biopsy samples of BE dysplasia and EAC (73,77,81) and this was confirmed 1 in an EAC mouse model (82). Several mechanisms could contribute to WNT signaling activation during BE malignant transformation. Increased expression of the WNT2 ligand and hypermethylation of the promoter of the genes encoding WNT antagonists WNT Inhibitory Factor 1 (WIF1) and Secreted Frizzled-like Receptor Proteins were described in EAC (83-85), and promoter hypermethylation of the WIF1 gene in BE metaplasia was associated with progression towards EAC (84). Interestingly, while APC gene mutations are a common driver of colon cancer development (86), these mutations are rare (<5%) in EAC (87). In summary, the absence of activated canonical WNT signaling in BE suggests that WNT signaling is not involved in BE development. On the other hand, the progressive increase of WNT signaling in the metaplasia-dysplasia-carcinoma sequence may be a driving factor in the subsequent malignant transformation of BE. During embryological development RA signaling contributes to a squamous differentiation of the esophageal foregut domain. In apparent contrast, RA biosynthesis is upregulated in BE (88,89). Incubation of human squamous epithelium biopsies with RA led to replacement of the squamous epithelium with a columnar, BE-like epithelium (89) and although RA failed to induce complete columnar differentiation in a squamous esophageal human cell line, it did increase the expression of MUC2, a glycoprotein typically expressed in BE but absent in SE (90). The contradictory effects of RA in esophageal embryology and BE development might be explained by differences in retinoic receptor subtype expression, which may lead to differences in response to RA. Compared to squamous epithelium RAR-alpha and RXR-gamma were upregulated and RAR-gamma and RXR- beta were downregulated in BE (91,92). Morphologically normal squamous epithelium from patients with BE and EAC already had alterations in retinoic acid receptor subtype expression (91,92). Given that only a minority of GERD patients develop BE (5%), pre- existing alterations in retinoic acid receptor subtype expression may predispose normal squamous epithelium to develop a metaplastic response after persistent GERD. In contrast to increased RA signaling pathway activity in BE metaplasia, RA signaling is reduced in BE dysplasia and EAC (88). RA signaling seems to a have tumor suppressor effect, and RA administration inhibited cell proliferation in a variety of cancer cell lines (93-95) and induced apoptosis in a BE metaplasia cell line (96). Downregulation of the RAR-beta has been described in a variety of solid malignancies (97-99) including EAC (100). In addition, enhanced expression of the RA catabolizing enzyme CYP26A1 during the malignant transformation of BE has been linked to decreased RA levels in EAC, and overexpression of CYP26A1 in a BE dysplastic cell line promoted cell proliferation

Proefschrift_Kirill.indb 25 10-05-15 23:47 and survival by depletion of intracellular RA (88). Overall, RA signaling seems to have a dual role in BE; high levels of RA contribute to the development of BE but also prevent the development of EAC. In summary, dysregulated signaling of the BMP, HH and RA pathways is associated with the development of BE metaplasia. During the progression of BE towards malignancy the SHH and WNT signaling pathways are upregulated while the RA and probably the BMP signaling pathways are downregulated.

Therapeutic targeting of embryological signaling pathways in Barrett’s metaplasia and esophageal adenocarcinoma.

Given the important roles of the BMP, HH, WNT and RA pathways in the development of BE metaplasia and progression towards EAC, therapeutic modulation of these pathways may be a valuable complimentary tool in the management of BE and EAC (Figure 2). Several possible therapeutic strategies are discussed below.

Preventing Barrett’s metaplasia development The HH pathway can be targeted both at the level of the cell membrane receptors PTCH and SMO as well as more downstream at the GLI transcription factors (101,102). Preclinical studies suggest that inhibition of HH signaling might prevent the development of BE. Intraperitoneal injections of cyclopamine (a HH-antagonist) decreased gastric BMP4 expression in a mouse model (101). Another study used a surgical rat model of GERD in which rats treated with an oral SMO antagonist (Bristol-Meyers Squibb 833923) for 28 weeks had a significantly lower incidence of Barrett’s metaplasia (59% versus 95% in the control group) (102). Since both animal groups underwent the same procedure to induce persistent GERD, blocking HH signaling can reduce the incidence of BE even in the continuous presence of GERD. The current lack of biomarkers capable of predicting which GERD patients will develop BE limits the options for chemoprevention in GERD patients. However, given the central role of SHH in BE development, modulating the HH-BMP axis might cause regression of established BE. Although this has not been tested, the HH-BMP axis is an attractive therapeutic target in BE.

Blocking Barrett’s metaplasia progression towards malignancy The HH signaling pathway seems to play a role in the progression of BE towards dysplasia and EAC, and provides an attractive therapeutic target in the prevention of BE malignant transformation. While clinical studies are currently lacking, suppression of GLI in a rat model using a combination of ursodeoxycholic acid and aspirin significantly decreased the incidence of EAC (65).

Proefschrift_Kirill.indb 26 10-05-15 23:47 WNT signaling is increased in BE dysplasia and EAC and presents a second therapeutic opportunity. Several clinically tested WNT antagonists are available, including NSAIDs. NSAIDs are known to downregulate WNT-signaling (103-106) and were found to decrease the incidence of several epithelial cancers including colon, gastric and breast 1 (107,108). It is currently unclear whether NSAID can prevent EAC development in patients with BE. NSAIDs prevented malignant transformation of BE in cohort studies (15), but the single reported randomized trial failed to find any chemopreventive effect of Celecoxib (24), despite the fact that Celecoxib inhibited WNT signaling in a colon cancer cell line (109). To clarify the potential value of NSAIDs in preventing EAC development an ongoing phase III trial (the AspECT trial) is investigating the effect of aspirin in combination with a proton pump inhibitor (esomeprazole) on progression towards malignancy (Table 2) (110). RA signaling is a third potential signaling pathway that can be modulated to prevent malignant transformation of BE. The decrease of RA levels during malignant transformation suggests that RA may act as a tumor supressor. One explanation may be that RA signaling can antagonize canonical WNT signaling, possibly via competition with p300, a common co-activator factor for both RARs and beta-Catenin (111). The finding that RA levels decrease while canonical WNT signaling progressively increases during malignant transformation of BE suggests that RA may repress WNT signaling and thus prevent EAC development, but this remains to be tested. While previous chemoprevention trials showed inconclusive results, this could be due to an incomplete coverage of the deregulated pathways. Targeting multiple deregulated signaling pathways by combining NSAIDs and HH antagonists with strategies to prevent a decrease in RA levels may thus translate into more effective chemoprevention strategies.

Therapeutic targets in esophageal adenocarcinoma As described above, only limited evidence suggests that BMP4 is downregulated in EAC, and the exact mechanisms and clinical consequence of reduced BMP4 expression in EAC are still unclear. In colon cancer BMP4 induced differentiation and cell death of colorectal cancer stem cells and sensitized them to oxaliplatin and 5-fluorouracil, possibly as a result of WNT pathway inhibition (112-115). Several studies have suggested that a stem cell subpopulation exists in EAC (116,117), but thus far the effect of targeting the BMP pathway has not been studied in EAC. HH signaling is upregulated in EAC. In vitro inhibition of the HH pathway with cyclopamine reduced tumor proliferation in EAC cell lines and caused regression of established cholagiocarcinoma tumors in a mouse xenograft model (72). A number of HH inhibitors are in development including small molecules targeting SMO (102,118) and blocking antibodies against the SHH ligand (72). In vitro these agents showed promising results by reducing cell viability (63,72). Several clinical trials are underway to evaluate

Proefschrift_Kirill.indb 27 10-05-15 23:47 their effects in various solid tumors as a single agent or in combination with chemotherapy (118-123) including one trial specifically evaluating the effect of an oral SMO antagonist (Bristol-Meyers Squibb 833923) in EAC (124) (Table 2). Furthermore, enhanced HH signaling was observed in patients with residual EAC tumors after chemoradiation, and blocking HH signaling augmented the efficacy of radiation in EAC cell lines (125), suggesting that modulation of HH signaling might also be used to augment the efficacy of current treatment modalities. As mentioned earlier, canonical WNT signaling is deregulated in EAC. In vitro, restoration of the expression of WNT-antagonist WIF1 decreased cell proliferation and sensitized EAC cells to carboplatin treatment (84). Two ongoing phase II trials in EAC are investigating the added value of Celecoxib in neoadjuvant chemotherapy (126) or chemoradiation (127). In addition, several small molecule inhibitors of WNT signaling are undergoing phase I trials in solid tumors (128,129) (Table 2). Downregulation of WNT signaling by clinically available (NSAIDs) or experimental WNT-antagonists may increase the efficacy of current treatment strategies. RA stimulates differentiation and limits cell proliferation in BE. EAC is characterized by low levels of RA (88,130,131) that may contribute to loss of differentiation and uncontrolled proliferation. Restoring RA levels could be an attractive therapeutic strategy in EAC. However, a small (13 EAC patients) phase II trial of RA and Interferon did not show any anti-tumor effect while being associated with considerable toxicity (132). An alternative strategy might be to use RA metabolism blocking agents (RAMBAs). RAMBAs counteract the elevated RA degradation and thus lead to increased RA levels. These agents increase tumor differentiation and apoptosis in vitro and reduce tumor growth in xenograft models of breast and prostate cancer (93,95). Moreover, whereas treatment with RA was associated with weight loss in the mouse models, RAMBAs were well tolerated in animal models and had no cytotoxic effect on normal, noncancerous cell lines (93-95).

Conclusion and perspectives

The increased understanding of embryological signaling pathways and their role in the development and malignant transformation of BE and EAC as outlined above suggests a number of novel therapeutic opportunities at all stages of the metaplasia-dysplasia- adenocarcinoma sequence. However, translation of the promising results obtained in preclinical studies into clinical applications faces several critical challenges. These include potential side effects from manipulating such fundamental pathways, difficulties in selective delivery of the drugs to the target tissue and the selection of patients that will benefit from treatment. Further research is required to address these issues and results from clinical studies testing the efficacy of these novel treatment strategies are eagerly awaited.

Proefschrift_Kirill.indb 28 10-05-15 23:47 Acknowledgements

We would like to thank Dr. E.M E. van Straten for help with the figures. K. Pavlov is supported by a MD-PhD grant from the Junior Scientific Masterclass Groningen. This 1 sponsor had no involvement in the preparation of this manuscript.

Proefschrift_Kirill.indb 29 10-05-15 23:47 References

(1) Toruner M, Soykan I, Ensari A, Kuzu I, Yurdaydin C, Ozden A. Barrett’s esophagus: prevalence and its relationship with dyspeptic symptoms. J Gastroenterol Hepatol 2004;19:535-40. (2) Veldhuyzen van Zanten SJ, Thomson AB, Barkun AN, Armstrong D, Chiba N, White RJ et al. The prevalence of Barrett’s oesophagus in a cohort of 1040 Canadian primary care patients with uninvestigated dyspepsia undergoing prompt endoscopy. Aliment Pharmacol Ther 2006;23:595- 9. (3) Balasubramanian G, Singh M, Gupta N, Gaddam S, Giacchino M, Wani SB et al. Prevalence and predictors of columnar lined esophagus in gastroesophageal reflux disease (GERD) patients undergoing upper endoscopy. Am J Gastroenterol 2012;107:1655-61. (4) Rex DK, Cummings OW, Shaw M, Cumings MD, Wong RK, Vasudeva RS et al. Screening for Barrett’s esophagus in colonoscopy patients with and without heartburn. Gastroenterology 2003;125:1670-7. (5) Theisen J, Nigro JJ, DeMeester TR, Peters JH, Gastal OL, Hagen JA et al. Chronology of the Barrett’s metaplasia-dysplasia-carcinoma sequence. Dis Esophagus 2004;17:67-70. (6) Hvid-Jensen F, Pedersen L, Drewes AM, Sorensen HT, Funch-Jensen P. Incidence of adenocarcinoma among patients with Barrett’s esophagus. N Engl J Med 2011;365:1375-83. (7) de Jonge PJ, van Blankenstein M, Looman CW, Casparie MK, Meijer GA, Kuipers EJ. Risk of malignant progression in patients with Barrett’s oesophagus: a Dutch nationwide cohort study. Gut 2010;59:1030-6. (8) Desai TK, Krishnan K, Samala N, Singh J, Cluley J, Perla S et al. The incidence of oesophageal adenocarcinoma in non-dysplastic Barrett’s oesophagus: a meta-analysis. Gut 2012;61:970-6. (9) American Gastroenterological Association, Spechler SJ, Sharma P, Souza RF, Inadomi JM, Shaheen NJ. American Gastroenterological Association medical position statement on the management of Barrett’s esophagus. Gastroenterology 2011;140:1084-91. (10) Peters FT, Ganesh S, Kuipers EJ, Sluiter WJ, Klinkenberg-Knol EC, Lamers CB et al. Endoscopic regression of Barrett’s oesophagus during omeprazole treatment; a randomised double blind study. Gut 1999;45:489-94. (11) Rees JR, Lao-Sirieix P, Wong A, Fitzgerald RC. Treatment for Barrett’s oesophagus. Cochrane Database Syst Rev 2010;doi:CD004060. (12) El-Serag HB, Aguirre TV, Davis S, Kuebeler M, Bhattacharyya A, Sampliner RE. Proton pump inhibitors are associated with reduced incidence of dysplasia in Barrett’s esophagus. Am J Gastroenterol 2004;99:1877-83. (13) Hillman LC, Chiragakis L, Shadbolt B, Kaye GL, Clarke AC. Proton-pump inhibitor therapy and the development of dysplasia in patients with Barrett’s oesophagus. Med J Aust 2004;180:387-91. (14) de Bortoli N, Martinucci I, Piaggi P, Maltinti S, Bianchi G, Ciancia E et al. Randomised clinical trial: twice daily esomeprazole 40 mg vs. pantoprazole 40 mg in Barrett’s oesophagus for 1 year. Aliment Pharmacol Ther 2011;33:1019-27. (15) Liao LM, Vaughan TL, Corley DA, Cook MB, Casson AG, Kamangar F et al. Nonsteroidal anti- inflammatory drug use reduces risk of adenocarcinomas of the esophagus and esophagogastric junction in a pooled analysis. Gastroenterology 2012;142:442,452-5. (16) Pandeya N, Williams GM, Sadhegi S, Green AC, Webb PM, Whiteman DC. Associations of duration, intensity, and quantity of smoking with adenocarcinoma and squamous cell carcinoma

Proefschrift_Kirill.indb 30 10-05-15 23:47 of the esophagus. Am J Epidemiol 2008;168:105-14. (17) Vaughan TL, Dong LM, Blount PL, Ayub K, Odze RD, Sanchez CA et al. Non-steroidal anti- inflammatory drugs and risk of neoplastic progression in Barrett’s oesophagus: a prospective study. Lancet Oncol 2005;6:945-52. 1 (18) Fortuny J, Johnson CC, Bohlke K, Chow WH, Hart G, Kucera G et al. Use of anti-inflammatory drugs and lower esophageal sphincter-relaxing drugs and risk of esophageal and gastric cancers. Clin Gastroenterol Hepatol 2007;5:1154-9. (19) Nguyen DM, Richardson P, El-Serag HB. Medications (NSAIDs, statins, proton pump inhibitors) and the risk of esophageal adenocarcinoma in patients with Barrett’s esophagus. Gastroenterology 2010;138:2260-6. (20) Corley DA, Kerlikowske K, Verma R, Buffler P. Protective association of aspirin/NSAIDs and esophageal cancer: a systematic review and meta-analysis. Gastroenterology 2003;124:47-56. (21) Kastelein F, Spaander MC, Biermann K, Steyerberg EW, Kuipers EJ, Bruno MJ et al. Nonsteroidal anti-inflammatory drugs and statins have chemopreventative effects in patients with Barrett’s esophagus. Gastroenterology 2011;141:2000-8. (22) Buttar NS, Wang KK, Leontovich O, Westcott JY, Pacifico RJ, Anderson MA et al. Chemoprevention of esophageal adenocarcinoma by COX-2 inhibitors in an animal model of Barrett’s esophagus. Gastroenterology 2002;122:1101-12. (23) Gatenby PA, Ramus JR, Caygill CP, Winslet MC, Watson A. Aspirin is not chemoprotective for Barrett’s adenocarcinoma of the oesophagus in multicentre cohort. Eur J Cancer Prev 2009;18:381-4. (24) Heath EI, Canto MI, Piantadosi S, Montgomery E, Weinstein WM, Herman JG et al. Secondary chemoprevention of Barrett’s esophagus with celecoxib: results of a randomized trial. J Natl Cancer Inst 2007;99:545-57. (25) Bouvier AM, Binquet C, Gagnaire A, Jouve JL, Faivre J, Bedenne L. Management and prognosis of esophageal cancers: has progress been made? Eur J Cancer 2006;42:228-33. (26) Worni M, Castleberry AW, Gloor B, Pietrobon R, Haney JC, D’Amico TA et al. Trends and outcomes in the use of surgery and radiation for the treatment of locally advanced esophageal cancer: a propensity score adjusted analysis of the surveillance, epidemiology, and end results registry from 1998 to 2008. Dis Esophagus 2013;doi: 10.1111/dote.12123 (27) Jacobs IJ, Ku WY, Que J. Genetic and cellular mechanisms regulating anterior foregut and esophageal development. Dev Biol 2012;369:54-64. (28) Que J, Okubo T, Goldenring JR, Nam KT, Kurotani R, Morrisey EE et al. Multiple dose-dependent roles for Sox2 in the patterning and differentiation of anterior foregut endoderm. Development 2007;134:2521-31. (29) Wang X, Ouyang H, Yamamoto Y, Kumar PA, Wei TS, Dagher R et al. Residual embryonic cells as precursors of a Barrett’s-like metaplasia. Cell 2011;145:1023-35. (30) Daniely Y, Liao G, Dixon D, Linnoila RI, Lori A, Randell SH et al. Critical role of p63 in the development of a normal esophageal and tracheobronchial epithelium. Am J Physiol Cell Physiol 2004;287:C171-81. (31) Litingtung Y, Lei L, Westphal H, Chiang C. Sonic hedgehog is essential to foregut development. Nat Genet 1998;20:58-61. (32) Mendelsohn C, Lohnes D, Decimo D, Lufkin T, LeMeur M, Chambon P et al. Function of the

Proefschrift_Kirill.indb 31 10-05-15 23:47 retinoic acid receptors (RARs) during development (II). Multiple abnormalities at various stages of organogenesis in RAR double mutants. Development 1994;120:2749-71. (33) Que J, Choi M, Ziel JW, Klingensmith J, Hogan BL. Morphogenesis of the trachea and esophagus: current players and new roles for noggin and Bmps. Differentiation 2006;74:422-37. (34) Jacobs IJ, Ku WY, Que J. Genetic and cellular mechanisms regulating anterior foregut and esophageal development. Dev Biol 2012;369:54-64. (35) Rodriguez P, Da Silva S, Oxburgh L, Wang F, Hogan BL, Que J. BMP signaling in the development of the mouse esophagus and forestomach. Development 2010;137:4171-6. (36) Domyan ET, Ferretti E, Throckmorton K, Mishina Y, Nicolis SK, Sun X. Signaling through BMP receptors promotes respiratory identity in the foregut via repression of Sox2. Development 2011;138:971-81. (37) Que J, Choi M, Ziel JW, Klingensmith J, Hogan BL. Morphogenesis of the trachea and esophagus: current players and new roles for noggin and Bmps. Differentiation 2006;74:422-37. (38) Ioannides AS, Henderson DJ, Spitz L, Copp AJ. Role of Sonic hedgehog in the development of the trachea and oesophagus. J Pediatr Surg 2003;38:29-36. (39) Arsic D, Cameron V, Ellmers L, Quan QB, Keenan J, Beasley S. Adriamycin disruption of the Shh-Gli pathway is associated with abnormalities of foregut development. J Pediatr Surg 2004;39:1747-53. (40) Litingtung Y, Lei L, Westphal H, Chiang C. Sonic hedgehog is essential to foregut development. Nat Genet 1998;20:58-61. (41) Motoyama J, Liu J, Mo R, Ding Q, Post M, Hui CC. Essential function of Gli2 and Gli3 in the formation of lung, trachea and oesophagus. Nat Genet 1998;20:54-7. (42) Mahlapuu M, Enerback S, Carlsson P. Haploinsufficiency of the forkhead gene Foxf1, a target for sonic hedgehog signaling, causes lung and foregut malformations. Development 2001;128:2397- 406. (43) van Dop WA, Rosekrans SL, Uhmann A, Jaks V, Offerhaus GJ, van den Bergh Weerman MA et al. Hedgehog signalling stimulates precursor cell accumulation and impairs epithelial maturation in the murine oesophagus. Gut 2013;62:348-57. (44) Madison BB, Braunstein K, Kuizon E, Portman K, Qiao XT, Gumucio DL. Epithelial hedgehog signals pattern the intestinal crypt-villus axis. Development 2005;132:279-89. (45) Anastas JN, Moon RT. WNT signalling pathways as therapeutic targets in cancer. Nat Rev Cancer 2013;13:11-26. (46) Niehrs C. The complex world of WNT receptor signalling. Nat Rev Mol Cell Biol 2012;13:767-79. (47) Goss AM, Tian Y, Tsukiyama T, Cohen ED, Zhou D, Lu MM et al. Wnt2/2b and beta-catenin signaling are necessary and sufficient to specify lung progenitors in the foregut. Dev Cell 2009;17:290-8. (48) Harris-Johnson KS, Domyan ET, Vezina CM, Sun X. beta-Catenin promotes respiratory progenitor identity in mouse foregut. Proc Natl Acad Sci U S A 2009;106:16287-92. (49) Woo J, Miletich I, Kim BM, Sharpe PT, Shivdasani RA. Barx1-mediated inhibition of Wnt signaling in the mouse thoracic foregut controls tracheo-esophageal septation and epithelial differentiation. PLoS One 2011;doi: 10.1371/journal.pone.0022493. (50) Kim BM, Buchner G, Miletich I, Sharpe PT, Shivdasani RA. The stomach mesenchymal transcription factor Barx1 specifies gastric epithelial identity through inhibition of transient Wnt

Proefschrift_Kirill.indb 32 10-05-15 23:47 signaling. Dev Cell 2005;8:611-22. (51) Luo J, Sucov HM, Bader JA, Evans RM, Giguere V. Compound mutants for retinoic acid receptor (RAR) beta and RAR alpha 1 reveal developmental functions for multiple RAR beta isoforms. Mech Dev 1996;55:33-44. 1 (52) Sasaki Y, Iwai N, Tsuda T, Kimura O. Sonic hedgehog and bone morphogenetic protein 4 expressions in the hindgut region of murine embryos with anorectal malformations. J Pediatr Surg 2004;39:170-3. (53) Milano F, van Baal JW, Buttar NS, Rygiel AM, de Kort F, DeMars CJ et al. Bone morphogenetic protein 4 expressed in esophagitis induces a columnar phenotype in esophageal squamous cells. Gastroenterology 2007;132:2412-21. (54) Quante M, Bhagat G, Abrams JA, Marache F, Good P, Lee MD et al. Bile acid and inflammation activate gastric cardia stem cells in a mouse model of Barrett-like metaplasia. Cancer Cell 2012;21:36-51. (55) Zhou G, Sun YG, Wang HB, Wang WQ, Wang XW, Fang DC. Acid and bile salt up-regulate BMP4 expression in human esophageal epithelium cells. Scand J Gastroenterol 2009;44:926-32. (56) van Baal JW, Milana F, Rygiel AM, Sondermeijer CM, Spek CA, Bergman JJ et al. A comparative analysis by SAGE of gene expression profiles of esophageal adenocarcinoma and esophageal squamous cell carcinoma. Cell Oncol 2008;30:63-75. (57) van den Brink GR, Hardwick JC, Nielsen C, Xu C, ten Kate FJ, Glickman J et al. Sonic hedgehog expression correlates with fundic gland differentiation in the adult gastrointestinal tract. Gut 2002;51:628-33. (58) Ma X, Sheng T, Zhang Y, Zhang X, He J, Huang S et al. Hedgehog signaling is activated in subsets of esophageal cancers. Int J Cancer 2006;118:139-48. (59) Yamanaka Y, Shiotani A, Fujimura Y, Ishii M, Fujita M, Matsumoto H et al. Expression of Sonic hedgehog (SHH) and CDX2 in the columnar epithelium of the lower oesophagus. Dig Liver Dis 2011;43:54-9. (60) Wang DH, Clemons NJ, Miyashita T, Dupuy AJ, Zhang W, Szczepny A et al. Aberrant epithelial- mesenchymal Hedgehog signaling characterizes Barrett’s metaplasia. Gastroenterology 2010;138:1810-22. (61) Isohata N, Aoyagi K, Mabuchi T, Daiko H, Fukaya M, Ohta H et al. Hedgehog and epithelial- mesenchymal transition signaling in normal and malignant epithelial cells of the esophagus. Int J Cancer 2009;125:1212-21. (62) Yang L, Wang LS, Chen XL, Gatalica Z, Qiu S, Liu Z et al. Hedgehog signaling activation in the development of squamous cell carcinoma and adenocarcinoma of esophagus. Int J Biochem Mol Biol 2012;3:46-57. (63) Zaidi AH, Komatsu Y, Kelly LA, Malhotra U, Rotoloni C, Kosovec JE et al. Smoothened inhibition leads to decreased proliferation and induces apoptosis in esophageal adenocarcinoma cells. Cancer Invest 2013;31:480-9. (64) Wang DH, Clemons NJ, Miyashita T, Dupuy AJ, Zhang W, Szczepny A et al. Aberrant epithelial- mesenchymal Hedgehog signaling characterizes Barrett’s metaplasia. Gastroenterology 2010;138:1810-22. (65) Rizvi S, Demars CJ, Comba A, Gainullin VG, Rizvi Z, Almada LL et al. Combinatorial chemoprevention reveals a novel smoothened-independent role of GLI1 in esophageal carcinogenesis. Cancer Res 2010;70:6787-96.

Proefschrift_Kirill.indb 33 10-05-15 23:47 (66) Kawai S, Sugiura T. Characterization of human bone morphogenetic protein (BMP)-4 and -7 gene promoters: activation of BMP promoters by Gli, a sonic hedgehog mediator. Bone 2001;29:54-61. (67) Bien-Willner GA, Stankiewicz P, Lupski JR. SOX9cre1, a cis-acting regulatory element located 1.1 Mb upstream of SOX9, mediates its enhancement through the SHH pathway. Hum Mol Genet 2007;16:1143-56. (68) Formeister EJ, Sionas AL, Lorance DK, Barkley CL, Lee GH, Magness ST. Distinct SOX9 levels differentially mark stem/progenitor populations and enteroendocrine cells of the small intestine epithelium. Am J Physiol Gastrointest Liver Physiol 2009;296:1108-18. (69) Bastide P, Darido C, Pannequin J, Kist R, Robine S, Marty-Double C et al. Sox9 regulates cell proliferation and is required for Paneth cell differentiation in the intestinal epithelium. J Cell Biol 2007;178:635-48. (70) Ramalingam S, Daughtridge GW, Johnston MJ, Gracz AD, Magness ST. Distinct levels of Sox9 expression mark colon epithelial stem cells that form colonoids in culture. Am J Physiol Gastrointest Liver Physiol 2012;302:10-20. (71) Clemons NJ, Wang DH, Croagh D, Tikoo A, Fennell CM, Murone C et al. Sox9 drives columnar differentiation of esophageal squamous epithelium: a possible role in the pathogenesis of Barrett’s esophagus. Am J Physiol Gastrointest Liver Physiol 2012;303:1335-46. (72) Berman DM, Karhadkar SS, Maitra A, Montes De Oca R, Gerstenblith MR, Briggs K et al. Widespread requirement for Hedgehog ligand stimulation in growth of digestive tract tumours. Nature 2003;425:846-51. (73) Bian YS, Osterheld MC, Bosman FT, Fontolliet C, Benhattar J. Nuclear accumulation of beta- catenin is a common and early event during neoplastic progression of Barrett esophagus. Am J Clin Pathol 2000;114:583-90. (74) Krishnadath KK, Tilanus HW, van Blankenstein M, Hop WC, Kremers ED, Dinjens WN et al. Reduced expression of the cadherin-catenin complex in oesophageal adenocarcinoma correlates with poor prognosis. J Pathol 1997;182:331-8. (75) Chen X, Jiang K, Fan Z, Liu Z, Zhang P, Zheng L et al. Aberrant expression of Wnt and Notch signal pathways in Barrett’s esophagus. Clin Res Hepatol Gastroenterol 2012;36:473-83. (76) Korinek V, Barker N, Morin PJ, van Wichen D, de Weger R, Kinzler KW et al. Constitutive transcriptional activation by a beta-catenin-Tcf complex in APC-/- colon carcinoma. Science 1997;275:1784-7. (77) Moyes LH, McEwan H, Radulescu S, Pawlikowski J, Lamm CG, Nixon C et al. Activation of Wnt signalling promotes development of dysplasia in Barrett’s oesophagus. J Pathol 2012;228:99-112. (78) Reveiller M, Ghatak S, Toia L, Kalatskaya I, Stein L, D’Souza M et al. Bile exposure inhibits expression of squamous differentiation genes in human esophageal epithelial cells. Ann Surg 2012;255:1113-20. (79) Ali I, Rafiee P, Zheng Y, Johnson C, Banerjee B, Haasler G et al. Intramucosal distribution of WNT signaling components in human esophagus. J Clin Gastroenterol 2009;43:327-37. (80) Kong J, Crissey MA, Stairs DB, Sepulveda AR, Lynch JP. Cox2 and beta-catenin/T-cell factor signaling intestinalize human esophageal keratinocytes when cultured under organotypic conditions. Neoplasia 2011;13:792-805. (81) Osterheld MC, Bian YS, Bosman FT, Benhattar J, Fontolliet C. Beta-catenin expression and its association with prognostic factors in adenocarcinoma developed in Barrett esophagus. Am J Clin Pathol 2002;117:451-6.

Proefschrift_Kirill.indb 34 10-05-15 23:47 (82) Quante M, Bhagat G, Abrams JA, Marache F, Good P, Lee MD et al. Bile acid and inflammation activate gastric cardia stem cells in a mouse model of Barrett-like metaplasia. Cancer Cell 2012;21:36-51. (83) Clement G, Braunschweig R, Pasquier N, Bosman FT, Benhattar J. Alterations of the Wnt signaling 1 pathway during the neoplastic progression of Barrett’s esophagus. Oncogene 2006;25:3084-92. (84) Clement G, Guilleret I, He B, Yagui-Beltran A, Lin YC, You L et al. Epigenetic alteration of the Wnt inhibitory factor-1 promoter occurs early in the carcinogenesis of Barrett’s esophagus. Cancer Sci 2008;99:46-53. (85) Zou H, Molina JR, Harrington JJ, Osborn NK, Klatt KK, Romero Y et al. Aberrant methylation of secreted frizzled-related protein genes in esophageal adenocarcinoma and Barrett’s esophagus. Int J Cancer 2005;116:584-91. (86) Morin PJ, Sparks AB, Korinek V, Barker N, Clevers H, Vogelstein B et al. Activation of beta- catenin-Tcf signaling in colon cancer by mutations in beta-catenin or APC. Science 1997;275:1787- 90. (87) Choi YW, Heath EI, Heitmiller R, Forastiere AA, Wu TT. Mutations in beta-catenin and APC genes are uncommon in esophageal and esophagogastric junction adenocarcinomas. Mod Pathol 2000;13:1055-9. (88) Chang CL, Hong E, Lao-Sirieix P, Fitzgerald RC. A novel role for the retinoic acid-catabolizing enzyme CYP26A1 in Barrett’s associated adenocarcinoma. Oncogene 2008;27:2951-60. (89) Chang CL, Lao-Sirieix P, Save V, De La Cueva Mendez G, Laskey R, Fitzgerald RC. Retinoic acid- induced glandular differentiation of the oesophagus. Gut 2007;56:906-17. (90) Cooke G, Blanco-Fernandez A, Seery JP. The effect of retinoic acid and deoxycholic acid on the differentiation of primary human esophageal keratinocytes. Dig Dis Sci 2008;53:2851-7. (91) Brabender J, Lord RV, Metzger R, Park J, Salonga D, Danenberg KD et al. Role of retinoid X receptor mRNA expression in Barrett’s esophagus. J Gastrointest Surg 2004;8:413-22. (92) Lord RV, Tsai PI, Danenberg KD, Peters JH, Demeester TR, Tsao-Wei DD et al. Retinoic acid receptor-alpha messenger RNA expression is increased and retinoic acid receptor-gamma expression is decreased in Barrett’s intestinal metaplasia, dysplasia, adenocarcinoma sequence. Surgery 2001;129:267-76. (93) Huynh CK, Brodie AM, Njar VC. Inhibitory effects of retinoic acid metabolism blocking agents (RAMBAs) on the growth of human prostate cancer cells and LNCaP prostate tumour xenografts in SCID mice. Br J Cancer 2006;94:513-23. (94) Patel JB, Huynh CK, Handratta VD, Gediya LK, Brodie AM, Goloubeva OG et al. Novel retinoic acid metabolism blocking agents endowed with multiple biological activities are efficient growth inhibitors of human breast and prostate cancer cells in vitro and a human breast tumor xenograft in nude mice. J Med Chem 2004;47:6716-29. (95) Godbole AM, Purushottamachar P, Martin MS, Daskalakis C, Njar VC. Autophagy inhibition synergistically enhances anticancer efficacy of RAMBA, VN/12-1 in SKBR-3 cells, and tumor xenografts. Mol Cancer Ther 2012;11:898-908. (96) Hormi-Carver K, Feagins LA, Spechler SJ, Souza RF. All trans-retinoic acid induces apoptosis via p38 and caspase pathways in metaplastic Barrett’s cells. Am J Physiol Gastrointest Liver Physiol 2007;292:18-27. (97) Suh YA, Lee HY, Virmani A, Wong J, Mann KK, Miller WH,Jr et al. Loss of retinoic acid receptor beta gene expression is linked to aberrant histone H3 acetylation in lung cancer cell lines. Cancer

Proefschrift_Kirill.indb 35 10-05-15 23:47 Res 2002;62:3945-9. (98) Hayashi K, Yokozaki H, Goodison S, Oue N, Suzuki T, Lotan R et al. Inactivation of retinoic acid receptor beta by promoter CpG hypermethylation in gastric cancer. Differentiation 2001;68:13- 21. (99) Yang Q, Sakurai T, Kakudo K. Retinoid, retinoic acid receptor beta and breast cancer. Breast Cancer Res Treat 2002;76:167-73. (100) Qiu H, Zhang W, El-Naggar AK, Lippman SM, Lin P, Lotan R et al. Loss of retinoic acid receptor- beta expression is an early event during esophageal carcinogenesis. Am J Pathol 1999;155:1519- 23. (101) van den Brink GR, Hardwick JC, Tytgat GN, Brink MA, Ten Kate FJ, Van Deventer SJ et al. Sonic hedgehog regulates gastric gland morphogenesis in man and mouse. Gastroenterology 2001;121:317-28. (102) Gibson MK, Zaidi AH, Davison JM, Sanz AF, Hough B, Komatsu Y et al. Prevention of barrett esophagus and esophageal adenocarcinoma by smoothened inhibitor in a rat model of gastroesophageal reflux disease. Ann Surg 2013;258:82-8. (103) Bos CL, Kodach LL, van den Brink GR, Diks SH, van Santen MM, Richel DJ et al. Effect of aspirin on the Wnt/beta-catenin pathway is mediated via protein phosphatase 2A. Oncogene 2006;25:6447-56. (104) Dihlmann S, Siermann A, von Knebel Doeberitz M. The nonsteroidal anti-inflammatory drugs aspirin and indomethacin attenuate beta-catenin/TCF-4 signaling. Oncogene 2001;20:645-53. (105) Greenspan EJ, Madigan JP, Boardman LA, Rosenberg DW. Ibuprofen inhibits activation of nuclear {beta}-catenin in human colon adenomas and induces the phosphorylation of GSK- 3{beta}. Cancer Prev Res 2011;4:161-71. (106) Barker N, Clevers H. Mining the Wnt pathway for cancer therapeutics. Nat Rev Drug Discov 2006;5:997-1014. (107) Boon EM, Keller JJ, Wormhoudt TA, Giardiello FM, Offerhaus GJ, van der Neut R et al. Sulindac targets nuclear beta-catenin accumulation and Wnt signalling in adenomas of patients with familial adenomatous polyposis and in human colorectal cancer cell lines. Br J Cancer 2004;90:224-9. (108) Algra AM, Rothwell PM. Effects of regular aspirin on long-term cancer incidence and metastasis: a systematic comparison of evidence from observational studies versus randomised trials. Lancet Oncol 2012;13:518-27. (109) Deng Y, Su Q, Mo J, Fu X, Zhang Y, Lin EH. Celecoxib downregulates CD133 expression through inhibition of the Wnt signaling pathway in colon cancer cells. Cancer Invest 2013;31:97-102. (110) University of Oxford. A Phase III, Randomized, Study of Aspirin and Esomeprazole Chemoprevention in Barrett’s Metaplasia. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT0035768, accessed 26- 2-2014 (111) Shah S, Hecht A, Pestell R, Byers SW. Trans-repression of beta-catenin activity by nuclear receptors. J Biol Chem 2003;278:48137-45. (112) Kallioniemi A. Bone morphogenetic protein 4-a fascinating regulator of cancer cell behavior. Cancer Genet 2012;205:267-77. (113) He XC, Zhang J, Tong WG, Tawfik O, Ross J, Scoville DH et al. BMP signaling inhibits

Proefschrift_Kirill.indb 36 10-05-15 23:47 intestinal stem cell self-renewal through suppression of Wnt-beta-catenin signaling. Nat Genet 2004;36:1117-21. (114) Lombardo Y, Scopelliti A, Cammareri P, Todaro M, Iovino F, Ricci-Vitiani L et al. Bone morphogenetic protein 4 induces differentiation of colorectal cancer stem cells and increases 1 their response to chemotherapy in mice. Gastroenterology 2011;140:297-309. (115) Kosinski C, Li VS, Chan AS, Zhang J, Ho C, Tsui WY et al. Gene expression patterns of human colon tops and basal crypts and BMP antagonists as intestinal stem cell niche factors. Proc Natl Acad Sci U S A 2007;104:15418-23. (116) Smit JK, Faber H, Niemantsverdriet M, Baanstra M, Bussink J, Hollema H et al. Prediction of response to radiotherapy in the treatment of esophageal cancer using stem cell markers. Radiother Oncol 2013;107:434-41 (117) Zhao R, Quaroni L, Casson AG. Identification and characterization of stemlike cells in human esophageal adenocarcinoma and normal epithelial cell lines. J Thorac Cardiovasc Surg 2012;144:1192-9. (118) LoRusso PM, Rudin CM, Reddy JC, Tibes R, Weiss GJ, Borad MJ et al. Phase I trial of hedgehog pathway inhibitor vismodegib (GDC-0449) in patients with refractory, locally advanced or metastatic solid tumors. Clin Cancer Res 2011;17:2502-11. (119) Genentech. A Study of the Hedgehog Pathway Inhibitor Vismodegib in Patients With Advanced Solid Malignancies Including Hepatocellular Carcinoma With Varying Degrees of Renal or Hepatic Function. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT01546519, accessed 26-2-2014 (120) Millennium Pharmaceuticals I. A Study of TAK-441 in Adult Patients With Advanced Nonhematologic Malignancies. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT01204073, accessed 26-2-2014 (121) Pfizer. A Study Of PF-04449913 Administered Alone In Select Solid Tumors. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ ct2/show/NCT01286467, accessed 26-2-2014 (122) Novartis Pharmaceuticals. A Dose Finding and Safety Study of Oral LEQ506 in Patients With Advanced Solid Tumors. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT01106508, accessed 26-2-2014 (123) National Cancer Institute. Combination Chemotherapy With or Without GDC-0449 in Treating Patients With Advanced Stomach Cancer or Gastroesophageal Junction Cancer. ClinicalTrials. gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials. gov/ct2/show/NCT00982592, accessed 26-2-2014 (124) Bristol-Myers Squibb. A Study of BMS-833923 With Cisplatin and Capecitabine in Inoperable, Metastatic Gastric, Gastroesophageal, or Esophageal Adenocarcinomas. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ ct2/show/NCT00909402, accessed 26-2-2014 (125) Sims-Mourtada J, Izzo JG, Apisarnthanarax S, Wu TT, Malhotra U, Luthra R et al. Hedgehog: an attribute to tumor regrowth after chemoradiotherapy and a target to improve radiation response. Clin Cancer Res 2006;12:6565-72. (126) Weill Medical College of Cornell University. Celecoxib, Paclitaxel, and Carboplatin in Treating Patients With Cancer of the Esophagus. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT00066716, accessed 26-

Proefschrift_Kirill.indb 37 10-05-15 23:47 2-2014 (127) UNC Lineberger Comprehensive Cancer Center. Irinotecan, Cisplatin, and Radiation Therapy With or Without Celecoxib in Treating Patients With Stage II, Stage III, or Stage IV Esophageal Cancer. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT00520091, accessed 26-2-2014 (128) Prism Pharma Co. L. Safety and Efficacy Study of PRI-724 in Subjects With Advanced Solid Tumors. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT01302405, accessed 26-2-2014 (129) OncoMed Pharmaceuticals I. A Dose Escalation Study of OMP-54F28 in Subjects With Solid Tumors. ClinicalTrials.gov [Internet] Bethesda (MD): National Library of Medicine (US);2013. http://clinicaltrials.gov/ct2/show/NCT01608867, accessed 26-2-2014 (130) Dowlatshahi K, Mehta RG, Levin B, Cerny WL, Skinner DB, Moon RC. Retinoic-acid-binding protein in normal and neoplastic human esophagus. Cancer 1984;54:308-11. (131) Liang ZD, Lippman SM, Wu TT, Lotan R, Xu XC. RRIG1 mediates effects of retinoic acid receptor beta2 on tumor cell growth and gene expression through binding to and inhibition of RhoA. Cancer Res 2006;66:7111-8. (132) Enzinger PC, Ilson DH, Saltz LB, Martin LK, Kelsen DP. Phase II clinical trial of 13-cis-retinoic acid and interferon-alpha-2a in patients with advanced esophageal carcinoma. Cancer 1999;85:1213-7.

Proefschrift_Kirill.indb 38 10-05-15 23:47 Reference [35] [36] [37] [41] [31] [42] [48] [47] [49] [51] -/-

Gli3 1 -/- positive + mice +/- Sox2 + Gli3 -/- Outcome Clusters of simple columnar cells in the esophagus Failure of foregut separation and p63 domain. cells in both the ventral and dorsal foregut Failure of proper foregut esophageal fistula separation with trachea- Esophageal lining has characteristics of trachea Dorsal foregut domain epithelium covered with separation in Gli2 of foregut Failure columnar No esophagus or trachea development in Gli2 separation of foregut Failure Loss of Nkx2.1 and expansion of Sox2 in the domain foregut ventral Loss of Sox2 and expansion of Nkx2.1 domain foregut in the dorsal Reduced Bmp4 signalling in the ventral foregut domain in the ventral foregut of Nkx2.1 expression Loss separation of foregut Failure Loss of p63 and expansion of domain foregut Nkx2.1 in the dorsal lined is foregut separation. Undivided foregut of Failure with ciliated epithelium mice mice and mice mice -/- +/- -/- / -/- -/- β mice mice mice Gli3 Gli3 -/- mice -/- +/- -/- -/- -/- -/- α1 Model Bmpr1 constitutive with mice Mutant the foregut throughout expression inactivation Bmpr1 with mice Mutant Nog Shh Gli2 Gli2 Foxf1 Mutant mice with constitutive beta- the throughout catenin expression foregut Mutant mice with inactivation of beta-catenin in the foregut Wnt2/2b Barx1 Rar Protein function Protein Bone Morphogenetic Receptor 1A Protein antagonist BMP Signaling protein in hedgehog pathway factors downstream Transcription of Shh. downstream factor Transcription of Shh. Central signal transducer in the Wnt signaling pathway canonical Wnt ligand Antagonist of canonical Wnt signaling Retinoic acid receptor gene) Ctnnb1 Gene Bmpr1a Nog Shh Gli2 Gli3 Foxf1 Beta-catenin (encoded by the Wnt2/2b Barx1 Rar Pathway BMP SHH WNT RA Table 1: Pathways involved in foregut embryology and differentiation of the trachea esophagus. Table

Proefschrift_Kirill.indb 39 10-05-15 23:47 Table 2: Clinical trials investigating modulators of SHH and WNT signaling. WNT SHH Pathway and Aspirin (AspECT) Esomeprazole versus carboplatin Carboplatin versus Paclitaxel and Celecoxib with Paclitaxel and therapy Irinotecan, Cisplatin and Radiation and Radiation therapy versus Celecoxib with Irinotecan, Cisplatin PRI-724 (Wnt pathway inhibitor) OMP-54F28 (Wnt pathway inhibitor) Vismodegib Vismodegib (Hedgehog inhibitor) TAK-441 (Hedgehog inhibitor) LEQ506 (Hedgehog inhibitor) PF-04449913 (Hedgehog inhibitor) with Cisplatin and Capecitabine BMS-833923 (Hedgehog inhibitor) Agent Phase 3 Phase 2 Phase 2 Phase 1 Phase 1 Phase 2 Phase 1 Phase 1 Phase 1 Phase 1 Phase 1 typev Study adenocarcinoma or high grade dysplasia All causes of mortality or conversion rate from Barrett's metaplasia to esophageal cancer Pathological response at the time of surgical resection in patients with in patients with esophageal cancer Rate of complete pathologic remission in patients with resectable disease Rate of cellular apoptosis and proliferation tumors solid with patients in PRI-724 of dose therapeutic maximum Determine Determine safety profile of OMP-54F28 in patients with solid tumors with advanced stomach cancer or gastro esophageal junction of patients survival progression-free on median Vismodegib of Effect Evaluate pharmacokinetics of Vismodegib in patients with solid tumors in of TAK-441 dose patients with solid tumors tolerated maximum and profile safety Determine of LEQ506 in patients with solid tumors Determine maximum tolerated dose and characterize dose limiting toxicity solid tumors with patients in PF-04449913 of toxicity limiting dose cycle first Determine adenocarcinomas Capecitabine in patients with gastric, gastroesophageal and esophageal and Cisplatin with combination in administered BMS-833923 of profile Determine maximum therapeutic dose, dose limiting toxicity and safety Primary outcome measure Ongoing Completed Completed Recruiting Recruiting Ongoing Recruiting Completed Ongoing Completed Completed Status NCT00357682 NCT00066716 NCT00520091 NCT01302405 NCT01608867 NCT00982592 NCT01546519 NCT01204073 NCT01106508 NCT01286467 NCT00909402 identifier ClinicalTrials.gov [110] [126] [127] [128] [129] [123] [119] [120] [122] [121] [124] Reference

Proefschrift_Kirill.indb 40 10-05-15 23:47 1

Proefschrift_Kirill.indb 41 10-05-15 23:47 Proefschrift_Kirill.indb 42 10-05-15 23:47 Chapter 2 New models of neoplastic progression in Barrett’s oesophagus Authors: Pavlov K.1 and Maley C.C.2

Affiliations: 1Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Molecular and Cellular Oncogenesis Program, Genomics and Computational Biology Graduate Program and the Cell and Molecular Biology Graduate Program, The Wistar Institute, Philadelphia, USA

Biochem Soc Trans. 2010 Apr;38(2):331-6

Proefschrift_Kirill.indb 43 10-05-15 23:47 Abstract

Research in Barrett’s oesophagus, and neoplastic progression to OAC (oesophageal adenocarcinoma), is hobbled by the lack of good pre-clinical models that capture the evolutionary dynamics of Barrett’s cell populations. Current models trade off tractability for realism. Computational models are perhaps the most tractable and can be used both to interpret data and to develop intuitions and hypotheses for neoplastic progression. Tissue culture models include squamous cell lines, Barrett’s oesophagus cell lines and OAC cell lines, although it was recognized recently that BIC-1, SEG-1 and TE-7 are not true OAC cell lines. Some of the unrealistic aspects of the micro-environment in two-dimensional tissue culture may be overcome with the development of three-dimensional organotypic cultures of Barrett’s oesophagus. The most realistic, but least tractable, model is a canine surgical model that generates reflux and leads to an intestinal metaplasia. Alternatively, rat surgical models have gained popularity and should be tested for the common genetic features of Barrett’s oesophagus neoplastic progression in humans including loss of CDKN2A (cyclin-dependent kinase inhibitor 2A) and TP53 (tumor protein 53), generation of aneuploidy and realistic levels of genetic diversity. This last feature will be important for studying the effects of cancer-prevention interventions. In order to study the dynamics of progression and the effects of an experimental intervention, there is a need to follow animals longitudinally, with periodic endoscopic biopsies. This is now possible and represents an exciting opportunity for the future.

Proefschrift_Kirill.indb 44 10-05-15 23:47 Introduction

Barrett’s oesophagus is an intestinal metaplasia of squamous oesophageal epithelium and is important clinically because it increases the chance of progression to OAC (oesophageal adenocarcinoma) 30–125-fold compared with people without Barrett’s oesophagus (1). Injury from persistent gastroduodenal reflux is considered to be the causative agent, and treatment consists of medical therapy, aimed at lowering the frequency and acidity of reflux, and ablation. Unfortunately, even acid suppression by proton pump inhibitors does not usually induce regression of the Barrett’s metaplasia (2,3), and ablative procedures 2 often fail to both completely eliminate all of the Barrett’s oesophagus tissue and to prevent recurrence of Barrett’s dysplasia (4). Furthermore, there is evidence that Barrett’s oesophagus itself is an adaptation to acid reflux and may serve to protect the oesophagus from the development of life-threatening strictures and infections (5–7). Since only 0.7% of people with Barrett’s oesophagus will progress to OAC each year (8), there is a critical need to distinguish patients at high risk of progression from low-risk patients, and to develop non-toxic cancer-prevention strategies. Both of these goals would be substantially aided by the development of good pre-clinical models of Barrett’s oesophagus. Research during the last few decades has shown that neoplastic progression is not just a simple transition from normal to disease state, but a complex dynamic of clonal competition and evolution among somatic cells (9,10). Thus cancer-prevention efforts are essentially efforts to affect and change the evolutionary dynamics of the pre-malignant cells. Because Barrett’s oesophagus can be followed longitudinally with multiple samples at each time point, Barrett’s oesophagus offers a unique opportunity for dissecting the evolutionary process of neoplastic progression, studying the impact of cancer-preventive interventions and extending those results to other cancers. The best pre-clinical models of Barrett’s oesophagus would reproduce the evolutionary dynamics of progression and the impact of interventions. Currently, there are no ideal pre-clinical models of Barrett’s oesophagus that capture the evolutionary dynamics of neoplastic progression. Those that exist range from the most tractable models, such as computational simulations and two-dimensional tissue culture, that lack many of the significant details of the human disease in vivo, to animal models that are not physiologically similar to the human disease and require long periods of time to develop OAC. Important opportunities remain to better characterize our current pre-clinical models and to improve them.

Computational models

Computational models provide the ultimate level of control and information. Every detail of a computational model is by definition both available for observation and modification.

Proefschrift_Kirill.indb 45 10-05-15 23:47 Their main drawback is that many of the details of a biological system are not yet understood, so the representation of those details and dynamics in a computational model is in essence a hypothesis for what may be true of the biological system. In addition, representation of a biological system in a computation simulation requires the abstraction or exclusion of much of the complexity of the biology, and so the generality of the results of the models are always in question. Nevertheless, computational models have proved useful for the exploration of current theories, hypothesis generation, and the discovery of important holes in our understanding that are likely to be critical to the dynamics of the biological system. When one has to write down the details of the essential aspects of a biological system, one quickly realizes how little is known about that system. Computational models can simulate the evolution of somatic cells over decades. We discovered that there is an important (and counterintuitive) interaction between the number of cancer genes (tumour-suppressor genes or oncogenes) that must be mutated in a single allele (dominant mutations) for the development of malignancy, and mutator lesions that increase the rate of (epi)genetic lesions (11). The requirement of more mutations for the development of malignancy actually increased the chance of progression to malignancy, because they provided more opportunities for a mutator lesion to hitchhike on an expansion of the clone with a dominant cancer gene mutation. The generation of a large genetically unstable clone greatly increases the probability of progression in the model. This has been supported in a cohort of Barrett’s oesophagus patients in which the size of a clone with p53 LOH (loss of heterozygosity), aneuploidy or tetraploidy was significantly associated with progression to OAC (12). Recently, we used a computational model to examinethe dynamics of clonal expansion and found that clones on two-dimensional surfaces, like Barrett’s oesophagus, are likely to expand quadratically rather than exponentially (13). This model fitted p53 mutant clone size data from a skin cancer mouse model better than an exponential model of clonal expansion. We also found that the shape of a clone differed depending on whether it had a proliferative fitness advantage or a survival fitness advantage. If the lesion driving the clonal expansion gave the clone a proliferative advantage over its neighbouring clones, then it tended to have a rougher, or concave, border (looked more ‘invasive’) than clones with a survival advantage, which produced a relatively smooth convex shape. A variety of models have been developed to study the dynamics of cells within a crypt (14,15), which probably also apply to Barrett’s oesophagus. Most of these have been used to test the implications of alternative hypotheses for stem cell dynamics and differentiation, as well as lesions that may initiate carcinogenesis with uncontrolled growth (16–21). Earlier models of stem cell dynamics were fitted to data on the conversion of polyclonal murine intestinal crypts into monoclonal crypts as well as crypt density and were used to infer crypt life-cycle dynamics (22–24).

Proefschrift_Kirill.indb 46 10-05-15 23:47 Tissue culture models

Squamous cell culture models Since Barrett’s oesophagus is hypothesized to originate from squamous epithelium, representative squamous cell lines could be of great value in research aimed at inducing transdifferentiation of squamous epithelium towards Barrett’s oesophagus. Using EPC2 cells, an oesophageal epithelial cell line transformed with hTERT (human telomerase reverse transcription), Kong et al. (25) showed that demethylation and CDX2(Caudal- type homeobox 2) expression induced intestinalization. A weakness of this approach is 2 the possibility that Barrett’s oesophagus does not derive from squamous cells.

Barrett’s oesophagus cell culture models Palanca-Wessels et al. (26,27) described the derivation (26) and immortalization using an hTERT transfection (27) of four Barrett’s oesophagus cell lines. Three of those cell lines (CP- B, CP-C and CP-D) were derived from patients with high-grade dysplasia and exhibited lesions in both CDKN2A (cyclin-dependent kinase inhibitor 2A) (p16/INK4A) and TP53 (tumour protein 53) (p53), whereas the fourth cell line (CP-A), from a patient with only metaplastic Barrett’s oesophagus, also had inactivated CDKN2A, but is wildtype for TP53 (26), although it has extensive LOH on chromosome 5q, including APC (adenomatous polyposis coli). Another hTERT immortalized cell line (BAR-T) was described by Jaiswal et al. (28). In contrast with the cell lines described by Palanca-Wessels et al. (26,27), this cell line initially showed both intact CDKN2A and TP53, and lost CDKN2A during adaptations to conditions in vitro (28). Owing to functioning TP53, this cell line may be a good model of early Barrett’s oesophagus progression.

OAC cell culture models A number of OAC cell lines have been described over the years (29–31). However, recent research has called into question the identity of three of the most commonly used OAC cell lines (BIC-1, SEG-1 and TE-7) (29,32). This caused a shortage of reliable DNA-fingerprinted OAC cell lines, although FLO and OE33 appear to be true OAC cell lines. Alvarez et al. (29) described the isolation of JH-EsoAd1, a cell line derived from moderately to poorly differentiated OAC. Because DNA fingerprinting confirmed its identity, JH-EsoAd1 also holds promise for future research in OAC.

Organotypic models

One of the major drawbacks of two-dimensional tissue culture models is that they fail to represent the microenvironment of the tumour, which has been shown to be important in progression (33). In order to mimic aspects of the tumour micro-environment, organotypic

Proefschrift_Kirill.indb 47 10-05-15 23:47 models of oesophageal keratinocytes have been developed (34,35). In these models, the epithelial cells are cultured on top of a layer of collagen and fibroblasts, with medium being fed into the system from below. Such organotypic models have been used to study both squamous cell carcinoma (34) and Barrett’s oesophagus (35) development. Stairs et al. (35) found that co-expression of c-myc and CDX1 of EPC2-hTERT cells in an organotypic model induced early Barrett’s esophagus characteristics such as KRT8 (cytokeratin 8) and MUC5 (mucin 5) expression (35). Organotypic models offer more realistic conditions for studying neoplastic progression than two-dimensional models, and the cells can remain viable and proliferative for more than 3 weeks without passaging. However, they typically lack important aspects of the micro-environment of Barrett’s oesophagus, including inflammatory cells and endothelial cells. Furthermore, they do not form into crypts, and so lack realistic stem cell and differentiation dynamics. Recent work on LGR5 (leucine- rich-repeat-containing G-protein-coupled receptor 5)-positive stem cells in colon crypts has shown that crypts can be induced to develop in a three-dimensional spheroid model (36). This may provide an alternative approach for developing organotypic models of Barrett’s oesophagus.

Animal models

The most popular animals models of Barrett’s esophagus are rat surgical models. There are also some mouse models and an older, but perhaps more physiologically realistic, canine model.

Rat models There are a variety of rat surgical models, some involving a gastrectomy, and often an anastamosis between the jejunum or duodenum and the oesophagus (37,38). In some cases, these models produce more Barrett’s esophagus or OAC with the addition of iron (39). Perhaps the most promising animal model of Barrett’s oesophagus is the OGDA (oesophagogastroduodenal anastamosis) model (40–42). After 40 weeks, 25.6% of the animals develop OAC (41). The OGDA model is particularly intriguing in that the contents of the duodenum are no longer acidic, and so this model shows that bile reflux is sufficient to generate both Barrett’s oesophagus and OAC. This result questions the emphasis on acid reflux and acid suppression in the treatment of Barrett’s oesophagus and the prevention of OAC. Important future questions for the OGDA model include whether or not the genetics of Barrett’s oesophagus and progression to OAC in this model match those observed in the human condition, such as the inactivation of CDKN2A (p16) and TP53 (p53), as well as the development of aneuploidy and tetraploidy (43). The large size of the rat oesophagus

Proefschrift_Kirill.indb 48 10-05-15 23:47 and the feasibility of endoscopic biopsy in this model suggest that the OGDA rats could be followed longitudinally with serial sampling to assess the genetic of progression in this model.

Mouse models There has been some attempt to develop mouse surgical models, similar to the rat model, including an oesophagojejunostomy model with N-methyl-N-benzylnitrosamine, which resulted in 37% of the mice developing OAC (44). However, this model appears to be more 2 difficult to implement than the rat model and has not been as popular. Genetic models have often been used to study the genetic underpinnings of carcinogenesis. There are currently no commonly used genetic models of Barrett’s oesophagus in mice. However, one team has reported the development of intestinal metaplasia of the oesophagus in mice defective for TSP1 (thrombospondin-1) (45). The wide availability of tools and reagents for manipulating mouse models makes this an attractive avenue for research. Unfortunately, the apparent development of Barrett’s oesophagus in the TSP1 mouse model was but one of many observed phenotypes, which did not develop in all cases and did not progress to OAC in the period of study. Both the mouse and rat gastrointestinal tract have significant differences from the human gastrointestinal tract. In particular, rats and mice have a squamous fore-stomach which may result in different dynamics of the development of intestinal metaplasia in the oesophagus compared with humans. Furthermore, the inactivation of a gene throughout the entire organism, or even an entire organ, poorly represents the process of somatic evolution where a gene is inactivated in a single cell and whether or not that clone expands depends on the fitness effects of the inactivated gene on the clone, as well as stochastic dynamics of cell turnover in the tissue.

Dog model One of the earliest animal models of Barrett’s esophagus was a canine model which involved surgical removal of the mucosa of the distal oesophagus as well as generation of a hiatal hernia and cardioplasty, which induced columnar epithelium after 8 weeks (37). This is perhaps the most realistic model of Barrett’s oesophagus since it is driven by reflux and wounding. Experiments with this model, leaving a barrier of squamous cells between the stomach and denuded region of the oesophagus, suggested that Barrett’s oesophagus does not develop by migration of columnar gastric epithelium (46) and that islands of columnar cells seem to develop initially around submucosal glands (47). Unfortunately, the canine model of Barrett’s oesophagus seems to have fallen out of use.

Proefschrift_Kirill.indb 49 10-05-15 23:47 The ideal model

An ideal model of Barrett’s oesophagus would capture the dynamics of progression from normal squamous epithelium to OAC. This would include a transition from squamous to Barrett’s oesophagus and would involve expansions of clones with lesions in CDKN2A as well as TP53 and the development of tetraploidy and aneuploidy late in progression (43). It seems likely that inclusion of the chronic inflammation observed in the human condition will be necessary for a model to replicate the dynamics of the human disease, although this has not yet been shown. Importantly, an ideal model would allow for longitudinal sampling within the same organism so that progression to OAC could be tracked over time. In addition, the model should allow for multiple samples at any one time point so that clonal expansions could be observed. It would be helpful if progression to OAC was both reliable and rapid, although accelerated progression may be incompatible with realistic evolutionary dynamics of progression. Finally, the genetic diversity of human Barrett’s oesophagus should be captured in the model so that realistic responses to interventions may be modelled. It may be the case that the OGDA rat model or the dog model may fit all (or most) of these criteria. Genetic characterization of the animal models will be important for their validation.

Major open questions

Good pre-clinical models of Barrett’s oesophagus would facilitate the investigation of a host of important questions in the field. A basic question in cancer biology for any tumour is to determine the sets of (epi)genetic lesions that are both necessary and sufficient to generate a malignancy in sporadic cancers. Genome-wide analyses of longitudinal samples from the same animal or patient would facilitate the discovery of potential lesions that could then be tested by introducing them sequentially into the Barrett’s oesophagus cells of an animal model. Observational studies, using genetic or epigenetic markers in multiple biopsies from each time point of a longitudinal study, could also be used to measure the frequency of clonal expansions during neoplastic progression. There is considerable disagreement in the literature over the frequency of such expansions (e.g. 1 compared with 20) (48,49), and very little data to resolve the debate. Associating particular lesions with those clonal expansions would also help to determine which of the necessary lesions in neoplastic progression increase the fitness of a clone, and which are evolutionarily neutral or deleterious. Lesions that cause clonal expansions make for good biomarkers that can be detected with few samples and can dramatically increase the probability of further progression (11, 12). Because clonal expansions dramatically increase the probability of progression to malignancy and can lead to secondary cancers, the inhibition of clonal expansions

Proefschrift_Kirill.indb 50 10-05-15 23:47 represents an attractive approach to cancer prevention. A first critical step in such efforts will be to discover the mechanism of such expansions. Does a clone expand by crypt bifurcation/budding, and, if so, do the new crypts replace their neighbours by some mechanism, or just generate more densely packed crypts? Alternatively, clones may spread by surviving the wounding effects of reflux and then repopulating the denuded regions (13,50). In addition, local ‘metastases’ of stem cells within a tissue have been hypothesized to spread a clone (51). Because neoplastic progression is a process of somatic evolution, cancer prevention 2 is fundamentally an attempt to have an impact on and control that evolutionary process. However, the evolutionary effects of our interventions are virtually unstudied. A critical question remains what (epi)genetic lesions are selected by an intervention? That is, what (epi)genetic lesions provide a relative fitness advantage in the altered micro-environment induced by the intervention? Some of those lesions may generate resistance to the cancer- preventive intervention, and an understanding of the mechanism of that resistance should help us to identify patients unlikely to benefit from the intervention and to develop second line or multidrug cancer-prevention treatments that can overcome the evolution of resistance (9,52).One approach to cancer prevention would be to alter the fitness of oesophageal squamous cells relative to Barrett’s oesophagus cells, so that the squamous epithelium could out-compete and replace the Barrett’s oesophagus tissue. This might be done by the traditional approach of reducing the fitness of Barrett’s oesophagus cells below the fitness of squamous cells or by increasing the fitness of squamous cells above that of Barrett’s oesophagus cells (53).

Conclusions

Current pre-clinical models of Barrett’s oesophagus range from the most tractable and least realistic computational and tissue culture models to a set of physiologically unrealistic animal models (with the possible exception of the canine model). Some of these models show promise, but will require further genetic and epigenetic characterization to validate their utility for studying the human disease. Important opportunities remain for both improving upon current models and the development of new models of Barrett’s oesophagus. Such pre-clinical models hold the promise of advancing both our knowledge and management of Barrett’s oesophagus, but will only be truly effective if they are used in longitudinal studies that can characterize the somatic evolution that drives progression and the effects of cancer-preventive interventions.

Proefschrift_Kirill.indb 51 10-05-15 23:47 Funding

This work was supported in part by the London AACR (American Association for Cancer Research) Innovator Award for Cancer Prevention and the National Institutes of Health [grant numbers R03 CA137811, P01 CA91955, P30 CA010815, R01 CA119224 and R01 CA140657].

Proefschrift_Kirill.indb 52 10-05-15 23:47 References

(1) Haggitt, R.C. (1994) Barrett’s esophagus, dysplasia, and adenocarcinoma. Hum. Pathol. 25, 982–993 (2) Jankowski, J.A. and Anderson, M. (2004) Management of oesophageal adenocarcinoma: control of acid, bile and inflammation in intervention strategies for Barrett’s oesophagus. Aliment. Pharmacol. Ther. 20, (Suppl. 5), 71–80 (3) Leedham, S. and Jankowski, J. (2007) The evidence base of proton pump inhibitor chemopreventative agents in Barrett’s esophagus: the good, the bad, and the flawed! Am. J. Gastroenterol. 102, 21–23 2 (4) Adler, D.C., Zhou, C., Tsai, T.H., Lee, H.C., Becker, L., Schmitt, J.M., Huang, Q., Fujimoto, J.G. and Mashimo, H. (2009) Three-dimensional opticalcoherence tomography of Barrett’s esophagus and buried glandsbeneath neosquamous epithelium following radiofrequency ablation. Endoscopy 41, 773–776 (5) Orlando, R.C. (2006) Mucosal defense in Barrett’s esophagus. In Barrett’s Esophagus and Esophageal Adenocarcinoma (Sharma, P. and Sampliner, R., eds) pp. 60–72, Blackwell Publishing, Oxford (6) Jovov, B., Van Itallie, C.M., Shaheen, N.J., Carson, J.L., Gambling, T.M., Anderson, J.M. and Orlando, R.C. (2007) Claudin-18: a dominant tight junction protein in Barrett’s esophagus and likely contributor to its acid resistance. Am. J. Physiol. Gastrointest. Liver Physiol. 293, G1106– G1113 (7) Tobey, N.A., Argote, C.M., Vanegas, X.C., Barlow, W. and Orlando, R.C. (2007) Electrical parameters and ion species for active transport in human esophageal stratified squamous epithelium and Barrett’s specialized columnar epithelium. Am. J. Physiol. Gastrointest. Liver Physiol. 293, G264–G270 (8) Thomas, T., Abrams, K.R., De Caestecker, J.S. and Robinson, R.J. (2007) Meta analysis: cancer risk in Barrett’s oesophagus. Aliment. Pharmacol. Ther. 26, 1465–1477 (9) Merlo, L.M., Pepper, J.W., Reid, B.J. and Maley, C.C. (2006) Cancer as an evolutionary and ecological process. Nat. Rev. Cancer 6, 924–935 (10) Pepper, J., Findlay, C.S., Kassen, R., Spencer, S. and Maley, C. (2009) Cancer research meets evolutionary biology. Evol. Appl. 2, 62–70 (11) Maley, C.C. and Forrest, S. (2000) Exploring the relationship between neutral and selective mutations in cancer. Artif. Life 6, 325–345 (12) Maley, C.C., Galipeau, P.C., Li, X., Sanchez, C.A., Paulson, T.G., Blount, P.L. and Reid, B.J. (2004) The combination of genetic instability and clonal expansion predicts progression to esophageal adenocarcinoma. Cancer Res. 64, 7629–7633 (13) Chao, D.L., Eck, J.T., Brash, D.E., Maley, C.C. and Luebeck, E.G. (2008) Preneoplastic lesion growth driven by the death of adjacent normal stem cells. Proc. Natl. Acad. Sci. U.S.A. 105, 15034–15039 (14) van Leeuwen, I.M., Byrne, H.M., Jensen, O.E. and King, J.R. (2006) Crypt dynamics and colorectal cancer: advances in mathematical modelling. Cell Prolif. 39, 157–181 (15) van Leeuwen, I.M., Edwards, C.M., Ilyas, M. and Byrne, H.M. (2007) Towards a multiscale model of colorectal cancer. World J. Gastroenterol. 13, 1399–1407 (16) Pepper, J.W., Sprouffske, K. and Maley, C.C. (2007) Animal cell differentiation patterns suppress

Proefschrift_Kirill.indb 53 10-05-15 23:47 somatic evolution. PLoS Comput. Biol. 3, e250 (17) Loeffler, M. and Roeder, I. (2004) Conceptual models to understand tissue stem cell organization. Curr. Opin. Hematol. 11, 81–87 (18) Johnston, M.D., Edwards, C.M., Bodmer, W.F., Maini, P.K. and Chapman, S.J. (2007) Mathematical modeling of cell population dynamics in the colonic crypt and in colorectal cancer. Proc. Natl. Acad. Sci. U.S.A. 104, 4008–4013 (19) d’Onofrio, A. and Tomlinson, I.P. (2007) A nonlinear mathematical model of cell turnover, differentiation and tumorigenesis in the intestinal crypt. J. Theor. Biol. 244, 367–374 (20) Boman, B.M., Fields, J.Z., Cavanaugh, K.L., Guetter, A. and Runquist, O.A. (2008) How dysregulated colonic crypt dynamics cause stem cell overpopulation and initiate colon cancer. Cancer Res. 68, 3304–3313 (21) Meineke, F.A., Potten, C.S. and Loeffler, M. (2001) Cell migration and organization inthe intestinal crypt using a lattice-free model. Cell Prolif. 34, 253–266 (22) Loeffler, M., Birke, A., Winton, D. and Potten, C. (1993) Somatic mutation, monoclonality and stochastic models of stem cell organization in the intestinal crypt. J. Theor. Biol. 160, 471–491 (23) Loeffler, M., Bratke, T., Paulus, U., Li, Y.Q. and Potten, C.S. (1997) Clonality and life cycles of intestinal crypts explained by a state dependent stochastic model of epithelial stem cell organization. J. Theor. Biol. 186, 41–54 (24) Totafurno, J., Bjerknes, M. and Cheng, H. (1987) The crypt cycle: crypt and villus production in the adult intestinal epithelium. Biophys. J. 52, 279–294 (25) Kong, J., Nakagawa, H., Isariyawongse, B.K., Funakoshi, S., Silberg, D.G., Rustgi, A.K. and Lynch, J.P. (2009) Induction of intestinalization in human esophageal keratinocytes is a multistep process. Carcinogenesis 30, 122–130 (26) Palanca-Wessels, M.C., Barrett, M.T., Galipeau, P.C., Rohrer, K.L., Reid, B.J. and Rabinovitch, P.S. (1998) Genetic analysis of long-term Barrett’s esophagus epithelial cultures exhibiting cytogenetic and ploidy abnormalities. Gastroenterology 114, 295–304 (27) Palanca-Wessels, M.C., Klingelhutz, A., Reid, B.J., Norwood, T.H., Opheim, K.E., Paulson, T.G., Feng, Z. and Rabinovitch, P.S. (2003) Extended lifespan of Barrett’s esophagus epithelium transduced with the human telomerase catalytic subunit: a useful in vitro model. Carcinogenesis 24, 1183–1190 (28) Jaiswal, K.R., Morales, C.P., Feagins, L.A., Gandia, K.G., Zhang, X., Zhang, H.Y., Hormi-Carver, K., Shen, Y., Elder, F., Ramirez, R.D. et al. (2007) Characterization of telomerase-immortalized, non-neoplastic, human Barrett’s cell line (BAR-T). Dis. Esophagus 20, 256–264 (29) Alvarez, H., Koorstra, J.B., Hong, S.M., Boonstra, J.J., Dinjens, W.N., Wu, T.T., Montgomery, E., Eshleman, J.R. and Maitra, A. (2008) Establishment and characterization of a bona fide Barrett esophagus-associated adenocarcinoma cell line. Cancer Biol. Ther. 7, 1753–1755 (30) Soldes, O.S., Kuick, R.D., Thompson, II, I.A., Hughes, S.J., Orringer, M.B., Iannettoni, M.D., Hanash, S.M. and Beer, D.G. (1999) Differential expression of Hsp27 in normal oesophagus, Barrett’s metaplasia and oesophageal adenocarcinomas. Br. J. Cancer 79, 595–603 (31) Souza, R.F., Shewmake, K., Beer, D.G., Cryer, B. and Spechler, S.J. (2000) Selective inhibition of cyclooxygenase-2 suppresses growth and induces apoptosis in human esophageal adenocarcinoma cells. Cancer Res. 60, 5767–5772 (32) Boonstra, J.J., van der Velden, A.W., Beerens, E.C., van Marion, R., Morita-Fujimura, Y., Matsui,

Proefschrift_Kirill.indb 54 10-05-15 23:47 Y., Nishihira, T., Tselepis, C., Hainaut, P., Lowe, A.W. et al. (2007) Mistaken identity of widely used oesophageal adenocarcinoma cell line TE-7. Cancer Res. 67, 7996–8001 (33) Beacham, D.A. and Cukierman, E. (2005) Stromagenesis: the changing face of fibroblastic microenvironments during tumor progression. Semin. Cancer Biol. 15, 329–341 (34) Okawa, T., Michaylira, C.Z., Kalabis, J., Stairs, D.B., Nakagawa, H., Andl, C.D., Johnstone, C.N., Klein-Szanto, A.J., El-Deiry, W.S., Cukierman, E. et al. (2007) The functional interplay between EGFR overexpression, hTERT activation, and p53 mutation in esophageal epithelial cells with activation of stromal fibroblasts induces tumor development, invasion, and differentiation. Genes Dev. 21, 2788–2803 (35) Stairs, D.B., Nakagawa, H., Klein-Szanto, A., Mitchell, S.D., Silberg, D.G., Tobias, J.W., Lynch, J.P. 2 and Rustgi, A.K. (2008) Cdx1 and c-Myc foster the initiation of transdifferentiation of the normal esophageal squamous epithelium toward Barrett’s esophagus. PLoS ONE 3, e3534 (36) Sato, T., Vries, R.G., Snippert, H.J., van de Wetering, M., Barker, N., Stange, D.E., van Es, J.H., Abo, A., Kujala, P., Peters, P.J. and Clevers, H. (2009) Single Lgr5 stem cells build crypt-villus structures in vitro without a mesenchymal niche. Nature 459, 262–265 (37) Koak, Y. and Winslet, M. (2002) Changing role of in vivo models in columnar-lined lower esophagus. Dis. Esophagus 15, 271–277 (38) Li, Y. and Martin, 2nd, R.C. (2007) Reflux injury of esophageal mucosa: experimental studies in animal models of esophagitis, Barrett’s esophagus and esophageal adenocarcinoma. Dis. Esophagus 20, 372–378 (39) Goldstein, S.R., Yang, G.Y., Chen, X., Curtis, S.K. and Yang, C.S. (1998) Studies of iron deposits, inducible nitric oxide synthase and nitrotyrosine in a rat model for esophageal adenocarcinoma. Carcinogenesis 19, 1445–1449 (40) Chen, X., Wang, S., Wu, N., Sood, S., Wang, P., Jin, Z., Beer, D.G., Giordano, T.J., Lin, Y., Shih, W.C. et al. (2004) Overexpression of 5-lipoxygenase in rat and human esophageal adenocarcinoma and inhibitory effects of zileuton and celecoxib on carcinogenesis. Clin. Cancer Res. 10, 6703– 6709 (41) Chen, X., Yang, G., Ding, W.Y., Bondoc, F., Curtis, S.K. and Yang, C.S. (1999) An esophagogastroduodenal anastomosis model for oesophageal adenocarcinogenesis in rats and enhancement by iron overload. Carcinogenesis 20, 1801–1808 (42) Su, Y., Chen, X., Klein, M., Fang, M., Wang, S., Yang, C.S. and Goyal, R.K. (2004) Phenotype of columnar-lined esophagus in rats with esophagogastroduodenal anastomosis: similarity to human Barrett’s esophagus. Lab. Invest. 84, 753–765 (43) Galipeau, P.C., Li, X., Blount, P.L., Maley, C.C., Sanchez, C.A., Odze, R.D., Ayub, K., Rabinovitch, P.S., Vaughan, T.L. and Reid, B.J. (2007) NSAIDs modulate CDKN2A, TP53, and DNA content risk for progression to esophageal adenocarcinoma. PLoS Med. 4, e67 (44) Xu, X., LoCicero, 3rd, J., Macri, E., Loda, M. and Ellis, Jr, F.H. (2000) Barrett’s esophagus and associated adenocarcinoma in a mouse surgical model. J. Surg. Res. 88, 120–124 (45) Crawford, S.E., Stellmach, V., Murphy-Ullrich, J.E., Ribeiro, S.M., Lawler, J., Hynes, R.O., Boivin, G.P. and Bouck, N. (1998) Thrombospondin-1 is a major activator of TGF-β1 in vivo. Cell 93, 1159–1170 (46) Gillen, P., Keeling, P., Byrne, P.J., West, A.B. and Hennessy, T.P. (1988) Experimental columnar metaplasia in the canine oesophagus. Br. J. Surg. 75, 113–115 (47) Pollara, W.M., Cecconello, I., Zilberstein, B., Iria, K. and Pinotti, H.W. (1983) Regeneration of

Proefschrift_Kirill.indb 55 10-05-15 23:47 esophageal epithelium in the presence of gastroesophageal reflux. Arq. Gastroenterol. 20, 53–59 (48) Meza, R., Jeon, J., Moolgavkar, S.H. and Luebeck, E.G. (2008) Age-specific incidence of cancer: phases, transitions, and biological implications. Proc. Natl. Acad. Sci. U.S.A. 105, 16284–16289 (49) Beerenwinkel, N., Antal, T., Dingli, D., Traulsen, A., Kinzler, K.W., Velculescu, V.E., Vogelstein, B. and Nowak, M.A. (2007) Genetic progression and the waiting time to cancer. PLoS Comput. Biol. 3, e225 (50) Zhang, W., Hanks, A.N., Boucher, K., Florell, S.R., Allen, S.M., Alexander, A., Brash, D.E. and Grossman, D. (2005) UVB-induced apoptosis drives clonal expansion during skin tumor development. Carcinogenesis 26, 249–257 (51) van Oijen, M.G. and Slootweg, P.J. (2000) Oral field cancerization: carcinogen-induced independent events or micrometastatic deposits? Cancer Epidemiol. Biomarkers Prev. 9, 249–256 (52) Potter, J.D. (1997) β-Carotene and the role of intervention studies. Cancer Lett. 114, 329–331 (53) Maley, C.C., Reid, B.J. and Forrest, S. (2004) Cancer prevention strategies that address the evolutionary dynamics of neoplastic cells: simulating benign cell boosters and selection for chemosensitivity. Cancer Epidemiol. Biomarkers Prev. 13, 1375–1384

Proefschrift_Kirill.indb 56 10-05-15 23:47 2

Proefschrift_Kirill.indb 57 10-05-15 23:47 Proefschrift_Kirill.indb 58 10-05-15 23:47 Chapter 3 Development and Characterization of an Organotypic Model of Barrett’s Esophagus Authors: Kosoff R.E.1, Gardiner K.L.2, Merlo L.M.F.2, Pavlov K.3, Rustgi A.K.4, Maley C.C.5

Affiliations: 1Cancer Biology Program, University of Pennsylvania, Philadelphia, Pennsylvania, USA 2Molecular and Cellular Oncogenesis Program, The Wistar Institute, Philadelphia, Pennsylvania,USA 3Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 4Gastroenterology Division, Departments of Medicine and Genetics, Abramson Cancer Center, University of Pennsylvania, Philadelphia, Pennsylvania, USA 5UCSF Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, San Francisco, California, USA

J Cell Physiol. 2012 Jun;227(6):2654-9

Proefschrift_Kirill.indb 59 10-05-15 23:47 Abstract

Understanding the molecular and cellular processes underlying the development, maintenance, and progression of Barrett’s oesophagus (BE) presents an empirical challenge because there are no simple animal models and standard 2D cell culture can distort cellular processes. Here we describe a three-dimensional (3D) cell culture system to study BE. BE cell lines (CP-A, CP-B, CP-C, and CP-D) and oesophageal squamous keratinocytes (EPC2) were cultured on a matrix consisting of esophageal fibroblasts and collagen. Comparison of growth and cytokeratin expression in the presence of all-trans retinoic acid or hydrochloric acid was made by immunohistochemistry and Alcian Blue staining to determine which treatments produced a BE phenotype of columnar cytokeratin expression in 3Dculture. All-trans retinoic acid differentially affected the growth of BE cell lines in 3D culture. Notably, the non-dyplastic metaplasia-derived cell line (CP-A) expressed reduced squamous cytokeratins and enhanced columnar cytokeratins upon ATRA treatment. ATRA altered the EPC2 squamous cytokeratin profile towards a more columnar expression pattern. Cell lines derived from patients with high-grade dysplasia already expressed columnar cytokeratins and therefore did not show a systematic shift toward a more columnar phenotype with ATRA treatment. ATRA treatment, however, did reduce the squamoid-like multilayer stratification observed in all cell lines. As the first study to demonstrate long-term 3D growth of BE cell lines, we have determined that BE cells can be cultured for at least 3 weeks on a fibroblast/ collagen matrix and that the use of ATRA causes a general reduction in squamous-like multilayered growth and an increase in columnar phenotype with the specific effects cell-line dependent.

Proefschrift_Kirill.indb 60 10-05-15 23:47 Introduction

Mechanistic studies of cancer initiation, progression, and therapeutic discovery require cell culture and/or animal model systems. Cell culture systems have typically grown cells on plastic, and often these results conflict within vivo experiments. Two-dimensional culture models do not capture many effects of the tissue microenvironment, including cell morphology, polarity and junctions, interactions with the extracellular matrix (ECM) and adjacent cells, and mechanotransduction, which affect intracellular signalling and cell fate (1). Three-dimensional (3D) co-culture, consisting of a substratum prepared with collagen and fibroblast upon which epithelial cells are seeded, better imitates the in vivo tissue microenvironment, and may help to bridge the gap between traditional cell culture systems and animal studies. Interest in the premalignant condition Barrett’s oesophagus (BE) stems from its association with oesophageal adenocarcinoma (EA). BE is the only known precursor of EA, 3 increasing the relative risk of progression 30- to 120-fold (2). The incidence of EA among individuals with BE is 6–7 per 1,000 person-years (3,4). BE is a condition of the esophagus associated with acid and bile reflux where normal multilayer, stratified oesophageal squamous tissue is replaced by a single layer of columnar cells with a crypt architecture (specialized intestinal metaplasia) (5). The presence of mucous-producing goblet cells in BE is thought to protect the esophagus from insult of acid and bile (5). However, this metaplasia increases a patient’s risk of cancer, likely because constant reflux increases cellular proliferation, telomere shortening, and DNA damage, resulting in chromosomal instability and genomic alterations (5). Building on 3D studies of skin and normal oesophageal epithelium (6,7), we provide the first 3D cell culture model of BE, allowing us to overcome some shortcomings associated with other models. Currently, the major BE animal model is a surgical anastomosis model in the rat, which creates reflux by connecting the jejunum of the small intestine with the distal esophagus (8), allowing the development of intestinal metaplasia. This model has various limitations, including atypical bile reflux, species differences, duration, expense, and the requirement of surgical expertise. Other studies have used explant tissue biopsies to study BE in more physiological conditions compared with 2D cell culture (9,10), however, biopsies are viable for only a short period of time ex vivo, preventing observation of cellular transformation, such as goblet cell formation, under different culture conditions. To develop this organotypic reconstruct model, we grew hTERT-immortalized BE cell lines, derived from BE patients (11,12) and a control normal squamous esophageal cell line, EPC2-hTERT (13), on a fibroblast-collagen matrix. The reconstructs were examined for the presence of goblet cells and various squamous and columnar cytokeratin markers,

Proefschrift_Kirill.indb 61 10-05-15 23:47 a reliable tool to describe cell type (14). Proper expression of cytokeratins in cell culture depends strongly on the method for culturing, including medium type, serum content, calcium concentration, and the presence of a feeder layer (15,16). Here, we studied the effects of acid pulses to simulate gastric reflux and continuous exposure to all-trans retinoic acid (ATRA) on expression of cytokeratins and morphology of BE cells in 3D culture.

Materials and Methods

Cell lines Four hTERT-transformed BE cell lines, CP-A (p53 wild-type, 9p loss of heterozygosity (LOH), 5q LOH), CP-B (17q LOH, 9p LOH), CP-C (17q LOH, 9p LOH), and CP-D (17p LOH, 9p homozygous deletion) (11,12) were obtained from University of Washington, Seattle, WA (gift from P. Rabinovich). Cells were adapted to serum-free conditions in keratinocyte serum-free medium (KSFM, Invitrogen, Carlsbad, CA). BE cell line identities were verified with the STR-based identifier PCR Amplification kit (Applied Biosystems, Carlsbad, CA) by comparison to original samples. EPC2-hTERT and primary fetal esophageal fibroblasts have been described previously (6,13).

Cell culture Organotypic culture was performed as previously published for esophageal squamous cell reconstructs (6) with some modifications. The fibroblast feeder layer was incubated in bovine collagen for 4 days, after which 5105 epithelial cells were seeded on top. Media used for epithelial cell growth on the feeder layer was the same as previously published (6) with 2% FBS (Atlanta Biologicals, Lawrenceville, CA). Fresh medium was added every 2–3 days. After seeding of the epithelial cells, medium was maintained on both the top and bottom of the 0.4 mm polycarbonate tissue-culture treated transwell membrane (Corning, Lowell, MA) for an additional 5 days, after which the media on top of the transwell was removed, while maintaining media on the bottom. At this point, epithelial cells were exposed to a liquid–air interface for the remainder of the experiment to enhance differentiation. Organotypic cultures were sampled weekly, fixed in 10% neutral buffered formaldehyde and embedded in paraffin for H&E and Alcian Blue staining.

All-trans retinoic acid organotypic culture Culture conditions were as described above with some alterations (Supplemental Methods). ATRA was supplemented in culture medium at a concentration of 0.33 mM and replenished every 2 days.

Proefschrift_Kirill.indb 62 10-05-15 23:47 Acid pulsing on organotypic culture Five days after epithelial seeding media was removed and replaced with pH 3.5 PBS for 1 h every other day (3 days per week) in the top and bottom chambers. The acid was washed out of the chambers with PBS and fresh media was replaced. Samples were taken for fixation 24 h after the last pulse. Week 1 samples received 3 pulses, week 2 received 6 pulses, and week 3 received 10 pulses total.

Immunohistochemistry Immunohistochemical staining was performed on 5mm sections of paraffin-embedded reconstructs. Overnight incubation with primary antibodies was done at 48Cwith the following: cytokeratin 13 (NovaCastra/Leica Microsystems, Bannockburn, IL), cytokeratin 19 (TROMA-III, University of Iowa Developmental Studies Hybridoma Bank), cytokeratin 8 (RDI-PRO61038, Research Diagnostics, Acton, MA), and cytokeratin 14 (PRB155P, 3 Covance, Princeton, NJ). Slide development was completed using DAB substrate kit for peroxidases (Vector Laboratories, Burlingame, CA), with development times monitored with a dissecting microscope. Hematoxylin counterstain was then applied. In order to determine the specificity of an antibody towards BE in 3D culture, each antibody was tested on samples of Barrett’s metaplasia, gastric and squamous tissue from four patients provided by the laboratory of Dr. Brian Reid (Fred Hutchinson Cancer Research Center, Seattle, WA). Antibodies were chosen based on specificity towards BE or squamous tissue. Slides were mounted with cytomount, and imaged using a Nikon E600 Upright microscope with ACT-1 software. Multiple images from each slide were taken and independently evaluated. REK, KLG, and LMFM independently scored each image based on proportion of positive cells and intensity of staining. Any discrepancies were resolved by examination of the original slides and, when necessary, repeating the staining procedure. Similarities of 3D BE cell culture with metaplastic BE tissue were evaluated by comparing patient biopsies of BE with 3D culture of BE using both IHC for specific columnar and squamous cytokeratin expression, hematoxylin and eosin for tissue architecture and Alcian blue for the presence of goblet cells.

Results

The patient-derived BE cell lines CP-A, CP-B, CP-C, and CP-D grown on a fibroblast/ collagen matrix vary widely in their phenotype and growth pattern. Co-culture of epithelial cells with fibroblast-embedded collagen allows for in vitro growth with more similarities to in vivo tissue than with 2D growth. CPB and CP-D cells, both derived from a region of high-grade dysplasia (HGD), grow as an even multilayered stratified epithelium, similar to squamous keratinocytes, but without keratin production (Fig. 1A,C,E). Additionally,

Proefschrift_Kirill.indb 63 10-05-15 23:47 CP-D cells have a mildly invasive phenotype when grown on a fibroblasts/collagen matrix. CP-C cells are also derived from a region of HGD but typically do not form a multilayered stratified epithelium like CP-B and CP-D over the 3-week course ofthe experiment. However, occasionally there are small patches where CP-C cells grow in more than one layer, possibly due to patchiness of seeding (Fig. 1D). The remaining cell line, CP-A, derived from a region of non-dysplastic metaplasia, is distinct from the other cell lines (Fig. 1B). CP-A cells grown in 3D culture form goblet cells, invade into the matrix, and express both squamous and columnar cytokeratins. BE cell culture on a fibroblast/collagen matrix was adapted to culture conditions originally developed for keratinocyte growth (6). Therefore, in this study we identify growth conditions necessary for the appropriate culture of BE cell lines to simulate in vivo growth. BE cells in vivo grow as a single layer of columnar epithelium with crypt architecture, with some goblet cells depending on the state of disease. BE cells cultured under conditions developed for keratinocyte growth resulted in a stratified epithelium. This study attempts to modify the stratified epithelium and squamous cytokeratin expression profile seen in control conditions with the differentiation factor, all-trans retinoic acid (ATRA), and physiologically by recreating gastric reflux with acid pulsing to promote more typical BE growth.

Effect of all-trans retinoic acid (ATRA) We evaluated the effect of all-trans retinoic acid, the irreversibly oxidized form of vitamin A, on BE cells in 3D culture over 3 weeks for expression of squamous and columnar cytokeratins, goblet cells, and altered cell morphology. Cells exposed to ATRA did not grow in the typical stratified pattern as seen in the absence of ATRA (Fig. 1; Supplemental Fig. 1). Without ATRA exposure, most cell lines formed a thick epithelium, as much as 10 cells thick (Fig. 1, ATRA). With the addition of ATRA, the squamoid stratified 3D structure of these cells was inhibited, and most cell lines (with the exception of CP-D) grew only 1–2 cells thick on the fibroblast/collagen matrix (Fig. 1, +ATRA). Even in the CP-D cell line, stratification was greatly reduced by ATRA treatment (Fig. 1E, +ATRA). EPC2 treated with ATRA resulted in reduced superficial keratin (Fig. 1A). Additionally, keratin production was evident by week 3 in CP-A cultures without ATRA and absent with ATRA (Fig. 1B, Supplemental Fig. 1C). Goblet cells were detected in very abundant numbers in CP-A cells in control conditions, however, with ATRA treatment, their numbers diminished greatly (Fig. 2). Reconstructs of cells derived from HGD lack goblet cells, consistent with the in vivo observation that BE goblet cells are typically present in the non-dysplastic metaplasia phase of BE and not in HGD (17).

Proefschrift_Kirill.indb 64 10-05-15 23:47

Figure 1

Figure 1: H&E staining of organotypic reconstructs demonstrate ATRA generally reduces epithelial thickness in BE and squamous cells at 3 weeks (see brackets). A:EPC2 cell line, (B) CP-A cell line, showing invasion (arrows) and67 goblet cells, (C) CP-B cell line, (D) CP-C cell line, (E) CP-D cell line. Scale bar=75μm

Squamous specific cytokeratins and ATRA CK13 and CK14 are typically expressed in the suprabasal layer and epidermis of squamous tissue, and therefore were used in this study to identify changes in the squamous phenotype with ATRA. The results demonstrate that at 3 weeks of growth on a fibroblast/ collagen matrix with ATRA, EPC2 cells (Fig. 3A) and CP-A cells (Fig. 3B; Supplemental Fig. 2A,B) show an appreciable decrease in squamous cytokeratin CK13 expression. It

Proefschrift_Kirill.indb 65 10-05-15 23:47 should be noted that there was no CK13 expression in CP-B, CP-C, and CP-D cells in control conditions, and no change in expression with ATRA was detected (Supplemental Fig. 2C–E). CK14 expression in CP-A cells is detected in control conditions, and completely abolished at 3 weeks with ATRA (Fig. 3C). Contrary to our expectations, CP-B and CP-D both showed slightly increased expression of CK14 with ATRA at week 3 (Fig. 3D,E). EPC2 expression of CK14 is not altered with ATRA treatment, and the CP-C cell line does not express CK14 (Supplemental Fig. 2F,I). Taken together, these results indicate that ATRA can reduce the squamous phenotype observed in the CP-A cells in 3D culture. CP-A cells were the only BE cell line studied here which expressed squamous cytokeratins (CK13 and CK14) under control conditions.

Figure 2

Figure 2: Alcian Blue staining of CP-A organotypic reconstructs at 3 weeks with or without ATRA (A), demonstrates presence of goblet cells preferentially in control conditions. Scale bar=100μm; arrows denote cytokeratin-positive cells.

Columnar specific cytokeratins and ATRA Columnar cytokeratins 8 and 19 (CK8 and CK19) are expressed in glandular, non-squamous

epithelium including BE (18), and therefore were used to analyze whether ATRA caused a shift towards a columnar phenotype in squamous and BE cells. Here, all five cell lines

expressed CK19 under control conditions. ATRA treatment with CP-D cells resulted in decreased CK19 expression along with increased patchiness of expression throughout the

reconstruct (Fig. 4A). EPC2 keratinocytes express CK19 in the basal layers, when cultured

with ATRA there is a reduction in stratified epithelium towards a single layer of CK19 positive cells (Fig. 4B). Expression of CK19 was not affected by ATRA exposure in the CP-

A, CP-B, and CP-C cell lines (Supplemental Fig. 3A–E). Evaluation of cytokeratin 8 expression in BE cells demonstrated that cells from

HGD (CP-B, CP-C, and CP-D) express moderate to high levels of CK8, independent of ATRA treatment (Supplemental Fig. 3F–H). CP-A cells express a constant low level of

CK8 with ATRA during weeks 1 and 2, however, by 3 weeks CP-A cells grown in the absence of ATRA also express CK8 (Fig. 4D,E). Therefore, there are temporal changes in

68 Proefschrift_Kirill.indb 66 10-05-15 23:47

expression of columnar cytokeratins in 3D culture, supporting the hypothesis that long- term culturing of these cells may be critical to create a system more similar to in vivo conditions. Additionally, ATRA may play a role in transition from squamous to columnar phenotype as evidenced in the treatment on EPC2 cells. Typically, EPC2 cells express no CK8, however, with ATRA treatment slight CK8 expression was detected (Fig. 4C).

Figure 3

Figure 3: Squamous cytokeratin expression is reduced with ATRA treatment in squamous and CP- A cells, but not in high-grade dysplasia. A: EPC2 CK13, (B) CP-A CK13, (C) CP-A CK14, (D) CP-B CK14, (E) CP-D CK14. Scale69 bar=100mm; arrows denote cytokeratin-positive cells.

Proefschrift_Kirill.indb 67 10-05-15 23:47 In summary, the effects of ATRA on cytokeratin expression and non-stratified growth are greatest on the growth pattern of CP-A cells derived from non-dysplastic metaplasia, rather than cells from HGD. Changes observed in the metaplasia-derived CP-A cell line with ATRA indicate a transition from a squamous - like phenotype to a more columnar, BE-like phenotype. This effect is not generally applicable to cell lines derived

from HGD, though multilayer stratification was inhibited by ATRA in all cell lines. Figure 4

Figure 4: Columnar CK19 expression in ATRA-treated CP-D cells and CK8 in CP-A cells were diminished, without measurable differences in other BE cells. A: CP-D week 3 CK19, (B) EPC2 week 3 CK19, (C) EPC2 week 3 CK8,70 (D) CP-A week 1 CK8, (E) CP-A week 3 CK8. Scale bar=100μm; arrows denote cytokeratin-positive cells.

Proefschrift_Kirill.indb 68 10-05-15 23:47 Effect of pH 3.5 hydrochloric acid pulses To study the effects of long-term acid reflux and BE in vitro, we cultured cells for 3 weeks on a fibroblast/collagen matrix to study morphological and cytokeratin changes to the CP-A cell line, derived from non-dysplastic metaplasia cells. Cultures were pulsed with pH 3.5 PBS for 1 h every other day. Acidic PBS was added to the top and bottom of the transwell chamber. Reconstructs were measured for CK8, 13, 14, and 19 expression by IHC and morphology measured with H&E over the course of 3 weeks. Some morphological differences were observed with acid. Frequent acid insult to CP-A cells resulted ina temporary reduction in cell number and epithelial thickness, however, by week 3, there was no observable difference, indicating perhaps an adaptation to acid flux (Fig. 5A, Supplemental Fig. 4A–D,F–K).

Squamous cytokeratins and acid 3 As demonstrated in Figure 3B,C, CP-A cells express squamous cytokeratins CK13 and 14 in the 3D model used here. Acid exposure does not alter the squamous cytokeratin profile of CP-A cells. CK13 expression in acid pulsed CP-A cells shows no difference

Figure 5

Figure 5: CP-A organotypic reconstructs demonstrated no effect of acid pulsing over 3 weeks on squamous and columnar cytokeratin expression. A: CP-A week 3 H&E, (B) CP-A week 1 CK14, (C) CP-A week 3 CK14. Scale bar=100μm; arrows denote cytokeratin-positive cells.

Proefschrift_Kirill.indb 69 10-05-15 23:47

over 3 weeks compared with controls (Supplemental Fig. 4B–D). CK14 expression was initially evident in only some cells of the acid pulsed reconstruct at week 1 and not in the controls, however, by week 2 and continuing to week 3, control and acid pulsed CP-A cells expressed similar levels of CK14 (Fig. 5B,C Supplemental Fig. 4E). In summary, CP-A cells retain their squamous cytokeratin profile independent of acid exposure.

Columnar cytokeratins and acid CP-A cells express a combination of both squamous and columnar cytokeratins in basal 3D culture conditions. Acid pulsed CP-A cells stained for columnar cytokeratins CK8 and CK19 demonstrate no significant change in expression compared with their controls over 3 weeks with acid exposure (Supplemental Fig. 4F–K). CK8 expression initially was particularly low, however, by week 2 expression increased substantially, independent of acid exposure. By week 3, control and acid-treated cells demonstrated equally reduced CK8 expression compared to week 2. CK19 expression was elevated throughout 3 weeks of culture, independent of acid exposure. In general, acid pulses of 1 h duration do not induce significant changes in squamous and columnar cytokeratin expression in CP-A cells.

Conclusion

In standard conditions designed to grow keratinocytes in 3D culture (6), BE cells express both squamous and columnar cytokeratins and exhibit a more squamous-like, multilayered growth morphology. Previous studies demonstrated that ATRA was observed to be important for columnar differentiation of esophageal tissue in vivo (10). Chang et al. demonstrated that squamous biopsies grown as explant cultures differentiate to a columnar phenotype when treated with ATRA and that this change was caused by altered differentiation, and not enhanced proliferation of a particular cell type within the explant tissue. Vitamin A was previously demonstrated to be critical to the development of chick esophageal epithelium where it strongly inhibited the development of a thick, stratified oesophageal epithelium in favor of a psuedostratified esophageal epithelium consisting of columnar, ciliated, and mucosal cells (19). Based on these studies, we used ATRA as a differentiation factor for BE cells. Our 3D model of BE established all-trans retinoic acid as a differentiation factor for the BE cell line CP-A, leading to a reduction in squamous cytokeratin expression and an increase in columnar cytokeratin expression after 3 weeks. Among the other BE cell lines, ATRA did not alter cytokeratin expression, but it did alter morphology leading to more single-layer BE-like growth. Further analysis of biological differences between these four BE cell lines will help elucidate the differences observed with ATRA treatment.

Proefschrift_Kirill.indb 70 10-05-15 23:47 The development of BE in vivo is a response to the constant reflux of acid and bile salts from the stomach into the oesophagus in patients with gastro-esophageal reflux disease (GERD). It is hypothesized that BE may actually be a protective mechanism against esophageal tissue damage during reflux (5). As the esophagus is constantly wounded with this milieu of irritants, the lining responds by producing a cell similar to the intestines, which can withstand low pH, and which produces mucous for protection (5). Over time, the squamous lining is replaced with a more protective columnar lining in some GERD patients. Acid exposure on ex vivo biopsies from BE patients resulted in enhanced survival of BE cells over squamous cells (20). Feagins et al. (21), however, demonstrated that acid exposure on BE cells in 2D culture had anti-proliferative effects. Other studies, however, demonstrated enhanced proliferation and decreased apoptosis in vitro and ex vivo with acid (22,23). Acid pulsing of a non-neoplastic BE cell line (BAR-T) demonstrated a change 3 towards a colonic phenotype, indicated by an increase in CK8/ 18 expression (24,25), and increased colony formation and tumorigenicity (25). Based on the potential effects of acid on various BE tissue and cell culture models, we demonstrated here that application of acid on non-dysplastic metaplasia CP-A cells for 3 weeks did not affect cytokeratin profiles nor growth patterns. The lack of altered expression in squamous and columnar cytokeratins with acid suggests that these cells may have an upregulated protective pathway to retain normal levels of cytokeratin expression under stress conditions and protect against changes induced by the toxic effects of acid. Alternatively, the use of bile salts along with acid, and more frequent pulses could reveal more effects on BE growth in 3D culture. In basic growth conditions without ATRA or acid, CP-A cells, derived from non- dysplastic metaplasia, demonstrated a highly invasive phenotype, unlike in vivo growth of BE metaplasia. This suggests that researchers should be cautious in interpreting CPA cells as a model of metaplasia and note that unlike the other cell lines tested here, they have LOH at chromosome 5q, which contains the APC tumor suppressor gene (11). APC blocks cell proliferation and induces differentiation (26). Tests for the relationship between mutation of APC and disruption to cellular growth in 3D culture are beyond the scope of this study but warrant further investigation. The model developed in this study opens up many potential applications to study BE in vitro. Most importantly, a 3D model of BE, with modifications, can be used to test potential therapeutics in a more physiologically realistic (though still artificial) culture environment. In addition, the model can be used to study factors necessary in vivo for malignant transformation of EA derived from benign BE. By understanding this transition, potential agents can be developed that interfere with this conversion. These cell lines may provide a platform to test the generality of potential therapies or treatments

Proefschrift_Kirill.indb 71 10-05-15 23:47 in a 3D culture environment as a stepping stone between standard 2D cell culture and other models. The heterogeneity in cell line responses ensures that the diversity present between patients and within a single patient can be captured experimentally.

Proefschrift_Kirill.indb 72 10-05-15 23:47 Acknowledgments

We thank James Hayden and Fred Keeney, Wistar Microscopy Core; Gary Swain, University of Pennsylvania Morphology Core Facility; Brian Reid and Rissa Sanchez of the Seattle Barrett’s Esophagus Program for supplying control biopsies, Peter Rabinovitch, University of Washington for supplying the BE cell lines, and Doug Stairs for providing antibodies. This work was supported in part by the Commonwealth Universal Research Enhancement Program, Pennsylvania Department of Health; the Pew Charitable Trust; the Landon AACR Innovator Award for Cancer Prevention; the PhRMA Foundation Bioinformatics Research Starter Grant; the American Cancer Society Research Scholar Grant (grant number 117209-RSG-09-163- 01-CNE) to C.C.M. and the National Institutes of Health (grant numbers F32 CA132450 to L.M.F.M., P30 DK050306 to A.R. and CCM, R03 CA137811, P01 CA91955, P30 CA010815, R01 CA119224, and R01 CA140657 to C.C.M.). 3

Proefschrift_Kirill.indb 73 10-05-15 23:47 Reference list

(1) Nelson CM, Bissell MJ. Of extracellular matrix, scaffolds, and signaling: tissue architecture regulates development, homeostasis, and cancer. Annu Rev Cell Dev Biol 2006;22:287-309. (2) Haggitt RC. Barrett’s esophagus, dysplasia, and adenocarcinoma. Hum Pathol 1994 Oct;25(10):982-993. (3) Thomas T, Abrams KR, De Caestecker JS, Robinson RJ. Meta analysis: Cancer risk in Barrett’s oesophagus. Aliment Pharmacol Ther 2007 Dec;26(11-12):1465-1477. (4) Yousef F, Cardwell C, Cantwell MM, Galway K, Johnston BT, Murray L. The incidence of esophageal cancer and high-grade dysplasia in Barrett’s esophagus: a systematic review and meta-analysis. Am J Epidemiol 2008 Aug 1;168(3):237-249. (5) Reid BJ, Li X, Galipeau PC, Vaughan TL. Barrett’s oesophagus and oesophageal adenocarcinoma: time for a new synthesis. Nat Rev Cancer 2010 Feb;10(2):87-101. (6) Okawa T, Michaylira CZ, Kalabis J, Stairs DB, Nakagawa H, Andl CD, et al. The functional interplay between EGFR overexpression, hTERT activation, and p53 mutation in esophageal epithelial cells with activation of stromal fibroblasts induces tumor development, invasion, and differentiation. Genes Dev 2007 Nov 1;21(21):2788-2803. (7) Kalabis J, Patterson MJ, Enders GH, Marian B, Iozzo RV, Rogler G, et al. Stimulation of human colonic epithelial cells by leukemia inhibitory factor is dependent on collagen-embedded fibroblasts in organotypic culture. FASEB J 2003 Jun;17(9):1115-1117. (8) Su Y, Chen X, Klein M, Fang M, Wang S, Yang CS, et al. Phenotype of columnar-lined esophagus in rats with esophagogastroduodenal anastomosis: similarity to human Barrett’s esophagus. Lab Invest 2004 Jun;84(6):753-765. (9) Ouatu-Lascar R, Fitzgerald RC, Triadafilopoulos G. Differentiation and proliferation in Barrett’s esophagus and the effects of acid suppression. Gastroenterology 1999 Aug;117(2):327-335. (10) Chang CL, Lao-Sirieix P, Save V, De La Cueva Mendez G, Laskey R, Fitzgerald RC. Retinoic acid- induced glandular differentiation of the oesophagus. Gut 2007 Jul;56(7):906-917. (11) Palanca-Wessels MC, Barrett MT, Galipeau PC, Rohrer KL, Reid BJ, Rabinovitch PS. Genetic analysis of long-term Barrett’s esophagus epithelial cultures exhibiting cytogenetic and ploidy abnormalities. Gastroenterology 1998 Feb;114(2):295-304. (12) Palanca-Wessels MC, Klingelhutz A, Reid BJ, Norwood TH, Opheim KE, Paulson TG, et al. Extended lifespan of Barrett’s esophagus epithelium transduced with the human telomerase catalytic subunit: a useful in vitro model. Carcinogenesis 2003 Jul;24(7):1183-1190. (13) Harada H, Nakagawa H, Oyama K, Takaoka M, Andl CD, Jacobmeier B, et al. Telomerase induces immortalization of human esophageal keratinocytes without p16INK4a inactivation. Mol Cancer Res 2003 Aug;1(10):729-738. (14) Nurgalieva Z, Lowrey A, El-Serag HB. The use of cytokeratin stain to distinguish Barrett’s esophagus from contiguous tissues: a systematic review. Dig Dis Sci 2007 May;52(5):1345-1354. (15) Biddle D, Spandau DF. Expression of vimentin in cultured human keratinocytes is associated with cell - extracellular matrix junctions. Arch Dermatol Res 1996 Sep;288(10):621-624. (16) Paladino G, Marino C, La Terra Mule S, Civiale C, Rusciano D, Enea V. Cytokeratin expression in primary epithelial cell culture from bovine conjunctiva. Tissue Cell 2004 Oct;36(5):323-332. (17) Glickman JN, Blount PL, Sanchez CA, Cowan DS, Wongsurawat VJ, Reid BJ, et al. Mucin core polypeptide expression in the progression of neoplasia in Barrett’s esophagus. Hum Pathol 2006

Proefschrift_Kirill.indb 74 10-05-15 23:47 Oct;37(10):1304-1315. (18) Boch JA, Shields HM, Antonioli DA, Zwas F, Sawhney RA, Trier JS. Distribution of cytokeratin markers in Barrett’s specialized columnar epithelium. Gastroenterology 1997 Mar;112(3):760- 765. (19) AYDELOTTE MB. The effect of vitamin A and citral on epithelial differentiation in vitro. I. The chick tracheal epithelium. J Embryol Exp Morphol 1963 Mar;11:279-291. (20) Dvorakova K, Payne CM, Ramsey L, Bernstein H, Holubec H, Chavarria M, et al. Apoptosis resistance in Barrett’s esophagus: ex vivo bioassay of live stressed tissues. Am J Gastroenterol 2005 Feb;100(2):424-431. (21) Feagins LA, Zhang HY, Hormi-Carver K, Quinones MH, Thomas D, Zhang X, et al. Acid has antiproliferative effects in nonneoplastic Barrett’s epithelial cells. Am J Gastroenterol 2007 Jan;102(1):10-20. (22) Fitzgerald RC, Omary MB, Triadafilopoulos G. Dynamic effects of acid on Barrett’s esophagus. An ex vivo proliferation and differentiation model. J Clin Invest 1996 Nov 1;98(9):2120-2128. (23) Shirvani VN, Ouatu-Lascar R, Kaur BS, Omary MB, Triadafilopoulos G. Cyclooxygenase 2 3 expression in Barrett’s esophagus and adenocarcinoma: Ex vivo induction by bile salts and acid exposure. Gastroenterology 2000 Mar;118(3):487-496. (24) Bajpai M, Liu J, Geng X, Souza RF, Amenta PS, Das KM. Repeated exposure to acid and bile selectively induces colonic phenotype expression in a heterogeneous Barrett’s epithelial cell line. Lab Invest 2008 Jun;88(6):643-651. (25) Das KM, Kong Y, Bajpai M, Kulkarni D, Geng X, Mishra P, et al. Transformation of benign Barrett’s epithelium by repeated acid and bile exposure over 65 weeks: a novel in vitro model. Int J Cancer 2011 Jan 15;128(2):274-282. (26) Jankowski JA, Odze RD. Biomarkers in gastroenterology: between hope and hype comes histopathology. Am J Gastroenterol 2009 May;104(5):1093-1096.

Proefschrift_Kirill.indb 75 10-05-15 23:47 Proefschrift_Kirill.indb 76 10-05-15 23:47 Chapter 4 Transcription factors associated with the development of Barrett’s esophagus; in vivo expression and modulation by Retinoic Acid Authors: Pavlov K.1, Boersma-van Ek W.1, Meijer C.2, Peters F.T.M.1, van den Berg A.3, Kleibeuker J.H.1, Kruyt F.A.E.2

Chapter

Proefschrift_Kirill.indb 77 10-05-15 23:47 Abstract

Background: Barrett’s esophagus (BE) is a columnar metaplasia that is the main risk factor for esophageal adenocarcinoma. Altered expression of transcription factors that direct differentiation could be important in BE development, but little is known regarding their expression pattern and the signaling pathways that regulate them. Retinoic acid (RA) is a potent morphogen that has previously been implicated in the development of BE. Aim: To investigate the expression of squamous (p63, SOX2), early (GATA6) and late (CDX2) columnar transcription factors in squamous epithelium and BE in vivo and study the effect of RA on the expression of these transcription factorsin vitro. Methods: The expression of the transcription factors p63, SOX2, GATA6 and CDX2 and the glycoprotein MUC2 was studied in formalin fixed and paraffin-embedded tissue biopsies of squamous epithelium (N=47), non-intestinal type (non-IM) BE (N=14) and intestinal- type (IM) BE (N=24). The effect of all-trans retinoic acid (ATRA) on the expression of squamous and columnar transcription factors was studied using the immortalized esophageal squamous cell line EPC2-hTERT. Results: P63 and SOX2 were expressed in the majority of cells in squamous epithelium. P63 expression was lost completely (p<0,001) In BE. The percentage of SOX2 positive cells was strongly reduced in BE, but a minority of cells retained SOX2 expression (p<0,001). The number of GATA6 and CDX2 positive cells was very low in squamous epithelium and significantly increased in non-IM BE (p<0,001 for both transcription factors). The number of CDX2 positive cells further increased in IM BE (p<0,001), whereas the percentage of GATA6 positive cells remained constant in IM BE. The percentage of MUC2 positive cells was close to zero in squamous epithelium, low in non-IM, while the majority of the cells in IM BE cells were MUC-positive. In vitro, ATRA treatment induced downregulation of the p63 isoform ΔNp63α protein (p=0,037) but had no significant effect on p63 mRNA levels. ATRA treatment also induced significant upregulation of GATA6 and SOX9 mRNA expression (p=0,015 for both) but did not alter the expression levels of CDX2 or MUC2 mRNA. Conclusion: Squamous epithelium and BE are each characterized by distinct patterns of transcription factor expression. RA can partially induce a columnar-type expression pattern of transcription factors in vitro.

Proefschrift_Kirill.indb 78 10-05-15 23:47 Introduction

Barrett’s metaplasia of the esophagus (BE) is a precursor lesion of esophageal adenocarcinoma. BE is defined by the replacement of squamous epithelium in the distal end of the esophagus with columnar epithelium. Histologically, BE is broadly classified as metaplastic columnar epithelium without goblet cells (non-intestinal metaplasia, non-IM) or with goblet cells (intestinal metaplasia, IM). This distinction is important as non-IM is thought to precede IM during BE development (1). Persistent gastro-esophageal reflux disease (GERD) shows a strong association with BE and is thought to be the main driving factor in its development (2). Several transcription factors, most notably CDX2, GATA6 and SOX9 have been associated with columnar differentiation of esophageal epithelium. However, no single columnar transcription factor is sufficient to drive a complete transdifferentiation of squamous epithelium towards a Barrett’s-like columnar metaplasia (3-5). Loss of transcription factors associated with squamous differentiation could be another necessary step for BE development. The transcription factors p63 and SOX2 are essential for the development and maintenance of squamous epithelium. Animal models showed that in mice lacking either p63 or SOX2 the esophagus is covered with a columnar epithelium resembling BE (6-8). The essential role of p63 in squamous epithelium is further supported 4 by the finding that a single p63 isoform, ΔNp63α, is responsible for the maintenance of squamous stem cells in skin (9-11). The signaling pathways responsible for the altered transcription factor expression during BE development are incompletely understood. Retinoic acid (RA) and its isoform All-Trans Retinoic Acid (ATRA) have been suggested as a possible inducer of columnar differentiation in the esophagus. ATRA is a known inducer of GATA6 during embryological development (13,14). In adult tissues ATRA deficiency induces a squamous metaplasia in the columnar lining of the trachea in rabbits and of the endocervix in mice (15,16). RA biosynthesis is increased in BE compared with normal squamous epithelium (17). In esophageal squamous epithelial cells RA induced the expression of MUC2, a glycoprotein that is a direct downstream target of CDX2 and is typically associated with IM, and ATRA caused loss of the squamous cytokeratin CK13 (18,19). While these data indicate that RA and its isoform ATRA play a role in inducing and maintaining columnar differentiation, it is unknown whether RA treatment has an effect on transcription factors associated with squamous or columnar differentiation of the esophagus. In this study we hypothesized that the development of BE is associated with loss of the squamous transcription factors p63 and SOX2 and gain of columnar transcription factors GATA6, SOX9 and CDX2, and that ATRA can induce this shift in transcription factor expression in vitro. To test this hypothesis first investigated the expression of p63,

Proefschrift_Kirill.indb 79 10-05-15 23:47 SOX2, CDX2 and MUC2 in tissue biopsies of squamous epithelium and BE. In addition, we studied the effect of ATRA on the expression of p63, miR-203, SOX2, GATA6, SOX9 and CDX2 in a squamous esophageal cell line.

Material and Methods

Tissue collection A total of 73 paraffin-embedded biopsies containing samples of squamous epithelium (N=47), non-IM BE (N=14) and IM BE (N=24) was retrieved from the archives of the Department of Pathology, University Medical Center Groningen, the Netherlands. All biopsy samples were collected from patients undergoing gastro-esophageal endoscopy at the University Medical Center Groningen. The biopsy samples were fixed in formalin overnight and embedded in paraffin.

Immunohistochemistry Slides were deparaffinized using xylene and rehydrated by a series of ethanol dilutions. Antigen retrieval was performed by heating the slides in 10mM citrate buffer (pH 6.0) for 15 minutes using a 400W microwave. Endogenous peroxidase blocking was performed by

incubating the slides for 30 minutes with 0,3% H202 in phosphate-buffered saline (PBS). All primary antibodies, pan-p63 (Cat. # sc-8431, 1:500, Santa Cruz, USA), ΔNp63 (Cat. # 619001, 1:500 Biolegend, USA), SOX2 (Cat. # 4900, 1:400 Cell Signaling, USA), CDX2 (Cat. # ab76541, 1:500 Abcam, UK), MUC2 (Cat. # sc59859 1:100 Santa Cruz, USA) were diluted in PBS with 1% bovine serum albumin. Slides were incubated for 60 minutes at room temperature with the primary antibody. As a negative control slides were stained with an isotype-matched IgG. After washing, slides were incubated for 30 minutes at room temperature with horseradish peroxidase-conjugated secondary and tertiary antibodies (all from Dako, Denmark) diluted 1:50 in PBS with 1% bovine serum albumin and 1% human serum. Peroxidase was visualized using 3,3’diaminobenzidine (Sigma-Aldrich, USA) and slides were counterstained with hematoxilin. The percentage of positive cells was calculated from an average of three crypts. For squamous epithelium, one representative area was evaluated. The final percentage was calculated by averaging the estimates of two observers (KP and WB).

Cell culturing and ATRA treatment The non-malignant squamous esophageal epithelial keratinocyte cell line EPC2-hTERT (kind gift from Dr. A.Rustgi, Division of Gastroenterology, University of Pennsylvania Perelman School of Medicine, Philadelphia, USA) was cultured in Keratinocyte Serum- Free Medium (KSFM) supplemented with KSFM Growth Supplement (Life Technologies,

Proefschrift_Kirill.indb 80 10-05-15 23:47 Table 1: PCR primers Primers (5’-3’) Forward Reverse Product size Pan-p63 TAACACAGACCACGCGCAGA GAATACGTCCAGGTGGCCGA 183 bp ΔNp63 GAAGAAAGGACAGCAGCATTGA GGGACTGGTGGACGAGGAG 156 bp SOX2 GGCAGCTACAGCATGATGCAGGAGC CTGGTCATGGAGTTGTACTGCAGG 131 bp GATA6 GCCAACTGTCACACCACAAC TGGAGTCATGGGAATGGAAT 216 bp SOX9 GACCAGTACCCGCACTT TTCACCGACTTCCTCCG 176 bp CDX2 GGCAGCCAAGTGAAAACCAG GGTGATGTAGCGACTGTAGTGAA 105 bp MUC2 GCTGCTATGTCGAGGACACC GGGAGGAGTTGGTACACACG 90 bp β-actin GTTGTCGACGACGAGCG GCACAGAGCCTCGCCTT 93 bp hsa-miR-203 Applied Biosystems, assay ID 000507

Bleiswijk, The Netherlands) using plastic disposables from Greiner (Greiner Bio One, Alphen aan den Rijn, the Netherlands). Cells were maintained at a maximum of 80% confluency to avoid terminal differentiation. Cells were counted at low magnification under a brightfield microscope using trypan blue to distinguish living from dead cells. ATRA (Sigma-Aldrich, USA) was dissolved in 96% ethanol to a final concentration of 0,5.10-2 M. For treatment ATRA was further diluted in KSFM to a concentration of 1 μM 4 ATRA. The control vehicle consisted of 96% ethanol diluted in KSFM. EPC2-hTERT cells were harvested at 24, 48 and 72 hours after treatment with 1 μM ATRA or control. Data points for each experimental condition were gathered in duplicate and experiments were performed in triplicate.

Western blotting 5μg denatured protein from whole cell lysates was separated on a 10% SDS-PAGE gel and transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore, The Netherlands). ΔNp63α protein level was examined using two primary antibodies: the pan-p63 antibody (Cat. # sc8431 1:1000, Santa Cruz, USA) and the ΔNp63 antibody (Cat. # 619001 dilution 1:500 Biolegend, USA). Both antibodies showed a band corresponding with the molecular weight of the ΔNp63α isoform (~70 kDa). β-actin (IZN Pharmaceuticals, Zoetermeer, The Netherlands) was used as a loading control for quantification of the relative ΔNp63 protein levels. Densitometric analysis was performed using the ImageJ software 1.47 (U. S. National Institutes of Health, Bethesda, Maryland).

RNA isolation, primers, PCR Total RNA was extracted using TRIzol (Invitrogen, Life Technologies, Bleiswijk, the Netherlands). RNA was reverse-transcribed using random primers (Invitrogen, Life Technologies, Bleiswijk, the Netherlands). For pan-p63, ΔNp63, SOX2 and GATA6 qRT-

Proefschrift_Kirill.indb 81 10-05-15 23:47 PCR was performed on a Biorad thermocycler using Sybergreen (Life Technologies, Bleiswijk, The Netherlands). Since it is not possible to design a primer set that is specific for the ΔNp63α isoform, we used one set that recognizes all six p63 isoforms and a second primer pair that is specific for all three ΔNp63 isoforms. β-actin was used as a control. Primer pairs are listed in Table 1. Relative expression levels were calculated according to the 2-ΔCt formula.

Statistical analysis Mann Whitney test was used for the immunohistochemical data and cell number analysis and the Wilcoxon signed rank test was used for quantification of Western blot and PCR data. A p-value of p <0.05 was considered significant. Statistical analysis was performed using the Prism 5.0 software (GraphPad Software, San Diego, CA, USA).

Results

In vivo expression of squamous and columnar transcription factors We examined the expression pattern of squamous transcription factors in biopsies from squamous epithelium, non-IM and IM BE. P63 stained positive in 70% of the nuclei in squamous epithelium specimens, but was completely absent in non-IM and IM (p<0,001, Figure 1A and 1B). Similar to p63, nuclear SOX2 expression was present in 50% of the nuclei in squamous epithelium specimens (Figure 1C). In contrast to p63, we observed a subset of SOX2-expressing cells in both non-IM and IM BE. Compared to squamous epithelium the percentage of SOX2-positive cells was significantly lower (8%) in non-IM BE (p<0,001). The percentage of SOX2 positive cells decreased further in IM specimens where it was present in approximately 3% of cells (p=0,017, Figure 1D). Next, we studied the expression of GATA6 and CDX2, two known markers of columnar differentiation. GATA6 showed nuclear expression in a minority (10%) of squamous epithelial cells and the percentage of GATA6-positive cells increased in non-IM BE to 30% (p<0,001) while there was no further increase in IM (28% positive cells, p=0,834, Figure 1E and F). CDX2 was almost absent in squamous epithelium and the percentage of CDX2-positive cells was significantly higher in IM BE (90%) compared to non-IM BE (21%) (p<0,001, Figure 1G, H). MUC2 was absent in squamous epithelium, while 6% of the cells were MUC2-positive in non-IM BE specimens, and this increased to 71% IM BE specimens (Figure 1I, J).

The in vitro effect of ATRA on cell proliferation and transcription factor expression We used the EPC2-hTERT as a model to explore the effects of ATRA. EPC2-hTERT cells showed a significant decrease in cell number at 48 and 72 hours compared to cells treated with the control vehicle (Figure 2A). ATRA did not induce cell death based on trypan blue

Proefschrift_Kirill.indb 82 10-05-15 23:47 Figure 1

Squamous epithelium non-IM IM A B p<0,001

pan-p63

Percentage of positi ve cells of positi Percentage non-IM Squamous C D p<0,001

p=0,017 SOX2

Percentage of positi ve cells of positi Percentage non-IM Squamous E F

p<0,001 p=0,834

GATA6

Percentage of positi ve cells ve positi of Percentage

non-IM Squamous 4 G H p<0,001

p<0,001 CDX2

Percentage of positi ve cells ve positi of Percentage non-IM Squamous

I J p<0,001

MUC2

Percentage of positi ve cells ve positi of Percentage non-IM Squamous

Figure 1: Expression of squamous and columnar transcription factors in patient biopsies Representative images and quantification of immunohistochemical staining for pan-p63, SOX2, GATA6, CDX2 and MUC2. Both p63 and SOX2 show nuclear expression in basal and suprabasal layers of squamous epithelium. P63 is completely lost in Barrett’s esophagus in both the non- intestinal metaplasia (non-IM) as well as in intestinal-type metaplasia (IM, panel 1A, B). SOX2 is retained in a minority of the non-IM and IM BE cells (panel 1C,D). GATA6 is expressed in a minority of squamous cells, but the percentage of positive cells significantly increases in non-IM and IM as compared to squamous epithelium. However, the percentage of GATA6-positive cells does not differ between non-IM and IM (panel C,D). CDX2 expression is almost absent in squamous epithelium, but the percentage of positive cells increases progressively in non-IM and IM (panel G, H). The percentage of MUC2 positive cells mirrors that of CDX2, being absent in squamous epithelium and progressively increasing in non-IM and IM (panel I, J). The bars in the boxes indicate the mean and error bars denote the standard deviation.

Proefschrift_Kirill.indb 83 10-05-15 23:47 Figure 2

A B Treated Control

24h 48h 72h 24h 48h 72h

◄70 kDa

pan-p63 AB

b-actin Normalized 0,502 0,114 0,373 0,708 0,421 0,669 expression Figure 2 ◄70 kDa

A B Treated Control �Np63 AB

24h 48h 72h 24h 48h 72h

◄70 kDa b-actin

Normalized 0,133 0,018 0,069 0,514 0,270 0,493 expression pan-p63 AB pan-p63 WB �Np63 WB b-actin Normalized 0,502 0,114 0,373 0,708 0,421 0,669 expression C D

◄70 kDa

�Np63 AB

b-actin

Normalized 0,133 0,018 0,069 0,514 0,270 0,493 expression pan-p63 PCR �Np63 PCR pan-p63 WB �Np63 WB Figure 2: ATRA treatmentE reduces cell proliferation and induces a shift Fin expression from squamous to columnar transcription factors C D ATRA blocks cell proliferation (panel A). ΔNp63α protein expression was examined using two different antibodies, one recognizing all p63 isoforms (pan-p63 AB) and one specific for the ΔNp63 isoforms (ΔNp63 AB). For both antibodies, the lane corresponding to ΔNp63α was used for analysis. An example of a western blot is given in panel B and quantification of results in panels C and D. ATRA treatment reduced ΔNp63α protein level 72 hours after

treatment (p=0,037 for the pan-p63 antibody and p=0,078 for the ΔNp63 antibody). ATRA treatment had no effect

on pan-p63 or ΔNp63 mRNA levels (panel E and F). ATRA treatment had no effect on SOX2 mRNA levels (panel

G), but increased GATA6 mRNA at 24 and 48 hours after treatment (panel H) and SOX9 mRNA at 72 hours after

treatment (panel I). Control conditions are represented as white columns, and treated samples as black columns. The

bars in the boxes indicate the mean and error bars denote the standard deviation. Data points for each experimental

condition were gathered in duplicate and experiments were performed in triplicate.

SOX2 PCR GATA6 PCR SOX9 PCR 84 pan-p63 PCR �Np63 PCRG H I

E F Proefschrift_Kirill.indb 84 10-05-15 23:47

SOX2 PCR GATA6 PCR SOX9 PCR G H I

Figure 2

A B Treated Control

24h 48h 72h 24h 48h 72h

◄70 kDa

pan-p63 AB

Figure 2 b-actin Normalized 0,502 0,114 0,373 0,708 0,421 0,669 expression

A B Treated Control ◄70 kDa

24h 48h 72h 24h 48h 72h �Np63 AB

◄70 kDa

b-actin pan-p63 AB Normalized 0,133 0,018 0,069 0,514 0,270 0,493

expression

b-actin Normalized 0,502 0,114 0,373 0,708 0,421 0,669 expression pan-p63 WB �Np63 WB

C pan-p63 WB ◄70 kDaD �Np63 WB

�Np63 AB

b-actin

0,133 0,018 0,069 0,514 0,270 0,493

Normalized

expression

pan-p63 WB �Np63 WB pan-p63 PCR �Np63 PCR C D E pan-p63 PCR F �Np63 PCR

4 pan-p63 PCR �Np63 PCR SOX2 PCR GATA6 PCR SOX9 PCR E F G SOX2 PCR H GATA6 PCR I

SOX2 PCR GATA6 PCR SOX9 PCR G H I SOX9 PCR

Proefschrift_Kirill.indb 85 10-05-15 23:47 dye exclusion, suggesting that the lower cell numbers were the result of an ATRA-induced block in proliferation (data not shown). Next, we examined the effect of ATRA on the squamous transcription factors p63 and SOX2. We focused this analysis on the ΔNp63α isoform, since previous studies showed that this isoform was responsible for the induction and maintenance of squamous epithelium. ATRA treatment had little effect on ΔNp63α protein expression at 24 and 48 hours, but at 72 hours we observed a downregulation in ATRA-treated cells compared with control cells (p=0.037 for the pan-p63 antibody and p=0,078 for the ΔNp63α specific antibody, Figure 2C,D). Next, we studied mRNA levels of pan-p63 and ΔNp63 (all three isoforms) by qRT-PCR. No differences in pan p63 or ΔNp63 mRNA levels were observed (Figure 2E, F). ATRA treatment had also no effect on SOX2 mRNA levels at all time points (Figure 2G). Finally, we examined whether ATRA treatment could induce the expression of markers of columnar differentiation. Compared to time-matched controls, ATRA treatment significantly increased the mRNA expression of the columnar transcription factors GATA6 at 24 and 48 hours after treatment (p=0,002 and p=0,015 respectively, Figure 2H) and of SOX9 at 48 and 72 hours (p=0,015 and p=0,002 respectively, Figure 2I). However, ATRA treatment did not induce expression of the intestinal markers CDX2 or its downstream target MUC2 (data not shown).

Discussion

We have shown that p63 expression is lost completely in BE, while a minority of BE cells retain SOX2 expression. The number of GATA6 and CDX2-positive cells is strongly increased in BE compared to squamous epithelium. Interestingly, the percentage of CDX2 positive cells is higher in IM compared to non-IM specimens, while the percentage of GATA6-positive cells does not differ between these two subtypes of BE.In vitro, treatment of EPC2-hTERT keratinocytes with ATRA induces loss of ΔNp63α protein and upregulation of the columnar transcription factors GATA6 and SOX9. RA signaling has long been implicated in columnar differentiation of various tissues including cervix, esophagus and bronchia (15,16,19,20). This is the first study that examined the effect of ATRA on a squamous and columnar transcription factors in an esophageal squamous epithelial cell line. ATRA downregulated ΔNp63α protein expression, but it had no effect on the mRNA level of p63 and SOX2. While we cannot exclude the possibility that ATRA affects SOX2 protein expression, the finding of a minority of SOX2-positive cells observed in BE biopsies suggests that loss of SOX2 expression is not essential for columnar differentiation. This is remarkable, since both p63 and SOX2 are necessary for the development of squamous epithelium during embryological development of the esophagus (6,8). The discrepancy between the observed levels of p63 protein and mRNA

Proefschrift_Kirill.indb 86 10-05-15 23:47 might be explained by post-transcriptional regulation by for example miR-203, which has been shown to target the p63 transcript (21-23). ATRA treatment enhanced the expression of GATA6 and SOX9 at the mRNA level. GATA6 is a zinc finger transcription factor that has been associated with proliferation and differentiation in intestinal epithelium (24-26), and a recent series of papers described GATA6 as a central regulator of stem cells in colon (27,28). While little is known about possible stem cells in BE, the crypt-based architecture of BE is suggestive of their presence, since it mimics the crypt structure of intestinal tissue that is known to be controlled by stem cells (29). SOX9 is another marker of intestinal stem cells and plays a role in intestinal homeostasis and Paneth cell differentiation (30-32). Previous studies showed that SOX9 and GATA6 expression are early events in BE development (4,5), suggesting that ATRA- treated EPC2 cells could provide a model for studying the early steps in BE development. ATRA treatment did not alter CDX2 expression levels. CDX2 is required for intestinal differentiation and widely expressed in IM-type BE (1). However, forced CDX2 expression in squamous epithelium did not induce a columnar differentiation and co-expression of both SOX9 and CDX2 had the same phenotype as forced expression of SOX9 alone. This suggests that CDX2 is dispensable in the development of simple, non-intestinal columnar epithelium (5). In a previous study RA was reported to induce MUC2 mRNA expression 4 after 24 hours of incubation in primary keratinocytes (33). Despite a longer incubation period, we did not detect MUC2 mRNA expression after ATRA treatment. This difference could possibly be explained by the fact that we used an established keratinocyte cell line as opposed to primary keratinocytes, but the lack of MUC2 expression is in line with the absence of CDX2 and the notion that our model (partially) may mimic the early stages of (non-IM) BE development. This is further supported by the finding that the percentage of positive cells of both CDX2 and MUC2 was low in non-IM type BE but significantly higher in IM-type BE. While this study is the first to examine the in vitro effect of RA signaling on transcription factors implicated in BE development, it has several limitations. First, we noted that the in vitro model had significant variation with regard to transcription factor expression. Second, it is currently unclear whether the upregulation of the GATA6 and SOX9 is a direct consequence of ΔNp63α loss. In mouse models loss of p63 was associated with the development of columnar epithelium independent of RA signaling (6), suggesting that p63 loss induces a default columnar differentiation. Functional studies to further examine the relation between p63 and GATA6 and SOX9 were beyond the scope of the current study, but could be part of a further characterization of our in vitro model. Antagonizing the effect of RA signaling could be a feasible therapeutic strategy to induce regression of BE towards normal squamous epithelium. Chang et al. used citral, a natural antagonist of RA signaling to successfully induce a columnar-to squamous

Proefschrift_Kirill.indb 87 10-05-15 23:47 differentiation in explants of BE (17). Our finding that ATRA causes a loss of squamous and upregulation of columnar transcription factors provides mechanistic clues for further analysis of factors that drive the development of BE and could be used to study the therapeutic effect of RA-antagonists in vitro. In conclusion, this paper shows for the first time that in contrast to p63, a minority of BE cells retain SOX2 expression and that ATRA treatment downregulates ΔNp63α expression, while inducing the expression of GATA6 and SOX9. These findings contribute to a better understanding of the molecular mechanisms involved in BE development, and offer a foundation for further studies aimed at BE chemoprevention through antagonizing RA signaling.

Proefschrift_Kirill.indb 88 10-05-15 23:47 References

(1) Mari L, Milano F, Parikh K, Straub D, Everts V, Hoeben KK, et al. A pSMAD/CDX2 complex is essential for the intestinalization of epithelial metaplasia. Cell Rep 2014 May 22;7(4):1197-1210. (2) Balasubramanian G, Singh M, Gupta N, Gaddam S, Giacchino M, Wani SB, et al. Prevalence and predictors of columnar lined esophagus in gastroesophageal reflux disease (GERD) patients undergoing upper endoscopy. Am J Gastroenterol 2012 Nov;107(11):1655-1661. (3) Tamagawa Y, Ishimura N, Uno G, Yuki T, Kazumori H, Ishihara S, et al. Notch signaling pathway and Cdx2 expression in the development of Barrett’s esophagus. Lab Invest 2012 Jun;92(6):896- 909. (4) van Baal JW, Verbeek RE, Bus P, Fassan M, Souza RF, Rugge M, et al. microRNA-145 in Barrett’s oesophagus: regulating BMP4 signalling via GATA6. Gut 2013 May;62(5):664-675. (5) Clemons NJ, Wang DH, Croagh D, Tikoo A, Fennell CM, Murone C, et al. Sox9 drives columnar differentiation of esophageal squamous epithelium: a possible role in the pathogenesis of Barrett’s esophagus. Am J Physiol Gastrointest Liver Physiol 2012 Dec 15;303(12):G1335-46. (6) Daniely Y, Liao G, Dixon D, Linnoila RI, Lori A, Randell SH, et al. Critical role of p63 in the development of a normal esophageal and tracheobronchial epithelium. Am J Physiol Cell Physiol 2004 Jul;287(1):C171-81. (7) Wang X, Ouyang H, Yamamoto Y, Kumar PA, Wei TS, Dagher R, et al. Residual embryonic cells as precursors of a Barrett’s-like metaplasia. Cell 2011 Jun 24;145(7):1023-1035. (8) Que J, Okubo T, Goldenring JR, Nam KT, Kurotani R, Morrisey EE, et al. Multiple dose- 4 dependent roles for Sox2 in the patterning and differentiation of anterior foregut endoderm. Development 2007 Jul;134(13):2521-2531. (9) Oh JE, Kim RH, Shin KH, Park NH, Kang MK. DeltaNp63alpha protein triggers epithelial- mesenchymal transition and confers stem cell properties in normal human keratinocytes. J Biol Chem 2011 Nov 4;286(44):38757-38767. (10) Okuyama R, Ogawa E, Nagoshi H, Yabuki M, Kurihara A, Terui T, et al. p53 homologue, p51/ p63, maintains the immaturity of keratinocyte stem cells by inhibiting Notch1 activity. Oncogene 2007 Jul 5;26(31):4478-4488. (11) Parsa R, Yang A, McKeon F, Green H. Association of p63 with proliferative potential in normal and neoplastic human keratinocytes. J Invest Dermatol 1999 Dec;113(6):1099-1105. (12) Lena AM, Shalom-Feuerstein R, Rivetti di Val Cervo P, Aberdam D, Knight RA, Melino G, et al. miR-203 represses ‘stemness’ by repressing DeltaNp63. Cell Death Differ 2008 Jul;15(7):1187- 1195. (13) Mauney JR, Ramachandran A, Yu RN, Daley GQ, Adam RM, Estrada CR. All-trans retinoic acid directs urothelial specification of murine embryonic stem cells via GATA4/6 signaling mechanisms. PLoS One 2010 Jul 13;5(7):e11513. (14) Capo-Chichi CD, Rula ME, Smedberg JL, Vanderveer L, Parmacek MS, Morrisey EE, et al. Perception of differentiation cues by GATA factors in primitive endoderm lineage determination of mouse embryonic stem cells. Dev Biol 2005 Oct 15;286(2):574-586. (15) Darwiche N, Celli G, Sly L, Lancillotti F, De Luca LM. Retinoid status controls the appearance of reserve cells and keratin expression in mouse cervical epithelium. Cancer Res 1993 May 15;53(10 Suppl):2287-2299. (16) Inayama Y, Kitamura H, Shibagaki T, Usuda Y, Ito T, Nakatani Y, et al. In vivo growth and

Proefschrift_Kirill.indb 89 10-05-15 23:47 differentiation potential of tracheal basal cells of rabbits in vitamin A deficiency. Int J Exp Pathol 1996 Apr;77(2):89-97. (17) Chang CL, Hong E, Lao-Sirieix P, Fitzgerald RC. A novel role for the retinoic acid-catabolizing enzyme CYP26A1 in Barrett’s associated adenocarcinoma. Oncogene 2008 May 8;27(21):2951- 2960. (18) Kosoff RE, Gardiner KL, Merlo LM, Pavlov K, Rustgi AK, Maley CC. Development and characterization of an organotypic model of Barrett’s esophagus. J Cell Physiol 2012 Jun;227(6):2654-2659. (19) Cooke G, Blanco-Fernandez A, Seery JP. The effect of retinoic acid and deoxycholic acid on the differentiation of primary human esophageal keratinocytes. Dig Dis Sci 2008 Nov;53(11):2851- 2857. (20) Oshima T, Gedda K, Koseki J, Chen X, Husmark J, Watari J, et al. Establishment of esophageal- like non-keratinized stratified epithelium using normal human bronchial epithelial cells. Am J Physiol Cell Physiol 2011 Jun;300(6):C1422-9. (21) Fassan M, Volinia S, Palatini J, Pizzi M, Fernandez-Cymering C, Balistreri M, et al. MicroRNA Expression Profiling in the Histological Subtypes of Barrett’s Metaplasia. Clin Transl Gastroenterol 2013 May 16;4:e34. (22) Wu X, Ajani JA, Gu J, Chang DW, Tan W, Hildebrandt MA, et al. MicroRNA expression signatures during malignant progression from Barrett’s esophagus to esophageal adenocarcinoma. Cancer Prev Res (Phila) 2013 Mar;6(3):196-205. (23) Saad R, Chen Z, Zhu S, Jia P, Zhao Z, Washington MK, et al. Deciphering the unique microRNA signature in human esophageal adenocarcinoma. PLoS One 2013 May 28;8(5):e64463. (24) Beuling E, Aronson BE, Tran LM, Stapleton KA, ter Horst EN, Vissers LA, et al. GATA6 is required for proliferation, migration, secretory cell maturation, and gene expression in the mature mouse colon. Mol Cell Biol 2012 Sep;32(17):3392-3402. (25) Beuling E, Baffour-Awuah NY, Stapleton KA, Aronson BE, Noah TK, Shroyer NF, et al. GATA factors regulate proliferation, differentiation, and gene expression in small intestine of mature mice. Gastroenterology 2011 Apr;140(4):1219-1229.e1-2. (26) Jonckheere N, Vincent A, Franquet-Ansart H, Witte-Bouma J, Korteland-van Male A, Leteurtre E, et al. GATA-4/-6 and HNF-1/-4 families of transcription factors control the transcriptional regulation of the murine Muc5ac mucin during stomach development and in epithelial cancer cells. Biochim Biophys Acta 2012 Aug;1819(8):869-876. (27) Tsuji S, Kawasaki Y, Furukawa S, Taniue K, Hayashi T, Okuno M, et al. The miR-363-GATA6- Lgr5 pathway is critical for colorectal tumourigenesis. Nat Commun 2014;5:3150. (28) Whissell G, Montagni E, Martinelli P, Hernando-Momblona X, Sevillano M, Jung P, et al. The transcription factor GATA6 enables self-renewal of colon adenoma stem cells by repressing BMP gene expression. Nat Cell Biol 2014 Jul;16(7):695-707. (29) Yui S, Nakamura T, Sato T, Nemoto Y, Mizutani T, Zheng X, et al. Functional engraftment of colon epithelium expanded in vitro from a single adult Lgr5(+) stem cell. Nat Med 2012 Mar 11;18(4):618-623. (30) Bastide P, Darido C, Pannequin J, Kist R, Robine S, Marty-Double C, et al. Sox9 regulates cell proliferation and is required for Paneth cell differentiation in the intestinal epithelium. J Cell Biol 2007 Aug 13;178(4):635-648. (31) Gracz AD, Ramalingam S, Magness ST. Sox9 expression marks a subset of CD24-expressing

Proefschrift_Kirill.indb 90 10-05-15 23:47 small intestine epithelial stem cells that form organoids in vitro. Am J Physiol Gastrointest Liver Physiol 2010 May;298(5):G590-600. (32) Ramalingam S, Daughtridge GW, Johnston MJ, Gracz AD, Magness ST. Distinct levels of Sox9 expression mark colon epithelial stem cells that form colonoids in culture. Am J Physiol Gastrointest Liver Physiol 2012 Jan 1;302(1):G10-20. (33) Cooke G, Blanco-Fernandez A, Seery JP. The effect of retinoic acid and deoxycholic acid on the differentiation of primary human esophageal keratinocytes. Dig Dis Sci 2008 Nov;53(11):2851- 2857.

Proefschrift_Kirill.indb 91 10-05-15 23:47 Proefschrift_Kirill.indb 92 10-05-15 23:47 Chapter 5 GATA6 expression in Barrett’s oesophagus and oesophageal adenocarcinoma Authors: Pavlov K.1, Honing J.2, Meijer C.3, Boersma-van Ek W.1, Peters F.T.M.1, van den Berg A.4, Karrenbeld A.4, Plukker J.T.M.2, Kruyt F.A.E.3, Kleibeuker J.H.1

Affiliations: 1Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Department of Surgery, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 3Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 4Department of Pathology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.

Dig Liver Dis. 2014 Oct 16. doi: 10.1016/j.dld.2014.09.014. [Epub ahead of print]

Proefschrift_Kirill.indb 93 10-05-15 23:47 Abstract

Background: Barrett’s oesophagus can progress towards oesophageal adenocarcinoma through a metaplasia-dysplasia-carcinoma sequence, but the underlying mechanisms are poorly understood. The transcription factor GATA6 is known to be involved in columnar differentiation and proliferation, and GATA6 gene amplification was recently linked with poor survival in oesophageal adenocarcinoma. Aim: To study the expression of GATA6 during Barrett’s oesophagus development and malignant transformation. To determine the prognostic value of GATA6 in oesophageal adenocarcinoma. Methods: Two retrospective cohorts were derived from the pathological archive of the University Medical Center Groningen. The first cohort contained 136 tissue samples of normal squamous epithelium, metaplasia, dysplasia and oesophageal adenocarcinoma. The second cohort consisted of a tissue microarray containing tissue from 92 oesophageal adenocarcinoma patients. Immunohistochemistry was used to examine GATA6 protein expression and to correlate GATA6 expression in oesophageal adenocarcinoma with overall and disease-free survival. Results: The percentage of GATA6-positive cells was low in squamous epithelium (10%) but increased progressively in Barrett’s oesophagus (30%, P<0.001) and high-grade dysplasia (82%, P=0.005). GATA6 expression was not associated with overall or disease- free survival in oesophageal adenocarcinoma patients (P=0.599 and P=0.700 respectively). Conclusion: GATA6 expression is progressively increased during Barrett’s oesophagus development and its malignant transformation. However, no prognostic value of GATA6 expression could be found in oesophageal adenocarcinoma.

Proefschrift_Kirill.indb 94 10-05-15 23:47 Introduction

Oesophageal adenocarcinoma (OAC) is a disease with a poor prognosis and rising incidence (1,2). Barrett’s oesophagus (BO) is the only known precursor lesion of OAC and is characterized by the replacement of the squamous epithelium by columnar epithelium in the oesophagus. BO is categorized as either non-intestinal metaplasia (non-IM) or intestinal metaplasia (IM) (3). IM differs from non-IM by the presence of goblet cells, and it has been suggested that non-IM might precede IM during BO development (4,5). Persistent oesophagitis due to gastro-oesophageal reflux disease (GORD) is the main risk factor for the development of BO. However, only a minority of GORD patients develops BO and an even smaller percentage shows subsequent malignant transformation through a metaplasia-dysplasia-carcinoma sequence. Chemoprevention and therapeutic options for BO and OAC are currently limited, in part by the lack of knowledge regarding the factors that drive the development of BO and OAC. GATA6 is a member of the GATA transcription factor family. GATA1, GATA2 and GATA3 are involved in hematopoiesis while GATA4, GATA5 and GATA6 play a role in endodermal differentiation (6,7). In particular, GATA6 is involved in intestinal development. Mouse models showed that GATA6 is involved in regulating proliferation and differentiation in stomach, small intestine and colon (8-10). Besides its role in intestinal development, GATA6 seems to be involved as an oncogene in various gastrointestinal cancers. GATA6 protein expression is increased in pancreatic and colon cancer (11-13). GATA6 gene amplification is associated with an adverse survival in cholangiocarcinoma and pancreatic cancer (14,15). In addition, in vitro studies have shown that GATA6 is 5 associated with tumor proliferation and invasion in pancreatic, gastric and colon cancer (11,14-17). Interestingly, GATA6 expression is also increased in pancreatitis suggesting that inflammation may induce GATA6 expression (11). The role of GATA6 in a variety of gastrointestinal malignancies, and its association with inflammation has raised interest in a possible role of GATA6 in BO and OAC. While the presence of GATA6 in normal squamous oesophageal epithelium is disputed, several studies showed that GATA6 is expressed in nondysplastic BO, and is further increased in dysplasia and OAC (18-20). GATA6 gene amplification has been associated with the transformation of BO towards OAC (20) and a worse survival of OAC patients (21), but it is unknown whether high GATA6 protein expression correlates with adverse survival in OAC patients. Also, little is known regarding the spatial expression pattern of GATA6 along the crypt axis in BO and how this changes during transformation towards OAC. In this study we hypothesize that GATA6 protein expression correlates with columnar differentiation and subsequent development of dysplasia and OAC, and that GATA6 plays a role in the biologic behavior of OAC. To test these hypotheses the

Proefschrift_Kirill.indb 95 10-05-15 23:47 expression of GATA6 during all stages of BO development and OAC progression was examined and its association with patient survival was determined.

Materials and Methods

Patient material All tissue samples were paraffin-embedded and derived from patients undergoing treatment or surveillance at the University Medical Center Groningen, the Netherlands. The expression of GATA6 during BO development and malignant transformation was studied using a tissue cohort consisting of a total of 136 patient samples (80 biopsy samples and 50 endomucosal resection specimens). GATA6 expression during BO development was analyzed using 86 biopsies containing squamous epithelium (N=43), inflamed oesophageal epithelium due to GORD (N=16), non-IM BO (N=16) and IM BO (N=26). All biopsies were collected during diagnostic or surveillance endoscopies, fixated in formalin, embedded in paraffin and stored at ambient temperature at the archives of the Pathology department of the University Medical Center Groningen. GATA6 expression during BO malignant transformation was analyzed using a total of 50 endomucosal resection (EMR) specimens containing samples of metaplasia (N=19), high grade dysplasia (HGD, N=24) and OAC (stage ˂ IB, N=22). Some samples contained more than one epithelial type. All EMR samples were collected during EMR-procedures and processed similar to the biopsy specimens. To correlate GATA6 expression with survival outcome in OAC patients we constructed a tissue microarray (TMA) containing triplicate cores of 104 OAC tumors from formalin-fixated, paraffin-embedded resection material. The tumor specimens were derived from patients with pathologic tumor stage ≥ 1B. These patients underwent an oesophagectomy without neoadjuvant treatment between 1998 and 2008. Clinical (age, sex, overall survival, disease-free survival) and pathological (TNM stage, resection status, angioinvasion, perineural growth, recurrence and localization) parameters were available for all patients whose tissue was included on the TMA and were analyzed in relation to GATA6 expression.

Immunohistochemistry Slides were deparaffinized and antigen retrieval was performed by heating the slides in 10mM citrate buffer (pH 6.0) for 15 minutes using a 400W microwave. Endogenous

peroxidase was blocked by incubating the slides for 30 minutes in 3% H202 in phosphate- buffered saline (PBS). Slides were incubated at room temperature with the primary antibody against GATA6 (polyclonal rabbit antibody against GATA6, sc-9055, Santa Cruz, USA; dilution 1:100) for 60 minutes. Control slides containing samples of normal squamous epithelium, metaplastic, dysplastic BO and OAC were incubated with rabbit

Proefschrift_Kirill.indb 96 10-05-15 23:47 IgG (Dako, Glostrup, Denmark) or PBS. A subset of EMR samples (N=5) and the TMA were also stained against Ki-67 (mouse monoclonal antibody against Ki-67, Dako, Glostrup, Denmark; dilution 1:350) for 60 minutes. Afterwards slides were incubated with a horseradish-peroxidase conjugated goat-anti rabbit secondary and rabbit-anti goat tertiary antibodies (Dako, Glostrup, Denmark; dilution of both secondary and tertiary antibody 1:50). The staining was visualized using 3,3’-diaminobenzidine (Sigma-Aldrich, Saint Louis, USA) and slides were counterstained with haematoxylin.

Staining evaluation All tissue samples were scored for localization, intensity and percentage of positive cells. Scoring was performed by two independent observers (KP and WB for the biopsy and endomucosal resection specimens and KP and JH for the TMA). Random slides were validated by an experienced gastro-intestinal pathologist (AK). Localization was scored as either nuclear or cytoplasmic and intensity was scored as absent (0) weak (1) medium (2) or strong (3). For the biopsy and endomucosal resection specimens, an absolute percentage of positive cells was estimated as a continuous variable and the score for each tissue slide was calculated from an average of three whole crypts. For squamous epithelium, one representative area was evaluated. The final percentage was calculated by averaging the estimates of the two observers. For the TMA, the percentage of positive cells was divided into six categories: no staining (0), 1≥5% (1), 5≥25% (2), 25≥50% (3), 50≥75% (4) and 75- 100% (5). The score for each tumor sample consisted of an average of three tumor cores. To correlate GATA6 expression on the TMA samples with survival outcome, an Immuno Reactivity Score (IRS) was obtained by multiplying the intensity score times the score for 5 percentage of positive cells, as described previously (22,23).

Statistical analysis The difference in percentage of positive cells in the biopsy and EMR specimens was analyzed using the Mann-Whitney test or a Wilcoxon signed rank test when comparing paired samples and an ANOVA test with Bonferroni correction when comparing three groups. Overall Survival (OS) was defined as the time interval between surgery and death or last known follow-up. Disease-Free Survival (DFS) was defined as the time interval between surgery and date of recurrence, death or last known follow-up. Patient characteristics were analyzed using Fisher’s exact test. A quartile analysis was used to study the association between GATA6 expression and survival outcome. Survival curves were calculated according to the Kaplan-Meier method and compared using the log-rank test. Univariate Cox-regression analysis was performed to identify prognostic factors for OS and DFS. A P-value of <0.05 was considered as significant in all statistical analyses. Ten patients from the TMA were excluded from analysis due to in-hospital death. Two

Proefschrift_Kirill.indb 97 10-05-15 23:47 patients were excluded due to insufficient tumor tissue on the TMA. Statistical analysis was performed using the Prism 5.0 software (GraphPad Software, San Diego, CA, USA).

Ethics Statement The study was conducted according to the guidelines of our ethics hospital board (www. ccmo.nl). Archival tissue was handled according to the Dutch Code for proper use of Human Tissue (www.federa.org).

a b

Squamous epithelium Oesophagitis c d

Non-intestinal metaplasia Intestinal metaplasia e e f f " " e f % % # # " "

# # ! !

" #"" #" Metaplasia (non-IM and IM)

$$" #$ $$" #$

# "# " Figure 1: GATA6 expression during Barrett’s oesophagus development. Nuclear GATA6 expression is absent or low in squamous epithelium (a,e) and increased in oesophagitis (with characteristic infiltration of inflammatory cells, denoted by arrows) (b,e). GATA6 expression in Barrett’s oesophagus metaplasia was significantly higher compared to normal squamous epithelium or oesophagitis (e). GATA6 expression is similar in both non-intestinal metaplasia (non-IM and intestinal metaplasia (IM) and preferentially localized in the lower half of the crypt (c,d,f).

Proefschrift_Kirill.indb 98 10-05-15 23:47 Results

GATA6 expression in BO metaplasia development GATA6 expression during BE development was studied in a cohort of 86 patient biopsies containing normal and inflamed squamous epithelium and non-IM and IM BO. GATA6 expression was localized in the nucleus in all tissue types, and no cytoplasmic expressio was observed. The IgG and PBS control slides used did not show any signal, confirming the specificity of the staining signal. Weak GATA6 expression was observed in both normal and inflamed squamous epithelium, but the intensity of GATA6 expression cells was significantly increased in BO metaplasia (P<0.001). We did not find a statistical difference in GATA6 intensity between non-IM and IM tissue (P=0.089). The mean percentage of GATA6-positive cells progressively increased from normal (10%) to oesophagitis (17%) (p=0.014) and BO metaplasia (30%, P=0.017) samples (Figure 1). Percentage of GATA6 expression was similar in both non-IM and IM BO tissue. GATA6 was preferentially expressed in the lower half of the crypts (P=0.004 and P=0,010, Figure 1).

GATA6 expression during BO malignant transformation GATA6 expression during BO malignant transformation was studied using a total of 50

Table 1: Patient characteristics of patients in the tissue microarray cohort

IRS quartile 1st quartile 2nd quartile 3rd quartile 4th quartile 5 (N=23) (N=23) (N=23) (N=23) P-value Sex (Male) 21 19 22 19 0.488 Age (<65 years) 10 11 8 10 0.879 pTa (pT1-T2) 6 8 9 6 0.755 pNb (pN0) 4 7 9 8 0.420 Resectionc (R0) 20 19 22 19 0.531 Angioinvasion 7 6 4 7 0.729 Perineural growth 7 7 5 7 0.900 Lymph invasion 3 5 2 5 0.597 Recurrence 18 13 12 15 0.262 Localization 0.928 Mid 1 1 0 1 Distal 14 16 17 17 GEJ 8 6 6 5 apT= pathological tumour size stage. bpN= pathological lymph node stage. cResection= Complete resection (RO) or incomplete resection (R1)

Proefschrift_Kirill.indb 99 10-05-15 23:47 Table 2: Univariate analysis of patients in the tissue microarray cohort

N Overall survival Disease-free survival (Total N=92) HR 95% CI P-value HR 95% CI P-value Sex (Male) 81 1.24 0.62-2.49 0.551 1.24 0.61-2.45 0.551 Age (<65 years) 39 0.51 0.31-0.87 0.009 0.66 0.41-1.05 0.080 pTa (pT1-T2) 29 0.36 0.20-0.64 0.001 0.35 0.20-0.61 0.000 pNb (pN0) 28 0.37 0.20-0.67 0.001 0.38 0.22-0.66 0.001 Resectionc (R0) 80 0.59 0.31-1.13 0.120 0.44 0.23-0.82 0.010 Angioinvasion 24 0.56 0.33-0.95 0.032 0.57 0.34-0.95 0.029 Perineural growth 26 0.78 0.47-1.30 0.359 0.68 0.42-1.10 0.116 Lymph invasion 15 1.05 0.55-1.99 0.873 1.06 0.57-1.97 0.847 Localization 0.059 0.102 Mid 3 4.62 1.30-16.42 3.84 1.10-13.41 Distal 64 1.35 0.76-2.37 1.09 0.65-1.85 GEJ 25 Reference Reference GATA6 expression 0.752 0.823 1st quartile 23 1.37 0.70-2.66 1.33 0.70-2.53 2nd quartile 23 1.01 0.51-2.00 1.04 0.54-2.00 3rd quartile 23 1.24 0.64-2.40 1.10 0.57-2.11 4th quartile 23 Reference Reference a pT= pathological tumour size stage. b pN= pathological lymph node stage. c Resection= Complete resection (RO) or incomplete resection (R1)

EMR specimens. The intensity of GATA6 expression did not differ between the metaplastic crypts, HGD and OAC in the EMR slides. However, the number of GATA6 positive cells showed an increase during malignant transformation of BO metaplasia towards OAC. The percentage of GATA6 positive cells was higher in HGD compared to metaplastic crypts (P=0.005, Figure 2). Comparison of metaplastic and HGD crypts from the same EMR specimens (n=16) revealed an expansion of GATA6 positive cells throughout the crypt in HGD compared to metaplasia (mean percentage 68% vs 82%, P=0.002, Figure 2). Within the EMR specimens, there was no significant difference in the percentage of GATA6-positive cells between HGD and OAC. We found that HGD-associated nondysplastic metaplastic crypts from EMR samples had a higher number of GATA6 positive cells compared to metaplastic crypts in samples from biopsies without HGD or OAC (mean percentage 30% vs 68%, P<0.001). To investigate the relation between GATA6 expression and proliferation we compared GATA6 and Ki-67 expression patterns in the EMR specimens. We found that

100

Proefschrift_Kirill.indb 100 10-05-15 23:47 HGD-associated High grade dysplasia Oesophageal Metaplasia adenocarcinoma a b c GATA6

d e f Ki-67

g g h h

! !

" "

Figure 2: GATA6 expression during Barrett’s oesophagus malignant progression. Compared to high grade dysplasia-associated metaplasia (HGD-metaplasia) nuclear GATA6 expression is increased in both high grade dysplasia and oesophageal adenocarcinoma (a-c,g). GATA6 expression partially co-localizes with Ki-67 and both show a progressive increase during Barrett’s oesophagus malignant development (d-f). The increase of GATA6 during progression of Barrett’s oesophagus towards malignancy is further confirmed by comparing GATA6 expression in metaplastic and dysplastic crypts from the same patient (h).

Ki-67 expression partially co-localized with GATA6 expression in metaplasia, HGD and OAC (Figure 2). Similar to the GATA6 expression pattern, we observed nuclear Ki-67 expression predominantly in the lower part of the metaplastic crypts, with expansion of Ki-67 positive cells along the crypt axis in HGD.

101

Proefschrift_Kirill.indb 101 10-05-15 23:47 Association of GATA6 and survival outcome in OAC To investigate the association between GATA6 expression and survival outcome in OAC patients we studied GATA6 expression in a TMA containing tumor samples of 104 OAC patients. 12 patients had insufficient tissue material present on the TMA and were excluded, leaving 92 patients for further analysis. GATA6 intensity was strong in 12 tumors (13.0%), medium in 29 tumors (31.5%), weak in 36 tumors (39.1%) and lost in 15 tumors (16.3%, Figure 3). The majority of tumors showed nuclear GATA6 expression in 75-100% of all tumor cells (67%). We found no association between GATA6 expression and Ki-67 positivity in the tumor samples (data not shown). Patient characteristics associated with the TMA are described in Table 1. Clinicopathological characteristics did not differ between the different IRS score quartiles (Table 1). Univariate analysis showed that pT stage, pN stage and R1 resection are significantly associated with a decreased OS and DFS (Table 2). However, we did not find an association between GATA6 expression and OS or DFS (Table 2, Figure 3B). To further validate our finding that GATA6 expression is not related with survival outcome we compared survival outcome between the lowest and highest IRS quartile and found no association with survival outcome (Supplementary figure 1).

Discussion This study presents the first thorough examination of GATA6 protein expression at all stages of BO and OAC. We show that GATA6 protein expression is progressively increased during BO development and progression towards OAC. GATA6 is not associated with survival outcome in our large cohort of OAC patients. Previous studies indicate that GATA6 is involved in the induction of columnar differentiation and proliferation (8-10,16). Both are important in BO, as altered differentiation of stem cells in squamous oesophageal epithelium may underlie the development of BO metaplasia, and deregulated proliferation is essential for malignant transformation (24,25). As inflammation was correlated with GATA6 expression in the pancreas (11), we hypothesized that inflammation due to persistent GORD may similarly induce GATA6 expression in the oesophagus. We found that GATA6 expression is increased in oesophagitis when compared to samples of normal squamous oesophageal epithelium, but lower than found in Barrett’s metaplasia. Given the role of GATA6 in intestinal development (8-10) our results warrant further studies to examine whether components of reflux or inflammation- associated factors contribute to GATA6 expression in squamous epithelium, and whether GATA6 expression might contribute to the induction of a columnar phenotype. Currently, little is known regarding GATA6 spatial expression in BO crypts and possible differences between IM and non-IM. GATA6-positive cells were preferentially

102

Proefschrift_Kirill.indb 102 10-05-15 23:47 Absent Weak a b

Medium Strong c d

e e ff f

Figure 3: GATA6 expression in oesophageal adenocarcinoma. Nuclear GATA6 expression varies in intensity and is absent in a subset of oesophageal adenocarcinoma samples (a-d). GATA6 expression, stratified by IRS quartile does not affect overall survival or disease-free survivale,f ( ).

located in the lower half of the BO metaplastic crypts, and this pattern was similar in both non-IM and IM, suggesting that increased GATA6 expression is an early event in the development of BO. In metaplastic crypts most, but not all cells expressing GATA6 were also Ki-67 positive. Proliferation was increased in HGD compared to non-IM and IM, and this was mirrored by an expansion of GATA6-positive cells along the crypt axis. A possible functional relationship between GATA6 and proliferation is also consistent with the finding that conditional intestinal deletion of GATA6 in a mouse model resulted in decreased proliferation and less Ki-67 positive cells in the intestinal crypts (10).

103

Proefschrift_Kirill.indb 103 10-05-15 23:47 However, it is likely that GATA6 expression is not only a marker of proliferating cells. Two recent studies showed that GATA6 is crucially important in the regulation of the stem cell niche in colon cancer. GATA6 has a central role in maintaining colonic stem cells through binding to the promotor of leucine-rich repeat containing G protein- coupled receptor 5 (LGR5), a crucial marker of intestinal stem cells (26) and by restricting microenvironmental signals that stimulate differentiation, such as the bone morphogenetic protein signaling pathway (17). In the light of these novel data, our finding of expansion of GATA6-positive cells along the crypt axis in HGD could also reflect impaired maturation of presumed BO stem cells and expansion of the stem cell compartment throughout the crypt axis. Since BMP4 could antagonize the effect of GATA6 in animal models of colon cancer (17), it would be of interest to examine whether administration of BMP4 could reduce the risk of malignancy in OAC animal models. The percentage of GATA6 positive cells in HGD-associated metaplastic crypts was increased compared to metaplastic crypts from patients without dysplasia. This may be explained by a “field effect” in which histologically normal tissue already has molecular alterations characteristic of dysplasia or malignancy. The existence of a field effect in BO has been proposed by several authors on the basis of altered expression of genes on mRNA level, but to our best knowledge this study provides the first evidence for his notion at protein level (27-31). This finding could be of potential clinical use as an early marker of malignant transformation of BO metaplastic crypts, but further research is required to validate the diagnostic value of increased GATA6 expression and caution is warranted since our data show a wide spread in the number of GATA6-positive cells in BO crypts. We did not find an association between GATA6 expression and survival outcome in OAC patients. Previously, Lin et al. reported that GATA6 gene amplification is associated with a shortened OS in OAC patients (21). Although we did not examine GATA6 gene amplification, previous studies reported thatGATA6 gene amplification is associated with increased GATA6 protein expression (20,21). While we did account for staining intensity in the calculation of the IRS score, immunohistochemistry is semiquantitative at best, and this is a limitation of the current study. Furthermore, GATA6 protein levels do not always correlate with mRNA level (32), making it difficult to correlate GATA6 protein expression with GATA6 gene amplification. The results of this study show that GATA6 is progressively upregulated during BO development and malignant progression. While GATA6 protein expression does not influence survival outcome in OAC, GATA6 may be a driver of BO development and malignant transformation and deserves further study.

104

Proefschrift_Kirill.indb 104 10-05-15 23:47 Sources of support

The authors would like to thank the Junior Scientific Masterclass and the Jan Kornelis de Cock foundation for financial support.

105

Proefschrift_Kirill.indb 105 10-05-15 23:47 FigureFigure S1 S1

a a b b

Supplementary figure 1: Kaplan-Meier survival curves comparing survival outcome of the 1st and 4thIRS quartiles. There is no difference in overall survival (a) or disease-free survival (b) between the lowest and highest IRS quartile.

106

Proefschrift_Kirill.indb 106 10-05-15 23:47 Reference list

(1) Buas MF, Vaughan TL. Epidemiology and risk factors for gastroesophageal junction tumors: understanding the rising incidence of this disease. Semin.Radiat.Oncol. 2013; 23: 3-9. (2) Mathieu LN, Kanarek NF, Tsai HL, et al. Age and sex differences in the incidence of esophageal adenocarcinoma: results from the Surveillance, Epidemiology, and End Results (SEER) Registry (1973-2008). Dis.Esophagus 2013; doi: 10.1111/dote.12147 Accessed online 10-3-2014 (3) Fitzgerald RC, di Pietro M, Ragunath K, et al. British Society of Gastroenterology guidelines on the diagnosis and management of Barrett’s oesophagus. Gut 2014; 63: 7-42. (4) Gatenby PA, Ramus JR, Caygill CP, et al. Relevance of the detection of intestinal metaplasia in non-dysplastic columnar-lined oesophagus. Scand.J.Gastroenterol. 2008; 43: 524-30. (5) Hahn HP, Blount PL, Ayub K, et al. Intestinal differentiation in metaplastic, nongoblet columnar epithelium in the esophagus. Am.J.Surg.Pathol. 2009; 33: 1006-15. (6) Ayanbule F, Belaguli NS, Berger DH. GATA factors in gastrointestinal malignancy. World J.Surg. 2011; 35: 1757-65. (7) Molkentin JD. The zinc finger-containing transcription factors GATA-4, -5, and -6. Ubiquitously expressed regulators of tissue-specific gene expression. J.Biol.Chem. 2000; 275: 38949-52. (8) Jonckheere N, Vincent A, Franquet-Ansart H, et al. GATA-4/-6 and HNF-1/-4 families of transcription factors control the transcriptional regulation of the murine Muc5ac mucin during stomach development and in epithelial cancer cells. Biochim.Biophys.Acta 2012; 1819: 869-76. (9) Beuling E, Aronson BE, Tran LM, et al. GATA6 is required for proliferation, migration, secretory cell maturation, and gene expression in the mature mouse colon. Mol.Cell.Biol. 2012; 32: 3392- 402. (10) Beuling E, Baffour-Awuah NY, Stapleton KA, et al. GATA factors regulate proliferation, differentiation, and gene expression in small intestine of mature mice. Gastroenterology 2011; 140: 1219-29. 5 (11) Kwei KA, Bashyam MD, Kao J, et al. Genomic profiling identifies GATA6 as a candidate oncogene amplified in pancreatobiliary cancer. PLoS Genet. 2008; 4: e1000081. Accessed online 10-3-2014 (12) Shureiqi I, Zuo X, Broaddus R, et al. The transcription factor GATA-6 is overexpressed in vivo and contributes to silencing 15-LOX-1 in vitro in human colon cancer. FASEB J. 2007; 21: 743-53. (13) Belaguli NS, Aftab M, Rigi M, et al. GATA6 promotes colon cancer cell invasion by regulating urokinase plasminogen activator gene expression. Neoplasia 2010; 12: 856-65. (14) Tian F, Li D, Chen J, et al. Aberrant expression of GATA binding protein 6 correlates with poor prognosis and promotes metastasis in cholangiocarcinoma. Eur.J.Cancer 2013; 49: 1771-80. (15) Zhong Y, Wang Z, Fu B, et al. GATA6 activates Wnt signaling in pancreatic cancer by negatively regulating the Wnt antagonist Dickkopf-1. PLoS One 2011; 6: e22129. Accessed online 10-3-2014 (16) Chen WB, Huang FT, Zhuang YY, et al. Silencing of GATA6 suppresses SW1990 pancreatic cancer cell growth in vitro and up-regulates reactive oxygen species. Dig.Dis.Sci. 2013; 58: 2518-27. (17) Whissell G, Montagni E, Martinelli P, et al. The transcription factor GATA6 enables self-renewal of colon adenoma stem cells by repressing BMP gene expression. Nat.Cell Biol. 2014; 16: 695-707. (18) Kimchi ET, Posner MC, Park JO, et al. Progression of Barrett’s metaplasia to adenocarcinoma is associated with the suppression of the transcriptional programs of epidermal differentiation. Cancer Res. 2005; 65: 3146-54.

107

Proefschrift_Kirill.indb 107 10-05-15 23:47 (19) Haveri H, Westerholm-Ormio M, Lindfors K, et al. Transcription factors GATA-4 and GATA-6 in normal and neoplastic human gastrointestinal mucosa. BMC Gastroenterol. 2008; 8: 9,230X-8-9. Accessed online 10-3-2014 (20) Alvarez H, Opalinska J, Zhou L, et al. Widespread hypomethylation occurs early and synergizes with gene amplification during esophageal carcinogenesis. PLoS Genet. 2011; 7: e1001356. Accessed online 10-3-2014 (21) Lin L, Bass AJ, Lockwood WW, et al. Activation of GATA binding protein 6 (GATA6) sustains oncogenic lineage-survival in esophageal adenocarcinoma. Proc.Natl.Acad.Sci.U.S.A. 2012; 109: 4251-6. (22) Taubert H, Heidenreich C, Holzhausen HJ, et al. Expression of survivin detected by immunohistochemistry in the cytoplasm and in the nucleus is associated with prognosis of leiomyosarcoma and synovial sarcoma patients. BMC Cancer 2010; 10: 65,2407-10-65. Accessed online 14-8-2014 (23) Kaemmerer D, Peter L, Lupp A, et al. Comparing of IRS and Her2 as immunohistochemical scoring schemes in gastroenteropancreatic neuroendocrine tumors. Int.J.Clin.Exp.Pathol. 2012; 5: 187-94. (24) Nicholson AM, Graham TA, Simpson A, et al. Barrett’s metaplasia glands are clonal, contain multiple stem cells and share a common squamous progenitor. Gut 2012; 61: 1380-9. (25) Volkweis BS, Gurski RR, Meurer L, et al. Ki-67 Antigen Overexpression Is Associated with the Metaplasia-Adenocarcinoma Sequence in Barrett’s Esophagus. Gastroenterol.Res.Pract. 2012; doi: 10.1155/2012/639748. Accessed online 10-3-2014 (26) Tsuji S, Kawasaki Y, Furukawa S, et al. The miR-363-GATA6-Lgr5 pathway is critical for colorectal tumourigenesis. Nat.Commun. 2014; 5: 3150. (27) Brabender J, Lord RV, Danenberg KD, et al. Increased c-myb mRNA expression in Barrett’s esophagus and Barrett’s-associated adenocarcinoma. J.Surg.Res. 2001; 99: 301-6. (28) Brabender J, Lord RV, Metzger R, et al. Differential SPARC mRNA expression in Barrett’s oesophagus. Br.J.Cancer 2003; 89: 1508-12. (29) Brabender J, Marjoram P, Lord RV, et al. The molecular signature of normal squamous esophageal epithelium identifies the presence of a field effect and can discriminate between patients with Barrett’s esophagus and patients with Barrett’s-associated adenocarcinoma. Cancer Epidemiol. Biomarkers Prev. 2005; 14: 2113-7. (30) Selaru FM, Wang S, Yin J, et al. Beyond Field Effect: Analysis of Shrunken Centroids in Normal Esophageal Epithelia Detects Concomitant Esophageal Adenocarcinoma. Bioinform Biol. Insights 2007; 1: 127-36. (31) Lord RV, Salonga D, Danenberg KD, et al. Telomerase reverse transcriptase expression is increased early in the Barrett’s metaplasia, dysplasia, adenocarcinoma sequence. J.Gastrointest. Surg. 2000; 4: 135-42. (32) Tsuji S, Kawasaki Y, Furukawa S, et al. The miR-363-GATA6-Lgr5 pathway is critical for colorectal tumourigenesis. Nat.Commun. 2014; 5: 3150.

108

Proefschrift_Kirill.indb 108 10-05-15 23:47 5

109

Proefschrift_Kirill.indb 109 10-05-15 23:47 Proefschrift_Kirill.indb 110 10-05-15 23:47 Chapter 6 Circulating miRNA’s in Barrett’s esophagus, high-grade dysplasia and esophageal adenocarcinoma Authors: Pavlov K.1, Kluiver J.L.2, Boersma-van Ek W.3, Meijer C.3, Kruyt F.A.E.3, Karrenbeld A.2, Kleibeuker J.H.1, Peters F.T.M.1, van den Berg A.2

Affiliations: 1Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Department of Pathology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 3Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.

In preparation

Proefschrift_Kirill.indb 111 10-05-15 23:47 Abstract

Background: Early detection of premalignant esophageal lesions could drastically improve overall survival, but screening is hampered by the absence of biomarkers indicative of metaplastic and/or malignant transformation. MiRNA’s are small non-coding regulatory RNA species that are characterized by high tissue specificity and stability in serum. Little is known regarding levels of serum miRNA’s in patients with Barrett’s esophagus, dysplasia or esophageal adenocarcinoma and their potential use as biomarkers. Methods: Six samples from each of the four study groups, patients without endoscopic esophageal abnormalities (SE group), patients with Barrett’s metaplasia (BE group), high-grade dysplasia (HGD group) and esophageal adenocarcinoma (EAC group), were profiled using the Nanostring platform. Identified miRNA’s were validated inserum samples of an extended cohort consisting of a total of 69 patients using qRT-PCR. Results: 10 miRNA’s were identified with a fold change>2 between the study groups on the Nanostring platform. Of these, 6 miRNA’s were selected for validation in the extended cohort. In the extended cohort MiR-199a-3p levels were significantly decreased in the BE group compared to the SE group (p≤0.001) and could discriminate between these groups with an area under the curve (AUC) of 0.813. Compared to the SE group, serum miR- 320e levels were significantly decreased in the BE (p≤0.001, AUC 0.790) and HGD groups (p≤0.05, AUC 0.786). Conclusion: Serum miRNA levels of miR-199a-3p and miR-320e could help in identifying patients with BE and HGD.

112

Proefschrift_Kirill.indb 112 10-05-15 23:47 Introduction

Esophageal adenocarcinoma (EAC) is a deadly disease with a rising incidence. While early EAC lesions have a 5-year survival of 78%, this drops to a mere 20% in patients with advanced disease (1,2). Early detection significantly improves survival outcome (3,4), but the majority of EAC cases are diagnosed at an advanced stage. EAC is thought to develop through a metaplasia-dysplasia-carcinoma sequence. The metaplastic lesion is known as Barrett’s esophagus (BE). Patients with BE are advised to undergo regular endoscopic surveillance to detect early malignant transformation. Persistent gastro-esophageal reflux disease (GERD) is the main risk factor for BE development. Currently, BE is diagnosed by histologic examination of endoscopic biopsies, but the low absolute incidence of BE in GERD patients and the invasive nature of the endoscopy preclude its use as a screening tool in patients at risk for BE development (5,6). A screening strategy that at least partially mitigates these drawbacks could contribute to an early detection of Barrett’s esophagus, allowing surveillance and/or early treatment and so decrease morbidity and mortality. Blood-based biomarkers have an established role in the screening of various solid malignancies, such as prostate and ovarian cancer. However, no blood-based biomarkers are currently available for BE, high-grade dysplasia (HGD) or EAC. MiRNAs are a class of small non-coding RNAs that act as negative regulators of gene expression by modulating mRNA translation. MiRNAs are highly tissue-specific and various studies indicate that alterations in miRNA expression have key roles in carcinogenesis (7,8). Moreover, miRNAs are stable in the circulation and alterations in circulating miRNA levels have been detected in various solid malignancies (9). Circulating miRNAs might thus be attractive blood- based biomarkers of BE and EAC. Several studies showed significant differences in tissue miRNA expression patterns between normal squamous epithelium, BE and EAC (10-13). However, it is currently not known whether these differences are also reflected in the circulating miRNA profile. The aim of this study was to investigate whether circulating miRNAs could be used to identify patients with BE, HGD or EAC. 6

Materials and Methods

Patients and samples Patients scheduled to undergo an esophagogastroduodenoscopy were prospectively recruited into one of four study groups: the control group consisting of patients without endoscopic esophageal abnormalities (SE group). A group consisting of patients with Barrett’s metaplasia but no dysplasia or adenocarcinoma (BE group), a group consisting of patients with columnar high-grade dysplasia or early stage (≤T1a) esophageal adenocarcinoma (HGD group) and a group consisting of patients with advanced EAC (stage ≥T1b, EAC group). For each participant, the diagnosis was confirmed based on

113

Proefschrift_Kirill.indb 113 10-05-15 23:47 endoscopic appearance, pathological biopsy analysis and imaging (EUS and/or PET/CT for the HGD and EAC groups). Each participant was also asked about smoking. General exclusion criteria were a history of previous malignancy in the past 5 years, concomitant inflammation or active auto-immune disease that was insufficiently controlled by medication. The study protocol was approved by the Ethics Committee of the University Medical Center Groningen. Written consent was obtained from all participants.

Blood collection Serum samples were collected from each participant. Using venapuncture, approximately 9ml blood was collected from each participant in a serum vacutainer tube (BD, Franklin Lakes, USA). The blood was left to clot for one hour at room temperature and serum was isolated by centrifugation and stored at -80°C.

RNA extraction Serum samples were thawed and filtered through a 0,2 uM filter (Corning, Corning, USA) to eliminate any residual platelets or cell debris. Total RNA was isolated using the miRNeasy micro kit (Qiagen, Venlo, the Netherlands).RNA isolation was performed using 200uL of serum according to the manufacturer’s protocol. Phase Lock Gel Heavy gel tubes (5 Prime, Hilden, Germany) were used for phase separation during RNA isolation. RNA was eluted in 10uL of RNAse-free water and stored at -80°C until further analysis.

Nanostring nCounter miRNA assay A high-throughput serum miRNA profile was generated from a subset of study participants. We selected serum samples from 6 patients of each study group, resulting in a total of 24 samples. Patients from the four study groups were matched for sex, age and comorbidity, as much as possible given the limited number of patient samples. To perform a comprehensive assay of circulating miRNAs in serum we used the nCounter® miRNA Expression Assay version 2 which contains probes for 800 mature human miRNAs (Nanostring Technologies, Seattle, USA). The Nanostring assay was performed by Nanostring Technologies.

Data analysis Data analysis was performed using GeneSpring GX version 12.5 software (Agilent Technologies, Santa Clara, USA). The upper range for negative control oligo’s on the Nanostring platform was 30 counts. Therefore, a detection cutoff was defined as at least four out of six samples in at least one of the four study groups having a count number >30. This cutoff yielded 63 endogenous miRNA’s, the levels of which were normalized by a 90-percentile shift normalization.

114

Proefschrift_Kirill.indb 114 10-05-15 23:47 qRT-PCR MiRNA-specific cDNA synthesis was performed using the Taqman miRNA transcription kit (Life Technologies, Bleijswijk, the Netherlands) in a multiplexed reaction as described previously (14) using reverse transcription primers and TaqMan microRNA assays: miR- 122-5p (ID 002245), miR-144-5p (ID 002148), miR-150-5p (ID 000473), miR-199a-3p (ID 002304), miR-223-3p (ID 002295), miR-320e (ID 243005_mat) (Life Technologies). qRT-PCR was performed using the TaqMan Universal Master Mix II without UNG (Life Technologies) and the abovementioned TaqMan microRNA assays. All qPCR reactions were run in triplicates on the ABI7900HT thermo cycler (Life Technologies). Cycle threshold (Ct) values for all miRNAs were quantified using the Sequence Detection software (version 2.3, Life Technologies). Serum miRNA levels were normalized according to the geometric mean of the Ct values of the SE group and relative levels were calculated according to the 2-ΔCt formula.

Statistical analysis Differences in patient characteristics were analyzed using ANOVA with Scheffé's method. The data from the Nanostring array were analyzed using ANOVA with Benjamini– Hochberg correction and fold changes were calculated based on the geometric mean. MiRNA levels in the serum of the four patient groups as determined by qPCR did not follow a normal distribution (evaluated using the D'Agostino-Pearson omnibus test) and were therefore examined using the Kruskall-Wallis test with Dunn’s multiple testing correction. Statistical analysis was performed using GeneSpring GX version 12.5 software for the Nanostring array data (Agilent Technologies, Santa Clara, USA) and Prism 5.0 statistical package for analysis of qPCR data and ROC curve construction (GraphPad Software, San Diego, USA)

Results 6 Patient groups Patient characteristics are described in Table 1. The extended patient characteristics are described in Supplementary table 1. The only significant difference was in diabetes incidence between the SE and HGD and between the BE and HGD groups (p=0.036 and p=0.013 respectively).

Identifying differences in circulating miRNA levels Using the Nanostring platform we found that 63 miRNA’s showed levels above background (Supplementary figure 1). Analysis of differences in miRNA levels between the four study groups using ANOVA with Benjamini–Hochberg correction for multiple testing yielded no differentially expressed miRNA’s. Analysis of differences in miRNA levels between the

115

Proefschrift_Kirill.indb 115 10-05-15 23:47 Table 1: Patient characteristics for each study group

Study group SE (N=19) BE (N=27) HGD (N=17) EAC (N=6) p-value Gender (Male) 8 (44%) 19 (70%) 13 (77%) 5 (84%) 0.067 Mean Age (Range) 54,3 (23-70) 58,7 (25-75) 65,1 (42-82) 62,5 (46-73) 0.086 Smoking 4 (22%) 5 (19%) 5 (29%) 4 (67%) 0.790 Cardiovascular disease 3 (17%) 6 (22%) 4 (24%) 2 (34%) 0.834 Autoimmune disease 1 (6%) 1 (4%) 1 (6%) 1 (17%) 0.691 Diabetes 1 (6%) 1 (4%) 6 (35%) 0 (0%) 0.004* *After multiple testing correction SE study group vs HGD study group: p=0.036, BE study group vs HGD study group: p=0.013

Table 2: Differences in miRNA expression on the Nanostring platform and in the extended cohort.

Study group SE vs BE SE vs HGD SE vs EAC BE vs HGD BE vs EAC HGD vs EAC p-value miR-223-3p ↓ - ↓ ↑ - ↓ n.s. miR-320e ↓ - - ↑ ↑ - >0.005 miR-122-5p - - ↓ - - - n.s. miR-144-3p - - ↓ - ↓ ↓ - miR-150-5p - - ↓ - - ↓ 0.045 miR-199a-3p - - ↓ - - - >0.001 miR-16-5p - - - ↑ - ↓ n.d. miR-93-5p - - - - - ↓ n.d. miR-4516 - - - ↑ - - n.d. miR-451a - - - ↑ - ↓ n.d. MiRNA’s denoted selected for further validation in the extended cohort are denoted in bold. Significance was determined using the Kruskall-Wallis test with Dunn’s multiple comparison correction. MiR-144-3p was excluded due to technical reasons. n.s.=not significant n.d.=not done

four study groups using a fold change >2 revealed 10 differentially expressed miRNA’s (Table 2). MiRNA’s with a fold change >2 between the SE group and any of the other three study groups were selected for further validation: miR-122-5p, miR-144-3p, miR- 150-5p, miR-199a, miR-233-3p and miR-320e. The pattern of the selected miRNA’s was examined in an expanded cohort consisting of the 24 patients included in the Nanostring array and an additional 45 patients (13 SE, 21 BE and 11 HGD serum samples). Of the six miRNA’s selected for validation in the extended cohort, three miRNA’s, miR-150-5p, miR-199a-3p and miR-320e, showed significant differences between the study groups based on the Kruskall-Wallis test (Table 2). The level of miR-199a-3p was significantly lower in the BE group compared to the SE group (p≤0.001) and the level of miR-320e was significantly lower in the BE and HGD groups compared to the SE group (p≤0.001 and p≤0.05 respectively, Figure 1). Dunn’s test did not reveal significant differences in miR-

116

Proefschrift_Kirill.indb 116 10-05-15 23:47 Figure 1 Figure 1 Figure 1 Figure 1 FigureA 1 FigureA 1 B B A A B B

A A B B

C C D D C C D D .

C C D D .

E E F F

E E F F . .

E E F F

Figure 1: Serum levels of miR-150-5p, miR-199a-3p and miR-320e in the expanded patient cohort. 6 The levels of the miRNA’s were determined by qRT-PCR. No significant differences in miR-150-5p levels were identified between the study groups with Dunn’s test (panel A). Compared to the SE group, serum levels of miR-199a-3p were significantly lower in the BE group (panel B) and serum levels of miR-320e were significantly lower in the BE and HGD groups (panel C). SE: squamous epithelium study group. BE: Barrett’s metaplasia study group. HGD: high-grade dysplasia study group. EAC: esophageal adenocarcinoma study group.

150-5p levels between study groups. We did not identify significant differences in miRNA levels between the four study groups for miR-122-5p and miR-233-3p (Supplementary figure 2). Of note, none of the miRNA’s validated in the extended cohort had significantly different levels in the EAC study group. MiR-144-3p was excluded from further analysis due to insufficient quality of the qRT-PCR assay data.

117

Proefschrift_Kirill.indb 117 10-05-15 23:47 Potential of the significantly decreased serum-derived miRNA’s as disease biomarkers. Based on the significant differences in levels of miR-199a-3p and miR-320e between the SE and the BE/HGD groups, we investigated whether these miRNA’s could be used to identify patients with premalignant changes in the esophagus (Figure 2). Serum miR- 199a-3p levels could discriminate between the SE and BE groups with an area under the curve (AUC) of 0.813 (optimal sensitivity of 85,2% and specificity of 68,4% at relative levels <0.766). Similarly, serum miR-320e levels could discriminate between the SE and BE groups with an AUC of 0.790 (optimal sensitivity of 85,2% and specificity of 63,2% at relative levels <0.844) and between the SE and HGD groups with an AUC of 0.786 FigureFigure 2Figure 2 2 (optimal sensitivity of 82,3% and specificity of 62,2% at relative level <0.838). FigureFigure 2Figure 2 2

A A SEA groupSE groupSE vs group BE vs group BE vs group BE groupB B SEB groupSE groupSE vs group BE vs group BE vs group BE groupC C SEC groupSE groupSE vs group HGD vs HGD groupvs HGD group group A A SEA groupASE groupSE vs group BE vs group BE vs group BE groupB B BSEB groupSE groupSE vs group BE vs group BE vs group BE groupC CC SEC groupSE groupSE vs group HGD vs HGD groupvs HGD group group

. . .

Figure 2: Receiver operating curves (ROC) of miR-199a-3p and miR-320e. Serum miR-199a-3p levels could discriminate between the SE and BE groups with an area under the curve (AUC) of 0.813 (optimal sensitivity of 85,2% and specificity of 68,4% at relative level <0.766, panel A). Serum miR-320e level could discriminate between SE and BE with an AUC of 0.790 (optimal sensitivity of 85,2% and specificity of 63,2% at relative level <0.844, pane; B) and serum miR- 320e levels could discriminate between the SE and HGD groups with an AUC of 0,786 (optimal sensitivity of 82,3% and specificity of 62,2% at relative level <0.838, panel C). AUC: area under curve. CI: 95% confidence interval. SE: squamous epithelium study group. BE: Barrett’s metaplasia study group. HGD: high-grade dysplasia study group.

Discussion

The current lack of easily accessible diagnostic biomarkers of BE and EAC is a significant contributor to the poor overall survival of EAC patients. MiRNA’s are attractive as potential biomarkers, but currently there are no data regarding circulating miRNA levels in patients with BE or HGD. This is the first study to investigate the levels of serum miRNA’s in patients at all stages of the metaplasia-dysplasia-carcinoma sequence. Using the Nanostring platform we identified 10 miRNA’s with a fold change >2 between different study groups. Using qRT-PCR in an extended group of patients we were able to validate

118

Proefschrift_Kirill.indb 118 10-05-15 23:47 that levels of circulating miR-199a-3p and miR-320e are decreased in patients with BE and HGD compared to the SE study group. MiR-199a-3p is a known tumor suppressor miRNA. In vitro studies showed that miR-199a-3p reduces cell proliferation and migration through a direct repression of mTOR and c-MET (15-18) and miR-199a-3p tissue levels were decreased in endometrial, thyroid, prostate and hepatocellular carcinoma (15-19). A single previous study studied miR- 199a-3p expression during the malignant transformation of BE to EAC and found that compared to samples of SE, miR-199a-3p levels were significantly increased in EAC tissue samples, but not in BE or HGD samples (20). This finding makes it less likely that the decreased serum levels of miR-199a-3p observed in the current study is due to a passive release of tumor cell miRNA’s into the circulation. The observed decrease could be due to a selective decrease of miR-199a-3p secretion by the metaplastic cells, or alternatively the source of miR-199a-3p could be other than the esophageal epithelium. Currently little is known regarding the precise function of miR-320e in cancer, but one of its family members, i.e. miR-320a, has been studied in more detail. Mir-320a prevents neovascularization and reduces WNT signaling by targeting β-catenin, and miR- 320a expression is reduced in tissue of prostate cancer, osteosarcoma and oral squamous cell carcinoma (21-23). Moreover, low miR-320a tissue levels were associated with recurrence in colon cancer (24). The sequence similarity between miR-320e and miR-320a suggests that miR-320e may also function as a tumor-suppressor miRNA, however, this needs to be proven experimentally. We did not detect significant differences in miRNA levels between the four patient groups using the Nanostring platform, possibly due to the relatively small size of the study groups. Despite this, in the expanded cohort we were able to identify two miRNA’s with a known tumor-suppressor function whose serum levels were significantly decreased in BE and HGD patients. The lack of miRNA’s with increased levels during the metaplasia- dyslasia-carcinoma sequence is surprising, since high levels of tumor-specific circulating 6 miRNA’s were reported in other solid malignancies. The current study was limited by the number of EAC serum samples, and it is possible that larger future studies will identify miRNA’s whose levels increase in EAC compared with SE or BE. The origin of serum miRNA’s remains poorly understood. A paper by Pritchard et al. (25) showed that the majority of cancer-associated circulating miRNA’s, of note not including miR-199a-3p and miR-320e, were also strongly expressed in hematopoietic cells and were correlated with blood count levels, suggesting the possibility that changes in blood cell composition and/or integrity (e.g. hemolysis) can affect circulating miRNA levels. It has been recently shown that tumor-derived exosomes contain miRNA’s and are able to act as paracrine signals (26). On the basis of our study we are not able to discern the origin of miR-199a-3p and miR-320e, but miRNA’s from tumor-derived exosomes could

119

Proefschrift_Kirill.indb 119 10-05-15 23:47 provide a more reliable source of circulating miRNA’s compared to vesicle-free miRNA’s, and should be studied further in the context of EAC. Regardless of its biological function and origin, the decreased levels of miR-199a- 3p and miR-320e might identify patients with BE and HGD. While this finding needs further validation, the possibility that patients at risk of BE could be further stratified using a simple venapunture is exciting, since there is currently no blood-based biomarker for BE. Endoscopic screening of patients at risk of BE is currently not effective due to the low absolute incidence of BE, but risk stratification within this group using serum-derived miRNA’s could make endoscopic surveillance feasible for a subpopulation of patients most at risk for BE development. In conclusion, our study suggests that serum miRNA’s could help in identifying patients with BE and HGD, thus providing possible biomarkers enabling better screening, earlier detection and ultimately, better survival for EAC patients.

120

Proefschrift_Kirill.indb 120 10-05-15 23:47 Supplementary table 1: Extended patient characteristics

Age Study group Gender Smoking Cardiovascular Autoimmune Diabetes disease disease 69 SE Female Yes No No No 58 SE Male No No No No 59 SE Male No No No No 52 SE Male Yes Yes No No 61 SE Male No No No No 24 SE Male No No No No 70 SE Male No Yes No Yes 53 SE Female Yes No No No 53 SE Female No No No No 69 SE Male No No No No 23 SE Female No No No No 65 SE Female No No No No 55 SE Female No No No No 42 SE Female No No No No 47 SE Male No No No No 64 SE Female No Yes No No 61 SE Female No No No No 47 SE Female Yes No No No 59 SE Female No No Yes No 67 BE Male No No No No 68 BE Male No No No No 67 BE Male No Yes No No 48 BE Male Yes No No No 74 BE Male No Yes No No 69 BE Female Yes Yes No No 57 BE Female No No No Yes 6 63 BE Female Yes No No No 67 BE Female No No No No 50 BE Female No No No No 48 BE Male Yes No No No 59 BE Male No No No No 42 BE Female No No No No 61 BE Male No No No No 59 BE Male No No No No 70 BE Male No No No No 25 BE Male No No No No 68 BE Female No No No No

121

Proefschrift_Kirill.indb 121 10-05-15 23:47 Supplementary table 1 (continued)

Age Study group Gender Smoking Cardiovascular Autoimmune Diabetes disease disease 25 BE Male No No No No 65 BE Male No No No No 59 BE Female No No No No 60 BE Male No Yes No No 67 BE Male No Yes No No 75 BE Male No Yes No No 55 BE Male No No No No 73 BE Male No No Yes No 44 BE Male Yes No No No 68 HGD Male Yes No No Yes 53 HGD Female Yes No No No 70 HGD Male No No No No 68 HGD Male No No No No 51 HGD Male Yes No No No 69 HGD Male No No No No 74 HGD Female No Yes No Yes 62 HGD Female No No No Yes 62 HGD Male Yes No No Yes 67 HGD Male No Yes No Yes 65 HGD Male No No No Yes 71 HGD Male No No No No 58 HGD Male Yes Yes Yes No 42 HGD Female No No No No 82 HGD Male No No No No 69 HGD Male No Yes No No 75 HGD Male No No No No 61 EAC Male No No No No 65 EAC Female Yes No No No 73 EAC Male No No Yes No 66 EAC Male No No No No 64 EAC Male Yes Yes No No 46 EAC Male No Yes No No Samples used for the Nanostring array are denoted in bold.

122

Proefschrift_Kirill.indb 122 10-05-15 23:47

SupplementarySupplementary figure figure 1: Normalized 1: Normalized expression expression values of values63 serum of 63miRNA’s serum with miRNA’sexpression with above expression back- groundabove in the background Nanostring in platform.the Nanostring Cutoff platform. of expression was defined as 4 out of 6 of samples in at least one studyCutoff group havingof expression >30 counts. was defined Horizontal as 4 linesout of indicate 6 of samples the geometric in at least mean one study and thegroup error having bars >30denote the 95% confidencecounts. Horizontal interval. lines SE: indicate squamous the geometric epithelium mean study and group. the error BE: bars Barrett’s denote the metaplasia 95% confidence study group. HGD: high-gradeinterval. dysplasia SE: squamous study group.epithelium EAC: study esophageal group. BE: adenocarcinoma Barrett’s metaplasia study study group. group. HGD: high- grade dysplasia study group. EAC: esophageal adenocarcinoma study group.

A A B B A A B B

C C D D C C D D 6

Supplementary figure 2: Nanostring and full cohort graphs for miR-122-5p and miR-233-3p. SupplementaryY-axisSupplementary denotes figure normalized 2:figure Nanostring 2: levels Nanostring andon thefull Nanostring andcohort full graphscohort platform forgraphs (panelsmiR-122-5p for A,C)miR-122-5p orand RT-qPCR miR-233-3p. and (panelsmiR-233-3p. Y-axis B,D). denotes Y-axis denotes normalizedSupplementaryNumbersnormalizedSupplementary levels figure indicate on levels the 2:figure foldNanostringonNanostring the changes 2: NanostringNanostring platform and >2 observedfull platform andcohort (panels full on graphs thecohort (panelsA,C) Nanostring fororgraphs RT-qPCRA,C)miR-122-5p foror platform. RT-qPCR(panelsmiR-122-5p and Horizontal (panels B,D).miR-233-3p. andNumbers B,D). barsmiR-233-3p. denoteNumbersY-axis indicate denotes Y-axis indicate fold denotes fold changesnormalizedgeometricchangesnormalized >2 levelsobserved >2 mean. on levelsobserved theon SE: theNanostring on squamous Nanostring theon theNanostring epitheliumNanostringplatform platform. platform (panelsstudy platform. Horizontal group. (panelsA,C) BE:Horizontalor barsRT-qPCRBarrett’sA,C) denoteor barsRT-qPCRmetaplasia(panels geometricdenote (panelsB,D). study geometric Numbers mean. group.B,D). SE:Numbers HGD:indicatemean. squamous SE: indicate fold squamous fold epitheliumchangeshigh-gradeepitheliumchanges >2 studyobserved >2 group. dysplasia studyobserved on BE: thegroup. study Barrett’sNanostring on BE: thegroup. Barrett’sNanostring metaplasiaEAC: platform. esophageal metaplasia platform. study Horizontal adenocarcinomagroup. study Horizontal bars HGD:group. denote high-grade bars studyHGD: geometricdenote group.high-grade dysplasia geometric mean. dysplasia study SE:mean. squamousgroup. study SE: EAC:squamousgroup. EAC: esophagealepitheliumesophagealepithelium studyadenocarcinoma group. studyadenocarcinoma BE: group. Barrett’s study BE: group. Barrett’s study metaplasia group. metaplasia study group. study HGD:group. high-grade HGD: high-grade dysplasia dysplasia study group. study EAC:group. EAC: esophagealesophageal adenocarcinoma adenocarcinoma study group. study group. 123

Proefschrift_Kirill.indb 123 10-05-15 23:47 References

(1) van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med. 2012;366(22):2074-2084. (2) Qumseya BJ, Panossian AM, Rizk C, et al. Survival in esophageal high-grade dysplasia/ adenocarcinoma post endoscopic resection. Dig Liver Dis. 2013;45(12):1028-1033. (3) Fountoulakis A, Zafirellis KD, Dolan K, Dexter SP, Martin IG, Sue-Ling HM. Effect of surveillance of barrett’s oesophagus on the clinical outcome of oesophageal cancer. Br J Surg. 2004;91(8):997- 1003. (4) Corley DA, Levin TR, Habel LA, Weiss NS, Buffler PA. Surveillance and survival in barrett’s adenocarcinomas: A population-based study. Gastroenterology. 2002;122(3):633-640. (5) Shaheen N, Ransohoff DF. Gastroesophageal reflux, barrett esophagus, and esophageal cancer: Scientific review. JAMA. 2002;287(15):1972-1981. (6) Spechler SJ, Souza RF. Barrett’s esophagus. N Engl J Med. 2014;371(9):836-845. (7) Croce CM. Causes and consequences of microRNA dysregulation in cancer. Nat Rev Genet. 2009;10(10):704-714. (8) Lujambio A, Lowe SW. The microcosmos of cancer. Nature. 2012;482(7385):347-355. (9) Schwarzenbach H, Nishida N, Calin GA, Pantel K. Clinical relevance of circulating cell-free microRNAs in cancer. Nat Rev Clin Oncol. 2014;11(3):145-156. (10) Wijnhoven BP, Hussey DJ, Watson DI, et al. MicroRNA profiling of barrett’s oesophagus and oesophageal adenocarcinoma. Br J Surg. 2010;97(6):853-861. (11) Revilla-Nuin B, Parrilla P, Lozano JJ, et al. Predictive value of MicroRNAs in the progression of barrett esophagus to adenocarcinoma in a long-term follow-up study. Ann Surg. 2013;257(5):886- 893. (12) Yang H, Gu J, Wang KK, et al. MicroRNA expression signatures in barrett’s esophagus and esophageal adenocarcinoma. Clin Cancer Res. 2009;15(18):5744-5752. (13) Feber A, Xi L, Luketich JD, et al. MicroRNA expression profiles of esophageal cancer. J Thorac Cardiovasc Surg. 2008;135(2):255-60; discussion 260. (14) Kluiver J, Slezak-Prochazka I, van den Berg A. Studying microRNAs in lymphoma. Methods Mol Biol. 2013;971:265-276. (15) Minna E, Romeo P, De Cecco L, et al. miR-199a-3p displays tumor suppressor functions in papillary thyroid carcinoma. Oncotarget. 2014;5(9):2513-2528. (16) Wu D, Huang HJ, He CN, Wang KY. MicroRNA-199a-3p regulates endometrial cancer cell proliferation by targeting mammalian target of rapamycin (mTOR). Int J Gynecol Cancer. 2013;23(7):1191-1197. (17) Fornari F, Milazzo M, Chieco P, et al. MiR-199a-3p regulates mTOR and c-met to influence the doxorubicin sensitivity of human hepatocarcinoma cells. Cancer Res. 2010;70(12):5184-5193. (18) Qu KZ, Zhang K, Li H, Afdhal NH, Albitar M. Circulating microRNAs as biomarkers for hepatocellular carcinoma. J Clin Gastroenterol. 2011;45(4):355-360. (19) Qu Y, Huang X, Li Z, et al. miR-199a-3p inhibits aurora kinase A and attenuates prostate cancer growth: New avenue for prostate cancer treatment. Am J Pathol. 2014;184(5):1541-1549. (20) Streppel MM, Pai S, Campbell NR, et al. MicroRNA 223 is upregulated in the multistep progression of barrett’s esophagus and modulates sensitivity to chemotherapy by targeting

124

Proefschrift_Kirill.indb 124 10-05-15 23:47 PARP1. Clin Cancer Res. 2013;19(15):4067-4078. (21) Hsieh IS, Chang KC, Tsai YT, et al. MicroRNA-320 suppresses the stem cell-like characteristics of prostate cancer cells by downregulating the wnt/beta-catenin signaling pathway. Carcinogenesis. 2013;34(3):530-538. (22) Cheng C, Chen ZQ, Shi XT. MicroRNA-320 inhibits osteosarcoma cells proliferation by directly targeting fatty acid synthase. Tumour Biol. 2014;35(5):4177-4183. (23) Wu YY, Chen YL, Jao YC, Hsieh IS, Chang KC, Hong TM. miR-320 regulates tumor angiogenesis driven by vascular endothelial cells in oral cancer by silencing neuropilin 1. Angiogenesis. 2014;17(1):247-260. (24) Schepeler T, Reinert JT, Ostenfeld MS, et al. Diagnostic and prognostic microRNAs in stage II colon cancer. Cancer Res. 2008;68(15):6416-6424. (25) Pritchard CC, Kroh E, Wood B, et al. Blood cell origin of circulating microRNAs: A cautionary note for cancer biomarker studies. Cancer Prev Res (Phila). 2012;5(3):492-497. (26) Melo SA, Sugimoto H, O’Connell JT, et al. Cancer exosomes perform cell-independent MicroRNA biogenesis and promote tumorigenesis. Cancer Cell. 2014;26(5):707-721.

125

Proefschrift_Kirill.indb 125 10-05-15 23:47 Proefschrift_Kirill.indb 126 10-05-15 23:47 Chapter 7 Loss of CD44 and SOX2 expression is correlated with a poor prognosis in esophageal adenocarcinoma patients Authors: Honing J.1, Pavlov K.2 , Meijer C.3 , Smit J.K.1, Boersma van Ek W.2, Karrenbeld A.4, Burgerhof J.G.M.5, Kruyt F.A.E.3, Plukker J.T.M.1

Affiliations: 1Department of Surgery, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 3Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 4Department of Pathology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 5Department of Epidemiology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.

Ann Surg Oncol. 2014 Dec;21 Suppl 4:657-64.

Proefschrift_Kirill.indb 127 10-05-15 23:47 Abstract

Background: It has been suggested that markers associated with cancer stem cells (CSC) may play a role in esophageal cancer. Our aim was to investigate the expression pattern of proposed CSC markers ALDH1, Axin2, BMI1, CD44 and SOX2 in esophageal adenocarcinoma (EAC) and to relate their expression to survival. Methods: In this study we included 94 EAC patients, and examined the expression of the above mentioned markers using immunohistochemistry on tissue microarrays (TMA’s). Expression was scored as positive or negative, or categorized as low or high in terms of an Immuno-Reactivity-Score (IRS). Expression rates were related to clinicopathological characteristics, overall and disease free survival (OS and DFS). Results: In a multivariate analysis, negative expression of CD44 and of SOX2 were both significant prognostic factors for DFS (HR 1.73 CI 1.00-2.96 P=0.046 and HR 2.06 CI 1.14- 3.70 P=0.016). When CD44 and SOX2 expression were analyzed together, negative SOX2 expression was an independent prognostic factor for DFS (HR 1.91, CI 1.05-3.46, P=0.034). Low IRS scores for ALDH1 or Axin2 were associated with a reduced median survival (12.8 vs. 28.7 months and 12.1 vs. 25.5 months). However, these markers and BMI1 were not prognostic factors for survival. Conclusion: Loss of CD44 expression and loss of SOX2 expression are prognostic factors of poor survival in EAC patients. This suggests a role of these proteins in EAC that requires further investigation.

128

Proefschrift_Kirill.indb 128 10-05-15 23:47 Introduction

Esophageal carcinoma (EC) is a lethal malignancy with a rising incidence. Despite better staging methods and selection of surgically treated patients, the prognosis remains low, with an overall 5-year survival rate of 15-35%.(1, 2) Current standard treatment consists of neoadjuvant chemoradiation (CRT) followed by curative intended surgery, leading to an increased 5-year survival rate of up to 47%.(3) Research aimed to unravel the biological mechanisms driving tumorigenesis and disease progression in EC is expected to improve prognosis. In this regard, the cancer stem cell (CSC) model is of potential interest. According to this model a subgroup of cells, the CSCs, are at the basis of tumor formation, progression and metastatic spread, and give rise to tumor heterogeneity and therapy resistance. The CSC theory has been studied in various cancers including glioblastoma, breast and colon cancer, whereas several proteins specifically showed to characterize CSCs.(4-6) The WNT-pathway has been associated with maintenance of CSCs in several tumor types.(5, 7, 8) Target genes of the WNT/β-catenin pathway have been used as markers for detecting CSCs, including CD44.(9, 10) CD44 is a membrane protein, involved in cell adhesion and was shown to be associated with CSCs in gastric and colon cancer.(8, 11) Recently, high CD44 expression has been suggested as a possible marker of CSCs in EC and might also play a role in radiotherapy resistance.(12) Axin2 is a scaffold protein that functions as an antagonist of WNT signalling. Interestingly, in esophageal squamous cell cancer (ESCC) reduced Axin protein expression was related to a worse overall survival rate.(13) Besides targets of the WNT-pathway, other markers, such as Aldehyde dehydrogenase (ALDH1), BMI1 and SOX2, also have been used to identify CSCs in various tumor types.(14-16) ALDH1 is an enzyme that converts retinol into retinoic acid. Retinoic acid regulates gene expression and is involved in normal cell growth and differentiation. (17) In ESCC, high ALDH1 expression was related to a worse cause-specific survival rate. (18) BMI1 is an important regulator of self-renewal in haematopoietic, neuronal and small intestine stem cells.(16, 19-21) BMI1 expression was reported to increase during esophageal adenocarcinoma (EAC) development and may indicate a worse survival rate in ESCC and EAC patients.(22, 23) SOX2 is a transcription factor expressed in the endoderm during esophageal embryogenesis.(24) In the esophagus, SOX2 is expressed in progenitor cells in the basal layer and plays a role in proliferation.(25, 26) In several cancers, including 7 pancreatic cancer, SOX2 expression was associated with CSC properties.(15) Although it has been suggested that the above mentioned proteins are involved in EC, their expression and possible relation with patient outcome has scarcely been investigated in EAC. Therefore, our aim was to investigate the expression of ALDH1, Axin2, BMI1, CD44 and SOX2 within a large cohort of EAC patients, and to determine their relation to patient outcome.

129

Proefschrift_Kirill.indb 129 10-05-15 23:47 Patients and Methods

Patient material This retrospective study consisted of 104 patients with EAC, who underwent a curative intended esophagectomy without neoadjuvant treatment in our institute, between 1998 and 2008. Patients with in-hospital death (N=10) were excluded from further analysis. Follow-up was recorded until September 2013. The mean follow up time was 40.5 months ± 36.7 months. The study was conducted according to the guidelines of our Ethical Board (www.ccmo.nl). Archival tissue of all patients was handled according to the Dutch Code for proper use of Human Tissue (www.federa.org).

Immunohistochemistry A tissue microarray (TMA) was constructed of paraffin-embedded tumor resection specimens. From each tumor three representative cores were taken. In all cases a minimum of two cores per tumor were available for adequate evaluation. Antigen-retrieval was performed using citrate or EDTA buffer, followed by blocking of endogenous peroxidase. The sections were incubated with the primary antibody for 1 hour at room temperature or overnight at 4°C (BMI1). Primary antibodies used were: ALDH1 (clone 44/ALDH1, 1:1000, BD biosciences, San Jose, USA), Axin2 (ab 32197, 1:1000, Abcam, Cambridge, UK), BMI1 (clone 384509, 1:400, R&D systems, Abingdon, UK), CD44 (clone IM7, 1:100, Biolegend, London, UK.), SOX-2 (clone L1D6A2, 1:200; Cell Signaling, Leiden, the Netherlands) and KI-67 (clone MIB-1, 1:350, Dako, Heverlee, Belgium). Subsequently, tissue sections were incubated for 30 minutes with HRP-conjugated secondary and tertiary antibodies (1:50 dilution, all from Dako). Staining was visualized using 3,3'-diaminobenzidine and haematoxylin counterstaining. Adequate positive and negative controls (immunoglobulin class-matched control sera) for each staining were included.

Analysis of immunohistochemistry TMA slides were digitized using the Digital Slide Scanner NanoZoomer and NDP software (Hamamatsu, Shizuoka, Japan). Scoring was performed on the digitalized images using the same monitor and screen settings. Protein expression was scored by determining subcellular localization, intensity and percentage of positive cells. Scoring was performed by two independent observers, who were blinded for the clinicopathological characteristics at the time of scoring. Random samples of each staining were validated by a blinded expert gastrointestinal pathologist. The percentage of positive cells was scored into six categories: 0) no staining, 1) 1-5%, 2) 5-25%, 3) 25-50%, 4) 50-75% and 5) 75-100% positive cells. Intensity was scored as 0) negative, 1) weak, 2) medium and 3) strong. An Immuno-Reactivity Score (IRS) was calculated by multiplying the percentage of positive

130

Proefschrift_Kirill.indb 130 10-05-15 23:47 cells times the intensity score, resulting in a scale from 0 to 15. The IRS was divided into two groups: negative or low staining (IRS 0-5), and positive or high staining (IRS 6-15). For each staining the most appropriate scoring system was determined, either the presence of the marker or the IRS-score. KI-67 staining was scored as percentage of positive cells, and divided in a low (<50% positivity) and high (>50%) group.

Statistics Overall survival (OS) was defined as the time interval between the date of surgery and the documented day of death or last follow-up alive. Disease Free Survival (DFS) was defined as the time between date of surgery and date of recurrence, or death, or last follow up alive. Differences in patient characteristics per group and co-expression of markers were calculated using the Fisher’s exact test. Survival curves were calculated according to the Kaplan-Meier method and compared using the log-rank test. To identify prognostic factors for OS and DFS, univariate Cox-regression analyses were performed. Factors with multiple categories were made dichotomous by creating two subgroups. pT-stage was subdivided into low (pT1-T2) and high (pT3-T4) stage. pN-stage was divided into negative (pN0) and positive (pN1-3). Factors with a P-value <0.100 for either the OS or DFS, and which met the proportional hazards criteria in the univariate analysis were included in a stepwise multivariate Cox-regression. A P-value <0.05 was considered significant. Statistical analysis was performed by using the International Business Machines Statistical Package for Social Sciences (IBM SPSS, Armonk, New York, USA) version 20.0.

Results

Patient and clinicopathological characteristics Patient characteristics are shown in table 1. There was no difference between the groups for all tested markers (supplementary table 1), except for a higher percentage of microscopically irradical (R1) resections in SOX2 negative patients (P=0.049). The average age was 64.4 years. CD44 negative samples correlated with systemic recurrence in patients (P=0.006).

Marker expression pattern Representative images of all tested markers are shown in figure 1. ALDH1 expression was 7 detected in the majority of patient samples (66%) with 75-100% being positive, whereas only a small percentage of patients were negative for this marker (7.4%). Axin2 staining was detected in all patient samples and 25.5% of the patients had lower expression levels. BMI1 was detected in almost all patients (92.6%) in 75-100% of the cells, with a reduced expression in 48% of the patient samples. CD44 expression was variable, with 38.3% of

131

Proefschrift_Kirill.indb 131 10-05-15 23:47 Table 1: Patient characteristics. Bold values indicate significant P value (P<0.05). GEJ=Gastroesophageal Junction

CD44 SOX2 Low High P-value Low High P-value N 22 72 18 76 Sex 1.000 1.000 Male 20 63 16 67 Female 2 9 2 9 Age 0.811 0.063 <65 9 32 4 37 ≥65 13 40 14 39 pT 0.610 0.162 pT1-T2 6 25 3 28 pT3-T4 16 47 15 48 pN 0.064 0.050 pN0 3 26 2 27 ≥pN1 19 46 16 49 Resection 0.466 0.049 R1 4/22 8/72 5/18 7/76 Angio-invasion 0.274 0.076 8/22 17/72 8/18 17/76 Perineural growth 0.423 0.576 8/22 19/72 4/18 23/76 Lymphe-invasion 0.331 0.728 5/22 10/72 2/18 13/76 Localization 0.223 0.432 Mid 0 3 1 2 Distal 13 53 11 55 GEJ 9 16 6 19 Recurrence 0.006 0.421 19/22 39/72 5/18 45/76 Local recurrence * 0.406 0.050 7/21 17/71 6/18 18/74 Systemic recurrence * 0.006 0.604 16/21 29/71 10/18 35/74 * Of 3 patients site of recurrence was missing and 14 patients had both local and systemic recurrence.

132

Proefschrift_Kirill.indb 132 10-05-15 23:47 ALDH1

High Low Axin2

High Low BMI1

High Low CD44

Positive Negative 7 SOX2

Positive Negative

Figure 1: Immunohistochemical staining patterns of ALDH1, Axin2, BMI1, CD44 and SOX2

133

Proefschrift_Kirill.indb 133 10-05-15 23:47 the tumor samples having 75-100% positive stained cells, and no staining was detected in 23.4% of the samples. SOX2 expression was detected in both the nucleus and the cytoplasm of the tumor cells. A majority of patient samples (81%) was positive for SOX2.

Prognostic value of the tested markers Univariate analysis showed that age>65 years (P=0.011), high pT-stage (P=0.001), positive pN-stage (P=0.001) and angioinvasion (P=0.039) were predictors for poor OS (table 2). Except for patients with age>65 years (P=0.106), all above described parameters and R1-resection, were also related to a poor DFS (P<0.001; P=0.001; P=0.022 and P=0.011 respectively). The number of events was 71 for OS and 76 for DFS. Loss of ALDH1 (P=0.064/ P=0.015), Axin2 (P=0.027/ P=0.056), CD44 (P=0.044/ P=0.005) and SOX2 (P=0.002/ P=0.001) were related to a poor OS and DFS, respectively. BMI1 was not related to a poor OS or DFS (P=0.819 and P=0.777). In the multivariate analysis, all parameters that were related to either OS or DFS were included: age, pT-stage, pN-stage, R1-resection and angioinvasion. Both negative CD44 expression and negative SOX2 expression were

Table 2 Univariate analysis for overall survival and disease-free survival. Bold values indicate significant P value (P<0.100) and which met proportional hazards criteria were included in the multivariate analysis. OS=Overall Survival DFS=Disease-Free Survival GEJ=Gastroesophageal Junction

OS DFS N HR 95% CI P-value HR 95% CI P-value Sex (M/F) 83/11 1.33 0.64-2.79 0.445 1.26 0.63-2.54 0.512 Age (<65/≥65) 41/53 0.53 0.33-0.86 0.011 0.68 0.43-1.08 0.106 pT (pT1-T2/pT3-T4) 31/63 0.39 0.22-0.67 0.001 0.38 0.23-0.65 0.000 pN (pN0/≥pN1) 29/65 0.39 0.22-0.69 0.001 0.39 0.23-0.67 0.001 Resection (R0/R1) 82/12 0.60 0.32-1.14 0.120 0.45 0.24-0.83 0.011 Angio-invasion (No/Yes) 69/25 0.60 0.35-0.97 0.039 0.56 0.34-0.92 0.022 Perineural growth 67/27 0.79 0.48-1.31 0.359 0.67 0.42-1.09 0.103 Lymphe-invasion 79/15 1.05 0.55-2.00 0.873 1.08 0.58-2.00 0.816 Localization 3 0.052 0.097 Mid 66 4.78 1.35-17.0 3.91 1.12-13.64 Distal 25 1.36 0.77-2.38 1.11 0.66-1.88 GEJ Reference Reference ALDH1 (Low/High*) 27/66 1.61 097-2.65 0.064 1.83 1.12-2.96 0.015 Axin2 (Low/High*) 23/71 1.81 1.07-3.06 0.027 1.64 0.99-2.72 0.056 BMI1 (Low/High*) 46/48 1.06 0.66-1.68 0.819 1.07 0.68-1.68 0.777 CD44 (Neg/Pos*) 22/72 1.75 1.02-3.02 0.044 2.15 1.27-3.64 0.005 SOX2 (Neg/Pos*) 18/76 2.38 1.36-4.15 0.002 2.46 1.42-4.26 0.001 * Reference

134

Proefschrift_Kirill.indb 134 10-05-15 23:47 significant adverse predictors of DFS (table 3, HR 1.73 CI 1.00-2.96 P=0.046 andHR 2.06 CI 1.14-3.70 P=0.016). When both CD44 and SOX2 expression were included in the multivariate analysis only SOX2 was an independent prognostic factor for DFS (HR1.91, CI 1.05-3.46, P=0.034). Figure 2 shows the Kaplan Meier curves for CD44 and SOX2 in relation to OS and DFS. Median DFS for CD44 positive patients in comparison to CD44 negative patients was 24.9 vs. 12.1 months, respectively. Median DFS for SOX2 positive patients was 26.9 months, and 10.6 months for patients with negative SOX2 expression. Patients displaying low ALDH1 or low Axin2 expression showed a trend for poor OS and DFS (supplementary

a) a 100 100

P=0.042 P=0.004

80 80

60 60

CD44 positive CD44 positive (N=72) (N=72) Survival (%) 40 Survival (%) 40

20 20 CD44 negative CD44 negative (N=22) (N=22)

0 0

0 12 24 36 48 60 72 0 12 24 36 48 60 72 OS (months) DFS (months) b) b 100 100

P=0.002 P=0.001

80 80

60 60

SOX2 positive (N=76) SOX2 positive (N=76) Survival (%) Survival (%) 40 40 7

20 20

SOX2 negative SOX2 negative (N=18) (N=18)

0 0

0 12 24 36 48 60 72 0 12 24 36 48 60 72 OS (months) DFS (months)

Figure 2: Overall survival and disease-free survival for CD44 (panel A) and SOX2 (panel B). OS=Overall Survival DFS=Disease-Free Survival

135

Proefschrift_Kirill.indb 135 10-05-15 23:47 Table 3 Multivariate analysis for OS and DFS. Bold values indicate significant P value (P<0.05). OS=Overall Survival DFS=Disease-Free Survival

Multivariate analysis OS DFS

N=94 HR 95% CI P-value HR 95% CI P-value

Age≥65 2.10 1.28-3.46 0.003 1.63 1.02-2.60 0.042

High pT-stage (T3-T4) 2.14 1.17-3.89 0.013 1.99 1.11-3.56 0.021

pN1 1.78 0.95-3.34 0.074 1.73 0.95-3.15 0.074

R1 resection 1.07 0.54-2.13 0.846 1.31 0.67-2.56 0.436

Angio-invasion 1.39 0.81-2.39 0.236 1.36 0.80-2.30 0.256

Negative CD44 1.32 0.76-2.30 0.326 1.73 1.00-2.96 0.046

Age≥65 1.95 1.17-3.23 0.010 1.45 0.90-2.35 0.126

High pT-stage (T3-T4) 1.95 1.07-3.55 0.029 1.71 0.95-3.07 0.074

pN1 1.98 1.06-3.70 0.033 2.03 1.12-3.69 0.020

R1 resection 1.12 0.57-2.22 0.741 1.52 0.78-2.99 0.223

Angio-invasion 1.34 0.77-2.32 0.296 1.32 0.77-2.25 0.311

Negative SOX2 1.67 0.93-3.00 0.087 2.06 1.14-3.70 0.016

Age≥65 1.95 1.18-3.25 0.016 1.48 0.92-2.39 0.109

High pT-stage (T3-T4) 1.97 1.08-3.58 0.027 1.73 0.96-3.12 0.067

pN1 1.91 1.01-3.61 0.047 1.88 1.03-3.45 0.040

R1 resection 1.11 0.56-2.19 0.767 1.44 0.73-2.84 0.292

Angio-invasion 1.33 0.77-2.30 0.307 1.34 0.79-2.28 0.278

Negative CD44 1.19 0.66-2.12 0.564 1.59 0.92-2.74 0.096

Negative SOX2 1.59 0.87-2.93 0.135 1.91 1.05-3.46 0.034

figure 1). However, in a multivariate analysis, low ALDH1 (HR 1.28 P=0.351 andHR 1.40 P=0.200) and low Axin2 expression (HR 1.60 P=0.088 and HR 1.51 P=0.134) did not significantly correlate with OS and DFS (supplementary table 1).

Correlations between markers We further explored possible correlations between the markers using a Fisher’s exact test. Low ALDH1 and low BMI1 expression were significantly related (supplementary table 2, P=0.012). High Axin2 and CD44 expression were also often co-expressed (P=0.051). SOX2 expression significantly correlated with high ALDH1 and BMI1 expression (P<0.001 and P<0.001). KI-67 staining was not related to the other markers in the study.

136

Proefschrift_Kirill.indb 136 10-05-15 23:47 Discussion In this study we examined the prognostic value of a number of proposed CSC markers in EAC patient samples. We show that loss of CD44 and loss of SOX2 expression are associated with a poor DFS. In addition, negative CD44 expression was related to systemic tumor recurrence. Furthermore, patients with either low ALDH1 or Axin2 expression demonstrated a lower median survival (12.8 vs. 28.7 months and 12.1 vs. 25.5 months); however, this was not significantly related to OS and DFS in a multivariate analysis. These results are of interest, as the expression of these CSC markers has not been investigated in such a large group of EAC patients and in relation to survival outcome. Interestingly, loss of CD44 and SOX2 expression was related to worse outcome. This seems counterintuitive, as both markers are regarded CSC markers, and high CSC marker expression is expected to relate to a worse patient outcome. However, increasing evidence suggests that SOX2 could have multiple functions both an oncogenic as well as a tumor suppressor function. (25, 27) While little is known regarding the exact mechanisms through which CD44 and SOX2 exert their function, loss of CD44 and SOX2 expression has been related to poor patient outcome in several cancer types.(28, 29) Loss of CD44 expression has been related to malignant progression in various tumors, including breast, colon and esophageal cancer.(30-32) Decreased CD44 expression has been associated with tumor progression in breast cancer, where low CD44 expression correlated with invasive tumors.(32) In esophageal cancer, CD44 expression has been described as both a positive and negative prognostic factor in ESCC.(18, 33) In accordance with our results, low CD44 expression in EAC has previously been associated with poor prognosis in a group of 41 patients.(28) The notion that CD44 has a tumor suppressor role in the carcinogenesis of EAC is further supported by the finding that CD44 expression is also decreasesd in the malignant sequence from Barrett’s esophagus to EAC, however it is still unclear how loss of CD44 contributes to malignant progression.(30) CD44 is a downstream target of WNT signaling and associated with CSCs.(8, 34) Active WNT signaling is expected to be associated with an adverse prognosis, possibly due to a larger CSC fraction. However, De Sousa et al. reported that high expression of WNT target genes was associated with a favourable prognosis in colorectal cancer patients, possibly due to methylation of these genes during tumor progression.(35) Further research is required to understand the role of WNT signalling in EAC, and especially the effect of 7 epigenetic modulation of WNT and CSC target genes on survival outcome. We found that loss of SOX2 expression relates to a poor survival in EAC patients. Previous studies reported the expression pattern of SOX2 in EAC and ESCC patients, but expression was not correlated to patient outcome.(36, 37) In agreement with our results, high SOX2 expression was shown to be an independent positive prognostic factor in squamous lung carcinoma and non-small cell lung cancer (NSCLC).(29, 38, 39) Contrary

137

Proefschrift_Kirill.indb 137 10-05-15 23:47 to these results, other studies in breast and ovarian cancer indicated that high SOX2 expression had a negative prognostic impact.(40, 41). It is not known how loss of SOX2 contributes to a poor prognosis, but a tumor suppressor role of SOX2 has previously been described.(27) SOX2 is known to serve various functions in embryological development, and is implicated in several cancers.(24, 25, 42) Several studies suggest that SOX2 acts as a tumor suppressor in gastrointestinal cancers. In gastric cancer, SOX2 expression inhibited proliferation and induced apoptosis in vitro, and methylation of the SOX2 promoter was related to a worse prognosis.(27) Similarly in colon cancer, SOX2 overexpression reduced cell proliferation and tumor growth through induction of autophagy.(42) While in vitro studies on the role of SOX2 in EAC are lacking, these results suggest that SOX2 could also act as a tumor suppressor. This might explain our finding that loss of SOX2 is associated with an adverse survival outcome in EAC patients. However, overexpression of SOX2 in mice could induce squamous cell tumors (26), and an oncogenic role for SOX2 has been reported in squamous cell cancers of the esophagus and lung.(25) Tissue-specific interaction of SOX2 with other transcription factors could modulate the function of SOX2. A recent paper reported that SOX2 interacts with p63 in ESCC, while in embryonic stem cells SOX2 preferentially interacts with OCT4, a known inducer of pluripotency.(43) A tissue-specific interaction with other transcription factors could explain how SOX2 might play a dual role in stem cells and the development of different cancer types. In our study we detected both nuclear and cytoplasmic SOX2 staining. The functional role of SOX2 in the cytoplasm is currently not well understood, but cytoplasmic SOX2 has been detected in mouse embryonic stem cells,(44) and also in residual cells in colorectal cancer after chemoradiation.(45) Considering the reported contradictory functions of SOX2 and different subcellular localization, it would be interesting to study the role of SOX2 in EAC carcinogenesis in more detail. A limitation of this study is that some patients with recurrence were treated, following standard care, with palliative chemotherapy or radiotherapy. This may have influenced the OS and we therefore consider the DFS as a more reliable outcome measure. In conclusion, in this study we showed that loss of CD44 and SOX2 expression are associated with a poor prognosis in a homogenous cohort of EAC patients. The functional role of these proteins in esophageal carcinogenesis remains to be further investigated.

138

Proefschrift_Kirill.indb 138 10-05-15 23:47 Supplementary table 1: Patient characteristics per group

ALDH1 Axin2 BMI1 Low High P-value Low High P-value Low High P-value N 27 67 23 71 46 48 Sex 0.724 1.000 0.350 Male 24 60 20 63 39 44 Female 3 7 3 8 7 4

Age 0.109 0.639 1.000 <65 8 33 9 32 20 21 ≥65 19 34 14 39 26 27

pT 0.226 0.213 0.192 pT1-T2 6 25 5 26 12 19 pT3-T4 21 42 18 45 34 29

pN 0.326 0.616 0.659 pN0 6 23 6 23 13 16 ≥pN1 21 44 17 48 33 32

Resection 0.096 0.724 0.759 R1 6/27 6/67 2/23 10/71 5/46 7/48

Angio-invasion 0.197 0.415 0.487 10/27 15/67 8/23 17/71 14/46 11/48

Perineural growth 0.616 0.288 0.652 9/27 18/67 9/23 18/71 12/46 15/48

Lymphe-invasion 0.757 1.000 0.576 5/27 10/67 4/23 11/71 6/46 9/48

Localization 0.073 0.083 0.929 Mid 2 1 2 1 2 1 Distal 15 51 13 53 32 34 GEJ 10 15 8 17 12 13

Recurrence 0.641 0.807 0.672 18/27 40/67 15/23 43/71 27/46 31/48

Local recurrence * 0.311 0.579 0.477 7 9/27 15/65 7/22 17/70 10/46 14/46

Systemic recurrence * 0.650 1.000 0.677

12/27 33/65 11/22 34/70 21/46 24/46

139

Proefschrift_Kirill.indb 139 10-05-15 23:47 Supplementary Table 2: Multivariate analysis ALDH1 and Axin2

Multivariate analysis OS DFS

N=94 HR 95% CI P-value HR 95% CI P-value

Age≥65 2.05 1.25-3.39 0.005 1.54 0.95-2.48 0.078

High pT-stage (T3-T4) 2.05 1.11-3.76 0.021 1.85 1.01-3.37 0.045

pN1 1.88 1.01-3.53 0.048 1.88 1.03-3.43 0.039

R1 resection 1.05 0.53-2.11 0.882 1.36 0.69-2.68 0.373

Angio-invasion 1.38 0.79-2.38 0.255 1.28 0.74-2.21 0.381

Low ALDH1 1.28 0.76-2.16 0.351 1.40 0.84-2.34 0.200

Age≥65 2.09 1.27-3.45 0.004 1.59 0.99-2.55 0.053

High pT-stage (T3-T4) 2.00 1.09-3.69 0.026 1.82 0.99-3.36 0.054

pN1 1.79 0.95-3.37 0.071 1.80 0.98-3.31 0.057

R1 resection 1.22 0.60-2.40 0.604 1.58 0.79-3.16 0.194

Angio-invasion 1.44 0.86-2.48 0.194 1.36 0.79-2.32 0.265

Low Axin2 1.60 0.93-2.75 0.088 1.51 0.89-2.58 0.134

140

Proefschrift_Kirill.indb 140 10-05-15 23:47 0.110 0.454 0.107 0.168 0.252 P-value 9 3 3 7 26 17 23 16 10 23 19 High KI-67 11 68 18 50 20 48 30 38 19 49 57 Low 0.000 0.763 0.000 0.120 P-value 76 15 61 18 58 30 46 15 61 High SOX2 6 5 2 7 11 18 12 13 16 Low 0.181 0.051 0.468 P-value 72 18 54 14 58 37 35 High CD44

9 9 9 22 13 13 13 Low 0.012 0.233 P-value 8 9 48 40 39 High BMI1 46 19 27 14 32 Low 0.596 P-value 71 19 52 High Axin2 8 23 15 Low 7 27 67 23 71 46 48 22 72 18 76 Low High Low High Low High Low High Low High ALDH1 Axin2 BMI1 CD44 SOX2 Supplementary table 3: Marker co-expression

141

Proefschrift_Kirill.indb 141 10-05-15 23:47 Supplementary figure 1: Overall survival and disease-free survival for ALDH1, Axin2 and BMI1

100 100 P=0.062 P=0.014

80 80

60 60

ALDH1 high ALDH1 high (N=67) (n=67) Survival (%) Survival (%) 40 40

20 20

ALDH1 low ALDH1 low (N=27) (n=27)

0 0

0 12 24 36 48 60 72 0 12 24 36 48 60 72 OS (months) DFS (months)

100 100 P=0.025 P=0.053

80 80

60 60

Axin2 high (N=71) Axin2 high Survival (%) Survival (%) 40 40 (N=71)

20 20

Axin2 low Axin2 low (N=23) (N=23) 0 0

0 12 24 36 48 60 72 0 12 24 36 48 60 72 OS (months) DFS (months)

100 100 P=0.819 P=0.777

80 80

60 60

BMI1 high BMI1 high

(N=48) Survival (%) (N=48) Survival (%) 40 40

BMI1 low BMI1 low (N=46) (N=46) 20 20

0 0

0 12 24 36 48 60 72 0 12 24 36 48 60 72 OS (months) DFS (months)

142

Proefschrift_Kirill.indb 142 10-05-15 23:47 References

(1) Pennathur A, Gibson MK, Jobe BA, Luketich JD. Oesophageal carcinoma. Lancet 2013: 381:400- 412. (2) http://www.cijfersoverkanker.nl/selecties/Dataset_2/img512d09a0bce07. [accessed 08-09- 2013, 2013). (3) van Hagen P, Hulshof MC, van Lanschot JJ et. al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 2012: 366:2074-2084. (4) Ponti D, Costa A, Zaffaroni N et. al. Isolation and in vitro propagation of tumorigenic breast cancer cells with stem/progenitor cell properties. Cancer Res 2005: 65:5506-5511. (5) Vermeulen L, De Sousa E Melo F, van der Heijden M et. al. Wnt activity defines colon cancer stem cells and is regulated by the microenvironment. Nat Cell Biol 2010: 12:468-476. (6) Singh SK, Hawkins C, Clarke ID et. al. Identification of human brain tumour initiating cells. Nature 2004: 432:396-401. (7) Barker N, Ridgway RA, van Es JH et. al. Crypt stem cells as the cells-of-origin of intestinal cancer. Nature 2009: 457:608-611. (8) Yeung TM, Gandhi SC, Wilding JL, Muschel R, Bodmer WF. Cancer stem cells from colorectal cancer-derived cell lines. Proc Natl Acad Sci U S A 2010: 107:3722-3727. (9) Lustig B, Jerchow B, Sachs M et. al. Negative feedback loop of Wnt signaling through upregulation of conductin/axin2 in colorectal and liver tumors. Mol Cell Biol 2002: 22:1184-1193. (10) Zeilstra J, Joosten SP, van Andel H et. al. Stem cell CD44v isoforms promote intestinal cancer formation in Apc(min) mice downstream of Wnt signaling. Oncogene 2013. (11) Takaishi S, Okumura T, Tu S et. al. Identification of gastric cancer stem cells using the cell surface marker CD44. Stem Cells 2009: 27:1006-1020. (12) Smit JK, Faber H, Niemantsverdriet M et. al. Prediction of response to radiotherapy in the treatment of esophageal cancer using stem cell markers. Radiother Oncol 2013: 107:434-441. (13) Li AF, Hsu PK, Tzao C, Wang YC, Hung IC, Huang MH, Hsu HS. Reduced axin protein expression is associated with a poor prognosis in patients with squamous cell carcinoma of esophagus. Ann Surg Oncol 2009: 16:2486-2493. (14) Deng S, Yang X, Lassus H et. al. Distinct expression levels and patterns of stem cell marker, aldehyde dehydrogenase isoform 1 (ALDH1), in human epithelial cancers. PLoS One 2010: 5:e10277. (15) Herreros-Villanueva M, Zhang JS, Koenig A et. al. SOX2 promotes dedifferentiation and imparts stem cell-like features to pancreatic cancer cells. Oncogenesis 2013: 2:e61. (16) Tian H, Biehs B, Warming S, Leong KG, Rangell L, Klein OD, de Sauvage FJ. A reserve stem cell population in small intestine renders Lgr5-positive cells dispensable. Nature 2011: 478:255-259. (17) Ghiaur G, Yegnasubramanian S, Perkins B, Gucwa JL, Gerber JM, Jones RJ. Regulation of 7 human hematopoietic stem cell self-renewal by the microenvironment’s control of retinoic acid signaling. Proc Natl Acad Sci U S A 2013: 110:16121-16126. (18) Minato T, Yamamoto Y, Seike J et. al. Aldehyde dehydrogenase 1 expression is associated with poor prognosis in patients with esophageal squamous cell carcinoma. Ann Surg Oncol 2013: 20:209-217. (19) Lessard J, Sauvageau G. Bmi-1 determines the proliferative capacity of normal and leukaemic

143

Proefschrift_Kirill.indb 143 10-05-15 23:47 stem cells. Nature 2003: 423:255-260. (20) Molofsky AV, He S, Bydon M, Morrison SJ, Pardal R. Bmi-1 promotes neural stem cell self- renewal and neural development but not mouse growth and survival by repressing the p16Ink4a and p19Arf senescence pathways. Genes Dev 2005: 19:1432-1437. (21) Yan KS, Chia LA, Li X et. al. The intestinal stem cell markers Bmi1 and Lgr5 identify two functionally distinct populations. Proc Natl Acad Sci U S A 2012: 109:466-471. (22) Choy B, Bandla S, Xia Y et. al. Clinicopathologic characteristics of high expression of Bmi-1 in esophageal adenocarcinoma and squamous cell carcinoma. BMC Gastroenterol 2012: 12:146- 230X-12-146. (23) Yoshikawa R, Tsujimura T, Tao L, Kamikonya N, Fujiwara Y. The oncoprotein and stem cell renewal factor BMI1 associates with poor clinical outcome in oesophageal cancer patients undergoing preoperative chemoradiotherapy. BMC Cancer 2012: 12:461-2407-12-461. (24) Que J, Okubo T, Goldenring JR et. al. Multiple dose-dependent roles for Sox2 in the patterning and differentiation of anterior foregut endoderm. Development 2007: 134:2521-2531. (25) Bass AJ, Watanabe H, Mermel CH et. al. SOX2 is an amplified lineage-survival oncogene in lung and esophageal squamous cell carcinomas. Nat Genet 2009: 41:1238-1242. (26) Liu K, Jiang M, Lu Y et. al. Sox2 cooperates with inflammation-mediated Stat3 activation in the malignant transformation of foregut basal progenitor cells. Cell Stem Cell 2013: 12:304-315. (27) Otsubo T, Akiyama Y, Yanagihara K, Yuasa Y. SOX2 is frequently downregulated in gastric cancers and inhibits cell growth through cell-cycle arrest and apoptosis. Br J Cancer 2008: 98:824- 831. (28) Bottger TC, Youssef V, Dutkowski P, Maschek H, Brenner W, Junginger T. Expression of CD44 variant proteins in adenocarcinoma of Barrett’s esophagus and its relation to prognosis. Cancer 1998: 83:1074-1080. (29) Chen Y, Huang Y, Huang Y, Chen J, Wang S, Zhou J. The prognostic value of SOX2 expression in non-small cell lung cancer: a meta-analysis. PLoS One 2013: 8:e71140. (30) Darlavoix T, Seelentag W, Yan P, Bachmann A, Bosman FT. Altered expression of CD44 and DKK1 in the progression of Barrett’s esophagus to esophageal adenocarcinoma. Virchows Arch 2009: 454:629-637. (31) Ngan CY, Yamamoto H, Seshimo I et. al. A multivariate analysis of adhesion molecules expression in assessment of colorectal cancer. J Surg Oncol 2007: 95:652-662. (32) Park SY, Lee HE, Li H, Shipitsin M, Gelman R, Polyak K. Heterogeneity for stem cell-related markers according to tumor subtype and histologic stage in breast cancer. Clin Cancer Res 2010: 16:876-887. (33) Gotoda T, Matsumura Y, Kondo H et. al. Expression of CD44 variants and prognosis in oesophageal squamous cell carcinoma. Gut 2000: 46:14-19. (34) Dalerba P, Dylla SJ, Park IK et. al. Phenotypic characterization of human colorectal cancer stem cells. Proc Natl Acad Sci U S A 2007: 104:10158-10163. (35) de Sousa E Melo F, Colak S, Buikhuisen J et. al. Methylation of cancer-stem-cell-associated Wnt target genes predicts poor prognosis in colorectal cancer patients. Cell Stem Cell 2011: 9:476-485. (36) DiMaio MA, Kwok S, Montgomery KD, Lowe AW, Pai RK. Immunohistochemical panel for distinguishing esophageal adenocarcinoma from squamous cell carcinoma: a combination of p63, cytokeratin 5/6, MUC5AC, and anterior gradient homolog 2 allows optimal subtyping.

144

Proefschrift_Kirill.indb 144 10-05-15 23:47 Hum Pathol 2012: 43:1799-1807. (37) Long KB, Hornick JL. SOX2 is highly expressed in squamous cell carcinomas of the gastrointestinal tract. Hum Pathol 2009: 40:1768-1773. (38) Velcheti V, Schalper K, Yao X et. al. High SOX2 levels predict better outcome in non-small cell lung carcinomas. PLoS One 2013: 8:e61427. (39) Wilbertz T, Wagner P, Petersen K et. al. SOX2 gene amplification and protein overexpression are associated with better outcome in squamous cell lung cancer. Mod Pathol 2011: 24:944-953. (40) Huang YH, Luo MH, Ni YB et. al. Increased SOX2 expression in less differentiated breast carcinomas and their lymph node metastases. Histopathology 2014: 64:494-503. (41) Zhang J, Chang DY, Mercado-Uribe I, Liu J. Sex-determining region Y-box 2 expression predicts poor prognosis in human ovarian carcinoma. Hum Pathol 2012: 43:1405-1412. (42) Cho YY, Kim DJ, Lee HS et. al. Autophagy and cellular senescence mediated by Sox2 suppress malignancy of cancer cells. PLoS One 2013: 8:e57172. (43) Watanabe H, Ma Q, Peng S et. al. SOX2 and p63 colocalize at genetic loci in squamous cell carcinomas. J Clin Invest 2014. (44) Li J, Pan G, Cui K, Liu Y, Xu S, Pei D. A dominant-negative form of mouse SOX2 induces trophectoderm differentiation and progressive polyploidy in mouse embryonic stem cells. J Biol Chem 2007: 282:19481-19492. (45) Saigusa S, Tanaka K, Toiyama Y et. al. Correlation of CD133, OCT4, and SOX2 in rectal cancer and their association with distant recurrence after chemoradiotherapy. Ann Surg Oncol 2009: 16:3488-3498.

145

Proefschrift_Kirill.indb 145 10-05-15 23:47 Proefschrift_Kirill.indb 146 10-05-15 23:47 Chapter 8 CD44, SHH and SOX2 as novel biomarkers in esophageal cancer patients treated with neoadjuvant chemoradiotherapy Authors: Honing J.1, Pavlov K.2, Karrenbeld A.3, Meijer C.4 , Faiz Z.1, Smit J.K.1, Mul V.E.M.5, Hospers G.A.P.4, Burgerhof J.G.M.6, Kruyt F.A.E.4, Kleibeuker J.H.2, Plukker J.T.M.1

Affiliations: 1Department of Surgery, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 2Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 3Department of Pathology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 4Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 5Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. 6Department of Epidemiology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands.

Submitted

Proefschrift_Kirill.indb 147 10-05-15 23:47 Abstract

Background: Neoadjuvant chemoradiotherapy (nCRT) improves survival in esophageal cancer (EC) patients, but the response is heterogeneous and little is known regarding prognostic and predictive markers in OC patients undergoing nCRT. CD44, SHH and SOX2 have been implicated in resistance to CRT, possibly through an association with a cancer stem cell phenotype. Methods: A total of 101 EC patients, treated with nCRT and surgery, were included. Sufficient pre-treatment biopsy material was present in 71 patients, of which 53 patients were non-complete responders on nCRT (nCR). Protein expression was examined using immunohistochemistry (IHC). Prognostic factors were determined using Cox regression analysis for disease free survival (DFS) and cause specific survival (CSS) in the complete cohort, the pre-treatment biopsies group and post-treatment nCR group. Results: Low CD44 expression in the nCR group was an independent prognostic factor for both DFS and CSS (DFS HR 2.81, p=0.002 and CSS HR 3.48, p=0.002). Low SHH in pretreatment biopsies was an independent prognostic factor for a poor DFS (HR 2.27, p=0.036) while absent SOX2 expression in preoperative biopsies was related to systemic recurrence (p=0.029). No relation between marker expression and response to nCRT was observed. Conclusion: Low expression of CD44 and SHH are associated with a poor survival outcome in OC patients treated with nCRT.

148

Proefschrift_Kirill.indb 148 10-05-15 23:47 Introduction

Eesophageal cancer (EC) is a common cancer with a poor prognosis. Two major histological EC subtypes are recognized: Esophageal Adenocarcinoma (EAC) and Esophageal Squamous Cell Carcinoma (ESCC). The treatment for both subtypes is similar, and historically consisted of a radical esophagectomy. In the past decade neoadjuvant chemoradiotherapy (nCRT) has emerged as a valuable addition to surgery, increasing median survival from 24.0 to 49.4 months (1). However, the response to nCRT is heterogenous. Whereas approximately 25% of EAC patients will achieve a pathological complete response (pCR) after nCRT, others respond partially or not at all (1). Currently, the differences underlying the response to therapy are poorly understood, and predictive and prognostic biomarkers are lacking in EC patients treated with CRT. The glycoprotein CD44, the Sex determining region Y-box 2 (SOX2) transcription factor and the hedgehog (HH) signalling pathway ligand Sonic Hedgehog (SHH) have been associated with resistance to chemo- and radiotherapy. CD44 and SOX2 are potential markers of Cancer Stem Cells (CSCs), a subpopulation of tumour cells with stem cell characteristics such as the ability to self-renew and generate multilineage offspring. CD44 and SOX2 have been associated with CSCs in colon cancer and glioblastoma (2,3). CSCs have been linked to chemo- and radiotherapy resistance (4), and CD44 expression has been associated with radiation-resistance in colon, head and neck and esophageal cancers (5-7) while SOX2 expression has been associated with both good response to radiotherapy (8,9) as well as with radiotherapy resistance (10). Previous work from our and other groups revealed that low CD44 and SOX2 expression are related with poor prognosis in EC patients treated with surgery alone (11,12), but the prognostic and predictive value in EC patients treated with nCRT is unknown. The hedgehog signalling pathway is a known regulator of the CSC in various solid malignancies (reviewed in (13)). Increased expression of SHH has been implicated in the development of EC and EC radiotherapy resistance in vitro (14-16), but the prognostic and predictive value of SHH expression in tissue of EC patients undergoing nCRT is unknown. The aim of this study was to determine the prognostic value and the predictive value for response to nCRT of the expression of CD44, SHH and SOX2 in pre-treatment biopsies and post-nCRT resection material of EC patients treated with combined nCRT and surgery.

149

Proefschrift_Kirill.indb 149 10-05-15 23:47 Patients and methods

Patient material This retrospective study consisted of 101 patients with EC who were treated with nCRT followed by an esophagectomy with curative intent at the University Medical Center Groningen, Groningen, the Netherlands between 2006 and 2013. Patients who died within 90 days after surgery or those treated with surgery alone were not included. Follow-up was recorded until March 2013, with a minimum follow-up of one year. The median follow-up was 22.2 (inter-quartile range 12.8-35.5) months. All data were prospectively collected from the patients’ medical records.

Total patients (N=101)

Pre-treatment biopsy group (N=71) Pre-treatment biopsy absent (N=30)

Complete pathological No complete pathologi- response (N=14) cal response (N=56)#

Post-treatment nCR Only tumor in lymph group nodes ypT0N1 (N=3) ypT≥1 N≥1 (N=53)

Positive lymph nodes (N=23)

Material from recurrence (N=12)

Figure 1: Schematic representation of patient material included. Patient material used in this study consisted of: 1) diagnostic endoscopic biopsies obtained before therapy; 71 of the 101 samples had sufficient material to be used for research purposes, 2) resection material of the patients with microscopic residual disease (N=53) and 3) positive lymph nodes of patients (N=36) with microscopic residual disease and 4) material of recurrent disease (N=18). The pre-treatment biopsy group (N=71) and post-treatment non-complete responder (nCR) group (N=53) are highlighted in grey. # One patient with a Mandard 2 score had no vital tumour cells in the resection specimens. * This patient group had no tumour tissue available and was thus excluded from further analysis in the marker expression analysis.

150

Proefschrift_Kirill.indb 150 10-05-15 23:47 pCR was defined as the absence of microresidual disease (mRD) in the resection specimen. mRD was defined as the presence of vital tumour cells, either at the primary tumour site (ypT) and/or in the lymph nodes (ypN) at postoperative pathological evaluation. Pathological response to therapy was determined on a 1-5 scale according to the Mandard scoring system (17). A positive circumferential margin (CRM) was defined by the presence of tumour cells within 1mm (≤1mm) from the resection margin. Figure 1 gives a schematic representation of the patient material included. Thirty patients were excluded because insufficient biopsy material was present to be used for research purposes as determined by the Dutch Code for proper use of Human Tissue (www.federa.org). Two groups were analyzed for marker expression in relation to survival; the pre-treatment biopsy group (n=71) and the post-treatment non-complete response (nCR) group. The latter group consisted of 53 of the 71 patients that had a non-complete response to nCRT. The study was conducted according to the guidelines of the Ethical Board of our institute (www.ccmo.nl). Archival tissue of all patients was handled according to the Dutch Code for proper use of Human Tissue (www.federa.org).

Neoadjuvant chemoradiotherapy Neoadjuvant chemoradiotherapy was given according to the CROSS-scheme (1). Carboplatin (AUC 2) and paclitaxel (50mg/m2) were given weekly during radiotherapy for five weeks. Radiotherapy was given in 23 fractions of 1.8Gy with a total of 41.4Gy. Three patients with a GEJ tumour received 45Gy.

Surgical procedure Patients were operated by an experienced surgical team at the University Medical Center Groningen, Groningen, the Netherlands. Resection was performed via a transthoracic route and a laparotomy with a two-field nodal dissection in the mediastinum and abdomen respectively, as previously described (18).

Immunohistochemistry Paraffin-embedded tumor specimens were retrieved from pathological archives of hospitals in the Northeast of the Netherlands. Antigen-retrieval was performed using citrate (pH 6.0) or EDTA buffer (pH 9.0), followed by blocking of endogenous peroxidase. The sections were incubated with primary antibody for 1 hour at room temperature. Primary antibodies used were directed against: Cytokeratin AE1/AE3 (1:400, Dako, Heverlee, Belgium), CD44 (clone IM7 recognizing all isoforms of CD44, 1:100, Biolegend, London, UK), SOX-2 (clone L1D6A2, 1:400; Cell Signaling, Leiden, the Netherlands) and SHH (ab53281, 1:750, Abcam, Cambridge, UK). Subsequently, tissue sections were 8 incubated for 30 minutes with HRP-conjugated secondary and tertiary antibodies (1:50

151

Proefschrift_Kirill.indb 151 10-05-15 23:47 dilution, all from Dako). Staining was visualized using 3,3'-diaminobenzidine and haematoxylin counterstaining. Positive and negative controls (immunoglobulin class- matched control sera) were included for each staining.

Analysis of immunohistochemistry Slides were digitized using the Digital Slide Scanner NanoZoomer and NDP software (Hamamatsu, Shizuoka, Japan). HE and cytokeratin staining was used to identify vital tumour areas. Scoring was performed by two independent observers (JH, KP) blinded for patient outcome and random samples were validated by a blinded expert gastrointestinal pathologist (AK). Protein expression of CD44 and SHH was determined by scoring localization, intensity and percentage of positive cells. CD44,and SHH expression was initially scored according to the Immunoreactivity Score (IRS); the percentage of positive cells was scored into six categories: 0) no staining, 1) 1-5%, 2) 5-25%, 3) 25-50%, 4) 50-75% and 5) 75-100% positive cells. Intensity was scored as 0) negative, 1) weak, 2) medium and 3) strong. The IRS was calculated by multiplying the percentage of positive cells times the intensity score, resulting in a scale from 0 to 15. For evaluation of the CD44 staining the IRS was divided into two groups: negative or low staining (IRS 0-5), and high staining (IRS 6-15). Since almost all specimens showed SHH expression in 75-100% of the tumour cells, only intensity for SHH was determined as the consensus of 2 out of 3 observers (JH, KP, AK) and dichotomized into low (intensity 0-1) and high (intensity 2-3). SOX2 staining was determined by either nuclear or cytoplasmic staining and groups were divided in either no staining or positive staining as described previously (12).

Statistics Cause-specific survival (CSS) was defined as the time between end of treatment i.e. surgery and the documented day of death or last follow-up alive, with death due to recurrence of the primary tumour. Disease-free survival (DFS) was defined as the time between date of surgery and date of recurrence, death, or last follow up alive without a recurrence. Differences in patient characteristics per marker expression group and co-expression of markers were calculated using the Fisher’s exact test. Differences in expression of markers between samples of one patient (pre-treatment biopsy, resection specimen, lymph node, recurrence) were calculated using a McNemar test. Survival curves were calculated with the Kaplan-Meier method using the log-rank test. Univariate Cox-regression analyses were performed to identify prognostic factors for CSS and DFS. Factors with multiple categories were made dichotomous by creating two subgroups. Clinical and pathological tumour stage (cT-stage and pT-stage) were subdivided into 2 groups: low (cT1-2/pT0-T2) and high (cT3-4/pT3). Clinical and pathological nodal stage (cN-stage and pN-stage) were divided into negative (cN0/pN0) and positive nodal involvement (cN≥1/pN≥1).

152

Proefschrift_Kirill.indb 152 10-05-15 23:47 Factors with a P-value of p<0.200 for either the CSS or DFS in the univariate analysis were included in a stepwise (backwards conditional method) multivariate Cox-regression. A P-value of p <0.05 was considered significant. Two groups were analyzed in the univariate and multivariate analysis: 1) patients with pre-treatment biopsy material (n=71), 2) patients with residual tumour after nCRT in the resection material (n=53). Statistical analysis was performed by using the International Business Machines Statistical Package for Social Sciences (IBM SPSS, Armonk, New York, USA) version 22.0.

Results

Patient and tumour characteristics In Table 1 patient and tumour characteristics are shown for patients with pre-treatment biopsies available for IHC (n=71) and the post-treatment nCR group (n=53). The median age of both study groups combined was 63 years and the majority of tumours were EAC (83%). In total 20 patients (20%) had a pCR.

Table 1: Patient and tumor characteristics.

Pre-treatment biopsies (n=71) Post- treatment nCR group (n=53) Sex Male 57 (80%) 45 (85%) Female 14 (20%) 8 (15%) Age ≥65 35 (50%) 27 (51%) Tumor localisation Mid 2 (3%) 1 (2%) Distal 58 (82%) 44 (83%) GEJ 11 (15%) 8 (15%) Differentiation grade High 20 (28%) 12 (23%) Histology EAC 61 (86%) 48 (91%) ESCC 10 (14%) 5 (9%) cT- stage* cT1-2 11 (15%) 8 (15%) cT3-4 59 (83%) 44 (83%) cN-stage cN≥1 53 (75%) 39 (74%) pT-stage pT0-2 41 (58%) 25 (47%) 8 pT3 30 (42%) 28 (53%)

153

Proefschrift_Kirill.indb 153 10-05-15 23:47 Table 1: Patient and tumor characteristics (continued).

Pre-treatment biopsies (n=71) Post- treatment nCR group (n=53) pN-stage ypN≥1 29 (41%) 26 (49%) Mandard 1 16 (23%) 2 23 (32%) 21 (40%) 3 22 (31%) 22 (41%) 4 8 (11%) 8 (15%) 5 2 (3%) 2 (4%) pCR 14 (20%) Positive CRM (≤1mm) 16 (23%) 15 (28%)

Angioinvasion 16 (23%) 16 (30%) Perineural 11 (15%) 11 (21%) Completed CRT 60(85%) 45 (85%)

Recurrence 38 (54%) 32 (60%) Systemic recurrence 29 (41%) 23 (43%) Local recurrence 15 (21%) 13 (25%)

* information missing of 1 patient Tumor regression rates classified according to the Mandard scoring system. EAC= esophageal adenocarcinoma. ESCC= esophageal squamous cell carcinoma. pCR= pathological complete response. CRM= circumferential resection margin, defined by a cut-off ≤1mm. CRT= chemoradiotherapy.

Clinicopathological characteristics in relation to marker expression pattern Figure 2 shows representative pictures of the immunohistochemical stainings. Supplementary Table 1 shows patient and tumour characteristics in relation to the expression of CD44, SHH and SOX2. Absent SOX2 expression in the pre-treatment biopsies group was related to systemic recurrence (P=0.029).

Predictive value of marker expression for disease-free and cause-specific survival In the group of patients in whom pre-treatment biopsies were available positive pN-stage (HR 2.40 95% Confidence interval (CI) 1.21-4.74 P=0.012) and positive CRM (HR 2.17 CI 1.02-4.61 P=0,044) were independent prognostic factors for a poor DFS in both univariate and multivariate analysis (Table 2a). Of note, low SHH expression was not significantly associated with a poor DFS or CSS in a univariate analysis (P=0,117 for DFS and P=0,054 for CSS, Table 2a and Figure 3), but low SHH expression was an independent prognostic

154

Proefschrift_Kirill.indb 154 10-05-15 23:47 Supplementary figure 1 CD44

High Low SOX2

Present Absent SHH

High Low

Figure 2: Representative images of Immunohistochemical stainings

factor for a poor DFS in the multivariate analysis (HR 2.27 CI 1.05-4.89 P=0.036, Table 2a). In a univariate analysis, absent SOX2 expression in pre-treatment biopsies showed a trend for a worse DFS (HR 1.99 1.04-4.89 P=0.039, Table 2a and Figure 3) and CSS (HR 2.16 1.08- 4.34 P= 0.030, Table 2a and Figure 4) but did not retain significance in a multivariate analysis. Analyzing only nuclear (instead of both cytoplasmic and nuclear) SOX2 expression did not alter the findings of our study (data not shown). CD44 expression was not correlated with DFS or CSS in the pre-treatment biopsy group. In the post-treatment nCR group positive pN-stage (HR 3.75 CI 1.67-8.42 P=0,001 for DFS and HR 2.96 CI 1.27-6.89 P=0,012 for CSS) and low CD44 expression (HR3.25 CI 1.46-7.24 P=0,004 for DFS and HR 4.08 CI 1.69-9.86 P=0,002 for CSS) were independent prognostic factors for poor survival (Table 2b, Figures 3 and 4) and remained independent prognostic factors for DFS and CSS when a subgroup analysis was performed on only EAC patients (data not shown). In contrast to the pre-treatment biopsy group, SHH expression 8 was not correlated with DFS or CSS in the post-treatment nCR group (Table 2b, Figure 3

155

Proefschrift_Kirill.indb 155 10-05-15 23:47 backward multivariate correction. a after (P<0.05) CSS or DFS for values representsignificant analysis multivariate the in bold in P-values analysis. multivariate the in included were which representP-values<0.200, analysis univariate the in bold in P-values groupresponder(n=53). post-treatment(nCR) non-complete pre-treatmentb) a) (n=71) for biopsies prognosticfactors Clinicopathological SOX2 (absent vs present) SHH (low vs high) CD44 (low vs high) CRM (pos) ypN ( ≥pN1) cT (cT3-4/ cT1-T2) Differentiation (high vs low) N=53 b) Post-treatment resection nCR group SOX2 (absent vs present) SHH (low vs high) CD44 (low vs high) pCR Mandard CRM (pos) ypN ( ≥pN1) ypT (pT3/T0-T2) cT (cT3-4/ cT1-T2) Histology ( ESCC vs EAC) Sex (female>male) N=71 a) Pre-treatment biopsies Table 2: Univariate and multivariate analysis for Disease Free Survival (DFS) Cause Specific (CSS). 1.42 1.85 2.66 2.15 3.05 3.27 1.75 1.99 1.79 1.21 1.75 1.39 2.31 2.54 1.76 2.59 0.42 0.38 HR HR Univariate Univariate 0.70-2.89 0.82-4.21 1.25-5.63 1.04-4.47 1.44-6.43 0.78-13.7 0.80-3.81 1.04-3.84 0.86-3.70 0.62-2.35 0.72-2.21 1.04-1.89 1.15-4.66 1.32-4.88 0.93-3.34 0.79-8.44 0.13-1.36 0.13-1.06 95% CI 95% CI P-value P-value 0.326 0.141 0.040 0.003 0.106 0.159 0.039 0.576 0.215 0.029 0.019 0.005 0.084 0.146 0.065 0.011 0.117 0.115 DFS DFS 3.25 3.75 2.27 2.17 2.40 HR HR Multivariate Multivariate 1.46-7.24 1.67-8.42 1.05-4.89 1.02-4.61 1.21-4.74 95% CI 95% CI P-value P-value 0.004 0.001 0.036 0.044 0.012 1.39 1.47 3.85 2.68 1.99 2.16 2.10 1.40 1.89 1.37 1.97 2.53 1.81 2.90 0.48 0.41 HR HR Univariate Univariate 0.69-12.20 0.64-3.02 0.62-3.53 1.64-9.07 1.21-5.94 0.86-4.63 1.08-4.34 0.99-4.46 0.69-2.85 0.72-4.95 1.00-1.90 0.94-4.12 1.26-5.09 0.90-3.64 0.15-1.59 0.14-1.17 95% CI 95% CI P-value P-value 0.407 0.385 0.002 0.015 0.109 0.030 0.054 0.356 0.196 0.054 0.073 0.009 0.096 0.146 0.233 0.096 CSS CSS 4.08 2.96 2.04 2.64 HR HR Multivariate Multivariate 1.69-9.86 1.27-6.89 0.96-4.36 1.30-5.34 95% CI 95% CI P-value P-value 0.002 0.012 0.064 0.007

156

Proefschrift_Kirill.indb 156 10-05-15 23:47 Figure 2

Pre-treatment biopsy group Post-treatment nCR group P=0.576 100 100 P=0.008

80 80

CD44 high (N=47) CD44 high (N=32) 60 60

40 40

CD44 low (N=24) CD44 Survival (%) Survival Survival (%) Survival CD44 low (N=20) 20 20

0 0

0 12 24 36 48 60 0 12 24 36 48 60 DFS (months) DFS (months)

100 P=0.035 100 P=0.324

80 80

60 SOX2 present (N=52) 60 SOX2 present (N=28)

40 40 SOX2 Survival (%) Survival Survival (%) Survival SOX2 absent (N=19) 20 20 SOX2 absent (N=25)

0 0

0 12 24 36 48 60 0 12 24 36 48 60 DFS (months) DFS (months)

100 P=0.112 100 P=0.135

80 80

60 SHH high (N=55) 60 SHH high (N=43)

40 40 SHH Survival (%) Survival (%) Survival SHH low (N=15) 20 20 SHH low (N=10)

0 0

0 12 24 36 48 60 0 12 24 36 48 60 DFS (months) DFS (months)

Figure 3: Disease Free Survival (DFS) curves based on expression of CD44, SOX2 and SHH in the pre-treatment biopsy group and post-treatment non-complete responder (nCR) group. 8

157

Proefschrift_Kirill.indb 157 10-05-15 23:47 Figure 3

Pre-treatment biopsy group Post-treatment nCR group P=0.354 100 100 P=0.001

80 80 CD44 high (N=32) CD44 high (N=47)

60 60

40 40 CD44 CD44 low (N=24) Survival (%) Survival (%) Survival CD44 low (N=20) 20 20

0 0

0122436 48 60 0 12 24 36 48 60 CSS (months) CSS (months) P=0.027 100 100 P=0.404

80 80 SOX2 present (N=52)

60 60 SOX2 present (N=28)

40 40

SOX2 absent (N=19) SOX2 Survival (%) Survival Survival (%) Survival SOX2 absent (N=25)

20 20

0 0

0 12 24 36 48 60 0 12 24 36 48 60 CSS (months) CSS (months) P=0.049 100 100 P=0.382

80 80

SHH high (N=55) SHH high (N=43) 60 60

40 40 SHH

Survival (%) Survival SHH low (N=15) (%) Survival SHH low (N=10) 20 20

0 0

0 12 24 36 48 60 0 12 24 36 48 60 CSS (months) CSS (months)

Figure 4: Cause Specific Survival (CSS) curves based on expression of CD44, SOX2 and SHH in the pre-treatment biopsy group and post-treatment non-complete responder (nCR) group.

158

Proefschrift_Kirill.indb 158 10-05-15 23:47 and 4). SOX2 expression in the post-treatment nCR group was also not correlated with DFS or CSS in either uni-or multivariate analysis (Table 2b, Figures 3 and 4)

Predictive value of marker expression for response to neoadjuvant CRT No predictive value of the tested markers was found for either pCR or response to nCRT therapy as determined by the Mandard score (Table 3).

Coexpression and marker expression in the patient-matched samples A trend for a correlation between low CD44 expression and absence of SOX2 expression was observed in both the pre-treatment biopsies group and the post-treatment nCR group (Supplementary Table 2, P=0.052 and P=0.087 respectively). To study the changes in expression of CD44, SHH and SOX2 during treatment, we used a subset of patients within the post-treatment nCR group (n=53) for whom samples of lymph nodes (n=23) and material of recurrent disease (n=12) were available. The proportion of patients with absent SOX2 expression increased in the sequence of biopsy, resection specimen, lymph node and recurrence (25%, 47%, 57%, and 70% respectively). This increase was significant between biopsies and resection specimens (McNemar test, P=0.029). No difference in the expression of CD44 or SHH was found between the patient matched samples.

Discussion

To our best knowledge this is the first study to determine the expression of CD44, SOX2 and SHH in both pre- and post-treatment samples of EC patients treated with nCRT. Low expression of CD44, SHH and SOX2 was associated with either recurrence or a poor survival in EC patients. The results of our study suggest that CD44 and SHH may be of use to stratify EC patients treated with nCRT and the use of these markers may help identifying patients eligible for adjuvant therapy. CD44 and SOX2 have been associated with CSCs and resistance to therapy in other solid tumours (2,3). Nevertheless, we did not find a relation with response to nCRT in this study. In line with our previous study, in which a cohort of surgically treated EAC patients was examined (12), low CD44 expression was also an independent prognostic factor for poor survival in the post-treatment nCR group, while in the pre-treatment biopsy group low CD44 expression showed a trend for poor survival that did not reach statistical significance. Low CD44 expression has previously been related to a worse survival in EAC patients and a decrease of CD44 expression was observed in EAC development (11,19). However, another study found that a high percentage of CD44-positive tumour cells was 8 associated with poor survival in EAC patients treated with nCRT and surgery, but this

159

Proefschrift_Kirill.indb 159 10-05-15 23:47 nCR= non-Complete Response pCR= pathological Complete Response. * SHH 1 biopsy staining failed determined using a Fisher exact test (P<0.05 was considered significant). was A Mandard system score of 2 and 3 was defined as a partial response and a score Mandard of 4-5 as absence response. by the or a pCR by either defined therapy neoadjuvant to response the and expression marker to according stratified groups different between relation The Pos (n=52) Neg (n=19) SOX2 High (n=55) Low (n=15) SHH* High (n=47) Low (n=24) CD44 Table 3: Response to neoadjuvant chemoradiotherapy in pre-treatment biopsies (n=71). nCR (n=57) 43 14 44 12 38 19 pCR (n=14) 11 9 5 3 9 5 P-value 0.502 1.000 1.000 (Complete responder n=16) Mandard 1 13 10 11 5 3 6 (Partial responder n=45) Mandard 2-3 32 13 33 30 15 11 (Non-responder n=10) Mandard 4-5 9 1 9 1 7 3 P-value 0.501 0.704 0.936

160

Proefschrift_Kirill.indb 160 10-05-15 23:47 study used only resection material without pre-treatment biopsies and had a smaller cohort of patients (n=39) (20). Our finding that loss of CD44 is an adverse prognostic factor in EC is unexpected given the previously reported association between CD44 with CSCs and radiotherapy resistance. A possible explanation could be that the CSC compartment might be heterogeneous with CD44-positive CSCs being primarily important in EAC tumour initiation, but less so in established EAC. CD44 is a downstream target of the Wnt signalling pathway, which is known to regulate CSCs in colon cancer (21). Activation of the Wnt pathway occurs during the malignant transformation of BE towards dysplasia, with no further increase of Wnt activity in EAC (22). The subsequent emergence of more aggressive, CD44-negative clones in established EC could possibly explain the negative correlation between CD44 expression and survival observed in the current study. Alternatively, CD44 may be a marker for CSC in vitro, but not in vivo. This possibility is supported by a recent study in non-small cell lung cancer that found that CD44 expression was associated with stem cell characteristics in vitro, while loss of CD44 expression in patient biopsies was associated with a poor survival (23). Our finding that absence of SOX2 was correlated with systemic recurrence combined with the progressive decrease of SOX2 expression in patient-matched samples of different stages of disease progression and the association between absent SOX2 staining and poor survival outcome (though not significant in multivariate analysis) all suggest that SOX2 has a tumour suppressor role in EC. This finding fits well with our previous observation that absence of SOX2 in EAC patients treated with surgery alone is related to poor survival in EAC patients (12). A tumour-suppressor role of SOX2 is further supported by in vitro studies in gastric and colon cancer cells showing that overexpression of SOX2 has a tumour suppressor effect by reducing cell proliferation (24,25) and by a recent study that reported loss of SOX2 expression during progression of BE towards EAC and found this to be an independent predictive factor for malignant progression (26). Taken together, low SOX2 expression might contribute to the development of EAC and a poor prognosis of EAC patients. HH signalling increases significantly in both ESCC and EAC development (14,15). High SHH expression has been described in post-treatment resection specimens of EAC treated with nCRT (16). However, thus far the prognostic and predictive value of SHH expression in EC has not been examined. Our finding that low SHH expression was a prognostic factor for a decreased DFS is in contrast with previous in vitro studies reporting that HH antagonists reduce cell growth and induced apoptosis in EAC and ESCC cell lines (14). The mechanisms underlying the discrepancy between our in vivo and previous in vitro findings are currently unknown. Differences between in vitro and in vivo tissue microenvironment could possibly explain these contradictory findings, but further 8 research is required to establish the role of SHH in EC biology.

161

Proefschrift_Kirill.indb 161 10-05-15 23:47 In conclusion, this is the first study to report that low CD44 and SHH expression are independent prognostic factors for poor survival in EC patients treated with nCRT, and that loss of SOX2 is related to disease recurrence. While the elucidation of the function of CD44, SOX2 and SHH in EC requires further study, our results suggest that these markers have a potential to identify EC patients with a poor prognosis.

162

Proefschrift_Kirill.indb 162 10-05-15 23:47 Acknowledgements

The authors would like to thank the van der Meer-Boerema foundation for its financial support. Special gratitude is extended to the technicians of the department of Pathology at the University Medical Center Groningen for their help generating all tumor slides. We thank the pathology departments of the hospitals in the Northeast Netherlands: Martini Hospital Groningen, Pathology Friesland, Pathology laboratory East Netherlands (Pathologie laboratorium Oost-Nederland), Pathology laboratory Collaborating Hospitals Oost- Groningen (Stichting Samenwerking Ziekenhuizen Oost-Groningen) and the Bethesda Hospital Hoogeveen for providing the patient material.

163

Proefschrift_Kirill.indb 163 10-05-15 23:47 Supplementary table 1: Patient and tumor characteristics in relation to marker expression

CD44 Pre-treatment biopsies Post-treatment nCR group (n=71) (n=52*) Low High P-value Low High P-value (n=24) (n=47) (n=20) (n=32)

Sex Male 17 40 16 28 Female 7 7 0.209 4 4 0.695 Age <65 11 25 8 18 ≥65 13 22 0.621 12 14 0.393 Tumor localisation Mid 1 1 0 1 Distal 19 39 15 28 GEJ 4 7 1.000 5 3 0.235 Differentiation grade** Low 18 33 14 26 High 6 14 0.784 6 6 0.500 Histology EAC 24 37 20 27 ESCC 0 10 0.013 0 5 0.143 cT- stage** cT1-2 2 9 1 7 cT3-4 22 37 0.309 18 25 0.231 cN-stage N0 7 11 3 11 N1 17 36 0.774 17 21 0.200 pT-stage pT0-2 15 26 8 16 pT3 9 21 0.619 12 16 0.573 pN-stage pN0 14 28 8 18 pN≥1 10 19 1.000 12 14 0.393 Resection margins pCR 5 9 1.000 Positive CRM (≤1mm) 6 10 0.769 5 10 0.757 Angioinvasion 5 11 1.000 8 8 0.356 Perineural invasion 5 6 0.490 4 7 1.000 Completed CRT 21 39 0.739 17 27 1.000 Recurrence 14 24 0.621 14 17 0.260

164

Proefschrift_Kirill.indb 164 10-05-15 23:47 Supplementary table 1 (continued)

CD44 Pre-treatment biopsies Post-treatment nCR group (n=71) (n=52*) Low High P-value Low High P-value (n=24) (n=47) (n=20) (n=32) Systemic recurrence 12 17 0.321 12 11 0.090 Local recurrence 5 10 1.000 4 9 0.743

*The CD44 staining failed in 1 resection specimens; these were excluded from analysis ** 1 patient missing information GEJ: gastro-esophageal junction EAC: esophageal adenocarcinoma ESCC: esophageal squamous cell carcinoma pCR: complete pathological resection CRM: circumferential resection margin

SOX2 Pre-treatment biopsies Post-treatment nCR group (n=71) (n=53) Neg (n=19) Pos (n=52) P-value Neg (n=25) Pos (n=28) P-value Sex Male 18 39 20 25 Female 1 13 0.093 5 3 0.453 Age <65 11 25 13 13 ≥65 8 27 0.594 12 15 0.786 Tumor localisation Mid 0 2 0 1 Distal 14 44 21 23 GEJ 5 6 0.260 4 4 1.000 Differentiation grade* Low 13 38 20 21 High 6 14 0.769 5 7 0.750 Histology EAC 18 43 23 25 ESCC 1 9 0.270 2 3 1.000 cT- stage* cT1-2 1 10 4 4 cT3-4 18 41 0.267 20 24 1.000 cN-stage N0 4 14 5 9 8 N1 15 38 0.762 20 19 0.365

165

Proefschrift_Kirill.indb 165 10-05-15 23:47 Supplementary table 1 (continued)

SOX2 Pre-treatment biopsies Post-treatment nCR group (n=71) (n=53) Neg (n=19) Pos (n=52) P-value Neg (n=25) Pos (n=28) P-value pT-stage pT0-2 10 31 11 14 pT3 9 21 0.601 14 14 0.785 pN-stage pN0 9 33 12 15 pN≥1 10 19 0.279 13 13 0.786 Resection margins pCR 5 9 0.502 Positive CRM (≤1mm) 5 11 0.750 5 10 0.237 Angioinvasion 4 12 1.000 8 8 1.000 Perineural invasion 1 10 0.267 6 5 0.737 Completed CRT 17 43 0.715 20 25 0.453 Recurrence 15 23 0.015 16 16 0.779 Systemic recurrence 12 17 0.029 12 11 0.586 Local recurrence 6 9 0.324 7 6 0.751

* 1 patient missing value GEJ: gastro-esophageal junction EAC: esophageal adenocarcinoma ESCC: esophageal squamous cell carcinoma pCR: complete pathological resection CRM: circumferential resection margin

SHH Pre-treatment biopsies Post-treatment nCR group (n=70) (n=53) Low High P-value Low High P-value (n=15) (n=55) (n=10) (n=43) Sex Male 12 44 8 37 Female 3 11 1.000 2 6 0.636 Age <65 10 26 3 23 ≥65 5 29 0.247 7 20 0.162 Tumor localisation Mid 2 0 0 1 Distal 10 47 8 36 GEJ 3 8 0.028 2 6 0.709

Proefschrift_Kirill.indb 166 10-05-15 23:47 Supplementary table 1 (continued)

SHH Pre-treatment biopsies Post-treatment nCR group (n=70) (n=53) Low High P-value Low High P-value (n=15) (n=55) (n=10) (n=43) Differentiation grade Low 10 40 8 33 High 5 15 0.435 2 10 1.000 Histology EAC 11 49 10 38 ESCC 4 6 0.205 0 5 0.570 cT- stage* cT1-2 2 8 1 7 cT3-4 13 46 1.000 9 35 1.000 cN-stage N0 3 15 3 11 N1 12 40 0.744 7 32 1.000 pT-stage pT0-2 7 33 3 22 pT3 8 22 0.391 7 21 0.302 pN-stage pN0 8 34 3 24 pN≥1 7 21 0.567 7 19 0.175 Resection margins pCR 3 11 1.000 Positive CRM (≤1mm) 1 15 0.163 5 10 0.124 Angioinvasion 2 14 0.492 5 11 0.148 Perineural invasion 0 11 0.105 5 6 0.023 Completed CRT 12 47 0.691 7 38 0.163 Recurrence 10 28 0.383 8 24 0.282 Systemic recurrence 9 20 0.140 7 16 0.082 Local recurrence 3 12 0.551 3 10 0.692

* 1 patient missing information The SHH staining failed in 1 resection specimens; this specimen was excluded from analysis GEJ: gastro-esophageal junction EAC: esophageal adenocarcinoma ESCC: esophageal squamous cell carcinoma pCR: complete pathological resection CRM: circumferential resection margin 8

167

Proefschrift_Kirill.indb 167 10-05-15 23:47 Supplementary table 2: Marker co-expression

a) Coexpression in pre-treatment biopsies CD44 P-value SHH P-value Low (n=24) High (n=47) Low (n=15) High (n=55) SHH 0.760 Low (n=15) 6 9 High (n=55) 18 37 SOX2 0.052 0.325 Neg (n=19) 10 9 6 13 Pos (n=52) 14 38 9 42

b) Coexpression in post-treatment nCR group CD44 P-value SHH P-value Low (n=20) High (n=32) Low (n=10) High (n=43) SHH 1.000 Low (n=10) 4 6 High (n=42) 16 26 SOX2 0.087 0.162 Neg (n=25) 13 12 7 18 Pos (n=28) 7 20 3 25

168

Proefschrift_Kirill.indb 168 10-05-15 23:47 References

(1) van Hagen P, Hulshof MC, van Lanschot JJ, et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med. 2012;366(22):2074-2084. (2) Dalerba P, Dylla SJ, Park IK, et al. Phenotypic characterization of human colorectal cancer stem cells. Proc Natl Acad Sci U S A. 2007;104(24):10158-10163. (3) Suva ML, Rheinbay E, Gillespie SM, et al. Reconstructing and reprogramming the tumor- propagating potential of glioblastoma stem-like cells. Cell. 2014;157(3):580-594. (4) Maugeri-Sacca M, Vigneri P, De Maria R. Cancer stem cells and chemosensitivity. Clin Cancer Res. 2011;17(15):4942-4947. (5) Sahlberg SH, Spiegelberg D, Glimelius B, Stenerlow B, Nestor M. Evaluation of cancer stem cell markers CD133, CD44, CD24: Association with AKT isoforms and radiation resistance in colon cancer cells. PLoS One. 2014;9(4):e94621. (6) Smit JK, Faber H, Niemantsverdriet M, et al. Prediction of response to radiotherapy in the treatment of esophageal cancer using stem cell markers. Radiother Oncol. 2013;107(3):434-441. (7) La Fleur L, Johansson AC, Roberg K. A CD44high/EGFRlow subpopulation within head and neck cancer cell lines shows an epithelial-mesenchymal transition phenotype and resistance to treatment. PLoS One. 2012;7(9):e44071. (8) Lemke D, Weiler M, Blaes J, et al. Primary glioblastoma cultures: Can profiling of stem cell markers predict radiotherapy sensitivity? J Neurochem. 2014. (9) Attramadal CG, Halstensen TS, Dhakal HP, et al. High nuclear SOX2 expression is associated with radiotherapy response in small (T1/T2) oral squamous cell carcinoma. J Oral Pathol Med. 2014. (10) Ghisolfi L, Keates AC, Hu X, Lee DK, Li CJ. Ionizing radiation induces stemness in cancer cells. PLoS One. 2012;7(8):e43628. (11) Bottger TC, Youssef V, Dutkowski P, Maschek H, Brenner W, Junginger T. Expression of CD44 variant proteins in adenocarcinoma of barrett’s esophagus and its relation to prognosis. Cancer. 1998;83(6):1074-1080. (12) Honing J, Pavlov KV, Meijer C, et al. Loss of CD44 and SOX2 expression is correlated with a poor prognosis in esophageal adenocarcinoma patients. Ann Surg Oncol. 2014;21 Suppl 4:657-664. (13) Merchant AA, Matsui W. Targeting hedgehog--a cancer stem cell pathway. Clin Cancer Res. 2010;16(12):3130-3140. (14) Ma X, Sheng T, Zhang Y, et al. Hedgehog signaling is activated in subsets of esophageal cancers. Int J Cancer. 2006;118(1):139-148. (15) Wang DH, Clemons NJ, Miyashita T, et al. Aberrant epithelial-mesenchymal hedgehog signaling characterizes barrett’s metaplasia. Gastroenterology. 2010;138(5):1810-1822. (16) Sims-Mourtada J, Izzo JG, Apisarnthanarax S, et al. Hedgehog: An attribute to tumor regrowth after chemoradiotherapy and a target to improve radiation response. Clin Cancer Res. 2006;12(21):6565-6572. (17) Mandard AM, Dalibard F, Mandard JC, et al. Pathologic assessment of tumor regression after preoperative chemoradiotherapy of esophageal carcinoma. clinicopathologic correlations. Cancer. 1994;73(11):2680-2686. 8 (18) Dikken JL, Lemmens VE, Wouters MW, et al. Increased incidence and survival for oesophageal cancer but not for gastric cardia cancer in the netherlands. Eur J Cancer. 2012;48(11):1624-1632.

169

Proefschrift_Kirill.indb 169 10-05-15 23:47 (19) Darlavoix T, Seelentag W, Yan P, Bachmann A, Bosman FT. Altered expression of CD44 and DKK1 in the progression of barrett’s esophagus to esophageal adenocarcinoma. Virchows Arch. 2009;454(6):629-637. (20) Turner JR, Torres CM, Wang HH, et al. Preoperative chemoradiotherapy alters the expression and prognostic significance of adhesion molecules in barrett’s-associated adenocarcinoma.Hum Pathol. 2000;31(3):347-353. (21) de Sousa EM, Vermeulen L, Richel D, Medema JP. Targeting wnt signaling in colon cancer stem cells. Clin Cancer Res. 2011;17(4):647-653. (22) Moyes LH, McEwan H, Radulescu S, et al. Activation of wnt signalling promotes development of dysplasia in barrett’s oesophagus. J Pathol. 2012;228(1):99-112. (23) Leung EL, Fiscus RR, Tung JW, et al. Non-small cell lung cancer cells expressing CD44 are enriched for stem cell-like properties. PLoS One. 2010;5(11):e14062. (24) Otsubo T, Akiyama Y, Yanagihara K, Yuasa Y. SOX2 is frequently downregulated in gastric cancers and inhibits cell growth through cell-cycle arrest and apoptosis. Br J Cancer. 2008;98(4):824-831. (25) Cho YY, Kim DJ, Lee HS, et al. Autophagy and cellular senescence mediated by Sox2 suppress malignancy of cancer cells. PLoS One. 2013;8(2):e57172. (26) van Olphen SH, Kastelein F, Kastelein F, et al. SOX2 as a novel marker to predict neoplastic progression in Barrett’s esophagus. 2014.

170

Proefschrift_Kirill.indb 170 10-05-15 23:47 8

171

Proefschrift_Kirill.indb 171 10-05-15 23:47 Proefschrift_Kirill.indb 172 10-05-15 23:47 Summary, Discussion and Future perspectives

Proefschrift_Kirill.indb 173 10-05-15 23:47 Proefschrift_Kirill.indb 174 10-05-15 23:47 Summary and discussion

Esophageal adenocarcinoma (EAC) is thought to develop from its precursor lesion, Barrett’s esophagus (BE) through a metaplasia-dysplasia-carcinoma sequence. While persistent gastro-esophageal reflux disease (GERD) is the main risk factor for the development of BE, only a minority of GERD patients will develop BE during their lifetime. Patients with BE are enrolled in endoscopic surveillance programs aimed at early detection of BE malignant transformation, and early detection has been shown to significantly improve survival. However, BE is asymptomatic, and in the absence of BE biomarkers the majority of BE patients will remain undiagnosed and present with EAC symptoms, often at a disease stage that precludes curative treatment. Even in EAC patients undergoing treatment with curative intent consisting of neoadjuvant chemoradiotherapy (nCRT) followed by an esophagectomy the 5-year survival is only around 50%. The current management of BE and EAC thus has several challenges, some of which are addressed in this thesis. First, the underlying molecular mechanisms of BE development are unclear, and this limits the development of pharmacological treatment modalities. Second, the absence of a BE biomarker hampers the detection of BE patients. Third, a better understanding of EAC biology is required to further improve patient survival. Understanding the signaling pathways that drive BE development could yield novel therapeutic targets. In chapter 1 we summarized the role of four main signaling pathways implicated in BE development and malignant progression: the Bone Morphogenetic Protein (BMP), Hedgehog (HH), Wingless-Type MMTV Integration Site Family (WNT) and Retinoic Acid (RA) signaling pathways. Activation of BMP, HH and RA signaling contributes to BE development, while high WNT and HH drive BE malignant transformation. Interestingly, these four signaling pathways also play a central role in the embryological development of the esophagus. Thus, studying the involvement of embryonic signaling pathways in the context of BE development and EAC carcinogenesis could serve as a potent approach to better understand the biology of BE and EAC. The availability of representative experimental models are critical to the study of pathways driving BE and EAC. In chapter 2 we discuss several of the current BE and EAC models and describe their advantages and limitations. Traditional in vitro cell culture models are accessible and easy to use, but the number of representative BE and EAC cell lines is limited. Cell culture in the absence of a representative microenvironment could influence cell behavior and precludes the possibility to study epithelial-stromal interactions in EAC development. Surgical esophagojejunostomy animal models using mice and rats have been developed that incorporate a stromal component, but it is not clear whether they develop the same molecular alterations that are commonly found in human EAC cases. In addition, these models require the animal to be sacrificed in order

175

Proefschrift_Kirill.indb 175 10-05-15 23:47 to take esophageal tissue samples, and do not allow biopsies at multiple time points. Two important novel developments have been the development of in vitro organotypic models of co-cultured stromal and epithelial cells that enable the study of stromal-epithelial interactions in vitro, and the use of computational models that simulate the dynamics of BE expansion and malignant transformation, that offer tractability and can be easily manipulated to reflect novel insights in BE and EAC. In chapter 3 we studied the ability of RA to induce a columnar morphology in an organotypic model. We found that the addition of RA to the organotypic model caused loss of the multilayered squamous architecture and was associated with loss of the squamous cell-specific cytokeratin 13. When the organotypic model was used to culture a metaplastic BE cell line, goblet cell differentiation, a diagnostic feature of BE was observed that has not been seen in traditional cell culture methods. Dysplastic BE cell lines cultured in the organotypic model displayed invasion into the stromal compartment, suggesting that the model could also be used to study tissue invasion in EAC. RA has previously been suggested to contribute to the development of the metaplastic columnar esophageal epithelium (1,2), but so far the transcription factors involved in this process were not identified. The aim of chapter 4 was to study the expression of squamous and columnar transcription factors during BE development using patient biopsies and examine the potential effect of RA on these transcription factors using an in vitro model of immortalized esophageal keratinocytes. We found that the squamous transcription factors p63 and SOX2 were expressed in the majority of cells in biopsies of squamous epithelium. In BE, p63 expression was lost completely while SOX2 expression persisted in a minority of BE biopsies. Both the columnar transcription factors GATA6 and CDX2 were expressed in a small minority of cells in squamous biopsies, but the percentage of positive cells increased significantly in biopsy samples of BE. While the percentage of GATA6-positive cells did not differ between BE biopsies containing goblet cells (intestinal type BE, IM) or without goblet cells (non-intestinal type BE, non-IM) we found that the percentage of CDX2-positve cells was significantly higher in IM biopsies compared to non- IM biopsies. In vitro, RA treatment significantly decreased protein expression of ΔNp63α, an isoform of the squamous transcription factor p63. However, RA treatment had no effect on the expression of SOX2. RA treatment also increased mRNA expression of the columnar transcription factors GATA6 and SOX9 but did not induce CDX2 mRNA expression. This can be explained by assuming that induction of CDX2 expression is a later event during BE development and associated with intestinal-type differentiation, but is not required for the differentiation of non-intestinalized columnar epithelium. These findings suggest that squamous epithelium and BE are each characterized by distinct patterns of transcription factor expression, and identify transcription factors whose expression is modulated by RA in squamous epithelium.

176

Proefschrift_Kirill.indb 176 10-05-15 23:47 As described in chapter 4, the zinc finger transcription factor GATA6 is regulated by RA signaling. The aim of chapter 5 was to study GATA6 protein expression during BE development, malignant transformation and evaluate the prognostic significance of GATA6 protein expression in EAC. Using a series of 130 biopsy samples containing normal squamous epithelium, esophagitis, non-IM, IM and high-grade dysplasia we found that GATA6 expression is low in normal squamous epithelium, but progressively increased in metaplastic and dysplastic BE samples. Interestingly, GATA6 expression was already significantly increased in inflamed squamous epithelium due to GERD, indicating that inflammation-driven GATA6 induction could be an early event in BE development. While a previous study reported that GATA6 gene amplification was associated with a poor survival in EAC patients (3), we found no prognostic value of GATA6 protein expression in a cohort of 92 EAC resection specimens. This discrepancy may be related to protein expression levels being regulated by multiple post-transcriptional mechanisms and this could explain the discrepancy between the previously reported prognostic value of GATA6 gene amplification and our findings. In Chapter 6 we set out to investigate the potential of circulating miRNA as biomarkers of BE, BE dysplasia and EAC. Using a Nanostring microarray we identified 6 miRNA’s with a fold change >2 between sera of control patients and patients with either BE, dysplasia or EAC (miR-122-5p, miR-144-3p, miR-150-5p, miR-199a-3p, miR-233-3p and miR-320e). In an extended patient cohort the expression of miR-199a was significantly decreased in patients with BE compared with control patients, while miR-302e expression was significantly reduced in patients with BE and patients with dysplasia compared to control patients. Despite the poor overall prognosis of EAC, there is considerable variation in survival between individual patients. This variation provides an opportunity to correlate biological characteristics with patient survival, thereby increasing the knowledge of EAC biology. In Chapter 7 the association of a panel of presumed Cancer Stem Cell (CSC) markers was studied in relation to patient survival in a cohort of 94 EAC patients that underwent a radical esophagectomy without neoadjuvant therapy. Protein expression of ALDH1, Axin2, BMI1, CD44 and SOX2 was measured semiquantitatively using immunohistochemistry and correlated with overall and disease-free survival. While no association was observed between ALDH1, AXIN2, BMI1 and patient survival, we found that loss of CD44 and SOX2 was associated with a worse prognosis. In a multivariate analysis, loss of both CD44 and SOX2 were significant prognostic factors for a shorter disease-free survival. Several explanations could account for this counter-intuitive finding. First, CD44 and SOX2 expression could not be valid markers of CSC in EAC, or the CSC compartment might be heterogeneous with CD44/SOX2-positive CSCs being primarily important in EAC tumor initiation, but less so in established EAC. The poor prognostic

177

Proefschrift_Kirill.indb 177 10-05-15 23:47 value of CD44 and SOX2 loss in established EAC could then be explained by the other known functions of CD44 and SOX2. CD44 plays a role in adhesion to extracellular matrix and in stimulating inflammation (4,5) while SOX2 has been associated with autophagy and cellular senescence (6). A second explanation could be that CD44 and/or SOX2 could be valid markers for CSC in vitro, but not in vivo. Such an association has been recently described in non-small-cell lung cancer, where CD44+ cells were enriched for stem cell properties and SOX2 expression in vitro, but CD44 expression in patient tissues was associated with a favorable prognosis (7). The lack of functional validation of CD44 and SOX2 as CSC markers in EAC precludes a definitive answer, but the results presented in Chapter 7 highlight the need for functional validation of proposed CSC markers, even if a marker has previously been validated in other tumor types. In Chapter 8 we explored the predictive and prognostic value of a panel of proteins that were previously associated with radiotherapy resistance in EAC patients treated with nCRT. The protein expression of SHH, CD44 and SOX2 was evaluated in two cohorts: 71 pre-treatment biopsies and 74 post-treatment resection specimens. In pre- treatment biopsies low SHH was predictive of poor disease-free survival and loss of SOX2 was associated with recurrence and poor disease-free survival in a univariate analysis, but it did not retain significance after multivariate analysis. While CD44 expression has previously been implicated in resistance to chemo-and radiotherapy (8-10), we found that loss of CD44 was associated with poor survival in the post-treatment resection specimens cohort. This finding is in concordance with our finding in chapter 7 that loss of CD44 is associated with a poor survival outcome in EAC patients treated with surgery only. Thus, the effect of CD44 loss appeared to be independent of treatment, suggesting that levels of CD44 expression reflect differences in cell behavior and not just resistance to treatment.

Future perspectives

Understanding Barrett’s esophagus development Currently, no effective pharmacological treatment exists for BE. Targeting dysregulated signaling pathways could provide novel therapeutic targets for BE and EAC. Increased signaling through the BMP, HH and RA pathways contributes to the induction of columnar differentiation (Chapters 1-3) and antagonizing their activity could induce a regression of BE. Previous studies showed that Citral, a naturally occurring RA antagonist present in citrus fruits, could partially reverse RA-induced columnar epithelium towards a squamous epithelium in an ex vivo study (1) and cyclopamine, a HH antagonist, decreased the incidence of BE in an animal model (11). These preliminary results are encouraging, but several challenges need to be solved before this approach can be tested in humans. First, further progress requires better models of BE development and malignant progression.

178

Proefschrift_Kirill.indb 178 10-05-15 23:47 An ideal experimental model would offer tractability, easy access at multiple time points and a representative microenvironment. Such a model is presently not available, but optimizing and combining existing models could significantly mitigate the limitations of current models. Organotypic models could be further developed to allow a crypt-based in vitro culture, while animal models that offer the possibility to trace cell fate and identity would enable elucidation of a hierarchy of BE cells. While BE is presumed to have a stem cell-based hierarchy analogous to columnar epithelium, the stem cell compartment has not been identified. Leucine-rich repeat-containing G-protein coupled receptor 5 (LGR5) is a marker of stem cells in colonic crypts, and a recently developed BE mouse model has the option to trace LGR5+ cells in the cardia/esophagus region (12). A better understanding of the BE cell hierarchy could in turn be used to improve current computational models of EAC tumor development. Due to the small luminal diameter of the esophagus in small rodents biopsies at multiple time points are a challenging task, but advanced imaging techniques such as microPET/CT could allow for non-invasive monitoring of BE and EAC development and response to therapy and enable in vivo validation of data from in vitro models. Avoiding off-target effects is a second challenge, since each of the described pathways play an important role in intestinal homeostasis. Extensive validation using animal models of BE is therefore required before novel therapeutic therapies based on modulation of the BMP, HH and RA signaling pathways could be tested in humans. A better understanding of BE hierarchy would in turn improve computational in silico models, and these could simulate BE development in time periods that reflect human disease (years) instead of the months that are possible with in vitro or animal models.

Improving detection of Barrett’s esophagus and its malignant transformation Despite a well-established pattern of EAC development through a metaplasia-dysplasia- carcinoma sequence, most patients with EAC are diagnosed at the stage of advanced disease. One of the main reasons for this is that BE itself is asymptomatic and therefore often not diagnosed. Several blood-based biomarkers of BE and EAC have been proposed, including serum proteomic patterns, serum leptin levels and cell-free circulating DNA (13-17). However, these potential biomarkers are limited by the need for complex equipment or variation due to other processes in the body (such as serum leptin levels). In contrast, miRNA’s are highly tissue-specific and easily measured. Chapter 6 identifies two miRNA’s, miR-199a-3p and miR-320e with the potential to discriminate between subjects with and without BE and dysplasia. These results suggest the possibility to use circulating miRNA’s to further stratify members of an at-risk population for endoscopic screening. However, the findings of chapter 6 need to be replicated in other independent, larger cohorts. While using miRNA’s as biomarkers appears an attractive strategy, two major challenges need to be addressed in future studies. The first challenge is the current

179

Proefschrift_Kirill.indb 179 10-05-15 23:47 lack of a reliable reference gene, preferably a miRNA the expression of which is not affected by gender, age or disease. The second challenge is the origin of miRNA’s found to be aberrantly expressed in serum of patients with disease. MiR-199a-3p is a known tumor suppressor miRNA in other solid malignancies, but was previously reported to be increased in EAC compared to squamous epithelium. In contrast, little is known about the function or tissue expression of miR-320e, making it difficult to functionally correlate circulating miRNA levels with tissue expression. Moreover, a recent paper reported that the majority of cell-free circulating miRNA’s previously reported as cancer biomarkers were highly expressed in blood cells (18), suggesting the possibility that changes in blood cell composition and/or integrity (e.g. hemolysis) can affect circulating miRNA levels. It has been recently shown that tumor-derived exosomes contain miRNA’s and are able to perform paracrine signaling (19). MiRNA’s from tumor-derived exosomes could provide a more reliable source of circulating miRNA’s, and should be studied further in the context of EAC.

Correlating esophageal adenocarcinoma survival with biomarker expression; promises and challenges of the cancer stem cell model Loss of CD44 and SOX2 was found to be associated with poor survival in two independent cohorts, but the mechanism behind this finding is unknown. The results of chapter 7 and 8 raise two main questions for future studies: do CD44 and/or SOX2 mark subpopulations of EAC cells with stem cell-like properties, and what is the mechanism behind the observed adverse effect of low CD44 and SOX2 expression on survival? Stemness assays using (preferably) patient tumor samples could be used to investigate whether expression of CD44 and/or SOX2 is associated with characteristics of stem cells, such as self-renewal and enhanced colony formation in a limited dilution assay. Animal models implanted with CD44/SOX2 knockout EAC cell lines can be used to investigate the effect of CD44 and SOX2 loss on tumor growth and in vivo behavior, possibly explaining the mechanisms underlying the poor prognostic value of CD44 and SOX2 loss found in chapters 7 and 8. Recently, two papers reported the identification of a circuit consisting of the WNT, BMP and GATA6 that regulates CSC expansion in colon cancer. WNT signaling promotes the expansion of colon CSC’s while BMP signaling induces CSC differentiation, antagonizing the oncogenic effect of WNT (20,21). GATA6 is a central mediator of this circuit and maintains stemness by stimulating WNT and antagonizing BMP signaling. Several findings support the notion that WNT/BMP/GATA6 circuit may also bea regulator of presumed CSC’s in EAC, and thus a potential treatment target. We describe a progressive increase in GATA6 expression during BE malignant transformation in vivo (chapter 5), while other studies described an increase in WNT signaling in BE and EAC (22,23) and a downregulation of the BMP4, a ligand of the BMP signaling pathway in EAC

180

Proefschrift_Kirill.indb 180 10-05-15 23:47 (24). Animal models allow modulation of WNT and BMP signalling in the esophagus and can be used in future studies to validate the existence of the WNT/BMP/GATA6 circuit in BE and EAC. If this signaling axis also exists in EAC, it can be targeted through several strategies. First, as reviewed in Chapter 1, a number of WNT antagonists are currently undergoing trials in EAC, and results of their effect on tumor growth and potentiation of chemotherapy effect are eagerly awaited. A second approach to target EAC CSC could be the addition of BMP4. In colon cancer BMP4 suppletion antagonized WNT signaling induced cell death and sensitized colon cancer CSC to traditional chemotherapy (20), and a similar approach should also be tested in EAC models. However, while the CSC theory is an attractive conceptual approach to understand EAC biology and could yield novel therapies based on the selective targeting of the stem cell compartment, the absence of functionally validated CSC markers in EAC is an important limitation, and requires further research. In conclusion, a better understanding of the molecular mechanisms of BE and EAC could provide new solutions for the current clinical challenges in BE and EAC management. In this thesis we show that aberrant activity of embryological signaling pathways, such as the RA pathway, could be a driver of BE development and this provides novel therapeutic opportunities. We found two circulating miRNA’s that can identify patients with BE and HGD. In addition we show that low expression of CD44 and SOX2 predicts a poor survival outcome in EAC patients. With further research and better models, the increased understanding of BE and EAC biology will likely translate into new surveillance options and therapeutic targets, and thus into improvement of preventive strategies, surveillance and treatment of our patients.

181

Proefschrift_Kirill.indb 181 10-05-15 23:47 References

(1) Chang CL, Lao-Sirieix P, Save V, De La Cueva Mendez G, Laskey R, Fitzgerald RC. Retinoic acid- induced glandular differentiation of the oesophagus. Gut 2007 Jul;56(7):906-917.

(2) Kosoff RE, Gardiner KL, Merlo LM, Pavlov K, Rustgi AK, Maley CC. Development and characterization of an organotypic model of Barrett’s esophagus. J Cell Physiol 2012 Jun;227(6):2654-2659.

(3) Lin L, Bass AJ, Lockwood WW, Wang Z, Silvers AL, Thomas DG, et al. Activation of GATA binding protein 6 (GATA6) sustains oncogenic lineage-survival in esophageal adenocarcinoma. Proc Natl Acad Sci U S A 2012 Mar 13;109(11):4251-4256.

(4) DeGrendele HC, Estess P, Siegelman MH. Requirement for CD44 in activated T cell extravasation into an inflammatory site. Science 1997 Oct 24;278(5338):672-675.

(5) Lopez JI, Camenisch TD, Stevens MV, Sands BJ, McDonald J, Schroeder JA. CD44 attenuates metastatic invasion during breast cancer progression. Cancer Res 2005 Aug 1;65(15):6755-6763.

(6) Cho YY, Kim DJ, Lee HS, Jeong CH, Cho EJ, Kim MO, et al. Autophagy and cellular senescence mediated by Sox2 suppress malignancy of cancer cells. PLoS One 2013;8(2):e57172.

(7) Leung EL, Fiscus RR, Tung JW, Tin VP, Cheng LC, Sihoe AD, et al. Non-small cell lung cancer cells expressing CD44 are enriched for stem cell-like properties. PLoS One 2010 Nov 19;5(11):e14062.

(8) Sahlberg SH, Spiegelberg D, Glimelius B, Stenerlow B, Nestor M. Evaluation of cancer stem cell markers CD133, CD44, CD24: association with AKT isoforms and radiation resistance in colon cancer cells. PLoS One 2014 Apr 23;9(4):e94621.

(9) Yoon C, Park do J, Schmidt B, Thomas NJ, Lee HJ, Kim TS, et al. CD44 Expression Denotes a Subpopulation of Gastric Cancer Cells in Which Hedgehog Signaling Promotes Chemotherapy Resistance. Clin Cancer Res 2014 Aug 1;20(15):3974-3988.

(10) Gomez-Casal R, Bhattacharya C, Ganesh N, Bailey L, Basse P, Gibson M, et al. Non-small cell lung cancer cells survived ionizing radiation treatment display cancer stem cell and epithelial- mesenchymal transition phenotypes. Mol Cancer 2013 Aug 16;12(1):94-4598-12-94.

(11) Gibson MK, Zaidi AH, Davison JM, Sanz AF, Hough B, Komatsu Y, et al. Prevention of Barrett esophagus and esophageal adenocarcinoma by smoothened inhibitor in a rat model of gastroesophageal reflux disease. Ann Surg 2013 Jul;258(1):82-88.

(12) Mari L, Milano F, Parikh K, Straub D, Everts V, Hoeben KK, et al. A pSMAD/CDX2 complex is essential for the intestinalization of epithelial metaplasia. Cell Rep 2014 May 22;7(4):1197-1210.

(13) Zaidi AH, Gopalakrishnan V, Kasi PM, Zeng X, Malhotra U, Balasubramanian J, et al. Evaluation of a 4-protein serum biomarker panel-biglycan, annexin-A6, myeloperoxidase, and protein S100-A9 (B-AMP)-for the detection of esophageal adenocarcinoma. Cancer 2014 Dec 15;120(24):3902-3913.

(14) Hammoud ZT, Dobrolecki L, Kesler KA, Rahmani E, Rieger K, Malkas LH, et al. Diagnosis of esophageal adenocarcinoma by serum proteomic pattern. Ann Thorac Surg 2007 Aug;84(2):384- 92; discussion 392.

(15) Thompson OM, Beresford SA, Kirk EA, Bronner MP, Vaughan TL. Serum leptin and adiponectin levels and risk of Barrett’s esophagus and intestinal metaplasia of the gastroesophageal junction.

182

Proefschrift_Kirill.indb 182 10-05-15 23:47 Obesity (Silver Spring) 2010 Nov;18(11):2204-2211.

(16) Kendall BJ, Macdonald GA, Hayward NK, Prins JB, Brown I, Walker N, et al. Leptin and the risk of Barrett’s oesophagus. Gut 2008 Apr;57(4):448-454.

(17) Zhai R, Zhao Y, Su L, Cassidy L, Liu G, Christiani DC. Genome-wide DNA methylation profiling of cell-free serum DNA in esophageal adenocarcinoma and Barrett esophagus. Neoplasia 2012 Jan;14(1):29-33.

(18) Pritchard CC, Kroh E, Wood B, Arroyo JD, Dougherty KJ, Miyaji MM, et al. Blood cell origin of circulating microRNAs: a cautionary note for cancer biomarker studies. Cancer Prev Res (Phila) 2012 Mar;5(3):492-497.

(19) Melo SA, Sugimoto H, O’Connell JT, Kato N, Villanueva A, Vidal A, et al. Cancer Exosomes Perform Cell-Independent MicroRNA Biogenesis and Promote Tumorigenesis. Cancer Cell 2014 Nov 10;26(5):707-721.

(20) Lombardo Y, Scopelliti A, Cammareri P, Todaro M, Iovino F, Ricci-Vitiani L, et al. Bone morphogenetic protein 4 induces differentiation of colorectal cancer stem cells and increases their response to chemotherapy in mice. Gastroenterology 2011 Jan;140(1):297-309.

(21) Whissell G, Montagni E, Martinelli P, Hernando-Momblona X, Sevillano M, Jung P, et al. The transcription factor GATA6 enables self-renewal of colon adenoma stem cells by repressing BMP gene expression. Nat Cell Biol 2014 Jul;16(7):695-707.

(22) Bian YS, Osterheld MC, Bosman FT, Fontolliet C, Benhattar J. Nuclear accumulation of beta- catenin is a common and early event during neoplastic progression of Barrett esophagus. Am J Clin Pathol 2000 Oct;114(4):583-590.

(23) Osterheld MC, Bian YS, Bosman FT, Benhattar J, Fontolliet C. Beta-catenin expression and its association with prognostic factors in adenocarcinoma developed in Barrett esophagus. Am J Clin Pathol 2002 Mar;117(3):451-456.

(24) van Baal JW, Milana F, Rygiel AM, Sondermeijer CM, Spek CA, Bergman JJ, et al. A comparative analysis by SAGE of gene expression profiles of esophageal adenocarcinoma and esophageal squamous cell carcinoma. Cell Oncol 2008;30(1):63-75.

183

Proefschrift_Kirill.indb 183 10-05-15 23:47 Proefschrift_Kirill.indb 184 10-05-15 23:47 Nederlandstalige samenvatting (Summary in Dutch)

Proefschrift_Kirill.indb 185 10-05-15 23:47 Proefschrift_Kirill.indb 186 10-05-15 23:47 Nederlandstalige samenvatting

Het adenocarcinoom van de slokdarm (AC) ontwikkelt zich via de metaplasie-dysplasie- carcinoom sequentie uit metaplastisch cilindrische slokdarmepitheel, het z.g. Barrett’s oesofagus (BO). Toegenomen reflux van maagzuur in de slokdarm, oftewel gastro- oesofagealerefluxziekte, is de belangrijkste risicofactor voor het ontwikkelen van BO, maar slechts een minderheid van de patiënten met refluxziekte zal gedurende het leven BO ontwikkelen. Onderzoek heeft aangetoond dat vroege detectie en behandeling van (pre)maligne afwijkingen de prognose voor EAC patiënten significant verbetert. Daarom worden patiënten met BO regelmatig gecontroleerd door middel van endoscopie. Echter, patiënten met BO hebben vaak weining of geen symptomen, waardoor de diagnose niet gesteld wordt. Een deel van deze patiënten zal zich later presenteren met een AC, en vaak is dan een curatieve behandeling niet meer mogelijk. De huidige in opzet curatieve behandeling van AC patiënten bestaat uit neoadjuvante chemoradiotherapie gevolgd door een chirurgische resectie van de slokdarm. Ondanks dit ingrijpende behandeltraject is de 5-jaars overleving van AC patiënten die in opzet curatief zijn behandeld slechts 50%. Het huidige beleid ten aanzien van BO en AC heeft dus verschillende tekortkomingen, waarvan er een aantal in dit proefschrift besproken worden. Ten eerste zijn de mechanismes die leiden tot de ontwikkeling van BO onvoldoende bekend. Hierdoor wordt de ontwikkeling van nieuwe behandelmogelijkheden beperkt. Ten tweede wordt de detectie van BO sterk beperkt voor de afwezigheid van een goede niet-invasief te bepalen biomarker, en ten derde is een beter begrip van de biologie van het AC noodzakelijk om de overleving te verbeteren. Een beter begrip van de signaaltransductieroutes die verantwoordelijk zijn voor de ontwikkeling van BO zou nieuwe aangrijpingspunten voor therapie kunnen opleveren. In hoofdstuk 1 wordt de rol van de vier belangrijkste signaaltransductieroutes in de ontwikkeling van BO en AC besproken: de Bone Morphogenetic Protein (BMP), Hedgehog (HH), Wingless-Type MMTV Integration Site Family (WNT) en Retinoic Acid (RA) transductieroutes. Deze signaaltransductieroutes spelen ook een centrale rol in de embryonale ontwikkeling van de slokdarm en een beter begrip van de biologie van BO en AC kan mogelijk verkregen worden vanuit het perspectief van de afwijkingen die in deze embryonale signaaltransductieroutes optreden. Goede modellen zijn onmisbaar om het effect van de modulatie van signaaltransductieroutes te bestuderen. De verschillende experimentele modellen van BO en AC hebben elk hun eigen voor- en nadelen en deze worden besproken in hoofdstuk 2. Traditionele cellijnmodellen zijn makkelijk toegankelijk en relatief eenvoudig op te zetten, maar worden beperkt door het ontbreken van een stromacompartiment. Chirurgische diermodellen van BO en AC hebben wel een stromacompartiment, maar het is nog

187

Proefschrift_Kirill.indb 187 10-05-15 23:47 onduidelijk in hoeverre deze modellen dezelfde mutaties hebben als humane tumoren en daarnaast kunnen deze modellen vaak maar op één tijdspunt worden bekeken. Twee belangrijke ontwikkelingen in modellen die beter de humane situatie nabootsen zijn de organotypische modellen die het mogelijk maken om een stroma compartiment in vitro te bestuderen en de in silico modellen die langdurig vervolgd kunnen worden. De laatste kunnen gemakkelijk aangepast kunnen worden aan nieuwe inzichten in de ontwikkeling van BO en de carcinogenese van AC. Een dergelijk organotypisch model wordt gebruikt in hoofdstuk 3, waar het effect van retoïnezuur (RA) op squameuze en cilindrische differentiatie wordt onderzocht. Het toevoegen van RA aan een organotypisch model van squameus epitheel induceerde het verlies van de kenmerkende meerlagige squameuze differentiatie en het verlies van het squameuze eiwit cytokeratine 13. In een organotypisch model van BO induceerde toevoeging van RA de differentiatie tot slijmbekercellen. Slijmbekercellen zijn een diagnostisch criterium voor intestinale metaplasie in BO, maar ze ontbreken in klassieke cellijnmodellen. Hoewel RA in verband werd gebracht met een cilindrische differentiatie, was het tot nog toe onduidelijk welke transcriptiefactoren hierbij betrokken waren. Het doel van hoofdstuk 4 was om de expressie van een panel van transcriptiefactoren die betrokken zijn bij een squameuze (p63, SOX2) of een cilindrische (GATA6, CDX2, SOX9) differentiatie in biopten van patiënten te bestuderen en het effect van RA op de expressie van deze transcriptiefactoren te onderzoeken in een squameuze cellijn. In een serie van 47 biopten van squameus epitheel kwamen p63 en SOX2 tot expressie in de meerderheid van de cellen. De expressie van p63 was compleet afwezig in een serie van 38 BO biopten. Het aantal SOX2-positieve cellen was veel lager in BO, maar zowel in intestinaal-type BO als in niet-intestinaal type BO biopten was een kleine minderheid van cellen SOX2-positief. De cilindrische transcriptiefactoren GATA6 en CDX2 waren aanwezig in een klein percentage squameuze cellen en het aantal positieve cellen voor deze factoren was significant hoger in BO biopten. Het percentage GATA6-positieve cellen verschilde niet tussen intestinaal-type BO en niet-intestinaal type BO, terwijl het percentage CDX2-positieve cellen in intestinaal- type BO hoger was dan in niet-intestinaal type BO. RA behandeling van een squameuze cellijn verminderde de eiwit-expressie van ΔNp63α, een isoform van p63, en zorgde voor een significante toename van GATA6 en SOX9 mRNA. Behandeling met RA had echter geen invloed op mRNA van SOX2 of CDX2. Deze bevindingen suggereren dat squameus en cilindrisch epitheel een verschillend expressieprofiel van transcriptiefactoren hebben, en dat RA het expressieprofiel deels richting een cilindrische differentiatie kan sturen. GATA6 is een transcriptiefactor die betrokken is bij de ontwikkeling van onder andere darmepitheel. De expressie van GATA6 is verhoogd in verschillende tumoren en amplificatie van het GATA6 gen is geassocieerd met een slechte prognose bij OAC

188

Proefschrift_Kirill.indb 188 10-05-15 23:47 patiënten. Het is echter niet bekend of de eiwitexpressie van GATA6 ook een prognostische waarde in AC heeft. Het doel van hoofdstuk 5 was om het expressiepatroon van GATA6 gedurende de ontwikkeling van BO en de maligne transformatie ervan te bestuderen, en te onderzoeken of GATA6 een prognostische waarde heeft in AC. Hiervoor is een cohort van 130 biopten met weefsel van normaal en ontstoken squameus epitheel, BO en BO dysplasie en een tissue microarray met weefsel van 92 AC patiënten gebruikt. Het aantal GATA6- positieve cellen was laag in normaal squameus epitheel, maar nam progressief toe in de metaplasie-dysplasie sequentie. Opvallend was dat het aantal GATA6-positieve cellen al significant was verhoogd in squameus epitheel dat ontstoken was als gevolg van zure reflux. Dit suggereert dat toename van GATA6 een vroege gebeurtenis is in het ontstaan van BO. De eiwit-expressie van GATA6 was echter niet gerelateerd aan overleving en kan dus niet gebruikt worden als prognostische factor. Dit lijkt in tegenstelling tot een eerdere studie die liet zien dat amplificatie van het GATA6-gen wel een negatieve prognostische factor was. Het doel van hoofdstuk 6 was om te onderzoeken of circulerende miRNA’s gebruikt zouden kunnen worden als biomarkers van de metaplasie-dysplasie-carcinoom sequentie in de slokdarm. Hiervoor werden serum samples verzameld van deelnemers zonder afwijkingen in de slokdarm en van patiënten met BO, BO dysplasie en AC. Met behulp van een Nanostring microarray werden 10 miRNA’s geïdentificeerd die een meer dan 2-voudig verschil in serum levels hadden tussen een van de vier studiegroepen. Van deze 10 miRNA’s werden vervolgend 6 gevalideerd in een groter cohort. Na validatie bleek dat het niveau van miR-199a-3p significant verlaagd was in sera van patiënten met BO en het niveau van miR-320e significant was verlaagd in sera van patiënten met BO en BO dysplasie vergeleken met de groep van deelnemers zonder afwijkingen in de slokdarm. Dit is de eerste studie die laat zien dat circulerende miRNA’s wellicht kunnen bijdragen aan een betere identificatie van patiënten met BO en BO dysplasie. Ondanks de matige 5-jaarsoverleving van AC patiënten is er een aanzienlijke variatie in de individuele overleving. Deze variatie biedt de mogelijkheid om biologische karakteristieken te correleren met overleving en hierdoor een beter begrip te krijgen van de biologie van het AC. Kankerstamcellen zijn een veronderstelde subgroep van cellen in de tumor die zijn geassocieerd met agressieve tumorkenmerken en recidivering. In hoofdstuk 7 is de prognostische waarde van een panel van veronderstelde stamcelmarkers (ALDH1, Axin2, BMI1, CD44 en SOX2) onderzocht in een cohort van 94 OAC patiënten die behandeld zijn met een primaire chirurgische resectie. Een lage expressie van CD44 en SOX2 bleek significant gecorreleerd te zijn met een slechte overleving. Hoewel deze bevinding onverwacht is gezien de algemene veronderstelling dat een hoge expressie van stamcelmarkers geassocieerd is met meer agressieve tumorkarakteristieken, kan dit mogelijk verklaard worden op grond van andere functies van CD44 en SOX2. CD44 is

189

Proefschrift_Kirill.indb 189 10-05-15 23:47 onder andere betrokken bij adhesie van cellen onderling met de extracellulaire matrix en moduleert het de ontstekingsrespons, terwijl SOX2 betrokken is bij autofagie en onderdrukking van de celdeling. Een andere mogelijke verklaring is dat CD44 en/of SOX2 goede markers zijn van cellen met stamcelkarakteristieken in vitro, maar niet in vivo. Neoadjuvante chemoradiotherapie is sinds enkele jaren onderdeel geworden van de standaard in opzet curatieve behandeling van AC, maar er is nog maar weinig bekend over mogelijke prognostische en predictieve markers in AC patiënten die behandeld worden met neoadjuvante chemoradiotherapie gevolgd door een chirurgische resectie. In hoofdstuk 8 is de predictieve en prognostische waarde van CD44, SOX2 en SHH onderzocht in twee cohorten: 71 biopten genomen voorafgaand aan de behandeling en 53 resectiepreparaten. Een lage expressie van SHH in de biopten en lage CD44 expressie in de resectiepreparaten waren gerelateerd aan een kortere overleving. De bevinding dat lage expressie van CD44 een negatieve prognostische factor is in AC behandeld met neoadjuvante chemoradiotherapie is in overeenstemming met de resultaten van hoofdstuk 7. Samenvattend kan een beter begrip van de biologie van Barrett’s oesophagus en het adenocarcinoom van de slokdarm nieuwe inzichten verschaffen om de huidige beperkingen in de diagnostiek en behandeling van deze aandoeningen op te lossen. Dit proefschrift laat zien dat embryologische signaaltransductieroutes als de RA signaaltransductieroute mogelijk een belangrijke rol spelen in de ontwikkeling en maligne transformatie van BO. Circulerende miRNA’s kunnen in de toekomst de diagnostiek van BO wellicht verbeteren en een lage expressie van CD44 en SOX2 lijken nieuwe prognostische factoren in AC te zijn.

190

Proefschrift_Kirill.indb 190 10-05-15 23:47 191

Proefschrift_Kirill.indb 191 10-05-15 23:47 Proefschrift_Kirill.indb 192 10-05-15 23:47 Dankwoord (Acknowledgements)

Proefschrift_Kirill.indb 193 10-05-15 23:47 Proefschrift_Kirill.indb 194 10-05-15 23:47 Dankwoord

Hoewel het dankwoord zich aan het einde van een proefschrift bevindt, is het meestal het eerste onderdeel dat gelezen wordt. Hoewel het ‘mijn’ promotieonderzoek was, zou dit proefschrift er niet zijn geweest zonder hulp van mijn begeleiders, collega’s en vrienden, waarvan ik er een aantal in het bijzonder wil noemen.

Beste prof. Kleibeuker, beste Jan, vanaf het moment dat ik vier jaar geleden bij je aanklopte met het idee ‘onderzoek te doen naar Barrett’s’ kon ik altijd bij je terecht voor vragen of voor de zoveelste versie van een artikel, welke jij geduldig had gecorrigeerd. Ik bewonder jouw rust en aandacht voor detail, zowel op wetenschappelijk als sociaal vlak.

Beste prof. Kruyt, beste Frank, ons maandagochtend overleg was altijd een leuk begin van de week. Achteraf denk ik dat je vaak meewarig hebt zitten luisteren als ik weer eens een nieuw experiment had bedacht, maar de ruimte die ik van je kreeg heeft een centrale rol gespeeld in mijn ontwikkeling als wetenschapper.

Beste prof. Van den Berg, beste Anke, ondanks het feit dat Barrett’s niet jouw directe vakgebied is, wist je altijd de juiste vragen te stellen. Hoewel ik onze meetings aan het begin wel spannend vond, heb ik gaandeweg ontzettend veel van je geleerd en hebben we samen veel gelachen. Samen met Joost hebben jullie me kennis laten maken met de wondere wereld van miRNA’s en tijdens het schrijven van mijn proefschrift heb ik veel aan jouw werkethos en scherpe blik gehad.

Beste dr. Peters, beste Frans, of het nou gaat over de endoscopische behandeling van Barrett’s oesofagus of Russische linguïstiek, we vinden altijd wel een onderwerp om over te praten. Dankzij jouw inzet voor de miRBAR-studie hebben we in korte tijd een mooi cohort kunnen includeren, en hopelijk zitten we in de toekomst nog vaker samen aan tafel bij een congres!

Beste dr. Meijer, beste Coby, jouw inbreng was vanaf het begin een drijvende kracht achter het project. Of het nou ging om het last-minute bestellen van reagentia of het nadenken over de opzet van een experiment, altijd was je betrokken en bereikbaar. Op de moeilijke momenten stond je altijd klaar om een oplossing te bedenken, en je hebt een groot aandeel in het succesvol afronden van mijn proefschrift.

Beste ing. Boersma, beste Wytske, zonder jou was ik een promovendus met twee linker handen. Ik heb de meeste praktische vaardigheden in het lab van jou geleerd, en daarnaast

195

Proefschrift_Kirill.indb 195 10-05-15 23:47 hielp jouw nuchtere kijk op de wereld en je humor de mislukte experimenten te relativeren. Ik zal onze samenwerking missen.

Beste dr. Karrenbeld, beste Arend, jouw expertise op het gebied van de slokdarmpathologie was onmisbaar voor dit proefschrift. Hoewel ik uiteindelijk niet voor de pathologie heb gekozen, hebben de gezamenlijke woensdagochtenden achter de microscoop me wel flink doen twijfelen!

Beste prof. Plukker, beste John, mijn eerste stappen in het onderzoek zette ik onder jouw begeleiding. Ik zal met plezier terugdenken aan de discussies op jouw kamer en ons gezamenlijk congres in San Francisco.

Ik wil prof. dr. K. K. Kirshnadath, prof. dr. H. Hollema en prof. dr. S. de Jong danken voor hun deelname in de leescomissie van dit proefschrift.

De (financiele) steun van de Junior Scientific Masterclass, de J.K. de Cock stichting en de van der Meer-Boerema stichting was cruciaal voor het verwezenlijken van dit promotieonderzoek. Vanaf het eerste jaar van mijn geneeskundestudie heb ik via de JSM de mogelijkheid gehad om eerst kennis te maken met wetenschappelijk onderzoek, en daarna de eerste stappen als student-onderzoeker en promovendus te zetten.

De ondersteuning die ik via het GIPS-M traject kreeg heeft mijn wetenschappelijke stage aan het Wistar Institute in de Verenigde Staten mogelijk gemaakt. Dear dr. Maley, dear Carlo, the time spent at your lab was crucial in shaping my interest in cancer research and Barrett’s esophagus in particular. Your approach to cancer from an evolutionary perspective is an ongoing inspiration for me.

De uitvoering van de miRBAR-studie was niet mogelijk geweest zonder de hulp van de medewerkers van de endoscopie-afdeling en de secretaresses van de MDL, die enorm hebben geholpen met de patiënteninclusie. Daarnaast wil ik ook alle patiënten bedanken die hebben meegewerkt aan de miRBAR-studie; dankzij jullie medewerking is het onderzoek mogelijk geworden.

De medewerkers van de afdeling pathologie wil ik bedanken voor de hulp met het selecteren van de juiste coupes, het leren snijden van vriesmateriaal en het beantwoorden van mijn eindeloze stroom aan vragen.

196

Proefschrift_Kirill.indb 196 10-05-15 23:47 Beste Niels, dank voor de mooie lay-out van het boekje. Ik zal nooit meer Enschede en Eindhoven door elkaar halen.

Na mijn promotie begin ik aan de volgende stap in mijn carriere, als AIOS MDL in de regio Nijmegen. Ik wil prof. dr. Drenth, dr. Van Kouwen en al mijn nieuwe collega’s in het Jeroen Bosch ziekenhuis bedanken voor het warme welkom, en ik hoop over een tijdje het onderzoek weer op te pakken in de nieuwe regio.

Ik wil mijn collega’s van de medische oncologie bedanken voor de ondersteuning en de gezelligheid tijdens mijn promotieonderzoek. Justin, Millind, Ingrid, we’ve had a lot of fun during our get-togethers. Esther, Marlous, Jennifer, Linda, Roland, Elly, Annechien, Ines, Roeliene, Martha en alle collega’s van de afdelingen medische oncologie, chirurgie, pathologie en MDL, zonder jullie was mijn promotietijd een stuk saaier geweest!

Beste clubgenoten, jullie zijn een belangrijk deel van mijn studententijd. Samen zijn we (min of meer) volwassen geworden. Ik weet zeker dat we dankzij de vriendschappen die we hebben opgebouwd elkaar zullen blijven zien, ondanks het feit dat we niet meer in dezelfde stad wonen.

Beste huisgenoten (Stijn, Bas, Tuut, Jeroen, Harm, Knoef), dankzij jullie was de Korreweg 202a ook echt een thuis, waar ik ook na mijn afstuderen vaak en graag kon logeren. Het feit dat mijn eerste congresposter nog steeds aan de muur hangt zal ik maar beschouwen als een positieve toevoeging aan het huis.

Beste Joost, Roderick en Allan, toen we elkaar leerden kennen waren we beginnende pubers in de eerste klas van de middelbare school. Intussen zijn we allemaal afgestudeerd maar telkens weten we elkaar te vinden. Onze vriendschap betekent veel voor me - binnenkort weer naar Alanya?

Beste Carolien en Jan, jullie waren mijn collega’s, studiegenoten, maar vooral vrienden. Ik kijk uit naar onze reis naar New York.

Beste Wim en Yvonne, Vincent en Gaby, dankzij jullie was Maarssen altijd een warme, fijne plek. Met jullie heb ik leren kamperen, kanoën en slechte Sinterklaasgedichten schrijven. Ik kom nog graag lang over de vloer.

Lieve moeder, jij hebt me altijd gestimuleerd in mijn wetenschappelijke ambities en onvoorwaardelijk gesteund tijdens mijn studie en promotie. Ik dacht altijd dat we een

197

Proefschrift_Kirill.indb 197 10-05-15 23:47 totaal verschillend karakter hadden, maar ik zie steeds meer dat we op veel vlakken op elkaar lijken. Ik heb een diep respect voor je zelfstandigheid, doorzettingsvermogen en principes. Ik ben ontzettend trots op je.

Nicolette en Arne, het is een eer om samen met jullie voor de corona te staan.

Beste Nicolette, je was mijn beste labmaatje in mijn laatste onderzoeksjaar. We bleken een onverwacht gedeelde liefde voor dieren te hebben, en ik pas graag weer op jouw katten, zolang jij een chocoladereep in je la hebt.

Beste Arne, we kunnen elkaars grappen afmaken, tot menige frustratie van de rest van onze collega’s. Jouw scherpe, kritische blik (woarom dat den?) heeft mijn onderzoek beter gemaakt, en ik bewonder je brede interesse en ambitie. Ik hoop dat we in de toekomst onze gedeelde kwaliteiten kunnen inzetten in de medische wereld (en daarbuiten).

Lieve Judith, we hebben niet veel woorden nodig om te omschrijven wat we voor elkaar betekenen. In het UMCG was je mijn studiebuddy en collega, maar daarbuiten ben je zoveel meer. Op naar het volgende avontuur?

Kirill

198

Proefschrift_Kirill.indb 198 10-05-15 23:47 199

Proefschrift_Kirill.indb 199 10-05-15 23:47 Proefschrift_Kirill.indb 200 10-05-15 23:47