oulu 2019 d 1547 university of oulu p.o. box 8000 fi...

UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

University Lecturer Tuomo Glumoff

University Lecturer Santeri Palviainen

Senior research fellow Jari Juuti

Professor Olli Vuolteenaho

University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

University Lecturer Anu Soikkeli


Publications Editor Kirsti Nurkkala

ISBN 978-952-62-2439-8 (Paperback)ISBN 978-952-62-2440-4 (PDF)ISSN 0355-3221 (Print)ISSN 1796-2234 (Online)

U N I V E R S I TAT I S O U L U E N S I S

MEDICA

ACTAD

D 1547

AC

TAM

adhura Patankar

OULU 2019

D 1547

Madhura Patankar

MODELING AND HISTOPATHOLOGICAL RECOGNITION OF ANOIKIS RESISTANCE IN COLORECTAL CARCINOMA

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF MEDICINE

ACTA UNIVERS ITAT I S OULUENS I SD M e d i c a 1 5 4 7

MADHURA PATANKAR


Academic dissertation to be presented with the assent ofthe Doctoral Training Committee of Health andBiosciences of the University of Oulu for public defence inAuditorium P117 (Aapistie 5B), on 13 Decamber 2019, at12 noon

UNIVERSITY OF OULU, OULU 2019

Copyright © 2019Acta Univ. Oul. D 1547, 2019

Supervised byProfessor Tuomo J. KarttunenDocent Sinikka Eskelinen

Reviewed byAssistant Professor Pekka TaimenDocent Keijo Viiri

ISBN 978-952-62-2439-8 (Paperback)ISBN 978-952-62-2440-4 (PDF)

ISSN 0355-3221 (Printed)ISSN 1796-2234 (Online)

Cover DesignRaimo Ahonen

JUVENES PRINTTAMPERE 2019

OpponentDocent Juha Klefström

Patankar, Madhura, Modeling and histopathological recognition of anoikisresistance in colorectal carcinoma. University of Oulu Graduate School; University of Oulu, Faculty of MedicineActa Univ. Oul. D 1547, 2019University of Oulu, P.O. Box 8000, FI-90014 University of Oulu, Finland

Abstract

Colorectal carcinoma (CRC) is an important cause of cancer-associated deaths. About 30–50% ofCRCs show KRAS or BRAF mutation. In many cancers, anoikis, i.e. apoptosis induced by loss ofextracellular matrix (ECM) contact, is disturbed. Anoikis resistance is essential for the formationof metastases, and since anoikis resistance assessment is based on in vitro cell cultures, theprognostic value of anoikis resistance is largely unknown. We aimed to identify thehistopathological features indicating anoikis resistance in CRC and analyze their prognostic value.The roles of BRAF and KRAS mutations and survivin in anoikis resistance were analyzed and 3-Dcell culture was used to model the histopathology of anoikis resistant (AR) structures.

The two cohorts of CRC cases used in the study consisted of 62 (series 1) and 137 patients(series 2). Immunohistochemistry for ECM proteins enabled identification of tumor cells with andwithout ECM contact, and in both populations, apoptosis was determined with staining forcaspase-cleaved keratin 18. Based on absence of ECM contact and decreased apoptosis rate, weidentified micropapillary (MIP), cribriform and solid structures to represent the putative ARpopulations. High areal density of AR structures associated independently with short survival andwas an independent prognostic factor. MIPs showed lower survivin expression, proliferation andapoptosis rates than non-MIP cells, and low apoptosis rate was associated with poor prognosis instage I and II cases. For 3-D in vitro model of AR structures, we transfected Caco-2 cells withmutated KRAS or BRAF genes; both induced anoikis resistance as measured with Annexin V testin suspension culture. In 3-D cultures, native Caco-2 cells formed polarized cysts. In contrast,mutated cell lines formed partially filled cysts or solid structures, and inverted polarity in KRASmutant cells.

In conclusion, it is possible to identify putative AR structures by conventional histopathologyand their number is associated with poor prognosis. MIPs represent a distinct subpopulation ofCRC cells with features of quiescence. KRAS and BRAF mutations induce anoikis resistance inCaco-2 cells. In 3-D cultures, oncogenes KRAS and BRAF induce solid structures and cell piling,with structural resemblance to putative AR structures observed by histopathology. The mutatedCaco-2 cells thus serve as a model to study the manifestation of anoikis resistance as a distincthistological feature with oncological significance.

Keywords: anoikis resistance, apoptosis, BRAF, carcinoma, colon, extracellular matrix,KRAS, proliferation, survivin

Patankar, Madhura, Anoikisresistenssin mallintaminen ja histopatologinen tunnis-taminen kolorektaalisyövässä. Oulun yliopiston tutkijakoulu; Oulun yliopisto, Lääketieteellinen tiedekuntaActa Univ. Oul. D 1547, 2019Oulun yliopisto, PL 8000, 90014 Oulun yliopisto

Tiivistelmä

Paksu- ja peräsuolisyöpä on yleinen syöpäkuoleman aiheuttaja. KRAS- tai BRAF-geenien mutaa-tio todetaan 30–50 prosentissa suolisyövistä. Anoikis tarkoittaa apoptoosia, jonka käynnistää solun irtoaminen soluväliaineesta. Anoikisresistenssi on etäpesäkkeen synnyn edellytys. Anoi-kisresistenssiä voidaan todeta vain soluviljelyssä, joten sen merkitystä syövässä in vivo ei ole aiemmin arvioitu. Tässä työssä pyrittiin tunnistamaan anoikisresistenssiin viittaavat muutokset histopatologisista suolisyöpänäytteistä ja selvittämään niiden vaikutusta potilaan ennusteeseen. Lisäksi tutkittiin BRAF- ja KRAS-mutaatioiden yhteyttä anoikisresistenssiin ja mallinnettiin anoikisresistenttejä (AR) rakenteita kolmiulotteisessa soluviljelmässä.

Potilasaineisto koostui 199 suolisyöpäpotilaasta. Kudosleikkeistä värjättiin soluväliaineen komponentteja sekä määritettiin apoptoottiset ja jakautuvat solut (M30- ja Ki-67-värjäykset). AR-solupopulaatioiden tunnistamisessa käytettiin kriteereinä soluväliainekontaktin puuttumista ja vähentynyttä apoptoositiheyttä. AR-populaatioiksi osoittautuivat mikropapillaariset (MIP), seulamaiset ja solidit rakenteet. Näiden rakenteiden korkea kokonaisesiintyvyys osoittautui itse-näiseksi huonon ennusteen tekijäksi. MIP-rakenteissa surviviinin ilmentyminen ja apoptoosi- ja proliferaatiotiheys olivat vähentyneet muihin kasvainsoluihin verrattuna. Lisäksi apoptoottisten solujen pieni määrä MIP-rakenteissa liittyi huonoon ennusteeseen paikallisessa syövässä. Mal-linnusta varten Caco-2 solut transfektoitiin mutatoiduilla KRAS- tai BRAF-geeneillä. Onkogee-nien transfektion todettiin indusoivan anoikisresistenssiä. Kolmiulotteisessa soluviljelyssä polarisoituneet Caco-2 solut muodostivat säännöllisiä rauhasmaisia rakenteita. Onkogeeneillä transfektoidut solut muodostivat puolestaan osittain tai kokonaan täyttyneitä rakenteita ja KRAS-transfektio aiheutti solujen polariteetin kääntymistä.

Havainnot osoittavat, että anoikisresistenssiä edustavat rakenteet voidaan tunnistaa kudos-leikkeestä ja niiden runsas määrä viittaa huonoon ennusteeseen. MIP-rakenteissa todettiin lepoti-lan (quiescence) piirteitä. KRAS- ja BRAFmutaatiot aiheuttavat Caco-2 soluissa anoikisresistens-siä. Kolmiulotteisissa soluviljelmissä onkogeenien vaikutus näkyy solujen pinoutumisena, mikä muistuttaa syöpäkudosnäytteissä todettuja AR-rakenteita. Tulosten perusteella modifioituja Caco-2 soluja voidaan hyödyntää anoikisresistenssin mallintamiseen ja tarkempien mekanismi-en tutkimiseen.

Asiasanat: anoikisresistanssi, apoptoosi, BRAF, karsinooma, kooolon, KRAS, proliferaatio, soluväliaine

To Aai

There is no greater Tapasya (penance) than peace, there is no greater happiness than satisfaction, there is no

greater disease than greed, and there is no greater dharma (religion) than mercy.

– Chanakya.

“Neither money pays, not name, nor fame, nor learning; it is CHARACTER than can cleave through adamantine

walls of difficulties.” – Swami Vivekananda.

9

Acknowledgements

The study was conducted at the Department of Pathology, Cancer and Translational

Medicine Research Unit at University of Oulu under the expert supervision of

Professor Tuomo J. Karttunen.

I would like to start with thanking my mentor, Professor Tuomo J. Karttunen

for his strong support and expert guidance. From him, I have learned the importance

of keeping one eye on the details and the other on the larger picture. He has been a

patient coach with a very clear vision, and, to top it all, his kind attitude,

commitment to science and constant availability, made this journey of science

possible and enjoyable for me.

I am thankful to Professor Markus Mäkinen for providing scientific guidance

and support for my thesis. I am very grateful to Docent Sinikka Eskelinen, my co-

supervisor, for introducing me to the interesting field of experimental cell biology,

for sharing her scientific knowledge and for her kind and loving support all these

years.

I am thankful to Assistant Professor Pekka Taimen and Docent Keijo Viiri for

their careful review and expert comments on the thesis. I am grateful to Anna

Vuolteenaho for her excellent linguistic editing of the thesis and making time for

this work.

I wish to thank my former and current thesis follow-up group, Professor Tuomo

Glumoff, Docent Jussi Koivunen, Johanna Mäkinen, MD, PhD, Kristian Koski,

PhD, Irina Raykhel, PhD for their encouragement and guidance towards the

completion of my thesis. I would like to thank former Medical Research Center

(MRC-Oulu) coordinator Mirja Peltola for her kind support. Sincere thanks to

senior colleagues of my department, Docent Anne Tuomisto for her kind support

and troubleshooting experimental halts, and Docent Katja Porvari and Dr Helena

Kaija, DMD, PhD, for their friendship and kind support.

I sincerely thank Professor Pentti Nieminen and Risto Bloigu, MA, for

introducing me to the essentials of medical statistics and their help with data

analysis. I would like to thank Professor Aki Manninen for providing the

infrastructure and his expert guidance on viral-mediated cell culture work. I express

my gratitude to Docent Virpi Glumoff for walking me through flow cytometry

experimentation and analysis. I would like to thank Veli-Pekka Ronkainen, PhD,

for his expert advice on imaging acquisition and analysis. Many thanks to Antti

Viklund and Mika Kaakinen, PhD, for helping with and fixing unexpected

experimental encounters.

10

I want to thank my co-authors Professor Jrki Mäkelä, Sara Väyrynen, MD, PhD,

Taneli Mattila, MD, Vesa-Matti Pohjanen, MD, PhD, Juha Väyrynen, MD, PhD,

Kai Klintrup, MD. I wish to thank former and current friends and colleagues Janne

Capra, Päivi Sirniö, Siina Vaarala, Ilkka Minkkinen, Anne-Mari Moilanen, and

Tiina Kantola; it has been a pleasure to share office space and work with you. I am

grateful to Riitta Vuento, Mirja Mäkeläinen and Erja Tomperi for their technical

expertise and kind support all these years. Sincere thanks to Jaana Träskelin and

Riitta Jokela for their technical help with viral mediated work.

I acknowledge the financial support from the Cancer and Translational

Medicine Research Unit, Medical Research Center Oulu, Oulu University Hospital,

Scholarship fund of University of Oulu and Thelma Mäkikyrö Foundation.

I want to thank my dear friend Sara Väyrynen for her friendship, optimism,

and unconditional support. Humble thanks to Hanna-Riikka Teppo and Laura

Luukkainen for their friendship and optimistic point of view. Heartfelt thanks also

to Susanna Kaisto and Lea Rahtu-Korpela for their caring friendship and being

there! My warm thanks to Virve Pääkkonen, Cynthia Paulin, Satu Myllymäki and

Salla Kangas for kind friendship and thoughtful scientific-non-scientific

discussions. Many thanks to Seija Leskelä for her friendship and teaching me tricks

and tips in graphic design software. I would like to thank Miia Hillilä, Julia

Kniazeva and Päivi Taskila, friends from outside work, for their caring and kind

friendship,

My Mother, Bharati Patankar, has been a rock for us and has beautifully juggled

the multiple roles of mother, father and friend all these years. Her optimism,

kindness and integrity are contagious and enriching. To thank my parents, Vikram

and Bharati Patankar, is similar to thanking God and there aren’t enough words to

do that. Warmest thanks to my little sister Mayura for her unconditional love,

support and believing in me. Finally, extending my warm affection to li’l

munchkins Legolas and Frodo, my sister’s kittens, for their innocent mischief that

kept all the nightmares at bay, and my humble thanks and regards to all the elders

of my family for their love and blessings.

14.11.2019 Madhura Patankar

11

Abbreviations

AI Apoptosis index

AJC Apical Junctional Complex

Akt Protein kinase B

APC Adenomatous polyposis coli

AR Anoikis resistant

Bak Bcl-2 homologous antagonist/killer

BCL-2 B-cell lymphoma 2

BCL-XL B-cell lymphoma-extra large

BH3 Bcl-2 homology domain 3

Bid BH3 interacting-domain death agonist

Bim Bcl-2-like protein 11

BIR Baculovirus inhibitor of apoptosis protein repeat

BRAF v-Raf murine sarcoma viral oncogene homolog B

Caco-2 Human colon carcinoma cell line

CD3 Cluster of differentiation 3

CD8 Cluster of differentiation 8

CD147 Extracellular matrix metalloproteinase inducer

Cdc-42 Cell division control protein 42 homolog

CIMP CpG island methylator phenotype

CIMP-H CpG island methylator phenotype-High

CIN Chromosomal instability

CPT1A Carnitine Palmitoyltransferase 1A

CRC Colorectal carcinoma

CSC Cancer Stem cells

CSS Cancer specific survival

CTC Circulating tumor cells

DFS Disease-Free Survival

DISC Death-inducing signaling complex

DMEM Dulbecco Modified Eagle Medium

DNA Deoxyribonucleotide Acid

E-cadherin Epithelial cadherin

ECM Extracellular matrix

EDTA Ethylenediaminetetraacetic Acid

EGFR Epidermal growth factor receptor

EMT Epithelial Mesenchymal Transition

12

ER Endoplasmic Reticulum

ERK Extracellular signal-regulated kinase

FACS Fluorescence-activated cell sorter

FAK Focal adhesion kinase

GCHP Goblet cell-rich hyperplastic polyp

GFP Green Fluorescent Protein

HIF1α Hypoxia-inducible factor 1-alpha

HP Hyperplastic polyp

HRP Horseradish peroxidase

IAP Inhibitor of Apoptosis Proteins

IBD Inflammatory bowel disease

ICC Interclass correlation coefficient

ILK Integrin-linked kinase

KRAS Kristen rat sarcoma viral oncogene homolog

MCF10A Human breast cancer cells

MCL-1 Induced Myeloid Leukemia Cell Differentiation Protein

MDCK Madine-Darby Canine Kidney cells

MEK Mitogen-activated extracellular signal regulated kinase

MEM Minimal Essential Medium Eagle

MGMT O (6)-methylguanine-DNA-methyl transferase

MIP Micropapillary structures

miRNA Micro-ribonucleic acid

MLH1 MutL homolog 1 gene

MPHP Mucin-poor hyperplastic polyp

MSI-L Microsatellite instable-low

MSS Microsatellite stable

MVHP Microvesicular hyperplastic polyp

N-Cadherin Neural cadherin

Non-MIP Non-micropapillary epithelium

Nox 4 NADPH oxidase 4

PBS Phosphate buffer saline

PIK3CA Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit

alpha

PI3K Phosphoinositide 3-kinases

PIP2 Phosphatidylinositol-4-5-bisphosphate

PIP3 Phosphatidylinositol-3-4-5- triphosphate

poly-HEMA Poly (2-hydroxyethyl methacrylate)

13

PTEN Phosphatase and tensin homolog gene

RIPA Radioimmunoprecipitation assay buffer

ROC Receiver operating characteristic curve

ROS Reactive oxygen species

RTK Receptor tyrosine kinase

SDS-PAGE Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis

snail Zinc finger protein SNAI1

SNORA 42 Box H/ACA small nucleolar RNAs

Sox2 Sex determining region Y-box 2

Src Proto-oncogene tyrosine-protein kinase

SSA Sessile serrated adenoma

TBS Tris-buffered saline

TGFβ Transforming growth factor-beta

TIL Tumor infiltrating lymphocytes

TNF Tumor necrosis Factor

TMA Tissue microarray

TNM Tumor, Node, Metastasis staging

TP53 Tumor protein p53

TrKB Tyrosine kinase B

TSA Traditional Serrated Adenoma

WHO World Health Organization

Wnt Wingless-related integration site

15

Original publications

This thesis is based on the following publications cited by their Roman numerals:

I Patankar M, Väyrynen S, Tuomisto A, Mäkinen M, Eskelinen S and Karttunen TJ. (2018). Micropapillary structures in colorectal cancer: An anoikis-resistant subpopulation. Anticancer Research 38, 2915-2921.

II Patankar M, Mattila T, Väyrynen JP, Klintrup K, Mäkelä J, Tuomisto A, Nieminen P, Mäkinen M J, Karttunen TJ. (2019). Histopathological assessment of anoikis resistance in colorectal carcinoma: Abundance of putative anoikis resistant subpopulations indicate poor prognosis. Manuscript.

III Patankar M, Eskelinen S, Tuomisto A, Mäkinen M, Karttunen TJ. (2019). KRAS and BRAF mutations induce anoikis resistance in Caco-2 cells and characteristic 3D phenotypes in Caco-2 cells. Molecular Medicine Reports, 20: 4634-4644.

17

Contents

Abstract

Tiivistelmä

Acknowledgements 9

Abbreviations 11

Original publications 15

Contents 17

1 Introduction 21

2 Review of Literature 23

2.1 Colon and Rectum ................................................................................... 23

2.1.1 Anatomy and physiology of the colon and rectum ....................... 23

2.1.2 Histology ...................................................................................... 24

2.2 Colorectal cancer ..................................................................................... 25

2.2.1 Conventional carcinoma ............................................................... 26

2.2.2 Serrated carcinoma ....................................................................... 27

2.2.3 Colorectal premalignant lesions ................................................... 27

2.2.4 Prognostic factors in colorectal carcinoma ................................... 29

2.3 Apoptosis ................................................................................................ 34

2.3.1 Pathways of apoptosis .................................................................. 34

2.3.2 Anoikis ......................................................................................... 35

2.4 Mechanism of anoikis resistance ............................................................ 36

2.4.1 Integrin signaling and cell survival .............................................. 36

2.4.2 Cell-cell contacts .......................................................................... 37

2.4.3 Activation of anti-apoptotic pathways .......................................... 38

2.4.4 Apical-basal polarity aberration ................................................... 39

2.4.5 Survivin ........................................................................................ 40

2.4.6 Other factors ................................................................................. 41

2.5 Anoikis resistance in cancer .................................................................... 42

2.5.1 Anoikis resistance and formation of metastasis ............................ 43

2.5.2 Anoikis resistance and prognosis in carcinoma ............................ 43

2.6 Assessment and modeling of anoikis resistance ..................................... 44

2.6.1 Anti-adhesion tests ....................................................................... 44

2.6.2 3-D in vitro culture ....................................................................... 44

2.6.3 Biomarkers associated with anoikis resistance .............................. 45

3 Aims of the study 47

18

4 Materials and Methods 49

4.1 Histopathology ........................................................................................ 49

4.1.1 Patients and tumors ....................................................................... 49

4.1.2 Histopathological analysis ............................................................ 51

4.1.3 Immunohistochemical stainings ................................................... 53

4.1.4 Immunohistochemical analysis ..................................................... 53

4.2 In vitro experiments ................................................................................ 55

4.2.1 Cell culture, staining and imaging ................................................ 55

4.2.2 Image collection and analysis ....................................................... 55

4.2.3 Paraffin embedding of 3-D cultures ............................................. 58

4.2.4 Protein expression analysis ........................................................... 58

4.3 Statistical analyses .................................................................................... 59

5 Results 61

5.1 Identification and quantification of putative AR structures in

carcinomas (I, II) ..................................................................................... 61

5.2 Immunohistochemical assessment of ECM proteins, apoptosis

and proliferation rates and survivin expression ....................................... 62

5.2.1 Expression of type IV collagen and laminin (II) .......................... 62

5.2.2 Proliferation and Apoptosis indexes (I, II) ................................... 62

5.2.3 Survivin expression (I) ................................................................. 63

5.3 Areal density of putative AR structures and their association with

clinical features of carcinomas (II) .......................................................... 66

5.3.1 Putative AR structures and clinicopathological features .............. 66

5.3.2 Putative AR structures associate with prognosis (II) .................... 66

5.4 Effects of BRAF and KRAS mutations on anoikis resistance in

vitro (III) .................................................................................................. 67

5.4.1 Transfection of Caco-2 cells with mutant KRAS or BRAF

genes ............................................................................................. 67

5.4.2 Anoikis resistance assessment with Annexin V test ..................... 68

5.4.3 Cell cycle stage estimation in different Caco-2 cell forms ........... 69

5.4.4 Bim expression in AR cells .......................................................... 70

5.5 3-D in vitro modeling of AR cell subpopulations ................................... 70

5.5.1 Live monitoring: evolution of 3-D structures ............................... 71

5.5.2 Organization and polarization of Caco-2 cells in 3-D

structures: influence of KRAS and BRAF mutations ................... 71

5.5.3 Paraffin embedment of 3-D cultures ............................................. 73

19

6 Discussion 75

6.1 Identification of putative AR subpopulations .......................................... 76

6.2 Putative AR structures and prognosis of carcinoma ................................ 77

6.3 Survivin, an anti-apoptotic protein influencing tumorigenesis ............... 78

6.4 3-D in vitro model with mutated KRAS and BRAF oncogenes

transfected Caco-2 cells to model anoikis resistance .............................. 79

6.5 KRAS and BRAF mutations induce anoikis resistance in adhesion

free culture in Caco-2 cells ..................................................................... 82

6.6 Anoikis-resistant cells and putative AR cells present in

carcinomas show features of dormancy or quiescence ........................... 84

6.7 A model untangling the fates of AR subpopulations in cancer

formation ................................................................................................. 84

6.8 Future perspectives ................................................................................. 85

7 Conclusions 89

List of references 91

Original publications 107

21

1 Introduction

Colorectal cancer (CRC) is the third most common cause of cancer-related deaths.

CRC usually develops slowly, and tumor formation involves many steps comprised

of genetic, epigenetic, histological and morphological changes that accumulate

over a long time. This often results in presentation of symptoms quite late when the

tumor is of considerable size, affecting the passage of feces, along with cramps and

bleeding. Surgery is the main treatment option for non-metastatic CRC (Haddad,

Eid, Kourie, Barouky, & Ghosn, 2018; Simon, 2016). Thorough clinical evaluation

with imaging and histological assessment of surgical resections allows

determination of stage and estimation of other factors for outcome assessment and

consideration of possible adjuvant therapy. However, the benefits of adjuvant

therapy after surgery are debatable, since it is difficult to identify the patients at risk

of recurrence (Tyrrell & Kerr, 2018). Therefore, there is a need for identification

and validation of prognostic and predictive markers to assess tumor recurrence in

stage II CRC patients to enable appropriate decision for adjuvant treatment course

(Eriksen et al., 2018; Haddad et al., 2018). For example, a low number of CD3+

and CD8+ tumor infiltrating lymphocytes (TILs) at the invasive front represents

poor prognosis for stage II CRC patients (Eriksen et al., 2018). Some other

predictive and prognostic markers, such as aberrant miRNA expression, mutations

of KRAS or BRAF genes, and high proportions of Ki-67 expressing cells are

promising, but they require thorough clinical testing and reproducibility validation

for decision of optimal treatments.

In normal healthy tissues, unwanted or faulty cells are removed by a process

called cellular apoptosis in order to maintain tissue homeostasis. Anoikis is a form

of apoptosis induced when a cell loses its contact with the extracellular matrix.

Anoikis was described in the year 1994 by Frisch and Francis (Frisch & Francis,

1994). Resistance to anoikis is important in the formation of cancer metastases, as

otherwise, cells detached to circulation would die by induction of apoptosis.

Anoikis resistance can be a result of disturbed integrin signaling, faulty cell death

mechanism, disrupted apical-basal cell polarity, epithelial-mesenchymal transition,

cellular dormancy, hypoxia or others factors.

Related to the essential role of anoikis resistance in metastasis, identification

of features indicating it would likely be important in the assessment of prognosis

and selection of optimal therapy of malignant tumors, and therefore, likely to be

important in routine clinical assessment of malignant tumors. So far, detection of

anoikis resistance has been limited to anti-adhesion in vitro cell culture-based tests.

22

Conventional histopathological methods have not been adopted to assess anoikis

resistance. To address this issue, we aimed to identify and quantitate cells with

anoikis resistance by conventional histopathological methods based on the basic

premise of anoikis resistance, survival of tumor cells without contact with the ECM.

To get indirect support for the concept, we evaluated the prognostic significance of

the amount of anoikis-resistant tumor cell subpopulations. Finally, we tested

whether corresponding putative anoikis resistant (AR) structures can be formed by

an in vitro 3-D model.

Firstly, we identified micropapillary (MIP) structures representing a

subpopulation of putative AR structures in a series of colorectal polyps and

carcinomas. In the same study, we assessed proliferation and apoptosis rates and

survivin expression in polyps and cancers in MIPs and non-MIP epithelium.

Built on the same idea, we extended our search of putative AR subpopulations

to other subpopulations besides MIPs by analyzing another series of CRC and

evaluated the prognostic significance of these subpopulations.

To study AR structures similar to actual carcinomas, we generated an in vitro

model with a modified Caco-2 colorectal cancer cell line by transfecting the cells

with mutated KRAS and BRAF oncogenes and assessed the growth pattern. In

addition, we investigated anoikis resistance ability by anti-adhesion tests and

studied the survival mechanisms in AR subpopulations.

23

2 Review of Literature

2.1 Colon and Rectum

2.1.1 Anatomy and physiology of the colon and rectum

The lower gastrointestinal tract consists of the large intestine, appendix and anus.

The large intestine is further divided into the first section, i.e. the caecum, followed

by the ascending and transverse colon. After the splenic flexure, the colon continues

as descending colon, and when the intestine enters the pelvis, it turns into the S-

shaped sigmoid colon, followed by the rectum, and ends at the anus, which opens

to the exterior (Ponz de Leon & Di Gregorio, 2001) (Figure 1). The surface area of

the large intestinal mucosa of an adult human is about 2 m2 (Helander & Fandriks,

2014). In contrast to the small intestine, the large intestine lacks

Fig. 1. Anatomy of the large bowel adapted from (Ponz de Leon & Di Gregorio, 2001).

villi and large volume of contents, and it absorbs fluids, processes luminal contents,

and acts as a reservoir (San Roman & Shivdasani, 2011).

The main functions are absorption and storage. The colon and rectum absorb

water and nutrients, including those remaining after absorption in the small

intestine and those formed by microbes. They store indigestible food residue before

24

it can be eliminated from the body via the anus in the form of feces (Cheng et al.,

2010; Greenwood-Van Meerveld, Johnson, & Grundy, 2017).

The blood supply to the colon derives from the superior mesenteric artery (caecum

to splenic flexure) and the inferior mesenteric artery (descending colon, sigmoid

colon and rectum) (San Roman & Shivdasani, 2011) whereas the middle and

inferior artery along with the internal iliac artery supply blood to the lower rectum.

2.1.2 Histology

The most luminal part of the colon is the mucosa, which consists of epithelial cells

and the supportive lamina propria. The epithelial component consists of a single

columnar epithelium lining the surface and crypts opening to the surface. The

epithelium is comprised of goblet cells, absorptive cells, undifferentiated cells, and

endocrine cells. At the base of the intestinal crypt lies a stem cell pool that divides

continuously; the daughter cells differentiate into new specialized cells of the

intestinal epithelium (Barker et al., 2009) The lamina propria contains pericryptal

cells supporting the epithelial cells, occasional smooth muscle cells, blood and

lymphatic vessels, nerves and the cells of innate and acquired immune system.

Fig. 2. Colon and rectum histology.

25

A thin smooth muscle layer, muscularis mucosae, separates the mucosa from the

submucosa composed mainly of loose connective tissue with fibrocytes and

supportive extracellular matrix (ECM), nerves, occasional lymphocytes, and mast

cells. Smooth muscle layers known as muscularis interna and externa extend from

the submucosa to the serosa and form the main bulk of the wall. (Ponz de Leon &

Di Gregorio, 2001) (Figure 2).

2.2 Colorectal cancer

In general, carcinogenesis may result from inactivation of tumor suppressor genes

including gatekeepers that regulate tumor growth by inhibiting cell proliferation or

enhancing death (e.g. APC or TP53 in CRC), caretakers that maintain genomic

stability and, in their absence, cause increased mutation in other genes (MLH1) or

landscapers which influence cellular microenvironment (e.g. PTEN loss). On the

other hand, cancer could also be a result of activation of pro-tumorigenic oncogenes

(e.g. KRAS) which are altered or transformed by mutation, chromosomal

translocation or gene amplification, enhancing expression or causing structural

changes in protein product that functions in proliferation, differentiation, growth

and apoptosis (De Rosa et al., 2015; Nguyen & Duong, 2018).

The development of CRC begins when normal colorectal epithelium transforms

into premalignant lesions composed of genetic alterations and some alterations in

their microscopic structure enabling their recognition. Premalignant lesions include

aberrant crypt foci and adenomas, and progress into cancer by accumulative genetic

alterations (mutation or amplification of gene) or epigenetic alterations, i.e.

inheritable changes (chromatin modification, aberrant DNA methylation, histone

modification and noncoding RNAs) modifying gene expression without changes in

the DNA sequence (Grady & Carethers, 2008; Pancione, Remo, & Colantuoni,

2012). CRC arises through a sporadic or hereditary pathway, the latter accounting

for about 5% of CRCs including Lynch syndrome (mutation in mismatch repair

genes resulting in accumulative DNA mutations) or familial adenomatous

polyposis (a syndrome caused by a mutation in adenomatous polyposis coli (APC))

gene essential for Wnt signaling (Nguyen & Duong, 2018).

Hereditary pathway to CRC results from germline mutations via the two-hit

hypothesis where a germ cell mutation in a tumor suppressor gene causes the first

hit followed by a second hit at somatic progenitor cells in target tissues (DA Silva,

Wernhoff, Dominguez-Barrera, & Dominguez-Valentin, 2016; Jass, 2007). CRC

may also develop via sporadic pathway by a somatic genetic or epigenetic change

26

during tumor initiation. By definition, sporadic cancers occur in patients without

history suggesting familial CRC and with no germline mutation (Mundade,

Imperiale, Prabhu, Loehrer, & Lu, 2014; Raskov, Pommergaard, Burcharth, &

Rosenberg, 2014).

Sporadic pathway can lead to three types of colorectal carcinoma: conventional

carcinoma (CAC) via adenoma-carcinoma sequence, serrated CRC (SAC) that

evolves via serrated pathway, and inflammatory bowel disease (IBD) associated

pathway (Beaugerie & Itzkowitz, 2015; Leggett B & Whitehall V, 2010).

2.2.1 Conventional carcinoma

For many decades, it was thought that colon cancer can only form by the adenoma-

carcinoma sequence, which is mostly initiated by chromosomal instability (CIN)

leading to cumulative genomic alterations. Fearon and Vogelstein proposed a

multistage model for colon cancer progression in 1990 (Fearon & Vogelstein, 1990).

In a normal colon epithelial cell, the Wnt signaling pathway is activated when the

APC gene undergoes mutation (due to frameshift or point mutation). The activation

results in accumulation of beta-catenin, which supports the maintenance of

undifferentiated stem cell phenotype with restricted migration ability (Armaghany,

Wilson, Chu, & Mills, 2012; Leslie et al., 2003). As a result, the accumulated

undifferentiated cells may lead to development of adenomas (early adenoma),

which in turn, via additional mutations (such as activation of proto-oncogene

KRAS), form late adenomas including aberrant cell adhesion and proliferative

signaling network (Armaghany et al., 2012).

According to the model, a late adenoma turns into carcinoma when it

encounters mutation in tumor suppressor genes such as TP53, inactivation of

PIK3CA oncogene, or loss of heterozygosity (e.g. loss of long arm of chromosome

18) mutation occurring in the remaining allele (Borowsky et al., 2018; Leslie et al.,

2003; Pinto, Jacobsen, & Horwitz, 2011).

Cancers forming through conventional adenomas often harbor chromosomal

instability (CIN) and are CpG island methylator phenotype (CIMP) negative and

microsatellite stable (MSS) (Figure 3). Mutated APC is common in CAC compared

to serrated malignancy; however, non-APC-related Wnt signaling has been

implicated in the serrated pathway (Borowsky et al., 2018).

27

2.2.2 Serrated carcinoma

SAC develops via serrated neoplasia pathway with molecular aberrations partially

distinct from the conventional adenoma-carcinoma sequence. This type of

carcinoma is more common in women, particularly in older age, possibly through

the post-menopause effect (low estrogen and folate level) (Noffsinger, 2009).

Histologically, these are glands with serrations, accompanied by frequent mucin

pools, no necrosis and large eosinophilic cytoplasm (Makinen, 2007; Tuppurainen

et al., 2005).

SACs may originate by respective serrated polyps via genetic and epigenetic

mutations. One route involves traditional serrated adenoma where early change

involves KRAS or BRAF mutation accompanied by O6-methyl guanine-DNA-

methyl transferase (MGMT) activation and methylation of other genes, eventually

giving rise to CIMP-High, chromosomally stable, MSI-Low or MSS SAC

predominantly found in the distal colon and rectum. Another route involves sessile

serrated adenomas with early BRAF mutation along with mutations in p16, a tumor

suppressor protein and Insulin-like growth factor-binding protein (IGFBP) 7

promoter hypermethylation and mutL homolog 1 (MLH1) epigenetic alteration

leading to CIMP-High phenotype. Such SACs have BRAF mutations, are

chromosomally stable and MSS, and are predominantly found in the proximal

colon. (De Palma et al., 2019; Makinen, 2007) (Figure 3).

Another route to formation of SAC may arise from adenomatous polyps or

serrated polyps with mutation in the KRAS oncogene. These cancers are CIMP-

Low, chromosomally unstable, harbor MGMT methylation, KRAS mutation and are

MSI-Low or MSS (Leggett B & Whitehall V, 2010; Snover, 2011; Tuppurainen et

al., 2005).

2.2.3 Colorectal premalignant lesions

CAC arises from conventional adenoma (CA) with characteristic features such as

variable dysplasia of epithelium and is more common in males and in patients over

50 years of age. Macroscopically, CAs appear polypoid or sessile, and

histologically, they comprise three subtypes: tubular adenoma (composed of

tubular/gland like structures), tubulovillous adenoma (combination of tubular

structures and villous projections at surface), and villous adenoma (predominantly

villous projections).

28

SAC mostly has a different set of precursor lesions, serrated polyps, which are

comprised of traditional serrated adenomas (TSA), sessile serrated adenomas

(SSA), and hyperplastic polyps (HP). HPs, considered to have less malignant

Fig. 3. Developmental pathways of colorectal cancer adapted from (Armaghany et al.,

2012; Leggett B & Whitehall V, 2010; Noffsinger, 2009)

potential than other serrated polyp types, mainly associate with polyps in the

proximal to sigmoid colon (Yang, Mitchell, Sepulveda, & Sepulveda, 2015). TSAs

are formed of cells with eosinophilic cytoplasm lining the glands and villous

structures. They often have ectopic crypts, i.e. crypts that have lost contact with the

muscularis mucosae, perpendicular to the longitudinal axis of villous projections

(Kalimuthu, Chelliah, & Chetty, 2016). TSA are accompanied by either KRAS or

BRAF mutation and may show features of cellular senescence (Fleming, Ravula,

Tatishchev, & Wang, 2012; Kalimuthu et al., 2016).

29

SSAs typically have crypts with serrated epithelium broadened bases

accompanied by basal branching and are often accompanied by dysplasia (admixed

polyp), a precursor of CRC with MSI (Makinen, 2007). HPs show non-dysplastic

saw tooth-like serrations on the epithelial folds of the upper parts of the crypt and

narrow, non-branching crypt bases. HPs are further divided into goblet cell

hyperplastic polyps (GCHP), microvesicular hyperplastic polyps (MVHP), and

mucin poor hyperplastic polyps (MPHP). GCHP are composed of abundant goblet

cells, lack BRAF mutation, and include less serrations in the upper half compared

to other subtypes. MVHP presents characteristic features of mucin in the cytoplasm

of epithelial cells and often harbor BRAF mutation. Therefore, it is speculative

whether MVHP progress into SSA/P and eventually serrated carcinoma, or whether

the cancer originates directly from MVHP without any intermediary phase (Groff,

Nash, & Ahnen, 2008; Rosty, Hewett, Brown, Leggett, & Whitehall, 2013). Finally,

MPHP are a HP type with the absence of cytoplasmic mucin (Fleming et al., 2012;

Sweetser, Smyrk, & Sinicrope, 2013).

2.2.4 Prognostic factors in colorectal carcinoma

Prognostic factors are defined as measurements showing associations with clinical

outcome either in the absence of therapy or upon standard therapy received by the

patient (Clark, 2008). In addition to estimating overall outcome, prognostic factors

are used to optimize treatment. Currently, in CRC, one of the main objectives is to

find prognostic markers to identify those patients with stage II disease who are at

increased risk of recurrence after surgery and who would benefit from adjuvant

therapies, such as adjuvant chemotherapy (van de Velde et al., 2014).

TNM class and stage

Classification based on the anatomical extent of the tumor is based on the widely

accepted TNM classification system and comprises the extent of local growth of

carcinoma (T; Figure 4), presence of distant metastases (M), and involvement of

lymph nodes (N). The AJCC and UICC TNM classification systems are reviewed

and updated constantly and the latest 8th edition was released in 2017.

Determination of stage provides information about prognosis and is very useful

from the viewpoint of surgical aspects. In addition to detailed TNM class,

anatomical extent can be summarized by translating TNM class to stages I-IV.

30

Stages I and II represent local carcinomas limited to the intestinal wall and adjacent

tissues. Stages III and IV are more advanced carcinomas with local lymph node

metastasis (stage III) or with distant metastases (stage IV). For colon cancer, 5-year

survival in stages I-II is 90%, in stage III 71%, and in stage IV, 14% (Carroll &

Zhao, 2019). (Table 1)

Fig. 4. Histological image showing invasion of colorectal cancer into mucscularis

propria, indicating TNM stage T2.

Histopathological prognostic factors

The characteristics determined by the differentiation of a tumor are recorded as a

histological grade. This is based on WHO standards (Bosman, Carneiro, Hruban,

& Theise, 2010) and the percentage of well-differentiated glands in the tumor as

follows: well-differentiated tumors (> 95% gland forming) are classified as grade

1, moderately differentiated (50–95% gland forming) as grade 2, poorly

differentiated (< 50% gland forming) as grade 3, and undifferentiated as grade 4

tumors (Derwinger, Kodeda, Bexe-Lindskog, & Taflin, 2010). Determination of

grade has rather poor reproducibility and limited prognostic value, but higher

31

Table 1. Tumor staging of colorectal carcinoma

Stage Main pathological and clinical features

Stage I T1 (Tumor invading submucosa), T2 (Tumor invading muscularis

propria)

Absence of lymph node metastasis

No distant metastasis

Stage II T3 (Tumor invading subserosa or into pericolic or perirectal

tissues), T4 (tumor directly invading organ or structures and/ or

perforates visceral peritoneum)

Absence of lymph node metastasis


Stage III Any tumor class

Presence of lymph node metastasis


Stage IV Any tumor class

Presence or absence of lymph node metastasis

Presence of distant metastasis

grade has rather poor reproducibility and limited prognostic value, but higher

grades predict less favorable prognosis (Barresi, Reggiani Bonetti, Ieni, Caruso, &

Tuccari, 2015).

Blood vessel and lymphatic vessel invasion are directly associated with nodal

metastasis and poor prognosis. Lymphatic vessel invasion alone displays strong

association with lymph node metastasis and poor outcome (Barresi et al., 2012; Lin

et al., 2010). Most importantly, lymphatic invasion in CRC is a stage-independent

prognostic factor and crucial for determining adjuvant therapy for stage II patients

(Harris et al., 2008).

Tumor budding comprises small discrete clusters of less than 5 tumor cells

mainly detected and assessed at the invasive front. In addition to being associated

with poor prognosis, the presence of buds is an indicator of epithelial mesenchymal

transition (EMT) and invasive phenotype (Lugli et al., 2017; Marks, West, Morris,

& Quirke, 2018). Moreover, Dawson et al. (Dawson et al., 2015) suggested that

based on low apoptosis rate and expression of a pro-survival molecule, anoikis

resistance might be related with the prognostic significance of budding. Detection

of tumor budding is currently an important prognostic feature.

32

Detection of poorly differentiated clusters (PDC) or solid cancer cell nests is a

relatively new feature. Their abundance is a better indicator of poor prognosis than

the assessment of conventional grade. PDC are defined as five or more cancer cells

that form clusters and are completely devoid of gland-like structures (Reggiani

Bonetti, Barresi, Bettelli, Domati, & Palmiere, 2016; Ueno et al., 2014). The

biology of PDC is not known.

Increased inflammatory cell response at the invasive border associates with

better outcome (Klintrup et al., 2005). The quality of immune response is

determined by immune cell profiling based on IHC assessment (Marks et al., 2018;

Vayrynen et al., 2013; Vayrynen et al., 2014). Tumor infiltrating lymphocyte (TIL)

response is also known to be associated with primary and metastatic colorectal

cancer (Kong et al., 2019). In addition, systemic level inflammation as detected by

blood C reactive protein level and albumin level associates with less favorable/poor

prognosis (Kantola et al., 2012; Proctor et al., 2011). Similarly, systemic

inflammation assessed by blood neutrophil-to-lymphocyte ratio (NLR) has

prognostic value, low NLR favoring poor survival (M. X. Li et al., 2014). NLR

combined with serum albumin concentration gives a systemic inflammation score

(SIS) and this score represents a prognostic factor (Shibutani et al., 2018).

Apoptosis and proliferation

Estimation of proliferation by detecting Ki-67 antigen-positive cells with

immunohistochemistry is a well-known independent prognostic factor, high rates

associating with disease progression and poor clinical outcome (Luo et al., 2019;

Melling et al., 2016). Additionally, low cellular apoptosis rate as shown by a low

number of cells expressing cleaved caspase 3 or caspase-cleaved Keratin 18

associates with poor prognosis (Reimers et al., 2014; Saigusa et al., 2014).

Molecular prognostic factors

Along with efforts to understand the mechanism of diseases and find effective

diagnosis and treatment, several genetic and epigenetic markers have been reported.

Genetic alterations in oncogenes are associated with patient outcome. For drugs

targeting the EGFR pathway, mutational analysis of BRAF, KRAS and NRAS

oncogenes is necessary to predict treatment response.

Mutations in KRAS oncogene are found in about 40% of CRCs and commonly

harbor at codon 12 or 13. Such mutations constitutively activate the EGFR pathway,

33

leading to resistance to anti-EGFR therapies (Lievre et al., 2006). KRAS-mutated

carcinomas are less likely to display MSI but associate with less favorable

prognosis among microsatellite stable (MSS) carcinomas (Phipps et al., 2013;

Taieb et al., 2016). Mutations in the BRAF oncogene are present in about 10% of

CRCs, and in MSS carcinomas they are associated with increased risk of disease

progression. BRAF mutation is a negative prognostic marker independent of stage

that is often observed in recurrent colorectal cancer (De Divitiis et al., 2014;

Haddad et al., 2018; Sepulveda et al., 2017).

Mutations of the TP53 gene are linked to poor patient outcome, but its

usefulness as a clinical biomarker has not been validated (V. D. Li, Li, & Li, 2019;

Nguyen & Duong, 2018). Microsatellite instability status (MSI) predicts improved

clinical outcome in colorectal cancer patients (Kang et al., 2018). Whether MSI-H

patients (i.e. patients with high instability) with stage II tumors should receive

adjuvant chemotherapy is still controversial (Nazemalhosseini Mojarad et al., 2016;

Saridaki, Souglakos, & Georgoulias, 2014).

Other biomarkers in blood, plasma and tissues

With extensive demand and ongoing research in the field of diagnostics, new

biomarkers for CRC are reported for clinical utility. At the same time, continuous

efforts are made to validate the prognostic/predictive value of already known

markers. These markers include genetic, epigenetic, and proteomic markers studied

biochemically as well as with histological and cell biological techniques utilizing

samples from tissues and body fluids. Due to the extensive amount of literature,

systematic review of the topic is not possible. As an example of the growing field

of markers are microRNA (miRNA) members of the non-coding RNA family. In

colorectal cancer, miRNA species may act as tumor suppressors or have oncogenic

effects as well as prognostic value (Yiu & Yiu, 2016).

Of the body fluids, blood is a valuable source of novel biomarkers, including

genetic material from CRC (such as cell-free circulating tumor DNA), various

protein markers, and circulating tumor cells. In the quest of easily accessible and

non-invasive tests, detection of biomarkers from blood, plasma and other body

fluids is of great potential value in the future (Fung et al., 2015; Hauptman &

Glavac, 2017; J. J. Jones et al., 2016; Nikolaou et al., 2018).

34

2.3 Apoptosis

Apoptosis is a form of programmed cell death enabling cellular homeostasis and

tissue organization. Two distinct pathways can trigger the complex process of

apoptosis via activation of caspases, leading to fragmentation of DNA with the help

of endonucleases. The process results in elimination of cells, often leaving cell

fragments in the form of small membrane-enclosed apoptotic bodies (Elmore,

2007).

2.3.1 Pathways of apoptosis

Cellular apoptosis may result via two main pathways, intrinsic and extrinsic

pathway. The intrinsic pathway is triggered by DNA damage or as a result of

endoplasmic reticulum stress. In this pathway, mitochondrial disturbance or

detachment of cells from the extracellular matrix induces a direct cascade of events,

including activation of activator BH3 proteins. These proteins promote Bax-Bak

oligomerization, permeabilization of mitochondrial membrane, and release of

cytochrome C from the mitochondrial pores into cytosol. These proteins form the

apoptosome complex leading to caspase activation and cell death. On the other

hand, sensitizer BH3 proteins can also indirectly activate the pathway by

counteracting the anti-apoptotic functions of Bcl-2 protein, leading to Bax-Bak

oligomerization and caspase-activated cell death (Jin & El-Deiry, 2005; Paoli,

Giannoni, & Chiarugi, 2013).

Another route, an alternative to the sequential activation of caspase-mediated

cellular apoptosis, is the extrinsic pathway initiated by ligand binding to death

receptor (member of TNF superfamily). The binding leads to formation of death-

inducing signaling complex (DISC) and activation of caspase-8. Alternately,

extrinsic and intrinsic pathways may link via caspase-8 activation mediated Bid

cleavage and activation resulting in apoptotic cascade (Figure 5) (Paoli et al., 2013).

35

Fig. 5. Schematic summary of the main apoptosis pathways.

2.3.2 Anoikis

Anoikis is a form of programmed cell death induced by loss of attachment of

cells to an appropriate matrix (Frisch & Francis, 1994). Anoikis contributes to

36

developmental processes such as hollowing of the glands and the involution

process and is often disturbed in cancer. Anoikis ensures elimination of misplaced

for development of metastasis (Gilmore, 2005). Both the intrinsic and extrinsic

apoptosis pathway are involved in the initiation and execution of anoikis (Paoli,

Giannoni, & Chiarugi, 2013). Furthermore, during anoikis, death signal receptor

activity may be modified by interaction with regulatory factors, such as their

known ligands promoting the progression of anoikis. (Bunek, Kamarajan, &

Kapila, 2011; Laguinge et al., 2008; Paoli et al., 2013).

2.4 Mechanism of anoikis resistance

After detachment of malignant cells from their original location, inhibition of

anoikis is an essential mechanism for the formation of metastases. Alterations in

cell death mechanism, faulty integrin signaling or epithelial-mesenchymal

transition (EMT) may result in anchorage-independent survival of tumor cells (Y.

N. Kim, Koo, Sung, Yun, & Kim, 2012; Paoli et al., 2013).

2.4.1 Integrin signaling and cell survival

Integrins are main mediators of anoikis, as these cell surface receptors support anti-

apoptotic and pro-survival signaling when associated with relevant matrix proteins.

In adherent cells, anoikis is inhibited by specific integrin activation and

downstream signaling mediators including integrin linked kinase (ILK), non-

receptor tyrosine kinase (Src), and focal adhesion kinases (FAK). The latter may

contribute to suppression of anoikis in undifferentiated cells as well as to EMT.

Unligated integrins in adherent cells may induce death receptor-independent

(caspase-mediated) cellular death. However, when integrins are ligated, the

integrin-caspase complex is disrupted, enabling cell survival (Stupack, Puente,

Boutsaboualoy, Storgard, & Cheresh, 2001). Integrins can be subdivided based on

their function and location. For instance, some integrins are essential for cell-cell

crosstalk while others contribute to adhesive forces between cells and ECM. The

most common and vital integrins in mammalian cells are fibronectin binding

integrin α5β1, collagen-binding integrin α2β1, and laminin-binding integrin

α3β1 receptors (Desgrosellier & Cheresh, 2010; Paoli et al., 2013; Vachon, 2011).

Basement membrane components maintain structural integrity, function in

tissues, and include type IV collagen, laminin, and other ECM components, as well

as growth factors. In addition to other receptors, integrins link epithelial cells to

37

basement membrane components. The absence of integrin signaling leads to

hyperactive survival and proliferation cascade. In addition, it leads to changed cell

shape as the shape is controlled in normal conditions by the cytoskeleton through

its connection with integrin at cell/ECM and cell/cell junctions.

Integrin α2β1 depletion is implicated in anoikis resistance of tumor cells via

Akt downregulation. Inhibition of α2β1 in melanoma cells grown on non-

adhesive surface leads to upregulation of p53, decreased Bcl-2, and increase in cell

cycle inhibitors, all of which are essential for the proliferation and survival of cells

(Kozlova, Morozevich, Ushakova, & Berman, 2019).

Interestingly, α3β1 is important for glandular phenotype in 3-D matrigel

cultures of intestinal epithelial cells. Lack of β1 Integrin instantly inhibits

glandular morphology along with disturbed cytoskeletal organization, whereas

integrin α3 induces glandular phenotype during the course of development of cyst

in 3-D environment (Zhang, Cromwell, Kunjummen, Yee, & Garcia-Aguilar, 2003).

2.4.2 Cell-cell contacts

Intercellular contacts play a key role in establishing many cellular functions

including proliferation, survival and apoptosis, to maintain tissue homeostasis.

These contacts enable transport of signals and factors to maintain exchange and

normal function. Actin and tubulin polymers responsible for organizing

cytoplasmic organelles, define cell polarity and intact intracellular

compartmentalization in epithelial cells. They determine cell shape and polarity

and promote stability in contact between cells or with ECM via cadherins and

integrins (Hall, 2009).

Cell-cell contact is supported by the adherens junctions located on the lateral

sides of the cells. These junctions provide plasticity and adhesion to cell groups,

allowing tolerance to mechanical stress along with supporting intercellular

transport ions, enzymes and other molecules. Cadherins are essential components

of adherens junctions and are involved in the control of tumor growth (Delva &

Kowalczyk, 2009). A cadherin type identified in epithelial cells is termed E-

Cadherin. It’s decreased expression associates with cancer metastasis and is one of

the hallmarks of EMT (Kourtidis, Lu, Pence, & Anastasiadis, 2017).

Desmosomes employ cadherin binding to intermediate filaments such as

keratins providing mechanical resilience (Garcia, Nelson, & Chavez, 2018).

Hemidesmosomes have very different molecular organization and are structurally

dissimilar from desmosomes. Via conjugation to integrins, they form a link between

38

the cell and the ECM (Garcia et al., 2018; Green & Jones, 1996; J. C. Jones, Yokoo,

& Goldman, 1986). Tight junctions, on the other hand, line the apical part of the

lateral membranes and create a strong bond without any extracellular space to

attach to actin filament networks (Anderson & Van Itallie, 2009). Tight junctions

also regulate a tumor suppressor protein, Scribble, which is important for the

plasticity and function of the intestinal epithelium (Ivanov et al., 2010).

Along with integrins and growth factors, contact and communication between

cells are key in terms of survival and proliferation. Furthermore, this contact

between cells is supported by E-cadherins, and a loss of E-cadherin expression

therefore indicates a loss of cell-cell contact. Thus, in primary or metastatic cancer

cell populations, decreased E-cadherin expression is indicative of active anoikis

resistance-based survival (Fouquet et al., 2004; Hofmann et al., 2007; Taddei,

Giannoni, Fiaschi, & Chiarugi, 2012).

2.4.3 Activation of anti-apoptotic pathways

In cancers, several signaling players promote cell survival and proliferation, in turn

counteracting the induction of apoptosis. Activation of pro-survival signaling

pathways such as Ras-Raf-MEK-ERK and PI3K/Akt enable abnormal regulation

of growth factor receptor-based pro-survival and pro-proliferation signaling and

promotes anoikis resistance (Y. N. Kim et al., 2012). The roles of these signaling

cascades are well acknowledged in the establishment and progression of

malignancy. Furthermore, these pathways are known to be a potential mechanism

towards resistance to chemo, radiation or hormone therapy and may cause cellular

drug resistance (Knickelbein & Zhang, 2015; Saif & Chu, 2010).

RAS-RAF-ERK pathway

In normal condition, mitogen-activated protein kinases (MAPK) are regulated by

phosphatases and can function as tumor suppressor or pro-oncogenic signals, which

is determined by activated signal in the respective context/tissue or can depend on

the strength of the signal. The MAPK pathway is essential for human cell survival,

dissemination and drug resistance and interacts with several other pathways. DNA

damage and metabolic stress may trigger the pathway in conjunction with cellular

contact with the matrix, other cells or external growth factors. Mutations may occur

upstream at receptor (EGFR), transducer (RAS) or regulatory player level

(McCubrey et al., 2007). Post MAPK/ERK activation, ERK detaches from

39

cytoplasm and translocates to the nucleus for transcriptional regulation. The effects

of the pathway in a carcinogenic situation enables anoikis resistance and lead to

metastasis depending on the intensity, duration and localization of the signal

(McCubrey et al., 2007).

The MAPK/ERK pathway is very active during proliferation and invasion,

driving further cell proliferation, dissemination and survival. The pathway

activation can occur upstream at receptor level (receptor tyrosine kinase) or at

downstream targets such as PI3K, Src or RAS. It is rare for two mutations to occur

within the same tumor at once in the RAS/RAF/MEK/ERK pathway (McCubrey et

al., 2007).

PI3K/Akt Pathway

The PI3K pathway can be positively regulated by gain-of-function mutations of

PIK3CA gene whereas loss-of-function mutation triggers negative regulation of

PI3K. Alternatively, overexpression of receptor tyrosine kinase (RTK) or mutation

of downstream effector (e.g. Akt) can lead to activation of PI3K cascade.

Modulators of the pathway include hormones, vitamins, growth factors, integrin,

ras-dependent MAPK pathway, and intracellular calcium (Martini, De Santis,

Braccini, Gulluni, & Hirsch, 2014).

When growth factor binds to receptor tyrosine kinase, it recruits the P85

regulatory subunit to the intracellular part. This results in the formation of a dimer

with the p110 catalytic subunit, leading to subsequent enzymatic activation of PI3K

and generation of PIP3. (Osaki, Oshimura, & Ito, 2004). Negative regulators of

PI3K involve dephosphorylating of PI3K to inactive PIP2 form via phospholipid

phosphatases. The PI3K/Akt pathway is crucial for anoikis suppression mediated

cell survival and Phosphatase and tensin homolog (PTEN), tumor suppressor

frequently mutated in many human cancers and inducer of anoikis, negatively

regulate the PI3K/Akt pathway (Y. N. Kim et al., 2012). Additionally,

immunohistochemical studies have shown Akt upregulation in cancers and

association of phosphorylation of AKT at serine 473 with poor prognosis in many

cancer (Martini et al., 2014).

2.4.4 Apical-basal polarity aberration

Correct apical-basal polarity and mitotic division are essential for tissue integrity

and cellular homeostasis. Epithelial cells establish apical domain already at first

40

cell division between two daughter cells, and this apical domain is maintained

during the course of growth (Jaffe, Kaji, Durgan, & Hall, 2008). Cell division

control protein 42 homolog (Cdc42) is implicated in controlling mitotic spindle

orientation during cell division, and this process regulates epithelial cell

morphogenesis in 3-D cultures.

An aberration of such polarity may lead to uncontrolled cell division in random

direction. Epithelial cells connect apical junction proteins to actin filaments. Apical

junction complex formation enables epithelial phenotype and regulates gene

transcription via apical polarity proteins (Lelievre, 2010).

Nuclear localization of apical protein dictates cell-cell junctional integrity. A

disturbed AJC results in free movement, turning nuclear repressors to promoters,

which drives claudin-2 mediated colon tumorigenesis (Buchert et al., 2010;

Coradini, Casarsa, & Oriana, 2011). A study with mouse gut organoids showed that

tissue polarity was lost in the early stages; however, apical-basal polarity was

initially maintained and changed only during later stages (Fatehullah, Appleton, &

Nathke, 2013). Absence of some integrin receptor proteins has been shown to alter

apical-basal polarity in colon cancer cells resulting in mispositioning of the apical

domain and piling of cells upon proliferation (Zhang, Cromwell, Kunjummen, Yee,

& Garcia-Aguilar, 2003). When cells are attached to ECM, integrins play important

role in controlling anoikis by regulating cell-ECM crosstalk. In the absence of such

contact in detached cells, anoikis can be overcome by cytoskeletal rearrangements

with the help of inside-out mechanism aided by signaling molecules. (Guadamillas,

Cerezo, & Del Pozo, 2011). When cells detach due to stress, lack of integrin

partners or in the absence of matrix stiffness, the inside-out signal is induced to

control anoikis (Debnath, 2010). Not surprisingly, loss of cell polarity and anoikis

resistance have been implicated in the formation of solid carcinomas (Halaoui et

al., 2017; Leung & Brugge, 2012).

2.4.5 Survivin

Survivin is classified as the smallest member of the inhibitor of apoptosis (IAP)

family with Baculovirus IAP Repeat (BIR). Survivin is often upregulated in several

cancer types, yet its role in apoptosis in a stressed cell is not completely understood.

Furthermore, survivin is essential for chromosomal segregation during mitosis

(Altieri, 2008).

The localization of survivin in different cell compartments may deliver

different functions. For example, when expressed in nuclei, survivin implies cell

41

division (Gianani et al., 2001; Mita, Mita, Nawrocki, & Giles, 2008), whereas

cytoplasmic localization is associated with cytoprotective role (Stauber RH, Mann

W, & Knauer SK, 2007). In one study, cytoplasmic survivin levels were inversely

correlated with anoikis in colorectal cancer cell lines, suggesting a role in anoikis

resistance. In carcinoma specimens, high cytoplasmic survivin levels associated

with lymph node and distant metastasis, suggesting that cytoplasmic survivin is a

possible biomarker for metastatic disease (Hori et al., 2013).

2.4.6 Other factors

Epithelial-mesenchymal transition

EMT associates with several aspects of tumor progression, including increased

metastatic potential and anoikis resistance (Frisch, Schaller, & Cieply, 2013). A

combination of events and forces may contribute to disruption of epithelial tissue

organization and acquisition of EMT. EMT is vital for many processes such as

embryonic development, wound repair and tissue remodeling. Indicators of EMT

include loss of epithelial characteristics along with disturbed apical-basal polarity,

loose cell-cell contacts, and acquisition of mesenchymal phenotype, often detected

by mesenchymal markers such as vimentin and N-cadherin (Macara & McCaffrey,

2013).

Suppression of epithelial marker E-cadherin by the transcriptional suppressors

Slug and Snail is often associated with cancer progression and drug resistance in

CRC (Bolos et al., 2003; Vu & Datta, 2017). Along with EMT, a single carcinoma

cell transitioned to a mesenchymal phenotype and showing features resembling

those in the stem cells may enter into lymphatic or blood stream. At a suitable

ectopic site the cell can extravasate, colonize and form metastasis (Macara &

McCaffrey, 2013).

During active EMT, members of integrin families such as αvβ6 are detected at

the cell-ECM junction. Expression of αvβ6 and TGFβ are indicators of invasion

and poor prognosis (Bates & Mercurio, 2005). A recent report by Fessler and

colleagues showed that TGFβ can induce EMT in early serrated pathway lesions

with BRAF mutation, but not in conventional adenomas (Fessler et al., 2016).

EMT has also been extensively studied in circulating tumor cells (CTC) as

CTCs are shown to present mesenchymal phenotype complemented with

overexpression of EMT transcriptional factors (Zeb1, Snail and Slug). Cell surface

42

vimentin detection in CTC from blood of CRC patients is a promising biomarker

for EMT (Satelli et al., 2015).

Reactive oxygen species (ROS)

Solid tumors are often accompanied by oxidative stress and reactive oxygen species

(ROS) that elicit pro-survival signals in tumor cells. Anoikis resistance

accompanied by Ras-induced ERK and Akt activation occurs via ROS induced Src

activation with phosphorylation of epidermal growth factor receptors in ligand-

independent manner (Guadamillas, Cerezo, & Del Pozo, 2011; Y. N. Kim, Koo,

Sung, Yun, & Kim, 2012). However, Ras is not an absolute requirement for ROS-

activated ERK pathway because without any ligand, it can still induce activation of

a growth factor (PDGF) leading to Ras activation and increased ERK1/2 activity

(Y. N. Kim et al., 2012). Studies show that MEK inhibitors can block oxidative

stress-induced ERK activation by inactivating upstream targets (Tan & Chiu, 2013).

Hypoxia

Carcinomas are often comprised of areas with insufficient oxygen supply, which in

normal cells would lead to apoptosis. However, cancer cells acquire anoikis

resistance in hypoxic conditions. HIF-1α induction occurs in tumor cells during

hypoxia (Y. N. Kim et al., 2012) and it seems to promote anoikis resistance through

several mechanisms, including upregulation of integrin α5 (Rohwer et al., 2008),

modulation of autophagy, and activation of nuclear factor-κB (NFκB) (Paoli et

al., 2013). As an alternative to independent effects of HIF1α activation on anoikis

resistance, it is possible that oncogene-induced signaling aberrations lead to both

HIF pathway activation and simultaneous anoikis resistance through separate

mechanisms (Paoli et al., 2013).

2.5 Anoikis resistance in cancer

Anoikis resistance is a key feature enabling cancer cells to survive during formation

of metastasis. Anoikis resistance in cancer cells can be acquired in many ways

including disturbed apoptotic mechanisms, integrin disengagement and EMT.

During stress, induced by detachment of cells from neighboring cells or from ECM,

other survival mechanisms providing anoikis resistance may appear. For example,

autophagy, which can provide resistance to metabolic stress, may support tumor

43

growth by replenishing nutrient loss and increasing energy production. Another

mechanism is entosis, a non-phagocytic process in which a cell that has lost contact

with the ECM can invade into another cell. The host cell may either kill the

invading cell by lysosomal mechanisms or keep the cell alive and release it later.

Although this is a potential mechanism to avoid anoikis, there is so far no evidence

of its significance in the formation of metastasis (Guadamillas et al., 2011; Lugini

et al., 2006).

2.5.1 Anoikis resistance and formation of metastasis

Formation of metastasis involves many steps beginning with cancer cell

detachment from the primary site, followed by migration and survival within the

lymphatic or blood system, attachment, extravasation and invasion to the secondary

site, and finally, proliferation in the distant organ (Bates & Mercurio, 2005). For

survival, the detached cancer cell without any support or signaling from the matrix

must escape the natural defense system and anoikis (Y. N. Kim et al., 2012).

Without anoikis resistance, a detached cancer cell undergoes anoikis, preventing

colony formation elsewhere (Paoli et al., 2013).

Studies on circulating tumor cells support the importance of anoikis resistance

in the formation of metastasis. Human studies have shown that several anoikis

resistance mechanism are activated in circulating tumor cells (Strilic & Offermanns,

2017; Wang et al., 2018) while animal studies suggest that metastatic efficiency

correlates with the activation of anoikis resistance mechanisms (Strilic &

Offermanns, 2017).

2.5.2 Anoikis resistance and prognosis in carcinoma

There are no clinical studies showing a direct association between anoikis

resistance levels and outcome. The expression levels of some biomarkers

functionally associated with anoikis resistance have been assessed for prognostic

value (as shown in 2.6.2). For example, tumor buds show features consistent with

anoikis resistance (such as low apoptosis rate and low expression of caspase-3) and

associate with poor prognosis (Dawson et al., 2014).

44

2.6 Assessment and modeling of anoikis resistance

Anoikis resistance has been addressed by several in vitro cell culture methods. The

methods include culturing of cells in suspension, by plating them on anchorage-

resistant surfaces or by culturing cells in soft gel assays, and by measuring the

apoptosis rate of the cells. Assessment of dysregulated anoikis in gut

pathophysiology has been reported in detail (Vachon, 2018). A colorimetric method

was further developed to monitor anoikis resistance in ovarian cancer cells (He,

Chien, & Shridhar, 2013).

2.6.1 Anti-adhesion tests

Anoikis resistance assessment for single cancer cells or for cells in aggregates has

been performed by different anti-adhesion tests (J. L. Chen et al., 2017). One such

test involves culturing the cells at single cell confluency on anti-adhesive surfaces.

To obtain such a surface, one can coat the cell culture dish with material such as

Poly-HEMA-ethanol solution. This can be complemented by a hypoxic chamber or

by addition of specific substrates such as laminin or poly-L-lysine (Attwell,

Roskelley, & Dedhar, 2000; Bozzo et al., 2006).

Another rigorous and widely utilized method is coating culture plates with agar

solution and employment of colony forming assay. However, this method has

limitations, such as heat stress if the solution is too hot, as well as difficulties in

cell plating and cell retrieval (Borowicz et al., 2014).

2.6.2 3-D in vitro culture

Cell culture in 3-dimension (3-D) is a promising model that mimics several features

of in vivo conditions better compared to 2-dimensional culture (2-D) (Pereira,

Lechanteur, & Sarmento, 2018), including studies for drug permeability, invasion

and migration properties (N. Li et al., 2013; Pereira et al., 2018). 3-D cultures are

generated by growing cells in semisolid media composed of basement membrane

extracts containing laminin, and growth factors or collagen gel, or mix of both.

Hydrogels are another alternative employed for spheroid cultures (Dosh, Essa,

Jordan-Mahy, Sammon, & Le Maitre, 2017; Dosh, Jordan-Mahy, Sammon, & Le

Maitre, 2019).

When epithelial cells such as MDCK cells, MCF10A cells or Caco-2 cells

achieve polarization cue from laminin-rich media, they form well-polarized round

45

cysts with lumen formation at the center via clearing of cells through active anoikis

(Andrew & Ewald, 2010; N. Li et al., 2013). The cysts resemble glandular

structures in vivo (Magudia, Lahoz, & Hall, 2012; Makrodouli et al., 2011; Martin-

Belmonte et al., 2008; Zhang et al., 2003). Due to physiological resemblances to

normal intestinal epithelium, Caco-2 cells are commonly used for studies involving

epithelial transport and permeability (Anabazhagan et al., 2017; Liang et al., 2001).

Lumen formation in epithelial 3-D cysts occurs by either cavitation or hollowing.

Cavitation involves apoptosis of cells trapped within large cell clusters and their

eventual clearance into the lumen, and polarity acquisition occurs post proliferation

and cavitation (Debnath et al., 2002). Hollowing is a mechanism where lumen is

formed through exocytosis and separation of membrane. In this mechanism, cells

acquire apical-basal polarity and intercellular junctions form between the cells.

Released vacuoles then fuse at apical surfaces, forming a small lumen, which

subsequently expands by addition of apical membranes (Lubarsky & Krasnow,

2003; Martin-Belmonte et al., 2008).

2.6.3 Biomarkers associated with anoikis resistance

Several in vitro studies have dissected the regulation of anoikis resistance and found

putative key regulatory molecules. The prognostic value of these molecules has

been assessed in several carcinoma types, and association with poor prognosis has

been constantly shown. Some of these biomarkers and their prognostic/predictive

values are presented in Table 2.

46

Table 2. Biomarkers associated with anoikis resistance in cancer.

Biomarker Putative effect Prognosis References

BCLXL expression

maintains anoikis

resistance

unfavorable prognosis in

ovarian cancer

(Frankel, Rosen,

Filmus, & Kerbel,

2001)

miRNA 451

downregulation

enables anoikis resistance poor prognosis in non-small

cell lung cancer (NSCLC)

(Bian, Pan, Yang,

Wang, & De, 2011)

CD147 upregulation

acquisition of anoikis

resistance

Metastasis of hepatocellular

carcinoma cells

(Ke, Li, Dong, &

Chen, 2012)

RAD21 overexpression

anoikis resistance feature

not studied

poor prognosis in KRAS

mutant CRC

(Deb et al., 2014)

Sox2 expression

induces anoikis resistance

poor prognosis in colorectal

cancer with increased

metastatic capacity

(Lundberg et al.,

2014)

TrKB (Tyrosine kinase

receptor) overexpression

promotes anoikis

resistance


cancer

(Dawson et al.,

2015)

miR-181a

increases anoikis

resistance

poor prognosis in CRC

(Niu et al., 2016)

SNORA42 overexpression

increases anoikis

resistance


cancer with increased

metastatic capacity

(Okugawa et al.,

2017)

MiR10a upregulation

increases anoikis

resistance

poor prognosis in metastatic

breast cancer

(Liu et al., 2017)

NOX4 (NADPH oxidase 4)

overexpression

promotes anoikis

resistance

poor prognosis in CRC

critical for metastatic

progression of gastric

cancer

(Du et al., 2018)

CPT1A-mediated fatty

acid oxidation

increases anoikis

resistance

promotes colorectal cancer

cell metastasis

(Wang et al., 2018)

47

3 Aims of the study

The main aim of the present study was to adopt the concept of anoikis resistance to

histopathological analysis of CRC. For this aim, morphological and

immunohistochemical criteria for putative anoikis-resistant (AR) carcinoma cell

subpopulations were constructed, and an in vitro model was developed. We also

assessed the role of some potential biological mechanisms of anoikis resistance by

immunohistochemistry and by using anti-adhesion cell culture-based tests. Finally,

the clinical significance of the concept was assessed by evaluating the prognostic

significance of the amount of the putative AR subpopulations present within the

tumors.

The specific aims were the following:

I To identify and quantify putative AR tumor cell subpopulations in

histopathological colorectal cancer specimens and to analyze their prognostic

significance (I, II).

II To study the mechanisms behind anchorage-independent survival of AR

subpopulations in colorectal cancer cells in vitro and in tissues using patient

samples (I, II, III).

III To generate a 3-D in vitro model corresponding to the multicellular structures

of putative AR structures as seen in carcinoma tissue specimens (III).

49

4 Materials and Methods

4.1 Histopathology

4.1.1 Patients and tumors

Study I included a case series consisting of 59 samples of normal mucosa, 56 benign

polyps and 62 carcinomas (1986–1996; Cohort 1) collected from the Department of

Pathology, Oulu University Hospital (Table 3). We obtained a representative series of

serrated carcinomas (n = 30) and a comparable number of unselected conventional

carcinomas (n = 32) matched by gender, location, Dukes’ stage and WHO histological

grade. The research project was approved by the Ethical Council of Oulu University

Hospital (25/2002, 58/2005, 184/2009).

Study II included a prospectively recruited series of 149 CRC patients who had

signed a written informed consent to participate and had been operated on in Oulu

University Hospital during 2006–2010 (Cohort 2; Table 3). Out of 149 cases, 12

(8%) were excluded from the study due to insufficiency of the sample material

available, leaving 137 cases for the assessment of anoikis resistance. The research

project was approved by the Ethical Council of Oulu University Hospital (58/2005,

184/2009). The latter study included TMAs constructed previously (Vayrynen et

al., 2012).

For the colorectal carcinoma series (I, II), we obtained clinical data such as age,

gender, treatment, and recurrence, from the clinical records and questionnaires. The

Finnish Cancer Registry provided survival data with 60-month follow-up.

A total of 56 colorectal polyps and polyps adjacent to carcinoma were selected

to represent different types of polyps (I) including 21 CAs (tubular adenomas and

tubulovillous adenomas), 20 SAs, and 10 HPs. In addition, we also assessed 5

adenomas next to the cancer area in specimens collected during 2006–2010 (Table

4).

50

Table 3. Colorectal cancer cohorts in studies I and II

Variable Cohort 1 (n = 62) Cohort 2 (n = 147)

Age, mean (SD) 66.49 (11.65) 66.8 (11.19)

Gender, n (%)

Male

32 (51.6%)

80 (53.69%)

Female 30 (48.38%) 69 (46.30%)

Tumor location, n(%)

Proximal colon 31 (50.8%) 49 (32.8%)

Distal colon 13 (21.3%) 28 (18.7%)

Rectum 18 (29.03%) 72 (48.3%)

WHO grade, n (%)

G1 (well) 27 (44.2%) 21 (14.3%)

G2 (moderately) 25 (40.9%) 108 (72.4%)

G3 (high) 10 (16.1%) 20 (13.6%)

TNM stage, n (%)

Stage I 9 (14.7%) 27 (18.1%)

Stage II 27 (44.2%) 55 (36.9%)

Stage III 18 (29.03%) 44 (29.5%)

Stage IV 8 (13.1%) 22 (14.7%)

T class, n (%)

T 1 5 (3.35%)

T 2 29 (19.46%)

T 3 103 (69,1)

T 4 11 (7.38%)

Lymphatic invasion, n (%)

Yes 61 (44.06%)

No 85 (57.04%)

Infiltrating border, n (%)

Yes 43 (29.05%)

No 105 (70.94%)

Metastasis, n (%)

Yes 19 (12.9%)

No 129 (87.16%)

Cancer type, n (%)

Serrated 30 34 (22.81%)

Conventional 32 115 (77.18%)

In some cases, the following information was missing: Cancer stage (n = 1), T class (n = 1), lymphatic

invasion (n = 3), presence of infiltrating borders (n = 1), metastasis (n = 1).

51

Table 4. Colorectal polyps in study I

Lesion Type N (%) Mean age(range), years

Serrated polyps

35 (62.5%)

63 (45-85)

Sessile serrated adenomas (SSA) 20 (35.7%) 66 (45-81)

Hyperplastic polyps (HP)

Serrated adenomas adjacent to cancer

10 (17.85%)

5 (8.92%)

62 (48-85)

60 (47-72)

Conventional adenomas

Tubular adenomas (TA) and tubulovillous

adenomas (TVA)

Conventional adenomas adjacent to

cancer

21 (37.5%)

16 (28.57%)

5 (8.92 %)

67(45-81)

65 (45-81)

68 (54-77)

4.1.2 Histopathological analysis

Carcinomas and polyps were classified according to WHO 2010 criteria (Bosman

et al., 2010).

Hematoxylin-eosin stained sections from formalin-fixed, paraffin-embedded

specimens were used in studies I and II for stage and grade determination and for

classification of polyp carcinomas. In study II, a TMA was constructed and the

most optimal tumor blocks used for TMAs were selected based on hematoxylin-

eosin stained diagnostic slides. One to four cores of 3.0 mm diameter were

manually sampled from the tumor bulk region in each case (Paarnio et al., 2019).

Stage and Grade

Staging and grades of differentiation were determined by TNM 6 (Greene & Sobin,

2002) and WHO criteria (Bosman et al., 2010), respectively.

52

Recognition and quantitation of the putative AR carcinoma cell

subpopulations

For carcinoma cases, we identified carcinoma cell populations in contact with the

ECM or with stromal areas of the tumors, and cells without such contact. The first

subpopulation included cells of single-layered columnar epithelium and the

outermost layer of multilayered epithelium. The latter subpopulations were defined

as putative AR tumor cell subpopulations, and included cells in micropapillary

structures (MIPs), cribriform structures and solid structures as shown below. In

Study I, only MIPs were identified while in Study II, all three types of putative AR

subpopulations were identified.

Criteria for putative AR tumor cell subpopulations

1. Micropapillary structures. These are composed of piled cells, the minimum

thickness of the pile being one cell (at least one cell piled over the basal cell

layer). The lateral extent may be two or more cells. Basal cells have contact

with the ECM, and as they are in contact with the ECM or stromal areas, are

not included in the putative AR subpopulation.

2. Cribriform structures. Inner cells in confluent groups of cells at least four cells

thick with the innermost cells not forming solid sheets; instead, there are

scattered, seemingly empty, spaces without cells. The outermost cells have

contact with the ECM or stromal areas and are not included in the putative AR

subpopulation.

3. Solid islands. Inner cells in the groups of cells forming solid sheets, with a

minimum thickness of four cells (at least one cell should be without contact

with ECM or stromal area). The outermost cells have contact with the ECM or

stromal area and are not included in the putative AR subpopulation.

For quantitation of the putative AR tumor cell subpopulations (II) we first measured

the area occupied by the tumor in each TMA core representing tumor bulk by using

the virtual microscope software Imagescope (Leica Aperio, Germany). Then, each

occurrence of AR subpopulations was determined in each core and areal density

(structures/mm2) determined.

In the first study (cohort 1), we assessed cells in MIPs and cells with contact

to ECM or stromal areas (non-MIP cells) separately in tumor bulk and at the

invasive front area. In the second study (cohort 2), we studied all three AR

53

subpopulations (i.e. cribriform and solid structures) and cells in contact with ECM

or stromal areas (non-AR cells).

4.1.3 Immunohistochemical stainings

Immunohistochemical stainings were performed on 4 µm sections cut from

respective tumor tissue on Superfrost slides followed by de-paraffinization in

xylene and rehydration through graded alcohol. After rinsing with distilled water

and PBS-Tween, the slides were immunohistochemically stained for the respective

proteins using the reagents and methods described in Table 5. All staining protocols

comprised treatment of sections for antigen retrieval. This was done with Tris-

EDTA or citrate buffer in microwave oven at 800W for 2 min followed by 150W

for 15 min in respective buffers or with pepsin for 30 min at 37°C (Table 5). After

washes, the sections were incubated with primary antibodies as specified in Table

5. Bound antibodies were detected with polymer-based detection reagents (Table 5)

using diaminobenzidine (DAB) as a color reagent. The sections were

counterstained with Hematoxylin and cover-slipped.

4.1.4 Immunohistochemical analysis

Staining patterns were assessed separately in the putative AR subpopulations, including

suprabasal cells of the MIPs (I and II), inner cells of cribriform and solid structures

(II) if present, and tumor cell populations in contact with the ECM or stromal areas (I,

II).

In the first study (I), we assessed survivin expression separately in the cytoplasm

and nucleus. Both the percentage of positively stained cells (0–100%) and intensity of

staining (negative: 0; weak: 1; moderate: 2; strong: 3) were recorded. Due to absence

of significant variation in the intensity of staining, only proportion of positively stained

cells was used for this study. Proliferation and apoptosis indexes were based on the

percentage of cells with positive nuclear Ki-67 staining and cytoplasmic cleaved

keratin-18 (M30) staining, respectively. The scoring was performed manually by

two independent observers blinded to clinical and pathological information. For

each polyp case, the superficial half and deep half of the lesion were scored

separately, and in carcinomas, the invasive front and bulk were scored separately.

For each case, a minimum of five of each subpopulation types was evaluated. Cases

with more than 30% disagreement between evaluators were evaluated jointly and

54

an agreement score was determined. For other cases, the average values of the

immunohistochemistry scores from both observers were used for statistical analysis.

Table 5. Details of immunohistochemical staining methods

Antigen Antigen

retrieval

Antibody Clone Manufacturer Dilution Incubation

time

Detection Paper

Survivin Citrate

buffer

Rabbit

polyclonal

Abcam469 Abcam,

Cambridge, UK

1:1000 30 min Envision I

Ki67 Tris-EDTA Mouse

monoclonal

NCL-Ki67-

MM1

Leica

Biosystems,

UK

1:50 30 min Envision I, II

M30 Tris-EDTA Mouse

monoclonal

M30

Cytodeath

Enzo Life

Sciences,

Lausen,

Switzerland

1:1000 30 min Envision I, II

Collagen

IV

0.4%

Pepsin

Monoclonal

mouse anti-

human

CIV 22 Dako,

Copenhagen,

Denmark

1:50 1 h Ultravision

Large

Volume

Detection

system TP-

125-HL

II

Laminin Leica Bond

Epitope

Retrieval

Solution 1

anti-

LAMC1

Atlas

Antibodies,

Stockholm,

Sweden

1:150 2h Leica Bond

Polymer

Refine

detection kit

DS 9800

II

The second study (II) involved assessment of Ki-67 stained slides, scanned with

Leica Aperio microscope scanner (Leica Biosystems, Nussloch, Germany) by using

the automated digital image analysis platform QuPath (Bankhead et al., 2017). The

putative AR cell subpopulations and tumor cells in contact with ECM or stromal

areas were identified and annotated. We aimed to annotate at minimum 10

subpopulations of each type. This was followed by image analysis workflow

composed of TMA dearraying, cell segmentation, identification, and counting of

positive and negative tumor cells in each annotated area. Mean proportion (%) of

Ki-67 positive cells for each tumor cell subpopulation was determined for each case.

For the assessment of interobserver agreement, a pilot series of ten cases was

studied by three observers. Otherwise, the assessment was shared by three

investigators. Apoptosis index assessment was carried out with scanned sections

stained for M30 by using the Aperio ImageScope program. Positive and negative

55

cells were manually counted from representative areas of each of the putative AR

subpopulation types and similarly from the cell populations with ECM contact. For

M30, one investigator (MP) carried out the initial analysis and an expert pathologist

(TJK) contributed to evaluation of cases with inconclusive staining patterns.

To assess the presence of ECM protein immunoreactivity within the putative

AR tumor cell groups (II), virtual images of sections stained for laminin and type

IV collagen were analyzed using the Aperio ImageScope program, and extracellular

immunoreaction within MIPs, cribriform or solid structures was recorded as

positive or negative. One investigator (MP) performed the assessment, and cases

with inconclusive staining patterns were assessed together with a pathologist (TJK),

who also analyzed a random subset of 30 cases.

4.2 In vitro experiments

We used Caco-2 colon cancer cells known to form well-polarized cysts when

cultured in Matrigel to generate an in vitro 3-D cell culture model for anoikis-

resistant structures.

4.2.1 Cell culture, staining and imaging

We generated Caco-2 cell lines with KRAS G12V and BRAF V600E mutations by

retroviral transfection at the Biocenter Viral Core Facility, Oulu, Finland. Prof. Alan

Hall (Sloan Kettering Institute, New York, USA) kindly gifted us plasmids with

GFP-tagged KRAS G12V and BRAF V600E mutations (Magudia et al., 2012).

Generation of cell lines, assessment of anoikis resistance by anchorage independent

assay, and performance of 3-D cultures is summarized in Table 6, and the reagents

used in these assays in Table 7.

4.2.2 Image collection and analysis

For live cell imaging, Zeiss Cell Observer Spinning Disk confocal microscope set

at 37°C was used. We collected images with 0.6 um step size using 40X objective

and excitation wavelengths of 405 nm, EC Plan NeoFluar 40X/0.75 DIC air

objective and BP 450/50nm (blue) emission filters. We collected images from 5–

15 cysts from each sample using spinning disk confocal microscope and each

experiment was repeated three times.

56

Table 6. In vitro cell manipulation and culture methods

Experiments

Generation of mutated cell lines

Retrovirus production with plasmids PQCXIP-GFP, pQCXIP-KRAS-GFP, pQCFLAP-

BRAF-GFP.

Transfection by Lipofectamine 2000.

Enhancement of viral titer by ultracentrifugation.

Infections to Caco-2 cells with respective retroviruses.

Selection of transfected cells:

Antibiotic selection for 2 days-Caco-2 control and Caco-2 KRAS

cells by Puromycin (4 ug/mL) and for Caco-2 BRAF cells by

G418 (0.8 mg/mL)

Sorting of GFP-positive clones by flow cytometry based sorting

Anchorage independent assay

-Anti-adhesion plate preparations: Poly-HEMA-ethanol solution (0.5 g Poly-HEMA /25

mL 95% Ethanol) plated to 10 cm bacterial cell culture dish.

- Cells at single cell suspension were plated on each 10 cm poly-HEMA coated dish

and cultured at 37°C for 24 and 48 hours, respectively.

- Anoikis-resistant (non-apoptotic) subpopulations were determined with Annexin V

flow cytometric assay with Annexin-negative cell population representing living cells.

3D culture

- 70,000 cells per well were plated on pre-coated (80% matrigel + 20% collagen I)

4-well chambered slides, grown for 10 days, and finally fixed and stained.

- Cultures for live imaging: 10,000 cells were plated on pre-coated (80% matrigel +

20% collagen I) 35 mm glass bottom well dish.

- To provide non-stressed cells for imaging, 3 dishes were initially plated with each cell

line simultaneously. One dish was imaged for 3 days, followed by the second dish for

the next 3 days and third dish for the last four days.

57

Table 7. Cell visualization and characterization methods

Experiment Dye Concentration/

dilution

Incubation

time

Manufacturer Detection Analysis

program

3-D culture

(Live

imaging)

Hoechst,

CellMask TM

plasma

membrane

stain

1:1000 30 min Thermo

Scientific

Zeiss cell

observer

spinning

disk

confocal

microscope

Hygen Pro

Zen

Fixed 10-

day old 3-D

structures

TRITC-

phalloidin

1:10 1 h Invitrogen Olympus

flow-view

confocal

microscope

Zen 2012

Fixed 10-

day old 3-D

structures

Dapi 1:1000 1 h Invitrogen Olympus

flow-view

confocal

microscope

Zen 2012

Anchorage

independent

assay

Annexin V 500 5ul/

100 ul cells

15 min BD

Biosciences

BD LSR

Fortessa

FACS diva

version 7.2

FlowJo X

Cell cycle

phase

estimation

ReadiDropTM

Propidium

Iodide

a few drops/

100 ul cells

15 min Bio-RAD BD LSR

Fortessa

FACS diva

version 7.2

FlowJo X

We viewed fixed 3-D culture specimens with Olympus Flow-view confocal

microscope and collected z-stack images using UPLSAPO 60x/1.35 oil immersion

objective. The excitation wavelengths of 405 and 543 nm and appropriate emission

filters were used for dapi and TRITC-phalloidin, respectively.

3-D cell growths were classified as cysts if they had an outer wall formed by a

single layer of cells and had a lumen without cells or cell debris. Structures filled

completely or partially with cells without a clear lumen as present in cysts, were

identified as solid growth or filled structures. We manually quantified the structures

using 20X objective, classifying and counting 145–212 structures from the fixed 3-

D specimens. A minimum of three independent experiments was performed and the

average proportion was used for graphical representation.

58

Fig. 6. Embedding of 3-D cultures in paraffin adaptation from (Pinto et al., 2011)

4.2.3 Paraffin embedding of 3-D cultures

3-D cultures were generated similarly as for fixed 3-D specimens for 8 days and

transferred to agarose, fixed and embedded in paraffin (Figure 6).

4.2.4 Protein expression analysis

We sorted annexin-negative cells with flow cytometry (BD FACS Aria) and lysed

them by sonication in 1 ml of radio immunoprecipitation assay (RIPA) buffer

59

(consisting of 50 mM Tris-HCL (pH 7.4), 150 mM NaCl, 4 mM EDTA, 1% Nonidet

NP-40 and 0.1% SDS in 200 mL dH2O) with a cocktail of proteinase inhibitors.

Lysates were incubated on ice for 30 min with vortexing every 5 min and cleared

by centrifugation (10,000g, 2min, +4°C). We measured protein concentration with

Bradford Protein assay (Bio-Rad). For western blot analysis, equal amounts (20 ug)

of protein samples were separated on 15% SDS-PAGE and electro-blotted to

nitrocellulose membrane (2 h). We blocked the membrane in 5% skim milk in TBS-

Tween buffer (1 h) followed by washes in 1X TBS-tween buffer, incubated with

primary antibody [Bim rabbit mAb (C34C5, Cell Signaling Technology, USA),

1:1,000; Tubulin mouse mAb (MS-719-P1, Neomarkers, Fremont, CA), 1:5,000;

rabbit polyclonal anti GFP (Ab11122, Invitrogen, UK), 1:1,000 and mouse

monoclonal anti-BRAFV600E (VE1, Roche Diagnostics, Indianapolis, IN, USA),

1:1,000] overnight at +4°C. After washing, the membranes were incubated in HRP

conjugated secondary antibody [goat anti-rabbit IgG (sc-2004, Santa Cruz

Biotechnology, Dallas, USA), 1:5,000; goat anti-mouse IgG (sc2005), Santa Cruz

Biotechnology, Dallas, USA) 1:10,000 for 1.5 h. Antibodies were detected by

incubating the membranes in a detection solution comprising 250 mM Luminol, 90

mM Cumoric and 3 uL H2O2 in 10 mL 0.1 M Tris-HCL (pH 8.3) for 2 min and

chemiluminescence was detected using Fuji-Las-3000 system. Bands were

quantified for presentation with Quantity One (Bio-Rad, Hercules, CA, USA)

software.

4.3 Statistical analyses

IBM SPSS 22 and 25 (IBM Corp., Armonk, NY, USA) software was used for

statistical analyses (Table 8). Reproducibility of the assessments was evaluated by

using interclass correlation coefficient (ICC) based on a mean-rating (k = 2),

absolute-agreement, two-way mixed-effects model. For all the tests, two-tailed

exact p values less than 0.05 were considered statistically significant.

60

Table 8. Statistical methods used

Assessment Test

To assess different parts of same lesion Wilcoxon matched pair test

To assess association of two independent variables Mann-Whitney test

To assess association of three or more independent

variables

To assess correlation between

different AR structures

To assess association to cancer-specific survival or

disease-free survival

To test for independent prognostic factors

To find cutoff value discriminating survival and non-survival

to assess capacity of AR quantification

To compare cell lineages for anoikis resistance and 3-D

growth patterns and for protein expression

Kruskal-Wallis test

Spearman correlation test

Kaplan-Meir curve method or

Hazard plot

Cox regression test: univariate and bivariate

ROC curve

ANOVA (Tukey for post hoc comparisons)

61

5 Results

5.1 Identification and quantification of putative AR structures in

carcinomas (I, II)

We assessed the numbers of putative AR structures: MIPs, cribriform and solid

structures from Cohort 2 material with image analysis techniques as described in

Materials and Methods.

From a total of 137 cases analyzed, 126 (92.0%) cases showed at least one MIP

structure, 125 (91.2%) a cribriform structure, and 121 (88.3%) a solid structure,

respectively. Furthermore, 104 (75.9%) cases presented all three types of the

putative AR subpopulations, 27 (18.7%) cases showed at least two structures, and

6 cases (4.4%) presented only one subtype with solid structures alone in four cases

and MIPs in two cases.

Fig. 7. The figure panel shows collagen IV staining in cribriform (left), solid (center) and

MIP (right) structures in colorectal carcinoma. Positive staining (brown) can be seen in

the interface of tumor cell islands and stroma, surrounding some of the spindle-shaped

stromal cells, and in the walls of blood vessels. No positive staining is present within

the putative AR structures.

We found that areal densities (structures/mm2) of each putative AR subpopulation

type correlated with the total areal density of all AR structures (II; p < 0.001). The

areal density of cribriform structures correlated positively with the densities of

solid structures (II; p = 0.017; rho = 0.204) whereas MIP areal density correlated

inversely with that of solid structures (II; p = 0.006; rho = 0.234).

62

5.2 Immunohistochemical assessment of ECM proteins, apoptosis

and proliferation rates and survivin expression

5.2.1 Expression of type IV collagen and laminin (II)

We assessed the presence of ECM proteins collagen IV and laminin in the tumors

(Cohort 2; II). Positive staining, often discontinuous, was seen along the outer

borders of cancer cell islands of all subtypes (Figure 7). However, putative AR

subpopulations in MIPs, cribriform and solid structures showed no extracellular

positivity, suggesting the absence of organized ECM within these structures (Figure

7).

5.2.2 Proliferation and Apoptosis indexes (I, II)

Proliferation index (PI) and apoptosis index (AI) were based on the counts of Ki-

67 and M30 positive cells (I: Figure 2), respectively. As expected, in normal

colorectal mucosa (I) epithelial cells in the upper half of the crypts showed lower

PI (p < 0.001) and higher AI as compared with lower half (p < 0.001).

In carcinomas (I, II) AI and PI were separately determined in different tumor

cell subpopulations, including cells in contact with the ECM, and different

populations of the putative AR cells without such contact (putative AR structures,

Table 3). In Study I, only MIPs were studied whereas in Study II, all AR

subpopulations were analyzed. Additionally, MIPs were analyzed in polyps (I). AI

was lower in all three putative AR structures as compared with cells in contact with

ECM (Table 9A and 9B, p < 0.001; II: Figure 4), suggesting that these

subpopulations are resistant to anoikis. Proliferation rates were mostly similar in

different subpopulations. Only suprabasal cells in the MIPs presented lower

proliferation rates than the tumor cells with ECM contact (Table 9A and 9B) (I, II).

Serrated and conventional colorectal carcinoma did not show any differences for

PI or AI in MIP and non-MIP epithelium (I, II). When comparing precursor lesion

subtypes, PI and AI rates in MIPs in serrated polyps (I, Table II; PI, 25 (0-75); AI, 0 (0-

4)) were significantly lower than in conventional adenomas (PI, 38 (10-75), p = 0.045;

AI, 1 (0-3), p = 0.045). A similar difference was seen in non-MIP epithelium with lower

PI in SPP than CA (SPP, 28 (7-50); CA, 42 (16-52), p = 0.002), but for AI no significant

difference emerged. In addition, when precursor lesions (combined) were compared to

malignant ones (combined) AI was lower in non-MIP epithelium of polyps compared,

63

but this difference was not present for MIPs in the compared lesion types (Figure 8).

Table 9 A. Distribution of AI and PI in different tumor cell subpopulations in CRC. AI and

PI represent proportion (%) of cells positive for M30 and Ki-67, respectively.

Tumor subpopulations AI (%) median (range) PI (%) median (range)

MIP 2.6 (0-15.23) 41 (0-100)

Cribriform 6.40 (0.43-35.40) 61 (0-100)

Solid 2.59 (0-11.92) 63.5 (0-100)

Total AR 4.57 (1.3-9.07) 56.38 (1-100)

Non-AR cells 12 (2.50- 54.31) 65 (1-100)

Table 9 B. Statistical comparison of PI and AI in different subpopulations in CRC (refers

to Table 9 A).

Compared subpopulations AI (p values) PI (p values)

Putative AR vs. non-AR cells

MIP vs. non-AR cells

p < 0.001 [<]

p < 0.001 [<]

Cribriform vs. non-AR cells p < 0.001 [<] p = 0.287 [=]

Solid AR vs. non-AR cells p <0.001 [<] p = 0.875 [=]

AR subpopulation comparison

MIP vs. Cribriform p < 0.001 [<] p < 0.001 [<]

MIP vs. Solid p = 0.842 [=] p = 0.001 [<]

Cribriform vs. Solid p < 0.001 [>] p = 0.718 [=]

<, > or = in brackets indicates apoptosis or proliferation rates compared between subpopulations

Total AR indicates sum of areal densities of MIPs, cribriform and solid structures.

Non-AR cells refer to tumor cells in contact with ECM or stromal areas.

5.2.3 Survivin expression (I)

We assessed nuclear and cytoplasmic survivin expression in the upper and lower

halves of normal colon mucosa (I). Nuclear survivin staining was higher

(proportion of positively stained cells) in the lower half of the crypts (p < 0.001)

whereas cytoplasmic survivin expression was significantly higher in the upper half (p

< 0.001) (I: Figure 2).

64

Fig. 8. Proliferation and apoptosis indexes in polyps and cancers. Significant differences

are represented by p-values in the graph. The symbols (asterisk) are outliers.

Nuclear survivin expression in MIP structures was significantly higher in cancers

than in polyps (p < 0.001, Figure 9) but there was no such difference in non-MIP

epithelium. In addition, survivin expression in MIPs was lower than in non-MIP

65

Fig. 9. Survivin expression in polyps and cancers. Significant differences between MIP and

non-MIP epithelium are represented by p-values in the graph.

epithelium in both cancers (nuclear survivin, p = 0.026; cytoplasmic survivin, p =

0.012) and in polyps (nuclear survivin, p < 0.001; cytoplasmic survivin, p < 0.001;

Figure 9). In subgroup analysis, polyp subtypes showed similarly lower survivin

66

expression in MIPs than in the non-MIP epithelium [nuclear survivin, SPP, p <

0.001; CA, p < 0.001; cytoplasmic survivin: SPP, p < 0.001; CA, p < 0.001].

5.3 Areal density of putative AR structures and their association

with clinical features of carcinomas (II)

5.3.1 Putative AR structures and clinicopathological features

Low tumor grade associated with high areal density of MIPs (II, Table 3; p = 0.005)

and high tumor grade with that of solid structures (p = 0.029). ). Although BRAF

or KRAS mutations did not associate with total areal density of AR structures, KRAS

mutation associated significantly with low abundance of solid structures (p = 0.048)

whereas BRAF mutation tended to associate with solid structures (p = 0.102). MIPs,

cribriform, solid structures, independently or combined, did not associate with

tumor stage (II: Table 3). No association was seen between putative AR structures

and tumor type (serrated vs. conventional).

5.3.2 Putative AR structures associate with prognosis (II)

We determined cutoff values for low and high areal densities of different AR types

and their sum by ROC curve analysis. Areal densities of MIPs did not associate

with prognosis in the whole case series. However, in subgroup analysis comprised

of only stage I and II tumors cases there was a trend for association of high MIP

areal density with worse survival (p = 0.058). Interestingly, areal densities of

cribriform structures associated with poor cancer-specific survival (CSS, p = 0.004,

II: Figure 5) and poor disease-free survival (DFS; p=0.015, II: Supplementary

figure 1). Similarly, for solid structures, areal density presented a strong association

with worse CSS (p = 0.017, II: Figure 5); however, the association was not

significant for DFS. High total putative AR areal density showed a significant

association with poor prognosis (CSS; p = 0.001; II, Figure 5; Figure 10).

67

Fig. 10. Hazard plot showing cancer survival (60 months; log rank) in patients (n = 138)

with high and low areal density (structure/mm2) of putative AR structures (total AR

count/mm2).

To investigate the independent prognostic value of areal densities of putative AR

structures, we used a multivariate analysis modification described by Kleinbaum

and Klein, since the number of cases in some subgroups was low (Kleinbaum &

Klein, 2012). Accordingly, we assessed prognostic value against one variable at a

time, and found that the total AR count based on MIP, cribriform and solid counts

was an independent prognostic factor when tested against several clinical features

such as age, location, stage, grade and others (II; Table 4). Such total putative AR

areal density was a stronger factor than the combined areal density of cribriform

and solid structures (II).

5.4 Effects of BRAF and KRAS mutations on anoikis resistance in

vitro (III)

5.4.1 Transfection of Caco-2 cells with mutant KRAS or BRAF genes

Caco-2 colorectal carcinoma cells natively not having KRAS or BRAF mutations

were transfected with retroviral vectors encoding GFP-tagged KRAS G12V, BRAF

V600E or GFP only as a control (III). The success of the transfection was confirmed

68

by flow cytometric detection of GFP and by showing the presence of mutant BRAF

protein and GFP-KRAS composite protein by western blotting (III; Supplementary

Figure 1). Control cells, transfected with GFP, were similarly shown to express GFP

(III).

5.4.2 Anoikis resistance assessment with Annexin V test

To assess if mutations have any effect on anchorage-independent survival ability of

colon cancer cells, we cultured control Caco-2 cells and Caco-2 cells transfected

with mutated KRAS or BRAF oncogenes on poor attachment (poly-HEMA coated)

surfaces (III). Annexin V500 FACS assay showed a significantly lower apoptosis

rate in Caco-2 KRAS and Caco-2 BRAF cells when compared to control cells at 24

h (Figure 11). On the contrary, cells grown in suspension for 48 h presented higher

apoptosis rates independent of the vector used (Figure 11).

Fig. 11. Proportions of Annexin V negative (non-apoptotic) cells among control Caco-2

cells and Caco-2 cells transfected with KRAS G12V or BRAF V600E. Cells were cultured

in non-adhesion conditions for 24 h or 48 h.

69

5.4.3 Cell cycle stage estimation in different Caco-2 cell forms

To assess the cell cycle phase of anoikis-resistant cell populations, we cultured

Caco-2 control cells and cells transfected with KRAS G12V or BRAF V600E

mutated oncogenes in suspension for 24 h and 48 h. Propidium iodide assay was

performed on fixed and permeabilized cells. At G0/G1 (pre-replicative) phase,

Caco-2 KRAS cells showed a strong peak with the highest proportion of cells,

followed by Caco-2 BRAF cells, Caco-2 control cells showing clearly the lowest

proportion at both time points, at 24 h and 48 h (Figure 12). In contrast, S-phase

fractions in Caco-2 KRAS, Caco-2 BRAF, and Caco-2 control cells were similar at

both time points (Figures 12 and 13).

Fig. 12. Cell cycle phase plots from annexin-negative sorted Caco-2 control cells (left),

Caco-2 cells transfected with mutated KRAS (center), and Caco-2 cells transfected with

mutated BRAF (right) oncogene. The upper row represents results from 24 h

suspension culture and the lower row results from 48 h suspension cultures.

70

Fig.13. Proportions of anoikis-resistant cells in different cell cycle phases in anchorage-

independent assays (24 h). Readings for native Caco-2 cells are based on one

experiment, and those for Caco-2 KRAS and Caco-2 BRAF are from three identical

experiments. Median (horizontal line), interquartile range (box) and range are shown.

5.4.4 Bim expression in AR cells

To survey possible mechanisms behind anchorage-independent survival in annexin

negative sorted cells, we assessed the expression of Bim, a pro-apoptotic protein

and a member of BCL-2 family. Quite unexpectedly, based on three experimental

replicates, we found a trend for overexpression of Bim in the Caco-2 BRAF and

Caco-2 KRAS cell lysates when compared to Caco-2 control lysates but the

differences was not statistically significant (III; Figure 7).

5.5 3-D in vitro modeling of AR cell subpopulations

To study the effect of KRAS and BRAF mutations in multicellular glandular

formation in vitro we generated a 3-D model with Caco-2 colon cancer cells and

Caco-2 cells transfected with either KRAS G12V or BRAF V600E (III). For 3-D

cultures, we used a mixture of basement membrane extract (Matrigel) plus collagen

I gel and grew the respective cells in the mixture for 10 days.

71

5.5.1 Live monitoring: evolution of 3-D structures

We followed the cells in the 3-D cultures by staining with DNA dye (Hoechst) (III).

Although such live imaging provided a close to normal environment for the cells,

the specimens were still prone to some toxicity induced by the dye after 3 days.

Therefore, cells were grown on three parallel sets and each set was only monitored

for a sequence of 3 to 4 days.

During the initial days, cells proliferated and formed round cell clusters

growing and expanding within the matrix, whereas later, i.e. post 8th day, most

Caco-2 control cells formed round cysts with epithelial cell lining around a fluid-

filled lumen (III; Figure 1). The structures formed by Caco-2 KRAS and Caco-2

BRAF cell lines were larger than those formed by the control cells. In addition,

some structures formed from Caco-2 KRAS cells were cysts like those in Caco-2

controls (III; Figure 2) with a clear lumen while others formed cystic structures

partially filled with cells. Finally, some cysts formed by Caco-2 KRAS cells

showed focal cell piling (Figure 14) or apoptotic cells in the luminal space adjacent

to the cyst wall. In contrast, Caco-2 BRAF cells dominantly formed solid structures

with cells filling up the lumen, resembling solid tumor cell clusters seen in some of

the clinical carcinoma specimens. Growth patterns were consistent in terms of cyst

formation or solid growth from early on and remained consistent after Day 7 in

culture for both control and mutated 3-D cultures (Figure 14).

5.5.2 Organization and polarization of Caco-2 cells in 3-D structures:

influence of KRAS and BRAF mutations

To understand the organization and polarization of cells in 3-D cultures, we imaged

10-day-old fixed cultures and stained the specimens for actin with TRITC-

phalloidin and dapi (III). We noted well-polarized 3-D cysts with regular apical-

basal polarity formed by control Caco-2 cells as indicated by regular apically

polarized localization of actin stress fibers (Figure 14). On the other hand, Caco-2

KRAS cells formed partially filled structures with irregular polarity of actin fibers

or structures with inverted cell polarity, with the apical surface pointing out of the

clusters (III: Figure 2). There was also irregularity in cell polarity present in the

piling cell clusters towards the luminal side. In contrast, Caco-2 BRAF cells

predominantly formed filled structures with irregular cell polarity as indicated by

irregular actin fibers between the cells within the clusters, but no clear inversion of

polarity (III).

72

Since the cystic or solid growth patterns did not entirely relate to the cell

lineage, we quantitated growth patterns in the 10-day-old fixed culture specimens

at 20X magnification. The majority (about 70%) of the structures formed by the

Fig.14. 3-D cultures of conventional and modified Caco-2 cells assessed by different

techniques. A) Nuclear DNA was stained with Hoechst (blue) and Caco-2-control cells

and Caco-2 BRAF cells additionally stained with CellMask TM plasma membrane stain

(orange) to show actin filaments. B) Formalin-fixed cultures were sectioned and stained

with hematoxylin and eosin. In this particular experiment, only a few cyst-like structures

were found in paraffin blocks of Caco-2 BRAF cells. C) Formalin fixed cultures were

sectioned and stained with Ki-67. In this experiment Ki-67 staining show proliferating

cells in the growing cysts, images representing 3-D structures such as Caco-2 control

might appear solid but this could be outermost layer of the cyst. Scale bar represents

20 µm.

control Caco-2 cells were polarized cysts with occasional solid structures (Figure

15). In Caco-2 KRAS 3-D cultures, about half of the structures were solid and the

other half were like those observed in control Caco-2 cysts with occasional cell

73

piling. In contrast, the Caco-2 BRAF cultures consisted predominantly (> 80%) of

solid structures and these differences were found to be statistically significant

(Figure 15).

Fig. 15. Proportions (%) of filled (solid) structures of all multicellular clusters formed by

Caco-2 control cells (Caco-2 GFP cells) and Caco-2 cells transfected with KRAS G12V

or BRAF V600E. Cells were cultured for 10 days. About 145–212 structures/experiments

were quantified from fixed specimens. Mean proportions (SD) based on three

experiments are shown.

5.5.3 Paraffin embedment of 3-D cultures

Preservation of samples in paraffin and possibility for conventional

immunohistochemical assessment would yield advantage over the storage of 3-D

samples and their use in research. Therefore, we adapted a protocol by Pinto et al.

with some changes (Pinto et al., 2011) (III). We analyzed the specimens for

architectural differences with H&E stainings. Reasonably, the structures formed by

the respective cell lines were similar and well preserved, as noted earlier from fixed

or live cell imaging (Figure 14). In addition, the Ki-67 stainings showed functional

utility of immunohistochemical applications of these specimens (Figure 14).

75

6 Discussion

By definition, anoikis is a form of cell death induced when cells lose contact with

the ECM (Gilmore, 2005), and resistance to anoikis is essential for cancer

metastasis (Wang et al., 2018). Anoikis resistance has direct links to poor prognosis,

as shown by prognostic marker reports functionally linked to anoikis resistance (F.

Chen et al., 2018; Dawson et al., 2015). However, so far it has only been possible

to evaluate anoikis resistance by using various anchorage-independent in vitro cell

cultures (Frisch & Francis, 1994). We aimed at identification and understanding of

anoikis resistance as a set of histopathological features by employing

histopathological tools and cell biology with 3-D modeling approaches. In addition,

the roles of some factors, including KRAS and BRAF mutations and anti-apoptotic

protein survivin in anoikis resistance were evaluated. The main hypothesis was that

a carcinoma cell’s ability to resist anoikis manifests as histopathologically

characteristic features. More specifically, we presumed that anoikis resistance

would allow generation of multicellular tumor cell clusters with cells surviving

without contact with the ECM. Such features have not been identified before by

using the anoikis resistance paradigm. However, they might have prognostic and

predictive value as anoikis resistance enables metastasis formation.

For the identification of putative anoikis-resistant (AR) subpopulations in

histopathological specimens, we searched for tumor cell subpopulations without

contact with the ECM, and then evaluated apoptosis rates in cell populations with

and without ECM contact. The absence of increased apoptosis rate in the latter

would imply the presence of anoikis resistance. With such a strategy, we were able

to identify subpopulations of tumor cells fulfilling the set criteria, and these

subpopulations were defined as putative AR populations. Supporting the validity

of the concept, high areal density of such subpopulations associated with adverse

prognosis in CRC.

To gain additional evidence to conceptualize the structural manifestations and

mechanistic information of anoikis resistance, we used genetically modified Caco-

2 cells. Transfection of the cells with mutated KRAS or BRAF oncogenes induced

anoikis resistance as assessed with a classical anchorage-independent survival test

(I), and in 3-D culture, oncogene transfection induced the formation of solid

clusters instead of crypts, suggestive of putative AR structures seen in

histopathology.

These observations potentially define new histopathological criteria with

prognostic value. If validated in large external cohorts, criteria based on a well-

76

defined biological paradigm, anoikis resistance, may provide clinically useful

prognostic and even predictive information. It is possible that such a paradigm can

be adopted to carcinomas of other organs as well.

6.1 Identification of putative AR subpopulations

For identification of putative AR subpopulations, we first spotted carcinoma cells

in contact with mesenchymal areas of the tumors and cells without such contact.

The latter included suprabasal cells of micropapillary structures (MIP) and inner

cells from cribriform and solid structures. The rationale for this was the reported

presence of basement membrane or interstitial collagens in the stromal interface.

Although many studies have evaluated the expression of basement membrane

proteins at the tumor stroma interface in CRC, there was lack of studies

commenting on the presence of intercellular basement membrane protein

expression within the clusters formed by carcinoma cells, such as between the inner

cells of solid or cribriform structures. Therefore, we used immunohistochemistry

to evaluate the presence of ECM proteins in these structures. Stainings for collagen

IV and laminin showed that although the tumor stroma interface encloses these

proteins, they are absent from regions between the cells of the putative AR

structures. We used M30 staining to access cellular apoptosis. M30 antibody binds

to the caspase-cleaved epitope of cytokeratin 18 and is confined to the cytoplasm

of apoptotic cells (Saigusa et al., 2014; West, Courtney, Poullis, & Leicester, 2009).

Our assessment showed decreased apoptosis rates in all putative AR structures as

compared with tumor cells in contact with mesenchymal areas. Additionally, MIPs,

a subpopulation in which we identified features supporting the presence of anoikis

resistance, also presented lower proliferation rates than cells in contact with ECM.

As non-dividing cells do not express Ki-67 antigen, lower PI could possibly be

indicative of features of quiescence in these cells.

In summary, we were able to identify subpopulations of carcinoma cells in

CRC characterized by the absence of ECM contact and showing decreased

apoptosis rate, and therefore representing putative AR cells. One should note that

all these subpopulations are multicellular clusters of cells and always next to tumor

cells in contact with the ECM. Based on our experience, identification of the

putative AR structures in tissue sections was straightforward, and by using image

analysis, it was convenient to define their areal density for quantification and

further analysis. For convenient and accurate quantification of the putative AR

subpopulation in histological sections, it might be possible to generate an

77

automated image analysis system based on recognition of tumor cell islands and

their characteristics, which could provide faster means for histopathological

evaluation.

6.2 Putative AR structures and prognosis of carcinoma

With the aim to extend understanding of the biology and importance of putative AR

subpopulations, we assessed if the quantities of single AR subpopulation types or

total quantity of all AR subpopulation types associated with carcinoma prognosis.

We found that high areal density of both cribriform and solid structures shows an

independent association with poor prognosis. The sum of areal densities of all three

types of AR subpopulations showed the strongest association with poor prognosis.

Such associations are plausible considering the significance of anoikis resistance

in the formation of metastasis, and indirectly support the current concept of

evaluating anoikis resistance by histopathological analysis.

There are no previous prognostic studies in any human carcinoma type based

on the concept of histopathological evaluation of anoikis resistance. In CRC, only

few studies have assessed the prognostic value of histopathological features related

with the current concepts of putative AR subpopulations. Cribriform carcinoma is

a rare type of CRC (Lino-Silva et al., 2015) presenting with unspecified abundance

of cribriform structures and association with poor prognosis. This is consistent with

our finding of association between abundance of cribriform structures and poor

prognosis. Additionally, our results provide evidence that even lesser abundance of

cribriform structures not fulfilling the criteria of cribriform carcinoma may

associate with poor survival.

Solid structures, another putative AR structure type, were associated with high

conventional tumor grade (Bosman et al., 2010), as expected based on conventional

grading criteria. Solid structures as defined in our study may be partially analogous

to poorly differentiated clusters in the recent grading system by Ueno et al. (Ueno

et al., 2012). A marker of poor prognosis, poorly differentiated clusters were

defined to be solid clusters comprised of a minimum five cells and lacking any

glandular phenotype. Accordingly, larger poorly differentiated clusters might

contain cells without contact with ECM and thereby fulfill the current criteria of

putative AR cells in solid clusters. Ueno did not consider any biological

mechanisms for the prognostic effect of poorly differentiated clusters. Based on our

observations, anoikis resistance would be a plausible mechanism.

78

MIPs showed no general association with prognosis. However, in grade 1 and

2 tumors, there was a trend (p = 0.058) for association with poor prognosis, and

interestingly, inclusion of MIP areal density into the total putative AR areal density

strengthened the association with poor prognosis. These findings suggest that MIPs

contribute to the prognostic effect of anoikis resistance. There are no previous

studies about prognostic effect of MIPs in CRC. In another non-invasive carcinoma,

ovarian borderline carcinoma, micropapillary architecture is associated with poor

prognosis (Hannibal et al., 2017).

We found both decreased proliferation rates and apoptosis rates in MIPs as

compared with cells in contact with the ECM, suggesting that these cells are at least

partly quiescent. Anoikis resistance and quiescence are potential cell escape

mechanisms; together, they may interfere with responses to chemotherapy and

radiotherapy (Kyle, Baker, & Minchinton, 2012; Masunaga & Ono, 2002).

Our study indicates that the current concept of histopathological assessment of

anoikis resistance provides new, biologically plausible criteria and a new system

for prognostic assessment in CRC. However, putative AR subpopulations

including MIPs should be explored further to understand their biology. Prognostic

value should be confirmed based on independent case series. In addition, being

functionally distinct subpopulations, the possible roles of the putative AR

populations as predictive markers warrant exploring. Finally, as tumor cell

subpopulations analogous to the current putative AR cells are found the most in

many other carcinoma types, it may be possible to expand the concept to those

malignancies as well.

6.3 Survivin, an anti-apoptotic protein influencing tumorigenesis

Survivin is a classified member of the inhibitor of apoptosis family and possesses

dual function in both proliferation and apoptosis (Altieri, 2008; Mita et al., 2008).

Although survivin is a short-lived protein with T1/2 of 30 minutes, it is essential for

chromosomal segregation and present in all stages of mitosis (Altieri, 2008). The

physiological role of survivin in colorectal mucosa has been a matter of controversy

(ref see II). Our detailed analysis of the subcellular localization of survivin in

normal colorectal mucosa showed heavy nuclear survivin expression in the cells

located in basal parts of the crypts of normal colon mucosa, indicating survivin’s

role in cell proliferation (Gianani et al., 2001; Mita et al., 2008). There is strong

cytoplasmic survivin expression in the cells in the superficial parts of crypt

epithelium, a location in the colon where cells mature, and eventually undergo

79

apoptosis and shed into lumen. This observation could indicate a pro-apoptotic role

of survivin contrary to the cyto-protective role decoded when survivin is localized

in the cytoplasm (Stauber, Mann, & Knauer, 2007). In summary, our results suggest

that in normal colorectal epithelium, survivin plays a role in both proliferation and

apoptosis depending on its subcellular location in cells.

Based on the suggested anti-apoptotic role of survivin (Mita et al., 2008) we

were interested to analyze its role in MIPs, a type of putative AR subpopulation.

Both nuclear and cytoplasmic expression were lower in the cells of MIPs as

compared with other tumor cells. Survivin expression, however, did not correlate

with apoptosis or proliferation rates in the tumor subpopulation types, and

accordingly, the functional role of survivin in anoikis resistance remains

speculative and warrants additional studies.

In line with earlier reports (Parfitt & Driman, 2007), we observed a bottom-

heavy staining pattern of nuclear survivin in serrated polyps whereas in

conventional adenomas, survivin expression was most abundant in the upper half

of the lesion. This distribution correlates with cell proliferation in these polyp types

and is consistent with the role of survivin in cell proliferation (Altieri, 2008). The

characteristic difference in the staining patterns suggests that survivin staining

might provide some, albeit minor, help in the classification of polyps.

The overall expression pattern of survivin was in line with previous findings

in CRC, illustrating higher cytoplasmic survivin and lower nuclear survivin in

cancer than in normal colon mucosa or benign polyps (Fang YJ et al., 2009; Gianani

et al., 2001). An association between high expression of survivin and adverse

prognosis has been previously reported (Fang YJ et al., 2009). Our finding of

absence of association with prognosis could be related to the limited number of

patients and selection bias related to emphasis on serrated carcinomas.

6.4 3-D in vitro model with mutated KRAS and BRAF oncogenes

transfected Caco-2 cells to model anoikis resistance

3-D culture has been developed for many decades and it holds advantages over

conventional 2-D cultures. 3-D culturing of normal or cancer cells in basement

membrane extract (matrigel), rat-tail collagen I or synthetic hydrogels has been

utilized to study the pathogenesis of several diseases. Moreover, this system allows

assessment of biological and functional properties of cells in an environment close

to that of actual in vivo situation.

80

We aimed to create an in vitro model for the putative AR subpopulations seen

in actual carcinomas to get additional evidence of association between anoikis

resistance and the histopathological features identified to represent it, including

MIPs, solid and cribriform structures.

3-D culture was prepared by plating Control-Caco2 cells and Caco-2 cells with

KRAS or BRAF mutations in a mixture of Matrigel and collagen as described

previously (Debnath, Muthuswamy, & Brugge, 2003; Magudia et al., 2012).

Monitoring and imaging of 3-D live cultures for 10 days showed that by Day 7, all

cell lineages had grown to constant patterns of cysts or filled structures, with no

visible changes in the distribution of these patterns at later time points. At Day 10,

well-polarized cysts with a clear lumen were formed by Caco-2 control cells as

expected (Magudia et al., 2012; Makrodouli et al., 2011). However, Caco-2 cells

with mutant KRAS formed filled structures (partial or complete solid cysts) with

inverted polarity shown by actin staining (Hall, 2009) at the apical pole outside of

the cell clusters. In addition, Caco-2 KRAS cells formed cysts with some cell piling

at the luminal side with disorganized cytoskeletal organization marked by

fragments of actin strands. Quite recently, inverted polarity was reported to be the

mechanism enabling survival and dissemination of cancer cell clusters during

peritoneal metastasis especially in serrated CRC (Zajac et al., 2018). Therefore,

solid structures formed by Caco-2 KRAS cells with inverted polarity might serve

as a useful model to provide additional mechanistic knowledge of inverted polarity

and anoikis resistance. Caco-2 BRAF cells, on the other hand, predominantly

formed solid structures with irregular polarity in the inner cells. The formation of

solid structures or intraluminal piling by Caco-2 cells with KRAS and BRAF

mutations is structurally reminiscent of the putative AR subpopulations seen in

histopathological sections of CRC (I, II, III, Figure 16), mostly with solid islands

and micropapillary structures, respectively. Emergence of these structures in vitro

along with emergence of anoikis resistance in adhesion-free cultures supports the

concept that the formation of these structures in vitro is a manifestation of anoikis

resistance. Further supporting this concept is the finding that the inner cells in the

cell clusters in vitro clearly survive without contact with the ECM. However, to

confirm the occurrence of anoikis resistance, detailed comparisons of inner and

outer cells of these multicellular cluster are needed to see whether there is any

suppression of apoptosis in the inner cells. In addition, functional studies would be

needed to identify the anti-apoptotic mechanisms involved in survival in such

organized multicellular clusters.

81

Fig. 16. Comparison of histopathology of CRC cases (left column; Hematoxylin & eosin

staining) and confocal images of 3-D cultures of different Caco-2 cell lines (right column;

DNA stained with dapi; actin with TRITC-phalloidin) to illustrate similarities between

putative non-AR (1st row) and AR (2nd and 3rd row) subpopulations in CRC cases, and

corresponding non-AR (1st row) and AR (2nd and 3rd rows) cells in 3-D in vitro cultures.

To provide possibilities for a more comprehensive comparison of in vitro 3-D

structures and actual tumors, we adapted a method by Pinto et al. by transferring 3-

D cultures into an agarose sandwich followed by embedding in paraffin. The

method not only allows preservation of experimental samples for long duration but

also provides an opportunity to study markers related to anoikis resistance and

progression by using conventional immunohistochemistry (Figure 14).

82

6.5 KRAS and BRAF mutations induce anoikis resistance in

adhesion free culture in Caco-2 cells

There are several methods for in vitro assessment of anoikis resistance. These

methods include anti-adhesion tests such as soft agar hanging drop cultures, utilized

to study colony formation, whereas cultures grown in suspension alone or with

additives such as poly-L lysine, collagen or other cell adhesion molecules are used

to study anchorage-independent mechanisms. The most frequently applied tool for

suspension cultures involves using commercially manufactured low attachment

dishes or cell culture dishes coated with poly HEMA-ethanol solution providing an

anchorage-independent surface. We utilized this cost-effective and robust method

to study anoikis resistance in suspension environment.

We used Caco-2 cells, a well-characterized colon cancer cell line that does not

harbor KRAS or BRAF oncogene mutations, which are commonly spotted in CRC.

Additional Caco-2 cells form cysts with a regular lumen and polarized cells around

in 3-D matrigel, resembling glandular structures of normal intestinal mucosa

(Magudia et al., 2012; Makrodouli et al., 2011). Caco-2 cells are adherent cell

lines and not resistant to anoikis (Kozlova, Morozevich, Chubukina, & Berman,

2001). To modify the cells we selected KRAS G12V and BRAF V600E mutations,

since mutations in KRAS G12V and BRAF V 600E are common in CRC and hold

prognostic significance in CRC (Richman et al., 2009; Taieb et al., 2016). We

modified Caco-2 cells with these oncogenic mutations. The Caco-2 cells with

KRAS or BRAF mutations have been shown to alter polarized cysts formation.

(Magudia et al., 2012; Makrodouli et al., 2011).

We noted a significant increase in the number of anoikis-resistant (Annexin

negative) cells in Caco-2 KRAS and Caco-2 BRAF cells compared to Caco-2

control cells in serum-free suspension cultures. KRAS G12V-mutated Caco-2 cells

showed higher resistance to anoikis than BRAF V600E-mutated Caco-2 cells; this

difference was more evident at 24 h and less pronounced at 48 h, possibly due to

stress induced by the culture conditions (i.e. cells in suspension). These

observations indicate, for the first time, that KRAS G12V and BRAF V600E induce

anoikis resistance in Caco-2 cells. In support of this observation, earlier literature

indicated the presence of anoikis resistance in all reported (n = 4) intestinal

carcinoma cell lines with either inherent KRAS G12V or BRAF V600E mutation

(III: Table 1).

Since Caco-2 cells harbor APC mutation (De Bosscher, Hill, & Nicolas, 2004;

Magudia et al., 2012), it is important to consider whether APC mutation contributes

83

to induction of anoikis resistance by KRAS or BRAF mutations. Extension of our

literature survey of CRC cell lines with inherent mutation of KRAS or BRAF and

either anoikis resistance or in vitro growth as solid structures in 3-D (III; Table 1)

indicated that all these cell lines except for HCT116 have APC mutation (Heinen,

Richardson, White, & Groden, 1995; Ilyas, Tomlinson, Rowan, Pignatelli, &

Bodmer, 1997). The latter reports provide experimental evidence that the combined

effect of APC mutation together with KRAS or BRAF may contribute to anoikis

resistance. However, we are not aware of studies confirming, for example by RNA

interference, that APC mutation has any essential role in anoikis resistance, and this

warrants additional studies. Biologically, combined effects of two mutations in the

generation of anoikis resistance seem plausible, as there is previous experimental

evidence indicating the existence of regulatory links between WNT and ERK

signaling (Jeong, Ro, & Choi, 2018). Furthermore, indirectly supporting the

concept, animal model studies have shown cumulative effects of APC and KRAS

on metastasis development (Sakai et al., 2018). Finally, in our clinical series, we

did not find clear associations between KRAS or BRAF mutation and the extent of

putative anoikis resistant subpopulations, suggesting that mutations additional to

KRAS or BRAF might be involved in anoikis resistance in CRC. Accordingly, it

would be interesting to assess the role of the combined effects of APC and KRAS

or BRAF mutations on the formation of putative anoikis resistant populations in a

clinical series of CRC.

To get some insight into the mechanisms behind anoikis resistance related to

transfection of mutated KRAS or BRAF, we studied Bim protein expression in our

three Caco-2 cell lineages surviving in adhesion-free conditions. Bim is a well-

characterized pro-apoptotic protein, which blocks anti-apoptotic proteins and

induces caspase-mediated apoptosis (Boisvert-Adamo & Aplin, 2008; Goldstein et

al., 2009). We found a non-significant trend for increased expression of Bim in the

cells with mutant oncogenes, suggesting an anti-apoptotic role for this protein.

Although the observation is in disagreement with the well-addressed pro-apoptotic

role of Bim protein, a prosurvival role of Bim has also been identified (Gogada et

al., 2013). Obviously, dissection of the mechanisms of anoikis resistance induced

by oncogene transfections requires additional research.

84

6.6 Anoikis-resistant cells and putative AR cells present in

carcinomas show features of dormancy or quiescence

Among several factors that may aid or induce anoikis resistance in cancer cells,

tumor quiescence stands strong. Cellular quiescence is represented when a cell is

detected at G0 phase. It can be reversible (quiescent stage) or irreversible

(senescent and differentiated, non-dividing). This phenomenon is commonly

explored in stem cell research and cancer research. Quiescence can be detected in

tissues by Ki-67 immunohistochemical staining, in cells by estimation of cell cycle

phase with Propidium iodide FACS assay (J. H. Kim et al., 2015), and with label

retention experiments using bromodeoxyuridine (BrdU) (Glauche et al., 2009).

We found that the majority of Annexin-negative (anoikis-resistant) cells from

Caco-2 cells transfected with mutant KRAS or BRAF oncogenes in suspension were

arrested at G0/G1 phase compared to control Caco-2 cells. This difference was

present in the 24 h suspension cultures and became stronger in the 48 h cultures. In

histopathological studies (I, II) we found that cells in the MIP structures showed

lower Ki-67 protein expression as compared to tumor epithelium in contact with

ECM, possibly indicating features of quiescence. Importantly, quiescent cancer

stem cells (CSCs) often present higher chemo resistance and tumor formation

ability than non-quiescent CSCs (Kojima et al., 2017), which is why their efficient

identification and elimination would be beneficial (Becker, 2018).

Further studies are needed to assess whether MIPs contain quiescent cells and

whether Caco-2 cells with BRAF or KRAS mutation serve as a model for quiescence.

6.7 A model untangling the fates of AR subpopulations in cancer

formation

Identification and assessment of cancer subpopulations with anoikis resistance

from human cancer tissues might provide vital information improving our

understanding of the mechanism for acquisition of this feature and its link to

survival and progression in cancer. We present here a theoretical model proposing

different fates AR cells might encounter (Figure 17). With the current knowledge

deciphered in this thesis, we have displayed possible routes of progression by any

anoikis resistant tumor cell population. MIPs were used here as example, since

they seem to present both anoikis resistance and cellular quiescence. There could

be three prospective fates they might run into. 1) Cells might remain quiescent at

the primary location for decades without being detected or removed, as has been

85

reported for cells of breast cancer. 2) Alternately, in response to a stimulus, these

cells might proliferate extensively towards the luminal space and, complemented

by decreased apoptosis, fill up the gland, resulting in formation of partially

(cribriform) or completely filled (solid) structures, contributing to poorly

differentiated cancer with poor clinical outcome. Accordingly, speculatively, MIPs

could be the origin of formation of other AR structures. 3) It is possible that

utilizing their anoikis resistance, cells from MIPs and other anoikis resistant

subpopulations might disseminate and travel through the blood and lymphatic

vessel system, resulting in the formation of metastases.

Although such sequences are still speculative, if real, any of these fates may

lead to poor clinical outcome and should therefore be investigated to confirm or

exclude any such possibility. An elaborate understanding of such cells and/or

structures, including a deeper understanding of the molecular mechanism by which

KRAS or BRAF mutations contribute to cancer metastasis via anoikis resistance and

associated aberrations, would be helpful in improving patient survival. The ways

for such improvement might include development of reliable methods for detection

of features indicating anoikis resistance and the concomitant risks as well as

development of drugs and other treatment modalities effective for elimination of

AR cells.

6.8 Future perspectives

In Study II, we identified and assessed putative AR structures in colorectal cancer

tissues with histopathology and immunohistochemical analysis. However, the

study was based on a single cohort. Therefore, analysis of larger independent

patient material is needed for confirmation of the prognostic value.

Our study was performed using samples from the tumor bulk region, since the

tumor front contains carcinoma cell buds in almost 40% of cases and budding is a

strong prognostic factor (Lugli et al., 2017b). By definition, all cells in the buds are

in contact with mesenchymal areas of the tumor (Lugli et al., 2017b) and therefore

do not qualify for any AR population. However, it would be interesting to analyze

the front region for putative AR subpopulations.

We used digital image analysis for the assessment of areal proportions of

putative AR populations and for Ki-67 expression, and it showed good interrater

reproducibility in the latter. However, since the program required manual

annotation of different cell populations, there is clearly a need for development of

software capable of recognizing different tumor cell subpopulations.

86

Fig. 17. Schematic representation of the fates of anoikis-resistant cells in disease

formation or progression.

We reported (study I, II) that cells in micropapillary structures present distinct

biology with anoikis resistance property and features of quiescence. In cell-cycle

stage assessments of anoikis-resistant colon cancer cells transfected with mutant

KRAS or BRAF we found that the cells were arrested at G0/G1 phase (III),

indicative of quiescence. Based on these observations, the presence of quiescence

should be confirmed by separation of G0 cells from G1 cells. The significance of

putative quiescent populations in clinical tumor samples should be studied with

87

larger patient material to assess their prognostic value, the effect on treatment

responses, and finally, to optimize treatment regimens. Similarly, there is a need

for characterization of the oncological significance of other AR subpopulations as

well by using both retrospective analyses focusing on the relationship between

treatment response and occurrence of AR subpopulations and by using in vitro

models.

We showed (III) that transfection with mutant KRAS or BRAF induces anoikis

resistance feature in colon cancer cells in 3-D cultures and in anti-adhesion tests.

Although we did not find any association of KRAS or BRAF mutations with the

occurrence of putative AR subpopulations in CRC, histopathological assessment of

AR subpopulations in larger material and assessment of APC mutation in addition

to KRAS and BRAF might enable evaluation of any such association if also present

in actual carcinoma.

Finally, regardless of the validity and significance of the current anoikis

resistance paradigm in terms of the putative AR subpopulations, it is very likely

that these subpopulations in CRC tissue and in the 3-D model represent a

biologically different cell population as compared to cells in contact with the ECM.

The patterns of biological differences remain speculative at present, and likely

involve mechanisms analogous for any intra-tumoral heterogeneity, including

genetic and epigenetic as well as non-hereditary functional differences (Stanta &

Bonin, 2018). We are not aware of studies in CRC combining any feature of

heterogeneity of the microscopic structure of the tumor and mutational, epigenetic

or mRNA expression related heterogeneity. Furthermore, it is likely that there are

differences in the overall biology of tumors with and without these subpopulations.

More importantly, the presence of these subpopulations seems to have prognostic

value. Further characterization of structures will help to understand the underlying

mechanism of disease progression and their prognostic effects and may be

informative in designing optimal treatment.

89

7 Conclusions

CRC is a common cause of death in most western countries and resistance to

anoikis is considered one key feature enabling the formation of metastasis in CRC

and in other carcinomas. We aimed to identify putative AR carcinoma cell

subpopulations in CRC and to assess their prognostic role. Additionally, we

searched support for the concept by using a 3-D in vitro cell culture model and anti-

adhesion tests.

I We identified putative AR subpopulations in micropapillary, cribriform and

solid structures based on the absence of basement membrane proteins and

lower apoptosis indexes in all three subpopulations. High areal density sum of

these subpopulations was an independent prognostic factor indicating

shortened cancer-specific survival in CRC. (I, II).

II We studied the mechanisms behind survival of putative AR cells and found

decreased survivin expression by immunohistochemistry along with lower

proliferation and apoptosis rates in MIPs, a putative AR subpopulation in CRC.

KRAS and BRAF mutations induced acquisition of anoikis resistance in Caco-

2 colon cancer cells in vitro, while in colon carcinoma tissues, no clear

association was detected between these mutations and the occurrence of the

putative AR subpopulations (I, II, III).

III We generated a 3-D in vitro model by transfecting Caco-2 cells with KRAS or

BRAF mutations. Along with generation of anoikis resistance in adhesion free

culture, these oncogenes induced the formation of solid multicellular clusters

or cystic structures with focal piling of cells in 3-D cultures, both patterns

indicating survival of cells without ECM contact, and accordingly, anoikis

resistance. Histopathologically, these structures showed features reminiscent

of solid and micropapillary structures in CRC, respectively. These findings

suggest that solid growth and micropapillary-like structures are related with

anoikis resistance both in vitro and in actual CRC cases in vivo (III).

91

List of references

Altieri, D. C. (2008). Survivin, cancer networks and pathway-directed drug discovery. Nature Reviews.Cancer, 8(1), 61-70. doi:nrc2293 [pii]

Anabazhagan, A. N., Chatterjee, I., Priyamvada, S., Kumar, A., Tyagi, S., Saksena, S., . . . Gill, R. K. (2017). Methods to study epithelial transport protein function and expression in native intestine and caco-2 cells grown in 3D. Journal of Visualized Experiments : JoVE, (121). doi(121), 10.3791/55304. doi:10.3791/55304 [doi]

Anderson, J. M., & Van Itallie, C. M. (2009). Physiology and function of the tight junction. Cold Spring Harbor Perspectives in Biology, 1(2), a002584. doi:10.1101/cshperspect.a002584 [doi]

Andrew, D. J., & Ewald, A. J. (2010). Morphogenesis of epithelial tubes: Insights into tube formation, elongation, and elaboration. Developmental Biology, 341(1), 34-55. doi:10.1016/j.ydbio.2009.09.024 [doi]

Armaghany, T., Wilson, J. D., Chu, Q., & Mills, G. (2012). Genetic alterations in colorectal cancer. Gastrointestinal Cancer Research : GCR, 5(1), 19-27.

Attwell, S., Roskelley, C., & Dedhar, S. (2000). The integrin-linked kinase (ILK) suppresses anoikis. Oncogene, 19(33), 3811-3815. doi:10.1038/sj.onc.1203711 [doi]

Bankhead, P., Loughrey, M. B., Fernandez, J. A., Dombrowski, Y., McArt, D. G., Dunne, P. D., . . . Hamilton, P. W. (2017). QuPath: Open source software for digital pathology image analysis. Scientific Reports, 7(1), 16878-017-17204-5. doi:10.1038/s41598-017-17204-5 [doi]

Barker, N., Ridgway, R. A., van Es, J. H., van de Wetering, M., Begthel, H., van den Born, M., . . . Clevers, H. (2009). Crypt stem cells as the cells-of-origin of intestinal cancer. Nature, 457(7229), 608-611. doi:10.1038/nature07602 [doi]

Barresi, V., Reggiani Bonetti, L., Ieni, A., Caruso, R. A., & Tuccari, G. (2015). Histological grading in colorectal cancer: New insights and perspectives. Histology and Histopathology, 30(9), 1059-1067. doi:10.14670/HH-11-633 [doi]

Barresi, V., Reggiani Bonetti, L., Vitarelli, E., Di Gregorio, C., Ponz de Leon, M., & Barresi, G. (2012). Immunohistochemical assessment of lymphovascular invasion in stage I colorectal carcinoma: Prognostic relevance and correlation with nodal micrometastases. The American Journal of Surgical Pathology, 36(1), 66-72. doi:10.1097/PAS.0b013e31822d3008 [doi]

Bates, R. C., & Mercurio, A. M. (2005). The epithelial-mesenchymal transition (EMT) and colorectal cancer progression. Cancer Biology & Therapy, 4(4), 365-370. doi:1655 [pii]

Beaugerie, L., & Itzkowitz, S. H. (2015). Cancers complicating inflammatory bowel disease. The New England Journal of Medicine, 372(15), 1441-1452. doi:10.1056/ NEJMra1403718 [doi]

Becker, W. (2018). A wake-up call to quiescent cancer cells - potential use of DYRK1B inhibitors in cancer therapy. The FEBS Journal, 285(7), 1203-1211. doi:10.1111/ febs.14347 [doi]

92

Bian, H. B., Pan, X., Yang, J. S., Wang, Z. X., & De, W. (2011). Upregulation of microRNA-451 increases cisplatin sensitivity of non-small cell lung cancer cell line (A549). Journal of Experimental & Clinical Cancer Research : CR, 30, 20-9966-30-20. doi:10.1186/1756-9966-30-20 [doi]

Boisvert-Adamo, K., & Aplin, A. E. (2008). Mutant B-RAF mediates resistance to anoikis via bad and bim. Oncogene, 27(23), 3301-3312. doi:10.1038/sj.onc.1211003 [doi]

Bolos, V., Peinado, H., Perez-Moreno, M. A., Fraga, M. F., Esteller, M., & Cano, A. (2003). The transcription factor slug represses E-cadherin expression and induces epithelial to mesenchymal transitions: A comparison with snail and E47 repressors. Journal of Cell Science, 116(Pt 3), 499-511. doi:10.1242/jcs.00224 [doi]

Borowicz, S., Van Scoyk, M., Avasarala, S., Karuppusamy Rathinam, M. K., Tauler, J., Bikkavilli, R. K., & Winn, R. A. (2014). The soft agar colony formation assay. Journal of Visualized Experiments : JoVE, (92):e51998. doi(92), e51998. doi:10.3791/51998 [doi]

Borowsky, J., Dumenil, T., Bettington, M., Pearson, S. A., Bond, C., Fennell, L., . . . Whitehall, V. (2018). The role of APC in WNT pathway activation in serrated neoplasia. Modern Pathology : An Official Journal of the United States and Canadian Academy of Pathology, Inc, 31(3), 495-504. doi:10.1038/modpathol.2017.150 [doi]

Bosman, F. T., Carneiro, F., Hruban, R. H., & Theise, N. D. (2010). WHO classification of tumours of the digestive system (Fourth ed.) WHO press.

Bozzo, C., Sabbatini, M., Tiberio, R., Piffanelli, V., Santoro, C., & Cannas, M. (2006). Activation of caspase-8 triggers anoikis in human neuroblastoma cells. Neuroscience Research, 56(2), 145-153. doi:S0168-0102(06)00169-6 [pii]

Buchert, M., Papin, M., Bonnans, C., Darido, C., Raye, W. S., Garambois, V., . . . Hollande, F. (2010). Symplekin promotes tumorigenicity by up-regulating claudin-2 expression. Proceedings of the National Academy of Sciences of the United States of America, 107(6), 2628-2633. doi:10.1073/pnas.0903747107 [doi]

Bunek, J., Kamarajan, P., & Kapila, Y. L. (2011). Anoikis mediators in oral squamous cell carcinoma. Oral Diseases, 17(4), 355-361. doi:10.1111/j.1601-0825.2010.01763.x [doi]

Carroll, R., & Zhao, S. (2019). Trends in colorectal cancer incidence and survival in iowa SEER data: The timing of it all. Clinical Colorectal Cancer, 18(2), e261-e274. doi:S1533-0028(18)30448-1 [pii]

Chen, F., Zhang, L., Wu, J., Huo, F., Ren, X., Zheng, J., & Pei, D. (2018). HCRP-1 regulates EGFR-AKT-BIM-mediated anoikis resistance and serves as a prognostic marker in human colon cancer. Cell Death & Disease, 9(12), 1176-018-1217-2. doi:10.1038/s41419-018-1217-2 [doi]

Chen, J. L., David, J., Cook-Spaeth, D., Casey, S., Cohen, D., Selvendiran, K., . . . Hays, J. L. (2017). Autophagy induction results in enhanced anoikis resistance in models of peritoneal disease. Molecular Cancer Research : MCR, 15(1), 26-34. doi:10.1158/ 1541-7786.MCR-16-0200-T [doi]

Cheng, L. K., O'Grady, G., Du, P., Egbuji, J. U., Windsor, J. A., & Pullan, A. J. (2010). Gastrointestinal system. Wiley Interdisciplinary Reviews.Systems Biology and Medicine, 2(1), 65-79. doi:10.1002/wsbm.19 [doi]

93

Clark, G. M. (2008). Prognostic factors versus predictive factors: Examples from a clinical trial of erlotinib. Molecular Oncology, 1(4), 406-412. doi:10.1016 / j.molonc.2007.12.001 [doi]

Coradini, D., Casarsa, C., & Oriana, S. (2011). Epithelial cell polarity and tumorigenesis: New perspectives for cancer detection and treatment. Acta Pharmacologica Sinica, 32(5), 552-564. doi:10.1038/aps.2011.20 [doi]

DA Silva, F. C., Wernhoff, P., Dominguez-Barrera, C., & Dominguez-Valentin, M. (2016). Update on hereditary colorectal cancer. Anticancer Research, 36(9), 4399-4405. doi:36/9/4399 [pii]

Dawson, H., Grundmann, S., Koelzer, V. H., Galvan, J. A., Kirsch, R., Karamitopoulou, E., . . . Zlobec, I. (2015). Tyrosine kinase receptor B (TrkB) expression in colorectal cancers highlights anoikis resistance as a survival mechanism of tumour budding cells. Histopathology, 66(5), 715-725. doi:10.1111/his.12603 [doi]

Dawson, H., Koelzer, V. H., Karamitopoulou, E., Economou, M., Hammer, C., Muller, D. E., . . . Zlobec, I. (2014). The apoptotic and proliferation rate of tumour budding cells in colorectal cancer outlines a heterogeneous population of cells with various impacts on clinical outcome. Histopathology, 64(4), 577-584. doi:10.1111/his.12294 [doi]

De Bosscher, K., Hill, C. S., & Nicolas, F. J. (2004). Molecular and functional consequences of Smad4 C-terminal missense mutations in colorectal tumour cells. The Biochemical Journal, 379(Pt 1), 209-216. doi:10.1042/BJ20031886 [doi]

De Divitiis, C., Nasti, G., Montano, M., Fisichella, R., Iaffaioli, R. V., & Berretta, M. (2014). Prognostic and predictive response factors in colorectal cancer patients: Between hope and reality. World Journal of Gastroenterology, 20(41), 15049-15059. doi:10.3748/wjg.v20.i41.15049 [doi]

De Palma, F. D. E., D'Argenio, V., Pol, J., Kroemer, G., Maiuri, M. C., & Salvatore, F. (2019). The molecular hallmarks of the serrated pathway in colorectal cancer. Cancers, 11(7), 10.3390/cancers11071017. doi:E1017 [pii]

De Rosa, M., Pace, U., Rega, D., Costabile, V., Duraturo, F., Izzo, P., & Delrio, P. (2015). Genetics, diagnosis and management of colorectal cancer (review). Oncology Reports, 34(3), 1087-1096. doi:10.3892/or.2015.4108 [doi]

Deb, S., Xu, H., Tuynman, J., George, J., Yan, Y., Li, J., . . . Fox, S. B. (2014). RAD21 cohesin overexpression is a prognostic and predictive marker exacerbating poor prognosis in KRAS mutant colorectal carcinomas. British Journal of Cancer, 110(6), 1606-1613. doi:10.1038/bjc.2014.31 [doi]

Debnath, J. (2010). p66(shc) and ras: Controlling anoikis from the inside-out. Oncogene, 29(41), 5556-5558. doi:10.1038/onc.2010.347 [doi]

Debnath, J., Mills, K. R., Collins, N. L., Reginato, M. J., Muthuswamy, S. K., & Brugge, J. S. (2002). The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini. Cell, 111(1), 29-40. doi:S0092867402010012 [pii]

94

Debnath, J., Muthuswamy, S. K., & Brugge, J. S. (2003). Morphogenesis and oncogenesis of MCF-10A mammary epithelial acini grown in three-dimensional basement membrane cultures. Methods (San Diego, Calif.), 30(3), 256-268. doi:S104620230300032X [pii]

Delva, E., & Kowalczyk, A. P. (2009). Regulation of cadherin trafficking. Traffic (Copenhagen, Denmark), 10(3), 259-267. doi:10.1111/j.1600-0854.2008.00862.x [doi]

Derwinger, K., Kodeda, K., Bexe-Lindskog, E., & Taflin, H. (2010). Tumour differentiation grade is associated with TNM staging and the risk of node metastasis in colorectal cancer. Acta Oncologica (Stockholm, Sweden), 49(1), 57-62. doi:10.3109/ 02841860903334411 [doi]

Desgrosellier, J. S., & Cheresh, D. A. (2010). Integrins in cancer: Biological implications and therapeutic opportunities. Nature Reviews.Cancer, 10(1), 9-22. doi:10.1038/ nrc2748 [doi]

Dosh, R. H., Essa, A., Jordan-Mahy, N., Sammon, C., & Le Maitre, C. L. (2017). Use of hydrogel scaffolds to develop an in vitro 3D culture model of human intestinal epithelium. Acta Biomaterialia, 62, 128-143. doi:S1742-7061(17)30549-4 [pii]

Dosh, R. H., Jordan-Mahy, N., Sammon, C., & Le Maitre, C. L. (2019). Long-term in vitro 3D hydrogel co-culture model of inflammatory bowel disease. Scientific Reports, 9(1), 1812-019-38524-8. doi:10.1038/s41598-019-38524-8 [doi]

Du, S., Miao, J., Zhu, Z., Xu, E., Shi, L., Ai, S., . . . Xia, X. (2018). NADPH oxidase 4 regulates anoikis resistance of gastric cancer cells through the generation of reactive oxygen species and the induction of EGFR. Cell Death & Disease, 9(10), 948-018-0953-7. doi:10.1038/s41419-018-0953-7 [doi]

Elmore, S. (2007). Apoptosis: A review of programmed cell death. Toxicologic Pathology, 35(4), 495-516. doi:779478428 [pii]

Eriksen, A. C., Sorensen, F. B., Lindebjerg, J., Hager, H., dePont Christensen, R., Kjaer-Frifeldt, S., & Hansen, T. F. (2018). The prognostic value of tumor-infiltrating lymphocytes in stage II colon cancer. A nationwide population-based study. Translational Oncology, 11(4), 979-987. doi:S1936-5233(18)30064-0 [pii]

Fang YJ, Lu ZH, Wang GQ, Pan ZZ, Zhou ZW, Yun JP, . . . Wan DS. (2009). Elevated expressions of MMP7, TROP2, and survivin are associated with survival, disease recurrence, and liver metastasis of colon cancer. International Journal of Colorectal Disease, 24(8), 875-884. doi:http://dx.doi.org/10.1007/s00384-009-0725-z

Fatehullah, A., Appleton, P. L., & Nathke, I. S. (2013). Cell and tissue polarity in the intestinal tract during tumourigenesis: Cells still know the right way up, but tissue organization is lost. Philosophical Transactions of the Royal Society of London.Series B, Biological Sciences, 368(1629), 20130014. doi:10.1098/rstb.2013.0014 [doi]

Fearon, E. R., & Vogelstein, B. (1990). A genetic model for colorectal tumorigenesis. Cell, 61(5), 759-767. doi:0092-8674(90)90186-I [pii]

Fessler, E., Drost, J., van Hooff, S. R., Linnekamp, J. F., Wang, X., Jansen, M., . . . Medema, J. P. (2016). TGFbeta signaling directs serrated adenomas to the mesenchymal colorectal cancer subtype. EMBO Molecular Medicine, 8(7), 745-760. doi:10.15252/emmm.201606184 [doi]

95

Fleming, M., Ravula, S., Tatishchev, S. F., & Wang, H. L. (2012). Colorectal carcinoma: Pathologic aspects. Journal of Gastrointestinal Oncology, 3(3), 153-173. doi:10.3978/ j.issn.2078-6891.2012.030 [doi]

Fouquet, S., Lugo-Martinez, V. H., Faussat, A. M., Renaud, F., Cardot, P., Chambaz, J., . . . Thenet, S. (2004). Early loss of E-cadherin from cell-cell contacts is involved in the onset of anoikis in enterocytes. The Journal of Biological Chemistry, 279(41), 43061-43069. doi:10.1074/jbc.M405095200 [doi]

Frankel, A., Rosen, K., Filmus, J., & Kerbel, R. S. (2001). Induction of anoikis and suppression of human ovarian tumor growth in vivo by down-regulation of bcl-X(L). Cancer Research, 61(12), 4837-4841.

Frisch, S. M., & Francis, H. (1994). Disruption of epithelial cell-matrix interactions induces apoptosis. The Journal of Cell Biology, 124(4), 619-626.

Frisch, S. M., Schaller, M., & Cieply, B. (2013). Mechanisms that link the oncogenic epithelial-mesenchymal transition to suppression of anoikis. Journal of Cell Science, 126(Pt 1), 21-29. doi:10.1242/jcs.120907 [doi]

Fung, K. Y., Tabor, B., Buckley, M. J., Priebe, I. K., Purins, L., Pompeia, C., . . . Cosgrove, L. J. (2015). Blood-based protein biomarker panel for the detection of colorectal cancer. PloS One, 10(3), e0120425. doi:10.1371/journal.pone.0120425 [doi]

Garcia, M. A., Nelson, W. J., & Chavez, N. (2018). Cell-cell junctions organize structural and signaling networks. Cold Spring Harbor Perspectives in Biology, 10(4), 10.1101/cshperspect.a029181. doi:a029181 [pii]

Gianani, R., Jarboe, E., Orlicky, D., Frost, M., Bobak, J., Lehner, R., & Shroyer, K. R. (2001). Expression of survivin in normal, hyperplastic, and neoplastic colonic mucosa. Human Pathology, 32(1), 119-125. doi:S0046-8177(01)61387-X [pii]

Gilmore, A. P. (2005). Anoikis. Cell Death and Differentiation, 12 Suppl 2, 1473-1477. doi:4401723 [pii]

Glauche, I., Moore, K., Thielecke, L., Horn, K., Loeffler, M., & Roeder, I. (2009). Stem cell proliferation and quiescence--two sides of the same coin. PLoS Computational Biology, 5(7), e1000447. doi:10.1371/journal.pcbi.1000447 [doi]

Gogada, R., Yadav, N., Liu, J., Tang, S., Zhang, D., Schneider, A., . . . Chandra, D. (2013). Bim, a proapoptotic protein, up-regulated via transcription factor E2F1-dependent mechanism, functions as a prosurvival molecule in cancer. The Journal of Biological Chemistry, 288(1), 368-381. doi:10.1074/jbc.M112.386102 [doi]

Goldstein, N. B., Johannes, W. U., Gadeliya, A. V., Green, M. R., Fujita, M., Norris, D. A., & Shellman, Y. G. (2009). Active N-ras and B-raf inhibit anoikis by downregulating bim expression in melanocytic cells. The Journal of Investigative Dermatology, 129(2), 432-437. doi:10.1038/jid.2008.227 [doi]

Grady, W. M., & Carethers, J. M. (2008). Genomic and epigenetic instability in colorectal cancer pathogenesis. Gastroenterology, 135(4), 1079-1099. doi:10.1053/ j.gastro.2008.07.076 [doi]

96

Green, K. J., & Jones, J. C. (1996). Desmosomes and hemidesmosomes: Structure and function of molecular components. FASEB Journal : Official Publication of the Federation of American Societies for Experimental Biology, 10(8), 871-881. doi:10.1096/fasebj.10.8.8666164 [doi]

Greene, F. L., & Sobin, L. H. (2002). The TNM system: Our language for cancer care. Journal of Surgical Oncology, 80(3), 119-120. doi:10.1002/jso.10114 [doi]

Greenwood-Van Meerveld, B., Johnson, A. C., & Grundy, D. (2017). Gastrointestinal physiology and function. Handbook of Experimental Pharmacology, 239, 1-16. doi:10.1007/164_2016_118 [doi]

Groff, R. J., Nash, R., & Ahnen, D. J. (2008). Significance of serrated polyps of the colon. Current Gastroenterology Reports, 10(5), 490-498.

Guadamillas, M. C., Cerezo, A., & Del Pozo, M. A. (2011). Overcoming anoikis--pathways to anchorage-independent growth in cancer. Journal of Cell Science, 124(Pt 19), 3189-3197. doi:10.1242/jcs.072165 [doi]

Haddad, F. G., Eid, R., Kourie, H. R., Barouky, E., & Ghosn, M. (2018). Prognostic and predictive biomarkers in nonmetastatic colorectal cancers. Future Oncology (London, England), 14(21), 2097-2102. doi:10.2217/fon-2017-0708 [doi]

Halaoui, R., Rejon, C., Chatterjee, S. J., Szymborski, J., Meterissian, S., Muller, W. J., . . . McCaffrey, L. (2017). Progressive polarity loss and luminal collapse disrupt tissue organization in carcinoma. Genes & Development, 31(15), 1573-1587. doi:10.1101/gad.300566.117 [doi]

Hall, A. (2009). The cytoskeleton and cancer. Cancer Metastasis Reviews, 28(1-2), 5-14. doi:10.1007/s10555-008-9166-3 [doi]

Hannibal, C. G., Vang, R., Junge, J., Frederiksen, K., Kurman, R. J., & Kjaer, S. K. (2017). A nationwide study of ovarian serous borderline tumors in denmark 1978-2002. risk of recurrence, and development of ovarian serous carcinoma. Gynecologic Oncology, 144(1), 174-180. doi:S0090-8258(16)31563-3 [pii]

Harris, E. I., Lewin, D. N., Wang, H. L., Lauwers, G. Y., Srivastava, A., Shyr, Y., . . . Washington, M. K. (2008). Lymphovascular invasion in colorectal cancer: An interobserver variability study. The American Journal of Surgical Pathology, 32(12), 1816-1821. doi:10.1097/PAS.0b013e3181816083 [doi]

Hauptman, N., & Glavac, D. (2017). Colorectal cancer blood-based biomarkers. Gastroenterology Research and Practice, 2017, 2195361. doi:10.1155/2017/2195361 [doi]

He, X., Chien, J., & Shridhar, V. (2013). Assessment of resistance to anoikis in ovarian cancer. Methods in Molecular Biology (Clifton, N.J.), 1049, 347-354. doi:10.1007/978-1-62703-547-7_26 [doi]

Heinen, C. D., Richardson, D., White, R., & Groden, J. (1995). Microsatellite instability in colorectal adenocarcinoma cell lines that have full-length adenomatous polyposis coli protein. Cancer Research, 55(21), 4797-4799.

Helander, H. F., & Fandriks, L. (2014). Surface area of the digestive tract - revisited. Scandinavian Journal of Gastroenterology, 49(6), 681-689. doi:10.3109/ 00365521.2014.898326 [doi]

97

Hofmann, C., Obermeier, F., Artinger, M., Hausmann, M., Falk, W., Schoelmerich, J., . . . Grossmann, J. (2007). Cell-cell contacts prevent anoikis in primary human colonic epithelial cells. Gastroenterology, 132(2), 587-600. doi:S0016-5085(06)02472-3 [pii]

Hori, M., Miki, T., Okamoto, M., Yazama, F., Konishi, H., Kaneko, H., . . . Tatsuka, M. (2013). The detergent-soluble cytoplasmic pool of survivin suppresses anoikis and its expression is associated with metastatic disease of human colon cancer. PloS One, 8(2), e55710. doi:10.1371/journal.pone.0055710 [doi]

Ilyas, M., Tomlinson, I. P., Rowan, A., Pignatelli, M., & Bodmer, W. F. (1997). Beta-catenin mutations in cell lines established from human colorectal cancers. Proceedings of the National Academy of Sciences of the United States of America, 94(19), 10330-10334. doi:10.1073/pnas.94.19.10330 [doi]

Ivanov, A. I., Young, C., Den Beste, K., Capaldo, C. T., Humbert, P. O., Brennwald, P., . . . Nusrat, A. (2010). Tumor suppressor scribble regulates assembly of tight junctions in the intestinal epithelium. The American Journal of Pathology, 176(1), 134-145. doi:10.2353/ajpath.2010.090220 [doi]

Jaffe, A. B., Kaji, N., Durgan, J., & Hall, A. (2008). Cdc42 controls spindle orientation to position the apical surface during epithelial morphogenesis. The Journal of Cell Biology, 183(4), 625-633. doi:10.1083/jcb.200807121 [doi]

Jass, J. R. (2007). Heredity and DNA methylation in colorectal cancer. Gut, 56(1), 154-155. doi:56/1/154-a [pii]

Jeong, W. J., Ro, E. J., & Choi, K. Y. (2018). Interaction between wnt/beta-catenin and RAS-ERK pathways and an anti-cancer strategy via degradations of beta-catenin and RAS by targeting the wnt/beta-catenin pathway. NPJ Precision Oncology, 2(1), 5-018-0049-y. eCollection 2018. doi:10.1038/s41698-018-0049-y [doi]

Jin, Z., & El-Deiry, W. S. (2005). Overview of cell death signaling pathways. Cancer Biology & Therapy, 4(2), 139-163. doi:1508 [pii]

Jones, J. C., Yokoo, K. M., & Goldman, R. D. (1986). Is the hemidesmosome a half desmosome? an immunological comparison of mammalian desmosomes and hemidesmosomes. Cell Motility and the Cytoskeleton, 6(6), 560-569. doi:10.1002/cm.970060604 [doi]

Jones, J. J., Wilcox, B. E., Benz, R. W., Babbar, N., Boragine, G., Burrell, T., . . . Blume, J. E. (2016). A plasma-based protein marker panel for colorectal cancer detection identified by multiplex targeted mass spectrometry. Clinical Colorectal Cancer, 15(2), 186-194.e13. doi:10.1016/j.clcc.2016.02.004 [doi]

Kalimuthu, S. N., Chelliah, A., & Chetty, R. (2016). From traditional serrated adenoma to tubulovillous adenoma and beyond. World Journal of Gastrointestinal Oncology, 8(12), 805-809. doi:10.4251/wjgo.v8.i12.805 [doi]

Kang, S., Na, Y., Joung, S. Y., Lee, S. I., Oh, S. C., & Min, B. W. (2018). The significance of microsatellite instability in colorectal cancer after controlling for clinicopathological factors. Medicine, 97(9), e0019. doi:10.1097/MD.0000000000010019 [doi]

98

Kantola, T., Klintrup, K., Vayrynen, J. P., Vornanen, J., Bloigu, R., Karhu, T., . . . Makinen, M. J. (2012). Stage-dependent alterations of the serum cytokine pattern in colorectal carcinoma. British Journal of Cancer, 107(10), 1729-1736. doi:10.1038/bjc.2012.456 [doi]

Ke, X., Li, L., Dong, H. L., & Chen, Z. N. (2012). Acquisition of anoikis resistance through CD147 upregulation: A new mechanism underlying metastasis of hepatocellular carcinoma cells. Oncology Letters, 3(6), 1249-1254. doi:10.3892/ol.2012.658 [doi]

Kim, J. H., Hwang, Y. J., Han, S. H., Lee, Y. E., Kim, S., Kim, Y. J., . . . Kim, S. H. (2015). Dexamethasone inhibits hypoxia-induced epithelial-mesenchymal transition in colon cancer. World Journal of Gastroenterology, 21(34), 9887-9899. doi:10.3748/ wjg.v21.i34.9887 [doi]

Kim, Y. N., Koo, K. H., Sung, J. Y., Yun, U. J., & Kim, H. (2012). Anoikis resistance: An essential prerequisite for tumor metastasis. International Journal of Cell Biology, 2012, 306879. doi:10.1155/2012/306879 [doi]

Kleinbaum, D. G., & Klein, M. (2012). Survival analysis : A selft-learning text (Third ed.) doi:10.1007/978-1-4419-6646-9_2

Klintrup, K., Makinen, J. M., Kauppila, S., Vare, P. O., Melkko, J., Tuominen, H., . . . Makinen, M. J. (2005). Inflammation and prognosis in colorectal cancer. European Journal of Cancer (Oxford, England : 1990), 41(17), 2645-2654. doi:S0959-8049(05)00729-X [pii]

Knickelbein, K., & Zhang, L. (2015). Mutant KRAS as a critical determinant of the therapeutic response of colorectal cancer. Genes & Diseases, 2(1), 4-12. doi:10.1016/j.gendis.2014.10.002 [doi]

Kojima, H., Okumura, T., Yamaguchi, T., Miwa, T., Shimada, Y., & Nagata, T. (2017). Enhanced cancer stem cell properties of a mitotically quiescent subpopulation of p75NTR-positive cells in esophageal squamous cell carcinoma. International Journal of Oncology, 51(1), 49-62. doi:10.3892/ijo.2017.4001 [doi]

Kong, J. C., Guerra, G. R., Pham, T., Mitchell, C., Lynch, A. C., Warrier, S. K., . . . Heriot, A. G. (2019). Prognostic impact of tumor-infiltrating lymphocytes in primary and metastatic colorectal cancer: A systematic review and meta-analysis. Diseases of the Colon and Rectum, 62(4), 498-508. doi:10.1097/DCR.0000000000001332 [doi]

Kourtidis, A., Lu, R., Pence, L. J., & Anastasiadis, P. Z. (2017). A central role for cadherin signaling in cancer. Experimental Cell Research, 358(1), 78-85. doi:S0014-4827(17)30206-9 [pii]

Kozlova, N. I., Morozevich, G. E., Chubukina, A. N., & Berman, A. E. (2001). Integrin alphavbeta3 promotes anchorage-dependent apoptosis in human intestinal carcinoma cells. Oncogene, 20(34), 4710-4717. doi:10.1038/sj.onc.1204619 [doi]

Kozlova, N. I., Morozevich, G. E., Ushakova, N. A., & Berman, A. E. (2019). Implication of integrin alpha2beta1 in anoikis of SK-mel-147 human melanoma cells: A non-canonical function of akt protein kinase. Oncotarget, 10(19), 1829-1839. doi:10.18632/oncotarget.26746 [doi]

99

Kyle, A. H., Baker, J. H., & Minchinton, A. I. (2012). Targeting quiescent tumor cells via oxygen and IGF-I supplementation. Cancer Research, 72(3), 801-809. doi:10.1158/0008-5472.CAN-11-3059 [doi]

Laguinge, L. M., Samara, R. N., Wang, W., El-Deiry, W. S., Corner, G., Augenlicht, L., . . . Jessup, J. M. (2008). DR5 receptor mediates anoikis in human colorectal carcinoma cell lines. Cancer Research, 68(3), 909-917. doi:10.1158/0008-5472.CAN-06-1806 [doi]

Leggett B, & Whitehall V. (2010). Role of the serrated pathway in colorectal cancer pathogenesis. Gastroenterology, 138(6), 2088-2100. doi:http://dx.doi.org/ 10.1053/ j.gastro.2009.12.066

Lelievre, S. A. (2010). Tissue polarity-dependent control of mammary epithelial homeostasis and cancer development: An epigenetic perspective. Journal of Mammary Gland Biology and Neoplasia, 15(1), 49-63. doi:10.1007/s10911-010-9168-y [doi]

Leslie, A., Pratt, N. R., Gillespie, K., Sales, M., Kernohan, N. M., Smith, G., . . . Steele, R. J. (2003). Mutations of APC, K-ras, and p53 are associated with specific chromosomal aberrations in colorectal adenocarcinomas. Cancer Research, 63(15), 4656-4661.

Leung, C. T., & Brugge, J. S. (2012). Outgrowth of single oncogene-expressing cells from suppressive epithelial environments. Nature, 482(7385), 410-413. doi:10.1038/ nature10826 [doi]

Li, M. X., Liu, X. M., Zhang, X. F., Zhang, J. F., Wang, W. L., Zhu, Y., . . . Lv, Y. (2014). Prognostic role of neutrophil-to-lymphocyte ratio in colorectal cancer: A systematic review and meta-analysis. International Journal of Cancer, 134(10), 2403-2413. doi:10.1002/ijc.28536 [doi]

Li, N., Wang, D., Sui, Z., Qi, X., Ji, L., Wang, X., & Yang, L. (2013). Development of an improved three-dimensional in vitro intestinal mucosa model for drug absorption evaluation. Tissue Engineering.Part C, Methods, 19(9), 708-719. doi:10.1089/ ten.TEC.2012.0463 [doi]

Li, V. D., Li, K. H., & Li, J. T. (2019). TP53 mutations as potential prognostic markers for specific cancers: Analysis of data from the cancer genome atlas and the international agency for research on cancer TP53 database. Journal of Cancer Research and Clinical Oncology, 145(3), 625-636. doi:10.1007/s00432-018-2817-z [doi]

Liang, E., Kabcenell, A. K., Coleman, J. R., Robson, J., Ruffles, R., & Yazdanian, M. (2001). Permeability measurement of macromolecules and assessment of mucosal antigen sampling using in vitro converted M cells. Journal of Pharmacological and Toxicological Methods, 46(2), 93-101. doi:S1056-8719(02)00163-6 [pii]

Lievre, A., Bachet, J. B., Le Corre, D., Boige, V., Landi, B., Emile, J. F., . . . Laurent-Puig, P. (2006). KRAS mutation status is predictive of response to cetuximab therapy in colorectal cancer. Cancer Research, 66(8), 3992-3995. doi:66/8/3992 [pii]

Lin, M., Ma, S. P., Lin, H. Z., Ji, P., Xie, D., & Yu, J. X. (2010). Intratumoral as well as peritumoral lymphatic vessel invasion correlates with lymph node metastasis and unfavourable outcome in colorectal cancer. Clinical & Experimental Metastasis, 27(3), 123-132. doi:10.1007/s10585-010-9309-0 [doi]

100

Lino-Silva, L. S., Salcedo-Hernandez, R. A., Herrera-Gomez, A., Padilla-Rosciano, A., Ramirez-Jaramillo, M., Herrera-Goepfert, R. E., & Meneses-Garcia, A. (2015). Colonic cribriform carcinoma, a morphologic pattern associated with low survival. International Journal of Surgical Pathology, 23(1), 13-19. doi:10.1177/1066896914542125 [doi]

Liu, Y., Zhang, Y., Wu, H., Li, Y., Zhang, Y., Liu, M., . . . Tang, H. (2017). miR-10a suppresses colorectal cancer metastasis by modulating the epithelial-to-mesenchymal transition and anoikis. Cell Death & Disease, 8(4), e2739. doi:10.1038/cddis.2017.61 [doi]

Lubarsky, B., & Krasnow, M. A. (2003). Tube morphogenesis: Making and shaping biological tubes. Cell, 112(1), 19-28. doi:S0092867402012837 [pii]

Lugini, L., Matarrese, P., Tinari, A., Lozupone, F., Federici, C., Iessi, E., . . . Fais, S. (2006). Cannibalism of live lymphocytes by human metastatic but not primary melanoma cells. Cancer Research, 66(7), 3629-3638. doi:66/7/3629 [pii]

Lugli, A., Kirsch, R., Ajioka, Y., Bosman, F., Cathomas, G., Dawson, H., . . . Quirke, P. (2017). Recommendations for reporting tumor budding in colorectal cancer based on the international tumor budding consensus conference (ITBCC) 2016. Modern Pathology : An Official Journal of the United States and Canadian Academy of Pathology, Inc, 30(9), 1299-1311. doi:10.1038/modpathol.2017.46 [doi]

Lundberg, I. V., Lofgren Burstrom, A., Edin, S., Eklof, V., Oberg, A., Stenling, R., . . . Wikberg, M. L. (2014). SOX2 expression is regulated by BRAF and contributes to poor patient prognosis in colorectal cancer. PloS One, 9(7), e101957. doi:10.1371/ journal.pone.0101957 [doi]

Luo, Z. W., Zhu, M. G., Zhang, Z. Q., Ye, F. J., Huang, W. H., & Luo, X. Z. (2019). Increased expression of ki-67 is a poor prognostic marker for colorectal cancer patients: A meta analysis. BMC Cancer, 19(1), 123-019-5324-y. doi:10.1186/s12885-019-5324-y [doi]

Macara, I. G., & McCaffrey, L. (2013). Cell polarity in morphogenesis and metastasis. Philosophical Transactions of the Royal Society of London.Series B, Biological Sciences, 368(1629), 20130012. doi:10.1098/rstb.2013.0012 [doi]

Magudia, K., Lahoz, A., & Hall, A. (2012). K-ras and B-raf oncogenes inhibit colon epithelial polarity establishment through up-regulation of c-myc. The Journal of Cell Biology, 198(2), 185-194. doi:10.1083/jcb.201202108 [doi]

Makinen, M. J. (2007). Colorectal serrated adenocarcinoma. Histopathology, 50(1), 131-150. doi:HIS2548 [pii]

Makrodouli, E., Oikonomou, E., Koc, M., Andera, L., Sasazuki, T., Shirasawa, S., & Pintzas, A. (2011). BRAF and RAS oncogenes regulate rho GTPase pathways to mediate migration and invasion properties in human colon cancer cells: A comparative study. Molecular Cancer, 10, 118-4598-10-118. doi:10.1186/1476-4598-10-118 [doi]

Marks, K. M., West, N. P., Morris, E., & Quirke, P. (2018). Clinicopathological, genomic and immunological factors in colorectal cancer prognosis. The British Journal of Surgery, 105(2), e99-e109. doi:10.1002/bjs.10756 [doi]

101

Martin-Belmonte, F., Yu, W., Rodriguez-Fraticelli, A. E., Ewald, A. J., Werb, Z., Alonso, M. A., & Mostov, K. (2008). Cell-polarity dynamics controls the mechanism of lumen formation in epithelial morphogenesis. Current Biology : CB, 18(7), 507-513. doi:10.1016/j.cub.2008.02.076 [doi]

Martini, M., De Santis, M. C., Braccini, L., Gulluni, F., & Hirsch, E. (2014). PI3K/AKT signaling pathway and cancer: An updated review. Annals of Medicine, 46(6), 372-383. doi:10.3109/07853890.2014.912836 [doi]

Masunaga, S., & Ono, K. (2002). Significance of the response of quiescent cell populations within solid tumors in cancer therapy. Journal of Radiation Research, 43(1), 11-25.

McCubrey, J. A., Steelman, L. S., Chappell, W. H., Abrams, S. L., Wong, E. W., Chang, F., . . . Franklin, R. A. (2007). Roles of the raf/MEK/ERK pathway in cell growth, malignant transformation and drug resistance. Biochimica Et Biophysica Acta, 1773(8), 1263-1284. doi:S0167-4889(06)00315-6 [pii]

Melling, N., Kowitz, C. M., Simon, R., Bokemeyer, C., Terracciano, L., Sauter, G., . . . Marx, A. H. (2016). High Ki67 expression is an independent good prognostic marker in colorectal cancer. Journal of Clinical Pathology, 69(3), 209-214. doi:10.1136/jclinpath-2015-202985 [doi]

Mita, A. C., Mita, M. M., Nawrocki, S. T., & Giles, F. J. (2008). Survivin: Key regulator of mitosis and apoptosis and novel target for cancer therapeutics. Clinical Cancer Research : An Official Journal of the American Association for Cancer Research, 14(16), 5000-5005. doi:10.1158/1078-0432.CCR-08-0746 [doi]

Mundade, R., Imperiale, T. F., Prabhu, L., Loehrer, P. J., & Lu, T. (2014). Genetic pathways, prevention, and treatment of sporadic colorectal cancer. Oncoscience, 1(6), 400-406. doi:59 [pii]

Nazemalhosseini Mojarad, E., Kashfi, S. M., Mirtalebi, H., Taleghani, M. Y., Azimzadeh, P., Savabkar, S., . . . Zali, M. R. (2016). Low level of microsatellite instability correlates with poor clinical prognosis in stage II colorectal cancer patients. Journal of Oncology, 2016, 2196703. doi:10.1155/2016/2196703 [doi]

Nguyen, H. T., & Duong, H. Q. (2018). The molecular characteristics of colorectal cancer: Implications for diagnosis and therapy. Oncology Letters, 16(1), 9-18. doi:10.3892/ol.2018.8679 [doi]

Nikolaou, S., Qiu, S., Fiorentino, F., Rasheed, S., Tekkis, P., & Kontovounisios, C. (2018). Systematic review of blood diagnostic markers in colorectal cancer. Techniques in Coloproctology, 22(7), 481-498. doi:10.1007/s10151-018-1820-3 [doi]

Niu, J., Xue, A., Chi, Y., Xue, J., Wang, W., Zhao, Z., . . . Wu, Z. H. (2016). Induction of miRNA-181a by genotoxic treatments promotes chemotherapeutic resistance and metastasis in breast cancer. Oncogene, 35(10), 1302-1313. doi:10.1038/onc.2015.189 [doi]

Noffsinger, A. E. (2009). Serrated polyps and colorectal cancer: New pathway to malignancy. Annual Review of Pathology, 4, 343-364. doi:http://dx.doi.org/ 10.1146/ annurev.pathol.4.110807.092317

102

Okugawa, Y., Toiyama, Y., Toden, S., Mitoma, H., Nagasaka, T., Tanaka, K., . . . Goel, A. (2017). Clinical significance of SNORA42 as an oncogene and a prognostic biomarker in colorectal cancer. Gut, 66(1), 107-117. doi:10.1136/gutjnl-2015-309359 [doi]

Osaki, M., Oshimura, M., & Ito, H. (2004). PI3K-akt pathway: Its functions and alterations in human cancer. Apoptosis : An International Journal on Programmed Cell Death, 9(6), 667-676. doi:5394700 [pii]

Paarnio, K., Tuomisto, A., Vayrynen, S. A., Vayrynen, J. P., Klintrup, K., Ohtonen, P., . . . Karttunen, T. J. (2019). Serum TLR2 and TLR4 levels in colorectal cancer and their association with systemic inflammatory markers, tumor characteristics, and disease outcome. APMIS : Acta Pathologica, Microbiologica, Et Immunologica Scandinavica, 127(8), 561-569. doi:10.1111/apm.12971 [doi]

Pancione, M., Remo, A., & Colantuoni, V. (2012). Genetic and epigenetic events generate multiple pathways in colorectal cancer progression. Pathology Research International, 2012, 509348. doi:10.1155/2012/509348 [doi]

Paoli, P., Giannoni, E., & Chiarugi, P. (2013). Anoikis molecular pathways and its role in cancer progression. Biochimica Et Biophysica Acta, 1833(12), 3481-3498. doi:10.1016/j.bbamcr.2013.06.026 [doi]

Parfitt, J. R., & Driman, D. K. (2007). Survivin and hedgehog protein expression in serrated colorectal polyps: An immunohistochemical study. Human Pathology, 38(5), 710-717. doi:S0046-8177(06)00704-0 [pii]

Pereira, I., Lechanteur, A., & Sarmento, B. (2018). 3D model replicating the intestinal function to evaluate drug permeability. Methods in Molecular Biology (Clifton, N.J.), 1817, 107-113. doi:10.1007/978-1-4939-8600-2_11 [doi]

Phipps, A. I., Buchanan, D. D., Makar, K. W., Win, A. K., Baron, J. A., Lindor, N. M., . . . Newcomb, P. A. (2013). KRAS-mutation status in relation to colorectal cancer survival: The joint impact of correlated tumour markers. British Journal of Cancer, 108(8), 1757-1764. doi:10.1038/bjc.2013.118 [doi]

Pinto, M. P., Jacobsen, B. M., & Horwitz, K. B. (2011). An immunohistochemical method to study breast cancer cell subpopulations and their growth regulation by hormones in three-dimensional cultures. Frontiers in Endocrinology, 2, 15. doi:10.3389/fendo.2011.00015 [doi]

Ponz de Leon, M., & Di Gregorio, C. (2001). Pathology of colorectal cancer. Digestive and Liver Disease : Official Journal of the Italian Society of Gastroenterology and the Italian Association for the Study of the Liver, 33(4), 372-388. doi:S1590-8658(01)80095-5 [pii]

Proctor, M. J., Morrison, D. S., Talwar, D., Balmer, S. M., Fletcher, C. D., O'Reilly, D. S., . . . McMillan, D. C. (2011). A comparison of inflammation-based prognostic scores in patients with cancer. A glasgow inflammation outcome study. European Journal of Cancer (Oxford, England : 1990), 47(17), 2633-2641. doi:10.1016/j.ejca.2011.03.028 [doi]

Raskov, H., Pommergaard, H. C., Burcharth, J., & Rosenberg, J. (2014). Colorectal carcinogenesis--update and perspectives. World Journal of Gastroenterology, 20(48), 18151-18164. doi:10.3748/wjg.v20.i48.18151 [doi]

103

Reggiani Bonetti, L., Barresi, V., Bettelli, S., Domati, F., & Palmiere, C. (2016). Poorly differentiated clusters (PDC) in colorectal cancer: What is and ought to be known. Diagnostic Pathology, 11, 31-016-0481-7. doi:10.1186/s13000-016-0481-7 [doi]

Reimers, M. S., Zeestraten, E. C., van Alphen, T. C., Dekker, J. W., Putter, H., Saadatmand, S., . . . Kuppen, P. J. (2014). Combined analysis of biomarkers of proliferation and apoptosis in colon cancer: An immunohistochemistry-based study using tissue microarray. International Journal of Colorectal Disease, 29(9), 1043-1052. doi:10.1007/s00384-014-1930-y [doi]

Richman, S. D., Seymour, M. T., Chambers, P., Elliott, F., Daly, C. L., Meade, A. M., . . . Quirke, P. (2009). KRAS and BRAF mutations in advanced colorectal cancer are associated with poor prognosis but do not preclude benefit from oxaliplatin or irinotecan: Results from the MRC FOCUS trial. Journal of Clinical Oncology : Official Journal of the American Society of Clinical Oncology, 27(35), 5931-5937. doi:10.1200/ JCO.2009.22.4295 [doi]

Rohwer, N., Welzel, M., Daskalow, K., Pfander, D., Wiedenmann, B., Detjen, K., & Cramer, T. (2008). Hypoxia-inducible factor 1alpha mediates anoikis resistance via suppression of alpha5 integrin. Cancer Research, 68(24), 10113-10120. doi:10.1158/0008-5472.CAN-08-1839 [doi]

Rosty, C., Hewett, D. G., Brown, I. S., Leggett, B. A., & Whitehall, V. L. (2013). Serrated polyps of the large intestine: Current understanding of diagnosis, pathogenesis, and clinical management. Journal of Gastroenterology, 48(3), 287-302. doi:10.1007/s00535-012-0720-y [doi]

Saif, M. W., & Chu, E. (2010). Biology of colorectal cancer. Cancer Journal (Sudbury, Mass.), 16(3), 196-201. doi:10.1097/PPO.0b013e3181e076af [doi]

Saigusa, S., Inoue, Y., Tanaka, K., Okugawa, Y., Toiyama, Y., Uchida, K., . . . Kusunoki, M. (2014). Lack of M30 expression correlates with factors reflecting tumor progression in rectal cancer with preoperative chemoradiotherapy. Molecular and Clinical Oncology, 2(1), 99-104. doi:10.3892/mco.2013.189 [doi]

Sakai, E., Nakayama, M., Oshima, H., Kouyama, Y., Niida, A., Fujii, S., . . . Oshima, M. (2018). Combined mutation of apc, KRAS, and Tgfbr2 effectively drives metastasis of intestinal cancer. Cancer Research, 78(5), 1334-1346. doi:10.1158/0008-5472.CAN-17-3303 [doi]

San Roman, A. K., & Shivdasani, R. A. (2011). Boundaries, junctions and transitions in the gastrointestinal tract. Experimental Cell Research, 317(19), 2711-2718. doi:10.1016/ j.yexcr.2011.07.011 [doi]

Saridaki, Z., Souglakos, J., & Georgoulias, V. (2014). Prognostic and predictive significance of MSI in stages II/III colon cancer. World Journal of Gastroenterology, 20(22), 6809-6814. doi:10.3748/wjg.v20.i22.6809 [doi]

Satelli, A., Mitra, A., Brownlee, Z., Xia, X., Bellister, S., Overman, M. J., . . . Li, S. (2015). Epithelial-mesenchymal transitioned circulating tumor cells capture for detecting tumor progression. Clinical Cancer Research : An Official Journal of the American Association for Cancer Research, 21(4), 899-906. doi:10.1158/1078-0432.CCR-14-0894 [doi]

104

Sepulveda, A. R., Hamilton, S. R., Allegra, C. J., Grody, W., Cushman-Vokoun, A. M., Funkhouser, W. K., . . . Nowak, J. A. (2017). Molecular biomarkers for the evaluation of colorectal cancer: Guideline from the american society for clinical pathology, college of american pathologists, association for molecular pathology, and the american society of clinical oncology. Journal of Clinical Oncology : Official Journal of the American Society of Clinical Oncology, 35(13), 1453-1486. doi:10.1200/JCO.2016.71.9807 [doi]

Shibutani, M., Maeda, K., Nagahara, H., Fukuoka, T., Matsutani, S., Kimura, K., . . . Ohira, M. (2018). The prognostic value of the systemic inflammatory score in patients with unresectable metastatic colorectal cancer. Oncology Letters, 16(1), 666-672. doi:10.3892/ol.2018.8628 [doi]

Simon, K. (2016). Colorectal cancer development and advances in screening. Clinical Interventions in Aging, 11, 967-976. doi:10.2147/CIA.S109285 [doi]

Snover, D. C. (2011). Update on the serrated pathway to colorectal carcinoma. Human Pathology, 42(1), 1-10. doi:http://dx.doi.org/10.1016/j.humpath.2010.06.002

Stanta, G., & Bonin, S. (2018). Overview on clinical relevance of intra-tumor heterogeneity. Frontiers in Medicine, 5, 85. doi:10.3389/fmed.2018.00085 [doi]

Stauber RH, Mann W, & Knauer SK. (2007). Nuclear and cytoplasmic survivin: Molecular mechanism, prognostic, and therapeutic potential. Cancer Research, 67(13), 5999-6002.

Strilic, B., & Offermanns, S. (2017). Intravascular survival and extravasation of tumor cells. Cancer Cell, 32(3), 282-293. doi:S1535-6108(17)30293-3 [pii]

Stupack, D. G., Puente, X. S., Boutsaboualoy, S., Storgard, C. M., & Cheresh, D. A. (2001). Apoptosis of adherent cells by recruitment of caspase-8 to unligated integrins. The Journal of Cell Biology, 155(3), 459-470. doi:10.1083/jcb.200106070 [doi]

Sweetser, S., Smyrk, T. C., & Sinicrope, F. A. (2013). Serrated colon polyps as precursors to colorectal cancer. Clinical Gastroenterology and Hepatology : The Official Clinical Practice Journal of the American Gastroenterological Association, 11(7), 760-7; quiz e54-5. doi:10.1016/j.cgh.2012.12.004 [doi]

Taddei, M. L., Giannoni, E., Fiaschi, T., & Chiarugi, P. (2012). Anoikis: An emerging hallmark in health and diseases. The Journal of Pathology, 226(2), 380-393. doi:10.1002/path.3000 [doi]

Taieb, J., Le Malicot, K., Shi, Q., Penault-Llorca, F., Bouche, O., Tabernero, J., . . . Sinicrope, F. A. (2016). Prognostic value of BRAF and KRAS mutations in MSI and MSS stage III colon cancer. Journal of the National Cancer Institute, 109(5), 10.1093/jnci/djw272. Print 2017 May. doi:10.1093/jnci/djw272 [doi]

Tan, B. J., & Chiu, G. N. (2013). Role of oxidative stress, endoplasmic reticulum stress and ERK activation in triptolide-induced apoptosis. International Journal of Oncology, 42(5), 1605-1612. doi:10.3892/ijo.2013.1843 [doi]

Tuppurainen, K., Makinen, J. M., Junttila, O., Liakka, A., Kyllonen, A. P., Tuominen, H., . . . Makinen, M. J. (2005). Morphology and microsatellite instability in sporadic serrated and non-serrated colorectal cancer. The Journal of Pathology, 207(3), 285-294. doi:10.1002/path.1850 [doi]

Tyrrell, H. E. J., & Kerr, D. (2018). Prognostic markers for colorectal cancer. Oncotarget, 9(69), 33060-33061. doi:10.18632/oncotarget.26012 [doi]

105

Ueno, H., Hase, K., Hashiguchi, Y., Shimazaki, H., Tanaka, M., Miyake, O., . . . Sugihara, K. (2014). Site-specific tumor grading system in colorectal cancer: Multicenter pathologic review of the value of quantifying poorly differentiated clusters. The American Journal of Surgical Pathology, 38(2), 197-204. doi:10.1097/PAS.0000000000000113 [doi]

Ueno, H., Kajiwara, Y., Shimazaki, H., Shinto, E., Hashiguchi, Y., Nakanishi, K., . . . Hase, K. (2012). New criteria for histologic grading of colorectal cancer. The American Journal of Surgical Pathology, 36(2), 193-201. doi:10.1097/PAS.0b013e318235edee [doi]

Vachon, P. H. (2011). Integrin signaling, cell survival, and anoikis: Distinctions, differences, and differentiation. Journal of Signal Transduction, 2011, 738137. doi:10.1155/2011/738137 [doi]

Vachon, P. H. (2018). Methods for assessing apoptosis and anoikis in normal intestine/colon and colorectal cancer. Methods in Molecular Biology (Clifton, N.J.), 1765, 99-137. doi:10.1007/978-1-4939-7765-9_7 [doi]

van de Velde, C. J., Boelens, P. G., Borras, J. M., Coebergh, J. W., Cervantes, A., Blomqvist, L., . . . Valentini, V. (2014). EURECCA colorectal: Multidisciplinary management: European consensus conference colon & rectum. European Journal of Cancer (Oxford, England : 1990), 50(1), 1.e1-1.e34. doi:10.1016/j.ejca.2013.06.048 [doi]

Vayrynen, J. P., Sajanti, S. A., Klintrup, K., Makela, J., Herzig, K. H., Karttunen, T. J., . . . Makinen, M. J. (2014). Characteristics and significance of colorectal cancer associated lymphoid reaction. International Journal of Cancer, 134(9), 2126-2135. doi:10.1002/ijc.28533 [doi]

Vayrynen, J. P., Tuomisto, A., Klintrup, K., Makela, J., Karttunen, T. J., & Makinen, M. J. (2013). Detailed analysis of inflammatory cell infiltration in colorectal cancer. British Journal of Cancer, 109(7), 1839-1847. doi:10.1038/bjc.2013.508 [doi]

Vayrynen, J. P., Vornanen, J., Tervahartiala, T., Sorsa, T., Bloigu, R., Salo, T., . . . Makinen, M. J. (2012). Serum MMP-8 levels increase in colorectal cancer and correlate with disease course and inflammatory properties of primary tumors. International Journal of Cancer, 131(4), E463-74. doi:10.1002/ijc.26435 [doi]

Vu, T., & Datta, P. K. (2017). Regulation of EMT in colorectal cancer: A culprit in metastasis. Cancers, 9(12), 10.3390/cancers9120171. doi:E171 [pii]

Wang, Y. N., Zeng, Z. L., Lu, J., Wang, Y., Liu, Z. X., He, M. M., . . . Xu, R. H. (2018). CPT1A-mediated fatty acid oxidation promotes colorectal cancer cell metastasis by inhibiting anoikis. Oncogene, 37(46), 6025-6040. doi:10.1038/s41388-018-0384-z [doi]

West, N. J., Courtney, E. D., Poullis, A. P., & Leicester, R. J. (2009). Apoptosis in the colonic crypt, colorectal adenomata, and manipulation by chemoprevention. Cancer Epidemiology, Biomarkers & Prevention : A Publication of the American Association for Cancer Research, Cosponsored by the American Society of Preventive Oncology, 18(6), 1680-1687. doi:10.1158/1055-9965.EPI-09-0006 [doi]

Yang, H. M., Mitchell, J. M., Sepulveda, J. L., & Sepulveda, A. R. (2015). Molecular and histologic considerations in the assessment of serrated polyps. Archives of Pathology & Laboratory Medicine, 139(6), 730-741. doi:10.5858/arpa.2014-0424-RA [doi]

106

Yiu, A. J., & Yiu, C. Y. (2016). Biomarkers in colorectal cancer. Anticancer Research, 36(3), 1093-1102. doi:36/3/1093 [pii]

Zajac, O., Raingeaud, J., Libanje, F., Lefebvre, C., Sabino, D., Martins, I., . . . Jaulin, F. (2018). Tumour spheres with inverted polarity drive the formation of peritoneal metastases in patients with hypermethylated colorectal carcinomas. Nature Cell Biology, 20(3), 296-306. doi:10.1038/s41556-017-0027-6 [doi]

Zhang, X., Cromwell, J. W., Kunjummen, B. D., Yee, D., & Garcia-Aguilar, J. (2003). The alpha2 and alpha3 integrins are required for morphologic differentiation of an intestinal epithelial cell line. Surgery, 133(4), 429-437. doi:10.1067/msy.2003.107 [doi]

107

Original publications

I Patankar M, Väyrynen S, Tuomisto A, Mäkinen M, Eskelinen S and Karttunen TJ. (2018). Micropapillary structures in colorectal cancer: An anoikis-resistant subpopulation. Anticancer Research 38, 2915-2921.

II Patankar M, Mattila T, Väyrynen JP, Klintrup K, Mäkelä J, Tuomisto A, Nieminen P, Mäkinen M J, Karttunen TJ. (2019). Histopathological assessment of anoikis resistance in colorectal carcinoma: Abundance of putative anoikis resistant subpopulations indicate poor prognosis. Manuscript.

III Patankar M, Eskelinen S, Tuomisto A, Mäkinen M, Karttunen TJ (2019). KRAS and BRAF mutations induce anoikis resistance in Caco-2 cells and characteristic 3D phenotypes in Caco-2 cells. Molecular Medicine Reports, 20: 4634-4644.

Reprinted with permission from published articles (I, III).

Original publications are not included in the electronic version of the dissertation


Book orders:Virtual book store

http://verkkokauppa.juvenesprint.fi

S E R I E S D M E D I C A

1532. Luukkonen, Jani (2019) Osteopontin and osteoclasts in rheumatoid arthritis andosteoarthritis

1533. Niemelä, Maisa (2019) Temporal patterns of physical activity and sedentary timeand their association with health at mid-life : The Northern Finland Birth Cohort1966 study

1534. Hakala, Mervi (2019) Ihokontaktin, ensi-imetyksen, vierihoidon ja täysimetyksentoteutuminen synnytyssairaaloissa

1535. Jussila, Katja (2019) On the autism spectrum? : recognition and assessment ofquantitative autism traits in high-functioning school-aged children. An epidemio-logical and clinical study

1536. Terho, Henri (2019) Electrocardiographic risk markers for cardiac events inmiddle-aged population

1537. Kyösti, Elina (2019) Long-term outcome, quality of life, and socio-economicconsequences of surviving pediatric intensive care

1538. Ylönen, Susanna (2019) Genetic risk factors for movement disorders in Finland

1539. Sirniö, Päivi (2019) Determinants of systemic inflammation in colorectal cancer

1540. Ylitalo-Katajisto, Kirsti (2019) Paljon palveluja tarvitsevien asiakkaiden yksilöitysosiaali- ja terveyspalvelujen yhteen kokoaminen

1541. Tiri, Hannu (2019) Comorbidities and mortality of hidradenitis suppurativa inFinland

1542. Hynynen, Johanna (2019) Status epilepticus in mitochondrial diseases and the roleof POLG1 variants in the valproic-acid induced hepatotoxicity

1543. Urpilainen, Elina (2019) The role of metformin and statins in ovarian and breastcancer in women with type 2 diabetes

1544. Remes, Tiina (2019) Signs of radiation-induced accelerated ageing in survivors ofchildhood brain tumors : The incidence of cerebrovascular disease, neurocogni-tive impairment, secondary neoplasms, and low bone mineral density after 18years of follow-up

1545. Karjula, Heikki (2019) Diagnosis, treatment and prophylaxis of pancreatic fistulasin severe necrotizing pancreatitis and the long-term outcome of acute pan-creatitis

1546. Huhtakangas, Jaana (2019) Evolution of obstructive sleep apnea after ischemicstroke

UNIVERSITY OF OULU P .O. Box 8000 F I -90014 UNIVERSITY OF OULU FINLAND


University Lecturer Tuomo Glumoff

University Lecturer Santeri Palviainen

Senior research fellow Jari Juuti


University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

University Lecturer Anu Soikkeli


Publications Editor Kirsti Nurkkala

ISBN 978-952-62-2439-8 (Paperback)ISBN 978-952-62-2440-4 (PDF)ISSN 0355-3221 (Print)ISSN 1796-2234 (Online)

U N I V E R S I TAT I S O U L U E N S I S

MEDICA

ACTAD

D 1547

AC

TAM

adhura Patankar

OULU 2019

D 1547

Madhura Patankar


UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF MEDICINE

oulu 2019 d 1547 university of oulu p.o. box 8000 fi...

Documents