The Dynamic Interplay of Epigenetics and Genetics in ......Jamie and Liyang, thank you for welcoming...

The Dynamic Interplay of Epigenetics and Genetics in Selected DNA Mismatch Repair and Wnt Signaling

Pathway Genes in Colorectal Cancer

by

Andrea Josephine Savio

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Graduate Department of Laboratory Medicine and Pathobiology University of Toronto

© Copyright by Andrea Josephine Savio 2017

ii

The Dynamic Interplay of Epigenetics and Genetics in Selected

DNA Mismatch Repair and Wnt Signaling Pathway Genes in

Colorectal Cancer

Andrea Josephine Savio

Doctor of Philosophy

Laboratory Medicine and Pathobiology

University of Toronto

2017

Abstract

Colorectal cancers (CRC) undergo distinct genetic and epigenetic alterations, contributing

towards chromosomal instability (CIN), microsatellite instability (MSI), and/or epigenetic

instability (CpG island methylator phenotype). Two major pathways disrupted in CRC include

upregulation of Wnt signaling and deficiency in DNA mismatch repair (MMR). MLH1, a MMR

gene that corrects DNA replication errors, is frequently lost in CRC due to promoter CpG island

hypermethylation leading to the MSI phenotype. A single nucleotide polymorphism (SNP,

rs1800734) in the MLH1 promoter is associated with MLH1 CpG island hypermethylation,

decreased expression, and MSI in tumours. In this thesis, a novel association was discovered

between variant SNP genotype and MLH1 CpG shore hypomethylation in normal tissues of CRC

cases and control individuals, specifically in peripheral blood mononuclear cells and normal

colon. Interestingly, no SNP-associated differences in methylation were observed in tumour

DNA at the MLH1 shore. Chromatin immunoprecipitation studies of histone modifications in

CRC cell lines revealed that the MLH1 shore may be a poised enhancer. The transcription factor

AP4 was shown to bind only at the MLH1 promoter in wildtype and heterozygous cell lines but

iii

not in homozygous variant carriers of rs1800734, demonstrating a functional consequence of this

polymorphism. To further explore epigenetic events between CRC subtypes the methylation

status of two genes involved in Wnt signaling were assessed. Methylation of ITF2, a Wnt target

gene, and APC, which inhibits Wnt signaling and is frequently mutated in CRC, was measured in

two CRC patient populations. Both genes incurred tumour-specific methylation, with APC

methylation occurring equally across CRC subtypes while ITF2 methylation was associated with

MSI tumours. This thesis has investigated the functional epigenetic regulation occurring at key

genes involved in DNA MMR and Wnt signaling, two pathways often dysregulated in CRC. This

work provides insight into the regulation of CpG shores, how DNA variants play a role in

epigenetics, and the contributions of DNA methylation to CRC subtypes and cancer

susceptibility.

iv

Acknowledgments

First and foremost, I would like to thank my supervisor and mentor, Dr. Bharati Bapat, for giving me the opportunity to join her laboratory. She has provided me an immense amount of guidance, support, training, and encouragement not only in my research, but also in all matters of life. I have grown so much, both professionally and personally, thanks to her.

I would also like to thank my thesis advisory members, Dr. Julia Knight and Dr. Philip Marsden, for their guidance throughout my time as a graduate student. I value the support that Dr. Knight provided, especially her help with analysis and interpretation of results. I also appreciate the thought-provoking questions that Dr. Marsden always provided. Thank you also to Dr. Mathieu Lemire, whose expertise and help proved invaluable to me throughout my PhD experience. His curiosity reminds me to always be willing to keep learning throughout life.

Thank you to the many amazing lab members that I have had the pleasure of spending time with over the years. Each and every one of you has helped me and influenced me in your own way. Jamie and Liyang, thank you for welcoming me to the lab; even though our time together was short, you ensured that my first few months in the lab were great. Ken, you’ve always been there for answers and advice, for which I am grateful. Linh, thanks for everything. Your personality (and taste in restaurants) made my time in the lab with you so much fun. Thank you Julia for your seemingly endless knowledge of laboratory techniques. Thomas, it was great having you and your knowledge in the lab, I will never forget you or your medical horror stories. Fang, your unique and stimulating ideas bring interest to every conversation we share, inside the lab or out. Thank you Shivani for your patience with me and my non-existent coding abilities. Carmelle, Madonna, and Renu, the drive and initiative that you all possess are an inspiration to me. Richard, thank you for the laughs. Last, but by no means least, thank you to Ekaterina. My PhD experience would not have been nearly as wonderful without your presence. Our constant conversations, no matter the subject, made every day in the lab better. I can only hope to be as smart, diligent, and strong as you are one day.

To my parents, Joe and Diane Savio, there aren’t enough words in the dictionary to thank you for everything you have given me in life. I consider myself so incredibly lucky to have been raised by such hard-working, dedicated, generous people. Thank you for allowing me the freedom to choose my own path through life. Your constant support, in all of its forms, means so much to me. Anthony, thank you for becoming the best big brother a girl could ask for, even if you used to cheat at Monopoly. Marisa and Ashley, you have been my best friends for my entire life, and I would not be where I am today without the two of you. To the rest of my family and friends, thank you for all of the love and encouragement.

v

List of Contributions

Chapters 2 and 5 of this thesis have been published in peer-reviewed journals and are included in

the appendix. Chapters 3 and 4 have been submitted to peer-reviewed journals.

Chapter 2: Mismatch repair gene polymorphisms are significantly associated with DNA

methylation in colorectal cancer cases and controls

Reference: Andrea J. Savio, Mathieu Lemire, Miralem Mrkonjic, Steven Gallinger, Brent W.

Zanke, Thomas J. Hudson, Bharati Bapat. MLH1 region polymorphisms show a significant

association with CpG island shore methylation in a large cohort of healthy individuals. PLoS

One. 2012;7(12):e51531.

Author contributions were as follows: The work in this chapter was primarily contributed by

Andrea J. Savio. Biostatistician Mathieu Lemire performed normalization of and curated

methylation array data, provided statistical expertise, and created Figure 2.3. Miralem Mrkonjic,

a former PhD student, performed SNP genotyping. Steven Gallinger provided DNA samples and

relevant clinical and pathological data. Brent W. Zanke and Thomas J. Hudson performed

GWAS for the ARCTIC group. Bharati Bapat supervised the project (including development,

experimental design, analysis and interpretation of data) and critically reviewed the manuscript.

Chapter 3: The dynamic DNA methylation landscape of the mutL homolog 1 CpG shore is

altered by MLH1-93G>A polymorphism in normal tissues and colorectal cancer

Chapter 3 has been submitted to a peer-reviewed journal. Authors: Andrea J. Savio, Miralem

Mrkonjic, Mathieu Lemire, Steven Gallinger, Julia Knight, Bharati Bapat.


Andrea J. Savio. Miralem Mrkonjic, a former PhD student, performed genotyping of SNP

rs1800734 and assessed MLH1 CpG island methylation. Mathieu Lemire and Julia Knight

provided statistical expertise. Steven Gallinger provided DNA samples and relevant clinical and

pathological data. Bharati Bapat supervised the project (including development, experimental

design, analysis and interpretation of data) and critically reviewed the manuscript.

vi

Chapter 4: Exploration of DNA methylation, histone modifications, and transcription

factors of the MLH1 CpG island and shore in colorectal cancer cell lines.

Chapter 4 has been submitted to a peer-reviewed journal. Authors: Andrea J. Savio & Bharati

Bapat.


Andrea J. Savio. Bharati Bapat supervised the project (including development, experimental

design, analysis and interpretation of data) and critically reviewed the manuscript.

Chapter 5: Promoter methylation of the Wnt signaling gene ITF2, but not APC, is

associated with microsatellite instability in two populations of colorectal cancer patients.

Reference: Andrea J. Savio, Darshana Daftary, Elizabeth Dicks, Daniel D. Buchanan, Patrick S.

Parfrey, Joanne P. Young, Daniel Weisenberger, Roger C. Green, Steven Gallinger, John R.

McLaughlin, Julia A. Knight, Bharati Bapat. Promoter methylation of ITF2, but not APC, is

associated with microsatellite instability in two populations of colorectal cancer patients. BMC

Cancer. 2016;16:113.


Andrea J. Savio. Darshana Daftary and Elizabeth Dicks participated in patient recruitment and

coordination. Daniel D. Buchanan and Joanne P. Young performed mutation analysis. Daniel

Weisenberger performed CIMP analysis. Patrick S. Parfrey, Roger C. Green, Steven Gallinger,

and John R. McLaughlin participated in study design and patient recruitment. Julia A. Knight

participated in study design, statistical analysis, and interpretation of data. Bharati Bapat

supervised the project (including development, experimental design, analysis and interpretation

of data) and critically reviewed the manuscript.

vii

Table of Contents

Acknowledgments .......................................................................................................................... iv!

List of Contributions ...................................................................................................................... iv!

Table of Contents .......................................................................................................................... vii!

List of Tables ................................................................................................................................. xii!

List of Figures .............................................................................................................................. xiv!

List of Appendices ........................................................................................................................ xv!

List of Abbreviations .................................................................................................................... xvi!

Chapter 1 General Introduction ....................................................................................................... 1!

1.1! Colorectal Cancer ................................................................................................................ 1!

1.1.1! Anatomy and physiology of the colorectum ........................................................... 1!

1.1.2! Natural history and epidemiology of colorectal cancer .......................................... 1!

1.1.3! Risk Factors ............................................................................................................. 2!

1.1.4! Screening, staging, and prognosis ........................................................................... 4!

1.2! Molecular pathways in CRC ............................................................................................. 10!

1.2.1! Chromosomal instability ....................................................................................... 12!

1.2.2! Microsatellite instability ........................................................................................ 17!

1.2.3! Epigenetic instability ............................................................................................. 19

1.2.3.1 Epigenetics………………………….…………………………………..19

1.2.3.2 CpG island methylator phenotype.……………………………………..24

1.3! Hereditary syndromes, pathology, and presentations ........................................................ 24!

1.3.1! Lynch syndrome .................................................................................................... 25!

1.3.2! Familial adenomatous polyposis ........................................................................... 28!

1.3.3! Other polyposis syndromes ................................................................................... 28!

1.4! Mismatch Repair (MMR) .................................................................................................. 29!

viii

1.4.1! Sources of DNA nucleotide mismatches ............................................................... 29!

1.4.2! Mismatch repair system overview ......................................................................... 31

1.4.2.1 Replication error repair………………………….……………….……..31

1.4.2.2 MMR in DNA damage signaling, cell cycle arrest, and apoptosis……..36

1.4.3! Other roles of MMR .............................................................................................. 37

1.4.3.1 Fidelity of genetic recombination…………………...…………………37

1.4.3.2 Generation of immunoglobulin diversity………...……………………37

1.4.4! Knockout mouse models of MMR genes .............................................................. 38!

1.5! Genetic variation ............................................................................................................... 41!

1.5.1! Polymorphisms ...................................................................................................... 41!

1.5.2! Utility and roles of SNPs in genetic epidemiology ............................................... 42!

1.5.3! Identification of low-penetrance alleles in CRC ................................................... 42!

1.6! Rationale ............................................................................................................................ 46

1.7 Hypothesis and objectives…………………………………………………………….…46

Chapter 2 Mismatch repair gene polymorphisms are significantly associated with DNA

methylation in colorectal cancer cases and controls…………………………………………..…48

2.1 Summary………………………………………………………………………………...48

2.2 Introduction……………………………………………………………………………...49

2.3 Materials and methods…………………………………………………………………..51

2.3.1 Study subjects………………………………………………………………………51

2.3.2 Single nucleotide polymorphism genotyping………………………………………52

2.3.3 Methylation microarray……………………….……………………………………52

2.3.4 Selection of CpG sites………………………………………………….…………...53

2.3.5 Adjusting for cell type proportions…………………………………………………56

2.3.6 Statistics…………………………………………………………………………….56

ix

2.4 Results…………………………………………………………………………………...57

2.4.1 PBMC methylation differences among MLH1-region SNP genotypes…………….59

2.4.2 Age-related decrease in methylation at the MLH1 shore…………………………...62

2.4.3 PBMC methylation differences among males and females at the MLH1 region.…..64

2.4.4 PBMC methylation differences among CRC cases and controls at MLH1………...66

2.4.5 No association between MSI status and DNA methylation in PBMCs………..…...69

2.4.6 Methylation levels of the MLH1 CpG island and shore in PBMCs………….……..69

2.4.7 Association between MSH2/MSH6 SNPs and DNA methylation………………….69

2.4.8 Association between methylation, age, and sex at the MSH2/MSH6 region…….…70

2.4.9 Differential PBMC DNA methylation between CRC cases and controls at the

MSH2/MSH6 region………………………………………………………………………70

2.5 Discussion…………………………………………………………………………..….....70

Chapter 3 The dynamic DNA methylation landscape of the mutL homolog 1 shore is altered by

MLH1-93G>A polymorphism in normal tissues and colorectal cancer…………………………76

3.1 Summary………………………………………………………………………..………...76

3.2 Introduction…………………………………………………………………………….....77

3.3 Materials and methods……………………………………………………………………80

3.3.1 Study subjects………………………………………………………………………80

3.3.2 Single nucleotide polymorphism genotyping………………………………………80

3.3.3 Microsatellite instability status………………………………………………..……81

3.3.4 MethyLight…………………………………………………………………………81

3.3.5 Bisulphite sequencing………………………………………………………………82

3.3.6 Statistics…………………………………………………………………………….84

3.4 Results………………………………………………………………………………….....84

3.4.1 The MLH1 shore is hypermethylated in tumour DNA……………………………..84

3.4.2 MLH1 shore methylation in tumours is correlated with methylation in PMBC

and normal colorectal DNA……………………………...………………………………87

3.4.3 The MLH1 shore is hypomethylated in variant SNP carriers in normal DNA……..89

3.4.4 Tumour hypermethylation at the MLH1 shore is driven by variant SNP allele…….89

3.4.5 MLH1 shore is hypomethylated in metastatic CRC, not associated with CpG island

methylation or MSI……………………………………………………………………….90

x

3.4.6 Bisulphite sequencing confirms SNP-associated hypomethylation of MLH1 shore in

normal colorectal DNA…………………………………………………………………...91

3.4.7 Examination of CpG island and shore methylation in normal and tumour tissues…95

3.5 Discussion…………………………………………………………………………..….....97

3.6 Acknowledgements……………………………………………………………………...101

Chapter 4 Exploration of DNA methylation, histone modifications, and transcription factors of

the MLH1 CpG island and shore in colorectal cancer cell lines……………………………..…102

4.1 Summary………………………………………………………………………………...102

4.2 Introduction……………………………………………………………………………...103

4.3 Materials and methods………………………………………………………………..…104

4.3.1 Cell lines…………………………………………………………………………..104

4.3.2 Cell line genotyping……………..……………..……………..…………………...104

4.3.3 MethyLight………………………………………………………………………..107

4.3.4 Bisulphite sequencing……………..……………..……………..………...……….107

4.3.5 Selection of candidate proteins for chromatin immunoprecipitation………….….108

4.3.6 Chromatin immunoprecipitation……………..……………..……………………..108

4.3.7 Confirmation of genotype from ChIP experiments……………..………………...109

4.3.8 Statistics…………………………………………………………………………...109

4.4 Results…………………………………………………………………………………...109

4.4.1 Genotype and methylation status of CRC cell lines……………..………………..109

4.4.2 Sequence-specific binding of AP4……………..……………..…………………...113

4.4.3 Histone modifications and Pol II are consistent across genotypes of rs1800734…117

4.4.4 Lack of CTCF at MLH1 CpG island and shore……………..…………………….122

4.5 Discussion……………..……………..……………..……………..……………..……...124

4.6 Acknowledgements……………..……………..……………..……………..…………...129

Chapter 5 Promoter methylation of the Wnt signaling gene ITF2, but not APC, is associated with

microsatellite instability in two populations of colorectal cancer patients……………………..131

5.1 Summary………………………………………………………………………………...131

5.2 Introduction……………………………………………………………………………...132

5.3 Materials and methods…………………………………………………………………..133

xi

5.3.1 Study subjects……………………………………………………………………..133

5.3.2 Molecular analysis………………………………………………………………...137

5.3.3 MethyLight………………………………………………………………………..137

5.3.4 Statistics…………………………………………………………………………...138

5.4 Results…………………………………………………………………………………...138

5.4.1 ITF2 promoter methylation in CRC tumours and normal colorectal mucosa…….138

5.4.2 APC promoter methylation in CRC tumours and normal colorectal mucosa……..139

5.4.3 ITF2 methylation and clinicopathological features, including MSI subtype, in

two distinct CRC cohorts..………………...……..……………..……………..………...140

5.4.4 APC methylation and clinicopathological features, including MSI subtype, in

two distinct CRC cohorts..………………...……..……………..……………..………...143

5.5 Discussion..………………...……..……………..……………..………..........................146

Chapter 6 Discussion and Future Directions ................................................................................. 52

6.1 Further exploration of MLH1-93G>A SNP……………………………………………..152

6.2 Other genes and/or polymorphisms associated with MLH1-93G>A……………………155

6.3 Differential methylation of Wnt signaling genes………………………………….….…157

6.4 The future of epigenetics in the clinic……………………………………………….…..158

References ................................................................................................................................... 163!

Appendices ................................................................................................................................. 213!

xii

List of Tables

Chapter 1

Table 1.1 Screening tests for the detection of colorectal cancer…………………………………7

Table 1.2 Tumour stage classification according to the American Joint Committee on Cancer

cancer staging manual, 7th Edition…………………………………………………….8

Table 1.3 Tumour staging defined by American Joint Committee on Cancer Staging Manual,

7th Edition……………………………………………………………………………...9

Table 1.4 Histone H3 writers, erasers, and readers……………………………………………..23

Table 1.5 Clinical criteria for Lynch syndrome………………………………………………...27

Table 1.6 Mismatch repair proteins in E. coli and their eukaryotic homologues………………34

Table 1.7 Phenotypes of MMR-deficient knockout mice………………………………………40

Table 1.8 Risk loci and SNPs for colorectal cancer…………………….………………………45

Chapter 2

Table 2.1 Characteristics of study population…………………………………………………..58

Table 2.2 Methylation between SNP genotypes in controls by ANOVA………………………60

Table 2.3 Methylation between SNP genotypes in CRC cases by ANOVA…………………...61

Table 2.4 Correlation between age and MLH1 shore methylation……………………………..63

Table 2.5 Association between sex and methylation by logistic regression………..…………..65

Table 2.6 Association between MLH1 shore methylation and CRC risk………………………67

Chapter 3

Table 3.1 Distribution of clinicopathological features in primary colorectal carcinomas

from Ontario, including distribution among genotypes of rs1800734……………….85

Table 3.2 Pearson correlation indicating association between methylation in MLH1 shore in

PBMC, normal colorectal, and tumour DNA from the same cases……….…………88

Chapter 4

Table 4.1 Primers used for PCR reactions…………………………………………………....106

Table 4.2 MLH1 promoter SNP genotype and methylation of its CpG island and shore in

CRC cell lines………………………………………………………………………111

Table 4.3 Selected genetic and epigenetic features of colorectal cancer cell lines utilized in

xiii

chromatin immunoprecipitation assays……………………………………………..114

Chapter 5

Table 5.1 Clinicopathological features of primary colorectal carcinomas of patients from

Ontario and Newfoundland…………………………………………………………135

Table 5.2 Associations between ITF2 methylation and clinicopathological features in tumour

DNA.………………………………………………………..............................……141

Table 5.3 Associations between APC methylation and clinicopathological features in tumour

DNA. .……………………………………………………................................……144

xiv

List of Figures

Chapter 1

Figure 1.1 Instability pathways of colorectal cancer…………………………………………...11

Figure 1.2 Overview of Wnt signaling pathway.…..…………………………………………...16

Figure 1.3 Overview of the main epigenetic regulatory mechanisms………………………….22

Figure 1.4 Simplified overview of mammalian mismatch repair………………………………35

Chapter 2

Figure 2.1 Region selected for DNA methylation analysis on chromosome 3…………..……..54

Figure 2.2 Region selected for DNA methylation analysis on chromosome 2…………………55

Figure 2.3 Locations of CpG sites and methylation in CRC cases and controls……………….68

Chapter 3

Figure 3.1 MLH1 gene region, including its CpG shore, CpG island, and SNPs of interest…...79

Figure 3.2 Mean percent methylated reference (PMR) of the MLH1 shore……………………86

Figure 3.3 Bisulphite sequencing results for six CpG sites of the MLH1 shore in six normal

colorectal tissue samples and matched colorectal tumours from the same cases…..93

Figure 3.4 Schematic model of DNA methylation at the MLH1 CpG island and shore………..96

Chapter 4

Figure 4.1 Bisulphite sequencing of MLH1 CpG island and shore in colorectal cancer cell

lines………………………………………………………………………………...112

Figure 4.2 ChIP analysis of AP4 occupancy at the MLH1 CpG island and shore region…….115

Figure 4.3 Representative chromatogram traces from AP4 chromatin immunoprecipitation

pull-down in heterozygous SNU-C2B cells at rs1800734………………………...116

Figure 4.4 ChIP analysis of histone modifications at the MLH1 CpG island and shore……...119

Figure 4.5 ChIP analysis of histone modifications at the MLH1 CpG island and shore

normalized to histone H3………………………………………………………….120

Figure 4.6 ChIP analysis of Pol II and CTCF at the MLH1 CpG island and shore…………...121

Figure 4.7 ChIP analysis of controls H3 and IgG at the MLH1 CpG island and shore……….123

Figure 4.8 Schematic models of transcription factors and epigenetic regulation at SNP

rs1800734………………………………………………………………………….130

xv

Chapter 5

Figure 5.1 Schematic of methylation events occurring in MSS and MSI CRC………………151

Chapter 6

Figure 6.1 The past, present, and future of epigenetics in colorectal cancer………………….162

xvi

List of Appendices

Table A1 Methylation between SNP genotypes of rs1800734, rs749072, and rs13098279

in controls by ANOVA………….………….…………..………….………….……213

Table A2 Methylation between SNP genotypes of rs1800734, rs749072, and rs13098279

in CRC cases by ANOVA.………….…………….………….………….………….218

Table A3 Correlation between age and DNA methylation at MLH1 region………………….223

Table A4 Association between sex and methylation by logistic regression……..……………224

Table A5 Logistic regression analysis for association with methylation between CRC cases

and controls.………….………….………….………….………………..………….227

Table A6 Methylation between SNP genotypes of MSH2-118T>C in controls by ANOVA...229

Table A7 Methylation between SNP genotypes of MSH2-118T>C in CRC cases by

ANOVA.…………………….………….………….………….……………………232

Table A8 Methylation between SNP genotypes of MSH6-159C>T in controls by

ANOVA…………………………………………………………………………….235

Table A9 Methylation between SNP genotype of MSH6-159C>T in CRC cases by

ANOVA…………………………………………………………………………….238

Table A10 Correlation between age and DNA methylation at MSH2 and MSH6…………….241

Table A11 Association between sex and methylation by logistic regression at

MSH2- MSH6 region………………………………………………………………244

Table A12 Logistic regression analysis for association with methylation between CRC

cases and controls…………………………………………………………………247

Publication 1 MLH1 region polymorphisms show a significant association with CpG

island shore methylation in a large cohort of healthy individuals………….......251

Publication 2 Promoter methylation of ITF2, but not APC, is associated with

Microsatellite instability in two populations of colorectal cancer patients……..260

xvii

List of Abbreviations

5-aza - 5-aza-2’-deoxycytidine

5-FU – 5-fluorouracil

5hmC – 5-hydroxymethylcytosine

95% CI – 95% confidence interval

ACF – aberrant crypt focus

AFAP – attenuated familial adenomatous polyposis

ANOVA – analysis of variance

AP4 – transcription factor AP-4 (activating enhancer binding protein 4)

APC – adenomatous polyposis coli

APC/C – anaphase-promoting complex/C

ARCTIC – Assessment of Risk of Colorectal Tumours in Canada

ATF2 – activating transcription factor 2

ATP – adenosine triphosphate

AXIN1 – axis inhibition protein 1

AXIN2 – axis inhibition protein 2

βTRCP – β-transducin-repeat-containing protein

BER – base excision repair

bp – base pair

BRAF – B-Raf proto-oncogene, serine/threonine kinase

CH3 – methyl group

xviii

ChIP – chromatin immunoprecipitation

CIMP – CpG island methylator phenotype

CIN – chromosomal instability

CK1 – casein kinase 1

CNV – copy number variation

CRC – colorectal cancer

CT – computed tomography

CTC – circulating tumour cell

CTCF – CCCTC-binding factor

CTNNB1 – catenin beta 1

dam – deoxyadenine methylase

dGTP – deoxyguanosine triphosphate

DKK1 – dickkopf WNT signaling pathway inhibitor 1

DNA – deoxyribonucleic acid

DNMT1 – DNA methyltransferase 1

DNMT3A – DNA methyltransferase 3 alpha

DNMT3B – DNA methyltransferase 3 beta

dNTP – deoxynucleotide triphosphate

EMSA – electrophoretic mobility shift assay

EPCAM – epithelial cell adhesion molecule

xix

EPM2AIP1 – EPM2A interacting protein 1

EXO1 – exonuclease 1

FAP – familial adenomatous polyposis

FFPE – formalin-fixed paraffin embedded

FIT – fecal immunochemical test

gFOBT – guaiac fecal occult blood test

GI – gastrointestinal

GSK3β – glycogen synthase kinase-3β

GWAS – genome-wide association studies

H2A – histone H2A

H2B – histone H2B

H3 – histone 3

H4 – histone 4

HAT – histone acetyltransferase

HDAC – histone deacetylase

HMT – histone methyltransferase

HNPCC – hereditary nonpolyposis colorectal cancer

IDL – insertion/deletion loop

ITF2 – immunoglobulin transcription factor 2

JPS – juvenile polyposis syndrome

xx

KCNK12 – potassium two pore domain channel subfamily K member 12

KDM – histone demethylase

KRAS – KRAS proto-oncogene, GTPase

LD – linkage disequilibrium

LEF – lymphoid enhancer factor

LRRFIP2 – LRR binding FLII interacting protein 2

MAF – minor allele frequency

MAP – MUTYH-associated polyposis

MGMT – O-6-methylguanine-DNA methyltransferase

MLH1 – mutL homolog 1

MMR – mismatch repair

MNNG – N-methyl-N’-nitro-N-nitrosoguanidine

MRC – magnetic resonance colonoscopy

MSH2 – mutS homolog 2





MSI – miscrosatellite instability

MSI-H – microsatellite instability high

xxi

MSI-L – microsatellite instability low

MSS – microsatellite stable

mut - mutator

MUTYH – mutY DNA glycosylase

MYC – v-myc avian myelocytomatosis viral oncogene homolog

NFCCR – Newfoundland Familial Colorectal Cancer Registry

OFCCR – Ontario Familial Colorectal Cancer Registry

PBMC – peripheral blood mononuclear cell

PCNA – proliferating cell nuclear antigen

PCR – polymerase chain reaction

PJS – Peutz-Jeghers syndrome

PMS1 – PMS1 homolog 1, mismatch repair system component

PMS2 – PMS1 homolog 2, mismatch repair system component

Pol – polymerase

PRC2 – Polycomb repressive complex 2

PTEN – phosphatase and tensin homolog

QUMA – Quantification tool for Methylation Analysis

RFC – replication factor C

RNA – ribonucleic acid

ROS – reactive oxygen species

xxii

RPA – replication protein A

SD – standard deviation

SFRP1 – secreted frizzled related protein 1

SNP – single nucleotide polymorphism

ssDNA – single-stranded DNA

TCAG – The Centre for Applied Genomics

TCF – T-cell factor

TCF7L2 – transcription factor 7 like 2

TCGA – The Cancer Genome Atlas

TET – ten-eleven translocation

TIS – translation initiation site

TNM – tumour-node-metastasis

TP53 – tumor protein p53

TS – thymidylate synthetase

TSS – transcription start site

U – uracil

WIF1 – WNT inhibitory factor 1

Wnt – Wingless-type MMTV integration site family

1

Chapter 1 General Introduction

1.1 Colorectal Cancer

1.1.1 Anatomy and physiology of the colorectum

The colorectum is a component of the digestive system that functions to absorb water, sodium,

and chloride, while secreting potassium and bicarbonate. The colorectum is divided into two

regions with different embryological origins, the proximal and distal colon, defined as regions

proximal and distal to the splenic flexure (1). The proximal colon is derived from the midgut and

consists of the appendix, caecum, ascending colon, hepatic flexure, and transverse colon while

the distal colon is derived from the hindgut and consists of the splenic flexure, descending colon,

sigmoid colon, and rectum.

The colonic mucosa lining the lumen of the colorectum contains infoldings approximately fifty

cells deep, which form crypts. The bases of the crypts contain a stem cell compartment; these

cells are marked by the receptor leucine-rich repeat-containing G-protein coupled receptor 5

(LGR5) and divide approximately once a day (2–4). The area of the crypt above the stem cells is

occupied by transit-amplifying cells, an intermediate progenitor pool, which divide twice a day

(5,6). Cells migrate through the transit-amplifying zone to the lumen where they become

terminally differentiated into absorptive enterocytes or secretory goblet and enteroendocrine cells

(6,7). Colonic cells undergo rapid turnover and the entire process takes five to seven days,

making the intestinal epithelium one of the most rapidly renewing tissues in the body (5,8).

1.1.2 Natural history and epidemiology of colorectal cancer

Colorectal cancer (CRC) results from the accumulation of genetic and epigenetic alterations, as

well as biochemical changes in the macro- and microenvironment, which lead to the

transformation of normal colonic epithelium into benign adenomas and eventually

adenocarcinoma, known as the adenoma-carcinoma sequence (9,10). The earliest identifiable

lesion of the colonic mucosa is the aberrant crypt focus (ACF), characterized by histological

features including darker staining, raised appearance, and crypt size at least three times larger

than adjacent normal mucosa (11). Individual cells in ACF are morphologically normal, however

the accumulation of cells within crypts leads to crowding and mucosal folding. A subset of these

2

microscopic mucosal abnormalities are believed to be the precursors of CRC (12,13). Some of

these lesions harbour mutations in APC or KRAS and have increased expression of proliferative

markers (14–16).

Polyps are benign gland-forming mucosal projections, which can be categorized as adenomatous

or serrated. Adenomatous polyps, or adenomas, may be tubular, tubulovillous, or villous

adenomas (17,18). Adenomas that are larger than 1 cm in size, predominantly villous, or

containing high grade dysplasia are considered advanced adenomas (18). The serrated pathway is

distinct from the conventional adenoma-carcinoma pathway, characterized by serrated

architecture of the epithelial compartment (17,19). Hyperplastic polyps are the most common

type of serrated lesion and are generally less than 5 mm in size, found in the distal colon, and

rarely progress to CRC (20). Serrated adenomas are commonly mutated at BRAF and show high

levels of CpG island methylator phenotype (CIMP, further discussed in Section 1.2.3.2) (21).

Promoter methylation of mutL homolog 1 (MLH1) is a late event in the serrated pathway and this

event may mark the transition from serrated adenoma to CRC (22,23).

Approximately 1.36 million cases of CRC are diagnosed each year worldwide, however, the

overall rates are higher in developed countries compared to developing nations (24–26). In

Canada, it is the second leading cause of cancer-related deaths affecting 1 in 14 males and 1 in

16 females during their lifetime (27). An estimated 25,100 cases of CRC and 9,300 deaths were

attributed to CRC in Canada in 2016. Among the provinces and territories of Canada the

incidence and mortality rates of CRC differ, with Atlantic provinces having the highest rates.

Newfoundland and Labrador in particular have nearly twice the mortality rate of CRC compared

to Ontario. Ontario has the third lowest incidence and lowest mortality rates of CRC of the

Canadian provinces. The estimated five-year net survival for all of Canada is 64% (27).

1.1.3 Risk Factors

Defined genetic predispositions account for approximately 5-10% of CRCs, leaving a large

proportion of CRC risk that may be contributed to by the environment, familial predisposition

due to germline variants or mutations in currently unidentified genes, or a combination of genetic

and environmental factors (28–31). The greatest risk factor for CRC is family history (32). An

individual’s risk of CRC more than doubles if one first-degree relative also has CRC, and

quadruples if two first-degree relatives are affected by CRC (32,33). Strong evidence also exists

3

for the environmental contribution to CRC. Cancer incidence and mortality rates can change

within the span of one generation in migrants. This has been demonstrated for European, East

Asian, and Southeast Asian migrants after 30 years of residence in Australia (34,35). Migrants

born in Europe, Asia, and Africa incur increased risk of CRC after 30 years of residence in Israel

(36). Additionally, migrants from the former Soviet Union to Israel have also been shown to

incur higher incidence rates of colon, but not rectal, cancer within 20 years (37).

Diet has been show to play an important role in CRC etiology through epidemiological studies.

Diets high in red and/or processed meat have been shown to increase risk of CRC (38–40). The

process of cooking meat at high heat forms heterocyclic aromatic amines and polycyclic

aromatic hydrocarbons in meat, which, in addition to nitroso compounds in processed meats and

heme iron found in red meats, are proposed to form DNA adducts, induce double stranded breaks

in the DNA, and stimulate epithelial proliferation (39,41,42). Diets rich in fats have not

conclusively shown an association with CRC, but may be associated with risk of colon tumours

with activating KRAS mutations (43,44). Diets rich in fruits, vegetables and fibre have been

shown to lower CRC risk (38,40,43). Vitamin D, calcium, and folate have also demonstrated

protective effects against CRC (40,43,45). Folate, and its synthetic form folic acid, are involved

in one-carbon metabolism which provides one-carbon units for nucleotide synthesis and methyl

groups for DNA methylation (46). Vitamin B6, vitamin B12, riboflavin, choline, and betaine are

also required for proper DNA methylation and may be associated with reduced risk of CRC (46–

49).

High alcohol consumption is associated with increased risk of CRC (40,43,50–52). The role of

alcohol or ethanol in carcinogenesis is not fully understood, but it may act as a co-carcinogen,

alter metabolic pathways and cell structures, and increase cell proliferation (53). Malnutrition in

chronic alcohol users has been shown to lead to deficiency in folate and other factors in one-

carbon metabolism leading to DNA methylation alterations (54,55). Cigarette smoking is

associated with elevated CRC risk, and potentially more so for the microsatellite instability and

CpG island methylator phenotype subtypes of CRC (40,52,56,57).

Physical activity and healthy body mass have been shown to be associated with reduced CRC

risk. Individuals who perform high levels of activity throughout their lives have significantly

lower risk of CRC than inactive individuals (38,40). Individuals with high body mass index may

4

also have a higher risk of CRC (38,40). Body fat distribution has been linked to increased risk of

CRC in men with gains of abdominal adiposity, even without gains in total weight (58).

Another risk factor for CRC is history of inflammatory bowel disease including ulcerative colitis

and Crohn’s disease (38,45). This increased risk is believed to be due to persistent inflammation

of the colon, however, this risk is attenuated by regular colonoscopy screening (59,60).

1.1.4 Screening, staging, and prognosis

Several screening strategies exist for asymptomatic persons at average risk for developing CRC,

summarized in Table 1.1. These strategies have allowed for earlier detection at more curable

stages and have resulted in reduced mortality rates (61–64). The guaiac fecal occult blood test

(gFOBT) is a non-invasive procedure able to detect small amounts of blood in stool samples that

can be performed at home. In randomized trials annual or biennial gFOBT reduced CRC

mortality by 15-33% (65–68). However, gFOBT has low sensitivity for CRC (25-38%) and

advanced adenomas (16-31%) (69). Another limitation is that this test detects the peroxidase

activity of heme so is not specific for human blood and may detect dietary sources of blood

(70,71). Further, blood from any source throughout the gastrointestinal tract may result in a

positive test, not only from the colon (72). Despite these limitations, clinical trials have

demonstrated that individuals with positive occult blood tests have three to four times higher risk

for developing CRC compared to those with negative tests (73). The fecal immunochemical test

(FIT) improves upon gFOBT by using antibodies against human hemoglobin that are more

specific for human blood. FIT has demonstrated higher sensitivity than gFOBT for CRC (61-

91%) and advanced adenomas (27-67%) (69). FIT requires at-home testing on one day without

dietary restrictions, compared to gFOBT in which measurements are required over a three day

period. Thus, several studies have shown higher participation rates for FIT than gFOBT (74,75).

Examinations of the colon and rectum are another method of screening utilized for CRC

detection. Virtual colonoscopy, including computed tomography (CT) and magnetic resonance

colonoscopy (MRC), are used to generate two-dimensional and three-dimensional images of the

colon and rectum. While this is less invasive than endoscopic methods, any positive results of

virtual colonoscopy must be followed up by endoscopy (69). Limitations of CT include potential

harm from ionizing radiation, while MRC is associated with higher costs and limited availability

(69,76). These methods lack sensitivity and specificity, especially for polyps smaller than 6 mm

5

(73). Endoscopy to detect colorectal adenomas and carcinomas includes flexible sigmoidoscopy

and colonoscopy. Flexible sigmoidoscopy is a technique allowing examination of the distal

portion of the colon. Although this method cannot detect proximal lesions, it does not require

sedation and requires less preparation for patients. Polypectomy can be performed during

flexible sigmoidoscopy. Randomized controlled trials demonstrate that screening by flexible

sigmoidoscopy can reduce CRC incidence by 18-23% and CRC mortality by 22-31% (77–79).

While the risk of complications is possible, a large survey of 109,534 flexible sigmoidoscopies

resulted in hospitalizations in only 0.02% of patients (80). The gold standard of examination is

colonoscopy, which allows for full colonic examination combined with polypectomy in a single

session. Studies have shown that colonoscopy reduces CRC incidence by 67-77% and CRC

mortality by 31-65% (81–84). This method has the highest level of sensitivity and specificity of

any screening method and is the final assessment step of any current screening program (85,86).

Limitations include operator variability among endoscopists and the risk of complications (87).

Also, colonoscopy is invasive and requires rigorous preparation, which are factors in low patient

acceptance (88).

Recommended screening varies between nations. The Canadian Task Force on Preventive Health

Care gives a strong recommendation for gFOBT or FIT screening every two years for adults

aged 60 to 74 or flexible sigmoidoscopy every 10 years (85). There is a weak recommendation

for the same screening regimen in individuals aged 50 to 59. Strong recommendations are made

when the Canadian Task Force on Preventive Health Care is confident that the desirable effects

of an intervention outweigh its undesirable effects while weak recommendations are made when

the desirable effects probably outweigh the undesirable effects but appreciable certainty exists.

These recommendations are for asymptomatic individuals without family history of CRC.

Screening differs in individuals with family history of CRC, as they are at higher risk of disease.

Recommended screening is more stringent and varies based on whether an individual is positive

for hereditary syndromes, and on the number, age and closeness of relatives who have developed

cancer, which will be discussed further in Section 1.3.

The stage at which CRC is diagnosed determines the treatment options and is also the strongest

predictor of survival. The tumour-node-metastasis (TNM) staging system was developed by the

American Joint Committee on Cancer and International Union Against Cancer in order to

standardize staging allowing for uniform evaluation across specialties and nations, described in

6

Tables 1.2 and 1.3 (89). T refers to the local extent of untreated primary tumour at the time of

diagnosis and initial workup. N refers to the status of the regional lymph nodes. M refers to

distant metastatic disease. Pathological classification is based on gross/microscopic examination

of the resected specimen of an untreated primary tumour and clinical classification is based on a

variety of techniques including physical examination, radiologic imaging, endoscopy, biopsy,

and surgical exploration (90). Dukes staging classification system relies on pathological

classification of disease stages and is not as useful as it cannot be used to preoperatively evaluate

the patient (90). The five-year disease-specific survival rate of CRC is around 90% for stage I

CRCs, 85% for stage II, 70% for stage III, and 10-15% for stage IV disease (91,92).

Histological grade is another feature of CRCs that can be assessed. Tumours are graded into

categories based on degree of gland-like structures and resemblance to original tissue. Grade 1

tumours are well differentiated, Grade 2 tumours are moderately differentiated, Grade 3 tumours

are poorly differentiated, and Grade 4 tumours are undifferentiated. Several limitations of the

grading system exist. For example, approximately 70% of CRCs are diagnosed as Grade 2, or

moderately differentiated (17). Additionally, tumours with microsatellite instability may show

high grade morphology but behave as low grade tumours (17). Nevertheless, patient prognosis

correlates with both higher stage and higher grade, and grade also correlates to TNM staging

(89,93,94).

7

Table 1.1 Screening tests for the detection of colorectal cancer. References: (69–71,73,76–

87,95–97).

Test Sensitivity Advantages Limitations Guaiac fecal occult blood test

Cancer: 25-38% Advanced adenoma: 16-31%

- Low initial cost - Can be performed at home

- Not specific for human hemoglobin - Dietary and medical restrictions - Repeat annual testing recommended

Fecal immunochemical test


- Specific for human hemoglobin - Can be performed at home - No dietary or medical restrictions

- Variation in performance

Computed tomography colonography


- High sensitivity for detection of lesions >6mm in diameter - Less invasive than endoscopy

- Detection of polyps <6mm in diameter uncertain - Requires bowel preparation and special resources - Radiation exposure

Magnetic resonance colonoscopy


- Less invasive than endoscopy

- High cost - Limited availability

Sigmoidoscopy Cancer: >95% in distal colon Advanced adenoma: 70%

- Office-based - No sedation required - 18-23% reduction in CRC incidence 22-31% reduction in CRC mortality

- Does not detect proximal colon cancer - Risk of complications

Colonoscopy Cancer: >95% Advanced adenoma: 88-98%

- 67-77% reduction in CRC incidence 31-65% reduction in CRC mortality

- High initial cost - Requires bowel preparation - Invasive - Risk of complications

8

Table 1.2 Tumour stage classification according to the American Joint Committee on

Cancer cancer staging manual, 7th Edition. Reference: (89).

TNM Stage Description TX Primary tumour cannot be assessed T0 No evidence of primary tumour Tis Carcinoma in situ; intraepithelial or invasion of lamina propria T1 Tumour invades submucosa T2 Tumour invades muscularis propria T3 Tumour invades through the muscularis propria into pericolorectal tissues T4a Tumour penetrates to the surface of the visceral peritoneum T4b Tumour directly invades or is adherent to other organs or structures NX Regional lymph nodes cannot be assessed N0 No regional lymph node metastasis N1 Metastasis in 1-3 regional lymph nodes N1a Metastasis in one regional lymph node N1b Metastasis in 2-3 regional lymph nodes N1c Tumour deposit(s) in the subserosa, mesentery, or nonperitonealized

pericolic or perirectal tissues without regional nodal metastasis N2 Metastasis in 4 or more regional lymph nodes N2a Metastasis in 4-6 regional lymph nodes N2b Metastasis in 7 or more regional lymph nodes M0 No distant metastasis M1 Distant metastasis M1a Metastasis confined to one organ or site M1b Metastases in more than one organ/site or the peritoneum

9

Table 1.3 Tumour staging defined by American Joint Committee on Cancer Staging

Manual, 7th Edition Reference: (89).

Stage T N M 0 Tis N0 M0 I T1

T2 N0 N0

M0 M0

IIA T3 N0 M0 IIB T4a N0 M0 IIC T4b N0 M0 IIIA T1-T2

T1 N1/N1c N2a

M0 M0

IIIB T3-T4a T2-T3 T1-T2

N1/N1c N2a N2b

M0 M0 M0

IIIC T4a T3-T4a T4b

N2a N2b N1-N2

M0 M0 M0

IVA Any T Any N M1a IVB Any T Any N M1b

10

1.2 Molecular pathways in CRC

CRC is a heterogeneous disease with a variety of diverse pathways involved in its initiation,

progress, invasion and metastasis. The sequential accumulation of mutations, which activate

oncogenes and disrupt tumour suppressor genes, combined with multiple cycles of clonal

selection and evolution facilitate carcinogenesis. On average, colorectal tumours contain

approximately 150 nonsynonymous mutations (98). Baseline mutation rates, estimated to be 10-9

mutations per base per generation do not account for the number of mutations observed (99).

Thus, cancer cells must acquire some form of genetic instability to facilitate this process

(100,101). In CRC, there are at least three major pathways contributing to instability:

chromosomal instability (CIN), microsatellite instability (MSI) and epigenetic instability, known

as the CpG island methylator phenotype (CIMP). The genetic and/or epigenetic alterations

common to the progression of these three pathways are demonstrated in Figure 1.1. These three

pathways are not mutually exclusive. 25% of MSI CRCs exhibit chromosomal instability and

12% of CIN tumours have high level of MSI (102). CIMP is most often found in MSI positive

CRCs, but up to 35% of CIMP positive tumours exhibit high levels of chromosomal

abnormalities (103–105). The biological relevance of these overlapping pathways and their role

in prognosis are not yet fully understood.

11

Figure 1.1 Instability pathways of colorectal cancer. Progression of colorectal cancers with

chromosomal instability (CIN), microsatellite instability (MSI), and CpG island methylator

phenotype (CIMP, epigenetic instability) from normal colonic epithelium to cancer. Several

representative genes altered in each pathway are shown for each stage of progression. Genes

indicated in CIN and MSI pathways are altered by mutation, except for MLH1, which may be

methylated. Genes in the CIMP pathway are methylated, except for BRAF, which is mutated.

Overlap between the three pathways exists. Adapted from Grady and Carethers,

Gastroenterology 2008;135(4):1079-99 and Walther et al. Nature Reviews Cancer

2009;9(7):489-499. [References (9,106)].

12

1.2.1 Chromosomal instability

Chromosomal instability (CIN) refers to an accelerated rate of chromosomal alterations resulting

in whole chromosomal gains or losses, inversions, deletions, duplications and translocations of

large chromosomal segments leading to large karyotypic variability from cell to cell (107,108).

CIN promotes carcinogenesis through loss of tumour suppressors and copy number gains of

oncogenes and is observed in approximately 85% of CRC (108,109).

The most commonly mutated gene in CRC is adenomatous polyposis coli (APC), located on

chromosome 5p21. One mechanism by which APC mutations contribute to CRC progression is

by over-activation of the Wingless-type MMTV integration site family (Wnt) signaling pathway

(110,111). Under normal conditions the Wnt signaling pathway is required for organ

development, cellular proliferation, morphology, motility, and cell fate determination (112–114).

An overview of the canonical Wnt signaling pathway is shown in Figure 1.2. When Wnt

signaling is activated by the presence of extracellular Wnt ligand, cytoplasmic β-catenin

accumulates, translocates to the nucleus, and drives the transcription of multiple genes by

binding to T-cell factor (TCF) and lymphoid enhancer factor (LEF) transcription factors

(115,116). Other factors involved in transcription include Pygopus and p300 (117). The role of

APC is to abolish the Wnt signaling cascade when appropriate by binding to a complex of

proteins including β-catenin, glycogen synthase kinase-3β (GSK3β), and casein kinase 1 (CK1)

on an axis inhibition protein 1 (AXIN1) scaffold. This is known as the β-catenin destruction

complex, which facilitates the phosphorylation of β-catenin by GSK3β and CK1 leading to its

proteasomal degradation, aided by β-transducin-repeat-containing protein (βTRCP).

Extracellular Wnt antagonists include dickkopf Wnt signaling pathway inhibitor 1 (DKK1), Wnt

inhibitory factor 1 (WIF-1), and secreted frizzled related proteins (SFRPs). Mutations to APC,

occurring in >75% of CRC tumours, increase transcriptional activity of β-catenin (111,118).

Alternatively, mutations can occur in β-catenin (CTNNB1) or other Wnt components to bring

about the same activation (119,120).

The second mechanism by which APC can lead to chromosomal instability is through its role in

mitosis. APC accumulates at the kinetochore during mitosis where it is able to bind to

microtubule-associated protein RP/EB family member 1 (MAPRE1), which is associated with

the ends of microtubules and centrosomes (7,121). Apc mutant murine cells form mitotic

13

spindles with a profusion of microtubules unable to properly connect with the kinetochores

(122,123). These mice demonstrate CIN and polyploidy. It is proposed that APC inactivation

creates a permissive state allowing tumours to tolerate development of aneuploidy and CIN

(9,124).

In addition to the potential role of APC in CIN and aneuploidy, our understanding of how and

why tumours develop abnormal numbers of chromosomes is incomplete. There are more than

100 genes known to regulate chromosome fidelity that are involved in centrosome and

microtubule formation, kinetochore structure and function, chromosome condensation, sister

chromatid cohesion, and cell cycle checkpoint control (124). Exactly which of these genes may

be responsible for CIN in CRC, in addition to APC, is still being investigated.

The major cell cycle control mechanism to ensure high fidelity of chromosome segregation is the

mitotic checkpoint. This checkpoint delays anaphase until all pairs of chromatids are properly

aligned at the metaphase plate. Failure of this process leads to incorrect segregation and unequal

distribution of chromosomes to daughter cells. A subset of CRCs experience deregulation of

MAD and BUB family proteins, two highly conversed components of the mitotic spindle

checkpoint that act as sensors and signal transducers (125,126). MAD and BUB inhibit the

anaphase-promoting complex/C (APC/C) and lead to cell cycle arrest. At anaphase, APC/C will

normally ubiquitinate securin leading to its eventual proteasomal degradation (127,128). The

function of securin is to interact with separins, a class of caspase-related proteases that regulate

cohesins, multi-protein complexes that create physical links maintaining sister chromatids during

metaphase (129,130). In addition to MAD and BUB, other important proteins in this cascade

show evidence of dysregulation in CRC. For example, polo-like kinases are involved in the auto

feedback loop regulating the securin-separin complex (131,132). Polo-like kinase 1 (PLK1) was

shown to be overexpressed in 73% of primary CRCs and PLK1 level was associated with tumour

invasion, lymph node involvement, and Dukes stage (133).

Another proposed mechanism causing CIN is abnormal centrosome number and/or function. The

centrosome plays a central role in chromosomal segregation during mitosis by serving as an

anchor for the reorganization of microtubules into the mitotic spindle apparatus. Extra

centrosomes result in unequal chromosome distribution (134). The centrosome-associated

Aurora kinase A (AURKA) is amplified and overexpressed in CRC cell lines (135,136).

14

Amplification of AURKA has been shown to induce aneuploidy in NIH/3T3 fibroblast cells

(137). Further evidence of the role of centrosome number in tumourigenesis was demonstrated

by transplanting extra centrosomes into the larval brain cells of flies, which induced metastatic

tumour formation (138).

Telomere stability may also play a role in CIN. Telomeres are the segments of DNA bound by a

capping structure at the ends of chromosomes, which act as a buffer to prevent loss of genomic

sequence during replication (139,140). They also prevent chromosomes from fusing at the ends

(139,140). A RNA primer is required for DNA replication, thus there will be a segment of

unreplicated DNA at the 5’ ends of chromosomes. This gradually leads to loss of telomeric

repeats and shortening of telomeres after each round of replication (141). Telomerase is the

specific enzyme able to maintain telomere length, consisting of a telomerase reverse

transcriptase (TERT) component and a ribonucleoprotein moiety (TERC) (141). In adults,

telomerase activity is only observed in immature germ cells, certain stem/progenitor cells, and a

subset of somatic cells such as fibroblasts (142). Upon reaching a critical telomere length, DNA

damage checkpoints will initiate crisis, leading to activation of senescence pathways and

apoptosis (143,144). Cells that escape this checkpoint activate telomerase or, more rarely, cells

will activate the telomerase independent mechanism, alternative lengthening of telomeres

pathway (145,146). It has been proposed that cancer cells must regain telomere maintenance to

avoid senescence and extensive chromosome fusion during crisis, and an estimated 85-90% of

human cancers have reactivated telomerase and can maintain telomere length (140,146).

Interestingly, these findings conflict with other reports of telomere length and function. It has

been found that telomeres of invasive cancers are often shorter than their normal counterparts

(147). When telomere end protection is compromised, chromosomes enter breakage-fusion-

bridge cycles that can dramatically reorganize the genome (108,140). Telomere shortening in

telomerase deficient mice (mTERC-/-) has been shown to lead to increased spontaneous tumour

formation and initiation of microadenomas of the gastrointestinal tract (148). Telomeres in

colorectal polyps and tumour cells are shorter than those in adjacent normal tissue (149–151).

These conflicting results may be explained by the timing of telomere shortening. Telomere

shortening may promote CIN, driving early carcinogenesis, whereas telomerase activation and

subsequent telomere stability may confer immortality at the latter stages of tumourigenesis

(108,151,152).

15

Genomic stability is maintained by checkpoints that regulate DNA fidelity maintenance. In

response to exogenous or endogenous genotoxic stress, checkpoints will initiate a cascade of

events culminating in either cell cycle arrest, if damage can be repaired, or apoptosis/cellular

senescence, if damage cannot be repaired. DNA damage checkpoints either recognize and

respond to DNA damage or monitor the fidelity of replicated DNA (111,153). Among the

complement of DNA repair proteins, a number have been demonstrated to play roles in various

human cancers, including TP53, ATM, ATR, BRCA1, BRCA2, PARP, and RAD proteins

(154,155). Of these, TP53 mutations have been directly implicated in CRC and CIN, occurring in

40-50% of CRCs (156,157). ATM mutations have also been demonstrated in CRC (118,157).

Mutations or dysregulation of other DNA repair proteins likely affect processes such as

apoptosis, postulated to create a permissive state allowing CIN to occur without directly causing

CIN (158). Studies have shown similar types of rearrangements in yeast as in primary human

cancers when checkpoint proteins are inactivated, indicating that these mechanisms are

conserved (159). These observations provide indirect evidence for the role of replication

checkpoints in CIN.

16

Figure 1.2 Overview of Wnt signaling pathway. Left panel: In the absence of Wnt signaling, a

β-catenin destruction complex forms in the cytoplasm including APC, GSK3β, AXIN1, and CK1

leading to the phosphorylation of β-catenin. β-catenin is then degraded by the proteasome. Co-

repressors such as Groucho prevent TCF/LEF transcription factors from transcribing target

genes. Extracellular Wnt antagonists including SFRPs, DKKs, and WIF-1 also inhibit activation

of this cascade. Right panel: When Wnt signaling is activated, Wnt ligand binds the receptor

complex. LRP5 or LRP6 is phosphorylated by GSK3β and CK1, and AXIN1 is recruited to the

plasma membrane. β-catenin is not degraded and enters the nucleus. Along with factors

including p300 and PYGO (Pygopus), β-catenin and TCF/LEF transcription factors initiate

transcription of target genes. Figure adapted from Gregorieff and Clevers, Genes Dev

2005:19;877-890, Staal et al. Nat Rev Immunol 2008;8(8):581-593, and Takebe et al. Nat Rev

Clin Oncol 2015;12(8):445-464. [References (117,160,161)].

17

1.2.2 Microsatellite instability

A microsatellite is a tract of repetitive DNA found at locations throughout the genome, made of

repeating units one to six base pairs in length. Microsatellite instability (MSI) is the change in

length, through either insertion or deletion of repeating units, in a microsatellite sequence in a

tumour compared to normal tissue (162–164). This lengthening or shortening is due to increased

strand slippage of DNA polymerases at repetitive regions. MSI, caused by defective DNA

mismatch repair (MMR) activity, is seen in approximately 15% of sporadic CRCs (163,165).

MMR loss has been shown to increase the normal mutation rate of a cell by 100- to 600-fold

permitting the accumulation of mutations in many genes (166,167).

MSI may be caused due to mutations in mismatch repair genes mutL homolog 1 (MLH1), mutS

homolog 2 (MSH2), mutS homolog 6 (MSH6), or PMS1 homolog 2, mismatch repair system

component (PMS2), but is most frequently due to promoter CpG island hypermethylation of

MLH1 (118,168). Clinicopathological features associated with MSI include mucinous histology,

signet cell ring morphology, Crohn’s like reaction (several nodular lymphoid aggregates beyond

the advancing edge of the tumour), increased tumour-infiltrating lymphocytes, poor

differentiation, and proximal colonic location (169–172). The MSI-high phenotype is generally

associated with more favourable prognosis than microsatellite stable (MSS) CRC patients, and

this improved outcome may be more pronounced for localized disease (172–174). The reasons

for improved outcomes in MSI patients are not completely understood but may be in part due to

lymphocyte infiltration.

Repetitive microsatellite sequences occur throughout the genome. Mutations occurring within

coding regions may cause frameshift mutations leading to the generation of premature stop

codons. A number of critical genes contain microsatellite repeats and often display somatic

mutations in MSI-H tumours including genes involved in cell proliferation (TGFBR2, TCF7L2,

WISP3, IGF2R, AXIN2), cell cycle and apoptosis (BAX, CASP5, PTEN, FAS), DNA repair

(BLM, CHEK1, RAD50), and, interestingly, the MMR genes MSH3 and MSH6 (175).

MSI status in tumours is determined through PCR on tumour and normal DNA from the same

individual. Normal tissue must be used to compare as individuals may harbour heritable and

stable polymorphic variation within microsatellites (163,176,177). The National Cancer Institute

has recommended a panel of five microsatellite markers including two mononucleotide repeats

18

(BAT25 and BAT26) and three dinucleotide repeats (D5S346, D2S123 and D17S250) (163).

These guidelines were later revised to include testing of additional mononucleotide markers

(BAT40, MYCL) in tumours with instability at only a dinucleotide marker, as mononucleotide

markers were found to be more reliable for identification of MSI-H tumours (178). Tumours

with instability in two or more markers are termed microsatellite instability high (MSI-H).

Tumours with instability in one out of five microsatellites are microsatellite instability low (MSI-

L). Tumours with zero unstable markers are considered microsatellite stable (MSS). If more than

five markers are used to assess MSI, MSI-H is defined as over 30-40% of markers unstable,

MSI-L is defined as >0% and <30-40% unstable, and MSS is 0% unstable.

The MSI-L phenotype has often been shown to be more similar to MSS tumours than MSI-H

tumours. If enough markers are tested, eventually some degree of microsatellite instability will

be detected (179). A small number of studies have described distinct gene expression profiles in

MSI-L CRC compared to MSI-H or MSS, however many studies have not demonstrated such a

distinction (180,181). MSI-L may also be an independent negative prognostic factor in stage III

CRC patients not undergoing chemotherapy or patients with mutation in RIS1 (182,183). Further

studies are required to determine what biological significance this MSI-L phenotype has, if any.

Most studies group MSS and MSI-L together and some have recommended the discontinuation

of using MSI-L as a distinct subgroup of CRC (184).

The major chemotherapeutic agent used in the treatment of CRC is 5-fluorouracil (5-FU). 5-FU

works by targeting thymidine synthetase and/or direct incorporation of its metabolites into DNA

and RNA, initiating apoptosis (185,186). This agent is recognized by the intact MMR system,

leading to cell death in cells with proficient MMR while MMR-deficient cells are resistant (187).

Though successful at treating MSS CRC, a number of studies have demonstrated that 5-FU does

not confer a survival advantage in stage II and stage III patients with MSI-H in retrospective and

prospective studies (188–190). Several alternative treatment options have demonstrated greater

effectiveness in treating MSI CRC compared to MSS. Studies have shown that irinotecan, a

topoisomerase I inhibitor, improves five year disease free survival in stage III MSI-H CRC

compared to MSS when given in combination with 5-FU and Leucovorin (191). The US Food

and Drug Administration has also recently approved pembrolizumab, an anti-PD-1 drug, for the

treatment of rare metastatic MSI-H CRC (192). This drug works as a PD-1 (programmed death

1) immune checkpoint inhibitor, offering clinical benefit to patients with MSI, but not MSS,

19

tumours.

1.2.3 Epigenetic instability

1.2.3.1 Epigenetics

Epigenetics is the study of heritable changes to gene expression caused by mechanisms other

than changes to the DNA sequence itself, including DNA methylation, DNA

hydroxymethylation, histone modifications, and non-coding RNAs (193–195). An overview of

epigenetic regulatory mechanisms is demonstrated in Figure 1.3. Epigenetic modifications

affecting the DNA and chromatin influence gene expression and genomic stability. DNA

methylation is the covalent addition of a methyl group (CH3) to the C-5 position of cytosine

residues, primarily occurring in a CpG dinucleotide. Cytosine may also be methylated in a non-

CpG context, in a CpA, CpT, or CpC dinucleotide, also called CpH methylation. Most

differentiated cell types contain very low levels of non-CpG methylation, for example, it

comprises just 0.28% of the total number of methylated Cs in rectal mucosa (196). Certain cell

types incur higher levels of non-CpG methylation including embryonic stem cells, pluripotent

stem cells, and adult brain cells (~10-25% of methylated Cs) (196–198).

CpG islands are regions dense in CpG dinucleotides and are found in the promoters of 60% of

human genes, are greater than 200 base pairs in length, contain a GC percentage greater than

50%, and an observed/expected CpG ratio above 0.60 (199,200). This observed/expected CpG

ratio is calculated by dividing the proportion of CpG dinucleotides in the given region by what is

expected by chance. CpG dinucleotides are depleted throughout much of the genome due to

spontaneous deamination of both unmethylated and methylated cytosines, yielding uracil and

thymine, respectively, thus observed rates of CpGs are lower than would be expected due to

chance (201). The majority of CpG sites in the genome are methylated in most cell types,

however, most of these sites are located outside of CpG islands and <10% of CpG islands are

methylated in normal cells (202–204). Beyond CpG islands, methylation may occur at regions of

lower CpG density lying in close proximity to CpG islands, termed CpG island shores (within

2000 base pairs of an island) and CpG shelves (within 4000 base pairs) (205,206). Despite lower

CpG content, CpG shore methylation is more closely associated with tissue of origin and more

frequently altered between normal and tumour cells than CpG island methylation.

20

DNA methylation plays a role in silencing gene expression and repetitive elements in the

genome (207,208). It also plays an important role in X chromosome inactivation, in which one of

the two X chromosomes in female somatic cells is stably silenced ensuring equal gene dosage in

males and females (209,210). DNA methylation is also responsible for silencing of imprinted

genes, in which one copy of a gene is silenced in a parent-of-origin manner, for example at the

IGF2-H19 locus in which IGF2 is expressed only from the paternally inherited allele while H19

is only expressed from the maternal allele (211,212).

Methylation is established de novo by DNA methyltransferase 3A and 3B (DNMT3A,

DNMT3B) and stably maintained following DNA replication by DNA methyltransferase 1

(DNMT1). Aberrant DNA methylation changes are a hallmark of cancer, characterized by global

hypomethylation coupled with specific CpG island hypermethylation (201). The first change in

DNA methylation implicated in carcinogenesis was DNA hypomethylation (213,214). CRCs in

particular have approximately 10-15% lower methylation levels than primary normal tissue

(202). Age-associated DNA hypomethylation also occurs in a tissue specific manner throughout

life, and may contribute to age-related increases in carcinogenesis (215–218).

More widely studied than genome-wide hypomethylation is promoter CpG island

hypermethylation, which silences gene expression. At the earliest stages of carcinogenesis

aberrant methylation can be detected, including methylation of genes such as MLH1, MGMT,

MINT1, MINT2, and MINT31 in ACF (219). Hundreds of studies of DNA methylation in CRC

have been published to date, and various genes have been suggested as methylation markers of

CRC diagnosis, prognosis, prediction, and classification.

DNA demethylation refers to the exchange of methylated cytosine back to the unmodified

cytosine base. DNA demethlyation was originally thought to be a passive process in which the

methylation mark is lost following DNA replication. While this passive process likely still plays

a role, the active demethylation model has also been demonstrated (220). In this multi-step

process the methylation mark is removed by ten-eleven translocation (TET) family enzymes,

which generates the intermediate oxidation derivative 5-hydroxymethylcytosine (5hmC). This

intermediate likely also plays a role in normal cell physiology, and is dysregulated in cancers

including CRC (221,222). 5hmC can then be further oxidized to form 5-formylcytosine, then 5-

carboxylcytosine, and finally cytosine, independent of DNA replication (223).

21

Histone modifications are another category of epigenetic regulation involved in regulating

chromatin structure and gene transcription. The basic unit of chromatin is the nucleosome, an

octamer of proteins consisting of two copies each of histones H2A, H2B, H3, and H4 (224). An

approximately 146 base pair section of DNA is wrapped around each nucleosome (225,226). The

protruding N-terminal tails of histones can undergo a variety of post-translational modifications

including methylation, acetylation, ubiquitylation, sumoylation, and phosphorylation (227).

Depending on the location and type of modification, these alterations can activate or repress

transcription by promoting chromatin remodeling into an open or closed state or by recruiting

other factors to the chromatin. There are a large number of proteins that establish, interpret, and

remove histone modifications that work together to maintain epigenetic regulation, broadly

classified as writers, readers, and erasers. Writers establish the histone modifications and are

recruited to appropriate loci by sequence context and the histone marks or other proteins already

present (228). Readers recognize histone modifications and aid in carrying out appropriate

responses and chromatin remodelers can alter nucleosome positioning (229). Erasers remove the

histone marks when required. Writers include histone methyltransferases (HMT), which deposit

methyl groups, and histone acetyltransferases (HAT), which deposit acetyl groups. HMTs and

HATs are counteracted by histone demethylases (KDM) and histone deacetylases (HDAC), of

which there are many that catalyze reactions at specific histone residues. Known histone writers,

erasers, and readers of histone H3 modifications are described in Table 1.4. Alterations in the

expression or function of histone modifications and their regulators may drive tumourigenesis

(230–232). Different histone variants also exist which play different roles, for example histone

variants of H2A include H2A.X and H2A.Z, and the H3 variant H3.3 is deposited in a

replication-independent manner at active promoters and enhancers (233,234).

An additional level of epigenetic control is exerted through non-coding RNAs (ncRNA). The

most widely studied type of ncRNA are microRNAs, which are small, single-stranded ncRNAs

that post-transcriptionally repress expression of mRNA in a sequence specific manner (235,236).

Other categories of ncRNA include transcription initiation RNA, piwi-interacting RNA, small

interfering RNA, enhancer RNA, small nucleolar RNA, and long non-coding RNA (237,238).

22

Figure 1.3 Overview of the main epigenetic regulatory mechanisms. Numerous epigenetic

mechanisms exist to regulate the DNA, including features at DNA, RNA, and protein levels.

Modifications to the DNA include DNA methylation (5mC) and DNA hydroxymethylation

(5hmC), primarily occurring in CpG dinucleotides. Non-coding RNA-based mechanisms

include, but are not limited to, microRNA and long non-coding RNA. Histone proteins, histone

variants, and histone post-translational modifications can activate or repress transcription by

promoting chromatin remodelling into an open or closed state. Adapted from HO LT. Genome-

wide distribution and regulation of DNA methylation and hydroxymethylation in prostate cancer.

Master’s thesis 2014, University of Toronto, Toronto, Canada.

23

Table 1.4 Histone H3 writers, erasers, and readers. Writers, erasers, and readers of well-

defined histone H3 lysine modifications are listed. While other modifications exist, including at

other residues and other histones, several of the best defined are listed. The major function and

location for each histone modification is listed. Adapted from Shen & Laird, Cell 2013;

153(1):38-55. Reference: (201).

Modification Function/Location Writers Erasers Readers H3K4me1/3 Activation of

enhancers (me1) or promoters (me3)

MLL MLL2-5 SETD1A/B SMYD1-3 PRDM9

KDM1A/B KDM2B KDM5A-D

MLL TAF3 RAG2 BPTF PHF2/6/8

H3K9me3 Repression of heterochromatin, centromeres, telomeres

SUV39H1/2 PRDM2/3/16 SETDB1/2 PRDM3 EHMT1/2 (G9A)

KDM1A KDM3A/B KDM4A-D

CBX5 EHMT1/2 (G9A) UHRF1

H3K9ac Activation of promoters

CREBBP/EP300 GNAT family MYST family

HDAC1-11 TAF1 BRD4/8 SMARCA4 KAT2B

H3K27ac Activation of promoters and enhancers

CREBBP/EP300 GNAT family MYST family

HDAC1-11 TAF1 BRD4/8 SMARCA4 KAT2B

H3K27me3 Repression of promoters

EZH1/EZH2 KDM6A/B/7A PHF8

EED CBX7

H3K36me3 Transcriptional elongation at gene bodies

SETD2 NSD1 WHSC1/1L1 SMYD1/2

KDM2A KDM4A/C

MORF4L1 BRPF1 DNMT3A

H3K79me2 Transcription at gene bodies

DOT1L Unknown TP53BP1

24

1.2.3.2 CpG island methylator phenotype

The idea of a CpG island methylator phenotype (CIMP) was first put forth in 1999 to describe

the subset of CRCs demonstrating widespread CpG island hypermethylation (130). There has

been considerable research since then to create a panel of markers that will clearly identify

CIMP. Two main panels of markers have emerged to discern CIMP tumours, the panel first

published by Weisenberger et al. (239), consisting of the markers CACNA1G, IGF2, NEUROG1,

RUNX3 and SOCS1, and the classic panel, which interrogates methylation at MLH1, MINT1,

MINT2, MINT12, MINT17, and MINT31 (200). The CIMP phenotype is most frequently found in

female CRC patients with MSI-H and BRAF mutation, while a low level of CIMP (CIMP-low) is

associated with KRAS mutation and male sex (240,241).

The biological consequences of CIMP are still being elucidated. There are conflicting reports of

CIMP having both a positive and negative impact on prognosis (242–244). Prognostic

implications of CIMP status are likely modified by CIN or MSI status, as well as mutation of

BRAF and KRAS. In a study of 2,050 participants, individuals with CIMP-positive, MSS tumours

with BRAF mutation had the highest disease specific mortality compared to all other

combinations of CIMP, MSI, and BRAF/KRAS mutation status (244). The lowest disease-specific

mortality was seen in CIMP-negative MSI-H patients without BRAF or KRAS mutations.

Another group reported significantly shorter disease free survival in patients with CIMP

independent of MSI status (242). Despite its potential ability to prognosticate CRC patients, to

date researchers have not consistently utilized a single unified CIMP panel nor is CIMP status

routinely used in a clinical setting.

1.3 Hereditary syndromes, pathology, and presentations

70-85% of CRCs are sporadic, arising from somatic alterations while ~25% of CRC cases

display familial aggregation. Less than 10% of all CRCs can be attributed to inherited

susceptibility syndromes (28–31). These hereditary syndromes are broadly classified into two

categories based on the presence or absence of multiple colorectal polyps. Polyposis syndromes

are distinguished by large numbers of polyps, which can be further subdivided into adenomatous

or harmartomatous syndromes (245).

25

1.3.1 Lynch syndrome

Lynch syndrome [formerly hereditary nonpolyposis colorectal cancer (HNPCC)] is the most

common form of hereditary CRC, accounting for less than 5% of all CRCs (30,31). This

autosomal dominant syndrome is caused by germline mutations in MMR genes. Over 90% of

Lynch syndrome families have mutations in MLH1, MSH2, or MSH6 (246,247). The mutational

spectrum is distributed across all regions of the genes without obvious hotspots (246,247). Less

commonly mutated genes include PMS2 and PMS1. There are also several reported cases of

Lynch syndrome in individuals with constitutional epimutations, in which methylation of MLH1

or MSH2 is found throughout normal somatic tissues. Epimutation of MLH1 manifests as

promoter CpG island methylation of a single allele while MSH2 epimutations are caused by

deletions containing the transcription termination signal of the neighbouring EPCAM gene,

resulting in EPCAM-MSH2 fusion transcripts and methylation of the MSH2 promoter (248–251).

Though the number of polyps occurring in Lynch syndrome patients is modest, especially

compared to individuals with polyposis syndromes such as familial adenomatous polyposis,

compared to the general population Lynch syndrome patients exhibit earlier onset of polyps

(third and fourth decade of life), greater size and frequency, more villous and dysplastic

characteristics, and more rapid progression to cancer (252–254). Characteristics also include

multiple CRCs occurring either synchronously (18%) or metachronously (24%) that are

generally located in the proximal colon (70%), have poor differentiation, mucinous histology,

and MSI (95%) (30,252,255). The risk of developing CRC by age 70 is approximately 35-54%

for women and 40-97% for men harbouring MLH1 mutations (246,256–258). In MSH2 mutation

carriers, the risk of CRC by age 70 is 35-64% for women and 40-92% for men (246,256–259).

For MSH6 mutation carriers the risk of CRC by age 70 is 25-30% for women and 25-69% for

men (246,260). Lynch syndrome patients may also present with extracolonic tumours including

endometrial, stomach, small bowel, ovarian, hepatobiliary epithelium, kidney epithelium, ureter

epithelium, bladder, and brain (30,255,261).

Without screening, many Lynch syndrome patients remain asymptomatic until cancer

development so several diagnostic criteria were developed to recognize patients at risk earlier.

Amsterdam criteria were established in 1990 to define families with the syndrome for research

purposes (262). Since then, the criteria have been modified to include extracolonic cancers and to

26

account for smaller families (263). The Bethesda Guidelines and revised Bethesda Guidelines

were subsequently developed to identify patients who would benefit from MSI testing (178,264).

Amsterdam and Bethesda criteria are described in Table 1.5. Familial segregation of MSS/MSI-L

tumours is termed “familial colorectal cancer type X” syndrome (265). The exact mechanisms

are unknown but may potentially result from common variant alleles with low to moderate

penetrance of key candidate genes already associated with colorectal cancer (266,267).

27

Table 1.5 Clinical criteria for Lynch syndrome. References: (178,262–264).

Name Criteria Amsterdam 1. Three or more relatives with CRC, one of whom is a first-degree

relative of the other two; familial adenomatous polyposis should be excluded; 2. CRC involving at least two generations; 3. One or more diagnosed at age <50 years

Amsterdam II 1. Three or more relatives with Lynch-associated cancer (colorectal, endometrial, small intestine, ureter, renal pelvis), one of whom is a first-degree relative of the other two; familial adenomatous polyposis should be excluded; 2. CRC involving at least two generations; 3. One or more CRCs diagnosed at age <50 years

Bethesda 1. Meets Amsterdam criteria 2. Two Lynch-related cancers, including synchronous and metachronous CRCs or associated extracolonic cancers 3. CRC and a first-degree relative with CRC and/or a Lynch-related extracolonic cancer and/or a colorectal adenoma; one of the cancers diagnosed at age <45 years, and the adenoma diagnosed at age <40 years 4. CRC or endometrial cancer diagnosed at age <45 years 5. Right-sided CRC with an undifferentiated pattern on histology diagnosed at age <45 years 6. Signet-ring cell-type CRC diagnosed at age <45 7. Adenomas diagnosed at age <40 years

Revised Bethesda

1. CRC diagnosed at age <50 years 2. Synchronous or metachronous CRC or other Lynch-associated tumours regardless of age 3. CRC diagnosed at age <60 years with histologic findings of infiltrating lymphocytes, Crohn’s like lymphocytic reaction, mucinous/signet ring differentiation or medullary growth pattern 4. CRC in ≥1 first-degree relative(s) with a Lynch-related tumour, with one of the cancers being diagnosed at age <50 years 5. CRC diagnosed in ≥2 first- or second-degree relatives with Lynch-related tumours, regardless of age

28

1.3.2 Familial adenomatous polyposis

Familial adenomatous polyposis (FAP) is the most well-characterized and first recognized

polyposis syndrome. Accounting for less than 1% of all CRCs, FAP is an autosomal dominant

syndrome with nearly 100% penetrance caused by germline mutations in the APC gene (29,245).

It is characterized by hundreds to thousands of colorectal adenomatous polyps, which develop

during the second or third decade of life (29,245,268). The number and size of the polyps

increases over time with a slight predisposition for the distal colon (17,245). APC mutations in

FAP patients are usually located in codons 169-1600 with a hotspot in exon 15 between codons

1286 and 1513, and over 95% of germline mutations are truncating or nonsense (268,269). FAP

patients also develop polyps in the upper gastrointestinal tract including the gastric fundus and

body, jejunum, and ileum, but these polyps have low malignant potential (245,270,271). FAP

patients may also develop extracolonic tumours, including papillary carcinoma of the thyroid,

hepatoblastoma, adrenal hyperplasia and carcinoma, and central nervous system (CNS) tumours

(272). FAP cases may also have congenital hypertrophy of the retinal pigment epithelium,

usually in patients with mutations in codons 463-1387 of APC (273,274).

Variants of FAP have been identified that have unique characteristics. Attenuated FAP (AFAP)

arises from mutations in APC. While originally reported to be caused by mutations at the 5’

region of the gene, subsequent studies demonstrated mutations located throughout the gene

(275). Compared to FAP, individuals with AFAP develop CRC at a later age, have a better

prognosis, and fewer numbers of adenomas, usually less than 100 (276,277). Gardner syndrome

is characterized by osteomas, epidermoid cysts, skin fibromas, dental anomalies, and desmoid

tumours in addition to colorectal adenomas (17,245,268). Mutations between codons 1403 and

1587 in APC have been identified as the causal mutation for Gardner syndrome, and in rarer

cases, β-catenin mutations (278). Turcot syndrome is characterized by colorectal adenomas and

CNS tumours including medulloblastomas, astrocytomas, and ependymomas (245). 70% of

Turcot syndrome patients have a mutation in APC, and the remaining 30% have mutations in the

mismatch repair genes (279).

1.3.3 Other polyposis syndromes

The base excision repair (BER) process is responsible for repairing DNA damage induced by

endogenous processes such as methylation, deamination, reactive oxygen species, and hydrolysis

29

(280,281). The process is initiated by DNA glycosylases, which recognize and cleave the N-

glycosylic bond connecting the damaged base to the DNA backbone (280,282). Faulty BER

predisposes to point mutations causing genomic instability at the base pair level (9,283). Biallelic

inactivation of the BER gene mutY DNA glycosylase (MUTYH) leads to the inherited form of

CRC called MUTYH-associated polyposis (MAP) (284,285). The phenotype in MAP patients is

similar to FAP, but differs in several ways. Polyposis occurs later, in the fifth and sixth decades

of life, and extra-colonic tumours are less frequent, including increased risk of duodenal polyps,

urinary bladder, ovarian, and skin cancers (286,287).

Harmartomatous polyposis syndromes are characterized by multiple benign, nodular growths in

the mucous lining of the intestinal wall. These develop at a young age, and are very rare, making

up less than 1% of all CRCs (288).

Peutz-Jeghers syndrome (PJS) is characterized by diffuse intestinal hamartomatous polyps with

distinctive mucocutaneous pigmentation. PJS patients have black or brown macules on the oral

and perioral mucosa as well as on the face (288). In addition, PJS patients have increased risk of

gastrointestinal tumours and extraintestinal cancers such as pancreatic, breast, ovarian, cervical,

thyroid, lung, and prostate cancers (289,290). This autosomal dominant syndrome has high

penetrance caused by mutations in serine/threonine kinase 11 (STK11) (291).

Juvenile polyposis syndrome (JPS) is characterized by smooth, spherical polyps composed of

cystically dilated crypts with increased malignant potential (292,293). In addition to CRC,

gastric, duodenal, and pancreatic cancers are observed in JPS (294). JPS is caused by mutations

to SMAD4 and BMPR1A which are involved in TGFβ signaling (295,296). Several other closely

related syndromes to JPS include the PTEN hamartoma tumour syndromes, Cowden’s syndrome

and Bannayan-Riley-Ruvalcaba syndrome (288,297).

1.4 Mismatch Repair (MMR)

1.4.1 Sources of DNA nucleotide mismatches

Canonical base pairing of nucleotides, also known as Watson-Crick pairing, follows the strict

rule by which adenine (A) pairs with thymine (T) and cytosine (C) pairs with guanine (G) (298).

The majority of DNA replication in eukaryotes is performed by Polδ and Polε of the B family of

DNA polymerases (299,300). The average fidelity of these enzymes is approximately one error

30

for every 105 nucleotides synthesized, which is further enhanced to one error in 107 bases due to

their inherent proofreading ability (301,302). At this error rate, hundreds of mispaired bases will

be generated through DNA replication, as the size of the human genome is approximately 3x109

bases (301). Such errors, if not rectified, will become fixed as mutations in the DNA upon a

subsequent round of DNA replication, which threatens the integrity of the DNA genetic code

(303,304).

DNA nucleotide mismatches may also occur at sites of DNA damage, which can be caused by

exogenous sources such as ionizing radiation, sunlight, chemical compounds, reactive oxygen

species (ROS), reactive nitrogen species, or genotoxic drugs (305). The most common form of

oxidative DNA damage is oxidation of deoxyguanosine triphosphate (dGTP) to 8-oxo-dGTP by

ROS (305,306). 8-oxo-GTP competes with deoxythymidine triphosphate (dTTP) for

incorporation in the DNA strand opposite to A during DNA synthesis. This leads to mispairing

of 8-oxo-G:A and subsequent T>G transversion mutations if not repaired (307,308). Another

way in which mismatches may occur is by nucleotide pool alkylation. This generates base

derivatives including the mutagenic O6-methylguanine (MeG), which is more favourably paired

with T, rather than C (309). These MeG:T mismatches give rise to G>A transition mutations after

subsequent replication (309,310). Alkylating agents such as N-methyl-N’-nitro-N-

nitrosoguanidine (MNNG) generate a range of DNA lesions including MeG (311,312).

Mismatches may also be generated if there are alterations in the dNTP pool ratios, as DNA

polymerases will increase incorporation of incorrect bases in this situation (313,314).

Chemotherapeutic drugs such as 5-FU exploit this effect. 5-FU inhibits thymidylate synthetase

(TS), the enzyme that catalyzes the conversion of deoxyuridine monophosphate (dUMP) to

deoxythymidine monophosphate (dTMP). Inhibition of TS leads to accumulation of dUMP,

which is phosphorylated to deoxyuridine triphosphate (dUTP) and misincorporated into the

DNA. This may lead to mismatches of U:G (315,316). DNA mismatches may also arise due to

the presence of an uneven number of nucleotides in one DNA strand relative to the other. If one

or more DNA bases are left unpaired this will form a small nucleotide insertion/deletion loop

(304,312).

31

1.4.2 Mismatch repair system overview

1.4.2.1 Replication error repair

The DNA mismatch repair (MMR) system guards the integrity of the genome by correcting

mispaired bases and insertion/deletion loops (IDLs), which may form during replication or

homologous recombination as well as due to DNA damage (317,318). The process of MMR is

highly conserved from bacteria to humans (319,320). Components of the MMR system were first

identified in Escherichia coli and termed “mutator” (Mut) genes since inactivation of these genes

generated hypermutable strains of E. coli (321,322).

In newly synthesized DNA of E. coli, adenines are methylated when located in the sequence

GATC by deoxyadenine methylase (dam) after a transient delay, following behind the replication

fork with a lag of approximately two minutes (323–325). Mismatch repair of hemi-methylated

DNA occurs on the newly synthesized strand (303,326). Deficiency or overproduction of dam

methylase in E. coli results in cells with a mutator phenotype (327,328).

MutS, MutL, and MutH are three essential proteins required for detecting mismatches and

directing the repair process in E. coli (304,317,325). MutS contains a DNA-binding domain and

an ATPase/dimerization domain. It forms a homodimer able to detect mismatches by binding

DNA non-specifically and bending the DNA in search of mismatches (320,329). Upon detecting

a mismatch, MutS undergoes a conformational change and initiates MMR by direct and indirect

interactions with other proteins, which include MutL, MutH, and MutU/UvrD (330,331). MutL,

an ATPase, is recruited in a MutS- and ATP-dependent manner (332,333). It dimerizes with

MutS, which activates the latent endonuclease activity of MutH (304,317). Monomeric MutH

recognizes the newly synthesized, temporarily unmethylated DNA strand and cleaves it at

hemimethylated GATC sequence within 1000 bp either 3’ or 5’ to the mismatch (334,335). MutL

also recruits MutU/UvrD, a DNA helicase II, to the mismatch (336). UvrD unwinds the DNA

from the nick generated by MutH to approximately 100 bp past the mismatch (336,337). Single-

stranded DNA binding (SSB) protein stabilizes the single-stranded DNA (ssDNA) gap (338).

The ssDNA flap is degraded in a 5’ to 3’ direction by ExoVII or RecJ exonuclease if the nick is

5’ to the mismatch (339–341). Conversely, ExoI, ExoVII, or ExoX exonucleases degrade the

DNA in a 3’ to 5’ direction if the nick is 3’ to the mismatch (339–341). The single-stranded gap

is then filled in by DNA polymerase III holoenzyme and LigI ligase seals the DNA ends (9,342).

32

β clamp, a polymerase processivity factor, and γ complex, which loads the β clamp onto the

DNA, are also required for MMR (343,344).

Eukaryotic MMR is not as well understood as in prokaryotes. It follows the same general steps

with some notable differences. A list of prokaryotic factors, their functions in MMR, and their

eukaryotic homologs is summarized in Table 1.6. MutS and MutL are homodimers in bacterial

cells whereas their eukaryotic homologs form heterodimers composed of two related but distinct

proteins. Eukaryotic cells have several MutS homologs (MSHs) and MutL homologs (MLHs).

The specific heterodimer partners dictate substrate specificity and functions (303,345).

In mammalian MMR DNA mismatches are recognized by the heterodimeric complex of either

MutSα (MSH2 and MSH6) or MutSβ (MSH2 and MSH3), depending on the DNA lesion. MutSα

recognizes single base-base mismatches and small IDLs up to 10 nucleotides in length while

MutSβ only recognizes larger IDLs up to 16 bases in length (346,347). Next, a complex

consisting of the MutL homologs, termed either MutLα (MLH1 and PMS2) or MutLγ (MLH1

and MLH3) binds MutSα or MutSβ. The MutLβ complex consisting of MLH1 and PMS1 has

also been identified, which may act as a non-essential accessory factor that enhances the activity

of MutLα (348,349). The ternary complex of MutS and MutL heterodimers undergoes an ATP-

dependent conformation change allowing it to slide away from the mismatch in the 5’ or 3’

direction until it reaches proliferating cell nuclear antigen (PCNA) (304,319). PCNA is loaded

onto the 3’ end of an Okazaki fragment or the leading strand by replication factor C (RFC)

(350,351). Binding of MutSα to PCNA and RFC activates ATP-dependent endonuclease activity

located in the PMS2 subunit of MutLα (352,353). This introduces nicks in the discontinuous

strand generating 5’ entry points for exonuclease 1 (EXO1) regardless of whether initial strand

discontinuity was 5’ or 3’ to the mismatch (305,354). Replication protein A (RPA) regulates the

rate of DNA resection by MutSα and EXO1 complex and also binds and protects the ssDNA

generated (355,356). EXO1 hydrolysis is terminated by regulation through MutLα once it lacks a

mismatched base pair (353). Polδ fills in the gaps with cofactors PCNA and RFC (303). DNA

ligase I completes the repair process by sealing the remaining nick (304,325). A simplified

overview of this process is outlined in Figure 1.4.

Unlike in E. coli, which uses DNA methylation to direct strand-specific DNA repair, the method

by which eukaryotic MMR system discriminates parental and newly synthesized DNA strands is

33

unknown. It has been proposed that the eukaryotic system uses strand breaks such as the 3’

terminus of the leading strand or the 5’ or 3’ termini of Okazaki fragments of the lagging strand.

It has also been shown that ribonucleotides are misincorporated into the DNA during replication

which are recognized by the MMR system to mark the nascent DNA strand (357,358).

34

Table 1.6 Mismatch repair proteins in E. coli and their eukaryotic homologues. References:

(300,304,326,341,347,352,354).

Protein in E. coli

Function Eukaryotic homologues

Function

MutS Binds mismatches MSH2-MSH6 (MutSα) MSH2-MSH3 (MutSβ)

Repairs single base-base mismatches and 1-2 base IDLs Repairs some single base IDLs and IDLs ≥ 2 bases

MutL Matchmaker, coordinates multiple MMR steps

MLH1-PMS2 (MutLα) MLH1-MLH3 (MutLγ)

Matchmaker for coordinating events from mismatch recognition to DNA repair synthesis Suppresses some IDLs, participates in meiosis

MutH Nicks nascent unmethylated strand at hemimethylated GATC sequence

None

β-clamp Interacts with MutS and may recruit it to mismatches and/or replication fork Enhances DNA pol III processivity

PCNA Interacts with MutS and MutL homologues Recruits MMR proteins to mismatches Increases mismatch binding specificity of MSH2-MSH6

γ-δ Complex Loads β-clamp onto DNA

RFC Complex Loads PCNA, modulates excision polarity

MutU/UvrD Loaded onto DNA at nick by MutS and MutL Unwinds DNA to allow excision of ssDNA

None

ExoI/ExoX 3’ to 5’ excision of ssDNA

EXO1 Excision of dsDNA

RecJ 5’ to 3’ excision of ssDNA

3’ exo of Polδ Excision of ssDNA

ExoVII 3’ to 5’ and 5’ to 3’ excision of ssDNA

3’ exo of Polε Excision of ssDNA

DNA pol III Accurate resynthesis of DNA

DNA polδ Accurate repair synthesis

SSB Participates in excision and DNA synthesis

RPA Participates in excision and DNA synthesis

DNA ligase Seals nicks after completion of DNA synthesis

DNA ligase Seals nicks after completion of DNA synthesis

35

Figure 1.4 Simplified overview of mammalian mismatch repair. Bidirectional mismatch

repair (MMR) requires strand discontinuities either 3’ or 5’ to the mismatch. MutSα heterodimer

(MSH2-MSH6) recognizes and binds the mismatch, recruits MutLα heterodimer (MLH1-PMS2),

and undergoes an ATP-dependent conformational change allowing the complex to slide along

the DNA until it reaches PCNA. If the complex diffuses upstream it will displace RFC. PMS2

endonuclease activity introduces nicks in the discontinuous strand allowing a 5’ entry point for

exonuclease 1 (EXO1), which degrades the nicked strand. Single stranded DNA gaps are

protected by RPA. DNA polymerase δ (Polδ) fills in the gap and DNA ligase completes repair by

sealing the remaining nick. Adapted from J. Jiricny. Nat Rev Mol Cell Biol 2006;7(5):335-46

[Reference: (325)].

36

1.4.2.2 MMR in DNA damage signaling, cell cycle arrest, and apoptosis

The MMR system repairs mismatches during DNA replication, but is also involved in the

cellular response to certain types of DNA damage, including lesions produced by methylating

agents, alkylating agents, ultraviolet radiation, and other carcinogens (303,305). Processing of

these lesions activates DNA damage signaling pathways, which result in cell cycle arrest or, in

cases of high lesion load, apoptosis to prevent cells with DNA damage from replicating

(303,359). Cells deficient in MutSα and MutLα are defective in cell cycle arrest in response to

DNA damaging agents (360,361). MMR-deficient cells are 100 times more resistant to killing by

alkylating agents and 2-4 times more resistant to killing by cisplatin, a chemotherapeutic agent

causing DNA crosslinks (362,363). Treating human cells with alkylating agents will normally

induce a G2/M cell cycle arrest in the second cell cycle after exposure (318,359). The kinase

ATM and Rad3-related (ATR) becomes activated and licenses G2/M cell cycle arrest, which is

mediated by downstream target checkpoint kinases including CHEK1 and CHEK2 and the

CDC25A phosphatase (364,365). This eventually results in phosphorylation of p53 and the

induction of apoptosis (366). Cells deficient in MutSα or MutLα fail to phosphorylate p53 or

undergo apoptosis in response to DNA damaging agents (366). ATR interacting protein (ATRIP)

is recruited to damaged DNA, where ATR phosphorylates proteins that carry out DNA repair

(359,367). MutSα and MutLα have been shown to physically interact with ATM and ATRIP in

response to DNA damaging agents, implicating a role for these proteins in MutSα- and MutLα-

mediated cellular response to damage (368,369). MSH2 also binds to CHEK1 and CHEK2 in

response to DNA damage and interacts with ATR leading to CHEK1 phosphorylation

(368,370,371).

Based on these observations, MMR proteins are implicated in a signaling cascade leading from

DNA damage to cell cycle arrest and/or apoptosis. Two models have been proposed to describe

the role of MMR. The first model, futile DNA repair cycle, suggests that MMR engages in futile

DNA repair when it encounters DNA lesions in the template strand, rather than the newly

replicated strand which it usually targets. The futile cycling activates DNA damage signaling

pathways, which lead to cell cycle arrest/apoptosis (372,373). The second model is the direct

signaling model, which suggests that MutSα and MutLα directly trigger damage response by

recruiting ATM or ATR/ATRIP to the lesion (369).

37

1.4.3 Other roles of MMR

1.4.3.1 Fidelity of genetic recombination

DNA recombination involves pairing of single strands of DNA derived from different parental

duplexes, which can often lead to mismatch generation (374). MMR ensures efficiency and

fidelity of both mitotic and meiotic DNA recombination (375,376). Failure to repair mismatches

during pairing will result in segregation of non-identical DNA strands during the next mitotic

division (377). MMR is believed to prevent strand exchange between non-identical homeologous

sequences during mitotic recombination by blocking the strand-exchange process when a

mismatch is detected (378,379). Consequently, MMR loss increases the gene duplication rate by

50-100 fold creating a destabilizing effect, which may contribute to cancer predisposition (380).

Several MMR proteins play a role in meiosis, which is the process by which the diploid

chromosome content of gametocytes is reduced by half to create haploid gametes. After one

round of DNA replication meiosis is achieved by two cycles of chromosome divisions (381).

Crossover between homologous non-sister chromatids occurs, followed by homology-dependent

repair of double strand breaks in the DNA (382). Defects in crossover formation lead to

chromosome segregation failure, which results in inefficient progression through meiosis and

poor viability of gametes (383). Inactivation of the MutS homologs MSH4 or MSH5 leads to

meiotic defects (384,385). MSH4 and MSH5 form a heterodimeric sliding clamp, which loads

onto synthetic Holliday junctions (386,387). Repeated loading of MSH4-MSH5 stabilizes the

recombination intermediate, facilitates maturation to a double Holliday junction, and yields

reciprocal crossovers associated with interference (386,387). MSH4-MSH5 may aid in

positioning a Holliday junction resolvase to cut the junctions in opposite directions (388). The

MLH1-MLH3 complex MutLγ may also play a role in meiosis, as both male and female mice

lacking these proteins are sterile (389–391). Male mouse knockouts of PMS2 are sterile,

indicating a potential role for PMS2 in synaptonemal-complex formation in spermatogenesis

(392,393).

1.4.3.2 Generation of immunoglobulin diversity

B cells utilize MMR and BER pathways in order to generate mutations in the antibody

diversification process. Mature B cells, upon antigen activation, undergo somatic hypermutation

(SHM) and class switch recombination (CSR) (394). During SHM, activation-induced deaminase

38

(AID) generates G:U mismatches through deamination of cytosine to uracil, which are processed

by MMR (395–397). However, rather than utilizing high-fidelity polymerase δ or ε, the

polymerase is replaced by error-prone polymerase η, which introduces base substitutions and

frameshift mutations (398,399). MMR proteins also play a role in class switch recombination

(CSR) in which the IgM constant region gets substituted by downstream constant sequences

(394). MMR proteins use strand breaks generated by uracil DNA glycosylase to repair AID-

induced G:U mispairs in a strand-indiscriminate manner which results in double strand DNA

breaks, stimulating CSR (400,401). Mice deficient in Msh2 and Msh6 accumulate five-fold

fewer mutations in the V region of antibody genes (395,396).

1.4.4 Knockout mouse models of MMR genes

Knockout mouse models have been developed for many MMR genes including Msh2, Msh3,

Msh4, Msh5, Msh6, Mlh1, Mlh3, Pms1, and Pms2 (Table 1.7). Most knockout mice display a

mutator phenotype, MSI, and are prone to cancer (402). Mlh1, Msh2, and Pms2 knockout mice

experience a high incidence of tumours, primarily developing lymphomas, with secondary

susceptibility to gastrointestinal (GI) tumours, skin neoplasms, and sarcomas (390,403–405).

These mice display MSI-H. Mlh1 knockouts are infertile, while only male Pms2 knockout mice

are infertile. Msh6 knockouts, similarly, have high tumour incidence of lymphomas, GI tumours,

and skin neoplasms, are fertile for both sexes, and have low MSI (406). Msh3 knockout mice are

fertile and develop lymphomas and GI tumours (404,407). These mice can repair single base-

base mismatches but lack ability to repair IDLs (408). Msh4 and Msh5 knockouts do not develop

tumours and do not display MSI, with sterility seen in both sexes for both genes (409,410). Pms1

knockout mice do not develop tumours nor do they incur MSI (390). The diverse phenotypes

seen in the mice with regards to tumour type and frequency, MSI status, and fertility provide

insights into the various functions of MMR genes. For example, Msh3 knockout mice have a

much lower tumour burden than Msh2 or Msh6 knockout mice. MutSα, consisting of MSH2 and

MSH6, repairs single base mispairs or IDLs while MutSβ, consisting of MSH2 and MSH3,

predominantly repairs IDLs greater than one base pair in length. Thus, Msh3-/- mice can still

repair single-base mispairs or IDLs thus these mice have more moderate repair defects and fewer

tumours (408). The infertility observed in Msh4, Msh5, Mlh1, Mlh3, and male Pms2 knockouts

indicates the roles for these MMR factors in meiosis (411).

39

Interestingly, the tumour spectra of these knockouts do not recapitulate that seen in humans, with

lymphomas being more common than GI tumours. While lymphomas are not commonly

observed in Lynch syndrome families with MMR germline defects, they have been observed in

rare cases (412,413). Several mouse models of CRC have been developed harbouring Apc

mutations that more closely resemble CRC in humans (414). A mouse model of FAP, termed

ApcMin for Multiple Intestinal Neoplasia, develops numerous adenomas throughout the GI tract

which eventually progress to carcinomas (415). The Apc mutation in these mice is a T>A

transversion mutation at nucleotide 2549, which creates a premature stop codon (416).

Subsequent mouse models of Apc mutations have since been developed, including the Apc1638N

model containing a heterozygous mutation in Apc leading to its truncation at amino acid 1638

(417). MMR-deficient mouse models have also been developed containing heterozygous Apc

mutations and these models are able to more closely mimic human phenotypes of CRC. The

combination of homozygous Msh2, Msh6, Mlh1, or Pms2 mutations with Apc1638N confines

tumour development to the intestinal tract in these mice (418,419).

40

Table 1.7 Phenotypes of MMR-deficient knockout mice. References: (390,403–410,418).

Genotype Tumour Spectrum

Tumour Incidence

MSI Status Fertility Male/Female

MSH2-/- Lymphoma, GI, skin, and others

High High +/+

MSH3-/- Lymphoma, GI Low Low +/+ MSH6-/- Lymphoma, GI,

skin, others High Low +/+

MSH4-/- None None Stable -/- MSH5-/- None None Stable -/- MLH1-/- Lymphoma, GI,

skin, others High High -/-

MLH3-/- Lymphoma, GI, skin, others

High High -/-

PMS1-/- None None Stable +/+ PMS2-/- Lymphoma,

sarcoma High High -/+

GI – gastrointestinal; MSI – microsatellite instability

41

1.5 Genetic variation

1.5.1 Polymorphisms

Single nucleotide polymorphisms (SNPs) are the most common form of genetic variation in the

human genome, where a single nucleotide at a fixed position in the genome is substituted with

another (420). Traditionally, SNPs are defined as alterations present in the population at a minor

allele frequency (MAF) greater than 1% resulting in neutral or benign phenotypic consequences

(421). Common SNPs are defined as having a MAF > 5% (420,422). Of the three billion bases of

the human genome, over 84 million SNPs have been identified occurring every 300-1000 base

pairs (420,423,424). The majority of SNPs are shared between populations, but some are specific

to populations or continental groupings of populations sharing a recent history (420). For

example, three populations from Puerto Rico, Colombia, or with Mexican ancestry share several

hundred common SNPs not observed in other populations (422). This subset of SNPs specific to

populations likely gives rise to the observable phenotypic differences in and between

populations, including disease susceptibility and outcome (423,425). SNPs can be used to

measure admixture in populations and map genes that could account for differences in disease

incidence between populations (426,427).

Over 29,000 SNPs associated with numerous traits and diseases have been discovered thus far

through genome-wide association studies (GWAS) (428). SNPs can exert their influence on

disease pathogenesis in a variety of ways. If located within a gene a SNP may have serious

consequences on the function or structural stability of a protein if it changes the primary

structure (429). Exonic SNPs resulting in amino acid substitutions, called non-synonymous

SNPs, are the most well characterized genetic polymorphisms. They are subject to detection bias

and can usually be assayed for their functional effects (423,430). Synonymous exonic SNPs that

do not alter protein structure may still affect mRNA stability and alter splicing signals (431,432).

SNPs located in introns, promoters, enhancers, or any other non-coding regions can also be

functionally important through alteration of gene regulation. Methodologies are currently being

developed for predicting the function of SNPs located in introns or regulatory regions (433,434).

SNPs may also disrupt or create CpG dinucleotides, causing altered methylation patterns

(435,436).

Another form of genetic variation is copy number variations (CNV), which are structural variants

42

containing large regions of variable copy numbers. CNVs are intermediate in size when

compared to large genomic rearrangements or single base variations of SNPs. CNVs are

typically defined as being larger than 50 bp, but can range in size up to 3 Mb and may

encompass entire genes, promoter regions, and have dose effects (437,438). 4.8-9.5% of the

genome can be classified as CNVs (438). A pair of individuals from a population is estimated to

differ by at least 11 CNVs (439). Tumours have an increased number of CNVs compared to

normal tissue (440,441).

1.5.2 Utility and roles of SNPs in genetic epidemiology

Though the majority of SNPs have no phenotypic role, a number of SNPs have been identified

which can alter an individual’s risk for certain diseases, including colorectal cancer, and predict

response to certain drugs (284,434,442,443). When SNPs within susceptibility loci are

discovered to be associated with incidence of a certain disease, the SNP within that locus may be

the causal functional variant responsible or it may be a proxy for another SNP in the same region

that is the true functional variant (444). Markers on the same chromosome that remain associated

with one another are in linkage disequilibrium (LD). This LD persists between many SNPs

because meiotic recombination does not occur completely at random, but is concentrated in

hotspots (445,446). If LD is plotted across the genome it demonstrates blocks of common genetic

variation, termed haplotype blocks, which are separated by hotspots of recombination (447).

Haplotypes vary in size and may be as large as hundreds of kilobases in length, demonstrating

that some regions of the genome are relatively protected from recombination (448,449). Pairwise

LD between two SNPs can be measured using D’ (standardized LD coefficienct, D) or by r2

(correlation coefficient). D’ and r2 values range from zero to one (423). A SNP best representing

a haplotype block is referred to as a tag SNP. A tag SNP in an LD block may be statistically

associated with a certain trait, disease, or phenotypic outcome, but may not be the precise variant

in the block that has a causal role (450). Fine mapping of the region, in silico annotation, and

functional studies are required to determine the exact functional variant (433,451).

1.5.3 Identification of low-penetrance alleles in CRC

Family history of CRC is one of the most important factors in determining an individual’s risk of

developing CRC (28,38,57). However, <10% of CRCs can be attributed to mutations in highly

penetrant genes associated with hereditary CRC syndromes. There may still be currently

43

undefined high-penetrance genes contributing to familial CRC risk. Alternatively, it is likely that

the accumulation of a number of low-risk, low-penetrance alleles contributes to significant CRC

risk.

GWAS allow the assessment of millions of SNPs per individual, which facilitates a hypothesis-

free approach to genetic epidemiological testing of common diseases. GWAS have identified

novel loci in a wide variety of cancer types and other diseases (452). SNPs identified as being

associated with CRC or other disease may be risk-associated variants, or modifier

variants/alleles. Modifier variants are coding or non-coding regulatory elements that interact with

the genome, and modifier variants can act together to modulate a phenotype in complex diseases

(453).

The National Human Genome Research Institute (NHGRI) and European Bioinformatics

Institute (EMBL-EBI) provide a quality controlled, manually curated collection of published

GWAS. Currently 2,610 studies have been curated and 29,382 SNP-trait associations with P-

values < 1.0x10-5 can be explored in their GWAS catalog (as of most recent release date

November 13, 2016) (428). 30 published studies meet the GWAS catalog criteria for association

between SNPs and colorectal cancer, for a total of 193 SNPs at 145 genetic loci (454). Utilizing

a P-value cut-off of < 1.0x10-8 to minimize false positives leaves 38 significant SNPs at 28 loci

from 20 studies. These significant SNPs are summarized in Table 1.8. The most commonly

reported loci was 8q24.21, including the SNPs rs7014346, rs10505477, and rs6983267

(284,455–462). This region is a gene desert and the closest protein coding gene in the region is

MYC. SNPs at 10p14 (rs10795668, rs11255841) have also been implicated in CRC in a number

of studies (284,459,460,462). These SNPs are located near GATA3, a transcription factor known

to be a cancer driver (463). SNPs in the SMAD7 locus at 18q21 (rs7226855, rs7229639,

rs4939827) are also associated with increased risk of CRC, with SMAD7 being involved in the

TGF-β family signaling pathway (284,455–457,464,465).

Several CRC susceptibility SNPs have also been identified that are located within MMR genes.

The intronic MSH2 SNP rs2303428 (IVS12-6T>C) has been shown to be associated with

familial and sporadic CRC as well as acute myeloid leukemia, non-Hodgkin lymphoma, high-

grade dysplasia and cancer in ulcerative colitis patients, endometrial cancer, and lung cancer

(466–473). MSH6 G39E, located in exon 1, (116G>A, rs1042821) was shown to be associated

44

with increased risk of CRC and triple negative breast cancer (474,475). This SNP was also found

to be associated with increased risk of CIMP positive CRC among MSS tumours (476). The

MLH1 founder polymorphism D132H is associated with susceptibility to sporadic CRCs, found

only in Israeli populations (477). This polymorphism attenuates the ATPase activity of MLH1

without affecting the MMR function. This uncoupling of the MLH1 functions results in

increased risk for MSS, but not MSI-H, CRCs (477).

Of particular interest to my research project is the MLH1 promoter SNP rs1800734 (-93G>A),

located in the CpG island of MLH1. The variant A allele of this SNP, as well as two downstream

SNPs in LD with rs1800734 (rs749072 and rs13098279), shows a strong association with MLH1

CpG island hypermethylation, loss of MLH1 expression, MMR deficiency, and MSI (478,479).

Further, the allelic variant of rs1800734 has a functional consequence in that it decreases the

transcriptional activity of MLH1 in CRC cell lines (480). This SNP has also demonstrated an

association with increased risk of glioblastoma, gastric cancer, lung cancer, and ovarian cancer

(481–484).

45

Table 1.8 Risk loci and SNPs for colorectal cancer. SNPs with P-value < 1.0x10-8 curated

from the NHGRI-EBI GWAS catalog are listed, along with chromosomal locations and

associated genes.

Risk Locus Risk SNP Associated Gene References 1q41 rs6691170

rs6687758 DUSP10 (485)

(462,485) 3p22.1 rs35360328 CTNNB1 (486) 4q32.2 rs35509282 FSTL5, NAF1 (487) 5q31.1 rs647161 PITX1 (462,488) 6p21.1 rs1321311 CDKN1A (489) 6q25.3 rs7758229 SLC22A3 (490) 8q23.3 rs16892766 EIF3H (456,459) 8q24.11 rs76316943 EIF3H (284) 8q24.21 rs7014346

rs10505477 rs6983267

MYC

(284,457) (460,461) (455,456,458,459,462)

10p14 rs10795668 rs11255841

GATA3 (459,462) (284,460)

10q24.2 rs1035209 SLC25A28 – NKX2-3 (460) 10q25.2 rs11196172

rs12241008 TCF7L2 VTI1A

(462) (491)

11q12.2 rs174537 MYRF (462) 11q13.4 rs3824999 POLD3 (284,489) 11q23.1 rs3802842 C11orf93 (456,457) 12p13.31 rs10849432 PLEKHG6 (462) 12p13.32 rs10774214 CCND2 (462,488) 12q13.12 rs11169552 DIP2B (485) 14q22.2 rs4444235 BMP4 (492) 15q13.3 rs2293582

rs73376930 GREM1 - SCG5 (284)

(460) 18q21.1 rs7226855

rs7229639 rs4939827

SMAD7 (284) (465) (455–457,464)

19q13.11 rs10411210 RHPN2 (492) 20p12.3 rs961253

rs2423279 FGFR3P3 - CASC20 HAO1

(492) (462,488)

20q13.13 rs6066825 PREX1 (456) 20q13.33 rs4925386

rs2427308 LAMA5 (485)

(284,460) Xp22.2 rs5934683 SHROOM2 (489)

46

1.6 Rationale

A diverse array of processes and pathways interact within cells, for example, the genome can

disrupt the epigenome. MLH1 CpG island methylation is influenced by the SNP rs1800734,

located within its promoter CpG island. This genetic variant also likely impacts DNA

methylation beyond the CpG island to include the MLH1 CpG shore, histone modifications

controlling chromatin activity, and transcription factor binding, in both normal and tumour DNA.

The epigenetic and regulational consequences of rs1800734 affecting the critical MLH1 gene

have yet to be completely determined. Two other pathways that interact in CRC are the DNA

mismatch repair and Wnt signalling pathways. Alterations in DNA mismatch repair are

associated with differential accumulation of methylation in a panel of Wnt signalling genes.

Other Wnt candidate genes also offer potential for methylation-mediated regulation of distinct

CRC subtypes, for example APC and ITF2. By investigating methylation of these candidate Wnt

signalling pathway genes in MSI and MSS colorectal tumours differentially epigenetically

regulated genes in CRC subtypes can be determined.

1.7 Hypothesis and objectives

I hypothesize that epigenetic regulation of the upstream regulatory regions of the mismatch

repair gene MLH1, including the CpG island and shore, contribute to colorectal tumourigenesis

and distinct subtypes of CRC and are further modified by the functional MLH1 promoter SNP

rs1800734. Further, I hypothesize that aberrant epigenetic regulation of candidate Wnt signaling

genes significantly contributes to distinct subtypes of colorectal cancer. Taken together, dynamic

interactions between genetic and epigenetic regulatory mechanisms may contribute to colorectal

cancer. In order to identify and characterize subtype-specific and genotype-specific epigenetic

changes in CRC, the following objectives were formulated:

I. Explore DNA methylation of the MLH1 shore and its relationship to SNP genotype of

rs1800734.

IA: Explore DNA methylation of the MLH1 shore and its relationship to SNP

genotype of rs1800734 in peripheral blood mononuclear cell DNA of CRC cases and

controls.

47

IB: Explore DNA methylation of the MLH1 shore and its relationship to SNP

genotype of rs1800734 in peripheral blood mononuclear cell, normal colorectal tissue,

and tumour DNA of CRC cases.

II. Discover epigenetic factors and transcription factors regulating MLH1 and how their presence

is modulated by SNP genotype of rs1800734.

III. Investigate the contribution of DNA methylation of candidate Wnt signaling genes to

colorectal cancer subtypes.

48

Chapter 2 Mismatch repair gene polymorphisms are significantly associated

with DNA methylation in colorectal cancer cases and controls

2.1 Summary

Single nucleotide polymorphisms (SNPs) are the most common form of genetic variation. It has

previously been demonstrated that SNPs in the mismatch repair gene mutL homolog 1 (MLH1)

region (rs1800734, rs749072, and rs13098279) are associated with MLH1 promoter CpG island

methylation, loss of MLH1 protein expression, and microsatellite instability (MSI) in colorectal

cancer (CRC) patients. Recent studies have identified CpG island ‘‘shore’’ regions flanking

many CpG islands, which are sequences less enriched in CpGs than islands. These shores often

exhibit distinct methylation profiles between different normal tissues and also normal versus

matched tumour cells of patients. To date, most epigenetic studies have focused on somatic

methylation events occurring within solid tumours; less is known of the contributions of

peripheral blood mononuclear cell (PBMC) methylation to processes such as aging and

tumourigenesis. To address whether MLH1 methylation in PBMCs is correlated with

tumourigenesis, the Illumina Infinium HumanMethylation450 BeadChip microarrays were

utilized to assess methylation in PBMC DNA of 846 controls and 884 CRC patients from

Ontario, Canada. Analysis of a region of chromosome 3p21 spanning the MLH1 locus in controls

revealed that a CpG shore 1 kb upstream of the MLH1 gene exhibits different methylation

profiles when stratified by SNP genotypes (rs1800734, rs749072, and rs13098279). Individuals

with wildtype genotypes incur significantly higher PBMC shore methylation than heterozygous

or homozygous variant carriers (P < 2.45x10-5). This significant association is also seen in CRC

cases (P < 6.81x10-6). Shore methylation also decreases significantly with increasing age in

controls and cases. Hypomethylation of the MLH1 shore is also significantly associated with an

increased risk of CRC. Similar associations were not observed in mismatch repair genes MSH2

or MSH6 with respect to SNPs located within these genes or age-associated hypomethylation,

however, methylation level at a number of CpGs was associated with CRC risk. This is the first

study of its kind to integrate PBMC methylation at a CpG shore with SNP genotype status in

CRC cases and controls. These results indicate that SNP genotype as well as the normal aging

process may influence CpG shore methylation in PBMCs.

49

2.2 Introduction

Epigenetic mechanisms induce functionally relevant changes to the genome without changing

the nucleotide sequence itself. These mechanisms include DNA methylation, DNA

hydroxymethylation, histone modifications and non-coding RNAs (224,232). DNA methylation

is among the most studied epigenetic mark, with clear links to a variety of diseases established.

In healthy individuals, genome-wide methylation levels are generally elevated at intergenic

regions and repetitive sequences (eg. ALU and LINE1 repeats), while methylation is low or non-

existent in the promoter CpG islands of most genes. These methylation patterns reverse with

increasing age, as well as in cancer (216,493,494). CpG islands, the sites of age- and cancer-

specific epigenetic changes, are defined by a length of at least 200 base pairs containing a GC

percentage greater than 50%, and an observed/expected CpG ratio over 0.60 (199). Recent

studies suggest that CpG island shores, which are less dense in CpG content than islands, flank

many CpG islands. Shores exhibit more readily distinguishable methylation levels than islands

between different tissues as well as between cancer and matched normal DNA

(202,205,495,496). The vast majority of epigenetic studies have investigated methylation at CpG

islands; however, the role of shore methylation is only just beginning to be understood.

Most published studies have investigated DNA methylation changes occurring at the tissue level

in normal and diseased states. While less is known about DNA methylation occurring in

peripheral blood mononuclear cells (PBMCs), a number of studies have demonstrated

associations between methylation in PBMCs and various diseases and cancer types including

CRC (497–505). Peripheral blood mononuclear cell types include lymphocytes and monocytes.

Lymphocytes comprise 70-90% of PBMCs, and of these approximately 70-85% are T cells, 20%

are B cells, and 5-20% are natural killer (NK) cells. Among these cell types, each has been

shown to harbour unique methylation profiles (506–508). PBMC cell type populations have been

shown to change with age, and their methylation has also been shown to change over time

(497,509). PBMC cell type proportions may also change in response to disease or toxic

exposures (507). These methylation and cell type proportion differences between disease states

and over time must be controlled for in epigenetic studies of PBMCs. Since blood samples are

collected easily from patients, and can be measured at multiple time points during disease

progression, DNA methylation changes in PBMCs can potentially be used as biomarkers for

50

various disease outcomes. Utilizing blood samples also allows for a minimally invasive

comparison between patients affected with disease and unaffected individuals. Using PBMCs as

an alternate biological source has potential which requires further systematic investigation, such

as integrating PBMC methylation with knowledge of the genetic and epigenetic landscape of

tissue DNA.

MLH1, MSH2, and MSH6 are key members of the DNA mismatch repair (MMR) pathway

(165,345). Function of MLH1 is lost in a subset of CRC tumours due to its inactivation through

mutation or methylation (57,241,510). MSH2 and MSH6 may also be mutated in CRC. Mismatch

repair deficiency leads to genome-wide accumulation of copy number alterations at short tandem

repeats, or microsatellites, termed microsatellite instability (MSI). Approximately 15% of

sporadic CRCs exhibit MSI and the majority of these occur due to promoter CpG island

methylation of the MLH1 gene in colorectal tumours (165,168). Our lab has previously

demonstrated associations between MMR gene polymorphisms, methylation, and cancer risk

(479,511). Single nucleotide polymorphisms, or SNPs, are the most common form of genetic

variation, with upwards of 84.7 million SNPs characterized by the 1000 Genomes Project (420).

Many SNPs have apparently no phenotypic consequences, while others may predispose to

various diseases (284,434,443,492). The underlying mechanism of action of these SNP variants

is not always understood.

In previous studies, the Bapat Lab examined 102 SNPs spanning 500 kb surrounding the MLH1

locus (478,479). Among these, three SNPs were significantly associated with MLH1

methylation, loss of MLH1 protein expression, and tumour MSI. These SNPs were in strong

linkage disequilibrium with one another spanning 197 kb of the genomic region on chromosome

3. This genomic region includes MLH1, constituting a haplotype block. The SNPs include

rs1800734 located 93 base pairs upstream of MLH1 (MLH1-93G>A), and rs749072 and

rs13098279 which are located further downstream of MLH1. Previous studies from our lab have

also shown through in vitro studies in transformed CRC cell lines that the risk variant (A allele)

of rs1800734 decreases MLH1 promoter CpG island-mediated transcriptional activity, thereby

providing insight into its potential role as a functional SNP (480). Promoter SNPs in MSH2

(MSH2-118T>C) and MSH6 (MSH6-159C>T) were also investigated and it was found that the

MSH2 SNP was associated with family history of CRC among CRC cases, especially in women

(511).

51

Taken together, there is a link between MLH1-region SNPs and MLH1 CpG island methylation

in CRC tumours, but the potential correlation of these three SNPs with MLH1 shore methylation

has never been investigated, nor has it been analyzed in peripheral blood mononuclear cells of

CRC cases or controls. MSH2-118T>C and MSH6-159C>T have also not been examined in

PBMC DNA. Since SNPs are static germline alterations, their potential effects on modulating

methylation may be exerted with varying capacity on all tissues of the body, including PBMCs.

Thus, the goal of this study was to examine the relationship between the aforementioned SNPs

and methylation status of the mismatch repair genes MLH1, MSH2, and MSH6 in PBMC DNA

obtained from a cohort of over 1,700 population-based controls and CRC patients.

2.3 Materials and methods

2.3.1 Study subjects

Study participants were recruited through the Ontario Familial Colorectal Cancer Registry

(OFCCR), one of six participating cancer registries that are part of the Colon Cancer Family

Registry, a US National Cancer Institute-supported consortium. Both primary CRC cases and

unaffected controls were accrued through population-based recruitment methods. A detailed

account of patient accrual, data collection, and biological specimen collection has been

previously described (461,512). Briefly, population control subjects were recruited via randomly

selected residential telephone numbers in 1999–2000 and by population-based Tax Assessment

Rolls of the provincial government allowing the identification of age- and sex-matched controls.

Due to the high proportion of self-reported Caucasians, patients with non-white, unknown or

mixed ethnic backgrounds were excluded. Of 2,736 individuals who agreed to participate, 1,336

controls completed family, personal, and diet questionnaires, provided blood samples, and were

self-reported as Caucasian. Ontario residents diagnosed with primary CRC from June 1, 1997 to

June 30, 2000 between the ages of 20 and 74 were eligible for recruitment to the OFCCR. Cases

of the inherited syndrome familial adenomatous polyposis were excluded from the study and no

related cases were used. A total of 1,257 case patients remained after exclusion. Blood and tissue

samples from CRC cases and controls were obtained with informed written consent, following

protocols approved by the research ethics board of Mount Sinai Hospital and the University of

Toronto.

52

2.3.2 Single nucleotide polymorphism genotyping

The SNPs chosen for study were selected based on extensive database and literature searches of

polymorphisms present on the Affymetrix GeneChip Human Mapping 100K and 500K

platforms. rs749072 and rs13098279 were chosen because these SNPs are in strong linkage

disequilibrium with rs1800734 as well as each other (r2 = 73 and D’ = 0.98). These 3 SNPs are

all in Hardy-Weinberg equilibrium (p = 1024) (479). The SNPs MSH2-118T>C and MSH6-

159C>T were selected as they are located in the upstream regulatory regions of mismatch repair

genes. SNP genotyping was performed as described previously (479,511). Briefly, peripheral

blood mononuclear cells (PBMCs) were isolated from the blood samples provided by CRC case

and control individuals using Ficoll-Paque gradient centrifugation according to manufacturer’s

protocol (Amersham Biosciences, Baie d’Urfé, Quebec, Canada). Genomic DNA was extracted

from PBMCs by phenol-chloroform or Qiagen DNA extraction kit (Qiagen Inc., Hilden,

Germany). The fluorogenic 5’ nuclease polymerase chain reaction (PCR) assay was used to

genotype rs1800734. This SNP was also genotyped using the Affymetrix GeneChip Human

Mapping 100K and 500K platforms as part of the Assessment of Risk of Colorectal Tumours in

Canada (ARCTIC) project and this data was used as a cross-validation measure (478). In all, 11

of 1,884 (0.58%) samples genotyped gave discordant results between the two platforms. Primer

and probe sequences have been described previously (478,479). rs749072 and rs13098279 as

well as MSH2 and MSH6 SNPs were genotyped using the Eurogentec qPCR kit (Eurogentec,

Liège, Belgium) and the 5’ nuclease PCR assay or TaqMan assay using Applied Biosystems

7900HT Sequence Detection System (Applied Biosystems, Foster City, CA).

2.3.3 Methylation microarray

CpG methylation was measured using Infinium HumanMethylation450 BeadChips from

Illumina (San Diego, CA). 1,207 control samples and 1,235 CRC samples were assayed on 96-

well plates; a subset of 65 samples were analyzed in duplicate or triplicate with data available for

a total of 136 possible pairs. Bisulphite conversion of DNA was performed using the EZ DNA

Methylation-Gold Kit (Zymo Research, Orange, CA). 500 ng of bisulphite converted DNA was

used for hybridization to the array following Illumina Infinium HD Methylation Protocol. The

efficiency of bisulphite conversion was verified using internal control probes. Samples that were

outliers with respect to internal control probes were excluded from analysis, leaving 998 controls

53

and 1,103 CRC cases. 12 of the DNA samples from CRC cases came from Epstein-Barr virus-

transformed lymphoblastoid cell lines. These were removed from analysis, leaving 998 controls

and 1,091 CRC cases. The methylation was measured at each CpG site using the fluorescent

intensity ratio. After normalization using the internal normalization probes, the resulting value

was represented by a β-value ranging from 0 (no methylation) to 1 (complete methylation).

Values with a detection p-value above 0.01 were removed from analysis.

2.3.4 Selection of CpG sites

The Infinium HumanMethylation450 BeadChip captures methylation measurements at over

450,000 CpG sites across the entire genome. I selected every CpG site on chromosome 3

between nucleotide positions 37,018,029 and 37,239,890 (Genome Build 37) spanning a 221 kb

region for further analysis. There are 70 CpG sites within this region, encompassing the genes

EPM2A (laforin) interacting protein 1 (EPM2AIP1), MLH1, and LRR binding FLII interacting

protein 2 (LRRFIP2). The SNPs rs1800734, rs749072, and rs13098279 also occur within this

region. This chromosomal region contains a CpG shore upstream of MLH1 within the coding

region of EPM2AIP1. The entire shore spans from nucleotide 37,033,373 to 37,034,166 and 13

CpG sites are located on the array. However, a section of the shore from 37,033,373 to

37,034,166, which exhibited the most significant associations containing 7 CpG sites, is the

focus of the results. A figure indicating the gene positions, CpG islands, and common SNPs is

shown in Figure 2.1.

CpGs located within and nearby the MSH2 and MSH6 genes were also selected, encompassing

106 CpG sites located from nucleotide positions 47,595,507 to 48,013,473 on chromosome 2

(Genome Build 37) spanning nearly 418 kb. Within this genomic span are the genes epithelial

cell adhesion molecule (EPCAM), MSH2, potassium two pore domain channel subfamily K

member 12 (KCNK12), and MSH6. The SNPs MSH2-118T>C and MSH6-159C>T are located

within this region. A figure indicating the gene positions, CpG islands, and common SNPs is

shown in Figure 2.2.

54

Figu

re 2

.1 R

egio

n se

lect

ed fo

r D

NA

met

hyla

tion

anal

ysis

on

chro

mos

ome

3. A

ll C

pG p

robe

s on

the

Illum

ina

Hum

anM

ethy

latio

n450

Bea

dChi

p K

its w

ere

anal

yzed

loca

ted

in th

e ge

nes EPM2AIP1,

MLH1,

and

LRRFIP2

. The

297

,540

bp

geno

mic

loca

tion

(Gen

ome

build

37/

hg19

) fro

m p

ositi

on 3

7,00

0,38

4 to

37,

297,

923

on c

hrom

osom

e 3

is sh

own

for t

hese

gen

es. C

pG is

land

s, an

d

the

num

ber o

f CpG

din

ucle

otid

es w

ithin

eac

h is

land

, are

indi

cate

d. C

pG is

land

s wer

e de

fined

as s

eque

nces

gre

ater

than

200

bp

in le

ngth

with

GC

con

tent

>50

% a

nd o

bser

ved/

expe

cted

CpG

ratio

>0.

6. C

omm

on S

NPs

foun

d in

≥ 1

% o

f sam

ples

from

dbS

NP

build

147

are

show

n. O

nly

SNPs

val

idat

ed b

y H

apM

ap o

r by

the

1000

Gen

omes

Pro

ject

are

indi

cate

d. S

NPs

are

col

ored

by

alle

le fr

eque

ncy

on a

red-

blue

spec

trum

, with

red

repr

esen

ting

rare

alle

les a

nd b

lue

repr

esen

ting

com

mon

alle

les.

Figu

re fr

om U

CSC

Gen

ome

Bro

wse

r.

55

Figu

re 2

.2 R

egio

n se

lect

ed fo

r D

NA

met

hyla

tion

anal

ysis

on

chro

mos

ome

2. A

ll C

pG p

robe

s on

the

Illum

ina

Hum

anM

ethy

latio

n450

Bea

dChi

p K

its w

ere

anal

yzed

loca

ted

in th

e ge

nes EPCAM

, MSH2,

KCNK12

, and

MSH6.

The

456

,645

bp

geno

mic

loca

tion

(Gen

ome

build

37/

hg19

) fro

m p

ositi

on 4

7,58

8,02

7 to

48,

046,

669

on c

hrom

osom

e 2

is sh

own

for t

hese

gen

es.

FBXO11

is sh

own

in th

e fig

ure

but C

pG p

robe

s wer

e no

t int

erro

gate

d fo

r thi

s gen

e. C

pG is

land

s, an

d th

e nu

mbe

r of C

pG d

inuc

leot

ides

with

in e

ach

isla

nd, a

re in

dica

ted.

CpG

isla

nds w

ere

defin

ed a

s seq

uenc

es g

reat

er th

an 2

00 b

p in

leng

th w

ith G

C c

onte

nt >

50%

and

obse

rved

/exp

ecte

d C

pG ra

tio >

0.6.

Com

mon

SN

Ps fo

und

in ≥

1%

of s

ampl

es fr

om d

bSN

P bu

ild 1

47 a

re sh

own.

Onl

y SN

Ps v

alid

ated

by H

apM

ap o

r by

the

1000

Gen

omes

Pro

ject

are

indi

cate

d. S

NPs

are

col

ored

by

alle

le fr

eque

ncy

on a

red-

blue

spec

trum

, with

red

repr

esen

ting

rare

alle

les a

nd b

lue

repr

esen

ting

com

mon

alle

les.

Figu

re fr

om U

CSC

Gen

ome

Bro

wse

r.

56

2.3.5 Adjusting for cell type proportions

Of the CRC cases profiled for methylation, 304 patients received chemotherapeutic treatment

(99% received combination of 5-fluorouracil and Leucovorin), 252 patients did not receive any

chemotherapeutic treatment, and 328 had unknown treatment status. As described in Lemire et

al.(513), a linear regression model was used to assess methylation differences between the two

groups using age, sex, alcohol consumption, smoking status, the location of the sample on the

array, and the first two principal components calculated from control probes which assess

technical variation. Factors were computed derived from 600 CpG sites previously shown to

discriminate between different blood cell types – CD4+ T cells, CD8+ T cells, B cells, NK cells,

monocytes, and granulocytes (506,508). The first six components of this analysis were utilized in

(513) as there were six cell types used to derive the 600 sites. The percentages of each cell type

in each patient’s blood sample were estimated using the estimateCellCounts function from the

minfi R package v.1.14.0. Patients who received treatment and patients who did not receive

treatment were compared and it was found that those undergoing treatment had a significant

decrease in CD4+ and CD8+ T cells (P = 9.8x10-7 and P = 0.04, respectively) and a significant

increase of NK cells (P = 0.05) and monocytes (P = 0.0006) (513). It has also been shown that

cell type proportions can be derived from 49 CpG probes present on the HumanMethylation450

arrays (503). These 49 sites together explain over 90% of the total variance. In order to reduce

the number of variables, the first two principal components associated with these 49 sites were

utilized for statistical analyses as described in Lemire et al. (514).

2.3.6 Statistics

Genotypic frequencies for the five SNPs investigated were compared between the sexes by

Pearson’s chi-square test and compared among ages by ANOVA. Methylation was compared

between groups using analysis of variance (ANOVA) with a significance level adjusted for

multiple comparisons. Groups compared were wildtype, heterozygous, and homozygous variant

groups of the SNP genotypes. Partial Pearson correlation was utilized to compare age and

methylation, controlling for sex and peripheral blood cell type proportions. Differences in

methylation among males and females were tested for association using logistic regression with

age at study recruitment and peripheral blood mononuclear cell type proportions as covariates.

Colon cancer diagnosis status was tested for association with percentage methylation for each

57

CpG site by logistic regression. Sex and age at study recruitment were used as covariates along

with peripheral blood mononuclear cell type proportions. To account for multiple comparisons of

the MLH1-region SNPs (70 CpG sites analyzed in total), a significant P-value of 7.14x10-4 was

applied to all analyses. To account for multiple comparisons of the MSH2 and MSH6 region (106

CpG sites analyzed in total), a significant P-value of 4.72x10-4 was used. All statistical analysis

was performed using IBM SPSS Statistics 21 (Armonk, NY).

2.4 Results

998 controls and 1,091 CRC cases from the Ontario Familial Colorectal Cancer Registry were

successfully analyzed for methylation levels across the genome spanning 450,000 CpG sites. A

mean correlation coefficient of 99.45% (range: 95.0– 99.9%) was calculated from the

comparison of methylation β-values between all duplicate pairs. Of these, 846 controls and 884

cases were successfully genotyped for rs1800734. 766 controls and 627 cases were genotyped

for rs749072, rs13098279, MSH2-118T>C, and MSH6-159C>T. Clinicopathological

characteristics of the population used in this study are shown in Table 2.1 along with genotypic

frequencies for the SNPs interrogated. There were no differences in SNP genotype frequencies

between different sexes (Pearson’s chi-square test, P = 0.06 for rs1800734, P = 0.07 for

rs749072, P = 0.08 for rs13098279, P = 0.20 for MSH2-118T>C, P = 0.14 for MSH6-159C>T) or

any associations between genotype and age (ANOVA, P = 0.48 for rs1800734, P = 0.64 for

rs749072, P = 0.58 for rs13098279, P = 0.35 for MSH2-118T>C, P = 0.12 for MSH6-159C>T).

No significant correlation was observed at any CpG probe between methylation and number of

years after study recruitment that blood was drawn. Results for the seven CpG sites in the MLH1

shore are discussed in the text, while results for all 70 CpG sites analyzed are shown in the

Appendix. Results of the MSH2-MSH6 region CpG sites are also reported in the Appendix.

Originally, there were methylation differences fou66yynd, genome-wide, between CRC cases

receiving chemotherapy prior to collection of blood compared to those without chemotherapy.

These differences were no longer apparent once peripheral blood mononuclear cell heterogeneity

was accounted for (513,514). The following results are given for all 884 CRC cases regardless

of chemotherapy status. Results using only cases without chemotherapy have been published in

Savio et al. (2012) (515). Regardless of chemotherapy status, PBMC type proportions using the

first two principal components, were controlled for where indicated.

58

Table 2.1 Characteristics of study population. Distribution of clinical features and SNP

genotypes in primary colorectal carcinoma cases and controls. Blood was drawn an average of

less than one year and no more than six years after study recruitment.

Controls CRC cases Characteristic N (%) N (%) Sex Female Male

356 (42.1) 490 (57.9)

516 (58.9) 360 (41.1)

Age Mean (SD) 64.3 (8.2) 64.1 (8.3) rs1800734 GG GA AA

!530 (62.6) 264 (31.2) 53 (6.3)

!519 (58.8) 321 (36.4) 43 (4.9)

rs749072 AA AG GG

438 (57.2) 271 (35.4) 57 (7.4)

333 (53.1) 258 (41.1) 36 (5.7)

rs13098279 GG GA AA

491 (64.1) 233 (30.4) 42 (5.5)

387 (61.7) 217 (34.6) 23 (3.7)

MSH2-118T>C TT TC CC

566 (74.0) 185 (24.2) 14 (1.8)

467 (74.5) 150 (23.9) 10 (1.6)

MSH6-159C>T CC CT TT

609 (79.6) 142 (18.6) 14 (1.8)

520 (82.9) 98 (15.6) 9 (1.4)

Percentages may not add to 100.0 due to rounding. SD – standard deviation

59

2.4.1 PBMC methylation differences among MLH1-region SNP genotypes

Methylation was compared in the MLH1 shore region between different SNP genotypes of

rs1800734, rs749072, and rs13098279 in controls. The mean methylation for each SNP genotype

(wildtype, heterozygous, or homozygous variant) was compared using ANOVA at 70 CpG sites.

The results for these CpG sites are shown in Appendix Table A1. The results of this analysis for

the MLH1 shore are shown in Table 2.2. There are seven CpG sites in the shore region of

interest, henceforth to be referred to as sites S1 through S7. The mean methylation in the MLH1

shore among individuals stratified by SNP genotypes was highest among the wildtype genotype

(GG) for rs1800734. The heterozygous genotype (GA) had intermediate levels of methylation

while the homozygous variant group (AA) had the lowest methylation. These differences in

methylation among genotypes were statistically significant for all 7 CpG sites localized to the

MLH1 shore region. For example, at S1 mean wildtype methylation was 0.746, heterozygous

methylation was 0.701, and homozygous variant methylation was 0.657 (P = 6.0x10-18). Similar

results were obtained for rs749072 and rs13098279. At S1 rs749072 mean wildtype methylation

was 0.746, heterozygous methylation was 0.707, and homozygous variant methylation was 0.666

(P = 1.5x10-13). For rs13098270 at S1 mean wildtype methylation was 0.746, heterozygous

methylation was 0.698, and homogyzous variant methylation was 0.644 (P = 3.0x10-17).

The same analysis was performed for CRC cases, shown in Table 2.3 for the MLH1 shore region

CpG sites. The results for all 70 CpG sites are shown in Appendix Table A2. By stratifying CRC

cases by genotype of rs1800734, the same significant pattern was observed as in controls, with

wildtype genotypes incurring significantly higher methylation than those with other genotypes.

For example, for rs1800734 at S1 in CRC cases, wildtype methylation was 0.742, heterozygous

methylation was 0.701, and homozygous variant methylation was 0.652 (P = 4.1x10-29).

Comparable significant results were found for rs749072 and rs13098279 (Table 2.3).

60

Table 2.2 Methylation between SNP genotypes in controls by ANOVA. Mean β-value of each

genotype of the SNPs rs1800734, rs749072, and rs13098279 in controls from Ontario at seven

CpG sites in the MLH1 shore. Chromosome 3 location (Genome Build 37) and probe IDs are the

same for CpG sites S1-S7 in subsequent tables.

Chr 3 Location

Probe ID CpG Site

Wild-type mean (SD)

Heterozygous Mean (SD)

Homozygous variant mean (SD)

P-value

rs1800734 N=530 N=264 N=53 37,033,373 cg02103401 S1 0.746 (0.081) 0.701 (0.092) 0.657 (0.102) 6.0x10-18

37,033,625 cg24607398 S2 0.885 (0.057) 0.855 (0.066) 0.836 (0.070) 3.9x10-17

37,033,632 cg10990993 S3 0.866 (0.053) 0.834 (0.063) 0.810 (0.067) 3.5x10-21

37,033,791 cg04726821 S4 0.220 (0.067) 0.186 (0.061) 0.150 (0.061) 1.5x10-22

37,033,894 cg11291081 S5 0.066 (0.031) 0.058 (0.024) 0.048 (0.020) 1.1x10-6

37,033,903 cg05670953 S6 0.167 (0.069) 0.147 (0.063) 0.120 (0.058) 2.4x10-8

37,033,980 cg18320188 S7 0.072 (0.021) 0.067 (0.019) 0.061 (0.017) 2.5x10-5

rs749072 N=438 N=271 N=57 37,033,373 cg02103401 S1 0.746 (0.082) 0.707 (0.090) 0.666 (0.102) 1.5x10-13

37,033,625 cg24607398 S2 0.884 (0.058) 0.860 (0.066) 0.841 (0.067) 1.9x10-9

37,033,632 cg10990993 S3 0.864 (0.053) 0.842 (0.064) 0.820 (0.069) 1.0x10-9

37,033,791 cg04726821 S4 0.218 (0.068) 0.191 (0.063) 0.163 (0.065) 6.3x10-12

37,033,894 cg11291081 S5 0.065 (0.031) 0.059 (0.024) 0.053 (0.026) 0.001

37,033,903 cg05670953 S6 0.165 (0.069) 0.150 (0.063) 0.127 (0.063) 2.6x10-5

37,033,980 cg18320188 S7 0.072 (0.022) 0.068 (0.018) 0.061 (0.018) 2.2x10-5

rs13098279 N=491 N=233 N=42 37,033,373 cg02103401 S1 0.746 (0.081) 0.698 (0.091) 0.644 (0.099) 3.0x10-17

37,033,625 cg24607398 S2 0.885 (0.057) 0.855 (0.067) 0.828 (0.067) 6.4x1016

37,033,632 cg10990993 S3 0.866 (0.053) 0.834 (0.063) 0.807 (0.069) 2.3x10-18

37,033,791 cg04726821 S4 0.219 (0.068) 0.183 (0.058) 0.150 (0.063) 3.8x10-20

37,033,894 cg11291081 S5 0.065 (0.031) 0.058 (0.023) 0.048 (0.021) 8.5x10-5

37,033,903 cg05670953 S6 0.165 (0.068) 0.146 (0.062) 0.117 (0.060) 7.5x10-7

37,033,980 cg18320188 S7 0.072 (0.021) 0.067 (0.018) 0.059 (0.018) 2.4x10-5

Significant results are bolded if P < 7.14x10-4. SD – standard deviation.

61

Table 2.3 Methylation between SNP genotypes in CRC cases by ANOVA. Mean β-value of

each genotype of the SNPs rs1800734, rs749072, and rs13098279 in CRC cases from Ontario at

seven CpG sites in the MLH1 shore.

CpG Site Wild-type mean (SD)

Heterozygous Mean (SD)

Homozygous variant mean (SD)

P-value

rs1800734 N=519 N=321 N=43 S1 0.742 (0.082) 0.701 (0.096) 0.652 (0.094) 4.1x10-29

S2 0.884 (0.056) 0.849 (0.068) 0.828 (0.067) 6.6x10-33

S3 0.858 (0.054) 0.823 (0.066) 0.811 (0.060) 2.5x10-38 S4 0.207 (0.070) 0.163 (0.062) 0.134 (0.049) 3.8x10-40 S5 0.065 (0.033) 0.056 (0.029) 0.048 (0.024) 6.9x10-6

S6 0.156 (0.070) 0.127 (0.067) 0.107 (0.051) 1.1x10-10

S7 0.072 (0.026) 0.062 (0.019) 0.059 (0.013) 1.6x10-9

rs749072 N=333 N=258 N=36 S1 0.736 (0.087) 0.707 (0.095) 0.659 (0.086) 1.7x10-7

S2 0.889 (0.058) 0.857 (0.067) 0.844 (0.070) 2.9x10-10

S3 0.863 (0.055) 0.835 (0.068) 0.824 (0.064) 1.9x10-8

S4 0.209 (0.072) 0.176 (0.064) 0.138 (0.049) 8.9x10-13

S5 0.063 (0.032) 0.057 (0.030) 0.042 (0.016) 2.2x10-4

S6 0.155 (0.072) 0.136 (0.070) 0.100 (0.047) 4.4x10-6

S7 0.072 (0.027) 0.064 (0.022) 0.060 (0.015) 4.0x10-5

rs13098279 N=387 N=217 N=23 S1 0.735 (0.086) 0.700 (0.096) 0.653 (0.098) 5.7x10-8 S2 0.886 (0.059) 0.854 (0.067) 0.841 (0.079) 1.2x10-9 S3 0.861 (0.057) 0.831 (0.067) 0.816 (0.067) 1.6x10-9 S4 0.208 (0.071) 0.170 (0.063) 0.124 (0.043) 0.001 S5 0.062 (0.031) 0.056 (0.031) 0.041 (0.018) 1.4x10-6 S6 0.154 (0.070) 0.132 (0.071) 0.089 (0.036) 6.2x10-5 S7 0.072 (0.027) 0.063 (0.021) 0.060 (0.015) 0.96 Significant results are bolded if P < 7.14x10-4. SD – standard deviation.

62

2.4.2 Age-related decrease in methylation at the MLH1 shore

Normally, as individuals age, global hypomethylation of the genome occurs combined with

increases in methylation at specific genes (215,216). To investigate whether the MLH1 shore

region exhibits age-associated changes in methylation, correlation analysis was performed shown

in Table 2.4. Results for all 70 MLH1-region CpG sites analyzed are shown in Appendix Table

A3. This was done in cases and controls separately to confirm whether any age-associated

changes in methylation at the shore were exclusive to CRC, or whether they occur in all

individuals. Several CpG sites within the MLH1 shore incur hypomethylation in both CRC cases

and controls. In controls, S1, S4, and S6 are significantly hypomethylated with increasing age.

For example, at site S1 R = -0.051 and P = 4.7x10-6. The exact same CpGs were not correlated in

CRC cases; hypomethylation of S2, S3, S4, and S6 were significantly associated with age. For

example, as S2 R = -0.147 and P = 1.3x10-4.

63

Table 2.4 Correlation between age and MLH1 shore methylation. Partial correlation,

controlling for sex and peripheral blood mononuclear cell proportions, between age and

methylation at seven CpG sites in the MLH1 shore for CRC cases and controls.

CpG Site Controls R P-value CRC Cases P-value S1 -0.051 4.7x10-6 -0.094 0.008 S2 0.029 0.004 -0.134 1.3x10-4 S3 0.069 0.007 -0.147 2.8x10-5 S4 -0.056 5.2x10-6 -0.174 6.6x10-7 S5 -0.052 0.25 -0.024 0.50 S6 -0.006 5.4x10-7 -0.135 1.1x10-4 S7 -0.038 0.27 0.036 0.31 Significant results are bolded if P < 7.14x10-4.

64

2.4.3 PBMC methylation differences among males and females at the

MLH1 region

Previous studies have demonstrated that MLH1 tumour hypermethylation at the CpG island is

more prevalent among female, MSI positive CRC patients. Therefore, I compared MLH1

methylation levels in PBMCs between males and females to determine whether sex plays a role

in this regard. The results for the MLH1 shore are found in Table 2.5, and for all 70 CpG sites

analyzed in Appendix Table A4. This was tested using logistic regression with age and

peripheral blood mononuclear cell proportions as covariates in all cases and controls. For most

CpG sites there were no significant differences in methylation between sexes. However, at CpG

site S5 and S6 methylation in females is significantly higher than in males. For example, at S5

the mean female methylation was 0.065 compared to 0.058 in males (P = 3.6x10-6, with effect

size (95% confidence interval) of 0.922 (0.891-0.954).

65

Table 2.5 Association between sex and methylation by logistic regression. Mean β-value is

shown for females and males along with logistic regression analysis at seven CpG sites of the

MLH1 shore. Analysis of male versus female methylation is adjusted for age and peripheral

blood mononuclear cell proportions.

CpG Site

Female Mean (SD)

Male Mean (SD)

P-value Effect Size

Lower 95% CI

Upper 95% CI

S1 0.725 (0.086) 0.724 (0.094) 0.59 0.997 0.987 1.007 S2 0.868 (0.062) 0.873 (0.065) 0.03 1.018 1.002 1.034 S3 0.845 (0.058) 0.851 (0.064) 0.01 1.021 1.005 1.038 S4 0.193 (0.069) 0.199 (0.071) 0.03 1.016 1.002 1.031 S5 0.065 (0.033) 0.058 (0.028) 3.6x10-6 0.922 0.891 0.954 S6 0.158 (0.071) 0.142 (0.066) 3.5x10-7 0.960 0.945 0.975 S7 0.069 (0.023) 0.068 (0.022) 0.59 0.988 0.947 1.032 Significant results are bolded if P < 7.14x10-4. SD – standard deviation; CI – confidence interval.

66

2.4.4 PBMC methylation differences among CRC cases and controls at

MLH1

Methylation was compared in the MLH1 gene region between 884 CRC patients and 845

controls. This association was tested between methylation level and presence of CRC (vs.

controls), utilizing sex, age, and cell type proportions as covariates by logistic regression (Table

2.6 and Appendix Table A5). A visual representation of case and control methylation at each of

the 70 sites analyzed is shown in Figure 2.3. At five sites of the CpG shore (S1-S4 and S6) there

was a significant association between hypomethylation and CRC risk. For example, at S1 mean

control methylation was 0.726 compared to 0.723 in CRC cases (P = 8.8x10-5, with effect size

(95% confidence interval) of 1.022 (1.011-1.033).

67

Table 2.6 Association between MLH1 shore methylation and CRC risk. Mean β-value of

controls and CRC cases is listed along with logistic regression analysis results at seven CpG sites

in the MLH1 shore. Analysis of CRC cases versus controls is adjusted for age, sex, and

peripheral blood mononuclear cell proportions. Effect size represents the increased risk of CRC

per 1% reduction in methylation.

CpG Site

Control Mean (SD)

Case Mean (SD)

P-value Effect Size

Lower 95% CI

Upper 95% CI

S1 0.726 (0.090) 0.723 (0.091) 8.8x10-5 1.022 1.011 1.033 S2 0.873 (0.063) 0.868 (0.064) 3.2x10-4 1.030 1.014 1.047 S3 0.852 (0.060) 0.843 (0.061) 3.8x10-7 1.045 1.027 1.063 S4 0.205 (0.068) 0.187 (0.070) 4.4x10-6 1.035 1.020 1.051 S5 0.063 (0.029) 0.061 (0.032) 0.64 1.008 0.975 1.042 S6 0.158 (0.068) 0.143 (0.070) 1.1x10-4 1.032 1.015 1.048 S7 0.070 (0.021) 0.067 (0.024) 0.10 1.037 0.993 1.084 Significant results are bolded if P < 7.14x10-4. SD – standard deviation; CI – confidence interval

68

Figure 2.3 Locations of CpG sites and methylation in CRC cases and controls. Pictured are

70 of the CpG sites analyzed in the MLH1 gene region, with indicated chromosomal positions

(Genome Build 37) located on chromosome 3. The CpG sites are located within the EPM2AIP1,

MLH1, and LRRFIP2 genes, with gene exons and transcriptional directions indicated. CpG

islands are indicated in green. The seven CpG sites of the MLH1 shore are highlighted in red.

Each vertical bar represents a CpG site, with control methylation (N = 846) displayed to the left

and CRC case methylation (N = 884) displayed to the right of the white dotted line. Controls and

CRC case samples are displayed layered horizontally from highest methylation to lowest

methylation. The distribution of degree of methylation in cases and controls is represented by the

colour variation, according to the scale.

69

2.4.5 No association between MSI status and DNA methylation

Methylation of the MLH1 promoter CpG island is a common occurrence in tumour tissue in MSI

CRC (168). There was no association between tumour MSI status and methylation at either the

MLH1 CpG island or shore in PBMC DNA of CRC cases when tested using logistic regression

with age, sex and peripheral blood mononuclear cell type proportions as covariates (data not

shown).

2.4.6 Methylation levels of the MLH1 CpG island and shore in PBMCs

The promoter of MLH1 spans from 37,034,130 to 37,034,856 (Genome Build 37) on

chromosome 3, -2711 to +15 relative to the MLH1 transcriptional start site (516). I investigated

the methylation status of this promoter CpG island in PBMC DNA. The Illumina Infinium

HumanMethylation450 microarrays contain 16 CpG sites located within the MLH1 promoter.

Overall, CpG island methylation is very low among both cases and controls in PBMCs, it does

not differ significantly when stratified by SNP genotypes, and is not significantly correlated with

age. The mean methylation for the CpG sites ranges from 0.004 to 0.067. Differences in

methylation among cases, controls, sex, and SNP genotypes and correlations with age can be

found for the promoter CpG island region in Appendix Tables A1-A5.

CpG shores flank CpG islands of some genes, located upstream and/or downstream. In addition

to the shore located upstream of the promoter CpG island, which is the focus of this

investigation, MLH1 also has a shore downstream of its island. There are only two CpG sites on

the Illumina array that interrogated methylation at this region, at 37,035,399 and 37,036,726.

Results, though not significant, for this methylation at this downstream shore can be found in

Appendix Tables A1-A5.

2.4.7 Association between MSH2/MSH6 SNPs and DNA methylation

106 CpG sites were interrogated on chromosome 2 surrounding MSH2 and MSH6 to further

explore PBMC methylation of DNA MMR genes. The MSH2-118T>C polymorphism was

associated with DNA methylation at five CpG sites in controls (Appendix Table A6), though

these were not all located near each other, nor did they all follow the same pattern. For example,

two of these significant CpGs (cg06478094, cg15582102) incurred the lowest level of

70

methylation in the heterozygous individuals while cg04175739 showed a significant decline of

methylation in TT to TC to CC genotypes. In CRC cases, four CpG sites were significantly

associated with MSH2-118T>C genotype, including three that were also significant in controls

(Appendix Table A7). The MSH6-159C>T SNP was also tested in controls and cases for

association with methylation (Appendix Tables A8-A9). In controls, three CpGs were

significantly associated with SNP genotype, all three of which had highest mean methylation in

heterozygous individuals. In CRC cases only two CpGs were significantly associated with

genotype, one of which was also significantly differentially methylated in controls

(cg19570558).

2.4.8 Association between methylation, age, and sex at the MSH2/MSH6

region

DNA methylation in the MSH2/MSH6 region was analyzed to determine whether it was

associated with age in controls or CRC cases, shown in Appendix Table A10. Methylation of

none of the CpGs was associated with age. Methylation in PBMC DNA at the MSH2/MSH6

region was also not significantly associated with sex (Appendix Table A11).

2.4.9 Differential PBMC DNA methylation between CRC cases and controls

at the MSH2/MSH6 region

DNA methylation in the MSH2/MSH6 region was analyzed to determine whether it was

associated with cancer status in PBMC DNA (Appendix Table A12). 49 of the 106 CpGs

(46.2%) were significantly associated with cancer risk. The significant CpGs were distributed

across each of the genes and not concentrated to a specific gene or functional region (eg. CpG

island or shore). However, interestingly, not all CpGs followed the same pattern.

Hypomethylation was associated with CRC at 27 of the significant sites while hypermethylation

was associated with CRC at the remaining 22 sites.

2.5 Discussion

In this study, methylation was measured in PBMC DNA of a large series of CRC cases and

controls using the Illumina Infinium HumanMethylation450 arrays. This methylation data was

71

integrated with SNP profiling data previously generated for the same controls and cases and

novel, significant associations at the MLH1 shore were found (479). This study demonstrated

that differences in MLH1 shore region methylation among PBMCs were significantly associated

with distinct genotypic variants in the MLH1 gene region. Specifically, a shore ~1 kb upstream

of MLH1 exhibited associations between methylation in PBMC DNA in controls and cases with

wildtype genotypes of SNPs located over 1 kb away (rs1800734, rs749072) and up to 200 kb

away (rs13098279) from this shore region. The variant alleles of these three SNPs are associated

with reduced methylation at CpG sites within the MLH1 shore, significantly lower than either the

heterozygous or wildtype alleles in PBMCs. Results also demonstrate that methylation of this

shore decreases with age at some of its CpG sites in controls and cases. Such associations

between PBMC methylation and genetic variants in a shore region have until now not been

described.

Though the concept of CpG islands dates back to the 1980s, CpG shores are a newer element of

methylation phenomena that has emerged in recent years (205,517). Shores are regions of the

genome that flank some CpG islands and have a lower GC content than islands do. Despite this

distance from genes and decreased CpG content, methylation of shores are reported to display

more specificity between different tissues, and between normal and cancerous cells from the

same patients (205,496,518). Gene expression is also strongly related to shore methylation

(205,232,496,518,519). In genome-wide methylation analysis, over 50% of the differentially

methylated regions between normal colon tissue and tumour tissue were located in shores, rather

than islands (205). CpG shore methylation has also been reported to be the most variable

genomic region when compared across a variety of normal and cancerous tissue types (202).

Studies have also demonstrated that shore methylation decreases with increasing age,

concomitantly with global hypomethylation (215). This is consistent with these MLH1 results,

which showed a decrease in methylation with increasing age at the MLH1 shore. Though much

remains to be discovered about the importance and regulation of CpG shores, methylation at

these regions shows potential at discriminating among different tissues, between normal and

diseased states, different genotypes, and age.

A number of CpG sites were identified that incurred methylation changes in CRC cases

compared to controls. At MLH1, five CpGs of the shore were hypomethylated in PBMCs of

cases compared to controls. An additional five CpGs located further downstream were also

72

hypomethylated, while two sites were hypermethylated, in controls. At the MSH2/MSH6 region

differential methylation was more pronounced, with 22 CpG sites hypomethylated in cases

compared to controls and 27 sites hypermethylated. Aberrant methylation events in tumour tissue

are linked to downregulation of tumour suppressor genes and upregulation of oncogenes. The

functional significance of methylation changes in PBMCs of CRC cases requires further

research. Methylation in PBMCs may be influenced by genetic background, including SNPs and

other forms of genetic variation (520). Diet may also potentially play a role in methylation in

PBMCs (505). Thus, methylation changes measured in PBMCs may be indicative of changes

also incurred within the tumour tissue itself. However, it remains to be established whether these

changes occur before disease onset or at early stages of CRC.

DNA sequence can affect methylation at nearby loci, as demonstrated in these results and other

studies (514,521,522). More recently it was verified that SNP-dependent DNA methylation

alterations could also play a role in disease (523–525). Our lab previously reported a significant

association between the MLH1 promoter SNP (rs1800734) and MSI CRCs, and subsequently

showed this association being mediated via MLH1 promoter hypermethylation and loss of MLH1

protein expression contributing to MSI CRC tumours (478,479). The role of this variant was

further assessed by measuring transcriptional activity of the MLH1 promoter CpG island of a

variety of cell lines, including CRC cell lines. Cells possessing the variant allele of rs1800734

exhibited decreased transcription compared to wildtype (480). Though rs1800734 was not found

to increase the overall risk of CRC, only the risk of the MSI phenotype of CRC, a subsequent

study found that among 10,409 cases and 6,965 controls the variant allele of this SNP is a

modest but significant risk factor for CRC overall, with a per allele odds ratio (95% confidence

interval) of 1.06 (1.00–1.11; P = 0.04) (524). These results have clearly demonstrated differences

in shore methylation of MLH1 between CRC cases and controls in PBMC DNA. Also, MLH1-

region SNPs show a strikingly significant association with shore methylation in the peripheral

blood of controls and CRC cases. Perhaps this variant-associated hypomethylation alone does

not cause cancer, but in combination with other genetic, epigenetic, and environmental

alterations in an individual, it may serve as a low-penetrance susceptibility marker.

Mutations in MSH2 and MSH6, two critical mismatch repair genes, are implicated in both

sporadic and familial CRC leading to MMR deficiency and MSI. The MSH2-118T>C SNP had

previously been shown by the Bapat lab to be associated with a strong family history of CRC

73

based on Amsterdam criteria I and II, especially in female CRC patients (511). While these SNPs

may alter transcription factor binding, as has been reported, based on the results presented in this

chapter it does not appear that they alter DNA methylation within the region in the same manner

that the MLH1 promoter SNP does (526,527). Increasing age in either CRC cases or controls

does not alter methylation in this chromosomal region in PBMC DNA, nor does sex play a role

in methylation. These results indicate that not all genes incur similar epigenetic control, even

genes which function together such as in DNA MMR. It also highlights the unique way that

MLH1 is epigenetically regulated at its CpG shore in PBMC DNA, in contrast to its CpG island

methylation in tumour DNA.

In addition to MLH1, MSH2, and MSH6, DNA methylation was also assessed at genes located

nearby, including EPM2AIP1, LRRFIP2, KCNK12, and EPCAM. EPM2AIP1 and MLH1 share a

bidirectional promoter and the shore is located within the single coding exon of EPM2AIP1.

Though ubiquitously expressed, the function of this gene is not well known, however, it plays a

role in glycogen synthesis in mice (528,529). LRRFIP2, located downstream of MLH1, contains

SNP rs749072 in its intron 27. This SNP was not associated with DNA methylation within the

LRRFIP2 gene itself. LRRFIP2 is required for the co-localization of several proteins (NLRP3,

ASC, F-actin) in the formation of the inflammasome complex which responds to various stimuli

in the inflammation process to activate caspase-1, but also plays a role in negative regulation of

NLRP3 inflammasomes (530,531). Interestingly, it has also been shown to have a role in the Wnt

signaling pathway, which is upregulated in the majority of CRCs. LRRFIP2 interacts with

disheveled segment polarity protein 1 (DVL1) leading to increased levels of cellular β-catenin

and TCF/LEF-dependent transcriptional activity (532). EPCAM is located directly upstream of

MSH2 and functions in cell-cell adhesion. Germline deletions of the 3’ end of EPCAM can result

in methylation of the MSH2 promoter and this genetic defect has been implicated in Lynch

syndrome (250,251). KCNK12 is located between MSH2 and MSH6 and is a potassium channel

protein. Its methylation has been implicated in CRC and pancreatic cancer and its expression was

associated with prediction and prognosis of lymphoma and leukemia in a small number of

studies (533–536). Though most of the genes surrounding the critical MMR genes have not been

particularly well characterized, especially regarding their role in CRC, it cannot be ruled out that

they may play a role, and that DNA methylation and/or SNP variation alters their functional

regulation.

74

One caveat concerning these results is the inability to ascribe our measured PBMC methylation

to a specific blood cell type. Peripheral blood mononuclear cells consist of natural killer cells, B

cells, T cells, and monocytes, each with their own epigenetic profiles. Genome-wide methylation

measurements have highlighted regions differentially methylated between different peripheral

blood cell populations (503). Also, peripheral blood mononuclear cell subpopulations change

with increasing age (537,538). Thus, we cannot say for certain whether the methylation changes

seen at the MLH1 shore are present in all PBMC types, or perhaps just in a certain subpopulation

of the cells, which may also be affected by age. Although the different proportions of blood cell

types were controlled for in this analysis, these results are unable to discriminate whether, for

example, the variant-associated hypomethylation seen is particularly pronounced in some PBMC

types but not others. What these results do show is that overall in PBMC samples, regardless of

cell populations, there are noticeable significant changes in methylation at the MLH1 shore

region.

Overall, this study has numerous strengths. The large sample size offers high statistical power

utilizing both CRC cases and controls. With more than 800 controls and cases significant

differences in methylation based on cancer status, age and SNP genotype were able to be

distinguished. Patient and control clinicopathological features have been extensively

characterized, as have the epigenetic and genetic features of the MLH1 gene region. This study

has now further described the epigenomic landscape of MLH1 by assessing methylation at its

CpG shore. This study also benefits from the use of PBMC DNA. Blood is an easily accessible

biological patient material, which can offer information about permanent changes such as

germline genetic variation as well as the epigenetic changes resulting in response to both genetic

and environmental sources. What remains to be seen is whether these patterns exist in other

tissues, such as the normal colon and rectum, and colorectal tumour tissue. Additional work for

the future includes further analyzing the data garnered from the Infinium HumanMethylation450

BeadChips, arrays which offer comprehensive genome-wide methylation analysis at nearly half a

million CpG sites. Thus far a small region of the genome has been studied and exciting

associations were found. Further probing of the methylomes of these CRC cases and controls

may reveal other genomic regions with detectable differences in methylation between cancer and

control, SNP variants, sex, age, tumour subtype, and other variables.

In summary, this novel study has demonstrated associations between SNP variants at

75

chromosome region 3p21 with methylation at a CpG shore of MLH1 in peripheral blood

mononuclear cells of 1,700 population-based controls and CRC patients. These results have also

shown an association with decreasing methylation at the shore with age, which may add another

facet to potential roles of shores and how they can incur methylation changes based on tissue,

presence of cancer, and environment. It is clear that SNP variants in the MLH1 region play many

roles in colorectal tumourigenesis, including the regulation of MLH1 methylation at its shore and

island.

76

Chapter 3 The dynamic DNA methylation landscape of the mutL homolog 1 shore is altered by MLH1-93G>A polymorphism in normal tissues

and colorectal cancer

3.1 Summary

Colorectal cancers (CRC) undergo distinct genetic and epigenetic alterations. MutL

homolog 1 (MLH1), a mismatch repair gene that corrects DNA replication errors, is lost in up to

15% of sporadic tumours due to mutation, or more commonly, due to DNA methylation of its

promoter CpG island. A single nucleotide polymorphism (SNP) in the CpG island of MLH1

(MLH1-93G>A or rs1800734) is associated with CpG island hypermethylation and decreased

MLH1 expression in CRC tumours. Further, as outlined in Chapter 2 of this thesis, in peripheral

blood mononuclear cell (PBMC) DNA of both CRC cases and non-cancer controls, the variant A

allele of rs1800734 is associated with hypomethylation at the MLH1 shore. In order to validate

these results using an alternative technique, I investigated the status of methylation at the MLH1

shore using the semi-quantitative real time PCR-based MethyLight assay. Further, to determine

whether this genotype-epigenotype association is present in other tissue types, including

colorectal tumours, I assessed DNA methylation in matched normal colorectal tissue, tumour,

and PBMC DNA from 349 population-based CRC cases recruited from the Ontario Familial

Colorectal Cancer Registry (OFCCR). MLH1 shore methylation was significantly higher in

tumour tissue than normal colon or PBMCs (P < 0.01). When shore methylation levels were

assessed stratified by SNP genotype, normal colorectal DNA and PBMC DNA incurred

significant hypomethylation in association with variant SNP genotype (P < 0.05). However, this

association was lost in tumour DNA. Among distinct stages of CRC, metastatic stage IV CRC

tumours incurred significant hypomethylation compared to stage I-III cases, irrespective of

genotype status. Shore methylation of MLH1 was not associated with MSI status or promoter

CpG island hypermethylation, regardless of genotype. To confirm these results, bisulphite

sequencing was performed in matched tumour and normal colorectal specimens from six CRC

cases, including two cases per genotype (wildtype, heterozygous, homozygous variant).

Bisulphite sequencing results corroborated the methylation patterns found by MethyLight, with

significant hypomethylation in normal colorectal tissue of variant SNP allele carriers. These

results indicate that the normal tissue types tested (colorectum and PBMC) experience dynamic

77

genotype-associated epigenetic alterations at the MLH1 shore, whereas tumour DNA incurs

aberrant hypermethylation compared to normal DNA.

3.2 Introduction

CRC develops as a result of the accumulation of genetic and epigenetic alterations. Aberrant

hypermethylation of CpG islands along with genome-wide hypomethylation is a common

signature in CRC (539,540). While a number of genes have been shown to incur methylation in

CRC, one of the best studied of these is the DNA mismatch repair (MMR) gene mutL homolog 1

(MLH1) (105,165,171,510). Loss of MLH1 or other MMR genes leads to the accumulation of

mutations, particularly at repetitive microsatellite regions leading to microsatellite instability

(MSI) (165,541,542). Approximately 15% of CRCs exhibit the MSI-high (MSI-H) phenotype,

and the majority of these cases have deficient MMR function due to hypermethylation incurred

at the MLH1 promoter CpG island (105,118).

Germline mutations of MLH1 or other MMR genes, including MSH2, MSH6, and PMS2, lead to

Lynch Syndrome accounting for approximately 2-5% of CRCs (31,247). Mutations in APC cause

familial adenomatous polyposis (FAP), occurring in <1% of CRCs (9,29). While these and

several other rarer germline gene mutations are known contributors to ~10% of CRCs, twin and

family studies have estimated the heritability of CRC to be up to 35% (543). SNPs have been

estimated to account for at least 7.42% of this heritability (544).

A number of genome-wide association studies (GWAS) have established susceptibility loci for

CRC, including at 8q24, 11q23, and others (284,457,545–547). It has previously been

demonstrated that a SNP variant in the promoter CpG island of MLH1 (rs1800734, MLH1-

93G>A) is associated with MLH1 CpG island hypermethylation, loss of protein expression, MSI

and overall increased risk of MSI-H CRC (478,479). A subsequent study implicated this SNP as

contributing to increased risk of CRC, though two meta-analyses refuted its role in CRC overall

while confirming risk for MSI CRC (524,548). While the overall status of this SNP as a risk

factor for CRC needs further clarification, what is clear is that it plays a role in MSI-H CRC and

MLH1 methylation status. Interestingly, further study of SNP rs1800734 in peripheral blood

mononuclear cell (PBMC) DNA using Illumina 450K methylation arrays indicated a different

phenomenon occurring upstream of this SNP and the CpG island in which it is located. At the

78

MLH1 shore in PBMCs of both CRC cases and controls significant hypomethylation was

observed in association with variant SNP genotype [Chapter 2 and (515)].

Although the majority of DNA methylation research has focused on CpG islands, whole-genome

methylation studies have shown that methylation changes at other non-coding regulatory regions

such as CpG shores and enhancers may also be implicated in tumourigenesis (205,549). CpG

shores are regions flanking some CpG islands that are less dense in CpGs than islands are.

Differential shore methylation has been shown to discriminate between normal and tumour DNA

in colorectal, prostate, and breast cancer, among other diseases (205,518,519).

While the mechanisms that direct DNA methylation patterns are not yet completely understood,

it is guided at least in part by DNA sequence (550–553). Through my research studies as well as

others in the Bapat Lab, DNA variant-associated CpG shore hypomethylation was demonstrated

in PBMCs while CpG island hypermethylation was shown in CRC tumours, both of which occur

in association with the same single nucleotide change (479,515). A figure demonstrating the

location of the island, shore, and SNP is shown in Figure 3.1. In this study, DNA methylation of

the MLH1 shore was investigated in a large cohort of 349 population-based CRC cases to

determine its association with rs1800734 SNP genotype in normal colorectal tissue, colorectal

tumours, and PBMCs of the same patients. These results indicate that static genetic variants can

dynamically modulate epigenetic regulation at the MLH1 gene region, and may play a role in

colorectal tumourigenesis.

79

Figure 3.1 MLH1 gene region, including its CpG shore, CpG island, and SNPs of interest.

Top: Region of chromosome 3 containing the genes EPM2AIP1, MLH1, and LRRFIP2,

interrogated for DNA methylation status in Chapter 2. Direction of transcription for each gene is

indicated by direction of arrowheads. Three SNPs in linkage disequilibrium are indicated by

white circles. Bottom: Zoomed in region of MLH1 promoter indicated by dashed grey line. CpG

island and shore, as well as SNP rs1800734, are indicated. A portion of the CpG island and all of

the CpG shore are located within the EPM2AIP1 gene.

80



Participants in this study were recruited through the Ontario Familial Colorectal Cancer Registry

(OFCCR), which is part of the Colon Cancer Family Registry, a consortium supported by the US

National Cancer Institute. Recruitment of primary CRC cases and controls was population-based,

and has been described previously (554). Briefly, residents from Ontario, Canada diagnosed

with primary CRC between June 1, 1997 and June 30, 2000 between the ages of 20 and 74 were

eligible for recruitment during Phase I. For Phase II, individuals with incident CRC under the

age of 50 diagnosed in Ontario between January 2003 and December 2006 were recruited.

Additional clinic-based recruitment was performed to recruit individuals diagnosed with CRC

above the age of 49 with fresh frozen tumour specimens available at the biospecimen repository.

Familial adenomatous polyposis cases were excluded from both Phase I and II. Cases with non-

white, mixed ethnic, or unknown background were excluded from the current study due to the

high proportion of self-reported Caucasians. Participants provided blood, tumour, and non-

neoplastic mucosa samples. Matched tumour and non-neoplastic mucosa (henceforth referred to

as ‘normal’ colorectal tissue) came from resected surgical specimens. These blood and tissue

samples were obtained with informed written consent following protocols approved by the

research ethics board of Mount Sinai Hospital and the University of Toronto.

3.3.2 Single nucleotide polymorphism genotyping

SNP selection and genotyping have been previously described (479). Peripheral blood

mononuclear cells (PBMCs) were isolated from blood samples of cases and controls by Ficoll-

Paque gradient centrifugation following manufacturer’s protocol (Amersham Biosciences, Baie

d’Urfé, Quebec, Canada). DNA was extracted from PBMCs by phenol-chloroform or Qiagen

DNA extraction kit (Qiagen Inc.,Hilden, Germany). The SNP rs1800734 was genotyped using a

fluorogenic 5’ nuclease polymerase chain reaction (PCR) assay. It was also genotyped using

Affymetrix GeneChip Human Mapping 100K and 500K platforms through the Assessment of

Risk of Colorectal Tumours in Canada (ARCTIC) project. Genotype of the five OFCCR Phase II

samples used for bisulphite sequencing was confirmed by Sanger sequencing at The Centre for

Applied Genomics (TCAG), The Hospital for Sick Children, Toronto, Canada. DNA from

81

samples was amplified by PCR for the region in the MLH1 promoter encompassing rs1800734

using primer sequences: (forward) 5’CGCCACATACCGCTCGTAGTA-3’ and (reverse) 5’-

TCCGTACCAGTTCTCAATCATCTC-3’. PCR was performed in 50 ul total volume per

reaction, with each reaction containing 5 µl of 10x PCR buffer, 6 µl of 50 mM MgCl2, 1 µl of 10

mM dNTP mix, 1 µl of 10 µM forward primer, 1 µl of 10 µM reverse primer, 0.4 µl of Platinum

Taq DNA polymerase, and 50 ng of DNA. PCR was performed using the following conditions:

95°C for 60 seconds, 35 cycles of [denaturation at 95°C for 30 seconds, annealing at 57°C for 45

seconds, extension at 72°C for 60 seconds], 72°C for ten minutes, and hold at 4°C. A 10 ul

aliquot of each PCR reaction was run on 1% agarose gel to confirm amplification and the

remaining PCR product was purified using ChargeSwitch PCR Clean-Up Kit (Invitrogen,

Carlsbad, CA). DNA concentration was assessed using a spectrophotometer, and 50 ng in a 7 µl

total volume was used for Sanger sequencing. Sequencing was performed at TCAG using an

internal primer (forward) 5’-GTCATCCACATTCTGCGGGA-3’.

3.3.3 Microsatellite instability analysis

PCR was performed on tumour and matched normal colorectal tissue DNA to compare MSI

patterns as described previously (555). Briefly, paraffin-embedded colorectal tumour tissue and

normal colorectal tissue from the same patients were microdissected for areas with more than

70% cellularity in tumour and normal cell populations. The MSI status was determined using the

National Cancer Institute guidelines, assessing four or more markers of: ACTC, BAT-25, BAT-

26, BAT-40, BAT-34C4, D10S197, D18S55, D17S250, D5S346, and MYC-L. Regions

containing each of these loci were amplified by PCR. MSI was indicated by the presence of

altered or additional bands of the amplified PCR product in tumour compared to normal. MSI

status was defined as MSI high (MSI-H) if ≥ 30% markers were unstable; MSI low (MSI-L) if 1-

29% of markers were unstable; and microsatellite stable if 0% of markers were unstable (163).

3.3.4 MethyLight

MethyLight was used to determine the DNA methylation status of the MLH1 shore in PBMCs,

normal colorectal tissue, and colorectal tumours of CRC cases. 50 ng of DNA was subject to

bisulphite modification with the EZ DNA Methylation-Gold Kit according to manufacturer’s

protocol and eluted to a final concentration of 10 ng/µl (Zymo Research Corp., Orange, CA).

82

Primers and probe were used to amplify a region of the MLH1 shore, with ALU-C4 primers and

probe used as control. Probes contained a 5’ fluorescent reporter dye and a 3’ quencher dye.

Sequences for the MLH1 shore are as follows: (forward) 5’-ATAGTTTTGATTAAGATTAGA-

GGCG-3’, (reverse) 5’-CGATGTTTGAATAATTGGTTTAGG-3’, and (probe) 5’-AGGCGAT-

TTGAATTTTAGATTTTATTAACGGAA-3’. Sequences for ALU-C4 are as follows: (forward)

5’-GGTTAGGTATAGTGGTTTATATTTGTAATTTTAGTA-3’, (reverse) 5’-ATTAACTAAA-

CTAATCTTAAACTCCTAACCTCA-3’and (probe) 5’-CCTACCTTAACCTCCC-3’. Each 30

µl PCR reaction contained a final concentration of 1X buffer, 200 µM dNTPs, 0.3 µM forward

primer, 0.3 µM reverse primer, 0.1 µM probe, 3.5 mM MgCl2, 0.01% Tween-20, 0.05% gelatin,

and 0.5 units of Taq polymerase. Samples were analyzed in duplicate in 96-well plates on an

ABI 7500 RT-PCR thermocycler. The PCR conditions for MethyLight are: initial incubation at

95°C for 10 minutes, followed by 50 cycles of denaturation at 95°C for 15 seconds and

simultaneous annealing and extension at 60°C for 60 seconds. Percent methylated reference

(PMR) score was calculated using the following calculation: [Gene of Interest/ALU-

C4]sample/[Gene of Interest/ALU-C4]CpGenome x 100%, where CpGenome represents

commercially available fully methylated CpGenome Universal Methylated DNA (Millipore,

Billerica, MA). In order to ensure that DNA quality was adequate, samples with an ALU-C4

threshold cycle greater than 22 were deemed poor quality and reanalyzed or removed from the

study (556). The cases selected for MethyLight profiling in this study were those from Phase I

OFCCR genotyped for SNP rs1800734 with available peripheral blood mononuclear cell, normal

colonic mucosa, and tumour DNA. MLH1 CpG island methylation was determined previously

using MethyLight for these cases (479).

3.3.5 Bisulphite sequencing

Bisulphite sequencing was performed to analyze DNA methylation in tumour and matched

normal colonic DNA from six CRC cases. Genomic DNA from formalin-fixed paraffin

embedded (FFPE) (one case from Phase I) or fresh frozen (five cases from Phase II) tissue was

used. DNA was treated with EZ DNA Methylation-Gold Kit for bisulphite conversion. Primers

located within the MLH1 shore were designed as follows: (forward) 5’-TTTGTTTGAGAAGTG-

GATTGTTGTTG-3’ and (reverse) 5’-TTTCTTCACTTAAAACTATTAAACTCC-3’. DNA was

amplified by PCR for each tumour and normal DNA sample. Each 50 ul PCR reaction contained

83

5 µl of 10x PCR buffer, 2 µl of 50 mM MgCl2, 1 µl of 10 mM dNTP mix, 1 µl of 10 µM forward

primer, 1 µl of 10 µM reverse primer, 0.2 µl of Platinum Taq DNA polymerase, and 50 ng of

DNA. PCR was performed using the following conditions: 95°C for 60 seconds, 35 cycles of

[denaturation at 95°C for 45 seconds, annealing at 50°C for 45 seconds, extension at 72°C for 45

seconds], 72°C for ten minutes, and hold at 4°C. A 10 ul aliquot of each PCR reaction was run

on 1% agarose gel to confirm amplification and the remaining PCR product was purified using

ChargeSwitch PCR Clean-Up Kit (Invitrogen, Carlsbad, CA). PCR products were cloned using

the pGEM-T Easy Vector System (Promega, Madison, WI) and MAX Efficiency DH5α

Competent Cells (Life Technologies, Carlsbad, CA) according to manufacturer’s protocol. The

ligation reaction contained 5 µl 2X Rapid Ligation Buffer, 1 µl pGEM-T Easy Vector (50 ng), 1

µl T4 DNA Ligase (3 Weiss units/µl), 6.3 ng DNA, and water to a final volume of 10 µl. The

reagents were mixed, incubated for one hour at room temperature, and incubated overnight at

4°C to maximize the number of transformants. 2 µl of each ligation reaction was placed into a

separate 14 ml polypropylene tube and stored on ice. MAX Efficiency DH5α Competent Cells

were thawed in an ice bath then 100 µl of cells was transferred to each tube containing the

ligation reaction. Tubes were gently mixed and stored on ice for 30 minutes. The cells were heat

shocked for 45 seconds in a 42°C water bath, then returned to ice for 2 minutes. 900 µl of SOC

medium was added to each transformation then incubated for 60 minutes at 37°C with shaking at

225 rpm in a Benchmark Incu-Shaker Mini. 100 µl of each transformation was cultured on LB

plates containing ampicillin, isopropyl β-D-1-thiogalactopyranoside (IPTG), and 5-bromo-4-

chloro-3-indolyl-β-D-galactopyranoside (X-gal). Plates were stored overnight at 37°C. At least

30 white colonies were selected for each sample and added to 14 ml tubes each containing 1.5 ml

SOC medium, 1.5 ml LB broth, and 50 µg ampicillin, then incubated overnight with shaking at

225 rpm at 37°C. Bacterial culture was then prepared using the QIAprep Spin Miniprep Kit

according to protocol (Qiagen, Hilden, Germany). Each reaction was tested for ligation of the

PCR amplicon into the pGEM-T vector using the restriction enzyme EcoRI. 1 µl FastDigest

EcoRI restriction enzyme, 2 µl 10x Fast Digest Green Buffer, 7 µl DNA prepared from the

QIAprep kit, and 10 µl water were mixed and incubated for 20 minutes at 37°C (Thermo Fisher

Scientific, Waltham, MA). 10 µl of this reaction was run on a 1% agarose gel. DNA

concentration was assessed with a spectrophotometer and 300 ng in a 7µl total volume from each

successful clone was sequenced by Sanger sequencing at TCAG with a M13 forward primer. At

84

least 15 clones were sequenced for each sample.

3.3.6 Statistics

MethyLight PMR values among tissue type were compared using a linear mixed model,

controlling for random and fixed effects to account for different DNA sources from the same

individual (PBMC, normal colorectal tissue, tumour tissue). PMR values between genotypes

were compared using ANOVA. Independent samples T-tests were used to compare methylation

with clinicopathological variables of cases. Despite significant P-values (<0.05) for the

Kolmogorov-Smirnov test for normality, histograms of the PMRs appeared normally distributed

and parametric tests were used for the data (557,558). A 6x4 contingency table and Pearson’s

chi-square tests were used to compare the sum of methylated CpGs between genotypes for each

sample (6 CpGs per clone with 15-27 clones per sample). All tests were performed using IBM

SPSS Statistics 21 with two-sided P < 0.05 defined as statistically significant (Armonk, NY).

3.4 Results

3.4.1 The MLH1 shore is hypermethylated in tumour DNA

The 349 CRC cases utilized for this project constitute only a subset of the total cases recruited

for the OFCCR. The 349 cases selected had been previously profiled on the Illumina Infinium

HumanMethylation450 arrays (Chapter 2), had rs1800734 genotype information, and had

sufficient DNA available from PBMC, normal colorectal tissue, and tumour tissue. These 349

cases did not differ significantly from the entire cohort of cases by age, sex, stage, or MSI status,

assessed by T-test (to compare ages) and chi-square tests (to compare sex, stage, and MSI).

MethyLight was performed to measure methylation at the MLH1 shore in DNA extracted from

PBMCs, normal colorectal tissue, and colorectal tumours of 349 CRC patients. Genotype and

clinicopathological variables for these cases is shown in Table 3.1. Mean methylation was

compared using a linear mixed model across the three tissue types that were analyzed (Figure

3.2A). Mean PMR [standard deviation (SD)] was 29.1% (4.5) in PBMCs, 30.5% (5.8) in normal

colorectal mucosa, and 33.3% (7.2) in tumour. MLH1 shore methylation was significantly higher

in tumour than normal colorectal tissue (P = 0.04) and PBMCs (P = 0.001). Mean methylation

did not differ significantly between normal colorectal tissue and PBMCs (P = 0.22).

85

Table 3.1 Distribution of clinicopathological features in primary colorectal carcinomas

from Ontario, including distribution among genotypes of rs1800734.

Feature All Genotypes N (%)

GG N (%)

GA N (%)

AA N (%)

Cases of primary colorectal carcinoma

349 211 (60.5) 119 (34.1) 19 (5.4)

Mean age (± SD)

61.9 (8.8) 62.0 (9.1) 62.0 (8.4) 60.8 (8.1)

Sex Female Male

163 (46.7) 186 (53.3)

95 (45.0) 116 (55.0)

58 (48.7) 61 (51.3)

11 (57.9) 8 (42.1)

TNM Stage 1 2 3 4 Unavailable

22 (6.3) 81 (23.2) 208 (59.6) 21 (6.0) 17 (4.9)

15 (7.1) 52 (24.6) 125 (59.2) 12 (5.7) 7 (3.3)

6 (5.0) 27 (22.7) 70 (58.8) 8 (6.7) 8 (6.7)

1 (5.3) 2 (10.5) 13 (68.4) 1 (5.3) 2 (10.5)

MSI Status Stable/Low High

287 (82.2) 62 (17.8)

181 (85.8) 30 (14.2)

95 (79.8) 24 (20.2)

11 (57.9) 8 (42.1)

MLH1 CpG Island Unmethylated Methylated Unavailable

300 (86.0) 34 (9.7) 15 (4.3)

187 (88.6) 17 (8.1) 7 (3.3)

98 (82.3) 14 (11.8) 7 (5.9)

15 (78.9) 3 (15.8) 1 (5.3)

Percentages may not add up to 100.0 due to rounding. SD – standard deviation

86

Figure 3.2 Mean percent methylated reference (PMR) of the MLH1 shore. MethyLight was

utilized to determine PMR in PBMC, normal colorectal tissue, and colorectal tumour DNA of

349 population-based CRC cases. All cases were genotyped for SNP rs1800734. There were 211

wildtype (GG), 119 heterozygous (GA), and 19 homozygous variant (AA) carriers. A. Mean

MLH1 shore methylation in each DNA source. B. Mean MLH1 shore methylation in each DNA

source stratified by genotype of rs1800734. C. Mean MLH1 shore methylation for each genotype

of rs1800734 stratified by DNA source. Error bars represent standard deviation. * P < 0.05 and

** P < 0.01.

87

3.4.2 MLH1 shore methylation in tumours is correlated with methylation in

PBMC and normal colorectal DNA

MLH1 shore methylation was assessed for correlation between the three different tissue types

tested (Table 3.2). Among all cases combined, regardless of genotype, there was no correlation

between PMR in PBMCs and normal colorectal methylation (r = 0.091, P = 0.09). There was a

significant positive correlation between PBMC and tumour DNA, with r = 0.122 and P = 0.02.

There was also a significant correlation between methylation in tumour and normal colorectal

DNA (r = 0.132 and P = 0.01). Stratified by genotype, there was a significant association

between PBMC and normal colorectal DNA among individuals with GA genotype of rs1800734

(r = 0.302, P = 0.001). There was also an association in methylation between normal colorectal

and tumour tissue (r = 0.163, P = 0.02). Overall, there is a moderate, yet significant, positive

correlation between methylation in tumour DNA with PBMC and normal colorectal DNA.

However, these correlations are weaker and/or not apparent once stratified by genotype of

rs1800734.

88

Table 3.2 Pearson correlation indicating association between methylation in MLH1 shore in

PBMC, normal colorectal, and tumour DNA from the same cases.

Comparison All Cases GG Cases GA Cases AA Cases PBMC and Normal r 0.091 0.008 0.302 -0.087 P-value 0.09 0.91 0.001 0.72 PBMC and Tumour r 0.122 0.109 0.175 0.127 P-value! 0.02! 0.11! 0.06! 0.60!Normal and Tumour!r! 0.132! 0.163! 0.122! -0.309!P-value! 0.01! 0.02! 0.19! 0.20!

89

3.4.2 The MLH1 shore is hypomethylated in variant SNP carriers in normal

DNA

Mean methylation at the MLH1 shore was compared between each genotype of rs1800734 in

PBMCs, normal colorectal tissue, and colorectal tumours by ANOVA (Figure 3.2B). Of 349

cases total, 211 were wildtype (GG), 119 were heterozygous (GA), and 19 were homozygous

variant carriers (AA). In PBMCs, mean methylation (SD) of GG, GA, and AA cases was 30.1%

(4.8), 28.3% (3.6), and 22.6% (3.0), respectively. This SNP-associated hypomethylation was

significant (P = 0.04). These findings in PBMC DNA utilizing RT-PCR-based MethyLight

technique confirmed previously published Illumina array-based results indicating SNP-

associated hypomethylation of the MLH1 shore region [Chapter 2, and (515)]. Comparing

methylation among genotypes in normal colorectal tissue, mean methylation (SD) of GG, GA,

and AA cases was 32.8% (6.5), 27.3% (4.5), and 24.0% (3.2), respectively, and these differences

were also significant (P = 0.005). In colorectal tumour samples stratified by genotype, mean

methylation (SD) of individuals with the wildtype GG genotype was 33.3% (7.4), while in GA

individuals it was 32.8% (7.0), and 37.0% (7.3) in AA individuals. Methylation did not differ

significantly in tumour DNA of CRC cases by SNP rs1800734 genotype (P = 0.73). Thus, DNA

methylation differences are associated with SNP genotype of rs1800734 only in the non-tumour

tissues tested, including normal colorectal tissue and PBMCs of CRC cases, whereas colorectal

tumours do not exhibit this genotype-epigenotype association.

3.4.3 Tumour hypermethylation at the MLH1 shore is driven by variant SNP

allele

Mean methylation was compared within each genotype of rs1800734 using a linear mixed

model, shown in Figure 3.2C, to examine the methylation patterns across normal and tumour

DNA at the MLH1 shore. Among individuals with the wildtype GG genotype of rs1800734 mean

methylation (SD) was 30.1% (4.8) in PBMCs, 32.8% (6.5) in normal colorectal tissue, and

33.3% (7.4) in tumour tissue, which did not differ significantly between DNA from different

sources (P = 0.15). In heterozygous individuals carrying the GA genotype mean methylation

(SD) was 28.3% (3.6) in PBMCs, 27.3% (4.5) in normal colorectal tissue, and 32.8% (7.0) in

tumour, which differs significantly between tissues (P = 0.008). Lastly, in homozygous variant

90

individuals carrying the AA genotype, mean methylation (SD) was 22.6% (3.0) in PBMCs,

24.0% (3.2) in normal colorectal tissue, and 37.1% (7.3) in tumour tissue, which also varied

significantly between tissues (P = 0.01). Overall, significant hypermethylation in tumours

compared to normal DNA is incurred only in individuals carrying one or two variant alleles,

either GA or AA.

3.4.4 MLH1 shore is hypomethylated in metastatic CRC, not associated

with CpG island methylation or MSI

Shore methylation was tested for associations with various clinicopathological variables of the

349 CRC cases that were assessed by MethyLight. Mean methylation (SD) in tumours of cases

with MSI-H phenotype was 32.9% (5.7), which did not differ from the 33.2% (7.5) in MSS/MSI-

L cases (P = 0.90). Stratified by rs1800734 SNP genotype methylation still did not differ

significantly between MSI-H and MSS/MSI-L cases. In GG individuals mean MSI-H

methylation was 30.9% (5.5) compared to 33.7% in MSS/MSI-L (7.7) (P = 0.44); in GA

individuals mean MSI-H methylation was 35.9% (6.6) compared to 31.5% (6.8) in MSS/MSI-L

(P = 0.35); lastly, in AA individuals mean MSI-H methylation was 31.7% (3.2) compared to

41.0% (9.0) in MSS/MSI-L (P = 0.32). I next tested whether or not methylation of the MLH1

CpG island in tumour DNA was associated with MLH1 shore methylation in tumour DNA of the

same individuals. The region tested at the MLH1 CpG island was located at -193 to -277 base

pairs relative to the MLH1 translation initiation site (TIS, measured from the A of the AUG start

codon), using CpG island methylation data generated by a former PhD student in the lab (479).

The MLH1 shore region analyzed was located at -1382 to -1499 base pairs relative to the MLH1

TIS. Cases were considered methylated at the MLH1 CpG island if PMR was greater or equal to

10%, as has been previously established (479,559). This threshold was shown to discriminate

between tumour and adjacent normal tissue and is sufficiently above background measurements

of methylation, yet lower than the PMR values usually obtained for most markers in colorectal

tumours (559). Mean MLH1 shore methylation was 33.4% (7.4) in tumour DNA of cases that

were unmethylated at the island while methylation was 30.4% (5.9) at the shore of cases that

were methylated at the CpG island, which did not differ significantly (P = 0.41). Since it has

previously been demonstrated that MLH1 CpG island hypermethylation is associated with

rs1800734 variant genotype, I next tested whether shore methylation was associated with CpG

91

island methylation stratified by genotype. In GG individuals CpG shore methylation was 33.7%

(7.8) when unmethylated at the island and 28.6% (5.3) when methylated at the island (P = 0.27).

In GA individuals CpG shore methylation was 31.7% (6.5) when unmethylated at the island and

34.7% (7.2) when methylated at the island (P = 0.68). In AA individuals shore methylation was

40.2% (7.8) when unmethylated at the island and 24.5% (2.7) when methylated at the island (P =

0.09). Thus, regardless of genotype MLH1 shore methylation is not associated with CpG island

methylation.

Tumour stage and MLH1 shore methylation in tumour DNA was compared. The mean

methylation (SD) of stage I-III tumours was 34.1% (7.3) versus 20.8% (3.6) in stage IV cases,

which was highly significantly different (P = 6.2x10-5). Hypomethylation in stage IV cases was

apparent for all cases, regardless of SNP genotype. In GG cases mean methylation was 33.6%

(7.4) for stage I-III and 22.4% (4.3) for stage IV (P = 0.02). In GA cases mean methylation was

33.7% (7.1) for stage I-III and 18.4% (2.4) for stage IV (P = 0.001). In AA cases mean

methylation was 42.0% (7.5) for stage I-III and 19.4% for one stage IV case. Since there was

only one stage IV AA case a P-value cannot be calculated for this comparison.

MLH1 shore methylation level was not significantly associated with MSI status, tumour MLH1

CpG island hypermethylation, or tumour stage in PBMCs or normal colorectal tissue for all cases

or when stratified by SNP genotype of rs1800734 (all P > 0.05).

3.4.5 Bisulphite sequencing confirms SNP-associated hypomethylation of

MLH1 shore in normal colorectal DNA

Bisulphite sequencing was performed on a 232 base pair region of the MLH1 shore containing

six CpGs, four of which were also included in the MethyLight primer and probe sequences. The

CpGs were located at -1550, -1477, -1439, -1414, -1407, and -1377 relative to the MLH1 TIS

and will be referred to as CpGs 1-6, respectively (Figure 3.3A). CpGs 2-5 were included in the

MethyLight amplicon assessed. Normal colonic DNA and tumour DNA was analyzed from six

CRC cases, comprising of 2 samples with each genotype (GG, GA, and AA). 15 to 27 clones

were sequenced for each sample, and the methylation patterns are shown in Figure 3.3B. Visual

inspection of the methylation patterns shows hypomethylation in normal colorectal DNA in GA

and AA cases compared to GG.

92

The total number of methylated CpGs at each site was compared between genotypes for normal

and tumour samples, shown in Figure 3.3C. All genotypes were compared against each other at

CpGs 1 to 6 in normal colorectal tissue. DNA methylation as assessed by bisulphite sequencing

was statistically significantly different between genotypes. P-values for each of the 6 CpG sites

were: CpG 1 P = 2.7x10-4; CpG 2 P = 3.2x10-10; CpG 3 P = 0.001; CpG 4 P = 5.0x10-4; CpG 5 P

= 3.5x10-4; CpG 6 P = 3.6x10-7. These results follow the same significant pattern that was

observed using the MethyLight technique; normal colorectal tissue incurs hypomethylation at the

MLH1 shore in individuals with variant rs1800734 genotype.

The number of methylated CpGs was compared between genotypes in tumour samples.

Comparing GG vs. GA vs. AA genotypes CpG 5 was significantly differentially methylated (P =

0.002). Though CpG 5 showed differential methylation, all other CpG site comparisons between

genotypes were not significantly statistically different. This corresponds with the MethyLight

findings, in which tumour DNA methylation of the MLH1 shore did not differ amongst

genotypes of rs1800734. Methylation was also compared between normal and tumour DNA (not

shown), and methylation at none of the six sites was significantly different (all P > 0.05).

93

Figure 3.3 Description on next page.

94

Figure 3.3 Bisulphite sequencing results for six CpG sites of the MLH1 shore in six normal

colorectal tissue samples and matched colorectal tumours from the same cases. A. The

location of the bisulphite sequencing amplicon relative to the MLH1 island and shore is

indicated. Position of each CpG interrogated relative to the MLH1 translation initiation site is

shown. Each CpG is numbered 1 to 6 in a 5’ to 3’ direction when referred to subsequently in the

figure. B. Methylation patterns in six normal colorectal tissue samples and matched tumours at

the MLH1 shore, with rs1800734 genotype indicated. Empty circles represent unmethylated

CpGs and filled in circles represent methylated CpGs. CpG site (1-6) is indicated. QUMA

(QUantification tool for Methylation Analysis) was utilized in the creation of this figure. C.

Graphical representation of bisulphite sequencing results. For each sample at each CpG site, the

percent of methylated CpGs was calculated. The mean percent of methylated CpGs was then

calculated for each genotype and tissue source grouping: GG normal, GA normal, AA normal,

GG tumour, GA tumour, and AA tumour. Pearson’s chi-square test was used to compare the total

number of methylated CpGs at each CpG site. Error bars represent standard deviation. ** P <

0.01; *** P < 0.001

95

3.4.6 Examination of CpG island and shore methylation in normal and

tumour tissues

The results outlined in Chapter 2 and this chapter of my thesis have revealed the methylation

patterns among rs1800734 genotypes at the MLH1 shore in PBMCs, normal colorectal tissue,

and colorectal tumour tissue as well as methylation at the MLH1 CpG island in PBMCs (450K

results of Chapter 2). Previous data that was generated by the Bapat Lab has interrogated MLH1

CpG island methylation in normal colorectal and tumour DNA (479). Figure 3.4 integrates

methylation data from this project with previous data generated by the Bapat Lab to show the

shifting epigenetic patterns at the MLH1 region. SNP genotype is associated with the opposite

direction of methylation at the CpG shore and island in normal and tumour DNA.

96

Figure 3.4 Schematic model of DNA methylation at the MLH1 CpG island and shore. In

PBMCs and normal colorectal tissue the MLH1 shore incurs hypomethylation in association with

variant SNP genotype of rs1800734. No methylation is present at the CpG island in these DNA

sources. In colorectal tumour, DNA methylation at the CpG shore loses its association with

rs1800734 genotype whereas the CpG island incurs hypermethylation in association with variant

SNP genotype of rs1800734. TSS – transcription start site.

97

3.5 Discussion

Aberrant methylation changes are a hallmark of all cancers, including CRC. Though the general

pattern observed in tumour DNA includes CpG island hypermethylation with genome-wide

hypomethylation, the findings of the present study demonstrate that methylation may be altered

in a tissue-, locus-, and genotype-specific manner. The critical mismatch repair gene MLH1

incurs CpG island hypermethylation in a subset of CRC cases. I have investigated its upstream

shore and determined that there is dynamic interplay between genotype and epigenotype in

normal and tumour DNA of CRC patients. In normal colorectal tissue, DNA methylation is

present at the CpG shore but is hypomethylated in individuals carrying one or two variant alleles

of the rs1800734 SNP. I found this same association in PBMC DNA of the same cases,

validating previous array-based results discussed in Chapter 2. I also found that this SNP-

associated methylation pattern is lost in tumour DNA, due to increases in CpG shore methylation

in tumour compared to normal colonic DNA of SNP variant carriers. These findings establish

that the static genetic sequence can modulate epigenetic marks in normal tissues. These results

also provide further evidence of shore methylation changes in matched tumour versus normal

DNA.

This study delved into DNA methylation patterns at the MLH1 shore, which builds upon

previous studies of MLH1 and its promoter SNP. The variant A allele of rs1800734 was first

shown to be associated with MSI-H CRCs, then with MLH1 CpG island hypermethylation

(478,479). Subsequent studies have also shown an association between this SNP and endometrial

and lung cancer risk, as well as worse outcome in oral squamous cell carcinoma (560–562).

CpG shores were first described as regions up to 2 kilobases away from CpG islands that are less

dense in CpGs (205). The original publication and subsequent studies have demonstrated that

shore methylation differs between different tissue types (202,205,563). However, in the two

‘normal’ DNA sources assessed, from colorectum and PBMCs, there were no significant

differences in MLH1 shore methylation. Having only examined two sources of non-cancer

tissues, it cannot be said for certain what methylation patterns would be seen in other normal

tissues from these patients at this specific region. It has also been shown that methylation

significantly differs between normal and tumour DNA at CpG shore regions in multiple cancer

types (205,496,518,519). MethyLight results in this study investigating 349 CRC cases have

98

indicated tumour hypermethylation at the MLH1 shore, agreeing with other reports. Another key

feature of shores is that they have been shown to have a stronger negative correlation between

methylation and gene expression than CpG islands (205). However, it was also found that certain

subsets of genes with unmethylated islands and methylated shores, termed ‘ravines’, in fact had

high transcriptional activity and a more transcriptionally permissive state including higher

DNase sensitivity and RNA polymerase II binding (203). In normal DNA and CRC tumours

without MLH1 CpG island hypermethylation, a similar pattern is seen, with methylation at the

shore and no methylation at the CpG island. In fact, the MLH1 promoter region has been well

characterized, and the link between decreased expression and hypermethylation at specific

regions in its CpG island have already been established (516,564). Thus, methylation at the

MLH1 shore likely does not play a large role, if any, in MLH1 expression though perhaps it

functions in other ways to create a transcriptionally permissive state as in the aforementioned

‘ravines’ (203). The exact mechanism or functional role for shores remains to be elucidated.

Much research has been focused toward discerning disease-associated SNPs. The majority of

SNPs mapped in GWAS are located in non-coding regions of the genome, thus, establishing the

function of SNPs has been difficult (434,565,566). Rather than altering protein function it has

been postulated that variant SNPs cause changes in gene expression levels (434,567). rs1800734

has not clearly emerged from CRC GWAS, though a large study has provided evidence that this

SNP is a risk factor for CRC in a study of 10,409 CRC cases and 6,965 controls with a

significant per allele odds ratio of 1.06 (524). However, other meta-analyses have not

corroborated these findings (524,548). Regardless of its influence on overall CRC incidence,

substantial evidence exists for association of rs1800734 with the MSI-H subtype of CRC

(478,479). In addition to methylation changes, the Bapat Lab has also shown functional changes

incurred due to the variant SNP genotype. Specifically, promoter constructs were created with

either the G or A allele and transfected into a variety of cell lines including the CRC line HCT

116 and normal colonic cell line CCD-841-CoTr (480). The variant A allele exhibited

significantly less luciferase activity than the G allele in all cell lines tested. The lab also showed,

through electrophoretic mobility shift assay (EMSA) experiments in HCT 116 and CCD-841-

CoTr, the presence of a DNA binding factor(s) with high affinity for the G allele but not A (480).

This work was replicated in HeLa cell nuclear extract by others (568). Which exact protein(s) is

able to bind G but not A at this sequence has not yet been elucidated. Active promoters bound by

99

transcription factors and RNA polymerase II are more resistant to incurring DNA methylation

than inactive promoters (569,570). Therefore, if a transcription factor is unable to bind at the A

allele, this likely provides the link between variant SNP genotype and increased CpG island

methylation, but does not yet provide a mechanism for methylation changes at the shore.

These results have shown an association between a SNP and methylation at the MLH1 CpG

shore in normal colorectal tissue and PBMCs, in contrast to previous studies, which

demonstrated a SNP-methylation association at the CpG island in tumour DNA. Tumour

methylation of the CpG island is associated with MSI-H whereas shore methylation does not

show any such association. The reasons for this are unclear. Perhaps the methylation changes at

the CpG shore and CpG island are two independently regulated events. In addition to DNA

methylation, changes to histone modifications are also seen in cancer, and may precede DNA

methylation changes. For example, Polycomb repressive complex 2 is recruited to the DNA,

where it lays down repressive histone H3 lysine 27 trimethylation (H3K27me3) marks, then

recruits DNA methyltransferases, which methylate the DNA (571,572). Since underlying DNA

sequence plays a role in recruitment of these modifying complexes, SNP genotype may in fact

alter recruitment of writers, erasers, or readers of histone modifications, which in turn would

alter DNA methylation patterns at MLH1 (571,572). Histone modifications present at the MLH1

CpG island have been previously studied. For example, repressive H3K9me1/2/3 and

H3K27me2/3 are found at higher levels in the RKO CRC cell line, which is hypermethylated at

the MLH1 CpG island, compared to the unmethylated SW480 CRC cell line (573). Conversely,

acetylated histone H3, an active histone modification, is present at the MLH1 CpG island in the

unmethylated LD419 cell line but not in methylated RKO or SW48 CRC cell lines (574). Further

research into the histone modifications present at the shore, taking into account rs1800734

genotype, will elucidate whether epigenetic regulation beyond DNA methylation is altered at

MLH1 in CRC.

An advantage of this study is the availability of DNA from a large population-based cohort from

Ontario, Canada. This cohort is well established with DNA available from matched blood,

normal colorectal tissue, and tumour samples, which enabled the ascertainment of a

comprehensive view of methylation patterns. The methylation differences measured between

genotypes were significant, but relatively subtle. For example, in normal colorectal tissue

methylation was 32.8% in GG individuals, 27.3% in GA, and 24.0% in AA. Despite this, the

100

data still demonstrated a SNP association in just six samples used for the bisulphite sequencing

experiments. Across multiple techniques in two types of normal DNA sources, whether a large or

small sample size, the genotype-epigenotype association at the MLH1 shore is significant.

A caveat to this study is that every CpG of the MLH1 shore cannot be individually examined due

to limitations of the techniques used. MethyLight is a real time PCR-based method that is limited

to approximately 150 base pair amplicons. This technique is only able to detect methylation if all

CpGs in the primer and probe sequences are methylated and does not account for variable

methylation patterns. Bisulphite sequencing will account for variable methylation patterns that

MethyLight cannot, and one can sequence a region up to 300 base pairs. However, this method is

low-throughput and not well suited for analysis of a large number of samples. Despite

differences in technique, I was still able to detect significant SNP-associated hypomethylation at

the MLH1 shore in normal DNA. However, there are discrepancies in the methylation values.

The mean methylation in normal colorectal tissue as assessed by MethyLight in 211 GG

individuals was 32.8% and with bisulphite sequencing in 2 cases was ~80%. This may be due to

differences in the measurement of methylation with MethyLight versus bisulphite sequencing,

but may also be due to the fact that only a small number of cases were used for bisulphite

sequencing. Perhaps if all 349 cases were measured by bisulphite sequencing the results would

be more concordant. This same explanation can also be used to explain the fact that MethyLight

was able to show significant hypermethylation in colorectal tumour DNA compared to normal

colorectum, while bisulphite sequencing did not. Another limitation is that due to the large

distance of approximately 1.5 kilobases between the shore and the promoter SNP it would be

difficult to show a direct link between SNP genotype and hypomethylation in heterozygous

cases. The bisulphite conversion step to prepare DNA for analysis fragments the DNA, and the

samples used were FFPE, which also consists of shorter fragmented DNA sequences. Thus I was

unable to assess shore methylation and SNP genotype on the same DNA strand.

Another caveat to keep in mind is that circulating tumour cells (CTCs), after spreading from the

primary tumour, travel through the circulatory system and may be detected in the blood. In fact,

CTCs are a prognostic marker for worse outcome in patients with metastatic CRC, and have also

been shown to be significantly associated with worse outcome in stage I-III patients (575,576).

Though more specialized protocols have been developed to isolate CTCs, the traditional method

of Ficoll-Paque gradient centrifugation to isolate PBMC from whole blood (as performed for the

101

OFCCR samples used in this analysis) is also able to isolate CTCs (577). Thus, there is a chance

that a proportion of the methylation measured in PBMCs is actually coming from tumour DNA.

However, tumour DNA likely does not make a significant contribution, as CTCs have been

shown to have extremely low concentrations in the blood, as low as one cell out of one billion

(578). Another potential limitation of this study is that there were only 19 cases with the AA

genotype available for analysis. No SNP-associated hypomethylation was observed in tumour

DNA, nor were there any associations between shore methylation and island methylation or MSI

status. Performing similar studies with a larger number of AA cases would confirm these results

with more certainty.

In summary, these results demonstrate an association between the promoter MLH1-93G>A

polymorphism rs1800734 and DNA hypomethylation at the MLH1 shore in normal colorectal

tissue, and also confirmed this in PBMCs, building upon my previous findings (515). This

association is not evident in tumour DNA from the same cases, but instead, as previously

demonstrated, this polymorphism is associated with hypermethylation at the CpG island in MSI-

H CRC (479). These results reveal that the epigenetic landscape of MLH1 is dynamically

regulated at least in part by the static genetic sequence. Additional characterization of epigenetic

and/or transcriptional regulation at the MLH1 CpG island and shore, taking into account

rs1800734 genotype differences, may lead to insight into mechanisms by which polymorphisms

contribute to cancer risk.

3.6 Acknowledgements

We thank Graham Casey for providing initial genotype information for the five OFCCR Phase II

cases utilized for bisulphite sequencing, which was subsequently confirmed by Sanger

sequencing as described in Section 3.2.2.

102

Chapter 4 Exploration of DNA methylation, histone modifications, and transcription factors of the MLH1 CpG island and shore in

colorectal cancer cell lines

4.1 Summary

MutL homolog 1 (MLH1) plays an important role in DNA mismatch repair and its expression is

frequently lost due to CpG island hypermethylation in colorectal cancer (CRC). The promoter

polymorphism MLH1-93G>A (rs1800734) is associated with decreased promoter activity, MLH1

CpG island hypermethylation, and MLH1 expression loss in colorectal tumours. Conversely, this

same variant genotype is associated with MLH1 shore hypomethylation in normal tissues,

including peripheral blood mononuclear cell (PBMC) and normal colorectal DNA as outlined in

previous chapters. To further explore this inverse association and to gain insight into this

regulation I characterized DNA methylation across the entire promoter CpG island and shore

region of MLH1 in five CRC cell lines by bisulphite sequencing. Cell lines containing one or two

variant alleles of rs1800734 demonstrated hypomethylation at the MLH1 shore compared to

wildtype cell lines. Chromatin immunoprecipitation (ChIP) analysis for selected histone marks

and transcription factors was then performed on CRC cell lines representing each genotype of

rs1800734. There was significant enrichment of the transcription factor AP4 at the MLH1

promoter region of the two cell lines containing the G allele (GG and GA genotypes) but no

enrichment in the AA cell line. Sanger sequencing confirmed preferential binding of AP4 to the

G versus A allele in the GA cell line. By performing ChIP for histone modifications in these

same cell lines I detected active histone modifications H3K4me3 and H3K27ac present in equal

amounts across all three cell lines in a pattern indicating active transcription, despite differences

in MLH1 shore methylation or genotype. The enhancer-associated modification of H3K4me1

was enriched at the MLH1 shore, potentially in a genotype-specific manner, however, H3K27ac,

another enhancer associated mark, was low in the shore region. No enrichment for H3K27me3

was found in any cell line. These results demonstrate the role of variant rs1800734 in altering

DNA methylation and histone modifications at regions beyond the MLH1 CpG island in which it

is located, possibly playing a role in multi-dimensional regulation of this region. Results also

provide evidence for a genotype-specific relationship between lack of transcription factor

103

binding, decreased promoter activity, and CpG island hypermethylation of the essential

mismatch repair gene MLH1 in CRC.

4.2 Introduction

Three single nucleotide polymorphisms (SNPs) in the MLH1 gene region (rs1800734, rs749072,

and rs13098279) are associated with MLH1 promoter CpG island hypermethylation and MSI

CRC (478,479). Specifically, rs1800734, located 93 base pairs upstream of the MLH1 translation

initiation site (TIS), has also shown association with risk of the MSI CRC subtype, as well as

other neoplasms including glioblastoma, gastric, lung, and ovarian cancers (481–484,568). This

SNP may also be a risk SNP for CRC overall (524). It has also been found that the allelic variant

of rs1800734, located in the MLH1 promoter CpG island, has a functional consequence in that it

decreases the transcriptional activity of MLH1 (480). As described in the previous two chapters,

the MLH1 shore incurs hypomethylation in association with SNP genotype of rs1800734 in

peripheral blood mononuclear cell DNA of CRC cases and controls as well as normal colorectal

tissue DNA of CRC cases (515).

This SNP also likely impacts epigenetic control and transcriptional regulation beyond DNA

methylation including histone modifications controlling chromatin activity and transcription

factor binding. DNA is packaged in the nucleus by wrapping around nucleosomes consisting of

an octamer of histone proteins. Control of gene expression is in part regulated by the degree of

accessibility of the DNA to transcription factors or other transcriptional machinery, which is

facilitated by presence or absence of histones, histone modifications, and histone variants (234).

Histones may be post-translationally modified on their protruding amino-terminal tails by

acetylation, methylation, phosphorylation, ubiquitylation, glycosylation, sumoylation, ADP-

ribosylation, and carbonylation (579–581). Just as presence or absence of DNA methylation can

alter gene expression, histone modifications can also alter the repression or activation of DNA.

The most widely studied histone modifications are methylation and acetylation marks added to

various lysine residues of histone H3. Polycomb repressive complex 2 (PRC2) consists of a

complex of proteins including EZH2, SUZ12, and EED and catalyzes the histone H3 lysine 27

trimethylation (H3K27me3) mark (571,572,582). PRC2 also recruits other polycomb

complexes, histone deacetylases and DNA methyltransferases, which act together to further

repress transcription and compact the chromatin in that region. The H3K27me3 mark is

104

counteracted by the Trithorax group complex, in which the histone writer MLL lays down

activating H3K4 methylation marks and the histone demethylase KDM6A removes H3K27me3

(201,583,584). A wide variety of histone writers, readers, and erasers and chromatin remodelling

proteins exist to further regulate the DNA. H3K4me1 and H3K27ac usually mark active

enhancer regions while H3K4me3 and H3K27ac usually mark active promoter regions

(581,585). In addition to these marks, a variety of transcription factors and transcriptional

machinery such as RNA polymerase are also required for active transcription to occur.

Transcription factors contain DNA-binding domains, which allow them to bind to specific DNA

sequences. Disruption of this sequence, such as due to occurrence of a SNP, may prevent the

factor from binding. Other types of chromatin binding factors exist that prevent the interaction of

promoters and enhancers and/or silencer regions, called insulators. Proteins such as CCCTC-

binding factor (CTCF) act as insulators and can prevent spreading of DNA methylation to

maintain regions that are free of methylation (211,586). This chapter outlines the work that I

performed to assess DNA methylation, histone modifications, CTCF, and the transcription factor

AP4 at the CpG island and shore of MLH1 and demonstrates how their presence is modulated by

SNP genotype of rs1800734.


4.3.1 Cell lines

The colorectal carcinoma cell lines COLO 320HSR, HCT-15, HCT 116, LS 174T, and SNU-

C2B were purchased from American Type Culture Collection. HCT 116 cells were cultured in

McCoy’s 5A Medium Modified. LS 174T cells were cultured in Eagle’s Minimum Essential

Medium. COLO 320HSR, HCT-15, and SNU-C2B cells were cultured in RPMI-1640. All cell

culture media were supplemented with 10% fetal bovine serum. All cell lines were maintained in

a humidified incubator at 37°C with 5% CO2.

4.3.2. Cell line genotyping

DNA from COLO 320HSR, HCT-15, HCT 116, LS 174T, and SNU-C2B cell lines was isolated

using QIAamp Mini and Blood Kit (Qiagen, Hilden, Germany). The colorectal carcinoma cell

lines Caco-2, HT-29, LS 180, SW48, SW480, and SW620 were obtained for screening of

105

rs1800734 genotype, courtesy of Colleen Ash (Gallinger lab) and George Karagiannis

(Diamandis lab). DNA was isolated from these cell lines in the same manner using the QIAamp

Mini and Blood Kit. 40 ng of DNA per cell line was amplified by PCR to amplify the region

surrounding the MLH1 promoter SNP rs1800734, as described in Chapter 3. Primers are listed in

Table 4.1. PCR products were sequenced by Sanger sequencing at The Centre for Applied

Genomics (TCAG) DNA Sequencing Facility, The Hospital for Sick Children, Toronto, Canada.

An external forward primer was used for Sanger sequencing (Table 4.1).

106

Table 4.1 Primers used for PCR reactions. Forward, reverse, and probe (where applicable)

sequences for PCR-based applications.

Gene Region Forward Reverse Probe Application MLH1 CpG Island

CGCCACATACCGCTCGTAGTA

TCCGTACCAGTTCTCAATCATCTC

N/A rs1800734 Genotyping (External)

MLH1 CpG Island

GTCATCCACATTCTGCGGGA

N/A N/A rs1800734 Genotyping (Internal)

MLH1 CpG Island

GAGCGGATAGCGATTT

TCTTCGTCCCTCCTAAAACG

CCCGCTACCTAAAAAAATATACGCTTACGCG

MethyLight

MLH1 Shore ATAGTTTTGATTAAGATTAGAGGCG

CGATGTTTGAATAATTGGTTTAGG

AGGCGATTTGAATTTTAGATTTTATTAACGGAA

MethyLight

ALU-C4 GGTTAGGTATAGTGGTTTATATTTGTAATTTTAGTA

ATTAACTAAACTAATCTTAAACTCCTAACCTCA

CCTACCTTAACCTCCC

MethyLight

Amplicon A: MLH1 CpG Island

GTTAGATTATTTTAGTAGAGGTATATAAGT

CCTTCAACCAATCACCTCAATACC

N/A Bisulphite Sequencing

Amplicon B: MLH1 Middle

GTTAGGTTGATTATGGTTAGAAGA

ACTTATATACCTCTACTAAAATAATCTAAC


Amplicon C: MLH1 shore

TGAGGGTAGGAAAGTTTGTTAG

AAACTACCTCCTAATCTTTATCC


S1: Shore region 1

TTCTCGAAGGTGTTCCATAATGTCCA

CTCTCTAAGAAAAGAAATGGAAGCGTTCTTG

CAGTAGGGGCAACAACAGTCCACTTCTCAGA

ChIP

S2: Shore region 2

AGCTCAACAGTTCCAAGTGAAGAAATCC

GAGAACTCAGGACTCGTCTCATACATGAG

AGGCCGTAAGCCCCAACTGTTGGAATGTCAT

ChIP

M1: Middle region 1

GGATAAAGACCAGGAGGTAGTTCTC

ACAGGAATCTACGCAACCAGCTTT

AGCCTATTCTCTTGCCTTGGACGACCAGGCTT

ChIP

M2: Middle region 2

GGCAGTACCTCTCTCAGCAACAC

TTCACTCCTGAAGAGAGAGCTGCTC

GCCTCGGGCTCTGCCGCCTCTTGG

ChIP

P1: Promoter CpG island

AAGAGACCCAGCAACCCACAGA

CCTCGTGCTCACTGGCTTCCTT

CCTTCAGCGGCAGCTATTGATTGGACAGCT

ChIP

107

4.3.3 MethyLight

MethyLight was performed on CRC cell line DNA, similar to that described in Chapter 3. DNA

from COLO 320HSR, HCT-15, HCT 116, LS 174T, SNU-C2B, Caco-2, HT-29, LS 180, SW48,

SW480, and SW620 was subjected to bisulphite modification with the EZ DNA Methylation-

Gold Kit according to manufacturer’s protocol (Zymo Research Corp., Orange, CA). Primers and

probe were used to amplify regions of the MLH1 shore and CpG island with ALU-C4 primers

and probe used as control. Probes contained a 5’ fluorescent reporter dye and a 3’ quencher dye.

Primer and probe sequences are found in Table 4.1. Percent methylated reference (PMR) was

calculated using the following calculation: [Gene of Interest/ALU-C4]sample/[Gene of Interest/

ALU-C4]CpGenome x 100%, where CpGenome represents commercially available fully

methylated CpGenome Universal Methylated DNA (Millipore, Billerica, MA). Samples were

analyzed in duplicate in 96-well plates on an ABI 7500 RT-PCR thermocycler. In order to ensure

that DNA quality was adequate, samples with an ALU-C4 threshold cycle greater than 22 were

deemed poor quality and reanalyzed or removed from the study (556).

4.3.4 Bisulphite sequencing

DNA from COLO 320HSR, HCT-15, HCT 116, LS 174T, and SNU-C2B was bisulphite

modified using EZ DNA Methylation-Gold Kit (Zymo Research, Irvine, CA). Three overlapping

regions spanning the MLH1 promoter CpG island and adjacent shore region were amplified by

PCR in bisulphite modified DNA from each cell line. Amplicon A, within the MLH1 shore,

spans from -1782 to -1033 base pairs relative to the MLH1 TIS. Amplicon B spans from -1114 to

-347 relative to the MLH1 TIS. Amplicon C, spanning the CpG island, covers -377 to -49

relative to the MLH1 TIS and contains rs1800734. Primers are listed in Table 4.1. PCR products

for each reaction were purified with ChargeSwitch PCR Clean Up Kit (Invitrogen, Carlsbad,

CA). Molecular cloning of amplicons was performed using pGEM-T Easy Vector System

(Promega, Madison, WI) with Max Efficiency DH5α Competent Cells (Thermo Fisher

Scientific, Waltham, MA), following the same protocol described in Chapter 3. Plasmid DNA

was prepared using QIAprep Spin MiniPrep Kit (Qiagen, Hilden, Germany) and sequenced by

Sanger sequencing at TCAG. At least ten clones were utilized for Sanger sequencing at TCAG

for each region in each cell line.

108

4.3.5 Selection of candidate proteins for chromatin immunoprecipitation

Publically available databases and in silico transcription factor binding programs were explored

to select a candidate protein able to bind the wildtype DNA sequence surrounding SNP

rs1800734 at the G allele but not the A allele. HOMER Motif Analysis software predicted

binding of the basic helix-loop-helix (bHLH) transcription factor AP4 to the non-canonical E-

box sequence CAGCTG containing wildtype G but not the sequence CAGCTA containing

variant A. TFBIND software also predicted binding of AP4 at CAGCTG sequence motif.

TRANSFAC predicted the bHLH myogenin, a muscle-specific transcription factor, which binds

the canonical E-box CACGTG. Since this protein is likely only to be found in muscle tissues, it

was not pursued as a possible candidate. HaploReg also predicted a bHLH factor,

BHLHE40/DEC2, which binds the canonical E-box sequence CACGTG. However, former

students in the lab had previously investigated this protein, and it was not found to bind in this

region, thus was not pursued. TFBIND software also predicted that the leucine zipper DNA

binding protein ATF2/CREBP1 binds to the G allele but not the A allele. This protein has also

been shown to bind to the region surrounding SNP rs1800734 in GM12878 and H1-hESC cell

lines through ChIP-Seq experiments from ENCODE, available on the UCSC Genome Browser.

However, a suitable ChIP-grade antibody was unavailable for this protein (two antibodies were

tested). Binding of CTCF approximately 250 bp upstream of rs1800734 was demonstrated in a

number of cell lines through ENCODE ChIP-Seq experiments.

4.3.6 Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) experiments were performed in triplicate on three

successive passages for HCT-15, HCT 116, and SNU-C2B following protocols from the EZ-

Magna ChIP A/G kit (EMD Millipore, Merck KGaA, Darmstadt, Germany). Histone H3

(ab61251), H3K4me1 (ab8895), H3K4me3 (ab8580), H3K27ac (ab4729), H3K27me3 (ab6002)

and CTCF (ab70303) antibodies were purchased from Abcam (Cambridge, UK). Normal Mouse

IgG and RNA polymerase II, clone CTD4H8 (Pol II) antibodies were provided in the EZ-Magna

ChIP A/G kit (EMD Millipore). TFAP4 (HPA001912) antibody was purchased from Sigma-

Aldrich (St. Louis, MO, USA). Histone H3 and Normal Mouse IgG antibodies were used as

positive and negative controls, respectively. ChIP-qPCR was performed in triplicate for each

reaction at five regions upstream of MLH1 as well as positive and negative control regions with

109

the QuantStudio 6 Flex Real-Time PCR System (Applied Biosystems, Foster City, CA, USA).

The five regions interrogated contained two regions in the MLH1 shore (called S1 and S2), two

regions in between the island and shore (called M1 and M2), and one region in the promoter

CpG island (called P1). The list of primers used for ChIP-qPCR at the MLH1 region can be

found in Table 4.1. Primers and probes for control regions other than AP4 were purchased from

Abcam, including ALDOA, GAD1, GAPDH, and H19-IGF (Abcam, Cambridge, UK). Potential

positive control regions for AP4 were tested using sequences that were shown to be positive for

AP4 in the CRC cell line SW620 in a publication from Jackstadt et al. as this information was

unavailable for the cell lines used for ChIP in these experiments (587). Primers for SNAIL,

CDH1, CDK2 and CKS2 were tested as described in (587).

4.3.7 Confirmation of genotype from ChIP experiments

Immunoprecipitated DNA from the ChIP reaction for AP4 and input from SNU-C2B cells was

amplified by PCR using the same primers utilized for cell line genotyping. PCR products for

AP4 and input were purified with ChargeSwitch PCR Clean Up Kit (Invitrogen, Carlsbad, CA).

Molecular cloning of amplicons was performed using pGEM-T Easy Vector System (Promega,

Madison, WI) with Max Efficiency DH5α Competent Cells (Thermo Fisher Scientific, Waltham,

MA). The same protocol was utilized as for bisulphite sequencing, described previously, except

DNA was not bisulphite modified beforehand. Plasmid DNA was prepared using QIAprep Spin

MiniPrep Kit (Qiagen, Hilden, Germany) and at least ten clones were sequenced by Sanger

sequencing at TCAG for each reaction.

4.3.8 Statistics

For ChIP-qPCR experiments a Student’s t-test was used to compare groups. Statistical

significance was defined by a P-value < 0.05.

4.4 Results

4.4.1 Genotype and methylation status of CRC cell lines

A panel of 11 CRC cell lines was genotyped for SNP rs1800734 and subjected to methylation

assessment at the MLH1 CpG island and shore region. The results are shown in Table 4.2. As

110

HCT-15 was the only cell line with the AA genotype, it was chosen for further analysis. Two cell

lines that had the GA genotype that were selected for further analysis were COLO 320HSR and

SNU-C2B. SW48 was not selected because the entire region, including the CpG island and

shore, was 100% methylated and its promoter (but not its shore) has previously been investigated

for several histone modifications (574). I decided that further exploration of this cell line might

not reveal novel information compared to the other selected GA cell lines. Seven of the cell lines

had the wildtype GG genotype. The methylation of these GG cell lines ranged from 24.5% up to

100% at the MLH1 shore. HCT 116 and LS 174T were selected as they had two of the highest

methylation levels at the shore, which more closely mimics the SNP-associated methylation

pattern that was identified in specimens from OFCCR CRC cases and controls as outlined in

Chapters 2 and 3.

Bisulphite sequencing was performed on the five selected cell lines at three overlapping PCR

amplicons across the entire upstream regulatory region of MLH1 including its island and shore.

Amplicon A, in the shore region, was highly methylated in the two GG cell lines, HCT 116 and

LS 174T, whereas the GA cell lines, SNU-C2B and COLO 320HSR, and the AA cell line, HCT-

15, were hypomethylated compared to wildtype at the MLH1 shore (Figure 4.1). The last three

CpG sites in Amplicon A overlap with the first three CpG sites of Amplicon B. At Amplicon B

there was some methylation at the first five CpG sites, but CpGs were unmethylated downstream

in all cell lines except HCT 116. HCT 116 had a small number of clones (2/10, 20%) that were

methylated across Amplicon B. Amplicon B and Amplicon C overlap by one CpG site. All five

cell lines, regardless of rs1800734 SNP genotype, were unmethylated at Amplicon C in the CpG

island of MLH1. These results suggest that in CRC cell lines methylation at the MLH1 shore is

correlated with genotype of rs1800734.

111

Table 4.2 MLH1 promoter SNP genotype and methylation of its CpG island and shore in

CRC cell lines. A panel of 11 colorectal carcinoma cell lines was genotyped by Sanger

sequencing to determine genotype of rs1800734. MethyLight was utilized to determine the

percentage of fully methylated alleles at the CpG island and shore of MLH1. Average PMR

(percent methylated reference) of two duplicate reactions is shown.

Cell Line rs1800734 Genotype CpG shore PMR CpG island PMR Caco-2 GG 59.9% 0.5% COLO 320HSR GA 23.4% 0% HCT-15 AA 34.6% 0% HCT 116 GG 99.6% 0% HT-29 GG 24.5% 0.4% LS 174T GG 73.1% 0% LS 180 GG 67.3% 0% SNU-C2B GA 28.9% 0% SW48 GA 100.0% 100.0% SW480 GG 100.0% 0% SW620 GG 52.7% 0%

112

Figure 4.1 Bisulphite sequencing of MLH1 CpG island and shore in colorectal cancer cell

lines. A. Three overlapping amplicons upstream of MLH1 were amplified and utilized for

bisulphite sequencing in cell lines: Amplicon A, Amplicon B, and Amplicon C. Amplicon A and

B overlap at 3 CpGs. Amplicon B and C overlap at 1 CpG. B. Representative unmethylated

clone for each of the three amplicons, located in the CpG shore, middle region, and CpG island.

Empty circles represent unmethylated CpG sites and filled in circles represent methylated CpG

sites. C. Methylation patterns in the colorectal cancer cell lines HCT 116, LS 174T, SNU-C2B,

COLO 320HSR, and HCT-15 with rs1800734 genotype indicated. Each horizontal line

represents a single DNA strand and circles represent individual CpG sites. Figure created with

QUMA.

113

4.4.2 Sequence-specific binding of AP4

ChIP experiments were undertaken for a variety of transcription factors and histone

modifications in three of the cell lines that were profiled for DNA methylation at MLH1 by

bisulphite sequencing: HCT 116, SNU-C2B, and HCT-15. Known genetic and epigenetic

features of these three cell lines are listed in Table 4.3. All three cell lines display MSI and are

MMR deficient without MLH1 CpG island hypermethylation (588–590). HCT 116 has a

hemizygous mutation at codon 252 in MLH1, while HCT-15 has a 1 bp deletion at codon 252

and a 5 bp deletion/substitution at codon 1103 (591,592). SNU-C2B is MSI but no mutations in

the mismatch repair genes MLH1, MSH2, MSH6, or PMS2 have been reported (589,590).

ChIP for transcription factor AP4, predicted to bind the G allele but not A allele, in the three

abovementioned cell lines resulted in enrichment at promoter amplicon P1 in cell lines

containing G allele of rs1800734 but not in the AA cell line HCT-15 (Figure 4.2). HCT 116 had

significantly higher occupancy of AP4 at the promoter P1 than SNU-C2B (P = 0.01) and HCT-

15 (P = 0.003). SNU-C2B also had significantly higher enrichment for AP4 at the MLH1

promoter than HCT-15 (P = 4.2x10-4). There was no enrichment at regions S1, S2, M1, or M2 as

expected based on sequence specificity of AP4 binding. Immunoprecipitated DNA from SNU-

C2B was sequenced to confirm that enrichment was genotype-specific at the MLH1 promoter. Of

the 20 alleles sequenced from the AP4 pull-down, 19 contained the G allele and one contained

the A allele. Input DNA contained nearly equal numbers of G and A alleles, with 8 having the G

allele and 11 having the A allele out of 19 alleles sequenced. Representative G and A

chromatogram traces are shown in Figure 4.3.

Five sets of primers were tested to act as positive controls for AP4 binding, including GAPDH,

SNAIL, CDH1, CDK2 and CKS2. The positive control region that gave the highest enrichment

level was for the primers located in the CDH1 gene, shown in Figure 4.2. There was no ideal

positive control for AP4 to be found in the literature or from companies selling antibodies, in

either the cell lines used for this study or any other cell lines. Though the CDH1 region did not

result in levels of enrichment as high as that seen for region P1, the positive control enrichment

exceeds that of the negative control region at GAD1 and thus can be considered an appropriate

control.

114

Table 4.3 Selected genetic and epigenetic features of colorectal cancer cell lines utilized in

chromatin immunoprecipitation assays. MSI, CIMP, and MLH1 promoter methylation status

are shown, as well as mutation status of MMR and other commonly mutated genes in CRC.

References: (588,590–592).

Cell Line! MSI Status!

CIMP Status!

MLH1 P/romoter Methylation!

MMR Gene Mutation!

Description of MMR Mutation

Other Mutations!

HCT 116! MSI+! CIMP+! Unmethylated! MLH1 Hemizygous C>A Codon 252

AXIN2 CTNNB1 KRAS!

SNU-C2B! MSI+! CIMP-! Unmethylated! Unknown No mutation detected in MLH1, MSH2, MSH6, PMS2

KRAS TP53!

HCT-15! MSI+! CIMP+! Unmethylated! MSH6 1 bp deletion Codon 252 & 5 bp deletion/ substitution Codon 1103

APC KRAS!

115

Figure 4.2 ChIP analysis of AP4 occupancy at the MLH1 CpG island and shore region. A.

Chromatin immunoprecipitation followed by qPCR was performed at five regions upstream of

MLH1, located in the CpG shore (S1 and S2), middle region (M1 and M2), and promoter CpG

island (P1). MethyLight and bisulphite sequencing amplicons are also indicated. SNP location is

indicated by white circle. B. Experiments were performed in HCT 116 (GG), SNU-C2B (GA),

and HCT-15 (AA) cell lines to compare AP4 occupancy among genotypes of SNP rs1800734.

Three biological replicates of each cell line were averaged after ChIP-qPCR. Error bars represent

standard deviation. CDH1 and GAD1, the positive and negative control regions, are indicated in

shades of grey. From left to right, bars represent HCT 116, SNU-C2B, and HCT-15 as in the

legend. * P < 0.05, ** P < 0.01, ***P < 0.001.

116

Figure 4.3 Representative chromatogram traces from AP4 chromatin immunoprecipitation

pull-down in heterozygous SNU-C2B cells at rs1800734. A. DNA pulled down from SNU-

C2B immunoprecipitated with antibody against AP4. 20 alleles were sequenced by Sanger

sequencing. Top: The sequence surrounding the wildtype rs1800734 SNP is CAGCTGAAGGA,

pulled down in 19/20 alleles. Bottom: The variant sequence surrounding rs1800734 SNP is

CAGCTAAAGGA, pulled down in 1/20 alleles. B. Input DNA from SNU-C2B, not

immunoprecipitated with any antibody. 19 alleles were sequenced by Sanger sequencing. Top:

The sequence surrounding the wildtype rs1800734 SNP is CAGCTGAAGGA, found in 8/19

input alleles sequenced. Bottom: The variant sequence surrounding rs1800734 SNP is

CAGCTAAAGGA, found in 11/19 input alleles sequenced.

117

4.4.3 Histone modifications and Pol II are consistent across genotypes of

rs1800734

The histone modifications H3K4me1, H3K4me3, H3K27ac, and H3K27me3 were assessed at the

MLH1 CpG island and shore at the same five regions as for AP4 (Figure 4.4). H3K4me1 was

most enriched at the shore region of S1 and S2 compared to downstream regions M1, M2, and

P1. H3K4me1 occupancy was significantly higher in SNU-C2B compared to HCT 116 at S2 (P

= 0.03). Though not significantly different than HCT 116, there was similar enrichment for

H3K4me1 in HCT-15 cells as in SNU-C2B. Also, at S1 there is a trend towards higher

enrichment for H3K4me1 in cell lines containing the variant A allele. SNU-C2B compared to

HCT 116 enrichment yields a P-value of 0.053 at S1.

H3K4me3 and H3K27ac both had similar enrichment patterns across the region, with low

enrichment at promoter region P1, then a peak of enrichment upstream at the M2 region and

decreasing levels of both modifications further upstream. H3K27me3 was low across the entire

region tested.

ChIP-qPCR data for all antibodies was normalized using the percent of input method, in which

the qPCR signal for each sample is divided by the input sample, which has not been

immunoprecipitated with an antibody. The qPCR signal obtained from input is assumed to

directly relate to the amount of input chromatin for each reaction (593,594). A disadvantage of

this method is that differences in enrichment of histone modifications may be due, in part, to

differences in nucleosome density of the region. For example, the proximal promoters of actively

transcribed genes are depleted of nucleosomes, allowing various transcription factors access to

the DNA (595). Another option is to normalize ChIP signals by nucleosome density, using an

invariant domain of histone H3. A disadvantage of normalizing to nucleosome density is that it

requires comparison of qPCR signals obtained from two different antibodies. Every antibody has

different epitope binding kinetics (593,596). Figure 4.5 shows ChIP results for H3K4me1,

H3K4me3, H3K27ac, and H3K27me3 normalized to histone H3. H3K4me1 still shows a similar

pattern of increasing enrichment at the shore compared to the island, however, there is no

significant difference in enrichment between cell lines at S2. H3K4me3 has high enrichment

level at promoter region P1 when normalized to H3, which was not apparent in Figure 4.4, as the

nucleosome density is very low at P1. Similarly, H3K27ac also has higher levels of enrichment

118

at promoter region P1 after normalization to H3. Additionally, H3K27ac is significantly more

enriched in HCT-15 compared to HCT 116 at M1 (P = 0.03) and is more enriched in HCT 116

compared to SNU-C2B at M2 (P = 0.002). H3K27me3 relative to histone H3 remains low across

the entire MLH1 region tested.

Presence of RNA polymerase II (Pol II) was also tested in all three cell lines across region S1,

S2, M1, M2, and P1 (Figure 4.6). There were only very low levels observed, no higher than that

of the negative control, for Pol II at all regions in all three cell lines. Overall, there is a trend

towards increased H3K4me1 at the MLH1 shore in cells containing variant A allele of

rs1800734. Other histone modifications and factors tested, including H3K4me3, H3K27ac,

H3K27me3, and Pol II do not show such associations.

119

Figure 4.4 ChIP analysis of histone modifications at the MLH1 CpG island and shore.

Chromatin immunoprecipitation was performed in HCT 116 (GG), SNU-C2B (GA), and HCT-

15 (AA) cell lines to compare histone modifications among genotypes of SNP rs1800734. Three

biological replicates of each cell line were averaged after ChIP-qPCR at five regions of the

MLH1 CpG island and shore: S1, S2, M1, M2, and P1. Positive and negative control regions are

shown for each histone mark tested, indicated in shades of grey. From left to right, bars represent

HCT 116, SNU-C2B, and HCT-15. Histone modifications include: A. H3K4me1, B. H3K4me3,

C. H3K27ac, and D. H3K27me3. Error bars represent standard deviation. Different y-axis scales

are used for each antibody in ChIP figures as the affinity for epitopes differs between antibodies.

* P < 0.05.

120

Figure 4.5 ChIP analysis of histone modifications at the MLH1 CpG island and shore

normalized to histone H3. Chromatin immunoprecipitation was performed in HCT 116 (GG),

SNU-C2B (GA), and HCT-15 (AA) cell lines to compare histone modifications among

genotypes of SNP rs1800734. Three biological replicates of each cell line were averaged after

ChIP-qPCR at five regions of the MLH1 CpG island and shore: S1, S2, M1, M2, and P1 and

normalized to histone H3 (percent of input for histone modification divided by percent of input

for histone H3). Positive and negative control regions are shown for each histone mark tested,

indicated in shades of grey. From left to right, bars represent HCT 116, SNU-C2B, and HCT-15.

Histone modifications include: A. H3K4me1, B. H3K4me3, C. H3K27ac, and D. H3K27me3.

Error bars represent standard deviation. Different y-axis scales are used for each antibody in

ChIP figures as the affinity for epitopes differs between antibodies. * P < 0.05, ** P < 0.01.

121

Figure 4.6 ChIP analysis of Pol II and CTCF at the MLH1 CpG island and shore.


15 (AA) cell lines to compare Pol II and CTCF binding among genotypes of SNP rs1800734.

Three biological replicates of each cell line were averaged after ChIP-qPCR at five regions of the

MLH1 CpG island and shore: S1, S2, M1, M2, and P1. Pol II (top panel) and CTCF (bottom

panel) were assessed. Positive and negative control regions are shown for each factor tested, in

shades of grey. From left to right, bars represent HCT 116, SNU-C2B, and HCT-15 as in the

legend. Error bars represent standard deviation. * P < 0.05.

122

4.4.4 Lack of CTCF at MLH1 CpG island and shore

CTCF binding was tested across the MLH1 CpG island and shore and there was only minimal

enrichment for this protein in cell lines at regions S1, S2, M1, and M2 (Figure 4.6). At region P1

there was significantly higher enrichment for CTCF in HCT 116 compared to HCT-15 (P =

0.04). Although the enrichment at P1 was quite low, this may indicate differential binding of

CTCF at the MLH1 promoter CpG island, with decreased binding in the rs1800734 homozygous

variant HCT-15 cell line. A figure showing representative results for the positive control

antibody histone H3 and negative control antibody IgG is shown in Figure 4.7. These controls

were used for each immunoprecipitation performed.

123

Figure 4.7 ChIP analysis of controls H3 and IgG at the MLH1 CpG island and shore.


15 (AA) cell lines to compare among genotypes of SNP rs1800734. Three biological replicates

of each cell line were averaged after ChIP-qPCR at five regions of the MLH1 CpG island and

shore: S1, S2, M1, M2, and P1. Results from a representative experiment are shown for positive

control antibody histone H3 (top panel) and negative control antibody IgG (bottom panel).

Positive and negative control regions are also shown in shades of grey. From left to right, bars

represent HCT 116, SNU-C2B, and HCT-15 as in the legend. Error bars represent standard

deviation. The same y-axis scale is used for both antibodies.

124

4.5 Discussion

Despite the fact that the majority of SNPs are located in non-coding regions of the genome, their

diverse roles in disease pathogenesis, including CRC, are steadily becoming established through

both experimental and computational methods (433,434,546,547,566,567). Here, I have

demonstrated that various epigenetic and regulatory modifications are associated with variant

genotype of the MLH1-93G>A promoter SNP rs1800734. It plays a role in modifying DNA

methylation at the MLH1 shore in cell lines. I have also demonstrated, for the first time,

significantly diminished binding of the transcription factor AP4 in cell lines lacking the wildtype

G allele, which may play a role in the decrease in transcriptional activity previously shown

(480). There may also be decreased binding of the insulator protein CTCF at the promoter region

in cell lines lacking the G allele. The enhancer histone mark H3K4me1 appears to be increased at

the MLH1 shore in cell lines carrying variant alleles. Interestingly, the other histone

modifications tested of enhancers, active regions, and repressed regions do not differ according

to genotype and/or methylation.

A number of non-coding SNPs identified through genome-wide association studies (GWAS)

have been shown to alter consensus sequences, which alters binding ability of transcription

factors, thus altering enhancer or promoter activity. For example, a prostate cancer risk SNP at

6q22.1 leads to increased binding of transcription factor HOXB13, increased transcription of

RFX6, and increased deposition of the H3K4me2 mark in the region compared to the protective

allele in a panel of isogenic cell lines representing each genotype (433). The variant rs6983267 at

the 8q24 risk locus has increased binding of the transcription factor TCF7L2 in CRC cell lines,

leading to interactions with the MYC promoter (442). At the CRC risk locus 14q23.1, a SNP

variant was shown to be associated with reduced expression of RTN1 (546). Similarly, the Bapat

Lab previously demonstrated significant decreases in transcriptional activity for the variant allele

of rs1800734 compared to wildtype in CRC, normal colon, and endometrial cancer cell lines

(480). Our group and others have also identified, though EMSA, the binding of a certain factor(s)

to the G allele but not the variant A allele in CRC cell lines (480,568). In these previous studies,

the precise factor(s) were not identified, though several proteins were tested including Max and

DEC2/BHLHE40. I have now shown the presence of AP4 enrichment directly at the SNP locus

in cell lines with one or two G alleles. I also confirmed preferential binding of AP4 to the G

125

allele by sequencing the DNA pulled down in AP4 experiments in the GA cell line SNU-C2B.

AP4 also may be present in a dose-dependent manner, as the wildtype cell line HCT 116 has

significantly higher enrichment for AP4 compared to the heterozygous line SNU-C2B. However,

this may also potentially be due to other differences in regulation of AP4, MLH1, or other factors

between the two cell lines. Binding of transcription factors and/or RNA polymerase II has been

shown to block or inhibit deposition of DNA methylation (569,570,597). Thus, lack of AP4

binding at the promoter variant SNP may lead to decreased transcriptional activity and over time,

increased DNA methylation. AP4 may act together with MYC, which has a binding motif

CACGAG located 15 bp downstream of rs1800734, to activate transcription of MLH1 (587).

Future experiments to measure co-localization of MYC and AP4 in this region in a sequence-

specific manner would further serve to elucidate this. A proposed model for this series of events

is demonstrated in Figure 4.8.

While the variant of interest, rs1800734, has not been discovered through GWAS studies, it has

been identified as a risk SNP for CRC in a large number of individuals, including 10,409 cases

and 6,965 controls (524). However, a subsequent study has shown this to not be the case (548).

Though there is controversy surrounding the overall role of this SNP in cancer susceptibility, it

has been shown that rs1800734 is a risk SNP for the MSI subtype of CRC (478,479). Further,

previous chapters of this thesis as well as previous members of the Bapat Lab have demonstrated

a number of consequences of this SNP in cell lines, CRC patients, and even non-cancer controls

(480,515). In the five cell lines selected for bisulphite sequencing analysis I observed SNP-

associated MLH1 shore hypomethylation which was previously observed in peripheral blood

mononuclear cell DNA of CRC cases and controls, as well as normal colorectal tissue. The exact

mechanism of MLH1 shore hypomethylation remains to be elucidated. Potentially, MLH1 is part

of the subset of genes with hypermethylation of their CpG shores alongside unmethylated CpG

islands that have high transcriptional activity and a more transcriptionally permissive state (203).

A lack of AP4 binding at the promoter may decrease transcriptional activity, modifying the

balance of methylation in the region.

An interesting finding of these results is the fact that three of the histone marks tested

(H3K4me3, H3K27ac, H3K27me3) have similar enrichment levels in all three cell lines despite

having variable methylation patterns and SNP genotypes. Thus, methylation changes appear to

be largely independent of histone modifications present at the MLH1 region. Perhaps greater

126

methylation differences are required in order for histone changes to occur, for example 0% vs.

100%, rather than more subtle and variable levels in between the two extremes. In order to test

this, one could utilize a CRC cell line that is highly methylated across the CpG island and shore,

such as SW48 or RKO. The expected results include histone modification differences compared

to unmethylated cell lines, as there is a more drastic difference in DNA methylation levels.

Differences in acetylated histone H3 have previously been reported for methylated SW48 and

RKO cells compared to unmethylated normal human fibroblast LD419 cells at the CpG island of

MLH1 (574).

Of the histone modifications examined in HCT 116, SNU-C2B, and HCT-15 cell lines, similar

levels in each cell line were observed. H3K27me3, a repressive marker, was low across the entire

region tested. Perhaps this indicates that transcription factor presence and/or DNA methylation

plays a larger role in repression at this region, compared to histone modifications. H3K4me3 and

H3K27ac, markers of an active promoter when present concurrently, peak in enrichment just

upstream of the MLH1 TIS at region M2 in all three cell lines and diminish further upstream at

the shore (598). The P1 promoter region tested has low levels of these two modifications, yet this

was not the case once normalized to histone H3 levels, indicating a nucleosome-depleted region

(595). The final histone modification interrogated at MLH1 was H3K4me1. H3K4me1 is a

marker of enhancer regions when found concurrently with H3K27ac, among other factors such

as p300 (598). H3K4me1 showed highest enrichment at the shore S1 and S2 regions of MLH1

compared to further downstream at M1, M2, and P2. However, H3K27ac was not similarly

enriched at the shore region. The S2 region had enrichment in the SNU-C2B cell line compared

to HCT 116. A non-significant trend of increased enrichment in variant cell lines compared to

wildtype HCT 116 was also observed at S1. Perhaps this differential H3K4me1 level is due to

DNA hypomethylation at the shore, or vice versa. The presence of H3K4me1 without H3K27ac

indicates that this MLH1 shore region may be considered an inactive or potentially ‘poised’

enhancer region (201,585,598). Depending on the cell type, developmental, or regulatory cues,

this region may act as an enhancer, but does not appear to be active in these three cells lines.

ChIP was also performed to determine whether CTCF binding is present in any of the cell lines.

This protein may act as an insulator to prevent the spread of DNA methylation. As has been

established in the previous two chapters, the usual status of MLH1 in normal wildtype tissues is

relatively high methylation at the shore contrasted with no methylation at its island. This pattern

127

undergoes an epigenetic switch in a subset of CRC tumours. I hypothesized that a lack of CTCF

may be responsible for the dysregulation and spreading of DNA methylation from the MLH1

shore downstream to the CpG island. CTCF may also act as an insulator to prevent interactions

between promoters and enhancers. Based on ChIP results for histone modifications, it appears as

though the MLH1 shore may be an enhancer, however it is not active in the CRC cell lines tested.

I also hypothesized that CTCF may be an insulator to prevent interaction between the enhancer

and the MLH1 promoter, or perhaps even the promoter of another gene located further away in

cis or in trans. Though CTCF was very low at all five of the regions tested across the MLH1

shore and promoter CpG island, there was a significant difference in enrichment at the promoter

P1 region. There was significantly less CTCF in HCT-15 AA cell line compared to HCT 116 GG

cell line. Potentially, CTCF is present at the MLH1 promoter in wildtype cells, and a loss or lack

of CTCF in variant-containing cells is in part responsible for changing methylation patterns at

the CpG island and shore.

Of note, the methylation patterns observed in cell lines were heterogeneous both from cell to cell

(different clones in bisulphite sequencing) as well as from CpG site to CpG site along the DNA

strand. This is in contrast to the bimodal distribution of methylation normally observed, with

most CpGs either completely unmethylated or 100% methylated (599,600). While this epigenetic

heterogeneity may be due in part to the nature of immortalized cell lines grown in culture, it has

also been shown to be associated with cancer (600,601). The heterogeneous methylation pattern

was more pronounced in the cell lines with either GA or AA genotype compared to GG. Similar

patterns were also observed in patient samples, included in the previous chapter (Chapter 3 and

Figure 3.3). It is possible that the A allele influences DNA methylation at the MLH1 shore to be

more ‘cancer-like’ in its heterogeneity.

A limitation of this study is that the cell lines used in ChIP experiments were selected based on

genotype of one SNP. Though all three cell lines have microsatellite instability but no MLH1

CpG island hypermethylation, each cell line differs in mutational spectra and thus may have

altered epigenetic or transcriptional machinery. In order to decrease the variability between cell

lines and to experimentally establish the association between genotype, AP4 binding, and MLH1

expression, using CRISPR to create all three possible genotypes of rs1800734 in the same cell

line would be ideal. An added experiment would be to use a demethylating treatment in cell lines

and track DNA methylation, histone modifications, transcription factor binding, and gene

128

expression changes over time. Though time-consuming, these experiments could aid in

mimicking the natural history of a tumour. Another caveat to this study is that I selected two GG

cell lines for bisulphite sequencing analysis, HCT 116 and LS 174T, that had high levels of shore

methylation, as assessed by MethyLight. Selection of other cell lines with lower shore

methylation levels may have yielded different results with respect to ChIP studies of histone

modifications.

As mentioned above, all three cell lines utilized for ChIP are MSI, thus have deficient DNA

mismatch repair. According to RNA expression array data found on Gene Expression Omnibus

(GEO, GSE59857), MLH1 expression is low in HCT 116, higher in SNU-C2B, and highest in

HCT-15. Expression in HCT-15 was comparable in level to both MSI (eg. Caco-2) and MSS

(eg. COLO 320HSR, SW480) cell lines. The lowest levels of expression were observed in cell

lines with MLH1 promoter CpG island hypermethylation, including SW48 and RKO lines. Low

levels of Pol II were present at the promoter of all three cell lines assessed by ChIP despite

differences in previously established expression arrays. Yet, there is still significantly enriched

binding for AP4 in both cell lines containing the G allele of rs1800734. Thus, in these cell lines

AP4 binding may not have a dramatic effect on MLH1 expression. Future work on a larger panel

of cell lines, especially those with MSS status and containing the AA genotype, would be of

considerable interest to determine how AP4 and Pol II presence are affected. Selection of cell

lines with promoter CpG island hypermethylation would also provide insight into binding of

AP4 or other transcription factors, shore methylation status, and presence of histone

modifications.

Colorectal cancer is both a genetic and an epigenetic disease. While these two processes are

frequently considered separately, it is clear that at the MLH1 locus genetic changes disrupt

epigenetic and transcriptional regulation, with important consequences. In this chapter I have

comprehensively studied the epigenetic effects that a single nucleotide change in the MLH1

promoter can confer. Variant SNP genotype is associated with hypomethylation of the MLH1

shore. Despite methylation differences seen among CRC cell lines, the histone modifications

assessed do not incur similar changes. This variant also alters the binding site of AP4 leading to

diminished binding of this transcription factor. These results explore the functional epigenetic

regulation and molecular mechanisms occurring at the important MLH1 locus in CRC, shedding

new light on the epigenetic concept of CpG shores and how DNA variants play a role in

129

epigenetics and cancer susceptibility. If this example of genetic-epigenetic interaction is applied

to the whole genome, there are clearly multitudes of ways in which the genome and epigenome

may interact. Further studies of such interactions will lead to a better understanding of the

processes and changes incurred by the genome and epigenome under both normal circumstances

and cancer development.

4.6 Acknowledgements I would like to thank Dr. Ken Kron for assistance with HOMER Motif Analysis software, which

aided in selection of AP4 as a candidate transcription factor for ChIP analysis.

130

Figure 4.8 Schematic models of transcription factors and epigenetic regulation at SNP

rs1800734. Top: In cells with wildtype G allele, AP4 transcription factor binds its consensus

sequence, potentially interacting with MYC and Pol II to promote transcription of MLH1.

Though not demonstrated through experimental results in this study, DNA methyltransferases

(DNMT) may maintain DNA methylation at the shore upstream. DNMTs may be prevented from

methylating the CpG island in part due to presence of CTCF. Bottom: In cells with the variant A

allele, AP4 does not bind which may decrease promoter transcriptional activity. Without the

presence of AP4 (or possibly CTCF), DNMTs may methylate the exposed CpG island. This may

lead to decreased methylation at the CpG shore and increased H3K4me1, which is deposited by

MLL proteins. Other currently unidentified factors may also bind and repress the region further.

131

Chapter 5 Promoter methylation of the Wnt signaling gene ITF2, but not

APC, is associated with microsatellite instability in two populations of colorectal cancer patients

5.1 Summary

Aberrant Wnt signaling activation occurs commonly in colorectal carcinogenesis, leading to

upregulation of many target genes. APC (adenomatous polyposis coli) is an important

component of the β-catenin destruction complex that regulates Wnt signaling and is often

mutated in colorectal cancer (CRC). In addition to mutational events, epigenetic changes arise

frequently in CRC, specifically, promoter hypermethylation which silences tumour suppressor

genes. APC and the Wnt signaling target gene ITF2 (immunoglobulin transcription factor 2)

incur hypermethylation in various cancers, however, methylation-dependent regulation of these

genes in CRC has not been studied in large, well-characterized patient cohorts. The

microsatellite instability (MSI) subtype of CRC, featuring DNA mismatch repair deficiency and

often promoter hypermethylation of mutL homolog 1 (MLH1), has a favorable outcome and is

characterized by different chemotherapeutic responses than microsatellite stable (MSS) tumours.

Other epigenetic events distinguishing MSI versus MSS CRC subtypes have not yet been fully

elucidated. This chapter outlines the work that I performed to assess promoter methylation of

ITF2 and APC by MethyLight in two case-case studies nested in population-based CRC cohorts

from the Ontario Familial Colorectal Cancer Registry (N = 330) and the Newfoundland Familial

Colorectal Cancer Registry (N = 102) comparing MSI status groups. ITF2 and APC methylation

were significantly associated with tumour versus normal state (both P < 3.4x10-9). ITF2 was

methylated in 45.8% of MSI cases and 26.9% of MSS cases and was significantly associated

with MSI in Ontario (P = 0.002) and Newfoundland (P = 0.005) as well as the MSI-associated

feature of MLH1 promoter hypermethylation (P = 6.7×10-4). APC methylation, although tumour-

specific, was not significantly associated with tumour subtype, age, sex, or stage, indicating it is

a general tumour-specific CRC biomarker. This chapter demonstrates, for the first time, MSI-

associated ITF2 methylation, and further reveals the subtype-specific epigenetic events

modulating Wnt signaling in CRC.

132

5.2 Introduction

Colorectal cancer (CRC) is one of the most common cancers in the Western world and is marked

by a high mortality rate (24,27,92). Early detection of CRC is the key to improved survival rates

(62,602,603). Another factor affecting disease prognosis is CRC subtype (173,604,605). The

microsatellite instability (MSI) subtype of CRC accounts for approximately 15% of colorectal

cancers (163,165). MSI tumours are distinguished by defects in the DNA mismatch repair

(MMR) system which leads to mutational insertions and deletions in short tandem repeats

(microsatellites) of DNA (162,542). MSI is most often due to promoter hypermethylation and

silencing of the mutL homolog 1 (MLH1) mismatch repair gene. Microsatellite stable (MSS)

tumours account for the majority (up to 85%) of CRCs and exhibit chromosomal instability,

including numerous chromosomal duplications, deletions and rearrangements (111,606,607).

MSI tumours differ from MSS tumours in several ways; MSI CRCs exhibit proximal colonic

location, increased lymphocytic infiltration, and poorer response to chemotherapeutic drugs

(173,608). MSI CRCs also demonstrate better prognosis at stages I-III, however, some studies

suggest poor prognosis at stage IV, though metastatic MSI cases are rare (173,609). A third CRC

subtype, the CpG island methylator phenotype (CIMP), is characterized by widespread DNA

hypermethylation of CpG-rich promoter islands. CIMP can exist concurrently with either the

MSI or MSS phenotype, though it is more frequently found in tandem with MSI and MLH1

hypermethylation (105,200,244). The prognostic significance of CIMP is currently undefined

and may be modified by MSI status, presence of BRAF or KRAS mutation, and tumour stage

(244,610,611). Recently, a classification system for further subtyping of CRC has been

proposed, consisting of four subtypes (612). One subtype consists mostly of MSI cases, while the

other three subtypes categorize the remainder of cases by Wnt signaling activation, metabolic

dysregulation, or mesenchymal activation.

The vast majority (up to 94 %) of CRCs feature dysregulation in the Wnt signaling pathway

(118,613,614). Wnt signaling is important in normal development, cell growth, and proliferation,

but when inappropriately activated may also lead to tumour initiation and development

(112,117,613). In canonical Wnt signaling, β-catenin accumulates within the cell, enters the

nucleus, and activates transcription of target genes, such as MYC and ITF2 (immunoglobulin

transcription factor 2) (112,114,615). ITF2 is also known as transcription factor 4 (TCF4). In the

133

absence of Wnt signaling, a β-catenin destruction complex, which includes adenomatous

polyposis coli (APC), AXIN1, GSK3β, and CK1, targets β-catenin for ubiquitination followed

by proteasomal degradation. In many cases of CRC the APC gene is mutated, rendering it

incapable of binding to β-catenin, which leads to β-catenin accumulation followed by its nuclear

translocation and subsequent activation of downstream target genes (114,117).

Evidence for DNA methylation of the APC promoter has been found in CRC. However, to what

extent APC methylation plays a role in colorectal carcinogenesis is unclear, as a broad range of

methylation levels has been reported in the literature, from 11% up to 63% of tumours

methylated (616–619). Conflicting reports exist regarding the extent of APC methylation in MSI

CRCs. Some small-scale studies (MSI tumours N ≤ 29) have suggested that APC methylation

may be associated with the MSI subtype but others show no significant difference (MSI tumours

N ≤ 44) (614,617–622).

The role of ITF2, a Wnt signaling target gene, is less understood in CRC. It is a target of Wnt

signaling and is overexpressed in colon cancers with Wnt dysregulation (623). Its expression was

reported elevated in some adenomas and cancers with aberrant Wnt signaling activation but

decreased in others (624,625). Among gastrointestinal malignancies, ITF2 methylation has been

reported in gastric cancer, but its methylation status has not been investigated in CRC (626,627).

The Bapat Lab has previously demonstrated associations between the methylation status of key

Wnt signaling pathway regulatory genes and CRC subtype including the extracellular Wnt

antagonists DKK1 and SFRP1 as well as WNT5A which is involved in non-canonical Wnt

activity (628,629). In this study, I have examined the role of APC and ITF2 methylation in two

nested case-case studies in CRC cohorts. These patients were recruited from two distinct

Canadian populations and the case groups were stratified by their MSI status.

5.3 Materials and Methods


Participants of this study were population-based primary CRC cases recruited through the

Ontario Familial Colorectal Cancer Registry (OFCCR) and Newfoundland Familial Colorectal

Cancer Registry (NFCCR). Procedures for patient accrual, biospecimen collection, and data

134

collection for the OFCCR and NFCCR have been previously described (554,630). Briefly,

Ontario residents between the ages of 20 and 74 diagnosed with pathology-confirmed primary

CRC between 1997 and 2000 were eligible for recruitment. Familial adenomatous polyposis

cases were excluded and non-white patients were also excluded due to the high prevalence of

self-reported Caucasians (92.5%). A total of 1,168 participants have been analyzed for MSI

status (see Molecular analysis below) of which 184 are MSI-high (MSI-H). 165 of these MSI-H

cases had available DNA of high quality. A matched case-case selection strategy was utilized to

select 165 patients with MSS tumours to match 165 patients with MSI-H tumours by sex, stage

at diagnosis, and age quartile. The 165 MSS tumours were selected from a total of 384 MSS

tumours available at the time this study was undertaken. Population-based recruitment by the

NFCCR was similar to the OFCCR, with a recruitment period from 1999 to 2003 of cases from

provincial tumour registries (630). For the NFCCR, proxy consent from living family members

was obtained for deceased patients leading to a high frequency of late-stage patients. These

tumour samples were not utilized in order to maintain similar patient age and tumour stage

between the Ontario and Newfoundland populations. 102 tumour samples from 696 total CRC

cases were chosen from probands of the NFCCR, 51 of which were MSI-H, matched to 51 MSS

cases by sex, stage at diagnosis and age quartile. Overall survival status, along with other patient

clinicopathological features, is described in Table 5.1. Recurrence data was not available for all

cases used in this study, thus was not included in analysis. Non-neoplastic colorectal mucosa

(henceforth referred to as ‘normal’) was also available from resected surgical specimens of all

patients. Of the 330 OFCCR and 102 NFCCR patients’ tumour samples utilized, 47 were

selected randomly for methylation analysis of matched normal tissue. Randomization was

performed by generating random numbers using the RAND function in Excel. A random number

was assigned to each patient, and then random numbers were sorted smallest to largest. The 47

lowest numbers, and their corresponding patient, were selected for analysis. Patient data was

obtained through protocols approved by the Research Ethics Boards of Mount Sinai Hospital, the

University of Toronto, and Memorial University of Newfoundland. All patients or their proxies

provided informed consent.

135

Table 5.1 Clinicopathological features of primary colorectal carcinomas of patients from

Ontario and Newfoundland.

Number of cases (%)

Ontario Newfoundland

MSS MSI-H MSS MSI-H

Cases of primary colorectal carcinoma

165 165 51 51

Mean age (±SDa)

59.9 (9.3) 60.1 (9.8) 58.3 (10.2) 58.4 (10.2)

Age <50 50+

19 (11.5) 146 (88.5)

28 (17.0) 137 (83.0)

11 (21.6) 40 (78.4)

10 (19.6) 41 (80.4)

Sex Male Female

74 (44.8) 91 (55.2)

74 (44.8) 91 (55.2)

26 (51.0) 25 (49.0)

26 (51.0) 25 (49.0)

TNM Stage 1 2 3 4

38 (23.0) 84 (50.9) 34 (20.6) 9 (5.5)

38 (23.0) 85 (51.5) 39 (23.6) 3 (1.8)

12 (23.5) 26 (51.0) 10 (19.6) 3 (5.9)

11 (21.6) 27 (52.9) 12 (23.5) 1 (2.0)

Histological Grade Low Moderate High Unavailable

16 (9.7) 123 (74.5) 13 (7.9) 13 (7.9)

10 (6.1) 45 (27.3) 20 (12.1) 90 (54.5)

4 (7.8) 37 (72.6) 8 (15.7) 2 (3.9)

7 (13.7) 36 (70.6) 8 (15.7)

Locationb

Distal Proximal Unavailable

108 (65.5) 51 (30.9) 6 (3.6)

15 (9.1) 63 (38.2) 87 (52.7)

37 (72.5) 14 (27.5)

9 (17.6) 42 (82.4)

Histological Typec

Non-Mucinous Mucinous Unavailable

143 (86.7) 19 (11.5) 3 (1.8)

107 (64.8) 53 (32.1) 5 (3.0)

46 (90.2) 5 (9.8)

42 (82.4) 9 (17.6)

MMR Protein Status

136

Intact Deficient Unavailable

148 (89.7) 4 (2.4) 13 (7.9)

26 (15.8) 136 (82.4) 3 (1.8)

50 (98.0) 0 (0.0) 1 (2.0)

4 (7.8) 46 (90.2) 1 (2.0)

MMR Germline Mutation No Yes

164 (99.4) 1 (0.6)

124 (75.2) 41 (24.8)

51 (100.0) 0 (0.0)

39 (76.5) 12 (23.5)

MLH1 Methylation Unmethylated Methylated Unavailable

159 (96.4) 5 (3.0) 1 (0.6)

87 (52.7) 78 (47.3)

36 (70.6) 1 (2.0) 14 (27.4)

23 (45.1) 28 (54.9)

BRAF V600E Mutation No Yes Unavailable

146 (88.5) 15 (9.1) 4 (2.4)

95 (57.6) 66 (40.0) 4 (2.4)

46 (90.2) 2 (3.9) 3 (5.9)

25 (49.0) 20 (39.2) 6 (11.8)

CIMP Status Negative Positive Unavailable

133 (80.6) 10 (6.1) 22 (13.3)

79 (47.9) 63 (38.2) 23 (13.9)

51 (100.0)

51 (100.0)

Survival Status Alive Deceased

! 100 (60.6) 65 (39.4)

99 (60.0) 66 (40.0)

45 (88.2) 6 (11.8)

! 49 (96.1) 2 (3.9)

a Standard Deviation b Proximal tumour location includes lesions up to and including the splenic flexure. c Mucinous histology includes the presence of any mucin within the tumour stroma.

137

5.3.2 Molecular analysis

DNA used to assess MSI status was extracted from archival paraffin-embedded tumours

microdissected to enrich for tumour cells. MSI status was assessed using National Cancer

Institute guidelines using four or more of the following markers: ACTC, BAT-25, BAT-26,

BAT-40, BAT-34C4, D10S197, D18S55, D17S250, D5S346 and MYC-L. The numbers of

positive markers used to define MSI status were: MSI-high (MSI-H), ≥ 30% unstable markers;

MSI-low (MSI-L), 1–29% unstable markers; MSS, 0% unstable markers (631). Tumours with

MSI-L status were not included in this study.

Somatic T>A mutation of nucleotide 1,799 in the BRAF gene leading to the V600E mutation was

determined by allele-specific PCR as described previously (628).

Immunohistochemistry was used to determine presence of the mismatch repair proteins MLH1,

MSH2, MSH6 and PMS2. Protein staining was classified as either present, absent, or

inconclusive. Tumours without positive staining for any of these proteins were defined as

mismatch repair deficient, as described previously (629).

5.3.3 MethyLight

The sensitive, semi-quantitative high-throughput MethyLight assay was used to analyze the

methylation of APC and ITF2 in tumour and normal colorectal DNA of CRC patients. DNA was

treated with sodium bisulphite prior to MethyLight according to protocol using the EZ DNA

Methylation-Gold Kit (Zymo Research Corp, Orange, CA). Primers and probe were used to

amplify a region within the CpG island of promoter 1A of APC. Forward primer: 5′-

GAACCAAAACGCTCCCCAT-3′. Probe: 5′- CCCGTCGAAAACCCGCCGATTA-3′. Reverse

primer: 5′-TTATATGTCGGTTACGTGCGTTTATAT-3′. Primers and probe were designed

within the promoter region of ITF2. Forward primer: 5′-GAAGCGGTAATACGAATAAGAGC-

3′. Probe: 5′-ATTCCCGAAACCGAAATCGTTCGCAAACC-3′. Reverse primer: 5′-

AACTATTCTCGAATAAACGTCGC-3′. ALU-C4 was also amplified to normalize the DNA

input. Forward primer: 5′-GGTTAGGTATAGTGGTTTATATTTGTAATT-3′. Probe: 5′-

CCTACCTTAACCTCCC-3′. Reverse primer: 5′-ATTAACTAAACTAATCTTAAACTCCTA-

3′. Probes contained a 5′ fluorescent reporter dye and a 3′ quencher dye. Samples were analyzed

using the ABI 7500 RT-PCR thermocycler in 96-well plates as previously described (239). APC,

138

ITF2, and ALU-C4 were also amplified in exogenously methylated CpGenome DNA (Millipore,

Billerica, MA). The percent methylated reference (PMR) was calculated to assess methylation

using the formula: (Gene of Interest/ALU-C4)sample/(Gene of Interest/ALU-C4)CpGenome x100%. In

order to ensure that DNA quality was adequate, samples with an ALU-C4 threshold cycle greater

than 22 were deemed poor quality and reanalyzed or removed from the study (556).

MLH1 methylation status was assessed by MethyLight as described previously, with positive

methylation defined as PMR ≥ 10% (479). CIMP status was determined using the Weisenberger

panel of markers, described previously (57). Briefly, MethyLight was used to assess a 5-gene

signature consisting of CACNA1G, IGF2, NEUROG1, RUNX3 and SOCS1. Tumours were

classified as CIMP if 3 or more of the 5 genes had PMR ≥ 10% and non-CIMP if 2 or fewer

genes had PMR ≥ 10%. CIMP status was available for a subset of Ontario cases (285 of 330) and

unavailable for Newfoundland cases.

5.3.4 Statistics

Comparison of the methylation status of matched tumour and normal DNA samples was

performed using McNemar’s test. Results were considered statistically significant if two-sided P

< 0.05. Pearson’s chi-square test was used to measure associations between clinicopathological

variables and ITF2 and APC methylation in tumour DNA. Bonferroni correction was used to

account for multiple comparisons. All analyses were performed using IBM SPSS Statistics 21

(Armonk, NY).

5.4 Results

5.4.1 ITF2 promoter methylation in CRC tumours and normal colorectal

mucosa

Patient clinicopathological features are shown in Table 5.1 for MSI-H and MSS cases for the

Ontario and Newfoundland populations. Promoter methylation of ITF2 was quantified using

MethyLight in CRC tumours from Ontario and Newfoundland. To quantify ITF2 promoter

methylation in both normal mucosa and CRC tumour tissue 47 normal-tumour pairs from

Ontario and Newfoundland were randomly selected, of which 12 were MSI-H and 35 were MSS.

The mean PMR was 8.8% in tumour DNA and 1.6% in normal colorectal DNA. Methylation

139

levels ranged from 0–31.1% in tumour and 0–15.6% in normal colonic DNA.

In order to find an ideal value to dichotomize PMR values for ITF2, a variety of cut-offs were

tested including 5% PMR, 10% PMR, mean, median, and third quartile. A cut-off of 10% to

dichotomize methylated and unmethylated samples was the most optimal. This is in line with

previous reports showing that a threshold of 10% PMR can discriminate between tumour and

adjacent normal tissue for a variety of gene markers and is sufficiently above background

measurements of methylation, yet lower than the PMR values usually obtained for most markers

in colorectal tumours (559). McNemar’s test comparing methylation above this cut-off in tumour

and normal tissues in CRC patients determined that tumour methylation of ITF2 was

significantly higher than normal colorectal mucosa methylation (P = 3.4x10-9).

For ITF2, comparable methylation levels were seen in CRC tumours between the two

populations, comprised of 165 MSI-H and 165 MSS cases from Ontario and 51 MSI-H and 51

MSS cases from Newfoundland. In the Ontario cases the mean PMR for all 330 cases was 8.5%.

Methylation values ranged from 0–57.2%. Using the PMR cut-off of 10%, 34.2% (113/330) of

cases were considered positively methylated in the Ontario cohort. In Newfoundland the mean

PMR for all 102 cases was 15.4%. Methylation values ranged from 0–95.4%. Using the PMR

cut-off of 10%, 43.1% (44/102) of cases were considered positively methylated in the

Newfoundland cohort.

5.4.2 APC promoter methylation in CRC tumours and normal colorectal

mucosa

Methylation of the APC promoter region was quantified in tumour and matched normal

colorectal mucosa of 47 randomly selected patients from both Ontario and Newfoundland. Mean

methylation was 16.1% in tumour DNA and 2.6% in normal colon. Methylation values ranged

from 0–60.2% for tumour samples and from 0–11.9% for normal samples. A PMR cut-off of

10% was used to dichotomize methylated and unmethylated samples, determined the same way

as for ITF2, described above. McNemar’s test comparing methylation above this cut-off in

tumour and normal tissues in CRC patients determined that tumour methylation of APC was

significantly higher than normal colorectal mucosa methylation (P = 9.5x10-10).

Comparable methylation levels of APC were seen in CRC tumours between the two populations.

140

The mean PMR in Ontario was 11.5%. Methylation values ranged from 0–92.2%. For the

Newfoundland samples, the mean PMR was 13.9%. Methylation values ranged from 0–70.5%.

Using the PMR cut-off of 10%, 34.8% of tumours (115/330) were methylated in the Ontario

cohort and 40.2% of tumours (41/102) were methylated in the Newfoundland cohort.

5.4.3 ITF2 methylation and clinicopathological features, including MSI

subtype, in two distinct CRC cohorts

I next examined whether methylation of ITF2 in tumour DNA was associated with patient

clinicopathological features. Methylation status was compared by Pearson’s chi-square test

between MSI-H and MSS cases. In Ontario 26.1% (43/165) of MSS cases compared to 42.4%

(70/165) of MSI-H cases were methylated, with an odds ratio (OR) of 2.09 [95% confidence

interval (CI) 1.31– 3.33], and P = 0.002. Similarly in Newfoundland 29.4% (15/51) of MSS

cases and 56.9% (29/51) of MSI-H cases were methylated with OR of 3.16 (95 % CI 1.40–7.17),

and P = 0.005. Ontario and Newfoundland data was then pooled for further analysis.

Due to the selection strategy and the association between MSI and ITF2 methylation, further

clinicopathological associations were analyzed separately for MSI-H and MSS cases, shown in

Table 5.2. There was a significant association between ITF2 promoter methylation with MLH1

promoter methylation in MSI-H cases, P = 6.7×10-4, OR 1.88 (1.10–3.68). There was also a trend

towards an association between ITF2 methylation and female sex in MSI-H cases, and ITF2

methylation and CIMP in MSS cases, but this was not significant using a conservative P-value

adjusted for multiple comparisons. No other significant associations were found for either MSS

or MSI-H cases between ITF2 methylation and early age of onset (<50 years), stage, grade,

tumour location, histological type, MMR protein status, MMR germline mutation, BRAF V600E

mutation, or survival status. Stage-specific survival was also performed, to account for potential

differences in early stage survival compared to stage IV in MSI-H cases. There was a trend

toward higher overall survival in stage I MSI-H cases with ITF2 methylation, but results were

not significant after correction for multiple comparisons.

141

Table 5.2 Associations between ITF2 methylation and clinicopathological features in

tumour DNA. Analysis of 216 MSI-H and 216 MSS CRC patients from Ontario and

Newfoundland. MSS (n=216) MSI-H (n=216)

Unmethylated (%)

Methylated (%)

OR (95% CI)a

P-value Unmethylated (%)

Methylated (%)

OR (95% CI)a

P-value

Age <50 50+

24 (15.2) 134 (84.8)

6 (10.3) 52 (89.7)

1.552 (0.600-4.014)

0.36

24 (20.5) 93 (79.5)

14 (14.1) 85 (85.9)

1.567 (0.761-3.225)

0.22

Sex Male Female

73 (46.2) 85 (53.8)

26 (44.8) 32 (55.2)

1.057 (0.577-1.935)

0.86

64 (54.7) 53 (45.3)

36 (36.4) 63 (63.6)

2.113 (1.222-3.655)

0.007

TNM Stageb

1 2 3 4

36 (22.8) 78 (49.4) 32 (20.3) 12 (7.6)

14 (24.1) 32 (55.2) 12 (20.7) 0 (0.0)

0.720 (0.606-0.856)

0.19

31 (26.5) 57 (48.7) 28 (23.9) 1 (0.9)

18 (18.2) 55 (55.6) 23 (23.2) 3 (3.0)

5.167 (0.499-53.450)

0.32

Histological Gradeb

Low Moderate High

16 (10.8) 118 (79.7) 14 (9.5)

4 (7.5) 42 (79.2) 7 (13.2)

2.000 (0.482-8.295)

0.62

9 (13.8) 39 (60.0) 17 (26.2)

8 (13.1) 42 (68.9) 11 (18.0)

0.728 (0.215-2.459)

0.51

Locationc

Proximal Distal

58 (37.9) 95 (62.1)

29 (50.9) 28 (49.1)

0.589 (0.319-1.089)

0.09

60 (89.6) 7 (10.4)

54 (87.1) 8 (12.9)

1.270 (0.432-3.735)

0.66

Histological Typed

Non-Mucinous Mucinous

140 (90.3) 15 (9.7)

49 (84.5) 9 (15.5)

1.714 (0.705-4.167)

0.23

76 (66.7) 38 (33.3)

73 (75.3) 24 (24.7)

0.658 (0.360-1.202)

0.17

MMR Protein Status Intact Deficient

143 (97.9) 3 (2.1)

55 (98.2) 1 (1.8)

0.867 (0.088-8.511)

0.90

21 (18.6) 92 (81.4)

9 (9.1) 90 (90.9)

2.283 (0.992-5.251)

0.05


157 (99.4) 1 (0.6)

58 (100.0) 0 (0.0)

0.730 (0.673-0.792)

0.54

89 (76.1) 28 (23.9)

74 (74.7) 25 (25.3)

1.074 (0.577-1.999)

0.82

MLH1 Methylation Unmethylated Methylated

143 (96.0) 6 (4.0)

52 (100.0) 0 (0.0)

0.733 (0.674-0.798)

0.14

68 (58.1) 49 (41.9)

42 (42.4) 57 (57.6)

1.883 (1.095-3.238)

6.7x10-4

BRAF V600E Mutation No Yes

141 (92.2) 12 (7.8)

51 (91.1) 5 (8.9)

1.152 (0.387-3.431)

0.80

70 (63.1) 41 (36.9)

50 (52.6) 45 (47.4)

1.537 (0.088-2.683)

0.13

CIMP Status Negative Positive

100 (96.2) 4 (3.8)

33 (84.6) 6 (15.4)

4.545 (1.208-17.100)

0.02

50 (61.7) 31 (38.3)

29 (47.5) 32 (52.5)

1.780 (0.908-3.489)

0.09

142


110 (69.6) 48 (30.4)

35 (60.3) 23 (39.7)

1.475 (0.790-2.755)

0.22

78 (66.7) 39 (33.3)

70 (70.7) 29 (29.3)

0.829 (0.464-1.478)

0.52

Survival Status Stage I Alive Deceased Stage II Alive Deceased Stage III Alive Deceased Stage IV Alive Deceased

30 (81.1) 7 (18.9) 56 (71.8) 22 (28.2) 24 (75.0) 8 (25.0) 0 (0.0) 12 (100.0)

10 (71.4) 4 (28.6) 19 (59.4) 13 (40.6) 6 (50.0) 6 (50.0) 0 0

1.714 (0.414-7.105) 1.742 (0.736-4.119) 3.00 (0.750-11.995) n/a

0.45 0.20 0.11 n/a

19 (61.3) 12 (38.7) 44 (77.2) 13 (22.8) 15 (53.6) 13 (46.4) 0 (0.0) 1 (100.0)

16 (88.9) 2 (11.1) 36 (65.5) 19 (34.5) 16 (69.6) 7 (30.4) 2 (66.7) 1 (33.3)

0.198 (0.039-1.018) 1.786 (0.778-4.104) 0.505 (0.159-1.607) 2.00 (0.500-7.997)

0.04 0.17 0.24 0.25

a Odds ratio and 95% confidence interval for methylated versus unmethylated. b OR and 95% CI given for lowest stage/grade versus highest stage/grade. c Proximal tumour location includes lesions up to and including the splenic flexure. d Mucinous histology includes the presence of any mucin within the tumour stroma.

143

5.4.4 APC methylation and clinicopathological features, including MSI

subtype, in two distinct CRC cohorts

I next examined whether methylation of APC in tumour DNA was associated with patient

clinicopathological features in both cohorts. Methylation status was compared by Pearson’s chi-

square test between MSI-H and MSS cases. In Ontario 32.1% (53/165) of MSS cases were

methylated while 37.6% (62/165) of MSI-H cases were methylated with an OR (95% CI) of 1.27

(0.81–2.00), and P = 0.30. Similarly, in Newfoundland 35.3% (18/51) of MSS cases were

methylated and 45.1% (23/51) of MSI-H cases methylated with OR (95% CI) of 1.51 (0.68–

3.34), and P = 0.31.

I then tested whether APC methylation was associated with patient clinicopathological features

in pooled CRC cases from Ontario and Newfoundland. Results are shown in Table 5.3.

Methylation of APC was not associated with early age of onset (<50 years), sex, stage, grade,

tumour location, histological type, CIMP status, MMR protein status, MMR germline mutation,

MLH1 methylation, BRAF V600E mutation, or survival status.

144

Table 5.3 Associations between APC methylation and clinicopathological features in

tumour DNA. Analysis of 216 MSI-H and 216 MSS CRC patients from Ontario and

Newfoundland. MSS (n=216) MSI-H (n=216)

Unmethylated (%)

Methylated (%)

OR (95% CI)a

P-value Unmethylated (%)

Methylated (%)

OR (95% CI)a

P-value

Age <50 50+

20 (13.8) 125 (86.2)

10 (14.1) 61 (85.9)

0.976 (0.431-2.213)

0.95

23 (17.6) 108 (82.4)

15 (17.9) 70 (82.4)

0.994 (0.485-2.035)

0.99

Sex Male Female

64 (44.1) 81 (55.9)

35 (49.3) 36 (50.7)

0.813 (0.460-1.436)

0.48

59 (45.0) 72 (55.0)

41 (48.2) 44 (51.8)

0.879 (0.509-1.520)

0.65

TNM Stageb

1 2 3 4

27 (18.6) 79 (54.5) 30 (20.7) 9 (6.2)

23 (32.4) 31 (43.7) 14 (19.7) 3 (4.2)

0.391 (0.095-1.619)

0.15

32 (24.4) 64 (48.9) 31 (23.7) 4 (3.1)

17 (20.0) 48 (56.5) 20 (23.5) 0 (0.0)

0.653 (0.533-0.801)

0.31

Histological Gradeb

Low Moderate High

12 (8.9) 108 (80.0) 15 (11.1)

8 (12.1) 52 (78.8) 6 (9.1)

0.600 (0.163-2.207)

0.72

8 (11.0) 50 (68.5) 15 (20.5)

9 (17.0) 31 (58.5) 13 (24.5)

0.770 (0.230-2.578)

0.47

Locationc

Proximal Distal

62 (44.0) 79 (56.0)

25 (36.2) 44 (63.8)

1.381 (0.763-2.499)

0.29

63 (85.1) 11 (14.9)

51 (92.7) 4 (7.3)

0.449 (0.135-1.495)

0.18

Histological Typed

Non-Mucinous Mucinous

129 (89.6) 15 (10.4)

60 (87.0) 9 (13.0)

1.290 (0.534-3.114)

0.57

87 (68.5) 40 (31.5)

62 (73.8) 22 (26.2)

0.772 (0.418-1.426)

0.41

MMR Protein Status Intact Deficient

132 (97.8) 3 (2.2)

66 (98.5) 1 (1.5)

0.668 (0.068-6.533)

0.73

21 (16.3) 108 (83.7)

9 (10.8) 74 (89.2)

1.599 (0.694-3.685)

0.27


144 (99.3) 1 (0.7)

71 (100.0) 0 (0.0)

0.670 (0.610-0.736)

0.48

104 (79.4) 27 (20.6)

59 (69.4) 26 (30.6)

1.697 (0.908-3.175)

0.10

MLH1 Methylation Unmethylated Methylated

131 (96.3) 5 (3.7)

64 (98.5) 1 (1.5)

0.409 (0.047-3.577)

0.41

62 (47.3) 69 (52.7)

48 (56.5) 37 (43.5)

0.693 (0.400-1.199)

0.19

BRAF V600E Mutation No Yes

126 (90.6) 13 (9.4)

66 (94.3) 4 (5.7)

0.587 (0.184-1.873)

0.36

67 (53.2) 59 (46.8)

53 (66.3) 27 (33.8)

0.579 (0.324-1.034)

0.06

CIMP Status Negative Positive

91 (92.9) 7 (7.1)

42 (93.3) 3 (6.7)

0.929 (0.229-3.770)

0.92

45 (52.3) 41 (47.7)

34 (60.7) 22 (39.3)

0.710 (0.359-1.406)

0.33

145


95 (65.5) 50 (34.5)

50 (70.4) 21 (29.6)

0.782 (0.424-1.444)

0.43

91 (69.5) 40 (30.5)

57 (67.1) 28 (32.9)

1.118 (0.622-2.007)

0.71

Survival Status Stage I Alive Deceased Stage II Alive Deceased Stage III Alive Deceased Stage IV Alive Deceased

20 (71.4) 8 (28.6) 54 (68.4) 25 (31.7) 21 (70.0) 9 (30.0) 0 (0.0) 9 (100.0)

20 (87.0) 3 (13.0) 21 (67.7) 10 (32.3) 9 (64.3) 5 (35.7) 0 (0.0) 3 (100.0)

0.375 (0.087-1.622) 1.029 (0.422-2.504) 1.296 (0.338-4.968) n/a

0.18 0.95 0.71 n/a

24 (75.0) 8 (25.0) 45 (70.3) 19 (29.7) 20 (64.5) 11 (35.5) 2 (50.0) 2 (50.0)

11 (64.7) 6 (35.3) 35 (72.9) 13 (27.1) 11 (55.0) 9 (45.0) 0 0

1.636 (0.457-5.865) 0.880 (0.383-2.022) 1.488 (0.472-4.688) n/a

0.45 0.76 0.50 n/a

a Odds ratio and 95% confidence interval for methylated versus unmethylated. b OR and 95% CI given for lowest stage/grade versus highest stage/grade. c Proximal tumour location includes lesions up to and including the splenic flexure. d Mucinous histology includes the presence of any mucin within the tumour stroma.

146

5.5 Discussion

Understanding the genetic and epigenetic differences amongst colorectal cancer subtypes is

essential, as CRC subtypes differ in their treatment options and offer distinct survival outcomes.

The Wnt signaling pathway is dysregulated in a majority of colorectal tumours and can be

altered at the extracellular, intracellular and gene target level (118,629,632). These changes to

the Wnt pathway also differ based upon microsatellite instability status, as demonstrated in the

results of this Chapter. I quantified the methylation status of the ITF2 and APC promoter CpG

islands in a nested case-case study in two cohorts of colorectal carcinoma from two different

populations, comparing cases by MSI status. The ITF2 promoter is hypermethylated in tumour

tissues compared with matched non-neoplastic mucosa, and further, MSI-H tumours are more

likely to incur promoter methylation compared with MSS tumours. ITF2 promoter methylation

was also significantly associated with MLH1 promoter methylation, a common occurrence in

MSI-H tumours. Conversely, it was found that APC, an important intracellular regulator of Wnt

signaling marked by both genetic mutations and hypermethylation in CRC, acquires DNA

methylation equally across subtypes. A schematic diagram indicating methylation events in the

MSS and MSI stubtypes of CRC is shown in Figure 5.1.

This is the first study to investigate DNA methylation of ITF2 in CRC cases. Here, it was

established that ITF2 methylation is a tumour-associated event, being a rare occurrence in

normal colorectal DNA. One sample out of 47 normal tissue samples was methylated, but this

rare occurrence may possibly be due to the field effect, or field cancerization, in which

apparently normal cells acquire genetic and/or epigenetic alterations and may eventually

progress to cancer. With regards to tumour methylation of ITF2, I showed that it was associated

with the MSI-H phenotype. ITF2 has been reported to be a tumour suppressor that can induce

cell cycle arrest and is sometimes lost due to loss of heterozygosity at 18q21 (633). However,

ITF2 expression has been found to be upregulated in some cancers with aberrantly activated Wnt

signaling but decreased in others (624,633). Further research is required to elucidate the role of

ITF2 in tumourigenesis. Treatment of gastric cancer cell lines with the DNA methyltransferase

inhibitor 5-aza-2’-deoxycytidine (5-aza) restored mRNA expression in cell lines that had ITF2

hypermethylation demonstrating methylation-dependent regulation of this gene (627). Thus,

transcriptional silencing in CRC through methylation would likely lead to a decrease in its

147

cellular expression levels potentially contributing to tumourigenesis.

APC promoter methylation is rarely observed in normal colorectal tissue compared with CRC

tumour tissue in our study population, which replicates the findings of a recent meta-analysis

(634). However, contrary to ITF2 methylation, there was no association with MSI-H CRC or any

other clinical features. APC expression is at least partially regulated by DNA methylation, as its

expression increases in CRC cell lines after treatment with 5-aza (635). Several other studies

have investigated the correlation between APC methylation and MSI with varying results.

Studies have shown wide variation in overall APC methylation, regardless of subtype, from as

low as 18% to as high as 63.4% (616,617). These current results demonstrate a more moderate

level of 34–40% of cases methylated. Findings in the literature for the correlation between APC

methylation and MSI are even less clear, with methylation in MSI-H tumours ranging from 14.3–

72.7% (617,622). However, these studies analyzed small numbers of patient samples, with a

maximum of 44 MSI tumours used (620). This current study, on the other hand, employed a total

of 432 samples, 216 of which were MSI-H. This sample size is many times larger than in any

other study of its kind, giving more statistical power and robustness to the results.

There were no differences between level of methylation at different stages of CRC diagnosis for

either APC or ITF2, indicating these may be early epigenetic events in tumourigenesis.

Additionally, APC methylation has been detected in colon adenoma, further evidence that it is an

early event (636,637). Detection of APC may be further exploited as a potential biomarker by

detection in other biospecimens, as its methylation has been detected in both stool and plasma

(638–640). Further investigation of the presence of ITF2 methylation in adenomas should be

undertaken, as well as whether its methylation can be detected in stool or plasma. This research

will indicate the potential of utilizing ITF2 and APC, perhaps in combination with other

methylation markers, as non-invasive stool- or plasma-based methylation markers for CRC

detection and/or subtype discrimination.

Data from colon and rectal tumours from The Cancer Genome Atlas (TCGA) shows that APC

mutation rates differ among the 224 tumours sequenced by exome sequencing. TCGA data

described hypermutated tumours, which have a mutation rate of 12 mutations per 106 bases and

consist mostly of MSI-H tumours. The prevalence of APC mutation in these hypermutated

tumours is 51% (118). Alternatively, non-hypermutated tumours, defined by a mutation rate

148

<8.24 per 106 bases and consisting mostly of MSS tumours, incurred APC mutations in 81% of

cases. This disparity in APC mutation rates may be explained by DNA methylation to inactivate

APC leading to constitutive ligand-independent Wnt signaling. In this same data set ITF2 is

genetically altered in only 3% of tumours, thus, methylation is likely to play a larger role in ITF2

dysregulation in cancer than mutation (641,642).

While MSI-H tumours are a largely well-defined subtype of CRC, MSS tumours comprise the

majority of cases and exhibit a wide variety of molecular characteristics. Thus, there is an

emerging research focus to further classify molecular subtypes of CRC. Recently, four consensus

molecular subtypes were defined. The first subtype consists mostly of MSI cases (612). The

remaining three subtypes are defined by ‘canonical’ Wnt and MYC activation, metabolic

dysregulation, or mesenchymal activation. The results in this chapter indicate that some MSS

cases incur methylation of the Wnt genes studied, so perhaps these cases belong to the subtype

characterized by Wnt activation. It would be interesting to see which sub-classification the MSS

cases used in this study belong to, and how ITF2 or APC methylation profiles differ among the

four subtypes.

MSI-H tumours often overlap with CIMP-positive status. Thus, the association found between

MSI-H and ITF2 methylation may in fact be part of the widespread hypermethylation of CpG

islands that characterizes CIMP tumours. CIMP status is available only for some Ontario cases

and none of the Newfoundland cases utilized in this study, thus there is an incomplete picture of

CIMP for the cohorts. From the available data a trend was observed between CIMP-positive

status and ITF2 methylation among MSS cases. However, there were only ten cases in this

group. From the current findings as well as previous investigation into epigenetic regulation of

Wnt signaling genes it has been found that dysregulation through aberrant methylation is

implicated in all subtypes of CRC, not solely in CIMP-positive cases. APC, which abrogates Wnt

signaling intracellularly, is methylated in a proportion of CRCs, regardless of subtype while

ITF2, a downstream target of Wnt signaling, is methylated more often in MSI-H tumours.

Previous work from the Bapat Lab has determined that DKK1 and SFRP1, coding for two

extracellular Wnt antagonists, incur promoter methylation that segregates strongly with different

CRC subtypes. DKK1 methylation is associated with the MSI-H phenotype and other MSI-

associated features, while SFRP1 methylation is associated with MSS tumours (632). It was also

found that WNT5A methylation, which codes for an extracellular ligand of the non-canonical

149

Wnt pathway, is associated with MSI-H (629). These results were found in the same cohort of

Ontario and Newfoundland patients used in this study. These observations underscore the

importance of both Wnt signaling and the role of DNA methylation in CRC.

One limitation of this study to bear in mind is that only a subset of available MSS cases was

chosen for analysis by matching to MSI-H cases by age quartile, stage and sex. Individuals with

MSI-H CRC are generally a younger age, more frequently female, have a lower tumour stage

and are more frequently CIMP-positive than those individuals with MSS tumours. Thus, the

MSS cases analyzed in this study do not wholly represent all MSS cases from the Ontario and

Newfoundland populations. Additionally, MSS cases were not selected from the entire Ontario

cohort, but only a subset available at the time this study was undertaken. The subset that was

selected from did not differ in age, sex, stage or CIMP rates from the entire cohort.

The strengths of this study include large sample size, the inclusion of two independent well-

characterized population-based cohorts, and the choice of technology. The use of MethyLight

technology is superior to methylation-specific PCR (MSP) and offers several advantages

including a quantitative, high-throughput methylation-specific real-time PCR-based technique,

which is amenable to using small quantities of DNA extracted from archival tissue specimens.

MSP is a more qualitative and subjective method that has been used in many prior studies of

APC methylation.

The findings outlined in this chapter demonstrate the importance of DNA methylation in the

regulation of genes selected for analysis and its differing effects based on tumour subtype. ITF2

is not yet well studied in CRC and I have now shown that this gene incurs MSI-associated

hypermethylation. For APC, both mutation and methylation play a role in its dysregulation. It is

likely that methylation of APC plays a secondary role in CRC to more commonly occurring

mutations and may act to fine-tune Wnt signaling. With both mutation and methylation

contributing to regulation of this gene, it is possible the sequence of events may dictate the way

CRC evolves. Based on its high specificity for CRC, APC methylation may offer usefulness as a

marker within a panel of other genes for CRC detection and ITF2 may be useful for detection of

MSI-H tumours. Future studies to independently validate these findings are warranted. Overall,

this study has investigated methylation of the Wnt genes APC and ITF2 in two large cohorts of

MSI-H and matched MSS CRC tumours to find that ITF2 methylation is significantly associated

150

with MSI-H tumours while APC methylation is a tumour-specific event in CRC, which does not

differ significantly between MSI-H and MSS subtypes or other clinicopathological variables.

151

Figure 5.1 Schematic of methylation events occurring in MSS and MSI CRC. In

microsatellite stable/chromosomal instability (MSS/CIN) colorectal cancer, the APC gene is

frequently mutated leading to increased levels of Wnt signaling. In CRC with microsatellite

instability (MSI), MLH1 is often methylated, which is associated with the variant A genotype of

the polymorphism rs1800734. ITF2 also incurs methylation in MSI tumours. The CpG island

methylator phenotype (CIMP) is frequently present in MSI tumours, which may also contribute

to these methylation events. APC incurs methylation equally among both MSS and MSI

tumours, regardless of its mutation status, which may fine-tune expression of this gene.

152

Chapter 6 Discussion and Future Directions

6.1 Further exploration of MLH1-93G>A SNP

The research outlined in this thesis has explored the epigenetic regulation and molecular

mechanisms occurring at the important MLH1 locus in CRC. These results have shed new light

on the epigenetic concept of CpG shores as well as further described how DNA variation can

play a role in epigenetics and cancer susceptibility. I have shown that the MLH1 promoter SNP

rs1800734 is associated with shore hypomethylation in peripheral blood mononuclear cell DNA

of CRC cases and control individuals, as well as normal colonic DNA of CRC cases, in contrast

to its previously defined role in CpG island hypermethylation in CRC tumours.

In the past several years there have been a growing number of studies linking SNP genotype to

DNA methylation changes, not only in cis but also associated with regions located on other

chromosomes (480,514,520,643,644). However, to my knowledge, rs1800734 is the only SNP

reported in the literature having differential effects on DNA methylation not only at two distinct

but closely spaced regulatory regions, the MLH1 CpG island and its upstream shore, but also

differing between normal and tumour DNA. It would be interesting to see whether this

phenomenon is unique to the MLH1 region, as it is such a critical gene required for DNA repair

that frequently undergoes tumour-associated CpG island hypermethylation or mutation in CRC.

Studies to address whether this phenomenon occurs at other regions of the genome, and perhaps

in other cancer types, would require large patient and control cohorts with multiple tissue sources

available for DNA methylation assessment, as well as genome-wide SNP genotyping analysis.

My project benefitted from its unique, well-characterized cohorts, and there are likely other such

SNPs in the genome that display this distinct genotype-epigenotype association.

ChIP experiments testing for the presence of the transcription factor AP4 revealed that AP4 is

present in the wildtype GG cell line HCT 116 at the region in the MLH1 promoter containing its

DNA-binding sequence. Enrichment for AP4 was significantly higher in HCT 116 than the

heterozygous GA cell line SNU-C2B. Binding of AP4 in both of these cell lines was statistically

significantly higher than the AA cell line HCT-15. These experiments demonstrate a functional

consequence of DNA variation, in that the A allele appears to abolish the ability of AP4 to bind

153

to this site. Future confirmation of this occurrence in a variety of other cell lines would further

verify these findings.

There is a putative MYC binding site located fifteen base pairs downstream of rs1800734.

Together, AP4 and MYC may act to activate MLH1 transcription. In addition, ChIP experiments

for the insulator protein CTCF revealed that binding of this protein might also be genotype-

specific as there was significantly less enrichment in the AA cell line HCT-15 compared to

wildtype cell line HCT 116. CTCF may ensure proper DNA methylation patterns are maintained

at MLH1, confining methylation to the shore and preventing it from spreading to the island.

There are likely a variety of other factors involved that are yet to be elucidated. The work of

previous students in the Bapat lab has demonstrated up to five unknown factors binding in the

vicinity of rs1800734 by EMSA experiments (480). The transcription factors may also be other

members of the basic helix-loop-helix family of transcription factors that AP4 belongs to that can

also bind to the same DNA sequence. For example, myogenin (MYOG) was predicted to bind at

the DNA sequence containing G, but not A, by the TRANSFAC program. This factor is muscle-

specific, thus I did not pursue ChIP experiments for MYOG in colorectal cancer cell lines.

Potentially different tissue types utilize different tissue-specific factors at MLH1.

Another transcription factor that was predicted to bind the G but not the A allele using the

TFBIND program was activating transcription factor 2 (ATF2). ATF2 binds cAMP-responsive

elements and can form a homodimer or a heterodimer with c-Jun. Interestingly it is also a histone

acetyltransferase for histones H2B and H4. ChIP-seq datasets from ENCODE (UCSC Genome

Browser) indicate that ATF2 is present at or close to rs1800734 in two cell lines tested, human

embryonic stem cells H1-hESC and human lymphoblastoid cells GM12878. GM12878 cells also

demonstrate binding of ATF2 further upstream of the promoter within the shore of MLH1. This

would be another interesting protein to test for, both for its ability to activate transcription as

well as its acetyltransferase activity. However, at the time of carrying out ChIP experiments there

were no suitable ChIP-grade antibodies available for ATF2.

Rather than performing ChIP or other related experiments on a large panel of potential proteins,

another innovative approach to determine the full complement of proteins binding the MLH1

region in a sequence-specific manner would be to utilize a mass spectrometry based technique.

By utilizing a pull-down assay for specific DNA sequence followed by two-dimensional gel

154

electrophoresis, proteins bound to the DNA locus of interest can be isolated then subjected to

mass spectrometry for identification (645–647). While this technique is not feasible for genome-

wide studies, it can be incredibly useful for protein identification at genes or other regions of

importance to tumourigenesis and disease.

ChIP experiment results did not show statistically significant differences between cell lines with

different rs1800734 genotypes for some of the histone modifications tested. The selected marks

included markers of repressed regions (H3K27me3), enhancers (H3K4me1 and H3K27ac) and

active promoters (H3K4me3 and H3K27ac), however this was not an exhaustive analysis. It is

possible that other histone modifications differ amongst cell lines, which may be modulated by

genotype, DNA methylation, or other cell type-specific factors. In addition, the proteins

responsible for writing or erasing these marks may differ, which were not tested for. There are a

large number of histone methyltransferases, histone demethylases, histone acetyltransferases, and

histone deacetylases that have yet to be assayed in any of the cell lines utilized in this thesis.

There was enrichment for H3K4me1 in the MLH1 shore that was not accompanied by increased

H3K27ac. Thus, in HCT 116, SNU-C2B, and HCT-15 cell lines this region does not appear to be

an active enhancer. However, SNU-C2B had significantly higher enrichment for H3K4me1 than

HCT 116 at the shore region S2. SNU-C2B was also hypomethylated at the shore compared to

HCT 116. Perhaps the rs1800734 SNP variant promotes both of these occurrences creating a

more enhancer-like state at the MLH1 shore. ChIP-seq data generated by the Broad

Institute/ENCODE annotates the shore region of MLH1 as a strong enhancer in several cell lines

tested, including K562 (myelogenous leukemia), HUVEC (human umbilical vein endothelial

cell), and NHEK (keratinocyte) cell lines. The annotations are based on computational

integration of ChIP-seq data for nine different factors (648,649). Thus, the MLH1 shore may

indeed be an enhancer in certain cell types, and this should be explored further to determine

under what circumstances or in which tissues it enhances transcription of MLH1 itself or perhaps

another gene located elsewhere in the genome. MLH1 incurs tumour-associated promoter

hypermethylation in only some tumour types, such as colorectal, endometrial, ovarian, and

gastric cancers (650–652). The MLH1-93G>A SNP has also shown association with increased

risk for several cancer types, including glioblastoma, lung, endometrial, and ovarian cancers

(481–483,560,561). The reasons why only some tissue types incur MLH1 methylation and MSI

155

are unclear, but may be a complex combination of genetic variants, environmental factors, and

tumour microenvironment.

The focus of this thesis has been on the upstream regulatory regions of MLH1, including its CpG

shore and island. Shores are defined as regions flanking CpG islands, located up to 2000 bp

away, which includes both the 5’ and 3’ direction. Upstream shores, located 5’ to islands, are

referred to as north shores, while downstream shores, located 3’ to islands, are referred to as

south shores. MLH1 does in fact have a south shore which is located within intron 1 of the gene.

Only two CpG sites within this shore were interrogated on the Illumina 450K arrays in PBMCs

of controls and CRC cases, as outlined in Chapter 2. Methylation at these probes was not

associated with SNP genotype of rs1800734 thus the south shore of MLH1 was not investigated

further. It has been demonstrated that many unmethylated CpG islands are flanked both upstream

and downstream by hypermethylated shores in normal tissues (203,653). This was observed in

our PBMC samples at MLH1, with maximum methylation of the north shore of 0.758 and

maximum south shore methylation of 0.885, whereas the CpG island was unmethylated, with a

maximum methylation β-value of 0.067. However, not every single CpG site within the shore

was on the 450K array, thus it cannot be ruled out that some CpGs in this south shore are

modified by variant SNP genotype in a similar way to the upstream north shore which was the

focus of my research.

6.2 Other genes and/or polymorphisms associated with MLH1-

93G>A

MLH1 shares a bidirectional promoter with the gene EPM2AIP1 (EPM2A interacting protein 1),

which is transcribed on the opposite strand 321 base pairs upstream of the MLH1 TIS (574). The

MLH1 shore is located within exon 1 of EPM2AIP1, which is its only exon. It is possible that the

SNP rs1800734 and/or differential shore methylation studied in this thesis may play a role in

EPM2AIP1 regulation in some way, rather than only at MLH1. The function of EPM2AIP1 is not

well defined in cancer, normal colorectal cells, or other tissues. It is known to bind to EPM2A

(aka laforin), a gene which when mutated causes an autosomal recessive form of adolescent

progressive myoclonus epilepsy. Both EPM2A and EPM2AIP1 appear to be ubiquitously

expressed (528). Epm2aip1-/- knockout mice have been generated and their phenotype

156

demonstrates that Epm2aip1 plays a role in glycogen synthesis (529). Absence of Epm2aip1 in

mice leads to impaired activation of glycogen synthase by glucose 6-phosphate, decreased

hepatic glycogen synthesis, increased liver fat, and hepatic insulin resistance. No phenotypes

related to the colon, rectum, or cancer have been reported.

MLH1 and EPM2AIP1 are expressed in CRC cell lines that lack MLH1 CpG island

hypermethylation, while methylation-induced silencing of MLH1 also silences EPM2AIP1

expression (574). It was previously shown in our lab lab that the G allele of rs1800734 leads to

higher MLH1 expression than the A allele in a number of cell lines including colorectal cancer

(HCT 116, SW620, HT-29, SW480), endometrial cancer (HEC-1-A, SK-UT-1B), normal colon

(CCD 841 CoTr), and embryonic kidney (293T). Interestingly, EPM2AIP1 had significantly

higher expression with the A allele compared to the G allele for several of these same cell lines,

HCT 116, SW620, HEC-1-A, and SK-UT-1B. Perhaps the A allele leads to recruitment of some

factor(s) which promotes preferential expression of EPM2AIP1 rather than MLH1.

My research focus has been specifically for rs1800734, as well as the SNPs in LD rs749072 and

rs13098279. While rs1800734 appears to be responsible for differential binding of AP4, there is

a possibility that these or other variants in the region may be contributing to DNA methylation

differences at the shore in normal tissues or the island in tumours. The SNP rs749072 is located

within intron 27 of the LRRFIP2 gene. Several transcription factors show enrichment in this

region including ATF2, BCL11A, FOS, JUN, MEF2C, NFIC, and SMC3 in a variety of cells

lines from ENCODE ChIP-seq datasets. rs13098279 is located between the genes LRRFIP2 and

GOLGA4 and ChIP-seq data sets indicate binding of transcription factors GATA2, NR2F2,

RCOR1, TAL1, and TEAD4 in K562 cells as well as H3K4me1 in the same cell line. Further

functional studies of these SNPs and their effect on transcription factor binding or other

epigenetic regulation would be of interest.

It is also possible that there are other SNPs in the MLH1 region that have a role in epigenetic

regulation at MLH1 and/or CRC susceptibility. To address this, the biostatistician Mathieu

Lemire confirmed that there were no other SNPs in the case or control populations utilized from

the OFCCR that were associated with MLH1 shore methylation results from Illumina 450K

profiling, based on genotyping previously performed on Affymetrix 100K and 500K platforms.

However, there are a number of other SNP genotyping platforms available that may cover other

157

SNPs. I utilized the publicly available SNAP (SNP Annotation and Proxy Search) tool from

Broad Institute to test whether there were other SNPs in linkage disequilibrium with rs1800734,

checking SNPs assayed on a variety of Illumina and Affymetrix arrays using the 1000 Genomes

Pilot 1 data. According to SNAP, the only SNPs in LD with rs1800734 within 500 kb, using a r2

cut-off of 0.80, were the SNPs rs749072 and rs13098279. I also utilized the online tool LDLink

from National Institutes of Health (NIH) to look for other potential SNPs in LD with rs1800734.

Using a cut-off value for r2 of 0.80 and utilizing all populations from Phase 3 of the 1000

Genomes Project again resulted in rs749072 and rs13098279 as candidate SNPs in LD.

Interestingly, a one base pair indel rs35149869 was the top hit, located 5,838 base pairs upstream

of rs1800734 in the 3’ untranslated region of EPM2AIP1. The major allele is C while the minor

allele is a deletion of the C with a minor allele frequency in European 1000 Genomes

populations of 0.265. Since the results from LDLink are the most up-to-date, utilizing Phase 3 of

the 1000 Genomes Project, future research on rs35149869 may be worthwhile, as none currently

exists. However, ENCODE/UCSC GenomeBrowser ChIP-seq data for 161 factors in 91 cell

lines do not show any enrichment for transcription factors or other DNA binding proteins at this

SNP location. Thus, based on this and the proximity of the MLH1-93G>A SNP to both the CpG

island and shore of MLH1, it is likely that rs1800734 is in fact the SNP responsible for the results

assessed in this thesis.

6.3 Differential methylation of Wnt signaling genes

Wnt signaling is required for a variety of processes including generation of cells of the

colorectum. Wnt signaling is most active at the base of the colonic crypt, decreases in activity

upward through the crypt, and is inactive in terminally differentiated cells at the top of the crypt.

A lack of Wnt signaling is normally ensured by a lack of extracellular Wnt ligand, extracellular

Wnt antagonists, and the intracellular β-catenin destruction complex containing, among other

factors, APC. Early in sporadic CRC development, genetic and epigenetic changes occur leading

along a path of chromosomal instability, microsatellite instability, or epigenetic instability.

Among these, CIN tumours incur aberrant upregulation of Wnt signaling the vast majority of the

time, and this is most frequently due to mutations of APC. On the other hand, MSI tumours

feature mismatch repair deficiency, most frequently due to promoter methylation of MLH1.

158

While CIMP may be seen amongst both CIN and MSI subtypes, it is more frequently seen in

MSI CRCs.

APC is mutated in many CRCs, but also in a variety of other cancer types, including up to 20%

of melanomas, prostate, gastric, uterine, and lung cancers (641,642). APC is also methylated in

several cancer types in addition to CRC, including breast and prostate cancers (654,655). Among

CRC subtypes, I have shown that its methylation occurs equally despite the fact that mutation is

more common in tumours with the CIN subtype (118,632). It is likely that APC dysregulation in

CRC is primarily dictated by genetic disruption of the gene, whereas methylation is a secondary

mechanisms that may fine-tune its expression. The frequency of both genetic and epigenetic

alterations to this gene indicates the critical function of APC in tumour suppression.

Other Wnt signaling genes that incur mutation in CRC less frequently, but incur

hypermethylation, do not appear to be dysregulated in the same manner as APC. I have shown

that the Wnt signaling target gene ITF2 incurs methylation more frequently in the MSI-H

subtype of CRC. Our lab has also demonstrated MSI-associated hypermethylation of the

extracellular Wnt antagonist DKK1 as well as WNT5A, involved in non-canonical Wnt signaling

(628,629). This epigenetic dysregulation of Wnt signaling genes incurred in MSI tumours may in

part be due to the significant overlap between the MSI and CIMP phenotype. Conversely,

hypermethylation of SFRP1, another extracellular Wnt antagonist, is inversely associated with

MSI (628). The reasons for distinctive methylation patterns at genes involved in the same

signaling cascade are currently unknown. These differential methylation events may be

specifically adjusted and selected for to create a permissive environment for tumourigenesis. Or

perhaps they are simply passenger epimutations resulting from aberrant epigenetic regulation

irrespective of their function. Further investigation into MSI-associated events may provide

insight into subtype-specific events as well as reasons for improved prognosis in MSI CRC.

6.4 The future of epigenetics in the clinic

All CRCs incur aberrant methylation changes, including genome-wide hypomethylation and

specific CpG island hypermethylation. The average CRC methylome incurs hundreds to

thousands of methylated genes (243,656). DNA hypomethylation was first reported in cancer in

1983, and shortly after in colorectal adenomas and CRC in 1985 (213,214). It was another

159

decade before specific hypermethylated genes were discovered in CRC (657). Since then, there

have been a number of important discoveries of DNA methylation associated with CRC

including MLH1 hypermethylation in MSI CRC and the CpG island methylator phenotype

(168,200). A brief outline of the history of epigenetic discoveries in CRC is described in Figure

6.1.

Thousands of articles have since been published regarding genes that are methylated in CRC,

including potential screening, diagnostic, prognostic, and predictive markers. Thus far, in the

clinic methylation testing for MLH1 is performed in some patients suspected of having Lynch

Syndrome if initial IHC screening for MMR genes is negative (658). For average risk

individuals, the FDA has approved two non-invasive methylation-based biomarker tests for CRC

screening. Cologuard® is a stool-based screening test that detects methylation of BMP3 and

NDRG4 as well as mutant KRAS (659). This test has shown sensitivity of 92.3% and specificity

of 89.8% (660). Recently, Epi proColon® also received approval as a serum-based biomarker

screening test for CRC, which tests for methylation of the SEPT9 gene in circulating DNA (661).

Sensitivity and specificity of the Epi proColon test range from 68.2-98.6% and 80.0-96.7%,

respectively (662). While these tests demonstrate the promising possibilities of epigenetic

contributions in a diagnostic setting, there are currently no non-invasive tests that can not only

detect cancer but also determine cancer subtype in the general population at average risk of CRC.

Identification of patient subtype upfront is critical in selecting suitable treatment and patient

management options.

While a number of studies have been published attempting to further subcategorize CRC, these

have usually been based on MSI, MSS, and CIMP status, sometimes including KRAS or BRAF

mutation. However, the MSS/CIMP-negative category is quite large and varied, comprising at

least 50% percent of cases (244,611). The prognostic implications for these subtypes have been

studied and suggest CIMP-positive tumours are independently associated with worse outcome

(242,663). One study found that MSS CIMP-positive cases with BRAF mutation have highest

disease-specific mortality rates, yet MSS CIMP-negative cases with KRAS mutation also

demonstrated significantly higher mortality rates than other cases in the study (244). Despite the

potential prognostic impact of CIMP, to date no single marker panel has been agreed upon in the

literature, nor is CIMP status utilized in the clinic. Thus, other methods of subtyping CRC into

clinically relevant groups are of interest. Recently, four consensus molecular subtypes were

160

described for CRC from a combination of 18 different data sets, representing 3,104 CRC

tumours (612). The subtypes differ in mutation count, somatic copy number alteration counts

(SCNA), and mRNA gene set enrichment signatures. While MSI/CIMP-high cases primarily

belonged to one subgroup, as may be expected, the remaining cases were broken down into three

separate groupings. Of these, the subgroup marked by high SCNAs, prominent growth factor-β

activation, stromal invasion and angiogenesis had significantly lower overall survival and

relapse-free survival than the other three groups. Whether this method of subgrouping CRCs

gains acceptance remains to be seen. Adoption of a robust classification system will aid in

patient management decisions for this heterogeneous disease.

As described in Chapter 5, APC methylation is a general phenomenon in CRC and its

methylation has been detected in stool samples of CRC patients. Upon further testing for ITF2

methylation in biological fluids, this gene could potentially be used for MSI subtype distinction.

Validated CRC consensus subtypes, such as those described above, with clear differences in

patient survival rates and/or ideal treatment regimens will likely incur differentially methylated

genes. Determining the optimal combination of biomarkers for diagnosis, prognosis, and

prediction of CRC within one non-invasive test will be a challenge for the future. However, once

accomplished, this could greatly improve patient screening rates and decrease CRC-associated

mortality rates (61,95,664,665).

Another interesting facet of epigenetic dysregulation in CRC is the potential to use drugs

targeting epigenetic marks. The DNA methyltransferase inhibitor azacytidine was given FDA

approval for the treatment of myelodysplastic syndrome in 2004 (666). Since then five other

epigenome-modifying agents have been approved for the treatment of lymphomas and myelomas

(667). The use of these agents in the treatment of solid tumours has not progressed as quickly.

Guadecitabine (SGI-110), a second-generation DNA demethylating drug, is currently undergoing

clinical trials in combination with other treatments for metastatic CRC. It remains to be seen

whether these studies are successful, whether the correct combination of guadecitabine and

traditional CRC chemotherapeutic drugs can be used to treat non-metastatic cancers, and what

influence CIMP status will have on efficacy of DNA demethylating agents.

In summary, colorectal cancer is both a genetic and an epigenetic disease. While these two

processes are frequently considered separately, it is clear that at MLH1 genetics and epigenetics

161

interact, with important consequences. I have comprehensively studied the epigenetic effects that

a single nucleotide change in the MLH1 promoter can have. Variant SNP genotype is associated

with hypomethylation of the CpG shore of MLH1 in normal tissues, yet this effect is not seen in

tumours. Despite methylation differences among CRC cell lines, the histone modifications of

H3K4me3, H3K27me3, and H3K27ac do not incur similar changes, while H3K4me1 is

increased at the shore in cell lines with the variant SNP allele. AP4 transcription factor binds at

the MLH1 promoter at the wildtype allele of rs1800734 but not the variant. CTCF binding may

also be altered at the promoter. Here, I have studied the functional epigenetic regulation and

molecular mechanisms occurring at the important MLH1 locus in CRC, shedding new light on

the epigenetic concept of CpG shores and how DNA variants play a role in epigenetics and

cancer susceptibility. If this example of genetic-epigenetic interactions is applied to the whole

genome, there are multitudes of ways in which the genome and epigenome may interact.

Knowledge of such interactions leads to a better understanding of the processes and changes

incurred by the genome and epigenome under both normal circumstances and cancer

development.

162

Figure 6.1 The past, present, and future of epigenetics in colorectal cancer. A brief timeline

of DNA methylation discoveries and advances in colorectal cancer, including potential events in

the future. Figure adapted from Okugawa et al. Gastroenterology 2015; 149(5): 1204-1225.

[Reference: (243)].

163

References

1. Gervaz P, Bucher P, Morel P. Two colons-two cancers: Paradigm shift and clinical implications. J Surg Oncol. 2004;88(4):261–6.

2. Barker N, van Es JH, Kuipers J, Kujala P, van den Born M, Cozijnsen M, et al. Identification of stem cells in small intestine and colon by marker gene Lgr5. Nature. 2007;449(7165):1003–7.

3. Barker N, Clevers H. Leucine-Rich Repeat-Containing G-Protein-Coupled Receptors as Markers of Adult Stem Cells. Gastroenterology. 2010;138(5):1681–96.

4. Snippert HJ, van der Flier LG, Sato T, van Es JH, van den Born M, Kroon-Veenboer C, et al. Intestinal crypt homeostasis results from neutral competition between symmetrically dividing Lgr5 stem cells. Cell. 2010;143(1):134–44.

5. Carulli A, Samuelson L, Schnell S. Uraveling intestinal stem cell behavior with models of crypt dynamics. Integr Biol. 2014;6(3):243–57.

6. Medema JP, Vermeulen L. Microenvironmental regulation of stem cells in intestinal homeostasis and cancer. Nature. 2011;474(7351):318–26.

7. Boman BM, Fields JZ. An APC:WNT Counter-Current-Like Mechanism Regulates Cell Division Along the Human Colonic Crypt Axis: A Mechanism That Explains How APC Mutations Induce Proliferative Abnormalities That Drive Colon Cancer Development. Front Oncol. 2013;3:244.

8. Barker N. Adult intestinal stem cells: critical drivers of epithelial homeostasis and regeneration. Nat Rev Mol Cell Biol. 2014;15(1):19–33.

9. Grady, William M., Carethers JM. Genomic and Epigenetic Instability in Colorectal Cancer Pathogenesis. Gastroenterology. 2008;135(4):1079–99.

10. Cho KR, Vogelstein B. Genetic alterations in the adenoma-carcinoma sequence. Cancer. 1992;70(6):1727–31.

11. Pretlow TP, Barrow BJ, Ashton WS, Riordan MAO, Pretlow TG, Jurcisek J a, et al. Aberrant Crypts+: Putative Preneoplastic Foci in Human Colonic Mucosa. Cancer Res. 1991;51:1564–7.

12. Sakai E, Nakajima A, Kaneda A. Accumulation of aberrant DNA methylation during colorectal cancer development. World J Gastroenterol. 2014;20(4):978–87.

13. Takayama T, Katsuki S, Takahashi Y, Ohi M, Nojiri S, Sakamaki S, et al. Aberrant crypt foci of the colon as precursors of adenoma and cancer. N Engl J Med. 1998;339(18):1277–84.

14. Shpitz B, Bomstein Y, Mekori Y, Cohen R, Kaufman Z, Neufeld D, et al. Aberrant crypt

164

foci in human colons: Distribution and histomorphologic characteristics. Hum Pathol. 1998;29(5):469–75.

15. Takayama T, Ohi M, Hayashi T, Miyanishi K, Nobuoka a, Nakajima T, et al. Analysis of K-ras, APC, and beta-catenin in aberrant crypt foci in sporadic adenoma, cancer, and familial adenomatous polyposis. Gastroenterology. 2001;121(3):599–611.

16. Smith AJ, Stern HS, Penner M, Hay K, Mitri A, Bapat B V., et al. Somatic APC and K-ras codon 12 mutations in aberrant crypt foci from human colons. Cancer Res. 1994;54(21):5527–30.

17. Fleming M, Ravula S, Tatishchev SF, Wang HL. Colorectal carcinoma: Pathologic aspects. J Gastrointest Oncol. 2012;3(3):153–73.

18. Heitman SJ, Ronksley PE, Hilsden RJ, Manns BJ, Rostom A, Hemmelgarn BR. Prevalence of Adenomas and Colorectal Cancer in Average Risk Individuals: A Systematic Review and Meta-analysis. Clin Gastroenterol Hepatol. 2009;7(12):1272–8.

19. Silva P, Albuquerque C, Lage P, Fontes V, Fonseca R, Vitoriano I, et al. Serrated polyposis associated with a family history of colorectal cancer and/or polyps: The preferential location of polyps in the colon and rectum defines two molecular entities. Int J Mol Med. 2016;38(3):687–702.

20. Fu X, Li L, Peng Y. Wnt signalling pathway in the serrated neoplastic pathway of the colorectum: possible roles and epigenetic regulatory mechanisms. J Clin Pathol. 2012;65(8):675–9.

21. Kambara T, Simms LA, Whitehall VLJ, Spring KJ, Wynter CVA, Walsh MD, et al. BRAF mutation is associated with DNA methylation in serrated polyps and cancers of the colorectum. Gut. 2004;53(8):1137–44.

22. Burnett-Hartman A, Newcomb P, Potter J, Passarelli M, Phipps AI, Wurscher M, et al. Genomic Aberrations Occurring in Subsets of Serrated Colorectal Lesions but not Conventional Adenomas. Cancer Res. 2013;73(9):1–6.

23. Sakai E, Fukuyo M, Ohata K, Matsusaka K, Doi N, Mano Y, et al. Genetic and epigenetic aberrations occurring in colorectal tumors associated with serrated pathway. Int J Cancer. 2016;138(7):1634–44.

24. Siegel RL, Miller KD, Jemal A. Cancer statistics. 2016;66(1):7–30.

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

26. Bray F, Jemal A, Grey N, Ferlay J, Forman D. Global cancer transitions according to the Human Development Index (2008 – 2030): a population-based study. Lancet Oncol. 2012;13:790–801.

165

27. Canadian Cancer Society. Canadian Cancer Statistics Special topic: HPV-associated cancers. Public Heal Agency Canada Agency Canada. 2016;

28. Dunlop M, Tenesa A, Farrington S, Ballereau S, Brewster D, Koessler T, et al. Cumulative impact of common genetic variants and other risk factors on colorectal cancer risk in 42 103 individuals. Gut. 2012;1–12.

29. Tezcan G, Tunca B, Ak S, Cecener G, Egeli U, Tezcan G, et al. Molecular approach to genetic and epigenetic pathogenesis of early-onset colorectal cancer. World J Gastrointest Oncol. 2016;8(1):83–98.

30. Giardiello FM, Allen JI, Axilbund JE, Boland CR, Burke CA, Burt RW, et al. Guidelines on Genetic Evaluation and Management of Lynch Syndrome. Dis Colon Rectum. 2014;57(8):1025–48.

31. Aaltonen LA, Salovaara R, Kristo P, Canzian F, Hemminki A, Peltomaki P, et al. Incidence of Hereditary Nonpolyposis Colorectal Cancer and the. N Engl J Med. 1998;338(21):1481–7.

32. Johns LE, Houlston RS. A Systematic Review and Meta-Analysis of Familial Colorectal Cancer Risk. Am J Gastroenterol. 2001;96(10):2992–3003.

33. Butterworth AS, Higgins JPT, Pharoah P. Relative and absolute risk of colorectal cancer for individuals with a family history: A meta-analysis. Eur J Cancer. 2006;42(2):216–27.

34. McCredie M, Williams S, Coates M. Cancer mortality in East and Southeast Asian migrants to New South Wales, Australia, 1975-1995. Br J Cancer. 1999;79(7–8):1277–82.

35. McCredie M, Williams S, Coates M. Cancer mortality in migrants from the British Isles and continental Europe to New South Wales, Australia, 1975-1995. Int J Cancer. 1999;83(2):179.

36. Kaldor J, Khlat M, Parkin DM, Shiboski S, Steinitz R. Log-linear models for cancer risk among migrants. Int J Epidemiol. 1990;19(2):233–9.

37. Iscovich J, Howe GR. Cancer incidence patterns (1972-91) among migrants from the Soviet Union to Israel. Cancer Causes Control. 1998;9(1):29–36.

38. Johnson CM, Wei C, Ensor JE, Smolenski DJ, Christopher I, Levin B, et al. Meta-analyses of Colorectal Cancer Risk Factors. Cancer Causes Control. 2013;24(6):1207–22.

39. Bouvard V, Loomis D, Guyton K, Grosse Y, El Ghissassi F, Benbrahim-Tallaa L, et al. Carcinogenicity of consumption of red and processed meat. Lancet. 2015;16:1599–600.

40. Lee DH, Keum N, Giovannucci EL. Colorectal Cancer Epidemiology in the Nurses’ Health Study. Am J Public Health. 2016;106(9):1599–607.

41. Pinto E, Viegas O, Pinho O, Ferreira IMPLVO. Exposure risks to carcinogens in food: formation of heterocyclic aromatic amines (HAs) and polycyclic aromatic hydrocarbons

166

(PAHs) in grilled muscle foods. Food Chem Toxicol. 2012;50:2128–34.

42. Lewin MH, Bailey N, Bandaletova T, Bowman R, Cross AJ, Pollock J, et al. Red Meat Enhances the Colonic Formation of the DNA Adduct O6-Carboxymethyl Guanine: Implications for Colorectal Cancer Risk. Cancer Res. 2006;66(3):1859–65.

43. Vargas A, Thompson P. Diet and Nutrient Factors in Colorectal Cancer Risk. Nutr Clin Pract. 2012;27:613–23.

44. Weijenberg MP, Luchtenborg M, De Goeij AFPM, Brink M, Van Muijen GNP, De Bruine AP, et al. Dietary fat and risk of colon and rectal cancer with aberrant MLH1 expression, APC or KRAS genes. Cancer Causes Control. 2007;18(8):865–79.

45. Meeker S, Seamons A, Maggio-Price L, Paik J. Protective links between Vitamin D, inflammatory bowel disease and colon cancer. Vol. 22, World Journal of Gastroenterology. 2016. p. 933–48.

46. Porter K, Hoey L, Hughes CF, Ward M, Mcnulty H. Causes, Consequences and Public Health Implications of Low B-Vitamin Status in Ageing. Nutrients. 2016;8:725.

47. Shin W, Yan J, Abratte CM, Vermeylen F, Caudill MA. Choline Intake Exceeding Current Dietary Recommendations Preserves Markers of Cellular Methylation in a Genetic Subgroup of Folate-Compromised Men. J Nutr. 2010;140:975–80.

48. Crider K, Yang T, Berry R, Bailey L. Folate and DNA Methylation: A Review of Molecular Mechanisms and the Evidence for Folate’s Role. Adv Nutr. 2012;3:21–38.

49. Nitter M, Norgård B, Vogel S De, Eussen SJPM, Meyer K, Ulvik A, et al. Plasma methionine, choline, betaine, and dimethylglycine in relation to colorectal cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC). Ann Oncol. 2014;25:1609–15.

50. Moskal A, Norat T, Ferrari P, Riboli E. Alcohol intake and colorectal cancer risk: A dose-response meta-analysis of published cohort studies. Int J Cancer. 2007;120(3):664–71.

51. Wang, Y., Duan, H., Yang, H. and Lin J. A pooled analysis of alcohol intake and colorectal cancer. Int J Clin Exp Med. 2015;8(5):6878–89.

52. Knudsen MD, de Lange T, Botteri E, Nguyen D-H, Evensen H, Steen CB, et al. Favorable lifestyle before diagnosis associated with lower risk of screen-detected advanced colorectal neoplasia. World J Gastroenterol. 2016;22(27):6276–86.

53. Oyesanmi O, Snyder D, Sullivan N, Reston J, Treadwell J, Schoelles KM. Alcohol consumption and cancer risk: understanding causal mechanisms for breast and colorectal cancers. Evid Rep Technol Assess (Full Rep). 2010;(197):1–151.

54. Varela-Rey M, Woodhoo A, Martinez-Chantar M, Mato J, Lu S. Alcohol, DNA Methylation, and Cancer. Alcohol Res. 2011;35(1):25–36.

167

55. Nan H, Lee J, Rimm E, Fuchs C, Giovannucci E, Cho E. Prospective study of alcohol consumption and the risk of colorectal cancer before and after folic acid fortification in the United States. Ann Epidemiol. 2014;23(9):558–63.

56. Chen K, Xia G, Zhang C, Sun Y. Correlation between smoking history and molecular pathways in sporadic colorectal cancer: A meta-analysis. Int J Clin Exp Med. 2015;8(3):3241–57.

57. Weisenberger DJ, Levine AJ, Long TI, Buchanan DD, Walters R, Clendenning M, et al. Association of the Colorectal CpG Island Methylator Phenotype with molecular features, risk factors and family history. Cancer Epidemiol Biomarkers Prev. 2016;24(3):512–9.

58. Song M, Hu FB, Spiegelman D, Chan AT, Wu K, Ogino S, et al. Long-term status and change of body fat distribution, and risk of colorectal cancer: a prospective cohort study. Int J Epidemiol. 2015;59:1–13.

59. Ananthakrishnan A, Cagan A, Cai T, Gainer V, Shaw S, Churchill S, et al. Colonoscopy is Associated with a Reduced Risk for Colon Cancer and Mortality in Patients with Inflammatory Bowel Diseases Ashwin. Clin Gastroenterol Hepatol. 2015;13(2):322–9.

60. Flood B, Oficjalska K, Laukens D, Fay J, O’Grady A, Caiazza F, et al. Altered expression of caspases-4 and -5 during inflammatory bowel disease and colorectal cancer: Diagnostic and therapeutic potential. Clin Exp Immunol. 2015;181(1):39–50.

61. Shaukat A, Mongin SJ, Geisser MS, Lederle FA, Bond JH, Mandel JS, et al. Long-term mortality after screening for colorectal cancer. N Engl J Med. 2013;369(12):1106–14.

62. Moiel D, Thompson J. Early Detection of Colon Cancer-the Kaiser Permanente Northwest 30-Year History: How Do We Measure Success? Is It the Test, the Number of Tests, the Stage, or the Percentage of Screen-Detected Patients? Perm J. 2011;15(4):30–8.

63. Winawer SJ, Zauber AG, Ho M, O’Brien M, Gottlieb L, Sternberg S, et al. Prevention of colorectal cancer by colonoscopic polypectomy. The National Polyp Study Workgroup. N Engl J Med. 1993;329(27):1977–81.

64. Levin B, Lieberman DA, McFarland B, Smith RA, Brooks D, Andrews KS, et al. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps, 2008: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. CA Cancer J Clin. 2008;58(3):130–60.

65. Mandel JS, Church TR, Ederer F, Bond JH. Colorectal cancer mortality: effectiveness of biennial screening for fecal occult blood. J Natl Cancer Inst. 1999;91(5):434–7.

66. Mandel JS, Church TR, Bond JH, Ederer F, Geisser MS, Mongin SJ, et al. The Effect of Fecal Occult-Blood Screening on the Incidence of Colorectal Cancer. N Engl J Med. 2000;343(22):1022–3.

67. Hardcastle JD, Chamberlain JO, Robinson MHE, Moss SM, Satya SA, Balfour TW, et al.

168

Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. Lancet. 1996;348:1472–7.

68. Mandel JS, Bond JH, Church TR, Snover DC, Bradley M, Schuman LM, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med. 1993;328(19):1365–71.

69. Garborg K, Holme, Løberg M, Kalager M, Adami HO, Bretthauer M. Current status of screening for colorectal cancer. Ann Oncol. 2013;24(8):1963–72.

70. Duffy MJ, Van Rossum LGM, Van Turenhout ST, Malminiemi O, Sturgeon C, Lamerz R, et al. Use of faecal markers in screening for colorectal neoplasia: A European group on tumor markers position paper. Int J Cancer. 2011;128(1):3–11.

71. Konrad G. Dietary interventions for fecal occult blood test screening. Can Fam Physician. 2010;56:229–38.

72. Young GP. Population-based screening for colorectal cancer: Australian research and implementation. J Gastroenterol Hepatol. 2009;24 Suppl 3:S33–42.

73. Lieberman DA. Screening for colorectal cancer. N Engl J Med. 2009;361(12):1179–87.

74. Hol L, van Leerdam ME, van Ballegooijen M, van Vuuren AJ, van Dekken H, Reijerink JCIY, et al. Screening for colorectal cancer: randomised trial comparing guaiac-based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut. 2010;59(1):62–8.

75. van Rossum LG, van Rijn AF, Laheij RJ, van Oijen MG, Fockens P, van Krieken HH, et al. Random Comparison of Guaiac and Immunochemical Fecal Occult Blood Tests for Colorectal Cancer in a Screening Population. Gastroenterology. 2008;135(1):82–90.

76. Thornton E, Morrin M, Yee J. Current Status of MR Colonography. Radiographics. 2010;30:201–18.

77. Atkin WS, Edwards R, Kralj-Hans I, Wooldrage K, Hart AR, Northover JM, et al. Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet. 2010;375(9726):1624–33.

78. Schoen RE, Pinsky P, Weissfeld J, Yokochi L, Church T, Laiyemo A, et al. Colorectal-cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med. 2012;366(25):2345–57.

79. Segnan N, Armaroli P, Bonelli L, Risio M, Sciallero S, Zappa M, et al. Once-only sigmoidoscopy in colorectal cancer screening: Follow-up findings of the italian randomized controlled trial - SCORE. J Natl Cancer Inst. 2011;103(17):1310–22.

80. Levin TR, Conell C, Shapiro JA, Chazan SG, Nadel MR, Selby J V. Complications of screening flexible sigmoidoscopy. Gastroenterology. 2002;123(6):1786–92.

169

81. Baxter NN, Goldwasser MA, Paszat LF, Saskin R, Urbach DR. Association of colonoscopy and death from colorectal cacner. Ann Intern Med. 2009;150:1–8.

82. Brenner H, Chang-Claude J, Seiler C, Rickert A, Hoffmeister M. Protection From Colorectal Cancer After Colonoscopy A Population-Based, Case–Control Study. Ann Intern Med. 2011;154:22–30.

83. Kahi CJ, Imperiale TF, Juliar BE, Rex DK. Effect of Screening Colonoscopy on Colorectal Cancer Incidence and Mortality. Clin Gastroenterol Hepatol. 2009;7(7):770–5.

84. Zauber AG, Winawer SJ, O’Brien M, Lansdorp-Vogelaar I, van Ballegooijen M, Hankey B, et al. Colonoscopic Polypectomy and Long-Term Prevention of Colorectal-Cancer Deaths. N Engl J Med. 2012;366(8):687–96.

85. Canadian Task Force on Preventive Health Care. Recommendations on screening for colorectal cancer in primary care. CMAJ. 2016;188(5):1–9.

86. US Preventive Services Task Force. Screening for Colorectal Cancer US Preventive Services Task Force Recommendation Statement. JAMA. 2016;315(23):2564–75.

87. Pullens HJ, Siersema PD. Quality indicators for colonoscopy: Current insights and caveats. World J Gastrointest Endosc. 2014;6(12):571–83.

88. Pox C, Schmiegel W, Classen M. Current status of screening colonoscopy in Europe and in the United States. Endoscopy. 2007;39:168–73.

89. Edge S, Byrd D, Compton C, Fritz A, Greene F, Trotti A. AJCC Cancer Staging Handbook, 7th edition. 7th ed. Edge S, Byrd D, Compton C, Fritz A, Greene F, Trotti A, editors. Springer New York; 2010.

90. Compton CC, Greene FL. The staging of colorectal cancer: 2004 and beyond. CA Cancer J Clin. 2004;54(6):295–308.

91. Hari D, Leung A, Lee J-H, Sim M-S, Vuong B, Chiu C, et al. AJCC-7TH Edition Staging Criteria for Colon Cancer: Do the Complex Modifications Improve Prognostic Assessment? J Am Coll Surg. 2013;217(2):181–90.

92. Siegel R, Desantis C, Jemal A. Colorectal Cancer Statistics, 2014. CA Cancer J Clin. 2014;64(1):104–17.

93. Compton CC. Pathology report in colon cancer: What is prognostically important? Dig Dis. 1999;17(2):67–79.

94. Derwinger K, Kodeda K, Bexe-Lindskog E, Taflin H. Tumour differentiation grade is associated with TNM staging and the risk of node metastasis in colorectal cancer. Acta Oncol. 2010;49(1):57–62.

95. Jensen CD, Corley DA, Quinn VP, Doubeni CA, Zauber AG, Lee JK, et al. Fecal immunochemical test program performance over 4 rounds of annual screening: A

170

retrospective cohort study. Ann Intern Med. 2016;164(7):456–63.

96. Graser A, Melzer A, Lindner E, Nagel D, Herrmann K, Stieber P, et al. Magnetic resonance colonography for the detection of colorectal neoplasia in asymptomatic adults. Gastroenterology. 2013;144:743–50.

97. Plumb AA, Halligan S, Pends?? DA, Taylor SA, Mallett S. Sensitivity and specificity of CT colonography for the detection of colonic neoplasia after positive faecal occult blood testing: Systematic review and meta-analysis. Eur Radiol. 2014;24(5):1049–58.

98. Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz L a, Kinzler KW. Cancer Genome Lanscapes. Science (80- ). 2013;339(6127):1546–58.

99. Lynch M. Rate, molecular spectrum, and consequences of human mutation. Proc Natl Acad Sci U S A. 2010;107(3):961–8.

100. Loeb LA. A Mutator Phenotype in Cancer. Cancer Res. 2001;61:3230–9.

101. Loeb L. Mutator phenotype may be required for multiple stage carcinogenesis. Cancer Res. 1991;51:3075–9.

102. Sinicrope FA, Rego RL, Halling KC, Foster N, Sargent DJ, La Plant B, et al. Prognostic Impact of Microsatellite Instability and DNA Ploidy in Human Colon Carcinoma Patients. Gastroenterology. 2006;131(3):729–37.

103. Shen L, Toyota M, Kondo Y, Lin E, Zhang L, Guo Y, et al. Integrated genetic and epigenetic analysis identifies three different subclasses of colon cancer. Proc Natl Acad Sci U S A. 2007;104(47):18654–9.

104. Cheng Y-W, Pincas H, Bacolod M, Schemmann G, Giardina S, Huang J, et al. CpG island methylator phenotype associates with low-degree chromosomal abnormalities in colorectal cancer. Clin Cancer Res. 2008;14(19):6005–13.

105. Levine AJ, Phipps AI, Baron JA, Buchanan DD, Ahnen DJ, Cohen SA, et al. Clinicopathologic risk factor distributions for MLH1 promoter region methylation in CIMP-positive tumors. Cancer Epidemiol Biomarkers Prev. 2016;25(1):68–75.

106. Walther A, Johnstone EC, Swanton C, Tomlinson IPM, Midgley R, Kerr D. Genetic prognostic and predictive markers in colorectal cancer. Nat Rev Cancer. 2009;9(7):489–99.

107. Lengauer C, KW K, Vogelstein B. Genetic instability in colorectal cancers. Vol. 386, Nature. 1997. p. 623–7.

108. Pino MS, Chung DC. The Chromosomal Instability Pathway in Colon. 2010;138(6):2059–72.

109. Ogino S, Ogino S, Goel A, Goel A. Molecular Classification and Correlates in Colorectal Cancer. J Mol Diagnostics. 2010;10(1):13–27.

171

110. Roberts DM, Pronobis MI, Poulton JS, Kane EG, Peifer M. Regulation of Wnt signaling by the tumor suppressor adenomatous polyposis coli does not require the ability to enter the nucleus or a particular cytoplasmic localization. Mol Biol Cell. 2012;23(11):2041–56.

111. Grady WM. Genomic instability and colon cancer. Cancer Metastasis Rev. 2004;23(1–2):11–27.

112. Jankowski JA, Bruton R, Shepherd N, Sanders DS. Cadherin and catenin biology represent a global mechanism for epithelial cancer progression. Mol Pathol. 1997;50(6):289–90.

113. Humphries A, Wright NA. Colonic crypt organization and tumorigenesis. Nat Rev Cancer. 2008;8(6):415–24.

114. Willert K, Jones KA. Wnt signaling: is the party in the nucleus? Genes Dev. 2006;20(11):1394–404.

115. Rubinfeld B, Souza B, Albert I, Muller O, Chamberlain SH, Masiarz FR. Association of the APC gene product with beta-catenin. Science (80- ). 1993;262(5140):1731–4.

116. Su L, Vogelstein B, Kinzler K. Association of the APC tumor suppressor protein with catenins. Science (80- ). 1993;262(5140):1734–7.

117. Gregorieff A, Clevers H. Wnt signaling in the intestinal epithelium+: from endoderm to cancer Wnt signaling in the intestinal epithelium+: from endoderm to cancer. Genes Dev. 2005;19:877–90.

118. The Cancer Genome Atlas Research Network. Comprehensive molecular characterization of human colon and rectal cancer. Nature. 2012;487(7407):330–7.

119. Rubinfeld B, Robbins P, El-Gamil M, Albert I, Porfiri E, Polakis P. Stabilization of beta-catenin by genetic defects in melanoma cell lines. Science. 1997;275(5307):1790–2.

120. Segditsas S, Tomlinson I. Colorectal cancer and genetic alterations in the Wnt pathway. Oncogene. 2006;25(57):7531–7.

121. Su L, Burrell M, Hill DE, Gyuris J, Brent R, Wiltshire R, et al. APC Binds to the Novel Protein EB1. Cancer Res. 1995;55:2972–7.

122. Fodde R, Kuipers J, Rosenberg C, Smits R, Kielman M, Gaspar C, et al. Mutations in the APC tumour suppressor gene cause chromosomal instability. Nat Cell Biol. 2001;3(4):433–8.

123. Mogensen MM, Tucker JB, Mackie JB, Prescott AR, Näthke IS. The adenomatous polyposis coli protein unambiguously localizes to microtubule plus ends and is involved in establishing parallel arrays of microtubule bundles in highly polarized epithelial cells. J Cell Biol. 2002;157(6):1041–8.

124. Jallepalli P V, Lengauer C. Chromosome segregation and cancer: cutting through the

172

mystery. Nat Rev Cancer. 2001;1(2):109–17.

125. Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson J, Markowitz SD, et al. Mutations of mitotic checkpoint genes in human cancers. Nature. 1998;392(6673):300–3.

126. Abal M, Obrador-Hevia A, Janssen KP, Casadome L, Menendez M, Carpentier S, et al. APC Inactivation Associates With Abnormal Mitosis Completion and Concomitant BUB1B/MAD2L1 Up-Regulation. Gastroenterology. 2007;132(7):2448–58.

127. Peters J. The Anaphase-Promoting Complex-Proteolysis in Mitosis and Beyond. Mol Cell. 2002;9:931–43.

128. Yu H, Peters JM, King RW, Page A, Hieter P, Kirschner MW. Identification of a cullin homology region in a subunit of the anaphase-promoting complex. Science. 1998;279(5354):1219–22.

129. Nasmyth K. Segregating sister genomes: the molecular biology of chromosome separation. Science. 2002;297(5581):559–65.

130. Nasmyth K. Splitting the Chromosome: Cutting the Ties That Bind Sister Chromatids. Science (80- ). 2000;288(5470):1379–84.

131. Jallepalli P V, Waizenegger IC, Bunz F, Langer S, Speicher MR, Peters J, et al. Securin Is Required for Chromosomal Stability in Human Cells. 2001;105:445–57.

132. Herz C, Schlurmann F, Batarello D, Fichter CD, Schopflin A, Munch C, et al. Occurrence of Aurora A positive multipolar mitoses in distinct molecular classes of colorectal carcinomas and effect of Aurora A inhibition. Mol Carcinog. 2012;51(9):696–710.

133. Takahashi T, Sano B, Nagata T, Kato H, Sugiyama Y, Kunieda K, et al. Polo-like kinase 1 (PLK1) is overexpressed in primary colorectal cancers. Cancer Sci. 2003;94(2):148–52.

134. Ganem NJ, Godinho S, Pellman D. A Mechanism Linking Extra Centrosomes to Chromosomal Instability. Nature. 2009;460(7252):278–82.

135. Chuang T, Wang J, Jao S, Wu C. Colorectal Adenoma To Carcinoma Progression. Oncotarget. 2016;7(29):45803–18.

136. Bischoff JR, al. et. A homologue of Drosophila is oncogenic and amplified in human colorectal cancers. Embo J. 1998;17(11):3052–65.

137. Zhou H, Kuang J, Zhong L, Kuo W-L, Gray J, Sahin A, et al. Tumour amplified kinase STK15/BTAK induces centrosome amplification, aneuploidy and transformation. Nat Genet. 1998;20:189–93.

138. Basto R, Brunk K, Vinadogrova T, Peel N, Khodjakov A, Raff JW. Centrosome Amplification Can Initiate Tumorigenesis in Flies. Cell. 2008;133(6):1032–42.

139. Hoeijmakers J. Genome maintenance mechanisms for preventing cancer. Nature.

173

2001;411:366–74.

140. Jafri MA, Ansari SA, Alqahtani MH, Shay JW. Roles of telomeres and telomerase in cancer, and advances in telomerase-targeted therapies. Genome Med. 2016;8(1):69.

141. Blackburn EH, Greider CW, Henderson E, Lee MS, Shampay J, Shippen-Lentz D. Recognition and elongation of telomeres by telomerase. Genome. 1989;31(2):553–60.

142. Wright W, Piatyszek M, Rainey W, Byrd W, Shay J. Telomerase activity in human germline and embryonic tissues and cells. Dev Genet. 1996;18(2):173–9.

143. Castro-Vega LJ, Jouravleva K, Ortiz-Montero P, Liu WY, Galeano JL, Romero M, et al. The senescent microenvironment promotes the emergence of heterogeneous cancer stem-like cells. Carcinogenesis. 2015;36(10):1180–92.

144. Shay JW, Wright WE. Role of telomeres and telomerase in cancer. Semin Cancer Biol. 2012;21(6):349–53.

145. Bryan TM, Marusic L, Bacchetti S, Namba M, Reddel RR. The telomere lengthening mechanism in telomerase-negative immortal human cells does not involve the telomerase RNA subunit. Hum Mol Genet. 1997;6(6):921–6.

146. Kim NW, Piatyszek MA, Prowse KR, Harley CB, West MD, Ho PL, et al. Specific association of human telomerase activity with immortal cells and cancer. Science. 1994;266(5193):2011–5.

147. Suraweera N, Mouradov D, Li S, Jorissen RN, Hampson D, Ghosh A, et al. Relative telomere lengths in tumor and normal mucosa are related to disease progression and chromosome instability profiles in colorectal cancer. Oncotarget. 2016;7(24).

148. Rudolph K, Millard M, Bosenberg M, DePinho R. Telomere dysfunction and evolution of intestinal carcinoma and humans. Nat Genet. 2001;28:155–9.

149. Engelhardt M, Drullinksy P, Guillem J, Moore M. Telomerase and Telomere Length in the Development and Progression of Premalignant Lesions to Colorectal Cancer. Clin Cancer Res. 1997;3:1931–41.

150. Roger L, Jones RE, Heppel NH, Williams GT, Sampson JR, Baird DM. Extensive telomere erosion in the initiation of colorectal adenomas and its association with chromosomal instability. J Natl Cancer Inst. 2013;105(16):1202–11.

151. Gertler R, Rosenberg R, Stricker D, Friederichs J, Hoos A, Werner M, et al. Telomere length and human telomerase reverse transcriptase expression as markers for progression and prognosis of colorectal carcinoma. J Clin Oncol. 2004;22(10):1807–14.

152. Meeker AK, Hicks JL, Iacobuzio-donahue CA, Montgomery EA, Westra WH, Chan TY, et al. Telomere Length Abnormalities Occur Early in the Initiation of Epithelial Carcinogenesis Telomere Length Abnormalities Occur Early in the Initiation of Epithelial Carcinogenesis. 2004;10(410):3317–26.

174

153. Nyberg K a, Michelson RJ, Putnam CW, Weinert T a. Toward maintaining the genome: DNA damage and replication checkpoints. Annu Rev Genet. 2002;36:617–56.

154. Rabenau K, Hofstatter E. DNA Damage Repair and the Emerging Role of Poly(ADP-ribose) Polymerase Inhibition in Cancer Therapeutics. Clin Ther. 2016;38(7):1577–88.

155. Rocha JC, Busatto FF, Guecheva TN, Saffi J. Role of nucleotide excision repair proteins in response to DNA damage induced by topoisomerase II inhibitors. Mutat Res Mutat Res. 2016;768:68–77.

156. Li XL, Zhou J, Chen ZR, Chng WJ. P53 mutations in colorectal cancer - molecular pathogenesis and pharmacological reactivation. World J Gastroenterol. 2015;21(1):84–93.

157. Müller MF, Ibrahim A, Arends MJ. Molecular pathological classification of colorectal cancer. Virchows Arch. 2016;469:125–34.

158. Liu G, Parant JM, Lang G, Chau P, Chavez-Reyes A, El-Naggar AK, et al. Chromosome stability, in the absence of apoptosis, is critical for suppression of tumorigenesis in Trp53 mutant mice. Nat Genet. 2004;36(1):63–8.

159. Kolodner RD, Putnam CD, Myung K. Maintenance of genome stability in Saccharomyces cerevisiae. Science. 2002;297(5581):552–7.

160. Staal FJT, Luis TC, Tiemessen MM. WNT signalling in the immune system: WNT is spreading its wings. Nat Rev Immunol. 2008;8(8):581–93.

161. Takebe N, Miele L, Harris PJ, Jeong W, Bando H, Yang SX, et al. Targeting Nothc, Hedgehog, and Wnt Pathways in cancer stem cells: clinical update. Nat Rev Clin Oncol. 2015;12(8):445–64.

162. Thibodeau SN, French AJ, Cunningham JM, Tester D, Burgart LJ, Roche PC, et al. Microsatellite Instability in Colorectal Cancer: Different Mutator Phenotypes and the Principal Involvement of hMLH1. Cancer Res. 1998;58:1713–8.

163. Boland C, Thibodeau S, Hamilton S, Sidranksy D, Eshleman J, Burt R, et al. A National Cancer Institute Workshop on Microsatellite Instability for Cancer Detection and Familial Predisposition: Development of International Criteria for the Determination of Microsatellite Instability in Colorectal Cancer. Cancer Res. 1998;58(22):5248–57.

164. Ionov Y, Peinado MA, Malkhosyan S, Shibata D, Perucho M. Ubiquitous somatic mutations in simple repeated sequences reveal a new mechanism for colonic carcinogenesis. Nat Lett. 1993;363:558–61.

165. Boland CR, Goel A. Microsatellite Instability in Colorectal Cancer. Gastroenterology. 2010;138(6):2073–87.

166. Bhattacharyya NP, Skandalis A, Ganesh A, Groden J, Meuth M. Mutator phenotypes in human colorectal carcinoma cell lines. Proc Natl Acad Sci U S A. 1994;91(14):6319–23.

175

167. Parsons R, Li GM, Longley MJ, Fang W, Papadopoulos N, Jen J, et al. Hypermutability and mismatch repair deficiency in RER+ tumor cells. Cell. 1993;75(6):1227–36.

168. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, Goldman H, et al. Methylation of the hMLH1 promoter correlates with lack of expression of hMLH1 in sporadic colon tumors and mismatch repair-defective human tumor cell lines. Cancer Res. 1997;57(5):808–11.

169. Alexander J, Watanabe T, Wu TT, Rashid A, Li S, Hamilton SR. Histopathological identification of colon cancer with microsatellite instability. Am J Pathol. 2001;158(2):527–35.

170. Jenkins MA, Hayashi S, Anne-marie O, Burgart LJ, Tom C, Shimizu D, et al. Pathology Features in Bethesda Guidelines Predict Colorectal Cancer Microsatellite Instability: A Population-Based Study Mark. Gastroenterology. 2007;133(1):48–56.

171. Li X, Yao X, Wang Y, Hu F, Wang F, Jiang L, et al. MLH1 Promoter Methylation Frequency in Colorectal Cancer Patients and Related Clinicopathological and Molecular Features. 2013;8(3).

172. Popat S, Hubner RA, Houlston RS. Systematic Review of Microsatellite Instability and Colorectal Cancer Prognosis. J Clin Oncol. 2005;23(3):609–18.

173. Hong SP, Min BS, Kim T Il, Cheon JH, Kim NK, Kim H, et al. The differential impact of microsatellite instability as a marker of prognosis and tumour response between colon cancer and rectal cancer. Eur J Cancer. 2012;48(8):1235–43.

174. Merok MA, Ahlquist T, Røyrvik EC, Tufteland KF, Hektoen M, Sjo OH, et al. Microsatellite instability has a positive prognostic impact on stage II colorectal cancer after complete resection: Results from a large, consecutive norwegian series. Ann Oncol. 2013;24(5):1274–82.

175. Duval A, Hamelin R. Mutations at Coding Repeat Sequences in Mismatch Repair-deficient Human Cancers: Toward a New Concept of Target Genes for Instability. Cancer Res. 2002;62(9):2447–54.

176. Pemberton TJ, DeGiorgio M, Rosenberg NA. Population structure in a comprehensive genomic data set on human microsatellite variation. G3. 2013;3(5):891–907.

177. Pyatt R, Chadwick RB, Johnson CK, Adebamowo C, de la Chapelle A, Prior TW. Polymorphic variation at the BAT-25 and BAT-26 loci in individuals of African origin. Implications for microsatellite instability testing. Am J Pathol. 1999;155(2):349–53.

178. Umar A, Boland R, Terdiman J, Syngal S, de la Chapelle A, Ruschoff J, et al. Revised Bethesda Guidelines for Hereditary Nonpolyposis Colorectal Cancer (Lynch Syndrome) and Microsatellite Instability. J Natl Cancer Inst. 2004;96(4):261–8.

179. Tomlinson I, Halford S, Aaltonen L, Hawkins N, Ward R. Does MSI-low exist? J Pathol. 2002;197(1):6–13.

176

180. Mori Y, Selaru FM, Sato F, Yin J, Simms LA, Xu Y, et al. The impact of microsatellite instability on the molecular phenotype of colorectal tumors. Cancer Res. 2003;63(15):4577–82.

181. Ak S, Tunca B, Yilmazlar T, Tezcan G, Cecener G, Egeli U, et al. Microsatellite instability status affects gene expression profiles in early onset colorectal cancer patients. J Surg Res. 2013;185(2):626–37.

182. Iglesias D, Fernández-Peralta AM, Nejda N, Daimiel L, Azcoita MM, Oliart S, et al. RIS1, a gene with trinucleotide repeats, is a target in the mutator pathway of colorectal carcinogenesis. Cancer Genet Cytogenet. 2006;167(2):138–44.

183. Kohonen-Corish MRJ, Daniel JJ, Chan C, Lin BPC, Kwun SY, Dent OF, et al. Low microsatellite instability is associated with poor prognosis in stage C colon cancer. J Clin Oncol. 2005;23(10):2318–24.

184. Kim TM, Laird PW, Park PJ. The landscape of microsatellite instability in colorectal and endometrial cancer genomes. Cell. 2013;155(4):858–68.

185. Longley D, Harkin D, Johnston P. 5-fluorouracil: mechanisms of action and clinical strategies. Nat Rev Cancer. 2003;3(5):330–8.

186. Wilson PM, Danenberg P V, Johnston PG, Lenz H-J, Ladner RD. Standing the test of time: targeting thymidylate biosynthesis in cancer therapy. Nat Rev Clin Oncol. 2014;11(5):282–98.

187. Carethers JM, Chauhan D, Fink D, Nebel S, Bresalier R, Howell S, et al. Mismatch repair proficiency and in vitro response to 5-fluorouracil. Gastroenterology. 1999;117(1):123–31.

188. Ribic C, Sargent DJ, Moore M, Thibodeau SN, French AJ, Goldberg RM, et al. Tumor Microsatellite-Instability Status as a Predictor of Benefit from Fluorouracil-Based Adjuvant Chemotherapy for Colon Cancer. New English J Med. 2003;349(3):247–57.

189. Carethers JM, Smith EJ, Behling CA, Nguyen L, Tajima A, Doctolero RT, et al. Use of 5-Fluorouracil and Survival in Patients with Microsatellite-Unstable Colorectal Cancer. Gastroenterology. 2004;126(2):394–401.

190. Jover R, Zapater P, Castells A, Llor X, Andreu M, Cubiella J, et al. The efficacy of adjuvant chemotherapy with 5-fluorouracil in colorectal cancer depends on the mismatch repair status. Eur J Cancer. 2009;45(3):365–73.

191. Bertagnolli MM, Niedzwiecki D, Compton CC, Hahn HP, Hall M, Damas B, et al. Microsatellite instability predicts improved response to adjuvant therapy with irinotecan, fluorouracil, and leucovorin in stage III colon cancer: Cancer and leukemia group B protocol 89803. J Clin Oncol. 2009;27(11):1814–21.

192. Le DT, Uram JN, Wang H, Bartlett BR, Kemberling H, Eyring AD, et al. PD-1 Blockade in Tumors with Mismatch-Repair Deficiency. N Engl J Med. 2015;372(26):2509–20.

177

193. Li W, Liu M. Distribution of 5-Hydroxymethylcytosine in Different Human Tissues. J Nucleic Acids. 2011;2011:870726.

194. Allis CD, Jenuwein T. The molecular hallmarks of epigenetic control. Nat Rev Genet. 2016;17:487–500.

195. Van Engeland M, Derks S, Smits KM, Meijer GA, Herman JG. Colorectal cancer epigenetics: Complex simplicity. J Clin Oncol. 2011;29(10):1382–91.

196. Ziller MJ, Muller F, Liao J, Zhang Y, Gu H, Bock C, et al. Genomic distribution and Inter-Sample variation of Non-CpG methylation across human cell types. PLoS Genet. 2011;7(12):e1002389.

197. Varley KE, Gertz J, Bowling KM, Parker SL, Reddy TE, Pauli-Behn F, et al. Dynamic DNA methylation across diverse human cell lines and tissues. Genome Res. 2013;23(3):555–67.

198. Guo J, Su Y, Shin J, Shin J, Li H, Xie B, et al. Distribution, recognition and regulation of non-CpG methylation in the adult mammalian brain. Nat Neurosci. 2014;17(2):215–22.

199. Gardiner-Garden M, Frommer M. CpG Islands in Vertibrate Genomes. J Mol Biol. 1987;196:261–82.

200. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, Issa J-PJ. CpG island methylator phenotype in colorectal cancer: A current perspective. Proc Natl Acad Sci. 1999;96(15):8681–6.

201. H S, Laird P. Interplay between the cancer genome and epigenome. Cell. 2013;153(1):38–55.

202. Ziller MJ, Gu H, Müller F, Donaghey J, Linus T. Charting a dynamic DNA methylation landscape of the human genome. Nature. 2013;500(7463):477–81.

203. Edgar R, Tan PPC, Portales-Casamar E, Pavlidis P. Meta-analysis of human methylomes reveals stably methylated sequences surrounding CpG islands associated with high gene expression. Epigenetics Chromatin. 2014;7(1):28.

204. Stevens M, Cheng JB, Li D, Xie M, Hong C, Maire CL, et al. Estimating Absolute Methylation Levels at Single CpG Resolution from Methylation Enrichment and Restriction Enzyme Sequencing Methods. Genome Res. 2013;23:1541–53.

205. Irizarry R, Ladd-Acosta C, Wen B, Wu Z, Montao C, Onyango P, et al. The human colon cancer methylome shows similar hypo- and hypermethylation at conserved tissue-specific CpG island shores. Nat Genet. 2009;41(2):178–86.

206. Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, et al. High density DNA methylation array with single CpG site resolution. Genomics. 2011;98(4):288–95.

207. Yang X, Han H, DeCarvalho D, Lay F, Jones P, Liang G. Gene body methylation can alter

178

gene expression and is a therapeutic target in cancer. Cancer Cell. 2014;26:577–90.

208. Li F, Karlsson H. Expression and regulation of human endogenous retrovirus W elements. Acta Pathol Microbiol Immunol Scand. 2016;124:52–66.

209. Lee JT, Bartolomei MS. X-Inactivation, Imprinting, and Long Noncoding RNAs in Health and Disease. Cell. 2013;152:1308–23.

210. Park Y. Epigenetic Aspects of X-Chromosome Dosage Compensation. Science (80- ). 2001;293(5532):1083–5.

211. Bell AC, Felsenfeld G. Methylation of a CTCF-dependent boundary controls imprinted expression of the Igf2 gene. Nature. 2000;405(6785):482–5.

212. Soellner L, Begemann M, Mackay DJ, Grønskov K, Tümer Z, Maher ER, et al. Recent Advances in Imprinting Disorders. Clin Genet. 2016;1–11.

213. Feinberg A, Vogelstein B. Hypomethylation distinguishes genes of some human cancers from their normal counterparts. Nature. 1983;301:89–92.

214. Goelz SE, Vogelstein B, Hamilton SR, Feinberg a P. Hypomethylation of DNA from benign and malignant human colon neoplasms. Science. 1985;228(4696):187–90.

215. Heyn H, Li N, Ferreira HHJ, Moran S, Pisano DG, Gomez A, et al. Distinct DNA methylomes of newborns and centenarians. Proc Natl Acad Sci U S A. 2012;109(26):10522–10527.

216. Toyota M, Issa JP. CpG island methylator phenotypes in aging and cancer. Semin Cancer Biol. 1999;9(5):349–57.

217. Teschendorff AE, Menon U, Gentry-Maharaj A, Ramus SJ, Weisenberger DJ, Shen H, et al. Age-dependent DNA methylation of genes that are suppressed in stem cells is a hallmark of cancer. Genome Res. 2010;20(4):440–6.

218. Klutstein M, Nejman D, Greenfield R, Cedar H. DNA Methylation in Cancer and Aging. Cancer Res. 2016;76(12):3446–50.

219. Chan A, Broaddus RR, Houlihan PS, Issa J-PJ, Hamilton SR, Rashid A. CpG island methylation in aberrant crypt foci of the colorectum. Am J Pathol. 2002;160(5):1823–30.

220. Tahiliani M, Koh KP, Shen Y, Pastor W a, Brudno Y, Agarwal S, et al. Conversion of 5-Methylcytosine to 5-Hydroxymethylcytosine in Mammalian DNA by MLL Partner TET1. Science (80- ). 2009;324(5929):930–5.

221. Kamdar SN, Ho LT, Kron KJ, Isserlin R, van der Kwast T, Zlotta AR, et al. Dynamic interplay between locus-specific DNA methylation and hydroxymethylation regulates distinct biological pathways in prostate carcinogenesis. Clin Epigenetics. 2016;8:32.

222. Li M, Gao F, Xia Y, Tang Y, Zhao W, Jin C, et al. Filtrating colorectal cancer associated

179

genes by integrated analyses of global DNA methylation and hydroxymethylation in cancer and normal tissue. Sci Rep. 2016;6:31826.

223. Ito S, Shen L, Dai Q, Wu SC, Collins LB, Swenberg JA, et al. Tet proteins can convert 5-methylcytosine to 5-formylcytosine and 5-carboxylcytosine. Science. 2011;333(6047):1300–3.

224. Venkatesh S, Workman JL. Histone exchange, chromatin structure and the regulation of transcription. Nat Rev Mol Cell Biol. 2015;16(3):178–89.

225. Kornberg RD, Lorch Y. Twenty-five years of the nucleosome, fundmamental particle of the eukaryotic chromosome. Cell. 1999;98:285–94.

226. Richmond TJ, Davey CA. The structure of DNA in the nucleosome core. Nature. 2003;423(6936):145–50.

227. Kouzarides T. Chromatin Modifications and Their Function. Cell. 2007;128(4):693–705.

228. Zhang T, Cooper S, Brockdorff N. The interplay of histone modifications - writers that read. EMBO Rep. 2015;16(11):1467–81.

229. Musselman CA, Lalonde M-E, Côté J, Kutateladze TG. Perceiving the epigenetic landscape through histone readers. Nat Struct Mol Biol. 2012;19(12):1218–27.

230. Palmirotta R, Cives M, Della-Morte D, Capuani B, Lauro D, Guadagni F, et al. Sirtuins and Cancer: Role in the Epithelial-Mesenchymal Transition. Oxid Med Cell Longev. 2016;2016:3031459.

231. Fraga MF, Ballestar E, Villar-Garea A, Boix-Chornet M, Espada J, Schotta G, et al. Loss of acetylation at Lys16 and trimethylation at Lys20 of histone H4 is a common hallmark of human cancer. Nat Genet. 2005;37(4):391–400.

232. Savio AJ, Bapat B. Beyond the Island: Epigenetic biomarkers of colorectal and prostate cancer. In: Cancer Epigenetics: Risk Assessment, Diagnosis, Treatment, and Prognosis. Springer New York; 2014. p. 103–24.

233. Chen T, Dent S. Chromatin modifiers: regulators of cellular differentiation. Nat Rev Genet. 2014;15(2):93–106.

234. Zentner GE, Henikoff S. Regulation of nucleosome dynamics by histone modifications. Nat Struct Mol Biol. 2013;20(3):259–66.

235. Fire A, Xu S, Montgomery M, Kostas S, Driver S, Mello C. Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nat Lett. 1998;391:806–11.

236. He L, Hannon GJ. MicroRNAs: small RNAs with a big role in gene regulation. Nat Rev Genet. 2004;5(7):522–31.

180

237. Cech TR, Steitz JA. The Noncoding RNA Revolution—Trashing Old Rules to Forge New Ones.pdf. Cell. 2014;157(1):77–94.

238. Holoch D, Moazed D. RNA-mediated epigenetic regulation of gene expression. Nat Rev Genet. 2015;16(2):71–84.

239. Weisenberger DJ, Campan M, Long TI, Kim M, Woods C, Fiala E, et al. Analysis of repetitive element DNA methylation by MethyLight. Nucleic Acids Res. 2005;33(21):6823–36.

240. Ogino S, Cantor M, Kawasaki T, Brahmandam M, Kirkner GJ, Weisenberger DJ, et al. CpG island methylator phenotype (CIMP) of colorectal cancer is best characterised by quantitative DNA methylation analysis and prospective cohort studies. Gut. 2006;55(7):1000–6.

241. Hinoue T, Weisenberger DJ, Lange CPE, Shen H, Byun HM, Van Den Berg D, et al. Genome-scale analysis of aberrant DNA methylation in colorectal cancer. Genome Res. 2012;22(2):271–82.

242. Juo YY, Johnston FM, Zhang DY, Juo HH, Wang H, Pappou EP, et al. Prognostic value of CpG island methylator phenotype among colorectal cancer patients: a systematic review and meta-analysis. Ann Oncol. 2014;25(12):2314–27.

243. Okugawa Y, Grady WM, Goel A. Epigenetic Alterations in Colorectal Cancer: Emerging Biomarkers. Gastroenterology. 2015;149(5):1204–25.

244. Phipps AI, Limburg PJ, Baron JA, Burnett-Hartman A, Weisenberger DJ, Laird PW, et al. Association between molecular subtypes of colorectal cancer and patient survival. Gastroenterology. 2015;148(1):77–87.

245. Strate LL, Syngal S. Hereditary colorectal cancer syndromes. Cancer Causes Control. 2005;16(3):201–13.

246. Bonadona V, Bonaiti B, Olschwang S, Grandjouan S, Huiart L, Longy M, et al. Cancer risks associated with germline mutations in MLH1, MSH2, and MSH6 genes in Lynch syndrome. JAMA. 2011;305(22):2304–10.

247. Peltomäki P, Vasen H. Mutations associated with HNPCC predisposition -- Update of ICG-HNPCC/INSiGHT mutation database. Dis Markers. 2004;20(4–5):269–76.

248. Kwok CT, Ward RL, Hawkins NJ, Hitchins MP. Detection of allelic imbalance in MLH1 expression by pyrosequencing serves as a tool for the identification of germline defects in Lynch syndrome. Fam Cancer. 2010;9(3):345–56.

249. Hitchins MP, Rapkins RW, Kwok CT, Srivastava S, Wong JJL, Khachigian LM, et al. Dominantly Inherited Constitutional Epigenetic Silencing of MLH1 in a Cancer-Affected Family Is Linked to a Single Nucleotide Variant within the 5’UTR. Cancer Cell. 2011;20(2):200–13.

181

250. Ligtenberg MJL, Kuiper RP, Geurts Van Kessel A, Hoogerbrugge N. EPCAM deletion carriers constitute a unique subgroup of Lynch syndrome patients. Fam Cancer. 2013;12(2):169–74.

251. Kempers M, Kuiper R, Ockeloen C, Chappuis P, Hutter P, Rahner N, et al. High colorectal and low endometrial cancer risk in EPCAM deletion-positive Lynch syndrome: a cohort study. Lancet Oncol. 2011;12(1):49–55.

252. Lynch HT, Watson P, Lanspa SJ, Marcus J, Smyrk T, Fitzgibbons RJ, et al. Natural history of colorectal cancer in hereditary nonpolyposis colorectal cancer (Lynch syndromes I and II). Dis Colon Rectum. 1988;31(6):439–44.

253. Vasen HF, Mecklin J-P, Watson P, Utsunomiya J, Bertario L, Lynch P, et al. Surveillance in hereditary nonpolyposis colorectal cancer. Dis Colon Rectum. 1993;36(1):1–4.

254. Jass JR, Stewart SM. Evolution of hereditary non-polyposis colorectal cancer. Gut. 1992;33(6):783–6.

255. Lynch HT, Smyrk TC, Watson P, Lanspa S, Lynch J, Lynch PM, et al. Genetics, Natural History, Tumor Spectrum, and Pathology of Hereditary Nonpolyposis Colorectal Cancer. Gastroenterology. 1993;104:1535–49.

256. Vasen HF, Stormorken A, Menko FH, Nagengast FM, Kleibeuker JH, Griffioen G, et al. MSH2 mutation carriers are at higher risk of cancer than MLH1 mutation carriers: a study of hereditary nonpolyposis colorectal cancer families. J Clin Oncol. 2001;19(20):4074–80.

257. Stoffel E, Mukherjee B, Raymond VM, Kastrinos F, Sparr J, Wang F, et al. Calculation of Risk of Colorectal and Endometrial Cancer Among Patients with Lynch Syndrome. Gastroenterology. 2009;137(5):1621–7.

258. Choi Y-H, Cotterchio M, McKeown-Eyssen G, Neerav M, Bapat B, Boyd K, et al. Penetrance of colorectal cancer among MLH1/MSH2 carriers participating in the colorectal cancer familial registry in Ontario. Hered Cancer Clin Pract. 2009;7(1):14.

259. Green J, Driscoll M, Barnes A, Maher ER, Bridge P, Shields K, et al. Impact of gender and parent of origin on the phenotypic expression of hereditary nonpolyposis colorectal cancer in a large Newfoundland kindred with a common MSH2 mutation. Dis Colon Rectum. 2002;45(9):1223–32.

260. Hendriks YMC, Wagner A, Morreau H, Menko F, Stormorken A, Quehenberger F, et al. Cancer risk in hereditary nonpolyposis colorectal cancer due to MSH6 mutations: Impact on counseling and surveillance. Gastroenterology. 2004;127(1):17–25.

261. Douglas JA, Gruber SB, Meister KA, Bonner J, Watson P, Lynch HT. History and molecular genetics of Lynch syndrome in Family G a century later. JAMA. 2005;294(17):2195–202.

262. Vasen H, Mecklin J, Khan P, Lynch H. The International Collaborative Group on Hereditary Non-Polyposis Colorectal Cancer (ICG-HNPCC). Dis Colon Rectum.

182

1991;34(5):424–5.

263. Vasen HFA, Watson P, Mecklin JP, Lynch HT. New Clinical Criteria for Hereditary Nonpolyposis Colorectal Cancer (HNPCC, Lynch Syndrome) Proposed by the International Collaborative Group on HNPCC. Gastroenterology. 1999;116:1453–6.

264. Rodriguez-Bigas M, Boland C, Hamilton S, Henson D, Srivastava S, Jass J, et al. A National Cancer Institute Workshop on Hereditary Nonpolyposis Colorectal Cancer Syndrome: Meeting Highlights and Bethesda Guidelines. J Natl Cancer Inst. 1997;89(23):1758–62.

265. Lindor N, Rabe K, Petersen G, Haile R, Casey G, Baron J, et al. Lower cancer incidence in Amsterdam-I criteria families without mismatch repair deficiency: familial colorectal cancer type X. JAMA. 2005;293(16):1979–85.

266. de Jong MM, Nolte IM, Meerman GJ. Low-penetrance Genes and Their Involvement in Colorectal Cancer Susceptibility. Cancer Epidemiol Biomarkers Prev. 2002;11:1332–52.

267. Webb EL, Rudd MF, Sellick GS, El Galta R, Bethke L, Wood W, et al. Search for low penetrance alleles for colorectal cancer through a scan of 1467 non-synonymous SNPs in 2575 cases and 2707 controls with validation by kin-cohort analysis of 14 704 first-degree relatives. Hum Mol Genet. 2006;15(21):3263–71.

268. de la Chapelle A. Genetic predisposition to colorectal cancer. Nat Rev Cancer. 2004;4(10):769–80.

269. Miyoshi Y, Nagase H, Ando H, Horii A, Ichii S, Nakatsuru S, et al. Somatic mutations of the APC gene in colorectal tumors: mutation cluster region in the APC gene. Hum Mol Genet. 1992;1(4):229–33.

270. Wallace MH, Phillips RK. Preventative strategies for periampullary tumours in FAP. Ann Oncol. 1999;10(Suppl 4):201–3.

271. Björk J, Akerbrant H, Iselius L, Bergman a, Engwall Y, Wahlström J, et al. Periampullary adenomas and adenocarcinomas in familial adenomatous polyposis: cumulative risks and APC gene mutations. Gastroenterology. 2001;121(5):1127–35.

272. Galle T, Juel K, Bülow S. Causes of Death in Familial Adenomatous Polyposis. Scand J Gastroenterol. 1999;34(8):808–12.

273. Chen CS, Phillips KD, Grist S, Bennet G, Craig JE, Muecke JS, et al. Congenital hypertrophy of the retinal pigment epithelium (CHRPE) in familial colorectal cancer. Fam Cancer. 2006;5(4):397–404.

274. Nusliha A, Dalpatadu U, Amarasinghe B, Chandrasinghe P, Deen K. Congenital hypertrophy of retinal pigment epithelium (CHRPE) in patients with familial adenomatous polyposis (FAP); a polyposis registry experience. BMC Res Notes. 2014;7(1):734.

275. Knudsen AL, Bülow S, Tomlinson I, Möslein G, Heinimann K, Christensen IJ. Attenuated

183

familial adenomatous polyposis: results from an international collaborative study. Colorectal Dis. 2010;12:e243–9.

276. Kobayashi H, Ishida H, Ueno H, Hinoi T, Inoue Y, Ishida F, et al. Association between the age and the development of colorectal cancer in patients with familial adenomatous polyposis: a multi-institutional study. Surg Today. 2016;doi:10.1007/s00595-016-1398-1.

277. Aretz S, Vasen HF a, Olschwang S. Clinical utility gene card for: familial adenomatous polyposis (FAP) and attenuated FAP (AFAP). Eur J Hum Genet. 2011;19(7):1–4.

278. Juhn E, Khachemoune A. Gardner syndrome: skin manifestations, differential diagnosis and management. Am J Clin Dermatol. 2010;11(2):117–22.

279. Hamilton S, Liu B, Parsons R, Papadopoulos N, Jen J, Powell S, et al. The molecular basis of Turcot’s syndrome. N Engl J Med. 1995;332(13):839–47.

280. David SS, Shea VLO, Kundu S. Base Excision Repair of Oxidative DNA Damage. Nature. 2010;447(7147):941–50.

281. Wood R, Mitchell M, Sgouros J, Lindahl T. Human DNA Repair Genes. Science (80- ). 2001;291(5507):1284–9.

282. Dizdaroglu M. Oxidatively induced DNA damage: Mechanisms, repair and disease. Cancer Lett. 2012;327(1–2):26–47.

283. Aretz S, Hes FJ. Clinical utility gene card for: MUTYH-associated polyposis (MAP), autosomal recessive colorectal adenomatous polyposis. Eur J Hum Genet. 2010;18(9):0–4.

284. Al-Tassan NA, Whiffin N, Hosking FJ, Palles C, Farrington SM, Dobbins SE, et al. A new GWAS and meta-analysis with 1000Genomes imputation identifies novel risk variants for colorectal cancer. Sci Rep. 2015;5:10442.

285. Bolocan a, Ion D, Stoian R V, Serban MB. Map syndrome (MYH Associated Polyposis) colorectal cancer, etiopathological connections. J Med Life. 2011;4(1):109–11.

286. Win AK, Reece JC, Dowty JG, Buchanan DD, Clendenning M, Rosty C, et al. Risk of extracolonic cancers for people with biallelic and monoallelic mutations in MUTYH. Int J cancer. 2016;1563:1557–63.

287. Vogt S, Jones N, Christian D, Engel C, Nielsen M, Kaufmann A, et al. Expanded Extracolonic Tumor Spectrum in MUTYH-Associated Polyposis. Gastroenterology. 2009;137(6):1976–85.

288. Jelsig AM, Brusgaard K, Hansen TP, Qvist N, Larsen M, Bojesen A, et al. Germline variants in Hamartomatous Polyposis Syndrome-associated genes from patients with one or few hamartomatous polyps. Scand J Gastroenterol. 2016;51(9):1118–25.

289. Giardiello FM, Brensinger JD, Tersmette AC, Goodman SN, Petersen GM, Booker S V, et al. Very high risk of cancer in familial Peutz-Jeghers syndrome. Gastroenterology.

184

2000;119(6):1447–53.

290. Boardman LA, Thibodeau SN, Schaid DJ, Lindor NM, McDonnell SK, Burgart LJ, et al. Increased risk for cancer in patients with the Peutz-Jeghers syndrome. Ann Intern Med. 1998;128(11):896–9.

291. Hemminki A, Markie D, Tomlinson I, Avizienyte E, Roth S, Loukola A, et al. A serine/threonine kinase gene defective in Peutz-Jeghers syndrome. Nature. 1998;391(6663):184–7.

292. Brosens LAA, Langeveld D, van Hattem WA, Giardiello FM, Offerhaus GJA. Juvenile polyposis syndrome. World J Gastroenterol. 2011;17(44):4839–44.

293. Jansen M, de Leng WWJ, Baas AF, Myoshi H, Mathus-Vliegen L, Taketo MM, et al. Mucosal prolapse in the pathogenesis of Peutz-Jeghers polyposis. Gut. 2006;55(1):1–5.

294. Coburn MC, Pricolo VE, DeLuca FG, Bland KI. Malignant potential in intestinal juvenile polyposis syndromes. Ann Surg Oncol. 1995;2(5):386–91.

295. Howe JR. Mutations in the SMAD4/DPC4 gene in juvenile polyps. Science (80- ). 1998;280(5366):15–9.

296. Calva-Cerqueira D, Chinnathambi S, Pechman B, Bair J, Larsen-Haidle J, Howe JR. The rate of germline mutations and large deletions of SMAD4 and BMPR1A in juvenile polyposis. Clin Genet. 2009;75(1):79–85.

297. Liaw D, Marsh D, Li J, Dahia P, Wang S, Zheng Z, et al. Germline mutations of the PTEN gene in Cowden disease, an inherited breast and thyroid cancer syndrome. Nat Genet. 1997;16:64–7.

298. Watson J, Crick F. Genetical implications of the structure of deoxyribonucleic acid. Nature. 1953;171(4361):964–7.

299. Johnson R, Klassen R, Prakash L, Prakash S. A major role of DNA polymerase δ in replication of both the leading and lagging DNA strands. Mol Cell. 2015;59(2):163–75.

300. Yurieva O, O’Donnell M. Reconstitution of a eukaryotic replisome reveals the mechanism of asymmetric distribution of DNA polymerases. Nucleus. 2016;7(4):360–8.

301. McCulloch S, Kunkel T. The fidelity of DNA synthesis by eukaryotic replicative and translesion synthesis polymerases. Cell Res. 2008;18(1):148–61.

302. Miyabe I, Kunkel TA, Carr AM. The major roles of DNA polymerases epsilon and delta at the eukaryotic replication fork are evolutionarily conserved. PLoS Genet. 2011;7(12).

303. Iyer RR, Pluciennik A, Burdett V, Modrich PL. DNA mismatch repair: functions and mechanisms. Chem Rev. 2006;106(2):302–23.

304. Kunz C, Saito Y, Schär P. Mismatched repair: Variations on a theme. Cell Mol Life Sci.

185

2009;66(6):1021–38.

305. Altieri F, Grillo C, Maceroni M, Chichiarelli S. DNA damage and repair: from molecular mechanisms to health implications. Antioxid Redox Signal. 2008;10(5):891–937.

306. Freudenthal B, Beard W, Perera L, Shock D, Kim T, Schlick T, et al. Uncovering the Polymerase-induced Cytotoxicty of an Oxidized Nucleotide. Nature. 2015;517(7536):635–9.

307. Kouchakdjian M, Bodepudi V, Shibutani S, Eisenberg M, Johnson F, Grollman AP, et al. NMR structural studies of the ionizing radiation adduct 7-hydro-8-oxodeoxyguanosine (8-oxo-7H-dG) opposite deoxyadenosine in a DNA duplex. 8-Oxo-7H-dG(syn).dA(anti) alignment at lesion site. Biochemistry. 1991;30(5):1403–12.

308. Lipscomb LA, Peek ME, Morningstar ML, Verghis SM, Miller EM, Rich A, et al. X-ray structure of a DNA decamer containing 7,8-dihydro-8- oxoguanine. Proc Natl Acad Sci U S A. 1995;92:719–23.

309. Trantakis IA, Nilforoushan A, Dahlmann HA, Stäuble CK, Sturla SJ. In-Gene Quantification of O 6 -Methylguanine with Elongated Nucleoside Analogues on Gold Nanoprobes. J Am Chem Soc. 2016;138:8497–504.

310. Eadie J, Conrad M, Toorchen D, Topal M. Mechanism of mutagenesis by O6-methylguanine. Nature. 1984;308(5955):201–3.

311. Kaina B, Christmann M, Naumann S, Roos WP. MGMT: Key node in the battle against genotoxicity, carcinogenicity and apoptosis induced by alkylating agents. DNA Repair (Amst). 2007;6(8):1079–99.

312. Stojic L, Brun R, Jiricny J. Mismatch repair and DNA damage signalling. DNA Repair (Amst). 2004;3(8–9):1091–101.

313. Baptiste B, Jacob K, Eckert K. Genetic Evidence That Both dNTP-Stabilized and Strand Slippage Mechanisms May Dictate DNA Polymerase Errors Within Mononucleotide Microsatellites. DNA Repair. 2015;29:91–100.

314. Bebenek K, Kunkel TA. Frameshift errors initiated by nucleotide misincorporation. Proc Natl Acad Sci U S A. 1990;87(13):4946–50.

315. An Q, Robins P, Lindahl T, Barnes DE. 5-Fluorouracil incorporated into DNA is excised by the Smug1 DNA glycosylase to reduce drug cytotoxicity. Cancer Res. 2007;67(3):940–5.

316. Wyatt MD, Wilson D. Participation of DNA repair in the response to 5-fluorouracil. Cell Mol Life Sci. 2009;66(5):788–99.

317. Schofield MJ, Hsieh P. DNA mismatch repair: molecular mechanisms and biological function. Annu Rev Microbiol. 2003;57:579–608.

186

318. Jiricny J. Postreplicative mismatch repair. Cold Spring Harb Perspect Biol. 2013;5(4):1–23.

319. Groothuizen FS, Sixma TK. The conserved molecular machinery in DNA mismatch repair enzyme structures. DNA Repair (Amst). 2016;38:14–23.

320. Sameer AS, Nissar S, Fatima K. Mismatch repair pathway. Eur J Cancer Prev. 2014;23(4):246–57.

321. Treffers HP, Spinelli V, Belser N. A factor (or mutator gene) influencing mutation rates in Escherichia coli. Proc Natl Acad Sci. 1954;40(11):1064–71.

322. Siegel EC, Bryson V. Selection of Resistant Strains of Escherichia Coli by Antibiotics and Antibacterial Agents: Role of Normal and Mutator Strains. Antimicrob Agents Chemother. 1963;161:629–34.

323. Marinus MG. Adenine methylation of Okazaki fragments in Escherichia coli. J Bacteriol. 1976;128(3):853–4.

324. Geier GE, Modrich P. Recognition sequence of the dam methylase of Escherichia coli K12 and mode of cleavage of Dpn 1 endonuclease. J Biol Chem. 1979;254(4):1408–13.

325. Jiricny J. The multifaceted mismatch-repair system. Nat Rev Mol Cell Biol. 2006;7(5):335–46.

326. Josephs EA, Zheng T, Marszalek PE. Atomic force microscopy captures the initiation of methyl-directed DNA mismatch repair. DNA Repair (Amst). 2015;35:71–84.

327. Glickman B. Spontaneous mutagenesis in Escherichia coli strains lacking 6-methyladenine residues in their DNA. Mutat Res. 1979;61:153–62.

328. Herman GE, Modrich P. Escherichia coli K-12 clones that overproduce dam methylase are hypermutable. J Bacteriol. 1981;145(1):644–6.

329. Tan C, Terakawa T, Takada S. Dynamic Coupling among Protein Binding, Sliding, and DNA Bending Revealed by Molecular Dynamics. J Am Chem Soc. 2016;138:8512–22.

330. Qiu R, DeRocco VC, Harris C, Sharma A, Hingorani MM, Erie D a, et al. Large conformational changes in MutS during DNA scanning, mismatch recognition and repair signalling. EMBO J. 2012;31(11):2528–40.

331. Erie DA, Weninger KR. Single molecule Studies of DNA Mismatch Repair. DNA Repair (Amst). 2014;20(919):71–81.

332. Spampinato C, Modrich P. The MutL ATPase is required for mismatch repair. J Biol Chem. 2000;275(13):9863–9.

333. Galio L, Bouquet C, Brooks P. ATP hydrolysis-dependent formation of a dynamic ternary nucleoprotein complex with MutS and MutL. Nucleic Acids Res. 1999;27(11):2325–31.

187

334. Bruni R, Martin D, Jiricny J. d(GATC) sequences influence Escherichia coli mismatch repair in a distance-dependent manner from positions both upstream and downstream of the mismatch. Nucleic Acids Res. 1988;16(11):4875–90.

335. Welsh KM, Lu AL, Clark S, Modrich P. Isolation and characterization of the Escherichia coli mutH gene product. J Biol Chem. 1987;262(32):15624–9.

336. Yamaguchi M, Dao V, Modrich P. MutS and MutL activate DNA helicase II in a mismatch-dependent manner. J Biol Chem. 1998;273(15):9197–201.

337. Fishel R. Mismatch repair. J Biol Chem. 2015;290(44):26395–403.

338. Meyer RR, Laine PS. The Single-Stranded DNA-Binding Protein of Escherichia coli. Microbiol Rev. 1990;54(4):342–80.

339. Cooper DL, Lahue RS, Modrich P. Methyl-directed mismatch repair is bidirectional. J Biol Chem. 1993;268(16):11823–9.

340. Grilley M, Griffith J, Modrich P. Bidirectional excision in methyl-directed mismatch repair. J Biol Chem. 1993;268(16):11830–7.

341. Burdett V, Baitinger C, Viswanathan M, Lovett ST, Modrich P. In vivo requirement for RecJ, ExoVII, ExoI, and ExoX in methyl-directed mismatch repair. Proc Natl Acad Sci U S A. 2001;98(12):6765–70.

342. Uphoff S, Reyes-Lamothe R, Garza de Leon F, Sherratt DJ, Kapanidis AN. Single-molecule DNA repair in live bacteria. Proc Natl Acad Sci U S A. 2013;110(20):8063–8.

343. Pluciennik A, Burdett V, Lukianova O, O’Donnell M, Modrich P. Involvement of the beta clamp in methyl-directed mismatch repair in vitro. J Biol Chem. 2009;284(47):32782–91.

344. Monti MR, Miguel V, Borgogno M V, Argaraña CE. Functional analysis of the interaction between the mismatch repair protein MutS and the replication processivity factor β clamp in Pseudomonas aeruginosa. DNA Repair (Amst). 2012;11(5):463–9.

345. Kunkel TA, Erie DA. Eukaryotic Mismatch Repair in Relation to DNA Replication. Annu Rev Genet. 2015;49(1):291–313.

346. Harfe BD, Minesinger BK, Jinks-Robertson S. Discrete in vivo roles for the MutL homologs Mlh2p and Mlh3p in the removal of frameshift intermediates in budding yeast. Curr Biol. 2000;10(3):145–8.

347. McCulloch SD, Gu L, Li GM. Bi-directional processing of DNA loops by mismatch repair-dependent and -independent pathways in human cells. J Biol Chem. 2003;278(6):3891–6.

348. Räschle M, Marra G, Nyström-Lahti M, Schär P, Jiricny J. Identification of hMutLβ, a heterodimer of hMLH1 and hPMS1. J Biol Chem. 1999;274(45):32368–75.

188

349. Campbell CS, Hombauer H, Srivatsan A, Bowen N, Gries K, Desai A, et al. Mlh2 Is an Accessory Factor for DNA Mismatch Repair in Saccharomyces cerevisiae. PLoS Genet. 2014;10(5):13–5.

350. Marzahn M, Hayner J, Meyer J, Bloom L. Kinetic analysis of PCNA clamp binding and release in the clamp loading reaction catalyzed by Saccharomyces cerevisiae replication factor C. Biochim Biophys Acta. 2015;1854(1):31–8.

351. Shiomi Y, Usukura J, Masamura Y, Takeyasu K, Nakayama Y, Obuse C, et al. ATP-dependent structural change of the eukaryotic clamp-loader protein, replication factor C. Proc Natl Acad Sci U S A. 2000;97(26):14127–32.

352. Pluciennik A, Dzantiev L, Iyer RR, Constantin N, Kadyrov FA, Modrich P. PCNA function in the activation and strand direction of MutLα endonuclease in mismatch repair. Proc Natl Acad Sci U S A. 2010;107(37):16066–71.

353. Kadyrov FA, Dzantiev L, Constantin N, Modrich P. Endonucleolytic Function of MutLa in Human Mismatch Repair. Cell. 2006;126(2):297–308.

354. Tran PT, Erdeniz N, Symington LS, Liskay RM. EXO1-A multi-tasking eukaryotic nuclease. DNA Repair (Amst). 2004;3(12):1549–59.

355. Lin YL, Shivji MK, Chen C, Kolodner R, Wood RD, Dutta A. The evolutionarily conserved zinc finger motif in the largest subunit of human replication protein A is required for DNA replication and mismatch repair but not for nucleotide excision repair. J Biol Chem. 1998;273(3):1453–61.

356. Genschel J, Modrich P. Mechanism of 5’-directed excision in human mismatch repair. Mol Cell. 2003;12(5):1077–86.

357. Lujan S, Jessica S. Williams, Clausen AR, Clark AB, A. TK. Evidence that ribonucleotides are signals for mismatch repair of leading strand replication errors. Mol Cell. 2013;50(3):437–43.

358. Ghodgaonkar MM, Lazzaro F, Olivera-Pimentel M, Artola-Boran M, Cejka P, Reijns MA, et al. Ribonucleotides misincorporated into DNA act as strand-discrimination signals in eukaryotic mismatch repair. Mol Cell. 2013;50(3):323–32.

359. Li Z, Pearlman AH, Hsieh P. DNA mismatch repair and the DNA damage response. DNA Repair (Amst). 2015;38:94–101.

360. Chen X, Zhao Y, Li G-M, Guo L. Proteomic analysis of mismatch repair-mediated alkylating agent-induced DNA damage response. Cell Biosci. 2013;3(1):37.

361. Karran P. Mechanisms of tolerance to DNA damaging therapeutic drugs. Vol. 22, Carcinogenesis. 2001. p. 1931–7.

362. O’Brien V, Brown R. Signalling cell cycle arrest and cell death through the MMR System. Carcinogenesis. 2006;27(4):682–92.

189

363. Cejka P, Stojic L, Mojas N, Russell AM, Heinimann K, Pietro M. Methylation-induced G 2 / M arrest requires a full complement of the mismatch repair protein hMLH1. EMBO J. 2003;22(9):2245–54.

364. Zhang Y, Hunter T. Roles of Chk1 in cell biology and cancer therapy. Int J Cancer. 2014;134(5):1013–23.

365. Stojic L, Mojas N, Cejka P, Di Pietro M, Ferrari S, Marra G, et al. Mismatch repair-dependent G2 checkpoint induced by low doses of SN1 type methylating agents requires the ATR kinase. Genes Dev. 2004;18(11):1331–44.

366. Duckett DR, Drummond JT, Murchiet AIH, Reardont JT, Sancart A, Lilleyt DMJ, et al. Human MutSα recognizes damaged DNA base pairs containing. Proc Natl Acad Sci U S A. 1996;93:6443–7.

367. Zou L, Elledge SJ. Sensing DNA damage through ATRIP recognition of RPA-ssDNA complexes. Vol. 300, Science. 2003. p. 1542–8.

368. Brown KD, Rathi A, Kamath R, Beardsley DI, Zhan Q, Mannino JL, et al. The mismatch repair system is required for S-phase checkpoint activation. Nat Genet. 2003;33(1):80–4.

369. Yoshioka K, Yoshioka Y, Hsieh P. ATR Kinase Activation Mediated by MutSα and MutLα in Response to Cytotoxic O6 -Methylguanine Adducts. Mol Cell. 2006;22(4):501–10.

370. Adamson A, Beardsley DI, Kim W, Gao Y, Baskaran R, Brown K. Methylator-induced, Mismatch Repair-dependent G2 Arrest Is Activated through Chk1 and Chk2. Mol Biol Cell. 2005;16(8):1513–26.

371. Wang Y, Qin J. MSH2 and ATR form a signaling module and regulate two branches of the damage response to DNA methylation. Proc Natl Acad Sci U S A. 2003;100(26):15387–92.

372. Olivera Harris M, Kallenberger L, Artola Borán M, Enoiu M, Costanzo V, Jiricny J. Mismatch repair-dependent metabolism of O6-methylguanine-containing DNA in Xenopus laevis egg extracts. DNA Repair (Amst). 2015;28:1–7.

373. Li GM. The role of mismatch repair in DNA damage-induced apoptosis. Oncol Res. 1999;11(9):393–400.

374. Chakraborty U, Alani E. Understanding how mismatch repair proteins participate in the repair/anti-recombination decision. FEMS Yeast Res. 2016;1–33.

375. Tham KC, Kanaar R, Lebbink JHG. Mismatch repair and homeologous recombination. DNA Repair (Amst). 2016;38:75–83.

376. Hoffmann ER, Borts RH. Meiotic recombination intermediates and mismatch repair proteins. Cytogenet Genome Res. 2004;107(3–4):232–48.

190

377. Harfe BD, Jinks-Robertson S. DNA mismatch repair and genetic instability. Annu Rev Genet. 2000;34:359–99.

378. Worth L, Bader T, Yang J, Clark S. Role of MutS ATPase activity in MutS,L-dependent block of in vitro strand transfer. J Biol Chem. 1998;273(36):23176–82.

379. Worth L, Clark S, Radman M, Modrich P. Mismatch repair proteins MutS and MutL inhibit RecA-catalyzed strand transfer between diverged DNAs. Proc Natl Acad Sci U S A. 1994;91:3238–41.

380. Chen S, Bigner SH, Modrich P. High rate of CAD gene amplification in human cells deficient in MLH1 or MSH6. Proc Natl Acad Sci U S A. 2001;98(24):13802–7.

381. Ohkura H. Meiosis: An overview of key differences from mitosis. Cold Spring Harb Perspect Biol. 2015;7(5):1–15.

382. Wang S, Zickler D, Kleckner N, Zhang L. Meiotic crossover patterns: Obligatory crossover, interference and homeostasis in a single process. Cell Cycle. 2015;14(3):305–14.

383. Handel MA, Schimenti JC. Genetics of mammalian meiosis: regulation, dynamics and impact on fertility. Nat Rev Genet. 2010;11(2):124–36.

384. Ross-Macdonald P, Roeder GS. Mutation of a meiosis-specific MutS homolog decreases crossing over but not mismatch correction. Cell. 1994;79(6):1069–80.

385. Hollingsworth N, Ponte L, Halsey C. MSH5, a novel MutS homolog, facilitates meiotic reciprocal recombination between homologs in Saccharomyces cerevisiae but not mismatch repair. Genes Dev. 1995;9:1728–39.

386. Snowden T, Acharya S, Butz C, Berardini M, Fishel R. hMSH4-hMSH5 recognizes holliday junctions and forms a meiosis-specific sliding clamp that embraces homologous chromosomes. Mol Cell. 2004;15(3):437–51.

387. Snowden T, Shim K-S, Schmutte C, Acharya S, Fishel R. hMSH4-hMSH5 Adenosine Nucleotide Processing and Interactions with Homologous Recombination Machinery. J Biol Chem. 2008;283(1):145–54.

388. Kunz C, Schär P. Meiotic recombination: Sealing the partnership at the junction. Curr Biol. 2004;14(22):962–4.

389. Lipkin SM, Moens PB, Wang V, Lenzi M, Shanmugarajah D, Gilgeous A, et al. Meiotic arrest and aneuploidy in MLH3-deficient mice. Nat Genet. 2002;31(4):385–90.

390. Prolla T, Baker S, Harris A, Tsao J-L, Yao X, Bronner CE, et al. Tumour susceptibility and spontaneous mutation in mice deficient in Mlh1, Pms1, and Pms2 DNA mismatch repair. Nat Genet. 1998;18(3):231–6.

391. Avdievich E, Reiss C, Scherer SJ, Zhang Y, Maier SM, Jin B, et al. Distinct effects of the

191

recurrent Mlh1G67R mutation on MMR functions, cancer, and meiosis. Proc Natl Acad Sci U S A. 2008;105(11):4247–52.

392. Baker SM, Bronner CE, Zhang L, Plug AW, Robatzek M, Warren G, et al. Male mice defective in the DNA mismatch repair gene PMS2 exhibit abnormal chromosome synapsis in meiosis. Cell. 1995;82(2):309–19.

393. Kolas NK, Svetlanov A, Lenzi ML, Macaluso FP, Lipkin SM, Liskay RM, et al. Localization of MMR proteins on meiotic chromosomes in mice indicates distinct functions during prophase I. J Cell Biol. 2005;171(3):447–58.

394. Zanotti KJ, Gearhart PJ. Antibody diversification caused by disrupted mismatch repair and promiscuous DNA polymerases. DNA Repair (Amst). 2016;38:110–6.

395. Rada C, Ehrenstein MR, Neuberger MS, Milstein C. Hot spot focusing of somatic hypermutation in MSH2-deficient mice suggests two stages of mutational targeting. Immunity. 1998;9(1):135–41.

396. Wiesendanger M, Kneitz B, Edelmann W, Scharff M. Somatic Hypermutation in MutS Homologue (MSH)3-, MSH6-, and MSH3/MSH6-deficient Mice Reveals a Role for the MSH2–MSH6 Heterodimer in Modulating the Base Substitution Pattern. J Exp Med. 2000;191(3):579–84.

397. Wilson TM, Vaisman A, Martomo SA, Sullivan P, Lan L, Hanaoka F, et al. MSH2-MSH6 stimulates DNA polymerase eta, suggesting a role for A:T mutations in antibody genes. J Exp Med. 2005;201(4):637–45.

398. Neuberger MS, Rada C. Somatic hypermutation: activation-induced deaminase for C/G followed by polymerase eta for A/T. J Exp Med. 2007;204(1):7–10.

399. Delbos F, Aoufouchi S, Faili A, Weill J-C, Reynaud C-A. DNA polymerase eta is the sole contributor of A/T modifications during immunoglobulin gene hypermutation in the mouse. J Exp Med. 2007;204(1):17–23.

400. Schrader CE, Guikema JEJ, Linehan EK, Selsing E, Stavnezer J. Activation-induced cytidine deaminase-dependent DNA breaks in class switch recombination occur during G1 phase of the cell cycle and depend upon mismatch repair. J Immunol. 2007;179(9):6064–71.

401. Schrader CE, Edelmann W, Kucherlapati R, Stavnezer J. Reduced Isotype Switching in Splenic B Cells from Mice Deficient in Mismatch Repair Enzymes. J Exp Med. 1999;190(3):323–30.

402. Lee K, Tosti E, Edelmann W. Mouse models of DNA mismatch repair in cancer research. DNA Repair (Amst). 2016;38:140–6.

403. Reitmair A, Schmits R, Ewel A, Bapat B, Redston M, Mitri A, et al. MSh2 deficient mice are viable and susceptible to lymphoid tumours. Nat Genet. 1995;11:64–70.

192

404. Chen PC, Dudley S, Hagen W, Dizon D, Paxton L, Reichow D, et al. Contributions by MutL homologues Mlh3 and Pms2 to DNA mismatch repair and tumor suppression in the mouse. Cancer Res. 2005;65(19):8662–70.

405. de Wind N, Dekker M, Berns A, Radman M, te Riele H. Inactivation of the mouse Msh2 gene results in mismatch repair deficiency, methylation tolerance, hyperrecombination, and predisposition to cancer. Cell. 1995;82(2):321–30.

406. Edelmann W, Yang K, Umar A, Heyer J, Lau K, Fan K, et al. Mutation in the mismatch repair gene Msh6 causes cancer susceptibility. Cell. 1997;91:467–77.

407. de Wind N, Dekker M, Claij N, Jansen L, van Klink Y, Radman M, et al. HNPCC-like cancer predisposition in mice through simultaneous loss of Msh3 and Msh6 mismatch-repair protein functions. Nat Genet. 1999;23(3):359–62.

408. Edelmann W, Umar A, Yang K, Heyer J, Kucherlapati M, Lia M, et al. The DNA Mismatch Repair Genes Msh3 and Msh6 Cooperate in Intestinal Tumor Suppression. Cancer Res. 2000;60:803–7.

409. Kneitz B, Cohen PE, Avdievich E, Zhu L, Kane MF, Hou H, et al. MutS homolog 4 localization to meiotic chromosomes is required for chromosome pairing during meiosis in male and female mice. Genes Dev. 2000;14(9):1085–97.

410. Edelmann W, Cohen PE, Kneitz B, Winand N, Lia M, Heyer J, et al. Mammalian MutS homologue 5 is required for chromosome pairing in meiosis. Nat Genet. 1999;21(1):123–7.

411. Wei K, Kucherlapati R, Edelmann W. Mouse models for human DNA mismatch-repair gene defects. TRENDS Mol Med. 2002;8(7):346–53.

412. Nguyen A, Bougeard G, Koob M, Chenard MP, Schneider A, Maugard C, et al. MSI detection and its pitfalls in CMMRD syndrome in a family with a bi-allelic MLH1 mutation. Fam Cancer. 2016;15(4):571–7.

413. Cheah C, Dsouza L, Taggart M, Schlette E, Turturro F. Diffuse large B-cell lymphoma with microsatellite instability developing in the setting of Muir – Torre variant hereditary non-polyposis colon cancer. J Clin Pathol. 2015;68(9):755–7.

414. Clark CR, Starr TK. Mouse models for the discovery of colorectal cancer driver genes. World J Gastroenterol. 2016;22(2):815–22.

415. Moser A, Pitot H, Dove W. A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science (80- ). 1990;247(4940):322.

416. Su LK, Kinzler KW, Vogelstein B, Preisinger AC, Moser AR, Luongo C, et al. Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science (80- ). 1992;256(5057):668–70.

417. Fodde R, Edelmann W, Yang K, van Leeuwen C, Carlson C, Renault B, et al. A targeted

193

chain-termination mutation in the mouse Apc gene results in multiple intestinal tumors. Proc Natl Acad Sci U S A. 1994;91(19):8969–73.

418. Edelmann W, Yang K, Kuraguchi M, Heyer J, Lia M, Kneitz B, et al. Tumorigenesis in Mlh1 and Mlh1 / Apc1638N Mutant Mice. Cancer Res. 1999;59:1301–7.

419. Kuraguchi M, Yang K, Wong E, Avdievich E, Fan K, Kolodner RD, et al. The distinct spectra of tumor-associated Apc mutations in mismatch repair-deficient Apc1638N mice define the roles of MSH3 and MSH6 in DNA repair and intestinal tumorigenesis. Cancer Res. 2001;61(21):7934–42.

420. Auton A, Abecasis GR, Altshuler DM, Durbin RM, Bentley DR, Chakravarti A, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74.

421. Kruglyak L, Nickerson DA. Variation is the spice of life. Nat Genet. 2001;27(3):234–6.

422. Choudhury A, Hazelhurst S, Meintjes A, Achinike-Oduaran O, Aron S, Gamieldien J, et al. Population-specific common SNPs reflect demographic histories and highlight regions of genomic plasticity with functional relevance. BMC Genomics. 2014;15(1):437.

423. Orr N, Chanock S. Common Genetic Variation and Human Disease. Adv Genet. 2008;62(8):1–32.

424. Venter J, Adams M, Myers E, Li P, Mural R, Sutton G, et al. The sequence of the human genome. Science (80- ). 2001;291(5507):1304–51.

425. Orgogozo V, Morizot B, Martin A. The differential view of genotype–phenotype relationships. Front Genet. 2015;6:179.

426. Patterson N, Hattangadi N, Lane B, Lohmueller KE, Hafler DA, Oksenberg JR, et al. Methods for high-density admixture mapping of disease genes. Am J Hum Genet. 2004;74(5):979–1000.

427. Shriver M, Mei R, Parra E, Sonpar V, Halder I, Tishkoff S, et al. Large-scale SNP analysis reveals clustered and continuous patterns of human genetic variation. Hum Genomics. 2005;2(2):81–9.

428. Welter D, Macarthur J, Morales J, Burdett T, Hall P, Junkins H, et al. The NHGRI GWAS Catalog , a curated resource of SNP-trait associations. Nucleic Acids Res. 2014;42:1001–6.

429. Yates C, Sternberg M. The Effects of Non-Synonymous Single Nucleotide Polymorphisms (nsSNPs) on Protein–Protein Interactions. J Mol Biol. 2013;425:3949–63.

430. Timofeeva MN, Kinnersley B, Farrington SM, Whiffin N, Palles C, Svinti V, et al. Recurrent Coding Sequence Variation Explains Only A Small Fraction of the Genetic Architecture of Colorectal Cancer. Sci Rep. 2015;5:16286.

431. Capon F, Allen MH, Ameen M, Burden AD, Tillman D, Barker JN, et al. A synonymous

194

SNP of the corneodesmosin gene leads to increased mRNA stability and demonstrates association with psoriasis across diverse ethnic groups. Hum Mol Genet. 2004;13(20):2361–8.

432. Chamary J V, Parmley JL, Hurst LD. Hearing silence: non-neutral evolution at synonymous sites in mammals. Nat Rev Genet. 2006;7(2):98–108.

433. Spisak S, Lawrenson K, Fu Y, Csabai I, Cottman RT, Seo JH, et al. CAUSEL: an epigenome- and genome-editing pipeline for establishing function of noncoding GWAS variants. Nat Med. 2015;21(11):1357–63.

434. Chen J, Tian W. Explaining the disease phenotype of intergenic SNP through predicted long range regulation. Nucleic Acids Res. 2016;44(28):8641–54.

435. Zhi D, Aslibekyan S, Irvin MR, Claas SA, Borecki IB, Ordovas JM, et al. SNPs located at CpG sites modulate genome-epigenome interaction. Epigenetics. 2013;8(8):802–6.

436. Chen YA, Lemire M, Choufani S, Butcher DT, Grafodatskaya D, Zanke BW, et al. Discovery of cross-reactive probes and polymorphic CpGs in the Illumina Infinium HumanMethylation450 microarray. Epigenetics. 2013;8(2):203–9.

437. Iafrate AJ, Feuk L, Rivera MN, Listewnik ML, Donahoe PK, Qi Y, et al. Detection of large-scale variation in the human genome. Nat Genet. 2004;36(9):949–51.

438. Zarrei M, MacDonald JR, Merico D, Scherer SW. A copy number variation map of the human genome. Nat Rev Genet. 2015;16(3):172–83.

439. Sebat J, Lakshmi B, Troge J, Alexander J, Young J, Lundin P, et al. Large-Scale Copy Number Polymorphism in the Human Genome. Science (80- ). 2004;305(5683):525–8.

440. Yang R, Chen B, Pfutze K, Buch S, Steinke V, Holinski-Feder E, et al. Genome-wide analysis associates familial colorectal cancer with increases in copy number variations and a rare structural variation at 12p12.3. Carcinogenesis. 2014;35(2):315–23.

441. Marczok S, Bortz B, Wang C, Pospisil H. Comprehensive Analysis of Genome Rearrangements in Eight Human Malignant Tumor Tissues. PLoS One. 2016;11(7):e0158995.

442. Pomerantz MM, Ahmadiyeh N, Jia L, Herman P, Verzi MP, Doddapaneni H, et al. The 8q24 cancer risk variant rs6983267 demonstrates long-range interaction with MYC in colorectal cancer. Nat Genet. 2009;41(8):882–4.

443. Sur I, Tuupanen S, Whitington T, Aaltonen LA, Taipale J. Lessons from functional analysis of genome-wide association studies. Cancer Res. 2013;73(14):4180–4.

444. Johnson AD, Handsaker RE, Pulit SL, Nizzari MM, O’Donnell CJ, De Bakker PIW. SNAP: A web-based tool for identification and annotation of proxy SNPs using HapMap. Bioinformatics. 2008;24(24):2938–9.

195

445. McVean G, Myers S, Hunt S, Deloukas P, Bentley D, Bonnelly P. The Fine-Scale Structure of Recombination Rate Variation in the Human Genome. Science (80- ). 2004;304:581–4.

446. Baudat F, Buard J, Grey C, Fledel-Alon A, Ober C, Przeworski M, et al. PRDM9 is a Major Determinant of Meiotic Recombination Hotspots in humans and mice. Science (80- ). 2010;327(5967):836–40.

447. Myers S, Bottolo L, Freeman C, Mcvean G, Donnelly P. A Fine-Scale Map of Recombination Rates and Hotspots Across the Human Genome. Science (80- ). 2005;310(2005):321–4.

448. Vahdati A, Wagner A. Parallel or convergent evolution in human population genomic data revealed by genotype networks. BMC Evol Biol. 2016;16(1):154.

449. Chaisson M, Wilson R, Eichler E. Genetic variation and the de novo assembly of human genomes Mark. Nat Rev Genet. 2015;16(11):627–40.

450. George Priya Doss C, Rajasekaran R, Arjun P, Sethumadhavan R. Prioritization of candidate SNPs in colon cancer using bioinformatics tools: An alternative approach for a cancer biologist. Interdiscip Sci Comput Life Sci. 2010;2(4):320–46.

451. Edwards SL, Beesley J, French JD, Dunning M. Beyond GWASs: Illuminating the dark road from association to function. Am J Hum Genet. 2013;93(5):779–97.

452. Frazer K, Murray S, Schork N, Topol E. Human genetic variation and its contribution to complex traits. Nat Rev Genet. 2009;10(4):241–51.

453. Harper AR, Nayee S, Topol EJ. Protective alleles and modifier variants in human health and disease. Nat Rev Genet. 2015;16:689–701.

454. Burdett T, Hall P, Hastings E, Hindorff L, Junkins H, Klemm A, et al. The NHGRI-EBI Catalog of published genome-wide association studies. Available at www.ebi.ac.uk/gwas. p. v1.0.

455. Peters U, Jiao S, Schumacher FR, Hutter CM, Aaron K, Baron JA, et al. Identification of Genetic Susceptibility Loci for Colorectal Tumors in a Genome-wide Meta-analysis. Gastroenterology. 2013;144(4):799–807.

456. Schumacher F, Schmit S, Jiao S, Edlund C, Wang H, Zhang B, et al. Genome-wide association study of colorectal cancer identifies six new susceptibility loci. Nat Commun. 2016;6:7138.

457. Tenesa A, Farrington SM, Prendergast JGD, Porteous ME, Walker M, Haq N, et al. Genome-wide association scan identifies a colorectal cancer susceptibility locus on 11q23 and replicates risk loci at 8q24 and 18q21. Nat Genet. 2009;40(5):631–7.

458. Tomlinson I, Webb E, Carvajal-carmona L, Broderick P, Kemp Z, Penegar S, et al. A genome-wide association scan of tag SNPs identifies a susceptibility variant for colorectal

196

cancer at 8q24.21. Nat Genet. 2007;39(8):984–8.

459. Tomlinson IPM, Webb E, Carvajal-Carmona L, Broderick P, Howarth K, Pittman AM, et al. A genome-wide association study identifies colorectal cancer susceptibility loci on chromosomes 10p14. Nat Genet. 2008;40(5):623–30.

460. Whiffin N, Hosking FJ, Farrington SM, Palles C, Dobbins SE, Lloyd A, et al. Identification of susceptibility loci for colorectal cancer in a genome-wide meta-analysis. Hum Mol Genet. 2014;23(17):4729–37.

461. Zanke BW, Greenwood CMT, Rangrej J, Kustra R, Tenesa A, Farrington SM, et al. Genome-wide association scan identifies a colorectal cancer susceptibility locus on chromosome 8q24. Nat Genet. 2007;39(8):989–94.

462. Zhang B, Jia W, Matsuda K, Kweon S, Matsuo K, Xiang B, et al. Large-scale genetic study in East Asians identifies six new loci associated with colorectal cancer risk. Nat Genet. 2014;46(6):533–42.

463. Yao L, Shen H, Laird PW, Farnham PJ, Berman BP. Inferring regulatory element landscapes and transcription factor networks from cancer methylomes. Genome Biol. 2015;16(1):105.

464. Broderick P, Carvajal-carmona L, Pittman AM, Webb E, Howarth K, Rowan A, et al. A genome-wide association study shows that common alleles of. Nat Genet. 2007;39(11):1315–7.

465. Zhang B, Jia W, Matsuo K, Shin A, Xiang Y, Matsuda K, et al. Genome-wide association study identifies a new SMAD7 risk variant associated with colorectal cancer risk in East Asians. Int J Cancer. 2014;135(4):948–55.

466. Worrillow LJ, Travis LB, Smith AG, Rollinson S, Smith AJ, Wild CP, et al. An intron splice acceptor polymorphism in hMSH2 and risk of leukemia after treatment with chemotherapeutic alkylating agents. Clin Cancer Res. 2003;9(8):3012–20.

467. Paz-y-Mio C, Pérez JC, Fiallo BF, Leone PE. A polymorphism in the hMSH2 gene (gIVS12-6T>C) associated with non-Hodgkin lymphomas. Cancer Genet Cytogenet. 2002;133(1):29–33.

468. Palicio M, Blanco I, Tortola S, Gonzalez I, Marcuello E, Brunet J, et al. Intron splice acceptor site polymorphism in the hMSH2 gene in sporadic and familial colorectal cancer. Br J Cancer. 2000;82(3):535–7.

469. Goessl C, Plaschke J, Pistorius S, Hahn M, Frank S, Hampl M, et al. An intronic germline transition in the HNPCC gene hMSH2 is associated with sporadic colorectal cancer. Eur J Cancer. 1997;33(11):1869–74.

470. Brentnall TA, Rubin CE, Crispin DA, Stevens A, Batchelor RH, Haggitt RC, et al. A germline substitution in the human MSH2 gene is associated with high-grade dysplasia and cancer in ulcerative colitis. Gastroenterology. 1995;109(1):151–5.

197

471. Jung CY, Choi JE, Park JM, Chae MH, Kang H, Kim KM, et al. Polymorphisms in the hMSH2 Gene and the Risk of Primary Lung Cancer. Cancer Epidemiol Biomarkers Prev. 2006;15:762–8.

472. Doherty JA, Sakoda LC, Loomis MM, Barnett MJ, Julianto L, Mark D. DNA repair genotype and lung cancer risk in the beta- carotene and retinol efficacy trial. Int J Mol Epidemiol Genet. 2013;4(1):11–34.

473. Beiner ME. Endometrial Cancer Risk Is Associated with Variants of the Mismatch Repair Genes MLH1 and MSH2. Cancer Epidemiol Biomarkers Prev. 2006;15(9):1636–40.

474. Campbell PT, Curtin K, Ulrich CM, Samowitz WS, Bigler J, Velicer CM, et al. Mismatch repair polymorphisms and risk of colon cancer, tumour microsatellite instability and interactions with lifestyle factors. Gut. 2009;58(5):661–7.

475. Lee E, Levine EA, Franco VI, Allen GO, Gong F, Zhang Y, et al. Combined genetic and nutritional risk models of triple negative breast cancer. Nutr Cancer. 2014;66(6):955–63.

476. Curtin K, Samowitz W, Wolff R, Caan B, Ulrich C, Potter J, et al. MSH6 G39E Polymorphism and CpG Island Methylator Phenotype in Colon Cancer. Mol Carcinog. 2009;48(11):989–94.

477. Lipkin SM, Rozek LS, Rennert G, Yang W, Chen P-C, Hacia J, et al. The MLH1 D132H variant is associated with susceptibility to sporadic colorectal cancer. Nat Genet. 2004;36(7):694–9.

478. Raptis S, Mrkonjic M, Green RC, Pethe V V., Monga N, Chan YM, et al. MLH1 -93G>A promoter polymorphism and the risk of microsatellite-unstable colorectal cancer. J Natl Cancer Inst. 2007;99(6):463–74.

479. Mrkonjic M, Roslin NM, Greenwood CM, Raptis S, Pollett A, Laird PW, et al. Specific variants in the MLH1 gene region may drive DNA methylation, loss of protein expression, and MSI-H colorectal cancer. PLoS One. 2010;5(10):1–10.

480. Perera S, Mrkonjic M, Rawson JB, Bapat B. Functional effects of the MLH1-93G>A polymorphism on MLH1/EPM2AIP1 promoter activity. Oncol Rep. 2011;25(3):809–15.

481. Rodriguez-Hernandez I, Perdomo S, Santos-Briz A, Garcia JL, Gomez-Moreta JA, Cruz JJ, et al. Analysis of DNA repair gene polymorphisms in glioblastoma. Gene. 2013;536:79–83.

482. Zhu H, Li X, Zhang X, Chen D, Li D, Ren J, et al. Polymorphisms in mismatch repair genes are associated with risk and microsatellite instability of gastric cancer, and interact with life exposures. Gene. 2016;579(1):52–7.

483. Niu L, Li S, Liang H, Li H. The hMLH1 −93G>A Polymorphism and Risk of Ovarian Cancer in the Chinese Population. PLoS One. 2015;10(8):e0135822.

484. Lo YL, Hsiao CF, Jou YS, Chang GC, Tsai YH, Su WC, et al. Polymorphisms of MLH1

198

and MSH2 genes and the risk of lung cancer among never smokers. Lung Cancer. 2011;72(3):280–6.

485. Houlston RS, Cheadle J, Dobbins SE, Tenesa A, Angela M, Howarth K, et al. Meta-analysis of three genome-wide association studies identifies susceptibility loci for colorectal cancer at 1q41, 3q26.2, 12q13.13 and 20q13.33. Nat Genet. 2010;42(11):973–7.

486. Figueiredo JC, Hsu L, Hutter CM, Lin Y, Campbell PT, Baron JA, et al. Genome-Wide Diet-Gene Interaction Analyses for Risk of Colorectal Cancer. PLoS Genet. 2014;10(4).

487. Schmit SL, Schumacher FR, Edlund CK, Conti D V, Raskin L, Lejbkowicz F, et al. A novel colorectal cancer risk locus at 4q32.2 identified from an international genome-wide association study. Carcinogenesis. 2014;35(11):2512–9.

488. Jia W, Zhang B, Matsuo K, Shin A, Xiang Y, Ha S, et al. Genome-wide association analyses in East Asians identify new susceptiblity loci for colorectal cancer. Nat Genet. 2013;45(2):191–6.

489. Dunlop M, Dobbins S, Farrington S, Jones A, Palles C, Whiffin N, et al. Common variation near CDKN1A, POLD3, and SHROOM2 influences colorectal cancer risk. Nat Genet. 2016;44(7):770–6.

490. Cui R, Okada Y, Jang SG, Ku JL, Park JG, Kamatani Y, et al. Common variant in 6q26-q27 is associated with distal colon cancer in an Asian population. Gut. 2011;60:799–805.

491. Wang H, Burnett T, Kono S, Haiman C, Iwasaki M, Wilkens L, et al. Trans-ethnic genome-wide association study of colorectal cancer identifies a new susceptibility locus in VTI1A. Nat Commun. 2015;5:4613.

492. COGENT Study. Meta-analysis of genome-wide association data identifies four new susceptibility loci for colorectal cancer. Nat Genet. 2008;40(12):1426–35.

493. Herman JG, Baylin SB. Gene silencing in cancer in association with promoter hypermethylation. N Engl J Med. 2003;349:2042–54.

494. Jones P, Baylin SB. The epigenomics of cancer. Cell. 2007;128(4):683–92.

495. Doi A, Park I, Wen B, Murakami P, Aryee M, Irizarry R, et al. Differential methylation of tissue-and cancer-specific CpG island shores distinguishes human induced pluripotent stem cells, embryonic stem cells and fibroblasts. Nat Genet. 2009;41(12):1350–3.

496. Feber A, Wilson G, Zhang L, Presneau N, Idowu B, Down T, et al. Comparative methylome analysis of benign and malignant peripheral nerve sheath tumors. Genome Res. 2011;21:515–24.

497. Terry MB, Delgado-Cruzata L, Vin-Raviv N, Wu HC, Santella RM. DNA methylation in white blood cells: Association with risk factors in epidemiologic studies. Epigenetics. 2011;6(7):828–37.

199

498. Bosviel R, Michard E, Lavediaux G, Kwiatkowski F, Bignon YJ, Bernard-Gallon DJ. Peripheral blood DNA methylation detected in the BRCA1 or BRCA2 promoter for sporadic ovarian cancer patients and controls. Clin Chim Acta. 2011;412(15–16):1472–5.

499. Van Bemmel D, Lenz P, Liao LM, Baris D, Sternberg LR, Warner A, et al. Correlation of LINE-1 methylation levels in patient-matched buffy coat, serum, buccal cell, and bladder tumor tissue DNA samples. Cancer Epidemiol Biomarkers Prev. 2012;21(7):1143–8.

500. Bosviel R, Garcia S, Lavediaux G, Michard E, Dravers M, Kwiatkowski F, et al. BRCA1 promoter methylation in peripheral blood DNA was identified in sporadic breast cancer and controls. Cancer Epidemiol. 2012;36(3):1–6.

501. Marsit CJ, Koestler DC, Christensen BC, Karagas MR, Houseman EA, Kelsey KT. DNA methylation array analysis identifies profiles of blood-derived DNA methylation associated with bladder cancer. J Clin Oncol. 2011;29(9):1133–9.

502. Al-Moundhri MS, Al-Nabhani M, Tarantini L, Baccarelli A, Rusiecki JA. The prognostic significance of whole blood global and specific DNA methylation levels in gastric adenocarcinoma. PLoS One. 2010;5(12).

503. Koestler DC, Marsit CJ, Christensen BC, Accomando W, Langevin SM, Houseman EA, et al. Peripheral blood immune cell methylation profiles are associated with nonhematopoietic cancers. Cancer Epidemiol Biomarkers Prev. 2012;21(8):1293–302.

504. Li B, Gan A, Chen X, Wang X, He W, Zhang X, et al. Diagnostic Performance of DNA Hypermethylation Markers in Peripheral Blood for the Detection of Colorectal Cancer: A Meta-Analysis and Systematic Review. PLoS One. 2016;11(5):e0155095.

505. Luo X, Huang R, Sun H, Liu Y, Bi H, Li J, et al. Methylation of a panel of genes in peripheral blood leukocytes is associated with colorectal cancer. Sci Rep. 2016;6:29922.

506. Jaffe AE, Irizarry RA. Accounting for cellular heterogeneity is critical in epigenome-wide association studies. Genome Biol. 2014;15(2):R31.

507. Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13(1):86.

508. Reinius LE, Acevedo N, Joerink M, Pershagen G, Dahlén SE, Greco D, et al. Differential DNA methylation in purified human blood cells: Implications for cell lineage and studies on disease susceptibility. PLoS One. 2012;7(7).

509. Wu HC, Wang Q, Delgado-Cruzata L, Santella RM, Terry MB. Genomic methylation changes over time in peripheral blood mononuclear cell DNA: Differences by assay type and baseline values. Cancer Epidemiol Biomarkers Prev. 2012;21(8):1314–8.

510. Donehower LA, Creighton CJ, Schultz N, Shinbrot E, Gunaratne PH, Muzny D, et al. MLH1-silenced and non-silenced subgroups of hypermutated colorectal carcinomas have distinct mutational landscapes. J Pathol. 2014;229(1):99–110.

200

511. Mrkonjic M, Raptis S, Green RC, Monga N, Daftary D, Dicks E, et al. MSH2 -118T>C and MSH6 -159C>T promoter polymorphisms and the risk of colorectal cancer. Carcinogenesis. 2007;28(12):2575–80.

512. Newcomb PA, Baron J, Cotterchio M, Gallinger S, Grove J, Haile R, et al. Colon cancer family registry: An international resource for studies of the genetic epidemiology of colon cancer. Cancer Epidemiol Biomarkers Prev. 2007;16(11):2331–43.

513. Lemire M, Zaidi SHE, Zanke BW, Gallinger S, Hudson TJ, Cleary SP. The effect of 5-fluorouracil/leucovorin chemotherapy on CpG methylation, or the confounding role of leukocyte heterogeneity: An illustration. Genomics. 2015;106(6):340–7.

514. Lemire M, Zaidi SHE, Ban M, Ge B, Aïssi D, Germain M, et al. Long-range epigenetic regulation is conferred by genetic variation located at thousands of independent loci. Nat Commun. 2015;6:6326.

515. Savio AJ, Lemire M, Mrkonjic M, Gallinger S, Zanke BW, Hudson TJ, et al. MLH1 Region Polymorphisms Show a Significant Association with CpG Island Shore Methylation in a Large Cohort of Healthy Individuals. PLoS One. 2012;7(12):e51531.

516. Deng G, Chen A, Hong J, Chae HS, Kim YS. Methylation of CpG in a small region of the hMLH1 promoter invariably correlates with the absence of gene expression. Cancer Res. 1999;59(9):2029–33.

517. Bird A, Taggart M, Frommer M, Miller OJ MD. A fraction of the mouse genome that is derived from islands of nonmethylated, CpG-rich DNA. Cell. 1985;40:91–9.

518. Kim JW, Kim S-T, Turner AR, Young T, Smith S, Liu W, et al. Identification of new differentially methylated genes that have potential functional consequences in prostate cancer. PLoS One. 2012;7(10):e48455.

519. Rao X, Evans J, Chae H, Kim S, Liu Y, Huang T, et al. CpG island shore methylation regulates caveolin-1 expression in breast cancer. Cancer Res. 2012;72(8 Supplement):5013–5013.

520. Gaunt TR, Shihab HA, Hemani G, Min JL, Woodward G, Lyttleton O, et al. Systematic identification of genetic influences on methylation across the human life course. Genome Biol. Genome Biology; 2016;17(1):61.

521. Gibbs JR, van der Brug MP, Hernandez DG, Traynor BJ, Nalls MA, Lai SL, et al. Abundant quantitative trait loci exist for DNA methylation and gene expression in Human Brain. PLoS Genet. 2010;6(5):29.

522. Müller K, Heller H, Doerfier W. Foreign DNA integration. Genome-wide perturbations of methylation and transcription in the recipient genomes. J Biol Chem. 2001;276(17):14271–8.

523. Bell CG, Finer S, Lindgren CM, Wilson GA, Rakyan VK, Teschendorff AE, et al. Integrated genetic and epigenetic analysis identifies haplotype-specific methylation in the

201

FTO type 2 diabetes and obesity susceptibility locus. PLoS One. 2010;5(11).

524. Whiffin N, Broderick P, Lubbe SJ, Pittman AM, Penegar S, Chandler I, et al. MLH1-93G > a is a risk factor for MSI colorectal cancer. Carcinogenesis. 2011;32(8):1157–61.

525. Kuroiwa-Trzmielina J, Wang F, Rapkins RW, Ward RL, Buchanan DD, Win AK, et al. SNP rs16906252C>T is an expression and methylation quantitative trait locus associated with an increased risk of developing MGMT-methylated colorectal cancer. Clin Cancer Res. 2016;22(24):6266–77.

526. Shin K, Shin J, Kim J, Park J. Advances in Brief Mutational Analysis of Promoters of Mismatch Repair Genes hMSH2 and hMLH1 in Hereditary Nonpolyposis Colorectal Cancer and Early Onset Colorectal Cancer Patients: Identification of Three Novel Germ-line Mutations in Promoter of the hMSH2. Cancer Res. 2002;62:38–42.

527. Gazzoli I, Kolodner RD. Regulation of the Human MSH6 Gene by the Sp1 Transcription Factor and Alteration of Promoter Activity and Expression by Polymorphisms. Mol Cell Biol. 2003;23(22):7992–8007.

528. Ianzano L, Zhao XC, Minassian BA, Scherer SW. Identification of a novel protein interacting with laforin, the EPM2A progressive myoclonus epilepsy gene product. Genomics. 2003;81(6):579–87.

529. Turnbull J, Tiberia E, Pereira S, Zhao X, Pencea N, Wheeler AL, et al. Deficiency of a glycogen synthase-associated protein, Epm2aip1, causes decreased glycogen synthesis and hepatic insulin resistance. J Biol Chem. 2013;288(48):34627–37.

530. Burger D, Fickentscher C, de Moerloose P, Brandt KJ. F-actin dampens NLRP3 inflammasome activity via Flightless-I and LRRFIP2. Sci Rep. 2016;6:29834.

531. Jin J, Yu Q, Han C, Hu X, Xu S, Wang Q, et al. LRRFIP2 negatively regulates NLRP3 inflammasome activation in macrophages by promoting Flightless-I-mediated caspase-1 inhibition. Nat Commun. 2013;4:2075.

532. Liu J, Bang AG, Kintner C, Orth AP, Chanda SK, Ding S, et al. Identification of the Wnt signaling activator leucine-rich repeat in Flightless interaction protein 2 by a genome-wide functional analysis. Proc Natl Acad Sci. 2005;102(6):1927–32.

533. Kisiel JB, Raimondo M, Taylor WR, Yab TC, Mahoney WD, Sun Z, et al. New DNA methylation markers for pancreatic cancer: Discovery, tissue validation, and pilot testing in pancreatic juice. Clin Cancer Res. 2015;21(19):4473–81.

534. Kim SJ, Sohn I, Do I-G, Jung SH, Ko YH, Yoo HY, et al. Gene expression profiles for the prediction of progression-free survival in diffuse large B cell lymphoma: results of a DASL assay. Ann Hematol. 2014;93(3):437–47.

535. Kober P, Bujko M, Oledzki J, Tysarowski A, Siedlecki JA. Methyl-CpG binding column-based identification of nine genes hypermethylated in colorectal cancer. Mol Carcinog. 2011;50(11):846–56.

202

536. Gomes AQ, Correia D V, Grosso AR, Lanca T, Ferreira C, Lacerda JF, et al. Identification of a panel of ten cell surface protein antigens associated with immunotargeting of leukemias and lymphomas by peripheral blood gamma delta T cells. Haematologica. 2010;95(8):1397–404.

537. Arnold CR, Wolf J, Brunner S, Herndler-Brandstetter D, Grubeck-Loebenstein B. Gain and loss of T cell subsets in old age - Age-related reshaping of the T cell repertoire. J Clin Immunol. 2011;31(2):137–46.

538. Perez-Andres M, Paiva B, Nieto WG, Caraux A, Schmitz A, Almeida J, et al. Human peripheral blood B-Cell compartments: A crossroad in B-cell traffic. Cytom Part B - Clin Cytom. 2010;78(SUPPL. 1):47–60.

539. Yang Z, Jones A, Widschwendter M, Teschendorff AE. An integrative pan-cancer-wide analysis of epigenetic enzymes reveals universal patterns of epigenomic deregulation in cancer. Genome Biol. 2015;16:140.

540. Sahnane N, Magnoli F, Bernasconi B, Tibiletti MG, Romualdi C, Pedroni M, et al. Aberrant DNA methylation profiles of inherited and sporadic colorectal cancer. Clin Epigenetics. 2015;7:131.

541. Lin EI, Tseng L, Gocke CD, Reil S, Le DT, Azad NS, et al. Mutational profiling of colorectal cancers with microsatellite instability. 2015;6(39).

542. Moslein G, Tester DJ, Lindor NM, Honchel R, Cunningham JM, French AJ, et al. Microsatellite instability and mutation analysis of hMSH2 and hMLH1 in patients with sporadic, familial and hereditary colorectal cancer. Hum MolGenet. 1996;5(9):1245–52.

543. Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M, Pukkala E, Skytthe A HK. Environmental and heritable factors in the causation of cancer - analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med. 2000;343(2):78–85.

544. Jiao S, Peters U, Berndt S, Brenner H, Butterbach K, Caan BJ, et al. Estimating the heritability of colorectal cancer. Hum Mol Genet. 2014;23(14):3898–905.

545. Hutter CM, Slattery ML, Duggan DJ, Muehling J, Curtin K, Hsu L, et al. Characterization of the association between 8q24 and colon cancer+: gene-environment exploration and meta-analysis. 2010;

546. Lemire M, Qu C, Loo LWM, Zaidi SHE, Wang H, Berndt SI, et al. A genome-wide association study for colorectal cancer identifies a risk locus in 14q23.1. Hum Genet. 2015;134(11–12):1249–62.

547. Biancolella M, Fortini BK, Tring S, Plummer SJ, Mendoza-Fandino GA, Hartiala J, et al. Identification and characterization of functional risk variants for colorectal cancer mapping to chromosome 11q23.1. Hum Mol Genet. 2014;23(8):2198–209.

548. Chen H, Shen Z, Hu Y, Xiao Q, Bei D, Shen X, et al. Association between MutL homolog

203

1 polymorphisms and the risk of colorectal cancer: a meta-analysis. J Cancer Res Clin Oncol. 2015;141(12):2147–58.

549. Taberlay PC, Statham AL, Kelly TK, Clark SJ, Jones PA. Reconfiguration of nucleosome-depleted regions at distal regulatory elements accompanies DNA methylation of enhancers and insulators in cancer. 2014;1421–32.

550. Lienert F, Wirbelauer C, Som I, Dean A, Mohn F, Schübeler D. Identification of genetic elements that autonomously determine DNA methylation states. Nat Genet. 2011;43(11):1091–7.

551. Cedar H, Bergman Y. Programming of DNA methylation patterns. Annu Rev Biochem. 2012;81:97–117.

552. Baubec T, Schubeler D. Genomic patterns and context specific interpretation of DNA methylation. Curr Opin Genet Dev. 2014;25(1):85–92.

553. Hernando-Herraez I, Heyn H, Fernandez-Callejo M, Vidal E, Fernandez-Bellon H, Prado-Martinez J, et al. The interplay between DNA methylation and sequence divergence in recent human evolution. Nucleic Acids Res. 2015;43(17):8204–14.

554. Cotterchio M, McKeown-Eyssen G, Sutherland H, Buchan G, Aronson M, Easson AM, et al. Ontario Familial Colon Cancer Registry: Methods and first-year response rates. Chronic Dis Can. 2000;21(2):81–6.

555. Lindor NM, Burgart LJ, Leontovich O, Goldberg RM, Cunningham JM, Sargent DJ, et al. Immunohistochemistry versus microsatellite instability testing in phenotyping colorectal tumors. J Clin Oncol. 2002;20(4):1043–8.

556. Campan M, Weisenberger DJ, Trinh B, Laird PW. MethyLight. Methods Mol Biol. 2009;507(6):325–37.

557. Jekel J, Katz D, Elmore J, Wild D. Epidemiology, Biostatistics and preventive medicine. Saunders Elsevier; 1996.

558. Vickers AJ. Parametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data. BMC Med Res Methodol. 2005;5:35.

559. Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, Faasse M a, et al. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet. 2006;38(7):787–93.

560. Poplawski T, Sobczuk A, Sarnik J, Pawlowska E, Blasiak J. Polymorphism of DNA mismatch repair genes in endometrial cancer. Exp Oncol. 2015;37(1):44–7.

561. Shih CM, Chen CY, Lee IH, Kao WT WY. A polymorphism in the hMLH1 gene (-93G>A) associated with lung cancer susceptibility and prognosis. Int J Mol Med. 2010;25(1):165–70.

204

562. Lin L-H, Lin M-W, Mar K, Lin C-S, Ji D-D, Lee W-P, et al. The hMLH1 -93G>A Promoter Polymorphism is Associated with Outcomes in Oral Squamous Cell Carcinoma Patients. Ann Surg Oncol. 2014;21:4270–7.

563. Yang X, Shao X, Gao L, Zhang S. Systematic DNA methylation analysis of multiple cell lines reveals common and specific patterns within and across tissues of origin. Hum Mol Genet. 2015;24(15):4374–84.

564. Warnick CT, Dabbas B, Ilstrup SJ, Ford C, Strait K. Cell Type-Dependent Regulation of hMLH1 Promoter Activity Is Influenced by the Presence of Multiple Redundant Elements. Mol Cancer Res. 2003;1:610–8.

565. Maurano MT, Humbert R, Rynes E, Thurman RE, Wang H, Reynolds AP, et al. Systematic Localization of Common Disease-Associated Variation in Regulatory DNA. Science (80- ). 2012;337(6099):1190–5.

566. Butter F, Davison L, Viturawong T, Scheibe M, Vermeulen M, Todd JA, et al. Proteome-Wide Analysis of Disease-Associated SNPs That Show Allele-Specific Transcription Factor Binding. PLoS Genet. 2012;8(9).

567. Tak YG, Farnham PJ. Making sense of GWAS: using epigenomics and genome engineering to understand the functional relevance of SNPs in non-coding regions of the human genome. Epigenetics Chromatin. 2015;8(1):57.

568. Miyakura Y, Tahara M, Lefor AT, Yasuda Y, Sugano K. Haplotype defined by the MLH1-93G/A polymorphism is associated with MLH1 promoter hypermethylation in sporadic colorectal cancers. BMC Res Notes. 2014;7:835.

569. Gebhard C, Benner C, Ehrich M, Schwarzfischer L, Schilling E, Klug M, et al. General transcription factor binding at CpG islands in normal cells correlates with resistance to de novo DNA methylation in cancer cells. Cancer Res. 2010;70(4):1398–407.

570. Takeshima H, Yamashita S, Shimazu T, Niwa T, Ushijima T. The presence of RNA polymerase II, active or stalled, predicts epigenetic fate of promoter CpG islands. Genome Res. 2009;19(11):1974–82.

571. Mendenhall EM, Koche RP, Truong T, Zhou VW, Issac B, Chi AS, et al. GC-rich sequence elements recruit PRC2 in mammalian ES cells. PLoS Genet. 2010;6(12):1–10.

572. Van Kruijsbergen I, Hontelez S, Veenstra GJC. Recruiting polycomb to chromatin. Int J Biochem Cell Biol. 2015;67:177–87.

573. McGarvey KM, Fahrner JA, Greene E, Martens J, Jenuwein T, Baylin SB. Silenced tumor suppressor genes reactivated by DNA demethylation do not return to a fully euchromatic chromatin state. Cancer Res. 2006;66(7):3541–9.

574. Lin JC, Jeong S, Liang G, Takai D, Fatemi M, Tsai YC, et al. Role of Nucleosomal Occupancy in the Epigenetic Silencing of the MLH1 CpG Island. Cancer Cell. 2007;12(5):432–44.

205

575. Bork U, Rahbari NN, Schölch S, Reissfelder C, Kahlert C, Büchler MW, et al. Circulating tumour cells and outcome in non-metastatic colorectal cancer: a prospective study. Br J Cancer. 2015;112:1306–13.

576. Rahbari NN, Aigner M, Thorlund K, Mollberg N, Motschall E, Jensen K, et al. Meta-analysis Shows That Detection of Circulating Tumor Cells Indicates Poor Prognosis in Patients With Colorectal Cancer. Gastroenterology. 2010;138(5):1714–26.

577. Weitz J, Kienle P, Lacroix J, Willeke F, Benner A, Lehnert T, et al. Dissemination of tumor cells in patients undergoing surgery for colorectal cancer. Clin Cancer Res. 1998;4:343–8.

578. Harouaka R a., Nisic M, Zheng SY. Circulating tumor cell enrichment based on physical properties. J Lab Autom. 2013;18(6):1–24.

579. Khorasanizadeh S. The nucleosome. From genomic organization to genomic regulation. Cell. 2004;116(2):259–72.

580. Tan M, Luo H, Lee S, Jin F, Yang JS, Montellier E, et al. Identification of 67 histone marks and histone lysine crotonylation as a new type of histone modification. Cell. 2011;21(8):1016–28.

581. Kimura H. Histone modifications for human epigenome analysis. J Hum Genet. 2013;58(7):439–45.

582. Ram O, Goren A, Amit I, Shoresh N, Yosef N, Kellis M, et al. Combinatorial patterning of chromatin regulators uncovered by genome-wide location analysis in human cells. Cell. 2011;147(7):1628–39.

583. Geisler SJ, Paro R. Trithorax and Polycomb group-dependent regulation: a tale of opposing activities. Development. 2015;142(17):2876–87.

584. Schuettengruber B, Martinez A-M, Iovino N, Cavalli G. Trithorax group proteins: switching genes on and keeping them active. Nat Rev Mol Cell Biol. 2011;12(12):799–814.

585. Zentner GE, Tesar PJ, Scacheri PC. Epigenetic signatures distinguish multiple classes of enhancers with distinct cellular functions. Genome Res. 2011;21(8):1273–83.

586. Hark AT, Schoenherr CJ, Katz DJ, Ingram RS, Levorse JM, Tilghman SM. CTCF mediates methylation-sensitive enhancer-blocking activity at the H19/Igf2 locus. Nature. 2000;405(6785):486–9.

587. Jackstadt R, Röh S, Neumann J, Jung P, Hoffmann R, Horst D, et al. AP4 is a mediator of epithelial-mesenchymal transition and metastasis in colorectal cancer. J Exp Med. 2013;210(7):1331–50.

588. Ahmed D, Eide PW, Eilertsen IA, Danielsen SA, Eknæs M, Hektoen M, et al. Epigenetic and genetic features of 24 colon cancer cell lines. Oncogenesis. 2013;2(9):e71.

206

589. Ku J, Park J. Biology of SNU Cell Lines. Cancer Res Treat. 2005;37(1):1–19.

590. Mouradov D, Sloggett C, Jorissen RN, Love CG, Li S, Burgess AW, et al. Colorectal cancer cell lines are representative models of the main molecular subtypes of primary cancer. Cancer Res. 2014;74(12):3238–47.

591. Boyer JC, Umar A, Risinger JI, Lipford JR, Kane M, Yin S, et al. Microsatellite instability, mismatch repair deficiency, and genetic defects in human cancer cell lines. Cancer Res. 1995;55(24):6063–70.

592. Papdopolous N, Nicolaides NC, Liu B, Parsons R, Lengauer C, Palombo F. Mutations of GTBP in genetically unstable cells. Science (80- ). 1995;268(5219):1915–7.

593. Haring M, Offermann S, Danker T, Horst I, Peterhansel C, Stam M. Plant Methods. Plant Methods. 2007;3(11):1–16.

594. Nagaki K, Talbert PB, Zhong CX, Dawe RK, Henikoff S, Jiang J. Chromatin Immunoprecipitation Reveals That the 180-bp Satellite Repeat Is the Key Functional DNA Element of Arabidopsis thaliana Centromeres. Genetics. 2003;1225:1221–5.

595. Tolkunov D, Zawadzki KA, Singer C, Elfving N, Morozov A V, Broach JR. Chromatin remodelers clear nucleosomes from intrinsically unfavorable sites to establish nucleosome-depleted regions at promoters. Mol Biol Cell. 2011;22(12):2106–18.

596. Kristjuhan A, Svejstrup JQ. Evidence for distinct mechanisms facilitating transcript elongation through chromatin in vivo. EMBO J. 2004;23(21):4243–52.

597. Blattler A, Farnham PJ. Cross-talk between site-specific transcription factors and DNA methylation states. J Biol Chem. 2013;288(48):34287–94.

598. Shlyueva D, Stampfel G, Stark A. Transcriptional enhancers: from properties to genome-wide predictions. Nat Rev Genet. 2014;15(4):272–86.

599. Dedeurwaerder S, Defrance M, Calonne E, Denis H, Sotiriou C, Fuks F. Evaluation of the Infinium Methylation 450K technology. Epigenomics. 2011;3(6):771–84.

600. Pisanic T, Athamanolap P, Wang T-H. Defining, distinguishing and detecting the contribution of heterogeneous methylation to cancer heterogeneity. Semin Cell Dev Biol. 2016;

601. Putnik M, Wojdacz TK, Pournara A, Vahter M, Wallberg AE. MS-HRM assay identi fi es high levels of epigenetic heterogeneity in human immortalized cell lines. Gene. 2015;560:165–72.

602. Siegel R, Desantis C, Virgo K, Stein K, Mariotto A, Smith T, et al. Cancer Treatment and Survivorship Statistics, 2014. CA Cancer J Clin. 2013;64(4):252–71.

603. Brenner H, Bouvier AM, Foschi R, Hackl M, Larsen IK, Lemmens V, et al. Progress in colorectal cancer survival in Europe from the late 1980s to the early 21st century: The

207

EUROCARE study. Int J Cancer. 2012;131(7):1649–58.

604. Shima K, Morikawa T, Yamauchi M, Kuchiba A, Imamura Y, Liao X, et al. TGFBR2 and BAX mononucleotide tract mutations, microsatellite instability, and prognosis in 1072 colorectal cancers. PLoS One. 2011;6(9):1–9.

605. Bae JM, Kim JH, Kang GH. Molecular Subtypes of Colorectal Cancer and Their Clinicopathologic Features, With an Emphasis on the Serrated Neoplasia Pathway. Arch Pathol Lab Med. 2016;140(5):406–12.

606. Arriba M, García JL, Inglada-Pérez L, Rueda D, Osorio I, Rodríguez Y, et al. DNA copy number profiling reveals different patterns of chromosomal instability within colorectal cancer according to the age of onset. Mol Carcinog. 2016;55(5):705–16.

607. Michael Ghadimi B, Ried T. Chromosomal Instability in Cancer Cells. Chromosom Instab Cancer Cells. 2015;1–224.

608. Michel S, Benner A, Tariverdian M, Wentzensen N, Hoefler P, Pommerencke T, et al. High density of FOXP3-positive T cells infiltrating colorectal cancers with microsatellite instability. Br J Cancer. 2008;99(11):1867–73.

609. Goldstein J, Tran B, Ensor J, Gibbs P, Wong HL, Wong SF, et al. Multicenter retrospective analysis of metastatic colorectal cancer (CRC) with high-level microsatellite instability (MSI-H). Ann Oncol. 2014;25(5):1032–8.

610. Shen L, Catalano P, Benson A, O’Dwyer P, Hamilton SR, Issa J. Association between DNA methylation and shortened survival in patients with advanced colorectal cancer treated with 5-fluorouracil based chemotherapy. Clin Cancer Res. 2007;13(20):6093–8.

611. Ogino S, Nosho K, Kirkner GJ, Kawasaki T, Jeffrey A, Loda M, et al. CpG island methylator phenotype, microsatellite instability, BRAF mutation and clinical outcome in colon cancer. Gut. 2009;58(1):90–6.

612. Guinney J, Dienstmann R, Wang X, de Reyniès A, Schlicker A, Soneson C, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21(11):1350–6.

613. Kudryavtseva A V., Lipatova A V., Zaretsky AR, Moskalev AA, Fedorova MS, Rasskazova AS, et al. Important molecular genetic markers of colorectal cancer. Oncotarget. 2016;7(33):53959–83.

614. Thorstensen L, Lind GE, Løvig T, Diep CB, Meling GI, Rognum TO, et al. Genetic and epigenetic changes of components affecting the WNT pathway in colorectal carcinomas stratified by microsatellite instability. Neoplasia. 2005;7(2):99–108.

615. Huelsken J, Birchmeier W. New aspects of Wnt signaling pathways in higher vertebrates. Curr Opin Genet Dev. 2001;11:547–53.

616. Chen J, Röcken C, Lofton-Day C, Schulz H-U, Müller O, Kutzner N, et al. Molecular analysis of APC promoter methylation and protein expression in colorectal cancer

208

metastasis. Carcinogenesis. 2005;26(1):37–43.

617. Derks S, Postma C, Carvalho B, van den Bosch SM, Moerkerk PTM, Herman JG, et al. Integrated analysis of chromosomal, microsatellite and epigenetic instability in colorectal cancer identifies specific associations between promoter methylation of pivotal tumour suppressor and DNA repair genes and specific chromosomal alterations. Carcinogenesis. 2008;29(2):434–9.

618. Kumar K, Brim H, Giardiello F, Smoot DT, Nouraie M, Lee EL, et al. Distinct BRAF (V600E) and KRAS mutations in high microsatellite Instability sporadic colorectal cancer in African Americans. Clin Cancer Res. 2009;15(4):1155–61.

619. Naghibalhossaini F, Zamani M, Mokarram P, Khalili I, Rasti M, Mostafavi-Pour Z. Epigenetic and genetic analysis of WNT signaling pathway in sporadic colorectal cancer patients from Iran. Mol Biol Rep. 2012;39(5):6171–8.

620. Gay LJ, Mitrou PN, Keen J, Bowman R, Naguib A, Cooke J, et al. Dietary, lifestyle and clinicopathological factors associated with APC mutations and promoter methylation in colorectal cancers from the EPIC-Norfolk study. J Pathol. 2012;228(3):405–15.

621. Goel A, Nagasaka T, Arnold CN, Inoue T, Hamilton C, Niedzwiecki D, et al. The CpG Island Methylator Phenotype and Chromosomal Instability Are Inversely Correlated in Sporadic Colorectal Cancer. Gastroenterology. 2007;132(1):127–38.

622. Kim JC, Choi JS, Roh S, Cho DH, Kim TW, Kim YS. Promoter methylation of specific genes is associated with the phenotype and progression of colorectal adenocarcinomas. Ann Surg Oncol. 2010;17(7):1767–76.

623. Zhai Y, Wu R, Schwartz D, Darrah D, Reed H, Kolligs F, et al. Role of beta-catenin/T-cell factor-regulated genes in ovarian endometrioid adenocarcinomas. Am J Pathol. 2002;160(4):1229–38.

624. Kolligs FT, Nieman MT, Winer I, Hu G, van Mater D, Feng Y, et al. ITF-2, a downstream target of the Wnt/TCF pathway, is activated in human cancers with beta-catenin defects and promotes neoplastic transformation. Cancer Cell. 2002;1(2):145–55.

625. Herbst A, Bommer GT, Kriegl L, Jung A, Behrens A, Csanadi E, et al. ITF-2 Is Disrupted via Allelic Loss of Chromosome 18q21, and ITF-2B Expression Is Lost at the Adenoma-Carcinoma Transition. Gastroenterology. 2009;137(2):639–48.

626. Joo JK, Kim SH, Kim HG, Kim DY, Ryu SY, Lee KH, et al. CpG methylation of transcription factor 4 in gastric carcinoma. Ann Surg Oncol. 2010;17(12):3344–53.

627. Kim SK, Jang HR, Kim JH, Kim M, Noh SM, Song KS, et al. CpG methylation in exon 1 of transcription factor 4 increases with age in normal gastric mucosa and is associated with gene silencing in intestinal-type gastric cancers. Carcinogenesis. 2008;29(8):1623–31.

628. Rawson JB, Manno M, Mrkonjic M, Daftary D, Dicks E, Buchanan DD, et al. Promoter

209

methylation of Wnt antagonists DKK1 and SFRP1 is associated with opposing tumor subtypes in two large populations of colorectal cancer patients. Carcinogenesis. 2011;32(5):741–7.

629. Rawson JB, Mrkonjic M, Daftary D, Dicks E, Buchanan DD, Younghusband HB, et al. Promoter methylation of Wnt5a is associated with microsatellite instability and BRAF V600E mutation in two large populations of colorectal cancer patients. Br J Cancer. 2011;104(12):1906–12.

630. Green RC, Green JS, Buehler SK, Robb JD, Daftary D, Gallinger S, et al. Very high incidence of familial colorectal cancer in Newfoundland: A comparison with Ontario and 13 other population-based studies. Fam Cancer. 2007;6(1):53–62.

631. Woods MO, Hyde AJ, Curtis FK, Stuckless S, Green JS, Pollett AF, et al. High frequency of hereditary colorectal cancer in Newfoundland likely involves novel susceptibility genes. Clin Cancer Res. 2005;11(19):6853–61.

632. Savio A, Daftary D, Dicks E, Buchanan DD, Parfrey PS, Young JP, et al. Promoter methylation of ITF2, but not APC, is associated with microsatellite instability in two populations of colorectal cancer patients. BMC Cancer. 2016;16:113.

633. Herbst A, Helferich S, Behrens A, Göke B, Kolligs FT. The transcription factor ITF-2A induces cell cycle arrest via p21Cip1. Biochem Biophys Res Commun. 2009;387(4):736–40.

634. Ding Z, Jiang T, Piao Y, Han T, Han Y, Xie X. Meta-Analysis of the association between APC promoter methylation and colorectal cancer. Onco Targets Ther. 2015;8:211–22.

635. Deng G, Song G, Pong E, Sleisenger M, Kim Y. Promoter methylation inhibits APC gene expression by causing changes in chromatin conformation and interfering with the binding of transcription factor CCAAT-binding factor. Cancer Res. 2004;64(415):2692–8.

636. Yang Q, Dong Y, Wu W, Zhu C, Chong H, Lu J, et al. Detection and differential diagnosis of colon cancer by a cumulative analysis of promoter methylation. Nat Commun. 2012;3:1206.

637. Murakami T, Mitomi H, Saito T, Takahashi M, Sakamoto N, Fukui N, et al. Distinct WNT/β-catenin signaling activation in the serrated neoplasia pathway and the adenoma-carcinoma sequence of the colorectum. Mod Pathol. 2014;28(1):1–13.

638. Azuara D, Rodriguez-Moranta F, de Oca J, Soriano-Izquierdo A, Mora J, Guardiola J, et al. Novel methylation panel for the early detection of colorectal tumors in stool DNA. Clin Colorectal Cancer. 2010;9(3):168–76.

639. Bo BL, Eun JL, Eun HJ, Chun HK, Dong KC, Sang YS, et al. Aberrant methylation of APC, MGMT, RASSF2A, and Wif-1 genes in plasma as a biomarker for early detection of colorectal cancer. Clin Cancer Res. 2009;15(19):6185–91.

640. Rasmussen SL, Krarup HB, Sunesen KG, Pedersen IS, Madsen PH, Thorlacius-Ussing O.

210

Hypermethylated DNA, a Biomarker for colorectal cancer: A systematic review. Color Dis. 2016;18:549–61.

641. Gao J, Aksoy B, Dogrusov U, Dresdner G, Gross B, Sumer S, et al. Integrative analysis of complex cancer genomics and clinical profiles using cBioPortal. Sci Signal. 2013;6(269):pl1.

642. Cerami E, Gao J, Dogrusoz U, Gross BE, Onur S, Larsson E, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012;2(5):401–4.

643. Volkov P, Olsson AH, Gillberg L, Jørgensen SW, Brøns C, Eriksson K-F, et al. A Genome-Wide mQTL Analysis in Human Adipose Tissue Identifies Genetic Variants Associated with DNA Methylation, Gene Expression and Metabolic Traits. PLoS One. 2016;11(6):e0157776.

644. Do C, Lang CF, Lin J, Darbary H, Krupska I, Gaba A, et al. Mechanisms and Disease Associations of Haplotype-Dependent Allele-Specific DNA Methylation. Am J Hum Genet. 2016;98(5):934–55.

645. Nordhoff E, Krogsdam a M, Jorgensen HF, Kallipolitis BH, Clark BF, Roepstorff P, et al. Rapid identification of DNA-binding proteins by mass spectrometry. Nat Biotechnol. 1999;17(9):884–8.

646. Tran DH, Shishido Y, Chung SP, Trinh HTT, Yorita K, Sakai T, et al. Identification of DNA-binding proteins that interact with the 5’-flanking region of the human D-amino acid oxidase gene by pull-down assay coupled with two-dimensional gel electrophoresis and mass spectrometry. J Pharm Biomed Anal. 2015;116:94–100.

647. Drewett V, Molina H, Millar A, Muller S, von Hesler F, Shaw PE. DNA-bound transcription factor complexes analysed by mass-spectrometry: binding of novel proteins to the human c-fos SRE and related sequences. Nucleic Acids Res. 2001;29(2):479–87.

648. Ernst J, Kheradpour P, Mikkelsen T, Shoresh N, Ward L, Epstein C, et al. Mapping and Analysis of Chromatin State Dynamics in Nine Human Cell Types. Nature. 2011;473(7345):43–9.

649. Ernst J, Kellis M. Discovery and characterization of chromatin states for systematic annotation of the human genome. Nat Biotechnol. 2010;28(8):817–25.

650. Buchanan DD, Tan YY, Walsh MD, Clendenning M, Metcalf AM, Ferguson K, et al. Tumor mismatch repair immunohistochemistry and DNA MLH1 methylation testing of patients with endometrial cancer diagnosed at age younger than 60 years optimizes triage for population-level germline mismatch repair gene mutation testing. J Clin Oncol. 2014;32(2):90–100.

651. Li Y, Yang Y, Lu Y, Herman J, Brock M, Zhao P, et al. Predictive value of CHFR and MLH1 methylation in human gastric cancer. Gastric Cancer. 2015;18(2):280–7.

211

652. Shilpa V, Bhagat R, Premalata C, Pallavi V, Krishnamoorthy L. Microsatellite instability, promoter methylation and protein expression of the DNA mismatch repair genes in epithelial ovarian cancer. Genomics. 2014;104(4):257–63.

653. Wang D, Liu X, Zhou Y, Xie H, Hong X, Tsai HJ, et al. Individual variation and longitudinal pattern of genome-wide DNA methylation from birth to the first two years of life. Epigenetics. 2012;7(6):594–605.

654. Zhou D, Tang W, Wang W, Pan X, An H-X, Zhang Y. Association between aberrant APC promoter methylation and breast cancer pathogenesis: a meta-analysis of 35 observational studies. PeerJ. 2016;4:e2203.

655. Liu L, Kron KJ, Pethe V V., Demetrashvili N, Nesbitt ME, Trachtenberg J, et al. Association of tissue promoter methylation levels of APC, TGFβ2, HOXD3 and RASSF1A with prostate cancer progression. Int J Cancer. 2011;129(10):2454–62.

656. Lao V, Grady WM. Epigenetics and colorectal cancer. Nat Rev Gastroenterol Hepatol. 2011;8(12):686–700.

657. Herman JG, Merlo A, Mao L, Herman G, Lapidus G, Issa J, et al. Inactivation of the CDKN2 / p16 / MTS1 Gene Is Frequently Associated with Aberrant DNA Methylation in All Common Human Cancers DNA Methylation in All Common Human Cancers. Cancer. 1995;55(1):4525–30.

658. Assasi N, Blackhouse G, Campbell K, Weeks L, Levine M. Mismatch Repair Deficiency Testing for Patients with Colorectal Cancer+: A Clinical and Evaluation. CADTH Optim Use Rep. 2015;5(3a):1–54.

659. Imperiale TF, Ransohoff DF, Itzkowitz SH, Levin TR, Lavin P, Lidgard GP, et al. Multitarget stool DNA testing for colorectal-cancer screening. N Engl J Med. 2014;370(14):1287–97.

660. A Stool DNA Test ( Cologuard ) for Colorectal Cancer Screening. JAMA. 2014;312(23):2566.

661. Church TR, Wandell M, Lofton-Day C, Mongin SJ, Burger M, Payne SR, et al. Prospective evaluation of methylated SEPT9 in plasma for detection of asymptomatic colorectal cancer. Gut. 2014;63(2):317–25.

662. Song L, Li Y. SEPT9: A Specific Circulating Biomarker for Colorectal Cancer. Adv Clin Chem. 2015;72:171–204.

663. Dahlin AM, Palmqvist R, Henriksson ML, Jacobsson M, Eklöf V, Rutegård J, et al. The role of the CpG island methylator phenotype in colorectal cancer prognosis depends on microsatellite instability screening status. Clin Cancer Res. 2010;16(6):1845–55.

664. Mengual-Ballester M, Pellicer-Franco E, Valero-Navarro G, Soria-Aledo V, García-Marín JA, Aguayo-Albasini JL. Increased survival and decreased recurrence in colorectal cancer patients diagnosed in a screening programme. Cancer Epidemiol. 2016;43:70–5.

212

665. Wolf RL, Basch CE, Brouse CH, Shmukler C, Shea S. Patient preferences and adherence to colorectal cancer screening in an urban population. Am J Public Health. 2006;96(5):809–11.

666. Kaminskas E, Farrell A, Wang Y-C, Sridhara R, Pazdur R. FDA Drug Approval Summary: Azacitidine (5-azacytidine, VidazaTM) for Injectable Suspension. Oncologist. 2005;10(3):176–82.

667. American Association for Cancer Research. AACR Cancer Progress Report 2015. Clin Cancer Res. 2015;21(Supplement 1):S1–128.

213

Appendices

Table A1. Methylation between SNP genotypes of rs1800734, rs749072, and rs13098279 in

controls by ANOVA. Mean β-value of each genotype for SNPs in controls from Ontario.

Chr 3 Location

Probe ID WT mean

WT SD

Het Mean

Het SD

Hom Mean

Hom SD

P-value

rs1800734 N = 530 N = 264 N = 53 37018029 cg21595053 0.988 0.005 0.989 0.005 0.989 0.005 0.808 37033373 cg02103401 0.746 0.081 0.701 0.092 0.657 0.102 5.99x10-18 37033625 cg24607398 0.885 0.057 0.855 0.066 0.836 0.070 3.91x10-17 37033632 cg10990993 0.866 0.053 0.834 0.063 0.810 0.067 3.50x10-21 37033791 cg04726821 0.220 0.067 0.186 0.061 0.150 0.061 1.51x10-22 37033894 cg11291081 0.066 0.031 0.058 0.024 0.048 0.020 1.11x10-6 37033903 cg05670953 0.167 0.069 0.147 0.063 0.120 0.058 2.43x10-8 37033980 cg18320188 0.072 0.021 0.067 0.019 0.061 0.017 2.45x10-5 37034028 cg04841293 0.015 0.004 0.014 0.003 0.015 0.004 0.124 37034066 cg05845319 0.029 0.008 0.028 0.008 0.028 0.010 0.345 37034084 cg21109167 0.121 0.047 0.111 0.045 0.106 0.048 0.005 37034142 cg03901257 0.013 0.004 0.013 0.004 0.013 0.003 0.736 37034154 cg02279071 0.009 0.003 0.009 0.003 0.009 0.003 0.604 37034166 cg14751544 0.022 0.007 0.022 0.006 0.021 0.006 0.517 37034346 cg16764580 0.024 0.010 0.024 0.012 0.025 0.012 0.743 37034441 cg01302270 0.020 0.006 0.021 0.007 0.020 0.006 0.188 37034473 cg17641046 0.044 0.016 0.045 0.015 0.045 0.015 0.361 37034495 cg07101782 0.009 0.003 0.009 0.001 0.009 0.001 0.990 37034654 cg03497419 0.012 0.007 0.012 0.009 0.011 0.007 0.838 37034661 cg27586588 0.010 0.004 0.010 0.005 0.010 0.003 0.571 37034693 cg16433211 0.021 0.004 0.021 0.004 0.021 0.005 0.121 37034730 cg10769891 0.005 0.003 0.005 0.002 0.005 0.001 0.910 37034739 cg19132762 0.016 0.007 0.016 0.007 0.016 0.006 0.844 37034787 cg23658326 0.016 0.006 0.016 0.002 0.016 0.002 0.825 37034814 cg11600697 0.044 0.012 0.044 0.013 0.046 0.015 0.490 37034825 cg21490561 0.010 0.003 0.010 0.003 0.009 0.004 0.755 37034840 cg00893636 0.019 0.004 0.019 0.004 0.019 0.004 0.821 37034909 cg03192963 0.013 0.003 0.014 0.004 0.013 0.004 0.035 37034956 cg06791151 0.012 0.002 0.011 0.002 0.012 0.003 0.579 37034997 cg07064226 0.035 0.040 0.037 0.048 0.036 0.047 0.852 37035063 cg06108510 0.015 0.012 0.017 0.018 0.015 0.012 0.386 37035090 cg24985459 0.033 0.016 0.034 0.020 0.033 0.016 0.755 37035117 cg12790037 0.027 0.007 0.027 0.007 0.026 0.007 0.581 37035158 cg25202636 0.034 0.036 0.038 0.052 0.032 0.031 0.269 37035168 cg17621259 0.011 0.003 0.011 0.002 0.011 0.002 0.684 37035200 cg14671526 0.020 0.005 0.020 0.005 0.020 0.005 0.843 37035205 cg05906740 0.016 0.003 0.016 0.003 0.016 0.003 0.840

214

37035207 cg27331401 0.043 0.012 0.044 0.011 0.044 0.012 0.154 37035220 cg25837710 0.011 0.003 0.011 0.002 0.011 0.002 0.683 37035222 cg12851504 0.021 0.005 0.022 0.005 0.022 0.005 0.319 37035228 cg06590608 0.012 0.003 0.012 0.002 0.013 0.002 0.988 37035282 cg11224603 0.017 0.003 0.017 0.003 0.017 0.003 0.992 37035345 cg19208331 0.028 0.006 0.028 0.006 0.028 0.006 0.957 37035355 cg14598950 0.008 0.002 0.009 0.022 0.008 0.002 0.354 37035399 cg13846866 0.068 0.093 0.065 0.093 0.065 0.087 0.881 37036726 cg04777024 0.969 0.011 0.970 0.010 0.970 0.009 0.553 37038591 cg17024523 0.977 0.012 0.977 0.012 0.978 0.011 0.411 37048044 ch.3.753362R 0.093 0.032 0.094 0.029 0.092 0.028 0.837 37055414 cg25212762 0.968 0.020 0.967 0.022 0.968 0.022 0.889 37082315 cg11363877 0.990 0.004 0.990 0.004 0.990 0.004 0.952 37082380 cg03405026 0.987 0.004 0.987 0.004 0.986 0.004 0.690 37092193 cg16863190 0.889 0.125 0.901 0.113 0.903 0.108 0.364 37095036 cg27373390 0.979 0.009 0.977 0.022 0.978 0.008 0.106 37152029 cg01934787 0.912 0.046 0.907 0.050 0.910 0.047 0.451 37173546 cg06284479 0.983 0.007 0.982 0.007 0.981 0.006 0.023 37179823 cg24305555 0.979 0.009 0.979 0.008 0.978 0.009 0.588 37204814 cg05433805 0.589 0.129 0.572 0.118 0.557 0.087 0.059 37212084 cg15934958 0.925 0.038 0.917 0.039 0.903 0.038 3.34x10-4 37216510 cg06734169 0.039 0.024 0.041 0.028 0.039 0.025 0.699 37217087 cg12792366 0.033 0.043 0.034 0.047 0.032 0.047 0.945 37217675 cg00747698 0.033 0.008 0.035 0.008 0.034 0.008 0.019 37217993 cg22221026 0.012 0.002 0.012 0.002 0.012 0.002 0.582 37217996 cg11574180 0.027 0.005 0.027 0.005 0.028 0.006 0.360 37218128 cg09310383 0.055 0.011 0.056 0.012 0.055 0.012 0.371 37218150 cg15011249 0.025 0.007 0.025 0.007 0.024 0.006 0.882 37218212 cg17479303 0.021 0.011 0.022 0.016 0.021 0.014 0.536 37218771 cg06853609 0.058 0.059 0.062 0.062 0.067 0.078 0.514 37219077 cg22985146 0.556 0.084 0.584 0.076 0.588 0.082 1.01x10-5 37225266 cg12999063 0.985 0.007 0.973 0.052 0.949 0.145 2.67x10-8 37239890 cg11321190 0.710 0.087 0.710 0.094 0.724 0.097 0.547

rs749072 N = 438 N=271 N=57 37018029 cg21595053 0.988 0.005 0.989 0.005 0.988 0.005 0.781 37033373 cg02103401 0.746 0.082 0.707 0.090 0.666 0.102 1.47x10-13 37033625 cg24607398 0.884 0.058 0.860 0.066 0.841 0.067 1.86x10-9 37033632 cg10990993 0.864 0.053 0.842 0.064 0.820 0.069 1.02x10-9 37033791 cg04726821 0.218 0.068 0.191 0.063 0.163 0.065 6.33x10-12 37033894 cg11291081 0.065 0.031 0.059 0.024 0.053 0.026 0.001 37033903 cg05670953 0.165 0.069 0.150 0.063 0.127 0.063 2.58x10-5 37033980 cg18320188 0.072 0.022 0.068 0.018 0.061 0.018 2.15x10-5 37034028 cg04841293 0.015 0.004 0.014 0.004 0.014 0.005 0.360 37034066 cg05845319 0.029 0.008 0.028 0.008 0.028 0.010 0.600 37034084 cg21109167 0.120 0.046 0.113 0.046 0.105 0.047 0.025 37034142 cg03901257 0.013 0.004 0.013 0.004 0.013 0.004 0.895

215

37034154 cg02279071 0.009 0.003 0.009 0.003 0.009 0.003 0.940 37034166 cg14751544 0.023 0.008 0.022 0.006 0.022 0.007 0.297 37034346 cg16764580 0.024 0.011 0.025 0.011 0.026 0.013 0.456 37034441 cg01302270 0.020 0.007 0.020 0.007 0.021 0.006 0.462 37034473 cg17641046 0.043 0.016 0.045 0.016 0.045 0.015 0.359 37034495 cg07101782 0.009 0.004 0.009 0.001 0.009 0.001 0.976 37034654 cg03497419 0.012 0.007 0.012 0.009 0.013 0.012 0.657 37034661 cg27586588 0.010 0.004 0.010 0.004 0.010 0.004 0.616 37034693 cg16433211 0.021 0.004 0.021 0.004 0.020 0.004 0.080 37034730 cg10769891 0.005 0.003 0.005 0.001 0.005 0.001 0.645 37034739 cg19132762 0.016 0.007 0.017 0.007 0.015 0.006 0.364 37034787 cg23658326 0.016 0.007 0.016 0.002 0.015 0.002 0.739 37034814 cg11600697 0.043 0.012 0.044 0.013 0.044 0.013 0.879 37034825 cg21490561 0.010 0.003 0.010 0.003 0.010 0.004 0.866 37034840 cg00893636 0.019 0.004 0.018 0.004 0.019 0.004 0.195 37034909 cg03192963 0.013 0.003 0.014 0.004 0.013 0.003 0.003 37034956 cg06791151 0.012 0.002 0.011 0.002 0.011 0.002 0.493 37034997 cg07064226 0.034 0.039 0.037 0.046 0.041 0.058 0.453 37035063 cg06108510 0.015 0.012 0.017 0.016 0.017 0.019 0.401 37035090 cg24985459 0.033 0.016 0.034 0.019 0.034 0.022 0.625 37035117 cg12790037 0.027 0.007 0.027 0.007 0.026 0.007 0.856 37035158 cg25202636 0.033 0.035 0.038 0.048 0.040 0.056 0.268 37035168 cg17621259 0.011 0.003 0.011 0.002 0.011 0.002 0.589 37035200 cg14671526 0.020 0.005 0.020 0.004 0.020 0.005 0.976 37035205 cg05906740 0.016 0.003 0.016 0.003 0.016 0.003 0.527 37035207 cg27331401 0.043 0.012 0.044 0.011 0.043 0.012 0.199 37035220 cg25837710 0.011 0.003 0.011 0.002 0.011 0.002 0.670 37035222 cg12851504 0.021 0.005 0.022 0.005 0.022 0.005 0.736 37035228 cg06590608 0.012 0.003 0.013 0.002 0.013 0.002 0.879 37035282 cg11224603 0.017 0.003 0.017 0.003 0.017 0.003 0.966 37035345 cg19208331 0.028 0.006 0.028 0.006 0.028 0.006 0.818 37035355 cg14598950 0.008 0.002 0.009 0.022 0.008 0.002 0.360 37035399 cg13846866 0.068 0.092 0.068 0.094 0.066 0.090 0.995 37036726 cg04777024 0.969 0.011 0.970 0.010 0.970 0.009 0.319 37038591 cg17024523 0.976 0.012 0.978 0.012 0.978 0.011 0.300 37048044 ch.3.753362R 0.093 0.032 0.094 0.029 0.088 0.027 0.464 37055414 cg25212762 0.969 0.017 0.968 0.021 0.965 0.025 0.318 37082315 cg11363877 0.990 0.004 0.990 0.004 0.990 0.003 0.824 37082380 cg03405026 0.987 0.004 0.987 0.004 0.986 0.004 0.289 37092193 cg16863190 0.889 0.125 0.894 0.118 0.911 0.106 0.459 37095036 cg27373390 0.979 0.007 0.977 0.021 0.979 0.008 0.081 37152029 cg01934787 0.913 0.043 0.907 0.049 0.907 0.049 0.158 37173546 cg06284479 0.983 0.007 0.982 0.007 0.981 0.006 0.017 37179823 cg24305555 0.980 0.007 0.979 0.008 0.978 0.009 0.291 37204814 cg05433805 0.590 0.133 0.578 0.116 0.555 0.084 0.100 37212084 cg15934958 0.926 0.036 0.918 0.035 0.904 0.033 2.01x10-5

216

37216510 cg06734169 0.038 0.023 0.041 0.028 0.039 0.027 0.286 37217087 cg12792366 0.034 0.043 0.033 0.043 0.038 0.063 0.788 37217675 cg00747698 0.033 0.008 0.035 0.008 0.034 0.008 0.083 37217993 cg22221026 0.012 0.002 0.012 0.002 0.012 0.002 0.280 37217996 cg11574180 0.027 0.005 0.027 0.005 0.028 0.007 0.437 37218128 cg09310383 0.055 0.011 0.056 0.012 0.055 0.013 0.141 37218150 cg15011249 0.025 0.007 0.025 0.007 0.024 0.006 0.523 37218212 cg17479303 0.021 0.011 0.022 0.015 0.023 0.017 0.371 37218771 cg06853609 0.057 0.056 0.064 0.062 0.066 0.081 0.242 37219077 cg22985146 0.554 0.085 0.582 0.077 0.586 0.076 7.26x10-6 37225266 cg12999063 0.985 0.005 0.975 0.050 0.968 0.063 7.11x10-6 37239890 cg11321190 0.707 0.088 0.714 0.092 0.723 0.091 0.348

rs13098279 N = 491 N = 233 N = 42 37018029 cg21595053 0.988 0.005 0.988 0.005 0.989 0.005 0.845 37033373 cg02103401 0.746 0.081 0.698 0.091 0.644 0.099 3.04x10-17 37033625 cg24607398 0.885 0.057 0.855 0.067 0.828 0.067 6.43x1016 37033632 cg10990993 0.866 0.053 0.834 0.063 0.807 0.069 2.32x10-18 37033791 cg04726821 0.219 0.068 0.183 0.058 0.150 0.063 3.82x10-20 37033894 cg11291081 0.065 0.031 0.058 0.023 0.048 0.021 8.51x10-5 37033903 cg05670953 0.165 0.068 0.146 0.062 0.117 0.060 7.51x10-7 37033980 cg18320188 0.072 0.021 0.067 0.018 0.059 0.018 2.37x10-5 37034028 cg04841293 0.015 0.004 0.014 0.004 0.014 0.004 0.369 37034066 cg05845319 0.029 0.008 0.028 0.008 0.028 0.011 0.542 37034084 cg21109167 0.120 0.046 0.111 0.044 0.106 0.049 0.026 37034142 cg03901257 0.013 0.004 0.013 0.004 0.013 0.003 0.953 37034154 cg02279071 0.009 0.003 0.009 0.003 0.009 0.003 0.454 37034166 cg14751544 0.023 0.007 0.022 0.007 0.021 0.005 0.399 37034346 cg16764580 0.024 0.011 0.025 0.012 0.025 0.011 0.787 37034441 cg01302270 0.020 0.006 0.021 0.007 0.021 0.006 0.151 37034473 cg17641046 0.043 0.016 0.045 0.016 0.045 0.015 0.225 37034495 cg07101782 0.009 0.003 0.009 0.001 0.009 0.001 0.989 37034654 cg03497419 0.012 0.007 0.012 0.010 0.011 0.008 0.723 37034661 cg27586588 0.010 0.004 0.010 0.005 0.009 0.003 0.345 37034693 cg16433211 0.021 0.004 0.021 0.004 0.020 0.004 0.108 37034730 cg10769891 0.005 0.003 0.005 0.001 0.005 0.001 0.867 37034739 cg19132762 0.016 0.007 0.016 0.007 0.016 0.006 0.582 37034787 cg23658326 0.016 0.006 0.016 0.002 0.015 0.003 0.812 37034814 cg11600697 0.043 0.012 0.044 0.013 0.045 0.014 0.570 37034825 cg21490561 0.010 0.003 0.010 0.003 0.009 0.004 0.886 37034840 cg00893636 0.019 0.004 0.019 0.004 0.018 0.004 0.795 37034909 cg03192963 0.013 0.003 0.014 0.004 0.013 0.004 0.030 37034956 cg06791151 0.012 0.002 0.011 0.002 0.012 0.003 0.521 37034997 cg07064226 0.035 0.039 0.038 0.051 0.037 0.052 0.623 37035063 cg06108510 0.015 0.012 0.017 0.018 0.014 0.012 0.241 37035090 cg24985459 0.033 0.016 0.034 0.021 0.032 0.017 0.602 37035117 cg12790037 0.027 0.007 0.027 0.007 0.026 0.008 0.688

217

37035158 cg25202636 0.033 0.035 0.040 0.054 0.033 0.034 0.149 37035168 cg17621259 0.011 0.003 0.011 0.002 0.011 0.002 0.535 37035200 cg14671526 0.020 0.005 0.020 0.005 0.020 0.005 0.957 37035205 cg05906740 0.016 0.003 0.016 0.003 0.016 0.003 0.598 37035207 cg27331401 0.043 0.012 0.044 0.011 0.044 0.013 0.177 37035220 cg25837710 0.011 0.003 0.011 0.002 0.011 0.002 0.913 37035222 cg12851504 0.021 0.005 0.022 0.005 0.022 0.005 0.404 37035228 cg06590608 0.013 0.003 0.013 0.002 0.013 0.002 0.992 37035282 cg11224603 0.017 0.003 0.017 0.003 0.017 0.003 0.987 37035345 cg19208331 0.028 0.006 0.028 0.006 0.028 0.006 0.950 37035355 cg14598950 0.008 0.002 0.009 0.023 0.008 0.002 0.300 37035399 cg13846866 0.069 0.093 0.066 0.094 0.064 0.085 0.912 37036726 cg04777024 0.969 0.011 0.970 0.010 0.969 0.010 0.378 37038591 cg17024523 0.976 0.012 0.978 0.012 0.978 0.010 0.336 37048044 ch.3.753362R 0.093 0.031 0.093 0.029 0.092 0.029 0.954 37055414 cg25212762 0.969 0.017 0.967 0.022 0.967 0.024 0.723 37082315 cg11363877 0.990 0.004 0.990 0.004 0.990 0.003 0.674 37082380 cg03405026 0.987 0.004 0.987 0.004 0.986 0.004 0.794 37092193 cg16863190 0.888 0.125 0.899 0.115 0.909 0.102 0.339 37095036 cg27373390 0.979 0.007 0.977 0.022 0.979 0.008 0.130 37152029 cg01934787 0.912 0.043 0.907 0.051 0.908 0.049 0.331 37173546 cg06284479 0.983 0.007 0.982 0.007 0.981 0.006 0.044 37179823 cg24305555 0.979 0.007 0.979 0.008 0.978 0.009 0.447 37204814 cg05433805 0.590 0.131 0.574 0.116 0.554 0.087 0.093 37212084 cg15934958 0.925 0.036 0.918 0.036 0.902 0.030 1.07x10-4 37216510 cg06734169 0.039 0.023 0.041 0.029 0.039 0.027 0.433 37217087 cg12792366 0.034 0.043 0.034 0.048 0.032 0.051 0.948 37217675 cg00747698 0.033 0.008 0.035 0.008 0.034 0.008 0.043 37217993 cg22221026 0.012 0.002 0.012 0.002 0.012 0.002 0.532 37217996 cg11574180 0.027 0.005 0.027 0.005 0.028 0.007 0.398 37218128 cg09310383 0.055 0.011 0.057 0.012 0.053 0.012 0.082 37218150 cg15011249 0.025 0.007 0.025 0.007 0.024 0.007 0.841 37218212 cg17479303 0.021 0.011 0.022 0.016 0.022 0.015 0.427 37218771 cg06853609 0.058 0.057 0.063 0.063 0.067 0.079 0.418 37219077 cg22985146 0.556 0.084 0.583 0.076 0.591 0.076 2.12x10-5 37225266 cg12999063 0.985 0.005 0.973 0.053 0.962 0.072 7.83x10-8 37239890 cg11321190 0.711 0.088 0.707 0.093 0.718 0.097 0.708 Significant results are bolded when P<7.14x10-4. SD – standard deviation; WT – wildtype; Het – heterozygous; Hom – homozygous variant

218

Table A2. Methylation between SNP genotypes of rs1800734, rs749072, and rs13098279 in

CRC cases by ANOVA. Mean β-value of each genotype for SNPs in CRC cases from Ontario.

Chr 3 Location

Probe ID WT mean

WT SD

Het Mean

Het SD

Hom Mean

Hom SD

P-value

rs1800734 N=519 N=321 N=43 37018029 cg21595053 0.988 0.005 0.988 0.006 0.990 0.005 0.133 37033373 cg02103401 0.742 0.082 0.701 0.096 0.652 0.094 4.12x10-29

37033625 cg24607398 0.884 0.056 0.849 0.068 0.828 0.067 6.62x10-33

37033632 cg10990993 0.858 0.054 0.823 0.066 0.811 0.060 2.53x10-38 37033791 cg04726821 0.207 0.070 0.163 0.062 0.134 0.049 3.79x10-40 37033894 cg11291081 0.065 0.033 0.056 0.029 0.048 0.024 6.81x10-6 37033903 cg05670953 0.156 0.070 0.127 0.067 0.107 0.051 1.09x10-10 37033980 cg18320188 0.072 0.026 0.062 0.019 0.059 0.013 1.56x10-9 37034028 cg04841293 0.014 0.004 0.014 0.004 0.013 0.003 0.425 37034066 cg05845319 0.028 0.008 0.026 0.008 0.027 0.007 0.103 37034084 cg21109167 0.109 0.045 0.095 0.044 0.099 0.040 7.50x10-5 37034142 cg03901257 0.013 0.004 0.013 0.004 0.013 0.003 0.765 37034154 cg02279071 0.010 0.003 0.009 0.003 0.009 0.003 0.937 37034166 cg14751544 0.022 0.006 0.022 0.006 0.023 0.006 0.597 37034346 cg16764580 0.024 0.010 0.023 0.009 0.025 0.012 0.169 37034441 cg01302270 0.020 0.006 0.021 0.008 0.022 0.009 0.460 37034473 cg17641046 0.044 0.016 0.044 0.016 0.050 0.023 0.056 37034495 cg07101782 0.009 0.001 0.009 0.001 0.010 0.003 0.012 37034654 cg03497419 0.012 0.010 0.013 0.011 0.013 0.008 0.770 37034661 cg27586588 0.010 0.004 0.010 0.006 0.010 0.004 0.703 37034693 cg16433211 0.020 0.003 0.020 0.003 0.021 0.003 0.263 37034730 cg10769891 0.005 0.002 0.005 0.002 0.006 0.003 0.300 37034739 cg19132762 0.015 0.004 0.015 0.004 0.016 0.005 0.237 37034787 cg23658326 0.016 0.002 0.015 0.002 0.016 0.002 0.141 37034814 cg11600697 0.042 0.012 0.041 0.012 0.045 0.015 0.181 37034825 cg21490561 0.010 0.003 0.010 0.003 0.009 0.003 0.748 37034840 cg00893636 0.019 0.005 0.019 0.005 0.018 0.005 0.429 37034909 cg03192963 0.014 0.004 0.014 0.004 0.015 0.004 0.242 37034956 cg06791151 0.012 0.002 0.011 0.002 0.012 0.003 0.339 37034997 cg07064226 0.034 0.044 0.036 0.045 0.037 0.041 0.756 37035063 cg06108510 0.016 0.020 0.017 0.027 0.016 0.014 0.690 37035090 cg24985459 0.033 0.020 0.035 0.026 0.035 0.018 0.397 37035117 cg12790037 0.028 0.007 0.028 0.006 0.028 0.009 0.957 37035158 cg25202636 0.035 0.045 0.039 0.055 0.041 0.050 0.444 37035168 cg17621259 0.012 0.002 0.011 0.002 0.012 0.003 0.228 37035200 cg14671526 0.020 0.004 0.020 0.004 0.021 0.005 0.282 37035205 cg05906740 0.016 0.003 0.016 0.002 0.017 0.003 0.049 37035207 cg27331401 0.043 0.012 0.042 0.011 0.045 0.015 0.378 37035220 cg25837710 0.011 0.002 0.011 0.002 0.012 0.002 0.146 37035222 cg12851504 0.021 0.005 0.021 0.004 0.023 0.006 0.091

219

37035228 cg06590608 0.013 0.002 0.013 0.002 0.013 0.003 0.390 37035282 cg11224603 0.017 0.002 0.017 0.003 0.018 0.004 0.029 37035345 cg19208331 0.028 0.006 0.028 0.005 0.028 0.004 0.972 37035355 cg14598950 0.007 0.002 0.007 0.002 0.008 0.002 0.060 37035399 cg13846866 0.067 0.094 0.068 0.098 0.078 0.115 0.770 37036726 cg04777024 0.970 0.010 0.970 0.010 0.968 0.011 0.785 37038591 cg17024523 0.977 0.011 0.978 0.010 0.979 0.011 0.185 37048044 ch.3.753362R 0.093 0.028 0.091 0.027 0.092 0.024 0.571 37055414 cg25212762 0.965 0.034 0.962 0.036 0.966 0.022 0.563 37082315 cg11363877 0.989 0.004 0.989 0.004 0.990 0.004 0.787 37082380 cg03405026 0.987 0.004 0.987 0.004 0.987 0.004 0.707 37092193 cg16863190 0.893 0.123 0.899 0.124 0.896 0.123 0.783 37095036 cg27373390 0.980 0.009 0.979 0.008 0.978 0.007 0.436 37152029 cg01934787 0.905 0.055 0.898 0.064 0.892 0.055 0.123 37173546 cg06284479 0.982 0.007 0.981 0.008 0.981 0.007 0.200 37179823 cg24305555 0.979 0.012 0.979 0.010 0.978 0.007 0.763 37204814 cg05433805 0.584 0.143 0.564 0.122 0.557 0.112 0.083 37212084 cg15934958 0.918 0.040 0.910 0.039 0.895 0.045 1.11x10-4 37216510 cg06734169 0.041 0.043 0.041 0.039 0.044 0.034 0.880 37217087 cg12792366 0.033 0.048 0.040 0.069 0.036 0.052 0.213 37217675 cg00747698 0.034 0.009 0.033 0.008 0.036 0.008 0.150 37217993 cg22221026 0.012 0.002 0.012 0.002 0.012 0.002 0.400 37217996 cg11574180 0.027 0.005 0.026 0.005 0.027 0.006 0.053 37218128 cg09310383 0.055 0.011 0.054 0.010 0.056 0.011 0.206 37218150 cg15011249 0.025 0.006 0.025 0.006 0.024 0.008 0.947 37218212 cg17479303 0.021 0.015 0.022 0.020 0.020 0.010 0.768 37218771 cg06853609 0.054 0.060 0.059 0.069 0.061 0.060 0.416 37219077 cg22985146 0.543 0.083 0.561 0.090 0.586 0.076 4.05x10-4 37225266 cg12999063 0.984 0.008 0.975 0.049 0.984 0.005 1.70x10-5 37239890 cg11321190 0.711 0.096 0.713 0.097 0.741 0.080 0.141

rs749072 N=333 N=258 N=36 37018029 cg21595053 0.989 0.005 0.989 0.006 0.992 0.004 0.009 37033373 cg02103401 0.736 0.087 0.707 0.095 0.659 0.086 1.68x10-7

37033625 cg24607398 0.889 0.058 0.857 0.067 0.844 0.070 2.93x10-10

37033632 cg10990993 0.863 0.055 0.835 0.068 0.824 0.064 1.86x10-8

37033791 cg04726821 0.209 0.072 0.176 0.064 0.138 0.049 8.89x10-13

37033894 cg11291081 0.063 0.032 0.057 0.030 0.042 0.016 2.19x10-4

37033903 cg05670953 0.155 0.072 0.136 0.070 0.100 0.047 4.36x10-6

37033980 cg18320188 0.072 0.027 0.064 0.022 0.060 0.015 4.00x10-5

37034028 cg04841293 0.013 0.004 0.013 0.004 0.012 0.003 0.282 37034066 cg05845319 0.027 0.008 0.027 0.008 0.026 0.007 0.601 37034084 cg21109167 0.108 0.044 0.100 0.046 0.094 0.039 0.027 37034142 cg03901257 0.013 0.004 0.013 0.004 0.012 0.004 0.462 37034154 cg02279071 0.009 0.003 0.009 0.003 0.008 0.002 0.142 37034166 cg14751544 0.022 0.007 0.022 0.006 0.021 0.007 0.744 37034346 cg16764580 0.025 0.011 0.024 0.010 0.028 0.013 0.022

220

37034441 cg01302270 0.021 0.007 0.020 0.007 0.021 0.010 0.895 37034473 cg17641046 0.043 0.016 0.044 0.016 0.049 0.022 0.104 37034495 cg07101782 0.009 0.001 0.009 0.001 0.010 0.004 4.34x10-4 37034654 cg03497419 0.013 0.012 0.013 0.011 0.012 0.008 0.807 37034661 cg27586588 0.010 0.005 0.010 0.006 0.010 0.004 0.923 37034693 cg16433211 0.021 0.003 0.021 0.003 0.022 0.004 0.045 37034730 cg10769891 0.005 0.002 0.005 0.002 0.005 0.004 0.706 37034739 cg19132762 0.015 0.004 0.015 0.004 0.016 0.004 0.415 37034787 cg23658326 0.016 0.002 0.015 0.002 0.016 0.002 0.256 37034814 cg11600697 0.043 0.013 0.043 0.014 0.049 0.016 0.020 37034825 cg21490561 0.009 0.003 0.009 0.003 0.008 0.003 0.032 37034840 cg00893636 0.019 0.005 0.019 0.005 0.018 0.005 0.794 37034909 cg03192963 0.014 0.004 0.014 0.004 0.014 0.004 0.568 37034956 cg06791151 0.011 0.002 0.011 0.002 0.011 0.003 0.747 37034997 cg07064226 0.040 0.051 0.038 0.046 0.042 0.038 0.760 37035063 cg06108510 0.018 0.025 0.018 0.029 0.018 0.011 0.985 37035090 cg24985459 0.036 0.024 0.036 0.025 0.038 0.017 0.842 37035117 cg12790037 0.027 0.007 0.027 0.006 0.026 0.009 0.698 37035158 cg25202636 0.040 0.052 0.040 0.058 0.047 0.049 0.789 37035168 cg17621259 0.011 0.002 0.011 0.002 0.011 0.003 0.643 37035200 cg14671526 0.020 0.005 0.020 0.004 0.022 0.005 0.037 37035205 cg05906740 0.016 0.003 0.016 0.002 0.017 0.003 0.033 37035207 cg27331401 0.043 0.012 0.043 0.012 0.046 0.016 0.279 37035220 cg25837710 0.011 0.002 0.011 0.002 0.011 0.002 0.747 37035222 cg12851504 0.021 0.005 0.021 0.005 0.022 0.007 0.190 37035228 cg06590608 0.013 0.002 0.012 0.002 0.013 0.003 0.792 37035282 cg11224603 0.017 0.002 0.017 0.003 0.018 0.004 0.048 37035345 cg19208331 0.027 0.005 0.028 0.006 0.028 0.005 0.105 37035355 cg14598950 0.007 0.002 0.007 0.002 0.008 0.002 0.025 37035399 cg13846866 0.082 0.108 0.071 0.099 0.098 0.124 0.252 37036726 cg04777024 0.971 0.010 0.970 0.010 0.972 0.012 0.708 37038591 cg17024523 0.979 0.011 0.979 0.010 0.985 0.008 0.005 37048044 ch.3.753362R 0.096 0.030 0.092 0.025 0.096 0.023 0.276 37055414 cg25212762 0.961 0.040 0.961 0.038 0.958 0.025 0.927 37082315 cg11363877 0.990 0.004 0.990 0.004 0.991 0.003 0.169 37082380 cg03405026 0.987 0.004 0.987 0.004 0.987 0.003 0.623 37092193 cg16863190 0.873 0.135 0.892 0.124 0.841 0.149 0.056 37095036 cg27373390 0.979 0.010 0.979 0.008 0.977 0.007 0.550 37152029 cg01934787 0.898 0.061 0.896 0.066 0.875 0.060 0.118 37173546 cg06284479 0.983 0.007 0.982 0.008 0.982 0.007 0.565 37179823 cg24305555 0.978 0.014 0.978 0.010 0.974 0.014 0.205 37204814 cg05433805 0.586 0.145 0.576 0.126 0.573 0.105 0.615 37212084 cg15934958 0.916 0.043 0.911 0.040 0.887 0.047 0.001 37216510 cg06734169 0.046 0.052 0.043 0.040 0.052 0.041 0.529 37217087 cg12792366 0.040 0.058 0.040 0.066 0.043 0.053 0.938 37217675 cg00747698 0.034 0.009 0.034 0.008 0.036 0.008 0.343

221

37217993 cg22221026 0.012 0.002 0.012 0.002 0.011 0.002 0.686 37217996 cg11574180 0.027 0.005 0.026 0.005 0.027 0.006 0.304 37218128 cg09310383 0.056 0.011 0.055 0.010 0.059 0.010 0.043 37218150 cg15011249 0.024 0.007 0.024 0.006 0.024 0.007 0.928 37218212 cg17479303 0.022 0.017 0.022 0.021 0.023 0.010 0.993 37218771 cg06853609 0.062 0.067 0.061 0.068 0.068 0.051 0.839 37219077 cg22985146 0.546 0.085 0.570 0.085 0.577 0.079 0.001 37225266 cg12999063 0.984 0.009 0.976 0.044 0.983 0.005 0.003 37239890 cg11321190 0.706 0.101 0.716 0.097 0.756 0.085 0.013

rs13098279 N=387 N=217 N=23 37018029 cg21595053 0.989 0.005 0.989 0.006 0.992 0.004 0.063 37033373 cg02103401 0.735 0.086 0.700 0.096 0.653 0.098 5.72x10-8 37033625 cg24607398 0.886 0.059 0.854 0.067 0.841 0.079 1.19x10-9 37033632 cg10990993 0.861 0.057 0.831 0.067 0.816 0.067 1.59x10-9 37033791 cg04726821 0.208 0.071 0.170 0.063 0.124 0.043 0.001 37033894 cg11291081 0.062 0.031 0.056 0.031 0.041 0.018 1.40x10-6 37033903 cg05670953 0.154 0.070 0.132 0.071 0.089 0.036 6.16x10-5 37033980 cg18320188 0.072 0.027 0.063 0.021 0.060 0.015 0.966 37034028 cg04841293 0.013 0.004 0.013 0.004 0.013 0.004 0.686 37034066 cg05845319 0.027 0.007 0.027 0.008 0.027 0.008 0.966 37034084 cg21109167 0.108 0.043 0.099 0.047 0.096 0.041 0.043 37034142 cg03901257 0.012 0.004 0.013 0.004 0.012 0.004 0.576 37034154 cg02279071 0.009 0.003 0.009 0.003 0.008 0.003 0.450 37034166 cg14751544 0.022 0.007 0.022 0.006 0.021 0.007 0.852 37034346 cg16764580 0.025 0.011 0.024 0.010 0.029 0.015 0.030 37034441 cg01302270 0.020 0.006 0.020 0.008 0.023 0.011 0.337 37034473 cg17641046 0.043 0.016 0.044 0.016 0.053 0.024 0.030 37034495 cg07101782 0.009 0.001 0.009 0.001 0.010 0.004 4.93x10-4 37034654 cg03497419 0.013 0.012 0.013 0.012 0.013 0.009 0.983 37034661 cg27586588 0.010 0.005 0.010 0.006 0.011 0.005 0.786 37034693 cg16433211 0.021 0.003 0.021 0.004 0.021 0.003 0.456 37034730 cg10769891 0.005 0.002 0.005 0.002 0.005 0.004 0.485 37034739 cg19132762 0.015 0.004 0.015 0.003 0.017 0.004 0.052 37034787 cg23658326 0.016 0.002 0.015 0.002 0.016 0.003 0.121 37034814 cg11600697 0.043 0.013 0.042 0.013 0.051 0.016 0.012 37034825 cg21490561 0.009 0.003 0.009 0.003 0.008 0.003 0.256 37034840 cg00893636 0.019 0.005 0.019 0.005 0.019 0.006 0.968 37034909 cg03192963 0.014 0.004 0.014 0.003 0.014 0.005 0.616 37034956 cg06791151 0.011 0.002 0.011 0.002 0.011 0.002 0.635 37034997 cg07064226 0.039 0.050 0.039 0.047 0.045 0.044 0.853 37035063 cg06108510 0.018 0.023 0.019 0.031 0.018 0.012 0.905 37035090 cg24985459 0.035 0.023 0.037 0.027 0.038 0.018 0.708 37035117 cg12790037 0.027 0.007 0.027 0.006 0.026 0.010 0.887 37035158 cg25202636 0.039 0.050 0.042 0.061 0.048 0.056 0.645 37035168 cg17621259 0.011 0.002 0.011 0.002 0.011 0.003 0.684 37035200 cg14671526 0.020 0.004 0.020 0.004 0.021 0.005 0.279

222

37035205 cg05906740 0.016 0.003 0.016 0.002 0.017 0.003 0.016 37035207 cg27331401 0.043 0.012 0.042 0.012 0.047 0.019 0.163 37035220 cg25837710 0.011 0.002 0.011 0.002 0.011 0.002 0.509 37035222 cg12851504 0.021 0.005 0.021 0.005 0.023 0.007 0.208 37035228 cg06590608 0.013 0.002 0.012 0.002 0.013 0.004 0.497 37035282 cg11224603 0.017 0.002 0.017 0.003 0.019 0.004 0.039 37035345 cg19208331 0.027 0.006 0.028 0.005 0.028 0.005 0.784 37035355 cg14598950 0.007 0.002 0.007 0.002 0.008 0.002 0.043 37035399 cg13846866 0.081 0.107 0.071 0.099 0.097 0.131 0.381 37036726 cg04777024 0.970 0.010 0.971 0.010 0.972 0.011 0.666 37038591 cg17024523 0.979 0.011 0.979 0.010 0.984 0.008 0.051 37048044 ch.3.753362R 0.096 0.029 0.092 0.026 0.095 0.020 0.306 37055414 cg25212762 0.962 0.039 0.959 0.041 0.962 0.021 0.739 37082315 cg11363877 0.990 0.004 0.990 0.004 0.991 0.003 0.208 37082380 cg03405026 0.987 0.004 0.987 0.004 0.988 0.003 0.606 37092193 cg16863190 0.874 0.134 0.893 0.126 0.852 0.137 0.164 37095036 cg27373390 0.979 0.010 0.979 0.008 0.977 0.007 0.488 37152029 cg01934787 0.899 0.060 0.893 0.068 0.875 0.058 0.152 37173546 cg06284479 0.983 0.007 0.982 0.008 0.983 0.006 0.298 37179823 cg24305555 0.978 0.013 0.978 0.011 0.976 0.007 0.715 37204814 cg05433805 0.589 0.143 0.569 0.123 0.575 0.103 0.207 37212084 cg15934958 0.915 0.042 0.909 0.041 0.886 0.052 0.002 37216510 cg06734169 0.045 0.049 0.045 0.043 0.051 0.041 0.842 37217087 cg12792366 0.038 0.055 0.043 0.071 0.043 0.060 0.667 37217675 cg00747698 0.034 0.009 0.034 0.008 0.038 0.008 0.044 37217993 cg22221026 0.012 0.002 0.012 0.002 0.011 0.002 0.947 37217996 cg11574180 0.027 0.005 0.026 0.005 0.028 0.007 0.184 37218128 cg09310383 0.056 0.011 0.054 0.010 0.060 0.010 0.009 37218150 cg15011249 0.024 0.006 0.024 0.007 0.023 0.007 0.569 37218212 cg17479303 0.022 0.016 0.023 0.023 0.022 0.010 0.908 37218771 cg06853609 0.062 0.066 0.063 0.069 0.069 0.056 0.873 37219077 cg22985146 0.550 0.083 0.568 0.089 0.578 0.071 0.019 37225266 cg12999063 0.984 0.009 0.974 0.048 0.983 0.005 4.75x10-4 37239890 cg11321190 0.711 0.100 0.713 0.098 0.749 0.079 0.203 Significant results are bolded when P<7.14x10-4. SD – standard deviation; WT – wildtype; Het – heterozygous; Hom – homozygous variant

223

Table A3. Correlation between age and DNA methylation at MLH1 region. Partial

correlation, controlling for sex and peripheral blood mononuclear cell proportions, between age

and methylation in CRC cases and controls.

Probe ID Controls R P-value CRC Cases P-value cg21595053 -0.001 0.200 -0.052 0.141 cg02103401 -0.051 4.69x10-6 -0.094 0.008 cg24607398 0.029 0.004 -0.134 1.29x10-4 cg10990993 0.069 0.007 -0.147 2.83x10-5 cg04726821 -0.056 5.18x10-6 -0.174 6.57x10-7 cg11291081 -0.052 0.251 -0.024 0.502 cg05670953 -0.006 5.38x10-7 -0.135 1.12x10-4 cg18320188 -0.038 0.265 0.036 0.306 cg04841293 0.023 0.156 -0.026 0.461 cg05845319 0.034 0.537 -0.001 0.974 cg21109167 0.050 1.41x10-6 -0.169 1.41x10-6 cg03901257 -0.003 0.073 -0.075 0.034 cg02279071 0.042 0.432 0.012 0.725 cg14751544 -0.025 0.893 0.012 0.739 cg16764580 0.029 0.354 -0.003 0.925 cg01302270 0.049 0.782 0.070 0.046 cg17641046 0.040 0.064 0.071 0.044 cg07101782 0.024 0.810 0.057 0.107 cg03497419 0.057 0.865 -0.065 0.064 cg27586588 -0.026 0.619 -0.068 0.054 cg16433211 0.024 0.657 0.014 0.690 cg10769891 0.079 0.981 0.026 0.456 cg19132762 0.013 0.149 0.022 0.523 cg23658326 -0.006 0.408 0.013 0.713 cg11600697 -0.003 0.050 -0.015 0.679 cg21490561 0.028 0.113 0.075 0.032 cg00893636 0.138 0.145 0.047 0.183 cg03192963 -0.038 0.867 -0.020 0.576 cg06791151 0.038 0.277 -0.004 0.901 cg07064226 -0.019 0.514 -0.054 0.123 cg06108510 0.002 0.340 -0.027 0.439 cg24985459 0.002 0.155 -0.065 0.064 cg12790037 -0.036 0.944 0.108 0.002 cg25202636 -0.066 0.231 -0.034 0.330 cg17621259 -0.051 0.488 0.063 0.073 cg14671526 -0.126 0.420 -0.021 0.553 cg05906740 -0.105 0.169 -0.026 0.455 cg27331401 0.035 0.264 0.060 0.089 cg25837710 0.015 0.495 -0.030 0.394 cg12851504 0.054 0.108 0.017 0.627

224

cg06590608 0.032 0.461 0.034 0.332 cg11224603 0.054 0.500 -0.002 0.965 cg19208331 0.064 0.025 0.041 0.248 cg14598950 0.004 0.711 -0.013 0.710 cg13846866 0.026 0.862 -0.067 0.058 cg04777024 0.034 0.935 -0.014 0.696 cg17024523 -0.117 0.427 -0.028 0.419

ch.3.753362R 0.027 8.78x10-5 0.013 0.713 cg25212762 -0.237 0.288 0.034 0.332 cg11363877 -0.001 0.280 -0.052 0.142 cg03405026 -0.051 0.587 0.091 0.009 cg16863190 0.029 0.946 0.088 0.012 cg27373390 0.069 0.950 0.015 0.669 cg01934787 -0.056 0.310 0.033 0.350 cg06284479 -0.052 0.061 -0.109 0.002 cg24305555 -0.006 0.151 0.037 0.297 cg05433805 -0.038 3.53x10-4 -0.150 1.91x10-5 cg15934958 0.023 0.003 -0.098 0.005 cg06734169 0.034 0.317 -0.063 0.073 cg12792366 0.050 0.668 -0.072 0.040 cg00747698 -0.003 0.130 0.094 0.007 cg22221026 0.042 0.365 0.034 0.340 cg11574180 -0.025 0.127 -0.004 0.920 cg09310383 0.029 0.072 0.063 0.072 cg15011249 0.049 0.916 0.019 0.581 cg17479303 0.040 0.460 -0.042 0.238 cg06853609 0.024 0.331 -0.042 0.228 cg22985146 0.057 0.001 -0.133 1.53x10-4 cg12999063 -0.026 0.443 -0.036 0.300 cg11321190 0.024 1.12x10-11 -0.322 1.08x10-13

Significant results are bolded when P<7.14x10-4.

225

Table A4. Association between sex and methylation by logistic regression. Mean β-value is

shown for females and males along with logistic regression analysis at the MLH1 region.

Analysis of male versus female methylation is adjusted for age and peripheral blood

mononuclear cell proportions.

Probe ID Female Mean

Female SD

Male Mean

Male SD

P-value Effect Size

Lower 95% CI

Upper 95% CI

cg21595053 0.988 0.006 0.989 0.005 3.04x10-17 1.652 1.363 2.002 cg02103401 0.725 0.086 0.724 0.094 0.586 0.997 0.987 1.007 cg24607398 0.868 0.062 0.873 0.065 0.029 1.018 1.002 1.034 cg10990993 0.845 0.058 0.851 0.064 0.012 1.021 1.005 1.038 cg04726821 0.193 0.069 0.199 0.071 0.026 1.016 1.002 1.031 cg11291081 0.065 0.033 0.058 0.028 3.55x10-6 0.922 0.891 0.954 cg05670953 0.158 0.071 0.142 0.066 3.48x10-7 0.960 0.945 0.975 cg18320188 0.069 0.023 0.068 0.022 0.587 0.988 0.947 1.032 cg04841293 0.014 0.004 0.014 0.004 0.002 0.670 0.519 0.865 cg05845319 0.029 0.009 0.027 0.008 8.11x10-6 0.752 0.664 0.852 cg21109167 0.115 0.049 0.105 0.042 2.37x10-8 0.929 0.906 0.954 cg03901257 0.013 0.004 0.013 0.004 1.79x10-4 0.603 0.463 0.786 cg02279071 0.010 0.003 0.009 0.003 2.52x10-7 0.411 0.293 0.577 cg14751544 0.022 0.006 0.022 0.007 0.073 0.870 0.747 1.013 cg16764580 0.023 0.010 0.024 0.011 1.67x10-4 1.209 1.095 1.335 cg01302270 0.020 0.007 0.020 0.007 0.805 1.018 0.884 1.172 cg17641046 0.044 0.016 0.045 0.016 0.241 1.037 0.976 1.101 cg07101782 0.009 0.001 0.009 0.003 0.976 1.007 0.648 1.566 cg03497419 0.012 0.009 0.012 0.010 0.388 1.047 0.944 1.161 cg27586588 0.010 0.004 0.010 0.005 0.701 1.041 0.847 1.280 cg16433211 0.021 0.003 0.021 0.004 0.045 1.329 1.006 1.756 cg10769891 0.005 0.002 0.005 0.003 0.015 0.506 0.293 0.874 cg19132762 0.015 0.005 0.016 0.006 0.176 1.130 0.947 1.349 cg23658326 0.016 0.002 0.016 0.005 0.658 0.943 0.729 1.221 cg11600697 0.042 0.012 0.044 0.013 1.50x10-5 1.191 1.101 1.290 cg21490561 0.010 0.003 0.010 0.003 1.34x10-4 0.563 0.419 0.756 cg00893636 0.019 0.005 0.018 0.005 3.46x10-7 0.569 0.458 0.707 cg03192963 0.014 0.004 0.014 0.004 0.078 0.793 0.613 1.026 cg06791151 0.012 0.002 0.012 0.002 0.099 0.694 0.450 1.072 cg07064226 0.034 0.041 0.037 0.046 0.040 1.024 1.001 1.047 cg06108510 0.015 0.019 0.017 0.019 0.033 1.064 1.005 1.126 cg24985459 0.033 0.020 0.034 0.021 0.045 1.052 1.001 1.106 cg12790037 0.028 0.007 0.027 0.007 0.010 0.826 0.713 0.956 cg25202636 0.034 0.040 0.038 0.051 0.012 1.028 1.006 1.051 cg17621259 0.011 0.002 0.011 0.003 0.010 0.487 0.282 0.842 cg14671526 0.020 0.004 0.020 0.005 0.015 1.325 1.057 1.662 cg05906740 0.016 0.003 0.016 0.003 0.479 1.135 0.800 1.610

226

cg27331401 0.042 0.011 0.044 0.012 0.012 1.112 1.024 1.209 cg25837710 0.011 0.002 0.011 0.002 0.018 0.526 0.309 0.895 cg12851504 0.021 0.004 0.022 0.005 0.216 1.132 0.930 1.376 cg06590608 0.013 0.002 0.013 0.003 0.029 0.570 0.345 0.944 cg11224603 0.017 0.003 0.017 0.003 0.309 1.196 0.847 1.690 cg19208331 0.028 0.005 0.028 0.006 0.226 1.107 0.939 1.306 cg14598950 0.008 0.012 0.007 0.002 0.378 0.812 0.511 1.290 cg13846866 0.066 0.096 0.068 0.093 0.241 1.006 0.996 1.017 cg04777024 0.969 0.010 0.970 0.011 5.91x10-5 1.218 1.106 1.342 cg17024523 0.977 0.011 0.978 0.011 2.48x10-4 1.186 1.083 1.300 ch.3.753362R 0.092 0.029 0.094 0.029 0.053 1.033 1.000 1.067 cg25212762 0.967 0.025 0.964 0.032 0.118 0.975 0.945 1.006 cg11363877 0.989 0.004 0.990 0.004 3.99x10-4 1.562 1.220 1.999 cg03405026 0.987 0.004 0.987 0.004 0.112 1.208 0.957 1.526 cg16863190 0.900 0.119 0.889 0.125 0.062 0.996 0.992 1.000 cg27373390 0.979 0.007 0.979 0.015 0.025 0.874 0.776 0.983 cg01934787 0.909 0.049 0.903 0.057 0.001 0.973 0.957 0.989 cg06284479 0.982 0.008 0.982 0.007 0.006 1.215 1.059 1.394 cg24305555 0.979 0.008 0.979 0.011 0.018 0.874 0.782 0.977 cg05433805 0.586 0.131 0.572 0.128 0.037 0.992 0.985 1.000 cg15934958 0.919 0.037 0.916 0.043 0.030 0.974 0.951 0.997 cg06734169 0.039 0.029 0.042 0.039 0.007 1.045 1.012 1.078 cg12792366 0.034 0.050 0.036 0.053 0.161 1.014 0.995 1.033 cg00747698 0.034 0.008 0.034 0.008 0.395 1.051 .937 1.179 cg22221026 0.012 0.002 0.012 0.002 1.24x10-7 0.273 0.169 0.442 cg11574180 0.027 0.005 0.027 0.005 0.973 1.003 0.830 1.212 cg09310383 0.054 0.011 0.056 0.011 1.46x10-4 1.193 1.089 1.307 cg15011249 0.025 0.006 0.025 0.007 0.324 0.928 0.801 1.076 cg17479303 0.021 0.015 0.021 0.015 0.344 1.032 0.966 1.103 cg06853609 0.056 0.062 0.060 0.063 0.079 1.014 0.998 1.030 cg22985146 0.557 0.085 0.561 0.084 0.154 1.010 0.996 1.023 cg12999063 0.979 0.047 0.981 0.030 0.428 1.010 0.986 1.035 cg11321190 0.720 0.087 0.704 0.098 9.26x10-5 0.976 0.964 0.988 Significant results are bolded when P<7.14x10-4. SD – standard deviation; CI – confidence interval

227

Table A5. Logistic regression analysis for association with methylation between CRC cases

and controls. Mean β-value of controls and CRC cases is shown along with logistic regression

analysis at the MLH1 region. Analysis of CRC cases versus controls is adjusted for age, sex, and

peripheral blood mononuclear cell proportions. Effect size represents the increased risk of CRC

per 1% reduction in methylation.

Probe ID Control Mean

Control SD

Case Mean

Case SD

P-value Effect Size

Lower 95% CI

Upper

95% CI

cg21595053 0.988 0.005 0.988 0.006 0.147 1.154 0.951 1.401 cg02103401 0.726 0.090 0.723 0.091 8.80x10-5 1.022 1.011 1.033 cg24607398 0.873 0.063 0.868 0.064 3.20x10-4 1.030 1.014 1.047 cg10990993 0.852 0.060 0.843 0.061 3.78x10-7 1.045 1.027 1.063 cg04726821 0.205 0.068 0.187 0.070 4.38x10-6 1.035 1.020 1.051 cg11291081 0.063 0.029 0.061 0.032 0.640 1.008 0.975 1.042 cg05670953 0.158 0.068 0.143 0.070 1.13x10-4 1.032 1.015 1.048 cg18320188 0.070 0.021 0.067 0.024 0.103 1.037 0.993 1.084 cg04841293 0.015 0.004 0.014 0.004 1.56x10-5 1.805 1.381 2.359 cg05845319 0.029 0.008 0.027 0.008 3.59x10-4 1.258 1.109 1.427 cg21109167 0.117 0.047 0.103 0.045 2.13x10-7 1.072 1.044 1.101 cg03901257 0.013 0.004 0.013 0.004 0.086 1.261 0.968 1.643 cg02279071 0.009 0.003 0.009 0.003 0.119 0.781 0.572 1.066 cg14751544 0.022 0.007 0.022 0.006 0.533 0.953 0.819 1.109 cg16764580 0.024 0.011 0.023 0.010 0.006 1.151 1.041 1.272 cg01302270 0.020 0.007 0.021 0.007 0.065 0.870 0.750 1.009 cg17641046 0.044 0.016 0.044 0.016 0.441 0.976 0.918 1.038 cg07101782 0.009 0.003 0.009 0.001 0.717 0.921 0.590 1.438 cg03497419 0.012 0.008 0.012 0.010 0.383 0.954 0.860 1.060 cg27586588 0.010 0.004 0.010 0.005 0.771 0.969 0.785 1.197 cg16433211 0.021 0.004 0.021 0.003 0.008 1.479 1.108 1.974 cg10769891 0.005 0.003 0.005 0.002 0.133 0.689 0.424 1.120 cg19132762 0.016 0.007 0.015 0.004 0.018 1.252 1.039 1.508 cg23658326 0.016 0.005 0.015 0.002 0.034 1.583 1.036 2.421 cg11600697 0.044 0.013 0.042 0.012 3.13x10-4 1.162 1.071 1.260 cg21490561 0.010 0.003 0.010 0.003 0.162 0.808 0.599 1.090 cg00893636 0.019 0.004 0.019 0.005 0.861 0.981 0.789 1.219 cg03192963 0.014 0.004 0.014 0.004 0.786 0.964 0.740 1.255 cg06791151 0.012 0.002 0.012 0.002 0.341 0.806 0.518 1.256 cg07064226 0.035 0.043 0.035 0.044 0.606 1.006 0.983 1.029 cg06108510 0.016 0.014 0.016 0.023 0.608 0.987 0.937 1.039 cg24985459 0.033 0.018 0.034 0.023 0.830 1.005 0.957 1.056 cg12790037 0.027 0.007 0.028 0.007 2.22x10-4 0.752 0.647 0.875 cg25202636 0.035 0.041 0.037 0.049 0.454 0.992 0.971 1.013 cg17621259 0.011 0.002 0.011 0.002 0.016 0.509 0.294 0.881

228

cg14671526 0.020 0.005 0.020 0.004 0.066 1.244 0.985 1.570 cg05906740 0.016 0.003 0.016 0.003 0.016 1.557 1.084 2.235 cg27331401 0.043 0.012 0.043 0.012 0.481 1.031 0.948 1.120 cg25837710 0.011 0.002 0.011 0.002 0.578 1.149 0.705 1.872 cg12851504 0.022 0.005 0.021 0.005 0.259 1.123 0.918 1.375 cg06590608 0.012 0.003 0.013 0.002 0.093 0.667 0.416 1.070 cg11224603 0.017 0.003 0.017 0.003 0.355 1.183 0.829 1.688 cg19208331 0.028 0.006 0.028 0.005 0.476 1.064 0.898 1.260 cg14598950 0.008 0.012 0.007 0.002 9.76x10-5 2.696 1.637 4.439 cg13846866 0.067 0.092 0.068 0.097 0.455 1.004 0.933 1.015 cg04777024 0.969 0.011 0.970 0.010 0.870 1.008 0.915 1.111 cg17024523 0.977 0.012 0.977 0.011 0.756 1.015 0.925 1.114 ch.3.753362R 0.094 0.031 0.092 0.028 0.330 1.017 0.983 1.051 cg25212762 0.968 0.021 0.964 0.034 0.005 1.054 1.016 1.093 cg11363877 0.990 0.004 0.989 0.004 0.036 1.309 1.017 1.686 cg03405026 0.987 0.004 0.987 0.004 0.609 0.939 0.739 1.194 cg16863190 0.894 0.120 0.896 0.123 0.977 1.000 0.996 1.004 cg27373390 0.978 0.014 0.979 0.008 0.018 0.857 0.754 0.974 cg01934787 0.911 0.047 0.902 0.058 0.001 1.031 1.013 1.048 cg06284479 0.983 0.007 0.981 0.008 0.009 1.211 1.049 1.399 cg24305555 0.979 0.008 0.979 0.011 0.911 0.994 0.896 1.102 cg05433805 0.582 0.124 0.576 0.134 0.067 1.007 1.000 1.015 cg15934958 0.921 0.039 0.914 0.040 0.011 1.033 1.007 1.059 cg06734169 0.039 0.025 0.041 0.041 0.344 0.986 0.958 1.015 cg12792366 0.033 0.044 0.036 0.057 0.409 0.992 0.973 1.011 cg00747698 0.034 0.008 0.034 0.008 0.987 1.001 0.890 1.126 cg22221026 0.012 0.002 0.012 0.002 0.712 0.913 0.562 1.483 cg11574180 0.027 0.005 0.027 0.005 0.036 1.233 1.014 1.500 cg09310383 0.055 0.011 0.055 0.011 0.250 1.056 0.962 1.159 cg15011249 0.025 0.007 0.025 0.006 0.492 0.948 0.815 1.103 cg17479303 0.021 0.013 0.021 0.017 0.849 1.006 0.942 1.076 cg06853609 0.060 0.061 0.056 0.063 0.246 1.010 0.993 1.026 cg22985146 0.567 0.083 0.552 0.086 0.199 1.009 0.995 1.023 cg12999063 0.979 0.048 0.981 0.030 0.267 0.986 0.962 1.011 cg11321190 0.711 0.090 0.713 0.096 2.75x10-8 1.037 1.024 1.050 Significant results are bolded when P<7.14x10-4. SD – standard deviation; CI – confidence interval

229

Table A6. Methylation between SNP genotypes of MSH2-118T>C in controls by ANOVA.

Mean β-value of each genotype of the MSH2 promoter SNP in controls from Ontario.

Chr 3 Location

Probe ID WT mean

WT SD

Het Mean

Het SD

Hom Mean

Hom SD

P-val

47595507 cg24916165 0.965 0.015 0.965 0.012 0.972 0.009 0.550 47596073 cg07059875 0.061 0.028 0.058 0.026 0.059 0.025 0.078 47596136 cg07658533 0.055 0.019 0.054 0.019 0.066 0.024 0.681 47596185 cg06233293 0.020 0.004 0.020 0.008 0.023 0.012 0.407 47596410 cg21926875 0.045 0.016 0.043 0.017 0.046 0.015 0.609 47596430 cg02348634 0.041 0.037 0.040 0.040 0.066 0.062 0.476 47596787 cg17525856 0.143 0.033 0.144 0.033 0.155 0.025 0.230 47596912 cg12942414 0.171 0.052 0.169 0.053 0.206 0.073 0.917 47597118 cg16076328 0.492 0.065 0.492 0.062 0.510 0.063 0.438 47597188 cg03210866 0.548 0.062 0.553 0.057 0.543 0.063 0.206 47597331 cg03706175 0.908 0.061 0.913 0.044 0.914 0.031 0.628 47600173 cg01167408 0.785 0.056 0.794 0.054 0.767 0.062 0.488 47603759 cg13823166 0.925 0.114 0.930 0.108 0.942 0.023 0.656 47604176 cg17246929 0.979 0.009 0.979 0.006 0.980 0.006 0.483 47604206 cg15792957 0.956 0.019 0.958 0.015 0.951 0.022 0.874 47604745 cg10285618 0.931 0.031 0.937 0.028 0.929 0.026 0.587 47629792 cg25746226 0.050 0.013 0.049 0.012 0.053 0.013 0.215 47629949 cg17129141 0.006 0.002 0.006 0.002 0.006 0.001 0.819 47630147 cg22547404 0.018 0.004 0.018 0.004 0.019 0.005 0.885 47630172 cg25868465 0.029 0.006 0.029 0.006 0.032 0.012 0.166 47630224 cg22269526 0.034 0.013 0.034 0.013 0.034 0.010 3.29x10-11 47630251 cg06478094 0.021 0.009 0.020 0.008 0.023 0.009 3.66x10-5 47630294 cg00547758 0.015 0.002 0.015 0.002 0.017 0.010 0.976 47630433 cg14282180 0.018 0.003 0.018 0.003 0.017 0.002 0.427 47630550 cg14803009 0.044 0.109 0.044 0.110 0.015 0.006 1.31x10-26 47630787 cg03639557 0.017 0.014 0.017 0.014 0.015 0.002 0.914 47630845 cg11311499 0.027 0.006 0.027 0.006 0.028 0.008 0.959 47631120 cg02458113 0.014 0.009 0.014 0.005 0.012 0.004 0.767 47631145 cg09740554 0.076 0.032 0.073 0.027 0.073 0.037 0.009 47633216 cg23898128 0.981 0.006 0.981 0.005 0.984 0.005 0.323 47684718 cg19180827 0.875 0.054 0.874 0.057 0.880 0.054 0.130 47713462 cg07479270 0.872 0.140 0.876 0.154 0.889 0.056 0.680 47716097 cg03266686 0.415 0.065 0.404 0.063 0.432 0.078 0.038 47716517 cg15110473 0.979 0.010 0.979 0.010 0.982 0.010 0.386 47717244 cg04185310 0.855 0.079 0.850 0.078 0.860 0.074 0.034 47736753 cg00981060 0.889 0.037 0.890 0.034 0.882 0.024 0.013 47739740 cg10208034 0.975 0.012 0.975 0.011 0.975 0.011 0.087 47743741 cg11691189 0.928 0.043 0.930 0.036 0.947 0.023 0.292 47745038 cg00495909 0.969 0.016 0.969 0.015 0.974 0.011 0.336 47746468 cg15582102 0.942 0.037 0.939 0.041 0.954 0.033 5.80x10-16 47748042 cg04175739 0.088 0.047 0.086 0.046 0.085 0.025 2.93x10-5

230

47748407 cg14556391 0.040 0.012 0.039 0.008 0.041 0.007 0.354 47748839 cg00042186 0.152 0.042 0.149 0.041 0.173 0.051 0.499 47749060 cg19614911 0.308 0.039 0.304 0.041 0.318 0.028 0.005 47750631 cg26990681 0.948 0.114 0.951 0.104 0.941 0.119 0.019 47752514 cg00783525 0.945 0.017 0.944 0.018 0.955 0.012 0.385 47755888 cg16118678 0.883 0.043 0.885 0.038 0.881 0.028 0.710 47761452 cg02590088 0.947 0.030 0.948 0.025 0.938 0.031 0.021 47768650 cg13278115 0.912 0.030 0.913 0.026 0.912 0.020 0.458 47779198 cg17353680 0.786 0.099 0.788 0.099 0.804 0.058 0.307 47793276 cg22162312 0.940 0.029 0.943 0.024 0.925 0.036 0.156 47796781 cg27138584 0.041 0.013 0.041 0.016 0.052 0.034 0.226 47797133 cg13223402 0.024 0.011 0.024 0.009 0.047 0.053 0.273 47797415 cg02318629 0.066 0.026 0.065 0.027 0.092 0.072 0.067 47797488 cg03603951 0.039 0.026 0.041 0.047 0.057 0.066 0.257 47797569 cg04431946 0.126 0.072 0.125 0.064 0.165 0.120 0.272 47797590 cg08074851 0.107 0.056 0.105 0.058 0.140 0.086 0.376 47797953 cg01826863 0.133 0.041 0.132 0.041 0.157 0.069 0.113 47797963 cg13913015 0.027 0.023 0.029 0.028 0.056 0.073 0.137 47798396 cg27320127 0.251 0.054 0.249 0.051 0.272 0.076 0.263 47798477 cg04981611 0.053 0.020 0.052 0.018 0.070 0.037 0.021 47798679 cg04934807 0.081 0.022 0.080 0.028 0.087 0.025 0.384 47799165 cg04943225 0.367 0.067 0.371 0.067 0.394 0.065 0.133 47799268 cg14758072 0.518 0.077 0.510 0.080 0.531 0.050 0.327 47799405 cg10903274 0.592 0.061 0.583 0.063 0.591 0.063 0.849 47800047 cg09022430 0.891 0.052 0.891 0.047 0.897 0.037 0.145 47800393 cg04735632 0.868 0.051 0.865 0.054 0.851 0.051 0.052 47802717 cg11125315 0.989 0.009 0.989 0.004 0.991 0.002 0.406 47820094 cg13258091 0.946 0.055 0.946 0.063 0.957 0.013 0.598 47828743 cg13820434 0.966 0.032 0.967 0.022 0.967 0.028 0.342 47842922 cg14299961 0.990 0.003 0.989 0.003 0.991 0.003 0.917 47848563 cg08241610 0.985 0.005 0.986 0.004 0.986 0.003 0.473 47859049 cg09453252 0.987 0.005 0.986 0.005 0.987 0.007 0.712 47863922 cg06842253 0.979 0.008 0.979 0.008 0.978 0.006 0.861 47879734 cg10809134 0.899 0.051 0.899 0.052 0.916 0.034 0.115 47882601 cg23432368 0.788 0.063 0.791 0.057 0.768 0.061 0.445 47905207 cg24211994 0.908 0.047 0.908 0.047 0.905 0.046 0.287 47915862 cg10211062 0.791 0.053 0.792 0.049 0.811 0.043 0.588 47916139 cg07464408 0.964 0.019 0.963 0.019 0.969 0.019 0.038 47917841 cg01693539 0.977 0.006 0.977 0.006 0.978 0.008 0.785 47918027 cg00111466 0.723 0.042 0.721 0.042 0.733 0.038 0.186 47918138 cg27274072 0.975 0.008 0.975 0.007 0.979 0.008 0.555 47918165 cg22563036 0.899 0.019 0.898 0.019 0.901 0.014 0.981 47918397 cg25828259 0.986 0.003 0.986 0.002 0.987 0.003 0.257 47918413 cg12482860 0.985 0.006 0.985 0.006 0.989 0.006 0.236 47922175 cg22234368 0.805 0.066 0.811 0.062 0.792 0.070 0.574 47922605 cg02053451 0.931 0.036 0.932 0.035 0.936 0.038 0.319

231

47979809 ch.2.1149235F 0.095 0.055 0.091 0.050 0.084 0.018 0.904 47982632 cg07262620 0.917 0.048 0.919 0.046 0.918 0.058 0.267 47991493 ch.2.47844995F 0.010 0.002 0.010 0.002 0.009 0.002 0.840 47994785 ch.2.47848289F 0.012 0.003 0.011 0.003 0.010 0.002 0.794 48006758 cg14296255 0.986 0.006 0.985 0.006 0.988 0.005 0.894 48009698 cg05039065 0.040 0.013 0.039 0.014 0.038 0.010 0.187 48009730 cg02905881 0.020 0.013 0.019 0.013 0.018 0.003 0.608 48009809 cg11869233 0.021 0.030 0.024 0.047 0.017 0.005 0.354 48009866 cg09898070 0.180 0.038 0.180 0.040 0.183 0.037 0.518 48009933 cg16595246 0.034 0.008 0.033 0.008 0.033 0.008 0.246 48009935 cg18222961 0.027 0.004 0.027 0.004 0.025 0.004 0.872 48010097 cg00620552 0.019 0.003 0.019 0.002 0.017 0.002 0.685 48010105 cg21405109 0.025 0.004 0.025 0.004 0.023 0.003 0.785 48010117 cg19736286 0.019 0.007 0.019 0.007 0.015 0.003 0.597 48010336 cg19076255 0.012 0.002 0.012 0.002 0.011 0.001 0.745 48010362 cg07652213 0.011 0.002 0.011 0.001 0.010 0.001 0.676 48011201 cg15355298 0.061 0.045 0.061 0.051 0.062 0.026 0.802 48012925 cg19570558 0.862 0.050 0.855 0.050 0.873 0.038 0.778 48013473 cg04781916 0.832 0.113 0.824 0.124 0.841 0.098 0.611 Significant results are bolded when P<4.72x10-4. SD – standard deviation; WT – wildtype; Het – heterozygous; Hom – homozygous variant

232

Table A7. Methylation between SNP genotypes of MSH2-118T>C in CRC cases by

ANOVA. Mean β-value of each genotype of the MSH2 promoter SNP in CRC cases from

Ontario.

Chr 3 Location

Probe ID WT mean

WT SD

Het Mean

Het SD

Hom Mean

Hom SD

P-val

47595507 cg24916165 0.968 0.011 0.968 0.009 0.971 0.007 0.773 47596073 cg07059875 0.059 0.027 0.062 0.034 0.060 0.020 0.005 47596136 cg07658533 0.060 0.032 0.061 0.034 0.067 0.036 0.062 47596185 cg06233293 0.021 0.006 0.021 0.005 0.021 0.004 0.095 47596410 cg21926875 0.042 0.014 0.043 0.014 0.044 0.010 0.537 47596430 cg02348634 0.036 0.030 0.036 0.029 0.038 0.024 0.436 47596787 cg17525856 0.156 0.047 0.154 0.044 0.159 0.055 0.209 47596912 cg12942414 0.154 0.051 0.156 0.055 0.159 0.047 0.598 47597118 cg16076328 0.492 0.066 0.489 0.064 0.450 0.055 0.709 47597188 cg03210866 0.548 0.061 0.545 0.056 0.509 0.041 0.994 47597331 cg03706175 0.881 0.065 0.872 0.100 0.868 0.061 0.076 47600173 cg01167408 0.799 0.062 0.804 0.059 0.797 0.063 0.174 47603759 cg13823166 0.805 0.275 0.830 0.243 0.892 0.141 0.714 47604176 cg17246929 0.976 0.017 0.974 0.023 0.978 0.005 0.368 47604206 cg15792957 0.958 0.018 0.958 0.019 0.955 0.019 0.019 47604745 cg10285618 0.939 0.025 0.938 0.026 0.938 0.018 0.376 47629792 cg25746226 0.046 0.010 0.046 0.010 0.045 0.008 0.534 47629949 cg17129141 0.006 0.002 0.006 0.002 0.007 0.001 0.858 47630147 cg22547404 0.018 0.004 0.018 0.004 0.017 0.002 0.131 47630172 cg25868465 0.027 0.005 0.027 0.005 0.025 0.005 0.652 47630224 cg22269526 0.029 0.010 0.030 0.010 0.030 0.009 2.83x10-7 47630251 cg06478094 0.022 0.012 0.022 0.011 0.019 0.008 5.83x10-3 47630294 cg00547758 0.015 0.003 0.015 0.003 0.015 0.002 0.769 47630433 cg14282180 0.017 0.003 0.016 0.003 0.017 0.002 0.796 47630550 cg14803009 0.042 0.109 0.049 0.117 0.063 0.142 5.30x10-17 47630787 cg03639557 0.024 0.020 0.025 0.028 0.020 0.009 0.364 47630845 cg11311499 0.025 0.006 0.025 0.006 0.026 0.007 0.310 47631120 cg02458113 0.014 0.011 0.014 0.010 0.014 0.004 0.215 47631145 cg09740554 0.073 0.033 0.077 0.043 0.075 0.017 0.060 47633216 cg23898128 0.982 0.006 0.982 0.005 0.982 0.005 0.597 47684718 cg19180827 0.871 0.055 0.879 0.051 0.884 0.034 0.219 47713462 cg07479270 0.770 0.294 0.781 0.289 0.884 0.149 0.840 47716097 cg03266686 0.405 0.062 0.411 0.062 0.433 0.060 0.002 47716517 cg15110473 0.979 0.011 0.978 0.009 0.980 0.008 0.121 47717244 cg04185310 0.848 0.077 0.856 0.069 0.881 0.056 0.616 47736753 cg00981060 0.895 0.036 0.891 0.034 0.883 0.035 0.009 47739740 cg10208034 0.975 0.012 0.976 0.010 0.974 0.012 0.039 47743741 cg11691189 0.932 0.042 0.928 0.055 0.928 0.041 0.942 47745038 cg00495909 0.968 0.022 0.969 0.016 0.967 0.012 0.841

233

47746468 cg15582102 0.941 0.037 0.940 0.039 0.927 0.051 1.21x10-8 47748042 cg04175739 0.083 0.051 0.081 0.045 0.070 0.018 0.037 47748407 cg14556391 0.039 0.020 0.040 0.013 0.040 0.006 0.337 47748839 cg00042186 0.154 0.045 0.153 0.038 0.150 0.024 0.966 47749060 cg19614911 0.309 0.038 0.306 0.034 0.292 0.014 0.028 47750631 cg26990681 0.961 0.087 0.950 0.106 0.948 0.124 0.122 47752514 cg00783525 0.949 0.023 0.946 0.022 0.945 0.018 0.416 47755888 cg16118678 0.898 0.044 0.893 0.048 0.885 0.033 0.954 47761452 cg02590088 0.952 0.026 0.952 0.024 0.940 0.024 0.142 47768650 cg13278115 0.917 0.030 0.917 0.028 0.901 0.031 0.107 47779198 cg17353680 0.791 0.090 0.782 0.099 0.774 0.078 0.571 47793276 cg22162312 0.945 0.028 0.943 0.029 0.936 0.038 0.492 47796781 cg27138584 0.038 0.014 0.039 0.014 0.047 0.010 0.376 47797133 cg13223402 0.023 0.017 0.023 0.010 0.028 0.017 0.236 47797415 cg02318629 0.058 0.029 0.059 0.032 0.064 0.018 0.464 47797488 cg03603951 0.033 0.026 0.033 0.021 0.044 0.025 0.414 47797569 cg04431946 0.119 0.071 0.117 0.064 0.163 0.107 0.166 47797590 cg08074851 0.100 0.058 0.096 0.057 0.114 0.059 0.101 47797953 cg01826863 0.114 0.037 0.112 0.028 0.119 0.021 0.485 47797963 cg13913015 0.028 0.039 0.026 0.016 0.027 0.010 0.522 47798396 cg27320127 0.250 0.054 0.248 0.051 0.269 0.045 3.56x10-4 47798477 cg04981611 0.041 0.025 0.042 0.025 0.051 0.021 0.382 47798679 cg04934807 0.074 0.024 0.074 0.023 0.080 0.018 0.262 47799165 cg04943225 0.367 0.071 0.369 0.069 0.374 0.082 0.408 47799268 cg14758072 0.519 0.082 0.520 0.074 0.526 0.046 0.316 47799405 cg10903274 0.610 0.082 0.608 0.084 0.585 0.066 0.569 47800047 cg09022430 0.898 0.051 0.893 0.050 0.875 0.043 0.227 47800393 cg04735632 0.873 0.054 0.872 0.060 0.852 0.052 0.139 47802717 cg11125315 0.989 0.004 0.989 0.004 0.989 0.004 0.477 47820094 cg13258091 0.847 0.186 0.857 0.185 0.900 0.139 0.351 47828743 cg13820434 0.964 0.035 0.961 0.032 0.949 0.045 0.479 47842922 cg14299961 0.990 0.004 0.990 0.003 0.990 0.003 0.255 47848563 cg08241610 0.987 0.003 0.987 0.003 0.988 0.002 0.944 47859049 cg09453252 0.985 0.012 0.984 0.013 0.987 0.002 0.192 47863922 cg06842253 0.981 0.006 0.981 0.007 0.980 0.008 0.047 47879734 cg10809134 0.907 0.050 0.902 0.049 0.875 0.036 0.319 47882601 cg23432368 0.798 0.057 0.790 0.055 0.752 0.047 0.163 47905207 cg24211994 0.911 0.046 0.906 0.044 0.881 0.050 0.436 47915862 cg10211062 0.793 0.057 0.790 0.047 0.773 0.029 0.739 47916139 cg07464408 0.965 0.019 0.963 0.018 0.966 0.015 0.777 47917841 cg01693539 0.972 0.008 0.972 0.009 0.973 0.009 0.073 47918027 cg00111466 0.734 0.065 0.730 0.066 0.714 0.052 0.535 47918138 cg27274072 0.974 0.008 0.975 0.007 0.975 0.008 0.715 47918165 cg22563036 0.904 0.028 0.903 0.028 0.903 0.022 0.880 47918397 cg25828259 0.986 0.003 0.986 0.003 0.986 0.002 0.184 47918413 cg12482860 0.986 0.007 0.986 0.007 0.986 0.006 0.330

234

47922175 cg22234368 0.802 0.076 0.792 0.081 0.772 0.097 0.419 47922605 cg02053451 0.927 0.033 0.923 0.033 0.898 0.020 0.229 47979809 ch.2.1149235F 0.149 0.125 0.144 0.117 0.139 0.151 0.806 47982632 cg07262620 0.918 0.046 0.915 0.046 0.897 0.034 0.464 47991493 ch.2.47844995F 0.009 0.003 0.009 0.002 0.010 0.002 0.754 47994785 ch.2.47848289F 0.010 0.003 0.010 0.002 0.010 0.002 0.631 48006758 cg14296255 0.986 0.007 0.986 0.007 0.987 0.005 0.174 48009698 cg05039065 0.038 0.012 0.039 0.012 0.045 0.013 0.279 48009730 cg02905881 0.028 0.027 0.028 0.031 0.021 0.011 0.163 48009809 cg11869233 0.067 0.082 0.060 0.078 0.043 0.058 0.183 48009866 cg09898070 0.145 0.055 0.145 0.058 0.179 0.044 0.898 48009933 cg16595246 0.034 0.009 0.034 0.010 0.033 0.007 0.614 48009935 cg18222961 0.027 0.005 0.027 0.005 0.027 0.004 0.606 48010097 cg00620552 0.019 0.003 0.018 0.003 0.019 0.002 0.307 48010105 cg21405109 0.028 0.005 0.027 0.005 0.026 0.002 0.205 48010117 cg19736286 0.024 0.008 0.024 0.008 0.025 0.008 0.423 48010336 cg19076255 0.012 0.002 0.013 0.001 0.013 0.002 0.854 48010362 cg07652213 0.011 0.002 0.011 0.002 0.011 0.001 0.328 48011201 cg15355298 0.111 0.120 0.102 0.117 0.096 0.127 0.778 48012925 cg19570558 0.845 0.068 0.848 0.059 0.885 0.034 0.981 48013473 cg04781916 0.832 0.115 0.834 0.135 0.915 0.048 0.412 Significant results are bolded when P<4.72x10-4. SD – standard deviation; WT – wildtype; Het – heterozygous; Hom – homozygous variant

235

Table A8. Methylation between SNP genotypes of MSH6-159C>T in controls by ANOVA.

Mean β-value of each genotype of the MSH6 promoter SNP in controls from Ontario.

Chr 3 Location

Probe ID WT mean

WT SD

Het Mean

Het SD

Hom Mean

Hom SD

P-val

47595507 cg24916165 0.965 0.014 0.966 0.011 0.951 0.037 0.167 47596073 cg07059875 0.061 0.028 0.056 0.022 0.056 0.018 0.344 47596136 cg07658533 0.056 0.020 0.052 0.016 0.051 0.013 0.165 47596185 cg06233293 0.020 0.006 0.019 0.004 0.020 0.003 0.892 47596410 cg21926875 0.045 0.017 0.041 0.010 0.041 0.010 0.802 47596430 cg02348634 0.043 0.041 0.038 0.025 0.033 0.021 0.165 47596787 cg17525856 0.143 0.033 0.147 0.031 0.136 0.030 0.063 47596912 cg12942414 0.171 0.053 0.168 0.052 0.169 0.047 0.241 47597118 cg16076328 0.495 0.064 0.482 0.064 0.496 0.082 0.334 47597188 cg03210866 0.550 0.060 0.542 0.061 0.569 0.074 0.321 47597331 cg03706175 0.911 0.058 0.907 0.050 0.907 0.048 0.981 47600173 cg01167408 0.786 0.056 0.793 0.054 0.768 0.057 0.280 47603759 cg13823166 0.927 0.110 0.927 0.120 0.902 0.106 0.754 47604176 cg17246929 0.979 0.008 0.979 0.006 0.970 0.029 0.094 47604206 cg15792957 0.956 0.019 0.957 0.016 0.948 0.026 0.365 47604745 cg10285618 0.933 0.030 0.932 0.028 0.917 0.058 0.017 47629792 cg25746226 0.050 0.013 0.049 0.013 0.054 0.018 0.063 47629949 cg17129141 0.006 0.002 0.006 0.001 0.007 0.002 0.138 47630147 cg22547404 0.018 0.004 0.018 0.004 0.021 0.007 0.438 47630172 cg25868465 0.029 0.006 0.029 0.006 0.032 0.008 0.821 47630224 cg22269526 0.034 0.013 0.033 0.011 0.034 0.014 0.890 47630251 cg06478094 0.021 0.009 0.020 0.009 0.024 0.010 0.957 47630294 cg00547758 0.015 0.003 0.015 0.002 0.016 0.003 0.220 47630433 cg14282180 0.018 0.003 0.018 0.002 0.018 0.003 0.070 47630550 cg14803009 0.046 0.112 0.035 0.098 0.042 0.107 0.860 47630787 cg03639557 0.017 0.015 0.017 0.012 0.018 0.004 0.333 47630845 cg11311499 0.027 0.006 0.028 0.006 0.029 0.008 0.222 47631120 cg02458113 0.014 0.009 0.014 0.005 0.013 0.004 0.577 47631145 cg09740554 0.075 0.030 0.075 0.033 0.072 0.033 0.654 47633216 cg23898128 0.981 0.006 0.982 0.005 0.975 0.009 0.158 47684718 cg19180827 0.874 0.056 0.874 0.052 0.894 0.038 0.988 47713462 cg07479270 0.874 0.137 0.876 0.151 0.807 0.246 0.180 47716097 cg03266686 0.413 0.067 0.414 0.059 0.417 0.056 0.612 47716517 cg15110473 0.979 0.010 0.978 0.010 0.980 0.008 0.327 47717244 cg04185310 0.854 0.079 0.851 0.078 0.878 0.052 0.934 47736753 cg00981060 0.888 0.037 0.891 0.030 0.912 0.032 0.002 47739740 cg10208034 0.975 0.012 0.974 0.011 0.975 0.009 0.354 47743741 cg11691189 0.929 0.042 0.929 0.037 0.943 0.030 0.470 47745038 cg00495909 0.969 0.016 0.969 0.015 0.972 0.014 0.785 47746468 cg15582102 0.941 0.039 0.942 0.035 0.954 0.033 0.100 47748042 cg04175739 0.085 0.044 0.098 0.056 0.078 0.021 0.788

236


237

47979809 ch.2.1149235F 0.093 0.052 0.092 0.055 0.136 0.076 0.022 47982632 cg07262620 0.916 0.050 0.924 0.038 0.919 0.042 0.566 47991493 ch.2.47844995F 0.010 0.002 0.010 0.002 0.011 0.003 0.135 47994785 ch.2.47848289F 0.012 0.003 0.011 0.003 0.016 0.008 0.266 48006758 cg14296255 0.986 0.006 0.985 0.006 0.986 0.005 0.085 48009698 cg05039065 0.040 0.013 0.041 0.014 0.036 0.013 7.30x10-10 48009730 cg02905881 0.020 0.014 0.019 0.008 0.020 0.005 0.002 48009809 cg11869233 0.022 0.035 0.023 0.038 0.021 0.006 0.599 48009866 cg09898070 0.182 0.039 0.176 0.038 0.159 0.042 0.966 48009933 cg16595246 0.033 0.008 0.033 0.008 0.035 0.008 0.009 48009935 cg18222961 0.027 0.004 0.027 0.004 0.027 0.005 0.082 48010097 cg00620552 0.019 0.003 0.019 0.002 0.019 0.004 0.030 48010105 cg21405109 0.025 0.004 0.025 0.004 0.024 0.004 0.088 48010117 cg19736286 0.019 0.007 0.019 0.007 0.018 0.005 0.426 48010336 cg19076255 0.012 0.002 0.012 0.002 0.012 0.002 0.079 48010362 cg07652213 0.011 0.002 0.011 0.002 0.011 0.002 0.429 48011201 cg15355298 0.061 0.046 0.056 0.042 0.077 0.070 0.295 48012925 cg19570558 0.860 0.050 0.863 0.049 0.856 0.051 2.98x10-10 48013473 cg04781916 0.829 0.114 0.837 0.121 0.797 0.116 3.22x10-4 Significant results are bolded when P<4.72x10-4. SD – standard deviation; WT – wildtype; Het – heterozygous; Hom – homozygous variant

238

Table A9. Methylation between SNP genotype of MSH6-159C>T in CRC cases by ANOVA.

Mean β-value of each genotype of the MSH6 promoter SNP in CRC cases from Ontario.

Chr 3 Location

Probe ID WT mean

WT SD

Het Mean

Het SD

Hom Mean

Hom SD

P-value

47595507 cg24916165 0.968 0.011 0.970 0.009 0.969 0.013 0.182 47596073 cg07059875 0.059 0.026 0.062 0.033 0.087 0.074 0.323 47596136 cg07658533 0.060 0.032 0.062 0.031 0.075 0.055 0.751 47596185 cg06233293 0.021 0.006 0.021 0.006 0.023 0.009 0.364 47596410 cg21926875 0.042 0.014 0.043 0.013 0.045 0.013 0.516 47596430 cg02348634 0.036 0.031 0.037 0.025 0.024 0.005 0.219 47596787 cg17525856 0.155 0.046 0.158 0.046 0.191 0.051 0.710 47596912 cg12942414 0.155 0.053 0.154 0.047 0.146 0.047 0.114 47597118 cg16076328 0.493 0.064 0.477 0.068 0.502 0.067 0.843 47597188 cg03210866 0.548 0.059 0.536 0.063 0.563 0.057 0.868 47597331 cg03706175 0.880 0.066 0.880 0.069 0.778 0.298 0.406 47600173 cg01167408 0.799 0.062 0.809 0.051 0.780 0.084 0.278 47603759 cg13823166 0.808 0.271 0.840 0.232 0.756 0.328 0.776 47604176 cg17246929 0.976 0.017 0.976 0.020 0.956 0.059 0.228 47604206 cg15792957 0.958 0.018 0.956 0.021 0.963 0.010 0.122 47604745 cg10285618 0.938 0.026 0.938 0.024 0.948 0.011 0.126 47629792 cg25746226 0.046 0.010 0.046 0.010 0.042 0.010 0.402 47629949 cg17129141 0.006 0.002 0.007 0.002 0.006 0.002 0.998 47630147 cg22547404 0.018 0.004 0.018 0.004 0.019 0.002 0.377 47630172 cg25868465 0.027 0.005 0.026 0.006 0.030 0.007 0.737 47630224 cg22269526 0.030 0.010 0.030 0.009 0.027 0.004 0.965 47630251 cg06478094 0.022 0.012 0.021 0.010 0.026 0.015 0.339 47630294 cg00547758 0.015 0.003 0.015 0.003 0.014 0.004 0.122 47630433 cg14282180 0.017 0.003 0.017 0.003 0.017 0.004 0.401 47630550 cg14803009 0.043 0.111 0.056 0.128 0.017 0.012 0.005 47630787 cg03639557 0.024 0.021 0.024 0.019 0.050 0.071 0.598 47630845 cg11311499 0.025 0.007 0.025 0.005 0.026 0.007 0.577 47631120 cg02458113 0.014 0.012 0.014 0.006 0.016 0.005 0.910 47631145 cg09740554 0.074 0.036 0.074 0.034 0.074 0.016 0.720 47633216 cg23898128 0.982 0.005 0.981 0.006 0.981 0.007 0.423 47684718 cg19180827 0.873 0.054 0.872 0.055 0.889 0.040 0.724 47713462 cg07479270 0.773 0.291 0.797 0.270 0.605 0.461 0.533 47716097 cg03266686 0.407 0.062 0.409 0.067 0.416 0.049 0.374 47716517 cg15110473 0.979 0.010 0.978 0.013 0.981 0.012 0.288 47717244 cg04185310 0.851 0.075 0.847 0.074 0.882 0.058 0.609 47736753 cg00981060 0.893 0.036 0.892 0.034 0.910 0.040 0.203 47739740 cg10208034 0.975 0.012 0.974 0.012 0.979 0.009 0.062 47743741 cg11691189 0.930 0.047 0.935 0.038 0.942 0.048 0.033 47745038 cg00495909 0.968 0.021 0.968 0.016 0.973 0.014 0.117 47746468 cg15582102 0.941 0.036 0.936 0.046 0.953 0.040 0.260 47748042 cg04175739 0.083 0.050 0.079 0.045 0.084 0.075 0.631

239


240

47979809 ch.2.1149235F 0.150 0.123 0.136 0.124 0.149 0.141 0.429 47982632 cg07262620 0.917 0.046 0.914 0.046 0.929 0.055 0.956 47991493 ch.2.47844995F 0.009 0.002 0.009 0.003 0.011 0.004 0.967 47994785 ch.2.47848289F 0.010 0.003 0.010 0.002 0.010 0.002 0.553 48006758 cg14296255 0.986 0.007 0.986 0.006 0.987 0.008 0.282 48009698 cg05039065 0.038 0.012 0.039 0.015 0.041 0.011 1.82x10-5 48009730 cg02905881 0.027 0.024 0.028 0.036 0.050 0.075 0.070 48009809 cg11869233 0.064 0.079 0.066 0.086 0.102 0.113 0.319 48009866 cg09898070 0.145 0.055 0.148 0.055 0.141 0.071 0.317 48009933 cg16595246 0.035 0.010 0.032 0.008 0.037 0.010 0.307 48009935 cg18222961 0.027 0.005 0.028 0.005 0.027 0.005 0.295 48010097 cg00620552 0.019 0.003 0.019 0.003 0.019 0.003 0.400 48010105 cg21405109 0.027 0.005 0.028 0.006 0.030 0.006 0.259 48010117 cg19736286 0.024 0.008 0.025 0.008 0.029 0.012 0.566 48010336 cg19076255 0.012 0.002 0.013 0.002 0.013 0.002 0.143 48010362 cg07652213 0.011 0.002 0.011 0.002 0.011 0.001 0.063 48011201 cg15355298 0.108 0.119 0.105 0.118 0.134 0.163 0.638 48012925 cg19570558 0.846 0.068 0.847 0.057 0.838 0.078 2.51x10-5 48013473 cg04781916 0.832 0.123 0.843 0.101 0.810 0.134 0.010 Significant results are bolded when P<4.72x10-4. SD – standard deviation; WT – wildtype; Het – heterozygous; Hom – homozygous variant

241

Table A10. Correlation between age and DNA methylation at MSH2 and MSH6. Partial

correlation, controlling for sex and peripheral blood mononuclear cell proportions, between age

and CRC cases and controls.

Probe ID Controls R P-value CRC Cases P-value cg24916165 -0.057 0.095 -0.002 0.945 cg07059875 0.026 0.451 -0.034 0.310 cg07658533 -0.046 0.183 0.006 0.865 cg06233293 0.057 0.100 -0.007 0.843 cg21926875 0.046 0.181 0.006 0.856 cg02348634 0.056 0.101 0.017 0.620 cg17525856 0.064 0.064 -0.060 0.071 cg12942414 0.118 0.001 -0.021 0.529 cg16076328 0.108 0.002 -0.030 0.371 cg03210866 0.079 0.022 -0.050 0.135 cg03706175 0.149 0.000 -0.040 0.229 cg01167408 -0.163 0.000 0.023 0.486 cg13823166 0.082 0.017 -0.015 0.663 cg17246929 0.038 0.267 -0.021 0.535 cg15792957 0.006 0.866 -0.066 0.050 cg10285618 -0.036 0.300 0.020 0.541 cg25746226 0.135 0.000 -0.001 0.979 cg17129141 0.039 0.253 0.026 0.438 cg22547404 0.085 0.013 -0.028 0.396 cg25868465 0.056 0.101 0.001 0.980 cg22269526 0.142 0.000 0.011 0.736 cg06478094 0.101 0.003 0.020 0.550 cg00547758 -0.016 0.647 -0.025 0.448 cg14282180 0.013 0.703 -0.008 0.808 cg14803009 -0.042 0.222 0.058 0.082 cg03639557 -0.142 0.000 0.014 0.686 cg11311499 0.042 0.223 0.010 0.773 cg02458113 0.012 0.723 -0.013 0.688 cg09740554 -0.031 0.362 -0.052 0.118 cg23898128 -0.010 0.778 -0.045 0.183 cg19180827 -0.014 0.676 0.025 0.454 cg07479270 0.031 0.363 0.003 0.935 cg03266686 0.014 0.690 0.017 0.601 cg15110473 0.136 0.000 0.020 0.558 cg04185310 0.028 0.424 -0.044 0.187 cg00981060 0.051 0.137 -0.012 0.721 cg10208034 0.059 0.084 0.023 0.487 cg11691189 0.022 0.515 -0.008 0.804 cg00495909 0.108 0.002 -0.010 0.764 cg15582102 -0.029 0.405 0.019 0.570

242

cg04175739 0.046 0.183 -0.032 0.341 cg14556391 -0.007 0.845 0.045 0.180 cg00042186 0.040 0.248 0.047 0.160 cg19614911 0.016 0.650 0.046 0.167 cg26990681 0.014 0.693 0.007 0.844 cg00783525 0.081 0.018 -0.030 0.377 cg16118678 0.026 0.445 0.027 0.418 cg02590088 0.002 0.950 0.013 0.698 cg13278115 -0.005 0.896 0.056 0.095 cg17353680 0.013 0.705 0.002 0.951 cg22162312 -0.011 0.739 -0.007 0.824 cg27138584 -0.023 0.503 0.018 0.587 cg13223402 0.033 0.342 0.028 0.405 cg02318629 0.072 0.037 0.042 0.211 cg03603951 0.028 0.416 -0.016 0.640 cg04431946 0.058 0.091 0.038 0.254 cg08074851 0.063 0.068 0.011 0.735 cg01826863 0.166 0.000 -0.013 0.690 cg13913015 0.073 0.033 0.017 0.604 cg27320127 -0.030 0.382 -0.023 0.491 cg04981611 0.019 0.573 0.006 0.853 cg04934807 0.117 0.001 0.023 0.486 cg04943225 0.011 0.755 -0.049 0.144 cg14758072 -0.001 0.987 -0.011 0.739 cg10903274 0.054 0.115 -0.005 0.873 cg09022430 0.077 0.025 -0.007 0.842 cg04735632 0.083 0.016 -0.064 0.056 cg11125315 -0.075 0.030 0.002 0.958 cg13258091 0.078 0.023 0.010 0.775 cg13820434 0.096 0.005 -0.017 0.602 cg14299961 0.082 0.017 -0.049 0.146 cg08241610 0.022 0.526 -0.030 0.365 cg09453252 0.057 0.095 -0.015 0.650 cg06842253 -0.142 0.000 -0.015 0.657 cg10809134 0.048 0.166 0.023 0.486 cg23432368 0.077 0.024 0.015 0.656 cg24211994 0.069 0.045 -0.015 0.646 cg10211062 0.063 0.066 0.018 0.589 cg07464408 0.128 0.000 -0.025 0.453 cg01693539 0.108 0.002 -0.022 0.512 cg00111466 0.064 0.064 0.009 0.789 cg27274072 -0.056 0.103 0.008 0.815 cg22563036 0.039 0.262 -0.048 0.152 cg25828259 -0.017 0.622 -0.001 0.977 cg12482860 0.035 0.316 0.003 0.933 cg22234368 -0.023 0.496 -0.040 0.228

243

cg02053451 0.016 0.643 0.124 0.000 ch.2.1149235F -0.020 0.569 -0.026 0.435

cg07262620 0.025 0.475 0.047 0.158 ch.2.47844995F 0.030 0.385 -0.046 0.168 ch.2.47848289F 0.127 0.000 0.006 0.848

cg14296255 0.093 0.007 0.033 0.320 cg05039065 0.020 0.569 -0.016 0.631 cg02905881 -0.103 0.003 -0.006 0.858 cg11869233 -0.136 0.000 -0.012 0.724 cg09898070 -0.014 0.679 -0.017 0.612 cg16595246 0.049 0.158 -0.044 0.192 cg18222961 -0.087 0.011 -0.026 0.437 cg00620552 -0.018 0.600 -0.007 0.844 cg21405109 -0.114 0.001 -0.039 0.240 cg19736286 -0.206 0.000 0.021 0.532 cg19076255 -0.003 0.937 0.001 0.983 cg07652213 -0.014 0.679 -0.027 0.423 cg15355298 0.033 0.336 -0.004 0.903 cg19570558 0.109 0.002 -0.044 0.184 cg04781916 -0.085 0.013 0.000 0.992

244

Table A11. Association between sex and methylation by logistic regression at MSH2- MSH6

region. Mean β-value is shown for females and males along with logistic regression analysis at

the MSH2-MSH6 region. Analysis of male versus female methylation is adjusted for age and

peripheral blood mononuclear cell proportions.

Probe ID Female Mean

Female SD

Male Mean

Male SD

P-value

Effect Size

Lower 95% CI

Upper 95% CI

cg24916165 0.968 0.011 0.966 0.014 0.001 0.870 0.803 0.943 cg07059875 0.061 0.031 0.060 0.027 0.244 0.981 0.949 1.013 cg07658533 0.059 0.032 0.056 0.023 0.007 0.951 0.917 0.986 cg06233293 0.020 0.005 0.020 0.006 0.950 0.995 0.840 1.178 cg21926875 0.043 0.014 0.043 0.015 0.495 1.023 0.959 1.091 cg02348634 0.038 0.033 0.039 0.035 0.361 1.013 0.985 1.042 cg17525856 0.151 0.043 0.150 0.039 0.436 0.991 0.968 1.014 cg12942414 0.160 0.051 0.165 0.053 0.091 1.016 0.997 1.035 cg16076328 0.489 0.066 0.495 0.065 0.052 1.015 1.000 1.029 cg03210866 0.545 0.061 0.551 0.060 0.056 1.015 1.000 1.032 cg03706175 0.890 0.068 0.896 0.072 0.091 1.012 0.998 1.026 cg01167408 0.795 0.059 0.792 0.058 0.453 0.994 0.978 1.010 cg13823166 0.858 0.222 0.878 0.200 0.049 1.005 1.000 1.009 cg17246929 0.977 0.014 0.977 0.013 0.816 1.008 0.940 1.082 cg15792957 0.958 0.017 0.957 0.018 0.154 0.962 0.912 1.015 cg10285618 0.936 0.026 0.935 0.030 0.245 0.980 0.947 1.014 cg25746226 0.047 0.012 0.048 0.012 0.431 1.033 0.952 1.122 cg17129141 0.006 0.002 0.006 0.002 0.880 0.956 0.530 1.722 cg22547404 0.018 0.004 0.018 0.004 0.853 0.979 0.777 1.232 cg25868465 0.027 0.006 0.028 0.006 0.010 1.243 1.053 1.468 cg22269526 0.031 0.011 0.033 0.012 0.025 1.100 1.012 1.196 cg06478094 0.022 0.012 0.022 0.010 0.626 0.978 0.897 1.068 cg00547758 0.015 0.003 0.015 0.003 0.393 1.157 0.828 1.616 cg14282180 0.017 0.003 0.017 0.003 0.300 1.192 0.856 1.659 cg14803009 0.046 0.113 0.041 0.105 0.544 0.997 0.989 1.006 cg03639557 0.022 0.022 0.020 0.017 0.012 0.934 0.885 0.985 cg11311499 0.026 0.006 0.026 0.007 0.150 1.121 0.959 1.311 cg02458113 0.013 0.005 0.014 0.012 0.034 1.216 1.015 1.457 cg09740554 0.072 0.027 0.075 0.036 0.537 0.948 0.801 1.123 cg23898128 0.982 0.006 0.981 0.005 0.613 0.996 0.979 1.013 cg19180827 0.874 0.054 0.873 0.056 0.057 1.004 1.000 1.008 cg07479270 0.813 0.250 0.833 0.220 0.849 1.001 0.987 1.016 cg03266686 0.410 0.063 0.411 0.064 0.885 0.993 0.903 1.092 cg15110473 0.979 0.010 0.979 0.010 0.610 0.997 0.985 1.009 cg04185310 0.853 0.074 0.852 0.080 0.760 0.996 0.970 1.023 cg00981060 0.892 0.036 0.892 0.036 0.952 1.002 0.925 1.086 cg10208034 0.975 0.012 0.975 0.011 0.577 0.994 0.973 1.015

245

cg11691189 0.930 0.043 0.930 0.046 0.841 1.006 0.952 1.062 cg00495909 0.968 0.019 0.969 0.016 0.765 0.996 0.971 1.022 cg15582102 0.941 0.037 0.941 0.039 0.957 0.999 0.980 1.019 cg04175739 0.085 0.051 0.085 0.048 0.527 1.017 0.964 1.073 cg14556391 0.039 0.023 0.040 0.016 0.482 1.008 0.986 1.031 cg00042186 0.152 0.043 0.153 0.043 0.879 0.998 0.974 1.022 cg19614911 0.308 0.040 0.308 0.038 0.115 1.007 0.998 1.016 cg26990681 0.947 0.119 0.955 0.103 0.645 0.989 0.943 1.037 cg00783525 0.947 0.021 0.947 0.019 0.090 0.982 0.961 1.003 cg16118678 0.892 0.046 0.889 0.042 0.783 1.005 0.971 1.040 cg02590088 0.949 0.028 0.950 0.026 0.440 0.987 0.956 1.020 cg13278115 0.915 0.029 0.914 0.030 0.284 1.005 0.996 1.015 cg17353680 0.785 0.095 0.790 0.096 0.585 0.991 0.957 1.025 cg22162312 0.943 0.028 0.942 0.027 0.974 1.001 0.938 1.068 cg27138584 0.040 0.014 0.040 0.015 0.291 1.041 0.966 1.123 cg13223402 0.023 0.011 0.024 0.015 0.431 1.014 0.980 1.049 cg02318629 0.061 0.027 0.062 0.029 0.131 1.027 0.992 1.064 cg03603951 0.035 0.024 0.037 0.032 0.888 0.999 0.985 1.013 cg04431946 0.122 0.069 0.122 0.070 0.424 1.007 0.990 1.024 cg08074851 0.101 0.057 0.104 0.057 0.016 1.032 1.006 1.058 cg01826863 0.120 0.036 0.125 0.041 0.376 1.015 0.982 1.050 cg13913015 0.027 0.024 0.028 0.032 0.889 1.001 0.984 1.019 cg27320127 0.249 0.053 0.250 0.054 0.114 1.034 0.992 1.078 cg04981611 0.046 0.023 0.047 0.023 0.274 1.024 0.981 1.068 cg04934807 0.076 0.022 0.078 0.023 0.946 1.000 0.986 1.014 cg04943225 0.369 0.068 0.368 0.068 0.412 0.995 0.983 1.007 cg14758072 0.519 0.076 0.517 0.079 0.320 0.993 0.981 1.006 cg10903274 0.602 0.075 0.598 0.071 0.988 1.000 0.981 1.019 cg09022430 0.894 0.052 0.894 0.050 0.416 1.007 0.990 1.025 cg04735632 0.869 0.053 0.871 0.054 0.097 1.193 0.969 1.469 cg11125315 0.989 0.008 0.989 0.004 0.096 1.006 0.999 1.012 cg13258091 0.890 0.146 0.902 0.147 0.704 1.005 0.978 1.034 cg13820434 0.964 0.038 0.965 0.031 0.757 0.955 0.716 1.275 cg14299961 0.990 0.003 0.990 0.003 0.485 0.953 0.831 1.092 cg08241610 0.986 0.009 0.986 0.005 0.989 0.999 0.897 1.113 cg09453252 0.986 0.009 0.986 0.009 0.257 0.928 0.815 1.056 cg06842253 0.980 0.008 0.980 0.007 0.448 0.993 0.974 1.012 cg10809134 0.903 0.050 0.902 0.051 0.846 1.002 0.986 1.018 cg23432368 0.792 0.060 0.793 0.059 0.950 0.999 0.979 1.020 cg24211994 0.909 0.047 0.909 0.046 0.146 1.013 0.995 1.032 cg10211062 0.790 0.053 0.794 0.053 0.969 1.001 0.952 1.053 cg07464408 0.964 0.019 0.964 0.019 0.004 1.202 1.060 1.363 cg01693539 0.974 0.008 0.975 0.007 0.643 0.996 0.979 1.013 cg00111466 0.728 0.057 0.728 0.054 0.889 0.991 0.874 1.124 cg27274072 0.975 0.008 0.975 0.008 0.284 0.979 0.941 1.018 cg22563036 0.902 0.025 0.901 0.024 0.612 0.913 0.642 1.299

246

cg25828259 0.986 0.003 0.986 0.003 0.574 1.043 0.901 1.207 cg12482860 0.986 0.007 0.986 0.006 0.811 1.002 0.988 1.015 cg22234368 0.803 0.071 0.803 0.071 0.573 1.008 0.980 1.307 cg02053451 0.928 0.034 0.929 0.035 0.045 0.990 0.981 1.000

ch.2.1149235F 0.127 0.106 0.117 0.094 0.488 1.007 0.987 1.028 cg07262620 0.916 0.046 0.918 0.046 0.889 1.028 0.697 1.516

ch.2.47844995F 0.010 0.002 0.010 0.002 0.145 1.263 0.922 1.729 ch.2.47848289F 0.011 0.003 0.011 0.003 0.619 1.038 0.897 1.201

cg14296255 0.986 0.006 0.986 0.007 0.078 0.935 0.868 1.007 cg05039065 0.040 0.013 0.039 0.012 0.121 0.964 0.920 1.010 cg02905881 0.024 0.023 0.023 0.020 0.009 0.981 0.967 0.995 cg11869233 0.049 0.071 0.040 0.063 0.037 1.020 1.001 1.039 cg09898070 0.160 0.053 0.165 0.050 0.427 0.958 0.862 1.065 cg16595246 0.034 0.009 0.034 0.009 0.847 1.020 0.834 1.247 cg18222961 0.027 0.005 0.027 0.005 0.458 0.874 0.612 1.248 cg00620552 0.019 0.003 0.019 0.003 0.042 0.806 0.656 0.992 cg21405109 0.026 0.005 0.026 0.005 0.015 0.854 0.751 0.970 cg19736286 0.022 0.008 0.021 0.007 0.097 0.625 0.359 1.088 cg19076255 0.012 0.002 0.012 0.002 0.126 0.630 0.349 1.138 cg07652213 0.011 0.002 0.011 0.002 0.061 0.990 0.980 1.000 cg15355298 0.090 0.101 0.081 0.088 0.207 1.011 0.994 1.028 cg19570558 0.852 0.060 0.855 0.054 0.537 0.997 0.989 1.006 cg04781916 0.833 0.116 0.829 0.118 0.382 1.002 0.991 1.014

SD – standard deviation; CI – confidence interval

247

Table A12. Logistic regression analysis for association with methylation between CRC

cases and controls. Mean β-value of controls and CRC cases is shown along with logistic

regression analysis at the MSH2-MSH6 region. Analysis of CRC cases versus controls is adjusted

for age, sex, and peripheral blood mononuclear cell proportions. Effect size represents the

increased risk of CRC per 1% reduction in methylation.

Probe ID Control Mean

Control SD

Case Mean

Case SD

P-value Effect Size

Lower 95% CI

Upper 95% CI

cg24916165 0.965 0.014 0.968 0.010 3.59x10-7 0.797 0.730 0.869 cg07059875 0.060 0.027 0.060 0.031 0.874 1.003 0.970 1.037 cg07658533 0.055 0.019 0.061 0.034 1.92x10-5 0.912 0.874 0.951 cg06233293 0.020 0.006 0.021 0.006 3.37x10-4 0.704 0.581 0.853 cg21926875 0.044 0.016 0.042 0.014 0.016 1.090 1.016 1.169 cg02348634 0.041 0.037 0.036 0.029 0.011 1.041 1.009 1.073 cg17525856 0.144 0.033 0.157 0.047 2.58x10-10 0.920 0.896 0.944 cg12942414 0.170 0.052 0.154 0.051 5.23x10-7 1.052 1.031 1.072 cg16076328 0.492 0.065 0.492 0.066 0.355 0.993 0.978 1.008 cg03210866 0.548 0.062 0.548 0.060 0.697 0.997 0.981 1.013 cg03706175 0.910 0.055 0.877 0.079 1.04x10-13 1.106 1.081 1.131 cg01167408 0.788 0.056 0.798 0.060 0.004 0.976 0.959 0.992 cg13823166 0.928 0.107 0.810 0.264 1.83x10-13 1.047 1.037 1.057 cg17246929 0.979 0.008 0.976 0.017 5.09x10-7 1.377 1.216 1.560 cg15792957 0.956 0.018 0.959 0.018 0.039 0.943 0.892 0.997 cg10285618 0.932 0.030 0.939 0.025 1.08x10-4 0.931 0.898 0.965 cg25746226 0.049 0.013 0.046 0.010 1.21x10-5 1.213 1.112 1.322 cg17129141 0.006 0.002 0.006 0.002 0.945 1.021 0.558 1.871 cg22547404 0.018 0.004 0.018 0.004 0.510 1.082 0.855 1.371 cg25868465 0.029 0.006 0.027 0.005 5.36x10-9 1.709 1.428 2.047 cg22269526 0.034 0.013 0.030 0.010 2.23x10-9 1.329 1.210 1.458 cg06478094 0.021 0.009 0.023 0.012 1.65x10-5 0.807 0.732 0.890 cg00547758 0.015 0.003 0.015 0.003 9.22x10-7 2.491 1.730 3.587 cg14282180 0.018 0.003 0.017 0.003 2.11x10-13 3.874 2.710 5.539 cg14803009 0.044 0.110 0.044 0.110 0.889 1.001 0.992 1.010 cg03639557 0.017 0.013 0.025 0.024 1.04x10-13 0.617 0.546 0.696 cg11311499 0.027 0.006 0.025 0.006 1.11x10-12 1.906 1.596 2.275 cg02458113 0.014 0.008 0.014 0.010 0.723 1.021 0.911 1.145 cg09740554 0.074 0.030 0.073 0.034 0.367 1.014 0.984 1.046 cg23898128 0.981 0.006 0.982 0.005 3.65x10-5 0.687 0.575 0.821 cg19180827 0.875 0.054 0.872 0.055 0.395 1.008 0.990 1.026 cg07479270 0.876 0.137 0.771 0.292 1.00x10-13 1.024 1.019 1.030 cg03266686 0.413 0.064 0.407 0.063 0.126 1.012 0.997 1.028 cg15110473 0.979 0.010 0.979 0.010 0.079 0.917 0.832 1.010 cg04185310 0.854 0.077 0.851 0.076 0.484 1.005 0.992 1.017 cg00981060 0.889 0.035 0.894 0.036 1.91x10-4 0.948 0.922 0.975

248

cg10208034 0.975 0.012 0.976 0.012 0.021 0.906 0.834 0.985 cg11691189 0.928 0.045 0.932 0.044 0.079 0.980 0.959 1.002 cg00495909 0.969 0.015 0.968 0.019 0.956 0.998 0.945 1.055 cg15582102 0.942 0.037 0.940 0.038 0.462 1.010 0.984 1.036 cg04175739 0.087 0.047 0.083 0.052 0.059 1.019 0.999 1.040 cg14556391 0.039 0.011 0.039 0.025 0.848 0.995 0.948 1.045 cg00042186 0.151 0.041 0.154 0.045 0.082 0.980 0.958 1.003 cg19614911 0.307 0.039 0.309 0.039 0.169 0.983 0.959 1.007 cg26990681 0.950 0.110 0.951 0.113 0.351 0.996 0.987 1.005 cg00783525 0.945 0.017 0.949 0.023 1.09x10-5 0.895 0.852 0.940 cg16118678 0.883 0.041 0.897 0.045 1.64x10-12 0.920 0.899 0.941 cg02590088 0.947 0.029 0.952 0.026 1.60x10-5 0.922 0.888 0.957 cg13278115 0.912 0.029 0.916 0.030 2.83x10-4 0.940 0.909 0.972 cg17353680 0.788 0.097 0.788 0.095 0.695 0.998 0.988 1.008 cg22162312 0.941 0.027 0.945 0.028 0.010 0.955 0.922 0.989 cg27138584 0.041 0.014 0.039 0.015 0.001 1.129 1.051 1.214 cg13223402 0.024 0.013 0.023 0.014 0.030 1.107 1.010 1.213 cg02318629 0.066 0.027 0.057 0.028 5.28x10-8 1.126 1.079 1.175 cg03603951 0.040 0.032 0.032 0.023 1.31x10-6 1.115 1.067 1.166 cg04431946 0.126 0.071 0.118 0.068 0.047 1.014 1.000 1.029 cg08074851 0.106 0.056 0.099 0.057 0.036 1.019 1.001 1.036 cg01826863 0.132 0.041 0.112 0.034 1.24x10-13 1.158 1.122 1.195 cg13913015 0.028 0.025 0.027 0.031 0.846 1.003 0.969 1.039 cg27320127 0.250 0.054 0.248 0.053 0.497 1.006 0.988 1.025 cg04981611 0.053 0.020 0.041 0.024 4.10x10-14 1.295 1.232 1.361 cg04934807 0.080 0.023 0.074 0.022 1.36x10-5 1.118 1.063 1.175 cg04943225 0.368 0.067 0.368 0.070 0.555 1.004 0.990 1.019 cg14758072 0.516 0.078 0.520 0.078 0.324 0.994 0.981 1.006 cg10903274 0.589 0.062 0.611 0.081 1.24x10-9 0.958 0.945 0.971 cg09022430 0.891 0.050 0.897 0.051 0.002 0.970 0.952 0.989 cg04735632 0.866 0.052 0.874 0.054 0.001 0.970 0.952 0.988 cg11125315 0.989 0.008 0.989 0.005 0.161 0.864 0.704 1.060 cg13258091 0.947 0.055 0.847 0.185 3.02x10-16 1.086 1.071 1.101 cg13820434 0.966 0.029 0.963 0.039 0.203 1.020 0.990 1.050 cg14299961 0.990 0.003 0.990 0.003 0.169 0.811 0.603 1.093 cg08241610 0.985 0.010 0.987 0.003 1.66x10-7 0.494 0.379 0.643 cg09453252 0.987 0.005 0.985 0.011 1.73x10-4 1.438 1.190 1.739 cg06842253 0.979 0.008 0.981 0.007 1.15x10-6 0.693 0.598 0.804 cg10809134 0.899 0.051 0.906 0.050 3.68x10-4 0.965 0.946 0.984 cg23432368 0.789 0.061 0.795 0.057 0.002 0.974 0.958 0.990 cg24211994 0.908 0.047 0.910 0.046 0.119 0.983 1.633 1.004 cg10211062 0.791 0.051 0.793 0.055 0.219 0.989 0.970 1.007 cg07464408 0.963 0.019 0.964 0.019 0.025 0.942 0.895 0.992 cg01693539 0.977 0.006 0.972 0.008 1.05x10-21 2.820 2.375 3.348 cg00111466 0.722 0.042 0.734 0.065 1.57x10-7 0.952 0.935 0.970 cg27274072 0.975 0.008 0.974 0.007 0.006 1.202 1.055 1.369

249

cg22563036 0.899 0.019 0.904 0.028 1.12x10-5 0.912 0.875 0.950 cg25828259 0.986 0.003 0.986 0.003 0.047 0.691 0.480 0.995 cg12482860 0.985 0.006 0.986 0.007 2.78x10-8 0.646 0.554 0.754 cg22234368 0.808 0.064 0.799 0.077 0.004 1.021 1.007 1.035 cg02053451 0.931 0.035 0.926 0.033 0.024 1.034 1.004 1.064 ch.2.1149235F 0.093 0.053 0.149 0.124 2.46x10-16 0.937 0.925 0.948 cg07262620 0.917 0.047 0.917 0.046 0.512 0.993 0.972 1.014 ch.2.47844995F 0.010 0.002 0.009 0.002 0.001 1.982 1.325 2.967 ch.2.47848289F 0.012 0.003 0.010 0.003 6.66x10-11 3.192 2.253 4.523 cg14296255 0.986 0.006 0.986 0.007 0.140 0.893 0.768 1.038 cg05039065 0.040 0.013 0.038 0.012 0.001 1.141 1.056 1.232 cg02905881 0.019 0.013 0.028 0.027 1.53x10-13 0.677 0.612 0.750 cg11869233 0.022 0.033 0.067 0.082 1.30x10-13 0.834 0.806 0.862 cg09898070 0.181 0.038 0.145 0.056 7.06x10-19 1.158 1.133 1.184 cg16595246 0.033 0.008 0.034 0.009 0.003 0.845 0.757 0.944 cg18222961 0.027 0.004 0.027 0.005 0.696 1.042 0.848 1.281 cg00620552 0.019 0.003 0.019 0.003 0.144 1.315 0.911 1.897 cg21405109 0.025 0.004 0.028 0.005 1.18x10-17 0.225 0.172 0.294 cg19736286 0.019 0.006 0.024 0.008 2.04x10-14 0.324 0.269 0.391 cg19076255 0.012 0.002 0.013 0.002 4.40x10-10 0.150 0.083 0.273 cg07652213 0.011 0.002 0.011 0.002 0.715 1.119 0.612 2.047 cg15355298 0.060 0.046 0.109 0.120 2.04x10-8 0.934 0.921 0.946 cg19570558 0.861 0.049 0.846 0.064 2.67x10-7 1.051 1.031 1.071 cg04781916 0.832 0.114 0.831 0.120 0.398 1.004 0.995 1.012 Significant results are bolded when P<4.72x10-4. SD – standard deviation; CI – confidence interval

MLH1 Region Polymorphisms Show a SignificantAssociation with CpG Island Shore Methylation in a LargeCohort of Healthy IndividualsAndrea J. Savio1,2, Mathieu Lemire3, Miralem Mrkonjic1,2, Steven Gallinger1,2,4,5, Brent W. Zanke6,

Thomas J. Hudson3, Bharati Bapat1,2,7*

1Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada, 2 Samuel Lunenfeld Research Institute, Toronto, Ontario,

Canada, 3Ontario Institute for Cancer Research, Toronto, Ontario, Canada, 4Ontario Familial Colorectal Cancer Registry, Cancer Care Ontario, Toronto, Ontario, Canada,

5Department of Surgery, University of Toronto, Toronto, Ontario, Canada, 6Ottawa Hospital Research Institute, Ottawa, Ontario, Canada, 7Department of Pathology,

University Health Network, Toronto, Ontario, Canada

Abstract

Single nucleotide polymorphisms (SNPs) are the most common form of genetic variation. We previously demonstrated thatSNPs (rs1800734, rs749072, and rs13098279) in the MLH1 gene region are associated with MLH1 promoter islandmethylation, loss of MLH1 protein expression, and microsatellite instability (MSI) in colorectal cancer (CRC) patients. Recentstudies have identified less CpG-dense ‘‘shore’’ regions flanking many CpG islands. These shores often exhibit distinctmethylation profiles between different tissues and matched normal versus tumor cells of patients. To date, most epigeneticstudies have focused on somatic methylation events occurring within solid tumors; less is known of the contributions ofperipheral blood cell (PBC) methylation to processes such as aging and tumorigenesis. To address whether MLH1methylation in PBCs is correlated with tumorigenesis we utilized the Illumina 450 K microarrays to measure methylation inPBC DNA of 846 healthy controls and 252 CRC patients from Ontario, Canada. Analysis of a region of chromosome 3p21spanning the MLH1 locus in healthy controls revealed that a CpG island shore 1 kb upstream of the MLH1 gene exhibitsdifferent methylation profiles when stratified by SNP genotypes (rs1800734, rs749072, and rs13098279). Individuals withwild-type genotypes incur significantly higher PBC shore methylation than heterozygous or homozygous variant carriers(p,1.161026; ANOVA). This trend is also seen in CRC cases (p,0.096; ANOVA). Shore methylation also decreasessignificantly with increasing age in cases and controls. This is the first study of its kind to integrate PBC methylation at a CpGisland shore with SNP genotype status in CRC cases and controls. These results indicate that CpG island shore methylation inPBCs may be influenced by genotype as well as the normal aging process.

Citation: Savio AJ, Lemire M, Mrkonjic M, Gallinger S, Zanke BW, et al. (2012) MLH1 Region Polymorphisms Show a Significant Association with CpG Island ShoreMethylation in a Large Cohort of Healthy Individuals. PLoS ONE 7(12): e51531. doi:10.1371/journal.pone.0051531

Editor: William B. Coleman, University of North Carolina School of Medicine, United States of America

Received September 10, 2012; Accepted November 5, 2012; Published December 11, 2012

Copyright: ! 2012 Savio et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This study was funded by the Ontario Research Fund Global Leadership Round in Genomics and Life Sciences (GL2), ORF-RE File# GL2-01-043 (BWZ,TJH, SG, BB). This work was also supported by the National Cancer Institute, National Institutes of Health under RFA # CA-95-011 and through cooperativeagreements with members of the Colon Cancer Family Registry and PIs. The content of this manuscript does not necessarily reflect the views or policies of theNational Cancer Institute or any of the collaborating centers in the CFRs, nor does mention of trade names, commercial products, or organizations implyendorsement by the US government or the CFR. This work was also supported by funding within the Colon Cancer Familial Registry awarded to the OntarioRegistry for Studies of Familial Colorectal Cancer (Grant no. U01 CA074783, Principal Investigator SG). BWZ and TJH are recipients of Senior Investigator Awardsfrom the Ontario Institute for Cancer Research, through support from the Ontario Ministry of Research and Innovation. AJS was supported by the InterdisciplinaryHealth Research Team Program studentship funded by the Canadian Institutes of Health Research, the Samuel Lunenfeld Research Institute Studentship at MountSinai Hospital, and the University of Toronto Fellowship award. The funders had no role in study design, data collection and analysis, decision to publish, orpreparation of the manuscript.

Competing Interests: The authors have declared that no competing interests exist.

* E-mail: [email protected]

Introduction

Epigenetic mechanisms induce functionally relevant changes tothe genome without changing the nucleotide sequence itself. Thesemechanisms include DNA methylation, histone modifications andnon-coding RNAs. Of these, DNA methylation is the most studiedepigenetic mark, with clear links to a variety of diseasesestablished. In healthy individuals, genome-wide methylationlevels are generally elevated at intergenic regions and repetitivesequences (eg. ALU, LINE-1 repeats) while methylation is low ornon-existent in the promoter CpG islands of most genes. Thesemethylation patterns reverse with increasing age, as well as in

disease states, including cancer [1–3]. CpG islands, the sites of age-and cancer-specific epigenetic changes, are defined by a length ofat least 200 base pairs containing a GC percentage greater than50%, and an observed/expected CpG ratio over 0.60 [4]. Recentstudies suggest that many CpG islands are flanked by CpG island‘‘shores’’ which are less dense in CpG content than islands.Nonetheless, shores exhibit more readily distinguishable methyl-ation levels than islands between different tissues as well asbetween cancer and matched normal DNA [5]. The vast majorityof epigenetic studies have investigated methylation at CpG islands;however, the role of CpG island shore methylation is only justbeginning to be understood.

PLOS ONE | www.plosone.org 1 December 2012 | Volume 7 | Issue 12 | e51531

The majority of published studies have investigated DNAmethylation changes occurring at the tissue level in normal anddiseased states, while less is known about methylation occurring inperipheral blood cells (PBCs). Since blood samples are collectedeasily from patients, and can be measured at multiple time pointsduring disease progression, studying DNA methylation changes inPBCs can potentially be used as a biomarker for various diseaseoutcomes. Utilizing blood samples also allows comparison betweenhealthy controls with diseased patients. Using PBCs as an alternatebiological source has potential which requires further systematicinvestigation, such as integrating PBC methylation with knowledgeof the genetic and epigenetic landscape of tissue DNA.Single nucleotide polymorphisms, or SNPs, are the most

common form of genetic variation, with upwards of 3 millionSNPs characterized in the human genome by HapMap phase II[6]. Many SNPs have apparently benign phenotypic conse-quences, while others may predispose to various diseases such ascolorectal cancer (CRC) [7]. The underlying mechanism of actionof these SNP variants is not always understood. Recently, wedemonstrated that certain SNPs in the mutL homolog 1, colon cancer,nonpolyposis type 2 (E. coli) (MLH1) gene region are associated withMLH1 promoter CpG island methylation, loss of MLH1 proteinexpression, and tumour microsatellite instability (MSI) phenotypein CRC patients [8]. MLH1 is a key member of a group of DNAmismatch repair (MMR) genes [9]. Function of MLH1 is lost ina subset of CRC tumours, due to its inactivation through mutationor methylation. This leads to genome-wide accumulation of copynumber alterations at short tandem repeats, or microsatellites,termed microsatellite instability (MSI). Approximately 15% ofsporadic CRCs exhibit MSI and the majority of these occur due topromoter CpG island methylation of the MLH1 gene in colontumors [9,10].In previous studies, we examined 102 SNPs spanning 500 kb

surrounding the MLH1 locus [8,11]. Among these, we observedthree SNPs significantly associated with MLH1 methylation andtumour MSI, which were in strong linkage disequilibriumspanning 197 kb of the genomic region on chromosome 3 whichincludes MLH1, thus constituting a haplotype block at this region.These 3 SNPs include rs1800734 located 93 base pairs upstreamof the MLH1 start site, and rs749072 and rs13098279 which arelocated further downstream of MLH1. We have also shownthrough in vitro studies in transformed colon cancer cell lines thatthe allelic variant of rs1800734 decreases MLH1 promoter CpGisland-mediated transcriptional activity, thereby providing insightinto its potential role as a functional SNP [12].Taken together, we have demonstrated a link between these

SNPs and MLH1 CpG island methylation in CRC tumours, butthe potential correlation of these three SNPs with MLH1 shoremethylation has never been investigated, nor has it been analyzedin peripheral blood cells of normal healthy individuals. Since SNPsare static germline alterations, their potential modifier effects onmethylation may be exerted with varying capacity on all tissues ofthe body, including PBCs from patients. Thus, the goal of ourstudy was to examine the relationship between the threeaforementioned MLH1-region SNPs and the methylation statusof the MLH1 CpG island shore in PBCs obtained from a cohort of1,100 population-based healthy controls and CRC patients.

Materials and Methods

Ethics StatementBlood and tissue samples from CRC cases and controls were

obtained with informed written consent, following protocols

approved by the research ethics board of Mount Sinai Hospitaland the University of Toronto.

Study SubjectsStudy participants were recruited through the Ontario Familial

Colorectal Cancer Registry (OFCCR), one of six participatingcancer registries which are part of the Colon Cancer FamilyRegistry, a US National Cancer Institute-supported consortium.Both primary CRC cases and unaffected controls were accruedthrough population-based recruitment methods. A detailedaccount of patient accrual, data collection, and biologicalspecimen collection has been previously described [13,14]. Briefly,population control subjects were recruited via randomly selectedresidential telephone numbers in 1999–2000, and by population-based Tax Assessment Rolls of the provincial government,allowing the identification of age- and sex-matched controls.Due to the high proportion of self-reported Caucasians, patientswith non-white, unknown or mixed ethnic backgrounds wereexcluded. Of 2,736 individuals who agreed to participate, 1,336controls completed family, personal, and diet questionnaires,provided blood samples, and were self-reported as Caucasian.Ontario residents diagnosed with primary CRC from June 1, 1997to June 30, 2000 between the ages of 20 and 74 were eligible forrecruitment to the OFCCR. Cases of familial adenomatouspolyposis were excluded from the study and no related cases wereused. A total of 1,257 case patients remained after exclusion.

Single Nucleotide Polymorphism GenotypingThe SNPs chosen for study were selected based on extensive

database and literature searches of polymorphisms present on theAffymetrix GeneChip Human Mapping 100 K and 500 Kplatforms. rs749072 and rs13098279 were chosen because theseSNPs are in strong linkage disequilibrium with rs1800734 as wellas each other (r2.0.73 and D’.0.98). These 3 SNPs are all inHardy-Weinberg equilibrium (p,1024) [8].SNP genotyping was performed as described previously [8].

Briefly, peripheral blood cells (PBCs) were isolated from the bloodsamples provided by CRC cases and controls using Ficoll-Paquegradient centrifugation according to manufacturer’s protocol(Amersham Biosciences, Baie d’Urfe, Quebec, Canada). GenomicDNA was extracted from PBCs by phenol-chloroform or QiagenDNA extraction kit (Qiagen Inc., Montgomery Co., MD). Thefluorogenic 59 nuclease polymerase chain reaction (PCR) assaywas used to genotype rs1800734. This SNP was also genotypedusing the Affymetrix GeneChip Human Mapping 100 K and500 K platforms as part of the Assessment of Risk of ColorectalTumors in Canada (ARCTIC) project [11] and this data was usedas a cross-validation measure. In all, 11 of 1884 (0.58%) samplesgenotyped gave discordant results between the two platforms.Primer and probe sequences have been described previously[8,11]. The rs749072 and rs13098279 SNPs were genotyped usingthe Eurogentec qtPCR kit (Eurogentec, San Diego, CA).

Methylation MicroarrayCpG methylation was measured using Infinium HumanMethy-

lation450 BeadChips from Illumina (San Diego, CA). 998 controlsamples and 1,103 CRC samples were assayed on 96-well plates;a subset of 65 samples was analyzed in duplicate or triplicate withdata available for a total of 136 possible pairs. Bisulphiteconversion of DNA was performed using the EZ DNA Methyl-ation-Gold Kit (Zymo Research, Orange, CA). 500 ng ofbisulphite converted DNA was used for hybridization to the arrayfollowing Illumina Infinium HD Methylation Protocol. Theefficiency of bisulphite conversion was verified using internal

Association of SNPs with MLH1 Shore Methylation


control probes. We excluded from analysis samples that areoutliers with respect to internal control probes. Also excluded wereCRC cases given chemotherapeutic treatment prior to donation ofblood sample, and CRC cases with unknown chemotherapy status.After exclusion, 252 CRC samples and 846 controls remained.The methylation was measured at each CpG site using thefluorescent intensity ratio. After normalization using the internalnormalization probes, the resulting value was represented bya b value ranging from 0 (no methylation) to 1 (completemethylation). Values with a detection p-value above 0.01 wereremoved from analysis.

Selection of CpG SitesThe Infinium HumanMethylation450 BeadChip captures

methylation measurements at over 450,000 CpG sites across theentire genome. We chose every CpG site on chromosome 3between nucleotide positions 37,018,029 and 37,239,890 (GenomeBuild 37) spanning a 221 kb region for further analysis. There are70 CpG sites within this region, encompassing the genes EPM2A(laforin) interacting protein 1 (EPM2AIP1), MLH1, and leucine rich repeat(in FLII) interacting protein 2 (LRRFIP2). The SNPs rs1800734,rs749072, and rs13098279 also occur within this region. Thischromosomal region contains a CpG island shore upstream ofMLH1 within the coding region of EPM2AIP1. The entire shorespans from nucleotide 37,033,373 to 37,034,166 and contains13 CpG sites from the array. However, a section of the shore from37,033,373 to 37,034,166, which exhibited the most significantassociations and contains 7 CpG sites, will be the focus of ourresults.

StatisticsMethylation was compared between groups using analysis of

variance (ANOVA) with a significance level adjusted for multiplecomparisons. Groups compared were wild-type, heterozygous, andhomozygous variant groups of the three SNP genotypes. Partialcorrelation was utilized to compare age and methylation,controlling for sex. Gender differences in methylation were testedfor association using age at study recruitment as a covariate. Coloncancer diagnosis status was tested for association with percentagemethylation for each CpG site. Sex and age at study recruitmentwere used as covariates. All statistical analysis was performed usingSPSS PASW Statistics 18 (Chicago, IL).

Results

846 controls and 252 CRC cases from the Ontario FamilialColorectal Cancer Registry were successfully analyzed formethylation levels across the genome spanning 450,000 CpGsites. A mean correlation coefficient of 99.45% (range: 95.0–99.9%) was calculated from the comparison of methylationb values between all duplicate pairs. A 221 kb section of DNAfrom chromosome 3 containing MLH1 was chosen for statisticalanalysis based on the presence of 3 SNPs associated with MLH1promoter methylation and MSI. Of these, 846 controls and 252cases were successfully genotyped for rs1800734; 766 controls and235 cases were genotyped for rs749072 and rs13098279.Clinicopathological characteristics of the population used in thisstudy are shown in Table 1 along with genotypic frequencies forthe SNPs rs1800734, rs749072, and rs13098279 for cases andcontrols. There were no differences in SNP genotype frequenciesbetween different genders (Fisher’s exact test, p = 0.058 forrs1800734, p= 0.074 for rs749072, p = 0.081 for rs13098279) orany associations between genotype and age (ANOVA, p= 0.475for rs1800734, p = 0.637 for rs749072, p = 0.577 for rs13098279).

Results for the seven CpG sites in the MLH1 CpG island shore arediscussed in the text, while results for the entire 70 CpG sitesanalysed are shown in supplemental files.

PBC Methylation Differences among SNP GenotypesWe compared methylation in the MLH1 shore region between

different SNP genotypes of rs1800734, rs749072, and rs13098279in healthy individuals. The mean methylation for each SNPgenotype (wild-type, heterozygous, or homozygous variant) wascompared using ANOVA at 70 CpG sites. The results for thesesites from position 37,018,029 to 37,239,890 on chromosome 3 areshown in Table S1. The results of this analysis for the MLH1 CpGisland shore are shown in Table 2. There are seven CpG sites inthe shore region of interest, henceforth to be referred to as sites S1through S7. The mean methylation in the MLH1 shore amongindividuals stratified by SNP genotypes was highest among thewild-type genotype (GG) for rs1800734. The heterozygousgenotype (GA) had intermediate levels of methylation while thehomozygous variant allele (AA) had the lowest methylation. Thesedifferences in methylation among genotypes were statisticallysignificant for all 7 CpG sites localized to the MLH1 shore region.For example, at S1 mean wild-type methylation was 0.648,heterozygous was 0.607, and homozygous variant methylation was0.569 (p = 1.93610216). Similar results were obtained forrs749072 and rs13098279. At S1 rs749072 mean wild-typemethylation was 0.647, heterozygous methylation was 0.614,and homozygous variant methylation was 0.578 (p= 7.92610212).For rs13098270 at S1 mean wild-type methylation was 0.648,heterozygous methylation was 0.606, and homogyzous variantmethylation was 0.558 (p = 6.11610217).

Table 1. Characteristics of study population.

CRC Cases Controls

Characteristic N (%) N (%)

Female 122 (48.4) 356 (42.1)

Male 130 (51.6) 490 (57.9)

Age (in years) – mean(SD)

63.4 (8.4)/64.1 (8.3) 64.3 (8.2)

rs1800734 genotype

Homozygous wild-type(GG)

150 (59.5) 528 (62.5)

Heterozygous (GA) 96 (38.1) 264 (31.2)

Homozygous variant (AA) 6 (2.4) 53 (6.3)

rs749072 genotype

Homozygous wild-type(AA)

122 (51.9) 438 (51.8)

Heterozygous (AG) 104 (44.3) 271 (32.0)

Homozygous variant (GG) 9 (3.8) 57 (6.7)

rs13098279 genotype

Homozygous wild-type(GG)

147 (62.6) 491 (64.1)

Heterozygous (GA) 84 (35.7) 233 (27.5)

Homozygous variant (AA) 4 (1.7) 42 (5.5)

Distribution of clinicopathological features in primary colorectal carcinomasand controls from Ontario. Age at study recruitment is indicated for CRC casesand controls. Blood was drawn an average of less than one year and no morethan six years after study recruitment.SD = standard deviation.doi:10.1371/journal.pone.0051531.t001



We also performed the same analysis for CRC cases, shown inTable 3 for the MLH1 shore region. The results for all 70 CpGsites are shown in Table S2. However, some CRC cases hadundergone chemotherapy prior to providing blood samples for thisstudy (n= 292) and other cases had unknown chemotherapy status(n = 347). This left 252 CRC cases remaining who had definitivelynot received chemotherapy prior to blood donation. To ensurethat chemotherapy does not add a confounding factor to ouranalyses, we only included cases that had not been givenchemotherapy. Stratifying these remaining 252 CRC cases bySNP genotype, the same pattern was found as in controls: thoseindividuals with wild-type genotypes incur higher methylationthan those with any other genotype. For example, for rs1800734 atS1 in CRC cases, wild-type methylation was 0.631, heterozygousmethylation was 0.606, and homozygous variant methylation was0.550 (p= 0.01). Comparable significant results were found forrs749072 and rs13098279. Some, but not all, of the MLH1 shoreCpG sites show a significant association with SNP genotype. Thisis likely due to the smaller sample size of only 252 cases, comparedto the 846 controls utilized in a similar analysis.

Age-related Decrease in Methylation at the MLH1 ShoreRegionNormally, as individuals age, global hypomethylation of the

genome occurs combined with increases in methylation at specificgenes [3]. To investigate whether the MLH1 shore region exhibitsage-associated changes in methylation, correlation analysis wasperformed, controlling for sex, shown in Table 4. Results for all70 CpG sites analyzed are shown in Table S3. This was done incases and controls separately to confirm whether any age-associated changes in methylation at the shore were exclusive toCRC, or whether they occur in all individuals. There is a trendtowards decreasing methylation with increasing age in both casesand controls. In controls at sites S4 and S6 in the MLH1 shoreregion, there was a significant decrease in methylation with age.For example, at S4, R=20.170 (p= 1.3061026). Similarly,methylation also decreases with increasing age among our casepopulation, significantly so at site S6 [R=20.236(p = 4.0061024)].

Table 2. Methylation between SNP genotypes in healthy controls by ANOVA.

Chromosome 3Location Probe IDa CpG Site

Wild-typemean b value(SD)

Heterozygote meanb value (SD)

Homozygotevariantmean bvalue(SD) P-value

rs1800734 n=528 n=264 n=53

37,033,373 cg02103401 S1 0.644 (0.080) 0.607 (0.086) 0.567 (0.088) 5.99610218

37, 033,625 cg24607398 S2 0.786 (0.055) 0.758 (0.060) 0.740 (0.063) 3.91610217

37,033,632 cg10990993 S3 0.757 (0.051) 0.728 (0.056) 0.708 (0.054) 3.50610221

37,033,791 cg04726821 S4 0.255 (0.050) 0.229 (0.048) 0.204 (0.046) 1.51610222

37,033,894 cg11291081 S5 0.125 (0.035) 0.117 (0.031) 0.106 (0.029) 1.11610206

37,033,903 cg05670953 S6 0.210 (0.053) 0.194 (0.051) 0.176 (0.047) 5.88610209

37,033,980 cg18320188 S7 0.124 (0.021) 0.118 (0.020) 0.113 (0.019) 7.31610207

rs749072 n=438 n=271 n=57

37,033,373 cg02103401 S1 0.644 (0.082) 0.615 (0.085) 0.579 (0.086) 1.20610211

37, 033,625 cg24607398 S2 0.786 (0.056) 0.763 (0.059) 0.746 (0.057) 3.67610212

37,033,632 cg10990993 S3 0.755 (0.051) 0.735 (0.057) 0.717 (0.056) 2.53610211

37,033,791 cg04726821 S4 0.254 (0.051) 0.233 (0.049) 0.213 (0.048) 3.92610214

37,033,894 cg11291081 S5 0.125 (0.035) 0.118 (0.031) 0.110 (0.031) 1.29610204

37,033,903 cg05670953 S6 0.209 (0.054) 0.197 (0.051) 0.181 (0.049) 8.79610206

37,033,980 cg18320188 S7 0.124 (0.022) 0.118 (0.019) 0.112 (0.019) 2.20610207

rs13098279 n=491 n=233 n=42

37,033,373 cg02103401 S1 0.644 (0.081) 0.607 (0.084) 0.557 (0.086) 3.04610217

37, 033,625 cg24607398 S2 0.785 (0.055) 0.758 (0.059) 0.735 (0.059) 6.43610216

37,033,632 cg10990993 S3 0.756 (0.051) 0.729 (0.055) 0.705 (0.056) 2.32610218

37,033,791 cg04726821 S4 0.254 (0.051) 0.228 (0.047) 0.204 (0.048) 3.82610220

37,033,894 cg11291081 S5 0.124 (0.034) 0.117 (0.031) 0.106 (0.030) 2.84610205

37,033,903 cg05670953 S6 0.209 (0.053) 0.194 (0.052) 0.171 (0.047) 6.98610208

37,033,980 cg18320188 S7 0.124 (0.021) 0.118 (0.019) 0.111 (0.020) 2.92610207

Mean b value of each genotype of the SNPs rs1800734, rs749072, and rs13098279 in healthy controls from Ontario at seven sites in the MLH1 CpG island shore.Chromosome 3 locations and Probe IDs are the same for CpG sites S1–S7 in subsequent tables. Significant results are bolded when p,0.001.aProbe ID according to Illumina Infinium HumanMethylation450 array, used throughout in tables.CI = confidence interval.doi:10.1371/journal.pone.0051531.t002



PBC Methylation Differences among Males and FemalesPrevious studies have demonstrated that MLH1 tumor methyl-

ation is more prevalent among female, MSI positive CRC patients.Therefore, we compared MLH1 methylation levels in PBCsbetween males and females to determine whether gender plays

a role in this regard. The results for the MLH1 shore are found inTable 5, and for all 70 CpG sites analyzed in Table S4. We testedfor association using binomial logistic regression using age asa covariate in all cases and controls. For most CpG sites, there areno significant differences in methylation between genders. At S5and S6, methylation in females is significantly higher than in males(S5: 0.126 vs. 0.118; S6: 0.214 vs. 0.195). For S5, p = 7.05610204,95% CI: 0.939 (0.905–0.974); for S6, p = 2.94610204, 95% CI:0.939 (0.917–0.962).

PBC Methylation Differences among CRC Cases andControlsWe compared methylation in the MLH1 gene region between

CRC patients and healthy controls in 253 cases and 845 controls.A visual representation of case and control methylation at each ofthe 70 sites analyzed is shown in Figure 1. We tested forassociation between methylation level and presence of CRC (vs.controls), utilizing sex and age as covariates by binomial logisticregression. The results of this analysis and mean methylation ateach CpG site for the MLH1 CpG island shore in cases andcontrols are shown in Table 6, and for all 70 CpG sites in TableS5. Though mean methylation in controls is higher than in cases,

Table 3. Methylation between SNP genotypes in CRC cases by ANOVA.

CpG SiteWild-type mean b value(SD)

Heterozygote mean b value(SD)

Homozygote variant meanb value (SD) P-value

rs1800734 n=150 n=96 n=6

S1 0.631 0.606 0.550 0.010

S2 0.781 0.755 0.729 6.77610204

S3 0.746 0.725 0.716 0.006

S4 0.251 0.222 0.200 2.06610205

S5 0.125 0.114 0.118 0.096

S6 0.206 0.189 0.186 0.060

S7 0.124 0.116 0.120 0.028

rs749072 n=122 n=103 n=9

S1 0.630 0.617 0.590 0.282

S2 0.783 0.762 0.745 0.008

S3 0.747 0.732 0.716 0.042

S4 0.251 0.228 0.207 6.64610204

S5 0.126 0.116 0.106 0.059

S6 0.207 0.190 0.190 0.078

S7 0.125 0.116 0.111 0.007

rs13098279 n=147 n=84 n=4

S1 0.631 0.612 0.550 0.060

S2 0.781 0.758 0.746 0.010

S3 0.747 0.728 0.708 0.019

S4 0.250 0.222 0.189 9.65610205

S5 0.125 0.114 0.118 0.150

S6 0.206 0.189 0.170 0.047

S7 0.124 0.116 0.113 0.020

Mean b value of each genotype of the SNPs rs1800734, rs749072, and rs13098279 in CRC patients from Ontario at seven sites in the MLH1 CpG island shore. Significantresults are bolded when p,0.001.doi:10.1371/journal.pone.0051531.t003

Table 4. Correlation between age and methylation.

CpG Site Controls R P-value CRC Cases R P-value

S1 20.081 0.021 0.042 0.536

S2 20.085 0.016 20.037 0.582

S3 20.087 0.014 20.120 0.074

S4 20.170 1.30610206 20.209 0.002

S5 20.101 0.004 20.115 0.089

S6 20.200 1.16610208 20.236 4.00610204

S7 20.007 0.846 0.017 0.797

Partial correlation, controlling for sex, between age and methylation at sevensites in the MLH1 CpG island shore for CRC cases and controls. Significant resultsare bolded when p,0.001.doi:10.1371/journal.pone.0051531.t004



there is no significant association found between methylation andhealthy or diseased state.

No Association between MSI Status and MethylationMethylation of the MLH1 promoter CpG island is a common

occurrence in tumor tissue in MSI CRC [10]. We found noassociation between tumor MSI status and methylation at eitherthe MLH1 CpG island or shore in PBC DNA of CRC cases, whentested using binomial logistic regression with age and sex ascovariates (data not shown).

Methylation Levels of the MLH1 CpG Island and Shore inPBCsThe promoter of MLH1 spans from chromosome 3 nucleotide

position 37,034,130 to 37,034,856 (2711 to +15 relative to theMLH1 transcriptional start site) [15]. We investigated themethylation status of this promoter island in our PBC samples.The Illumina Infinium HumanMethylation450 microarrays con-tain 16 CpG sites located within the MLH1 promoter. We foundthat overall, methylation is very low among both cases andcontrols in PBCs, it does not differ significantly when stratified bySNP genotypes, and is not significantly correlated with age. Themean methylation for the CpG sites ranges from 0.004 to 0.064.

Differences in methylation among cases, controls, and SNPgenotypes and correlations with age can be found for thepromoter CpG island region in Tables S1, S2, S3, S4.CpG island shores can flank CpG islands of genes, being located

upstream and/or downstream. In addition to the shore locatedupstream of the promoter CpG island, which is the focus of thisinvestigation, MLH1 also has a shore downstream of its island.There are only two CpG sites on the Illumina microarrays whichinterrogated methylation at this region, at 37,035,399 and37,036,726. Results, though not significant, for this methylationat this downstream shore can be found in Tables S1, S2, S3, S4.

Discussion

In this study, we measured methylation in PBC DNA of a largeseries of healthy individuals, as well as CRC cases, using theIllumina Infinium HumanMethylation450 arrays. We integratedthis methylation data with SNP profiling data previously generatedby our group for the same controls and cases [8] and found novel,significant associations at the MLH1 CpG island shore. We havedemonstrated that differences in MLH1 shore region methylationamong PBCs are significantly associated with distinct genotypicvariants in theMLH1 gene region. Specifically, a CpG island shore1 kb upstream of the MLH1 start site exhibits associations between

Table 5. Associations between gender and methylation by logistic regression.

CpG SiteMale Mean b Value (SD)(n=617)

Female Mean b Value (SD)(n=476) P-value Effect Size Lower 95% CI Upper 95% CI

S1 0.623 (0.083) 0.632 (0.087) 0.058 1.014 1.000 1.028

S2 0.773 (0.060) 0.775 (0.059) 0.433 1.008 0.988 1.029

S3 0.745 (0.053) 0.745 (0.056) 0.624 1.006 0.984 1.028

S4 0.244 (0.051) 0.244 (0.053) 0.680 1.005 0.982 1.029

S5 0.126 (0.036) 0.118 (0.032) 0.001 0.939 0.905 0.974

S6 0.214 (0.055) 0.195 (0.050) 0.003 0.939 0.917 0.962

S7 0.123(0.022) 0.120 (0.020) 0.030 0.938 0.885 0.994

Mean b value of is shown for males and females along with logistic regression analysis at seven CpG sites in the MLH1 CpG island shore. Analysis of male versus femalemethylation is adjusted for age. Significant results are bolded when p,0.001.doi:10.1371/journal.pone.0051531.t005

Figure 1. Locations of CpG sites and methylation between cases and controls. Pictured are the 70 CpG sites analyzed, with indicatedchromosomal positions located on chromosome 3. The CpG sites are located within the EPM2AIP1, MLH1, and LRRFIP2 genes, with gene exons andtranscriptional directions indicated. CpG islands are indicated in green. The seven CpG sites of the MLH1 shore are highlighted in red. Each vertical barrepresents a CpG site, with control methylation, n = 846, displayed to the left and CRC case methylation, n = 252, displayed to the right of the whitedotted line. Controls and CRC case samples are displayed layered horizontally from highest methylation to lowest methylation. The distribution ofdegree of methylation in cases and controls is represented by the colour variation, according to the scale.doi:10.1371/journal.pone.0051531.g001



methylation in PBC DNA in controls with wild-type genotypes ofSNPs located over 1 kb away (rs1800734, rs749072) and up to200 kb away (rs13098279) from this shore region. The variantalleles of these three SNPs are associated with reduced methylationat CpG sites within the MLH1 shore, significantly lower thaneither the heterozygous or homozygous wild-type alleles in PBCs.Results also show that methylation of this shore decreases with age,in healthy individuals and CRC cases. Such associations betweenPBC methylation and genetic variants in a shore region have untilnow not been described.Though the concept of CpG islands dates back to the 1980s

[16], CpG island shores are a newer element of methylationphenomena that has emerged in recent years [5]. Shores areregions of the genome that flank some CpG islands and havea lower GC content than islands do. Despite this distance fromgenes and decreased CpG content, methylation of CpG islandshores are reported to display more specificity between differenttissues, and between normal and cancerous cells from the samepatients [5]. Gene expression is also strongly related to shoremethylation [5,17]. In genome-wide methylation analysis, over50% of the differentially methylated regions between normal colontissue and tumor tissue were located in shores, rather than islands[5]. Shore methylation has also been shown to discriminatebetween benign and malignant peripheral nerve sheath tumors[17]. Recent studies have demonstrated that shore methylationdecreases with increasing age, concomitantly with global hypo-methylation [18]. This is consistent with our results, which showeda decrease in methylation with increasing age at the MLH1 shore.Though much remains to be discovered about the importance andregulation of shores, methylation at these regions shows potentialat discriminating among different tissues, between normal anddiseased states, different genotypes, and age.Earlier studies have shown that DNA sequence can affect

methylation at nearby loci [19,20], as we have demonstrated inour results. More recently it was verified that SNP-dependentDNA methylation alterations can also play a role in disease[21,22]. We previously reported a significant association betweenthe MLH1 promoter SNP (rs1800734) and MSI CRCs, andsubsequently showed this association being mediated via MLH1promoter hypermethylation and loss of MLH1 protein expressioncontributing to MSI CRC tumors [8,11]. We further assessed therole of this variant by measuring transcriptional activity of theMLH1 promoter CpG island of transformed colon cancer celllines. Cells possessing the variant allele of rs1800734 exhibiteddecreased transcription compared to wild-type [12]. Though wedid not find that rs1800734 increased the overall risk of CRC, only

the risk of the MSI phenotype of CRC, a subsequent meta-analysiswas performed by another group, which included our data in theanalysis. It was found that indeed, the variant allele of this SNP isa modest but significant risk factor for CRC overall, with an oddsratio (95% confidence interval) of 1.06 (1.00–1.11; p = 0.037) [22].Though we did not find any associations between PBC shoremethylation and CRC status, we have clearly demonstrated thatthese MLH1-region SNPs show a strikingly significant associationwith shore methylation in the peripheral blood of healthyindividuals. Perhaps this variant-associated hypomethylation alonedoes not cause cancer, but in combination with other genetic,epigenetic, and environmental alterations of an individual, it mayserve as a low-penetrance susceptibility marker.Alternatively, there is a possibility that the SNPs rs1800734,

rs749072, and rs13098279 are actually linked to a different rarefunctional variant which is causing these outcomes. Though thereis currently no known rare variant in the MLH1 SNP haplotypeblock, other studies have analyzed chromosomal regions linked todisease in order to determine the underlying causative variants.For example, microsatellite fine mapping in an affected familydetermined that a 1.3 Mbp interval of chromosome 1 containeda rare mutation in the gene UbiA prenyltransferase domain containing 1,the cause behind Schnyder crystalline corneal dystrophy [23].Another possibility is that our SNPs serve a currently unknownfunction. For example, the 8q24 susceptibility locus for breast,prostate, and colorectal cancers [14,24] contains several SNPswith functional consequences. Rs378854 variant reduces bindingof the YY1 transcription factor, leading to increased expression ofPvt1 in prostate cancer cell lines [25] while rs6983267 affectsbinding of the transcription factor TCF4 in CRC cells [26]. Anyfunction of our SNPs or linkage to another variant is currentlyunknown, however, and warrants further investigation.One caveat concerning our results is the inability to ascribe our

measured PBC methylation to a specific blood cell type. Peripheralblood consists of natural killer cells, B cells, T cells, monocytes, andgranulocytes, each with their own epigenetic profiles. Genome-wide methylation measurements using Illumina 27 K arrays havehighlighted regions differentially methylated between differentperipheral blood cell populations [27]. Also, peripheral bloodsubpopulations change with increasing age [28,29]. Thus, wecannot say for certain whether the methylation changes we see atthe MLH1 shore are present in all PBC types, or perhaps just ina certain subpopulation of the cells, which may also be affected byage. Perhaps the variant-associated hypomethylation we see isparticularly pronounced in some PBC types but not others. Whatwe do know is that overall in PBC samples, regardless of cell

Table 6. Logistic regression analysis for association with methylation between CRC cases and controls.

CpG SiteControl Mean b Value(SD) (n =846)

Case Mean b Value (SD)(n=252) P-value Effect Size Lower 95% CI Upper 95% CI

S1 0.630 (0.085) 0.620 (0.086) 0.109 1.014 0.997 1.031

S2 0.775 (0.059) 0.770 (0.059) 0.182 1.016 0.992 1.041

S3 0.747 (0.055) 0.738 (0.055) 0.017 1.032 1.006 1.059

S4 0.245 (0.051) 0.238 (0.053) 0.040 1.031 1.001 1.06

S5 0.122 (0.032) 0.121 (0.038) 0.464 1.016 0.973 1.061

S6 0.204 (0.052) 0.199 (0.057) 0.058 1.028 0.999 1.057

S7 0.121 (0.020) 0.121 (0.023) 0.768 1.01 0.943 1.082

Mean b value of CRC cases and controls is shown along with logistic regression analysis at seven CpG sites in the MLH1 CpG island shore. Analysis of CRC cases versuscontrols is adjusted for age and sex. Effect size represents the increased risk of CRC per 1% reduction in methylation.doi:10.1371/journal.pone.0051531.t006



populations, there are noticeable significant changes in methyla-tion at the MLH1 shore region.Overall, this study has numerous strengths. Our large sample

size offers high statistical power utilizing both CRC cases andcontrols. With more than 800 control samples we were able todistinguish differences in methylation based on age and stratifiedby SNP genotype. Patient and control clinicopathological featureshave been extensively characterized, as has the epigenetic andgenetic features of the MLH1 gene region. We have now furtherdescribed the epigenomic landscape of MLH1 by assessingmethylation at its CpG island shore. Our study also benefits fromthe use of PBC DNA. Blood is an easily accessible biologicalpatient material which can offer information about permanentchanges such as germline genetic alterations (SNPs) as well as thevarying epigenetic changes resulting in response to both geneticand environmental sources. We have found associations in healthycontrols with age and SNP genotype in PBCs. What remains to beseen is whether these patterns exist in other tissues, such as thenormal colon, and colon tumour tissue. Additional work for thefuture includes further analyzing our data garnered from theInfinium HumanMethylation450 BeadChips, arrays which offercomprehensive genome-wide methylation analysis at nearly halfa million CpG sites. Thus far we have studied a small region of thegenome and found exciting associations. Further probing of themethylomes of our CRC cases and controls may reveal othergenomic regions with detectable differences in methylationbetween cancer and control, SNP variants, gender, age, tumorsubtype, and other variables.In summary, this novel study has demonstrated associations

between SNP variants at 3p21 with methylation at a CpG islandshore ofMLH1 in peripheral blood cells of 1,100 population-basedcontrols and CRC patients. Our results have also shown anassociation with decreasing methylation at the shore with age,which may add another facet to potential roles of shoremethylation and how it can incur changes based on tissue,presence of cancer, and environment. It is clear that these 3 SNPvariants in the MLH1 region play many roles in colorectal

tumorigenesis, including the regulation ofMLH1 methylation at itsCpG island shore and island.

Supporting Information

Table S1 Mean methylation between SNP genotypes forcontrols.(DOCX)

Table S2 Mean methylation between SNP genotypes for CRCcases.(DOCX)

Table S3 Correlation between age and methylation.(DOCX)

Table S4 Logistic regression analysis for gender.(DOCX)

Table S5 Regression analysis for CRC cases vs. controls.(DOCX)

Acknowledgments

This work was undertaken at the Department of Laboratory Medicine andPathobiology at the University of Toronto, Toronto, Ontario, Canada andwas conducted at the Samuel Lunenfeld Research Institute, Mount SinaiHospital, Toronto, Ontario, Canada. We sincerely thank the investigators,staff, and participants of the Colon Cancer Family Registry for theirdedicated contributions leading to this work. The content of thismanuscript does not necessarily reflect the views or policies of the NationalCancer Institute or any of the collaborating centers in the CFRs, nor doesmention of trade names, commercial products, or organizations implyendorsement by the US government or the CFR. We also thank ColleenAsh for her assistance.

Author Contributions

Conceived and designed the experiments: SG BWZ TJH BB. Performedthe experiments: AS ML MM. Analyzed the data: AS ML MM TJH BB.Contributed reagents/materials/analysis tools: SG. Wrote the paper: ASML MM BWZ TJH BB.

References

1. Herman JG, Baylin SB (2003) Gene silencing in cancer in association withpromoter hypermethylation. N Engl J Med 349: 2042–2054.

2. Toyota M, Issa JP (1999) CpG island methylator phenotypes in aging andcancer. Semin Cancer Biol 9: 349–357.

3. Jones PA, Baylin SB (2007) The epigenomics of cancer. Cell 128: 683–962.4. Gardiner-Garden M, Frommer M (1987) CpG islands in vertebrate genomes.

J Mol Biol 196: 261–282.5. Irizarry RA, Ladd-Acosta C, Wen B, Wu Z, Montano C, et al. (2009) The

human colon cancer methylome shows similar hypo- and hypermethylation atconserved tissue-specific CpG island shores. Nat Genet 41: 178–186.

6. The International HapMap Consortium (2007) A second generation humanhaplotype map of over 3.1 million SNPs. Nature 449: 851–861.

7. Houlston RS, Webb E, Broderick P, Pittman AM, Di Bernardo MC, et al. (2008)Meta-analysis of genome-wide association data identifies four new susceptibilityloci for colorectal cancer. Nat Gen 40: 1426–1435.

8. Mrkonjic M, Roslin NM, Greenwood CM, Raptis S, Pollett A, et al. (2010)Specific variants in the MLH1 gene region may drive DNA methylation, loss ofprotein expression, and MSI-H colorectal cancer. PLoS One 5: e13314.

9. Boland CR, Goel A (2010) Microsatellite instability in colorectal cancer.Gastroenterology 138: 2073–2087.

10. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, et al. (1997) Methylationof the hMLH1 promoter correlates with a lack of expression of hMLH1 insporadic colon tumors and mismatch repair-defective human tumor cell lines.Cancer Res 57: 808–811.

11. Raptis S, Mrkonjic M, Green RC, Pethe VV, Monga N, et al. (2007) MLH1–93G.A promoter polymorphism and the risk of microsatellite-unstablecolorectal cancer. J Natl Cancer Inst 99: 463–474.

12. Perera S, Mrkonjic M, Rawson JB, Bapat B (2011) Functional effects of theMLH1–93G.A polymorphism on MLH1/EPM2AIP1 promoter activity.Oncol Rep 25: 809–815.

13. Newcomb PA, Baron J, Cotterchio M, Gallinger S, Grove J, et al. (2007) ColonCancer Family Registry: an international resource for studies of the genetic

epidemiology of colon cancer. Cancer Epidemiol Biomarkers Prev 16: 2331–2342.

14. Zanke BW, Greenwood CM, Rangrej J, Kustra R, Tenesa A, et al. (2007)Genome-wide association scan identifies a colorectal cancer susceptibility locuson chromosome 8q24. Nat Genet 39: 989–994.

15. Deng G, Chen A, Hong J, Chae HS, Kim YS (1999) Methylation of CpG ina small region of the hMLH1 promoter invariably correlates with the absence ofgene expression. Cancer Res 59: 2029–2033.

16. Bird A, Taggart M, Frommer M, Miller OJ, Macleod D (1985) A fraction of themouse genome that is derived from islands of nonmethylated, CpG-rich DNA.Cell 40: 91–99.

17. Feber A, Wilson GA, Zhang L, Presneau N, Idowu B, et al. (2011) Comparativemethylome analysis of benign and malignant peripheral nerve sheath tumors.Genome Res 21: 515–524.

18. Heyn H, Li N, Ferreira HJ, Moran S, Pisano DG, et al. (2012) Distinct DNAmethylomes of newborns and centenarians. Proc Natl Acad Sci USA 109:10522–10527.

19. Gibbs JR, van der Brug MP, Hernandez DG, Traynor BJ, Nalls MA, et al.(2010) Abundant quantitative trait Loci exist for DNA methylation and geneexpression in human brain. PLoS Genet 6: e1000952.

20. Muller K, Heller H, Doerfler W (2001) Foreign DNA integration. Genome-wideperturbations of methylation and transcription in the recipient genomes. J BiolChem 276: 14271–14278.

21. Bell CG, Finer S, Lindgren CM, Wilson GA, Rakyan VK, et al. (2010)Integrated genetic and epigenetic analysis identifies haplotype-specific methyl-ation in the FTO type 2 diabetes and obesity susceptibility locus. PLoS One 5:e14040.

22. Whiffin N, Broderick P, Lubbe SJ, Pittman AM, Penegar S, et al. (2011) MLH1–93G.A is a risk factor for MSI colorectal cancer. Carcinogenesis 32: 1157–1161.



23. Jiang Haiyan, Orr A, Guernsey DL, Robitaille J, Asselin G, et al. (2009)Application of homozygosity haplotype analysis to genetic mapping with high-density SNP genotype data. PLoS ONE 4: e5280.

24. Schumacher FR, Feigelson HS, Cox DG, Haiman CA, Albanes D, et al. (2007)A common 8q24 variant in prostate and breast cancer from a large nested case-control study. Cancer Res 67: 2951–2956.

25. Meyer KB, Maia AT, O’Reilly M, Ghoussaini M, Prathalingam R, et al. (2011)A functional variant at a prostate cancer predisposition locus at 8q24 isassociated with PTVT1 expression. PLoS Genet 7: e1002165.

26. Tuupanen S, Turunen M, Lehtonen R, Hallikas O, Vanharanta S, et al. (2009)The common colorectal cancer predisposition SNP rs6983267 at chromosome8q24 confers potential to enhanced Wnt signaling. Nat Genet 41: 885–890.

27. Koestler DC, Marsit CJ, Christensen BC, Accomando W, Langevin SM, et al.(2012) Peripheral blood immune cell methylation profiles are associated withnonhematopoietic cancers. Cancer Epidemiol Biomarkers Prev 21: 1293–1302.

28. Arnold CR, Wolf J, Brunner S, Herndler-Brandstetter D, Grubeck-LoebensteinB (2011) Gain and loss of T cell subsets in old age – age-related reshaping of theT cell repertoire. J Clin Immunol 31: 137–146.

29. Perez-Andres M, Paiva B, Nieto WG, Caraux A, Schmitz A, et al. (2010)Human Peripheral blood B-cell compartments: a crossroad in B-cell traffic.Cytometry B Clin Cytom 78: S47–S60.



RESEARCH ARTICLE Open Access

Promoter methylation of ITF2, but not APC,is associated with microsatellite instabilityin two populations of colorectal cancerpatientsAndrea J. Savio1,2, Darshana Daftary2,3, Elizabeth Dicks4, Daniel D. Buchanan5,6, Patrick S. Parfrey4,Joanne P. Young7, Daniel Weisenberger8, Roger C. Green4, Steven Gallinger1,2,3,9, John R. McLaughlin2,3,10,Julia A. Knight2,10 and Bharati Bapat1,2,11*

Abstract

Background: Aberrant Wnt signaling activation occurs commonly in colorectal carcinogenesis, leading toupregulation of many target genes. APC (adenomatous polyposis coli) is an important component of the β-catenindestruction complex, which regulates Wnt signaling, and is often mutated in colorectal cancer (CRC). In addition tomutational events, epigenetic changes arise frequently in CRC, specifically, promoter hypermethylation which silencestumor suppressor genes. APC and the Wnt signaling target gene ITF2 (immunoglobulin transcription factor 2) incurhypermethylation in various cancers, however, methylation-dependent regulation of these genes in CRC has not beenstudied in large, well-characterized patient cohorts. The microsatellite instability (MSI) subtype of CRC, featuring DNAmismatch repair deficiency and often promoter hypermethylation of MutL homolog 1 (MLH1), has a favorable outcomeand is characterized by different chemotherapeutic responses than microsatellite stable (MSS) tumors. Other epigeneticevents distinguishing these subtypes have not yet been fully elucidated.

Methods: Here, we quantify promoter methylation of ITF2 and APC by MethyLight in two case-case studies nested inpopulation-based CRC cohorts from the Ontario Familial Colorectal Cancer Registry (n = 330) and the NewfoundlandFamilial Colorectal Cancer Registry (n = 102) comparing MSI status groups.

Results: ITF2 and APC methylation are significantly associated with tumor versus normal state (both P < 1.0×10-6). ITF2 ismethylated in 45.8 % of MSI cases and 26.9 % of MSS cases and is significantly associated with MSI in Ontario (P = 0.002)and Newfoundland (P = 0.005) as well as the MSI-associated feature of MLH1 promoter hypermethylation (P = 6.72×10-4).APC methylation, although tumor-specific, does not show a significant association with tumor subtype, age, gender, orstage, indicating it is a general tumor-specific CRC biomarker.

Conclusions: This study demonstrates, for the first time, MSI-associated ITF2 methylation, and further reveals thesubtype-specific epigenetic events modulating Wnt signaling in CRC.

Keywords: Colorectal cancer, DNA methylation, Microsatellite instability, Wnt signaling, MethyLight

* Correspondence: [email protected] of Laboratory Medicine and Pathobiology, University ofToronto, Toronto, ON, Canada2Lunenfeld-Tanenbaum Research Institute of Mount Sinai Hospital, Toronto,ON, CanadaFull list of author information is available at the end of the article

© 2016 Savio et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Savio et al. BMC Cancer (2016) 16:113 DOI 10.1186/s12885-016-2149-9

BackgroundColorectal cancer (CRC) is one of the most commoncancers in the Western world and is marked by a highmortality rate [1]. Early detection of CRC is the key toimproved survival rates [2]. Another factor affectingdisease prognosis is CRC subtype [3]. The microsatelliteinstability (MSI) subtype of CRC accounts for approxi-mately 15 % of colorectal cancers [4]. MSI tumors aredistinguished by defects in the DNA mismatch repairsystem which leads to mutational insertions and dele-tions in short tandem repeats (microsatellites) of DNA[5]. MSI is most often due to promoter hypermethyla-tion and silencing of the MutL homolog 1 (MLH1) mis-match repair gene. Microsatellite stable (MSS) tumorsaccount for 85 % of CRCs and exhibit chromosomalinstability, including numerous chromosomal duplica-tions, deletions and rearrangements [6]. MSI tumorsdiffer from MSS tumors in several ways; MSI CRCsexhibit proximal colonic location, increased lymphocyticinfiltration, and poorer response to chemotherapeuticdrugs [7, 8]. MSI CRCs also demonstrate better progno-sis at stages I-III, however, some studies suggest poorprognosis at stage IV, though metastatic MSI cases arerare [7, 9]. A third CRC subtype, the CpG island methyl-ator phenotype (CIMP), is characterized by widespreadDNA hypermethylation of CpG-rich promoter islands.CIMP can exist concurrently with either the MSI orMSS phenotype, though it is more frequently found intandem with MSI and MLH1 hypermethylation [10].The prognostic significance of CIMP is currently un-defined and may be modified by MSI status, presence ofBRAF mutation, tumor stage, or other factors [11–13].Recently, a classification system for further subtyping ofCRC has been proposed, consisting of four subtypes[14]. One subtype consists mostly of MSI cases, whilethe other three are able to categorize the remainder ofcases by Wnt signaling activation, metabolic dysregula-tion, or mesenchymal activation.The vast majority (up to 94 %) of CRCs feature dysreg-

ulation in the Wnt signaling pathway [15]. Wnt signalingis important in normal development, cell growth andproliferation, but when inappropriately activated mayalso lead to tumor initiation and development [16]. Incanonical Wnt signaling, β-catenin accumulates withinthe cell, enters the nucleus and activates transcription oftarget genes, such as c-Myc and ITF2 (immunoglobulintranscription factor 2) [17, 18]. ITF2 is also known astranscription factor 4 (TCF4). In the absence of Wntsignaling, a β-catenin destruction complex includingadenomatous polyposis coli (APC) targets β-catenin forubiquitination followed by proteasomal degradation[17, 18]. In many cases of CRC, the APC gene is mu-tated, rendering it incapable of binding to β-catenin,which leads to β-catenin accumulation followed by its

nuclear translocation and subsequent activation ofdownstream target genes [18].Evidence for DNA methylation of the APC promoter

has been found in CRC. However, to what extent APCmethylation plays a role in colorectal carcinogenesis isunclear, as a broad range of methylation levels has beenfound in the literature, from 11 up to 63 % of tumorsmethylated [19–23]. Conflicting reports exist regardingthe extent of APC methylation in MSI CRCs. Somesmall-scale studies (MSI n ≤ 29) have suggested thatAPC methylation may be associated with the MSI sub-type, but others show no significant difference [21–27].Still another study has found APC methylation to beinversely correlated with CIMP but not MSI [28].The role of ITF2, a Wnt signaling target gene, is less

understood in CRC. It is a target of Wnt signaling and isoverexpressed in colon cancers with Wnt dysregulation [29].Its expression was reported elevated in some cancers withaberrant Wnt signaling activation but decreased in others[30, 31]. Among gastrointestinal malignancies, ITF2 methy-lation has been reported in gastric cancer, but its methyla-tion status has not been investigated in CRC [32, 33].Our group has previously demonstrated associations

between the methylation status of key Wnt signalingpathway regulatory genes and CRC subtype includingthe extracellular Wnt antagonists DKK1 and SFRP1 aswell as Wnt5a which is involved in non-canonical Wntactivity [34, 35].In this study, we have examined the role of APC and

ITF2 methylation in two nested case-case studies inCRC cohorts. These patients were recruited from twodistinct Canadian populations and the case groups werestratified by their MSI status.

MethodsStudy participantsParticipants of this study were population-based primaryCRC cases recruited through the Ontario Familial Colo-rectal Cancer Registry (OFCCR) and NewfoundlandFamilial Colorectal Cancer Registry (NFCCR). Proce-dures for patient accrual, biospecimen collection anddata collection for the OFCCR and NFCCR have beenpreviously described [36, 37]. Briefly, Ontario residentsbetween the ages of 20 and 74 diagnosed withpathology-confirmed primary CRC between 1997 and2000 were eligible for recruitment. Familial adenoma-tous polyposis cases were excluded and in the currentstudy non-white patients were also excluded due tothe high prevalence of self-reported Caucasians in thestudy (92.5 %). A total of 1168 participants have beenanalyzed for MSI status (see Molecular analysis below)of which 184 are MSI high (MSI-H). 165 of these MSI-H cases had available DNA of high quality. A matchedcase-case selection strategy was utilized to select 165

Savio et al. BMC Cancer (2016) 16:113 Page 2 of 11

patients with MSS tumors to match 165 patients withMSI-H tumors by sex, stage at diagnosis and age quar-tile. The 165 MSS tumors were selected from a totalof 384 MSS tumors available at the time this studywas undertaken. Population-based recruitment by theNFCCR was similar to the OFCCR, with a recruitmentperiod from 1999 to 2003 of cases from provincialtumor registries [37]. For the NFCCR, proxy consentfrom living family members was obtained for deceasedpatients leading to a high frequency of late-stagepatients. These tumor samples were not utilized inorder to maintain similar patient age and tumor stagebetween the Ontario and Newfoundland populations.102 tumor samples from 696 total CRC cases werechosen from probands of the NFCCR, 51 of whichwere MSI-H, matched to 51 MSS cases by sex, stage atdiagnosis and age quartile. Overall survival status,along with other patient clinicopathological features,is described in Table 1. Recurrence data was not avail-able for all cases used in this study, thus was notincluded in analysis. DNA from normal colonic mu-cosa was also available for all patients. Of the 330OFCCR and 102 NFCCR patients’ tumor samplesutilized, 47 were selected randomly for methylationanalysis of normal adjacent tissue. Patient data wasobtained through protocols approved by the ResearchEthics Boards of Mount Sinai Hospital, the Universityof Toronto and Memorial University of Newfoundland.All patients or their proxies provided informed consent.

Molecular analysisDNA used to assess MSI status was extracted fromarchival paraffin-embedded tumors microdissected toenrich for tumor cells. MSI status was assessed usingNational Cancer Institute guidelines using four ormore of the following markers: ACTC, BAT-25,BAT-26, BAT-40, BAT-34C4, D10S197, D18S55,D17S250, D5S346 and MYC-L. The numbers of posi-tive markers used to define MSI status are: MSI high(MSI-H), ≥30 % unstable markers; MSI low (MSI-L),1–29 % unstable markers; MSS, 0 % unstablemarkers [38]. Tumors with MSI-L status were notincluded in this study.Somatic T > A mutation of nucleotide 1799 in the BRAF

gene leading to the V600E mutation was determined byallele-specific PCR as described previously [34].Immunohistochemistry was used to determine pres-

ence of the mismatch repair proteins MLH1, MSH2,MSH6 and PMS2. Protein staining was classified aseither present, absent, or inconclusive. Tumors with-out positive staining for any of these proteins weredefined as mismatch repair deficient, as describedpreviously [39].

MethyLight analysisThe sensitive, semi-quantitative high-throughput Methy-Light assay was used to analyze the methylation of APCand ITF2 in tumor and normal colonic DNA of CRCpatients. DNA was treated with sodium bisulfite prior toMethyLight according to protocol using the EZ DNAMethylation Gold Kit (Zymo Research Corp, Orange,CA). Primers and probe were used to amplify a regionwithin the CpG island of promoter 1A of APC. Forwardprimer: 5′-GAACCAAAACGCTCCCCAT-3′. Probe: 5′-CCCGTCGAAAACCCGCCGATTA-3′. Reverse primer:5′-TTATATGTCGGTTACGTGCGTTTATAT-3′. Primersand probe were designed within the promoter region ofITF2. Forward primer: 5′-GAAGCGGTAATACGAATAAGAGC-3′. Probe: 5′-ATTCCCGAAACCGAAATCGTTCGCAAACC-3′. Reverse primer: 5′- AACTATTCTCGAATAAACGTCGC-3′. Alu-C4 was also amplified to normalizethe DNA input. Forward primer: 5′-GGTTAGGTATAGTGGTTTATATTTGTAATT-3′. Probe: 5′-CCTACCTTAACCTCCC-3′. Reverse primer: 5′-ATTAACTAAACTAATCTTAAACTCCTA-3′. Probes contained a 5′ fluorescent re-porter dye and a 3′ quencher dye. Samples were analyzedusing the ABI 7500 RT-PCR thermocycler in 96-well platesas previously described [40]. APC, ITF2 and Alu-C4 werealso amplified in exogenously methylated CpGenome DNA(Millipore, Billerica, MA). The percent methylated refer-ence (PMR) was calculated to assess the methylation usingthe formula: (Gene of Interest/Alu-C4)sample/(Gene of Inter-est/Alu-C4)CpGenomex100%. In order to ensure that DNAquality was adequate, samples with an Alu-C4 thresholdcycle greater than 22 were deemed poor quality and re-analyzed or removed from the study [41].MLH1 methylation status was assessed by MethyLight

as described previously, with positive methylationdefined as PMR ≥ 10 % [42]. CIMP status was deter-mined using the Weisenberger panel of markers, de-scribed previously [43]. Briefly, MethyLight was used toassess a 5-gene signature consisting of CACNA1G, IGF2,NEUROG1, RUNX3 and SOCS1. Tumors were classifiedas CIMP if 3 or more of 5 genes had PMR ≥ 10 % andnon-CIMP if 2 or fewer genes had PMR ≥ 10 %. CIMPstatus was available for a subset of Ontario cases (285 of330) and unavailable for Newfoundland cases.

Statistical analysisComparison of the methylation status of matched tumorand normal DNA samples was performed using McNe-mar’s test. Results were considered statistically significant iftwo-sided P < 0.05. Pearson’s chi-square test was used tomeasure associations between clinicopathological variablesand ITF2 and APC methylation in tumor DNA. Bonferronicorrection was used to account for multiple comparisons.All analyses were performed using PASW Statistics 21(SPSS Inc., Chicago, IL).


ResultsITF2 promoter methylation in CRC tumors and normalcolonic mucosaPatient clinicopathological features are shown inTable 1 for MSI-H and MSS cases for the Ontario andNewfoundland populations. We quantified promotermethylation of ITF2 using MethyLight in CRC tumorsfrom Ontario and Newfoundland. To quantify ITF2promoter methylation levels in both normal mucosaand CRC tumor tissue we tested 47 randomly selectednormal-tumor pairs from Ontario and NewfoundlandCRC cases, of which 12 were MSI-H and 35 wereMSS. The mean PMR was 8.8 % in tumor DNA and1.6 % in normal colonic DNA. Methylation levelsranged from 0–31.1 % in tumor and 0–15.6 % innormal colonic DNA. A PMR cut-off of 10 % was usedto dichotomize methylated and unmethylated samples.McNemar’s test comparing methylation above thiscut-off in tumor and normal tissues in CRC patientsdetermined tumor methylation of ITF2 to be signifi-cantly higher than normal colonic mucosa methylation(P < 1.0×10-6). For ITF2 promoter methylation, com-parable methylation levels were seen in CRC tumorsbetween the two populations, comprised of 165 MSI-Hand 165 MSS cases from Ontario and 51 MSI-H and 51MSS cases from Newfoundland. In the Ontario cases themean PMR for all 330 cases was 8.5 %. Methylation valuesranged from 0–57.2 %. Using the PMR cut-off of 10 %,34.2 % (113/330) of cases were considered positivelymethylated in the Ontario cohort. In Newfoundland casesthe mean PMR for all 102 cases was 15.4 %. Methylation

Table 1 Clinicopathological features of primary colorectalcarcinomas of patients from Ontario and Newfoundland

No. of cases (%)

Ontario Newfoundland

MSS MSI-H MSS MSI-H

Cases of primarycolorectal carcinoma

165 165 51 51

Mean age (±SDa) 59.9 (9.3) 60.1 (9.8) 58.3 (10.2) 58.4 (10.2)

Age

<50 19 (11.5) 28 (17.0) 11 (21.6) 10 (19.6)

50+ 146 (88.5) 137 (83.0) 40 (78.4) 41 (80.4)

Sex

Male 74 (44.8) 74 (44.8) 26 (51.0) 26 (51.0)

Female 91 (55.2) 91 (55.2) 25 (49.0) 25 (49.0)

TNM Stage

1 38 (23.0) 38 (23.0) 12 (23.5) 11 (21.6)

2 84 (50.9) 85 (51.5) 26 (51.0) 27 (52.9)

3 34 (20.6) 39 (23.6) 10 (19.6) 12 (23.5)

4 9 (5.5) 3 (1.8) 3 (5.9) 1 (2.0)

Histological Grade

Low 16 (9.7) 10 (6.1) 4 (7.8) 7 (13.7)

Moderate 123 (74.5) 45 (27.3) 37 (72.6) 36 (70.6)

High 13 (7.9) 20 (12.1) 8 (15.7) 8 (15.7)

Unavailable 13 (7.9) 90 (54.5) 2 (3.9)

Locationb

Distal 108 (65.5) 15 (9.1) 37 (72.5) 9 (17.6)

Proximal 51 (30.9) 63 (38.2) 14 (27.5) 42 (82.4)

Unavailable 6 (3.6) 87 (52.7)

Histological Typec

Non-Mucinous 143 (86.7) 107 (64.8) 46 (90.2) 42 (82.4)

Mucinous 19 (11.5) 53 (32.1) 5 (9.8) 9 (17.6)

Unavailable 3 (1.8) 5 (3.0)

MMR Protein Status

Intact 148 (89.7) 26 (15.8) 50 (98.0) 4 (7.8)

Deficient 4 (2.4) 136 (82.4) 0 (0.0) 46 (90.2)

Unavailable 13 (7.9) 3 (1.8) 1 (2.0) 1 (2.0)

MMR Germline Mutation

No 164 (99.4) 124 (75.2) 51 (100.0) 39 (76.5)

Yes 1 (0.6) 41 (24.8) 0 (0.0) 12 (23.5)

MLH1 Methylation

Unmethylated 159 (96.4) 87 (52.7) 36 (70.6) 23 (45.1)

Methylated 5 (3.0) 78 (47.3) 1 (2.0) 28 (54.9)

Unavailable 1 (0.6) 14 (27.4)

BRAF V600E Mutation

No 146 (88.5) 95 (57.6) 46 (90.2) 25 (49.0)

Yes 15 (9.1) 66 (40.0) 2 (3.9) 20 (39.2)

Unavailable 4 (2.4) 4 (2.4) 3 (5.9) 6 (11.8)

Table 1 Clinicopathological features of primary colorectalcarcinomas of patients from Ontario and Newfoundland(Continued)

CIMP Status

Negative 133 (80.6) 79 (47.9)

Positive 10 (6.1) 63 (38.2)

Unavailable 22 (13.3) 23 (13.9) 51 (100.0) 51 (100.0)

Survival Status

Alive 100 (60.6) 99 (60.0) 45 (88.2) 49 (96.1)

Deceased 65 (39.4) 66 (40.0) 6 (11.8) 2 (3.9)

ITF2 Methylation

Unmethylated 122 (73.9) 95 (57.6) 36 (70.6) 22 (43.1)

Methylated 43 (26.1) 70 (42.4) 15 (29.4) 29 (56.9)

APC Methylation

Unmethylated 112 (67.9) 103 (67.9) 33 (64.7) 28 (54.9)

Methylated 53 (32.1) 62 (37.6) 18 (35.3) 23 (45.1)aStandard DeviationbProximal tumor location includes lesions up to and including thesplenic flexurecMucinous histology includes the presence of any mucin within thetumor stroma


values ranged from 0–95.4 %. Using the PMR cut-off of10 %, 43.1 % (44/102) of cases were considered positivelymethylated in the Newfoundland cohort.

APC promoter methylation in CRC tumors and normalcolonic mucosaWe quantified methylation of the APC promoter region intumor and matched normal colonic mucosa of 47 ran-domly selected patients from both Ontario and New-foundland. Mean methylation was 16.1 % in tumor DNAand 2.6 % in normal colon. Methylation values rangedfrom 0–60.2 % for tumor samples and from 0–11.9 % fornormal samples. A PMR cut-off of 10 % was used todichotomize methylated and unmethylated samples.McNemar’s test comparing methylation above this cut-offin tumor and normal tissues in CRC patients determinedtumor methylation of APC to be significantly higher thannormal colonic mucosa methylation (P < 1.0×10-6).For APC promoter methylation, comparable methyla-

tion levels were seen in CRC tumors between the twopopulations. The mean PMR in Ontario was 11.5 %.Methylation values ranged from 0–92.2 %. For theNewfoundland samples, the mean PMR was 13.9 %.Methylation values ranged from 0–70.5 %. Using thePMR cut-off of 10 %, 34.8 % of tumors (115/330) weremethylated in the Ontario cohort and 40.2 % of tumors(41/102) were methylated in the Newfoundland cohort.

ITF2 methylation and clinicopathological features,including MSI subtype, in two distinct CRC cohortsWe examined whether methylation of ITF2 in tumorDNA was associated with patient clinicopathological fea-tures. Methylation status was compared by Pearson’schi-square test between MSI-H and MSS cases. In On-tario 26.1 % (43/165) of MSS cases compared to 42.4 %(70/165) of MSI-H cases were methylated, with an oddsratio (OR) of 2.09 [95 % confidence interval (CI) 1.31–3.33], and P = 0.002. Similarly in Newfoundland 29.4 %(15/51) of MSS cases and 56.9 % (29/51) of MSI-H caseswere methylated with OR of 3.16 (95 % CI 1.40–7.17),and P = 0.005. Ontario and Newfoundland data was thenpooled for further analysis.Due to our selection strategy and the association

between MSI and ITF2 methylation, further clinicopath-ological associations were analyzed separately for MSI-Hand MSS cases, shown in Table 2. There was a signifi-cant association between ITF2 promoter methylationwith MLH1 promoter methylation in MSI-H cases, P =6.72×10-4, OR 1.88 (1.10–3.68). There was also a trendtowards an association between ITF2 methylation andfemale gender in MSI-H cases, and ITF2 methylationand CIMP in MSS cases, but this was not significantusing a conservative p-value adjusted for multiple com-parisons. No other significant associations were found

for either MSS or MSI-H cases between ITF2 methylationand early age of onset (<50 years), stage, grade, tumor loca-tion, histological type, MMR protein status, MMR germlinemutation, BRAF V600E mutation, or survival status. Stage-specific survival was also performed, to account for poten-tial differences in early stage survival compared to stage IVin MSI-H cases (data not shown). There was a trend to-ward higher overall survival in stage I MSI-H cases withITF2 methylation, but results were not significant aftercorrection for multiple comparisons.

APC methylation and clinicopathological features,including MSI subtype, in two distinct CRC cohortsWe examined whether methylation of APC in tumorDNA was associated with patient clinicopathological fea-tures in both cohorts. Methylation status was comparedby Pearson’s chi-square test between MSI-H and MSScases. In Ontario 32.1 % (53/165) of MSS cases weremethylated while 37.6 % (62/165) of MSI-H cases weremethylated with an OR (95 % CI) of 1.27 (0.81–2.00),and P = 0.298. Similarly in Newfoundland 35.3 % (18/51)of MSS cases were methylated and 45.1 % (23/51) ofMSI-H cases methylated with OR (95 % CI) of 1.51(0.68–3.34), and P = 0.313.We examined whether APC methylation was associated

with patient clinicopathological features in pooled CRCcases from Ontario and Newfoundland. Results are shownin Table 3. Methylation of APC was not found to be asso-ciated with early age of onset (<50 years), sex, stage, grade,tumor location, histological type, CIMP status, MMR pro-tein status, MMR germline mutation, MLH1 methylation,BRAF V600E mutation, or survival status.

DiscussionUnderstanding the genetic and epigenetic differencesamongst colorectal cancer subtypes is essential, as CRCsubtypes differ in their treatment options and offer dis-tinct survival outcomes. The Wnt signaling pathway isdysregulated in a majority of colorectal tumors and canbe altered at the extracellular, intracellular and gene tar-get level [15, 34, 35]. We have shown that these changesto the Wnt pathway also differ based upon microsatelliteinstability status. We quantified the methylation statusof the ITF2 and APC promoter CpG islands in a nestedcase-case study in two cohorts of colorectal carcinomafrom two different populations, comparing cases by MSIstatus. We have demonstrated that the ITF2 promoter ishypermethylated in tumor tissues compared with matchednormal mucosa, and further, MSI-H tumors are more likelyto incur promoter methylation compared with MSS tu-mors. ITF2 promoter methylation was also significantlyassociated with MLH1 promoter methylation, a commonoccurrence in MSI-H tumors. Conversely, we found thatAPC, an important intracellular regulator of Wnt signaling


Table 2 Associations between ITF2 methylation and clinicopathological features in tumor DNA. Analysis of 216 MSI-H and 216 MSSCRC patients from Ontario and Newfoundland

MSS (n = 216) MSI-H (n = 216)

Unmethylated(%)

Methylated(%)

OR (95 % CI)a P-value Unmethylated(%)

Methylated(%)

OR (95 % CI)a P-value

Age

<50 24 (15.2) 6 (10.3) 1.552 (0.600–4.014) 0.361 24 (20.5) 14 (14.1) 1.567 (0.761–3.225) 0.220

50+ 134 (84.8) 52 (89.7) 93 (79.5) 85 (85.9)

Sex

Male 73 (46.2) 26 (44.8) 1.057 (0.577–1.935) 0.857 64 (54.7) 36 (36.4) 2.113 (1.222–3.655) 0.007

Female 85 (53.8) 32 (55.2) 53 (45.3) 63 (63.6)

TNM Stageb

1 36 (22.8) 14 (24.1) 0.720 (0.606–0.856) 0.193 31 (26.5) 18 (18.2) 5.167 (0.499–53.450) 0.321

2 78 (49.4) 32 (55.2) 57 (48.7) 55 (55.6)

3 32 (20.3) 12 (20.7) 28 (23.9) 23 (23.2)

4 12 (7.6) 0 (0.0) 1 (0.9) 3 (3.0)

Histological Gradeb

Low 16 (10.8) 4 (7.5) 2.000 (0.482–8.295) 0.624 9 (13.8) 8 (13.1) 0.728 (0.215–2.459) 0.514

Moderate 118 (79.7) 42 (79.2) 39 (60.0) 42 (68.9)

High 14 (9.5) 7 (13.2) 17 (26.2) 11 (18.0)

Locationc

Proximal 58 (37.9) 29 (50.9) 0.589 (0.319–1.089) 0.090 60 (89.6) 54 (87.1) 1.270 (0.432–3.735) 0.664

Distal 95 (62.1) 28 (49.1) 7 (10.4) 8 (12.9)

Histological Typed

Non-Mucinous 140 (90.3) 49 (84.5) 1.714 (0.705–4.167) 0.230 76 (66.7) 73 (75.3) 0.658 (0.360–1.202) 0.172

Mucinous 15 (9.7) 9 (15.5) 38 (33.3) 24 (24.7)

MMR Protein Status

Intact 143 (97.9) 55 (98.2) 0.867 (0.088–8.511) 0.902 21 (18.6) 9 (9.1) 2.283 (0.992–5.251) 0.048

Deficient 3 (2.1) 1 (1.8) 92 (81.4) 90 (90.9)


No 157 (99.4) 58 (100.0) 0.730 (0.673–0.792) 0.544 89 (76.1) 74 (74.7) 1.074 (0.577–1.999) 0.822

Yes 1 (0.6) 0 (0.0) 28 (23.9) 25 (25.3)

MLH1 Methylation

Unmethylated 143 (96.0) 52 (100.0) 0.733 (0.674–0.798) 0.142 68 (58.1) 42 (42.4) 1.883 (1.095–3.238) 6.72x10-4

Methylated 6 (4.0) 0 (0.0) 49 (41.9) 57 (57.6)

BRAF V600E Mutation

No 141 (92.2) 51 (91.1) 1.152 (0.387–3.431) 0.799 70 (63.1) 50 (52.6) 1.537 (0.088–2.683) 0.130

Yes 12 (7.8) 5 (8.9) 41 (36.9) 45 (47.4)

CIMP Status

Negative 100 (96.2) 33 (84.6) 4.545 (1.208–17.100) 0.016 50 (61.7) 29 (47.5) 1.780 (0.908–3.489) 0.092

Positive 4 (3.8) 6 (15.4) 31 (38.3) 32 (52.5)

Survival Status

Alive 110 (69.6) 35 (60.3) 1.475 (0.790–2.755) 0.221 78 (66.7) 70 (70.7) 0.829 (0.464–1.478) 0.524

Deceased 48 (30.4) 23 (39.7) 39 (33.3) 29 (29.3)aOdds ratio and 95 % confidence interval for methylated versus unmethylatedbOR and 95 % CI given for lowest stage/grade versus highest stage/gradecProximal tumor location includes lesions up to and including the splenic flexuredMucinous histology includes the presence of any mucin within the tumor stroma


Table 3 Associations between APC methylation and clinicopathological features in tumor DNA. Analysis of 216 MSI-H and 216 MSSCRC patients from Ontario and Newfoundland

MSS (n = 216) MSI-H (n = 216)

Unmethylated(%)

Methylated(%)

OR (95 % CI)a P-value Unmethylated(%)

Methylated(%)

OR (95 % CI)a P-value

Age

<50 20 (13.8) 10 (14.1) 0.976 (0.431–2.213) 0.954 23 (17.6) 15 (17.9) 0.994 (0.485–2.035) 0.986

50+ 125 (86.2) 61 (85.9) 108 (82.4) 70 (82.4)

Sex

Male 64 (44.1) 35 (49.3) 0.813 (0.460–1.436) 0.475 59 (45.0) 41 (48.2) 0.879 (0.509–1.520) 0.645

Female 81 (55.9) 36 (50.7) 72 (55.0) 44 (51.8)

TNM Stageb

1 27 (18.6) 23 (32.4) 0.391 (0.095–1.619) 0.147 32 (24.4) 17 (20.0) 0.653 (0.533–0.801) 0.306

2 79 (54.5) 31 (43.7) 64 (48.9) 48 (56.5)

3 30 (20.7) 14 (19.7) 31 (23.7) 20 (23.5)

4 9 (6.2) 3 (4.2) 4 (3.1) 0 (0.0)

Histological Gradeb

Low 12 (8.9) 8 (12.1) 0.600 (0.163–2.207) 0.724 8 (11.0) 9 (17.0) 0.770 (0.230–2.578) 0.467

Moderate 108 (80.0) 52 (78.8) 50 (68.5) 31 (58.5)

High 15 (11.1) 6 (9.1) 15 (20.5) 13 (24.5)

Locationc

Proximal 62 (44.0) 25 (36.2) 1.381 (0.763–2.499) 0.285 63 (85.1) 51 (92.7) 0.449 (0.135–1.495) 0.183

Distal 79 (56.0) 44 (63.8) 11 (14.9) 4 (7.3)

Histological Typed

Non-Mucinous 129 (89.6) 60 (87.0) 1.290 (0.534–3.114) 0.570 87 (68.5) 62 (73.8) 0.772 (0.418–1.426) 0.408

Mucinous 15 (10.4) 9 (13.0) 40 (31.5) 22 (26.2)

MMR Protein Status

Intact 132 (97.8) 66 (98.5) 0.668 (0.068–6.533) 0.726 21 (16.3) 9 (10.8) 1.599 (0.694–3.685) 0.268

Deficient 3 (2.2) 1 (1.5) 108 (83.7) 74 (89.2)


No 144 (99.3) 71 (100.0) 0.670 (0.610–0.736) 0.483 104 (79.4) 59 (69.4) 1.697 (0.908–3.175) 0.096

Yes 1 (0.7) 0 (0.0) 27 (20.6) 26 (30.6)

MLH1 Methylation

Unmethylated 131 (96.3) 64 (98.5) 0.409 (0.047–3.577) 0.405 62 (47.3) 48 (56.5) 0.693 (0.400–1.199) 0.189

Methylated 5 (3.7) 1 (1.5) 69 (52.7) 37 (43.5)

BRAF V600E Mutation

No 126 (90.6) 66 (94.3) 0.587 (0.184–1.873) 0.364 67 (53.2) 53 (66.3) 0.579 (0.324–1.034) 0.064

Yes 13 (9.4) 4 (5.7) 59 (46.8) 27 (33.8)

CIMP Status

Negative 91 (92.9) 42 (93.3) 0.929 (0.229–3.770) 0.917 45 (52.3) 34 (60.7) 0.710 (0.359–1.406) 0.325

Positive 7 (7.1) 3 (6.7) 41 (47.7) 22 (39.3)

Survival Status

Alive 95 (65.5) 50 (70.4) 0.782 (0.424–1.444) 0.432 91 (69.5) 57 (67.1) 1.118 (0.622–2.007) 0.710

Deceased 50 (34.5) 21 (29.6) 40 (30.5) 28 (32.9)aOdds ratio and 95 % confidence interval for methylated versus unmethylatedbOR and 95 % CI given for lowest stage/grade versus highest stage/gradecProximal tumor location includes lesions up to and including the splenic flexuredMucinous histology includes the presence of any mucin within the tumor stroma


marked by both genetic mutations and hypermethylation inCRC, acquires DNA methylation equally across subtypes.This is the first study to investigate DNA methylation

of ITF2 in CRC cases. Here, we have established thatITF2 methylation is a tumor-associated event, being arare occurrence in normal tissue DNA. One sample outof 47 normal colonic tissue samples was methylated, butthis rare occurrence may possibly be due to the fieldeffect, or field cancerization, in which apparently normalcells acquire genetic and/or epigenetic alterations andmay eventually progress to cancer. With regards totumor methylation of ITF2, we showed that it is associ-ated with the MSI-H phenotype. ITF2 has been reportedto be a tumor suppressor that can induce cell cyclearrest and is sometimes lost due to loss of heterozygosityat 18q21 [31]. However, ITF2 expression has been foundto be upregulated in some cancers with aberrantly acti-vated Wnt signaling but decreased in others [30, 31].Further research is required to elucidate the role of ITF2in tumorigenesis. Treatment of gastric cancer cell lineswith the DNA methyltransferase inhibitor 5-aza-2’-deox-ycytidine (5-aza) restored mRNA expression in cell linesthat had hypermethylation demonstrating methylation-dependent regulation of this gene [33]. Thus, transcrip-tional silencing in CRC through methylation wouldlikely lead to a decrease in its cellular expression levelspotentially contributing to tumorigenesis.APC promoter methylation is rarely observed in nor-

mal colonic tissue compared with CRC tumor tissue inour study population, which replicates the findings of arecent meta-analysis [44]. However, contrary to ITF2methylation, we did not see an association with MSI-HCRC or any other clinical features. APC expression is atleast partially regulated by DNA methylation, as itsexpression increases in CRC cell lines after treatmentwith 5-aza [45]. Several other studies have investigatedthe correlation between APC methylation and MSI withvarying results. Studies have shown wide variation inoverall APC methylation, regardless of subtype, from aslow as 18 % to as high as 63.4 % [19, 20]. Our resultsshow a more moderate level of 34–40 % of cases methyl-ated. Findings in the literature for the correlationbetween APC and MSI are even less clear, with methyla-tion in MSI-H tumors ranging from 14.3–72.7 % [21, 26].However, these studies analyzed small numbers of patientsamples, with a maximum of 29 MSI tumors used [24].Our study, on the other hand, employed a total of 432samples, 216 of which were MSI-H. This sample size ismany times larger than any other of its kind, giving morestatistical power and certainty to our results.There are no differences between level of methylation

at different stages of CRC diagnosis for either APC orITF2, indicating these may be early epigenetic events intumorigenesis. Additionally, APC methylation has been

detected in colon adenoma, further evidence that it is anearly event [46]. Detection of APC may be furtherexploited as a potential biomarker by detection in otherbiospecimens, as its methylation has been detected inboth stool and plasma [47, 48]. Further investigation ofthe presence of ITF2 methylation in adenomas shouldbe undertaken, as well as whether its methylation can bedetected in stool or plasma. This research will indicatethe potential of utilizing ITF2 and APC, perhaps incombination with other methylation markers, as non-invasive stool- or plasma-based methylation markersfor CRC detection and/or subtype discrimination.Data from colon and rectal tumors from The Cancer

Genome Atlas (TCGA) shows that APC mutation ratesdiffer among the 224 tumors sequenced by exomesequencing. TCGA data described hypermutated tumors,which have a mutation rate of 12/106 and consist mostlyof MSI-H tumors. The prevalence of APC mutation inthese hypermutated tumors is 51 % [15]. Alternatively,non-hypermutated tumors, defined by a mutation rate<8.24/106 and consisting mostly of MSS tumors, in-curred APC mutations in 81 % of cases [15]. This dispar-ity in APC mutation rates may be explained by DNAmethylation to inactivate APC leading to constitutiveligand-independent Wnt signaling. In this same data setITF2 is genetically altered in only 3 % of tumors, thus,methylation is likely to play a larger role in ITF2 dysreg-ulation in cancer [49, 50].While MSI-H tumors are a largely well-defined sub-

type of CRC, MSS tumors comprise the majority of casesand exhibit a wide variety of molecular characteristics.Thus, there is an emerging research focus to furtherclassify molecular subtypes of CRC. Recently, four con-sensus molecular subtypes were defined. The first sub-type consists mostly of MSI cases [14]. The remainingthree subtypes are defined by ‘canonical’ Wnt and MYCactivation, metabolic dysregulation, or mesenchymalactivation. Our results indicated that some MSS casesincur methylation of the Wnt genes studied, so perhapsthese cases belong to the subtype characterized by Wntactivation. It would be interesting to see which sub-classification the MSS cases used in this study belong to,and how ITF2 or APC methylation profiles differ amongthe four subtypes.MSI-H tumors often overlap with CIMP-positive

status. Thus, the association we see between MSI-H andITF2 methylation may in fact be part of the widespreadhypermethylation of CpG islands that characterizesCIMP tumors. CIMP status information is unavailablefor some Ontario cases and all Newfoundland casesutilized in this study, thus we do not have a completepicture of CIMP for our cohort. From our available datawe did see a trend between CIMP-positive status andITF2 methylation among MSS cases. However, there


were only ten cases in this group. From our currentfindings as well as previous investigation into epigeneticregulation of Wnt signalling genes we have found thatdysregulation through aberrant methylation is implicatedin all subtypes of CRC, not solely in CIMP-positive cases.APC, which abrogates Wnt signaling intracellularly, ismethylated in a proportion of CRCs, regardless of subtypewhile ITF2, a downstream target of Wnt signaling, is meth-ylated more often in MSI-H tumors. Our lab has previouslyfound that DKK1 and SFRP1 promoter methylation, codingfor two extracellular Wnt antagonists, segregate stronglywith different CRC subtypes. DKK1 methylation is associ-ated with the MSI-H phenotype and other MSI-associatedfeatures, while SFRP1 methylation is associated with MSStumors [34]. We also found that Wnt5a methylation, whichcodes for an extracellular ligand of the non-canonical Wntpathway, is associated with MSI-H [35]. These results werefound in the same cohort of Ontario and Newfoundlandpatients used in this study. These observations underscorethe importance of both Wnt signaling and the role of DNAmethylation in CRC.One limitation of this study to bear in mind is that only

a subset of available MSS cases was chosen for analysis bymatching to MSI-H cases by age quartile, stage and sex.Individuals with MSI-H CRC are generally a younger age,more frequently female, have a lower tumor stage and aremore frequently CIMP-positive than those individualswith MSS tumors. Thus, the MSS cases analyzed in thisstudy do not wholly represent all MSS cases from ourOntario and Newfoundland populations. Additionally, wedid not select MSS cases from the entire Ontario cohort,but only a subset available at the time this study wasundertaken. The subset that we selected from did not dif-fer in age, sex, stage or CIMP rates from the entire cohort.The strengths of our study include large sample size,

the inclusion of two independent well-characterizedpopulation-based cohorts and the choice of technol-ogy. The use of MethyLight technology is superior tomethylation-specific PCR (MSP) and offers severaladvantages including a quantitative, high-throughputmethylation-specific real-time PCR-based technique,which is amenable to using small quantities of DNAextracted from archival tissue specimens. MSP is amore qualitative and subjective method that has beenused in many prior studies of APC methylation.

ConclusionsOur findings demonstrate the importance of DNA methyla-tion in the regulation of genes selected for analysis and itsdiffering effects based on tumor subtype. ITF2 is not yetwell studied in CRC and we have now shown that this geneincurs MSI-associated hypermethylation. For APC, bothmutation and methylation play a role in its dysregulation. Itis likely that methylation of APC plays a secondary role in

CRC to more commonly occurring mutations and may actto fine-tune Wnt signaling. With both mutation andmethylation contributing to regulation of this gene, it ispossible the sequence of events may dictate the way CRCevolves. Based on its high specificity for CRC, APC methy-lation may offer usefulness as a marker within a panel ofother genes for CRC detection and ITF2 may be useful fordetection of MSI-H tumors. Future studies to independ-ently validate these findings are warranted. Overall, thisstudy has investigated methylation of the Wnt genes APCand ITF2 in a large cohort of MSI-H and matched MSSCRC tumors to find that ITF2 methylation is significantlyassociated with MSI-H tumors while APC methylation is atumor-specific event in CRC, which does not differ signifi-cantly between MSI-H and MSS subtypes or other clinico-pathological variables.

Abbreviations5-aza: 5-aza-2’-deoxycytidine; APC: adenomatous polyposis coli; CI: confidenceinterval; CIMP: CpG island methylator phenotype; CRC: colorectal cancer;DKK1: dickkopf 1 homolog [Xenopus laevis]; ITF2: immunoglobulin transcriptionfactor 2/transcription factor 4; MLH1: MutL homolog 1; MSI: microsatelliteinstability; MSI-H: microsatellite instability high; MSI-L: microsatellite instability low;MSP: methylation-specific polymerase chain reaction; MSS: microsatellite stable;NFCCR: Newfoundland Familial Colorectal Cancer Registry; OFCCR: Ontario FamilialColorectal Cancer Registry; OR: odds ratio; PMR: percent methylated reference;SD: standard deviation; SFRP1: secreted frizzled-related protein 1; TCGA: TheCancer Genome Atlas; Wnt5a: wingless-type MMTV integration site family,member 5A.

Competing interestsThe authors declare that they have no competing interests.

Authors’ contributionsAJS carried out the methylation analysis, performed the statistical analysis,and drafted the manuscript. DD and ED participated in patient recruitmentand coordination. DDB and JPY performed mutation analysis. DW performedCIMP analysis. PSP, RCG, SG, and JRM participated in study design and patientrecruitment. JAK participated in the design of the study and statistical analysisand interpretation of data. BB conceived the study, contributed in analysis andinterpretation of results, and drafted the manuscript. All authors read andapproved the final manuscript.

AcknowledgementsThis work was undertaken at the Department of Laboratory Medicine andPathobiology at the University of Toronto, Toronto, Ontario, Canada and wasconducted at the Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital,Toronto, Ontario, Canada. We sincerely thank the investigators, staff, andparticipants of the Colon Cancer Family Registry for their dedicated contributionsleading to this work. We gratefully acknowledge The Jeremy Jass MemorialPathology Bank for the tissue samples and pathology data used in thisstudy. This work was supported by grant UM1 CA167551 from the NationalCancer Institute and through cooperative agreements with the followingCCFR centers: Ontario Registry for Studies of Familial Colorectal Cancer(U01/U24 CA074783) and Australasian Colorectal Cancer Family Registry(U01/U24 CA097735). This work was also supported by a Team Grant fromthe Canadian Institutes of Health Research (CTP-79845) awarded to BB, JAK,SG, RCG, and PSP by the NCI under Request For Applications (CA-95-011).The content of this manuscript does not necessarily reflect the views orpolicies of the National Cancer Institute or any of the collaborating centersin the Colon Cancer Family Registry (CCFR), nor does mention of tradenames, commercial products, or organizations imply endorsement by theU.S. Government or CFR. AJS was supported by the Interdisciplinary HealthResearch Team Program studentship funded by the Canadian Institutes ofHealth Research, the Lunenfeld-Tanenbaum Research Institute Studentshipat Mount Sinai Hospital, and the University of Toronto Fellowship award.


The funders had no role in study design, data collection and analysis, decisionto publish, or preparation of the manuscript.

Author details1Department of Laboratory Medicine and Pathobiology, University ofToronto, Toronto, ON, Canada. 2Lunenfeld-Tanenbaum Research Institute ofMount Sinai Hospital, Toronto, ON, Canada. 3Ontario Familial ColorectalCancer Registry, Toronto, ON, Canada. 4Faculty of Medicine, MemorialUniversity of Newfoundland, St John’s, Newfoundland, Canada.5Oncogenomics Group, Genetic Epidemiology Laboratory, Department ofPathology, The University of Melbourne, Parkville, VIC, Australia. 6Centre forEpidemiology and Biostatistics, Melbourne School of Population and GlobalHealth, The University of Melbourne, Parkville, VIC, Australia. 7Department ofHaematology and Oncology, The Queen Elizabeth Hospital, Woodville, SouthAustralia, Australia. 8USC Epigenome Center, University of Southern California,Los Angeles, CA, USA. 9Department of Surgery, University of Toronto,Toronto, ON, Canada. 10Dalla Lana School of Public Health, University ofToronto, Toronto, ON, Canada. 11Department of Pathology, University HealthNetwork, Toronto, ON, Canada.

Received: 10 August 2015 Accepted: 8 February 2016

References1. Siegel R, Naishadham D, Jemal A. Cancer Statistics, 2013. CA Cancer J Clin.

2011;60:69–90.2. Moiel D, Thompson J. Early detection of colon cancer – the Kaiser Permanente

Northwest 30-year history: how do we measure success? Is it the test, thenumber of tests, the stage, or the percentage of screen-detected patients? PermJ. 2011;15:30–8.

3. Shima K, Morikawa T, Yamauchi M, Kuchiba A, Imamura Y, Liao X, et al.TGFBR2 and BAX mononucleotide tract mutations, microsatellite instability,and prognosis in 1072 colorectal cancers. PLoS One. 2011;6:e25062.

4. Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR, Burt RW,et al. A National Cancer Institute Workshop on Microsatellite Instability forcancer detection and familial predisposition: development of internationalcriteria for the determination of microsatellite instability in colorectal cancer.Cancer Res. 1998;58:5248–57.

5. Moslein G, Tester DJ, Lindor NM, Honchel R, Cunningham JM, French AJ,et al. Microsatellite instability and mutation analysis of hMSH2 and hMLH1in patients with sporadic, familial and hereditary colorectal cancer. Hum MolGenet. 1996;5:1245–52.

6. Grady WM. Genomic instability and colon cancer. Cancer Metastasis Rev.2004;23:11–27.

7. Hong SP, Min BS, Kim TI, Cheon JH, Kim NK, Kim H, et al. The differential impactof microsatellite instability as a marker of prognosis and tumor responsebetween colon cancer and rectal cancer. Eur J Cancer. 2012;48:1235–43.

8. Michel S, Benner A, Tariverdian M, Wentzensen N, Hoefler P, PommerenckeT, et al. High density of FOXP3-positive T cells infiltrating colorectal cancerswith microsatellite instability. Br J Cancer. 2008;99:1867–73.

9. Goldstein J, Tran B, Ensor J, Gibbs P, Wong HL, Wong SF, et al. Multicenterretrospective analysis of metastatic colorectal cancer with high-level microsatelliteinstability (MSI-H). Ann Oncol. 2014;25:1032–8.

10. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, Issa JP. CpG islandmethylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A. 1999;96:8681–6.

11. Phipps AI, Limburg PJ, Baron JA, Burnett-Hartman AN, Weisenberger DJ, LairdPW, et al. Association between molecular subtypes of colorectal cancer andpatient survival. Gastroenterology. 2015;148:77–87.

12. Shen L, Catalano PJ, Benson AB, O’Dwyer P, Hamilton SR, Issa JP. Associationbetween DNA methylation and shortened survival in patients with advancedcolorectal cancer treated with 5-fluorouracil based chemotherapy. Clin CancerRes. 2007;13:6093–8.

13. Ogino S, Nosho K, Kirkner GJ, Kawasaki T, Meverhardt JA, Loda M, et al. CpGisland methylator phenotype, microsatellite instability, BRAF mutation andclinical outcome in colon cancer. Gut. 2009;58:90–6.

14. Guinney J, Dienstmann R, Wang X, de Reyniès A, Schlicker A, Soneson C, et al.The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21:1350–6.

15. Muzny DM, Bainbridge MN, Chang K, Dinh HH, Drummond JA, Fowler G,et al. Comprehensive molecular characterization of human colon and rectalcancer. Nature. 2012;487:330–7.

16. Gregorieff A, Clevers H. Wnt signaling in the intestinal epithelium: fromendoderm to cancer. Genes Dev. 2005;19:877–90.

17. Huelsken J, Birchmeier W. New aspects of Wnt signaling pathways in highervertebrates. Curr Opin Genet Dev. 2001;11:547–53.

18. Willert K, Jones KA. Wnt signaling: is the party in the nucleus? Genes Dev. 2006;20:1394–404.

19. Chen J, Rocken C, Lofton-Day C, Schulz HU, Muller O, Kutzner N, et al.Molecular analysis of APC promoter methylation and protein expression incolorectal cancer metastasis. Carcinogenesis. 2005;26:37–43.

20. Chen SP, Chiu SC, Wu CC, Lin SZ, Kang JC, Chen YL, et al. The association ofmethylation in the promoter of APC and MGMT and the prognosis ofTaiwanese CRC patients. Genet Test Mol Biomarkers. 2009;3:67–71.

21. Derks S, Postma C, Caralho B, van den Bosch SM, Moerkerk PT, HermanJG, et al. Integrated analysis of chromosomal, microsatellite andepigenetic instability in colorectal cancer identifies specific associationsbetween promoter methylation of pivotal tumor suppressor and DNArepair genes and specific chromosomal alterations. Carcinogenesis.2008;29:434–9.

22. Kumar K, Brim H, Giardiello F, Smoot DT, Nouraie M, Lee EL, et al. DistinctBRAF (V600E) and KRAS mutations in high microsatellite instability sporadiccolorectal cancer in African Americans. Clin Cancer Res. 2009;15:1155–61.

23. Naghibalhossaini F, Zamani M, Mokarram P, Khalili I, Rasti M, Mostafavi-PourZ. Epigenetic and genetic analysis of Wnt signaling pathway in sporadiccolorectal cancer patients from Iran. Mol Biol Rep. 2012;39:6171–8.

24. Gay LJ, Mitrou PN, Keen J, Bowman R, Naguib A, Cooke J, et al. Dietary, lifestyleand clinicopathological factors associated with APC mutations and promotermethylation in colorectal cancers from the EPIC-Norfolk study. J Pathol. 2012;228:405–15.

25. Goel A, Nagasaka T, Arnold CN, Inoue T, Hamilton C, Niedzwiecki D, et al. TheCpG island methylator phenotype and chromosomal instability are inverselycorrelated in sporadic colorectal cancer. Gastroenterology. 2007;132:127–38.

26. Kim JC, Choi JS, Roh SA, Cho DH, Kim TW, Kim YS. Promoter methylation ofspecific genes is associated with the phenotype and progression of colorectaladenocarcinomas. Ann Surg Oncol. 2010;17:1767–76.

27. Thorstensen L, Lind GE, Lovig T, Diep CB, Meling GI, Rognum TO, et al. Geneticand epigenetic changes of components affecting the WNT pathway in colorectalcarcinomas stratified by microsatellite instability. Neoplasia. 2005;7:99–108.

28. Iacopetta B, Grieu F, Li W, Ruszkiewicz A, Caruso M, Moore J, et al. APC genemethylation is inversely correlated with features of the CpG islandmethylator phenotype in colorectal cancer. Int J Cancer. 2006;119:2272–8.

29. Zhai Y, Wu R, Schwartz DR, Darrah D, Reed H, Kolligs FT, et al. Role ofbeta-catenin/T-cell factor-regulated genes in ovarian endometrioidadenocarcinomas. Am J Pathol. 2002;160:1229–38.

30. Kolligs FT, Nieman MT, Winer I, Hu G, Van Mater D, Feng Y, et al. ITF-2, adownstream target of the Wnt/TCF pathway, is activated in human cancerswith β-catenin defects and promotes neoplastic transformation. Cancer Cell.2002;1:145–55.

31. Herbst A, Helferich S, Behrens A, Goke B, Kolligs FT. The transcription factor ITF-2Ainduces cell cycle arrest via p21(Cip1). Biochem Biophys Res Commun. 2009;387:736–40.

32. Joo JK, Kim SH, Kim HG, Kim DY, Ryu SY, Lee KH, et al. Methylation oftranscription factor 4 in gastric carcinoma. Ann Surg Oncol. 2010;17:3344–53.

33. Kim JH, Kim M, Noh SM, Song KS, Kang GH, Kim HJ, et al. CpG methylation inexon 1 of transcription factor 4 increases with age in normal gastric mucosa andis associated with gene silencing in intestinal-type gastric cancers. Carcinogenesis.2008;29:1623–31.

34. Rawson JB, Manno M, Mrkonjic M, Daftary D, Dicks E, Buchanan DD, et al.Promoter methylation of Wnt antagonists DKK1 and SFRP1 is associatedwith opposing tumor subtypes in two large populations of colorectalcancer patients. Carcinogenesis. 2011;32:741–7.

35. Rawson JB, Mrkonjic M, Daftary D, Dicks E, Buchanan DD, YounghusbandHB, et al. Promoter methylation of Wnt5a is associated with microsatelliteinstability and BRAF V600E mutation in two large populations of colorectalcancer patients. Br J Cancer. 2011;104:1906–12.

36. Cotterchio M, McKeown-Eyssen G, Sutherland H, Buchan G, Aronson M, EassonAM, et al. Ontario familial colon cancer registry: methods and first-year responserates. Chron Dis Can. 2000;21:81–6.

37. Green RC, Green JS, Buehler SK, Robb JD, Daftary D, Gallinger S, et al. Very highincidence of familial colorectal cancer in Newfoundland: a comparison withOntario and 13 other population-based studies. Fam Cancer. 2007;6:53–62.


38. Woods MO, Hyde AJ, Curtis FK, Stuckless S, Green JS, Pollett AF, et al. Highfrequency of hereditary colorectal cancer in Newfoundland likely involvesnovel susceptibility genes. Clin Cancer Res. 2005;11:6853–61.

39. Lindor NM, Burgart LJ, Leontovich O, Goldberg RM, Cunningham JM,Sargent DJ, et al. Immunohistochemistry versus microsatellite instabilitytesting in phenotyping colorectal tumors. J Clin Oncol. 2002;20:1043–8.

40. Weisenberger DJ, Campan M, Long TI, Kim M, Woods C, Fiala E, et al. Analysisof repetitive element DNA methylation by MethyLight. Nucleic Acids Res. 2005;33:6823–36.

41. Campan M, Weisenberger DJ, Trinh B, Laird PW. MethyLight. Methods MolBiol. 2009;507:325–37.

42. Mrkonjic M, Roslin NM, Greenwood CM, Raptis S, Pollett A, Laird PW, et al.Specific variants in the MLH1 gene region may drive DNA methylation, loss ofprotein expression, and MSI-H colorectal cancer. PLoS One. 2010;5:e13314.

43. Weisenberger DJ, Levine AJ, Long TI, Buchanan DD, Walters R, ClendenningM, et al. Association of the colorectal CpG island methylator phenotypewith molecular features, risk factors, and family history. Cancer EpidemiolBiomarkers Prev. 2015;24:512–9.

44. Ding Z, Jiang T, Piao Y, Han T, Han Y, Xie X. Meta-analysis of the associationbetween APC promoter methylation and colorectal cancer. Onco TargetsTher. 2015;8:211–22.

45. Deng G, Song GA, Pong E, Sleisenger M, Kim YS. Promoter methylation inhibitsAPC gene expression by causing changes in chromatin conformation andinterfering with the binding of transcription factor CCAAT-binding factor.Cancer Res. 2004;64:2692–8.

46. Yang Q, Dong Y, Wu W, Zhu C, Chong H, Lu J, et al. Detection and differentialdiagnosis of colon cancer by a cumulative analysis of promoter methylation.Nat Commun. 2012;3:1206.

47. Azuara D, Rodriguez-Moranta F, de Oca J, Soriano-Izguierdo A, Mora J, GuardiolaJ, et al. Novel methylation panel for the early detection of colorectal tumors instool DNA. Clin Colorectal Cancer. 2010;9:168–76.

48. Lee BB, Lee EJ, Jung EH, Chun HK, Chang DK, Song SY, et al. Aberrantmethylation of APC, MGMT, RASSF2A, and WIF-1 genes in plasma as a biomarkerfor early detection of colorectal cancer. Clin Cancer Res. 2009;15:6185–91.

49. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al.Integrative analysis of complex cancer genomics and clinical profiles usingcBioPortal. Sci Signal. 2013;6:pl1.

50. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBiocancer genomics portal: an open platform for exploring multidimensionalcancer genomics data. Cancer Discov. 2012;2:401–4.

• We accept pre-submission inquiries • Our selector tool helps you to find the most relevant journal• We provide round the clock customer support • Convenient online submission• Thorough peer review• Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript atwww.biomedcentral.com/submit

Submit your next manuscript to BioMed Central and we will help you at every step:


The Dynamic Interplay of Epigenetics and Genetics in ......Jamie and Liyang, thank you for welcoming...

Documents

Transcript of The Dynamic Interplay of Epigenetics and Genetics in ......Jamie and Liyang, thank you for welcoming...