dna methylation in colorectal cancer...dna methylation in colorectal cancer sophia pei woon ang bsc...

DNA methylation in colorectal cancer

Sophia Pei Woon Ang

BSc (Hons)

2009

This thesis is presented for the degree of Doctor of Philosophy

in the School of Surgery, University of Western Australia

Abstract

Abstract

De novo methylation of CpG islands in gene promoter regions is an alternative gene

silencing mechanism that contributes to the development and progression of colorectal

cancer (CRC). CRCs that exhibit widespread hypermethylation of gene promoters are

referred to as having the CpG island methylator phenotype (CIMP). To improve our

understanding of the CIMP+ subgroup of CRC, the first aim of this research was to

address one of the technical limitations of methylation analysis by developing a quality

control system for sample DNA (Chapter 3). A ridge-regression approach was then

devised in an attempt to identify CpG loci whose methylation status was predictive of

gene expression (Chapter 4). Comprehensive methylation profiling was used to further

characterise CRC based on unbiased analysis of DNA methylation levels (Chapter 5).

Finally, CIMP+ was evaluated in a large series of young CRC patients with the longer

term aim of identifying possible familial cases of this CRC subtype (Chapter 6).

A growing number of studies have shown that aberrant tumour DNA methylation may

represent a promising molecular diagnostic, prognostic and predictive marker. This has

generated the need for accurate quality control systems for the analysis of DNA

methylation in clinical specimens. An adequate amount of DNA after bisulfite treatment

and the subsequent availability of sufficient target molecules for PCR based methylation

analyses are critical for ensuring accurate quantification. In Chapter 3, experiments that

tested multiple extractions, processing, analytical methods and sample types were

conducted. These demonstrated that cycle threshold (Ct) value was superior to

spectrophotometric analysis for predicting the suitability of DNA for methylation

analyses. The potential benefits of this approach are improvements in the reproducibility

of methylation analysis and in sample management.

Abstract

Since DNA hypermethylation of promoter regions is involved in the transcriptional

regulation of many genes, it is widely used as a surrogate marker for gene expression.

However, most existing DNA methylation assays have not been correlated with

corresponding gene expression. In Chapter 4, a ridge-regression model was developed

with the aim of identifying CpG loci that were predictive of RNA levels for the RUNX3

and DPYD genes in CRC. The model was developed using methylation and RNA

expression data obtained in a panel of CRC cell lines. When tested in primary CRC, the

methylation status of the candidate CpG sites failed to predict RNA expression. These

findings suggest that quantitative and qualitative differences between training and

validation samples should be considered when constructing models for the prediction of

methylation-dependent RNA expression.

The use of an unbiased and comprehensive methylation profiling approach described in

Chapter 5 identified three CIMP subgroups in CRC associated with distinct

clinicopathological and molecular characteristics. BRAF and KRAS mutations were

positively associated with the CIMP-high subgroup, suggesting these genetic and

epigenetic events occur within the same CRC pathway. In addition to aberrant

hypermethylation, CIMP-high tumours also showed demethylation of a small proportion

of CpG sites in comparison to normal tissue. Although these findings require validation in

additional CRC cohorts, it is likely that genome-wide epigenetic and parallel gene

expression studies will help to identify the molecular alterations that underlie the major

CIMP subgroups.

While the majority of CIMP+ tumours arise in elderly patients, the study described in

Chapter 6 found that that 8% of CRC from young patients were also CIMP+. These

Abstract

tumours shared hallmark clinicopathological and molecular characteristics with CIMP+

tumours from elderly patients, including a high frequency of BRAF mutation. The

presentation of CIMP+ CRC in patients aged less than 60 years suggests a hereditary

component for some of these tumours, possibly involving hyperplastic polyposis or

serrated pathway syndromes. This issue requires further prospective studies involving

collection of more detailed information on family history of cancer and pathological

information on the serrated adenoma/hyperplastic morphologies.

Table of contents

Chapter 1. Introduction ............................................................................................ 1

1.1 COLORECTAL CANCER...................................................................................... 1

1.2 MOLECULAR PATHWAYS IN COLORECTAL CANCER ............................... 2

1.2.1 The chromosomal instability pathway................................................................. 3

1.2.2 Microsatellite instability pathway........................................................................ 3

1.2.3 Epigenetic instability in colorectal cancer ........................................................... 4

1.3 THE CpG ISLAND METHYLATOR PHENOTYPE (CIMP) ............................... 5

1.3.1 CIMP as a distinct CRC phenotype ..................................................................... 5

1.3.2 Clinical features of CIMP+ CRC......................................................................... 6

1.3.3 BRAF and KRAS mutations in CIMP+ CRC ....................................................... 7

1.3.4 Non-genetic risk factors for the development of CIMP+ CRC ........................... 8

1.3.5 Genetic risk factors for the development of CIMP+ CRC ................................ 11

1.3.6 The serrated adenoma pathway in the development of CIMP+ CRC ............... 12

1.3.7 Serrated pathway syndrome............................................................................... 13

1.3.8 Molecular mechanisms in the serrated pathway................................................ 14

1.4 HERITABILITY OF DNA METHYLATION...................................................... 15

1.5 DNA HYPOMETHYLATION IN COLORECTAL CANCER............................ 16

1.6 DNA METHYLTRANSFERASES....................................................................... 17

1.7 OTHER COMPONENTS OF THE EPIGENETIC MACHINERY...................... 18

1.8 DNA HYPERMETHYLATION IN CANCER AS A CLINICAL BIOMARKER

AND THERAPEUTIC TARGET.......................................................................... 20

1.9 AIMS OF THIS RESEARCH ............................................................................... 22

Chapter 2. Methods & Materials......................................................................... 25

2.1 CLINICAL SAMPLES AND CRC CELL LINES................................................ 25

2.1.1 Recruitment of Caucasian CRC patients ........................................................... 25

2.1.2 Recruitment of Asian CRC patients .................................................................. 25

2.1.3 Formalin-fixed and paraffin embedded CRC tissues......................................... 25

2.1.4 CRC cell lines .................................................................................................... 26

2.2 DNA EXTRACTION ............................................................................................ 27

2.2.1 DNA extraction from frozen tissues .................................................................. 27

2.2.2 DNA extraction from paraffin-embedded tissues.............................................. 27

2.2.3 DNA extraction from cell lines.......................................................................... 28

2.3 RNA EXTRACTION ............................................................................................ 28

2.3.1 RNA extraction from frozen tissues .................................................................. 28

2.3.2 RNA extraction from cell lines.......................................................................... 29

2.4 QUALITY CONTROL OF NUCLEIC ACIDS .................................................... 29

2.4.1 Quantitation and qualitative assessment of genomic DNA ............................... 29

2.4.2 Quantitative and qualitative assessment of total RNA ...................................... 30

2.5 QUANTITATION OF RNA EXPRESSION LEVELS ........................................ 30

2.6 SCREENING FOR KRAS MUTATION, BRAF MUTATION AND MSI ........... 31

2.6.1 Fluorescent-single strand conformational polymorphism (F-SSCP)................. 31

2.6.2 PCR amplification of KRAS, BRAF and BAT26 ............................................... 32

2.6.3 F-SSCP screening for KRAS mutations, BRAF mutations and MSI.................. 32

2.7 DNA METHYLATION ANALYSIS.................................................................... 33

2.7.1 Sodium bisulfite conversion of genomic DNA ................................................. 33

2.7.1.1 Sodium bisulfite conversion of genomic DNA (in-house protocol).................. 33

2.7.1.2 Sodium bisulfite conversion of genomic DNA (EZ-DNA Methylation kitTM) . 34

2.7.2 CpG Methyltransferase (M.SssI) treatment of DNA......................................... 34

2.7.3 Methylation-specific PCR (MSP)...................................................................... 35

2.7.4 MethyLight ........................................................................................................ 36

2.7.5 Pyrosequencing.................................................................................................. 38

2.7.6 Cloning and sequencing of bisulfite-treated DNA ............................................ 40

2.7.6.1 PCR for bisulfite-cloning and sequencing (bsSEQ) .......................................... 40

2.7.6.2 PCR product purification and ligation............................................................... 41

2.7.6.3 Transformation and selection of clones............................................................. 41

2.7.6.4 Sequencing of plasmid DNA............................................................................. 42

2.7.7 Methylation analysis using Goldengate array.................................................... 43

2.8 STATISTICAL ANALYSIS ................................................................................. 44

Chapter 3. An improved quality control for bisulfite-PCR based DNA

methylation analysis: cycle threshold value ...................................................... 45

3.1 BACKGROUND ......................................................................................................... 45

3.2 METHODS & MATERIALS...................................................................................... 46

3.3 RESULTS.................................................................................................................... 47

3.4 DISCUSSION.............................................................................................................. 51

Chapter 4. Identification of CpG sites in the RUNX3 and DPYD genes

associated with expression level............................................................................. 55

4.1 BACKGROUND ................................................................................................... 55

4.2 METHODS AND MATERIALS .......................................................................... 57

4.2.1 Sample processing and analyses............................................................................... 57

4.2.2 Identifying individual CpG sites correlated with RNA expression.......................... 57

4.2.3 Identifying multiple adjacent CpG sites correlated with RNA expression............... 57

4.2.4 Correlation between methylation at candidate expression-linked CpG sites and

RNA levels in primary tumours......................................................................... 58

4.3 RESULTS.............................................................................................................. 59

4.3.1 Methylation of CpG sites in the RUNX3 and DPYD promoters correlates with

RNA expression in CRC cell lines .................................................................... 59

4.3.2 Methylation at multiple adjacent CpG sites is poorly correlated to RNA levels

in CRC cell lines................................................................................................ 60

4.3.3 Expression-related CpG sites identified in CRC cell lines failed to predict RNA

expression in primary tumours .......................................................................... 60

4.4 DISCUSSION........................................................................................................ 65

Chapter 5. Comprehensive profiling of DNA methylation in colorectal

cancer reveals three subgroups with distinct clinicopathological and

molecular features ...................................................................................................... 68

5.1 BACKGROUND ................................................................................................... 68


5.2.1 Tissue samples ................................................................................................... 70

5.2.2 BRAF mutation, KRAS mutation and microsatellite instability......................... 70

5.2.3 MethyLight determination of CIMPW status ..................................................... 71

5.2.4 DNA methylation profiling using Illumina GoldenGate® methylation bead

array ................................................................................................................... 71

5.2.5 Statistical analysis.............................................................................................. 72

5.3 RESULTS.............................................................................................................. 73

5.3.1 DNA methylation patterns in normal and tumour tissue ................................... 73

5.3.2 CRC subgroups show distinctive clinicopathological and molecular features.. 74

5.3.3 Differentially methylated genes in CRC subgroups .......................................... 75

5.4 DISCUSSION........................................................................................................ 78

Chapter 6. BRAF mutation is associated with the CpG island

methylator phenotype in colorectal cancer from young patients .............. 82

6.1 BACKGROUND ................................................................................................... 82


6.3 RESULTS.............................................................................................................. 85

6.4 DISCUSSION........................................................................................................ 88

Chapter 7. GENERAL DISCUSSION............................................................... 93

7.1 Contribution of this work to the understanding of DNA methylation in CRC...... 93

7.2 DNA quality and methylation analysis.................................................................. 94

7.3 Identification of CpG loci whose methylation status correlates with gene

expression .............................................................................................................. 95

7.4 Comprehensive DNA methylation profiling to define CIMP ............................... 97

7.5 CIMP+ in early onset CRC.................................................................................. 100

7.6 Major Findings and Conclusions ............................................................................... 102

References.................................................................................................................... 104

Appendices .................................................................................................................. 117

List of abbreviations

ACF Aberrant crypt foci

ACTB beta-actin

ASO Allele-specific oligonucleotide

ATP Adenosyl-triphosphate

bsDNA Bisulfite DNA

bsSEQ Bisulfite-cloning and sequencing

cDNA Complementary DNA

CGIs CpG islands

CIMP CpG Island methylator phenotype

CIMP-H CIMP-high

CIMP-M CIMP-mid

CIMP-L CIMP-low

CIN Chromosomal instability

CRC Colorectal cancer

Ct Cycle threshold

EMVI Extramural invasion

FAP Familial adenomatous polyposis

FDR False discovery rate

FFPE Formalin-fixed and paraffin embedded

F-SSCP Fluorescent-single strand conformational polymorphism

GAPDH Glyceraldehyde-3-phosphate dehydrogenase

gDNA Genomic DNA

GSH Glutathione

HNPCC Hereditary non-polyposis colorectal cancer

HPP Hyperplastic polyposis

KNN K-nearest neighbour

LOH Loss of heterozygosity

LOI Loss of imprinting

LSO Locus-specific oligonucleotide

LVI Lymphovascular invasion

MAP MUTYH associated polyposis

MBD Methyl-domain binding proteins

MINTs Methylated in tumour markers

miRNA MicroRNA

MMR Mismatch repair

MSI Microsatellite instability

MSP Methylation-specific PCR

M.SssI CpG methyltransferase

PCA Principal component analysis

PMR Percentage of methylated reference

PNI Perineural invasion

PPi Inorganic pyrophosphate

PRC2 Polycomb repressive complex 2

PS Predictive score

QPCR Real-time PCR

ROC Receiver operating characteristics

RR Ridge regression

RT-PCR Reverse-transcriptase real-time PCR

SA serrated adenomas

SAM S-adenosylmethionine

SAH S-adenosylhomocysteine

SNP Single nucleotide polymorphisms

SPS Serrated pathway syndromes

THF Tetrahydrofolate

TILS Tumour-infiltrating lymphocytic response

5-FU 5-fluorouracil

� Ridge parameter

Acknowledgements

Sincere thanks are extended to all the many people who make completion of this PhD

thesis possible.

I am most grateful to my principal supervisor Prof. Barry Iacopetta for his invaluable

insights and constructive feedback throughout the candidature. His extensive knowledge

of colorectal cancer and his enthusiasm as an educator have introduced me to the breath

and variety of the field.

I am also grateful to my co-supervisor Assoc. Prof Richie Soong for his technical advice

and constant guidance. It was a great privilege to be given the opportunity to carry out

experimental works in his laboratory.

I would like to thank Dr. Agus Salim, Dr. Arief Gustnanto and Marie Loh for providing

their expertise in statistical analysis.

My sincere gratitudes are extended to Fabienne Grieu for her technical assistance,

continuous support and many good advices.

I thank Dr. Natalia Liem for her technical support and constant encouragement.

Many thanks also go out to the past and present members of Translational Interface,

Cancer Science Institute for their assistance and friendship.

I greatly appreciate the administrative personnels of the School of Surgery for their

assistance in many ways.

A special thanks to my family for their endless support and understanding.

Statement of candidate contribution

This thesis contains published work and/or work prepared for publication, some of which

has been co-authored. The bibliographical details of the work and where it appears in the

thesis are outlined below.

Chapter 3. Ang PW, Toh HB, Iacopetta B, Soong R. An improved quality control for

bisulfite-PCR-based DNA methylation analysis: cycle threshold value. Clin Chem Lab

Med. 2008; 46(8):1117-21.

PWA performed all experimental works using clinical samples, drafted the manuscript,

analysed and interpreted the data.

Chapter 5. Ang PW, Loh M, Liem N, Grieu F, Vaithilingham A, Platell C, Yong WP,

Iacopetta B, Soong R. Comprehensive profiling of DNA methylation in colorectal cancer

reveals three subgroups with distinct clinicopathological and molecular features.

Submitted for publication

PWA extracted DNA, performed molecular analyses, drafted the manuscript, and

interpreted the data.

Chapter 6. Ang PW, Li WQ, Soong R, Iacopetta B. BRAF mutation is associated with

the CpG island methylator phenotype in colorectal cancer from young patients. Cancer

Lett. 2009; 273(2):221-4.

PWA extracted DNA, performed DNA methylation analyses, drafted the manuscript,

analysed and interpreted the data.

__________________(candidate) _________________ (Principal supervisor) Sophia Pei Woon Ang Prof. Barry Iacopetta

Chapter 1 Introduction 1

Chapter 1. Introduction

1.1 COLORECTAL CANCER

Colorectal cancer (CRC) affects more than a million individuals annually and is the fourth

leading cause of cancer mortality worldwide (Jemal et al., 2009; Stewart and Kleihues,

2003). The incidence of CRC is higher in developed nations than in the developing world

(Jemal et al., 2009; Stewart and Kleihues, 2003). Disparities in the incidence and

mortality of CRC in different racial groups and regions are likely to be related to genetic

variability, environmental exposure, access to high-quality regular screening, timely

diagnosis and treatment (Weitz et al., 2005). CRC is largely a disease of old age with

80% of cases diagnosed in those aged 60 years or more. Although the majority of CRC

are sporadic, 5-10% of cases arising in young patients are associated with familial

disorders with an underlying genetic predisposition (Weitz et al., 2005). The three major

forms of hereditary CRC syndromes are familial adenomatous polyposis (FAP),

hereditary non-polyposis colorectal cancer (HNPCC, or Lynch syndrome) and MUTYH

associated polyposis (MAP). Identification of the genes responsible for these familial

syndromes has allowed the implementation of genetic testing in routine clinical practice

to determine predisposition to CRC. Nonetheless, FAP, HNPCC and MAP together

account for less than 5% of all CRC cases (Aaltonen et al., 2007). It is probable that a

significant proportion of sporadic CRC patients have inherited a combination of relatively

common, low-penetrance gene variants that contribute to CRC risk (Tenesa et al., 2008;

Tomlinson et al., 2008), as exemplified by variants on locus 8q24 (Haiman et al., 2007;

Tomlinson et al., 2007; Zanke et al., 2007). CRC is therefore likely to result from a

combination of environmental factors, germline factors, somatic mutations and epigenetic

changes. Unravelling the pathogenic mechanisms involved in CRC should allow the


development of a more robust, molecular-based classification of these tumours, as well as

suggesting more effective prevention strategies and treatments.

1.2 MOLECULAR PATHWAYS IN COLORECTAL CANCER

The evolution of normal colonic epithelial cells to adenocarcinoma is driven by the

progressive accumulation of genetic mutations and epigenetic alterations. Two

predominant forms of genomic instability have been elucidated in CRC, representing the

chromosomal instability (CIN) and microsatellite instability (MSI) pathways (Figure 1.1)

(Søreide et al., 2009). In addition to genomic instability, the CpG Island methylator

phenotype (CIMP) pathway has also been identified as a major etiological factor in the

pathogenesis of CRC due to the widespread inactivation of tumour suppressor genes

(Figure 1.1) (Issa, 2004; Søreide et al., 2009). Tumours arising from CIN and CIMP

pathways have varied molecular and clinicopathological features, as discussed below.

Figure 1.1 Molecular pathways in CRC. CIMP, CpG island methylator phenotype

(adapted from Søreide et al., 2009)

Intermediateadenoma

Normalmucosa

Earlyadenoma

Lateadenoma

Carcinoma

Hyperplasticpolyps

Serratedadenoma

Serratedadenoma

with dysplasia

Chromosomal instability pathwayGenetic alterations through chromosomal losses and gains

Deletion 5q

APC

Deletion 8p

KRAS

LOH 18q

DCC, SMAD4

LOH 17p

p53

�-catenin BAX, TCF4 IGFIIR, TGF�RII

Microsatellite instability pathwayGenetic alterations through defective mismatch repair proteins

Hypermethylator phenotype pathwayEpigenetic alterations through aberrant promoter hypermethylation

Inactivation ofMLH1 by mutation or hypermethylation

BRAF, KRASHypomethylation, hypermethylation CIMP hypermethylation

Intermediateadenoma

Normalmucosa

Earlyadenoma

Lateadenoma

Carcinoma

Hyperplasticpolyps

Serratedadenoma

Serratedadenoma

with dysplasia

Chromosomal instability pathwayGenetic alterations through chromosomal losses and gains

Deletion 5q

APC

Deletion 8p

KRAS

LOH 18q

DCC, SMAD4

LOH 17p

p53

�-catenin BAX, TCF4 IGFIIR, TGF�RII

Microsatellite instability pathwayGenetic alterations through defective mismatch repair proteins

Hypermethylator phenotype pathwayEpigenetic alterations through aberrant promoter hypermethylation

Inactivation ofMLH1 by mutation or hypermethylation

BRAF, KRASHypomethylation, hypermethylation CIMP hypermethylation


1.2.1 The Chromosomal instability pathway

The progression of aberrant crypt foci (ACF) to frank carcinoma with concurrent

sequential acquisition of molecular changes in FAP forms the paradigm for the CIN

pathway proposed by Fearon and Volgelstein (Fearon and Vogelstein, 1990). Mutations

to APC, TP53 and KRAS, and the loss of heterozygosity (LOH) at 18q and 17p confer

growth advantages to the altered cells (Fearon and Vogelstein, 1990). Following

successive waves of uncontrolled clonal expansion, acquisition of additional genetic

changes leads to the formation of histologically advanced neoplasms (Fearon and

Vogelstein, 1990). Aneuploidy, multiple chromosomal rearrangements and the

accumulation of somatic mutations are hallmarks of CIN tumours (Grady and Carethers,

2008; Søreide et al., 2009). Deregulation of the WNT pathway is responsible for the high

malignant potential of dysplastic ACF harbouring APC mutations (Grady and Carethers,

2008). Although about 85% of CRCs are characterised by the CIN phenotype, only a

small minority (~7%) carry all three KRAS, TP53 and APC mutations (Rodriguez et al.,

2006).

1.2.2 Microsatellite instability pathway

Approximately 15% of sporadic CRC and almost all tumours from individuals with

HNPCC arise through the MSI pathway (Imai and Yamamoto, 2008). Germline mutations

of mismatch repair (MMR) genes (MLH1, MSH2, MSH6 and PMS2) underlie the

development of MSI+ tumours in HNPCC, while methylation-induced silencing of MLH1

occurs in the majority of sporadic MSI+ CRC (Jass, 2007b; Søreide et al., 2009).

Although MSI+ tumours are usually near-diploid, they frequently accumulate small

deletions or insertions in short repetitive sequences contained with the coding region of

genes including TGF�RII (transforming growth factor � receptor II), IGF2R (insulin-like

growth factor 2 receptor) and BAX (BCL2-associated X protein). Repetitive sequences are


susceptible to slippage by strand misalignment during replication and thus inactivation of

the MMR system results in the propagation of mutations during subsequent cell division

(Söreide et al., 2006).

In sporadic CRC, the MSI+ phenotype is typically associated with poorly differentiated

tumours, strong peritumoural lymphocytic infiltration, older age, female gender and

proximal location in the colon (Söreide et al., 2006). Despite having aggressive tumour

features including larger (T3) primary tumour, poor differentiation and deep tumour

invasion, MSI+ CRC are thought to have better prognosis than MSI- CRC (Söreide et al.,

2006). The survival benefit gained by MSI+ CRC patients receiving 5-fluorouracil (5-FU)

treatment is equivocal due to inconsistent results reported across different studies

(Carethers et al., 2004; Elsaleh et al., 2001; Hemminki et al., 2000; Jover et al., 2006;

Kim et al., 2007; Popat et al., 2005; Ribic et al., 2003).

1.2.3 Epigenetic instability in colorectal cancer

Epigenetic instability represents an alternative mechanism to genomic instability for

driving CRC tumourigenesis. The term “epigenetics” was first coined by Conrad

Waddington to describe heritable phenotypic changes without alterations in the genetic

sequence (reviewed in Probst et al., 2009; Tost, 2009). The epigenetic machinery consists

of chromatin remodelling, histone modification, DNA methylation and non-coding RNA.

Epigenetic modifications superimposed on the genetic sequence dictate the eventual

phenotypic traits of a multicellular organism. DNA methylation modulates normal

development via effects on tissue-specific gene expression, X-chromosome inactivation,

genomic imprinting and suppression of parasitic repetitive elements in the genome.

Abnormal DNA methylation patterns in comparison to normal colonic mucosa are

frequently observed in CRC and form the topic of investigation of this thesis.


Methylation of DNA involves addition of a methyl-group from S-adenosylmethionine

(SAM) to the 5’-carbon of the cytosine ring in a CpG dinucleotide. The distribution of

CpGs in the genome is disproportionate, with dense clustering of the CpG sequence in

regions termed CpG islands (CGIs). CGIs are defined as genomic domains of more than

0.5kb in length with a ratio of observed CpG to expected CpG of >0.65 (Grønbaek et al.,

2007; Takai and Jones, 2002). Approximately 60% of all genes in the human genome

contain a CGI in the promoter. The number of methylated CpGs are under-represented in

the genome, accounting for only 1% of total nucleotides as a consequence of spontaneous

deamination of methylated cytosine to thymine during the course of evolution (Grønbaek

et al., 2007). Inactivation of gene transcription by methylation of CGIs with concomitant

histone modification and formation of heterochromatin is a common pathogenic

phenomenon in various malignancies.

1.3 THE CpG ISLAND METHYLATOR PHENOTYPE (CIMP)

1.3.1 CIMP as a distinct CRC phenotype

A subset of CRCs exhibiting simultaneous hypermethylation of multiple gene promoters

has been classified as the CIMP+ CRC. CIMP was first described by Toyota et al in 1999

following the analysis of 50 tumours for methylation in 7 “methylated in tumour” (MINT)

markers (Toyota et al., 1999a). The bimodal distribution pattern for methylation of the 7

MINTs was used to segregate CRC into two groups, with the CIMP+ tumours showing

strong association with MSI+ and female patients (Toyota et al., 1999a). The existence of

CIMP in CRC was subsequently challenged by other groups who proposed that CIMP+

tumours were merely a subset of MSI+ tumours and that methylation levels amongst

CRCs were evenly distributed when a larger number of genes was analysed (Anacleto et

al., 2005; Yamashita et al., 2003).


The dispute over CIMP+ CRC as a distinct subtype was resolved when Laird and

coworkers carried out an unbiased, quantitative methylation analysis of 195 genes in 295

CRCs (Weisenberger et al., 2006a). They proposed the use of a 5-marker panel consisting

of RUNX3, IGF2, CACNA1G, NEUROG1 and SOCS1 for the standardized classification

of CIMP+. Using this panel to define CIMP+, a tight association was observed with

BRAF V600E mutation independently of MSI status (Weisenberger et al., 2006a). The

existence of CIMP+ as a distinct CRC subgroup has since been validated by independent

groups working on large population-based series. From these studies, it has emerged that

CIMP+ occurs in ~15% of all CRC and is frequently associated with MSI+, BRAF

mutation, older age, proximal tumour site and female gender (Barault et al., 2008a;

Hawkins et al., 2002a; Ogino et al., 2006a; Ogino et al., 2007a; Samowitz et al., 2005a;

van Rijnsoever et al., 2002; Weisenberger et al., 2006a). CIMP+ CRC also show

distinctive gene expression, as demonstrated in a series of CIMP+ MSI- CRC (Ferracin et

al., 2008b).

1.3.2 Clinical features of CIMP+ CRC

The prognostic significance of CIMP+ may depend on the MSI status. In MSI+ CRC,

CIMP+ has no prognostic value since the longer survival conferred by the MSI+

phenotype appears to predominate over the CIMP status (Ogino et al., 2009; Samowitz et

al., 2005a; Ward et al., 2003). In contrast, MSI- CIMP+ CRC patients have poorer

survival compared to MSI- CIMP- patients (Barault et al., 2008a; Ferracin et al., 2008b;

Hawkins et al., 2002a; Lee et al., 2008; Ogino et al., 2007c; Shen et al., 2007a; Ward et

al., 2003). When BRAF mutation and MSI status are taken into consideration, CIMP+ was

independently associated with lower cancer-specific mortality (Ogino et al., 2009).

CIMP+ has been reported to show predictive value for survival benefit from adjuvant

treatment with 5-FU in stage III CRC patients (Van Rijnsoever et al., 2003). The tight


associations of BRAF mutation and MSI+ with CIMP+ highlights the need to take these

factors into account when evaluating the influence of CIMP+ on clinical outcomes

(Iacopetta et al., 2008; Ogino et al., 2009).

1.3.3 BRAF and KRAS mutations in CIMP+ CRC

The RAS/RAF/MEK/ERK pathway is frequently deregulated in CRC through gain-of-

function mutations in KRAS and BRAF (Downward, 2003). The V600E substitution

represents more than 90% of reported mutations in BRAF and shows a 500-fold elevated

kinase activity compared to wild type BRAF (Crook et al., 2009). This mutation is found

in 5-15% of all CRC and up to 70% of CIMP+ CRC (Davies et al., 2002; Li et al., 2006a;

Nosho et al., 2008a; Rajagopalan et al., 2002; Samowitz et al., 2005a; Weisenberger et

al., 2006a; Yuen et al., 2002). BRAF mutant CRC exhibit a distinct gene expression

profile, DNA methylation pattern and specific clinicopathological features (Kim et al.,

2006; Li et al., 2006a; Nagasaka et al., 2008b). The proclivity of BRAF mutation to

cluster with CIMP+ and MSI+ may be linked to the ability of mutated BRAF protein to

induce methylation in MLH1 (Minoo et al., 2007). BRAF mutation has been associated

with poor survival of CIMP+ CRC patients (French et al., 2008; Lee et al., 2008;

Samowitz et al., 2005b) and a more aggressive tumour phenotype compared to BRAF

wildtype tumours (Lee et al., 2008; Minoo et al., 2007; Oliveira et al., 2007).

KRAS mutations occur early in colorectal tumourigenesis (Fearon and Vogelstein, 1990;

Velho et al., 2008). Mutations in codons 12, 13, and less frequently codon 61 result in

constitutive activation of the GTP-bound RAS protein and are found in ~40% of CRC

(Walther et al., 2009). Although KRAS mutations are only weakly prognostic in CRC

(Andreyev et al., 1998; Andreyev et al., 2001), they are highly predictive for response to

anti-EGFR therapies (Normanno et al., 2009). KRAS mutations have been associated with


distinct methylation signatures (Horii et al., 2009; Nagasaka et al., 2008a; Nagasaka et

al., 2008b; Ogino et al., 2006b; Shen et al., 2007b; Suehiro et al., 2008) and are almost

always mutually exclusive with BRAF mutation (Oliveira et al., 2007). Although KRAS

and BRAF mutations can induce similar effects via the same signalling pathway, BRAF

mutant cancer cells are preferentially dependent on MEK/ERK signalling whereas KRAS

mutations exert their effect through other effectors including phosphoinositide 3-kinases

(PI3K) and ral guanine nucleotide dissociation stimulator (RALGDS) (Preto et al., 2008).

The differential reliance of KRAS and BRAF mutant cells on activation of the

RAS/RAF/MEK/ERK pathway suggests a synergistic function for each mutation, as

observed in advanced CRC (Oliveira et al., 2007).

1.3.4 Non-genetic risk factors for the development of CIMP+ CRC

Enzymes linked to methionine metabolism and including folate and glutathione (GSH)

reactions are significantly downregulated in CIMP+ tumours (Ferracin et al., 2008b).

Consistent with this finding, CIMP+ CRC have lower gamma-glutamyl hydrolase mRNA

levels (Kawakami et al., 2008), which probably explains the higher levels of folate

intermediates observed in CIMP+ compared to CIMP- CRC (Kawakami et al., 2003).

Increased levels of GSH and GSH-related enzymes in the proximal colon of elderly

females (Hoensch et al., 2006) may signal an inherent predisposition to deregulation of

the GSH pathway and the development of CIMP+ CRC.

Intrinsic differences in methylation levels in the normal colonic mucosa may also underlie

the preponderance of CIMP+ CRC in elderly females. Methylation levels of MLH1 show

a marked decrease from the right to left colon with increasing age in females (Kawakami

et al., 2006; Menigatti et al., 2009). Expression of the DNA methylating enzyme

DNMT3B is also higher in the liver of older females compared to younger males (Xiao et


al., 2008). On the other hand, promoter-specific methylation levels in lymphocyte DNA

are higher in Asian males than Asian females (Sarter et al., 2005).

Exogenous factors may also modulate the risk for developing CIMP+ CRC. Samowitz et

al reported a positive association between smoking and CIMP+ CRC in a large

population-based study (Samowitz et al., 2006). The association between smoking and

CIMP+ CRC may be linked to activation of the aryl hydrocarbon receptor by dioxin in

cigarette to induce gene hypermethylation (Ray and Swanson, 2004) and to higher

DNMT1 expression observed in the liver of smokers compared to non-smokers

(Hammons et al., 1999).

Folate, a methyl donor for the remethylation of homocysteine to methionine, is an

essential substrate for DNA methylation. Methionine is converted to SAM, the universal

methyl donor used in all cellular one-carbon transfer reactions (Figure 1.2) (Kim, 2005;

Ulrich et al., 2008). However, studies that have examined the association between folate

intake and DNA methylation levels in animal models and human intervention trials have

produced discordant results (Kim, 2004; Kim, 2005; Ulrich et al., 2008). In a mouse

model, enhanced availability of methyl groups through folate supplementation

significantly increased both global DNA methylation and p16 promoter methylation,

particularly in older animals (Keyes et al., 2007). In contrast, folate depletion in rats was

associated with a 30% increase in genomic DNA methylation (Sohn et al., 2003).


Figure 1.2 Cellular methyl group metabolism, DNA synthesis and DNA methylation.

B12, vitamin B-12; DHFR, dihydrofolate reductase; CH3, methyl group; CpG cytosine-

guanine dinucleotide sequence; MTHFR, methylenetetrahydrofolate reductase; SAH, S-

adenosylhomocysteine; SAM, S-adenosylmethionine; THF, tetrahydrofolate (Kim, 2005).

Folate depletion in elderly humans (>60 years) and postmenopausal women led to

reduced global DNA methylation levels in lymphocytes (Jacob et al., 1998; Rampersaud

et al., 2000). Nonetheless, folate deficiency in young males did not affect the in vivo

methylation capacity of lymphocytes (Jacob et al., 1995). Low folate and high alcohol

consumption were associated with a higher frequency of multiple hypermethylated genes

in CRC patients compared to those with high folate and low alcohol intake (van Engeland

et al., 2003). However, no correlation was found between dietary folate levels and the

development of CIMP+ CRC in a large case-control study (Slattery et al., 2007). The

association between folate intake and development of CIMP+ cannot be ruled out as more

quantitative and direct measures of folate level such as plasma folate levels may be

needed as accurate indicators of folate levels in these patients. Further studies using

different CIMP+ markers with longer patient follow-up time may also be more


informative to determine the role of dietary folate in the development of CIMP+ CRC.

Other metabolites such as vitamin B12, choline and methionine can interact with folate to

influence the level of DNA methylation (Ulrich et al., 2008). The vitamin B12 level in

serum, but not folate level, was inversely correlated to ER� methylation in normal colonic

mucosa (Al-Ghnaniem et al., 2007). The extent to which folate modulates DNA

methylation is therefore likely to depend on gender, age, duration of folate depletion,

tissue type and other metabolites in the folate cycle.

1.3.5 Genetic risk factors for the development of CIMP+ CRC

Firm evidence for the existence of genetic risk factors for CIMP+ CRC is lacking.

Although Frazier et al reported a 14-fold higher likelihood of positive family history of

cancer in 47 patients with tumours showing frequent methylation (Frazier et al., 2003),

analysis in a larger cohort of 562 unselected CRC found no association between CIMP+

and family history of cancer (Ward et al., 2004). Nonetheless, the lower risk for CIMP+

CRC in Southern Europeans compared to Anglo-celtic Caucasians after adjusting for

other known risk factors suggests a possible genetic predisposition in the pathogenesis of

CIMP+ CRC (English et al., 2008).

Genetic variants in MLH1 (-93G>A; rs1800734) (Samowitz et al., 2008) and MSH6

(116G>A; rs1042821) (Curtin et al., 2009) have been associated with increased risk for

CIMP+ CRC. In conjunction with folate intake, polymorphisms in genes involved in one-

carbon metabolism pathways (Figure 1.2) could influence the level of DNA methylation

in normal colonic mucosa and therefore the risk of developing CIMP+ CRC. Carriers of

the methylene tetrahydrofolate reductase (MTHFR) A1298C variant allele with low folate

and high alcohol intake showed an increased risk for CIMP+ CRC (Curtin et al., 2007).

Although MTHFR 677TT homozygous individuals were most susceptible to reduced


DNA methylation under low folate conditions (Axume et al., 2007; Friso et al., 2002),

they did not show a decreased risk for CIMP+ CRC (Curtin et al., 2007). A recent genetic

epidemiology study has shown that the risks for CRC associated with common variants in

MTHFR and DNMT3b are specific for tumours that arise in the proximal colon (Iacopetta

et al., 2009), in keeping with the strong predilection for CIMP+ CRC to occur in this part

of the colon.

1.3.6 The serrated adenoma pathway in the development of CIMP+ CRC

The existence of a CRC pathway distinct from the classic Vogelstein pathway is

suggested by the finding of tumours which have frequent DNA hypermethylation and

BRAF mutations but which lack CIN-type features such as LOH, aneuploidy and TP53

mutation. To this end, Jass has proposed the serrated pathway involving serrated

adenomas (SA) as the precursor lesion for CIMP+ CRC (Jass, 2005; Jass et al., 2002; Jass

et al., 2000). SAs are a heterogeneous group of polyps with a characteristic serrated

histological architecture that distinguishes them from traditional adenomas. SAs

encompass a spectrum of admixed polyps, sessile serrated adenoma, traditional serrated

adenoma and some large hyperplastic polyps (Jass, 2007a; Mäkinen, 2007; Noffsinger,

2009). A signature gene expression profile for SAs supports their existence as a separate

subgroup of polyps (Kim et al., 2008). SAs share several molecular features with CIMP+

CRC including high frequencies of MSI+, BRAF mutation and promoter

hypermethylation that also distinguishes them from conventional adenomas (Carr et al.,

2009; Hawkins and Ward, 2001; Kambara et al., 2004; Konishi et al., 2004; O'Brien et

al., 2006; Spring et al., 2006; Wynter et al., 2004). SAs show a predilection for the

proximal colon and occur more frequently in women (Mäkinen, 2007; Noffsinger, 2009).

They comprise just 9% of all polyps detected in patients undergoing colonoscopy for


standard clinical indications and account for 8-22% of all serrated lesions (Goldstein et

al., 2003; Spring et al., 2006).

The juxtaposition of SA with serrated adenocarcinoma (Jass et al., 2000; Mäkinen et al.,

2001) and MSI+ CRC (Hawkins and Ward, 2001) is highly suggestive of the malignant

potential of these lesions. Consistent with these findings, the presence of large SAs was

shown to correlate tightly and independently with synchronous advanced CRC (Li et al.,

2009). Patients with SAs also showed a significant positive family history of CRC

(Spring et al., 2006). Risk factors for the development of SAs appear to differ according

to anatomical site in the colon, with folate supplementation associated with an increased

risk of proximal SA and smoking and obesity associated with increased risk of distal SA

(Wallace et al., 2009).

1.3.7 Serrated pathway syndrome

A familial syndrome for CIMP+ CRC has been proposed based on the hyperplastic

polyposis (HPP) and serrated pathway (SPS) syndromes (Young and Jass, 2006; Young et

al., 2007). SAs that arise in these conditions show frequent BRAF mutation and elevated

DNA methylation. HPP is an extremely rare condition with infrequent familial

segregation that suggests a recessive mode of inheritance (Young and Jass, 2006; Young

et al., 2007). The large, atypical and dysplastic SAs pose the highest risk for malignant

transformation in HPP individuals. Although the increased risk for CRC associated with

selected HPP patients has been described in several case reports (Abeyasundara and

Hampshire, 2001; Jass et al., 2000), the molecular pathway that underlies HPP remains

poorly understood (Carvajal-Carmona et al., 2007; Rashid et al., 2000). BRAF mutations

are frequently observed in the serrated lesions of HPP patients, together with increased

DNA hypermethylation in the normal colonic mucosa compared to patients with sporadic


serrated lesions (Chan et al., 2003; Minoo et al., 2006; Wynter et al., 2004). SPS is an

autosomal dominant condition proposed for familial predisposition to serrated neoplasia

(Young and Jass, 2006; Young et al., 2007). SPS was first described in individuals from

families fulfilling the Amsterdam I criteria and showing variable levels of MSI (Young et

al., 2005). SPS shows a predilection for development of advanced serrated polyps in the

proximal colon and predominantly in females (Young and Jass, 2006; Young et al.,

2007). The significant association reported between BRAF mutation and positive family

history of CRC provides circumstantial evidence to support the hypothesis that HPP and

SPS are familial syndromes for CIMP+ CRC (Vandrovcova et al., 2006; Young et al.,

2005).

1.3.8 Molecular mechanisms in the serrated pathway

Key molecular mechanisms in the serrated neoplastic pathway are DNA methylation,

inactivation of the MMR system and mutations in BRAF and KRAS (Jass, 2005; Jass et

al., 2002). Widespread DNA methylation in the normal colonic mucosa is a frequent

event observed in patients with CIMP+ CRC (Kawakami et al., 2006), HPP (Minoo et al.,

2006) and SAs (Kambara et al., 2004; O'Brien et al., 2006; Park et al., 2003; Velho et al.,

2008). Hypermethylation-induced inactivation of MLH1 may trigger the progression of

hyperplasia to dysplasia in MSI+ SAs (Jass, 2005; Jass et al., 2002). Methylation of

HPP1, coding for a transmembrane protein with follistatin and epidermal factor-like

domains, may also initiate the malignant transformation of SAs (Young et al., 2001).

Evidence to support the involvement of defective MMR repair in the evolution of serrated

neoplasia is provided by the segregation of MSI+ SA with MSI+ CRC (Hawkins and

Ward, 2001). Moreover, identical mutations in TGF�RII, BAX and IGF2R have been

identified in the serrated lesions and corresponding tumours from the same patients (Jass


et al., 2000). Inactivation of MLH1 and MGMT through hypermethylation is more

commonly observed in serrated lesions than traditional adenomas, indicating the

importance of DNA repair genes in driving tumour progression via the serrated pathway

(Oh et al., 2005).

Mutations in BRAF and KRAS, are over-represented in SAs and occur at a comparable

frequency to MSI+ CRC (Carr et al., 2009; Jass et al., 2000; Kim et al., 2008b;

Rosenberg et al., 2007; Velho et al., 2008). These mutations are thought to result in

failure to initiate shedding of epithelial cell leading to the serrated architecture

characteristic of SA (Jass, 2005; Jass et al., 2002). As mutations in BRAF and KRAS

results in pro-survival and pro-apoptotic responses respectively, aberrant signalling in the

RAS/RAF/MAPK pathway complemented by methylation-inactivation of other genes

such as DNA repair genes, MGMT and anti-adhesion gene HPP1 may together contribute

to the malignant transformation of SA.

1.4 HERITABILITY OF DNA METHYLATION

During normal development of a zygote, epigenetic marks are erased in the primordial

germ cells and then re-established before preimplantation through epigenetic

reprogramming (Cedar and Bergman, 2009; Probst et al., 2009). However,

hypermethylation of normally unmethylated genes, termed constitutional epimutation, has

been reported in the normal somatic cells of a small proportion of cancer patients

(Dobrovic and Kristensen, 2009). This includes hypermethylation of DAPK1 in chronic

lymphocytic leukaemia patients, BRCA1 in breast cancer patients and MLH1 in CRC

patients. Such findings raise the possibility of transgenerational inheritance of DNA

methylation in humans (Dobrovic and Kristensen, 2009). Epimutation has been proposed


as an alternative first-hit mechanism in Knudson’s theory of tumourigenesis (Cropley et

al., 2008).

The earliest report of epimutation was in HNPCC-like patients that lacked detectable

germline mutations in MMR genes. Hypermethylation of MLH1 in DNA derived from

peripheral blood cells, buccal mucosa and hair follicles of these patients suggested the

epimutation may be germline (Gazzoli et al., 2002; Suter et al., 2004). However, further

investigation of the heritability of MLH1 epimutation in the parents, offspring and

siblings of affected individuals showed that inheritance of epimutation was probabilistic

and did not conform to Mendelian inheritance (Hitchins et al., 2005; Hitchins et al., 2007;

Morak et al., 2008; Valle et al., 2007). Transgenerational inheritance of epimutation may

be influenced by cis-acting factors, as exemplified by the mosaic pattern of MSH2

epimutation with germline deletion of TACSTD1 (Ligtenberg et al., 2009). Most

epimutation carriers reported to date were diagnosed with CRC at a relatively young age

(Hitchins and Ward, 2009). A recent study suggests that epimutations may contribute as

many as 16% of suspected HNPCC cases (Niessen et al., 2009).

1.5 DNA HYPOMETHYLATION IN COLORECTAL CANCER

Global DNA hypomethylation in CRC occurs concurrently but independently of promoter

hypermethylation (Bariol et al., 2003; Frigola et al., 2005; Iacopetta et al., 2007). Loss of

methylation in repetitive elements is a major source of global hypomethylation

(Hoffmann and Schulz, 2005; Wilson et al., 2007). Chromosomal instability subsequent

to loss of global DNA methylation has been demonstrated in the development of T-cell

lymphoma in a mouse model with defective DNMT1 (Eden et al., 2003; Gaudet et al.,

2003). Loss of imprinting (LOI) at genes expressed in an allele-specific manner has also

been observed to contribute to cancer risk (Hoffmann and Schulz, 2005; Wilson et al.,


2007). Increased IGF2 expression due to LOI promotes the development of intestinal

tumours in a mouse model (Sakatani et al., 2005).

LINE1 hypomethylation occurs early in CRC and is detectable in preneoplastic lesions

(Suter et al., 2004). Demethylation of LINE1 in the normal colonic mucosa of CRC

patients is inversely associated with the presence of MSI+, CIMP+ and promoter

hypermethylation in the tumours (Estecio et al., 2007; Iacopetta et al., 2007; Ogino et al.,

2008). Tumours with LINE1 hypomethylation were associated with poor prognosis

(Frigola et al., 2005; Ogino et al., 2008b), possibly due to cumulative chromosomal

aberrations in the demethylated genome (Rodriguez et al., 2006). Global DNA

methylation levels are lower in the proximal compared to distal normal colonic mucosa

(Figueiredo et al., 2009) and may be modulated by dietary folate intake (Friso et al.,

2002; Mathers, 2005; Pufulete et al., 2005) and polymorphisms in folate metabolism

genes (Kawakami et al., 2006; Stern et al., 2000).

1.6 DNA METHYLTRANSFERASES

Three classes of DNA methyltransferase encoded by DNMT1, DNMT2 and DNMT3 are

found in mammalians, with the latter comprising DNMT3A, DNMT3B and DNMT3L

(Goll and Bestor, 2005). DNMT3L serves as a regulatory factor for DNMT3A and

DNMT3B (Cheng and Blumenthal, 2008; Schaefer et al., 2007) while DNMT2 acts as an

RNA methyltransferase (Goll et al., 2006; Rai et al., 2007; Rottach et al., 2009).

Although DNMT1 and DNMT3 have traditionally been classified as maintenance and de

novo methyltransferases, respectively, synergistic activities of both enzymes are required

for effective DNA hypermethylation (Esteller, 2007). Simultaneous inactivation of

DNMT1 and DNMT3 leads to loss of methylation and restoration of the expression of


genes previously silenced by hypermethylation (James et al., 2006; Paz et al., 2003; Rhee

et al., 2002).

Overexpression of DNMT1 in 40-70% of CRC (Kanai et al., 2001; Kang et al., 2007; Zhu

et al., 2007) is associated with elevated RNA levels of SUV39H1 histone

methyltransferase (Kang et al., 2007) and simultaneous hypermethylation of multiple

gene promoters (Kanai et al., 2001). The tumourigenic effect of DNMT1 in neoplastic

cells may not be mediated solely by its methyltransferase activity (Damelin and Bestor,

2007; Robert et al., 2003) as the loss of DNMT1 can induce chromosomal and mitotic

defects (Chen et al., 2007). Elevated levels of DNMT3B have also been reported in CRC

and may contribute to the hypermethylator phenotype (Kanai et al., 2001; Nosho et al.,

2009). DNMT3A expression increases in parallel with the extent of dysplasia from

adenoma to carcinoma (Schmidt et al., 2007). In APCmin/+ mice, DNMT3B promotes

tumourigenesis through targeted de novo methylation of gene promoters (Linhart et al.,

2007), consistent with the non-random and tumour-specific DNA methylation patterns

observed in cancer (Costello et al., 2000). De novo methylation of cancer genes is

orchestrated by the interaction of cis-acting DNA sequence and trans-acting protein

complexes that are able to recruit DNMTs for gene-specific methylation (Keshet et al.,

2006; McCabe et al., 2009).

1.7 OTHER COMPONENTS OF THE EPIGENETIC MACHINERY

Histone modifications and chromatin remodelling add another layer of epigenetic

transcriptional regulation (Figure 1.3). The methyl marks on DNA are recognised by a

group of methyl-domain binding proteins (MBDs) consisting of MeCP2, MBD1, MBD2,

MBD3, MBD4 and Kaiso (McCabe et al., 2009). MBDs bind to methylated DNA to


recruit transcriptional repressors including histone deacetylases, polycomb group proteins

and chromatin remodelling factors (Lopez-Serra and Esteller, 2008).

A combination of histone modifications including acetylation, methylation,

phosphorylation, ubiquitination, ADP-ribosylation, citrullination and sumoylation forms

the histone code to provide a dynamic transition platform for transcriptional inhibition

and initiation (Cedar and Bergman, 2009). Generally, histone acetylation is compatible

with active transcription while histone deacetylation leads to gene silencing (Cedar and

Bergman, 2009). Global loss of monoacetylation of histone 4 lysine 16 is a common

phenomenon observed in cancers (Fraga and Esteller, 2005; Iacobuzio-Donahue, 2009).

Truncating mutations in histone deacetylase 2 (HDAC2) have been observed in MSI+

CRC (Hanigan et al., 2008) and result in resistance to HDAC inhibitor (Ropero et al.,

2006). Histone methylation controls transcriptional activity according to the type of

methylation and the amino acid residue on which histone methylation occurs. Elevated

levels of histone methyltransferases are observed in several tumour types (Iacobuzio-

Donahue, 2009).

MicroRNA (miRNA) are a class of non-coding RNA that negatively regulate target gene

expression (He and Hannon, 2004). Complementary base-pairing of miRNA to its target

sequence degrades the target transcript by deadenylation and decapping of the mRNA

(Winter et al., 2009; Yang et al., 2008). Alternatively, competitive binding of the

Argonaute protein in the RNA-induced silencing complex to mRNA inhibits translation

initiation by enhancer (Kiriakidou et al., 2007). In CRC, miR-143 down-regulates

DNMT3A mRNA and protein expression (Ng et al., 2009). Abrogation of miR-122a

tumour suppressive function by deregulation of the APC/�-catenin pathway is critical for

cancer cell proliferation (Wang et al., 2009). The transcriptional regulatory role of


miRNA is subjected to stringent genetic and epigenetic regulation and is thus emerging as

a new target in the development of epigenetic therapies for various malignancies (Guil

and Esteller, 2009).

1.8 DNA HYPERMETHYLATION IN CANCER AS A CLINICAL

BIOMARKER AND THERAPEUTIC TARGET

Although DNA methylation is modifiable by environmental factors and is affected by the

aging process, it could potentially be used as a biomarker for early detection,

prognostication and therapeutic target in cancer (Mulero-Navarro and Esteller, 2008). For

example, DAPK and p16 methylation are useful prognostic markers in lung and CRC

patients (Esteller, 2002). MGMT methylation has predictive value for responsiveness to

carmustine or cyclophosphamide in glioma and lymphoma patients (Esteller, 2002).

Figure 1.3 The epigenetic events of DNA methylation, chromatin remodelling and

histone modification are altered in cancer. Unmethylated DNA is associated with open

chromatin conformation and active histone modification such as histone H3 and H4


acetylation and methylation of histone 3 lysine 4 (H3K4) to initiate transcription. In

contrast, methylated DNA is associated with closed chromatin conformation and

repressive histone modifications including di- or tri-methylation of H3K9 (H3K9me2/3),

H3K27me3, and/or H4K20me3 to inactivate transcription (McCabe et al., 2009).

GSTP1 methylation, identified in up to 80-90% of prostate cancers, could be a sensitive

and specific marker for the detection of this tumour type (Mulero-Navarro and Esteller,

2008). Methylation across multiple genes such as CIMP-specific genes may potentially be

predictive for response to 5-FU (Iacopetta et al., 2008).

The introduction of DNA methylation markers into clinical practice is logistically feasible

since methylation is a relatively stable covalent modification. Methylation analysis can be

performed on archival materials and surrogate tissues such as bodily fluids obtained

through minimal invasive procedures. The localisation of DNA hypermethylation to CpG

dinucleotides limits the assessment of alterations to CpG-rich regions only, effectively

reducing the size of genomic regions that require screening. Moreover, the analysis of

DNA methylation generates a positive signal that is readily detectable in a background of

normal DNA, as opposed to negative signals such as LOH that can be masked by the

presence of normal cells.

Owing to its high prevalence and reversible nature, DNA methylation is an ideal

therapeutic target. The demethylating agents 5-azacytidine and 5-aza-2’-deoxycytidine

have been approved by the US Food and Drug Administration for haematological

malignancies (Mulero-Navarro and Esteller, 2008). Although the therapeutic mechanism

of these agents has not been fully elucidated, they potentially restore the expression of

genes silenced by DNA methylation, thereby returning cells to their normal state (Cortez


and Jones, 2008). Despite concerns over the selective targeting of tumour cells, there have

so far been no reports of long-term deleterious effects in patients receiving these drugs

(Cortez and Jones, 2008). Future generations of demethylating agents such as zebularine

with a longer half-life are under development to overcome the acute haematological

toxicities caused by current methylation inhibitors (Cortez and Jones, 2008).

1.9 AIMS OF THIS RESEARCH

The general aim of this thesis was to address the shortcomings of DNA methylation

analysis by establishing robust and accurate quality controls. These quality controls and

quantitative DNA methylation analysis tools were then used to examine CIMP CRC in

greater detail, with particular emphasis on their etiology in young patients and on their

classification.

AIM 1

Background

Many existing assays for methylation detection are based on the analysis of bisulfite-

treated DNA and subsequent PCR amplification. For accurate characterization of

methylation, the number of available DNA templates following bisulfite reaction and

hence the availability of suitable target molecules for PCR amplification are critical. This

study was therefore undertaken to establish a standardised measure for ensuring

maximum efficiency of DNA recovery after sodium bisulfite conversion through accurate

quantification of genomic DNA.

Aim 1: To demonstrate use of the cycle threshold (Ct) value as a quality control

parameter for bisulfite PCR-based DNA methylation analysis.


This aim was addressed in Results Chapter 3. The results have been published in:

Ang PW, Toh HB, Iacopetta B, Soong R. An improved quality control for bisulfite-PCR-

based DNA methylation analysis: cycle threshold value. Clin Chem Lab Med. 2008;

46(8):1117-21.

AIM 2

Background

Current approaches for the quantitative evaluation of gene promoter methylation usually

assess CpG sites that are selected arbitrarily. This may impact upon the potential value of

promoter methylation as a clinically useful biomarker. Identification of CpG sites linked

to transcriptional silencing is therefore essential if methylation is to be used as a surrogate

marker for gene expression.

Aim 2: To develop and validate an approach for identifying individual CpG sites within

specific promoters whose methylation is most tightly linked with gene expression.

This aim was addressed in Results Chapter 4.

AIM 3

Background

CRC is a highly heterogeneous group of tumours displaying different methylation

patterns and clinicopathological characteristics. The lack of a consensus panel of

methylation markers to define the CIMP+ subgroup of tumours has led to inconsistent

results for the reported characteristics of CIMP+ tumours. Accurate classification of

CIMP+ CRC would facilitate further etiological and clinical investigation of this

important CRC subtype.


Aim 3: To profile DNA methylation in an unbiased and comprehensive fashion for

characterisation of CIMP subtypes in CRC in a consecutive series of CRC.

This aim was addressed in Results Chapter 5. The work is in preparation for publication:

Pei Woon Ang, Marie Loh, Natalia Liem, Fabienne Grieu, Aparna Vaithilingham,

Cameron Platell, Wei Peng Yong, Barry Iacopetta, Richie Soong. Comprehensive

profiling of DNA methylation in colorectal cancer reveals three subgroups with distinct

clinicopathological and molecular features (submitted for publication).

AIM 4

Background

Sporadic CIMP+ CRC with BRAF V600E mutation occur frequently in elderly CRC

patients but are less frequent in younger patients. Since hereditary factors are sometimes

implicated in young CRC patients, this study was undertaken to evaluate the prevalence

and the molecular and clinicopathological characteristics of CIMP+ in young patients

with a view to identifying possible familial cases.

Aim 4: To evaluate the incidence of CIMP+ and BRAF mutations in CRC from a

population-based study of young patients aged <60 years.

This aim was addressed in Results Chapter 6. The work has been published in:

Ang PW, Li WQ, Soong R, Iacopetta B. BRAF mutation is associated with the CpG island

methylator phenotype in colorectal cancer from young patients. Cancer Lett. 2009;

273(2):221-4.

Chapter 2 Methods and Materials 25

Chapter 2. Methods & Materials

2.1 CLINICAL SAMPLES AND CRC CELL LINES

2.1.1 Recruitment of Caucasian CRC patients

Colorectal tumour and adjacent normal colonic tissue samples were obtained from a

consecutive series of patients undergoing surgery for CRC at the St. John of God Hospital

(Subiaco, WA). This cohort has well-annotated clinicopathological features including

age, gender, tumour site, tumour stage, presence of lymphocytic infiltration and

assessment of perineural and extramural venous invasion. Assessment of genetic

alterations including MSI, BRAF V600E mutation and KRAS mutations in codons 12 and

13 were performed on all cases. All patients provided written, informed consent for

research to be carried out on their tissue samples.

2.1.2 Recruitment of Asian CRC patients

Frozen colorectal tumours were obtained from the National University Hospital Tissue

Repository, Singapore, according to institutionally approved protocols. Methylation

analysis of RUNX3 and DPYD were carried out on this set of primary tumours.

2.1.3 Formalin-fixed and paraffin embedded CRC tissues

Formalin-fixed and paraffin embedded (FFPE) CRC tissues were obtained from the

School of Surgery, University of Western Australia. These samples were from a large,

population-based study on MSI screening of tumours from young CRC patients for the

detection of HNPCC (Schofield et al., 2009). Four µM sections cut from FFPE tumour

blocks were stained with hematoxylin and eosin for tumour cellularity assessment. DNA

extracted from the FFPE tissues was analysed for methylation of the MLH1 promoter.


2.1.4 CRC cell lines

All CRC cell lines used in this study were obtained from American Type Culture

Collection (ATCC, Manassas, VA) and maintained under recommended conditions.

Information on the cell lines and culture media used in the study are shown in

Supplementary Table 2.1.

To recover cryopreserved cells from liquid nitrogen stocks, cells were thawed in a 37°C

water bath with continuous agitation and transferred into a sterile centrifuge tube

containing 2 ml of complete medium supplemented with 20% FBS, 2 mM L-glutamine,

100 U/ml penicillin and 100 �g/ml streptomycin sulfate. Cells were then centrifuged at

1,000 rpm for 10 minutes at room temperature. The supernatant was discarded and cells

were resuspended in 1 ml of complete medium. The recovered cells were then reseeded in

10 ml of complete culture medium appropriate for each cell lines in a humidified 5% CO2

atmosphere at 37°C, pH 7.2 in a T25 tissue culture flask (Corning, Lowell, MA) in an

incubator (Forma 310 Series, Thermo Fisher Scientific, Waltham, MA).

Cells were seeded into new plates when they reached 90% confluency and the media was

changed as appropriate to ensure normal cell growth. Used culture media was aspirated

and cells were washed with phosphate-buffered saline (PBS). For monolayer culture,

adherent cells were trypsinised with trypsin/EDTA for 5 minutes. Trypsinization was then

inhibited by adding 2 ml complete medium. The cell suspension was drawn into a pipette

and the cell layer rinsed several times to dissociate cells and dislodge any remaining

adherent cells. Where necessary, cells were counted with a hematocytometer and diluted

to an approximate number for subculture to reach confluence 3 to 5 days after the addition

of fresh medium. All cultures were regularly assessed for evidence of contamination and

aseptic technique was practiced throughout the procedures. Cells were harvested at the


fourth passage by transfer to a sterile tube followed by centrifugation at 1,200 rpm for 5

minutes. Cell pellets were stored at -80°C until nucleic acid extraction.

2.2 DNA EXTRACTION

2.2.1 DNA extraction from frozen tissues

DNA was extracted from frozen tissues using the phenol-chloroform method.

Approximately 150-200 mg of frozen tissue was ground with a pre-chilled mortar and

pestle to a fine powder in liquid nitrogen. The powdered tissue was then suspended in 1

ml of extraction buffer consisting of 0.5% SDS, 10 mM Tris-HCl (pH 8), 100 mM EDTA

(pH 8) and 20 �g/ml pancreatic RNase. Samples were thoroughly mixed to produce a

homogeneous solution. Four mg of Proteinase K was added to each sample to inactivate

proteolytic enzymes. Tubes were then incubated with shaking at 53°C for 24 hrs. The

lysed and digested samples were cooled to room temperature before addition of 2 ml of

phenol/chloroform/isoamyl alcohol (25:24:1) followed by mixing on a rotating wheel for

10 minutes. To resolve the aqueous and phenolic phases, samples were centrifuged for 30

minutes at 5,000 rpm at room temperature. Genomic DNA was precipitated from the

aqueous solution by adding half the volume of 7.5M ammonium acetate and 2 volumes

(of the original amount of aqueous layer) of 100% ice-cold ethanol. Tubes were then

inverted multiple times and incubated at -20°C for 15 minutes. The DNA pellet was

thoroughly rinsed with 70% ethanol and subsequently air-dried for 30 minutes at room

temperature. DNA was resuspended at room temperature in TE buffer at ~1 mg/ml until

dissolved and then stored at -20°C until use.

2.2.2 DNA extraction from paraffin-embedded tissues

DNA was extracted from paraffin-embedded tissues using the Proteinase K method

(Soong’97 Mod Path). Briefly, sections of 25 �m thickness cut from paraffin-embedded


tissue blocks were added to 300 �l of digestion buffer consisting of 50 mM Tris-HCl, 1

mM EDTA and 0.5% TWEEN20 (pH 8.5) and incubated at 94°C for 10 min to dissolve

the paraffin. The tissues were separated from the paraffin by centrifuging at 13,200 rpm

for 10 min. The tube was then cooled to 4°C to allow the paraffin to solidify, enabling the

underlying tissues to be removed from the solution and transferred to a new 1.5ml

microcentrifuge tube. Fresh digestion buffer (200 �l) and 20 �l of 20mg/ml Proteinase K

(Invitrogen; Carlsbad, California) were added to the tissue and the mixture incubated at

55°C for 48 hrs on a rotator. The Proteinase K was inactivated by heating at 94°C for 10

minutes, following which the solution was centrifuged at 13,200 rpm for 10 minutes. The

supernatant containing the extracted DNA was stored at -20°C until use.

2.2.3 DNA extraction from cell lines

Cell pellets were resuspended in PBS buffer and digested with 20 �l proteinase K before

undergoing DNA extraction using the DNeasy Blood & Tissue Kit (Qiagen, Valencia,

CA) as recommended by the manufacturer. DNA was eluted in 10 mM Tris-HCl, pH 8.0

at a concentration of approximately 1 mg/ml and stored at -20°C until use.

2.3 RNA EXTRACTION

2.3.1 RNA extraction from frozen tissues

Surgical tissue samples were immediately snap-frozen in liquid nitrogen and stored at -

80°C until use. Total RNA was isolated using TRI Reagent® (Molecular Research

Centre, Cincinnati, OH) in conjunction with further purification using the RNeasy® Mini

Kit (Qiagen, Valencia, CA). Approximately 50 mg of tissue was ground to a fine powder

under liquid nitrogen using a mortar and pestle. The suspension of tissue powder in liquid

nitrogen was transferred to a sterile 2 ml microcentrifuge tube. The liquid nitrogen was

allowed to evaporate without thawing of the samples and this was followed by addition of


1 ml of TRI Reagent® to lyse the tissues. Homogenisation of the tissue was carried out on

ice by multiple passing of the lysate through a 20-gauge needle until a homogeneous

lysate was produced. This was incubated at room temperature for 5 minutes before

addition of 200 �l chloroform. The resulting mixture was mixed vigorously for 15

seconds before further incubation at room temperature for 15 minutes, followed by

centrifugation at 15,000 rpm for 15 minutes at 4°C to separate the organic and aqueous

phases. RNA isolated from the aqueous phase was transferred to a new 2 ml

microcentrifuge tube and mixed with 500 �l isopropanol. The precipitation was mixed by

inversion and incubated at room temperature for 5 minutes. The mixture was then

transferred to a spin column from the RNeasy® Mini Kit (Qiagen, Valencia, CA) for

further purification. RNA was eluted in 10 mM Tris-HCl, pH 8.0 at a concentration of

approximately 1 mg/ml, dispensed in single use aliquot tubes and stored at -80°C until

use.

2.3.2 RNA extraction from cell lines

To isolate RNA from cell lines, residual culture medium was completely removed from

the cell pellets before proceeding with RNA extraction using the RNeasy® Mini Kit

(Qiagen, Valencia, CA) following the manufacturer’s instructions. RNA was eluted in 10

mM Tris-HCl, pH 8.0 at a concentration of approximately 1 mg/ml and stored at -80°C

until use.

2.4 QUALITY CONTROL OF NUCLEIC ACIDS

2.4.1 Quantitation and qualitative assessment of genomic DNA

The purity of DNA was assessed by the ratio of absorbance readings at 260 nm and 280

nm (A260/A280) and the yield by absorbance at 260 nm (A260). These readings were

performed using a spectrophotometer (Nanodrop ND-1000, Thermo Fisher, Wilmington,


DE) and with the DNA dissolved in 10 mM Tris-HCl, pH 7.5 buffer. Samples with ratios

from 1.7 to 1.9 were considered pure and used for subsequent analysis. Real-time

quantification of the housekeeping gene �-actin (ACTB) was performed on all samples to

assess the amount of amplifiable genomic DNA, as described in Chapter 3.

2.4.2 Quantitative and qualitative assessment of total RNA

RNA was assessed for purity and concentration based on the measurement of A260/A280

values as described for DNA. Samples with readings within the optimal range of 1.9 to

2.1 were considered to have passed quality control. The integrity of RNA was also

assessed using agarose gel electrophoresis to examine the intensities of 28S rRNA and

18S rRNA bands. Samples were considered intact when the ribosomal bands appeared

sharp and showed an approximately 2:1 ratio of 28s rRNA to 18S rRNA. The amount of

amplifiable RNA was quantified by reverse-transcriptase real-time PCR (RT-PCR) using

the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) housekeeping gene.

2.5 QUANTITATION OF RNA EXPRESSION LEVELS

The High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City,

CA) was used to reverse transcribe RNA to complementary DNA (cDNA). The total

reaction volume of 20 µl comprised of 2 µl of 10x RT buffer, 0.8 µl of 25x dNTP mix

(100 mM), 2 µl of 10x RT random primers, 1 µl of MultiScribe™ Reverse Transcriptase,

1 µl of RNase Inhibitor and 1 µg RNA. cDNA was synthesised by preincubation at 25°C

for 10 minutes, then reverse-transcribed for 120 minutes at 37°C, before denaturation at

85°C for 5 minutes.

Quantitative RT-PCR was performed using the TaqMan® Gene Expression Assays on the

7900HT Fast Real-Time PCR System (all from Applied Biosystems, Foster City, CA).


The inventoried gene expression assay IDs are shown in Supplementary Table 2.7.

Amplification was performed in a total volume of 20 µl containing 1 µl of TaqMan®

Gene Expression Assay (20x), 9 µl of cDNA template and 10 µl of TaqMan® Fast

Universal PCR Master Mix (2x) without AmpErase® UNG underwent the following

cycling conditions: 50°C for 2 minutes, 95°C for 10 minutes, followed by 60 cycles at

95°C for 15 seconds and 60°C for 1 minute. Cell lines expressing the gene of interest

were used as calibrators and GAPDH was used as a reference gene for each assay.

The RNA level was quantified using the comparative Ct method as described by

Schmittgen and Livak (Schmittgen and Livak, 2008). The fold change in expression,

normalised to the endogenous reference (GAPDH) and relative to a calibrator is expressed

as 2–��CT, where ��Ct = (Ctgene of interest – CtGAPDH) sample – (Ct gene of interest – CtGAPDH)

calibrator.

2.6 SCREENING FOR KRAS MUTATION, BRAF MUTATION AND MSI

2.6.1 Fluorescent-single strand conformational polymorphism (F-SSCP)

Single-strand conformational polymorphism (SSCP) is a rapid and efficient method for

the detection of sequence variation and is based on the differential electrophoretic

migration of single stranded DNA. This adopts different secondary structures as

determined by sequence-specific intramolecular base-pairing, thus giving rise to

differences in migration patterns between wildtype and mutant DNA fragments

(Humphries et al., 1997). Under optimized conditions, the detection sensitivity can

approach 100% (Ellison et al., 1993).

F-SSCP involves four major steps: PCR amplification of the target region with

fluorescent-labelled PCR primers typically giving rise to a 100 to 200bp fragment,


denaturation of the double-stranded PCR products, self-annealing of single stranded DNA

under low temperature followed by electrophoretic migration under non-denaturing

conditions (Makino et al., 1992).

2.6.2 PCR amplification of KRAS, BRAF and BAT26

Screening for KRAS codon 12 and 13 mutations, BRAF V600E mutations and MSI was

performed as described previously (Iacopetta and Grieu, 2000; Li et al., 2006a; Wang et

al., 2003). MSI was detected as deletions in the BAT26 quasi-monomorphic

mononucleotide repeat. The primer sequences and annealing temperatures used for PCR

are listed in Supplementary Table 2.2. The gel conditions (%Polyacrylamide/ %Glycerol)

used for F-SSCP analysis were for KRAS (12%/2%), BRAF and BAT26 (8%/2%). PCR

was performed in a total reaction volume of 16 �l consisting of 1x polymerization buffer,

1x Q-solution, 200 �M of each dNTP, 3 mM MgCl2, 0.5 �M of each forward and reverse

primer, 0.5U Taq DNA Polymerase (Qiagen, Hilden, Germany) and 1 �l of genomic

DNA. Hot-start PCR reaction was carried out using the following cycling conditions:

94°C for 5 min followed by 35 cycles of three-step cycling consisting of 94°C for 30 sec,

the appropriate annealing temperature for 30 sec, 70°C for 30 sec and a final extension at

70°C for 10 min.

2.6.3 F-SSCP screening for KRAS mutations, BRAF mutations and MSI

Two �l of fluorescent-labelled PCR product was diluted with 9 �l of deionized

formamide loading dye containing 0.05% dextran blue and denatured at 94°C for 5

minutes. One �l of this mixture was loaded onto a non-denaturing polyacrylamide gel (80

�m thickness, 18 cm long) mounted in a real-time DNA fragment analyzer (Gel-Scan

2000, Corbett Research, Sydney) for detection of the HEX fluorochrome incorporated

during PCR. Samples were pulse-loaded for 20 seconds at 1400 V, the wells were then


rinsed thoroughly before the gel was run for 2 hours at 1400 V in 0.8x TBE (Tris-borate-

EDTA) buffer at a constant temperature of 25°C as regulated by a built-in cooling unit.

ONE-Dscan 1.3 software (Scanalytics, Billerica, MA, USA) was used to analyse the

eletrophoretogram and samples displaying weak signal intensity were adjusted with the

aid of software to facilitate data interpretation.

2.7 DNA METHYLATION ANALYSIS

2.7.1 Sodium bisulfite conversion of genomic DNA

Sodium bisulfite converts unmethylated cytosine to uracil. Resistance of methylated

cytosine to the conversion results in methylation patterns that are preserved in the form of

genetic polymorphisms after bisulfite treatment that can be examined by various

methodologies. Sodium bisulfite conversion of genomic DNA was performed using the

EZ-DNA Methylation kitTM according to the manufacturer’s instructions (Zymo

Research, Orange, CA) apart from the study described in Chapter 3. The quality control

system established from that study was used for later experiments.

2.7.1.1 Sodium bisulfite conversion of genomic DNA (in-house protocol)

Sodium bisulfite conversion of genomic DNA was performed as described by Frommer et

al with slight modification (Frommer et al., 1992). For each batch of samples undergoing

sodium bisulfite treatment, a negative control (water sample without DNA template) was

included to check for contamination. DNA samples were sheared by passing through a

fine needle, followed by denaturation with 5 �l of 3 M NaOH added to a final

concentration of 0.3M. The DNA was then incubated at 37°C for 10 minutes and kept on

ice thereafter to maintain it in single-stranded form. In a total volume of 500 �l, the

denatured DNA, overlaid with mineral oil, was incubated with freshly prepared 3.1 M

sodium bisulfite, 0.5 mM hydroquinone (pH 5.0) at 50°C for 16 to 18 hours. Desalting


was carried out using the Wizard® DNA Clean-Up System (Promega, Madison, WI).

Briefly, the sample mixture was added to 1 ml of resin in a 1.5 ml microcentrifuge tube

and mixed by inversion. The mixture was then passed through a column followed by

washing with 2 ml of 80% isopropanol. To dry the resin, columns were then centrifuged

for 2 minutes at 13,000 rpm. Pre-warmed nuclease-free water (50 �l) was added to the

column and allowed to stand for 1 minute followed by centrifugation at 13,000 rpm for 1 minute

to elute the DNA. NaOH (1.6 �l of 10 M) was added to the eluted DNA to a final

concentration of 0.3 M and incubated at room temperature for 10 minutes. One �l of

glycogen (10 �g/ml) and 90 �l ammonium acetate (10 M, pH 7.0) to a final concentration

of 3 M was added to each tube, together with 200 �l of absolute ethanol. The reaction was

mixed by vortexing and subsequently centrifuged at 13,000 rpm at 4°C for 30 minutes.

The resulting supernatant was removed and the precipitated DNA pellet was washed

twice with 70% ethanol. Samples were air-dried at ambient temperature before being

resuspended in 20 µl of 10 mmol/L Tris-HCl (pH 8). All samples were stored at -20°C

and used within two months.

2.7.1.2 Sodium bisulfite conversion of genomic DNA (EZ-DNA Methylation kitTM)

Using procedures described in Chapter 3, samples were assessed to determine the

appropriate quantity of genomic DNA needed for sodium bisulfite conversion using the

EZ-DNA Methylation kitTM (Zymo Research, Orange, CA). Converted samples were

eluted in 50 µl of 10 mmol/L Tris-HCl (pH 8) and stored at -20°C for up to 2 months.

2.7.2 CpG Methyltransferase (M.SssI) treatment of DNA

Leukocyte DNA was incubated with M.SssI methylase (New England Biolabs, Ipswich,

MA) and SAM as the methyl donor to artificially methylate all CpG dinucleotides. A 20

�l reaction mix consisting of H2O, 1 �g genomic DNA, 0.16 mM SAM, 1x NE buffer 2


and 4 U M.SssI enzyme was incubated overnight at 37°C. Additional M.SssI enzyme and

SAM were added after 24 hours before a further 4-hour incubation to ensure complete

methylation of all CpG dinucleotides. The reaction was then heated to 65°C for 20

minutes to inactivate the enzyme. The completely methylated leukocyte DNA were stored

at -20°C and used as the universally methylated reference control in all methylation-

specific PCR (MSP), MethyLight and Pyrosequencing assays.

2.7.3 Methylation-specific PCR (MSP)

MSP is one of the earliest and most widely used bisulfite-dependent techniques for the

qualitative detection of differential DNA methylation in gene promoters (Licchesi and

Herman, 2009). Each assay comprises of “methylated” and “unmethylated” PCR in

separate reactions using primers that overlap multiple CG dinucleotides. Primer design is

critical for achieving maximal discrimination between methylated and unmethylated

sequences in MSP. Several parameters for the correct design of MSP primers are: (i) each

primer should contain at least two pairs of CG-dinucleotides and several cytosines outside

the CG-sequence at the 3’end to avoid false positives; (ii) the annealing temperature for

both sets of primers (methylated [M] and unmethylated [UM]) should be similar and

range between 55-65°C; (iii) the amplicon size should be less than 200 bp.

MSP of MLH1 was performed as described by Herman et al with minor modification

(Herman et al., 1998). PCR reactions were carried out in a total volume of 25 �l

consisting 1x PCR buffer, 200 �M of each dNTP, 1.5 mM MgCl2, 400 �M of each

primer, 0.04 U FastStart Taq polymerase (Roche, Mannheim, Germany) and 2 �l of

bisulfite-converted DNA. PCR conditions were as follows: 96oC for 3 minutes, 35 cycles

of 96oC for 20 seconds, annealing temperature for specific primers set at 45 seconds (M,

59°C; UM, 60°C) and 72oC for 30 seconds, followed by 72oC for 5 minutes. A negative


control (no template) and a positive control (fully methylated) were included in each MSP

run. PCR products were visualised on an automated, multicapillary DNA electrophoresis

system (QIAxcel system, Qiagen, Valencia, CA).

2.7.4 MethyLight

MethyLight quantifies DNA methylation events based on real-time fluorescent detection

of methylated sequence following conversion by sodium bisulfite. The flexibility in

design of methylation-specific primers and/or fluorescent probes allows highly sensitive

quantitation of DNA methylation at different resolution and the inclusion of inbuilt

controls (Campan et al., 2009).

To assess the efficiency and reproducibility of bisulfite conversion, three quality control

reactions were set up. Sample quantity and integrity were measured by a methylation-

independent assay that targeted cytosines outside the CG context. To monitor the

recovery efficiency for DNA following bisulfite treatment, probes and primers were

designed for amplification of a cytosine-free region. Completeness of bisulfite conversion

were indicated through amplification of a CG region using a set of common primers

lacking cytosine, in combination with multiple probe sets targeting non-CG cytosines

(Campan et al., 2009).

To measure DNA methylation levels at a specific gene locus using the MethyLight assay,

four PCR reactions were set up using bisulfite-treated DNA of the sample of interest and

a universally methylated reference control. For each sample, two PCR reactions were

performed: one PCR amplified the gene of interest to detect methylation events, while

another control PCR targeted CG-free sequences from a housekeeping gene in order to

account for the amount of input DNA. The level of methylated sequence discrimination


can be resolved at the probe hybridization level or amplification level, depending on the

CGs targeted by the primers and/or the probe. The versatility of the assay design thus

renders MethyLight capable of detecting as little as one methylated allele per 10,000

unmethylated alleles with high specificity (Campan et al., 2009; Eads et al., 2000). The

sensitivity can be further increased to a single-molecule level when digital PCR is

employed (Weisenberger et al., 2008).

MethyLight assays in this study were carried out as described by Eads et al with slight

modifications (Eads et al., 2000). Primer sequences for RUNX3, NEUROG1, IGF2,

CACNA1G, SOCS1 and APC used in the study described in Chapter 5 and 6 were

obtained from Weisenberger et al (Weisenberger et al., 2006a). MethyLight probe and

primers for RUNX3 and DPYD described in Chapter 4 were designed as follows:

RUNX3 forward primer 5’– CGTGGGGTTCGGAGGGCGCGTTCG – 3’

RUNX3 reverse primer 5’-ATACGCACGAACTCGCCTACG – 3’

RUNX3 probe 5’FAM – CGTTCGATGGTGGACGTG – 3’TAMRA

DPYD forward primer 5’ – GCGCGGGAGTCGTAGGATCGAGAGCG – 3’

DPYD reverse primer 5’ – ACCGACGACGCGAAAACGAAACGA – 3’

DPYD probe 5’FAM – CGCGAAACGACAACGCCCCCGAAACGA – 3’TAMRA

PCR amplification was carried out in a 20 �l reaction comprising of 10 �l Taqman® Fast

Universal PCR Master Mix (2X), 1 mM MgCl2, 0.25 �M probe and 0.75 M each of the

forward and reverse primers for the specific assay. Real-time PCR was carried out on an

ABI 7900HT (Applied Biosystems, Foster City, CA) using the following cycling

conditions: 95°C for 20 sec, followed by 40 cycles at 95°C for 1 sec and 60°C for 20 sec.

A methylation-independent assay targeting �-actin (ACTB) was used as an internal

control in combination with target-specific assays. DNA methylation values for each


locus (GENE) were based on the copy number derived from Ct values and converted to

percentage of methylation reference (PMR) by dividing the GENE/ACTB ratio of each

sample with the GENE/ACTB ratio of the universally methylated reference and

multiplying by 100 (Campan et al., 2009). For all genes the threshold value for

methylation was set at PMR � 4 as described by Weisenberger et al (Weisenberger et al.,

2006a).

2.7.5 Pyrosequencing

Pyrosequencing is a sequencing-by-synthesis method for quantitative measurement of

DNA methylation at a single CpG resolution level (Tost and Gut, 2007). The technology

is based on a four-enzyme cascade where inorganic pyrophosphate (PPi) is released

following nucleotide incorporation by DNA Polymerase I. Subsequently, adenosyl-

triphosphate (ATP) is produced by ATP sulfurylase using PPi as a substrate. The released

ATP then serves as the energy source for luciferase to generate light that is captured by a

charge-coupled device camera. Unincorporated nucleotides and excess ATP are removed

by Apyrase to ensure synchronised DNA synthesis and hence the precise assignment of a

specific nucleotide. Quantitative DNA methylation levels are deduced from the sequence

Pyrogram generated by the Q-CpG software (Dejeux et al., 2009).

Although Pyrosequencing is ideal for DNA methylation analysis that targets specific

sites, the short reading length of approximately 150 bp precludes analysis of a large

number of CpG sites. However, multiple CpGs within a target region of less than 350bp

can be interrogated using serial Pyrosequencing and the successive use of several

sequencing primers on the same template (Tost and Gut, 2007). The labelling of PCR

products by specific biotinylated primers for each assay can be circumvented with the use


of a single biotinylated primer that is simultaneously incorporated into a universally-

tagged PCR product during the amplification process (Royo et al., 2007).

Pyrosequencing for methylation analysis was performed in this study using methods

described by Tost et al (Tost and Gut, 2007). Briefly, PCR was carried out in a volume of

25 �l containing 1x PCR buffer, 200 �M of each dNTP, 2.5 mM MgCl2, 0.08 U FastStart

Taq polymerase (Roche, Mannheim, Germany), 400 �M each of PCR forward primer and

a 1:9 mixture of PCR reverse primer and universal biotinylated PCR primer based on a

published sequence (Tan et al., 2008) and 4 �l of bisulfite-converted DNA. PCR cycling

consisted of initial denaturation at 95oC for 7 minutes, 50 cycles of 95oC for 45 seconds,

annealing at the appropriate temperature for 45 seconds and extension at 72oC for 30

seconds, followed by 72oC for 5 minutes. A negative control without DNA template and a

positive methylated control were included in each PCR run. PCR products were verified

using the QIAxcel system (Qiagen, Valencia, CA).

Samples with successful amplification were subsequently analysed for DNA methylation

by Pyrosequencing using the PSQ96MA instrument (Biotage). The reaction mix

comprised of Pyro Gold Reagent kit (Biotage), 1x annealing buffer, binding buffer at pH

7.6 (10 mM Tris-HCl, 2 M NaCl, 1 mM EDTA, 1 ml/L Tween 20), 3 �l of Streptavidin

SepharoseTM High Performance beads (Amersham Biosciences, Uppsala, Sweden) and 15

�M pyrosequencing primer. Primer sequences, PCR annealing temperatures and the

pyrosequencing nucleotide dispensation order are shown in Supplementary Table 2.3.

Negative and positive PCR controls were always included in each run.


2.7.6 Cloning and sequencing of bisulfite-treated DNA

The cloning and sequencing of PCR products obtained using methylation-independent

primers from bisulfite-converted DNA template is considered the gold standard for

quantitative DNA methylation analysis. This approach results in methylation maps of

single DNA molecules. In contrast to direct sequencing of the PCR product which yields

average methylation values from a pool of molecules, bisulfite-cloning and sequencing

(bsSEQ) provides information about the methylation status of each CG site in single DNA

fragments of up to 500 bp in length (Zhang et al., 2009).

2.7.6.1 PCR for bisulfite-cloning and sequencing (bsSEQ)

To design primers for bsSEQ, CpG islands contained within the gene of interest were

identified using the MethPrimer software available online at

http://www.urogene.org/methprimer/index1.html (Li and Dahiya, 2002). Primer

sequences and optimal annealing temperatures for RUNX3 and DPYD studied by bsSEQ

in the current study are shown in Supplementary Table 2.4. For each gene, the 5’-end of

successive amplicons was design to overlap with the 3’end of the previous amplicon.

Methylation readings from the overlapping regions were compared and analysis was

performed based on results from the most 3’ end wherever discrepancies were observed.

The targeted region was amplified in a 25 �l PCR volume containing 1x PCR buffer, 200

�M of each dNTP, 2.5 mM MgCl2, 0.08 U FastStart Taq polymerase (Roche, Mannheim,

Germany), 500 nM of each forward and reverse primer and 4 �l of bisulfite-converted

DNA. The reaction was thermally-cycled as follows: 7 min at 95°C followed by 40 cycles

of 95°C for 30 sec, the specific annealing temperature for 45 sec and 72°C for 1 min; hold

at 72°C for 5 min.


2.7.6.2 PCR product purification and ligation

PCR products were resolved on 1.7% agarose gels and stained with SYBR® Safe DNA

gel stain (Invitrogen, Carlsbad, CA) diluted at 1:10,000 in 1X TBE buffer. The gel was

run for 20 minutes at 135V in 1X TBE and visualised with a UV transilluminator (Bio-

Rad Gel DocTM 2000 system, Bio-rad, Hercules, CA). Exposure to UV was kept to a

minimum to avoid the formation of pyrimidine dimers. Target bands were excised from

the gel and purified with the QIAquick DNA Purification kit (Qiagen, Valencia, CA) as

recommended by the manufacturer. Eluted DNA was quantified and assessed for purity

by spectrophotometry (Nanodrop ND-1000, Thermo Fisher, Wilmington, DE).

Purified PCR product was subcloned into pGEM®-T Vector using the pGEM®-T Easy

Vector Systems (Promega, Madison, WI) according to the manufacturer’s instructions. A

10 �l ligation reaction was set up for each sample along with a positive control using

control insert DNA provided with the kit and a negative control lacking DNA template.

Each reaction consisted of 5 �l of 2x rapid ligation buffer, 1 �l T4 DNA ligase, 1 �l

pGEM®-T vector (50 ng/ �l) and 150 ng of PCR product incubated overnight at 4°C.

2.7.6.3 Transformation and selection of clones

Transformation was performed using MAX Efficiency® DH5�™ Competent Cells

(Invitrogen, Carlsbad, CA). Competent cells were thawed on ice and a 50 �l suspension

was aliquoted into chilled microcentrifuge tubes containing 2 �l of ligation mixture.

Tubes were mixed by gentle flicking and incubated on ice for 30 minutes before

subjecting the transformation mix to heat shock at 42°C for 45 seconds in a water bath to

allow the uptake of plasmid into competent cells. Tubes were then returned to ice and

incubated for another 2 minutes, followed by the addition of 900 �l pre-warmed (37°C)

SOC medium. The transformed bacterial cells were allowed to recover by incubation at


37°C with shaking at 231 rpm for one hour. A volume of 100 µl of each transformation

mix was plated onto duplicate, pre-cast LB agar plates supplemented with 0.05 mg/ml

ampicillin, 0.5 mM IPTG and 80 µg/ml X-Gal. Plates were incubated for 16 hours at

37°C to allow bacterial cell growth.

Recombinant clones were identified by colour screening, with white colour indicating

successful cloning of the insert leading to disruption of the �-galactosidase coding

sequence. To verify the insert was the actual target amplicon, selected white clones were

PCR amplified using the same primers as for the initial PCR. PCR products were resolved

by electrophoresis (QIAxcel system, Qiagen, Valencia, CA) to confirm the presence of

insert with correct amplicon. Positive clones carrying the desired inserts were inoculated

into 4 ml LB broth supplemented with 0.01 mg/ml ampicillin and incubated overnight at

37°C with continuous agitation (231 rpm) for 16 hours. Plasmid DNA was isolated using

the Wizard® Plus SV Miniprep DNA Purification System (Promega, Madison, WI)

according to the manufacturer’s instructions.

2.7.6.4 Sequencing of plasmid DNA

Ten clones of each sample were sequenced using the BigDye Terminator v 3.1 Cycle

Sequencing kit (Applied Biosystems, Foster City, CA) as per the manufacturer’s

instructions. Sequencing reactions contained 2 µl plasmid DNA, 2 µl sequencing buffer, 2

µl RR-100 Sequencing enzyme and 320 nM of M13-reverse primer: 5’ -

CAGGAAACAGCTATGACC – 3’ in a 20 µl total reaction volume. Sequencing

reactions were thermally cycled at 96°C for 1 minute, followed by 27 cycles of 96°C for

20 seconds, 50°C for 10 seconds and 60°C for 4 minutes before a final extension at 72°C

for 7 minutes. The products were then purified and precipitated using the ethanol/sodium

acetate method. Briefly, 4 µl sodium acetate (3M, pH 4.6) and 50 µl 100% EtOH were


added to each sample and incubated on ice for 20 minutes to precipitate the extension

products. Tubes were then centrifuged for 20 min at 13,200 rpm and the resulting DNA

pellet was washed with 200 µl of 70% ethanol twice before drying in a vacuum

concentrator (MV-100 Micro Vac, Tomy Tech Inc, CA) for 15 minutes. Purified

sequencing products were analysed on a 3130xl Genetic Analyzer with a 50 cm array and

POP-7 polymer (Applied Biosystems, Foster City, CA). Resulting sequences were

processed for quality clipping using Sequencing Analysis 5.2 with the KB Basecaller

(Applied Biosystems, Foster City, CA) and analysed with Bioedit available at

http://www.mbio.ncsu.edu/BioEdit/bioedit.html (Ibis Bioscience, Carlsbad, CA).

2.7.7 Methylation analysis using Goldengate array

The Goldengate Cancer Panel I methylation array from Illumina (Illumina Inc., San

Diego, USA) is a hybridization-based array adapted from the single-nucleotide

polymorphism genotyping platform. It allows the analysis of methylation levels at 1,505

individual CpG sites contained within 808 cancer-related genes selected from published

literature. The array offers 96-sample throughput with 2.5% detection sensitivity and the

ability to distinguish differences of at least 17% in the methylation level (Bibikova and

Fan, 2009).

Each assay scheme uses two pairs of probes consisting of an allele-specific

oligonucleotide (ASO) and a locus-specific oligonucleotide (LSO) for interrogating CpG

dinucleotides in the methylated or unmethylated states. The 3’ end of the ASO binds to

the complementary “C” or “T” allele of the bisulfite-treated DNA (bsDNA) template,

while the 5’ end sequence is complementary to a universal primer sequence P1 or P2.

Complementary binding of the oligonucleotide is then extended and ligated to the LSO,

containing a locus-specific sequence at the 5’end, an address sequence in the middle


complementary to a capture sequence on the array, and a 3’ end sequence which serves as

a universal PCR priming site. Extension and ligation of the ASO to the LSO forms the

template for subsequent PCR utilizing the universal primers P1, P2, and P3. Each of the

universal primers P1 and P2 are fluorescently labelled and correspond to the methylated

“C” or unmethylated “T” allele, respectively. The methylation level at each locus is

calculated from the ratio of intensity of the methylated “C” dye to the total intensity of the

locus.

Array-based methylation analysis in the current study was performed using the Sentrix

Array Matrix format (Illumina Inc., San Diego, USA) according to the manufacturer’s

instructions. Briefly, genomic DNA was subjected to sodium bisulfite conversion using

the EZ-DNA Methylation kitTM (Zymo Research, Orange, CA) and biotinylated. The

biotinylated bisulfite-treated DNA (BS-DNA) was then reconstituted in solution after

removal of excess biotin. Query oligonucleotides were annealed to the bsDNA and ligated

and extended to form DNA template for amplification. Amplified products were

hybridised to the array and dried for imaging using the BeadArray Reader (Illumina). The

BeadScan and the BeadStudio software package incorporated in the BeadArray Reader

(Illumina) were used for image processing, intensity data extraction and exporting the

array intensity data for statistical analysis, respectively.

2.8 Statistical analysis

Statistical analysis for each data set and the bioinformatics approaches used in the study

are detailed in the relevant Results chapters as appropriate.

Chapter 3 Quality control for bisulfite-PCR-based methylation analysis 45

Chapter 3. An improved quality control for bisulfite-PCR-based DNA

methylation analysis: cycle threshold value

3.1 BACKGROUND

Methylation of DNA is an epigenetic mechanism of gene regulation involved in both

physiological and pathological cellular processes (Rashid and Issa, 2004). Increasingly,

DNA methylation is being associated with disease risk, prognosis and treatment response,

indicating its detection to have diagnostic potential. This potential has in turn created a

need for accurate quality control systems for methylation analysis.

Of particular value would be a system for estimating the likely analytical reliability of a

sample, and guiding sample loading for efficient sample management. Today, the

majority of methylation analyses are bisulfite PCR-based approaches, such as methylation

specific PCR (MSP) (Herman et al., 1996b) and MethyLight (Eads et al., 2000). In the

absence of a suitable control system, the practice in almost all studies using these methods

has been to load a standard, large quantity (usually 1-2�g) of DNA based on

spectrophotometric measurement.

However, spectrophotometric analysis of DNA is susceptible to variations introduced by

buffer components, pH and UV-absorbing contaminants (Ellison et al., 2006; Haque et

al., 2003). Previously, we tested the hypothesis that cycle threshold (Ct) values in real-

time PCR (QPCR) analyses would be more accurate predictors of PCR reliability than

spectrophotometric concentration, as Ct values should be more direct measures of

amplifiable DNA (Soong and Ladanyi, 2003). Using sets of limiting dilution sample

series, our results showed Ct values indeed have a closer correlation to replicate detection

(qualitative) and Ct value (quantitative) reproducibility (Soong and Ladanyi, 2003).


It follows that Ct values could also be improved controls to spectrophotometric

assessment for bisulfite-PCR methylation analysis. However, bisulfite treatment causes

significant degradation of DNA (Munson et al., 2007), and could introduce factors that

affect PCR kinetics. To our knowledge, no systematic, direct evidence to support the use

of Ct values as controls for methylation analysis exists. This study tests whether the

superiority of Ct values to spectophotometry in predicting PCR detection reliability could

be applied to bisulfite PCR-based methylation analysis.

3.2 METHODS & MATERIALS

To generate a series of samples with varying analytical reliability, undiluted and 1:10

diluted DNA samples from 9 colorectal cancer cell lines (ATCC, Manassas, VA), were

prepared using the PureGene kit (Gentra Systems, Minneapolis, MN). One microliter of

each sample was used to determine DNA concentration based on spectrophotometric

absorbance at 260nm and the 260/280 absorbance ratio (Nanodrop ND-1000, Thermo

Fisher, Wilmington, DE). An equivalent 1µl of each sample was then also analyzed to

determine genomic DNA (gDNA) beta-actin (ACTB) Ct values by real-time PCR on the

ABI 7900HT Real Time PCR System (Applied Biosystems, Foster City, CA). PCR

amplification was performed in a 20µl volume comprising 600nmol/L of each primer,

200nmol/L probe and 1x Taqman Fast Universal PCR Master Mix (Applied Biosystems).

The primers and probe sequences were as follows: forward primer: 5’-

TCAGATCATTGCTCCTCCTG-3’, reverse primer: 5’-CTTGCTGATCCACAT-3’,

probe: 5’-FAM-CATCCTGGCCTCGCTGTCCA-TAMRA- 3’.

A further 1µl of each sample then underwent bisulfite treatment as described by Frommer

et al (Frommer et al., 1992). Briefly, 1µl of DNA was added to 3.1mol/L sodium bisulfite


and 5mmol/L hydroquinone, pH 5.0 and incubated at 50°C for 16 hours. The DNA was

then purified using the Wizard® DNA Clean-Up System (Promega, Madison, WI) and

resuspended in 20µl of 10mmol/L Tris-HCl (pH 8). Bisulfite DNA (bsDNA) Ct values for

ACTB were then determined in quintuplicate from 1µl aliquots using QPCR according to

MethyLight protocol of Eads et al. (Eads et al., 2000).

3.3 RESULTS

The replicate detection frequencies and standard deviations of bsDNA ACTB Ct values,

and gDNA ACTB Ct and spectrophotometric values are displayed in Table 3.1, and

plotted in Figure 3.1. Optimal thresholds for maximizing the accuracy of predicting

qualitative (detection in 5/5 replicates) and quantitative (Ct value standard deviation of

less than 1.0) reliability by spectrophotometry and Ct values were determined by

observation.

Consistent with our previous findings, gDNA Ct values correlated better with detection

reliability than spectrophotometric measurement. Using a threshold gDNA Ct value of

23.50, samples with consistent and inconsistent detection were correctly identified in 6/7

(86%) and 11/11 (100%) cases respectively (Figure 3.1A). Using an optimal threshold of

350ng/µl from spectrophotometric measurement of gDNA, the corresponding frequencies

were 6/8 (75%) and 10/10 (100%) respectively (Figure 3.1B). A major discrepancy was

the detection of two DNA samples with concentrations of 1271ng/�l and 2445ng/�l in 4/5

(80%) replicates. These samples had Ct values of 36.64 and 34.90 respectively and were

thus correctly identified by Ct values as unreliable (Table 3.1). In terms of quantitative

reliability, all 7/7 (100%) samples with Ct values below 23.50 were reliable, and 4/4

(100%) above 23.50 were unreliable (Figure 3.1C). The corresponding frequencies for


spectrophotometry using a threshold of 350ng/µl were lower at 6/8 (75%) and 2/3 (67%)

respectively (Figure 3.1D).

Using a spectrophotometric threshold of 1000ng/ul did not improve reliability prediction

accuracy: only 3/5 (60%) samples above this threshold were detected reliably and these

all had Ct values standard deviation of <1.0 (Table 3.1). Furthermore, the exclusion of

samples with A260/A280 ratios of <1.8 or >2.0 did not improve the reliability predicted

using spectrophotometric assessment.

Table 3.1 Association between detection frequencies, Ct values and their standard

deviations from quintuplicate ACTB analysis and bisulfite-treated DNA (bsDNA) Ct

values, genomic (gDNA) Ct values and spectrophotometric measurement.

ACTB (bsDNA) gDNA Spectrophotometry

Cell Line Dilution Detection aCt bSD Ct ng/ul 260/280 RKO 1:1 5/5 27.55 0.48 18.36 1932.0 1.93 SW620 1:1 5/5 28.39 0.37 20.12 1238.8 2.03 WiDR 1:1 5/5 30.06 0.74 21.01 399.2 1.85 LS174T 1:1 5/5 29.51 0.52 21.14 678.2 1.97 RKO 1:10 5/5 33.25 0.47 21.19 356.1 1.91 DLD1 1:1 5/5 32.38 0.99 23.43 1196.2 2.05 HCT116 1:1 4/5 34.90 1.12 23.65 2445.4 2.13 HT29 1:1 4/5 36.64 1.63 25.76 1271.6 2.07 Colo205 1:1 3/5 34.94 0.28 23.25 203.1 2.04 LS174T 1:10 2/5 36.97 1.99 25.29 61.0 1.87 WiDR 1:10 2/5 34.63 1.81 25.89 36.7 1.72 SW620 1:10 1/5 36.40 ND 23.65 322.5 1.99 HT29 1:10 1/5 38.04 ND 29.26 112.6 2.03 Colo205 1:10 0/5 40.00 ND 27.12 18.9 1.74 HCT116 1:10 0/5 40.00 ND 27.67 233.9 2.13 DLD1 1:10 0/5 40.00 ND 27.71 100.7 2.08 SW480 1:1 0/5 40.00 ND 30.53 0.6 0.32 SW480 1:10 0/5 40.00 ND 34.04 ND 1.27 ND = not determinable a average Ct from 5 replicate analysis b Standard deviation (SD) from detected replicates


0.00

0.50

1.00

1.50

2.00

2.50

15 20 25 30 35

genomic DNA Ct

Ct

Sta

nda

rd D

evia

tion

0

1/5

2/5

3/5

4/5

1

1 1/5

15 20 25 30 35

genomic DNA Ct

0

1/5

2/5

3/5

4/5

1

1 1/5

0 1 2 3 4

log A260 concentration (ng/ul)

B

0.00

0.50

1.00

1.50

2.00

2.50

0 1 2 3 4


Ct

Sta

nd

ard

Dev

iati

on

D

A

C

0/5

Det

ectio

n of

AC

TBby

QP

CR

0/5

5/5D

etec

tion

of A

CTB

by Q

PC

R

0/5

5/5

0.00

0.50

1.00

1.50

2.00

2.50

15 20 25 30 35

genomic DNA Ct

Ct

Sta

nda

rd D

evia

tion

0

1/5

2/5

3/5

4/5

1

1 1/5

15 20 25 30 35

genomic DNA Ct

0

1/5

2/5

3/5

4/5

1

1 1/5

0 1 2 3 4


B

0.00

0.50

1.00

1.50

2.00

2.50

0 1 2 3 4


Ct

Sta

nd

ard

Dev

iati

on

D

A

C

0/5

Det

ectio

n of

AC

TBby

QP

CR

0/5

5/5D

etec

tion

of A

CTB

by Q

PC

R

0/5

5/5

Figure 3.1 Association between genomic DNA Ct values (A, C) and spectrophotometric

concentration (B, D) and detection of bsDNA ACTB by QPCR (A, B) and Ct value

standard deviation (C, D). Samples for which reliability is wrongly predicted by the

respective methods are circled with dashed lines.

To further test the validity of Ct values as a quality control, DNA was extracted from 40

formalin fixed and paraffin-embedded tissue (FFPET) samples of colorectal tumours as

described previously (Soong and Iacopetta, 1997). As before, 1�l of FFPET DNA was

quantified by spectrophotometry and 1�l by QPCR for ACTB gDNA. Another 1�l

underwent bisulfite treatment using the EZ DNA Methylation kit according to the


manufacturer’s instructions (Zymo Research, Orange, CA) and was eluted in 20�l

10mmol/L Tris-HCl (pH 8) and used immediately. Two microliters of the resulting

bsDNA was then analyzed in quintuplicate for MLH1 methylation by MSP with primers

as described previously (Herman et al., 1996):

methylated-sense, 5'-ACGTAGACGTTTTATTAGGGTCGC-3'

methylated-antisense, 5'-CCTCATCGTAACTACCCGCG-3'

unmethylated-sense, 5'-TTTTGATGTAGATGTTTTATTAGGGTTGT-3'

unmethylated antisense, 5'-ACCACCTCATCATAACTACCCACA-3')

FFPET DNA samples were deemed reliable if consistent results (detected in 5/5, or not

detected in 5/5) were observed using methylated (“M”) and unmethylated (“U”) primer

sets. Using this criteria, 29 of the 40 samples were reliable and 11 unreliable for MSP

analysis.

Table 3.2 displays the quintuplicate MSP results, spectrophotometric concentrations and

gDNA Ct values of respective samples. Optimal thresholds for maximally distinguishing

reliable from unreliable samples were determined by sorting for reliability status, Ct

values and spectrophotometric concentration. Corroborating the findings with cell line

DNA, gDNA Ct values were better indicators of sample reliability than

spectrophotometric concentration. All samples with gDNA Ct values below a threshold of

26.00 were reliably detected (25/25, 100%) and 11/15 (73%) above this threshold were

unreliably detected. Using spectrophotometry and a threshold of 350ng/µl, the

corresponding frequencies were 11/14 (79%) and 8/26 (31%). Even with the use of other

thresholds (1000, 400, 300, 200, 100ng/µl), the prediction accuracy did not attain the

level achieved by Ct values (results not shown).


Table 3.2 Analytical reliability of bisulfite-DNA as defined by MSP analysis of

MLH1 and its concordance with assessment by genomic DNA Ct values of ACTB and

spectrophotometry concentrations.

Case U1 U2 U3 U4 U5 M1 M2 M3 M4 M5 bStatus

cObserved Reliability Ct Value

Predicted Reliability ng/ul

Predicted Reliability

3 + + + + + + + + + + P R 22.46 R 360.09 R6 + + + + + + + + + + P R 23.14 R 372.69 R221 + + + + + + + + + + P R 24.72 R 394.30 R17 + + + + + + + + + + P R 25.78 R 509.60 R285 + + + + + + + + + + P R 25.32 R 584.12 R201 + + + + + + + + + + P R 24.99 R 655.99 R5 + + + + + - - - - - U R 24.06 R 414.39 R13 + + + + + - - - - - U R 21.97 R 544.89 R11 + + + + + - - - - - U R 23.37 R 647.30 R9 + + + + + - - - - - U R 22.64 R 1037.59 R16 + + + + + + + + + + P R 25.38 R 70.72 U14 + + + + + + + + + + P R 24.62 R 74.82 U19 + + + + + + + + + + P R 24.45 R 84.30 U57 + + + + + + + + + + P R 25.14 R 127.24 U8 + + + + + + + + + + P R 23.05 R 344.99 U30 + + + + + - - - - - U R 25.65 R 61.55 U31 + + + + + - - - - - U R 25.67 R 92.95 U7 + + + + + - - - - - U R 23.38 R 117.93 U18 + + + + + - - - - - U R 24.29 R 139.52 U4 + + + + + - - - - - U R 25.23 R 187.95 U1 + + + + + - - - - - U R 24.81 R 227.59 U12 + + + + + - - - - - U R 25.47 R 236.02 U15 + + + + + - - - - - U R 22.47 R 246.64 U10 + + + + + - - - - - U R 23.66 R 273.01 U2 + + + + + - - - - - U R 24.68 R 295.62 U273 + + + + + + + + + + P R 27.37 U 599.23 R64 + + + + + - - - - - U R 26.95 U 50.56 U58 + + + + + - - - - - U R 27.51 U 42.24 U111 + + + + + - - - - - U R 27.72 U 269.97 U

67 + + + + + - + + + + - U 26.34 U 131.02 U43 + + + + + + - + + + - U 26.39 U 104.20 U65 + + + + + - - + - - - U 26.70 U 113.46 U66 + + + + + + - - + + - U 27.38 U 87.56 U68 + + + - - + + + + + - U 29.01 U 48.30 U45 + - + - - - - - - - - U 26.28 U 59.96 U55 + - + + - - - - - - - U 26.31 U 41.60 U59 - + - + - + + - + + - U 26.43 U 128.28 U268 + + + + + - + + + - - U 26.80 U 399.91 R110 + + + + + + - + + + - U 31.38 U 618.49 R99 + + + + + + + - + + - U 40.00 U 878.41 Ra Results of 5 PCRs with primers specific to unmethylated (U) and methylated (M) DNA, - = negative, + = positiveb Unmethylated (U), Partially methylated (P), Methylated (M)c Reliable (R), Unreliable in replicate analyses (U)

QPCR of gDNA ACTB SpectrophotometryaMethylation Specific PCR of MLH1

3.4 DISCUSSION

Taken together, the above results indicate that Ct values could be useful indicators for

predicting the reliability of a sample for bisulfite-PCR based methylation analysis. In both

cell lines (Figure 3.1) and FFPET samples (Table 3.1), Ct values had a closer correlation

to the qualitative and quantitative reliability of samples than spectrophotometric

assessment. These correlations were observed using multiple methods of DNA extraction,

bisulfite treatment and methylation analysis, highlighting their robustness.


A key finding was that 18/26 (69%) of samples identified to be unreliable by

spectrophotometry were reliable in MSP analysis, compared to 4/15 (27%) using Ct

values (P<0.05). FFPET samples are a significant resource for methylation analysis

because of their central involvement in diagnostic workflows and the large repositories of

archival tissues. Formalin fixation leads to considerable fragmentation of DNA (Gilbert et

al., 2007); however its effect on methylation analytical processes has not been tested.

Since most standard bisulfite treatment protocols use 1-2µg of DNA measured

spectrophotometrically as starting material (Clark et al., 2006), the current results infer

that a large number of FFPET samples are being unnecessarily disqualified from analysis

or used in excess based on spectrophotometric analysis. This provides a strong case for

the use of Ct values.

Ct values were tested against spectrophotometric concentrations in this study, as they are

the most commonly used measure for controlling sample reliability in current methylation

analyses. Nevertheless, a Ct value system is also likely to have advantages over other

potential control systems. Its measure of PCR-amplifiable DNA should make it more

accurate than measures from fluorescent double-stranded DNA binding dyes, such as

PicoGreen, which are less sensitive to factors such as PCR inhibitors. Recently, Ehrich et

al. (Ehrich et al., 2007) proposed a system based on measuring the amplification success

of replicates of bsDNA PCR amplicons of decreasing length on mass spectrometers. In

comparison, the Ct value system stands to be simpler and faster as it involves a single

gDNA Ct measurement in a 1-2 hour QPCR run. Moreoever, QPCR instrumentation is

common in many laboratories today, making the system more amenable to

standardization. Critically, the Ct value system is based on gDNA compared to bsDNA,

for which a measurement would be too late to enable sample adjustment or sample

exclusion and reagent conservation.


An alternative approach to this study would have been to ascertain reliability thresholds

for both Ct and spectrophotometric approaches, adjust sample loading to these thresholds,

and then evaluate the qualitative and quantitative reliability of the samples. However, this

design was found to be impractical, given the low concentrations and volumes of many of

the samples in the study (Table 3.1, Table 3.2). Moreover, by loading 1 µl for each

control system, the current study design avoids the bias that would likely be introduced by

the irregular sample loading adjustments required.

The experiments in this small study were simply designed to test whether Ct or

spectrophotometric values would be better indicators of reliability for methylation

analysis. In this respect, they have made an important step in providing evidence showing

Ct values to have a better correlation with bisulfite-PCR reliability. The application of the

Ct value system is likely to be amplicon and assay specific. The optimal Ct value

thresholds for reliability differed for the measurement of ACTB by MethyLight (23.50)

and MLH1 bsDNA by MSP (26.00), likely reflecting the differing detection limits and

amplification efficiencies of the respective assays and PCR amplicons. Previously, we

found that Ct value thresholds for PCR reliability could be determined from the values at

which analytical unreliability occurred in replicate limiting dilution series (Soong and

Ladanyi, 2003). The evidence suggested that the thresholds were independent of the

sample used. This suggests a standard procedure for implementing a Ct value control

system could be to ascertain reliability thresholds for a given assay through limiting

dilutions of an abundant sample (eg. cell line DNA), followed by single analyses of test

samples. Based on PCR doubling kinetics and Ct values, the suitability of samples for

analysis could then be quantitatively assessed, and the suitable samples loaded to

maximize sample utility and reliability. The value of a Ct value control system will be

demonstrated by its use in future studies.

Chapter 4 RUNX3 and DPYD methylation 55

Chapter 4. Identification of CpG sites in the RUNX3 and DPYD genes

associated with expression level

4.1 BACKGROUND

DNA methylation of CGIs in gene promoters is an important gene regulatory mechanism

for tissue-specific gene expression during normal development. It is an alternative

mechanism for the disruption of many cellular pathways in complex diseases including

cancers. Most studies that have investigated possible mechanistic links between DNA

methylation and gene expression are limited by the non-quantitative assessment of

methylation at arbitrarily chosen CpG sites. Consequently, real associations between

DNA methylation and the end-point of interest could be masked by the lack of correlation

between gene expression and methylation at the probed CpG sites.

Methylation analysis at non-informative CpG sites can have significant implications, as

exemplified by studies on MLH1 and MGMT methylation in cancer. Hypermethylation-

induced silencing of MLH1 expression occurs when the proximal, but not distal, region of

the MLH1 promoter is methylated, resulting in the MSI+ phenotype in a subset of

sporadic CRC (Capel et al., 2007; Deng et al., 1999). Similarly, the selection of different

promoter regions for analysis of methylation may contribute to the discordant correlation

between MGMT hypermethylation and MGMT RNA expression observed in gliomas

(Brell et al., 2005; Everhard et al., 2009) and CRC (Nagasaka et al., 2008a; Shen et al.,

2005).

An approach is required to identify CpG sites whose methylation is most closely

associated with gene silencing. Validation of this approach would be of great benefit for


the design of methylation assays that assess possible functional effects of this epigenetic

alteration in tumour cells.

Hypermethylation of RUNX3 and DPYD have been implicated in the response and

toxicity to 5-fluorouracil (5-FU) (Ezzeldin et al., 2005; Schwab et al., 2008; Yu et al.,

2006) and in the pathogenesis of CRC (Goel et al., 2004; Ku et al., 2004), respectively.

Discrepant reports for the correlation between DPYD methylation and toxicity to 5-FU

(Schwab et al., 2008; Yu et al., 2006) or DPD defienciency (Ezzeldin et al., 2005) may be

due to the lack of consensus CpG methylation sites that are linked to RNA expression.

Methylation-induced transcriptional silencing and hence inactivation of RUNX3 is a

common phenomenon in CRC cell lines (Goel et al., 2004; Ku et al., 2004). RUNX3

methylation has been identified as one of the most sensitive and specific markers for the

CIMP+ subgroup of CRC (Ogino et al., 2007a; Weisenberger et al., 2006a) and has also

been proposed as an important tumour suppressor gene in gastrointestinal cancers (Ito et

al., 2008). Therefore, identification of the CpG sites within the RUNX3 promoter whose

methylation is most tightly linked to RNA expression would help to determine the

functional significance of RUNX3 methylation in CIMP+ tumours.

The current study used a bioinformatics approach to identify CpG sites in the RUNX3 and

DPYD promoters whose methylation regulates RNA expression in CRC cell lines. The

putative CpG sites whose methylation was linked to the expression of these genes in vitro

were then assessed in primary tumours for a similar link with RNA expression.


4.2 METHODS AND MATERIALS

4.2.1 Sample processing and analyses

CRC cell lines and clinical samples used for the study were obtained and processed as

described in Chapter 2. RNA quantification and DNA methylation analysis using bisulfite

clonal sequencing (bsSEQ), Pyrosequencing and MethyLight were carried out following

the protocols described in Chapter 2. Lengths of sequence 5000bp upstream and 5000bp

downstream of the transcription start site were evaluated for the presence of CGI as

described in Chapter 2. Identified CGIs were then analysed for methylation at individual

CpG sites using bsSEQ in 10 CRC cell lines. Single and multiple CpG loci whose

methylation was associated with RNA level in CRC cell lines were identified and further

evaluated in primary CRC.

4.2.2 Identifying individual CpG sites correlated with RNA expression

To identify individual CpG sites whose methylation regulates RNA expression, bsSEQ

methylation data derived from 10 CRC cell lines was evaluated by ridge regression (RR)

analysis as described by Myers (Myers, 1992). Subsets of candidate CpGs linked to

expression were selected based on ridge parameter, �, or Mallow’s Cp criterion where

appropriate (Whittaker et al., 2000) using bootstrap resampling. The performance of

candidate CpGs was then examined based on the goodness-of-fit of the predicted RNA

expression to the observed RNA expression. Methylation of candidate CpG sites in

RUNX3 and DPYD was assessed in primary CRC using Pyrosequencing assays.

4.2.3 Identifying multiple adjacent CpG sites correlated with RNA expression

The correlation between concurrent methylation at multiple adjacent CpG sites and RNA

level was evaluated using bsSEQ methylation data from 10 CRC cell lines. Simulated

MethyLight assays for multiple adjacent CpG sites were developed using optimal


specifications for each oligonucleotide (Weisenberger et al., 2006a) (Supplementary

Table 4.1). Lengths of the forward primer (F), probe (P) and reverse primer (R) were set

at 21 bp, 24 bp and 21 bp respectively. P was placed 7 bp from the end of F, while R was

positioned 13 bp after the end of P. For each amplicon, the first configuration was placed

at the first available CpG, corresponding to the start of the forward primer. Each assay

targeted a minimum of 5 CpGs in the whole amplicon. Successive configuration was then

started with the next adjacent CpG. Using the bsSEQ methylation data from cell lines, a

clone was classified as being methylated only when all targeted CpGs were methylated

for each of the simulated MethyLight assays. The methylation score for each cell line was

thus the proportion of clones methylated for the particular cell line. Leave-one-out cross

validation was performed on the simulated MethyLight methylation data to identify the

cluster of neighbouring CpG sites most strongly linked to RNA expression.

4.2.4 Correlation between methylation at candidate expression-linked CpG sites

and RNA levels in primary tumors

Pyrosequencing was used to measure the methylation of individual candidate CpG sites,

thus allowing correlation with RNA expression levels in tumours. Only samples with at

least two CpG sites successfully analysed were included in subsequent analysis. CpG sites

that gave readings of 0% in all samples were also excluded from subsequent analysis.

Independent t-test of RNA expression was performed to examine for potential systemic

bias in samples that had non-detectable levels of methylation at candidate CpG sites.

Missing methylation values were imputed using the K-nearest neighbour (KNN) method

as described by Hastie et al (http://cran.r-project.org/) (Hastie et al., 1999). Euclidean

distance was used as the metrics for identification of K closest. Bayesian analysis was

conducted to predict RNA expression based on methylation levels at individual CpG sites

measured using Pyrosequencing in the tumours. Bayesian analysis was carried out using


WinBUGS (Lunn et al., 2000). A predictive score derived from the Bayesian analysis was

then used to classify samples as having above- or below-median RNA expression based

on an optimal threshold, c, determined by a receiver operating characteristics (ROC)

curve.

4.3 RESULTS

4.3.1 Methylation of CpG sites in the RUNX3 and DPYD promoters correlates with

RNA expression in CRC cell lines

Ridge regression (RR) modelling as described in 4.2.2 demonstrated that the level of

overall methylation at all CpG sites was able to predict RNA expression. The RNA levels

predicted from overall methylation correlated very closely with the observed RNA levels

(RUNX3, r2=0.9928; DPYD, r2=0.9999; Figure 4.1).

To identify the combination of individual CpG sites whose methylation was most closely

associated with RNA expression, optimum ridge parameter (�) was determined from RR

analysis. This factor indicates the minimum number of CpG sites needed to predict RNA

expression based on their methylation level. � was determined to be e-2 and e-8 for RUNX3

and DPYD respectively (Figure 4.2), corresponding to methylation at 8 CpG sites for

RUNX3 and 9 CpG sites for DPYD. At these optimal � values, the predicted RNA

expression showed high correlation to the observed RNA expression (RUNX3, r2=0.859;

DPYD, r2=1.000; Figure 4.3). The 8 CpG sites for RUNX3 relative to the transcriptional

start site and ranked according to their contribution to the prediction of RNA expression

were +279, -789, -10, +180, -515, -719, -1035, -826. For DPYD the 9 CpG sites were

+152, -72, -86, +235, +196, -78, -31, +249, -20.


4.3.2 Methylation at multiple adjacent CpG sites is poorly correlated to RNA levels

in CRC cell lines

MSP and MethyLight are based upon the analysis of multiple adjacent CpG sites for

methylation status. Using methods described in 4.2.3, we evaluated methylation levels at

multiple adjacent CpG sites, or clusters, for correlation with RNA expression. The top

ranking cluster of CpG sites for RUNX3 (+132, +141, +148, +150, +154, +166, +170,

+180, +195, +201, +207, +211) was poorly predictive of RNA expression (r2=0.082,

Figure 4.4). The top ranking CpG cluster for DPYD (+10, +12, +17, +56, +84) was more

predictive but still relatively low (r2=0.444, Figure 4.4).

4.3.3 Expression-related CpG sites identified in CRC cell lines failed to predict

RNA expression in primary tumors

Of the 8 and 9 CpG sites identified in CRC cell lines as being linked to RNA expression

of RUNX3 and DPYD, respectively, one CpG site in each gene was excluded from

analysis in the primary tumours due to absence of methylation in all samples. The

methylation status of the remaining 7 and 8 CpG sites was used to predict RNA

expression using Bayesian analysis as described in 4.2.4. Independent t-test of RNA

expression found there was no systemic bias in samples with non-detectable methylation

at these CpG sites. A predictive score (PS) was formulated for each of RUNX3 and DPYD

to calculate RNA expression based on methylation levels at the expression-linked CpG

sites. Using the optimal threshold determined from the ROC, RNA levels in the primary

tumours were predicted with the following accuracy: RUNX3, sensitivity = 57.5%,

specificity = 47.8%; DPYD, sensitivity = 58.3%; specificity = 66.7% (Figure 4.5).

Comparison between methylation levels at CpG sites measured in both the cell lines and

tumours showed generally lower methylation levels of RUNX3 and DPYD in the tumours

than in the cell lines with less variation (Figure 4.6).


-14 -12 -10 -8 -6

3.70

3.75

3.80

3.85

Mallows

Cp

log �

RUNX3 DPYD

Ridge parameter estimated with Mallow’s Cp

Ridge parameter predicted with cross-validation

Mal

low

’s C

p

E(R

SS) v

alid

atio

n

log �-14 -12 -10 -8 -6

3.70

3.75

3.80

3.85

Mallows

Cp

log �

RUNX3 DPYD

Ridge parameter estimated with Mallow’s Cp

Ridge parameter predicted with cross-validation

Mal

low

’s C

p

E(R

SS) v

alid

atio

n

log �

Figure 4.1 Correlation between RNA levels predicted by ridge regression modelling of

methylation data at individual CpG sites and observed RNA expression in 10 CRC cell

lines for RUNX3 and DPYD. RNA levels are presented on a log scale.

Figure 4.2 Optimum ridge parameter, �, for RUNX3 was �=e-2 and DPYD, �=e-2 where

the residual sum of squared errors are at their lowest.

-6 -4 -2 0

-6-4

-20

RUNX3 DPYD

log

RN

A le

vel (

pred

icte

d)

log

RN

A le

vel (

pred

icte

d)

log RNA level (observed) log RNA level (observed)

r2=0.9928 r2=0.9999

-6 -4 -2 0

-6-4

-20

RUNX3 DPYD

log

RN

A le

vel (

pred

icte

d)

log

RN

A le

vel (

pred

icte

d)


r2=0.9928 r2=0.9999



log

RN

A le

vel (

pred

icte

d)

log

RN

A le

vel (

pred

icte

d)

DPYDRUNX3

r2=0.444r2=0.082


log

RN

A le

vel (

pred

icte

d)

log

RN

A le

vel (

pred

icte

d)

DPYDRUNX3

r2=0.444r2=0.082

Model fitted with 8 CpG sites Model fitted with 9 CpG sites

-6 -4 -2 0

-7-6

-5-4

-3-2

-10

log

RN

A le

vel (

pred

icte

d)

log RNA level (observed)

log

RN

A le

vel (

pred

icte

d)


RUNX3 DPYD

r2=0.859 r2=1.000

Model fitted with 8 CpG sites Model fitted with 9 CpG sites

-6 -4 -2 0

-7-6

-5-4

-3-2

-10

log

RN

A le

vel (

pred

icte

d)


log

RN

A le

vel (

pred

icte

d)


RUNX3 DPYD

r2=0.859 r2=1.000

Figure 4.3 Methylation of candidate CpGs sites identified by ridge-regression predicted

RNA levels that showed perfect correlation with observed RNA expression for RUNX3

and DPYD (r2=1.000).

Figure 4.4 Correlation between the observed RNA expression and RNA levels predicted

by methylation levels at multiple adjacent CpG sites for RUNX3 and DPYD in CRC cell

lines.


Area under ROC curve = 0.5056 Area under ROC curve = 0.6489

0.00

0.25

0.50

0.75

1.00

Sen

sitiv

ity

0.00 0.25 0.50 0.75 1.001 - Specificity

0.00

0.25

0.50

0.75

1.00

Sen

sitiv

ity

0.00 0.25 0.50 0.75 1.001 - Specificity

RUNX3 DPYD

Area under ROC curve = 0.5056 Area under ROC curve = 0.6489

0.00

0.25

0.50

0.75

1.00

Sen

sitiv

ity

0.00 0.25 0.50 0.75 1.001 - Specificity

0.00

0.25

0.50

0.75

1.00

Sen

sitiv

ity

0.00 0.25 0.50 0.75 1.001 - Specificity

RUNX3 DPYD

Figure 4.5 Receiver operating characteristic (ROC) curve showing sensitivity and

specificity of predicted RNA expression calculated using methylation predictive score for

RUNX3 and DPYD.


0

20

40

60

80

100

Perc

enta

ge o

f met

hyla

tion

(Pyr

oseq

uenc

ing)

-1035

T CL

-826

T CL

-789

T CL

-719

T CL

-515

T CL

-10

T CL

180

T CL

279

T CL

RUNX3 CpG sites

0

20

40

60

80

100

Perc

enta

ge o

f met

hyla

tion

(Pyr

oseq

uenc

ing)

-1035

T CL

-826

T CL

-789

T CL

-719

T CL

-515

T CL

-10

T CL

180

T CL

279

T CL

RUNX3 CpG sites

0

20

40

60

80

100

Perc

enta

ge o

f met

hyla

tion

(Pyr

oseq

uenc

ing)

-86

T CL

-78

T CL

-72

T CL

-31

T CL

-20

T CL

+152

T CL

+196

T CL

+235

T CL

+249

T CL

DPYD CpG sites

0

20

40

60

80

100

Perc

enta

ge o

f met

hyla

tion

(Pyr

oseq

uenc

ing)

-86

T CL

-78

T CL

-72

T CL

-31

T CL

-20

T CL

+152

T CL

+196

T CL

+235

T CL

+249

T CL

DPYD CpG sites

Figure 4.6 Distribution of methylation levels in cell lines and primary CRC for RUNX3

and DPYD at individual CpG sites. T, primary tumour; CL, CRC cell lines. Line bar

indicates median methylation value.


4.4 DISCUSSION

The present study identified CpG sites in the RUNX3 and DPYD promoters whose

methylation in CRC cell lines correlates with RNA expression. In silico modelling of

DNA methylation was used to identify CpG sites whose methylation was linked to RNA

expression. Although these candidate CpG sites were able to predict RNA expression

with high accuracy in CRC cell lines (Figure 4.3), they failed to predict RNA expression

in primary CRC (Figure 4.5). This may be due to fundamental differences in methylation

patterns between CRC cell lines and primary tumours (Paz et al., 2003; Suter et al.,

2003).

Methylation levels of RUNX3 and DPYD were generally lower in primary tumours

compared to CRC cell lines, particularly for DPYD (Table 4.1). This agrees with previous

reports in CRC cell lines for a number of CpG islands including hMLH1, p16, various

MINT’s and MGMT (Paz et al., 2003; Suter et al., 2003). Paz et al suggested the higher

methylation levels in cell lines may be due to the very high folate concentrations present

in culture media. Cultured cells are most susceptible to DNA methylation changes during

early population doublings (Allegrucci et al., 2007). In the present study, standardised

culture conditions and cell passages were used to try and minimize variation in

methylation. Another possible explanation for the lower overall methylation level of

primary tumours is because of their contamination with non-tumoural stromal cells and

lymphocytes having low methylation. As discussed further below, the poor correlation in

methylation levels obtained using the different techniques for methylation analysis

(bsSEQ and pyrosequencing) could also contribute to the differences observed between

cell lines and primary tumours.


In addition to lower overall methylation levels, primary CRC also showed less variation

in the methylation of individual CpG sites compared to cell lines (Table 4.1,

Supplementary Figure 4.1). This parallels the dichotomous methylation status (all or none

of the adjacent CpG sites methylated) reported previously in primary CRC for DPYD (Yu

et al., 2008). It also agrees with the previously reported interdependence in methylation of

adjacent CpG sites in tumours (Lacey and Ehrlich, 2009). The in silico model developed

in the present study was based on cell lines that exhibited highly variable levels of

methylation across different CpG sites. The limited variation in methylation level of

primary tumours may therefore explain the inability of the candidate panel of individual

CpG sites to predict RNA expression.

One of the limitations of this study was the detection sensitivity of Pyrosequencing for

methylation, estimated at approximately 10% (Reed et al., 2009; Tost et al., 2003). This

may have reduced the sensitivity for detection of low levels of RUNX3 and DPYD

methylation in primary tumours. It is unlikely that imputation of methylation data for

samples that failed to give a result in the Pyrosequencing assay could explain the lack of

correlation between observed and predicted RNA expression in primary tumours. KNN

imputation for 20% of missing data is associated with a 10% reduction in accuracy

(Troyanskaya et al., 2001). The KNN approach is better than other commonly used

methods for the imputation of missing data such as “row average method” or “filling

missing values with zeroes” (Scheel et al., 2005).

In conclusion, the overall methylation level at a panel of individual candidate CpG sites

could accurately predict RNA expression for RUNX3 and DPYD in CRC cell lines. This

study provides a theoretical basis for the objective selection of CpG sites whose

methylation is linked to RNA expression in cancer cell lines. Methylation levels across


multiple adjacent CpG loci failed to correlate with RNA expression in cell lines.

Moreover, the model built upon observations made in CRC cell lines was unable to

predict RNA expression in primary CRC. The discordance between cell lines and primary

tumours is likely to relate to the substantial qualitative and quantitative differences in

methylation patterns between these samples.

Table 4.1 Mean methylation levels of RUNX3 and DPYD at individual candidate CpG sites in cell

lines and primary CRC.

RUNX3 n1 Cell line Mean ± SD CRC mean ± SD P-value2

-1035 82 79.2 ± 32.5 38.0± 14.4 0.002

-826 61 45.8 ± 41.0 55.8 ± 35.6 0.650

-789 77 58.3 ± 33.5 36.4 ± 32.1 0.871

-719 77 72.0 ± 26.7 32.1 ± 31.6 0.479

-515 65 73.0 ± 30.4 5.3 ± 14.2 0.008

-10 84 55.8 ± 23.4 37.6 ± 19.6 0.267

180 75 72.0 ± 39.5 6.5 ± 15.5 0.000

279 72 88.2 ± 31.2 11.5 ± 23.1 0.751

DPYD

-86 70 53.3 ± 28.8 0.6 ± 4.8 0.000

-78 70 26.3 ± 20.2 0.6 ± 4.7 0.000

-72 69 97.0 ± 5.8 0.0 ± 0.0 0.000

-31 79 17.7 ± 24.4 1.3 ± 6.4 0.000

-20 79 12.0 ± 22.0 2.7 ± 12.5 0.013

152 95 5.3 ± 9.9 9.0 ± 12.4 0.259

196 93 59.1 ± 19.1 5.8 ± 8.5 0.000

235 85 13.8 ± 21.2 7.9 ± 12.5 0.057

249 76 9.6 ± 17.6 5.9 ± 8.9 0.045 1Total number of tumours successfully analysed using Pyrosequencing at each CpG site. A total

of 10 CRC cell lines was analysed at all CpG sites using bsSEQ.

2P-value derived from Levene’s test for equality of variances.

Chapter 5 Characterization of methylated subgroups in CRC 68

Chapter 5. Comprehensive profiling of DNA methylation in colorectal

cancer reveals three subgroups with distinct clinicopathological and

molecular features

5.1 BACKGROUND

DNA hypermethylation-induced gene silencing is a common event in many malignancies

and serves as an alternative mechanism to genetic mutation for the loss of tumour

suppressor functions (Esteller, 2008; Issa, 2004). Although the mechanisms that underlie

aberrant DNA methylation in cancer cells remain to be elucidated, current evidence

suggests that it may be an early and possibly even an initiating event in the development

of CRC.

A subset of CRC has been shown to exhibit frequent and concurrent hypermethylation at

specific gene promoters and is referred to as the CpG island methylator phenotype

(CIMP+) (Toyota et al., 1999b). CIMP+ CRC is associated with distinct

clinicopathological and molecular features including proximal tumour location,

preponderance in elderly females, poorly differentiated and mucinous tumour histology,

MSI and frequent BRAF V600E mutation (Barault et al., 2008b; Hawkins et al., 2002;

Ogino et al., 2006b; Ogino et al., 2007b; Samowitz et al., 2005c; Toyota et al., 1999b;

van Rijnsoever et al., 2002; Weisenberger et al., 2006b). CIMP+ CRC often lack the

hallmark genetic alterations in APC, p53 and 18q that characterize the classic adenoma-

carcinoma sequence. Instead, CIMP+ tumours are thought to develop along an alternate

serrated adenoma pathway in which hypermethylation rather than mutation is used to

inactivate tumour suppressor genes (Jass, 2007c).


In an effort to establish CIMP+ CRC as a distinct subgroup of CRC, Laird and colleagues

analysed the methylation of 195 individual gene promoter regions in 295 CRC using the

quantitative MethyLight assay (Weisenberger et al., 2006b). From their results, they

proposed a panel of 5 CpG island methylation markers to standardize the classification of

CIMP+ CRC. However, different groups have continued to use a variety of methylation

markers to define CIMP+ CRC (Ferracin et al., 2008a; Nagasaka et al., 2008b; Ogino et

al., 2007b; Shen et al., 2007b). The lack of consensus markers has led to reports of

several CIMP subgroups according to the frequency of CpG island methylation

(Kawasaki et al., 2008; Nagasaka et al., 2008b; Shen et al., 2007b). The investigators

who originally proposed CIMP recently described two subgroups of CIMP+, termed

CIMP-1 and CIMP-2, that displayed increased frequencies of BRAF and KRAS mutations,

respectively (Shen et al., 2007b). Similarly, Nagasaka et al described two distinct patterns

of gene methylation in CRC that also segregated with BRAF and KRAS mutations

(Nagasaka et al., 2008b; Nagasaka et al., 2004). Using a panel of 8 methylation markers,

Ogino et al identified a CRC subgroup which they termed CIMP-low that was associated

with frequent KRAS mutation, MGMT methylation and occurrence in males (Ogino et al.,

2006c).

Most previous studies of CIMP+ CRC have investigated a relatively small number of

CpG island markers for methylation. The GoldenGate Methylation BeadArray (Illumina,

Inc.) technology provides the opportunity for high-throughput methylation analysis of a

large number of CpG sites. In the present study the GoldenGate Methylation Cancer

Panel I containing 1,505 CpG loci within 807 cancer-related genes was used to study

methylation patterns in 91 unselected CRC. These genes were selected based on their

involvement in cell growth control, differentiation, migration, apoptosis, DNA damage

repair and oxidative metabolism. The GoldenGate technology allowed us to identify three


distinct CRC subgroups according to their methylation pattern which showed distinctive

clinicopathological and molecular characteristics and differed in their frequencies of

BRAF and KRAS mutation.


5.2.1 Tissue samples

Unselected cases of CRC and adjacent normal colonic mucosa were obtained from 91

patients undergoing surgical resection at St John of God Hospital, Subiaco, Western

Australia. All samples were snap-frozen in liquid nitrogen at the time of surgery and

stored at -80oC until use. This set of tumours contains well-annotated clinicopathological

information including age, gender, tumour location, staging, presence of lymphocytic

infiltration and careful pathological assessment of perineural (PNI), lymphovascular

(LVI) and extramural invasion (EMVI). Informed consent was obtained from all patients

and the project was approved by the Human Research Ethics Committee of St John of

God Hospital.

5.2.2 BRAF mutation, KRAS mutation and microsatellite instability

DNA was extracted from approximately 25mg of tissue using standard phenol-chloroform

extraction. Hotspot mutations in BRAF (V600E) and KRAS (codons 12 and 13) were

identified using fluorescent single strand conformation polymorphism (F-SSCP) as

described previously (Li et al., 2006b; Wang et al., 2003). Deletions in the BAT-26

mononucleotide repeat were detected using F-SSCP and this was used to establish MSI+

status (Iacopetta and Grieu, 2000).


5.2.3 MethyLight determination of CIMPW status

Sodium bisulfite modification was performed using the EZ DNA methylation kit

according to the manufacturer’s instructions (Zymo Research, Orange, CA) and eluted

into 20 �l of 10 mmol/L Tris-HCl (pH 8). The required amount of genomic DNA to

ensure reliable evaluation of DNA methylation following bisulfite modification was

determined as described previously (Ang et al., 2008). DNA methylation levels for the

panel of markers (RUNX3, CACNA1G, IGF2, NEUROG1, SOCS1) described by

Weisenberger et al (Weisenberger et al., 2006b) were measured using MethyLight as

described by the authors. The percentage of methylated reference (PMR) was calculated

and normalised against �-actin to account for variability in the amount of input bisulfite-

treated DNA. SssI methylase-treated DNA was used as the methylated standard. A

threshold PMR value of >4 was used to classify loci as methylated or non-methylated. In

the present study, CIMPW refers to the classification of CIMP using the panel of markers

described by Weisenberger et al., whereby CIMPW-high is defined as 3 or more

methylated loci, CIMPW-low as 1 or 2 methylated loci and CIMPW-negative as no

methylated loci.

5.2.4 DNA methylation profiling using Illumina GoldenGate® methylation bead

array

Comprehensive DNA methylation profiling using the Illumina Goldengate Methylation

Arrays® (Illumina, San Diego, CA) was carried out as described by Bibikova et al

(Bibikova and Fan, 2009) on 91 CRC and 28 randomly selected, matched normal colonic

mucosa samples. Briefly, DNA was quantified by real-time PCR and treated with bisulfite

as for the MethyLight assay. Human sperm DNA and Universal methylated DNA

(Chemicon, Temcula, CA) were included in each run as unmethylated and methylated

controls, respectively. The bisulfite-converted DNA was probed at 1,505 individual CpG


loci contained within 807 genes in the GoldenGate Methylation Cancer Panel I according

to the manufacturer’s instructions (Illumina). Hybridised arrays were scanned using the

BeadArray Reader (Illumina). Extraction and normalization of intensity data was

performed using the Beadscan software. To ensure adequate sample quality, only samples

having >75% loci with a detection P-value of <0.05 were included for analysis.

5.2.5 Statistical analysis

The methylation level at each CpG site, or �-value, was defined as the ratio of methylated

allele to the sum of methylated and unmethylated allele and ranged from 0 (completely

unmethylated) to 1 (completely methylated). All statistical analyses were carried out

using �-value as a continuous variable unless specified otherwise. To compare the number

of methylated genes between different tumour subgroups, �-values were binarized using a

cut-off value of 0.297. This threshold was set based on a 5% false discovery rate (FDR)

for the methylated control. A total of 84 CpG sites contained within 39 X-chromosome

genes were excluded from the analysis in order to eliminate gender-specific bias.

Unsupervised and supervised hierarchical clustering analyses were performed with the

heatmap.2 function in the gplots library. The robustness of clustering was assessed using

non-parametric bootstrap resampling analysis. Additional evidence to support the

delineation of clusters was obtained through unsupervised principal component analysis

(PCA). The frequency and level of CpG methylation across different clusters was

compared using a two-sample proportion test based on both binarised and continuous �-

values. The association of clinicopathological and molecular variables with each cluster

was analysed using continuous �-values and the two-sample proportion t-test. All

statistical analyses were performed in R version 2.7.1 (The R Foundation for Statistical


Computing) at 5% significance level unless otherwise stated. Where applicable,

Bonferroni correction was applied to adjust for multiple testing.

5.3 RESULTS

5.3.1 DNA methylation patterns in normal and tumour tissue

Unsupervised hierarchical clustering of DNA methylation data from 1,505 CpG sites in

28 samples of normal colonic mucosa revealed no distinct clusters [Supplementary Figure

5.1]. As expected, the methylation status of 84 CpG sites in 39 genes located in the X-

chromosome was perfectly correlated with gender [Supplementary Figure 5.1]. These

genes were excluded from subsequent analyses. For the 91 tumour samples, three distinct

clusters were observed when methylation data from all 1,505 loci were included in the

analysis [Supplementary Figure 5.2]. The frequencies of CIMPW, KRAS and BRAF

mutations of the corresponding tumours from these 28 patients were 18%, 36% and 18%

respectively. These frequencies were within the ranges of previously reported

frequencies. Although the frequency of MSI+ (25%) in these patients was higher than that

present in 15% of population-based series of CRC, overrepresentation of MSI+ was

unlikely to have altered the identification of CpGs differentially methylated between

normal colonic mucosa and tumours.

A total of 202 CpG sites, corresponding to 132 genes (90 hypermethylated and 42

hypomethylated), were differentially methylated between tumour and normal colonic

mucosa (P<0.001, FDR 5%) [Supplementary Table 5.1]. Unsupervised hierarchical

clustering of methylation data from these 202 tumour-specific markers identified three

major tumour groups (Figure 5.1), referred to here as highly methylated CRC (HM-CRC;

59/91, 65%), moderately methylated CRC (MM-CRC; 13/91, 14%) and lightly

methylated CRC (LM-CRC; 19/91, 21%). The mean methylation level (�-value) of the


202 CpG sites for these groups was 0.617, 0.506 and 0.370, respectively (P<0.001).

Binarization of the methylation readings using a �-value cut-off of 0.297 revealed a

decreasing number of methylated CpG sites for the three groups (167, 136 and 105

respectively; P<0.001).

Although branching of the dendogram suggested the existence of two subgroups within

HM-CRC (Figure 5.1), the mean methylation level and the frequency of methylation

between these groups were not significantly different (P=0.37 and P=0.90 respectively).

Additional evidence for the validity of tumour segregation was obtained through

unsupervised PCA. HM-CRC could be clearly segregated from LM-CRC. HM-CRC and

MM-CRC could also be discriminated from each other, although less distinctly. This is

presumably because of a greater similarity between these two groups [Supplementary

Figure 5.3].

5.3.2 CRC subgroups show distinctive clinicopathological and molecular features

The distribution of clinicopathological and molecular features for 91 CRC in relation to

the methylation pattern obtained from analysis of all 202 differentially methylated CpG

sites is shown in Figure 5.1. The frequencies of KRAS (32%) and BRAF (16%) mutations

and MSI+ (16%) cases in the total cohort are comparable to those previously reported in

large population based studies (Ogino et al.,2009; Samowitz et al., 2005c). Calculation of

associations between these features and the three CRC subgroups are shown in Table 5.1.

Similar to previous reports on CIMP+, the HM-CRCs in this study were significantly

associated with older age, proximal tumour location and BRAF mutation relative to MM-

CRCs and LM-CRCs. HM-CRC was also significantly associated with MSI+ when

compared to MM-CRC, but not LM-CRCs. Two of the 15 MSI+ tumours were observed

in the LM-CRC group and 13 in the HM-CRC group. Interestingly, the two patients with


LM-CRC MSI+ tumours were aged 44 and 60 years, suggesting the underlying cause of

the MSI+ phenotype was germline or somatic mutation of the mismatch repair genes

rather than hMLH1 methylation.

All 16 tumours classified as CIMPW-high using the panel of markers proposed by

Weisenberger et al (>3/5 sites methylated) were contained within the HM-CRC group,

while all 18 tumours classified as CIMPW-low (1/5 or 2/5 sites methylated) segregated

into the HM-CRC or MM-CRC groups. All 15 tumours with BRAF mutation were HM-

CRC. A significantly higher frequency of KRAS mutation was observed in HM-CRC

compared to LM-CRC or MM-CRC. None of the 13 MM-CRCs contained a KRAS

mutation. The presence of extramural vascular invasion (EMVI) was more frequent in

MM-CRC compared to HM-CRC or LM-CRC. The presence of a tumour-infiltrating

lymphocytic response (TILS) was not associated with any of the methylated CRC

subgroups.

5.3.3 Differentially methylated genes in CRC subgroups

Five clusters of CpG loci, termed A to E, were apparent following unsupervised

hierarchical clustering of methylation data for the 202 CpG loci that showed tumour-

specific methylation (Figure 5.1). CpG sites in clusters A and C were more highly

methylated in MM-CRC and HM-CRC compared to LM-CRC, while the converse was

true for the CpG sites in cluster D. CpG sites in cluster B and cluster E showed uniformly

high and low methylation, respectively, in each of the 3 CRC subgroups.

Using published data from studies on human stem cells (Lee et al., 2006), 50% (39/98) of

the genes within clusters A and C were found to be targets for binding by Polycomb

repressive complex 2 (PRC2) components and/or H3K27 trimethylation. In contrast, only


GenderAgeLocationStageLVIEMVIPNITILSCIMPW

MSI

L M H

E

D

C

B

A

CRC subgroups

CpG

clus

ters

BRAFKRAS

E

D

C

B

A

GenderAgeLocationStageLVIEMVIPNITILSCIMPW

MSI

L M H

E

D

C

B

A

CRC subgroups

CpG

clus

ters

BRAFKRAS

E

D

C

B

A

12% (5/41) of the genes within clusters B, D and E were targets (P<0.001). These

observations support previous reports that hypermethylated genes in cancer are frequent

targets of PRC2-mediated H3K27 trimethylation (Widschwendter et al., 2007).

Figure 5.1 Unsupervised hierarchical clustering of 202 tumour-specific probes (rows) in 91 CRC

(columns). The 3 tumour clusters generated by this analysis were termed highly methylated CRC

(HM-CRC), moderately methylated CRC (MM-CRC) and lightly methylated CRC (LM-CRC).

Clinicopathological and molecular features are shown above the heatmap. White rectangles are

cases with missing data. Gender: female (red), male (blue); Age: �67 years (red), <67 (blue);

Tumour location: proximal (red), distal (blue); Tumour stage (ACPS): A or B (blue), C or D (red);

Lymphovascular invasion (LVI): present (red), absent (blue); Extramural vascular invasion

(EMVI): present (red), absent (blue); Perineural invasion (PNI): present (red), absent (blue);

Tumour infiltrating lymphocytes (TILS): present (red), absent (blue); CIMPW: CIMPW-high (red),

CIMPW-low (yellow), CIMPW-negative (blue); BRAF: mutant (red), wildtype (blue); KRAS:

mutant (red), wildtype (blue); Microsatellite instability (MSI): positive (red), negative (blue). Five

CpG clusters (A-E) were apparent from the analysis and showed differential methylation amongst

the 3 CRC subgroups.


Table 5.1 Clinicopathological and molecular characteristics of CRC subgroups

CRC subgroup (n ,%) P

L M H L vs M M vs H L vs H 19 (21) 13 (14) 59 (65) Female 6 (32) 4 (31) 30 (51) Male 13 (68) 9 (69) 29 (49) 0.952 0.211 0.175 Age � 67 years 6 (32) 5 (38) 37 (63) Age < 67 years 13 (68) 8 (62) 22 (37) 0.570 0.003 0.012 Proximal tumour site1 5 (26) 1 (8) 29 (49) Distal tumour site 14 (74) 12 (92) 29 (49) 0.152 <0.001 0.001 ACPS Stage A or B 8 (42) 4 (31) 36 (61) ACPS Stage C or D 11 (58) 9 (69) 23 (39) 0.520 0.025 0.105 LVI Negative 15 (79) 6 (46) 39 (66) LVI Positive 4 (21) 7 (54) 20 (34) 0.049 0.188 0.126 EMVI Negative 19 (100) 8 (62) 52 (88) EMVI Positive 0 (0) 5 (38) 7 (12) 0.005 0.031 0.024 PNI Negative 17 (89) 11 (85) 52 (88) PNI Positive 2 (11) 2 (15) 7 (12) 0.744 0.506 0.708 TILS Negative2 9 (47) 5 (38) 26 (44) TILS Positive 7 (37) 8 (62) 33 (56) 0.326 0.547 0.521 CIMPW – negative3 19 (100) 10 (77) 28 (47) CIMPW – low 0 (0) 3 (23) 15 (25) CIMPW – high 0 (0) 0 (0) 16 (27) 1.000 <0.001 <0.001 MSI+ 2 (11) 0 (0) 13 (18) MSI- 17 (89) 13 (100) 46 (78) 0.125 <0.001 0.221 BRAF mutant 0 (0) 0 (0) 15 (25) BRAF wildtype 19 (100) 13 (100) 44 (75) 1.000 <0.001 <0.001 KRAS mutant 3 (16) 0 (0) 26 (44) KRAS wildtype 16 (84) 13 (100) 33 (56) 0.057 <0.001 0.014 L, LM-CRC; M, MM-CRC; H, HM-CRC; LVI, lymphovascular invasion; EMVI,

extramural vascular invasion; PNI, perineural invasion; TILS, tumour infiltrating

lymphocytes; CIMPW, classification of CIMP using the Weisenberger et al panel,

whereby CIMPW-high is defined as 3 or more methylated loci, CIMPW-low as 1 or 2

methylated loci and CIMPW-negative as no methylated loci. MSI, microsatellite

instability; 1Tumour location was unknown for 1 patient in HM-CRC, 2TILS data

unknown for 3 patients in LM-CRC, 3P value for CIMPW was generated from comparison

between CIMPW-high and CIMPW-low or CIMPW-negative.


5.4 DISCUSSION

The current study is the first to use array-based technology to enable comprehensive

methylation profiling of CRC. A total of 1,505 CpG sites contained within 807 genes

were assessed in 91 consecutive cases of CRC. The GoldenGate® arrays employed here

were recently used to profile methylation in head and neck cancer (Marsit et al., 2009),

renal cancer (McRonald et al., 2009), glioblastoma (Martinez et al., 2009) and

hematological neoplasms (Martin-Subero et al., 2009; O'Riain et al., 2009). The validity

of these arrays for the quantitative assessment of methylation was shown in several

previous studies by comparison with other quantitative methods (Bibikova et al., 2006;

Christensen et al., 2009; Ladd-Acosta et al., 2007). The finding that methylation of CpG

sites in X-linked genes correlated with gender provided further validation [Supplementary

Figure 5.1]. Many of the genes found to be hypermethylated in this study were previously

reported to be methylated in CRC [Supplementary Table 5.2]. Finally, in agreement with

earlier work on cancer (Widschwendter et al., 2007), many of the genes showing de novo

hypermethylation in this study of CRC (cluster A and C genes, Figure 5.1) are known

targets for PRC2 (Lee et al., 2006).

Similar to earlier studies in CRC that evaluated a limited number of methylation markers

(Barault et al., 2008b; Hawkins et al., 2002; Ogino et al., 2006b; Ogino et al., 2007b;

Samowitz et al., 2005c; Toyota et al., 1999b; van Rijnsoever et al., 2002; Weisenberger

et al., 2006b), comprehensive methylation profiling in the present study revealed the

existence of distinct tumour subgroups (Figure. 5.1). The three major subgroups identified

by unsupervised hierarchical clustering were classified as HM-CRC, MM-CRC and LM-

CRC according to the level and frequency of methylation. In agreement with previous

studies, HM-CRCs were associated with older patient age, proximal site and BRAF

mutation (Table 5.1). All 16 tumours identified as CIMPW-high using a proposed


consensus panel of 5 markers were contained within the HM-CRC group, as well as all 15

tumours containing a BRAF mutation. Using small numbers of methylation markers in

unselected CRC, the original studies by Toyota et al reported CIMP+ frequencies of 62%

(Toyota et al., 1999b) and 51% (Toyota et al., 2000) whereas subsequent studies reported

lower frequencies of 15-32% (Hawkins et al., 2002; Ogino et al., 2007b; Samowitz et al.,

2005c; Shen et al., 2007b; van Rijnsoever et al., 2002; Weisenberger et al., 2006b). In

contrast, by investigating a large number of methylation sites and using unsupervised

hierarchical clustering to analyze the results, we observed a relatively high proportion

(65%) of HM-CRCs in the present study. The lack of association of HM-CRC with other

clinicopathological features such as TILS typically associated with CIMPW was unlikely

to be due to inaccuracies in TILS assessment as TILS was significantly associated with

MSI+ (P<0.05), consistent with previous report (Ogino et al., 2006d).

Previous studies have reported inconsistent results for the association between CpG island

methylation and KRAS mutation (Nagasaka et al., 2008b; Nosho et al., 2008; Ogino et al.,

2006c; Samowitz et al., 2005c; Toyota et al., 2000), probably because of the different

methylation markers used in each study. Analysis of a large number of CpG sites in the

present study revealed that HM-CRCs showed a significantly higher KRAS mutation

frequency compared to both MM-CRCs and LM-CRCs (Table 5.1). This result agrees

with some studies (Barault et al., 2008b; Hawkins et al., 2002; Samowitz et al., 2005c;

Toyota et al., 2000) but not others that found an inverse association between KRAS

mutation and CIMP+ (Ogino et al., 2007b; van Rijnsoever et al., 2002; Weisenberger et

al., 2006b).

The absence of MSI and KRAS and BRAF mutations in the 13 MM-CRCs suggests this

subgroup may have a distinctive molecular and clinical phenotype. In support of this,


MM-CRCs showed a significantly higher frequency of EMVI compared to both LM-

CRCs and HM-CRCs and a significantly higher stage compared to HM-CRC (Table 5.1).

Moreover, almost all MM-CRCs were located in the distal colon or rectum (12/13, 92%).

The existence of a distinctive MM-CRC subgroup will however require confirmation in

independent tumour series.

After adjustment for multiple testing, 170 CpG sites were hypermethylated in HM-CRC

compared to LM-CRC. The 112 genes containing these CpG sites are ranked according to

significance in Supplementary Table 5.2. Of these, 54 were previously reported as

methylated in cancer, 38 as methylated in gastrointestinal cancers and 30 in CRC

[Supplementary Table 5.2]. Of the top 10 genes that were hypermethylated in HM-CRC

compared to LM-CRCs, 5 have previously been implicated in the pathogenesis of

gastrointestinal tumours (NTRK3, HS3ST2, TWIST1, CD40 and EYA4). Somatic mutation

of NTRK3 has been reported in human colon cancer (Wood et al., 2006), while

methylation of EYA4 has been documented previously in ulcerative colitis-associated

dysplasia (Osborn et al., 2006) and CRC (Schatz et al., 2006).

MM-CRCs were found to have a relatively high incidence of EMVI (38%) compared to

HM-CRCs and LM-CRCs (Table 5.1). Supervised analysis revealed that HS3ST2, also

known as 3-OST-2, was the only gene to be differentially methylated between tumours

showing presence or absence of EMVI. Methylation-associated silencing of HS3ST2

expression has been demonstrated in breast, lung, pancreatic and colon cancers

(Miyamoto et al., 2003). This gene encodes an enzyme that modifies heparin sulfate

proteoglycans (Shworak et al., 1999) involved in cell adhesion and migration (Perrimon

and Bernfield, 2000), thus suggesting a possible mechanistic link between HS3ST2

methylation and EMVI.


The use of Illumina GoldenGate® Beadarray technology in this study allowed a large

number of CpG sites to be evaluated for methylation in an unbiased fashion. However,

there are several limitations with this approach for the characterization of methylated

subgroups in CRC. Firstly, only a small fraction of all genes were investigated for

methylation and in 70% of these just one CpG sites per gene was evaluated. Secondly, it

is unclear whether the methylation level at these sites relates to expression of the genes.

Thirdly, some of the probes used in this assay contain single nucleotide polymorphisms

(SNPs) or repetitive elements that could influence methylation analysis (Byun et al.,

2009).

Methylation profiling of 807 cancer-related genes revealed the presence of three CRC

subgroups with distinct clinicopathological and molecular features. Similar to earlier

studies that investigated fewer methylation markers, HM-CRCs were associated with

older patient age, proximal location and mutations in BRAF and KRAS.

Chapter 6 BRAF mutation in young CRC 82

Chapter 6. BRAF mutation is associated with the CpG island

methylator phenotype in colorectal cancer from young patients

6.1 BACKGROUND

The CpG island methylator phenotype (CIMP+) in colorectal cancer (CRC) is

characterised by high levels of CpG methylation in multiple gene promoter regions.

Distinctive clinicopathological features of CIMP+ CRC include origin in the proximal

colon, more advanced age, female gender, mucinous histology and poor differentiation

(Hawkins et al., 2002a; Jass, 2007a; Samowitz et al., 2005a; Teodoridis et al., 2008; van

Rijnsoever et al., 2002). A panel of 5 markers was recently proposed to allow

standardized assessment of CIMP+ (Weisenberger et al., 2006a). This consists of CpG-

rich regions within 5 genes: RUNX3, CACNA1G, IGF2, NEUROG1 and SOCS1. Using

these markers, the frequency of CIMP+ in unselected series of CRC was estimated at

approximately 15-18% (Kambara et al., 2004; Ogino et al., 2006a; Ogino et al., 2007a;

Weisenberger et al., 2006a). Quantitative evaluation of methylation has also been

recommended for the study of CIMP+ (Ogino et al., 2006a), with MethyLight being the

most commonly used method. One of the distinguishing molecular features of CIMP+

tumours was a high incidence of mutation in the BRAF oncogene, particularly the V600E

hotspot mutation (Kambara et al., 2004; Ogino et al., 2006a; Ogino et al., 2007a;

Samowitz et al., 2005a; Weisenberger et al., 2006a). Another feature recently reported for

CIMP+ was an inverse correlation with APC methylation in an unselected CRC cohort

(Iacopetta et al., 2006).

Although CIMP+ CRCs share many clinicopathological features with the microsatellite

instability phenotype (MSI+), recent work has confirmed they comprise a distinct


subgroup independently of MSI status (Ogino et al., 2006a; Ogino et al., 2007a;

Samowitz et al., 2005a; Weisenberger et al., 2006a). Approximately half of all CIMP+

tumours are also MSI+ due to methylation-induced transcriptional silencing of the

hMLH1 mismatch repair gene (Ogino et al., 2006a; Ogino et al., 2007a; Weisenberger et

al., 2006a). Compared to CIMP+/MSI+ tumours, CIMP+/MSI- tumours display higher

stage at presentation, absence of lymphocytic infiltration and poor prognosis (Jass,

2007a). Recent reports also suggest the BRAF V600E mutation is associated with poor

prognosis in CIMP+/MSI- tumours (Ferracin et al., 2008b; Lee et al., 2008; Samowitz et

al., 2005b). These studies were carried out on CIMP+/MSI- tumours originating

predominantly in older patients and little is known of the frequency or the

clinicopathological and molecular characteristics of such tumours in younger patients.

The aim of the present study was therefore to investigate BRAF mutation and CIMP+ in

MSI- tumours from an exclusively younger cohort of sporadic CRC patients aged <60

yrs.


The 735 CRC samples investigated here were derived from a population-based screening

study of hereditary non-polyposis colorectal cancer (HNPCC) in patients aged <60 yrs at

diagnosis (Watson et al., 2007). Only tumours judged to be MSI- following analysis of

the BAT-26 mononucleotide repeat were included in this study, thereby excluding

HNPCC cases. Information on patient age and gender and on tumour characteristics

including stage, site and grade were obtained from pathology reports. BRAF V600E

mutations were identified by fluorescent-single strand conformation polymorphism as

described previously (Li et al., 2006a).


A randomly selected subset of 32 tumours with BRAF mutation was matched for patient

age and tumour site with 57 tumours having BRAF wildtype status. DNA was extracted

from formalin-fixed, paraffin embedded tissues as described previously (Soong and

Iacopetta, 1997). The minimum starting amount of genomic DNA to ensure reliable

evaluation of DNA methylation following bisulfite modification was estimated by real-

time quantification of amplifiable genomic DNA (Ang et al., 2008). Sodium bisulfite

modification was performed using the EZ DNA methylation kit (Zymo Research, Orange,

CA) and eluted into 20�l of 10mmol/L Tris-HCl (pH 8). DNA methylation levels for

RUNX3, CACNA1G, IGF2, NEUROG1, SOCS1 and APC were quantified using

MethyLight as described previously (Iacopetta et al., 2006; Weisenberger et al., 2006a).

The level of DNA methylation was calculated as a percentage of methylated reference

(PMR) using �-actin to normalize for the amount of input bisulfite-treated DNA and SssI

methylase-treated DNA as the methylated standard. A threshold PMR value of 4 was used

to classify methylated loci (Iacopetta et al., 2006; Ogino et al., 2007a; Weisenberger et

al., 2006a). Positive CIMP status (CIMP+) was defined as �3 of the RUNX3, CACNA1G,

IGF2, NEUROG1 and SOCS1 markers being methylated.

The �2 test was used to assess univariate relationships between the BRAF V600E

mutation and categorical variables including clinicopathological characteristics and

methylation at each locus. Fisher’s exact test was used when the sample size was less than

5. All P values were two-sided and the level of statistical significance was <0.05.

Student’s t-test was used for comparisons of the mean number of methylated loci between

the BRAF wildtype and BRAF mutant tumours.


6.3 RESULTS

The frequency of BRAF V600E mutation in 735 microsatellite stable (MSI-) CRCs from

patients aged <60 yrs was 7% (Table 6.1). The presence of BRAF mutation was strongly

associated with advanced stage, location in the proximal colon and high histological

grade. A trend was also apparent for higher BRAF mutation frequency in tumours from

very young patients (<40 yrs). Similar result was also observed when BRAF mutation

frequency was analysed using age as a continuous variable (P=0.0017).

The frequencies of gene hypermethylation and CIMP+ were investigated in 32 BRAF

mutant and 57 BRAF wildtype tumours that were matched for patient age and tumour site.

The median age of these patient groups was 52.0 and 51.5 years, respectively. A

significantly higher frequency of methylation in each of the markers comprising the

CIMP panel was observed in BRAF mutant compared to BRAF wildtype tumours (Table

6.2). Approximately half of the BRAF mutant tumours identified in this young cohort of

MSI- CRC were CIMP+, compared to just 4% for the BRAF wildtype tumours. Figure 6.1

shows the distribution of the number of methylated CIMP markers for BRAF wildtype

and BRAF mutant tumours. The mean number of methylated loci was higher in BRAF

mutant (2.4 ± 1.54) compared to BRAF wildtype (0.3 ± 0.79) tumours (P<0.0001, Student

t-test).

In contrast to the CIMP markers, a significant inverse correlation was observed between

the presence of APC methylation and BRAF mutation (Table 6.2). All 8 tumours with

APC hypermethylation were wildtype for BRAF and were amongst the 45 tumours that

showed no methylation of any of the CIMP markers (Figure 6.1).


Table 6.1 Associations between BRAF V600E mutation and clinicopathological features of

microsatellite stable CRC from patients aged <60 years.

Feature (n) a BRAF wildtype (%)

BRAF mutant (%)

P

Total (735) 685 50 Age (yrs) <30 (13) 10 (1) 3 (6) 30-39 (42) 38 (6) 4 (8) 40-49 (168) 159 (23) 9 (18) 50-59 (512) 478 (70) 34 (68) 0.07 b Sex Female (295) 272 (40) 23 (46) Male (440) 413 (60) 27 (54) 0.38 Stage (AJCC) In situ (38) 36 (6) 2 (5) I (116) 113 (20) 3 (8) II (115) 151 (27) 4 (11) III (223) 205 (37) 18 (49) IV (60) 50 (9) 10 (27) 0.0004 c Site Proximal (192) 163 (24) 29 (59) Distal (531) 511 (76) 20 (41) <0.0001 Grade Low (561) 536 (88) 25 (58) High (89) 71 (12) 18 (42) <0.0001 Mucinous histology Absent (599) 564 (82) 35 (70) Present (136) 121 (18) 15 (30) 0.03

a Data for stage, site, grade was not available for 183, 30 and 103 cases, respectively.

b Age <40 yrs vs 40-59 yrs.

c Stages III/IV vs in situ/I/II.


Table 6.2 Associations between gene hypermethylation and BRAF V600E mutation in

microsatellite stable CRC from patients aged <60 yrs.

Gene methylation (n) a

BRAF wildtype (%)

BRAF mutant (%)

P

Total 57 32 RUNX3 Yes (22) 2 (4) 20 (62) No (67) 55 (96) 12 (38) <0.0001 CACNA1G Yes (14) 3 (5) 11 (34) No (75) 54 (95) 21 (66) 0.0005 IGF2 Yes (23) 3 (5) 20 (61) No (66) 54 (95) 13 (39) <0.0001 NEU Yes (29) 9 (16) 20 (62) No (60) 48 (84) 12 (38) <0.0001 SOCS Yes (7) 1 (2) 6 (19) No (82) 56 (98) 26 (81) 0.007 CIMP b Positive (19) 2 (4) 17 (53) Negative (70) 55 (96) 15 (47) <0.0001 APC Yes (8) 8 (14) 0 (0) No (81) 49 (86) 32 (100) 0.02

a Yes, PMR �4; No, PMR <4

b CIMP+ was defined as 3 or more of RUNX3, CACNA1G, IGF2, NEU and SOCS

showing methylation.


Figure 6.1. Distribution of number of methylated CIMP markers for BRAF wildtype and

BRAF mutant tumours.

6.4 DISCUSSION

Approximately 20-25% of CRC cases diagnosed in Western countries occur in patients

aged <60 yrs (Morris et al., 2007). The pathways by which these tumours develop are of

special interest because the large majority of HNPCC cases arise in young patients. The

MSI+ phenotype occurs in about 8% of CRC from patients aged <60 yrs (Watson et al.,

2007) and many of these cases are associated with HNPCC. To investigate the pathways

by which sporadic CRCs develop in younger patients, the present study investigated only

MSI- tumours from a large and consecutive series of patients aged <60 yrs. We were

specifically interested in the associations between BRAF mutation, CIMP+ and APC

methylation in these tumours.

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5

No. of methylated CIMP+ markers

Per

cent

age

of tu

mor

s

BRAF mutant

BRAF wildtype

No. of methylated CIMP+ markersNo. of methylated CIMP+ markers

Per

cent

age

of tu

mor

s

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5


Per

cent

age

of tu

mor

s

BRAF mutant

BRAF wildtype

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5


Per

cent

age

of tu

mor

s

BRAF mutant

BRAF wildtype

No. of methylated CIMP+ markersNo. of methylated CIMP+ markers

Per

cent

age

of tu

mor

s


The frequency of BRAF mutation in this young patient cohort (7%) was within the range

of previous reports of 3-9% in other population-based studies of unselected cases

(Ferracin et al., 2008b; Lee et al., 2008; Li et al., 2006a; Oliveira et al., 2007; Samowitz

et al., 2005b). BRAF mutation was approximately 4�fold more frequent in tumours from

the proximal colon (29/192, 15%) compared to distal colon and rectal tumours (20/531,

4%; P<0.0001). In agreement with other studies, BRAF mutation was strongly associated

with the poor prognosis features of advanced stage and high grade (Ferracin et al., 2008b;

Lee et al., 2008; Oliveira et al., 2007; Samowitz et al., 2005b). Of the 37 BRAF mutant

tumours in which stage could be ascertained from the pathology report, 28 (76%) were

either stage III or IV. It is presently unclear why mutation of this oncogene is associated

with such an aggressive tumour phenotype. The proportion of proximal BRAF mutant

tumours with advanced stage was similar to that of distal BRAF mutant tumours, although

this analysis was limited by the relatively small sample size. Similar to the findings in

MSI- tumours by Samowitz et al (Samowitz et al., 2005b), no association was observed

between gender and CIMP+ in this cohort.

Previous workers have demonstrated a strong association between BRAF mutation and

CIMP+ (Nagasaka et al., 2004; Ogino et al., 2006a; Ogino et al., 2007a; Tanaka et al.,

2006; Weisenberger et al., 2006a) and indeed the former was used to help select markers

to standardize the classification of CIMP+ (Weisenberger et al., 2006a). The earlier

studies linking BRAF mutation to promoter methylation of multiple cancer-related genes

were carried out mostly on older subjects. The present study is the first to specifically

investigate a younger patient cohort and in the absence of possible confounding effects

from the MSI+ phenotype. A subset of 32 BRAF mutant and 57 BRAF wildtype tumours

matched for patient age and anatomical site were selected for the study of CIMP+ using

the quantitative MethyLight assay and a standardized panel of markers. All markers with


the exception of NEU showed a low frequency of methylation (5%) in BRAF wildtype

tumours, but significantly higher frequencies in BRAF mutant tumours (Table 6.2). Using

the criterion of �3 methylated markers to define CIMP+, just over half (53%) of the

BRAF mutant tumours were classified as CIMP+ compared to only 4% of the BRAF

wildtype tumours. These results provide strong support for the notion that BRAF mutation

and CIMP+ occur in the same pathway of CRC development, arising predominantly in

the proximal colon. Although the proportion of stage III and IV tumours was higher in

BRAF mutant (85%) compared to wildtype (58%) tumours, it is unlikely to account for

the large difference observed in CIMP+ frequency between the two groups (Table 6.2).

This is because previous studies have demonstrated only a weak association between

methylation and tumour stage (Hawkins et al., 2002a; Samowitz et al., 2005a; van

Rijnsoever et al., 2002).

The overall frequency of CIMP+ tumours in young CRC patients (<60 years) was

estimated for the first time in this study to be approximately 8%. This value comprises of

4% derived from patients with wildtype BRAF who were CIMP+, plus an additional 4%

derived from patients with BRAF mutation (7% frequency) who were also CIMP+ (53%).

This estimate is about half the frequency reported for CIMP+ in unselected CRC series

(Iacopetta et al., 2007; Ogino et al., 2007a; Weisenberger et al., 2006a). The occurrence

of CIMP+ CRC in young patients highlights the need to identify dietary and genetic risk

factors that are specific for this tumour subgroup, thus allowing targeted screening.

It was not possible to determine from the current results whether BRAF mutation precedes

CIMP+ or vice versa. Co-segregation of BRAF mutations with extensive DNA

methylation has been reported in serrated adenomas found mostly in the proximal colon

and which have been proposed as precursor lesions for CIMP+ tumours (Kambara et al.,


2004; Spring et al., 2006; Yang et al., 2004). The presence of BRAF mutation in serrated

adenomas and CRC has been associated with a positive family history of this disease

(Samowitz et al., 2005b; Spring et al., 2006). By extension, this may also imply genetic

predisposition to the development of CIMP+, possibly via the level of DNA methylation

in normal colonic mucosa (Kawakami et al., 2006).

In striking contrast to CIMP+ determined by the new panel of markers, methylation of

APC was inversely correlated with BRAF mutation (Table 6.2). This confirms a previous

observation in an unselected CRC series (Iacopetta et al., 2006). Moreover, Samowitz et

al have reported that mutations in APC are inversely correlated with the presence of

BRAF mutation and CIMP+ in CRC (Samowitz et al., 2007). Together, these results

suggest that inactivation of the APC gene, either by methylation-induced silencing or by

mutation, occur in a different pathway of CRC development to that which involves BRAF

mutation and CIMP+ as defined by the new markers. The mechanism underlying the

differential methylation of genes between these two tumour subtypes is unknown. Folate

depletion has been shown to alter the level of APC expression in a normal human colon

cell line (Crott et al., 2008), although it is unclear whether this is due to changes in the

methylation status of APC. Hypermethylation of APC may be inversely associated with

cellular folate status in comparison to methylation of the CIMP+ group of genes.

In conclusion, the present results demonstrate that BRAF mutation is rare in MSI- CRC

from young patients, however when present it is associated with an aggressive tumour

phenotype. Approximately half of all tumours with BRAF mutation are also CIMP+,

compared to only a very low frequency of CIMP+ in tumours with wildtype BRAF. The

inverse correlation observed here between APC methylation and BRAF mutation or

CIMP+ provides further support for the existence of alternate pathways of CRC


development. These findings demonstrate that the younger CRC population (<60 yrs) is

also susceptible to the CIMP+ pathway of colorectal tumourigenesis, albeit at

approximately half the frequency observed in older patients.

Chapter 7 General discussion 93

Chapter 7. GENERAL DISCUSSION

7.1 Contribution of this work to the understanding of DNA methylation in CRC

Most investigations of DNA methylation are based on bisulfite-PCR based methods.

However quality control of DNA methylation analysis is lacking. Using cycle threshold

(Ct) values, this study has demonstrated that Ct values derived from real-time PCR

analysis are better than conventional spectrophotometric analysis for determining the

reliability of DNA samples for methylation analysis (Chapter 3; Ang et al., 2008).

Although CpG island methylation is often evaluated as a surrogate marker for gene

silencing, simultaneous assessment of RNA or protein expression level is rarely

performed. In the current work, a ridge-regression based algorithm was developed to

predict RNA expression levels based on DNA methylation at individual CpG sites

(Chapter 4). This model showed that methylation of individual CpG sites identified from

the algorithm correlated with RNA expression in CRC cell lines but not in primary

tumours. Widespread and intense gene promoter methylation is the hallmark

characteristic of CIMP+ CRC. Using unbiased and comprehensive DNA methylation

profiling, the current study identified three CIMP subgroups (CIMP-L, CIMP-M and

CIMP-H) with distinctive clinicopathological and molecular features in CRC (Chapter 5;

Ang et al., submitted). CIMP-H CRCs were associated with KRAS and BRAF mutations

and with older patient age. Although the large majority of CIMP-H (CIMP+) CRCs occur

in older patients, approximately 8% of CRC from younger patients (<60 years) were also

observed to be CIMP+ in this study (Chapter 6; (Ang et al., 2009). Almost all CIMP+

tumours in young patients harbored a BRAF mutation (89%), suggesting these two

important molecular alterations are involved in the development of a minority of early

onset CRC.


7.2 DNA quality and methylation analysis

The availability of molecular techniques to evaluate DNA methylation in cancer-related

genes has prompted the development of potential clinical testing of CpG methylation for

cancer detection, prognosis and monitoring using cell-free biological samples (Kristensen

and Hansen, 2009; Sepulveda et al., 2009). The growing importance of methylation-based

clinical tests highlights the need for robust quality controls and for assay validation, as

dealt with in Chapters 3 and 4 respectively.

Most assays for methylation analysis are based on sodium bisulfite treatment followed by

PCR (Esteller, 2007a; Kristensen and Hansen, 2009). The bisulfite-conversion reaction is

the major source of variability in DNA methylation analysis (Genereux et al., 2008).

Optimal conditions for the bisulfite reaction achieve the complete conversion of

unmethylated cytosine to uracil while minimizing DNA degradation. Excessive DNA

degradation reduces the number of DNA molecules available for PCR amplification, thus

potentially giving rise to PCR amplification bias. Less aggressive methods risk

incomplete conversion of unmethylated cytosine, therefore leading to overestimation of

the methylation levels. To overcome these problems, measurement of genomic DNA to

give the Ct value can be used to gauge sample suitability for methylation analysis

(Chapter 3; (Ang et al., 2008). The Ct value establishes the appropriate amount of initial

genomic DNA required for bisulfite conversion to ensure reproducible methylation

analysis. The Ct method is particularly useful for analytical methods such as MSP that

lack built-in controls.

The Ct quality control system is amenable to various downstream methylation analysis

techniques and is ideally suited to applications involving poor quality or low amounts of

DNA. As shown by the results in Figure 3.1, the use of spectrophotometry to determine


the amount of DNA required for bisulfite conversion can result in the excessive use of

what is often a precious resource, particularly for clinical samples. Spectrophotometry can

also result in the exclusion of samples that may otherwise be found suitable for

methylation analysis by the use of Ct values. The Ct quality control assay therefore

facilitates the optimal use of various DNA methylation assays, such as those used in

Chapter 4 (pyrosequencing, bsSEQ), Chapter 5 (GoldenGate methylation array) and

Chapter 6 (MethyLight). It allows for failure to detect methylation because of inadequate

sample quality to be distinguished from the apparent absence of methylation.

7.3 Identification of CpG loci whose methylation status correlates with gene

expression

The choice of method for DNA methylation analysis is dependent on the goals of testing

and the required assay performance (Esteller, 2007a; Kristensen and Hansen, 2009). This

may be influenced by several factors including the number of individual CpG sites or

CpG islands to be analysed for each sample, the anticipated amount of target DNA within

samples and the quality of DNA. In order to develop DNA methylation assays as

surrogate markers for gene transcript levels, the assays must evaluate CpG sites that have

been shown to correlate with gene expression. The assay should be validated by

correlating methylation data with protein or RNA expression levels. Most assays reported

in the literature were designed to assess methylation at CpG sites that were originally

selected on the basis of optimal primer design. In many cases the functional significance

of methylation at these CpG loci remains to be determined.

In this study, the modeling of DNA methylation status at individual CpG sites allowed the

identification of panels of CpG loci whose methylation correlated with RNA expression

in CRC cell lines (Chapter 4). However, methylation of these CpG panels failed to predict


RNA expression for RUNX3 and DPYD in primary CRC. The inability to use methylation

status at these loci to predict RNA expression may be attributed to differences in the

underlying causal mechanisms and patterns of methylation between cancer cell lines and

primary tumours. In accordance with previous reports (Paz et al., 2003; Suter et al.,

2003), promoter methylation levels in CRC cell lines were generally higher than in

primary CRC (Table 4.1). This could be due to the abundance of folate in culture media

giving rise to a higher concentration of the methyl donor SAM (Figure 1.2). The

acquisition during cell culture of new mutations in genes involved in methyl group

metabolism could also account for differences in methylation between cell lines and

primary tumours.

Technical issues associated with the methylation assays employed in Chapter 4 may also

account for the inability to use methylation status at defined CpG panels to predict RNA

expression in primary tumours. For example, discordance in methylation results obtained

with bsSEQ and Pyrosequencing has been reported (Reed et al., 2009). Although both

assays generate methylation data with single base resolution, the cloning bias associated

with bsSEQ may introduce a degree of error. This problem can be circumvented by digital

bisulfite-sequencing (Weisenberger et al., 2008) whereby the dilution of samples to a

critical level allows PCR reactions starting with a single DNA template. RUNX3 and

DPYD have been reported to be methylated in 21% (Goel et al., 2004) and 13% (Yu et

al., 2008) of primary CRC, respectively. These workers used the MSP and

Pyrosequencing methods, respectively, but targeted different CpG sites to those in the

present study. The low sensitivity of Pyrosequencing (10%) (Tost et al., 2003) could

potentially preclude the detection of low methylation levels in the primary tumours.


The results from Chapter 4 show that methylation data derived from cell lines cannot be

used to predict RNA expression in primary tumours. The current study suggests that

modeling of methylation status in primary tumours is needed to identify panels of CpG

sites whose methylation correlates with RNA expression. Analysis of methylation at

single CpG resolution is required, preferably using digital bisulfite-sequencing. In

addition, the non-tumour cell content of each sample would need to be carefully

evaluated. Laser capture microdissection would likely be required to maximize tumour

cell content from clinical samples.

7.4 Comprehensive DNA methylation profiling to define CIMP

The CIMP+ subgroup of CRC is characterized by frequent hypermethylation of gene

promoter regions (Toyota et al., 1999a). Previous studies have used the MSP, COBRA,

pyrosequencing and MethyLight assays together with selected methylation markers to

investigate CIMP CRC. The present study is the first to use array-based methylation

profiling to characterize CIMP CRC (Chapter 5; Ang et al, submitted).

Using various candidate panels of methylation markers to define CIMP+ in CRC,

previous studies have reported frequencies ranging from 15%-62% (Barault et al., 2008a;

Hawkins et al., 2002a; Ogino et al., 2007a; Samowitz et al., 2005a; Toyota et al., 1999a;

Toyota et al., 2000; van Rijnsoever et al., 2002; Weisenberger et al., 2006a). In

comparison and based on unsupervised hierarchical clustering of methylation data from a

large number of CpG sites, this study found a higher frequency of CIMP+, or CIMP-H,

CRC (65%, Table 5.1). One of the major differences between the current work and

previous studies was the inclusion of hypomethylated CpG sites in tumours relative to

normal mucosa for the classification of CIMP. In the heatmap of methylation data shown

in Figure 5.1, the majority of these demethylated sites were found within CpG clusters D


and E. Interestingly, CpG sites contained within cluster D were mostly demethylated in

CIMP-H tumours but remained methylated in CIMP-L tumours relative to normal colonic

mucosa. This indicates that a small proportion of CpG sites undergo demethylation rather

than hypermethylation in CIMP-H tumours. In contrast to hypermethylated CpG sites, the

demethylated sites were mostly located outside of CpG islands and were not targets for

polycomb proteins.

Mutations in both KRAS and BRAF oncogenes were found in this study to segregate

strongly with CIMP-H CRC (Table 5.1). In contrast, CIMP-M tumours contained no

KRAS or BRAF mutations, while CIMP-L tumours contained no BRAF mutations and a

low 16% frequency of KRAS mutations. The association of BRAF V600E mutation with

CIMP+ CRC has been confirmed in multiple studies (Li et al., 2006a; Ogino et al.,

2007a; Weisenberger et al., 2006a), while the association of KRAS mutations with CIMP+

has been inconsistent (Barault et al., 2008a; Nagasaka et al., 2008a; Ogino et al., 2007a;

Shen et al., 2007b; van Rijnsoever et al., 2002; Weisenberger et al., 2006a). Although the

BRAF V600E mutation is often observed to be tightly associated with CIMP+ CRC, the

role of BRAF mutation in the pathogenesis of CIMP+ is unclear. In normal colon

epithelial cells, induced expression of BRAF V600E has been shown to lead to

hypermethylation of MLH1 during cellular transformation (Minoo et al., 2007). Similarly,

fibroblast cells transformed by RAS have been shown to up-regulate DNMT1 leading to

increased promoter methylation (Ordway et al., 2004).

The small subgroup of CIMP-M tumours identified in the current study (14%) showed

intermediate levels of CpG methylation relative to CIMP-L and CIMP-H. The complete

absence of KRAS and BRAF mutations, predilection for the distal colon and high

frequency of EMVI (Table 5.1) suggests that CIMP-M tumours may represent a clinically


distinctive subgroup of CRC. However, since CIMP-M was a relatively small subgroup,

further investigation is required to confirm these observations. The short follow-up time

of the CRC patient cohort in this study meant the prognostic significance of the three

CIMP subgroups could not be evaluated.

Available evidence suggests that mitotically heritable DNA methylation in normal cells

functions to determine cell fate (Reik, 2007; Shen et al., 2007a). Genes targeted for

hypermethylation in stem cells and in cancer cells show frequent association with PcG

proteins and a propensity for bivalent modification of associated histone proteins

(McCabe et al., 2009b; Ohm et al., 2007; Schlesinger et al., 2007; Zhao et al., 2007).

Hypermethylation of target genes may thus confer a survival advantage to cancer cells

and promote carcinogenesis through maintenance of epigenetic plasticity and cellular

pluripotency. To determine if specific methylation changes are involved in the

development of CIMP+ CRC, expression of the relevant genes would need to be assessed.

Since RNA expression was not evaluated in this tumour series, it was not possible to

assess the functional significance of aberrant CpG methylation.

A substantial number of genes showed highly significant, differential methylation

between CIMP-H and CIMP-L tumours (Supplementary Table 5.2). Methylation and

somatic mutations for some of these genes have been reported previously in CRC. Further

investigation of these genes in terms of their expression level and their role in

development of the CIMP-H phenotype is warranted. Integration of other important

genetic events in CRC such as mutations in APC and p53 and LOH of 18q and 17p may

also help to further define the three CIMP subgroups. As with any approach for the

classification of CIMP, the present results require confirmation in independent cohorts of

unselected CRC using the same methylation analysis platform.


Despite the “comprehensive” profiling of methylation in this study, only a subset of

cancer-related genes was analyzed (n=808). Methylation analysis at the genome-wide

scale is required to elucidate the entire methylome in CRC. This can be achieved using

commercially available products with greater coverage such as the Illumina Infinium II

methylation array or by combining chromatin immunoprecipitation-enriched DNA with

microarray platforms. Alternatively, massively-parallel sequencing technology (deep

sequencing) allows methylation analysis with high accuracy, throughput and coverage

and requires no a priori knowledge of the genome sequence (Lister and Ecker, 2009).

This approach will allow high resolution analysis of methylation changes within and

outside of CpG islands that may be involved in the development of CRC.

7.5 CIMP+ in early onset CRC

Although CIMP+ CRC occurs predominantly in elderly patients, the present study

estimated that 8% of CRC from patients aged <60 years were CIMP+ (Chapter 6; Ang et

al., 2009). CIMP+ in this cohort was determined using a consensus panel of methylation

markers and MethyLight analysis. Considering that about 25% of all CRCs occur in <60

year old patients (Morris et al., 2007) and that the overall frequency of CIMP+ in CRC is

approximately 20%, 1 in 10 CIMP+ CRC are extrapolated to occur in this young age

group. The clinicopathological and molecular characteristics of these CIMP+ tumours

closely resemble those found in older patients and include proximal tumour location,

advanced stage and association with BRAF V600E mutation.

The occurrence of CIMP+ tumours in early onset CRC patients, together with their

previously reported associations with serrated adenomas/hyperplastic polyps (Hawkins

and Ward, 2001; Jass et al., 2000; Mäkinen et al., 2001) suggests a possible familial


origin for this tumour phenotype. Jass and coworkers have proposed that hyperplastic

polyposis (HPP) and serrated polyposis syndrome (SPS) may predispose to CIMP+ CRC

(Young and Jass, 2006; Young et al., 2007). SPS and HPP are rare conditions in which

familial clustering is only sometimes observed, meaning the proportion of all CIMP+

CRCs with a hereditary origin is likely to be small (Young and Jass, 2006; Young et al.,

2007). Elevated methylation levels have been demonstrated in the normal mucosa of SPS

and HPP patients (Minoo et al., 2006; Wynter et al., 2004), paralleling observations made

in the normal colonic tissues of sporadic CIMP+ CRC patients (Kawakami et al., 2006).

Information on family history of CRC and detailed pathology information on serrated

architecture was not available for the young CIMP+ CRC patients identified in the present

study. This information should ideally be collected in further prospective research studies,

together with molecular screening for CIMP+ and BRAF mutation.

BRAF mutation in MSS CRC has been associated with positive family history of CRC

(Samowitz’05CR). In the present study (Chapter 6; Ang et al, 2009), BRAF mutations

were found in approximately half of the CIMP+ CRC from young patients. Serrated

adenomas also show a high frequency of BRAF mutation (Spring et al., 2006). These

findings suggest that screening for BRAF mutation in premalignant lesions and tumours

from young patients may be a convenient approach for the identification of possible

familial conditions such as SPS and HPP that could predispose to CIMP+ CRC. Another

approach may be to assess promoter methylation levels in the normal colonic mucosa of

individuals judged to be at high risk because of positive family history of CRC and/or the

presence of polyps with a serrated architecture. Because of the generally low methylation

levels present in normal colonic mucosa, techniques that quantify methylation with single

CpG site resolution and high sensitivity are needed. Digital bisulfite sequencing, mass

spectrometry and pyrosequencing should all be suitable.


7.6 Major Findings and Conclusions

Ct value quality control

The Ct value was established as a quality control to assess the suitability of DNA for

methylation analysis, thus permitting efficient sample management. In the present study,

the Ct value was useful for ensuring the reproducibility of results using several

methylation analysis techniques including MSP, MethyLight, pyrosequencing, bsSEQ and

GoldenGate methylation arrays. As real-time quantification of genomic DNA is a

relatively straight forward process requiring minimal optimization, the Ct value represents

an ideal quality control system for widespread use in research and clinical settings.

CpG site methylation and gene expression

To date there has been no systematic study aimed at identifying CpG sites whose

methylation correlates with gene expression. Methylation data from single CpG sites in

CRC cell lines was used to construct a predictive model for RNA expression. This model

failed to predict RNA expression in primary CRCs, suggesting the same experimental

system should be used for both the construction and validation of such models.

Analysis of CIMP using methylation arrays

Unbiased analysis of comprehensive methylation data revealed three CIMP subgroups,

each associated with distinct clinicopathological and molecular features. KRAS and BRAF

mutations were significantly associated with the CIMP-H subgroup. The use of

methylation arrays allows both hyper- and hypomethylated genes to be investigated in

tumours relative to normal tissue. Genome-wide analysis of methylation and gene

expression should help to identify the aberrant biological pathways that underlie the

different CRC CIMP subgroups.


Possible familial origin for CIMP+

Approximately 10% of all CIMP+ CRC were found to arise in relatively young patients,

of which half also showed BRAF mutation. Several groups have proposed that hereditary

factors may contribute to the development of some CIMP+ CRC. The lack of information

on family history of cancer and on pathological information relating to serrated

architecture precluded investigation of this concept in the present study.

References 104

References

Aaltonen L, Johns L, Jarvinen H, Mecklin JP, Houlston R (2007). Explaining the familial colorectal cancer risk associated with mismatch repair (MMR)-deficient and MMR-stable tumours. Clin Cancer Res 13: 356-61. Ang PW, Li WQ, Soong R, Iacopetta B (2009). BRAF mutation is associated with the CpG island methylator phenotype in colorectal cancer from young patients. Cancer Lett 273: 221-4. Ang PW, Toh HB, Iacopetta B, Soong R (2008). An improved quality control for bisulfite-PCR-based DNA methylation analysis: cycle threshold value. Clin Chem Lab Med 46: 1117-21. Barault L, Charon-Barra C, Jooste V, de la Vega M, Martin L, Roignot P et al (2008a). Hypermethylator phenotype in sporadic colon cancer: study on a population-based series of 582 cases. Cancer Res 68: 8541-6. Barault L, Charon-Barra C, Jooste V, de la Vega MF, Martin L, Roignot P et al (2008b). Hypermethylator phenotype in sporadic colon cancer: study on a population-based series of 582 cases. Cancer Res 68: 8541-6. Bibikova M, Fan JB (2009). GoldenGate assay for DNA methylation profiling. Methods Mol Biol 507: 149-63. Bibikova M, Lin Z, Zhou L, Chudin E, Garcia EW, Wu B et al (2006). High-throughput DNA methylation profiling using universal bead arrays. Genome Res 16: 383-93. Byun HM, Siegmund KD, Pan F, Weisenberger DJ, Kanel G, Laird PW et al (2009). Epigenetic profiling of somatic tissues from human autopsy specimens identifies tissue- and individual-specific DNA methylation patterns. Hum Mol Genet. Campan M, Weisenberger DJ, Trinh B, Laird PW (2009). MethyLight. Methods Mol Biol 507: 325-37. Cedar H, Bergman Y (2009). Linking DNA methylation and histone modification: patterns and paradigms. Nat Rev Genet 10: 295-304. Chen T, Hevi S, Gay F, Tsujimoto N, He T, Zhang B et al (2007). Complete inactivation of DNMT1 leads to mitotic catastrophe in human cancer cells. Nat Genet 39: 391-6. Christensen BC, Houseman EA, Marsit CJ, Zheng S, Wrensch MR, Wiemels JL et al (2009). Aging and environmental exposures alter tissue-specific DNA methylation dependent upon CpG island context. PLoS Genet 5: e1000602. Clark SJ, Statham A, Stirzaker C, Molloy PL, Frommer M (2006). DNA methylation: bisulphite modification and analysis. Nat Protoc 1: 2353-64.

References 105

Cortez CC, Jones PA (2008). Chromatin, cancer and drug therapies. Mutat Res 647: 44-51. Crook S, Seth R, Jackson D, Ilyas M (2009). Concomitant mutations and splice variants in KRAS and BRAF demonstrate complex perturbation of the Ras/Raf signalling pathway in Colorectal Cancer. Gut. Cropley JE, Martin DI, Suter CM (2008). Germline epimutation in humans. Pharmacogenomics 9: 1861-8. Crott JW, Liu Z, Keyes MK, Choi SW, Jang H, Moyer MP et al (2008). Moderate folate depletion modulates the expression of selected genes involved in cell cycle, intracellular signaling and folate uptake in human colonic epithelial cell lines. J Nutr Biochem 19: 328-35. Curtin K, Samowitz W, Wolff R, Caan B, Ulrich C, Potter J et al (2009). MSH6 G39E polymorphism and CpG island methylator phenotype in colon cancer. Mol Carcinog. Curtin K, Slattery ML, Ulrich CM, Bigler J, Levin TR, Wolff RK et al (2007). Genetic polymorphisms in one-carbon metabolism: associations with CpG island methylator phenotype (CIMP) in colon cancer and the modifying effects of diet. Carcinogenesis 28: 1672-9. Dejeux E, El abdalaoui H, Gut IG, Tost J (2009). Identification and quantification of differentially methylated loci by the pyrosequencing technology. Methods Mol Biol 507: 189-205. Dobrovic A, Kristensen LS (2009). DNA methylation, epimutations and cancer predisposition. Int J Biochem Cell Biol 41: 34-9. Downward J (2003). Targeting RAS signalling pathways in cancer therapy. Nat Rev Cancer 3: 11-22. Ehrich M, Zoll S, Sur S, van den Boom D (2007). A new method for accurate assessment of DNA quality after bisulfite treatment. Nucleic Acids Res 35: e29. Ellison J, Dean M, Goldman D (1993). Efficacy of fluorescence-based PCR-SSCP for detection of point mutations. Biotechniques 15: 684-91. Esteller M (2002). CpG island hypermethylation and tumour suppressor genes: a booming present, a brighter future. Oncogene 21: 5427-40. Esteller M (2007). Epigenetic gene silencing in cancer: the DNA hypermethylome. Hum Mol Genet 16 Spec No 1: R50-9. Fearon E, Vogelstein B (1990). A genetic model for colorectal tumourigenesis. Cell 61: 759-67. Ferracin M, Gafa R, Miotto E, Veronese A, Pultrone C, Sabbioni S et al (2008a). The methylator phenotype in microsatellite stable colorectal cancers is characterized by a distinct gene expression profile. J Pathol 214: 594-602.

References 106

Ferracin M, Gafà R, Miotto E, Veronese A, Pultrone C, Sabbioni S et al (2008b). The methylator phenotype in microsatellite stable colorectal cancers is characterized by a distinct gene expression profile. J Pathol 214: 594-602. Frazier M, Xi L, Zong J, Viscofsky N, Rashid A, Wu E et al (2003). Association of the CpG island methylator phenotype with family history of cancer in patients with colorectal cancer. Cancer Res 63: 4805-8. Frommer M, McDonald LE, Millar DS, Collis CM, Watt F, Grigg GW et al (1992). A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proc Natl Acad Sci U S A 89: 1827-31. Genereux DP, Johnson WC, Burden AF, Stoger R, Laird CD (2008). Errors in the bisulfite conversion of DNA: modulating inappropriate- and failed-conversion frequencies. Nucleic Acids Res 36: e150. Gilbert MT, Haselkorn T, Bunce M, Sanchez JJ, Lucas SB, Jewell LD et al (2007). The isolation of nucleic acids from fixed, paraffin-embedded tissues-which methods are useful when? PLoS One 2: e537. Goel A, Arnold CN, Tassone P, Chang DK, Niedzwiecki D, Dowell JM et al (2004). Epigenetic inactivation of RUNX3 in microsatellite unstable sporadic colon cancers. Int J Cancer 112: 754-9. Goll MG, Bestor TH (2005). Eukaryotic cytosine methyltransferases. Annu Rev Biochem 74: 481-514. Grady W, Carethers J (2008). Genomic and epigenetic instability in colorectal cancer pathogenesis. Gastroenterology 135: 1079-99. Grønbaek K, Hother C, Jones P (2007). Epigenetic changes in cancer. APMIS 115: 1039-59. Guil S, Esteller M (2009). DNA methylomes, histone codes and miRNAs: tying it all together. Int J Biochem Cell Biol 41: 87-95. Hammons G, Yan Y, Lopatina N, Jin B, Wise C, Blann E et al (1999). Increased expression of hepatic DNA methyltransferase in smokers. Cell Biol Toxicol 15: 389-94. Hastie T, Tibshirani R, Sherlock G, Eisen M, Brown P, Botstein D. (1999). Stanford University Statistics Department Technical report Stanford University Statistics Department. Hawkins N, Norrie M, Cheong K, Mokany E, Ku SL, Meagher A et al (2002). CpG island methylation in sporadic colorectal cancers and its relationship to microsatellite instability. Gastroenterology 122: 1376-87. Hawkins NJ, Ward RL (2001). Sporadic colorectal cancers with microsatellite instability and their possible origin in hyperplastic polyps and serrated adenomas. J Natl Cancer Inst 93: 1307-13.

References 107

He L, Hannon GJ (2004). MicroRNAs: small RNAs with a big role in gene regulation. Nat Rev Genet 5: 522-31. Herman J, Graff J, Myohanen S, Nelkin B, Baylin S (1996). Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc Natl Acad Sci U S A 93: 9821-6. Hitchins MP, Ward RL (2009). Constitutional (germline) MLH1 epimutation as an aetiological mechanism for hereditary non-polyposis colorectal cancer. J Med Genet. Hoensch H, Peters W, Roelofs H, Kirch W (2006). Expression of the glutathione enzyme system of human colon mucosa by localisation, gender and age. Curr Med Res Opin 22: 1075-83. Humphries SE, Gudnason V, Whittall R, Day IN (1997). Single-strand conformation polymorphism analysis with high throughput modifications, and its use in mutation detection in familial hypercholesterolemia. International Federation of Clinical Chemistry Scientific Division: Committee on Molecular Biology Techniques. Clin Chem 43: 427-35. Iacobuzio-Donahue CA (2009). Epigenetic changes in cancer. Annu Rev Pathol 4: 229-49. Iacopetta B, Grieu F (2000). Routine detection of the replication error phenotype in clinical tumour specimens using fluorescence-SSCP. Biotechniques 28: 566-8, 570. Iacopetta B, Grieu F, Li W, Ruszkiewicz A, Caruso M, Moore J et al (2006). APC gene methylation is inversely correlated with features of the CpG island methylator phenotype in colorectal cancer. Int J Cancer 119: 2272-8. Iacopetta B, Heyworth J, Girschik J, Grieu F, Clayforth C, Fritschi L (2009). The MTHFR C677T and DeltaDNMT3B C-149T polymorphisms confer different risks for right- and left-sided colorectal cancer. Int J Cancer 125: 84-90. Iacopetta B, Kawakami K, Watanabe T (2008). Predicting clinical outcome of 5-fluorouracil-based chemotherapy for colon cancer patients: is the CpG island methylator phenotype the 5-fluorouracil-responsive subgroup? . Int J Clin Oncol 13: 498-503. Imai K, Yamamoto H (2008). Carcinogenesis and microsatellite instability: the interrelationship between genetics and epigenetics. Carcinogenesis 29: 673-80. Jass J (2005). Serrated adenoma of the colorectum and the DNA-methylator phenotype. Nat Clin Pract Oncol 2: 398-405. Jass J (2007a). Classification of colorectal cancer based on correlation of clinical, morphological and molecular features. Histopathology 50: 113-30. Jass J (2007b). Molecular heterogeneity of colorectal cancer: Implications for cancer control. Surg Oncol 16: S7-9.

References 108

Jass J, Whitehall V, Young J, Leggett B (2002). Emerging concepts in colorectal neoplasia. Gastroenterology 123: 862-76. Jass JR (2007c). Classification of colorectal cancer based on correlation of clinical, morphological and molecular features. Histopathology 50: 113-30. Jass JR, Iino H, Ruszkiewicz A, Painter D, Solomon MJ, Koorey DJ et al (2000). Neoplastic progression occurs through mutator pathways in hyperplastic polyposis of the colorectum. Gut 47: 43-9. Kanai Y, Ushijima S, Kondo Y, Nakanishi Y, Hirohashi S (2001). DNA methyltransferase expression and DNA methylation of CPG islands and peri-centromeric satellite regions in human colorectal and stomach cancers. Int J Cancer 91: 205-12. Kawakami K, Ooyama A, Ruszkiewicz A, Jin M, Watanabe G, Moore J et al (2008). Low expression of gamma-glutamyl hydrolase mRNA in primary colorectal cancer with the CpG island methylator phenotype. Br J Cancer 98: 1555-61. Kawakami K, Ruszkiewicz A, Bennett G, Moore J, Grieu F, Watanabe G et al (2006). DNA hypermethylation in the normal colonic mucosa of patients with colorectal cancer. Br J Cancer 94: 593-8. Kawasaki T, Ohnishi M, Nosho K, Suemoto Y, Kirkner GJ, Meyerhardt JA et al (2008). CpG island methylator phenotype-low (CIMP-low) colorectal cancer shows not only few methylated CIMP-high-specific CpG islands, but also low-level methylation at individual loci. Mod Pathol 21: 245-55. Keyes MK, Jang H, Mason JB, Liu Z, Crott JW, Smith DE et al (2007). Older age and dietary folate are determinants of genomic and p16-specific DNA methylation in mouse colon. J Nutr 137: 1713-7. Kim K, Park U, Wang J, Lee J, Park S, Kim S et al (2008). Gene profiling of colonic serrated adenomas by using oligonucleotide microarray. Int J Colorectal Dis 23: 569-80. Kim YI (2005). 5,10-Methylenetetrahydrofolate reductase polymorphisms and pharmacogenetics: a new role of single nucleotide polymorphisms in the folate metabolic pathway in human health and disease. Nutr Rev 63: 398-407. Kiriakidou M, Tan GS, Lamprinaki S, De Planell-Saguer M, Nelson PT, Mourelatos Z (2007). An mRNA m7G cap binding-like motif within human Ago2 represses translation. Cell 129: 1141-51. Kristensen LS, Hansen LL (2009). PCR-based methods for detecting single-locus DNA methylation biomarkers in cancer diagnostics, prognostics, and response to treatment. Clin Chem 55: 1471-83. Lacey MR, Ehrlich M (2009). Modeling dependence in methylation patterns with application to ovarian carcinomas. Stat Appl Genet Mol Biol 8: Article 40.

References 109

Ladd-Acosta C, Pevsner J, Sabunciyan S, Yolken RH, Webster MJ, Dinkins T et al (2007). DNA methylation signatures within the human brain. Am J Hum Genet 81: 1304-15. Lee TI, Jenner RG, Boyer LA, Guenther MG, Levine SS, Kumar RM et al (2006). Control of developmental regulators by Polycomb in human embryonic stem cells. Cell 125: 301-13. Li D, Jin C, McCulloch C, Kakar S, Berger BM, Imperiale TF et al (2009). Association of large serrated polyps with synchronous advanced colorectal neoplasia. Am J Gastroenterol 104: 695-702. Li LC, Dahiya R (2002). MethPrimer: designing primers for methylation PCRs. Bioinformatics 18: 1427-31. Li W, Kawakami K, Ruszkiewicz A, Bennett G, Moore J, Iacopetta B (2006a). BRAF mutations are associated with distinctive clinical, pathological and molecular features of colorectal cancer independently of microsatellite instability status. Mol Cancer 5. Li WQ, Kawakami K, Ruszkiewicz A, Bennett G, Moore J, Iacopetta B (2006b). BRAF mutations are associated with distinctive clinical, pathological and molecular features of colorectal cancer independently of microsatellite instability status. Mol Cancer 5: 2. Licchesi JD, Herman JG (2009). Methylation-specific PCR. Methods Mol Biol 507: 305-23. Lister R, Ecker JR (2009). Finding the fifth base: genome-wide sequencing of cytosine methylation. Genome Res 19: 959-66. Lopez-Serra L, Esteller M (2008). Proteins that bind methylated DNA and human cancer: reading the wrong words. Br J Cancer 98: 1881-5. Lunn D, Thomas A, Best N, Spiegelhalter D (2000). WinBUGS -- a Bayesian modelling framework: concepts, structure, and extensibility. Statistics and Computing 10: 325-337. Mäkinen M, George S, Jernvall P, Mäkelä J, Vihko P, Karttunen T (2001). Colorectal carcinoma associated with serrated adenoma--prevalence, histological features, and prognosis. J Pathol 193: 286-94. Makino R, Yazyu H, Kishimoto Y, Sekiya T, Hayashi K (1992). F-SSCP: fluorescence-based polymerase chain reaction-single-strand conformation polymorphism (PCR-SSCP) analysis. PCR Methods Appl 2: 10-3. Marsit CJ, Christensen BC, Houseman EA, Karagas MR, Wrensch MR, Yeh RF et al (2009). Epigenetic profiling reveals etiologically distinct patterns of DNA methylation in head and neck squamous cell carcinoma. Carcinogenesis 30: 416-22. Martin-Subero JI, Ammerpohl O, Bibikova M, Wickham-Garcia E, Agirre X, Alvarez S et al (2009). A comprehensive microarray-based DNA methylation study of 367 hematological neoplasms. PLoS One 4: e6986.

References 110

Martinez R, Martin-Subero JI, Rohde V, Kirsch M, Alaminos M, Fernandez AF et al (2009). A microarray-based DNA methylation study of glioblastoma multiforme. Epigenetics 4: 255-64. McCabe MT, Brandes JC, Vertino PM (2009). Cancer DNA methylation: molecular mechanisms and clinical implications. Clin Cancer Res 15: 3927-37. McRonald FE, Morris MR, Gentle D, Winchester L, Baban D, Ragoussis J et al (2009). CpG methylation profiling in VHL related and VHL unrelated renal cell carcinoma. Mol Cancer 8: 31. Menigatti M, Truninger K, Gebbers J, Marbet U, Marra G, Schär P (2009). Normal colorectal mucosa exhibits sex- and segment-specific susceptibility to DNA methylation at the hMLH1 and MGMT promoters. Oncogene 28: 899-909. Minoo P, Baker K, Goswami R, Chong G, Foulkes WD, Ruszkiewicz AR et al (2006). Extensive DNA methylation in normal colorectal mucosa in hyperplastic polyposis. Gut 55: 1467-74. Minoo P, Moyer M, Jass J (2007). Role of BRAF-V600E in the serrated pathway of colorectal tumourigenesis. J Pathol 212: 124-33. Miyamoto K, Asada K, Fukutomi T, Okochi E, Yagi Y, Hasegawa T et al (2003). Methylation-associated silencing of heparan sulfate D-glucosaminyl 3-O-sulfotransferase-2 (3-OST-2) in human breast, colon, lung and pancreatic cancers. Oncogene 22: 274-80. Morris M, Platell C, Iacopetta B (2007). A population-based study of age-related variation in clinicopathological features, molecular. Markers and outcome from colorectal cancer. Anticancer Res 27: 2833-8. Mulero-Navarro S, Esteller M (2008). Epigenetic biomarkers for human cancer: the time is now. Crit Rev Oncol Hematol 68: 1-11. Munson K, Clark J, Lamparska-Kupsik K, Smith SS (2007). Recovery of bisulfite-converted genomic sequences in the methylation-sensitive QPCR. Nucleic Acids Res 35: 2893-903. Myers R (1992). Classical and Modern Regression Analysis, 2nd edn. Wiley. Nagasaka T, Koi M, Kloor M, Gebert J, Vilkin A, Nishida N et al (2008a). Mutations in both KRAS and BRAF may contribute to the methylator phenotype in colon cancer. Gastroenterology 134: 1950-60. Nagasaka T, Koi M, Kloor M, Gebert J, Vilkin A, Nishida N et al (2008b). Mutations in both KRAS and BRAF may contribute to the methylator phenotype in colon cancer. Gastroenterology 134: 1950-60, 1960 e1. Nagasaka T, Sasamoto H, Notohara K, Cullings HM, Takeda M, Kimura K et al (2004). Colorectal cancer with mutation in BRAF, KRAS, and wild-type with respect to both oncogenes showing different patterns of DNA methylation. J Clin Oncol 22: 4584-94.

References 111

Ng EK, Tsang WP, Ng SS, Jin HC, Yu J, Li JJ et al (2009). MicroRNA-143 targets DNA methyltransferases 3A in colorectal cancer. Br J Cancer 101: 699-706. Niessen RC, Hofstra RM, Westers H, Ligtenberg MJ, Kooi K, Jager PO et al (2009). Germline hypermethylation of MLH1 and EPCAM deletions are a frequent cause of Lynch syndrome. Genes Chromosomes Cancer 48: 737-44. Normanno N, Tejpar S, Morgillo F, De Luca A, Van Cutsem E, Ciardiello F (2009). Implications for KRAS status and EGFR-targeted therapies in metastatic CRC. Nat Rev Clin Oncol. Nosho K, Irahara N, Shima K, Kure S, Kirkner GJ, Schernhammer ES et al (2008). Comprehensive biostatistical analysis of CpG island methylator phenotype in colorectal cancer using a large population-based sample. PLoS One 3: e3698. O'Riain C, O'Shea DM, Yang Y, Le Dieu R, Gribben JG, Summers K et al (2009). Array-based DNA methylation profiling in follicular lymphoma. Leukemia 23: 1858-66. Ogino S, Cantor M, Kawasaki T, Brahmandam M, Kirkner G, Weisenberger D et al (2006a). CpG island methylator phenotype (CIMP) of colorectal cancer is best characterised by quantitative DNA methylation analysis and prospective cohort studies. Gut 55: 1000-6. Ogino S, Cantor M, Kawasaki T, Brahmandam M, Kirkner GJ, Weisenberger DJ et al (2006b). CpG island methylator phenotype (CIMP) of colorectal cancer is best characterised by quantitative DNA methylation analysis and prospective cohort studies. Gut 55: 1000-6. Ogino S, Kawasaki T, Kirkner G, Kraft P, Loda M, Fuchs C (2007a). Evaluation of markers for CpG island methylator phenotype (CIMP) in colorectal cancer by a large population-based sample. J Mol Diagn 9: 305-14. Ogino S, Kawasaki T, Kirkner GJ, Kraft P, Loda M, Fuchs CS (2007b). Evaluation of markers for CpG island methylator phenotype (CIMP) in colorectal cancer by a large population-based sample. J Mol Diagn 9: 305-14. Ogino S, Kawasaki T, Kirkner GJ, Loda M, Fuchs CS (2006c). CpG island methylator phenotype-low (CIMP-low) in colorectal cancer: possible associations with male sex and KRAS mutations. J Mol Diagn 8: 582-8. Ogino S, Odze RD, Kawasaki T, Brahmandam M, Kirkner GJ, Laird PW et al (2006d). Correlation of pathologic features with CpG island methylator phenotype (CIMP) by quantitative DNA methylation analysis in colorectal carcinoma. Am J Surg Pathol. 30:1175-83. Ogino S, Nosho K, Kirkner G, Kawasaki T, Meyerhardt J, Loda M et al (2009). CpG island methylator phenotype, microsatellite instability, BRAF mutation and clinical outcome in colon cancer. Gut 58: 90-6.

References 112

Oliveira C, Velho S, Moutinho C, Ferreira A, Preto A, Domingo E et al (2007). KRAS and BRAF oncogenic mutations in MSS colorectal carcinoma progression. Oncogene 26: 158-63. Ordway JM, Williams K, Curran T (2004). Transcription repression in oncogenic transformation: common targets of epigenetic repression in cells transformed by Fos, Ras or Dnmt1. Oncogene 23: 3737-48. Osborn NK, Zou H, Molina JR, Lesche R, Lewin J, Lofton-Day C et al (2006). Aberrant methylation of the eyes absent 4 gene in ulcerative colitis-associated dysplasia. Clin Gastroenterol Hepatol 4: 212-8. Paz MF, Fraga MF, Avila S, Guo M, Pollan M, Herman JG et al (2003). A systematic profile of DNA methylation in human cancer cell lines. Cancer Res 63: 1114-21. Perrimon N, Bernfield M (2000). Specificities of heparan sulphate proteoglycans in developmental processes. Nature 404: 725-8. Preto A, Figueiredo J, Velho S, Ribeiro A, Soares P, Oliveira C et al (2008). BRAF provides proliferation and survival signals in MSI colorectal carcinoma cells displaying BRAF(V600E) but not KRAS mutations. J Pathol 214: 320-7. Probst A, Dunleavy E, Almouzni G (2009). Epigenetic inheritance during the cell cycle. Nat Rev Mol Cell Biol 10: 192-206. Rashid A, Issa JP (2004). CpG island methylation in gastroenterologic neoplasia: a maturing field. Gastroenterology 127: 1578-88. Ray S, Swanson H (2004). Dioxin-induced immortalization of normal human keratinocytes and silencing of p53 and p16. J Biol Chem 279: 27187–93. Reed K, Poulin ML, Yan L, Parissenti AM (2009). Comparison of bisulfite sequencing PCR with pyrosequencing for measuring differences in DNA methylation. Anal Biochem. Reik W (2007). Stability and flexibility of epigenetic gene regulation in mammalian development. Nature 447: 425-32. Rodriguez J, Frigola J, Vendrell E, Risques RA, Fraga MF, Morales C et al (2006). Chromosomal instability correlates with genome-wide DNA demethylation in human primary colorectal cancers. Cancer Res 66: 8462-9468. Royo JL, Hidalgo M, Ruiz A (2007). Pyrosequencing protocol using a universal biotinylated primer for mutation detection and SNP genotyping. Nat Protoc 2: 1734-9. Samowitz W, Albertsen H, Herrick J, Levin T, Sweeney C, Murtaugh M et al (2005a). Evaluation of a large, population-based sample supports a CpG island methylator phenotype in colon cancer. Gastroenterology 129: 837-45. Samowitz W, Slattery M, Sweeney C, Herrick J, Wolff R, Albertsen H (2007). APC mutations and other genetic and epigenetic changes in colon cancer. Mol Cancer Res 5: 165-70.

References 113

Samowitz W, Sweeney C, Herrick J, Albertsen H, Levin T, Murtaugh M et al (2005b). Poor survival associated with the BRAF V600E mutation in microsatellite-stable colon cancers. Cancer Res 65: 6063-9. Samowitz WS, Albertsen H, Herrick J, Levin TR, Sweeney C, Murtaugh MA et al (2005c). Evaluation of a large, population-based sample supports a CpG island methylator phenotype in colon cancer. Gastroenterology 129: 837-45. Samowitz WS, Albertsen H, Sweeney C, Herrick J, Caan BJ, Anderson KE et al (2006). Association of smoking, CpG island methylator phenotype, and V600E BRAF mutations in colon cancer. J Natl Cancer Inst 98: 1731-8. Sarter B, Long T, Tsong W, Koh W, Yu M, Laird P (2005). Sex differential in methylation patterns of selected genes in Singapore Chinese. Hum Genet 117: 402-3. Schatz P, Distler J, Berlin K, Schuster M (2006). Novel method for high throughput DNA methylation marker evaluation using PNA-probe library hybridization and MALDI-TOF detection. Nucleic Acids Res 34: e59. Schmittgen TD, Livak KJ (2008). Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc 3: 1101-8. Schofield L, Watson N, Grieu F, Li WQ, Zeps N, Harvey J et al (2009). Population-based detection of Lynch syndrome in young colorectal cancer patients using microsatellite instability as the initial test. Int J Cancer 124: 1097-102. Sepulveda AR, Jones D, Ogino S, Samowitz W, Gulley ML, Edwards R et al (2009). CpG methylation analysis--current status of clinical assays and potential applications in molecular diagnostics: a report of the Association for Molecular Pathology. J Mol Diagn 11: 266-78. Shen L, Kondo Y, Guo Y, Zhang J, Zhang L, Ahmed S et al (2007a). Genome-wide profiling of DNA methylation reveals a class of normally methylated CpG island promoters. PLoS Genet 3: 2023-36. Shen L, Toyota M, Kondo Y, Lin E, Zhang L, Guo Y et al (2007b). Integrated genetic and epigenetic analysis identifies three different subclasses of colon cancer. Proc Natl Acad Sci U S A 104: 18654-9. Shworak NW, Liu J, Petros LM, Zhang L, Kobayashi M, Copeland NG et al (1999). Multiple isoforms of heparan sulfate D-glucosaminyl 3-O-sulfotransferase. Isolation, characterization, and expression of human cdnas and identification of distinct genomic loci. J Biol Chem 274: 5170-84. Sohn KJ, Stempak JM, Reid S, Shirwadkar S, Mason JB, Kim YI (2003). The effect of dietary folate on genomic and p53-specific DNA methylation in rat colon. Carcinogenesis 24: 81-90.

References 114

Soong R, Iacopetta BJ (1997). A rapid and nonisotopic method for the screening and sequencing of p53 gene mutations in formalin-fixed, paraffin-embedded tumours. Mod Pathol 10: 252-8. Soong R, Ladanyi A (2003). Improved indicators for assessing the reliability of detection and quantification by kinetic PCR. Clin Chem 49: 973-6. Söreide K, Janssen E, Söiland H, Körner H, Baak J (2006). Microsatellite instability in colorectal cancer. Br J Surg 93: 395-406. Søreide K, Nedrebø B, Knapp J, Glomsaker T, Søreide J, Kørner H (2009). Evolving molecular classification by genomic and proteomic biomarkers in colorectal cancer: potential implications for the surgical oncologist. Surg Oncol 18: 31-50. Spring K, Zhao Z, Karamatic R, Walsh M, Whitehall V, Pike T et al (2006). High prevalence of sessile serrated adenomas with BRAF mutations: a prospective study of patients undergoing colonoscopy. Gastroenterology 131: 1400-7. Suter CM, Martin DI, Ward RL (2004). Hypomethylation of L1 retrotransposons in colorectal cancer and adjacent normal tissue. Int J Colorectal Dis 19: 95-101. Suter CM, Norrie M, Ku SL, Cheong KF, Tomlinson I, Ward RL (2003). CpG island methylation is a common finding in colorectal cancer cell lines. Br J Cancer 88: 413-9. Tan YH, Liu Y, Eu KW, Ang PW, Li WQ, Salto-Tellez M et al (2008). Detection of BRAF V600E mutation by pyrosequencing. Pathology 40: 295-8. Tost J (2009). DNA methylation: an introduction to the biology and the disease-associated changes of a promising biomarker. Methods Mol Biol 507: 3-20. Tost J, Dunker J, Gut IG (2003). Analysis and quantification of multiple methylation variable positions in CpG islands by Pyrosequencing. Biotechniques 35: 152-6. Tost J, Gut IG (2007). DNA methylation analysis by pyrosequencing. Nat Protoc 2: 2265-75. Toyota M, Ahuja N, Ohe-Toyota M, Herman J, Baylin S, Issa J (1999a). CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 96: 8681-6. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, Issa JP (1999b). CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 96: 8681-6. Toyota M, Ohe-Toyota M, Ahuja N, Issa JP (2000). Distinct genetic profiles in colorectal tumours with or without the CpG island methylator phenotype. Proc Natl Acad Sci U S A 97: 710-5. Troyanskaya O, Cantor M, Sherlock G, Brown P, Hastie T, Tibshirani R et al (2001). Missing value estimation methods for DNA microarrays. Bioinformatics 17: 520-5. Ulrich CM, Reed MC, Nijhout HF (2008). Modeling folate, one-carbon metabolism, and DNA methylation. Nutr Rev 66 Suppl 1: S27-30.

References 115

Van Rijnsoever M, Elsaleh H, Joseph D, McCaul K, Iacopetta B (2003). CpG island methylator phenotype is an independent predictor of survival benefit from 5-fluorouracil in stage III colorectal cancer. Clin Cancer Res 9: 2898-903. van Rijnsoever M, Grieu F, Elsaleh H, Joseph D, Iacopetta B (2002). Characterisation of colorectal cancers showing hypermethylation at multiple CpG islands. Gut 51: 797-802. Wallace K, Grau MV, Ahnen D, Snover DC, Robertson DJ, Mahnke D et al (2009). The association of lifestyle and dietary factors with the risk for serrated polyps of the colorectum. Cancer Epidemiol Biomarkers Prev 18: 2310-7. Walther A, Johnstone E, Swanton C, Midgley R, Tomlinson I, Kerr D (2009). Genetic prognostic and predictive markers in colorectal cancer. Nat Rev Cancer 9: 489-99. Wang C, van Rijnsoever M, Grieu F, Bydder S, Elsaleh H, Joseph D et al (2003). Prognostic significance of microsatellite instability and Ki-ras mutation type in stage II colorectal cancer. Oncology 64: 259-65. Wang X, Lam EK, Zhang J, Jin H, Sung JJ (2009). MicroRNA-122a functions as a novel tumour suppressor downstream of adenomatous polyposis coli in gastrointestinal cancers. Biochem Biophys Res Commun 387: 376-80. Ward R, Williams R, Law M, Hawkins N (2004). The CpG island methylator phenotype is not associated with a personal or family history of cancer. Cancer Res 64: 7618-21. Watson N, Grieu F, Morris M, Harvey J, Stewart C, Schofield L et al (2007). Heterogeneous staining for mismatch repair proteins during population-based prescreening for hereditary nonpolyposis colorectal cancer. J Mol Diagn 9: 472-8. Weisenberger D, Siegmund K, Campan M, Young J, Long T, Faasse M et al (2006a). CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet 38: 787-93. Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, Faasse MA et al (2006b). CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nat Genet 38: 787-93. Weisenberger DJ, Trinh BN, Campan M, Sharma S, Long TI, Ananthnarayan S et al (2008). DNA methylation analysis by digital bisulfite genomic sequencing and digital MethyLight. Nucleic Acids Res 36: 4689-98. Weitz J, Koch M, Debus J, Hohler T, Galle PR, Buchler MW (2005). Colorectal cancer. Lancet 365: 153-65. Whittaker JC, Thompson R, Denham MC (2000). Marker-assisted selection using ridge regression. Genet Res 75: 249-52. Widschwendter M, Fiegl H, Egle D, Mueller-Holzner E, Spizzo G, Marth C et al (2007). Epigenetic stem cell signature in cancer. Nat Genet 39: 157-8.

References 116

Wood LD, Calhoun ES, Silliman N, Ptak J, Szabo S, Powell SM et al (2006). Somatic mutations of GUCY2F, EPHA3, and NTRK3 in human cancers. Hum Mutat 27: 1060-1. Wynter C, Walsh M, Higuchi T, Leggett B, Young J, Jass J (2004). Methylation patterns define two types of hyperplastic polyp associated with colorectal cancer. Gut 53: 573-80. Xiao Y, Word B, Starlard-Davenport A, Haefele A, Lyn-Cook B, Hammons G (2008). Age and gender affect DNMT3a and DNMT3b expression in human liver. Cell Biol Toxicol 24: 265-72. Young J, Jass JR (2006). The case for a genetic predisposition to serrated neoplasia in the colorectum: hypothesis and review of the literature. Cancer Epidemiol Biomarkers Prev 15: 1778-84. Young J, Jenkins M, Parry S, Young B, Nancarrow D, English D et al (2007). Serrated pathway colorectal cancer in the population: genetic consideration. Gut 56: 1453-9. Yu J, Freimuth RR, Culverhouse R, Marsh S, Watson MA, McLeod HL (2008). DNA methylotype analysis in colorectal cancer. Oncol Rep 20: 921-7. Zhang Y, Rohde C, Tierling S, Stamerjohanns H, Reinhardt R, Walter J et al (2009). DNA methylation analysis by bisulfite conversion, cloning, and sequencing of individual clones. Methods Mol Biol 507: 177-87.

Appendices 117

Appendices

Supplementary Table 2.1 CRC cell lines and corresponding culture medium as

recommended by American Type Culture Collection (ATCC).

Cell lines Culture medium

Colo 201 RPMI-1640

Colo 205 RPMI-1640

Colo 320 RPMI-1640

HCT 116 DMEM

HL 60 DMEM

LS 174T DMEM

SW 480 DMEM

SW 620 DMEM

SW 837 DMEM

RKO DMEM

Supplementary Table 2.2 PCR amplification conditions for F-SSCP analysis of BRAF,

KRAS, and MSI.

Primer

Sequence

Annealing Temperature

(°C)

Amplicon size (bp)

BAT26 F 5’ – TTGGATATTGCAGCAGTCAG – 3’ 46 136 BAT26 R 5’ – GCTCCTTTATAAGCTTCTTCA – 3’

BRAF F 5’ – TCATAATGCTTGCTCTGATAGGA – 3’ 60 224 BRAF R 5’ – GGCCAAAAATTTAATCAGTGGA – 3’

KRAS F 5’ – GACTGAATATAAACTTGTGG – 3’ 54 107 KRAS R 5’ – CTATTGTTGGATCATATTCG – 3’

Appendices 118

Supplementary Table 2.3 Pyrosequencing PCR conditions, PCR primers, pyrosequencing primers

Gene PCR Forward primer (5’ – 3’) PCR Reverse primer (5’ – 3’) tagged with universal seq GGGACACCGCTGATCGTTTA at the 5’ end

Sequencing primer (5’ – 3’)

Amplicon size (bp)

Annealing Temp (°C)

Sequence to analyse Dispensation order

RUNX3

R133 TAGGGTTTTTAGGAGATT CCCAAAAACCTCCACCCCACA TTTTAGGAGATTTTTTTT 133 55 ATCAGTG YGTTTGTTTT

R147 GGGACACCGCTGATCGTTTATTATYGGTTTTGAGATTT CTCRAATCGTAACTCCCT CAACCCCGAACTAAAAAC 147 60 TCGATCAGAGTC CRCRAAATCT

R147a TTATYGGTTTTGAGATTT CTCRAATCGTAACTCCCT GTTGTAGAAGTTATAGGT 147 50 ATCTGAGTA TYGAAGTAG GAAGGGGTTGYGTTTTAG ATCAGTAGA YGTTAGGGAGT R170 GYGGTTTTYGGTAGGTTT ATCCTCCAAAATCAAATA GTAGGTTTYGTTTTTTTT 170 55 ATCAGTCTGATCAGT YGYGAAYGTTAT

R140 GGGACACCGCTGATCGTTTAGTYGYGTYGTTTTYGTTTGTT AATCCCRCACTCACCTTA AACAACRACCRTCAAAAC 140 58 TGATCAGAGC RCCRAACAA

R181 AGGYGGTYGTGGGGTTYGGA AATCCCRCACTCACCTTA TYGAGGTGYGTTYGATGGTGG 181 60 GATCAGTGT AYGTGTTGG R148 TYGAGGTGYGTTYGATGGTGG AATCCCRCACTCACCTTA ATTGGYGTTGTAATAAGA 148 60 ATCAGTGTCAGTC YGTTGTTYGTYG

DPYD

D244 TTTTTGTTTGTAGGTTGGG TACCAATAACAAACCCTCCTTAC YGTGTTTGGYGYGGGAGT 244 60 ATCAGTAGATCTGAGAGTCTG YGTAGGATYGAGAGYGT TYGTTTYGYGTTYGT 244 ATCAGTCAGTCTG TTYGTTTTYGYGTYGT D230 AGTAAGGATTYGGYGGATAT AAAATCCTCCCCTAAACT YGTGTTTGGYGYGGGAGT 230 55 ATCAGTCAGTC YGYGTYGG GGTGGTGGGGAGTGTTTT 230 AGTCAGTAG GYGGGTAGGTG TTTTYGYGGTTTTYGTATTT 230 ATCTGAGAGTGAGAGATCAGTCTG YGGAGAGTGAGAGAYGYG

Appendices 119

Supplementary Table 2.4 Bisulfite-cloning sequencing PCR conditions, primers

Amplicon size (bp)

Forward primer (5’ – 3’)

Reverse primer (5’ – 3’)

PCR annealing

temperature (°C)

DPYD

289 AGAGATTAAAGGTTAGTTT CCAATAACAAACCCTCCTTACRTC 55

295 AGTYGTAGGATYGAGAGYGT ACACCTACCCRCAAAACACT 60

287 AAGGAGGGTTTGTTATTGGT AAAATCCTCCCCTAAACT 65

226 AGTGTTTTGYGGGTAGGTGT CAAACTTCCTAAAATCTCTT 60

RUNX3

428 GAAATTTGTTTTGAGGGGAGAGTAG ACCRCCCTCTCTCRAATCRTAACT 65

410 TAGAAAGAGTTGGGGAAGTT AACAAATCCTCCAAAATCAAATAAC 60

438 ATGGGGGTTTYGTYGATTGGT ACRCCTCCTCRACCRCCRCTA 65

338 GAGGATGYGGGATTAGTYGGGT CCRCTATTCTCRCCCATCTTA 65

422 GGYGTTTTGAYGGTYGTTGTTA ATATCRCCRCRACRTCTCRAA 57

347 TAAGGTGAGTGYGGGATT AAATCCTCTTCTCCRTTACCCRCA 65

Appendices 120

Supplementary Table 2.5 Reagents and suppliers

Reagents Supplier

Acrylamide (�99%) Sigma-Aldrich, St. Louis, MO Ammonium acetate Sigma-Aldrich, St. Louis, MO Ammonium persulfate (�98%) Sigma-Aldrich, St. Louis, MO Ampicilin Sigma-Aldrich, St. Louis, MO Boric Acid MERCK Pte Ltd. Victoria, Australia Chloroform Sigma-Aldrich, St. Louis, MO Dulbecco’s modified Eagle medium (DMEM)

GIBCO® Invitrogen Cell Culture, Invitrogen Corporation, Carlsbad, CA

EDTA MERCK Pte Ltd. Victoria, Australia Fetal bovine calf serum (FBS) GIBCO® Invitrogen Cell Culture, Invitrogen

Corporation, Carlsbad, CA Glycerol MERCK Pte Ltd. Victoria, Australia Glycogen Sigma-Aldrich, St. Louis, MO

L-glutamine GIBCO® Invitrogen Cell Culture, Invitrogen Corporation, Carlsbad, CA

Hydrochloric acid Sigma-Aldrich, St. Louis, MO IPTG Probiogen AG, Berlin, Germany Magnesium acetate Sigma-Aldrich, St. Louis, MO Penicillin Sigma-Aldrich, St. Louis, MO Phosphate buffer saline (PBS) Invitrogen Corporation, Carlsbad, CA Proteinase K Solution (20 mg/ml) Invitrogen Corporation, Carlsbad, CA Phenol:Chloroform:Isoamyl Alcohol

25:24:1 Saturated with 10 mM Tris, pH 8.0, 1 mM EDTA.

Sigma-Aldrich, St. Louis, MO

RPMI 1640 Invitrogen Corporation, Carlsbad, CA Streptomycin sulphate Sigma-Aldrich, St. Louis, MO Sodium chloride Sigma-Aldrich, St. Louis, MO Streptavidin SepharoseTM High

Performance beads Amersham Biosciences, Uppsala, Sweden

Sodium acetate Sigma-Aldrich, St. Louis, MO SYBR® Safe DNA gel stain Invitrogen Corporation, Carlsbad, CA Trypsin Gibco Tween® 20, Molecular Grade Promega, Madison, WI TRI Reagent® Molecular Research Centre, Cincinnati, OH) Tris MERCK Pte Ltd. Victoria, Australia Tris acetate MERCK Pte Ltd. Victoria, Australia X-gal Promega, Madison, WI

Appendices 121

Supplementary Table 2.6 Commercial kits and suppliers

Commercial kits/assays Manufacturer/supplier Cell lines Colorectal cancer cell lines American Type Culture Collection (ATCC),

Manassas, VA

Cloning & sequencing BigDye Terminator v 3.1 Cycle Sequencing kit Applied Biosystems, Foster City, CA MAX Efficiency® DH5�™ Competent Cells Invitrogen, Carlsbad, CA pGEM®-T Easy Vector Systems Promega, Madison, WI Wizard® DNA Clean-Up System Promega, Madison, WI Wizard® Plus SV Minipreps DNA Purification System

Promega, Madison, WI

Enzyme CpG Methyltransferase (M.SssI) New England Biolabs, Ipswich, MA GoldenGate methylation analysis Goldengate Cancer Panel I methylation array Illumina Inc. San Diego Nucleic acid extraction kits DNeasy Blood & Tissue Kit Qiagen, Valencia, CA RNeasy® Mini Kit Qiagen, Valencia, CA QIAquick DNA Purification kit Qiagen, Valencia, CA PCR reagents FastStart Taq polymerase Roche, Mannheim, Germany HEX-fluorescently tagged oligonucleotides GeneWorks Pty Ltd, Adelaide PCR-grade oligonucleotides 1st BASE Pte Ltd, Singapore Taq DNA Polymerase Qiagen, Valencia, CA 6FAM, TAMRA-labelled probes 1st BASE Pte Ltd, Singapore Pyrosequencing enzymes Pyro Gold Reagents Biotage RNA quantification High Capacity cDNA Reverse Transcription Kit

Applied Biosystems, Foster City, CA

TaqMan® Gene Expression Assays Hs00559279_m1 (DPYD), Hs00231709_m1 (RUNX3), Hs99999905_m1 (GAPDH)


TaqMan® Fast Universal PCR Master Mix (2x) without AmpErase® UNG


Appendices 122

Supplementary Table 2.7 Instrument, software, labware and manufacturers

Instruments, software, labware Manufacturer

BeadArray Reader Illumina Inc. San Diego BeadScan Illumina Inc. San Diego BeadStudio Illumina Inc. San Diego Bio-Rad Gel DocTM 2000 system Bio-Rad Laboratories, Inc, Hercules, CA Forma 310 Series Direct Heat CO2 Incubator Thermo Fisher Scientific Inc.,

Wilmington, DE , USA MicroAmp® Fast 96-Well Reaction Plate Applied Biosystems Inc, Foster City, CA MicroAmp® Optical Adhesive Film Applied Biosystems Inc, Foster City, CA Nanodrop ND-1000 Thermo Fisher, Wilmington, DE One-D gel analysis software (1.3) Scanalytics Inc., Fairfax, VA, USA PSQ96MA instrument Qiagen, Valencia, CA Sequencing Analysis 5.2 with KB basecaller Applied Biosystems, Foster City, CA Tomy MV-100 Micro Vac Tomy Tech USA Inc, Encyclopedia

Circle Fremont, CA T25 tissue culture flask Corning Incorporated Life Sciences,

Lowell, MA QIAxcel system Qiagen, Valencia, CA 3130xl Genetic Analyzer Applied Biosystems, Foster City, CA 7900HT Fast Real-Time PCR System Applied Biosystems, Foster City, CA

Supplementary Table 2.8 Constituents of buffers and solutions

Buffers/Solutions Constituents LB (Luria-Bertani) Agar with ampicilin, X-gal and IPTG

1% bacto tryptone, 0.5% yeast extract, 0.5% sodium choride, 0.08% sodium hydroxide 5 M, 100 �g/ml ampicilin, 0.08 mg/ml X-Gal, 5 �M IPTG, 1.5% agar

LB (Luria-Bertani) broth with ampicilin

1% bacto tryptone, 0.5% yeast extract, 0.5% sodium choride, 0.08% sodium hydroxide 5 M, 100 �g/ml ampicilin

SOC medium 2% bacto tryptone, 0.5% yeast extract, 0.05% sodium chloride, 0.02% potassium chloride, 2 M glucose, 2 M Mg2+sctock

TBE buffer 89 mM Tris, 89 M boric acid, 2 mM EDTA in H2O Tris-HCl 121 g Tris base, 800 ml H2O

Adjusted to desired pH with concentrated HCl, add H2O to 1 liter

30% Acrylamide (50:1) Acrylamide 30g, N-N’-methylene-bis acrylamide 0.6g in H2O

Appendices 123

Supplementary Table 4.1 Simulated MethyLight are designed based on the number of

CG sites probed by forward primer (F), probe (P) and reverse primer (R), the number of

CG not examined between forward primer and probe (CGs btw FP), probe and reverse

primer (CGs btw PR) and the distance between forward primer and probe (F to P), probe

and reverse primer (P to R), the length of each oligonucleotide and the entire amplicon.

Gene F CGs between

FP

P CGs between

PR

R F (bp)

F to P (bp)

P (bp)

P to R (bp)

R (bp)

Amplicon (bp)

RUNX 3 0 5 3 3 19 3 25 45 25 117 CACNA1G 3 0 4 0 4 22 1 25 1 18 67 IGF2 3 2 4 0 3 20 22 22 3 21 88 NEU 3 0 4 0 3 23 3 27 10 25 88 SOCS2 3 0 4 1 2 21 6 26 10 21 84 CDH1 3 0 3 3 1 20 3 21 31 21 96 CHFR 2 2 3 3 2 19 20 25 19 21 104 DAPK 4 0 3 0 3 21 2 21 5 19 68 hMLH1 2 0 4 0 2 21 4 31 7 21 84 p15 2 0 2 0 3 20 6 29 5 21 81 p16 1 1 4 0 2 22 9 18 2 19 70 p57 3 0 3 0 3 20 4 31 2 18 75 RASSF1A 1 1 2 0 3 20 6 18 1 20 65 TIMP3 2 2 4 4 3 21 8 19 25 22 95 ER 2 0 5 3 3 19 9 24 30 19 101 Mlh1 4 2 4 0 3 30 5 22 13 18 88 Average bp 3 1 4 1 3 21 7 24 13 21 86

Appendices 124

Supplementary Figure 5.1 Unsupervised hierarchical clustering of 1505 probes (rows)

in 28 normal colonic tissues (columns). Methylation of X-chromosome genes (enclosed

within yellow rectangle) showed 100% correlation to gender as indicated by the red

(female) and blue (male) bar above the heatmap.

Supplementary Figure 5.2 Unsupervised hierarchical clustering of 1505 probes (rows)

in 91 colorectal tumours (columns) revealed three tumour subgroups.

Appendices 125

Supplementary Figure 5.3 Principal component analysis of 202 CpG loci that were

differentially methylated between tumour and normal colonic tissue. This identified

principal component 2 as the top ranking dimension and which explained 20% of the

variability in the dataset. CIMP-H tumours are denoted in black, CIMP-M in black and

CIMP-L in red.

Appendices 126

Supplementary Table 5.1 202 CpG sites differentially methylated between normal and tumour tissues. Probe Gene P-value

(Normal vs Tumor)

Hypermethylation/Hypomethylation1

Overlap_with_Repeat2

SNP contained in the probe2

SNP frequency Variant Freq > 5%

EYA4_E277_F EYA4 1.27E-13 1 HS3ST2_E145_R HS3ST2 8.90E-13 1 TFPI2_P152_R TFPI2 3.20E-12 1 SLIT2_P208_F SLIT2 7.70E-12 1 NO rs2301252 G/T: G allele=64%, T

allele=36% Yes

SLIT2_E111_R SLIT2 2.43E-11 1 TMEFF2_P152_R TMEFF2 3.36E-11 1 DBC1_E204_F DBC1 3.39E-11 1 HTR1B_E232_R HTR1B 3.91E-11 1 TMEFF2_E94_R TMEFF2 1.02E-10 1 TWIST1_E117_R TWIST1 1.03E-10 1 NO rs34354639 ins/delG: freq unknown na NPY_E31_R NPY 1.09E-10 1 TWIST1_P44_R TWIST1 1.38E-10 1 NEFL_P209_R NEFL 1.53E-10 1 TFPI2_P9_F TFPI2 2.16E-10 1 YES rs2071458 G/T: G allele=79%, T

allele=21% Yes

SFRP1_P157_F SFRP1 2.72E-10 1 ESR1_P151_R ESR1 3.55E-10 1 NO rs34430742 A/G: G allele=100%, A

allele=0% No

FLI1_E29_F FLI1 4.34E-10 1 NTRK3_P752_F NTRK3 4.59E-10 1 TGFB2_E226_R TGFB2 5.18E-10 1 FLI1_P620_R FLI1 5.28E-10 1 NO rs2227636 A/G: G allele=100%, A

allele=0% No

RASGRF1_E16_F RASGRF1 5.39E-10 1 EYA4_P794_F EYA4 6.36E-10 1 SPP1_P647_F SPP1 7.39E-10 0 NGFB_E353_F NGFB 8.07E-10 1

Appendices 127

SOX17_P287_R SOX17 1.00E-09 1 NPY_P295_F NPY 1.02E-09 1 NPY_P91_F NPY 1.04E-09 1 NGFB_P13_F NGFB 1.31E-09 1 GSTM2_E153_F GSTM2 1.42E-09 1 SCGB3A1_E55_R SCGB3A1 1.48E-09 1 BMP4_P199_R BMP4 1.71E-09 0 DLK1_E227_R DLK1 1.88E-09 1 TPEF_seq_44_S88_R TPEF 2.30E-09 1 GABRB3_E42_F GABRB3 3.08E-09 1 TWIST1_P355_R TWIST1 4.06E-09 1 NO rs10717999 ins/delG: freq unknown na TNFSF8_P184_F TNFSF8 5.16E-09 0 NTRK3_E131_F NTRK3 5.78E-09 1 KDR_P445_R KDR 6.97E-09 1 NO rs39315 C/T: C allele=43%, T

allele=57% Yes

CDH13_E102_F CDH13 7.88E-09 1 NOS3_P38_F NOS3 8.02E-09 0 EYA4_P508_F EYA4 8.29E-09 1 ALK_E183_R ALK 8.43E-09 1 FLT4_E206_F FLT4 1.02E-08 1 GAS7_E148_F GAS7 1.12E-08 1 CDH13_P88_F CDH13 1.22E-08 1 WNT2_E109_R WNT2 1.23E-08 1 FGF5_P238_R FGF5 1.27E-08 1 NO rs2051548 A/G: G allele=100%, A

allele=0% No

HLA-DPA1_P205_R HLA-DPA1 1.43E-08 0 NO rs7047734 C/G: freq unknown na HLA-DPA1_P28_R HLA-DPA1 1.60E-08 0 ALK_P28_F ALK 1.75E-08 1 NO rs2303368 A/C: C allele=100%, A

allele=0% No

GSTM2_P453_R GSTM2 1.79E-08 1 THY1_P149_R THY1 1.83E-08 1 PTPRH_E173_F PTPRH 1.93E-08 0 GABRB3_P92_F GABRB3 1.96E-08 1 NO rs36005339 A/T: A allele=100%, T

allele=0% No

Appendices 128

KRT1_P798_R KRT1 2.02E-08 0 HS3ST2_P171_F HS3ST2 2.55E-08 1 NO rs1805088;rs45592943 C/T: C allele=100%, T

allele=0% No

NTRK3_P636_R NTRK3 3.42E-08 1 IGFBP3_P423_R IGFBP3 3.51E-08 1 NO rs9951523 C/T: C allele=97%, T

allele=3% No

MYOD1_E156_F MYOD1 3.58E-08 1 TUSC3_P85_R TUSC3 3.65E-08 1 NO rs34597537;rs3215003 ins/delC: del/del=100% No MEST_E150_F MEST 3.71E-08 0 NRG1_P558_R NRG1 3.82E-08 1 YES rs7420590 G/A: freq unknown na PGR_P790_F PGR 3.93E-08 0 GAS7_P622_R GAS7 4.34E-08 1 ADCYAP1_P398_F ADCYAP1 5.22E-08 1 EPHA5_E158_R EPHA5 5.23E-08 1 GSTM2_P109_R GSTM2 5.52E-08 1 CPA4_E20_F CPA4 5.88E-08 0 IFNG_E293_F IFNG 6.93E-08 0 NO rs1883832 C/T: C allele=78%, T

allele=22% Yes

MPO_P883_R MPO 7.70E-08 0 DAB2IP_E18_R DAB2IP 8.24E-08 1 ADAMTS12_E52_R ADAMTS12 8.45E-08 1 ADCYAP1_P455_R ADCYAP1 8.63E-08 1 DLC1_P695_F DLC1 9.66E-08 0 PI3_P274_R PI3 1.08E-07 0 BDNF_P259_R BDNF 1.19E-07 1 NO rs4906901 A/C: freq unknown na S100A2_E36_R S100A2 1.21E-07 0 MME_E29_F MME 1.31E-07 1 FLT4_P180_R FLT4 1.34E-07 1 TUSC3_E29_R TUSC3 1.35E-07 1 TNFSF10_E53_F TNFSF10 1.39E-07 0 WNT8B_E487_F WNT8B 1.48E-07 0 SOX17_P303_F SOX17 1.55E-07 1 CSPG2_E38_F CSPG2 1.60E-07 1 SFRP1_E398_R SFRP1 1.82E-07 1

Appendices 129

MEG3_E91_F MEG3 1.92E-07 1 NO rs6498979 C/T: freq unknown na KDR_E79_F KDR 1.97E-07 1 NTSR1_P318_F NTSR1 1.99E-07 1 CTLA4_P1128_F CTLA4 2.02E-07 0 WT1_P853_F WT1 2.05E-07 1 NO rs34456313 ins/delG: freq unknown na ESR1_E298_R ESR1 2.07E-07 1 RIPK3_P124_F RIPK3 2.09E-07 0 HS3ST2_P546_F HS3ST2 2.20E-07 1 PENK_P447_R PENK 2.28E-07 1 FGF5_E16_F FGF5 2.46E-07 1 NO rs41416852 A/G: G allele=100%, A

allele=0% No

CD40_P372_R CD40 2.50E-07 1 NO rs10234713 A/G: A allele=99%, G allele=1%

Yes

SPI1_P929_F SPI1 2.55E-07 0 COL1A2_P48_R COL1A2 2.89E-07 1 NO rs3810773 A/G: freq unknown na DCC_P471_R DCC 2.98E-07 1 HOXA5_P1324_F HOXA5 3.05E-07 1 RBP1_E158_F RBP1 3.13E-07 1 NEFL_E23_R NEFL 3.13E-07 1 HBII-52_E142_F HBII-52 3.22E-07 0 NO rs433139 C/G: freq unknown na AGTR1_P154_F AGTR1 3.58E-07 1 MEG3_P235_F MEG3 3.64E-07 1 MME_P388_F MME 3.65E-07 1 CARD15_P665_F CARD15 3.79E-07 0 NO rs28736786; rs4292323 A/C: freq unknown na ER_seq_a1_S60_F ER 3.85E-07 1 EPHA5_P66_F EPHA5 4.14E-07 1 DES_E228_R DES 4.35E-07 1 IRAK3_P13_F IRAK3 4.71E-07 1 LMO1_E265_R LMO1 4.96E-07 1 NO rs28372675 C/T: C allele=67%, T

allele=33% Yes

PEG10_P978_R PEG10 5.02E-07 1 EMR3_E61_F EMR3 5.19E-07 0 WNT2_P217_F WNT2 5.25E-07 1 NO rs10706596 ins/delG: freq unknown na TBX1_P885_R TBX1 5.59E-07 1

Appendices 130

MMP7_E59_F MMP7 5.68E-07 0 VAMP8_P114_F VAMP8 6.19E-07 0 AGTR1_P41_F AGTR1 7.64E-07 1 DBC1_P351_R DBC1 7.76E-07 1 ALOX12_P223_R ALOX12 8.39E-07 1 PI3_P1394_R PI3 9.20E-07 0 MMP2_E21_R MMP2 9.40E-07 1 DCC_P177_F DCC 9.58E-07 1 OPCML_E219_R OPCML 9.64E-07 1 EMR3_P39_R EMR3 1.18E-06 0 FGF3_P171_R FGF3 1.19E-06 1 PI3_E107_F PI3 1.28E-06 0 NO rs11736206 G/T: G allele=100%, T

allele=0% No

VAMP8_P241_F VAMP8 1.33E-06 0 CSPG2_P82_R CSPG2 1.35E-06 1 CDH11_P354_R CDH11 1.50E-06 1 NO rs4995341;rs4995342;rs49

95343 A/G: freq unknown na

MYH11_P22_F MYH11 1.95E-06 1 NO rs9340771 C/T: T allele=99%, C allele=1%

No

CEACAM1_P44_R CEACAM1 1.96E-06 0 NO rs34178679 A/G: freq unknown na ALOX12_E85_R ALOX12 2.00E-06 1 FLT3_E326_R FLT3 2.12E-06 1 NRG1_E74_F NRG1 2.12E-06 1 ERN1_P809_R ERN1 2.13E-06 0 WT1_E32_F WT1 2.17E-06 1 CCNA1_P216_F CCNA1 2.21E-06 1 CCNA1_E7_F CCNA1 2.24E-06 1 LYN_P241_F LYN 2.25E-06 1 EPHA7_E6_F EPHA7 2.27E-06 1 YES ;rs45537841 A/G: G allele = 73%, A

allele=27% Yes

CD40_E58_R CD40 2.51E-06 1 NO rs45521832; rs11765572 A/G: A allele=14%, G allele=86%

Yes

CD34_E20_R CD34 2.56E-06 1 MOS_E60_R MOS 2.75E-06 1

Appendices 131

CSF1R_E26_F CSF1R 2.92E-06 0 FGFR1_P204_F FGFR1 3.31E-06 1 IRAK3_E130_F IRAK3 3.37E-06 1 ABCB4_P892_F ABCB4 3.53E-06 0 NO rs28383487 A/C: A allele=3%, C

allele=97% No

PDE1B_P263_R PDE1B 3.67E-06 1 CAPG_E228_F CAPG 3.85E-06 0 SEZ6L_P299_F SEZ6L 4.48E-06 1 AIM2_P624_F AIM2 4.50E-06 0 VAMP8_E7_F VAMP8 4.67E-06 0 IGF2AS_E4_F IGF2AS 4.86E-06 1 SEZ6L_P249_F SEZ6L 4.92E-06 1 CCL3_P543_R CCL3 5.12E-06 0 NO rs3806469 A/G: A allele =1%, G

allele =99% No

HPN_P374_R HPN 5.26E-06 1 PTPRH_P255_F PTPRH 5.36E-06 0 MMP9_E88_R MMP9 5.46E-06 0 NO rs45546040 C/T: freq unknown na HCK_P858_F HCK 5.51E-06 1 IGF2AS_P203_F IGF2AS 6.16E-06 1 TMEFF2_P210_R TMEFF2 6.35E-06 1 NO rs2069711 G/T: T allele=100%, G

allele=0% No

SRC_E100_R SRC 6.87E-06 0 COL1A2_E299_F COL1A2 7.14E-06 1 MEST_P62_R MEST 7.76E-06 0 TFPI2_E141_F TFPI2 7.85E-06 1 EPHA7_P205_R EPHA7 7.99E-06 1 TSP50_P137_F TSP50 8.78E-06 1 PGR_E183_R PGR 8.84E-06 0 HLA-DPB1_E2_R HLA-DPB1 9.08E-06 0 SRC_P164_F SRC 9.43E-06 0 SERPINB2_P939_F SERPINB2 9.90E-06 0 TCF4_P317_F TCF4 1.04E-05 1 HTR1B_P222_F HTR1B 1.08E-05 1 NO rs12324391;rs2009256 C/T: C allele=73%; T

allele=27% Yes

Appendices 132

SGCE_E149_F SGCE 1.11E-05 1 MYOD1_P50_F MYOD1 1.16E-05 1 EPO_E244_R EPO 1.26E-05 1 GALR1_E52_F GALR1 1.36E-05 1 GLI3_P453_R GLI3 1.37E-05 1 CHFR_P501_F CHFR 1.41E-05 1 NO rs41282752;rs17333103 A/G: freq unknown na GML_E144_F GML 1.45E-05 0 NO rs34311679 A/T: A allele=100%, T

allele=0% No

IRAK3_P185_F IRAK3 1.49E-05 1 CHGA_E52_F CHGA 1.61E-05 1 CYP1B1_E83_R CYP1B1 1.66E-05 1 SERPINE1_E189_R SERPINE1 1.77E-05 1 NO rs12362192 G/T: G allele=100%, A

allele=0% No

PRSS1_E45_R PRSS1 2.06E-05 0 CARD15_P302_R CARD15 2.15E-05 0 RBP1_P150_F RBP1 2.24E-05 1 CCKBR_P480_F CCKBR 2.26E-05 1 NO rs13309040 A/C: freq unknown na LMTK2_P1034_F LMTK2 2.29E-05 0 SPI1_E205_F SPI1 2.49E-05 0 FGF3_E198_R FGF3 2.55E-05 1 VIM_P343_R VIM 2.57E-05 1 EPHB6_E342_F EPHB6 2.81E-05 1 MST1R_P392_F MST1R 2.88E-05 0 DCC_E53_R DCC 2.88E-05 1 NO rs2522207 A/G: freq unknown na PALM2-AKAP2_P420_R

PALM2-AKAP2

2.88E-05 1 NO rs34475065 G/T: freq unknwon na

CDH11_E102_R CDH11 3.00E-05 1 CCR5_P630_R CCR5 3.07E-05 0 MEST_P4_F MEST 3.23E-05 0 NO rs12420823 C/T: freq unknown na GLI2_P295_F GLI2 3.28E-05 0 1Differential methylation of genes in tumour relative to normal colonic tissues, "1" and "0" denotes hypermethylation and hypomethylation respectively. 2Data obtained from Byun et al (Byun'09HMG), “na” denotes no available data.

Appendices 133

Supplementary Table 5.2 Known functions of 112 genes differentially methylated between CIMP-H and CIMP-L and their reported methylation in cancer and putative roles in gastrointestinal cancer.

Gene Mean �-value in CIMP-H

Mean �-value in CIMP-L

P-value (CIMP-L vs

CIMP-H)

Reported methylation in cancer Known function Involvement in GI cancers References

NTRK3 0.81 0.10 1.15E-21 no a member of the neurotrophic tyrosine receptor kinase (NTRK) family. This kinase is a membrane-bound receptor that, upon neurotrophin binding, phosphorylates itself and members of the MAPK pathway. Signalling through this kinase leads to cell differentiation and may play a role in the development of proprioceptive neurons that sense body position. Mutations in this gene have been associated with medulloblastomas, secretory breast carcinomas and other cancers

In the highly metastatic CT26 murine colon cancer cell line, which expresses endogenous TrkC, silencing TrkC expression by small interfering RNA significantly enhanced BMP-2-induced Smad1 phosphorylation and restored BMP-2 growth inhibitory activity, somatic mutations found in colon cancer

Cancer Res. 2007 Oct 15;67(20):9869-77, Science. 2003 May 9;300(5621):949

HS3ST2 0.46 0.07 4.57E-21 methylated in hematological, breast and cervical cancer, methylated in CRC

heparan sulfate biosynthetic enzyme family. It is a type II integral membrane protein and possesses heparan sulfate glucosaminyl 3-O-sulfotransferase activity. This gene is expressed predominantly in brain and may play a role in the nervous system

unknown J Histochem Cytochem. 2009 May;57(5):477-89, PLoS One. 2009 Sep 11;4(9):e6986, Gynecol Oncol. 2007 Dec;107(3):549-53, Nat Genet. 2007 Feb;39(2):232-6, Oncogene. 2003 Jan 16;22(2):274-80.

FLT3 0.80 0.08 1.61E-19 no class III receptor tyrosine kinase that regulates hematopoiesis, activated receptor kinase subsequently phosphorylates and activates multiple cytoplasmic effector molecules in pathways involved in apoptosis, proliferation, and differentiation of hematopoietic cells in bone marrow

unknown na

TWIST1 0.70 0.18 2.76E-19 methylated in breast, bladder, gastric, lung, cervical,

Basic helix-loop-helix (bHLH) transcription factors have been implicated in cell lineage determination and differentiation

TWIST1 overexpression is associated with nodal invasion and male sex in primary colorectal cancer.

Eur Urol. 2009 Aug 5, Cancer Epidemiol Biomarkers Prev. 2008 Dec;17(12):3325-30, Lab Invest. 2008 Feb;88(2):161-70, Mol Cancer. 2007 Oct 29;6:70, Cancer Epidemiol Biomarkers Prev. 2007 Jun;16(6):1178-84, Ann Surg Oncol. 2009 Jan;16(1):78-87

Appendices 134

EPHA5 0.57 0.17 7.89E-19 methylated in breast Ca ephrin receptor subfamily of the protein-tyrosine kinase family. EPH and EPH-related receptors have been implicated in mediating developmental events, particularly in the nervous system

unknown Hum Pathol. 2009 Sep 4

NRG1 0.41 0.06 6.10E-18 no a signaling protein that mediates cell-cell interactions and plays critical roles in the growth and development of multiple organ systems

unknown na

SEZ6L 0.61 0.05 7.08E-18 methylated in Gastric Ca,

unknown unknown Lab Invest. 2008 Feb;88(2):161-70

AGTR1 0.66 0.09 1.03E-17 no Angiotensin II is a potent vasopressor hormone and a primary regulator of aldosterone secretion. It is an important effector controlling blood pressure and volume in the cardiovascular system.

unknown na

CD40 0.66 0.31 1.02E-16 no TNF-receptor superfamily. This receptor has been found to be essential in mediating a broad variety of immune and inflammatory responses including T cell-dependent immunoglobulin class switching, memory B cell development, and germinal center formation

CD40 is expressed in a proportion of established CRC lines in culture and that receptor expression is functional. Activation of CD40 by membrane-presented CD40L causes high levels of death in CD40-positive CRC cells and induces secretion of proinflammatory cytokines

Int J Cancer. 2007 Sep 15;121(6):1373-81

EYA4 0.50 0.18 2.41E-16 methylated in ulcerative-colitis-related dysplasia, Barrett's esophagus and esophageal adenocarcinoma, colon Ca

aberrant expression in the steady state level of EYA4 triggers distinct apoptotic mechanisms, implicating these genes as possible regulators of programmed cell death

Methylated in ulcerative colitis-associated dysplasia, CRC

Clin Gastroenterol Hepatol. 2006 Feb;4(2):212-8, Cancer Epidemiol Biomarkers Prev. 2005 Apr;14(4):830-4, Nucleic Acids Res. 2006 May 2;34(8):e59

NOS3 0.54 0.86 3.94E-16 no Nitric oxide is a reactive free radical which acts as a biologic mediator in several processes, including neurotransmission and antimicrobial and antitumoural activities

eNOS-deficient mice had greater leukocyte infiltration, gut injury, and expressed higher levels of the mucosal addressin, MAdCAM-1. These results demonstrate that eNOS plays an important role in limiting injury to the intestine during experimental colitis and altered eNOS content and/or activity may contribute to human IBD. The number of NOS-3-immunoreactive vascular profiles increased in the lamina propria of UC colon, overexpression of NOS3 may correlate with tumor growth and vascular invasion

Free Radic Biol Med. 2003 Dec 15;35(12):1679-87, Scand J Gastroenterol. 2001 Feb;36(2):180-9, Virchows Arch. 2000 Feb;436(2):109-14,

EPHA7 0.56 0.18 6.99E-16 methylated in CRC, prostate, gastric Ca, lymphoma

ephrin receptor subfamily of the protein-tyrosine kinase family. EPH and EPH-related receptors have been implicated in mediating developmental events, particularly in the

hypermethylation of colorectal cancers was more frequent in male than in female (P=0.0078), and in moderately differentiated than in well-differentiated adenocarcinomas

Oncogene. 2005 Aug 25;24(36):5637-47, Int J Cancer. 2009 Jan 1;124(1):88-94, Hum Pathol. 2007 Nov;38(11):1649-

Appendices 135

nervous system (P=0.0361). There was a tendency that hypermethylation in rectal cancers was more frequent than in colon cancers (P=0.0816). Hypermethylation was also observed in colorectal adenomas

56, Oncogene. 2007 Jun 21;26(29):4243-52

CSPG2 0.73 0.13 1.16E-15 methylated in astrocytoma, hepatocellular carcinoma, dysplastic epithelium from high-grade dysplasia (HGD)/cancer patients with UC

a member of the aggrecan/versican proteoglycan family. The protein encoded is a large chondroitin sulfate proteoglycan and is a major component of the extracellular matrix. This protein is involved in cell adhesion, proliferation, proliferation, migration and angiogenesis and plays a central role in tissue morphogenesis and maintenance.

CSPG2 is directly transactivated by p53 BMC Cancer. 2004 Sep 14;4:65, Cell Res. 2003 Oct;13(5):319-33, Cancer Res. 2001 May 1;61(9):3573-7, Proc Natl Acad Sci U S A. 2002 Nov 26;99(24):15632-7

RASGRF1 0.73 0.16 1.48E-15 no a guanine nucleotide exchange factor that stimulates the dissociation of GDP from RAS protein

unknown na

NGFB 0.82 0.41 1.82E-15 no This protein has nerve growth stimulating activity and the complex is involved in the regulation of growth and the differentiation of sympathetic and certain sensory neurons

unknown na

RBP1 0.49 0.06 2.12E-15 methylated in esophageal SCC, B-cell lymphoma, intrahepatic cholangiocarcinoma, prostate Ca, gastric Ca, CRC

encodes the carrier protein involved in the transport of retinol (vitamin A alcohol) from the liver storage site to peripheral tissue. Vitamin A is a fat-soluble vitamin necessary for growth, reproduction, differentiation of epithelial tissues, and vision

RBP1 is methylated in the aberrant crypt foci and tumour of CRC patients

Int J Cancer. 2005 Jul 10;115(5):747-51, Oncol Rep. 2009 Apr;21(4):1067-73, Leukemia. 2008 May;22(5):1035-43, Arch Pathol Lab Med. 2007 Jun;131(6):923-30, J Pathol. 2007 Feb;211(3):269-77, Cancer. 2005 Oct 15;104(8):1609-19

ALK 0.63 0.27 4.18E-15 no The 2;5 chromosomal translocation is frequently associated with anaplastic large cell lymphomas (ALCLs). The translocation creates a fusion gene consisting of the ALK (anaplastic lymphoma kinase) gene and the nucleophosmin (NPM) gene: ALK plays an important role in the development of the brain and exerts its effects on specific neurons in the nervous system

echinoderm microtubule-associated protein-like 4-anaplastic lymphoma kinase (EML4-ALK) fusion gene is an oncogene in CRC

Mol Cancer Res. 2009 Sep;7(9):1466-76

MYH11 0.72 0.14 4.93E-15 no It functions as a major contractile protein, converting chemical energy into mechanical energy through the hydrolysis of ATP.

mutated & differentially expressed in CRC of HNPCC patients, somatic protein-elongating frameshift mutations in 55% of CRCs displaying microsatellite instability and in the germ-line of one individual with PJS. All mutations resulted in unregulated molecules displaying constitutive motor activity. Unregulated MYH11 may affect the cellular

Br J Cancer. 2008 Nov 18;99(10):1726-8, Proc Natl Acad Sci U S A. 2008 Apr 8;105(14):5513-8, J Surg Res. 2008 Jan;144(1):29-35

Appendices 136

energy balance or disturb cell lineage decisions in tumour progenitor cells, downregulate in MSS-CRC

NPY 0.52 0.18 5.22E-15 no a neurotransmitter expressed in submucous and myenteric nerves

decrease in vasoconstrictor neurotransmitters NPY around submucosal arterioles of both early and advanced polyps, Gastrointestinal secretion, absorption, motility, cell proliferation, local immune defense and blood flow are all regulated by the neuroendocrine peptides, In 12-month-old mice, the concentrations of PYY, somatostatin, VIP, NPY, galanin and neurotensin decreased compared with those in 3-month-old mice

Colorectal Dis. 2006 Mar;8(3):230-4, Gerontology. 1999 Jan-Feb;45(1):17-22

LMO1 0.61 0.17 6.89E-15 no It is mapped to an area of consistent chromosomal translocation in chromosome 11, disrupting it in T-cell leukemia

unknown na

PEG10 0.69 0.22 7.40E-15 no the imprinted gene paternally expressed gene-10 (PEG10) has been reported to support proliferation in hepatocellular carcinomas

unknown na

GAS7 0.67 0.11 2.12E-14 no expressed primarily in terminally differentiated brain cells and predominantly in mature cerebellar Purkinje neurons. GAS7 plays a putative role in neuronal development

Expression downregulated in MSI+ CRC Mol Cancer. 2007 Aug 23;6:54

ADAMTS12 0.65 0.24 2.28E-14 methylated in cancer cell lines and CRC,

It may play roles in pulmonary cells during fetal development or in tumour processes through its proteolytic activity or as a molecule potentially involved in regulation of cell adhesion

Anti-tumour protease that can reduce the proliferative properties of tumour cells. This function is lost by epigenetic silencing in tumour cells, but concurrently induced in stromal cells, probably as part of a response of the normal tissue aimed at controlling the progression of cancer.

J Cell Sci. 2009 Aug 15;122(Pt 16):2906-13

PTPRH 0.64 0.91 2.44E-14 methylated in breast Ca a member of the protein tyrosine phosphatase (PTP) family. PTPs are known to be signaling molecules that regulate a variety of cellular processes including cell growth, differentiation, mitotic cycle, and oncogenic transformation.

Overexpressed in CRC Mol Endocrinol. 2009 Feb;23(2):176-87; Biochem Biophys Res Commun. 1997 Feb 24;231(3):705-11

IRAK3 0.54 0.09 2.50E-14 no IRAK-M is a negative regulator of Toll-like receptor signalling , deletion of IRAK-M enhances host anti-tumour immune response

unknown Cell. 2002 Jul 26;110(2):191-202, Mol Immunol. 2007 Jul;44(14):3453-61.

CDH13 0.69 0.16 5.09E-14 methylated in CRC, endometrial, prostate, breast Ca

a calcium dependent cell-cell adhesion glycoprotein, a putative mediator of cell-cell interaction in the heart and may act as a negative regulator of neural cell growth

CDH13 expression is frequently silenced by aberrant methylation in colorectal cancers and adenomas and that methylation of CDH13 commences at an early stage of multistep colorectal tumourigenesis; specifically methylated in poorly differentiated CRC; methylated in CRC without lymph node

Cancer Res. 2002 Jun 15;62(12):3382-6; Br J Cancer. 2004 Mar 8;90(5):1030-3; Dis Colon Rectum. 2005 Jun;48(6):1282-6; Anticancer Res. 2006 Jan-Feb;26(1A):55-8

Appendices 137

metastasis

FGF3 0.27 0.02 1.15E-13 no possess broad mitogenic and cell survival activities and are involved in a variety of biological processes including embryonic development, cell growth, morphogenesis, tissue repair, tumour growth and invasion. This gene was identified by its similarity with mouse fgf3/int-2, a proto-oncogene activated in virally induced mammary tumours in the mouse. Frequent amplification of this gene has been found in human tumours, which may be important for neoplastic transformation and tumour progression

FGF-3 is activated in tumours induced in nude mice by MYC-transfected cells from non-tumourigenic clones. However, in most of the cell lines established from these tumours, FGF-3 expression tends to be lost upon in vitro propagation, The FGF-3 gene is constitutively expressed in tumourigenic clones from the SW613-S human colon carcinoma cell line but is silent in non-tumourigenic clones, Int-2 was positively detected in only four tumours (i.e. 5% of the cases examined).

Oncogene. 1995 Jun 15;10(12):2331-42, J Biol Chem. 2000 Jun 9;275(23):17364-73, J Gastroenterol Hepatol. 2002 Oct;17(10):1084-6

ADCYAP1 0.60 0.17 1.35E-13 no adenylate cyclase activating polypeptide 1. Mediated by adenylate cyclase activating polypeptide 1 receptors, this polypeptide stimulates adenylate cyclase and subsequently increases the cAMP level in target cells. Adenylate cyclase activating polypeptide 1 is not only a hypophysiotropic hormone, but also functions as a neurotransmitter and neuromodulator. In addition, it plays a role in paracrine and autocrine regulation of certain types of cells

60% of the PACAP KO mice developed colorectal tumours with an aggressive-appearing pathology, PACAP is capable of increasing the number of viable cells and regulating Fas-R expression in a human colonic cancer cell line, suggesting that PACAP might play a role in the regulation of colon cancer growth and modulation of T lymphocyte anti-tumoural response via the Fas-R/Fas-L apoptotic pathway, Pituitary Adenylate-Cyclase Activating Polypeptide (PACAP), were expressed in 4 human colonic adenocarcinoma cell lines, HT29, SW403, DLD-1 and Caco-2, that spontaneously displayed variable phenotypic properties in culture

Int J Cancer. 2008 Apr 15;122(8):1803-9, Regul Pept. 2002 Nov 15;109(1-3):115-25, Cell Signal. 1998 Jan;10(1):13-26

SFRP1 0.52 0.14 1.45E-13 methylated in renal, breast, NSCLC, ovarian, esophageal, HCC, oral squamous cell carcinoma, CRC

a member of the SFRP family that contains a cysteine-rich domain homologous to the putative Wnt-binding site of Frizzled proteins. Members of this family act as soluble modulators of Wnt signaling; epigenetic silencing of SFRP genes leads to deregulated activation of the Wnt-pathway which is associated with cancer

methylated in CRC, associated with progression from inflammatory bowel disease (IBD) to IBD-related neoplasia, age-related methylation in normal colonic mucosa of CRC patients, 52% of syndromic hyperplastic polyps showed a reproducible and distinct staining pattern for secreted Frizzled receptor protein 1 that was not seen in control specimens and that was associated with larger polyp size and location in the proximal colon, the reduced activity or absence of sFRP1 allows the transduction of noncanonical Wnt signals, which contribute to tumour progression.

Cancer Res. 2004 Feb 1;64(3):883-8, Arch Pathol Lab Med. 2004 Sep;128(9):967-73, Br J Cancer. 2008 Jul 8;99(1):136-42, Hum Mol Genet. 2009 Apr 1;18(7):1332-42, Gynecol Oncol. 2009 Feb;112(2):301-6, Mol Cancer. 2008 Nov 6;7:83, Int J Cancer. 2009 Jan 15;124(2):387-93, Mol Cancer. 2008 Oct 2;7:75, Oncogene. 2007 Aug 16;26(38):5680-91, J Gastroenterol. 2008;43(5):378-89, Int J Oncol. 2008 Jun;32(6):1253-61, J Gastrointest Surg. 2008 Oct;12(10):1745-53

Appendices 138

VIM 0.46 0.11 3.00E-13 methylated in CRC a member of the intermediate filament family. Intermediate filamentents, along with microtubules and actin microfilaments, make up the cytoskeleton. The protein encoded by this gene is responsible for maintaining cell shape, integrity of the cytoplasm, and stabilizing cytoskeletal interactions. It is also involved in the immune response, and controls the transport of low-density lipoprotein (LDL)-derived cholesterol from a lysosome to the site of esterification. It functions as an organizer of a number of critical proteins involved in attachment, migration, and cell signaling

in HCT116 colon cancer cells treated with selenomethionine, VIM expression is increased

Clin Chem. 2006 Dec;52(12): Cancer Biol Ther. 2007 Apr;6(4):494-503

FLT4 0.62 0.22 3.29E-13 no a tyrosine kinase receptor for vascular endothelial growth factors C and D. The protein is thought to be involved in lymphangiogenesis and maintenance of the lymphatic endothelium

VEGF-D is a mitogen for endothelial cells, VEGF-D is a ligand for VEGFR-3 (Flt4) and can activate these receptors

Proc Natl Acad Sci U S A. 1998 Jan 20;95(2):548-53

GALR1 0.43 0.11 3.80E-13 methylated in HNSCC GALR1 inhibits adenylyl cyclase via a G protein of the Gi/Go family. GALR1 is widely expressed in the brain and spinal cord, as well as in peripheral sites such as the small intestine and heart

Expressed at the highest level in the large intestine of mice

Clin Cancer Res. 2008 Dec 1;14(23):7604-13, Neuropeptides. 2005 Jun;39(3):349-52,

CDH11 0.45 0.05 4.50E-13 no a type II classical cadherin from the cadherin superfamily, integral membrane proteins that mediate calcium-dependent cell-cell adhesion, Expression of this particular cadherin in osteoblastic cell lines, and its upregulation during differentiation, suggests a specific function in bone development and maintenance

Up-regulated in inflammatory bowel disease PLoS Med. 2005 Aug;2(8):e199

SRC 0.75 0.92 4.59E-13 no This proto-oncogene may play a role in the regulation of embryonic development and cell growth. The protein encoded by this gene is a tyrosine-protein kinase whose activity can be inhibited by phosphorylation by c-SRC kinase. Mutations in this gene could be involved in the malignant progression of colon cancer

dominant acting protooncogene in CRC, Induction of cyclooxygenase-2 overexpression in human gastric epithelial cells by Helicobacter pylori involves TLR2/TLR9 and c-Src-dependent nuclear factor-kappaB activation

Cancer. 1992 Sep 1;70(5 Suppl):1216-21, Mol Pharmacol. 2004 Dec;66(6):1465-77

MMP2 0.47 0.11 8.14E-13 methylated in CRC, lung, breast Ca,

This gene encodes an enzyme which degrades type IV collagen, the major structural component of basement membranes. The enzyme plays a role in endometrial menstrual breakdown, regulation of vascularization and the inflammatory response, MMP-2 protects against tissue damage and maintains gut barrier function

MMP2 activation may be required for tumour invasion in CRC

PLoS One. 2009 Sep 11;4(9):e7012, Am J Physiol Gastrointest Liver Physiol. 2009 Feb;296(2):G175-84, Mol Med. 2000 May;6(5):450-60

Appendices 139

WNT2 0.74 0.28 8.28E-13 no The WNT gene family consists of structurally related genes which encode secreted signaling proteins. These proteins have been implicated in oncogenesis and in several developmental processes, including regulation of cell fate and patterning during embryogenesis

WNT2 is overexpressed in CRC & Barrett's esophagus, WNT2 abrogates Fz4 expression in some CRC cell lines

Oncogene. 2006 May 18;25(21):3084-92, Neoplasma. 2009;56(2):119-23

SCGB3A1 0.78 0.19 1.08E-12 methylated in nasopharyngeal, oral, esophageal, breast, lung, gastric, testicular germ cell, ovarian Ca and neuroblastoma, CRC

HIN-1 is a potent inhibitor of anchorage-dependent and anchorage-independent cell growth, cell migration, and invasion. Expression of HIN-1 in synchronized cells inhibits cell cycle reentry and the phosphorylation of the retinoblastoma protein (Rb), whereas in exponentially growing cells, HIN-1 induces apoptosis without apparent cell cycle arrest and effect on Rb phosphorylation

Methylated in CRC, correlate with MSI+, Mol Cancer. 2008 Dec 31;7:94, Clin Cancer Res. 2009 Jun 15;15(12):4174-80, Cancer Epidemiol Biomarkers Prev. 2009 Mar;18(3):901-14, Epigenetics. 2008 Nov;3(6):336-41, Lab Invest. 2008 Feb;88(2):161-70, APMIS. 2007 Oct;115(10):1147-60, Mol Cancer. 2007 Jul 10;6:45, Clin Cancer Res. 2007 Jun 1;13(11):3191-7, Cancer Res. 2005 Nov 1;65(21):9659-69

WT1 0.71 0.13 1.60E-12 methylated in CRC, cervical, ovarian, breast, astrocytoma, hypomethylated in renal cancer,

It has an essential role in the normal development of the urogenital system, and it is mutated in a small subset of patients with Wilm's tumours

Expression of WT1 in CRC is inconsistent in different studies - 1. WT1 is overexpressed in CRC, no mutation/deletion found. 2. The expression of WT1 gene, as revealed by in situ hybridization, showed no differences between normal colonic mucosa and malignant carcinoma. The transcriptional activity of WT1 proteins and their ability to function as tumour suppressors or oncogenes depends on the cellular status of p53.

Stem Cells. 2008 Jul;26(7):1808-17, Int J Cancer. 2008 Jul 1;123(1):161-7, Cancer. 2005 Nov 1;104(9):1924-30, World J Gastroenterol. 2004 Dec 1;10(23):3441-54, BMC Cancer. 2004 Sep 14;4:65, Br J Cancer. 1997;76(9):1124-30, Cancer Sci. 2003 Aug;94(8):712-7, J Biol Chem. 2003 Jan 31;278(5):3474-82, Br J Cancer. 1997;76(9):1124-30

TFPI2 0.35 0.07 1.60E-12 methylated in CRC, esophageal, gastric, melanoma, cervical, hepatocellular Ca

the expression of TFPI-2 diminishes with an increasing degree of malignancy, which may suggest a role for TFPI-2 in the maintenance of tumour stability and inhibition of the growth of neoplasms

TFPI2 methylation was detected in stool DNA from stage I to III CRC patients

Cancer Res. 2009 Jun 1;69(11):4691-9, Oncol Rep. 2009 Apr;21(4):1067-73, Clin Cancer Res. 2009 Mar 1;15(5):1801-7, Eur J Cancer. 2009 May;45(7):1282-93, Cancer Epidemiol Biomarkers Prev. 2006 Jan;15(1):114-23, Hepatology. 2007 May;45(5):1129-38, Thromb Haemost. 2003 Jul;90(1):140-6

IGFBP3 0.69 0.30 1.87E-12 methylated in hepatocellular, prostate, renal, gastric, colorectal Ca

a member of the insulin-like growth factor binding protein (IGFBP) family, circulates in the plasma, prolonging the half-life of IGFs and altering their interaction with cell surface

Methylated in CIMP-low CRC at low level, IGFBP3 methylation is inversely associated with MSI in CIMP-high colorectal cancers, and this relationship is limited to p53-negative

Neoplasia. 2007 Dec;9(12):1091-8, Cancer Sci. 2006 Jan;97(1):64-71, Mod Pathol. 2008 Mar;21(3):245-55, J Clin Invest.

Appendices 140

receptors. induced by wild-type p53, regulates IGF and interacts with the TGF-beta pathway

tumours 2007 Sep;117(9):2713-22, Br J Cancer. 2007 May 21;96(10):1587-94, Cancer Res. 2006 May 15;66(10):5021-8,

GLI3 0.51 0.15 1.90E-12 no DNA-binding transcription factors and are mediators of Sonic hedgehog (Shh) signaling

expression of wild-type SMO is required for expression of GLI3 by a mechanism that is independent of conventional Hedgehog signalling, the repressive form of Gli3, characteristic of an inactive pathway, was detected in SW480 and Colo320 cells,

Cancer Lett. 2004 Apr 30;207(2):205-14, Int J Cancer. 2007 Dec 15;121(12):2622-7

CCKBR 0.58 0.08 1.91E-12 no a G-protein coupled receptor for gastrin and cholecystokinin (CCK), regulatory peptides of the brain and gastrointestinal tract. This protein is a type B gastrin receptor, which has a high affinity for both sulfated and nonsulfated CCK analogs and is found principally in the central nervous system and the gastrointestinal tract.

CCKBR was present in 96% of polyps. Expression of gastrin and CCKBR was seen in all histological types and sizes of polyps. Activation of cholecystokinin-2 receptor (CCK2R) by gastrin stimulates a rapid activation of focal adhesion kinase (FAK) pathway in human colon cancer cells. CCK2R regulating invasion and motility of colon cancer cells, and support a role of CCK2R in the progression of colon cancer. FAK play a critical role in this CCK2R-mediated effect

Gut. 2000 Dec;47(6):820-4, Int J Cancer. 2006 Dec 15;119(12):2724-32

TBX1 0.77 0.24 2.12E-12 no regulation of developmental processes unknown na

MME 0.65 0.11 2.55E-12 no a common acute lymphocytic leukemia antigen that is an important cell surface marker in the diagnosis of human acute lymphocytic leukemia (ALL). This protein is present on leukemic cells of pre-B phenotype, which represent 85% of cases of ALL, It is a glycoprotein that is particularly abundant in kidney.

CD10 expression is an integral part of colorectal carcinogenesis and seems to contribute to the invasion and thus probably facilitates metastasis, CD10 expression was detected in more than half of the cases of non-polypoid growth colorectal neoplasms

Hum Pathol. 2002 Aug;33(8):806-11, Histopathology. 2008 Apr;52(5):569-77

DBC1 0.71 0.25 3.23E-12 methylated in hematological malignancies, non-small cell lung cancer, oral squamous cell carcinoma, bladder cancer

unknown lost in many cancer cell lines and promotes TP53-mediated apoptosis through specific inhibition of SIRT1

PLoS One. 2009 Sep 11;4(9):e6986, Mod Pathol. 2008 May;21(5):632-8, Hum Mol Genet. 2005 Apr 15;14(8):997-1007, Br J Cancer. 2004 Aug 16;91(4):760-4, Oncogene. 2001 Jan 25;20(4):531-7, Genes Chromosomes Cancer. 2009 Aug 11;48(11):953-962

DES 0.42 0.25 3.80E-12 no a muscle-specific class III intermediate filament. Homopolymers of this protein form a stable intracytoplasmic filamentous network connecting myofibrils to each other and to the plasma membrane

DES and ENO1 act as the central elements in colon cancer susceptibility, and protein biosynthesis and the ribosome-associated function categories largely account for the colon cancer tumuorigenesis

BMC Syst Biol. 2008 Aug 10;2:72

SLIT2 0.78 0.29 4.10E-12 methylated in leukemia, breast, liver, lung,

a group of secreted glycoproteins that play a role in the regulation of cell migration

SLIT2 axon guidance molecule is frequently inactivated in colorectal cancer and suppresses

Neoplasia. 2008 Dec;10(12):1411-20, Cancer Res.

Appendices 141

cervical, renal Ca, glioma, CRC

growth of colorectal carcinoma cells. 2003 Mar 1;63(5):1054-8, Epigenetics. 2009 May;4(4):265-9, J Histochem Cytochem. 2009 May;57(5):477-89, Biochem Biophys Res Commun. 2009 Jan 30;379(1):86-91, Br J Cancer. 2004 Dec 13;91(12):2071-8, Oncogene. 2003 Jul 17;22(29):4611-6., Int J Cancer. 2003 Jan 20;103(3):306-15,

PDE1B 0.60 0.12 5.69E-12 no PDE1B2 regulates cGMP and a subset of the phenotypic characteristics acquired upon macrophage differentiation from a monocyte, likely to regulate cGMP in macrophages

unknown Proc Natl Acad Sci U S A. 2005 Jan 11;102(2):497-502, Proc Natl Acad Sci U S A. 2006 Jan 10;103(2):460-5

GABRB3 0.88 0.62 1.01E-11 no receptor for gamma-aminobutyric acid, the major inhibitory transmitter of the nervous system. This gene is located on the long arm of chromosome 15 in a cluster with two genes encoding related subunits of the family. Mutations in this gene may be associated with the pathogenesis of Angelman syndrome, Prader-Willi syndrome, and autism

unknown na

TMEFF2 0.75 0.19 1.62E-11 methylated in cholangiocarcinoma, gastric, esophageal, lung Ca, CRC

TMEFF2 contributes to cell proliferation in an ADAM17-dependent autocrine fashion in cells expressing this protein

HPP1 demonstrates tumour suppressive and pro-apoptotic activity, both in vitro and in vivo, activation of the STAT1 pathway likely represents the principal mediator of HPP1's tumour suppressive properties; frequent TPEF methylation in primary colorectal cancers and liver metastases, indicating that epigenetic alterations are not only present in the early phases of carcinogenesis, but are also common in metastatic lesions; methylation of HPP1 is a relatively common early event in UC-associated carcinogenesis;

J Biol Chem. 2007 Dec 28;282(52):37378-88, Virchows Arch. 2009 Sep 10, Mol Cancer. 2008 Oct 2;7:75, Dig Liver Dis. 2008 Dec;40(12):920-6, Lung Cancer. 2009 May;64(2):155-9, Clin Chem. 2008 Feb;54(2):414-23, Int J Cancer. 2006 Sep 15;119(6):1298-302, Neoplasia. 2005 Aug;7(8):771-8, Gastroenterology. 2005 Jul;129(1):74-85, Proc Natl Acad Sci U S A. 2001 Jan 2;98(1):265-70, Int J Cancer. 2008 Apr 1;122(7):1567-72, Cancer Res. 2002 Dec 1;62(23):6820-2

KDR 0.77 0.26 1.63E-11 methylated in stomach cancer, colon cancer and hepatocellular carcinoma

a type III receptor tyrosine kinase, VEGF receptor, mediates VEGF-induced endothelial proliferation, survival, migration, tubular morphogenesis and sprouting.

Expression of VEGF and KDR was higher in metastatic than in nonmetastatic neoplasms and directly correlated with the extent of neovascularization and the degree of proliferation

Epigenetics. 2009 Jul;4(5):313-21, Cancer Res. 1995 Sep 15;55(18):3964-8

FLI1 0.82 0.22 1.82E-11 no unknown presence of EWS-FLI1 chimeric mRNA in tumour arising in the mesentery helps distinguish pPNET fm other tumours

J Gastroenterol. 2002;37(7):543-9, Surg Today. 2006;36(2):193-7

Appendices 142

(Peripheral primitive neuroectodermal tumours (pPNETs - usually found in the soft tissue of the extremities, paravertebral region).

SPI1 0.62 0.77 3.49E-11 methylated in diffuse large B-cell lymphoma

an ETS-domain transcription factor that activates gene expression during myeloid and B-lymphoid cell development

unknown Int J Biochem Cell Biol. 2007;39(7-8):1523-38

WNT8B 0.74 0.93 5.28E-11 no Secreted signaling protein implicated in oncogenesis and in several developmental processes, including regulation of cell fate and patterning during embryogenesis, The expression patterns of the human and mouse genes appear identical and are restricted to the developing brain

unknown na

PALM2-AKAP2

0.59 0.17 6.39E-11 no a naturally occurring co-transcribed product of the neighboring PALM2 and AKAP2 genes. The significance of this co-transcribed mRNA and the function of its protein product have not yet been determined

unknown na

CHGA 0.67 0.19 9.34E-11 no a member of the chromogranin/secretogranin family of neuroendocrine secretory proteins. It is found in secretory vesicles of neurons and endocrine cells. This gene product is a precursor to three biologically active peptides; vasostatin, pancreastatin, and parastatin. These peptides act as autocrine or paracrine negative modulators of the neuroendocrine system

chromogranin A expression in some colon carcinomas suggests that a previously unrecognized subgroup of these tumours has neuroendocrine features, NE cell component, either diffusely scattered or occasional, occurs in about 15% of gastric and colorectal tumours; (2) there is no correlation between the presence of CgA-positive neuroendocrine (NE) cells and degree of tumour differentiation, Elevated serum chromogranin A is detectable in patients with carcinomas at advanced disease stages, CGA overexpression could reflect a more aggressive tumour

J Clin Invest. 1988 Aug;82(2):686-90, Int J Cancer. 1992 May 8;51(2):189-94, Ann Clin Lab Sci. 2000 Apr;30(2):175-8, Anticancer Res. 2002 Jan-Feb;22(1A):395-8, Hepatogastroenterology. 2005 May-Jun;52(63):731-41

SGCE 0.28 0.09 1.09E-10 no Sarcoglycans are transmembrane components in the dystrophin-glycoprotein complex which help stabilize the muscle fiber membranes and link the muscle cytoskeleton to the extracellular matrix.

unknown na

COL1A2 0.52 0.26 1.44E-10 methylated in melanoma, bladder Ca, medulloblastoma, hepatoma

Type I is a fibril-forming collagen found in most connective tissues and is abundant in bone, cornea, dermis and tendon

Pro-inflammatory gene COL1A2 was at the active/chronic inflammatory stages in a mouse model of chronic colitis, & in murine model of chronic inflammation-induced intestinal fibrosis

Genome Res. 2009 Aug;19(8):1462-70, Int J Oncol. 2009 Jun;34(6):1593-602, Neuro Oncol. 2008 Dec;10(6):981-94, Eur J Cancer. 2005 May;41(8):1185-94, J Immunol. 2007 Nov 15;179(10):6988-7000, Gastroenterology. 2003 Dec;125(6):1750-61

TCF4 0.81 0.32 2.08E-10 methylated in Gastric The encoded protein recognizes an Ephrussi- Frameshift mutation found in 46% of MSI+ Neoplasia. 2005 Feb;7(2):99-108;

Appendices 143

Ca box ('E-box') binding site ('CANNTG') - a motif first identified in immunoglobulin enhancers. This gene is expressed predominantly in pre-B-cells, although it is found in other tissues as well

CRC, rs6983267 on 8q24 affects a binding site for the Wnt-regulated transcription factor TCF4, c-jun is a direct target of the TCF4/beta-catenin complex, the control of tcf4 expression by JNK/c-Jun leads to a positive feedback loop that connects JNK and Wnt signalling, beta-catenin/TCF4 regulates cell cycle promoting (c-MYC, CYCLIN D(1)) and inhibiting genes (p16(INK4A)) at the same time in the mesenchymally differentiated tumour cells at the front of invasion, expression of a Paneth gene programme is critically dependent on TCF4 in embryonic intestine; In colorectal cancer, mutations in Wnt cascade genes such as APC lead to the inappropriate formation of beta-catenin/Tcf4 complexes; disruption of beta-catenin/TCF4 activity in CRC cells by the overexpression of dominant-negative TCF induces rapid G1 arrest and differentiation

Carcinogenesis. 2008 Aug;29(8):1623-31, Nat Genet. 2009 Aug;41(8):885-90, EMBO J. 2009 Jul 8;28(13):1843-54, Gastroenterology. 2009 Jan;136(1):196-205.e2, Oncogene. 2006 Dec 4;25(57):7531-7, Nat Cell Biol. 2005 Apr;7(4):381-6, Cancer Cell. 2004 Jan;5(1):5-6, EMBO Rep. 2003 Jun;4(6):609-15,

DCC 0.76 0.32 2.12E-10 methylated in CRC, esophageal squamous cell carcinoma, head & neck SCC, hematological malignancies,

unknown methylated in CRC with repressive histone marks, Decreased expression of DCC may be caused by LOH or hypermethylation

Carcinogenesis. 2009 Jun;30(6):1041-8, Int J Cancer. 2008 Jun 1;122(11), Gastroenterology. 2007 Dec;133(6):1849-57, Cancer Res. 2006 Oct 1;66(19):9401-7, J Nippon Med Sch. 2005 Oct;72(5):270-7,

KRT1 0.72 0.88 2.75E-10 no The type II cytokeratins consist of basic or neutral proteins which are arranged in pairs of heterotypic keratin chains coexpressed during differentiation of simple and stratified epithelial tissues. This type II cytokeratin is specifically expressed in the spinous and granular layers of the epidermis with family member KRT10 and mutations in these genes have been associated with bullous congenital ichthyosiform erythroderma

carcinogen 4-nitroquinoline 1-oxide (4-NQO) changed the expression patterns of the intermediate filament proteins K1 in the oral cavity of carcinogen-treated mice thus may play a role in esophageal carcinogenesis

Clin Cancer Res. 2004 Jan 1;10(1 Pt 1):301-13

MYOD1 0.75 0.29 2.92E-10 methylated in CRC, leukemia, breast, bladder, astrocytoma, cervical, liver, glioma (hypo), lung, prostate, esophageal, rhabdomyosarcoma, ovarian

regulates muscle cell differentiation by inducing cell cycle arrest, a prerequisite for myogenic initiation. The protein is also involved in muscle regeneration. It activates its own transcription which may stabilize commitment to myogenesis

The methylation of MYOD1 in the normal mucosa was significantly correlated with K-ras mutation in neoplastic tissue; Normal mucosa was more highly methylated in the distal than in the proximal colon in MYOD1; Significantly increased levels of Myf-3 methylation were observed in tumours which were more invasive, located in the proximal colon or from older patients, Patients without

Hum Pathol. 2009 Sep 4, Clin Biochem. 2008 Dec;41(18):1440-8, Int J Cancer. 1999 Apr 20;84(2):109-13, Int J Mol Med. 2004 Mar;13(3):413-7, PLoS One. 2009 Sep 11;4(9):e6986, Leukemia. 2008 May;22(5):1035-43, Clin Cancer Res. 2007 Dec 15;13(24):7296-

Appendices 144

MYOD1 hypermethylation showed significantly longer survival than those with hypermethylation (p=0.0077).

304, BMC Cancer. 2004 Sep 14;4:65, Clin Cancer Res. 2004 Jan 15;10(2):565-71, Int J Cancer. 2003 Aug 10;106(1):52-9., Cancer Epidemiol Biomarkers Prev. 2002 Mar;11(3):291-7, J Natl Cancer Inst. 2001 Nov 21;93(22):1747-52, Cancer Res. 2001 Apr 15;61(8):3410-8, Am J Pathol. 1998 Apr;152(4):1071-9, Br J Cancer. 1997;75(3):396-402

THY1 0.68 0.34 4.57E-10 methylated in nasopharyngeal Ca

the function of 'adhesion molecules' in particular of Thy-1, may not only be to provide mechanical support but also regulate neutrophil functions such as extravasation and recruitment of additional neutrophils

Up-regulation of THY1 and PHLAD1 was associated with the presence of anemia in colon cancer patients

Eur J Immunol. 2008 May;38(5):1391-403, Cancer Epidemiol Biomarkers Prev. 2005 Feb;14(2):437-43, Oncogene. 2005 Sep 29;24(43):6525-32

SOX17 0.83 0.35 5.08E-10 methylated in breast Ca & CRC

a member of the SOX (SRY-related HMG-box) family of transcription factors involved in the regulation of embryonic development and in the determination of the cell fate

Sox17 plays a tumour suppressor role through suppression of Wnt signaling. However, Sox17 is induced by Wnt activation in the early stage of gastrointestinal tumourigenesis, and Sox17 is down-regulated by methylation during malignant progression. It is therefore conceivable that Sox17 protects benign tumours from malignant progression at an early stage of tumourigenesis, and down-regulation of Sox17 contributes to malignant progression through promotion of Wnt activity; SOX17 is a negative modulator of canonical Wnt signaling, and that SOX17 silencing due to promoter hypermethylation is an early event during tumourigenesis and may contribute to aberrant activation of Wnt signaling in CRC, SOX17 expression is upregulated in the inflamed mucosa of inflammatory bowel disease pts. Sox17 is antagonistic to WNT signalling, Sox17 promotes the degradation of both beta-catenin and TCF proteins via a noncanonical, glycogen synthase kinase 3beta-independent mechanism,

Gastroenterology. 2009 Jun 21, Breast Cancer Res Treat. 2009 Mar 20, Cancer Res. 2008 Apr 15;68(8):2764-72, Dig Dis Sci. 2008 Apr;53(4):1013-9; Mol Cell Biol. 2007 Nov;27(22):7802-15

FGF5 0.75 0.19 6.87E-10 no involved in a variety of biological processes, including embryonic development, cell growth, morphogenesis, tissue repair, tumour growth and invasion. This gene was identified as an oncogene, which confers transforming potential when transfected into mammalian

unknown na

Appendices 145

cells. Targeted disruption of the homolog of this gene in mouse resulted in the phenotype of abnormally long hair, which suggested a function as an inhibitor of hair elongation.

PRSS1 0.45 0.67 8.47E-10 no encodes a trypsinogen, which is a member of the trypsin family of serine proteases. This enzyme is secreted by the pancreas and cleaved to its active form in the small intestine. It is active on peptide linkages involving the carboxyl group of lysine or arginine. Mutations in this gene are associated with hereditary pancreatitis

unknown na

CCNA1 0.71 0.25 8.90E-10 methylated in colon, lung, breast, prostate

The cyclin encoded by this gene is expressed in testis and brain, & several leukemic cell lines - primarily function in the control of the germline meiotic cell cycle. This cyclin binds both CDK2 and CDC2 kinases, which give two distinct kinase activities, one appearing in S phase, the other in G2, and thus regulate separate functions in cell cycle. This cyclin was found to bind to important cell cycle regulators, such as Rb family proteins, transcription factor E2F-1, and the p21 family proteins

unknown PLoS Med. 2006 Dec;3(12):e486

TPEF 0.75 0.31 1.00E-09 methylated in cholangiocarcinoma, gastric, esophageal, lung Ca, CRC

TMEFF2 contributes to cell proliferation in an ADAM17-dependent autocrine fashion in cells expressing this protein

HPP1 demonstrates tumour suppressive and pro-apoptotic activity, both in vitro and in vivo, activation of the STAT1 pathway likely represents the principal mediator of HPP1's tumour suppressive properties; frequent TPEF methylation in primary colorectal cancers and liver metastases, indicating that epigenetic alterations are not only present in the early phases of carcinogenesis, but are also common in metastatic lesions; methylation of HPP1 is a relatively common early event in UC-associated carcinogenesis;

J Biol Chem. 2007 Dec 28;282(52):37378-88, Virchows Arch. 2009 Sep 10, Mol Cancer. 2008 Oct 2;7:75, Dig Liver Dis. 2008 Dec;40(12):920-6, Lung Cancer. 2009 May;64(2):155-9, Clin Chem. 2008 Feb;54(2):414-23, Int J Cancer. 2006 Sep 15;119(6):1298-302, Neoplasia. 2005 Aug;7(8):771-8, Gastroenterology. 2005 Jul;129(1):74-85, Proc Natl Acad Sci U S A. 2001 Jan 2;98(1):265-70, Int J Cancer. 2008 Apr 1;122(7):1567-72, Cancer Res. 2002 Dec 1;62(23):6820-2

CYP1B1 0.69 0.12 1.22E-09 methylated in CRC, gastric, melanoma, breast Ca, hypomethylated in prostate Ca

The cytochrome P450 proteins are monooxygenases which catalyze many reactions involved in drug metabolism and synthesis of cholesterol, steroids and other lipids

The increase in expression of CYP1B1 occurred not only in colorectal carcinoma and but also in adenoma, Levels of CYP1B1 mRNA were elevated in the bronchial mucosa of human tobacco smokers versus never smokers; Cytochrome P450 1B1 (CYP1B1) is

Int J Oncol. 2009 Apr;34(4):1085-91, Lab Invest. 2008 Feb;88(2):161-70, Cancer Res. 2006 Dec 1;66(23):11187-93, Clin Cancer Res. 2005 Aug 15;11(16), Cancer Res. 2004 Jun

Appendices 146

overexpressed in human colon adenocarcinomas relative to normal colon, CYP1B1 is constitutively expressed in most human tissues, including colon and breast, and can activate numerous chemically diverse carcinogens - the formation of N2-OH-PhIP, a potent mutagen implicated in the etiology of human colon and breast cancer, indicates that CYP1B1 may play an important role in PhIP-mediated carcinogenesis,

1;64(11):3807-13, Cancer Detect Prev. 2005;29(6):562-9, Carcinogenesis. 2004 Nov;25(11):2275-81, Mol Cancer Ther. 2003 Jun;2(6):527-34, Carcinogenesis. 1997 Sep;18(9):1793-8

ER 0.60 0.27 1.31E-09 methylated in CRC, breast, prostate, gastric, lung Ca, acute myeloid leukemia,

estrogen receptor, a ligand-activated transcription factor composed of several domains important for hormone binding, DNA binding, and activation of transcription

The methylation status of the ER promoter in lymph nodes of UICC stage I and II CRC patients may be a useful marker for the identification of patients at a high risk for local recurrence, Methylation of estrogen receptor 1 in colorectal adenomas is not age-dependent, but is correlated with K-ras mutation; Serum vitamin B-12 but not folate status may be associated with ERalpha promoter methylation in normal-appearing colorectal mucosa; the 'all or none' mechanism for methylation of this gene, and shows how age-dependent methylation of the ESR1 CGI leads rapidly to silencing of the gene within the cells, and hence the colonic crypt within which it occurs. Preliminary studies with a rodent model suggest the rate of age-dependent methylation of ESR1 is modifiable by dietary folate.

Br J Cancer. 2009 Jan 27;100(2):360-5, Cancer Sci. 2009 Jun;100(6):1005-11, Oncol Rep. 2008 Nov;20(5):1137-42, Leuk Lymphoma. 2008 Jun;49(6):1132-41, Int J Surg Pathol. 2007 Jul;15(3):242-51, Biochem Soc Trans. 2005 Aug;33(Pt 4):709-11

DAB2IP 0.57 0.23 1.39E-09 methylated in endometrial, liver, bladder, prostate, colorectal, lung, breast Ca

a Ras GTPase-activating protein (GAP) that acts as a tumour suppressor gene and is inactivated by methylation in prostate and breast cancers

methylation in the m2b region of DAB2IP promoter was associated with location of the tumour in the stomach. In summary, our results demonstrated that hDAB2IP methylation is frequently present in gastrointestinal tumours and that the resulting gene silencing plays an important role in gastrointestinal carcinogenesis.

Int J Cancer. 2008 Jul 15;123(2):296-302, J Hepatol. 2007 Apr;46(4):655-63, Int J Mol Med. 2006 Jan;17(1):3-13, J Biol Chem. 2005 Jun 10;280(23), Br J Cancer. 2005 Mar 28;92(6):1117-25, Int J Cancer. 2005 Jan 1;113(1):59-66, Clin Cancer Res. 2004 Mar 15;10(6):2082-9

MMP9 0.72 0.96 1.88E-09 methylated in MLL rearrangement leukaemia

degrades type IV and V collagens, a role in tumour-associated tissue remodeling

Upregulated in inflammatory bowel disease, Matrix metalloproteinase-9-mediated tissue injury overrides the protective effect of matrix metalloproteinase-2 during colitis, protein expression is upregulated in primary CRC than normal adjacent tissues, and was characteristic of colorectal cancer with high invasive metastatic potential, significantly higher expression of MMP-9 in adenoma with high grade dysplasia-colorectal cancer sequence as

Cancer Res. 2009 Feb 1;69(3):1109-16, Am J Physiol Gastrointest Liver Physiol. 2009 Feb;296(2):G175-84, Mol Med. 2000 May;6(5):450-60, Bull Exp Biol Med. 2008 Nov;146(5):616-9, Pathol Oncol Res. 2008 Mar;14(1):31-7, Oncol Rep. 2008 May;19(5):1285-91, Bull Exp Biol Med. 2007 Apr;143(4):455-

Appendices 147

compared to normal tissue; over-expression of MMP-9 strongly suggests its association with colorectal carcinogenesis; in CRC MMP-9 expression correlates with venous invasion; increased expression of MMP9 was associated significantly with low histological differentiation of the tumour, deeper tumour invasion, and was more often observed in tumours of colorectal cancer patients with unfavorable prognosis. Increased levels of MMP9 in serum predicts CRC in symptomatic patients, 1562C>T polymorphism may increase the risk of lymphatic metastasis of colorectal cancer.

8, Hum Pathol. 2007 Nov;38(11):1603-10, Br J Cancer. 2007 Oct 8;97(7):971-7, World J Gastroenterol. 2007 Sep 14;13(34):4626-9

ESR1 0.67 0.28 2.52E-09 methylated in CRC, breast, prostate, gastric, lung Ca, acute myeloid leukemia,

estrogen receptor, a ligand-activated transcription factor composed of several domains important for hormone binding, DNA binding, and activation of transcription

The methylation status of the ER promoter in lymph nodes of UICC stage I and II CRC patients may be a useful marker for the identification of patients at a high risk for local recurrence, Methylation of estrogen receptor 1 in colorectal adenomas is not age-dependent, but is correlated with K-ras mutation; Serum vitamin B-12 but not folate status may be associated with ERalpha promoter methylation in normal-appearing colorectal mucosa; the 'all or none' mechanism for methylation of this gene, and shows how age-dependent methylation of the ESR1 CGI leads rapidly to silencing of the gene within the cells, and hence the colonic crypt within which it occurs. Preliminary studies with a rodent model suggest the rate of age-dependent methylation of ESR1 is modifiable by dietary folate.

Br J Cancer. 2009 Jan 27;100(2):360-5, Cancer Sci. 2009 Jun;100(6):1005-11, Oncol Rep. 2008 Nov;20(5):1137-42, Leuk Lymphoma. 2008 Jun;49(6):1132-41, Int J Surg Pathol. 2007 Jul;15(3):242-51, Biochem Soc Trans. 2005 Aug;33(Pt 4):709-11

TUSC3 0.71 0.22 2.71E-09 no a candidate tumour suppressor gene. It is located within a homozygously deleted region of a metastatic prostate cancer. The gene is expressed in most nonlymphoid human tissues including prostate, lung, liver, and colon. Expression was also detected in many epithelial tumour cell lines.

unknown na

EPO 0.52 0.08 5.51E-09 methylated in astrocytoma

encodes a secreted, glycosylated cytokine, The protein is found in the plasma and regulates red cell production by promoting erythroid differentiation and initiating hemoglobin synthesis. This protein also has neuroprotective activity against a variety of potential brain injuries and antiapoptotic functions in several tissue types

unknown BMC Cancer. 2004 Sep 14;4:65

Appendices 148

NEFL 0.84 0.34 5.75E-09 no maintain the neuronal caliber. They may also play a role in intracellular transport to axons and dendrites.

unknown na

VAMP8 0.47 0.80 9.75E-09 no The protein encoded by this gene is a member of the vesicle-associated membrane protein (VAMP)/synaptobrevin family. It is associated with the perinuclear vesicular structures of the early endocytic compartment.

unknown na

OPCML 0.77 0.29 2.24E-08 methylated in nasopharyngeal, esophageal, lung, gastric, colon, liver, breast, cervix, prostate), lymphoma cell lines (non-Hodgkin and Hodgkin lymphoma, nasal NK/T-cell lymphoma), neuroblastoma, ovarian,

a member of the IgLON subfamily in the immunoglobulin protein superfamily. The encoded protein is localized in the plasma membrane and may have an accessory role in opioid receptor function

unknown PLoS One. 2008 Aug 20;3(8):e2990, Mol Cancer. 2008 Jul 10;7:62, Cancer Invest. 2008 Jul;26(6):569-74, Endocr Relat Cancer. 2008 Sep;15(3):777-86, Eur J Gynaecol Oncol. 2007;28(6):464-7, Mol Cancer. 2007 Oct 29;6:70, Nat Genet. 2003 Jul;34(3):337-43

EMR3 0.68 0.91 2.43E-08 no encodes a member of the class B seven-span transmembrane (TM7) receptor family expressed predominantly by cells of the immune system, This protein may play a role in myeloid-myeloid interactions during immune and inflammatory responses

unknown na

CTLA4 0.76 0.85 2.53E-08 no a member of the immunoglobulin superfamily and encodes a protein which transmits an inhibitory signal to T cells

unknown na

GSTM2 0.70 0.24 4.93E-08 methylated in Barrett's adenocarcinoma

detoxification of electrophilic compounds, including carcinogens, therapeutic drugs, environmental toxins and products of oxidative stress, by conjugation with glutathione

butyrate, an important luminal component produced from fermentation of dietary fibers, is an efficient inducer of GSTs and especially of GSTM2, to activate defence against oxidative stress in human colon cells

Carcinogenesis. 2005 Jun;26(6):1064-76, Carcinogenesis. 2003 Oct;24(10):1637-44

PGR 0.52 0.79 5.22E-08 methylated in melanoma, leukemia, liver, ovarian, colorectal, prostate, beast, cervical Ca

a member of the steroid receptor superfamily. The encoded protein mediates the physiological effects of progesterone, which plays a central role in reproductive events associated with the establishment and maintenance of pregnancy

Methylated at a very low frequency in colorectal adenoma, Overexpression of progesterone receptor B increases sensitivity of human colon muscle cells to progesterone

Melanoma Res. 2009 Jun;19(3):146-55, Blood. 2008 Aug 15;112(4):1366-73, Hepatol Res. 2007 Nov;37(11):974-83, Oncogene. 2006 Apr 27;25(18):2636-45, Cancer Res. 2006 Jan 1;66(1):29-33, Methods Inf Med. 2005;44(4):516-9, Neoplasia. 2005 Aug;7(8):748-60, Cancer Res. 2004 Jun 1;64(11):3807-13, Cancer Epidemiol Biomarkers Prev. 2005 May;14(5):1219-23, Am J

Appendices 149

Physiol Gastrointest Liver Physiol. 2008 Sep;295(3):G493-502

FGFR1 0.57 0.24 6.16E-08 hypomethylated in rhabdomyosarcoma,

binds both acidic and basic fibroblast growth factors and is involved in limb induction

Hirschsprung's disease (HD) is characterised by the absence of ganglion cells and the presence of hypertrophic nerve trunks in the distal bowel - lack of FGFRI expression in neuronal tissue of both ganglionic and aganglionic bowel; mRNA for FGF receptors 1 and 2 were expressed in both the adenoma and carcinoma cells whereas immunocytochemistry showed that the expression of the FGF R1 was reduced significantly in the carcinoma cells

Genes Chromosomes Cancer. 2007 Nov;46(11):1028-38, Pediatr Surg Int. 2001 May;17(4):299-303, Int J Cancer. 1999 Jul 19;82(2):298-304

PENK 0.74 0.38 7.81E-08 no Enkephalins and other peptide products from the precursor protein proenkephalin (PENK) act as neurotransmitters, autocrine and paracrine factors, and hormones that modulate pain, behavior, cardiac function, organogenesis, and immunity

unknown Peptides. 2008 Jan;29(1):83-92

NTSR1 0.73 0.25 1.05E-07 methylated in pancreatic Ca

mediates the multiple functions of neurotensin, such as hypotension, hyperglycemia, hypothermia, antinociception, and regulation of intestinal motility and secretion

increased NTSR1 expression may be an early event during colonic tumourigenesis and also contribute to tumour progression and aggressive behavior in colonic adenocarcinomas.

Oncogene. 2004 Nov 11;23(53):8705-10, Peptides. 2008 Sep;29(9):1609-15.

HTR1B 0.64 0.20 1.28E-07 methylated in lung Ca The neurotransmitter serotonin (5-hydroxytryptamine; 5-HT) exerts a wide variety of physiologic functions through a multiplicity of receptors and may be involved in human neuropsychiatric disorders such as anxiety, depression, or migraine.

unknown Oncogene. 2001 Nov 8;20(51):7505-13

CHFR 0.68 0.17 1.41E-07 methylated in gastric, liver, nasopharyngeal Ca, leukemia, colon Ca

activate cell cycle checkpoint when cells are treated with microtubule depolymerizing agents. Furthermore, CHFR was reported to have E3 ligase activity and promote ubiquitination and degradation of oncogenic proteins such as Aurora A and polo-like kinase 1. the FHA domain of CHFR plays an important role in initiating a cell cycle arrest at G2/M, indicating a functional link exists between the anti-proliferative effects and checkpoint function of this tumour suppressor protein via this domain

polymorphisms in the CHFR gene are associated with colorectal cancer susceptibility; promoter methylation of CHFR, as well as a high methylation index (MI), was positively related to chromosomal gain at 8q23-qter, in early colorectal cancer the CHFR gene was more frequently methylated than in advanced cases; CHFR promoter hypermethylation in colon cancer correlates with the microsatellite instability phenotype, Methylation was associated with loss of Chfr mRNA and protein expression in cancer cell lines; CpG methylation and thus silencing of CHFR depended on the activities of two DNA methyltransferases, DNMT1 and DNMT3b, as

Cancer Res. 2008 Jun 15;68(12):4597-605, PLoS One. 2008 Mar 12;3(3):e1776, Cancer Lett. 2008 Feb 18;260(1-2):170-9, Carcinogenesis. 2008 Feb;29(2):434-9, Mod Pathol. 2008 Mar;21(3):245-55, Anticancer Res. 2006 May-Jun;26(3A):1791-5, Carcinogenesis. 2005 Jun;26(6):1152-6, Proc Natl Acad Sci U S A. 2003 Jun 24;100(13):7818-23, Mol Carcinog. 2005 Aug;43(4):237-45, J Huazhong Univ Sci

Appendices 150

their genetic inactivation restored CHFR expression; low level methylation of CHFR was not persistently more prevalent in CIMP-low tumours

Technolog Med Sci. 2005;25(3):240-2., Hepatogastroenterology. 2005 Nov-Dec;52(66):1854-7, World J Gastroenterol. 2008 Aug 28;14(32):5000-7

HCK 0.81 0.46 2.18E-07 methylated in acute lymphocytic leukaemia

a protein-tyrosine kinase that is predominantly expressed in hemopoietic cell types. The encoded protein may help couple the Fc receptor to the activation of the respiratory burst. In addition, it may play a role in neutrophil migration and in the degranulation of neutrophils

not expressed in colon Ca cell line Leukemia. 2007 May;21(5):906-11, Oncogene. 1993 Oct;8(10):2627-35

CEACAM1 0.23 0.56 3.45E-07 no a member of the carcinoembryonic antigen (CEA) gene family, The encoded protein mediates cell adhesion via homophilic as well as heterophilic binding to other proteins of the subgroup. Multiple cellular activities have been attributed to the encoded protein, including roles in the differentiation and arrangement of tissue three-dimensional structure, angiogenesis, apoptosis, tumour suppression, metastasis, and the modulation of innate and adaptive immune responses

CEACAM1 acts as a regulator of apoptosis in the colonic epithelium. Thus, failure of the maturing colon cell to express CEACAM1 is likely to contribute to the development of hyperplastic lesions, which may eventually pave the way to neoplastic transformation and colon cancer development; CEACAM1 is down-regulated in colon; loss or reduced expression of the adhesion molecule BGP is a major event in colorectal carcinogenesis.

Oncogene. 2004 Dec 16;23(58):9306-13, Oncogene. 1999 Sep 30;18(40):5563-72, Proc Natl Acad Sci U S A. 1993 Nov 15;90(22):10744-8

DLK1 0.88 0.41 4.96E-07 methylated in multiple myeloma, upregulation by imprinting (hypermeth) in hepatocellular Ca,

unknown Pref-1 inhibits colonocyte differentiation and proliferation and is expressed in a subset of human colon cancer cell lines

Carcinogenesis. 2007 May;28(5):1094-103, Hum Mol Genet. 2006 Mar 15;15(6):821-30, Carcinogenesis. 2004 Nov;25(11):2239-46

SPP1 0.40 0.70 5.99E-07 no a glyco-phosphoprotein that is expressed and secreted by numerous human cancers. Opn has pivotal role in cell adhesion, chemotaxis, prevention of apoptosis, invasion, migration and anchorage-independent growth of tumour cells

OPN, a downstream effector of PI3K, protects transformed intestinal epithelial cells from programmed cell death and stimulates their anchorage-independent growth, tumour-derived OPN may enhance tumour survival by down regulating expression of NO in the local microenvironment, Ets-1 and Runx2 are critical transcriptional regulators of OPN expression in CT26 colorectal cancer cells. Suppression of these transcription factors results in significant down-regulation of the OPN metastasis protein; RNA interference stably reduces CT26 tumour expression of OPN and significantly attenuates CT26 colon cancer metastasis by diminishing tumor cell motility and invasiveness,

Front Biosci. 2008 May 1;13:4276-84, Carcinogenesis. 2007 Dec;28(12):2476-83, Surgery. 2006 Aug;140(2):132-40, J Biol Chem. 2006 Jul 14;281(28):18973-82, Carcinogenesis. 2005 Apr;26(4):741-51,

GLI2 0.84 0.93 7.28E-07 no transcription factors which bind DNA through GLI2 not expressed in CRC cell lines, Int J Cancer. 2007 Dec

Appendices 151

zinc finger motifs, mediators of Sonic hedgehog (Shh) signaling and they are implicated as potent oncogenes in the embryonal carcinoma cell, play a role during embryogenesis

suggeting aberrant activation of the Hh signaling pathway is not common in colorectal cancer cell lines; In normal colon, GLI2 expression was detected along the whole crypts, suggesting Hedgehog signaling is involved in differentiation of normal colonic tissue rather than in tumour proliferation

15;121(12):2622-7, Virchows Arch. 2009 Apr;454(4):369-79

IGF2AS 0.48 0.30 7.69E-07 no a paternally imprinted antisense transcript of the insulin-like growth factor 2 gene. The transcript is overexpressed in Wilms' tumour. This gene is predicted to be non-coding because the predicted protein is not conserved in any other species and the majority of transcripts would be candidates for non-sense mediated decay (NMD) if a protein were expressed

unknown na

PI3 0.75 0.91 1.15E-06 no elastase-specific inhibitor that functions as an antimicrobial peptide against Gram-positive and Gram-negative bacteria

unknown na

CARD15 0.34 0.64 1.61E-06 no The protein is primarily expressed in the peripheral blood leukocytes. It plays a role in the immune response to intracellular bacterial lipopolysaccharides (LPS) by recognizing the muramyl dipeptide (MDP) derived from them and activating the NFKB protein. Mutations in this gene have been associated with Crohn disease and Blau syndrome

Mutation and/or polymorphism inconsistently linked to CRC risk, NOD2 promotes epithelial cell growth,

Int J Cancer. 2005 Apr 10;114(3):433-5, Cancer Res. 2006 Mar 1;66(5):2532-5, BMC Cancer. 2007 Mar 27;7:54, BMC Cancer. 2008 Apr 23;8:112, World J Gastroenterol. 2008 Oct 14;14(38):5834-41,

MOS 0.80 0.42 1.61E-06 methylated in hematological cancer,

a serine/threonine kinase that activates the MAP kinase cascade through direct phosphorylation of the MAP kinase activator MEK

unknown

AIM2 0.64 0.86 4.20E-06 methylated in CRC a member of the IFI20X /IFI16 family. It plays a putative role in tumourigenic reversion and may control cell proliferation. Interferon-gamma induces expression of AIM2

demonstrate that inactivation of AIM2 by genetic and epigenetic mechanisms is frequent in MMR-deficient colorectal cancers, thus suggesting that AIM2 is a mutational target relevant for the progression of MSI-H colorectal cancers

Genes Chromosomes Cancer. 2007 Dec;46(12):1080-9

ABCB4 0.83 0.94 6.71E-06 methylated in breast the superfamily of ATP-binding cassette (ABC) transporters. ABC proteins transport various molecules across extra- and intra-cellular membranes, function of this protein has not yet been determined; however, it may involve transport of phospholipids from liver hepatocytes into bile; Members of the MDR/TAP subfamily are involved in multidrug resistance as well as antigen

Upregulated in S1 colon cancer cell line with acquired resistance against five cytostatic drugs

Epigenetics. 2008 Sep;3(5):270-80; Anticancer Res. 2005 Jul-Aug;25(4):2661-8

Appendices 152

presentation

TGFB2 0.88 0.36 8.43E-06 methylated in breast Ca, melanoma

regulates key mechanisms of tumour development, namely immunosuppression, metastasis, angiogenesis, and proliferation

TGFB2 is overexpressed in poorly differentiated CRC, TGFB2 expression is upregulated in cancer-associated fibroblasts in metastatic colon cancer, moderate energy restriction attenuated TGF-beta and COX protein expression and the carcinogenic process in Zucker obese rats, Steady state levels of transforming growth factor-beta1 and -beta2 mRNA and protein expression are elevated in colonic tumours in vivo irrespective of dietary lipids intervention, colon carcinoma progression is associated with gradual and significant increases in expression of TGF-beta1 and TGF-beta2 mRNA and proteins, plasma levels of both TGF-beta1 and TGF-beta2 were significantly higher in cancer patients when compared with unaffected individuals

Cancer Res. 2007 Dec 15;67(24):11517-27, Cancer Res. 2006 Jun 15;66(12):6080-6, Anticancer Res. 2006 Jul-Aug;26(4B):2901-7, Oncogene. 2004 Sep 23;23(44):7366-77, Cancer Res. 2003 Oct 15;63(20):6595-601, Int J Cancer. 2002 Aug 20;100(6):635-41, Eur J Cancer. 2001 Jan;37(2):224-33

MEST 0.11 0.37 9.32E-06 LOI in breast, lung CRC a member of the [alpha]/[beta] hydrolase fold family and has isoform specific imprinting. The loss of imprinting of this gene has been linked to certain types of cancer and may be due to promotor switching. The encoded protein may play a role in development

Loss of imprinting (LOI) of PEG1/MEST was 35% CRC, Putative loss of imprinting (LOI) of PEG1/MEST has been implicated in the aetiology of colon cancer

Int J Oncol. 2000 Aug;17(2):317-22, Hum Mol Genet. 2002 Jun 1;11(12):1449-53

BMP4 0.27 0.49 1.35E-05 no important role in the onset of endochondral bone formation in humans, and a reduction in expression has been associated with a variety of bone diseases

Bone morphogenetic protein-4 is overexpressed in colonic adenocarcinomas and promotes migration and invasion of HCT116 cells, BMP4 mRNAs was detected in peripheral blood of patients with colon cancer by RT-PCR, Modulation of Wnt gene expression by BMP4 had several functional consequences--BMP4 treatment led to activation of TCF reporters; complete activation of at least one BMP4-responsive gene required TCF sites; and treatment with a Wnt ligand was sufficient to mimic several of the phenotypic effects of BMP4 treatment. These data demonstrate the tumour suppressive properties of BMP4 signaling, show that colon cancer cells are resistant to BMP4-induced differentiation and growth suppression, BMP4 is overexpressed and secreted by human colon cancer cells with mutant adenomatous polyposis coli genes

Exp Cell Res. 2007 Mar 10;313(5):1033-44, Int J Oncol. 2004 Oct;25(4):1049-56, Cancer Biol Ther. 2004 Jul;3(7):667-75, Cancer Res. 2002 May 15;62(10):2744-8

Appendices 153

LMTK2 0.45 0.55 1.40E-05 no Cprk is expressed in a number of tissues but is enriched in brain and muscle and within the brain is found in a wide range of neuronal populations

unknown J Neurosci. 2003 Jun 15;23(12):4975-83

RIPK3 0.13 0.46 1.67E-05 methylated in lung Ca receptor-interacting protein (RIP) family of serine/threonine protein kinases, a component of the tumour necrosis factor (TNF) receptor-I signaling complex, and can induce apoptosis and weakly activate the NF-kappaB transcription factor

Overexpressed in CRC, RIP3 beta and RIP3 gamma, two novel splice variants of receptor-interacting protein 3 (RIP3), downregulate RIP3-induced apoptosis.

J Hum Genet. 2006;51(4):368-74, Biochem Biophys Res Commun. 2005 Jun 24;332(1):181-7

BDNF 0.88 0.55 1.95E-05 no Endogenous BDNF enhances the peristaltic reflex by augmenting the release of serotonin and calcitonin gene-related peptide that mediate the sensory limb of the peristaltic reflex induced by mucosal stimulation

unknown Gastroenterology. 2006 Mar;130(3):771-80

DLC1 0.45 0.69 2.80E-05 methylated in liver, breast, colon, and prostate cancers

a RhoGTPase-activating protein (RhoGAP) domain containing tumour suppressor that is often down-regulated in various cancer types

One exonic missense mutation and three intronic insertions/deletions were identified in primary colorectal tumours, knocking down of DLC-1 gene expression promotes LoVo cell migration. Our observations suggest that the DLC-1 gene is associated with LoVo cell proliferation, migration and cell cycle distribution. DLC-1 is a potential suppressor gene in the colon cancer LoVo cell line, mutations in DLC-1 may lead to loss of function and contribute to the tumourigenesis

Cancer Genet Cytogenet. 2003 Jan 15;140(2):113-7, Hum Mutat. 2000;15(2):156-65, Oncol Rep. 2008 Mar;19(3):669-74, Cancer Res. 2008 Oct 1;68(19):7718-22

EPHB6 0.38 0.14 5.16E-05 methylated in prostate, breast, neuroblastoma

Ephrin receptors and their ligands, the ephrins, mediate numerous developmental processes, particularly in the nervous system

unknown Biochem Biophys Res Commun. 2006 Apr 21;342(4):1263-72, Biochem Biophys Res Commun. 2006 Feb 3;340(1):268-76, Clin Cancer Res. 2004 Sep 1;10(17):5837-44

ALOX12 0.80 0.55 7.16E-05 no encodes an enzyme in the eicosanoid pathway in the skin that plays an essential role in the establishment and/or maintenance of the epidermal barrier function

In the human colon, arachidonic acid is metabolized primarily by cyclooxygenase (COX) and arachidonate lipoxygenase (ALOX) to bioactive lipids, which are implicated in colon cancer risk. Gln261Arg in ALOX12 was not associated with colon cancer risk in Caucasians

Carcinogenesis. 2004 Dec;25(12):2467-72

MEG3 0.76 0.55 1.04E-04 methylated in multiple myeloma, pituitary adenomas, neuroblastoma,

a maternally expressed imprinted gene which appears to function as a non-coding RNA molecule, MEG3 may interact with the cAMP-dependent signaling pathway to be involved in the control of cell proliferation and other cAMP-related physiological functions.

unknown Int J Biochem Cell Biol. 2006;38(10):1808-20, Clin Lymphoma Myeloma. 2008 Jun;8(3):171-5, J Clin Endocrinol Metab. 2008 Oct;93(10):4119-25, Br J Cancer. 2005 Apr 25;92(8):1574-80

Appendices 154

IFNG 0.44 0.75 1.11E-04 no a cytokine critical for innate and adaptive immunity against viral and intracellular bacterial infections and for tumour control

In tumour-infiltrating CD4(+) T lymphocytes from patients with colon cancer, tumour-infiltrating lymphocytes cells are inappropriately hypermethylated at IFNG, and thus not confined to the Th1 lineage. In contrast, IFNG in CD4(+) T cells from the tumour draining lymph node were significantly more demethylated than tumour-infiltrating lymphocytes --> methylation of IFNG as an epigenetic mechanism of tumor-induced immunosuppression, LIGHT sensitizes IFN-gamma-mediated apoptosis of HT-29 human carcinoma cells

J Immunol. 2008 Aug 15;181(4):2878-86, Cell Res. 2004 Apr;14(2):117-24,

ERN1 0.39 0.50 1.12E-04 no This protein possesses intrinsic kinase activity and an endoribonuclease activity and it is important in altering gene expression as a response to endoplasmic reticulum-based stress signals

unknown na

CPA4 0.29 0.45 1.17E-04 no Carboxypeptidases are zinc-containing exopeptidases that catalyze the release of carboxy-terminal amino acids, and are synthesized as zymogens that are activated by proteolytic cleavage. This gene could be involved in the histone hyperacetylation pathway. It is imprinted and may be a strong candidate gene for prostate cancer aggressiveness

unknown na

CCR5 0.28 0.49 1.30E-04 no This protein is expressed by T cells and macrophages, and is known to be an important co-receptor for macrophage-tropic virus, including HIV, to enter host cells. Defective alleles of this gene have been associated with the HIV infection resistance, Expression of this gene was also detected in a promyeloblastic cell line, suggesting that this protein may play a role in granulocyte lineage proliferation and differentiation

CCR5, receptor for CCL5 (chemotactic for monocytes/macrophages and T cells) is upregulated in CRC, active recruitment of T cells expressing CCR5 or CXCR3 into the invasive margin of colorectal cancer,

Methods Enzymol. 2009;460:105-21, Int J Cancer. 2005 Oct 10;116(6):949-56

dna methylation in colorectal cancer...dna methylation in colorectal cancer sophia pei woon ang bsc...

Documents